Clinical MR Neuroimaging: Diffusion, Perfusion and Spectroscopy

This page intentionally left blank

Clinical MR Neuroimaging Diffusion, Perfusion and Spectroscopy The physiological magnetic resonance (MR) techniques of diffusion imaging, perfusion imaging and spectroscopy offer insights into brain structure, function and metabolism. Until recently, these were mainly applied within the realm of medical research but, with their increasing availability on clinical MR imaging (MRI) machines, they are now entering clinical practice for the evaluation of neuropathology. This book provides the reader with a thorough review of the underlying physical principles of each of these methods, as well as comprehensive coverage of their clinical applications. Topics covered include single- and multiple-voxel MRS techniques; MR perfusion based on both arterial spin labeling and dynamic bolus tracking approaches; and diffusion-weighted imaging, including techniques for mapping brain white-matter fiber bundles. Clinical applications are reviewed in depth for each technique, with case reports included throughout the book. Attention is also drawn to possible artifacts and pitfalls.

To Susan for making it happen, and Charlotte and Emily for making it worthwhile JHG

To Jess, Ella and Danny ADW

To Naomi, Catherine and Stephanie for their patience and support during the preparation of this book PBB

Clinical MR Neuroimaging Diffusion, Perfusion and Spectroscopy

Edited by

Jonathan H. Gillard Department of Radiology, University of Cambridge, UK

Adam D. Waldman Department of Imaging, Charing Cross Hospital, London, UK Institute of Neurology and Imperial College of Science, Technology and Medicine, London, UK

Peter B. Barker Department of Radiology, Johns Hopkins University School of Medicine, Baltimore, USA

CAMBRIDGE UNIVERSITY PRESS

Cambridge, New York, Melbourne, Madrid, Cape Town, Singapore, São Paulo Cambridge University Press The Edinburgh Building, Cambridge CB2 8RU, UK Published in the United States of America by Cambridge University Press, New York www.cambridge.org Information on this title: www.cambridge.org/9780521824576 © Cambridge University Press 2005 This publication is in copyright. Subject to statutory exception and to the provision of relevant collective licensing agreements, no reproduction of any part may take place without the written permission of Cambridge University Press. First published in print format 2004 ISBN-13 ISBN-10

978-0-511-26447-4 eBook (EBL) 0-511-26447-X eBook (EBL)

ISBN-13 ISBN-10

978-0-521-82457-6 hardback 0-521-82457-5 hardback

Cambridge University Press has no responsibility for the persistence or accuracy of urls for external or third-party internet websites referred to in this publication, and does not guarantee that any content on such websites is, or will remain, accurate or appropriate.

Contents

List of case studies List of contributors List of abbreviations Foreword

page viii x xvii xxiii

Jonathan H. Gillard, Adam D. Waldman and Peter B. Barker

Introduction R. Nick Bryan

1

SECTION 1 PHYSIOLOGICAL MR TECHNIQUES 1 Fundamentals of MR spectroscopy Peter B. Barker

7

2 Quantification and analysis in MR spectroscopy Thomas Ernst

27

3 Artifacts and pitfalls in MR spectroscopy Ralph E. Hurd

38

4 Fundamentals of diffusion MR imaging Derek K. Jones

54

5 MR tractography using diffusion tensor MR imaging Susumu Mori and Peter van Zijl

86

6 Artifacts and pitfalls in diffusion MR imaging Martin A. Koch and David G. Norris

99

7 Cerebral perfusion imaging by exogenous contrast agents Leif Østergaard

109

v

vi

Contents

8 MRI detection of regional blood flow using arterial spin labeling Alan P. Koretsky, S. Lalith Talagala, Shella Keilholz and Afonso C. Silva

119

9 Artifacts and pitfalls in perfusion MR imaging Fernando Calamante

141

21 Perfusion MR imaging in adult neoplasia Alan Jackson

329

SECTION 4 INFECTION, INFLAMMATION AND DEMYELINATION 22 Physiological imaging in infection, 353 inflammation and demyelination: overview Robert D. Zimmerman

SECTION 2 CEREBROVASCULAR DISEASE 10 Cerebrovascular disease: overview Brian M. Tress

163

11 MR spectroscopy in stroke Peter B. Barker and Jonathan H. Gillard

168

12 Diffusion and perfusion MR in stroke Joanna M. Wardlaw

182

13 Arterial spin labeling perfusion MRI in stroke Jiongjiong Wang and John A. Detre

207

14 MR diffusion-tensor imaging in stroke 223 Pamela W. Schaefer, Luca Roccatagliata and R. Gilberto Gonzalez 15 MR spectroscopy in severe obstructive carotid artery disease Jeroen van der Grond and Dirk R. Rutgers

234

16 Perfusion and diffusion imaging in chronic carotid disease Iain D. Wilkinson

246

17 Imaging migraine pathogenesis K. Michael Welch

263

23 MR spectroscopy in intracranial infection Monika Garg and Rakesh K. Gupta

380

24 The role of diffusion-weighted imaging in intracranial infection Christopher G. Fillipi

408

25 MR spectroscopy in demyelination 429 and inflammation Gioacchino Tedeschi and Simona Bonavita 26 Diffusion imaging in demyelination and inflammation Marco Rovaris and Massimo Filippi

444

27 Physiological MR to evaluate HIV-associated brain disorders Linda Chang and Thomas Ernst

460

SECTION 5 SEIZURE DISORDERS

SECTION 3 ADULT NEOPLASIA 18 Adult neoplasia: overview Tom Mikkelsen

279

19 MR spectroscopy of brain tumors in adults Jeffry R. Alger

288

20 Diffusion MR imaging in adult neoplasia Bradford A. Moffat, Thomas L. Chenevert and Brian D. Ross

312

28 Seizure disorders: overview Thomas R. Henry

481

29 MR spectroscopy in seizure disorders Regula S. Briellmann, Mark Wellard and Graeme D. Jackson

488

30 Diffusion and perfusion MR imaging 509 in seizure disorders Konstantinos Arfanakis and Bruce P. Hermann SECTION 6 PSYCHIATRIC AND NEURODEGENERATIVE DISEASES 31 Psychiatric and neurodegenerative disease: overview Adam D. Waldman

523

32 MR spectroscopy in psychiatry John D. Port

529

Contents

33 Diffusion MR imaging in neuropsychiatry and aging Adolf Pfefferbaum and Edith V. Sullivan

558

34 MR spectroscopy in aging and dementia Kejal Kantarci and Clifford R. Jack, Jr.

579

35 MR spectroscopy in neurodegeneration C.A. Davie

594

SECTION 7 TRAUMA 36 Potential role of MR spectroscopy, 609 diffusion-weighted/diffusion-tensor imaging and perfusion-weighted imaging in traumatic brain injury: overview John D. Pickard 37 MR spectroscopy in traumatic brain injury William M. Brooks

613

38 Diffusion- and perfusion-weighted MR imaging in head injury Peter G. Bradley and David K. Menon

626

SECTION 8 PEDIATRICS 39 Physiological MR of the pediatric brain: overview Elias R. Melhem and Xavier Golay

647

40 Physiological MRI of normal development and developmental delay A. James Barkovich, Pratik Mukherjee and Daniel B. Vigneron

674

41 MR spectroscopy of hypoxic brain injury Brian Ross, Cathleen Enriquez and Alexander Lin

690

42 The role of diffusion and perfusion weighted brain imaging in neonatology Mary A. Rutherford and Serena J. Counsell

706

43 Physiological MR imaging of pediatric brain tumors Jill V. Hunter

722

44 Physiological MRI techniques and pediatric stroke Dawn Saunders, W. Kling Chong and Vijeya Ganesan

736

45 MR spectroscopy in pediatric white matter disease Folker Hanefeld, Knut Brockmann and Peter Dechent

755

46 MR spectroscopy of inborn errors of metabolism Alberto Bizzi, Marianna Bugiani and Ugo Danesi

779

Index

805

vii

List of Case Studies

Cerebrovascular disease 11.1 MRSI in acute brain ischemia

181

12.1 Diffusion tensor imaging and tissue anisotropy

204

12.2 Reversal of diffusion lesion in acute stroke

205

12.3 Diffusion and perfusion MR in subarachnoid hemorrhage

206

17.1 Stroke or migraine? An MR perfusion study

275

Adult neoplasia 19.1 Metabolic heterogeneity of glioma

306

19.2 Tumefactive multiple sclerosis – MRSI

308

19.3 MRS in meningioma

309

19.4a Recurrent astrocytoma

310

19.4b Radiation necrosis

311

20.1 DWI of epidermoids and arachnoid cysts

326

20.2 Diffusion tensor imaging of glioma infiltration

327

20.3 Differentiating gliomas from metastases 328 with DTI 21.1 Anaplastic oligodendroglioma: ASL MR perfusion

349

21.2 Radiation necrosis vs. recurrence

350 viii

List of Case Studies

Trauma

Infection, inflammation and demyelination 23.1 MRS in variant Creutzfeldt–Jakob disease

407

38.1 Diffuse axonal injury

642 643

24.1 West Nile encephalitis

427

38.2 Occult brain damage in a professional boxer

24.2 Creutzfeldt–Jacob disease: DWI

428

25.1 Acute disseminated encephalomyelitis

442

25.2 Reversible posterior leukoencephalopathy: MRSI

443

26.1 Tumefactive MS: MR perfusion

459

27.1 Progressive multifocal leukoencephalopathy

478

Seizure disorders 29.1 Rasmussen’s encephalitis: MRSI

507

29.2 Temporal lobe epilepsy: MRSI

508

30.1 Identifying an epileptic focus with DTI

520

Psychiatric and neurodegenerative diseases 34.1 MRS for investigation of Alzheimer’s disease

593

35.1 DTI in primary lateral sclerosis

605

Pediatrics 41.1 Reye’s syndrome: MRSI

704

42.1 Perinatal asphyxic injury

721

43.1 Pediatric astrocytoma

735

44.1 Moyamoya disease: MR perfusion

753

44.2 Sturge–Weber syndrome: MR perfusion imaging

754

45.1 Adrenoleukodystrophy (ALD): MRSI

778

46.1 Mitochondrial encephalopathy, lactic acidosis and stroke like episodes (MELAS)

803

ix

Contributors

Dr. Elfar Adelsteinsson, SRI International, Room BN 168, 333 Ravenswood, Menlo Park, CA 94025, USA Dr. Jeffry Alger, University of California Los Angeles, Ahmanson-Lovelace Brain, Mapping Center Room 163, 660 Charles E Young Dr South, Los Angeles, CA 90095-7085, USA Dr. Konstantinos Arfanakis, Department of Biomedical Engineering, Illinois Institute of Technology, Chicago, IL 60616-3793, USA Dr. Peter B. Barker, Department of Radiology, MRI 143C, John Hopkins Medical Institutions, 600 N Wolfe Steet, Baltimore, MD 21287, USA Dr. A. James Barkovich, Department of Radiology L371, University of California San Francisco, 505 Parnassus Avenue, PO Box 0628, San Francisco, CA 94143-0628, USA x

List of contributors

Dr. Alberto Bizzi, Department of Neuroradiology, Istituto Nazionale Neurologico, Carlo Besta, via Celoria 11, Milano 20133, Italy

Dr. R. Nick Bryan, Department of Radiology, University of Pennsylvania Health System, 3400 Spruce Street, Philadelphia, PA 19104, USA

Dr. Simona Bonavita, Institute of Neurological Sciences, Second University of Naples, Piazza Miraglia 2, Naples 80131, Italy

Dr. Marianna Bugiani, Department of Neuroradiology, Istituto Nazionale Neurologico, Carlo Besta, via Celoria 11, Milano 20133, Italy

Dr. Peter Bradley, Division of Anaesthesia, University of Cambridge, PO Box 93, Level 4, E Block, Addenbrooke’s Hospital, Hills Road, Cambridge, CB2 2QQ, UK

Dr. Fernando Calamante, Radiology and Physics Unit, Institute of Child Health, University College London, 30 Guilford Street, London, WC1N 1EH, UK

Dr. Regula Briellmann, Brain Research Institute, Austin and Repatriation Medical Centre, Banksia Street, Heidelberg West, Victoria 3084, Australia

Dr. Linda Chang, Department of Medicine, University of Hawaii, John A. Burns School of Medicine, Honolulu, HI 96813, USA

Dr. Knut Brockmann, Georg-August-Universitat, Kinderklinik and Poliklinik, Gottingen 37075, Germany

Dr. Thomas Chenevert, Departments of Radiology and Biological Chemistry, MSRB III, Room 9301, 1150 W Medical Center Dr, Ann Arbor, MI 48109-0648, USA

Dr. William Brooks, Departments of Neurology and Molecular and Integrative Physiology, University of Kansas Medical Center, 3901 Rainbow Boulevard, Kansas City, KS 66160, USA

Dr. W. Kling Chong, Department of Radiology, Great Ormond Street Hospital, Great Ormond Street, London, WC1M 3JH, UK

xi

xii

List of contributors

Serena Counsell, Robert Steiner MR Unit, Imaging Sciences Department, Clinical Sciences Centre, Imperial College, Hammersmith Hospital, London W12 OHS, UK Dr. Ugo Danesi, Department of Neuroradiology, Istituto Nazionale Neurologico, Carlo Besta, via Celoria 11, Milano 20133, Italy Dr. Charles Davie, Floor 3, Royal Free and University College Medical School, Rowland Hill Street, London, NW3 2PF, UK Dr. Peter Dechent, Biomedizinische NMR, Forschungs GmbH am Max-Planck-Institut fur biophysikalische Chemie, Gottingen 37070, Germany Dr. John Detre, Departments of Neurology and Radiology, University of Pennsylvania, Philadelphia, PA 19104, USA Dr. Cathleen Enriquez, Magnetic Resonance Spectroscopy, Huntington Medical Research Institute, 660 South Fairoaks Avenue, Pasadena, CA 91105, USA Dr. Thomas Ernst, Department of Medicine, University of Hawaii, John A. Burns School of Medicine, Honolulu, HI 96813, USA

Dr. Christopher Fillipi, Fletcher Allen Health Care-University of Vermont, Department of Radiology, Division of Neuroradiology, 111 Colchester Avenue, Burlington, Vermont 05401 Dr. Massimo Filippi, Neuroimaging Research Unit, Scientific Institute and University, Ospedale San Raffaele, Via Olgettina 60, Milan 20132, Italy Dr. Vijeya Ganesan, Great Ormond Street Hospital for Children, Great Ormond Street, London, WC1N 3JH, UK Dr. Monika Garg, Type V-B/3, SGPGIMS Campus, Raebareli Road, Lucknow 226014, India Dr. Jonathan Gillard, University Department of Radiology, Addenbrooke’s Hospital, Cambridge, CB2 2QQ, UK Dr. Xavier Golay, National Neuroscience Institute, 11 Jalan Tan Tock Jeng, Singapore, 308433 Dr. R. Gilberto Gonzalez, Department of Radiology, Massachusetts General, Hospital, Gray 2, Room B285, 55 Fruit Street, Boston, MA 02114, USA

List of contributors

Dr. Jeroen van der Grond, Department of Radiology, University Medical Center Utrecht, Heidelberglaan 100, Postbus 85500, 3508 Ga Utrecht, The Netherlands Dr. Rakesh Gupta, Sanjay Gandhi Postgraduate Institute of Medical Sciences, Raebareli Road, Lucknow 226014, India

Professor Clifford Jack, Department of Radiology, Mayo Clinic, 200 First Street SW, Rochester, MN 55905, USA Professor Alan Jackson, Imaging Sciences and Biomedical Engineering, The Medical School, University of Manchester, Oxford Road, Manchester M13 9PT, UK

Dr. Folker Hanefeld, Georg-August-Universitat, Kinderklinik and Poliklinik, Gottingen 37075, Germany

Professor Graeme Jackson, Brain Research Institute, Austing and Repatriation Medical Centre, Banksia Street, Heidelberg West, Victoria 3084, Australia

Dr. Thomas Henry, Department of Neurology, Emory University, Woodruff Memorial Building, Suite 6000 PO Drawer V, 1639 Pierce Dr, Atlanta, GA 30322, USA

Dr. Derek Jones, Section on Tissue Biophysics and Biomimetics, Laboratory of Integrative Medicine and Biophysics, National Institute of Child Health and Development, Bethesda, USA

Dr. Bruce Hermann, Department of Neurology, University of Wisconsin, Madison, WI 53792, USA

Dr. Kejal Kantarci, Department of Radiology, Mayo Clinic, 200 First St SW, Rochester, MN 55905, USA

Dr. Jill Hunter, E. B. Singleton Department of Diagnostic Imaging, Texas Children’s Hospital, MC2-2521 6621 Fannin Street, Houston, TX 77030, USA Dr. Ralph Hurd, GE Medical Systems, 333 Ravenswood Avenue, Building 307, Menlo Park, CA 94025, USA

Dr. Shella Keilholz, Laboratory of Functional and Molecular Imaging, NINDS, NIH, 10/B1D118 MSC, Bethesda, MD 20817, USA Dr. Martin Koch, Universitatsklinikum Hamburg-Eppendorf, Klinik und Poliklinik für Neurologie Martinistr 52, 20246 Hamburg, Germany

xiii

xiv

List of contributors

Dr. Alan Koretsky, Laboratory of Functional and Molecular Imaging, NINDS, NIH, 10/B1D118 MSC, Bethesda, MD 20817, USA Dr. Alexander Lin, Magnetic Resonance Spectroscopy, Huntington Medical Research Institute, 660 South Fairoaks Avenue, Pasadena, CA 91105, USA Dr. Pratik Mukherjee, Department of Radiology L371, University of California San Francisco, 505 Parnassus Avenue, PO Box 0628, San Francisco, CA 94143-0628, USA Dr. Elias Melhem, Department of Radiology, University of Pennsylvania, Health System, 3400 Spruce Street, Pennsylvania, PA 19104, USA Professor David Menon, Division of Anaesthesia, University of Cambridge, Box 93, Level 4, E Block, Addenbrooke’s Hospital, Hills Road, Cambridge, CB2 2QQ UK Dr. Tom Mikkelsen, Hermelin Brain Tumor Center, Henry Ford Hospital, K-11, Room W1138, Grand Boulevard, Detroit, MI 48202, USA

Dr. Brad Moffat, Departments of Radiology and Biological Chemistry, MSRB III, Room 9301, 1150 W Medical Center Dr, Ann Arbor, MI 48109-0648, USA

Dr. Susumu Mori, Department of Radiology, Johns Hopkins Medical Institutions, 217 Traylor Building, 720 Rutland Avenue, Baltimore, MD 21205-2195, USA

Dr. David Norris, FC Donders Centre for Cognitive Neuroimaging, Trigon 181, PO Box 9101, 6500 HB Nijmegen, The Netherlands

Dr. Leif Østergaard, Department of Neuroradiology, CFIN, Arhus University Hospital, Building 30, Norrebrogade 44, DK-8000 Aarhus C, Denmark

Dr. Adolf Pfefferbaum, SRI International, Room BN 168, 333 Ravenswood, Menlo Park, CA 94025, USA Professor John Pickard, Academic Department of Neurosurgery, Addenbrooke’s Hospital, PO Box 167, Cambridge, CB2 2QQ, UK

List of contributors

Dr. John Port, Department of Radiology, Mayo Clinic, 200 First Street SW, Rochester, MN 55905, USA Dr. Luca Roccatagliata, Department of Radiology, Massachusetts General, Hospital, Gray 2, Room B285, 55 Fruit Street, Boston, MA 02114, USA Dr. Brian Ross, Departments of Radiology and Biological Chemistry, MSRB III, Room 9301, 1150 W Medical Center Dr, Ann Arbor, MI 48109-0648, USA Dr. Brian Ross, Magnetic Resonance Spectroscopy, Huntington Medical Research Institute, 660 South Fairoaks Avenue, Pasadena, CA 91105, USA Dr. Marco Rovaris, Neuroimaging Research Unit, Scientific Institute and University, Ospedale San Raffaele, Via Olgettina 60, Milan 20132, Italy Dr. Dirk Rutgers, Department of Radiology, University Medical Center Utrecht, Heidelberglaan 100, Postbus 85500, 3508 Ga Utrecht, The Netherlands

Dr. Mary Rutherford, Robert Steiner MR Unit, Imaging Sciences Department, Clinical Sciences Centre, Imperial College, Hammersmith Hospital, London W12 OHS, UK

Dr. Dawn Saunders, Great Ormond Street Hospital for Children, Great Ormond Street, London, WC1N 3JH, UK

Dr. Pamela Schaefer, Department of Radiology, Massachusetts General, Hospital, Gray 2, Room B285, 55 Fruit Street, Boston, MA 02114, USA

Dr. Afonso Silva, Laboratory of Functional and Molecular Imaging, NINDS, NIH, 10/B1D118 MSC, Bethesda, MD 20817, USA

Dr. Edith Sullivan, SRI International, Room BN 168, 333 Ravenswood, Menlo Park, CA 94025, USA

Dr. Lalith Talagala, Laboratory of Functional and Molecular Imaging, NINDS, NIH, 10/B1D118 MSC, Bethesda, MD 20817, USA

xv

xvi

List of contributors

Professor Gioacchino Tedeschi, Institute of Neurological Sciences, Second University of Naples, Piazza Miraglia 2, Naples 80131, Italy

Professor K.M.A. Welch, University of Kansas Medical, Center, 8002 Wescoe, 3901 Rainbow Boulevard, Mail Stop 1039, Kansas City, KS 66160-7300, USA

Dr. Brian Tress, Department of Radiology, c/o PD Royal Melbourne Hospital, Parkville, Victoria 3060, Australia

Dr. Mark Wellard, Brain Research Institute, Austin and Repatriation Medical Centre, Banksia Street, Heidelberg West, VIC. 3084, Australia

Dr. Daniel Vigneron, Department of Radiology L371, University of California San Francisco, 505 Parnassus Avenue, PO Box 0628, San Francisco, CA 94143-0628, USA

Dr. lain Wilkinson, Academic Radiology C Floor, Royal Hillamshire Hospital, Glossop Road, Sheffield, S10 2JF UK

Dr. Adam D. Waldman, Department of Imaging, Charing Cross Hospital, Fulham Palace Road, London, W6 8RF, UK

Dr. Peter van Ziji, Department of Radiology, Johns Hopkins Medical Institutions, 217 Traylor Building, 720 Rutland Avenue, Baltimore, MD 21205-2095, USA

Dr. Jiongjiong Wang, Departments of Neurology and Radiology, University of Pennsylvania, Philadelphia, PA 19104, USA

Professor Robert Zimmerman, Deparment of Radiology, New York Hospital/Cornell, 525 E 68th Street, New York, NY 10021, USA

Dr. Joanna Wardlaw, Department of Clinical Neurosciences, Western General Hospital, Crewe Road, Edinburgh, Scotland

List of abbreviations

AA

Anaplastic astrocytoma

ABC

ATP-binding assette

ACA

Anterior cerebral artery

Ace

Acetate

AComA

Anterior communicating artery

AD

Alzheimer’s disease

ADC

Apparent diffusion coefficient

ADEM

Acute disseminated encephalomyelitis

ADNFLE

Autosomal-dominant nocturnal frontal lobe epilepsy

ADP

Adenosine diphosphate

AED

Antiepileptic drugs

AFB

Acid fast bacilli

AFP

Adiabatic fast passage

AGAT

L-arginine : glycine amidinotransferase

a-glu

-glucose

AIDS

Acquired immunodeficiency syndrome

AIF

Arterial input function

Ala

Alanine

ALD

Adrenoleukodystrophy

ALS

Amyotrophic lateral sclerosis

ANE

Acute necrotizing encephalopathy

ASA

Arylsulfatase A

ASL

Arterial spin labeling

ASPA

Aspartoacylase

ATLS

Advanced trauma and life support xvii

xviii

List of abbreviations

ATP

Adenosine triphosphate

CPP

Cerebral perfusion pressure

ATRT

Atypical terato-rhabdoid tumour

Cr

Creatine

AUP

Area under peak

CSD

Cortical spreading depression

AVM

Arteriovenous malformations

CSF

Cerebrospinal fluid

AZT

Azidothymidine

CSI

Chemical shift imaging

BASING

Band-selective inversion with gradient dephasing

CT

Computed tomography

CTA

Computed tomography angiography

BAT

Bolus arrival times

CTX

Cerebrotendinous xanthomatosis

BBB

Blood–brain barrier

CVR

Cerebrovascular reserve

BCAA

Branched-chain amino acids

DAI

Diffuse axonal injury

BCKA

Branched-chain alpha-ketoacid

DANTE

bFGF

Basic fibroblast growth factor

Delays alternating with notations for tailored excitation

b-glu

-glucose

DAT

Dopamine uptake transporter

BGT

Basal ganglia and thalami

DEHSI

Diffuse excessive high signal intensity

BOLD

Blood oxygen level dependent

DLB

Dementia with Long Bodies

BVR

Basal vein of Rosenthal

DLPFC

Dorsolateral prefrontal cortex

CACH

Childhood ataxia with central hypomyelination

DNA

Deoxyribonucleic acid

DNET

CADASIL Cerebral autosomal dominant arteriopathy with subcortical infarcts and leukoencephalopathy

Dysembryoblastic neuroepithelial tumour

DRCE

Dynamic relaxivity contrast enhanced imaging

CASL

Continuous arterial spin labeling

DRESS

Depth resolved surface coil spectroscopy

CBD

Corticobasal degeneration

DRS

Disability rating scale

CBF

Cerebral blood flow

DRSTOT

Dementia rating scale total score

CBV

Cerebral blood volume

DSA

Digital subtraction angiography

CE

Carotid endarterectomy

DSCI

Dynamic susceptibility contrast imaging

CEA

Carotid endarterectomy

DSM

CHESS

Chemical shift selective water suppression

Diagnostic and statistical manual of mental disorders

DTI

Diffusion tensor imaging

Cho

Choline

DWI

Diffusion-weighted imaging

CJD

Creutzfeldt–Jakob disease

DWIS

Diffusion-weighted imaging spectra

CM

Chronic migraine

DZ

Dizygotic

CMRO2

Cerebral metabolic rate of oxygen metabolism

EBV

Epstein–Barr virus

ECG

Electrocardiogram

CNS

Central nervous system

EDAS

Encephaloduroarteriosynangiosis

CO

Carbon monoxide

EDE

Epidural empyemas

COSY

Correlation spectroscopy

EDSS

Expanded disability status scale

List of abbreviations

EEG

Electroencephalograms

GBM

Glioblastoma multiforme

EES

Extravascular/extracellular space

GCS

Glasgow coma scale

EITB

Enzyme-linked immunotransfer blot

ELISA

Enzyme linked immunosorbent assay

Gd-DTPA Gadolinium dimeglumine gadopentetate

EM

Episodic migraine

EPI

Echo planar imaging

EPISTAR

Echo planar imaging-signal tagging with alternating radio frequency

FA

Fractional anisotropy

FACT

Fiber assignment by continuous tracking

FADH

Flavin adenosine dinucleotide

FAIR

Flow-sensitive alternating inversion recovery

FAIRER

Flow-sensitive alternating inversion recovery with an extra radio frequency pulse

GE

Gradient echo

GFAP

Glial fibrillary acidic protein

GLD

Globoid cell leukodystrophy (Krabbe disease)

Gln

Glutamine

Glu

Glutamate

Glx

Glutamate and glutamine

Gly

Glycine

GM

Gray matter

GOS

Glasgow outcome score

GPC

Glycerophosphocholine

GPE

Glycerophosphoethanolamine

FASTMAP Fast automatic shimming technique by mapping long projections

GRASS

Gradient recalled echo acquisition at steady state

FDA

Food and drug administration

GSS

Gerstmann–Straussler–Scheinker disease

FDG

Fluoro-2-deoxyglucose

HAART

Highly active antiretroviral therapy

FEAST

Flow encoding arterial spin tagging

HD

Huntington’s disease

FEMN

“First episode, medication naive (schizophrenia)”

HIE

Hypoxic ischemic encephalopathy

HIV

Human immunodeficiency virus

FGF

Fibroblast growth factor

HMPAO

Hexamethylpropyleneamine oxime

FID

Free induction decay

HPE

Holoprosencephaly

FLAIR

Fluid attenuated inversion recovery

HS

Hippocampal sclerosis

fMRI

Functional magnetic resonance imaging

HSCT

Hematopoietic stem cell transplantation

FOV

Field of view

HSE

Herpes simplex encephalitis

FSE

Fast spin echo

HSV

Herpes simplex virus

FT

Fourier transform

ICA

Internal carotid artery

FTD

Frontotemporal degeneration

ICD

International classification of disease

FWHM

Full width at half maximum

ICP

Intracranial pressure



Gyromagnetic ratios

ILAE

International league against epilepsy

GAA

Guanidinoacetate

Ile

Isoleucine

GABA

-amino-butyric acid

ISIS

Image selective in vivo spectroscopy

GALC

Galactocerebroside -galactosidase

IPD

Idiopathic Parkinson’s disease

GAMT

Guanidinoacetate methyl transferase

IVF

Interstitial volume fraction

xix

xx

List of abbreviations

IVIM

Intra voxel incoherent motion

MRI

Magnetic resonance imaging

JPA

Juvenile pilocytic astrocytoma

MRS

Magnetic resonance spectroscopy

KD

Krabbe disease

MRSI

KSS

Kearns–Sayre syndrome

Magnetic resonance spectroscopic imaging

Lac

Lactate

MRUI

Magnetic resonance user interface

LACI

Lacunar infarction

MS

Multiple sclerosis

LC model Linear combination model

MSA

Multisystem atrophy

Leu

Leucine

MSM

Methylsulfonylmethane

LGN

Lateral geniculate nucleus

MSUD

Maple syrup urine disease

LHON

Leber’s hereditary optic atrophy

MT

Magnetization transfer

LI

Lattice index

MTC

Magnetization transfer contrast

LR

Logistic regression

MTI

Magnetization transfer imaging

LS

Leigh syndrome

MTR

Magnetization transfer ratio

MB

Medulloblastomas

MTT

Mean transit time

MCA

Middle cerebral artery

MZ

Monozygotic

MCD

Myelinopathia centralis diffusa

NAA

N-acetyl aspartate

MCI

Mild cognitive impairment

NAAG

N-acetyl aspartyl glutamate

MCMD

Minor cognitive motor disorder

NABT

Normal appearing brain tissue

MEG

Magneto-encephalography

NAGM

Normal appearing gray matter

MEGA

Mescher–Garwood

NANA

N-acetylneuraminic acid

MELAS

Mitochondrial encephalopathy with lactic acidosis and stroke

NASCET

North American symptomatic carotid endarterectomy trial

MERRF

Myoclonic epilepsy with ragged red fibers

NAWM

Normal appearing white matter

NBV

Normalized brain volume

mI

myo-inositol

NF

Neurofibromatosis

MITR

Maximal intensity change per time interval ratio

NICE

National Institute of Clinical Excellence

NIHSS

National Institute of Health Stroke Scale

MLC

Megalencephalic leukoencephalopathy with subcortical cysts

NINCDS–ADRDA

MLD

Metachromatic leukodystrophy

MMSE

Mini-mental state examination

MPC

Maximum peak concentration

MPCSI

Multi planar chemical shift imaging

NKH

Non-ketotic hyperglycinaemia

MPRAGE

Magnetization prepared rapid acquisition gradient echo

NMR

Nuclear magnetic resonance

NOESY

Nuclear overhauser effect

MR

Magnetic resonance

NOS

Nitric oxide synthase

MRA

Magnetic resonance angiography

NTP

Nucleoside triphosphate

National Institute of Neurological and Communicative Disorders and Stroke–Alzheimer’s Disease and Related Disorders Association

List of abbreviations

OCD

Obsessive compulsive disorder

OEF

Oxygen extraction fraction

OHG

PP

Primary progressive (multiple sclerosis)

Hydroxy glutaric acid

ppm

Parts per million

OphtA

Ophthalmic artery

PRESS

Point resolved spectroscopy

OVS

Outer volume suppression

PROBE

Proton brain exam

OXPHOS

Oxidative phosphorylation

PROPELLER

PAC

Pulmonary artery catheter

Periodically rotated overlapping parallel lines with enhanced reconstruction

PACE

Prospective acquisition and correction

PS

Permeability surface

PACI

Partial anterior circulation infarct

PSD

Periodic synchronous discharge

PAGM

Periaqueductal gray matter

PSP

Progressive supranuclear palsy

PASL

Pulsed arterial spin labeling

PTA

Post traumatic amnesia

PC

Phosphocholine

PTSD

Posttraumatic stress disorder

PCA

Posterior cerebral artery

PVE

Partial volume effect

PCD

Programmed cell death

PVL

Periventricular leukomalacia

PComA

Posterior communicating artery

PWI

Perfusion weighted imaging

PCR

Polymerase chain reaction

QUALY

Quality adjusted life years

PCr

Phosphocreatine

QUIPSS

PD

Proton density

Quantitative imaging of perfusion using a single subtraction

PDE

Phosphodiester

QUIPSSII

PDGF

Platelet derived growth factor

Quantitative imaging of perfusion using a single subtraction II

PDS

Paroxysmal depolarization shifts

RA

Relative anisotropy

PET

Positron emission tomography

RAA

Recently abstinent alcholics

PGSE

Pulsed gradient spin echo

rCBF

Relative cerebral blood flow

Phe

Phenylalanine

rCBV

Relative cerebral blood volume

Pi

Inorganic phosphate

RF

Radio frequency

PKU

Phenylketonuria

RFA

Reduced flip angle

PLIC

Posterior limb of the internal capsule

RN

Red nucleus

PLP

Proteolipid protein

ROI

Region of interest

PMD

Pelizaeus–Merzbacher disease

RPLS

Reversible posterior leukoencephalopathy syndrome

PME

Phosphomonoester

rR

Relative recirculation

PML

Progressive multifocal leukoencephalopathy

RR

Relapsing–remitting (multiple sclerosis)

PMN

Polymorphonuclear neutrophils

RT

Radiation therapy

PNET

Primative neuroectodermal tumor

rt-PA

POCI

Posterior circulation infarct

Recombinant tissue plasminogen activator

POI

Pixel of interest

SAT

Saturation

xxi

xxii

List of abbreviations

sCJD

Sporadic Creutzfeldt–Jakob disease

TLE

Temporal lobe epilepsy

SD

Salla disease

TM

Transverse myelitis

SDE

Subdural empyema

TM

Mixing time

SDH

Subdural hemorrhage

Tmax

Time to maximum peak

SDMT

Symbol digit modalities test

TMS

Transcranial magnetic stimulation

SE

Spin echo

TOAST

SENSE

Sensitivity encoding

Trial of Org 10172 in acute stroke treatment

SI

Spectroscopic imaging

TR

Repetition time

SIAM

Spectroscopic imaging acquisition mode

TS

Tuberous sclerosis

SLR

Shinnar–LeRoux

TSP

Trimethyl lysyl sodium propionate

SLS

Sjogren–Larsson syndrome

TTFM

Time to first moment

SMIT

Na/myo-inositol cotransporter

TTP

Time to peak

SN

Substantia nigra

UNFAIR

SNR

Signal-to-noise ratio

Perfusion imaging by un-inverted flowsensitive alternating inversion recovery

SOL

Space occupying lesion

USPIO

Ultra small particulates of iron oxide

SP

Secondary progressive (multiple sclerosis)

VaD

Vascular dementias

SPECT

Single photon emission computed tomography

Val

Valine

VC

Visual cortex

SPM

Statistical parametric mapping

vCJD

Variant Creutzfeldt–Jakob disease

SRO

Steele–Richardson–Olszewski syndrome

VEGF

Vascular endothelial growth factor

SSPE

Subacute sclerosing panencephalitis

VHL

von Hippel–Lindau syndrome

SSRIs

Selective serotonin reuptake inhibitors

VLCFA

Very long chain fatty acids

STEAM

Stimulated echo acquisition mode

VOI

Volume of interest

Suc

Succinate

VSS

Very selective saturation

SVD

Singular value decomposition

VWM

Vanishing white matter disease

T2W

T2-weighted

WET

TACI

Total anterior circulation infarct

Water suppression enhanced through T1 effects

TB

Tuberculosis

WHO

World Health Organization

TBI

Traumatic brain injury

WI

Weighted images

TCA

Tricarboxylic acid

WM

White matter

TCD

Transcranial Doppler sonography

WMH

White matter hyperintensities

TDL

Tumefactive demyelinating lesions

x-ALD

x linked adrenoleukodystrophy

TE

Echo time

Xe-CT

Xenon-enhanced computed tomography

TI

Inversion time

ZDV

Zidovudine

TIA

Transient ischemic attack

Foreword

The advent of clinical MR imaging (MRI) in the 1980s heralded a new era in the ability to image the brain in vivo. MRI allows the detailed depiction of brain anatomy and pathology with unprecedented spatial resolution and soft-tissue contrast. It is also relatively safe and completely non-invasive. Nevertheless, the sensitivity and specificity with which structural MRI alone can define the wide range of neurological disease is limited. The last decade has also seen the development of physiological MR techniques, whereby information concerning tissue function as well as structure is obtained. These techniques include diffusion, perfusion, and MR spectroscopy, which provide information on tissue ultra-structure, blood flow, and biochemistry, respectively. Information of this type supplements and complements that from clinical or structural imaging investigations, often providing important surrogate markers of disease pathophysiology or therapeutic response. These techniques, previously only available in a research environment, are now accessible on most MR systems and can readily be incorporated into clinical imaging examinations. To date, however, there has been a paucity of literature in a single volume to support those wishing to apply physiological imaging studies in a clinical context. The aim of this book is to address the appropriate clinical application and interpretation of diffusion, perfusion, and spectroscopy. The first section of the book describes the physical principles underlying each technique, as well as the potential associated artifacts and pitfalls. xxiii

xxiv

Foreword

The second section addresses applications in the different branches of clinical neuroscience. Chapters are grouped according to pathology, and are preceded by overviews that aim to place these methodologies in a broader clinical perspective. Illustrative case reports are included throughout the book. We recognize that the term “functional MRI” (fMRI) has become synonymous with studies of localized brain activation, mostly using “blood oxygen leveldependent” (BOLD) contrast. This approach, which continues to contribute to the understanding of the relationship between brain structure and function, is well covered in other texts and is not addressed in this volume. Likewise, magnetization transfer imaging, and methods for post-processing structural data, for example volumetric analysis, or MRI relaxometry, are not included. While these techniques are the subject of much research effort, they are not

widely available at the time of writing, and have yet to find a definitive clinical role. The aim of this book is to create a reference work for those techniques that can be widely applied, not just at academic medical centers. Currently, diffusion, perfusion, and spectroscopy are the physiological techniques most likely to be used routinely. Our hope is that this book will provide a balanced reference work for physiological MRI in real clinical practice. The overall aim is to optimize the use of these techniques to increase the sensitivity and specificity of the MR imaging examination, and thereby improve the management of individual patients. Jonathan H. Gillard, Cambridge Adam D. Waldman, London Peter B. Barker, Baltimore

Introduction R. Nick Bryan Department of Radiology, University of Pennsylvania Health System, 3400 Spruce Street, Philadelphia, USA

The last several decades have seen remarkable advances in the clinical neurosciences with some of the most remarkable achievements related to neuroimaging. Given the current depth of knowledge about the brain, it is difficult to appreciate that barely 300 years ago this organ was almost a complete mystery, particularly as to its function. While the brain has been recognized as an “organ” since antiquity, no functional role was ascribed to it until the early 1600s when Descartes placed the “soul” in one of its small parts, the pineal gland (Marshall and Magoun, 1998). Prior to this intriguing, but erroneous concept, much more functional importance had been attributed to the fluid in the ventricles than the brain itself. Descartes’ non-scientific attribution was, fortunately, quickly followed by the much more rigorous description of the structure of the brain by Thomas Willis (1664). While Willis’ application of the scientific method to the brain was seminal, the primitive scientific tools available at the time limited his direct observations to anatomy, which in and of itself does not convey function. Despite little direct evidence, Willis began to argue that mental functions reside in the brain, as do certain diseases such as epilepsy. The scientific tools necessary to prove his assertions by actual observation of physiology, molecular biology, and other “functional” aspects of the brain were still several centuries away. However, the brain was found to have a peculiarly strong correlation between structure (anatomy) and function (behavior). This intimate relationship provided the basis for the still robust field of “experimental” neuroanatomy. Experimental neuroanatomy, such as the destruction of a portion of the brain in an animal followed by observations of its behavior

allowed 18th and early 19th Century scientists such as Gall and Rolando to make structure/function correlations that documented the brain as a central control organ (Marshall and Magoun, 1998). Since it has never been appropriate to perform debilitating experiments on human beings, many fundamental questions pertaining to human brain function persisted until the “natural science” version of experimental neuroanatomy was introduced by clinicians such as Morgagni, who attributed neurological deficits such as hemiparesis to grossly destructive lesions of patients’ brains found at autopsy (Morgagni, 1760). Broca, in 1860, applied such lesion/deficit correlation to a patient who had suffered the acute onset of aphasia and whose brain at autopsy revealed an infarct in the right frontal operculum, thus localizing a component of speech to a particular cortical region (Broca, 1861). Such “dysfunctional” imaging was subsequently employed by many clinical scientists, particularly those 19th and early 20th Century neurologists whose names are attached to so many neurological syndromes. While lesion/deficit correlation has been a very informative means of studying the brain, it is limited by its anatomic basis that does not provide any direct information about the brain’s physiology or molecular makeup. Note that all of these early methods of studying the brain involved some form of imaging. Given the spatially heterogeneous nature of the brain (both structurally and functionally), imaging of the brain is an absolute necessity in order to document the location of an experimental or natural lesion. Only with this anatomic information could the observed neurological, psychological, or cognitive dysfunction be linked to its physical source. In human 1

2

R. Nick Bryan

beings these types of investigations were severely restricted by the unfortunate necessity of a patient having to suffer an insult to the brain and the additional burden that the patient either die and have permitted an autopsy or submit to a craniotomy. These were until very recently the only means of directly documenting the presence and extent of a brain lesion. Despite the many drawbacks, experimental anatomy and clinical lesion/deficit research in the first half of the 20th Century provided the basis of much of our current understanding of functional localization in the brain. During the second half of the 20th Century, these early, primitive, but informative techniques were increasingly supplemented by sophisticated histological, neurophysiological, and molecular biological techniques that have combined to yield the great depth of knowledge about the brain that we now have; knowledge that extends from single cell events to highly integrated cognitive functions. However, many of these newer techniques also have restrictions to their applications in human beings, particularly intact, functioning human beings. Histological techniques require tissue, never easily obtained from human brain and almost never from multiple or large regions. Many neurophysiological techniques require intrusion into the brain, as for electrode recordings or cortical stimulation. Molecular techniques are seldom feasible in intact functioning brain. While these powerful techniques provide extraordinarily detailed information about small parts of the brain, none provide data from the entire, functioning brain. This is a significant limitation as many functions of the brain involve composite actions of its many spatially, physiologically, and biochemically disparate components. This is particularly true of complex behavioral tasks and cognition. The spatial heterogeneity of the brain has always begged for imaging of the whole organ, preferably in the intact, functioning state. This has not been feasible until very recently. In 1974 clinical neuroscience experienced a profound change with the invention of the X-ray computed tomography (CT) scanner, an instrument that for the first time could non-invasively produce images of the whole, living human brain (Hounsfield, 1973). CT scans are based on electron density and there are only subtle differences of this parameter in

the brain. For instance, the electron density of gray matter (GM) and white matter (WM) differ by only 0.5%. Hence, clinical CT scans yield relatively crude images of the brain. While CT scanners can only image anatomy at a relatively low resolution it has allowed the traditional lesion/deficit methodology to be applied to living subjects contemporaneously with functional examinations. Autopsy and craniotomy are no longer necessary to demonstrate the anatomical correlates of functional deficits and the literature has become replete with lesion/deficit studies expanding our knowledge of how the function of the human brain is spatially distributed. Investigators such as the Damasio’s have used clinical CT, and later magnetic resonance (MR) scans of hundreds of neurologically, psychologically, and cognitively impaired subjects to better demonstrate the anatomic substrate of higher order mental tasks (Damasio and Damasio, 1989). However these images still show only static anatomy and do not reflect any physiological or molecular aspect of the brain. Indeed it can be difficult to tell a conventional CT or MR scan of a cadaver’s brain from that of a normal person. While we now understand many strong relationships between the gross structure and function of the brain, there remains the overpowering need to be able to directly “see” physiological and molecular function of the brain. After all, it is more important to know what the brain is doing than what it looks like! This need was initially met by the combination of positron emission tomography (PET) and metabolic radio tracers such as F18DG, H2O15, and CO15 (Fox et al., 1988). PET methodology allows non-invasive imaging of the whole brain under resting as well as task conditions. Physiological parameters, such as cerebral blood flow (CBF) can be imaged noninvasively in the clinical environment, as can responses of these parameters to activation of the brain by a task – direct imaging of dynamic brain physiology. In addition, radio ligands have been developed that produce images of the distribution of specific molecules in the brain, such as components of neurotransmitter systems. This methodology remains a powerful research tool, albeit expensive and logistically challenging. As a result of these advances, the 20th Century progressed from very limited, invasive anatomic

Introduction

imaging of a poorly understood human brain to widely applied, non-invasive, dynamic physiological and biochemical imaging of a richly appreciated organ. Continuing advances in neuroimaging will offer ever more information about the brain and its function. This book focuses on the important evolving methodology of MR imaging (MRI), specifically physiological MRI of the brain. MRI derived from nuclear MR (NMR), a physical phenomenon related to the behavior of nuclei in the presence of a magnetic field that was described by Felix Bloch, Hansen and Packard (1946). During the 1940s and 1950s many investigators developed techniques that allowed this physical phenomenon to be exploited for the study of chemical structure. Since the introduction of the Fourier transform (FT) technique by Ernst in 1966 and the development of high-field superconducting magnets, NMR has been able to elucidate the detailed chemical structure of even large molecules such as proteins (Ernst and Anderson, 1966). The addition of magnetic field gradients to the requisite static magnetic field of NMR can spatially define a sample, allowing MRI. This concept of the use of magnetic field gradients to generate images was first demonstrated in the landmark 1973 paper by Lauterbur; in 1976, Ernst introduced the principle of two-dimensional FT NMR which is now almost universally used for all MRI (Lauterbur, 1973; Aue et al., 1976). Conventional MRI relies on radio signals emitted by nuclei of molecules, particularly H2O, of relatively stationary tissue. Because of their different water content and relaxation times, there is typically more than 20% difference in this signal between GM and WM. Similar differences can be found between certain pathological tissues and normal brain. This accounts for the exquisite images of normal neuroanatomy or multiple sclerosis (MS) plaques produced by contemporary MRI. The first decade of clinical MRI was characterized by steady improvements in the morphological imaging capabilities of this quite remarkable and completely non-invasive and safe technology. However, there is little useful physiological information in conventional MRI signal, except for that related to fast-flowing fluids such as blood. Recent MRI advances have focused on the development and application of molecular and physiological

imaging capabilities. These new MRI methods are the subject of this volume and reflect the continuing evolution from purely anatomic to physiological and molecular imaging of the brain. The three main physiological MR methods to be presented are MR spectroscopy (MRS), diffusion, and perfusion MRI. MRS yields images of the distribution and concentration of naturally occurring molecules such as N-acetyl aspartate (NAA) (one of the most abundant amino acids in the brain, and believed to be localized predominantly in neurons and their processes), choline (Cho) (a key constituent of cell membranes) and lactate (Lac) (a reflection of anaerobic metabolism). Diffusion MRI demonstrates regions of normal and pathological micromolecular motion. Under appropriate conditions, these images can reflect patterns of axonal anatomy and when applied as “fiber tracking” this technique can turn the large homogeneously bland regions of WM of conventional MRI into dramatic threedimensional displays of the major axonal pathways. Using extrinsic contrast agents or intrinsic contrast agents, such as blood, perfusion MRI cannot only create qualitative, but quantitative maps of various perfusion parameters, including CBF, cerebral blood volume (CBV), and vascular permeability. With these techniques, at last, neuroscientists can painlessly, non-invasively, and safely study important physiological properties of a whole, living, functioning human brain. One can now actually see what the brain is doing, not just what it looks like. The clinical value of these physiological and molecular tools is becoming increasingly appreciated and can be illustrated by their applications to one disease – cerebral ischemia and stroke. Lac is an important metabolic molecule of which little is produced by the brain under aerobic conditions. However, under anaerobic conditions, such as ischemia, abundant Lac may be produced and is easily detected by proton MRS (Barker et al., 1994). The imaging of Lac by MRS is one of the most sensitive means of detecting even mild cerebral ischemia, its presence temporally preceding irreversible ischemia and stroke. Diffusion MRI is also very sensitive to ischemia, presumably because there is a shift of extracellular water molecules into the intracellular compartment where molecular diffusion is more

3

4

R. Nick Bryan

restricted (Le Bihan et al., 1986). Even if this theory is not correct, empirically it is well established that diffusion weighted images show some of the earliest changes of stroke and severe ischemia. It almost goes without saying that perfusion imaging is a powerful tool for evaluating cerebral ischemia. Perfusion MRI can easily, directly, and accurately document the reduction of CBF secondary to obstructive or nonobstructive cerebral ischemia as well as demonstrate changes in CBV that often provide additional information as to the physiological severity of the insult (Rempp et al., 1994). Such physiological tools are increasingly necessary for the management of acute cerebral ischemia when the traditional anatomic diagnosis of “live brain/dead brain” is not adequate for directing vascular or neuroprotective treatment. The authors of the chapters of this book describe the latest physiological MRI methodologies in detail and then illustrate their applications to major diseases of the brain, including cerebrovascular and degenerative diseases, neoplasia, inflammation, trauma, and even psychiatric disorders. These new techniques of the early 21st Century foreshadow even more remarkable advances in neuroimaging, but first, please appreciate the robust functional imaging capabilities so well described and illustrated in this volume.

REFERENCES Aue WP, Bartholdi E, Ernst RR. 1976. Two-dimensional spectroscopy. Application to nuclear magnetic resonance. J Chem Phys 64: 2229.

Barker PB, Gillard JH, van Zijl PCM, et al. 1994. Acute stroke: evaluation with serial proton magnetic resonance spectroscopy. Radiology 192: 723–732. Bloch F, Hansen WW, Packard M. 1946. The nuclear induction experiment. Phys Rev 70: 474–485. Broca P. 1861. Remarques sur le siège de la faculté du langage articule, suivies d’une observation d’aphémie. Bull Soc Anat Paris: 330–357. Trans. Von Boninn G. Some Papers on the Cerebral Cortex. Springfilld. Ill. C.C. Thomas, 1960. Damasio H, Damasio AR. 1989. Lesion Analysis in Neuropsychology. Oxford University Press, NY, USA. Ernst RR, Anderson WA. 1966. Rev Sci Instrum 37: 93. Fox PF, Raichle ME, et al. 1988. Nonoxidative glucose consumption during focal physiologic neural activity. Science 241: 462–464. Hounsfield GN. 1973. Computerized transverse axial scanning (tomography): part 1. description of system. Br J Radiol 46: 1016–1022. Lauterbur PC. 1973. Image formation by induced local interactions: examples employing nuclear magnetic resonance. Nature 242: 190–191. Le Bihan D, Breton E, Lallemand D. 1986. MR imaging of intravoxel incoherent motions: application to diffusion and perfusion in neurologic disorders. Radiology 161: 401–407. Marshall LH, Magoun HW. 1998. Discoveries in the Human Brain. Humana Press Inc, Totowa, NJ, USA. Morgagni JB. 1760. The Seats and Causes of Diseases. Trans. Alexander B. The Classics of Medicine Library, Birmingham, AL, USA. Rempp KA, Brix G, Wenz F, Becker CR, Guckel F, Lorenz WJ. 1994. Quantification of regional cerebral blood flow and volume with dynamic susceptibility contrast-enhanced MR imaging. Radiology 193: 637–641. Willis T. 1664. Cerebri Anatome. Trans. Pordage S. The Classics of Medicine Library, Birmingham, AL, USA.

Section 1 Physiological MR techniques

1

Fundamentals of MR spectroscopy Peter B. Barker Department of Radiology, Johns Hopkins University, School of Medicine, Baltimore, USA

Introduction Nuclear MR (NMR) spectroscopy in bulk matter was demonstrated for the first time in 1945 when Bloch and Purcell independently demonstrated that a strong magnetic field induced splitting of the energy levels and detected the resonance phenomena (Bloch, 1946; Purcell et al., 1946). The method was originally of interest only to physicists for the measurement of gyromagnetic ratios () of different nuclei, a constant specific to a particular nucleus, but applications of NMR to chemistry became apparent after the discovery of chemical shift and spinspin coupling effects in 1950 and 1951, respectively (Proctor and Yu, 1950; Gutowsky et al., 1951). The spectra of high-resolution liquid NMR contain fine structure information because the nuclear resonance frequency is influenced by both neighboring nuclei and the chemical environment which allows information on the structure of the molecule to be deduced. Hence, NMR spectroscopy rapidly became an important, and widely used, technique for chemical analysis and structure elucidation of chemical and biological compounds. Major technical advances in the 1960s included the introduction of superconducting magnets (1965), which were very stable and allowed higher field strengths than with conventional electromagnets to be attained, and in 1966 the use of the Fourier transform (FT) for signal processing. In nearly all contemporary spectrometers, the sample is subjected to periodic radio frequency (RF) pulses directed perpendicular to the external field and the signal is Fourier transformed to give a spectrum in the frequency domain. FT NMR provides increased

sensitivity compared to previous techniques, and also led to the development of a huge variety of pulsed NMR methods, including multi-dimensional NMR techniques. Biological and medical applications of MR were developed in the early 1970s with the introduction of MR imaging (MRI) and MR spectroscopy (MRS) of biological tissue. In vivo MRS of humans became possible in the early 1980s with the advent of whole body magnets with sufficiently high field strength and homogeneity (Radda, 1986). Early studies focused on the phosphorus nucleus, since this was the most technically feasible at that time. Methods were developed for spatially localized 31P MRS (Luyten et al., 1989), and studies of major neuropathology (such as stroke or brain tumors) were performed (Arnold et al., 1989; Cadoux et al., 1989; Levine et al., 1992). A significant problem with 31 P MRS, however, is its low sensitivity (mainly because of the relatively low  of 31P, and low concentrations of phosphorus containing compounds). Since the spatial resolution in in vivo spectroscopy is largely limited by the signal-to-noise ratio (SNR) the minimum voxel size for 31P spectroscopy of the human brain is typically 30 cm3 using conventional techniques and 1.5 T magnets. The technique can therefore only be applied to either very large lesions, diffuse or global diseases. In recent years, there has been more interest in proton MRS, particularly after it was demonstrated that it was possible to obtain high-resolution spectra from small, well-defined regions in reasonably short scan times (Frahm et al., 1989). The higher sensitivity of the proton is due to several factors, including higher , higher metabolite concentrations, and 7

8

Peter B. Barker

(c)

(a) NAA B

A

Cho Cr

NAA

Cho Cr

mI

(b)

“Glx” Lac Lipids

ppm 4.0

3.0

2.0

1.0

ppm 4.0

3.0

2.0

1.0

Fig. 1.1 Proton spectra of the human brain recorded at both (a, b) long (TE 272) and (c) short (TE 35 ms) TEs, in the long TE spectra from a patient with an acute right middle cerebral artery (MCA) stroke, the normal spectrum (a) from the left hemisphere shows signals from Choline (Cho), Creatine (Cr) and N-acetyl aspartate (NAA). In the ischemic left hemisphere (b) an additional signal due to Lactate (Lac) is apparent as well as a moderate decrease in NAA. In the short TE spectrum of normal frontal WM (c), in addition to NAA, Cr and Cho, signals can be detected from myo-inositol (mI), Glutamate and Glutamine (Glx), and lipids. (a) and (b) are from a multi-slice MR spectroscopic imaging (MRSI) data set (nominal voxel size 0.8 cm3), while (c) is recorded from an 8 cm3 single voxel using the Point resolved spectroscopy (PRESS) sequence.

more favorable relaxation times. Although proton spectroscopy has been demonstrated in a number of organ systems (in particular, recent studies show promise for the use of proton spectroscopy in the diagnosis of prostate and breast cancer), the overwhelming number of applications have been in the brain, because of the absence of free lipid signals in normal cerebrum, relative ease of shimming, and lack of motion artifacts. The proton is also a widely used nucleus because it is the same nucleus used for conventional MRI, and therefore it is usually possible to perform proton MRS on most 1.5 T or higher clinical MRI machines without the need to purchase additional scanner hardware or modifications. NMR spectroscopy can in fact be performed with many different nuclei, and in the brain, in addition to 1H and 31P, there have been reports of spectroscopy of deuterium (2D), carbon-13, nitrogen-15, lithium-7, sodium-23, and fluorine-19, using either

signals from endogenous nuclei and/or compounds, or via the administration of (sometimes isotopically enriched) exogenous substances. All of these studies fall into the context of advanced research at the current time, and therefore will not be considered further here. This chapter focuses on the information content of proton MR spectra of the brain, technical issues such as choice of localization technique, and normal age-related and anatomical variations.

Information content of proton MR spectra of the brain Figure 1.1 shows examples of proton spectra recorded at long and short echo times (TEs). The assignment and significance of each the resonances in the spectrum is discussed below, and summarized in Table 1.1.

Chemical shift normal concentration median (range) 2.02 ppm 7.8 mM (6.5–9.7)

3.2 ppm 1.3 mM (0.8–1.6)

3.0 ppm 4.5 mM (3.4–5.5)

3.56 ppm (short TE only) 3.8 mM (2.2–6.8) 2.1–2.4 ppm (short TE only) Glu ⬃ 10 mM Gln ⬃ 5 mM

1.35 ppm (doublet, 7 ppm separation) Detectable ⬃1 mM 0.9 and 1.3 ppm (short TE unless(↑↑)) Various Not normally detectable

Various

Metabolite

NAA (NAA, other N-acetyl moieties)

Cho Cho-containing compounds

Cr Creatine/ phosphocreatine

Myo Myo-inositol (mI) (other inositols)

Glx Glutamate (Glu)/ Glutamine(Gln)

Lactate

Lipids Mobile liquid moieties

Succinate, acetate, amino acids Acetoacetate, acetone

Mannitol, ethanol

Intermediary metabolites only pathologically elevated in inborn errors. Administered drugs and other substances

Not seen in normal brain. Membrane breakdown/lipid droplet formation. May precede histological necrosis. Products of bacterial metabolism.

Not seen in normal brain. End product of anaerobic respiration. May be energetic substrate of much brain metabolism. Thought to be elevated in foamy macrophages.

Complex overlapping J-coupled resonances difficult to separate and quantify at clinical field strengths (1.5–3 T). Amino acid neurotransmitters Glu excitatory, Gln inhibitory.

Pentose sugar. Involved in inositol triphosphate intra-cellular second messenger cycle, osmolyte, glial cell marker. High in infants.

Compounds related to energy storage; thought to be marker of energetic status of cells. Other metabolities are frequently expressed as ratio to Cr. Low in infants. Increases with age.

Detectable resonance is predominantly free Cho and derivatives. Marker of membrane turnover. Higher in W.M. than G.M. Increase with age.

Health neuronal cell marker. Only seen in nervous tissue. Exact physiological role uncertain.

Physiological significance

Table 1.1. Metabolites detected in the brain by proton MR spectoscopy

Pyogenic abcesses. Alanine: meningiomas Inborn errors of metabolism

High-grade tumors, abscesses, acute inflammation, acute stroke.

Ischaemia, inborn errors of metabolism (especially respiratory chain defects, tumors (all grades)), abscesses, inflammation.

Hepatic encephalopathy, severe hypoxia, OTC deficiency

Neonates, Alzheimer’s disease, diabetes, recovered encephalopathy, low grade glioma, hyperosmolar

Trauma, hypersomolar states

Tumors, inflammation, chronic hypoxia,

v.rarely Canavan’s disease

Increased

Possibly Alzheimer’s disease

Malignant tumors, Chronic hepatic encephalopathy, stroke

Hypoxia, stroke, tumors

Stroke, encephalopathy (hepatic human immunodeficiency virus (HIV)/liver disease.

Commonly: non-specific neuronal loss or dysfunction due to range of insults. Incl. Ischaemia, trauma, inflammation, infection, tumors, dementia, gliosis.

Decreased

10

Peter B. Barker

N-acetyl aspartate The largest metabolite signal, resonating at 2.02 ppm, occurs from the N-acetyl group of N-acetyl aspartate (NAA), with perhaps a small contribution from N-acetyl aspartyl-glutamate (NAAG) (Frahm et al., 1991). Despite being one of the most abundant amino acids in the central nervous system (CNS), NAA was not discovered until 1956 and its function has been the subject of considerable debate. It has been speculated to be a source of acetyl groups for lipid synthesis, a regulator of protein synthesis, a storage form of acetyl-CoA or aspartate, a breakdown product of NAAG, a “molecular water pump”, or an osmolyte (Barker, 2001). Using immunocytochemical techniques, NAA has been shown to be predominantly localized to neurons, axons, and dendrites within the central nervous system (Simmons et al., 1991), and studies of diseases known to involve neuronal and/or axonal loss (e.g. infarcts, brain tumors, seizure foci, multiple sclerosis (MS) plaques) have uniformly shown NAA to be decreased. In pathologies, such as MS, correlations between brain levels and clinical measures of disability have been shown (De Stefano et al., 2001). Animal models of chronic neuronal injury have also been shown to give good correlations between NAA levels (as measured by MRS) and in vitro measures of neuronal survival (Simmons et al., 1991; Guimaraes et al., 1995). For all these reasons, it has been tempting to “label” NAA as a neuronal marker, and to equate levels of NAA with neuronal density. However, there is increasing evidence that this may not be the case. NAA has been detected in non-neuronal cell types, such as mast cells or isolated oligodendrocyte preparations, suggesting that NAA may not be specific for neuronal processes (Urenjak et al., 1992; Burlina et al., 1997, Bhakoo and Pearce, 2000), although it is not completely clear if these cells are present in the brain or high concentrations, or if their metabolism is identical, in vivo. It is also well known that there are exceptions to the correlation between neuronal density and NAA levels. For instance, the pediatric leukoencephalopathy (Canavan’s disease) is associated with a large elevation of intracellular NAA, owing to deficiency of aspartoacylase (ASPA), the enzyme that degrades NAA to acetate and aspartate (Figure 1.2) (Barker et al., 1992).

(a)

Glucose, acetate, pyruvate, etc Glycolysis, TCA cycle Aspartate  acetyl CoA L-aspartate N-acetyl transferase “ANAT”, EC 2.3.1.17 NAA “ASPA” EC 3.5.1.15

NAAG “NAALADase” EC 3.4.17.21

Acetate  aspartate

(b)

NAA

ppm 4.0

3.0

2.0

1.0

3.0

2.0

1.0

(c)

ppm

4.0

Fig. 1.2 (a) Some biochemical pathways involving NAA, and (b, c) pathological processes involving NAA metabolism. (b) Long TE (270 ms) proton spectra of the frontal WM in a child with Canavan’s disease, showing a high ratio of NAA/Cr (and NAA to other metabolites) due to the lack of the enzyme ASPA which degrades NAA. T2-weighted MRI shows a near complete lack of myelination. (c) A 3-year old boy with mental retardation and complete absence of NAA on brain MRS (short TE). MRI is only mildly abnormal, while other metabolites in the spectrum are also in the normal range. A deficit in the NAA synthetic pathway was suspected, but not proven. Reproduced with permission from Martin et al. (2001).

11

Fundamentals of MR spectroscopy

(a) FLAIR

NAA

Lac

Cho

Cr NAA

Lac

ppm 4.0

3.0

2.0

1.0

ppm 4.0

3.0

2.0

1.0

(b)

Fig. 1.3 An example of a reversible reduction in NAA in a 6-year old child with ADEM. (a) 36 days after symptom onset, FLAIR MRI shows multiple, bilateral lesions which are characterized by reduced levels of NAA and increased Lac. Cho and Cr are within the normal range. (b) At day 137 after steroid treatment, the lesions have nearly resolved, and the spectra are more normal, in particular NAA has partially recovered and Lac is now undetectable.

In addition, there has been one remarkable case report of a young boy with mental retardation with an apparently global complete absence of NAA (Figure 1.2) (Martin et al., 2001). Clearly, in these subjects, the high levels or absent of NAA do not reflect changes in neuronal density, but rather a perturbation of the synthetic and degradation pathways of NAA metabolism (Figure 1.2). Further examples of the lack of direct correlation of NAA and neuronal density are various pathologies which have shown either spontaneous or treatmentrelated reversals of NAA decreases. Some examples include MS, mitochondrial diseases, acquired immuno deficiency syndrome (AIDS), temporal lobe epilepsy (TLE), amyotrophic lateral sclerosis (ALS) or acute disseminated encephalomyelitis (ADEM) (Bizzi et al., 2001; Barker, 2001) (Figure 1.3). Evidently, NAA does not appear to be essential for neuronal function.

How should changes (in most instances, decreases) in NAA be interpreted? It should be recognized that the macroscopic concentration of NAA (like that of any neurochemical) depends on the fluxes of synthetic and degradation pathways, cellular density, and brain water content and distribution. Sometimes, a decrease in NAA may be solely or largely attributable simply to increased extracellular water or cerebrospinal fluid (CSF) content within the localized MRS volume, although these factors can be corrected with appropriate analysis techniques (cf. Chapter 2). Neuronal and axonal dysfunction or loss should be considered when the tissue NAA content is reduced, because the balance of evidence suggests that the majority of NAA is located within neuronal processes. Whether the reduction represents an irreversible loss of cells or a potentially reversible

12

Peter B. Barker

metabolic process will in large part depend on the individual pathology in which it is encountered, and the prognosis for recovery of brain function is presumably also variable. In certain types of lesions (e.g. chronic infarction, brain tumors), it appears likely that in NAA do indeed correspond to irreversible neuronal loss. Overall, non-invasive MRS measurements of NAA appear to be one of the best surrogate markers currently available for neuronal integrity. However, it should be kept in mind that in some pathologies, NAA levels may vary independent of the state of the health and number of neurons. Choline The “choline” signal (Cho, 3.24 ppm) arises from the ˆN(CH3)3 groups of glycerophosphocholine (GPC), phosphocholine (PC), and a small amount of free Cho, compounds which are involved in membrane synthesis and degradation. Both increases and decreases in Cho have been reported in pathological conditions: processes leading to elevation of Cho signal include active demyelination (Davie et al., 1993), resulting from the degradation of myelin phospholipids primarily to GPC, or increased numbers of glial cells (Gill et al., 1989, 1990). Low Cho has been observed in hepatic encephalopathy (Kreis et al., 1992a), and there is also some evidence to suggest that dietary intake of Cho can modulate cerebral Cho levels (Stoll et al., 1995). Elevated Cho levels seem to be a characteristic of many types of neoplasms, including high-grade brain tumors (provided that they are not necrotic), prostate, breast, head and neck, and others. Creatine The “creatine” signal (Cr, 3.02 ppm) is a composite peak consisting of Cr and phosphocreatine, compounds which are involved in energy metabolism via the Cr kinase reaction generating ATP. Since Cr is synthesized in liver, chronic liver disease leads to lower cerebral Cr concentration (Ross and Michaelis, 1994a). There is also a rare group of diseases which involve total Cr deficiency in the brain, resulting from either lack of synthesis in the liver (guanidinoacetate methyl transferase (GAMT)

deficiency) or defective transport to the brain (Stockler et al., 1994; Cecil et al., 2001; Bizzi et al., 2002). In vitro, glial cells contain a two- to four-fold higher concentration of Cr than do neurons (Urenjak et al., 1993), although curiously white matter (WM) Cr levels are lower than those of gray matter (GM) in the normal brain. It has been suggested that the sum of Cr and phosphocreatine is relatively constant in the human brain, and for this reason Cr is often used as a reference signal, and it is a common practice for metabolite ratios to be expressed as a ratio relative to Cr. However, with the development of quantitative analysis techniques, it is clear that total Cr is not constant, both in different brain regions and in pathological processes, so the assumption of Cr as an invariate reference signal should be used with caution. Absolute metabolite quantification techniques are discussed in detail in Chapter 2. Lactate In normal human brain, lactate (Lac, 1.33 ppm) is below (or at the limit of) detectability in most studies. Any detectable increase in Lac can therefore be considered abnormal, except perhaps in CSF where it may be detectable at a low level in normal subjects with prominent ventricles. Increased Lac is usually the result of deranged energy metabolism, and has been observed in ischemia (both acute (highest) and chronic (Petroff et al., 1992; Barker et al., 1994)), brain tumors (Alger et al., 1990), mitochondrial diseases (Mathews et al., 1993), and other conditions. Small elevations of Lac have also been reported in the visual cortex (VC) during photic stimulation (Prichard et al., 1991), believed to be due to increased non-oxidative glycolysis, but this effect does not appear to be particularly reproducible (Merboldt et al., 1992). Myo-inositol At short TEs, additional compounds are detected which are not visible at long TEs, either because of short T2 relaxation times and/or the dephasing effects of J-coupling (Figure 1.1(c)). One of the largest signals occurs from myo-inositol (mI) at

Fundamentals of MR spectroscopy

3.56 ppm. mI is a pentose sugar, which is part of the inositol triphosphate intracellular second messenger system. Levels have been found to be reduced in hepatic encephalopathy (Ross et al., 1994b), and increased in Alzheimer’s dementia (Shonk et al., 1995) and demyelinating diseases (Kruse et al., 1993). The exact pathophysiological significance of alterations in mI is uncertain. A leading hypothesis is that elevated mI reflects increased populations of glial cells which are known to express higher levels of this metabolite than neurons (Brand et al., 1993; Flogel et al., 1994); this may be related to differences in mI/Na co-transporter activity which appears to play a key role in astrocyte osmoregulation (Strange et al., 1994). This would explain chronic disturbance in mI both in degenerative and inflammatory disease, and transiently in hypo- and hyper-osmolar states.

Glutamate and glutamine Glutamate (Glu) and glutamine (Gln) are difficult to separate in proton spectra at 1.5 T (and are often labeled as a composite peak glutamine and glutamate Glx), although some authors have attempted to distinguish them (Kreis et al., 1992b). At very high fields (at 4 T or above), the C4 resonances of Glu and Gln start to become resolved. Increased cerebral Gln has been found in patients with liver failure (hepatic encephalopathy (Ross et al., 1994b), and Reye’s syndrome (Kreis et al., 1995a)) as the result of increased blood ammonia levels, which increases Gln synthesis.

Less commonly detected compounds A survey of the literature reveals some 25 additional compounds that have been assigned in proton spectra of the human brain. Some of these compounds are present in normal circumstances, but because they are very small and/or overlapping peaks it is usually difficult to detect them. Some examples of these include NAAG, aspartate, taurine, scylloinositol, betaine, ethanolamine, purine nucleotides, histidine, glucose, and glycogen (van Zijl and Barker, 1997). Other compounds are yet more difficult to detect and require the use of special spectral editing

pulses (beyond the scope of the current chapter) to detect; example of these include -amino-butyric acid (GABA), glutathione, and certain macromolecules (Rothman et al., 1993; Terpstra et al., 2003). Under disease conditions, other compounds may become detectable because their concentration is pathologically increased. Examples of compounds that have been detected under pathological conditions include the ketone bodies -hydroxy-butyrate and acetone (Seymour et al., 1999; Pan et al., 2001), and other compounds such as phenylalanine (Phe) (in phenylketoneuria (PKU) (Kreis et al., 1995b)), galactitol, ribitol, arabitol in “polyol disease” (van der Knaap et al., 1999), succinate, pyruvate, alanine, glycine, and threonine. Finally, exogenous compounds which are able to cross the blood–brain barrier (BBB) may also reach sufficiently high concentrations to be detected by proton MRS. Examples of exogenous compounds, sometimes termed “xenobiotics”, include the drug delivery vehicle propan-1,2-diol (Cady et al., 1994), mannitol (used to reduce swelling and edema in neurosurgical procedures and intensive care), ethanol (Meyerhoff et al., 1996), and the health food supplement methylsulfonylmethane (MSM) (Lin et al., 2001). In order for a compound to be detectable by proton MRS in vivo, a rule of thumb is that its concentration should be 1 mM or greater, and it should be a small, mobile molecule. Hence large and/or membrane-associated molecules will not be detected. The ability to detect and quantify compounds should increase with increasing magnetic field strength; for instance, a recent study of the normal human brain at 7 T was able to detect more than 14 different compounds (Figure 1.4). Recently, measurements of brain temperature have also been made using the water–NAA chemical shift difference (the water chemical shift has a 0.01 ppm/°C temperature dependence) (Cady et al., 1995).

Technical issues: spatial localization Single-voxel techniques Generally, two different approaches are used for proton spectroscopy of the brain: single-voxel methods

13

14

Peter B. Barker

NAA

Vol  8 ml

Cr PCr NAAG

Cr PCr

Cho GSH

Glu Gln

MM mI

mI

4.5

4.0

Gln

NAA Asp

Tau

3.5

3.0

2.5

Glu Lac

2.0

1.5

1.0 ppm

Fig. 1.4 Proton MR spectrum from parietal WM measured at 7 T in the normal human brain. STEAM, TE 56 ms, TM 5 32 ms, TR 5 s, voxel size 5.8 ml, 160 averages (scan time approximately 13 min), resolution enhancement by a shifted Gaussian function. Inset: gradient echo (GE) transverse MRI with the voxel location. Reproduced with permission from Tkac et al. (2001).

based on the stimulated echo acquisition mode (STEAM) (Frahm et al., 1989) or point resolved spectroscopy (PRESS) (Bottomley, 1984) pulse sequences, or spectroscopic imaging (SI) (also known as chemical shift imaging (CSI)) studies usually done in two dimensions using a variety of different pulse sequences (spin-echo (SE), PRESS) (Brown et al., 1982; Luyten et al., 1990; Duyn et al., 1993). The basic principle underlying single-voxel localization techniques is to use (usually) three mutually orthogonal slice selective pulses and design the pulse sequence to collect only the echo signal from the point (voxel) in space where all three slices intersect (Figure 1.5). The two most commonly used sequence are called STEAM (Frahm et al., 1989) and PRESS. In STEAM (Figure 1.5(b)), three 90° pulses are used, and the stimulated echo is collected. All other signals (echoes) should be dephased by the large crusher gradient applied during the so-called mixing time (TM, from analogy with the two-dimensional

(2D) NMR nuclear overhauser efffect (NOESY) pulse sequence (Ernst et al., 1987)). Crusher gradients applied during TE on selected gradient channels are necessary for consistent formation of the stimulated echo and removal of unwanted coherences. In PRESS, the second and third pulses are refocusing (180°) pulses, and crusher gradients are applied around these pulses to select the desired SE signal arising from all three RF pulses, and dephasing unwanted coherences. STEAM and PRESS have been the subject of a detailed comparison (Moonen et al., 1989); they are generally similar but differ in a few key respects: 1. Slice profile (i.e. sharpness of edges of voxel): STEAM is somewhat better because it is easier to produce a 90° pulse with a sharp slice profile than a 180° pulse. 2. SNR: Provided that equal volumes of tissue are observed and using the same parameters (repetition time (TR), TE, number of averages, etc.),

Fundamentals of MR spectroscopy

(a)

(b)

90°

90°

90°

RF

TE/2

TM

TE/2

gX gY gZ (c)

90°

180°

180°

RF

TE/4

TE/2

TE/4

gX gY gZ Fig. 1.5 Single-voxel pulse sequences: (a) schematic illustration of three orthogonal slice selective pulses. The size and position of the voxel is controlled by the frequency and bandwidth of the slice selective pulses, as well as by the amplitude of the associated slice selective field gradients, (b) STEAM and (c) PRESS. Note that simplified diagrams are presented which do not show all crusher gradients, gradient lobes and RF pulse shapes.

PRESS should have approximately a factor of two better SNR than STEAM, because the stimulated echo is formed from only half the available equilibrium magnetization. 3. Minimum TE: STEAM should have a shorter minimum TE than PRESS, since it uses a TM time

period, and shorter 90° than 180° pulses may be possible. 4. Water suppression: STEAM may have slightly better water suppression factors, because water suppression (cf. below) pulses can be added during the TM period (this period does not occur in PRESS). Also, STEAM may have less spurious water signal from the 90° slice selective pulses than the 180° pulses in PRESS. 5. Coupled spin systems and zero-quantum interference: The complex phenomena that can occur in coupled spin systems (e.g. Lac, Glu, etc.), namely modulation of the echo signal by scalar couplings, and/or the creation of zero- or multiple-quantum coherences, may occur with both sequences. However, the detailed dependence of these compounds’ signals on TE and other experimental parameters will be different for STEAM and PRESS. STEAM is more susceptible for the creation of (usually unwanted) zero-quantum coherence because it uses 90° pulses. It should be recognized that the differences listed above are fairly subtle, and generally STEAM or PRESS are essentially interchangeable in clinical brain spectroscopy, and the choice of sequence in practice often mainly depends on the particular availability from the MRI vendor. It is important to recognize the importance of accurate spatial localization and suppression of signal from outside the desired voxel. The volume of the human head is two to three orders of magnitude larger than that of the volume of interest (VOI). Even a few percent outer-volume contamination can have a disastrous effect on spectral quality, particularly if field homogeneity is poor in remote regions, and if they contain large water and lipid signals. Methods for maximizing out-of-volume suppression (saturation pulses, optimal use of crusher gradients) are discussed in Chapter 3. Multiple-voxel (SI) techniques While single-voxel techniques are popular in clinical practice for several reasons (they have short scan times, are widely available, can be done at short TE, and are relatively easy to use and interpret), they do also suffer some limitations. Probably the greatest

15

16

Peter B. Barker

single limitation is the lack of ability to determine spatial heterogeneity of spectral patterns (often very important in brain tumors, for instance), and the fact that only a small number of brain regions can be covered within the time constraints of a normal clinical MR examination. Therefore, there has been considerable effort over the last decade and a half to develop clinically feasible MR spectroscopic (MRSI) techniques. Early attempts at MRSI in the human brain used onedimensional (1D)-MRSI (i.e. phase encoding in a single direction) (Petroff et al., 1992), and while these demonstrated proof-of-principle, generally 1D localization is insufficient for detailed studies of focal brain pathology. Therefore, MRSI techniques were extended to two dimensions by using phase-encoding gradients in two directions (Luyten et al., 1990; Duyn et al., 1993) (Figure 1.6), or, subsequently, with full three-dimensional (3D) encoding (Nelson et al., 1999). One widely used 2D-MRSI pulse sequence combines multi-slice capability with full-slice coverage using a combination of spin-echoes and outervolume suppression (OVS) pulses (Duyn et al., 1993). The sequence is illustrated schematically in Figure 1.7. Compared to PRESS-MRSI, this sequence can cover the whole slice out to the edge of the cortex, and also record multiple slices. Also, by interleaving multiple slices within one TR, the sequence is very efficient in terms of data collection, and generally, can acquire data at higher spatial resolution and brain coverage than comparable sequences using 3D-MRSI. One potential caveat when attempting wide coverage of brain regions, however, is the difficulty of obtaining sufficient magnetic field homogeneity over the full volume of the brain (simultaneously). For this reason, the sequence of Figure 1.7 is usually performed at long TE (e.g. 140 or 280 ms). These are optimum TEs for detecting the Lac signal (modulation due to scaler coupling causes the Lac signal to be inverted at TE 140 ms). Generally, field homogeneity requirements are less stringent for long TE spectra than short TE, because the spectra are simpler with less overlapping resonances (cf. Chapter 3). Recent technical advances to address this issue include slice-by-slice shimming (i.e. dynamic adjustment of the shim currents within the TR time period for each slice) and the

development of high-order shimming in vivo. An excellent approach for localized shimming in vivo is the fast automatic shimming technique by mapping along projections (FASTMAP) method of Gruetter (1993). An example of a representative multi-slice MRSI data set performed at long TE is given in Figure 1.8. Generally, good quality spectra can be obtained from most parts of the brain, with insufficient field homogeneity only present in regions adjacent to air–tissue interfaces inside the head (e.g. artifacts can be seen in the anterior, mesial temporal lobes and inferior frontal lobe). MRSI experiments are relatively time consuming, because there are usually a large number of phaseencoding gradient steps to collect. This is particularly true for 2D- or 3D-MRSI experiments that require both high spatial resolution and full (or large) brain coverage. Therefore, there have been various methods proposed to decrease scan time (Duyn and Moonen, 1994; Posse et al., 1995). The discussion of these methods is beyond the scope of this chapter, however increasingly it is expected that fast MRSI techniques will become used for human spectroscopy, such that ultimately MRSI sequences may have similar scan times to single-voxel methods (e.g. 5–10 min, cf. Table 1.2). Comparison of single-voxel vs. SI techniques Usually, but not exclusively, single-voxel scans are recorded at short TEs (35 ms) while MRSI studies are done at long TEs (e.g. TE  135–140 ms). Short TE spectra contain signals from more compounds and have better SNRs, but also have worse water and lipid contamination. Long TE spectra have lower SNR, fewer detectable compounds, and variable amount of T2-weighting, but are usually better resolved spectra with flatter baselines. Lac is usually best detected at long TEs (e.g. TE  140 or 280 ms, so that the J-modulation is rephased) to distinguish it from lipid signals. The relative advantages and disadvantages of single-voxel vs. SI techniques are listed in Table 1.1. The choice of method depends (in addition to availability) on the information required in the particular medical or research application. For instance, if spectroscopy is being used to search for

Fundamentals of MR spectroscopy

(a) RF TE

gX

gY gZ

(b) T2MRI

Cho

NAA

Cho Cho

NAA

Cr

Cr

ppm 4.0

3.0

NAA

2.0

1.0

ppm 4.0

3.0

2.0

1.0

Fig. 1.6 Pulse sequence for 2D PRESS-MRSI. (a) PRESS sequence (red) is used to select a large region of interest (ROI) within the brain (but avoiding unwanted lipid signals in the skull and scalp on this coronal example), and then phase-encoding gradients (green) are applied in two dimensions to encode spatial information inside the excited volume. Data is processed by 3D Fourier transformation (two spatial and one time domains). Full crusher gradients are shown, including those associated with the initial (black) water suppression pulse. Slice selective gradients are indicated in blue. Adapted with permission from (Moonen et al., 1992). (b) An example of the 2D PRESS MRSI pulse sequence in a 14-year old female presenting with seizures with a lesion in the left mesial temporal lobe. Data are presented as metabolic images of NAA and Cho, as well as selected spectra from the left and right hippocampi (voxel positions indicated on Cho images). The lesion has elevated Cho and Cr, and low NAA, typical of a glioma (and atypical for mesial temporal sclerosis which usually shows a selective reduction in NAA only).

the location of a stroke or a seizure focus, SI would be preferable since this generates maps of metabolite levels which can be screened for abnormalities in different locations. Alternatively, if the issue is to observe changes in compounds such as Gln/Glu or mI, which can only be detected in short TE spectra,

in global or diffuse diseases such as hepatic encephalopathy, then short TE single-voxel spectroscopy would be the method of choice. Other factors include the length of time available, and whether or not the required voxel location would be better viewed using localized shimming (i.e. single

17

18

Peter B. Barker

(a)

(b) CHESS

OVS

OVS

180°

90°

RF

gX gY

gZ TE

Slice 1

Slice 3

Slice 2

Slice 4

TR Fig. 1.7 (a) Schematic illustration of pulse sequence for multi-slice MRSI pulse with CHESS water suppression and outer-volume saturation bands for lipid suppression (Duyn et al., 1993) (for clarity, not all crusher gradients are illustrated). A slice selective spin-echo sequence is used, with interleaved acquisition (in this example) or four slices within one TR period. (b) The orientation and locations of the eight OVS pulses are schematically illustrated on sagittal and axial views; an octagonal pattern is prescribed in order to saturated as much peri-cranial lipid as possible while signal from brain is un-perturbed. Ideally, sharp profile, high bandwidth pulses (to minimize chemical shift effects) should be used for OVS.

voxel) or not. Short TE SI is becoming available in commercial sequences and as the techniques become more refined, will provide spatial maps of a greater range of metabolites. Water and lipid suppression Brain metabolite levels are on the order of 10 mM or less, whereas protons in brain water are approximately 80 M, and lipids in peri-cranial fat are also present in very high concentrations. Therefore, water and lipid suppression techniques are essential in proton spectroscopy in order to observe reliably the much smaller metabolite signals. Numerous methods for solvent (water) suppression have been developed in high-resolution NMR spectroscopy, and some of these methods have been applied to in vivo spectroscopy. The most common approach is to pre-saturate the water signal using

frequency-selective, 90° pulses (chemical shiftselective water suppression (CHESS) pulses (Haase et al., 1985)) prior to localization pulse sequence (Figure 1.7). By using more than one pulse, and with correct choice of flip angles (Moonen and van Zijl, 1990; Ogg, 1994), very good suppression factors can be attained (1000). Lipid suppression can be performed in several different ways. One approach is to avoid exciting the lipid signal using, e.g. STEAM or PRESS localization to avoid exciting lipid-containing regions (Figure 1.5). Alternatively (or in addition), OVS pulses can be used to pre-saturate the lipid signal (Duyn et al., 1993) (Figure 1.7). An inversion pulse can also be used for lipid suppression, exploiting the difference in T1s between lipid (typically 300 ms) and metabolites (typically 1000–2000 ms) (Spielman et al., 1992). Choice of a short inversion time (TI) of around 200 ms ( T1 ln[2]) will selectively null the lipid signal,

Fundamentals of MR spectroscopy

Table 1.2. Comparison of single-voxel and multi-voxel MRSI methodologies Single voxel

MRSI

TE

Short or long

Typical voxel sizes (cm3) Typical scan times (min) Shimming Water/lipid suppression Processing/quantitation

4–20 5–10 Localized Better Simple processing, can be quantified 3 or 4 at most

Usually long, can be short if field homogeneity is good (e.g. small region of coverage) 1–4 6–30 Global Worse Processing and quantification more time consuming Many voxels

Multiple voxels

1

4

NAA

Cho Cr

2

5

3

6

ppm 4.0

3.0

2.0 Cho

1 2

ppm 4.0 Cr

3.0

2.0

1.0 NAA

ppm 4.0

3.0

2.0

1.0

Lac

4 5

3

1.0

7

7

6

Fig. 1.8 MRSI data recorded using the pulse sequence of Figure 1.7. Metabolic images of Cho, Cr, NAA and lactate from one-slice at the level of the lateral ventricles in a normal 49 year adult are presented, as well as representative spectra from different brain regions. Scan parameters were TR 2300 ms, TE 272 ms, 15 mm slice thickness, field of view (FOV) 24 cm, matrix size 32 32, scan time 30 min with circular k-space encoding. The nominal voxel size is 0.8 cm3. NAA is fairly evenly distributed at this level, while Cho shows an increase from posterior to anterior brain regions (e.g. cf. posterior (6) to anterior (4) WM, and splenium (3) to genu (2) of corpus callosum). Cho is also lower in the lateral GM region (7) compared to WM voxels. No lactate is detectable above the noise floor of the data set.

19

20

Peter B. Barker

while most of the metabolite signal remains inverted. In MRSI, it is also possible to reduce lipid artifacts by post-processing methods (Haupt et al., 1996). Since both water and lipid resonances have shorter T2 relaxation times than many metabolites, suppression factors are also usually better in long TE compared to short TE spectra.

Data analysis and quantification Peak area measurements in in vivo spectroscopy are complicated by resonance overlap, baseline distortions, and non-Lorentzian lineshapes. Various methods have been used to measure peak areas, ranging from simple integration to fitting algorithms in the time or frequency domains (Raphael, 1991; de Beer and van Ormondt, 1992). One of the more widely used methods for spectral quantitation in recent years is the linear combination model (LC model) method developed by Provencher et al. The LC model fits the in vivo spectrum as a combination of pure, model spectra from each of the expected compounds in the brain (Provencher, 1993). The model also includes automatic phase correction and baseline correction, or the baseline may also be modeled as a combination of macromolecular resonances. Provided that each scanner is properly calibrated with the appropriate model solutions, the program returns metabolite concentrations as well as estimates of uncertainty (e.g. Cramer–Rao lower bounds). Quantification of in vivo spectra is discussed in detail in Chapter 2. Quantification is important for several reasons, but particularly so in clinical cases where all metabolites (or all regions of the brain) may be abnormal. Quantification methods based on internal or external standards have been extensively developed and tested for single-voxel spectroscopy (Henriksen, 1995) and can be used routinely. With care, it is also possible to quantify MR spectroscopic imaging data (Soher et al., 1996). Occasionally, ratios of peak areas may also be useful, for instance to account for partial volume effects (PVE) or to enhance spectroscopic “contrast” in conditions where metabolites may change in opposite directions (e.g. Cho increases, NAA decreases).

Anatomical variations in brain spectra: changes associated with brain development and aging Evidently, it is important to establish normal spectral variations associated with age and anatomical location in the healthy control population. Numerous studies have looked at anatomical variations in brain spectra, usually in young adult subjects. At the level of the lateral ventricles and above, brain spectra appear to be fairly homogeneous, with spectra which are characteristic of GM and WM (Kreis et al., 1993a; Michaelis et al., 1993; Hetherington et al., 1994; Soher et al., 1996). Depending on the quantification technique used (and if partial volume correction is applied or not), generally the Cho and NAA signals are found to be marginally higher in WM than cortical GM, with WM showing a lower Cr level than GM. At the level of the third ventricle and below, significant anatomical variations exist in brain spectra. High levels of Cho are found in the insular cortex, and in the region of the hypothalamus. Occipital Cho in the region of the visual cortex is generally low. The pons has high levels of NAA and Cho, and low levels of Cr, perhaps due to its high density of fiber bundles. Cerebellar levels of Cr and Cho are significantly higher than supratentorial values (Michaelis et al., 1993), and temporal lobe has been reported to have lower NAA values (Breiter et al., 1994). Significant anterior–posterior differences have also been reported in normal hippocampal metabolite concentrations, with low NAA and high Cho in the anterior regions of the hippocampus (Vermathen et al., 1997). Relatively fewer papers have addressed the issue of gender differences or metabolic asymmetries in normal brain. However, it appears that there are minimal spectral differences (Charles et al., 1994) with regard to these variables, at least in young adults (Figure 1.9). Several papers have been published on the changes that occur in proton spectra in the developing brain, and most of the results are in good agreement (van der Knaap et al., 1990; Huppi et al., 1991; Kreis et al., 1993b; Kimura et al., 1995). At birth, NAA is low, while Cho and mI are high, and over the first 1–2 years there is a gradual normalization towards adult values (Figure 1.10) (Kreis et al., 1993b).

(a)

NAA

NAA Cho Cr

Cho Cr

50

0 Hz

50 ppm 4.0

3.0 T1 MRI

(b)

2.0

1.0 Cho

4.0

3.0

2.0

Cr

1.0 NAA

3 2 1 3

Pons 2

Cerebellar hemisphere

Vermis 1

Fig. 1.9 Multi-slice MRSI of (a) temporal lobe and (b) posterior fossa brain regions recorded using the sequence of Figure 1.7. (a) MRSI in an oblique-axial plane parallel to the long axis of the temporal lobe. NAA levels decrease and Cho levels increase in the anterior mesial temporal lobe relative to posterior. Note also the severe field inhomogeneity caused by the sinuses in the frontal region on the B0 field map. Field homogeneity is also perturbed superior to the auditory canals. (b) MRSI of the posterior fossa. Note the high Cho and Cr levels in the cerebellar vermis and hemispheres. The pons also shows high levels of Cho, but low Cr.

22

Peter B. Barker

Parietal WM

Cr mI Glx

4

Occipital GM 4 days

5 months

5 months

4 years

4 years

NA

Cr Ch NA Glx

3

4 days

2 ppm

adult LA CH2 CH3

1

0

4

3

2 ppm

1

0

Fig. 1.10 Developmental changes in the human brain. Spectra recorded at short TE (TE 35 ms) are shown in posterior white and GM regions as a function of post-partum age; 4 days, 5 months, 4 years and adult. Note the high levels of Cho and mI at the earliest time point, which decline over the first 2 years of life. Also, NAA is low at birth and increases rapidly. By 4 years of age (in these brain regions), spectra are indistinguishable from those in adults. Reproduced and adapted with permission from (Kreis et al., 1993b).

Similar patterns are seen for both GM and WM, although regional developmental changes have yet to be studied in detail (e.g. using SI). Recent studies have suggested that although the major changes occur within the first year of life, slower changes occur thereafter, with full adult values not being reached until about 20 years of age (Pouwels et al., 1997), and that some regions (e.g. frontal lobe) may develop more slowly than posterior regions (cf. also Chapter 40). In contrast to studies of developing brain, fewer studies of normal aging have been reported, and the results are less concordant. Some groups find lower

NAA with increasing age (Christiansen et al., 1993; Lim and Spielman, 1997), which may reflect neuronal loss, while others find no changes (Chang et al., 1996; Soher et al., 1996). In one study, NAA was only reduced in subjects who also had cerebral atrophy as identified by MRI (Lundbom et al., 1997). Some groups have also found increased levels of Cr or Cho in older subjects, perhaps reflecting increased gliosis (Chang et al., 1996; Soher et al., 1996). This area is discussed further in Chapter 34. The discrepancies between different studies could be due to many different technical factors in data collection and analysis, but may also reflect the wide

Fundamentals of MR spectroscopy

physiological variations of normal human aging. More studies are required to definitively establish the spectroscopic characteristics of normal aging, but it is apparent that the changes associated with normal aging are appreciably more subtle than those associated with brain development. Due to significant technique-related, regional, or age-related changes, it is advisable that spectroscopic studies should have carefully age- and anatomically-matched spectra from control subjects for comparison. In addition, spectroscopic scans of focal brain lesions (for instance) are often much easier to interpret if spectra from normal brain in the contralateral hemisphere are available for comparison.

Summary Proton MRS and MRSI are now mature methodologies that can be applied routinely on 1.5 T (and higher) imaging systems for the study of neurological disease. The subsequent chapters of this book cover spectral quantification techniques, artifacts and pitfalls, and the clinical applications of these techniques. It is expected that advances in pulse sequence design, analysis methods, and the use of high magnetic fields will continue to occur.

REFERENCES Alger JR, Frank JA, Bizzi A, Fulham MJ, DeSouza BX, Duhaney MO, Inscoe SW, Black JL, van ZP, Moonen CT, et al. 1990. Metabolism of human gliomas: assessment with H-1 MR spectroscopy and F-18 fluorodeoxyglucose PET [see comments]. Radiology 177: 633–641. Arnold DL, Shoubridge EA, Emrich J, Feindel W, Villemure JG. 1989. Early metabolic changes following chemotherapy of human gliomas in vivo demonstrated by phosphorus magnetic resonance spectroscopy. Invest Radiol 24: 958–961. Barker PB, Bryan RN, Kumar AJ, Naidu S. 1992. Proton NMR spectroscopy of Canavan’s disease. Neuropediatrics 23: 263–267. Barker PB, Gillard JH, van Zijl PCM, Soher BJ, Hanley DF, Agildere AM, Oppenheimer SM, Bryan RN. 1994. Acute stroke: evaluation with serial proton magnetic resonance spectroscopic imaging. Radiology 192.

Barker PB. 2001. N-acetyl aspartate – a neuronal marker? Ann Neurol 49: 423–424. Bhakoo KK, Pearce D. 2000. In vitro expression of N-acetyl aspartate by oligodendrocytes: implications for proton magnetic resonance spectroscopy signal in vivo. J Neurochem 74: 254–262. Bizzi A, Ulug AM, Crawford TO, Passe T, Bugiani M, Bryan RN, Barker PB. 2001. Quantitative proton MR spectroscopic imaging in acute disseminated encephalomyelitis. Am J Neuroradiol 22: 1125–1130. Bizzi A, Bugiani M, Salomons GS, Hunneman DH, Moroni I, Estienne M, Danesi U, Jakobs C, Uziel G. 2002. X-linked creatine deficiency syndrome: a novel mutation in creatine transporter gene SLC6A8. Ann Neurol 52: 227–231. Bloch F. 1946. Nuclear induction. Phys Rev 70: 460–474. Bottomley P. 1984. In U.S. Patent, Vol. 4 480 228 USA. Breiter SN, Arroyo S, Mathews VP, Lesser RP, Bryan RN, Barker PB. 1994. Proton magnetic resonance spectroscopy in patients with seizure disorders. Am J Neuroradiol 15: 373–384. Brand A, Richter-Landsberg C, Leibfritz D. 1993. Multinuclear NMR studies on the energy metabolism of glial and neuronal cells. Dev Neurosci 15(305): 289–298. Brown T, Kincaid B, Ugurbil K. 1982. NMR Chemical shift imaging in three dimensions. Proc Natl Acad Sci USA, 79: 3523–3526. Burlina AP, Ferrari V, Facci L, Skaper SD, Burlina AB. 1997. Mast cells contain large quantities of secretagogue-sensitive N-acetylaspartate. J Neurochem 69: 1314–1317. Cadoux HT, Blackledge MJ, Rajagopalan B, Taylor DJ, Radda GK. 1989. Human primary brain tumour metabolism in vivo: a phosphorus magnetic resonance spectroscopy study. Br J Cancer 60: 430–436. Cady EB, Lorek A, Penrice J, Reynolds EO, Iles RA, Burns SP, Coutts GA, Cowan FM. 1994. Detection of propan-1,2-diol in neonatal brain by in vivo proton magnetic resonance spectroscopy. Magn Reson Med 32: 764–767. Cady EB, D’Souza PC, Penrice J, Lorek A. 1995. The estimation of local brain temperature by in vivo 1H magnetic resonance spectroscopy. Magn Reson Med 33: 862–867. Cecil KM, Salomons GS, Ball Jr WS, Wong B, Chuck G, Verhoeven NM, Jakobs C, DeGrauw, TJ. 2001. Irreversible brain creatine deficiency with elevated serum and urine creatine: a creatine transporter defect? Ann Neurol 49: 401–404. Chang L, Ernst T, Poland RE, Jenden DJ. 1996. In vivo proton magnetic resonance spectroscopy of the normal aging human brain. Life Sci 58: 2049. Charles HC, Lazeyras F, Krishnan KRR, Boyko OB, Patterson LJ, Doraiswamy PM, McDonald WM. 1994. Proton spectroscopy of human brain: effects of age and sex. Prog NeuroPsychopharmacol and Biol Psychiat 18: 995.

23

24

Peter B. Barker

Christiansen P, Toft P, Larsson HBW, Stubgaard M, Henriksen O. 1993. The concentration of N-acetyl aspartate, creatine  phosphocreatine, and choline in different parts of the brain in adulthood and senium. Magn Reson Imaging 11: 799. Davie CA, Hawkins CP, Barker GJ, Brennan A, Tofts PS, Miller DH, McDonald WI. 1993. Detection of myelin breakdown products by proton magnetic resonance spectroscopy. Lancet 341: 630–631. de Beer R, van Ormondt D. 1992. In NMR Basic Principles and Progress, Vol. 26 (Eds., Diehl P, Fluck E, Günther H, Kosfeld R, Seelig J, Rudin M), Springer-Verlag, Berlin, pp. 201–248. De Stefano N, Narayanan S, Francis GS, Arnaoutelis R, Tartaglia MC, Antel JP, Matthews PM, Arnold DL. 2001. Evidence of axonal damage in the early stages of multiple sclerosis and its relevance to disability. Arch Neurol 58: 65–70. Duyn JH, Gillen J, Sobering G, van Zijl PCM, Moonen CTW. 1993. Multislice proton MR spectroscopic imaging of the brain. Radiology 188: 277– 2 82 . Duyn JH, Moonen CTW. 1994. Fast proton spectroscopic imaging of the human brain using multiple spin-echoes. Magn Reson Med 30: 409–414. Ernst R, Bodenhausen G, Wokaun A. 1987. Principles of Nuclear Magnetic Resonance in One and Two Dimensions, Oxford University Press, Oxford. Flogel U, Willker W, Leibfritz D. 1994. Regulation of intracellular pH in neuronal and glial tumour cells, studied by multinuclear NMR spectroscopy. NMR Biomed 7: 157–166. Frahm J, Bruhn H, Gyngell ML, Merboldt KD, Hanicke W, Sauter R. 1989. Localized high-resolution proton NMR spectroscopy using stimulated echoes: initial applications to human brain in vivo. Magn Reson Med 9: 79–93. Frahm J, Michaelis T, Merboldt K.-D, Hanicke W, Gyngell ML, Bruhn H. 1991. On the N-acetyl methyl resonance in localized 1H NMR spectra of the human brain in vivo. NMR Biomed 4: 201–204. Gill SS, Small RK, Thomas DG, Patel P, Porteous R, van Bruggen N, Gadian DG, Kauppinen RA, Williams SR. 1989. Brain metabolites as 1H NMR markers of neuronal and glial disorders. NMR Biomed 2: 196–200. Gill SS, Thomas DG, Van BN, Gadian DG, Peden CJ, Bell JD, Cox IJ, Menon DK, Iles RA, Bryant DJ. 1990. Proton MR spectroscopy of intracranial tumours: in vivo and in vitro studies. J Comput Assist Tomogr 14: 497–504. Gruetter R. 1993. Automatic, localized in vivo adjustment of all first- and second-order shim coils. Magn Reson Med 29: 804–811. Guimaraes A, Schwartz P, Prakash MR, Carr CA, Berger UV, Jenkins BG, Coyle JT, Gonzalez RG. 1995. Quantitative in vivo 1H nuclear magnetic resonance spectroscopic imaging of neuronal loss in rat brain. Neuroscience 69: 1095.

Gutowsky HS, McCall DW, Slichter CP. 1951. Phys Rev 84: 589. Haase A, Frahm J, Hanicke W, Matthei D. 1985. 1H NMR chemical shift selective imaging. Phys Med Biol 30: 341–344. Haupt CI, Schuff N, Weiner MW, Maudsley AA. 1996. Removal of lipid artifacts in 1H spectroscopic imaging by data extrapolation. Magn Reson Med 35: 678–687. Henriksen O. 1995. In vivo quantitation of metabolite concentrations in the brain by means of proton MRS. NMR Biomed 8: 139–148. Hetherington HP, Mason GF, Pan JW, Ponder SL, Vaughan JT, Twieg DB, Pohost GM. 1994. Evaluation of cerebral gray and white matter metabolite differences by spectroscopic imaging at 4.1T. Magn Reson Med 32: 565–571. Huppi PS, Posse S, Lazeyras F, Burri R, Bossi E, Herschkowitz N. 1991. Magnetic resonance in preterm and term newborns: 1H-spectrscopy in developing brain. Pediatric Res 30: 574–578. Kimura H, Fujii Y, Itoh S, Matsuda T, Iwasaki T, Maeda M, Konishi Y, Ishii Y. 1995. Metabolic Alterations in the neonate and infant brain during development: evaluation with proton MR spectroscopy. Radiology 194: 483–489. Kreis R, Ross BD, Farrow NA, Ackerman Z. 1992a. Metabolic disorders of the brain in chronic hepatic encephalopathy detected with H-1 MR spectroscopy. Radiology 182: 19–27. Kreis R, Ross BD, Farrow NA, Ackerman Z. 1992b. Metabolic disorders of the brain in chronic hepatic encephalopathy detected with H-1 MR spectroscopy. Radiology 182: 19–27. Kreis R, Ernst T, Ross B. 1993a. Absolute quantitation of water and metabolites in the human brain. II metabolite concentrations. J Magn Reson B 102: 9–19. Kreis R, Ernst T, Ross BD. 1993b. Development of the human brain: In vivo quantification of metabolite and water content with proton magnetic resonance spectroscopy. Magn Reson Med 30: 424–437. Kreis R, Pfenninger J, Herschkowitz N, Boesch C. 1995a. In vivo proton magnetic resonance spectroscopy in a case of Reye’s syndrome. Intens Care Med 21: 266–269. Kreis R, Pietz J, Penzien J, Herschkowitz N, Boesch C. 1995b. Identification and quantitation of phenylalanine in the brain of patients with phenylketonuria by means of localized in vivo 1H magnetic-resonance spectroscopy. J Magn Reson B 107: 242–251. Kruse B, Hanefeld F, Christen HJ, Bruhn H, Michaelis T, Hanicke W, Frahm J. 1993. Alterations of brain metabolites in metachromatic leukodystrophy as detected by localized proton magnetic resonance spectroscopy in vivo. J Neurol 241: 68–74. Levine SR, Helpern JA, Welch KM, Vande LA, Sawaya KL, Brown EE, Ramadan NM, Deveshwar RK, Ordidge RJ. 1992.

Fundamentals of MR spectroscopy

Human focal cerebral ischemia: evaluation of brain pH and energy metabolism with P-31 NMR spectroscopy. Radiology 185: 537–544. Lim KO, Spielman DM. 1997. Estimating NAA in cortical gray matter with applications for measuring changes due to aging. Magn Reson Med 37: 372–377. Lin A, Nguy CH, Shic F, Ross BD. 2001. Accumulation of methylsulfonylmethane in the human brain: identification by multinuclear magnetic resonance spectroscopy. Toxicol Lett 123: 169–177. Lundbom N, Barnett A, Bonavita S, Patronas N, Rajapakse J, Tedeschi G, Di Chiro G. 1997. In ISMRM 5th Scientific Meeting and Symposium, Vancouver, BC, pp. 1209. Luyten PR, Groen JP, Vermeulen JW, den HJ. 1989. Experimental approaches to image localized human 31P NMR spectroscopy. Magn Reson Med 11: 1–21. Luyten PR, Marien AJ, Heindel W, van GP, Herholz K, den HJ, Friedmann G, Heiss WD. 1990. Metabolic imaging of patients with intracranial tumors: H-1 MR spectroscopic imaging and PET. Radiology 176: 791–799. Martin E, Capone A, Schneider J, Hennig J, Thiel T. 2001. Absence of N-acetylaspartate in the human brain: impact on neurospectroscopy? Ann Neurol 49: 518–521. Mathews PM, Andermann F, Silver K, Karpati G, Arnold DL. 1993. Proton MR spectroscopic characterization of differences in regional brain metabolic abnormalities in mitochondrial encephalomyopathies. Neurology 43: 2484–2490. Merboldt K-D, Bruhn H, Hanicke W, Michaelis T, Frahm J. 1992. Decrease of glucose in the human visual cortex during photic stimulation. Magn Reson Med 25: 187–194. Meyerhoff DJ, Rooney WD, Tokumitsu T, Weiner MW. 1996. Evidence of multiple ethanol pools in the brain: an in vivo proton magnetization transfer study. Alcohol Clin Exp Res 20: 1283–1288. Michaelis T, Merboldt K-D, Bruhn H, Hanicke W, Frahm J. 1993. Absolute concentrations of metabolites in the adult human brain in vivo: quantification of localized proton MR spectra. Radiology 187: 219–227. Moonen CT, van Zijl PCM, Gillen J, Daly P, von Kienlin M, Wolf J, Cohen J. 1989. Comparison of single-shot localization methods (STEAM and PRESS) for in vivo proton NMR spectroscopy. NMR Biomed 2: 201–208. Moonen CTW, van Zijl, PCM. 1990. Highly efficient water suppression for in vivo proton NMR spectroscopy. J Magn Reson 88: 28–41. Moonen CTW, Sobering G, van Zijl PCM, Gillen J, von Kienlin M, Bizzi A. 1992. Proton spectroscopic imaging of human brain. J Magn Reson 98: 556–575. Nelson SJ, Vigneron DB, Dillon WP. 1999. Serial evaluation of patients with brain tumors using volume MRI and 3D 1H MRSI. NMR Biomed 12: 123–138.

Ogg RJ, Kingsley PB, Taylor JS. 1994. WET, a T1 and B1-insensitive water suppression method for in vivo localized 1H NMR spectroscopy. J Magn Reson B104: 1–10. Pan JW, Telang FW, Lee JH, de Graaf RA, Rothman DL, Stein DT, Hetherington HP. 2001. Measurement of beta-hydroxybutyrate in acute hyperketonemia in human brain. J Neurochem 79: 539–544. Petroff OAC, Graham GD, Blamire AM, al Rayess M, Rothman DL, Fayad PB, Brass LM, Shulman RG, Prichard JW. 1992. Spectroscopic imaging of stroke in humans: histopathology correlates of spectral changes. Neurology 42: 1349–1354. Posse S, Tedeschi G, Risinger R, Ogg R, Le Bihan D. 1995. High speed 1H spectroscopic imaging in human brain by echo planar spatial-spectral encoding. Magn Reson Med 33: 34–40. Pouwels PJW, Kruse B, Hanefeld F, Frahm J. 1997. In ISMRM 5th Scientific Meeting and Exhibition, Vancouver, BC, pp. 482. Prichard J, Rothman D, Novotny E, Petroff O, Kuwabara T, Avison M, Howseman A, Hanstock C, Shulman R. 1991. Lactate rise detected by 1H NMR in human visual cortex during physiologic stimulation. Proc Natl Acad Sci USA 88: 5829–5831. Proctor WG, Yu FC. 1950. The dependence of a nuclear magnetic resonance frequency upon chemical compound. Phys Rev 77: 717. Provencher SW. 1993. Estimation of metabolite concentrations from localized in vivo proton NMR spectra. Magn Reson Med 30: 672–679. Purcell EU, Torrey HC, Pound RV. 1946. Resonance absorption by nuclear magnetic moments in a solid. Phys Rev 69: 37–38. Radda GK. 1986. The use of NMR spectroscopy for the understanding of disease. Science 233: 640–655. Raphael C. 1991. In vivo NMR spectral parameter estimation: a comparison between time and frequency domain methods. Magn Reson Med 18: 358–370. Ross BD, Michaelis T. 1994a. Clinical applications of magnetic resonance spectroscopy. Magn Reson Q 10: 191–247. Ross BD, Jacobson S, Villamil F, Korula J, Kreis R, Ernst T, Shonk T, Moats RA. 1994b. Subclinical hepatic encephalopathy: proton MR spectroscopic abnormalities. Radiology 193: 457–463. Rothman DL, Petroff OA, Behar KL, Mattson RH. 1993. Localized 1H NMR measurements of gamma-aminobutyric acid in human brain in vivo. Proc Natl Acad Sci USA 90: 5662–5666. Seymour KJ, Bluml S, Sutherling J, Sutherling W, Ross BD. 1999. Identification of cerebral acetone by 1H-MRS in patients with epilepsy controlled by ketogenic diet. Magma 8: 33–42.

25

26

Peter B. Barker

Shonk TK, Moats RA, Gifford P, Michaelis T, Mandigo JC, Izumi J, Ross BD. 1995. Probable Alzheimer disease: diagnosis with proton MR spectroscopy [see comments]. Radiology 195: 65–75. Simmons ML, Frondoza CG, Coyle JT. 1991. Immunocytochemical localization of N-acetyl-aspartate with monoclonal antibodies. Neuroscience 45: 37–45. Soher BJ, van Zijl PCM, Duyn JH, Barker PB. 1996. Quantitative proton spectroscopic imaging of the human brain. Magn Reson Med 35: 356–363. Spielman DM, Pauly JM, Macovski A, Glover GH, Enzmann DR. 1992. Lipid-suppressed single- and multisection proton spectroscopic imaging of the human brain. J Magn Reson Imaging 2: 253–262. Stockler S, Holzbach U, Hanefeld F, Marquardt I, Helms G, Requart M, Hanicke W, Frahm J. 1994. Creatine deficiency in the brain: a new, treatable inborn error of metabolism. Pediatr Res 36: 409–13. Stoll AL, Renshaw PF, De Micheli E, Wurtman R, Pillay SS, Cohen BM. 1995. Choline ingestion increases the resonance of choline-containing compounds in human brain: an in vivo proton magnetic resonance study. Biol Psychiatry 37: 170–174. Strange K, Emma F, Paredes A, Morrison R. 1994. Osmoregulatory changes in myo-inositol content and Na+/Myo-inositol cotransport in rat cortical astrocytes. Glia 12(1): 35–43. Terpstra M, Henry PG, Gruetter R. 2003. Measurement of reduced glutathione (GSH) in human brain using LC model

analysis of difference-edited spectra. Magn Reson Med 50: 19–23. Tkac I, Andersen P, Adriany G, Merkle H, Ugurbil K, Gruetter R. 2001. In vivo 1H NMR spectroscopy of the human brain at 7T. Magn Reson Med 46: 451–456. Urenjak J, Williams SR, Gadian DG, Noble M. 1992. Specific expression of N-acetylaspartate in neurons, oligodendrocyte- type-2 astrocyte progenitors, and immature oligodendrocytes in vitro. J Neurochem 59: 55–61. Urenjak J, Williams SR, Gadian DG, Noble M. 1993. Proton nuclear magnetic resonance spectroscopy unambiguously identifies different neural cell types. J Neurosc 13: 981. van der Knaap MS, van der Grond J, van Rijen PC, Faber JAJ, Valk J, Willemse K. 1990. Age-dependent changes in localized proton and phosphorus MR spectrscopy of the brain. Radiology 176: 509–515. van der Knaap MS, Wevers RA, Struys EA, Verhoeven NM, Pouwels PJ, Engelke UF, Feikema W, Valk J, Jakobs C. 1999. Leukoencephalopathy associated with a disturbance in the metabolism of polyols. Ann Neurol 46: 925–928. van Zijl PCM, Barker PB. 1997. In Imaging Brain Structure and Function, Vol. 820, New York, NY, pp. 75–96. Vermathen P, Laxer KD, El Din M, Matson GB, Weiner MW. 1997. In ISMRM 5th Scientific Meeting and Symposium, Vancouver, BC, pp. 36.

2

Quantification and analysis in MR spectroscopy Thomas Ernst Department of Medicine, University of Hawaii, John A. Burns School of Medicine, Honolulu, Hawaii

Key points • Spectral quantification allows detection of metabolite abnormalities that are not appreciated from visual inspection alone. • Metabolite ratio determination is robust and reproducible in a clinical environment, but prone to changes in the denominator metabolite concentration (commonly creatine). • Spectra may be fitted in the time or frequency domain. • Absolute metabolite quantification requires internal or external reference standards, and correction for tissue volumes (e.g. cerebrospinal fluid) within the voxel. Water is a commonly used internal reference signal. • Chemical shift imaging allows calculation of metabolite levels within different tissues, e.g. gray and white matter.

spectroscopic peaks reflect the concentrations of metabolites in the tissue; however, it is impossible to determine these concentrations visually. These points are illustrated in Figure 2.1, which shows proton spectra from a lymphoma lesion and a contralateral voxel in a patient with acquired immuno deficiency syndrome (AIDS). Since the spectra can be plotted with arbitrary vertical scaling, it is unclear if a given metabolite peak, and its associated concentration, in the lesion is higher or lower compared to the healthy brain tissue. It is even difficult to estimate the relative heights of the metabolite peaks within each voxel. Therefore, the ultimate goal of spectral analysis is to determine accurate estimates of metabolite peak areas that reflect metabolite concentrations.

Spectral analysis Overview

Introduction Why quantification and not visual interpretation? The quantification of spectral peaks plays an important role in MR spectroscopy (MRS), and pure visual readings of spectra are less common compared to MR imaging (MRI). The reason for this difference is that MRI relies on the detection of spatial or signal abnormalities as a result of disease conditions, whereas MRS interpretation commonly relies on the interpretation of differences in relative proportions of metabolite peaks at a given location. Furthermore,

The first major step in determining metabolite concentrations is to obtain the signal strength Sm of each metabolite in a given spectrum. Typically, sophisticated computer algorithms are used for this purpose. We will describe the major techniques and discuss major advantages and problems. However, the exact details of the analysis often will be completely hidden from the user, especially with some of the more recent automated programs, and most likely will have only minor influence on the quality of the analyses. Spectral analysis can be performed in the “time domain”, using the so-called “free-induction decay” 27

28

Thomas Ernst

Contralateral

Frontal lesion

NAA  8.52 mM Cr  7.4 mM Cho  1.97 mM mI  8.5 mM

NAA  6.92 mM ( 19%) Cr  3.53 mM ( 52%) Cho  3.06 mM (56%) mI  5.43 mM ( 36%)

Post-Gd T1MRI

Lac/lipids

NAA

Cho

Cr Cho

Glx

mI

4

NAA Glx

mI

3

2 ppm

1

0

4

3

2 ppm

1

0

Fig. 2.1 Localized 1H MR spectra from a lymphoma in the left frontal lobe and from a contralateral control region in normalappearing white matter (WM). While visual inspection shows clear differences between the two spectra, it does not allow an accurate, quantitative assessment of the metabolite abnormalities. Spectral quantification, using the water signal as a reference, makes it possible to calculate millimolar (mM) metabolite concentrations, and demonstrates that the lymphoma lesion has reduced concentrations of N-acetyl aspartate (NAA) compounds, total creatine (Cr), and myo-inositol (mI), whereas the concentration of Choline (Cho) compounds is increased in the lesion. Glx, glutamate (Glu)  glutamine (Gln).

(or FID), or the “frequency domain”, using “spectra” after Fourier transformation of the time-domain data. Of these two, analysis in the frequency domain (Mierisova and Ala-Korpela, 2001), i.e. the use of spectra, is more intuitive and will be discussed first. The steps involved in frequency-domain spectral analysis are exemplified in Figure 2.2 and outlined below. Historically, the steps were performed sequentially and manually by a spectroscopist; however, more recent spectral analysis programs are completely automatic.

potentially very large residual water signal (Coron et al., 2001). The resulting data are multiplied with a decaying function, such as a decaying exponential, to artenuate signals on the right side of the FID. This step is called “low-pass filtering” or “apodization,” and reduces noise in the spectrum (cf. Figure 2.2), but at the expense of increasing spectral linewidths. Finally, the apodized are padded with zeroes on the right side (cf. Figure 2.2); for instance, the total number of data points may be increased from 1024 to 2048 or 4096. This step is called “zero-filling” and improves the digital resolution of spectra.

Time domain pre-processing Spectral analysis typically involves several preprocessing steps in the time domain that are summarized in Figure 2.2 (top row). First, the digital time domain spectral data (FIDs) are corrected to remove phase variations due to residual gradient-induced eddy currents (Klose, 1990; Lin et al., 1994). Next, a digital filter is commonly applied that removes the

Fourier transformation The zero-filled, low-pass filtered, and eddy-current corrected time-domain data are then Fourier transformed, which yields frequency-domain data (spectra). The remaining processing steps are applied in the frequency domain (cf. Figure 2.2, bottom row).

Quantification and analysis in MR spectroscopy

Apodization

Zero-filling

Time domain

Fourier transformation

Phase correction

Baseline correction

Frequency domain

4 3.5 3 2.5 2 1.5 1 0.5 ppm

4 3.5 3 2.5 2 1.5 1 0.5 ppm

4 3.5 3 2.5 2 1.5 1 0.5 ppm

Fig. 2.2 Overview of the major processing steps for spectral analysis. The graphs show the typical appearance of MR signals (top row) and spectra (bottom row) after each step. Several pre-processing steps are performed in the time domain. Correction for gradient-induced eddy currents and removal of residual water signal yields the signal shown in the top left graph. This signal is multiplied with a decaying function (“apodization” or “low-pass filtering”) and padded with zeroes on the right side of the graph (“zero-filling”; result shown in top right box). For frequency-domain processing, the pre-processed signal is Fourier transformed. The resulting spectrum typically has distorted line shapes (bottom left), which can be adjusted with a phase-correction algorithm (bottom center). Next, baseline correction (manual or automatic) yields a spectrum with a well-defined, horizontal baseline (bottom right). This pre-processed, phase- and baseline-corrected spectrum is then used to estimate metabolite peak areas with integration or iterative peak fitting algorithms.

Phase correction The phase of the raw spectrum after Fourier transformation is usually incorrect, i.e. metabolite peaks may be inverted or have distorted line shapes (Figure 2.2, bottom left), and requires manual or automatic adjustment.

Baseline correction After phase correction, the baseline of the resulting spectrum is typically distorted or slanted, and has to be corrected. For manual baseline correction, the user defines several spectral points, typically between the major metabolite peaks, as “baseline”.

The computer then fits and subtracts a smooth curve through these points. The result is a spectrum with a flat baseline, which is better suited for determination of metabolite peak areas (cf. next step). Determination of metabolite peak areas The final step in spectral processing is the determination of the metabolite peak areas. The signal strength Sm of each metabolite relates to the size of the metabolite signal in the time domain (e.g. in millivolts). In the frequency domain, the corresponding measure is the area of the metabolite peak. Of note, the peak height alone does not represent the metabolite concentrations.

29

30

Thomas Ernst

Manual or automatic integration One of the most intuitive and earliest methods to determine peak areas is by means of numerical integration, either manually or automatically. The user or a computer program selects two frequency points, one to the left and one to the right of a metabolite peak. The amplitude at these two points is assumed to define the spectral “baseline” and set to zero. The computer then proceeds to integrate the area under the peak, using a numerical integration algorithm. While peak integration is very intuitive, it has substantial limitations, especially when applied to in vivo spectra. First, because in vivo spectra are commonly crowded and contain many overlapping metabolite peaks, it may be difficult to define the exact boundary between adjacent peaks. For instance, the choline (Cho) and creatine (Cr) peaks in proton spectra are separated by only 0.2 ppm, and there is often no clear boundary between the two. Of even greater concern is the overlap between some of the singlet resonances with broader multiplets or macromolecule resonances, such as that of the N-acetyl aspartate (NAA) peak at 2 ppm with the broad multiplet from glutamate (Glu) and glutamine (Gln) (between 2.0 and 2.4 ppm; visible at short echo times (TE)). In this situation, it is virtually impossible to separate the contributions from the different metabolites with simple peak integration algorithms. Peak fitting To resolve the problems of numerical peak integration, sophisticated computer algorithms have been developed that rely on the iterative fitting of ideal or experimental model spectra to in vivo spectra. Early algorithms commonly modeled in vivo spectra as a superposition of multiple individual peaks of a certain “ideal” line shape. For instance, one may assume that a given in vivo spectrum comprises multiple resonances of a Lorentzian line shape, and model it accordingly. Other common line shapes include Gaussian or variable mixtures of Lorentzian with Gaussian lines, which resemble Voigt functions. Incorporation of prior knowledge in fitting algorithms An important feature of the fitting algorithms is the inclusion of prior knowledge. The simplest

algorithms fit each spectral peak separately. Such an approach may be sufficient for analyzing spectra that contain only a few well-separated resonances, such as long TE proton spectra of the brain. However, it is generally advantageous to incorporate “prior knowledge” into fitting algorithms. “Prior knowledge” characterizes known information about spectral characteristics that are not variable among subjects, such as relative peak positions, relative intensity or phase of peaks for multiplets, etc. The use of prior knowledge reduces the number of free parameters that need to be determined by the fitting algorithm, and generally improves the quality of the fit and reduces fitting errors. Time-domain fitting While less intuitive than spectral fitting in the frequency domain, the actual fitting procedure may also be performed in the time domain. In fact, mathematically there is no substantial difference between frequency- and time-domain fitting. However, in the time domain, it is essentially impossible to perform manual phase or baseline correction. Fitting of model spectra A more recent program named linear combination model (“LC model”) (Provencher, 1993) fits in vivo spectra as a linear superposition of high-resolution “basis” spectra that are acquired from model solutions of metabolites that are present in the organ of interest. For instance, to model brain 1H MR spectra, an LC model basis set may include high-resolution spectra of the major metabolites NAA, Cr and phosphocreatine, Cho, myo-inositol (mI), and Glu, as well as those of minor metabolites, such as Gln, gamma-amino-butyric acid (GABA), glucose, NAA– Glu (NAAG), etc. The fitting program determines the contribution of each basis spectrum to a given in vivo spectrum, and thus determines the relative concentration of the various metabolites in the basis set. Advantages of LC model are that all pre-processing steps, automatic phase correction as well as modeling of a smooth baseline are included. The initial acquisition of the basis spectra requires substantial expertise and effort; however, standardized basis sets are available for the most common clinical MR machines (both 1.5 and 3 T).

Quantification and analysis in MR spectroscopy

Common problems Most difficulties with spectral analysis are related to the fact that in vivo spectra contain multiple overlapping peaks including those from macromolecules, have relatively low signal-to-noise ratio (SNR) and ill-defined or slanted spectral baselines. While the resonances of metabolites at high concentration, such as Cr, are generally sufficiently well defined to allow accurate peak area determination, it may be difficult or impossible to obtain reliable peak areas for minor resonances. For instance, it is essentially impossible to obtain a reliable estimate of the amount of GABA from a regular in vivo 1 H brain spectrum, since the major GABA signals co-resonate with the NAA and Cr resonances, which have 5–10 times higher concentration than GABA. In fact, some analysis programs determine a fitting error for each metabolite. An error of 20% or greater generally indicates that the peak area determination is unreliable; errors of 50% or greater imply that the area measure is entirely meaningless.

Quantification Theoretical considerations After metabolite peak areas have been determined with one of the methods described above, the second major step is to convert the peak areas, which are in arbitrary units, into metabolite concentrations. This quantification step relies on the fact that the strength of the MR signal Sm for a given metabolite “m” (or water) is proportional to the number of observed spins in the volume V of interest (VOI or voxel), which in turn is proportional to the concentration cm of the metabolite and the number ns of spins contributing to the resonance (e.g. 2 for water, which has 2 hydrogen nuclei). Formally, we can write for the signal of subject i at time t: Sm(i ,t )  (i ,t ) cm V ns Fm ( T1m , T2m , J m , TR, TE, B1(t ))

(2.1)

where (i,t) is a scaling factor and Fm is a “modulation” factor; both of these are discussed in detail

below. Eq. (2.1) yields the following solution for the metabolite concentration cm: cm  Sm(i ,t )/[ (i ,t ) V ns Fm ( T1m , T2m , J m , TR, TE, B1(t ))]

(2.2)

This equation shows that the metabolite concentration is proportional to the MR signal strength Sm and inverse to the volume, the number of spins, and the scaling and modulation factors. The modulation function Fm describes how the signal is modulated by the TE, recovery time (TR), the longitudinal and transverse relaxation times of the metabolite (T1m and T2m), the radio-frequency (RF) field strength B1 (i.e. flip angles), as well as the coupling constant Jm of the metabolite (if present). Some metabolites have simple, singlet spectra which do not contain couplings, whereas other have complicated spectral patterns with multiple coupling constants among the different protons in the molecule. The detailed dependence of Fm on these parameters may be very complicated, depends on the pulse sequence being used, and requires quantum mechanical calculations for coupled (i.e. Jm  0) spin systems (Ernst et al., 1990). However, the overall dependence of Fm on flip angles, relaxation times, and sequence timing (TE and TR) resembles that of MR imaging sequences in that shorter TR values or longer TE values generally attenuate metabolite signals, and act similarly to “T1 weighting” or “T2 weighting” in MRI. Likewise, the signal amplitude is dependent on the strength of the RF field (i.e. flip angles); maximum signal is only achieved when the RF field is adjusted correctly. Of note, since the relaxation times and J-coupling constants differ among metabolites, the factor Fm may vary from one metabolite to another. Another important parameter in Eq. (2.1) is the scaling factor (i,t). The scaling factor describes how the amplitude of the observed nuclear MR (NMR) signal relates to tissue-specific “internal” variables, such as metabolite concentrations, volume, etc.  depends on the parameters such as the size of the object to be imaged, the RF coil tuning and matching, and the gain of the RF receiver chain, all of which are difficult to control. Therefore, (i,t) may vary from subject to subject (index i), as well as

31

32

Thomas Ernst

within a subject from one study to another (i.e. over time t). The goal of quantification is to derive the concentration cm of a metabolite of interest from its signal strength Sm for a given subject and study. Therefore, according to Eq. (2.1), it is necessary to determine the scaling factor (i,t). Essentially all techniques rely on measuring the signal of a substance with known concentration (a “reference”) to determine (i,t). Several authors have reviewed quantification techniques (Tofts and Wray, 1988; Buchli et al., 1994). The subsequent paragraphs briefly describe the different choices for the “reference”, and how the calibration is performed. For simplicity, we assume that a single-voxel proton MRS experiment is performed; the principles described can, however, be extended to multi-voxel methods, such as spectroscopic imaging (SI), or other nuclei. Techniques for quantifying metabolite concentrations “Metabolite ratios”: use of a metabolite as a reference The most common approach to spectral quantification is to express metabolite levels relative to a “reference metabolite” in the same spectrum. This approach is referred to as the use of “metabolite ratios”. For 1H MRS, the most common reference metabolite is total Cr, and for 31P MRS, phosphocreatine is the most common reference. For example, 1 H MRS studies of the brain frequently report results as ratios between a metabolite peak and Cr, such as the ratio of the NAA resonance to total Cr, or the “NAA/Cr ratio”. Of note, “metabolite ratios” do not reflect true concentrations, unless the concentration of the reference metabolite is known. The use of metabolite ratios has several advantages. First, because the reference signal is acquired simultaneously with the metabolite of interest, many potential sources of systematic errors in Eq. (2.2), such as the scaling factor , the exact volume, partial volume with cerebrospinal fluid (CSF), or the flip angles (B1), are removed. Consequently, metabolite ratios are probably the most robust of all spectral quantitation techniques. For example, the intrasubject variability of metabolite ratios in 1H MR

spectra of the human brain may be below 10% within a single-site, and below 15% across sites (Webb et al., 1994) using similar, automated methodology. An added advantage of “metabolite ratios” is that their measurement does not require modifications to existing MRS sequences or additional series. Therefore, “metabolite ratios” may provide reliable markers of tissue biochemistry and be useful for clinical diagnosis. However, “metabolite ratios” are associated with a significant shortcoming: it is impossible to determine whether an abnormality in a ratio is due to a change in the numerator metabolite (e.g. NAA) or in the denominator metabolite (e.g. Cr), or both. Therefore, metabolite ratios are intrinsically ambiguous and prone to misinterpretation. Nonetheless, it is a frequent implicit assumption that the concentration of the reference metabolite is constant over time or across subjects and disease conditions. For example, reduced NAA/Cr ratio in brain tissue is commonly interpreted as decreased NAA concentration due to neuronal loss. However, reduced NAA/Cr may also be a result of increased Cr. In the brain, for instance, increased Cr concentrations (measured with one of the techniques described below) have been observed in multiple sclerosis (MS) (Inglese et al., 2003), HIV dementia (Chang et al., 1999), in myotonic dystrophy where the Cr concentration shows a dramatic and linear increase with the number of CTG repeats (a genetic marker of disease severity) (Chang et al., 1998) and in other brain diseases. Conversely, decreased Cr concentrations are common in conditions that are associated with the destruction of normal brain tissue, such as strokes (Saunders, 2000), abscesses (Chang et al., 1995), or neoplasms (Chang et al., 1995; Negendank et al., 1996; Preul et al., 1996). Furthermore, the cerebral Cr concentration also changes during neurodevelopment (Kreis et al., 1993a; Pouwels et al., 1999; Horska et al., 2002). Finally, Figure 2.3 demonstrates that the Cr concentration in the brain also increases during normal aging, at a rate of approximately 2.5% per decade in the white matter (WM) (Christiansen et al., 1993b; Chang et al., 1996; Pfefferbaum et al., 1999; Suhy et al., 2000). In summary, metabolite ratios provide robust in vivo markers of biochemistry. However, metabolite ratios have to be interpreted with caution since it is

Frontal WM [Cr] (mmole/kg)

Quantification and analysis in MR spectroscopy

6.5

5.5

4.5 R  0.6 P  0.0001

3.5 5

15 25 35 45 55 65 75 Age (years)

85 95

Fig. 2.3 Dependence of the total Cr concentration [Cr] in the healthy brain on age. The [Cr] in the WM increases by approximately 2.5% per decade throughout the adult life. Consequently, it is incorrect to assume that [Cr] is constant when interpreting metabolite ratios. For instance, the NAA/Cr ratio in a 65-year-old subject would be 10% lower relative to a 25-year-old subject as a result of the changes in [Cr], even without changes in the NAA concentration.

generally incorrect to assume that the concentration of the reference metabolite is unchanged across subjects and disease conditions. Use of spectrum from control region as a reference One clinically useful method to assess metabolite levels may be to express metabolite levels in a region of interest (ROI) relative to those in another region, for instance, a contralateral region (cf. Figure 2.1). This may be particularly useful for studies of focal abnormalities, and is commonly employed with chemical shift imaging (CSI). However, this approach provides little value in the evaluation of diseases that have a diffuse or global spatial distribution. Use of water as a reference signal To resolve the ambiguities associated with the use of metabolite ratios, the water signal from brain parenchyma is commonly used as a reference to determine the scaling factor  (Barker et al., 1993; Christiansen et al., 1993a; Ernst et al., 1993). Since the water content in a unit volume of brain tissue is almost a constant, the water signal is a good internal reference for measuring metabolite concentrations. Since the

water concentration in tissues are known accurately, the signal strength of the water signal can then be used to determine the scaling factor  for each subject and study, according to Eq. (2.1). The concentration of pure water is approximately 55 M, and since there are 2 protons per water molecule, the proton concentration is 110 M. In brain, the water content varies from 70% to 80% for WM and gray matter (GM), respectively, and therefore the proton concentration in brain is typically in the range of 77–88 M. Furthermore, because the water and metabolite signals are acquired from an identical VOI, and with the same pulse sequence and flip angles, many potential error sources are eliminated and the metabolite concentration measurement becomes relatively robust. An added benefit is that the time to acquire the water signal is negligible, and that it requires no substantial modifications to the MRS sequence. One of the potential drawbacks of this approach is that the water signal is invariably acquired at an TE greater than zero, typically 20 ms. As a result, the water signal always has some degree of “T2 weighting”, and changes in the transverse relaxation time of tissue water (T2) may lead to erroneous changes in the apparent water signal amplitude, and thus the scaling factor . Despite this drawback, the robustness of using a single unsuppressed water FID as a reference signal has been demonstrated in a multisite study that involved identical MR machines; typical variations in metabolite concentrations were approximately 15% (Soher et al., 1996). However, the use of the water signal as a concentration reference is more complex as it may appear. Relatively large size of MRS voxels (typically cm3) makes it likely that each voxel contains a mixture of several compartments. For instance, a typical MRS voxel in the human brain may contain GM, WM, as well as CSF; cf. basal-ganglia voxel in Figure 2.4. Each of these macroscopic compartments may contain a different concentration of each metabolite. This effect is particularly pronounced for CSF, which has markedly lower concentrations of the major brain metabolites (NAA, Cr, Cho, and mI) than brain tissue. As a result, significant amounts of CSF in a given MRS voxel may lead to an apparent reduction in metabolite concentrations, even if the

33

Brain water

Thomas Ernst

Y  0.05  42.9 exp( TE/85 ms)  7.3 exp(TE/1012 ms)

CSF

34

0

250

500

750 1000 TE (ms)

1250

1500

Fig. 2.4 The graph shows the dependence of the total water signal (vertical axis) from a typical MRS voxel (box on structural MR image) on the TE. Since the relatively large MRS voxel contains a mixture of brain parenchyma and CSF (cf. structural MRI), the T2 decay of the water signal is double exponential. The water signal amplitude in the graph was measured at 10 different TE (ranging from 30 to 1500 ms), and fitted with an iterative computer algorithm to extract the two T2 components. The slow component (T2 typically 2 s) is due to CSF, and the fast component (T2  100 ms) is due to water in the brain parenchyma. The amplitude of the fast component, interpolated back to TE  0 (“Brain water” signal), provides an excellent reference signal for calculating metabolite concentrations.

concentrations in the brain parenchyma are normal. Moreover, significant differences in the major brain metabolites also exist between GM and WM (Pouwels and Frahm, 1998; Schuff et al., 2001). Several techniques have been developed to resolve these problems. One of the earliest techniques aimed at eliminating the volume dilution effect due to CSF (Ernst et al., 1993). The technique is based on the measurement of the T2-decay curve of the water signal from within the VOI, using multiple TE ranging from the shortest value possible into the second range. This “T2 measurement” adds a few minutes of time to each voxel. As the T2 value of water in brain parenchyma (100 ms) is markedly shorter than that of CSF (2000–3000 ms), it is relatively easy to separate the fast-decaying brain water signal from the slowly-decaying CSF water signal, as shown in Figure 2.4. As an added benefit, it is possible to accurately extrapolate the brain water signal back to

TE  0. The resulting reference signal at TE  0 reflects the “true” amount of brain water within the VOI, without “T2 weighting” and without any mixture of CSF. This technique is robust enough to be used in multi-center drug trials (Table 2.1). It is rather instructive to study the T2-decay curve in Figure 2.4, since it identifies potential errors associated with some of the other quantification approaches. For instance, the simple technique described above measures the water signal at a single TE only. As a result, the measured water signal contains contributions from both brain parenchyma and CSF, and does not correct for the invariable T2 decay of the signals (which additionally is different between CSF and brain parenchyma). Other methods to correct for the partial volume of CSF are based on MRI. For example, T2-weighted images may be used to segment CSF from brain tissue, and determine the percentage of CSF (Suhy et al., 2000). This allows quantification of the amount of brain parenchyma in a given MRS voxel, and corrects the metabolite concentrations accordingly. Since T2-weighted images are commonly acquired during MRS acquisitions, this approach may require no additional time. However, since the MR images have to be segmented, the post-processing may be extensive and not applicable for routine use. This approach may also suffer from an inability to correct for imperfections in the adjustment of RF power, and to determine the “true” water signal at TE  0. Of note, while metabolite concentrations measured on a single MR machine with a single technique may yield highly reproducible results, comparisons of metabolite concentrations across different vendors and technical approaches are commonly inconsistent, as a result of systematic differences. However, the inter-subject variability of metabolite concentrations in carefully designed multi-center trials is similar to that of single-site studies (Lee et al., 2003). Other techniques A variety of other techniques for measuring metabolite concentrations have been devised; however, many of these are mostly used in academic research. For instance, an “external reference” can be used to determine the scaling factor  (Ernst et al., 1993; Kreis et al., 1993b; Buchli et al., 1994). The external

Quantification and analysis in MR spectroscopy

Table 2.1. Advantages and problems of techniques to quantify in vivo spectra Reference signal

Advantages

Drawbacks

Metabolite (e.g. Cr) “metabolite ratios”

No additional series Reduces systematic errors High reproducibility Robustness

Do not reflect metabolite concentrations Reference metabolite may not be stable Interpretation

Water (at same TE as spectrum)

Short imaging time for reference signal (single water FID) Reduces systematic errors Robustness

No intrinsic atrophy (CSF) correction Water reference signal is “T2 weighted”

Brain water (TE  0)

Reduces systematic errors Atrophy (CSF) correction Independent of water T2 Robustness

Additional imaging time for reference signal (multiple water FIDs)

MRI signals

No additional MRS series May correct for atrophy (CSF)

Image processing (segmentation, etc) Water reference signal is “T2 weighted” Susceptible to systematic errors’

“External reference”

Concentration of reference signal exactly known

Complicated handling Susceptible to B1 variations

Spectrum from control region

Subject provides his or her own normative value

Not applicable to diseases with diffuse spatial distribution No atrophy (CSF) correction Requires acquisition of second spectrum

reference contains a chemical of known concentration, and is placed into the MR machine simultaneously with human subjects. The MR signal from the external reference is acquired while the subject is still in the RF coil, to ensure identical coil loading and RF calibration. This approach measures metabolite concentrations as well as the water content of the human brain with great precision. External reference techniques may also be necessary for measuring metabolite concentrations in cases where only a single metabolite signal is present, such as lithium spectroscopy (Gonzalez et al., 1993). However, the technique requires additional time and is susceptible to variations in the RF field (Table 2.1). The presence of the external standard may also disrupt the magnet field homogeneity within the brain because of magnetic susceptibility effects. A variation of the external standard is the “phantom replacement” method, which requires the collection of a spectrum (under the same experimental conditions) from a phantom of known concentration,

either before or after the patient has been imaged. Various correction factors have to be applied, in particular corrections due to differences in RF coil loading that occur between the patient and the phantom. One way of determining the loading factor is from the amount of RF power needed for a reference 90° pulse (Michaelis et al., 1993) or water suppression pulse (Danielsen and Henriksen, 1994). Quantification of MR spectroscopic imaging data sets For MR spectroscopic imaging (MRSI), many of the same approaches that were developed for single voxel spectroscopy can also be applied; however, some approaches may be more practical than others. For instance, use of the water signal (or measurement of its multi-exponential T2) as an internal intensity reference is often not feasible, since the recording of multiple MRSI studies (with and without water suppression) may be too time consuming. Therefore,

35

36

Thomas Ernst

more often, MRSI data are quantified using either external standards or adaptations of the phantom replacement method. Since MRSI yields images that reflect the distribution of metabolites, or “metabolite maps”, quantitation approaches tend to be image based (such as the use of MR images to segment and correct for CSF). Also, the use of metabolite levels from a control region (such as a contralateral region) is ideally suited for use with MRSI, since the spectrum in the control region can be acquired simultaneously with that in the ROI (Table 2.1). Determination of “pure” gray or white matter concentrations with MRSI MRSI makes it possible to infer metabolite concentrations in pure GM or WM (Hetherington et al., 2001; Schuff et al., 2001). For each MRSI voxel, the metabolite concentrations are calculated as well as the percentage of GM and WM (using image segmentation). Next, the metabolite concentrations for all voxels are plotted vs. the gray (or white) matter percentage. Linear regression analysis can then determine the “endpoints” of the curve, which correspond to the theoretical metabolite concentrations for 100% GM or WM.

Summary The interpretation of in vivo MR spectra based on quantitative spectral analysis is preferable to visual inspection. Spectral analysis involves several processing steps in the time domain and/or frequency domain; modern analysis packages perform these steps in an automated fashion and hidden from the user. Spectra can be quantified in terms of metabolite ratios or metabolite concentrations. Metabolite ratios are calculated by dividing the area of each metabolite peak by the area of a reference peak (commonly total Cr) from the same spectrum. While the resulting metabolite ratios provide a robust quantitative biochemical value, metabolite ratios are difficult to interpret since abnormal ratios may be the result of changes in the numerator or denominator metabolite, or both. More than one decade of research demonstrates that none of the metabolites,

including total Cr, can be assumed to be constant across diseases or even during normal brain development or aging. Therefore, only metabolite concentrations allow unambiguous interpretation of in vivo spectra. The water signal from each MRS voxel can be used as a concentration reference to calculate metabolite concentrations; however, the various techniques differ in their ability to correct for partial volumes of macroscopic compartments, such as GM, WM, and CSF in the brain. Since some of these techniques are robust and relatively fast, clinical MRS reports should be based on the interpretation of metabolite concentrations. Future developments will likely provide standardized MRS protocols with libraries of normative metabolite concentrations that are readily available for comparison with in vivo spectra in patients.

REFERENCES Barker P, Soher B, et al. 1993. Quantitation of proton NMR spectra of the human brain using tissue water as an internal concentration reference. NMR Biomed 6(1): 89–94. Buchli R, Martin E, et al. 1994. Comparison of calibration strategies for the in vivo determination of absolute metabolite concentrations in the human brain by 31P MRS. NMR Biomed 7(5): 225–230. Chang L, Ernst T, et al. 1996. In vivo proton magnetic resonance spectroscopy of the normal human aging brain. Life Sci 58(22): 2049–2056. Chang L, Ernst T, et al. 1998. Proton Spectroscopy in myotonic dystrophy: correlation with CTG Repeats. Arch of Neurol 55(3): 305–311. Chang L, Ernst T, et al. 1999. Cerebral metabolite abnormalities correlate with clinical severity of HIV-cognitive motor complex. Neurology 52(1): 100–108. Chang L, Miller BL, et al. 1995. Brain lesions in patients with AIDS: H-1 MR spectroscopy. Radiology 197: 525–531. Christiansen P, Henriksen O, et al. 1993a. In vivo quantification of brain metabolites by 1H-MRS using water as an internal standard. Magn Reson Imaging 11: 107–118. Christiansen P, Toft P, et al. 1993b. The concentration of N-acetyl aspartate, creatine, phosphocreatine, and choline in different parts of the brain in adulthood and senium. Magn Reson Imaging 11: 799–807. Coron A, Vanhamme L, et al. 2001. The filtering approach to solvent peak suppression in MRS: a critical review. J Magn Reson 152(1): 26–40.

Quantification and analysis in MR spectroscopy

Danielsen E, Henriksen O. 1994. Absolute quantitative proton NMR spectroscopy based on the amplitude of the local water suppression pulse. Quantification of brain water and metabolites. NMR Biomed 7(7): 311–318. Ernst R, Bodenhausen G, et al. 1990. Principles of Nuclear Magnetic Resonance in One and Two Dimensions. Oxford University Press, Oxford. Ernst T, Kreis R, et al. 1993. Absolute quantitation of water and metabolites in the human brain. I: compartments and water. J Magn Reson B102: 1–8. Gonzalez R, Guimaraes A, et al. 1993. Measurement of human brain lithium in vivo by MR spectroscopy. Am J Neuroradiol 14(5): 1027–1037. Hetherington H, Spencer D, et al. 2001. Quantitative (31)P spectroscopic imaging of human brain at 4 Tesla: assessment of gray and white matter differences of phosphocreatine and ATP. Magn Reson Med 45(1): 46–52. Horska A, Kaufmann W, et al. 2002. In vivo quantitative proton MRSI study of brain development from childhood to adolescence. J Magn Reson Imaging 15(2): 137–143. Inglese M, Li B, et al. 2003. Diffusely elevated cerebral choline and creatine in relapsing–remitting multiple sclerosis. Magn Reson Med 50(1): 190–195. Klose U. 1990. In vivo proton spectroscopy in presence of eddy currents. Magn Reson Med 14(1): 26–30. Kreis R, Ernst T, et al. 1993a. Development of the human brain: in vivo quantification of metabolite and water content with proton magnetic resonance spectroscopy. Magn Reson Med 30: 424–437. Kreis R, Ernst T, et al. 1993b. Absolute quantitation of water and metabolites in the human brain. II: metabolite concentrations. J Magn Reson B102: 9–19. Lee P, Yiannoutsos LCT, et al. 2003. A multi-center 1H MRS study of the AIDS dementia complex: validation and preliminary analysis. J Magn Reson Imaging 17(6): 625–633. Lin C, Wendt R, et al. 1994. Eddy current correction in volume-localized MR spectroscopy. J Magn Reson Imaging 4(6): 823–827. Michaelis T, Merboldt K, et al. 1993. Absolute concentrations of metabolites in the adult human brain in vivo: quantification of localized proton MR spectra. Radiology 187(1): 219–227.

Mierisova S, Ala-Korpela M. 2001. MR spectroscopy quantitation: a review of frequency domain methods. NMR Biomed 14(4): 247–259. Negendank W, Sauter R, et al. 1996. Proton magnetic resonance spectroscopy in patients with glial tumors: a multicenter study. J Neurosurg 84(3): 449–458. Pfefferbaum A, Adalsteinsson E, et al. 1999. In vivo spectroscopic quantification of the N-acetyl moiety, creatine, and choline from large volumes of brain gray and white matter: effects of normal aging. Magn Reson Med 41(2): 276–284. Pouwels PJ, Brockmann K, et al. 1999. Regional age dependence of human brain metabolites from infancy to adulthood as detected by quantitative localized proton MRS. Pediatr Res 46(4): 474–485. Pouwels P, Frahm J. 1998. Regional metabolite concentrations in human brain as determined by quantitative localized proton MRS. Magn Reson Med 39(1): 53–60. Preul MC, Caramanos Z, et al. 1996. Accurate, noninvasive diagnosis of human brain tumors by using proton magnetic resonance spectroscopy. Nature Med 2(3): 323–325. Provencher S. 1993. Estimation of metabolite concentrations from localized in vivo proton NMR spectra. Magn Reson Med 30: 672. Saunders D. 2000. MR spectroscopy in stroke. Br Med Bull 56(2): 334–345. Schuff N, Ezekiel F, et al. 2001. Region and tissue differences of metabolites in normally aged brain using multislice 1H magnetic resonance spectroscopic imaging. Magn Reson Med 45(5): 899–907. Soher B, Hurd R, et al. 1996. Quantitation of automated singlevoxel proton MRS using cerebral water as an internal reference. Magn Reson Med 36(3): 335–339. Suhy J, Rooney W, et al. 2000. 1H MRSI comparison of white matter and lesions in primary progressive and relapsing– remitting MS. Mult Scler 6(3): 148–155. Tofts P, Wray S. 1988. A critical assessment of methods of measuring metabolite concentrations by NMR spectroscopy. NMR Biomed 1(1): 1–10. Webb P, Sailasuta N, et al. 1994. Automated single-voxel proton MRS: technical development and multi-site verification. Magn Reson Med 31: 365–373.

37

3

Artifacts and pitfalls in MR spectroscopy Ralph E. Hurd GE Medical Systems, Menlo Park, CA, USA

Key points • Ideal spectra require careful pre-scan prescription including shimming, particularly for longitudinal studies. • Optimal spectra production requires good fat and water suppression. • The effects of intravenous contrast medium on spectra are negligible.

proton spectroscopy and spectroscopic imaging (SI) of cerebral metabolites, but the basic principles can be applied to other nuclei, and other parts of the body. With a few notable exceptions, most of the advances in MRS artifact reduction were developed for proton neurological applications at 1.5 T using orthogonalslice localization methods, most notably stimulated echo acquisition mode (STEAM) and double spin echo (SE) point resolved spectroscopy (PRESS).

Pre-scan Introduction In general, proton MR spectroscopy (MRS), as implemented on modern clinical MR imaging (MRI) machines, provides a reliable adjunct to the growing cadre of imaging methods. Pitfalls can be minimized using automation and standard protocols (Webb et al., 1994). Given that the spectral patterns are well known, minor artifacts are relatively easy to identify and read through, at least for the large signals in the spectrum, such as choline (Cho), creatine (Cr) and N-acetyl aspartate (NAA). However, the current trend is toward the incorporation of second tier markers, such as lactate (Lac), glutamate (Glu), glutamine (Gln) and myo-inositol (mI), as well as a demand for longitudinal studies with narrow repeatability requirements. These applications require an understanding of potential artifacts, and the limits of the existing remedies. To achieve repeatability at the limit of biological variation, may, in fact require the development of new artifact reduction algorithms. This chapter details artifacts, remedies and trade-offs that impact the quantitative use of in vivo MRS. The focus is on 38

Pre-scan operations, including prescription, sequence, variable selection, shimming, transmitter gain and water suppression, all impact the limits of detection and repeatability of the examination. Even with automation and reproducible, pre-defined protocols, attention to the details of the pre-scan may improve repeatability. Some applications still benefit from manual pre-scan operations. Routine maintenance and quality control checks using a standard MRS phantom will also reduce the chance for system contributions to variance. Prescription For longitudinal studies, patient position and re-prescription using reliable imaging landmarks is an important contribution to repeatability (Brooks et al., 1999). In addition to re-prescription relative to localizer landmarks, the tilt of the head and reproducible shim are important factors. For prescriptions near subcutaneous lipid or near high susceptibility gradients, out-of-volume suppression is

Artifacts and pitfalls in MR spectroscopy

also recommended. This is especially important for large volume prescriptions as might be used in PRESS-based MR spectroscopic imaging (MRSI), and for studies in which local or array coils are used. It is well established that voxel size is important, and will impact the achievable signal to noise ratio (SNR), homogeneity and sensitivity to partial volume (Ernst et al., 1989; Spielman et al., 1998; Hanson et al., 2000). As a rule, if a focal prescription requires a volume of less than 3 cc, it is usually advantageous to use MRSI. Since longer signal averaging will be required for small volume scans, the ability to collect multiple voxels, and retrospectively shift the volume of interest, will ameliorate some of the partial volume variations of the smaller-voxel size. It is also useful to know the orientation of your systems’ chemical shift offset artifacts, in terms of patient coordinates. This information will help avoid picking up unwanted lipid signal outside of the desired prescription. For example, if the chemical shift artifact shifts lipid signals from right to left then placing a voxel too near the border with subcutaneous lipid on the left, will risk out-ofvolume lipid in the spectrum.

Shimming The use of automated homogeneity adjustment is key to reliable spectroscopic measurements. There are a number of methods, which quickly optimize homogeneity to the limits of the system (Webb and Macovski, 1991; Webb et al., 1994; Gruetter et al., 1998; Kim et al., 2002). Variation in shim will impact repeatability, as will residual inhomogeneity in excess of 0.1 ppm across a voxel for long echo time (TE), and in excess of 0.07 ppm for short TE data. Overlap of resonances and higher levels of broader macromolecule signals, at short TE, places an added demand on homogeneity. Given the limits in shim correction available on whole-body scanners (normally second-order plus Z3, at most), voxel location and size can still significantly impact the final homogeneity, and should remain a consideration during prescription. When inadequate linewidths are attained (e.g. 0.1 ppm equals approximately 6 Hz at 1.5 T), it may be necessary to adjust the location of the spectrum or repeat the shimming procedure in

order to attempt to improve the spectral resolution. As discussed in Chapter 1, because of naturally occurring susceptibility effects, some regions of the brain (e.g. inferior frontal lobe, mesial anterior temporal lobe) are much less amenable to spectroscopy than others.

Sampling time The SNR of a spectrum is directly proportional to the voxel volume, and the square root of averaging time. For any given coil and system, spatial resolution and sampling time can be evaluated with a simple phantom. For second tier metabolites, spatial resolution and sampling time trade-offs can be made based on the NAA or Cr SNR, taking into consideration the expected concentration limits, and coupling pattern of the desired metabolite. For a volume head coil at 1.5 T, expect to average 8–15 min for adequate SNR of the main metabolites at a volumetric resolution of 1 cc. Single-voxel studies with a nominal 8 cc volume of interest are normally acquired in 2–4 min at higher SNR. It should be noted that the detectable limits, as estimated by Cramer-Rao bounds (Cavassila et al., 2000), are not at the SNR limit, and resolution or other spectral simplification may be more valuable than additional sampling time. Sampling time and/or spatial resolution can be improved dramatically using surface or array coils (Nelson et al., 1997; Schaffter et al., 1998; Noworolski et al., 1999).

Transmitter gain setting For certain pulse designs, maximum signal does not mean that transmitter gain has been correctly set (Ryner et al., 1998). For this reason, manual setting should be made on voxel profile. Fast, automated setting of transmitter gain, as optimized for imaging, is often based on a full slice, and as such is not always optimal for a limited volume spectroscopic prescription. These algorithms should be avoided, especially at high field where the dielectric focusing effect can lead to regional variations. Lastly, the ability to collect a water image of the selected voxel is a very useful feature, especially, when searching for potential sources of system-dependent repeatability error.

39

40

Ralph E. Hurd

Water reduction Keeping some water is usually valuable, but how much? In practice 10–100 suppression will leave enough water for reference, without contributing to baseline errors. Thus, design of the water-suppression sequence to leave a consistent amount of residual water is desirable. Methods, which require some optimization during the pre-scan process, are usually not designed for much B1 or T1 insensitivity. Preset water-suppression strategies with some intolerance to B1 and T1 variation are ideal (Ernst and Hennig, 1995).

Data acquisition Fitting and quantification of the overlapping signals that make up an in vivo MR spectrum have been developed to deal with a certain level of baseline distortion and artifacts (cf. Chapter 2), but at a cost in variability, and higher Cramer-Rao bound limits of detection limits. For this reason, the best strategy is to eliminate the burden on the fitting software by improving data acquisition. Pulse design and pulse sequence improvements, as well as improved sampling strategies can all be used to reduce unnecessary artifacts in volume localized spectroscopy.

Pulse design and chemical shift Radio frequency (RF) pulse design is one of the most critical aspects of minimizing potential artifacts in spectroscopy. Pulses are designed to be selective in space and/or chemical shift. They are used to excite, refocus or eliminate spectroscopic coherences. Chemical shift bandwidth is a critical limiting factor in the design of these pulses. As illustrated in Figure 3.1, RF pulses can be described in terms of a stop, transition and a passband. In-band and outof-band ripple can also be defined. Effective bandwidth, the width at half-maximum, needs to be large relative to chemical shift bandwidth to avoid chemical shift registration artifacts. The ideal pulse for slice-selective excitation or refocusing in proton spectroscopy has a minimum transition, and a high effective bandwidth. It is also useful to maintain a

Common pass band Transition band In-band ripple Pass band

Rx bandwidth Out-of-band ripple

v/gG Chemical shift registration error Fig. 3.1 Slice selection can be described in terms of the RF-profiles. The prescribed slice width is adjusted using gradient strength G, based on the effective pulse bandwidth at half-height. The effective bandwidth also impacts the amount of chemical shift registration error. The selectivity of a pulse designed for spectroscopy can be described as the percent of passband common for the targeted chemical shift, relative to the bandwidth containing 95% of the signal. In-band and out-of-band ripple are also RF design specifications.

common slice profile for 90s and 180s in the same sequence. The major design limit for spectroscopic RF pulses is the peak B1 available. When whole-body coil excite systems are used, the RF field intensity is normally limited to peak B1 of 0.15–0.2G. This leads to trade-offs. Consider for example, a Shinnar-LeRoux (SLR) (Pauly et al., 1991) linear phase 180 refocusing pulse, designed to match a good 90 profile. Assume, as designed, this pulse has a time–bandwidth product of 7.2. At a peak B1 limit of 0.15G, the minimum width of this pulse is 6.5 ms, yielding a maximum effective bandwidth of 1100 Hz. This bandwidth can hardly be considered large, relative to a nominal 3 ppm proton range at 3 T (384 Hz). Peak B1 power requirements can be significantly reduced by using reduced flip angle (RFA) and refocusing pulses (Raidy et al., 1995). For example, if 5% loss in SNR is acceptable, the flip angle can be reduced by 14% to 154 degrees, the peak amplitude goes down by 37%

Artifacts and pitfalls in MR spectroscopy

Single-shot volume selection

TE1

TE2

TE1

RF TE1/2

TE1/2

TE2/2 TM

F2

TE1/2

TE1/2

SE-1,2 F1

TE1/2

TE2/2

STE

SEvol

SE-2,3

SE-1,3

F3

Signal

Gipi = 0 detected

2 1 p

0

1

2 b Grad G1

G2

G3

Fig. 3.2 Coherences generated by three orthogonal slice, volume–echo methods. The desired coherence for PRESS is the volume SEvol at TE1  TE2. The desired PRESS coherence order, p, pathway is given by the dotted line. All other coherences must remain dephased. The desired coherence for STEAM is the volume stimulated echo at TE1  TM. The desired coherence order pathway is illustrated with the dashed line. Again all other coherences must be completely dephased to avoid artifacts. The second RF pulse can also create double, p  2 and zero p  0, order coherences for coupled spin systems. Depending on gradient integrals, population of these pathways can lead to signal loss.

and peak power by 60%. Thus for the same limits in pulse time–bandwidth product of 7.2 and peak B1 field maximum of 0.15G, the 154-degree RFA pulse can be played out in 4.8 ms, with an effective bandwidth of 1500 Hz. Spectral–spatial pulses generally have even more favorable effective bandwidths, but are long and are not compatible with short TE acquisitions. However, within the limits of minimum TEs achievable, spectral–spatial pulses can be designed to avoid both water and lipid, or used for spectral editing.

Controlling coherences The general form and coherences from a three-pulse sequence, for example STEAM and PRESS, are illustrated in Figure 3.2. Each slice-selective RF pulse is capable of producing a free induction decay (FID) signal, and each pair of pulses a SE. The three sliceselective pulses produce the desired volume spin or volume-stimulated echo from the region in space where all three slices intersect. If any of the signals other than the desired volume–echo are not

41

42

Ralph E. Hurd

Slice order Axial–coronal–sagittal

Flip angles

Slice order Coronal–sagittal–axial NA ml

90–180–180

Cho Cr

0–180–180

90–0–180

90–180–0

90–0–0

0–180–0

0–0–180 4

3

2 ppm

1

0

4

3

2 ppm

1

0

Fig. 3.3 The source of unspoiled coherence artifacts, and impact of slice order is illustrated in this frontal lobe prescription of short TE PRESS. With a slice order of axial–coronal–sagittal, artifacts from SE-2,3 (0–180–180 trace), SE-1,3 (90–0–180 trace) and F3 (0–0–180 trace) are large. These artifacts are significantly reduced, albeit still observable, in the coronal–sagittal–axial slice order. Adapted from Ernst and Chang (1996) with permission of Lippincott Williams & Wilkins.

eliminated by the design of the pulse sequence, they will appear as artifacts. Therefore, strong-crusher (or “spoiler”) gradients are usually used in the pulse sequences to dephase as much as possible to undesired coherence pathways resulting from less than all three pulses. Scalar, J-coupled spins also form zero and multiple quantum coherences that can impact the final signal level, and represent a potential quantitative artifact. This latter effect is nearly eliminated using a fully asymmetric PRESS sequence, in which the interval between the 90 and first 180 pulse is minimized, thus avoiding any significant generation of zero and multiple quantum coherences. PRESS and STEAM timing can also be optimized to favor one coupled spin over another. Finally, when using a short repetition time (TR), coherences can also be generated from previous repetitions of the pulse sequence; these can generally be eliminated

by using a (variable) crusher gradient during the TR relaxation period. Slice order and spectroscopic imaging acquisition mode Since the three orthogonal pulses in a volumeselection sequence excite regions well beyond the volume of interest, it is possible to have large lipid and water coherences shifted out of the suppression stop band, placing a large burden on the spoiler gradients. This can be difficult if these regions are coincident with the SE formed by the intersection of the two refocusing pulses, SE-2,3, or the FID from the plane excited by the final RF pulse, F3. Thus, knowing the location of these high shift regions, it is possible to reduce the load on the final primer-crusher gradient pair, using re-ordering the slice selection

Artifacts and pitfalls in MR spectroscopy

(a)

(c)

6

(b)

4 Frequency (ppm)

2

0

A120.0

L120.0

R125.0

(d)

4 P120.0

3

2 1 Frequency (ppm)

0

Fig. 3.4 The elimination of out-of-band coherence order artifacts using SI acquisition mode (SIAM) is illustrated. (a) Out-of-band artifacts severely impact repeatability, especially near water and lipid regions in these repeated 3 T CPRESS TE 45 spectra. (b) and (c) The use of coarse 12 12 phase encoding NEX  1, in place of normal averaging, resolves the unspoiled coherences arising for the sinuses in this volunteer study, yielding an artifact free single-voxel spectrum shown in (d). Similar artifacts are also often created by susceptibility shifted water signals in the mouth and throat.

(Ernst and Chang, 1996). Slice re-ordering and the impact of the sub-components of volume selection are illustrated in Figure 3.3. Water, shifted out of the water stop band, and lipids at air–water–lipid interfaces are often the source of these hardto-spoil signals. As can be seen from the 0–180–180, 90–0–180 and 0–0–180 traces, optimal coronal– sagittal–axial slice order does not completely eliminate artifacts from SE-2,3, SE-1,3 and F3 coherences. In fact, these more subtle artifacts can be more difficult to “read” through, and can be confounding to quantification algorithms. Larger crusher gradients are generally not the answer, as that approach can

lead to other artifacts. Slice-order optimization can be augmented by phase cycling or spatial encoding, in addition to the use of outer-volume suppression (OVS – cf. below). Spatial encoded acquisitions MRSI, extending the field of view (FOV) to include the sources of unwanted signal, resolve and thus eliminate these artifacts. This can also be applied to single-voxel methods, using SI acquisition mode (SIAM). In this approach the normal signal average is replaced with phase encoding, at a coarser resolution than the selected volume (Hurd and Sailasuta, 1997). As shown in Figure 3.4(a), repeat studies that contain spatially

43

Ralph E. Hurd

remote artifacts can appear to be real signal, for example interpreted as macromolecular or lipid signals and can add a level of difficulty to reading and quantification. Replacing the normal signal average with a two dimensional (2D) MRSI sampling at a spatial resolution equal to the voxel dimensions (Figure 3.4(b) and (c)) resolves and eliminates artifacts from the sinuses in this 3 T example, yielding a spectrum free of this type of artifact (Figure 3.4(d)).

(a) In vivo representation numerically calculated (TE  30 ms)

Arbitrary units

2.5 2

TM  20 ms TM  34 ms

3

1.5 1 0.5 0 3.4 3.2

3

2.8

2.6 2.4

2.2

2

1.8

Chemical shift (ppm)

Arbitrary units

(b) In vivo spectrum of human brain (TE  30 ms) 2.5 2

TM  20 ms TM  34 ms

3

1.5 1 0.5 0 3.4 3.2

3

2.8

2.6 2.4

2.2

2

1.8

2

1.8

Chemical shift (ppm) (c) Calculated changes in key metabolites (TE  30 ms) 1.5

Arbitrary units

44

1 0.5 0 0.5 0 1 0.5

TM  20 ms TM  34 ms

TM  20 ms TM  34 ms NAA TM  20 ms TM  34 ms

0 3.4 3.2

3

2.8

3

Glu

Gln

3

3

2.6 2.4

2.2

Chemical shift (ppm) Fig. 3.5 Impact of TM on coupled spins observed between 2 and 3 ppm in STEAM spectra. Numerically calculated, and in vivo response of the coupled spins of Glu, Gln and NAA illustrates the dramatic, but predictable changes that TM can make on the measurement of these spins. Conditions can be optimized for one or more species, depending on need. From Thompson and Allen (2001) with permission of Wiley and Sons.

Coupled spins Metabolite signals of interest, which are spin coupled, notably Lac, alanine, Glu, Gln, mI, GABA, etc., are greatly impacted by sequence timing and chemical shift registration error. For example, if all spins in a coupled spin system do not experience the same RF pattern in the PRESS localization sequence, coupled spins in different regions of the volume will modulate differently, and in worst case, cancel. In SI of a PRESS volume, these effects are resolved, but are clearly difficult to interpret given the width of the transition bands and partial volume effects (PVE). Spin coupling behavior can also be used as an advantage, as in spectral editing (Hurd and Freeman, 1991; Allen et al., 1997), or just to make a desired metabolite easier to quantify. For example, the reduction of coupled spin signal between 1.9 and 2.5 ppm in STEAM, due to ZQ effects, can make it easier to quantify the NAA signal at 2.02 ppm. As illustrated in Figure 3.5, in STEAM, the choice of both TE and mixing time (TM) intervals has a profound, but predictable impact (Thompson and Allen, 2001). Similarly, the asymmetry of PRESS TE1 vs. TE2 has a significant but predictable impact on coupled spin response (Thompson and Allen, 1999). To illustrate, the impact of TE1 vs. TE2 on Glu signal at 3 T is presented in the contour plots shown in Figure 3.6. These numerical calculations are made with and without the impact of RF pulse design limits included in the simulation. Water-reduction methods Some suppression of water, at least a factor of 10 , is usually required to avoid baseline distortion in proton MRS. Unsuppressed data is clearly the exception, and usually collected at long TE. While early

Artifacts and pitfalls in MR spectroscopy

(a) 100

(b) 100 .4

90

90

.8

80

2.5

.7

70

80

2 ppm

2.5

.6

2 ppm

70 .6

60

2.5

50 .7

.4

40

60 2.5

50

2 ppm

.4

.2 40

.9 .8

.7

30

2 ppm

TE1 (ms)

TE1 (ms)

45

30

.6

20

.4

.6 .4

20

.8

.6

.6 10 0

0

10

.9

1

10

1 .8

.4 20

30

40

50

60

70

80

90 100

0 0

10

20

30

40

TE2 (ms)

2.5

2 ppm

2.5

2 ppm

2.5

50 60 TE2 (ms)

70

2 ppm

80

90 100

2.5

2 ppm

Fig. 3.6 The impact of TE1 vs. TE2 in PRESS is illustrated for the Glu C3 and C4 protons via contour plots of signal intensity. (a) Illustrates the impact of TE1 vs. TE2 with perfect pulses and (b) includes the result including the impact of a typical slice-selective RF pulse design. From Thompson and Allen (1999) with permission of Wiley and Sons.

water-reduction methods were optimized for “complete” suppression of the water signal, most recent methods are designed to leave a well-controlled residual water peak for phase and eddy current correction. Dual-band spectral–spatial (Star-Lack et al., 2000; Schricker et al., 2001), “Mescher-Garwood pulse” (MEGA) (Mescher et al., 1998) and “Bandselective inversion with gradient diphasing” (BASING) (Star-Lack et al., 1997) are the best for this purpose, but the trade-off is longer minimum TEs. Chemical shift selective water suppression (CHESS) (Frahm et al., 1987) and its derivatives, for example water suppression enhanced through T1 effects (WET) (Ogg et al., 1994), utilize pre-excite-spoil pulse trains, and thus, do not impact TE. However, they are harder to control, and can leave water, which varies spatially, when further resolved in an MRSI study. Another difference between spectrally selective excitation and the pre-excite-spoil approach is how these methods respond to peak-to-peak inhomogeneity over the full FOV. Although inhomogeneity across the voxel is the only concern for spectral resolution,

inhomogeneity across the full MRSI FOV can lead to water-suppression artifacts, and may impact quantification. This is illustrated in Figure 3.7 for a modest 30 Hz offset, comparing BASING vs. CHESS for PRESS at TE 144. Note the impact of the BASING approach on the Cho signal at 3.2 ppm and also on the spin coupled responses from 2 to 3 ppm. Magnetization transfer effects Water-to-metabolite magnetization transfer (MT) has been demonstrated for Cho, NAA, Lac and ethanol, and must be considered in the design of water-suppression schemes (Leibfritz and Dreher, 2001). However, these effects are minimal for three and four pulse CHESS schemes, and are not a factor in spectral–spatial or in MEGA/BASING type acquisitions. It is safe to ignore this effect. In fact, reverse metabolite-to-water transfer has been shown to be an advantage for enhanced detection of cerebral ethanol (Estilaei et al., 2003) and exchangeable amide protons (Zhou et al., 2003).

46

Ralph E. Hurd

Real

Real

2.0 104

2.0 104

1.5 104

1.5 104

1.0 104

1.0 104

5.0 103

5.0 103

0

0 4

3 2 Frequency (ppm) BASING

1

4

3 2 Frequency (ppm) CHESS

1

Fig. 3.7 The impact of 30 Hz of peak-to-peak inhomogeneity over the total FOV is illustrated in these overlayed 1.5 T PRESS 144 spectra of the standard GE MRS HD sphere. For BASING, this level of frequency variation leads to both loss of Cho signal at 3.2 ppm and changes in coupled spin responses from Glu, NAA and Lac. The CHESS version does not show these artifacts.

Water reference Collection of separate unsuppressed water reference is especially useful in single-voxel or highly accelerated MRSI studies, where this can be accomplished in a small fraction of the overall sampling time. Such references provide quantitative water signals, and can be used to determine phase and eddy current correction to be applied to the suppressed data (Klose, 1990). They may also be useful for coil combination in cases where phased-array acquisitions are prescribed. However, this approach does not account for frequency changes during the study, and does not address any contribution from physiological noise. This is why data to be averaged should always be collected in separate, time-resolved frames, retaining residual water for correction of phase and frequency changes that may occur during the course of the exam. These imaging time changes can include both physiological and system level drift (Felblinger et al., 1998; Henry et al., 1999).

Water sidebands Although they have improved in recent years, gradient systems can be responsible for unwanted sidebands in the spectral response. If the parent signal is

large relative to the rest of the spectrum (such as un- or under-suppressed water), the sidebands can interfere with the spectral region of interest (ROI). These gradient-induced sidebands are usually not a problem at long TE values. This is in part because the sidebands are relatively short lived, and in part because water T2 is short relative to most metabolite signals. Gradient cycling and post-processing algorithms have been developed to deal with these sidebands. Two methods are illustrated here. First, if an unsuppressed water reference is collected, the residual water and corresponding sidebands can be subtracted from the partially water-suppressed data. The water and sideband spectrum must first be separated from metabolites to avoid impacting the true metabolite signal. This pure water-subtraction algorithm is illustrated in Figure 3.8. Another approach to eliminating the water sidebands is by using a 2DJ acquisition oversampled in t1 (Hurd et al., 1998). The sidebands modulate and are resolved in the J-dimension ( f1). As illustrated in Figure 3.9, this can be accomplished without any water suppression at all. In many cases, the nonmodulated metabolite components are sufficient, and simple TE averaging can be used. TE-averaged PRESS has also been useful to eliminate similar sideband artifacts from spectra expected to contain a

Artifacts and pitfalls in MR spectroscopy

47

large lipid signal (Bolan et al., 2002), and have more recently been used to measure Glu and Gln more directly (Hurd et al., 2003).

OVS

4

3

2

1

0

Fig. 3.8 These spectra illustrate the application of “pure water subtraction” to remove parasitic water artifact. The top trace is a normal PRESS TE 35 3 T spectrum contaminated with water sidebands. The middle trace is the unsuppressed–suppressed signal scaled to match the residual water intensity. The bottom trace is the corrected spectrum.

(a)

As described in section Pulse design and chemical shift, volume definition is limited by the transition and by the effective bandwidth of the slice-selective RF pulses. One way to improve the volume definition of PRESS or STEAM makes use of the very selective saturation (VSS) pulses. As they can be designed with an extensive phase gradient across their bandwidths, short saturation (SAT) pulses can have high effective bandwidth and narrow transition bands (Le Roux). These VSS pulses have been combined with over-prescription of the PRESS volume to significantly reduce chemical shift registration error (Tran et al., 2000). This approach is illustrated in Figure 3.10. In the absence of sufficient lipid suppression, these signals can propagate into spectra within the volume of interest due to the limited point sampling of k-space and/or due to phase noise.

(b)

(c) F1  F2 F1  F2/2

Real

Real

4000 5 104 F1

3000 2000

0

1000

5 104 (0,0)

200 100 0

100 200

Frequency (Hz)

200 100

0

100

Frequency (Hz)

200

0 8

6

4

2

Frequency (ppm)

F2 Fig. 3.9 2DJ TE-averaged artifact reduction is illustrated. (a) A 100 vertical expansion of a 1.5 T PRESS TE 35 spectrum without water suppression, and in this case dominated by parasitic water artifacts. (b) Stack plot of a 2DJ t1-oversampled data without water suppression is shown. (c) t1  0 trace from the 2DJ spectrum showing resolution from artifacts without water suppression. Adapted from Hurd et al. (1998) with permission of Williams & Wilkins.

0

(a) Chemical shift errors Phantom PRESS box images acquired with   100, 0, 100 Hz Conventional PRESS X: 180 SLR (1.1 kHz) Y: 90 SLR (2.5 kHz)

Overprescribed PRESS box with VSS

Y

X

(b) Applications of VSS to brain MRSI study

No VSS

VSS

Lipids

Fig. 3.10 (a) Chemical shift error is illustrated on the left with a voxel image collected at 100, 0, 100 Hz offset, green, white, red in conventional PRESS. The same set of images are collected on the right using over-PRESS, where VSS pulses are used to select the common passband in an over-prescribed PRESS. (b) This figure illustrates the use of VSS pulses to avoid contamination of MRSI spectra with out-of-band lipid. The high selectivity, and high effective bandwidth, of the VSS pulses, allows for the use of relatively wide SAT bands, retaining spectral quality in prescriptions near regions of high lipid signal. The contamination of lipid within the volume of interest, in the No VSS case, is due to the spatial sampling point spread, plus any phase noise. Adapted from Tran et al. (2000), courtesy of Wiley and Sons.

Artifacts and pitfalls in MR spectroscopy

Moderate insensitivity to B1 inhomogeneity is achievable using three pulses for each SAT band. MRSI quality benefits from out-of-band SAT, even when using PRESS volume selection. VSS lipid suppression is particularly important when the suspected pathology may involve accumulation of brain Lac, since residual lipid signals are often mistaken for Lac (as they resonate in the same region of the spectrum). For a positive identification of Lac, a doublet with a coupling constant (splitting) of 7 Hz centered at 1.3 ppm should be accurately determined. This may be best visualized at long TE because of the longer T2 of Lac compared to lipids, and at TE 140 ms the Lac signal (but not lipid) should be inverted in STEAM or PRESS.

Relaxivity Since SNR is often a limiting factor in spectroscopy, spectra are often run under conditions, which are significantly T1 weighted (optimum SNR for PRESS is when TR  1.3T1). For simpler spectra, or to accommodate spectral–spatial pulse design, etc., most MRSI spectra are also run under moderately T2 weighted conditions. Thus, regional (Brief et al., 2000) and disease (Kamada et al., 1994; Ke et al., 2002) changes in relaxivity add a level of uncertainty to the spectral interpretation, and without additional measurements, prevent “absolute” quantification. Carr-Purcell (Hennig et al., 1997) and a J-refocused (van Zijl et al., 1990; Lee et al., 1995) versions of PRESS both limit modulation of coupled spins at longer TEs, and can be optimized to offer different contrast between metabolites, than standard PRESS and STEAM and in some cases can be used to achieve a better estimate of the T2 of coupled spins.

Post-contrast spectroscopy Commercial chelated-gadolinium contrast agents, such as Magnevist™ and Omniscan™, are often used to localize, and via uptake dynamics, characterize lesions. Spectroscopic prescription, from contrastenhanced images, is therefore important. Thus, the

possible effects of contrast on spectroscopy need to be anticipated. The impact of these agents on metabolites in free solution (Murphy et al., 1999) suggests that extra-cellular metabolites, especially Cho, may have altered T1 and T2 relaxation times after the administration of intravenous contrast medium. However, gadolinium does not enter the intracellular space, where the majority of the metabolite signals are believed to originate from. For spectroscopy run under typical T1 and T2 weighted conditions, contrast may alter the result. However, as discussed in the previous section, disease and regional changes, alone, indicate a need for a measure of these relaxation times for quantitative studies. The contrast agent effects reported in the literature have not been significant, at least for short TE brain tumor studies (Lin and Ross, 2001). Minor effects have been reported at longer TEs (Sijens et al., 1998), but these effects are not always observed or significant. Figure 3.11 shows spectroscopy of an enhancing infarct, pre- and post-contrast, at TE 30, 144 and 288. In these spectra, the most prominent signals are Lac, Glu and Cho. This is a worse case, in that these chemicals could well exist in extra-cellular space, but within the SNR of this spectroscopic exam, they are not impacted by use of contrast prior to the spectroscopic study. Provided that magnetic field homogeneity is adjusted, post-contrast administration and peak-area measurements (rather than peak height) are made to account for possible contrastrelated increases in linewidth; it appears generally unlikely that contrast agents have any significant effect on brain spectra. TE averaging Another application of 2DJ-resolved PRESS is spectral simplification of the f 1  0 trace. This is the same as averaging TEs. For example, at 3 T and above, Glu is fully resolved from Gln and NAA (Hurd et al., 2003). This method also enables metabolite T2 measurement, as part of the acquisition. A conventional TE 35 PRESS spectrum is compared with a TE-averaged spectrum of normal gray matter (GM) in Figure 3.12. As illustrated by the component spectra of Glu, Gln, mI and NAA solutions, less overlap is achieved by TE averaging.

49

50

Ralph E. Hurd

TE 144

TE 30

TE 288

Post

Pre

Fig. 3.11 Spectra obtained before and after the intravenous administration of contrast medium of an enhancing lesion at TE 30, 144 and 288. Spectra are dominated by Lac, Glu and Cho. Some Cr signal is observed at the shortest TE. Within the SNR of these scans, post-contrast spectra are not significantly different relative to pre-contrast. Study courtesy of Dr. Orest Boyko, Temple University.

TE averaged

TE 35

NAA

NAA

Glu

Glu Gln

Gln

mI

mI 4.0

3.5

3.0

2.5

2.0

1.5

4.0

3.5

3.0

2.5

2.0

1.5

Fig. 3.12 Left: Conventional 3 T TE 35 PRESS spectrum of normal GM. Overlap of NAA, Glu, Gln and mI are illustrated by component spectra. Right: TE-averaged 35–185 spectrum of normal GM. Note resolution of Glu, NAA and Glx achieved by averaging process.

Artifacts and pitfalls in MR spectroscopy

Acceleration and improved coverage methods With the availability of higher field strengths and phased-array coils, the interest in accelerated MRSI methods has increased. Given the moderately long metabolite T1 relaxation time, multi-slice MRSI (Spielman et al., 1992; Duyn et al., 1993) can be an efficient approach to expanded coverage. Lipid suppression is a key concern for reliable measurements. An advantage of multi-slice compared to volumetric methods is the ability to dynamically update shims, slice-to-slice and optimizing homogeneity over the extended volume (Blamire et al., 1996; Hurd, 1997; Morrell and Spielman, 1997; de Graaf et al., 2003). Phase encoding during spectroscopic sampling time, t2, with methods such as echo-planar SI (EPSI) (Webb et al., 1989; Posse et al., 1994) or spiral encoding spiral-SI (Adalsteinsson et al., 1998) have also been effective at extending coverage, and with sufficient SNR, faster studies. Since these methods are often targeted at extending coverage, shimming and out-of-band lipid suppression are the greatest concern with respect to artifacts and limits. Lipid reduction methods include the use of multiple VSS pulses as described in section Water sidebands, inversion recovery nulling of the lipid component (Ebel et al., 2003), variable density sampling (Adalsteinsson et al., 1999; Sarkar et al., 2002) and post-processing (Ebel and Maudsley, 2001). Methods, which use t2-sampling time for chemical shift and spatial sampling, also reduce some of the spectroscopic over-sampling advantage, and thus may have less effective dynamic range. Spatial encoding, as in sensitivity encoding (SENSE) SI (Dydak et al., 2003), is yet another way to reduce k-space sampling. Since this approach is always accomplished with very high receptivity of exterior lipid signal, suppression of these is perhaps the most important acquisition requirement of reliable results. Like EPSI and spiral-SI, there is high demand on avoiding artifacts from out-of-band lipid.

may be useful. Despite the ability of some fitting methods to deal with spectral artifacts, variability and repeatability will suffer. Before using quantification from any fitting routine, it is always a good idea to look at the residuals and component spectra, or to set a minimum fit condition. The term “absolute concentration” in spectroscopy and SI is most often used to refer to the use of units in moles-per-tissue weight. This loose use of the term “absolute” does not necessarily imply that these values are corrected for relaxivity, voxel shape or for any other variable in data sampling, and should not be assumed to be corrected, unless specifically indicated. For any longitudinal study, in which system variability may contribute to the overall result, it is a good idea to use a standard spectroscopy phantom as routine quality control. One example of a routine spectroscopy phantom is a mixture, informally dubbed “liquid braino”, made in large batches and transferred to 2.7 l polyethylene spheres. The phantom consists of 12.5 mM NAA, 10 mM Cr, 3 mM Cho, 12.5 mM Glu, 7.5 mM mI and 5 mM Lac in a 50 mM phosphate buffer, adjusted to pH 7.2; 1% Magnevist™ v/v is included to reduce T2, and 0.1% w/w sodium azide is included to aid shelf life.

Summary The description of artifacts and pitfalls is not intended to scare anyone away from using MRS in research of clinical applications. For the most part, the sources and contribution of artifacts to variability are understood, and not the primary limit of typical MRS and MRSI studies. On the other hand, the increased use of longitudinal studies, where absolute tissue levels are required, the attention to the details of data prescription, set-up, collection and reconstruction may be vital. REFERENCES

Quantification The methods of MRS quantification are beyond the scope of this article (cf. Chapter 2), but a few comments with respect to the interface between data collection and quantification, assumptions and limits

Adalsteinsson E, Irarrazabal P, et al. 1998. Volumetric spectroscopic imaging with spiral-based k-space trajectories. Magn Reson Med 39(6): 889–898. Adalsteinsson E, Star-Lack J, et al. 1999. Reduced spatial side lobes in chemical-shift imaging. Magn Reson Med 42(2): 314–323.

51

52

Ralph E. Hurd

Allen PS, Thompson RB, et al. 1997. Metabolite-specific NMR spectroscopy in vivo. NMR Biomed 10(8): 435–444. Blamire AM, Rothman DL, et al. 1996. Dynamic shim updating: a new approach towards optimized whole brain shimming. Magn Reson Med 36(1): 159–165. Bolan PJ, DelaBarre L, et al. 2002. Eliminating spurious lipid sidebands in 1H MRS of breast lesions. Magn Reson Med 48(2): 215–222. Brief E, Whittal K, et al. 2000. Metabolite T1 differs within and between regions of normal human brain. In: Proceedings of 8th Annual ISMRM, 1939. Brooks WM, Friedman SD, et al. 1999. Reproducibility of 1HMRS in vivo. Magn Reson Med 41(1): 193–197. Cavassila S, Deval S, et al. 2000. Cramer-Rao bound expressions for parametric estimation of overlapping peaks: influence of prior knowledge. J Magn Reson 143(2): 311–320. de Graaf RA, Brown PB, et al. 2003. Dynamic shim updating (DSU) for multislice signal acquisition. Magn Reson Med 49(3): 409–416. Duyn JH, Gillen J, et al. 1993. Multisection proton MR spectroscopic imaging of the brain. Radiology 188(1): 277–282. Dydak U, Pruessmann KP, et al. 2003. Parallel spectroscopic imaging with spin-echo trains. Magn Reson Med 50(1): 196–200. Ebel A, Govindaraju V, et al. 2003. Comparison of inversion recovery preparation schemes for lipid suppression in 1H MRSI of human brain. Magn Reson Med 49(5): 903–908. Ebel A, Maudsley AA. 2001. Comparison of methods for reduction of lipid contamination for in vivo proton MR spectroscopic imaging of the brain. Magn Reson Med 46(4): 706–712. Ernst T, Chang L. 1996. Elimination of artifacts in short echo time H MR spectroscopy of the frontal lobe. Magn Reson Med 36(3): 462–468. Ernst T, Hennig J. 1995. Improved water suppression for localized in vivo 1H spectroscopy. J Magn Reson B 106(2): 181–186. Ernst T, Hennig J, et al. 1989. The importance of the voxel size in clinical 1H spectroscopy of the human brain. NMR Biomed 2(5–6): 216–224. Estilaei MR, Matson GB, et al. 2003. Indirect imaging of ethanol via magnetization transfer at high and low magnetic fields. Magn Reson Med 49(4): 755–759. Felblinger J, Kreis R, et al. 1998. Effects of physiologic motion of the human brain upon quantitative 1H-MRS: analysis and correction by retro-gating. NMR Biomed 11(3): 107–114. Frahm J, Merboldt K, et al. 1987. Localized proton spectroscopy using stimulated echoes. J Magn Reson 72: 502–508. Gruetter R, Weisdorf SA, et al. 1998. Resolution improvements in in vivo 1H NMR spectra with increased magnetic field strength. J Magn Reson 135(1): 260–264.

Hanson LG, Adalsteinsson E, et al. 2000. Optimal voxel size for measuring global gray and white matter proton metabolite concentrations using chemical shift imaging. Magn Reson Med 44(1): 10–18. Hennig J, Thiel T, et al. 1997. Improved sensitivity to overlapping multiplet signals in in vivo proton spectroscopy using a multiecho volume selective (CPRESS) experiment. Magn Reson Med 37(6): 816–820. Henry PG, van de Moortele PF, et al. 1999. Field-frequency locked in vivo proton MRS on a whole-body spectrometer. Magn Reson Med 42(4): 636–642. Hurd R. 1997. Interleaved MR spectroscopy and imaging with dynamically updated acquisition parameters. US 5,657,757. Hurd RE, Freeman D. 1991. Proton editing and imaging of lactate. NMR Biomed 4(2): 73–80. Hurd R, Sailasuta N. 1997. Elimination of artifacts in short echo proton spectroscopy. In Proceedings 5th Annual Meeting of ISMRM, Vancouver, Canada, p. 1453. Hurd RE, Gurr D, et al. 1998. Proton spectroscopy without water suppression: the oversampled J-resolved experiment. Magn Reson Med 40(3): 343–347. Hurd R, Sailasuta N, et al. 2003. Measurement of brain glutamate at 3T using TE-averaged PRESS. Magn Reson Med (in press). Kamada K, Houkin K, et al. 1994. Localized proton spectroscopy of focal brain pathology in humans: significant effects of edema on spin–spin relaxation time. Magn Reson Med 31(5): 537–540. Ke Y, Coyle N, et al. 2002. Brain NAA T2 values are significantly lower in schizophrenia. In Proceedings 10th Annual Meeting of ISMRM, Honolulu, Hi p. 976. Kim DH, Adalsteinsson E, et al. 2002. Regularized higher-order in vivo shimming. Magn Reson Med 48(4): 715–722. Klose U. 1990. In vivo proton spectroscopy in presence of eddy currents. Magn Reson Med 14(1): 26–30. Lee HK, Yaman A, et al. 1995. Homonuclear J-refocused spectral editing technique for quantification of glutamine and glutamate by 1H NMR spectroscopy. Magn Reson Med 34(2): 253–259. Leibfritz D, Dreher W. 2001. Magnetization transfer MRS. NMR Biomed 14(2): 65–76. Lin AP, Ross BD. 2001. Short-echo time proton MR spectroscopy in the presence of gadolinium. J Comput Assist Tomogr 25(5): 705–712. Mescher M, Merkle H, et al. 1998. Simultaneous in vivo spectral editing and water suppression. NMR Biomed 11(6): 266–272. Morrell G, Spielman D. 1997. Dynamic shimming for multislice magnetic resonance imaging. Magn Reson Med 38(3): 477–483.

Artifacts and pitfalls in MR spectroscopy

Murphy PS, Leach MO, et al. 1999. Signal modulation in (1)H magnetic resonance spectroscopy using contrast agents: proton relaxivities of choline, creatine, and N-acetylaspartate. Magn Reson Med 42(6): 1155–1158. Nelson SJ, Vigneron DB, et al. 1997. High spatial resolution and speed in MRSI. NMR Biomed 10(8): 411–422. Noworolski SM, Nelson SJ, et al. 1999. High spatial resolution 1H-MRSI and segmented MRI of cortical gray matter and subcortical white matter in three regions of the human brain. Magn Reson Med 41(1): 21–29. Ogg RJ, Kingsley PB, et al. 1994. WET, a T1- and B1-insensitive water-suppression method for in vivo localized 1H NMR spectroscopy. J Magn Reson B 104(1): 1–10. Pauly J, Le Roux P, et al. 1991. Parameter relations for the Shinnar Le-Roux RF design algorithm. IEEE Trans Med Imag 10: 53–65. Posse S, DeCarli C, et al. 1994. Three-dimensional echo-planar MR spectroscopic imaging at short echo times in the human brain. Radiology 192(3): 733–738. Raidy T, Sailasuta N, et al. 1995. Application of reduced flip angle: 180-degree RF pulses in PRESS. In Proceedings 3rd Annual Meeting of ISMRM, Nice, France, p. 1020. Ryner LN, Ke Y, et al. 1998. Flip angle effects in STEAM and PRESS-optimized versus sinc RF pulses. J Magn Reson 131(1): 118–125. Sarkar S, Heberlein K, et al. 2002. Truncation artifact reduction in spectroscopic imaging using a dual-density spiral kspace trajectory. Magn Reson Imag 20(10): 743–757. Schaffter T, Bornert P, et al. 1998. Fast 1H spectroscopic imaging using a multi-element head-coil array. Magn Reson Med 40(2): 185–193. Schricker AA, Pauly JM, et al. 2001. Dualband spectral–spatial RF pulses for prostate MR spectroscopic imaging. Magn Reson Med 46(6): 1079–1087. Sijens PE, Oudkerk M, et al. 1998. 1H MR spectroscopy monitoring of changes in choline peak area and line shape after Gdcontrast administration. Magn Reson Imag 16(10): 1273–1280. Spielman DM, Pauly JM, et al. 1992. Lipid-suppressed singleand multisection proton spectroscopic imaging of the human brain. J Magn Reson Imag 2(3): 253–262.

Spielman DM, Adalsteinsson E, et al. 1998. Quantitative assessment of improved homogeneity using higher-order shims for spectroscopic imaging of the brain. Magn Reson Med 40(3): 376–382. Star-Lack JM, Adalsteinsson E, et al. 2000. In vivo 1H MR spectroscopy of human head and neck lymph node metastasis and comparison with oxygen tension measurements. AJNR Am J Neuroradiol 21(1): 183–193. Star-Lack J, Vigneron DB, et al. 1997. Improved solvent suppression and increased spatial excitation bandwidths for three-dimensional PRESS CSI using phase-compensating spectral/spatial spin-echo pulses. J Magn Reson Imag 7(4): 745–757. Thompson RB, Allen PS. 1999. Sources of variability in the response of coupled spins to the PRESS sequence and their potential impact on metabolite quantification. Magn Reson Med 41(6): 1162–1169. Thompson RB, Allen PS. 2001. Response of metabolites with coupled spins to the STEAM sequence. Magn Reson Med 45(6): 955–965. Tran TK, Vigneron DB, et al. 2000. Very selective suppression pulses for clinical MRSI studies of brain and prostate cancer. Magn Reson Med 43(1): 23–33. van Zijl PC, Moonen CT, et al. 1990. Homonuclear J refocusing in echo spectroscopy. J Magn Reson 89: 28–37. Webb P, Macovski A. 1991. Rapid, fully automatic, arbitraryvolume in vivo shimming. Magn Reson Med 20(1): 113–122. Webb P, Spielman D, et al. 1989. A fast spectroscopic imaging method using a blipped phase encode gradient. Magn Reson Med 12(3): 306–315. Webb PG, Sailasuta N, et al. 1994. Automated single-voxel proton MRS: technical development and multisite verification. Magn Reson Med 31(4): 365–373. Zhou J, Payen JF, et al. 2003. Using the amide proton signals of intracellular proteins and peptides to detect pH effects in MRI. Nat Med 9(8): 1085–1090.

53

4

Fundamentals of diffusion MR imaging Derek K. Jones Section on Tissue Biophysics and Biomimetics, Laboratory of Integrative Medicine and Biophysics, National Institute of Child Health and Development, Bethesda, MD, USA

Key points • Diffusion-weighted imaging probes local tissue microstructure reflected by the freedom of microscopic motion of water molecules. • Apparent diffusion coefficient (ADC) maps are independent of T1 and T2 relaxation effects. • Diffusion tensor imaging (DTI) allows the anisotropy of water diffusion in vivo to be estimated: i.e. the degree to which the diffusivity of water depends on the direction in which it is measured. • While myelin modulates anisotropy, diffusion anisotropy is not a direct marker of myelin content per se. The major determinant of diffusion anisotropy is the architectural paradigm of the underlying tissue. • Parallel imaging techniques, high bandwidth acquisition, interleaved echo planar imaging and PROPELLER sequences are strategies to minimize artifacts due to susceptibility effects. • There are limitations to the single diffusion tensor model in describing fiber tracks due to heterogeneity within the voxel, and partialvolume effects will affect results.

The basics of diffusion Brownian motion and diffusion In 1827, when looking through a microscope at pollen grains from Clarkia pulchella, Robert Brown (a botanist from Montrose in Scotland) observed tiny 54

particles within the grains that appeared to move randomly (Brown, 1828). Intrigued by these movements (and with “essence of life” investigations in vogue), he examined pollen – even dead pollen – from other species and found the same random motion. We now know that the motion that Brown observed was not, of course, arising from the “essence of life”, but was due to bombardment of the pollen grains by gas molecules. This random molecular motion, often known as Brownian motion, and more frequently as diffusion, is the topic of this chapter. Diffusion is an essential physical process for the normal functioning of living systems. For example, the transport of metabolites into cells is facilitated by diffusion. As we will see in later chapters, studying diffusion has the potential to provide insights into both cell physiology and cell structure. Unlike other MR parameters, such as the longitudinal and tranverse relaxation time constants (T1 and T2) that are affected by experimental MR parameters, diffusion is an intrinsic property that is independent of the MR procedure employed to measure it. Diffusion and concentration gradients All of us will remember one of our earliest schooldays experiments where a large glass container, with a nonporous divider placed vertically down its middle, is filled with water and then ink is poured into one side. After the system has stabilized (i.e. convection currents have subsided), the divider is removed. With time, the ink disperses from the region of high concentration to the region of low concentration; this process continues until equilibrium is reached (i.e. the concentration of the ink is uniform throughout the container).

Fundamentals of diffusion MR imaging

The rate of change of the concentration of the ink is given by Fick’s first law that states that the flux density is linearly proportional to the concentration gradient. The constant of proportionality is the diffusion coefficient and therefore, by measuring the concentration of the species of interest over time, the diffusion coefficient can be estimated. Such approaches work when a concentration gradient exists, but when the diffusing species is in equilibrium, no net flux will be observed. Thus, extra species (i.e. chemical or radioactive labeled tracers that mimic the species of interest) must be introduced and the diffusion coefficients of the tracers used as a proxy for the species of interest. Such techniques are clearly unsuitable for routine clinical practice and so alternative techniques must be used. Diffusion as a random walk The ideal method for measuring diffusion, were it to exist, would be to keep track of the individual diffusing molecules directly (thereby eliminating the need to infer diffusivity from concentration gradients). Consider a set of molecules that are diffusing. After a certain observation time, each molecule would have a net displacement – but one that would be impossible to predict exactly for a particular spin. However, it is possible to predict the distribution of final positions of an ensemble of molecules from the theory of the random walk. By random walk, we mean that a spin remains in a certain location for a time t, then moves in a random direction by a fixed amount. This process continues so that a random path is traced out by the diffusing particle. Figure 4.1(a) shows a simulation of a random walk for five single particles in an isotropic medium (i.e. a medium exhibiting properties with the same values when measured along axes in all directions). For a particular particle, it is clear that we cannot accurately predict its position at a given time. However, Figure 4.1(b) shows the results of a simulation involving 1,000,000 such particles all starting from the same initial position. The histogram shows the distribution of total displacements (i.e. the absolute distance between the final positions and the initial positions) for different observation times. The histograms can be thought of as representing the probability of a

particular displacement for a specified diffusion time and hence the form of this histogram is often referred to as the displacement probability profile. Two qualities of the displacement probability profiles in Figure 4.1(b) are noteworthy. First, they have a bell-shaped or Gaussian form. (It can be shown mathematically that the distribution of final displacements for an ensemble of molecules undergoing a large number of steps is Gaussian, but such a proof is outside the scope of this chapter.) Second, the width of the curves at half their maximum height or their “full-width at half maximum” (FWHM) increases in proportion to the square root of the total number of steps. This relationship between displacement and diffusion time was formalized by Albert Einstein (Einstein, 1905) in the Einstein equation: r 2  2Dt

(4.1)

where 〈r 2〉 is the mean-squared displacement, D is the diffusion coefficient and t is the observation time. The equation shows that the longer the diffusion time and the greater the diffusivity, the greater the meansquared displacement. The Einstein equation is useful for obtaining order of magnitude estimates of the length scale probed during an observation time or diffusion time. For example, the diffusivity of water in parenchyma is on the order of 1.0 10 3 mm2 s 1, hence for a diffusion time of 35 ms, which is typical of diffusion times in clinical diffusion experiments, the root meansquared displacement will be on the order of 8 m, which is comparable to the diameter of the axon in white matter (WM).

Diffusion and MR How does diffusion affect the MR signal? The effects of diffusion on the MR signal were first reported over half a decade ago by Erwin Hahn (1950) who observed that In the case of echo phenomena it is found that nuclear signals due to precessing nuclear moments contained in liquid molecules (particularly of low viscosity) are not only attenuated by the influence of T1 and T2, but also suffer a decay due to the

55

Derek K. Jones

(a)

N  10

N  20

10

10

5

5

0

0

5

10

10

5

10

10

10

0 0

10 10

N  40

N  80

10

10

5

5

0

0

5

10

10

0 0

10 10

5

10

10

10

0 0

10

10 10

10 0 0

10 10

7

(b)

6

Percentage of spins

56

5

100 steps 400 steps 900 steps 1600 steps

4 3 2 1

x 2x 3x 4x

0

100 80 60 40 20 0 20 40 60 80 Net displacement from origin (arbitrary units)

100

Fig. 4.1 (a) Demonstration of a random walk in three dimensions. The pathways of five different particles, all starting at the origin, are shown in different colors. The numbers correspond to the total number of steps taken by each particle. For a particular particle, it is impossible to predict where it will be after a given number of steps. However, for an ensemble of spins, the distribution of final displacements can be predicted according to the theory of the random walk. (b) Histogram of final displacements for an ensemble of 1,000,000 particles, each undergoing a random walk of 100, 400, 900 and 1600 steps. The distribution is Gaussian. The horizontal lines correspond to the full-width at half maximum height (FWHM) of the profiles. Note that the FWHM increases in proportion to the square root of the number of steps.

Fundamentals of diffusion MR imaging

self-diffusion of the molecules into differing local fields established by external field inhomogeneities. To understand this statement, we begin by considering the familiar Larmor equation, which states that the angular frequency of precessing spins, , is directly proportional to the magnitude of the magnetic field, B, i.e.:   B

(4.2)

where  is the gyromagnetic ratio. If, as suggested by Hahn, the B field is non-homogenous, such that there is a gradient in the B field, then we can write: B(x)  B0  Gxx

(4.3)

where B(x) is the field as a function of position along the x-axis, x, B0 is the uniform field and Gx is the gradient in B along the x-axis. The phase accumulated by precessing spins at position x in a time interval, t, x(t) is: t

t

0

0

x(t )  ∫  dt ′  ∫ (B0  G x x ) dt ′

(4.4)

Eq. (4.4) shows that the accumulated phase is position dependent. If the gradient, Gx, is present throughout a spin-echo (SE) experiment, then the phase acquired prior to the 180° radio frequency (RF) pulse will be exactly matched in magnitude to the phase acquired after the 180° RF pulse. However, due to the inverting effect of the 180° RF pulse, the signs of the phase changes before and after the 180° RF pulse are reversed. Hence, regardless of a spin’s position with respect to the field gradient, as long as the spin remains stationary at position x during the entire experiment, the net phase change will be zero. However, as we know from earlier discussion, in diffusing species, the spins are not stationary and the longer the observation time and/or the greater the diffusivity, the greater the spread of positions of the spins (cf. Eq. (4.1) and Figure 4.1). Consequently, as the diffusion coefficient or the observation time increases, the phase dispersion will increase and the signal will become progressively attenuated. Diffusion encoding While diffusion attenuation of the MR signal due to field inhomogeneities was almost considered a

nuisance by Bloch (1950), by purposely imposing a field inhomogeneity in a controlled manner (e.g. imposing a uniform field gradient), the effects of diffusion can be magnified. Early experiments to this end (e.g. Carr and Purcell, 1954), involved superimposing a field gradient for the duration of the experiment (the constant field gradient SE approach). A schematic of this sequence is depicted in Figure 4.2(a). Note that the diffusion-encoding gradient acts as an additional readout gradient leading to compression of the MR signals in the time domain. The result is increased bandwidth in the frequency domain and hence a poorer signal-to-noise ratio (SNR). Furthermore, combining this approach with MRI can prove problematic since the additional field gradient interferes with slice selection. A great improvement to the constant field gradient experiment was the pulsed gradient spin-echo (PGSE) experiment, proposed by Stejskal and Tanner (1965) and shown schematically in Figure 4.2(b). The diffusion-encoding gradient is applied in two matched pulses, one placed either side of the 180° RF pulse. With this design, the diffusion-encoding gradients do not need to be applied during the slice selection or readout parts of the sequence, therefore avoiding the increase in bandwidth and slice-selection problems associated with the constant field gradient approach. Additionally, the diffusion time in the pulsed gradient approach is more readily characterized than that of the constant gradient approach. The Stejskal–Tanner sequence is the most widely used today. The b-factor For a fixed diffusion weighting, and for a diffusing species with a single diffusivity, D, it can be shown that the signal in a diffusion-weighted experiment is given by: I  I 0 e − TE/T2 e − bD

(4.5)

where I0 is the signal intensity in the absence of any T2 or diffusion weighting, TE is the echo time (TE), T2 is the transverse relaxation time and D is the apparent diffusivity. b is the “b-factor” or “b-value”, a single scalar quantity that characterizes the diffusion sensitization. The

57

58

Derek K. Jones

(a)

90ⴗ

180ⴗ

Echo

(b)

90ⴗ

180ⴗ

Echo

δ

δ Δ

Fig. 4.2 Schematic representations of diffusion-weighted MR sequences. (a) Constant field gradient SE scheme. The diffusion-encoding gradient is applied during the entire experiment. (b) PGSE sequence. The diffusion-encoding gradients are applied in two matched pulses. The duration is , their temporal separation is .

first exponential term is the familiar weighting due to transverse (T2) relaxation. The second term shows that the attenuation of the signal due to diffusion is exponential. Detailed derivation of the b-factor for a given sequence is beyond the scope of this chapter, but a typical example for the Stejskal–Tanner sequence is given below: ⎛ ⎞ b  (G)2 ⎜ ⎟ ⎝ 3⎠

(4.6)

where is the temporal separation of the gradient pulses,  is their duration and G is the gradient amplitude. This expression – the Stejskal–Tanner expression – is often used to obtain an order of magnitude estimate of the b-value for a given diffusion experiment. The diffusion time is assigned as (

/3), where the second term in the accounts for the finite duration of the pulsed field gradients. The units for the b-factor are s mm 2 and the range of values typically used in clinical diffusion weighting is 800–1500 s mm 2.

Increasing the b-factor Eq. (4.6) shows that the b-factor is increased by either increasing the gradient strength, G, the temporal separation of the gradients, , or their duration, . However, varying either or , varies the diffusion time and, as discussed in section Diffusion as a random walk (cf. Eq. (4.1)), length scales probed are then also changed. Hence, while such experiments have merits in their own right (e.g. Norris et al., 1994), if diffusivities are to be estimated from a series of measurements taken at different b-values, it is important to keep the diffusion time constant, and to change the b-value only by varying the gradient amplitude.

Diffusion-weighted MRI Diffusion-weighted imaging In 1984, Wesbey et al. (1984) described the first studies to incorporate diffusion-encoding gradients into an MR imaging (MRI) sequence. The following year,

Fundamentals of diffusion MR imaging

(a)

(b)

Fig. 4.3 Results from a diffusion-weighted imaging (DWI) experiment. (a) T2-weighted image (TE  107 ms). (b) Diffusion-weighted image collected from a healthy male subject (age 39 years). The b-value was 1000 s mm 2 and the direction of the diffusion sensitization is along the left–right axis.

Le Bihan and Breton (1985), Merboldt et al. (1985) and Taylor and Bushell (1985) all described their independent investigations in this area. Figure 4.3 shows a typical diffusion-weighted image obtained from a healthy human subject. The brightness of each voxel in the image corresponds to the diffusion-weighted intensity obtained with the same amount of diffusion weighting. Why then, given that we are looking at the effects of the diffusion of water (which has a diffusivity of 3.0 10 3 mm2 s 1), has the signal in each voxel not been attenuated by the same amount? (i.e. why is the image contrast not that of a standard T2-weighted image?). For example, the signal attenuation in the fluid filled ventricles is much greater than that observed in the parenchyma. As the signal attenuation is dependent on the net displacement over the diffusion time, we must conclude that the net displacement is greater in cerebrospinal fluid (CSF) than in parenchyma. Thus, it appears that the self-diffusivity of water is higher in CSF than in parenchyma. In fact, the true diffusivity of water in CSF and parenchyma is the same; however, in the CSF water molecules are relatively unhindered, but in parenchyma their pathways are hindered by the

presence of cell membranes and cellular inclusions. Relating this to Eq. (4.1), the mean-squared displacement per unit time, 〈r2〉/t, is reduced and thus it appears that the diffusivity, D, is reduced. Hence, to reflect the fact that we are not talking about the intrinsic self-diffusivity of water per se in MR diffusion imaging, the term apparent diffusion coefficient (or ADC) was coined (Tanner, 1978; Le Bihan et al., 1986). Use of the term ADC also clarifies that the estimated diffusivity comes from an ensemble of spins, i.e. all the spins contained within a voxel, and that this is a bulk-averaged estimate of the diffusivity. T2-shine through Diffusion-weighted images are typically acquired with a b-value on the order of 800–1500 s mm 2. Assuming a gradient amplitude, G, in Eq. (4.6) of 22 mTm 1, and assuming that  , then by re-arranging Eq. (4.6) the duration of the diffusion-encoding gradients 3 must be at least 冪3b/(22G 2)  40 ms. Hence, the TE must be at least twice this value (and when imaging gradients are included, the TE can be considerably longer). Consequently, for gradient amplitudes that are available on clinical scanners (10–40 mTm 1),

59

60

Derek K. Jones

(a)

(b)

(c)

(d)

Fig. 4.4 Demonstration of the potential confusion arising from the T2-shine through effect. The top row shows (a) a T2-weighted and (b) corresponding diffusion-weighted (b  1000 s mm 2) image from a 71-year old male with a 36 h cerebral infarction in the left anterior parietal region. The infarct is only subtly visible on the T2-weighed image but is very conspicuous on the diffusion-weighted image as a region of hyperintensity. The bottom row, however, shows that hyperintensity on the diffusion-weighted image does not always mean reduced ADC. The T2-weighted (c) and diffusion-weighted images (d) in the bottom row were obtained from a 60-year old male with a 7-day old cerebral infarction in right occipital and temporal lobes. The majority of the hyperintensity in the diffusion-weighted image can be attributed to the hyperintensity in the T2-weighted image, or T2-shine through. These images are reproduced from Burdette et al. (1999) with kind permission of the author and the publishers (the Radiological Society of North America).

Fundamentals of diffusion MR imaging

diffusion-weighted images are heavily T2-weighted. This weighting can lead to ambiguities in diffusionweighted signals, commonly referred to as the “T2-shine through” effect (Burdette et al., 1999; Provenzale et al., 1999), depicted in Figure 4.4.

Quantitative estimates of diffusivity Diffusion-weighted images offer only qualitative insight into the diffusion process and, coupled with the effects of T2-shine through, can be difficult to interpret. However, as should be clear from Eq. (4.5), by acquiring at least two images, I1 and I2, with different b-values (b1 and b2), it is possible to obtain a quantitative estimate of the diffusivity: I1  I 0 e − TE/T2 e − b1D

(4.7)

I1  I 0 e − TE/T2 e − b2D

(4.8)

Taking logarithms of both sides of Eqs. (4.7) and (4.8) and dividing Eq. (4.7) by Eq. (4.8) yields: D

1 ln( I 2 I1 ) (b2 b1 )

(4.9)

Note that the term I0e TE/T2 appears in both Eqs. (4.7) and (4.8), and hence, when taking the ratio of I1/I2 in formulating Eq. (4.9), this term cancels out. Thus, the estimate of the diffusivity is free from the T2-shine through effect seen in diffusion-weighted images.

Anisotropic diffusion Implications for diffusion weighted imaging The clinical utility of diffusion weighed imaging (DWI) first became apparent in 1990, when Moseley et al. (1990a) reported an acute reduction of the ADC in ischemic tissue in the cat brain within the first few hours of onset of ischemia. The application of DWI to ischemia is discussed in Chapter 12. About the same time as Moseley’s initial observations, it was noted that the ADC in certain regions of the mammalian brain appeared to depend on the direction of the

applied diffusion-encoding gradient (Moseley et al., 1990b). In other words, the ADC was directionally dependent. This effect had been known for some time in ex vivo samples of muscle and brain tissue dating back to the pioneering work of Hansen (1971) and Cleveland et al. (1976). Shortly after Moseley’s observation in the cat brain, the directional dependence of the ADC was reported in human WM by Doran et al. (1990) and Chenevert et al. (1990). This is illustrated in Figure 4.5, which shows the same brain imaged three times, each time with the diffusion-encoding gradient applied along one of three orthogonal axes. In certain regions of the brain, the diffusionweighted intensity is the same in all three images suggesting that the ADC is the same in all directions. Diffusion in these cases is described as isotropic. However, in the regions highlighted by arrows this is not the case, and diffusion in these regions is referred to as anisotropic. From just these three diffusion-weighted images, we can infer a substantial amount of information about the structure indicated by the arrows. First, the large differences in diffusion-weighted intensities that are observed as the direction of the diffusionencoding gradient is changed, suggest that the tissue here is highly ordered on the voxel scale. Second, as there is high signal attenuation in Figure 4.5(a) (in which the diffusion-encoding gradients were applied in a left–right orientation), we can infer that diffusion is relatively unhindered along this direction. Conversely, in the two perpendicular orientations (Figures 4.5(b) and (c)), the signal attenuation is much less, indicating that the mean-squared displacement per unit time is reduced and that something is therefore hindering the displacement of water molecules along these orthogonal axes. Therefore, from just these three images, we are able to infer an ordered structure has predominantly a left–right orientation. These inferences are entirely consistent with the fibers of the corpus callosum, which passes through this region (Dejerine, 1895; Crosby et al., 1962). What is the source of diffusion anisotropy? Initial suggestions for the mechanisms underlying diffusion anisotropy in WM included the myelin sheath (Thomsen et al., 1987), local susceptibility

61

62

Derek K. Jones

(a)

(b)

(c)

(d)

(e)

(f)

Fig. 4.5 Effect of changing the direction of the diffusion-sensitizing gradients on the diffusion-weighted intensity (top row) and computed ADC (bottom row). The figure shows the same brain slice, with gradients applied in the left–right direction (a and d), anterior–posterior direction (b and e), and superior–inferior direction (c and f). The amount of diffusion weighting (b  1000 s mm 2) was the same in all three cases.

gradients (Hong and Dixon, 1992; Lian et al., 1994), axonal cytoskeleton and fast-axonal transport. Myelin itself, however, does not appear to be necessary in order for diffusion to be anisotropic in the brain. This conclusion was first suggested by the demonstration of anisotropic diffusion in the immature rat brain where there was no histological evidence of myelin (Wimberger et al., 1995; Prayer et al., 1997). Furthermore, Gulani et al. (2001) reported anisotropic diffusion in the spinal cord of a myelin-deficient rat.

In the mid-1990s, Christian Beaulieu and Peter Allen conducted a series of experiments to try to elucidate the origin of diffusion anisotropy in WM (Beaulieu and Allen, 1994a, 1994b, 1996) and were able to rule out the effects of susceptibility induced gradients, axonal cytoskeleton and fast-axonal transport. They concluded that the main determinant of anisotropy in nervous tissue is the presence of intact cell membranes and that myelination serves to modulate anisotropy. For a thorough and

Fundamentals of diffusion MR imaging

excellent review of the work done in this area, see the recent article by Beaulieu (2002).

The diffusion tensor Why is a single ADC inadequate for characterizing diffusion in vivo? If we wish to report the diffusivity in regions of anisotropic tissue such as those described above, which ADC should we report? Imagine that, in a multi-center study, we wished to compare measurements of the ADC with gradients applied in a left–right direction in the splenium of the corpus callosum. Clearly, a meaningful comparison of values obtained from different sites would require that each subject’s head be oriented in exactly the same way. If the diffusion-encoding gradients were applied along the x-axis and the subject rotated his head slightly, the estimated ADC would change. We say, therefore, that the ADC is a rotationally variant measure. Clearly an infinite number of ADC measures can be obtained within anisotropic tissue; it is, therefore, also clear that a single ADC is inadequate for characterizing diffusion and a more complex description is required. The next most complex description is the diffusion tensor matrix (Crank, 1956). This is a 3 3 symmetric matrix, D, ⎡Dxx ⎢ D  ⎢Dxy ⎢⎣Dxz

Dxy Dyy Dyz

Dxz ⎤ ⎥ Dyz ⎥ Dzz ⎥⎦

(4.10)

What do the elements of the diffusion tensor represent? The diagonal elements (Dxx, Dyy, Dzz) correspond to the diffusivities along three orthogonal axes and the off-diagonal elements relate diffusivities along those orthogonal axes. It is essential to realize that the offdiagonal elements of the diffusion tensor do not represent diffusivities per se. For example, the element Dxy is not the diffusivity in the xy direction. The off-diagonal elements reflect the correlation between molecular displacements in orthogonal directions,

hence Dxy correlates displacements along the x- and y-axes. Imagine an anisotropic medium oriented such that the axis of greatest diffusivity is at 45° to both the x- and y-axes (cf. Figure 4.6(a)). The diffusivity along the x-axis will be matched in amplitude to the diffusivity along the y-axis. Furthermore, displacements along the x-axis will be perfectly correlated with displacements along the y-axis. The fact that these displacements are correlated will be reflected by a non-zero value of the off-diagonal element of the diffusion tensor, Dxy. Imagine now that the anisotropic median is slowly rotated counter-clockwise so that the axis of greatest diffusivity becomes increasingly aligned with the y-axis. Displacements along the xaxis and y-axis will become increasingly less correlated, which will be reflected by an increasingly smaller value of Dxy. At the point that the axis of greatest diffusivity becomes perfectly aligned with the y-axis (and therefore diffusion along the x-axis is no longer correlated with diffusion along the y-axis), the off-diagonal element Dxy, will become zero. Note that although Dxy is zero, the diffusivity in the x–y direction, ADCxy, is not zero. When all three off-diagonal elements are zero, this means that the tensor is aligned with the principal axes of the anisotropic medium. In this condition, we say that the tensor is diagonalized and its diagonal elements correspond to its eigenvalues. The three eigenvalues (denoted as 1, 2 and 3) correspond to the three diffusivities along the principal axes of the diffusion tensor. The orientation of the principal axes is given by the three eigenvectors (denoted by 1, 2 and 3). By definition, the three eigenvectors are mutually orthogonal. The orientation of the tensor is taken to be parallel to the principal eigenvector, which is the eigenvector associated with the largest eigenvalue. The principal eigenvector is assumed to be collinear with the dominant fiber orientation within the voxel. The diffusion ellipsoid and its relation to the diffusion tensor Now consider a gedanken experiment in which we place a drop of ink in the center of a large vat of water. As the ink particles displace over time, the outer profile of the displacements would resemble a spherical

63

Derek K. Jones

(a) y

(b) y

ADCy ADCyy

ADCxy

ADCxy ADCx ADCx

u x

x

(c) ADCxy / Dxy ( 10 3 mm2s 1)

64

1.2 1.0

ADCxy

0.8 0.6 0.4

Dxy

0.2 0.0

0.2

0.4

0.6

0 15 30 45 60 75 90 105 120 135 150 165 180 u (degrees)

envelope, since diffusion in isotropic media is isotropic. However, in an anisotropic medium, the ink particles would diffuse further along the principal axis of the anisotropic medium than in a perpendicular orientation. Clearly, the displacement profile can no longer be described by a sphere and is more correctly described by an ellipsoidal envelope with the long axis parallel to the long axis of the anisotropic medium. The diffusion tensor is often thought of in terms of this ellipsoid – a surface representing the distance that a spin will diffuse to with equal probability from the origin. The orientations of the axes of the ellipsoid are given by the eigenvectors, and the lengths are given by the diffusion distance in a given

Fig. 4.6 Explanation of the difference between Dxy, the offdiagonal element of the diffusion tensor and the ADCxy, the ADC in the xy direction. In (a), the anisotropic medium is oriented at 45° to both the x- and y-axes. The diffusivity in the x-direction is equal to the diffusivity in the y-direction, and displacements along the two axes are perfectly correlated (reflected by Dxy taking its maximal value). In (b), with the anisotropic medium aligned with the y-axis, displacements along the x- and y-axes are no longer correlated and Dxy equals zero. However, the ADC in the direction (x,y), is not zero. Further, while Dxy can take negative values, the ADC in the direction (x,y) can, by definition, never take negative values. In (c), the variation of both Dxy and ADCxy is plotted as a function of the orientation of the principal axes of the anisotropic medium with respect to the y-axis ().

time, t. Eq. (4.1) shows that the displacement in a given time is proportional to the square root of the diffusivity. Hence, the ellipsoid’s axes are scaled according to the square root of the eigenvalues, 冪苶 1, 冪苶 2 and 冪苶 3. Example ellipsoids are shown in Figure 4.7. How is the diffusion tensor estimated? The fact that a tensor should be used to describe diffusion in anisotropic systems was recognized in the MR field nearly 30 years ago (Stejskal and Tanner, 1965), when it was incorporated into the Bloch equations (Bloch, 1946) that describe the fundamentals of NMR. However, no attempt was made to estimate

Fundamentals of diffusion MR imaging

εˆ 3

εˆ 1

λ3

λ1

λ3 εˆ 3

λ1

λ2 εˆ 2

λ2 εˆ 2

εˆ 1

Fig. 4.7 Schematic of the diffusion ellipsoid. The ellipsoid is the envelope where a spin – placed at its center – will diffuse with equal probability in a time t. The axes are scaled according to the square root of the eigenvalues, 兹苶 1, 兹苶 2, and 兹苶 3, and the principal axes are given by the corresponding eigenvectors, ˆ1 ˆ2 and ˆ3. The eigenvalues are sorted according to their magnitude such that 1  2 3. The tensor in a. is prolate, where 1  2 ⬇ 3. The principal eigenvector is designated as ˆ1. In b, the tensor is oblate, i.e. 1 ⬇ 2  3 and the principal eigenvector is therefore poorly defined.

the tensor directly from diffusion-weighted data until Basser and colleagues described the first diffusion tensor measurements in 1992 (Basser et al., 1992) and later described how this approach could be combined with imaging into what is now known as diffusion tensor MR imaging (DT). As the tensor is symmetric (i.e. Dxy  Dyx, Dxz  Dzx and Dyz  Dzy), there are only six unknown elements to determine. These are estimated from a series of diffusion-weighted images acquired with gradients applied in non-collinear and non-coplanar directions. We are all familiar with the idea that to find n unknowns in linear algebra, we need to solve at least n simultaneous equations, and the same applies when estimating the diffusion tensor from MR data. The minimum number of diffusion-encoding images required for estimating the tensor is six (with the addition of one non-diffusion-weighted image). Basser et al. (1994a) showed that by taking the logarithm of the diffusion-weighted intensities (cf. Eq. (4.5)), a set of simultaneous linear equations is established which can be solved using linear algebra, and estimates of the diffusion tensor obtained. In this case, the scalar b-value is replaced by the so-called b-matrix (Basser et al., 1994a), whose elements, bij, scale the attenuation of the signal by the

corresponding element of the diffusion tensor, Dij. Thus, Eq. (4.5) becomes: I  I 0 e TE/T2 e

b xxD xx b yyD yy bzzDzz 2b xyD xy 2b xzD xz 2b yzD yz

(4.11) Computation of the b-matrix is outside the scope of this chapter, but the interested reader is referred to Mattiello et al. (1994). Figure 4.8 shows nine images, each corresponding to an element of the diffusion tensor. Visually extracting information (such as mean diffusivity, degree of diffusion anisotropy and axis of greatest apparent diffusivity) from such a set of images, is difficult. However, the data in these images can be used to reconstruct the diffusion ellipsoid discussed earlier. Figure 4.9 shows some diffusion ellipsoids computed from actual DTI data. In the CSF-filled lateral ventricles, diffusion is isotropic and unhindered and hence the ellipsoids are spherical and have large radii compared with the other ellipsoids. In the cortex, diffusion is more hindered than in CSF, but at the voxel resolution typical of DT-MRI (2 2 2 mm), there is no preferred diffusion direction. Hence, the ellipsoids in cortex also appear spherical, albeit with a smaller radius than in CSF. In the WM

65

66

Derek K. Jones

Dxx

Dxy

Dxz

Dxy

Dyy

Dyz

Dxz

Dyz

Dzz

Fig. 4.8 Elements of the diffusion tensor. The images of Dxx, Dyy and Dzz show the diffusivity along the x-, y- and zaxes respectively, while the images of Dxy, Dxz and Dyz show correlations between displacements in orthogonal directions. Note that Dxy, Dxz and Dyz can take either positive or negative values, while Dxx, Dyy and Dzz (corresponding to diffusivities) take only non-negative values. Hence, while the three diagonal elements of the tensor have been windowed the same, and the three off-diagonal elements have been windowed the same, different windows have been used for the diagonal and off-diagonal elements.

of the splenium of the corpus callosum, diffusion is more hindered perpendicular to the long axis of the fibers and hence the ellipsoids become elongated along the fiber direction. The ellipsoids in this region have been likened to a string of link-sausages (Peter Basser, personal communication). By following the link-sausages by eye, an impression of the trajectory of the fibers in the corpus callosum is created.

Quantitative measurements from diffusion tensor MRI A key quality of many diffusion tensor measures is that they are rotationally invariant, i.e. the value that they take is independent of the orientation of the sample with respect to the laboratory frame of reference. As discussed above, this is an important

Fundamentals of diffusion MR imaging

Fig. 4.9 Diffusion ellipsoids reconstructed from real DTI data. The brain slice is a T2-weighted axial image and the box shows the location of a region of interest (ROI) centered over the splenium of the corpus callosum, but also containing portions of the lateral ventricles and occipital cortex. The zoomed region shows the ellipsoids computed from within this region. This figure originally appeared in Pierpaoli et al. (1996). The author is grateful to both the Radiological Society of North America and Dr. Carlo Pierpaoli, National Institutes of Health, Bethesda, MD, for permission to reproduce it here.

quality when comparing measures obtained from different subjects, or even when comparing measures from different regions of the brain of the same subject.

Trace Without doubt, the most clinically useful measure obtained from DTI is the trace. This is the sum of the three diagonal elements of the diffusion tensor (i.e. Dxx  Dyy  Dzz), which can be shown to be equal to the sum of its three eigenvalues. The trace/3 can be thought of as being equal to the orientationallyaveraged mean diffusivity. Note that, particularly in

the earlier diffusion MRI literature, many alternative phrases have been used to describe this measure, including trace ADC and mean trace ADC. These terms are nonsensical since the trace is a property of tensors, while an ADC is a scalar quantity and refers to the diffusivity along one axis; the use of such terms should therefore be avoided. An example of an image of the mean diffusivity (i.e. trace/3) is provided in Figure 4.10, along with estimates of the ADC along three orthogonal axes. A remarkable property of the trace is that, in the b-value range typically used in clinical studies (b  1500 s mm 2), the mean diffusivity is fairly

67

68

(a)

Derek K. Jones

(b)

(c)

(d)

Fig. 4.10 Images of the diagonal elements of the diffusion tensor (a) Dxx, (b) Dyy and (c) Dzz and (d) the mean diffusivity  1/3(Dxx  Dyy  Dzz  1/3 Trace(D). Notice that while it is possible to infer fiber orientation and anisotropy from the three individual images, the mean diffusivity is relatively uniform throughout parenchyma. While this homogeneity makes anatomical localization difficult, it can make regions of abnormal diffusivity (either elevated or lowered) conspicuous.

uniform throughout parenchyma (0.7 10 3 mm2 s 1). Although homogeneity makes it difficult to distinguish anatomical structures, it does offer the advantage that the effects of anisotropy do not confound detection of diffusion abnormalities, such as acute ischemic lesions (Lythgoe et al., 1997).

1996). To circumvent this problem, indices that do not require sorting (Basser and Pierpaoli, 1996; Pierpaoli and Basser, 1996) have been proposed and have been shown to be less sensitive to the SNR. The two most popular are the fractional anisotropy (FA) and relative anisotropy (RA), given by:

Anisotropy indices Prior to the introduction of the tensor model into MRI by Basser et al. (1992), several indices for diffusivity were proposed, such as the ratio of ADCs obtained in two orthogonal directions. The limitation of such indices can be understood by referring back to Figure 4.6. For the fibers oriented at 45° to the x- and y-axes, the ratio ADCy/ADCx is equal to unity, for the fibers oriented along the y-axis, the ratio ADCy/ADCx takes its maximal value, and for the fibers oriented along the x-axis, the ratio takes its minimal value. This is, therefore, another example of a measure that is rotationally variant. Anisotropy indices formed from the eigenvalues of the tensor will, by definition, be rotationally invariant. The simplest anisotropy index, analogous to the ratio ADCy/ADCx would be the ratio of the largest to the smallest eigenvalue (i.e. 1/3). However, it has been shown that sorting the eigenvalues according to their magnitude introduces a bias in the measurements at low SNRs (Pierpaoli et al.,

3 FA  2

(  )  ( 2

1

2



)  (  ) 2

2

3

(4.12)

21  22  23

and

(  )  ( 2

1 RA  3

1

2

 

)  (  ) 2

2

3

(4.13)

where  

1 ( 1  2  3 ) 3

(4.14)

The numerator for both terms is the same and is related to the variance of the three eigenvalues about their mean. The FA index normalizes this measure by the magnitude of the tensor as a whole. Just as the magnitude of a vector can be found from the sum of the squares of its individual components, the magnitude of the tensor is found from the sum of the squares of its eigenvalues. Thus, FA measures

Fundamentals of diffusion MR imaging

Table 4.1. Eigenvalues, trace, mean diffusivity, and anisotropy indices computed from different brain regions. The eigenvalues were taken from Pierpaoli et al. (1996) and were obtained from six monkey brains

1 2 3 Trace Mean diffusivity FA RA A

Splenium

Internal capsule

Optic radiation

Centrum semiovale

U fibers

Frontal cortex

CSF

1.685 0.287 0.109 2.081 0.694

1.320 0.447 0.139 1.906 0.635

1.460 0.496 0.213 2.169 0.723

0.995 0.602 0.349 1.946 0.649

1.200 0.545 0.208 1.953 0.651

1.002 0.810 0.666 2.478 0.826

3.600 3.141 2.932 9.673 3.224

0.873 1.016 0.718

0.758 0.787 0.557

0.727 0.738 0.522

0.464 0.410 0.290

0.655 0.633 0.447

0.201 0.167 0.118

0.106 0.087 0.061

the fraction of the tensor that can be assigned to anisotropic diffusion. The FA index is appropriately normalized so that it takes values from zero (when diffusion is isotropic) to one (when diffusion is constrained along one axis only). The denominator of the RA index is simply the mean diffusivity. This index is mathematically identical to a coefficient of variation, i.e. standard deviation divided by the mean. To ensure that this index scales from zero to one, Shimony et al. (1999) divided the RA index — by 冪 2, and named the index A , i.e.: A 

RA 2

(4.15)

Table 4.1 shows some example values of trace, FA, RA and A , in different regions of the brain. The most commonly used anisotropy index in the literature is the FA. Example images showing FA for the whole brain in axial, coronal and sagittal planes are presented in Figure 4.11. The relative merits of the various anisotropy indices have been discussed by Papadakis et al. (1999). It should be noted that even though measures such as FA and RA are still less sensitive to noise than measures such as 1/3, they are nevertheless sensitive to noise. As the SNR is lowered, the anisotropy indices become increasingly overestimated (Pierpaoli et al., 1996). Thus comparisons of anisotropy indices obtained from different studies in which different imaging parameters have been used should be treated with caution.

Tensor orientation We saw in Figure 4.5 that for structures predominantly oriented along the principal axes (x, y and z), it was possible to infer fiber orientation from three diffusion-weighted images or three ADC images in which the diffusion encoding was applied along those three orthogonal axes. Early work attempted to capitalize on this ability and approaches were proposed for creating fiber orientation maps based on ADC measurements (Douek et al., 1991; Nakada et al., 1994). However, as should now be clear from previous discussion, these maps are rotationally variant. Jones et al. (1997) and Pierpaoli and Pajevic (1997), showed how robust and readily interpreted fiber orientation maps could be derived by using the information contained within the diffusion tensor (more specifically, the eigenvector associated with the largest eigenvalue). The key idea is that components of the orientation of the fiber are represented using different primary colors. Figure 4.12 shows an example of the absolute direction scheme proposed by Pajevic and Pierpaoli (1999), the most commonly used scheme to date. By viewing fiber orientation in one voxel and following, by eye, a path of smooth transition in color from one voxel to the next, one can gain an impression of the trajectory of the major WM pathways. In fiber tracking or tractography (Conturo et al., 1999; Jones et al., 1999a; Mori et al., 1999; Basser et al., 2000; Poupon et al., 2000; Parker et al., 2000, 2002; Tuch et al., 2000; Koch et al., 2001; Behrens et al., 2003), algorithms are used to perform a similar task – i.e. following smooth pathways in the fiber orientation

69

70

Derek K. Jones

(a)

Fig. 4.11 Whole brain FA data collected from a healthy male volunteer (29 years) in (a) axial, (b) coronal and (c) sagittal sections. The intensity of the image is directly proportional to anisotropy. The CSF-filled regions (sulci and ventricles) and gray matter (GM) have low intensity as the self-diffusion of water is isotropic at the voxel resolution (2.5 2.5 2.5 mm). In the WM, where diffusion is more anisotropic, the image appears bright.

Fundamentals of diffusion MR imaging

(b)

(c)

Fig. 4.11 (cont.)

71

72

Derek K. Jones

cingulum

occipito-frontal fasciculus arcuate(or longitudinal superior) fasciculus uncinate fasciculus inferior longitudinal fasciculus middle cerebellar peduncle

pontine transverse fibers

cortico-spinal tract

forceps major

superior cerebellar peduncle middle cerebellar peduncle sensory pathways vertical occipital fibers

optic radiations

Fig. 4.12 Example of color encoded fiber orientation maps. Fibers that are predominantly oriented left–right are shown in red, anterior–posterior fibers are shown in green and superior–inferior fibers are shown in blue (see color wheel at lower right hand corner). Comparing these images with those in Figure 4.11(b), it is clear that the color maps provide more information than anisotropy maps alone. Figure taken from Pajevic and Pierpaoli (1999). © 1999. Reprinted by permission of Wiley-Liss, Inc. a subsidiary of John Wiley& Sons, Inc. The author is grateful to Dr. Carlo Pierpaoli, National Institute of Child Health and Human Development, Bethesda, MD, for supplying the figure.

Fundamentals of diffusion MR imaging

field to reconstruct WM pathways – in an automated way. Tractography will be discussed in detail in Chapter 5.

Optimization of diffusion-weighted and DT-MRI sequences As with all MR sequences, the experimental parameters should be chosen to yield the most reliable results possible in the time available. In DWI (particularly DTI), a number of quantities can be optimized. What is the optimal b-value? A simple calculation that minimizes the variance in the estimated ADC along a particular direction shows that the optimal diffusion weighting is of the order of b  1.1/ADC (Bito et al., 1995).1 Assuming an ADC of 0.7 10 3 mm2 s 1 (the mean diffusivity in parenchyma), the optimal b-factor according to this rule is 1571 s mm 2. However, it should be remembered that gradient power is limited and so, as the Stejskal–Tanner equation (Eq. (4.6)) shows, to achieve this b-value, the duration of the diffusion-encoding gradients must be increased, leading to a concomitant increase in TE and increased transverse (T2) relaxation. It has been shown that for typical imaging experiments, the optimal b-factor is about 77% of that value (Jones et al., 1999b), i.e. b 艐 1257 s mm 2. What is the optimal number of measurements at each b-value? Bito et al. (1995) and Eis and Hoehn-Berlage (1995) have shown that for a given number of diffusionweighted images, the most precise estimate of the ADC will be achieved when just two diffusionweighting amplitudes are used (as opposed to using more than two equally or unequally spaced amplitudes). Furthermore, Bito et al. (1995) showed that, for estimating an ADC, the optimal ratio of the number of measurements made at the higher b-value to the number made at the lower b-value, Nhigh/Nlow, 1

This is often approximated to a rule of thumb that states that the optimal diffusion weighting occurs when b. ADC ⬇ 1.

was 3.6 : 1. Jones et al. (1999b) later showed that the optimal value of Nhigh/Nlow for estimating the diffusion tensor was 11.3 : 1. However, to balance the effects of T2-weighting as discussed above, the practical optimal ratio is 77% of this value, i.e. 8.7 : 1. What is the optimal arrangement of the sampling vectors for DT-MRI? While it is possible to estimate the tensor from just six diffusion-weighted images and one non-diffusionweighted image, if time permits acquiring more images is beneficial since it boosts the precision of the experiment. When time permits the collection of only six diffusion-weighted images, the dual-gradient scheme proposed by Davis et al. (1993), and later popularized by Pierpaoli et al. (1996) is the most widely used gradient arrangement. This scheme involves placing gradients at full gradient amplitude, Gfull, on two axes simultaneously. The resultant gradient amplitude is 2 therefore 冪G 2full  G full  冪2Gfull. Again, the Stejskal– Tanner expression (Eq. (4.6)) shows that increasing G for a given b-factor, leads to a reduction in the required duration of the diffusion-encoding gradients, a reduction in the TE and therefore increased SNR per unit time. If time permits more measurements, then it has been shown (Jones, 2003) that it is beneficial to acquire as many unique sampling orientations as time will allow, with an asymptotic limit at 30 unique orientations. Skare et al. (2000) compared various DTMRI sampling schemes in which the number of unique sampling orientations varied, and concluded that the optimal scheme out of all those tested was a scheme in which 30 unique sampling directions (uniformly distributed over the sphere) are employed (Jones et al., 1999b). Later studies (Batchelor et al., 2002, 2003; Jones, 2003) have confirmed that schemes involving at least 30 unique sampling orientations (distributed according to the algorithm proposed in Jones et al., 1999b) are optimal for DT-MRI. What is the optimal TE? This question is difficult to answer directly, since it depends on the design of the pulse-sequence.

73

74

Derek K. Jones

However, the sequence should be designed to avoid deadtime, i.e. time when a gradient (imaging or diffusion) is not applied, i.e. the TE should be as short as possible for a given b-value. This design principle imposes a relationship between TE and b-factor, allowing the two to be optimized simultaneously, as shown in Jones et al. (1999b). What is the optimal repetition time? The introduction of diffusion-encoding gradient pulses to an imaging pulse sequence renders the image sensitive to both microscopic and macroscopic motion. Poncelet et al. (1992) and Enzmann and Pelc (1992) have demonstrated intrinsic pulsatile motion of the brain linked to the cardiac cycle, leading to spatial velocity gradients within the parenchyma. Such motion can be coherent and/or incoherent, each type of motion posing different potential problems for quantitative diffusion imaging. If motion is coherent (i.e. all the tissue within a voxel moves at the same velocity in the same direction), then if the separate images used to compute an ADC or diffusion tensor are acquired at different points in the cardiac cycle, local misregistration of tissue may occur. If the intra-voxel motion is incoherent, such that different parts of the tissue contained within a voxel move in different directions during the TE, then there will be additional attenuation of the signal – leading to overestimation of the diffusivity in the direction of the applied gradient for that voxel. Such problems can, and should, be ameliorated by gating the acquisition to the cardiac cycle (either using chest-leads or a peripheral pulse oximeter), ensuring that each diffusion-weighted image for a particular slice location is acquired at the same point in the cardiac cycle (to avoid problems of local misregistration) and also during the diastolic phase of the cardiac cycle (Chien et al., 1990; Turner et al., 1990; Conturo et al., 1995; Wieshmann et al., 1998; Skare et al., 2001; Pierpaoli et al., 2003).

have led some groups to consider alternative acquisition strategies. High T2 sensitivity As discussed earlier, the long TEs associated with DWI lead to high T2 sensitivity. This can be particularly problematic for experiments involving tissue with short T2 characteristics (e.g. muscle) since there can be very little signal left to measure at the end of the TE, but for all tissue types, increased T2 relaxation means lower SNR and therefore less reliable data. The problem of minimizing T2 relaxation can be addressed using the stimulated echo acquisition mode (STEAM) sequence (Tanner, 1970; Merboldt et al., 1985) in which the transverse magnetization is stored as longitudinal magnetization during a part of the sequence known as the mixing time (TM). During this time, T1 relaxation becomes the dominant decay mechanism, but as T1 is typically longer than T2, longer diffusion times can be established without the large T2 decay seen in conventional SE diffusion-weighted echo planar imaging (EPI). However, a major limitation of STEAM approaches is that the SNR is automatically reduced by a factor of two compared with SE type sequences. T*2 decay and low phase encode bandwidth A further problem with EPI based acquisitions is that image blurring occurs as a result of the T2* decay during the EPI readout, thereby limiting resolution. Furthermore, the bandwidth per pixel is very low in the phase encode direction with EPI readouts. Any inhomogeneities in the field (e.g. due to susceptibility induced gradients) tend to grossly distort the image along the phase encode direction. This effect becomes more pronounced at higher field strengths. Figure 4.13 shows an example of DTI data collected using a SE EPI acquisition on a 3 Tesla system. Remedies

Alternative acquisition strategies While the PGSE experiment with echo planar readout (Turner et al., 1990) is undoubtedly the most commonly used sequence, its associated limitations

Shortening the EPI train and increasing the speed of k-space trajectories remedies the problems associated with EPI. One approach for reducing the speed of k-space traversal is interleaved EPI, in which the traversal of k-space is divided into a

Fundamentals of diffusion MR imaging

(a)

(b)

Fig. 4.13 Example of susceptibility induced artifacts: (a) shows the T2-weighted image while (b) shows the FA image. These data were collected using a single-shot SE EPI acquisition sequence on a 3 Tesla machine. The huge distortion in the left–right direction (phase-encoding direction) results from susceptibility gradients present at air/tissue interfaces.

number of interleaves, resulting in faster traversal of k-space in the phase encode direction for each interleave. However, the main obstacle to robust use of interleaved EPI presents itself when the separate k-space acquisitions are to be interleaved prior to constructing the image. Since diffusion-weighted MR sensitizes the MR signal to microscopic motion, it also sensitizes the signal to any macroscopic motion, including both voluntary and involuntary physiological motion. Thus, in the presence of motion, each interleave of k-space may suffer different phase changes. When these interleaves are reconstituted into the full k-space sample and then Fourier transformed to create the image, the different phase changes will manifest as ghosting within the image. By eliminating differences in phase between each interleave, however, a ghost-free image can be obtained. Motion induced phase differences can be avoided completely by using imaging sequences that do not require phase encoding at all, such as the line-scan diffusion imaging approach suggested by Gudbjartsson et al. (1996, 1997) and Bammer et al. (2003). Alternatively, the phase differences between successive interleaves can be corrected by carefully

monitoring the phase differences and correcting them using the so-called navigated diffusionweighted sequences developed by De Crespigny et al. (1995) and Butts et al. (1996, 1997). An example of a spiral navigated interleaved EPI image is presented in Figure 4.14(b) ( cf. Butts et al., 1997 for full details). The advantages of using interleaved EPI are clear in Figure 4.15, which shows conventional T2-weighted and diffusion-weighted EPI images together with the navigated interleaved EPI T2-weighted and diffusion-weighted images from a patient with an acute stroke. The susceptibility induced distortions are markedly reduced when using the interleaved acquisition. The artifacts associated with the low bandwidth in the phase encode direction in EPI can also be reduced using newly proposed parallel imaging techniques. By acquiring separate parts of k-space using multiple coils in parallel, it is possible to reduce the time for traversing k-space in the phase-encoding direction, thereby increasing the bandwidth per voxel. Estimates of the coil sensitivities are used to correct for the aliasing that occurs as a result of the parallel acquisition (Pruessman et al., 1999). Using this

75

76

Derek K. Jones

(a)

(b)

Fig. 4.14 Images acquired using the interleaved spiral EPI acquisition of Butts et al. (1997): (a) before and (b) after navigator correction. Note the ghosting artifacts in the un-navigated image, which make the image useless. In (b), the navigated image appears to be free of ghosting artifact and lacks the usual susceptibility induced artifacts seen in regular single-shot EPI images. Figure adapted from Butts et al. (1997). © 1997. Reprinted by permission of Wiley-Liss, Inc. a subsidiary of John Wiley & Sons, Inc. The author is grateful to Dr. Kim Butts, Lucas MRS/I Center, Stanford University, Stanford, California for supplying the original of this figure.

approach also reduces the duration of the EPI readout, thereby reducing T2* decay and permitting higher resolution data to be acquired. Bammer et al. (2001, 2002a) have recently demonstrated the use of the sensitivity-encoding (SENSE) technique in combination with both DWI and DTI, an example is given in Figure 4.16. Finally we note that schemes in which the center of k-space is oversampled also enable the k-space acquisition to be fragmented and phase inconsistencies between these fragments to be corrected. One example is the use of spiral interleaved k-space trajectories (e.g. Glover and Lai, 1998). The spiral acquisition has the added advantage that the blurring due to T2* decay during the image readout is no longer constrained along one axis (as is the case with conventional interleaved EPI) and is effectively distributed over all directions. This approach has successfully been applied to mapping the ADC and diffusion tensor by Li et al. (1999) and Bammer et al. (2002b). A further approach to the problem is Periodically Rotated Overlapping ParallEL Lines with Enhanced Reconstruction (PROPELLER), which was recently proposed by Pipe et al. (2002). In this fast-spin echo (FSE) based approach, the collection of k-space is again fragmented but this time into a series of blades,

i.e. a band of lines of k-space centered about the origin. Each successive blade is rotated around the origin so as to provide complete coverage of k-space (Figure 4.17). In the process, the center of k-space is repeatedly sampled as the blades overlap, therefore allowing for the correction of phase inconsistencies between successive blades. Figure 4.18 shows some example data collected using the PROPELLER acquisition scheme. Note lack of the usual signal dropout and distortion at air/ tissue interfaces. The net result of all these approaches (interleaved EPI, parallel imaging, and PROPELLER) is increased bandwidth per voxel in the phase encode direction and therefore reduced artifacts arising from field inhomogeneities such as those induced by eddy currents and local susceptibility gradients.

Alternatives to the tensor model Limitations of the single diffusion tensor model As has been stated several times above, the estimate of diffusivity (or of the diffusion tensor) within a voxel represents the bulk-averaged diffusion properties

Fundamentals of diffusion MR imaging

(a)

(b)

(c)

(d)

Fig. 4.15 T2-weighted (a and c) and diffusion-weighted (b and d) images acquired from a patient with an acute ischemic lesion in the left posterior portion of the brain. Panels (a) and (b) were obtained using conventional single-shot EPI, while panels (c) and (d) were obtained using the spiral navigated interleaved EPI acquisition of Butts et al. (1997). Figure adapted from Butts et al. (1997). © 1997. Reprinted by permission of Wiley-Liss, Inc. a subsidiary of John Wiley & Sons, Inc. The author is grateful to Dr. Kim Butts, Lucas MRS/I Center, Stanford University, Stanford, California for supplying the original of this figure.

for that voxel. If the tissue within a voxel is homogeneous (uniform diffusivity, anisotropy and fiber orientation), then the bulk-averaged diffusion measure will reflect the underlying tissue microstructure. However, typical image resolutions used in DWI are on the order of 2.5 2.5 2.5 mm, i.e. a volume of

approximately 15 mm3. Since the diameter of an axon is on the order of 10 m, it is clear that a voxel can contain multiple fiber populations or tissues with different diffusivities. Since the mean diffusivity is relatively homogenous throughout parenchyma, the problem of voxels containing tissues with different

77

78

Derek K. Jones

Fig. 4.16 Demonstration of efficacy of parallel imaging in reducing the distortions inherent in conventional EPI. The data are collected from a 62-year-old male stroke patient with signal abnormalities in the basal ganglia and anterior portion of the left middle cerebral artery (MCA). The arrowheads point to susceptibility distortions in the frontal lobe, while the arrow points to a chemical shift artifact. The top row corresponds to conventional single-shot EPI, the second and third rows correspond to single shot SENSE acquisition with the phase-encode direction in the left–right and anterior–posterior directions respectively, while the final row shows the trace. In this case, set the phase-encoding direction to be left–right appears to produce better results than when applied in an anterior–posterior direction. Figure taken from Bammer et al. (2001). © 2001. Reprinted by permission of Wiley-Liss, Inc. a subsidiary of John Wiley & Sons, Inc. The author is grateful to Dr. Roland Bammer, Lucas MRS/I Center, Stanford University, Stanford, California for supplying the original of this figure.

Fundamentals of diffusion MR imaging

diffusivities becomes relevant at tissue/CSF interfaces, leading to partial-volume effects which include overestimation of the mean diffusivity of water in gray matter (GM) GM/CSF interfaces. These problems can be partially corrected by adopting a fluidsuppression strategy (e.g. fluid attenuated inversion recovery, FLAIR) as demonstrated by Bastin et al. (2001) and Papadakis et al. (2002). However, the use of FLAIR obviously prohibits the use of cardiac gating and so there is a risk of pulsatility artifacts corrupting the data. The problem of including multiple fiber populations within a voxel is more problematic and has many consequences for DTI. The anisotropy of the voxel-averaged tensor depends on the architectural paradigm of the tissue it contains (Pierpaoli et al., 1996). In Figure 4.19, several regions in the WM on four slices of a FA data set collected from a healthy male volunteer (age 29 years), appear to have low anisotropy. Given that these data were acquired from a healthy male volunteer with a clean bill of health, WM damage is not expected in these areas. To understand why the anisotropy appears low, it should be remembered that the diffusion tensor computed in each voxel represents the bulk-average of the diffusion properties within the voxel. Thus, if there are multiple fiber populations oriented at different angles to one another within the voxel, diffusion no longer occurs preferentially along one axis, and so the anisotropy of the voxel-average will be low. (This is another indication that anisotropy should not be considered a marker of myelination!) Now consider a voxel in which there are three fiber populations, one oriented along each of the x-, y- and z-axes. We would be hard-pressed to state in which direction diffusion is least hindered. Similarly, the estimated diffusion tensor will not appear cigar shaped, but oblate (cf. Figure 4.7(b)) or even spherical. Extracting the three individual fiber orientations from the tensor model would be impossible, since the tensor model only indicates one principal eigenvector. This is a major problem in the field of tractography where the aim is to reconstruct the pathways of WM fibers (Conturo et al., 1999; Jones et al., 1999a; Mori et al., 1999; Basser et al., 2000; Parker, 2000; Poupon et al., 2000; Tuch et al., 2000; Koch et al., 2001; Parker et al., 2002; Behrens et al.,

Fig. 4.17 Schematic illustration of how the PROPELLER acquisition samples k-space. The bold lines represent on “blade” of k-space (a band of lines that are collected in one echo train in the FSE experiment). Successive blades, rotated with respect to each other, are collected in subsequent repetition time (TRs). Note the overlap of the blades at the center of k-space, which allows for correction of phase inconsistencies between blades. Figure taken from Pipe et al. (2002). © 2002. Reprinted by permission of Wiley-Liss, Inc. a subsidiary of John Wiley & Sons, Inc. The author would like to thank James G Pipe, Ph.D., Barrow Neurological Institute, Phoenix, Arizona, for supplying the original of this figure.

2003). At points where fibers cross, twist, splay, or “kiss” within a voxel, the tensor model proves inadequate for extracting the fiber architecture. Recently, several groups have proposed methods that aim to resolve some of these problems (Tuch et al., 1999, 2002; Frank, 2001, 2002; Anderson and Ding, 2002; Alexander et al., 2002; Jansons and Alexander, 2002). For example, Frank (2001) suggested an approach to resolving the low apparent anisotropy problem by estimating the ADC in a large number of directions, and then simply taking the variance of these estimates about the mean value. This approach, however, requires data to be collected at

79

80

Derek K. Jones

(a)

(b)

(c)

Fig. 4.18 Data collected with the PROPELLER acquisition: (a) b  0 images, (b) b  1000 s mm 2 images and (c) computed ADC maps. Note the lack of distortion at tissue-air interfaces. The author would like to thank James G Pipe, Ph.D., Barrow Neurological Institute, Phoenix, Arizona for supplying this figure.

higher b-values than are typically used clinically, e.g. b ⬇ 3000 s mm 2 (Alexander et al., 2001; Frank, 2001) which results in noisier data. q-space imaging is an approach developed by Callaghan and colleagues (Callaghan et al., 1988; Callaghan, 1991), that aims to extract the displacement probability profile directly from diffusionweighted signals. The hardware requirements for performing robust and accurate q-space experiments are demanding (requiring very strong magnetic field

gradients). The practical limitations of obtaining such data on a clinical system have been discussed elsewhere (Basser, 2002). However, several groups have used the same analysis that is used for q-space imaging to analyze diffusion-weighted signals obtained from human brain on clinical scanners (Wedeen et al., 2000; Jansons and Alexander, 2002). For example, Wedeen et al. (2000) in an approach called Diffusion Spectrum Imaging, use q-space type data to infer the displacement probability profile directly.

Fundamentals of diffusion MR imaging

Fig. 4.19 Illustration of the problem of powder averaging of fiber orientation with a voxel. The figure shows four slices of a FA data set obtained from a healthy male volunteer (age 29 years). The arrows point to regions where the anisotropy appears to be very low. These regions of low apparent anisotropy indicate that the architectural paradigm of the underlying tissue is a key factor in determining the anisotropy of the voxel.

These techniques aim to resolve the problem of fiber crossing and are mostly targeted toward improving fiber tractography. However, robust application of these techniques often involves prohibitively long acquisition times. For these techniques to be useful in a clinical setting, they must be refined to reduce the time required to extract meaningful information (e.g. Tuch et al., 2003).

REFERENCES Alexander AL, Hasan KM, Lazar M, Tsuruda JS, Parker DL. 2001. Analysis of partial volume effects in diffusiontensor MRI. Magn Reson Med 45: 770–780. Alexander DC, Barker GJ, Arridge SR. 2002. Tissue structure complexity maps from high angular resolution diffusion weighted magnetic resonance measurements. In Book of Abstracts: Tenth Annual Meeting of the International Society

81

82

Derek K. Jones

for Magnetic Resonance in Medicine, ISMRM, Berkeley, CA, p. 1158. Anderson AW, Ding Z. 2002. Sub-voxel measurement of fiber orientation using high angular resolution diffusion tensor imaging. In Book of Abstracts: Tenth Annual Meeting of the International Society for Magnetic Resonance in Medicine, ISMRM, Berkeley, CA, p. 440. Bammer R, Keeling SL, Augustin M, Pruessman KP, Wolf R, Stollberger R, Hartung H-P, Fazekas F. 2001. Improved diffusion-weighted single-shot echo-planar imaging (EPI) in stroke using sensitivity encoding (SENSE). Magn Reson Med 46: 548–552. Bammer R, Auer M, Keeling SL, Augustin M, Stables LA, Prokesch RW, Stollberger R, Moseley ME, Fazekas F. 2002a. Diffusion tensor imaging using single-shot SENSE-EPI. Magn Reson Med 48: 128–136. Bammer R, Glover GH, Moseley ME. 2002b. Diffusion tensor spiral imaging. In Book of Abstracts: Tenth Annual Meeting of the International Society for Magnetic Resonance in Medicine, ISMRM, Berkeley, CA, p. 1111. Bammer R, Herneth AM, Maier SE, Butts K, Prokesch RW, Do HM, Atlas SW, Moseley ME. 2003. Line scan diffusion imaging of the spine. Am J Neuroradiol 24: 5–12. Basser PJ. 2002. Relationships between diffusion tensor and q-space MRI. Magn Reson Med 47: 392–397. Basser PJ, Pierpaoli C. 1996. Microstructural and physiological features of tissue elucidated by quantitative-diffusiontensor MRI. J Magn Reson B 111: 209–219. Basser PJ, Mattiello J, Le Bihan D. 1994a. Estimation of the effective self-diffusion tensor from the NMR spin echo. J Magn Reson B 103: 247–254. Basser PJ, Mattiello J, Le Bihan D. 1994b. MR diffusion tensor spectroscopy and imaging. Biophys J 66: 259–267. Basser PJ, Pajevic S, Pierpaoli C, Duda J, Aldroubi A. 2000. In vivo tractography using DT-MRI data. Magn Reson Med 44: 625–632. Basser PJ, Jones DK. 2002. Diffusion-tensor MRI: theory, experimental design and data analysis – a technical review. NMR Biomed 15: 456–467. Basser PJ, Le Bihan D. 1992. Fiber orientation mapping in an anisotropic medium with NMR diffusion spectroscopy. In Book of Abstracts: Eleventh Annual Meeting of the Society of Magnetic Resonance in Medicine, ISMRM, Berkeley, CA, p. 1221. Bastin ME. 2001. On the use of the FLAIR technique to improve the correction of eddy current induced artifacts in MR diffusion tensor imaging. Magn Reson Imaging 19: 937–950. Batchelor P. 2002. Optimisation of direction schemes for diffusion tensor imaging. In Proceedings of Workshop on Diffusion MRI: Biophysical Issues (What Can We Measure?), St Malo, France.

Batchelor PG, Atkinson D, Hill DLG, Calmante F, Connelly A. 2003. Anisotropic noise propagation in diffusion tensor MRI sampling schemes. Magn Reson Med 49: 1143–1151. Beaulieu C. 2002. The basis of anisotropic water diffusion in the nervous system – a technical review. NMR Biomed 15: 435–455. Beaulieu C, Allen PS. 1994a. Determinants of anisotropic water diffusion in nerves. Magn Reson Med 31: 394–400. Beaulieu C, Allen PS. 1994b. Water diffusion in the giant axon of the squid: implications for diffusion-weighted MRI of the nervous system. Magn Reson Med 32: 579–583. Beaulieu C, Allen PS. 1996. An in vitro evaluation of the effects of local magnetic-susceptibility-induced gradients on anisotropic diffusion in nerve. Magn Reson Med 36: 39–44. Behrens TEJ, Johansen-Berg H, Woolrich MW, Smith SM, Wheeler-Kingshott CAM, Boulby PA, Barker GJ, Sillery EL, Sheehan K, Ciccarelli O, Thompson AJ, Brady JM, Matthews PM. 2003. Non-invasive mapping of connections between human thalamus and cortex using diffusion imaging. Nat Neurosci 6: 750–757. Bito Y, Hirata S, Yamamoto E. 1995. Optimal gradient factors for ADC measurements. In Book of Abstracts: Third Annual Meeting of the International Society for Magnetic Resonance in Medicine, ISMRM, Berkeley, CA, p. 913. Bloch F. 1946. Nuclear induction. Phys Rev 70: 460–474. Brown R. 1828. A brief account of microscopical observations made in the months of June, July and August 1827 on the particles contained in the pollen of plants; and on the general existence of active molecules in organic and inorganic bodies. Philosoph Mag 4: 161. Burdette JH, Elster AD, Ricci PE. 1999. Acute cerebral infarction: quantification of spin-density and T2-shine-through phenomena on diffusion-weighted MR images. Radiology 212: 333–339. Butts K, de Crespigny A, Pauly JM, Moseley ME. 1996. Diffusion-weighted interleaved echo-planar imaging with a pair of orthogonal navigator echoes. Magn Reson Med 35: 763–770. Butts K, Pauly J, de Crespigny A, Moseley M. 1997. Isotropic diffusion-weighted and spiral-navigated interleaved EPI for routine imaging of acute stroke. Magn Reson Med 38: 741–749. Callaghan PT. 1991. Principles of Nuclear Magnetic Resonance Microscopy, Oxford University Press, Oxford. Callaghan PT, Eccles CD, Xia Y. 1988. NMR microscopy of dynamic displacements: k-space and q-space imaging. J Phys E: Sci Instrum 21: 820–822. Carr HY, Purcell EM. 1954. Effects of diffusion on free precession in nuclear magnetic resonance experiments. Phys Rev 94: 630–638.

Fundamentals of diffusion MR imaging

Chenevert TL, Brunberg JA, Pipe JG. 1990. Anisotropic diffusion within human white matter: demonstration with NMR techniques in vivo. Radiology 177: 401–405. Chien D, Buxton RB, Kwong KK, Brady T, Rosen BR. 1990. MR diffusion imaging of the brain. J Comput Assist Tomogr 14: 514–520. Cleveland GG, Chang DC, Hazelwood CF, Rorschach HE. 1976. Nuclear magnetic resonance measurement of skeletal muscle. Anisotropy of the diffusion coefficient of the intracellular water. Biophys J 16: 1043–1053. Conturo TE, Lori NF, Cull TS, Akbudak E, Snyder AZ, Shimony JS, McKinstry RC, Burton M, Raichle ME. 1999. Tracking neuronal fiber pathways in the living human brain. Proc Natl Acad Sci USA 96: 10422–10427. Conturo TE, McKinstry RC, Akbudak E, Robinson BH. 1995. Encoding of anisotropic diffusion with tetrahedral gradients: a general mathematical formalism and experimental results. Magn Reson Med 35: 399–412. Crank J. 1956. The Mathematics of Diffusion, Oxford University Press, Oxford. de Crespigny AJ, Marks MP, Enzmann DR, Moseley ME. 1995. Navigated diffusion imaging of normal and ischemic human brain. Magn Reson Med 33: 720–728. Crosby EC, Humphrey T, Lauer EW. 1962. Correlative Anatomy of the Nervous System, The Macmillian Company, New York. Davis TL, Wedeen VJ, Weisskoff, Rosen BR. 1993. White matter tract visualization by echo-planar MRI. In Book of Abstracts: Twelfth Annual Meeting of the Society of Magnetic Resonance in Medicine, ISMRM, Berkeley, CA, p. 289. Dejerine J. 1895. Anatomie des centres nerveux. Vol. 1, Rueff et Cie, Paris. Doran M, Hajnal J, Van Bruggen N, King MD, Young IR, Bydder GM. 1990. Normal and abnormal white matter tracts shown by MR imaging using directional diffusion weighted sequences. J Comput Assist Tomogr 14: 865–873. Douek P, Turner R, Pekar J, Patronas NJ, Le Bihan D. 1991. MR color mapping of myelin fiber orientation. J Comput Assist Tomogr 15: 923–929. Einstein A. 1905. Über die von der molekularkinetischen Theorie der Wärme gefordete Bewegung von in ruhenden Flüssigkeiten suspendierten Teilchen. Ann Physik 4: 549–590. Eis M, Hoehn-Berlage M. 1995. Correction of gradient crosstalk and optimisation of measurement parameters in diffusion MR imaging. J Magn Reson Ser B 107: 222–234. Enzmann DR, Pelc NJ. 1992. Brain motion: measurement with phase-contrast MR imaging. Radiology 185: 653–660. Frank LR. 2001. Anisotropy in high angular resolution diffusion-weighted MRI. Magn Reson Med 45: 935–939. Frank LR. 2002. Characterization of anisotropy in high angular resolution diffusion-weighted MRI. Magn Reson Med 47: 1083–1089.

Glover GH, Lai S. 1998. Self-navigated spiral fMRI: interleaved versus single-shot. Magn Reson Med 39: 361–368. Gudbjartsson H, Maier SE, Mulkern RV, Morocz IA, Patz S, Jolesz FA. 1996. Line scan diffusion imaging. Magn Reson Med 34: 509–519. Gudbjartsson H, Maier SE, Jolesz FA. 1997. Double line scan diffusion imaging. Magn Reson Med 38: 101–109. Gulani V, Webb AG, Duncan ID, Lauterbur PC. 2001. Apparent diffusion tensor measurements in myelin-deficient rat spinal cords. Magn Reson Med 45: 191–195. Hahn EL. 1950. Spin echoes. Phys Rev 80: 580–594. Hansen JR. 1971. Pulsed NMR study of water mobility in muscle and brain tissue. Biochim Biophys Acta 230: 482–486. Hong X, Dixon WT. 1992. Measuring diffusion in inhomogeneous systems in imaging mode using antisymmetric sensitizing gradients. J Magn Reson 99: 561–570. Jansons KH, Alexander DC. 2002. Spin echo attenuation to diffusion displacement density: a general inversion for measurements on a sphere. In Book of Abstracts: Tenth Annual Meeting of the International Society of Magnetic Resonance in Medicine ISMRM, Berkeley, CA, p. 1171. Jones DK. 2003. The effect of gradient sampling scheme on estimates of fiber orientation: implications for fiber tractography. In Book of abstracts: Eleventh Annual Meeting of the International Society of Magnetic Resonance in Medicine ISMRM, Berkeley, CA, p. 72. Jones DK, Horsfield MA, Simmons A. 1999b. Optimal strategies for measuring diffusion in anisotropic systems by magnetic resonance imaging. Magn Reson Med 42: 515–525. Jones DK, Simmons A, Williams SCR, Horsfield MA. 1999a. Non-invasive assessment of axonal fiber connectivity in the human brain via diffusion tensor MRI. Magn Reson Med 42: 37–41. Jones DK, Williams S, Horsfield MA. 1997. Full representation of white-matter fibre direction on one map via diffusion tensor analysis. In Book of Abstracts: Fifth Annual Meeting of the International Society of Magnetic Resonance in Medicine, ISMRM, Berkeley, CA, p. 1741. Koch M, Glauche V, Finsterbusch J, Nolte U, Frahm J, Buchel C. 2001. Estimation of anatomical connectivity from diffusion tensor data. Neuroimage 13: S176. Le Bihan D, Breton E. 1985. Imagerie de diffusion in vivo par résonance magnétique nucléaire. CR Acad Sci Paris 301: 1109–1112. Le Bihan D. et al. 1986. MR imaging of intravoxel incoherent motions: application to diffusion and perfusion in neurologic disorders. Radiology 161: 401–407. Li TQ, Takahashi AM, Hindmarsh T, Moseley ME. 1999. ADC mapping by means of a single-shot spiral MRI technique with application in acute cerebral ischemia. Magn Reson Med 41: 143–147.

83

84

Derek K. Jones

Lian J, Williams DS, Lowe IJ. 1994. Magnetic resonance imaging in the presence of background gradients and imaging of background gradients. J Magn Reson A 106: 65–74. Lythgoe MF, Busza AL, Calamante F, Sotak CH, King MD, Bingham AC, Williams SR, Gadian DG. 1997. Effects of diffusion anisotropy on lesion delineation in a rat model of cerebral ischemia. Magn Reson Med 38: 662–668. Mattiello J, Basser PJ, Le Bihan D. 1997. The b matrix in diffusion tensor echo-planar imaging. Magn Reson Med 37(2): 292–300. Merboldt KD, Hanicke W, Frahm J. 1985. Self-diffusion NMR imaging using stimulated echoes. J Magn Reson 64: 479–486. Mori S, Crain BJ, Chacko VP, van Zijl PC. 1999. Three dimensional tracking of axonal projections in the brain by magnetic resonance imaging. Ann Neurol 45: 265–269. Moseley ME, Cohen Y, Mintorovitch J, Chileuitt L, Shimizu H, Kucharczyk J, Wendland MF, Weinstein PR. 1990a. Early detection of regional brain ischemia in cats: comparison of diffusion- and T2-weighted MRI and spectroscopy. Magn Reson Med 14: 330–346. Moseley ME, Cohen Y, Kucharczyk J. 1990b. Diffusion weighted MR imaging of anisotropic water diffusion in cat central nervous system. Radiology 187: 439–446. Nakada T, Matsuwaza H, Kwee IL. 1994. Magnetic resonance axonography of the rat spinal-cord. Neuroreport 5: 2053–2056. Norris DG, Niendorf T, Hoehn Berlage M, Kohno K, Schneider EJ, Hainz P, Hropot M, Leibfritz D. 1994. Incidence of apparent restricted diffusion in 3 different models of cerebral infarction. Magn Reson Imaging 12: 1175–1182. Pajevic S, Pierpaoli C. 1999. Color schemes to represent the orientation of anisotropic tissues from diffusion tensor data: application to white matter fiber tract mapping in the human brain. Magn Reson Med 43: 526–540. (Erratum appears in Magn Reson Med 43: 921.) Papadakis NG, Martin KM, Mustafa MH, Wilkinson ID, Griffiths PD, Huang CLH, Woodruff PWR. 2002. Study of the effect of CSF suppression on white matter diffusion anisotropy mapping of healthy human brain. Magn Reson Med 48: 394–398. Papadakis NG, Xing D, Houston GC, Smith JM, Smith MI, James MF, Parsons AA, Huang CLH, Hall LD, Carpenter TA. 1999. A study of rotationally invariant and symmetric indices of diffusion anisotropy. Magn Reson Imaging 17: 881–892. Parker GJM. 2000. Tracing fiber tracts using fast marching. In Book of Abstracts: Eighth Annual Meeting of the International Society for Magnetic Resonance in Medicine, ISMRM, Berkeley, CA, p. 85. Parker GJM, Barker GJ, Buckley DL. 2002. A probabilistic index of connectivity (PICo) determined using a Monte Carlo approach to streamlines. ISMRM Workshop Diffusion MRI: Biophysical Issues, St Malo, France, March.

Pierpaoli C. 1997. Oh no! One more method for color mapping of fiber tract direction using diffusion MR imaging data. In Book of Abstracts: Fifth Annual Meeting of the International Society of Magnetic Resonance in Medicine, ISMRM, Berkeley, CA, p. 1743. Pierpaoli C, Basser PJ. 1996. Toward a quantitative assessment of diffusion anisotropy. Magn Reson Med. 36: 893–906. Pierpaoli C, Jezzard P, Basser PJ, Barnett AS. 1996. Diffusion tensor MR imaging of the human brain. Radiology 201: 637–648. Pierpaoli C, Marenco S, Rohde G, Jones DK, Barnett AS. 2003. Analyzing the contribution of cardiac pulsation to the variability of quantities derived from the diffusion tensor. In Book of Abstracts: Eleventh Annual Meeting of the International Society of Magnetic Resonance in Medicine , ISMRM, Berkeley, CA, p. 70. Pipe JG, Farthing VG, Forbes KP. 2002. Multishot diffusionweighted FSE using PROPELLER MRI. Magn Reson Med 47: 42–52. Poncelet BP, Wedeen VJ, Weisskoff RM, Cohen MS. 1992. Brain parenchyma motion: measurement with cine echo-planar MR imaging. Radiology 185: 645–651. Poupon C, Clark CA, Frouin V, Regis J, Bloch I, Le Bihan D, Mangin J. 2000. Regularization of diffusion-based direction maps for the tracking of brain white matter fasciculi. Neuroimage 12: 184–195. Prayer D, Roberts T, Barkovich AJ, Prayer L, Kucharczyk J, Moseley M, Arieff A. 1997. Diffusion-weighted MRI of myelination in the rat brain following treatment with gonadal hormones. Neuroradiology 39: 320–325. Provenzale JM, Sorenson AG. 1999. Diffusion-weighted MR imaging in acute stroke: theoretic considerations and clinical applications. Am J Roentgenol 173: 1459–1467. Pruessmann KP, Weiger M, Scheidegger MB, Boesiger P. 1999. SENSE: sensitivity encoding for fast MRI. Magn Reson Med 42: 952–962. Shimony JS, McKinstry RC, Akbudak E, Aronovitz JA, Snyder AZ, Lori NF, Cull TS, Conturo TE. 1999. Quantitative diffusiontenser anisotropy brain MR imaging: normative human data and anatomic analysis. Radiology 212: 770–784. Skare S, Andersson JLR. 2001. On the effects of gating in diffusion imaging of the brain using single shot EPI. Magn Reson Imaging 19: 1125–1128. Skare S, Hedehus M, Moseley ME, Li TQ. 2000. Condition number as a measure of noise performance of diffusion tensor data acquisition schemes with MRI. J Magn Reson 147: 340–352. Stejskal EO, Tanner JE. 1965. Spin diffusion measurements: spin echoes in the presence of a time-dependent field gradient. J Chem Phys 1965; 42: 288–292. Tanner JE. 1970. Use of the stimulated-echo in NMR diffusion studies. J Chem Physiol 52: 2523–2526.

Fundamentals of diffusion MR imaging

Tanner JE. 1978. Transient diffusion in a system partitioned by permeable barriers. Application to NMR measurements with a pulsed field gradient. J Chem Physiol 69: 1748–1754. Taylor DG, Bushell MC. 1985. The spatial mapping of translational diffusion coefficients by the NMR imaging technique. Phys Med Biol 42: 288–292. Thomsen C, Henriksen O, Ring P. 1987. In vivo measurement of water self diffusion in the human brain by magnetic resonance imaging. Acta Radiol 28: 353–361. Torrey HC. 1956. Bloch equations with diffusion terms. Phys Rev 104: 563–565. Tuch DS, Weisskoff RM, Belliveau JW, Wedeen VJ. 1999. High angular resolution diffusion imaging of the human brain. In Proc 7th Annual Meeting, ISMRM, Philadelphia, p. 321. Tuch DS, Belliveau JW, Wedeen VJ. 2000. A path integral approach to white matter tracotgraphy. In Book of Abstracts: Eighth Annual Meeting of the International Society for Magnetic Resonance in Medicine, ISMRM, Berkeley, CA, p. 791. Tuch DS, Reese TG, Wiegell MR, Makris N, Belliveau JW, Wedeen VJ. 2002. High angular resolution diffusion imaging reveals intravoxel white matter fiber heterogeneity. Magn Reson Med 48: 577–582. Tuch DS, Reese TG, Wiegell MR, Wedeen VJ. 2003. Q-ball imaging. In Book of Abstracts: Eleventh Annual Meeting of the

International Society for Magnetic Resonance in Medicine, ISMRM, Berkeley, CA, p. 63. Turner R, Le Bihan D, Maier J, Vavrek R, Hedges LK, Pekar J. 1990. Echo-planar imaging of intravoxel incoherent motions. Radiology 177: 407–414. Wedeen VJ, Reese TG, Tuch DS, Weigel MR, Dou J-G, Weisskoff RM, Chesler D. 2000. Mapping fiber orientation spectra in cerebral white matter with Fourier transform diffusion MRI. In Book of Abstracts: Eighth Annual Meeting of the International Society for Magnetic Resonance in Medicine, ISMRM, Berkeley, CA, p. 82. Wesbey GE, Moseley ME, Ehman RI. 1984. Translational molecular self-diffusion in magnetic resonance imaging: effects and applications. Invest Radiol 19: 491–498. Wieshmann UC, Symms MR, Franconi F, Clark CA, Barker GJ, Shorvon SD. 1998. The variability and accuracy of the apparent diffusion coefficient in diffusion weighted EPI. In Book of Abstracts: Sixth Annual Meeting of the International Society of Magnetic Resonance in Medicine, ISMRM, Berkeley, CA, p. 1748. Wimberger DM, Roberts TP, Barkovich AJ, Prayer LM, Moseley ME, Kucharczyk J. 1995. Identification of “premyelination” by diffusion-weighted MRI. J Comput Assist Tomogr 19: 28–33.

85

5

MR tractography using diffusion tensor MR imaging Sasumu Mori1 and Peter van Zijl2 1

Department of Radiology and Radiological Science, Johns Hopkins University School of Medicine, Baltimore, USA Kennedy Krieger Institute, F.M. Kirby Research Center for Functional Brain Imaging, Baltimore, USA

2

Introduction Key points • Water diffusion within brain tissue is anisotropic, with the greatest anisotropy in the major white matter (WM) axonal fiber bundles. • The directional variation in diffusion can be described by a 3 3 tensor; indices of average diffusion (apparent diffusion coefficient) and anisotropy (for instance, fractional anisotropy) can be calculated from the diffusion tensor. Eigenvalues and eigenvectors describe the magnitude and directions of the three principle axes of the diffusion ellipsoid. • Algorithms have been developed to reconstruct three-dimensional (3D) visualizations of the major axonal fiber bundles in the brain based on the eigenvectors and eigen values. • The most robust methods for fiber-tracking involve continuous number field line propagation at the sub-pixel level with exhaustive searches such multi-pixel “seed points”. • Fiber-track visualizations may vary due to different reconstruction algorithms and stopping criteria. • High-resolution 3D diffusion tensor imaging (DTI) data with good signal-to-noise and an absence of motion artifacts is required for successful fiber-tracking. • DTI can be used to segment WM from gray matter and cerebrospinal fluid (CSF); it can also be used to segment different fiber bundles within WM. 86

Experimental evidence has shown that water diffusion is anisotropic in organized tissues such as muscles (Tanner, 1979; Scollan et al., 1998) or brain white matter (WM) (Moseley et al., 1990). In the last decade, the quantitative description of this anisotropy with diffusion tensor imaging (DTI) has become well established in the research environment and its first applications in the clinic are now being reported (Basser et al., 1994a; Basser and Jones, 2002). For example, DTI is presently being explored as a research tool to study brain development (Neil et al., 1998; Neil et al., in press; Mori et al., 2001), multiple sclerosis (MS) (Tievsky et al., 1999; Clark et al., 2000), amyotrophic lateral sclerosis (ALS) (Ellis et al., 1999), stroke (Mukherjee et al., 2000a, 2000b; Sotak, 2002), schizophrenia (Lim et al., 1999; Horsfield and Jones, 2002), and reading disability (Klingberg et al., 2000). Based on fiber orientation information obtained from DTI, it has been also shown that in vivo fiber-tracking is possible (Basser, 1998; Mori et al., 1998, 1999, 2000; Conturo et al., 1999; Jones et al., 1999; Xue et al., 1999; Basser et al., 2000; Lazar et al., 2000; Parker, 2000; Poupon et al., 2000; Tuch et al., 2000; Wedeen et al., 2000; Werring et al., 2000; Stieltjes et al., 2001; Lazar et al., 2003). In order to better utilize this promising technology, it is important to understand the basis of the anisotropy contrast in DTI and the limitations imposed by using a macroscopic technique to visualize microscopic axonal structures. In this chapter, basic principles of the DTI-based tract reconstruction and its capability and limitations will be discussed.

MR tractography using diffusion tensor MR imaging

Isotropic and anisotropic diffusion

(a)

It has been known that MR imaging (MRI) can measure molecular diffusion; as outlined in the previous chapter. One of the unique and important features of the diffusion measurement by MR is that it always detects molecular movement along one predetermined axis (Figure 5.1), which is determined by the resultant orientation of applied magnetic field gradients. Every MR machine is equipped with three orthogonal, namely x-, y-, and z-gradients. By combining these three-axis gradients, diffusion along any arbitrary axis can be measured. For example, if equal strength of x- and y-gradient is applied simultaneously, diffusion along 45° from x and y axes can be measured. The orientation of the diffusion measurement is not important if we are interested in freely diffusing water because the results are independent of the measurement orientations. Such orientationindependent diffusion is called “isotropic” diffusion (e.g. the lower compartment of Figure 5.1(a)). In biological systems, diffusion process can be much more complicated because water molecules see many obstacles and barriers during the diffusion process. If the biological system has ordered alignment, such as muscle or axonal fibers, the extent of water diffusion may be different depending on the measurement orientation, which is called “anisotropic” diffusion (e.g. the upper compartment of Figure 5.1(a)). If the sample has anisotropic diffusion, results of MR diffusion measurements depend on which gradient axis is used. Figure 5.2 shows an example of diffusion measurements in a rat brain, in which it can be clearly seen that the apparent diffusion constants (ADC) change considerably if different gradient orientation is used. For example, a pixel indicated by pink arrows has low diffusion constants with x- and y-gradient but has a high diffusion constant when it is measured by z-gradient. In previous studies, there are ample evidences that water tends to diffuse preferably along fibers (Stejskal, 1965; Moseley et al., 1990; Douek et al., 1991; Basser et al., 1994b; van Gelderen et al., 1994; Scollan et al., 1998) probably because it sees fewer obstacles. From these rat brain studies, we can immediately conclude that the pixel indicated by the pink arrow contains axonal fibers

Ordered structure

87

(b) Diffusion measurement orientation

Random structure

Tensor calculation (c) Orientation

Anisotropy map

Orientation map

(d)

Fig. 5.1 (a) A schematic diagram of some example tissue structures. The upper region has an ordered structure due to fibers running along the curved arrow. The lower region, while the shape of the structure is the same, has random fiber structure. (b) Results of diffusion measurement along three different axes are indicated by pointed fingers. The ADC is faster (bright areas) when the fiber orientation coincides with the measurement orientation and slower (dark areas) when it is perpendicular to each other. This results in different diffusion constants that depend on the measurement orientation in the upper region while the lower region is insensitive to measurement orientation. (c) Anisotropy and color-coded orientation maps calculated from the measurement results in (b). Anisotropy (diffusion directionality) of the upper region is high because the diffusion constant of this region depends on measurement orientation. When anisotropy is high, the fiber angle can be calculated based on the information in (b), which can be represented by vectors or by color. In this 2D example, regions with fibers running horizontally are green and those running vertically are red. Transition areas become yellow, which is the mixture of green and red. (d) Although a vector was used to indicate the fiber and diffusion orientation in a–c, the actual water diffusion is a 3D process in which water molecules diffuse to all orientations, resulting is 3D ellipsoid shape (d) for probability of the displacement. In order to fully describe this 3D diffusion process, a more comprehensive mathematical description, a 3 3 tensor, is needed.

88

S. Mori and P. van Zijl

X

Y

Z

Fig. 5.2 ADC maps of a fixed rat brain measured by x-, y-, and z-gradient. The contrasts heavily depend on the measurement orientations (indicated by green arrows), suggesting water diffusion inside the brain is anisotropic (purple arrows).

that are running perpendicular to x and y axes and parallel to z. Obviously, the MR diffusion measurement has capability to provide information on fiber architectures within the sample.

How anisotropy is measured The results shown in Figure 5.2 clearly suggest that ADC measurement along multiple orientations contains important information on the axonal organization of the brain. The question is how we can fully characterize the anisotropic diffusion and, subsequently, fiber architecture. When diffusion is isotropic, the probability to find a water molecule after a certain amount of time is spherical, which can be described by one parameter (diameter). If water is confined in a homogeneously aligned system, we can assume that the diffusion process leads to elliptic shape of the probability with the longest axis aligned to the orientation of the fibers (Basser et al., 1994a, 1994b; Basser and Pierpaoli, 1996). Then, our task is to define the shape of the ellipsoid (called diffusion ellipsoid) and its orientation. The most intuitive way is to measure the ADCs along a very large number of orientations, from which a well-defined shape can be reconstructed. This direct measurement of the diffusion ellipsoid shape is actually becoming popular recently (Wedeen et al., 2000; Wiegell et al., 2000; Frank, 2001). An alternative way is to measure the ADCs along the smaller number of orientation, from which the shape of the ellipsoid is calculated (Basser et al., 1994a, 1994b; Basser and Pierpaoli, 1996). For this calculation, we need an

Fig. 5.3 An example of three-axes diffusion measurement in an anisotropic system (left) and isotropic system (right), which leads to the same results.

aide of mathematical procedure called “tensor” calculation and, thus, this process is called DTI (Figure 5.1). First of all, it is very important to realize that the anisotropic diffusion (or the diffusion ellipsoid) in tissue cannot be characterized by measurements along three orthogonal axes, x, y, and z. This is illustrated in Figure 5.3, where ADC measurements along three axes lead to the same results (the lengths of the ellipsoid and sphere along x, y, and z axes are the same) for two systems that have markedly different diffusion properties. Obviously, unlike a vector, diffusion cannot be described exactly by determining its dimensions along three orthogonal axes. In order to fully characterize the diffusion ellipsoid, we need at least six parameters as shown in Figure 5.4, namely 1, 2, 3, v1, v2, and v3. The lengths of its three principal axes (1, 2, and 3) define its shape and three vectors (v1, v2, and v3) define the orientations. In order to keep track of

MR tractography using diffusion tensor MR imaging

Process of DTI experiments and tensor calculation

λ1 v1

λ2

In diffusion measurements, the signal attenuation is described by:

λ3

v3 v2

Three numbers to define the shape

Three vectors to define the orientation

Fig. 5.4 Six parameters needed to define an ellipsoid.

λ2

y x

Shape and orientation

Fig. 5.5 Diffusion ellipsoid can be fully characterized from diffusion measurements along six independent axes.

these six parameters, we need a 3 3 tensor, called  diffusion tensor, D, which relates to the six parameters by a process called “diagonalization”: ⎡Dxx ⎢ D  ⎢Dyx ⎣⎢Dxz

Dxy Dyy Dyz

Dxz ⎤ ⎥ Dyz ⎥ ⎯diagonalization ⎯⎯⎯⎯ ⎯→ 1, 2 , 3 , v1, v 2 , v 3 Dxy ⎦⎥

where  is the gyromagnetic ratio; , the gradient pulse length; , the separation of a pair of gradients; and D, the diffusion constant. This equation is correct only for isotropic diffusion or for diffusion measurement along one axis. For a more complete expression in anisotropic media, we have to use the equation:

Again, if we solve this equation for the experiment with a pair of square-shape gradients, we obtain:

λ3

Measure diffusion along various directions (6)

(5.2)

t t' t' ⎡S⎤ ln ⎢ ⎥  − ∫  2 ⎡⎢ ∫ G(t")dt"⎤⎥ • D • ⎡⎢ ∫ G(t")dt"⎤⎥dt ' 0 0 0 S ⎦ ⎦ ⎣ ⎣ ⎣ 0⎦ (5.3)

λ1 Tensor calculation

S − 2G 22 ( −/3)D e  e − bd S0

(5.1)

 This diffusion tensor, D, is a symmetric tensor, which means Dij  Dji, and, thus, there are six independent parameters, which makes sense because it intrinsically contains the six parameters of the diffusion ellipsoid.  In order to determine these six elements of D, not surprisingly, we need to measure at least six diffusion constants along six independent axes (Figure 5.5). In the following section, the actual experimental  process to determine the D will be described.

S  e− S0

T

bD b

(5.4)

– – – – where b is 2G22 ( /3). Here G (and also b ) is a vector because it contains information of not only gradient strength but also the orientation. In actual experiments, we want to determine the  – six parameters in the D while parameters, G, , , and  are known parameters, and S0 and S are the experimental results. Please note that this equation  has total seven unknowns (six in D and S0) and we need at least seven experimental results (S) with – different G to solve the tensor.

Two-dimensional visualization There are many ways to characterize the diffusion anisotropy (Basser et al., 1994a, 1994b; Pierpaoli and Basser, 1996). The simplest and most intuitive method is to calculate the ratio of the length of the longest and shortest axes. This method, however, has several unwanted properties. For example, the range of its value is 1 (sphere)–infinity, which is difficult to visualize, and the length of the shortest axis (thus the ratio) is very susceptible to noise. It is preferable

89

90

S. Mori and P. van Zijl

(a)

(b)

(c)

Fig. 5.6 Comparison of T1-weighted (a), FA map (b) and color-coded orientation (c) maps. Images were acquired using a 1.5 T machine. Colors in (c) represent orientations of fibers; red: right–left, green: anterior–posterior, and blue: superior–inferior.

to use a parameter that ranges 0 (isotropy) to 1 (anisotropy) for the visualization purpose. The most widely used normalized parameters are

FA 

( 1 − 2 )2 + ( 2 − 3 )2 + ( 1 − 2 )2

RA 

( 1 − 2 )2 + ( 2 − 3 )2 + ( 1 − 2 )2

VR 

(

2 12 + 22 + 23

1 + 2 + 3

)

(5.5)

12 3

((  +  +  ) / 3)

3

1

2

3

here fractional anisotropy (FA), relative anisotropy (RA), and volume ratio (VR). The information provided by these parameters is essentially the same. They all indicate how elongated the diffusion ellipsoid is, but the contrast they provide is not identical (Ulug and van Zijl, 1999). Among these parameters, the FA is most widely used. In the example in Figure 5.6, it can be seen that WM has high FA (Figure 5.6(b)) values, which makes sense because it consists of densely packed axonal fibers. Segmentation of the WM and gray matter (GM) can also be achieved by conventional T1- and T2-weighted images (Figure 5.6(a)). However, detailed inspection of Figure 5.6 clearly

shows that the GM–WM contrasts of these two types of images are not identical. Anisotropy, which also shows a very high contrast between WM and GM, is based on a completely different contrasting mechanism; the directionality of water diffusion given by axonal fibers. The exact mechanism underlying the anisotropy map is not completely understood (Beaulieu and Allen, 1994; Henkelman et al., 1994; Stanisz et al., 1997; Beaulieu, 2002). What is known is that the anisotropy drastically increases during early development (Sakuma et al., 1991), and its time course is different from other conventional MRI parameters, such as T1/T2 relaxation properties. The changes during the development may suggest involvement of the myelin sheath (Sakuma et al., 1991). However, large anisotropy has been reported in axonal fibers without myelin sheaths (Beaulieu and Allen, 1994), implying that the increased anisotropy during development may be due to increased fiber density. Our recent studies on co-registered FA and T2 map revealed that these parameters are not always correlated (Stieltjes et al., 2001). In any case, it is apparent that the anisotropy provides a new contrasting mechanism that was formally inaccessible and, thus, it is worth pursuing its clinical possibilities as a new diagnostic tool.

MR tractography using diffusion tensor MR imaging

mcp ml

cst

cst

cp sn

ml scp icp Fig. 5.7 Examples of color maps and identification of prominent WM tracts (axial images at the level of the brain stem). ml, medial lemniscus; icp, inferior cerebellar peduncle; mcp, middle cerebellar peduncle; scp, superior cerebellar peduncle; cp, cerebral peduncle; sn, substatia nigra. The color scheme is the same as Figure 5.6(c).

Visualization of orientation of the ellipsoid Another parameter that can be obtained from DTI is the orientation of diffusion ellipsoids (Figure 5.4). The most intuitive way to show orientation is a vector presentation, in which small lines (vectors) indicate the orientations of the longest axis of diffusion ellipsoids. However, unless a small region is magnified, the vector orientation is often difficult to see. To overcome this problem, a color-coded scheme was proposed (Douek et al., 1991; Nakada and Matsuzawa, 1995; Pajevic and Pierpaoli, 1999), an example of which is shown in Figure 5.6(c). In the color map, three orthogonal axes (e.g. right–left, superior–inferior, and anterior–posterior) are assigned to three principal colors (red, green, and blue). If a fiber is running 45° from red and blue axes, it is assigned magenta, which is mixture of red and blue. When compared to conventional images such as T1- and T2-weighted images (Figure 5.6(a–c)), it can be clearly seen that the DTI-based color map carries detailed information about the anatomy of WM. Using the color maps, some prominent WM tracts can be immediately identified (Makris et al., 1997; Virta et al., 1999; Stieltjes et al., 2001; Mori et al., 2002b). To further illustrate this point, an atlas type presentation is shown in Figure 5.7.

Fiber reconstruction techniques Assuming that the orientation of the largest component of the diagonalized diffusion tensor represents the orientation of dominant axonal tracts, DTI can provide a 3D vector field, in which each vector presents the fiber orientation. Currently, there are several different approaches to reconstruct WM tracts, which can be roughly divided into two

types. Techniques classified in the first category are based on line-propagation algorithms that use local tensor information for each step of the propagation (Conturo et al., 1999; Jones et al., 1999; Mori et al., 1999; Xue et al., 1999; Basser et al., 2000; Lazar et al., 2000; Poupon et al., 2000; Mori and Van Zijl, 2002; Lori et al., 2002). The main differences among techniques in this class stem from the way information from neighboring pixels is incorporated to define smooth trajectories or to minimize noise contributions (Tuch et al., 2000; Parker et al., 2002). The second type of approach is based on global energy minimization to find the energetically most favorable path between two predetermined pixels. In this chapter, we will focus on the former approach, the line-propagation model.

Line-propagation approaches The most intuitive way to reconstruct a 3D trajectory from a 3D vector field is to propagate a line from a seed point by following the local vector orientation. However, if a line is propagated simply by connecting pixels, which are discrete entities, the vector information contained at each pixel may not be fully reflected in the propagation. In the simple example illustrated in Figure 5.8(a), axonal tracts are running along 30° from the vertical line. When applying the discrete “pixel connection” approach, a judgment has to be made about which pixel should be connected. No matter what the judgment is, it should be clear that this simple pixel connection scheme cannot represent the real tract even in such a simple case. The simplest way to convert the discrete voxel information into a continuous tracking line is to linearly propagate “a line”, in a continuous number field (Mori

91

92

S. Mori and P. van Zijl

(a)

(a)

(b)

(b)

3 2 2 1 1 0 0

1

2

0

1

2

3

Fig. 5.8 Schematic diagram of the linear line-propagation approach. Double-headed arrows indicate fiber orientations at each pixel. Tracking is initiated from the center pixel. In the discrete number field (a), the coordinate of the seed pixel is {1, 1}. If it is judged that the vector is pointing to {1, 2} and {1, 0}, shaded pixels are connected. In the continuous number field (b), the seed point is {1.50, 1.50} and a line, instead of a series of pixels, is propagated.

et al., 1999; Xue et al., 1999). This conversion from the discrete to continuous number field is shown in Figure 5.8(b). In this example, the seed point is {1.50, 1.50} and a line propagates from this point following the vector orientation of the pixel with discrete coordinate {1, 1}. This line exits the pixel (discrete coordinate {1, 1}) to the next pixel (discrete coordinate {1, 2}) at the location {1.79, 2.00} in the continuous coordinate. By repeating this process, it is easy to see that the line can follow the actual tract (or pixels can be connected) more precisely. This linear-propagation approach, which was dubbed fiber assignment by continuous tracking (FACT), was used for the first successful tract reconstruction, which was accomplished for a fixed rat brain and showed good agreement with histological knowledge of brain fiber orientations and locations (Mori et al., 1999; Xue et al., 1999).

Termination criteria Line propagation must be terminated at some point (Figure 5.9). The most intuitive termination criterion is the extent of anisotropy. In a low anisotropy region, such as GM, there may not be a coherent tract orientation within a pixel and the orientation of the largest principal axis is more sensitive to noise

Fig. 5.9 Two criteria for propagation termination. Propagation is terminated when it enters a low anisotropy area (a) or the angle between connected pixels is large (b).

errors (for isotropic diffusion, the anisotropy information is dominated by noise and becomes purely random). The FA of the GM is typically in the range 0.1–0.2. Therefore, one simple termination approach is to set the threshold for tracking termination at 0.2. Another important criterion is the angle change between pixels. For the linear line-propagation model, large errors occur if the angle transition is large. Even for the interpolation approach, it should be noted that the diffusion tensor calculation assumes that there is no consistent curvature of axonal tracts within a voxel. The presence of curvature violates the assumption that the diffusion process along any arbitrary axis is Gaussian, thereby invalidating the routine tensor calculation. It is preferable, therefore, to set a threshold that prohibits a sharp turn during line propagation. The significance of this angletransition threshold depends on the particular trajectories of tracts of interest and the image resolution. An image resolution of 1–3 mm, for example, is high enough to smoothly reconstruct the curvature of trajectories of major tracts in the brainstem (Stieltjes et al., 2001) and the cortico-cortical association fibers connecting the functional regions of the brain (Mori et al., 2002b). Under such favorable conditions, the angles between connected vectors are small and the termination criteria are dominated by the magnitude of the FA. However, for smaller tracts in environments that are structurally highly convoluted, such as sub-cortical U-fibers, the same resolution may be too coarse to smoothly represent the trajectories and, thus, analysis for the errors in tensor calculations and fiber-tracking due to the curvature becomes

MR tractography using diffusion tensor MR imaging

(a)

(b)

(c)

Fig. 5.10. Schematic diagram of the difference between the single tracking and exhaustive search approaches. Suppose (a) represents the shape of a WM tract of interest with an anatomical landmark indicated by a white circle. If tracking is initiated from the landmark, there are four possibilities for the results, each representing one branch of the tract (b). This is because a propagation result from one pixel can delineate only one line. Conversely, the line propagation can be initiated from all pixels (c) and all propagation results that penetrate the anatomical landmark are searched, which leads to more comprehensive delineation of the tract of interest. The right column compares the tracking results for these two approaches for the genu of the corpus callosum.

more important. While at this moment there are no comprehensive simulation studies in this regard, some preliminary data on the effect of the turning radius on the tracking results in particular examples can be found in recent reports (Lori et al., 1999; Lazar and Alexander, 2001).

Branching WM tracts often have extensive branching, which renders tracking computationally complex. For example, bifurcation of a line during propagation, is already a mathematically involved issue. From a programming point of view, this problem can be much more easily handled by merging two lines rather than splitting a line into two, for instance by using the brute force approach shown in figure 5.10 (Conturo et al., 1999; Stieltjes et al., 2001). In this approach,

tracking is initiated from all pixels within the brain and tracking results that penetrate the pixel of interest (POI) are kept. In other words, instead of using the POI as a seed pixel, all pixels in the brain are used as seed pixels. When using the linear-propagation model for a data size of 256 256 60, this exhaustive search takes about 15–30 minutes on a 900 MHz Pentium III processor.

Tract editing using multiple-region of interest Results of the line-propagation techniques depend heavily on the initial placement of reference POI. Suppose one is interested in the optic radiation, there are many choices to place a reference POI or a group of pixels (a region of interest, ROI). For instance, one can use the WM close to the lateral geniculate nucleus (LGN), close to the visual cortex (VC), or anywhere in

93

94

S. Mori and P. van Zijl

(a)

(b)

Fig. 5.11. Examples of tract editing by a multi-ROI approach. In (a), only one ROI is defined at the cerebral peduncle, which leads to the combined reconstruction of corticopontine and corticospinal tracts. By placing a second ROI at the pyramidal tract in caudal pons, only corticospinal tract was selected.

between the two areas to reconstruct the optic radiation. A powerful technique that can effectively search all seed pixels potentially containing the tracts of interest is the so-called multiple-ROI or tracking– editing technique (Conturo et al., 1999; Holodny et al., 2001; Stieltjes et al., 2001; Mori et al., 2002b). In this approach, relatively large reference ROIs are drawn that contain the WM close to the target GM. For example, the posterior thalamic radiation that contains the optic radiation can be defined by one ROI at the pulvinar and another at the occipital lobe. Then, an exhaustive search, such as described in Section Branching, is performed to identify all tracts that penetrate both ROIs. In this way, the reconstruction results for specific WM tracts become less dependent on the locations of the initial POI or ROI. This technique is knowledge-based and existing gross anatomical knowledge of the tract trajectory is required. It also does not allow the elucidation of branching patterns in between the multiple-ROIs.

Limitations and solutions The techniques discussed in the previous chapter are all based on the principle that a clear principal axis can be defined inside a MRI voxel, that this voxel

occupies a single tissue type, and that the vector can be connected to a neighboring voxel. In practice, voxels are more likely to consist of contributions from multiple tissues (different WM tracts, some CSF and GM, etc.) and signal to noise may be limited. In addition, there may not be a single predominant direction of water diffusion. Therefore, results obtained from DTI are to be considered as approximation of underlying biology. In Figure 5.11, an example was shown for multiROI tract editing. This approach imposes a significant constraint in the tract reconstruction, thereby effectively reducing the likelihood of the occurrence of erroneous results. For example, if only one instead of two ROIs were to be used, the result would be more likely to contain not only the tract of interest but also many other tracts. Some may be erroneous due to noise and/or partial volume effect (PVE), and some may be real trajectories of tracts that share the same ROI. These unwanted contributions can be effectively reduced by placing a second, or even third, reference ROI if the general trajectory of the tract of interest is known. The reason for using knowledgebased approaches is that once the tracking deviates from the real trajectory due to noise or PVE, it is highly unlikely that the real trajectory can be returned to by chance. A drawback of this approach

MR tractography using diffusion tensor MR imaging

is that it can, in many cases, be applied only to anatomically well-documented tracts, imposing limitations on the discovery of new tracts. The approach, however, has a significant advantage in that the location of many tracts can be identified in living human beings non-invasively (Conturo et al., 1999; Werring et al., 2000; Stieltjes et al., 2001; Mori et al., 2002b). In addition, in some cases even tracts that have large deviations, e.g. due to the presence of tissue deformations, can still be reconstructed, as was recently demonstrated for a patient in which the cortical spinal tract was displaced by tumors in the brain stem (Stieltjes et al., 2001) and cerebrum (Mori et al., 2002a) as well as for patients with several developmental abnormalities (Albayram et al., 2002; Hoon et al., 2002). There is no doubt that validation is of central importance for the development of tractography. For this purpose, we first have to evaluate what tractography provides us with and what the gold standard is to validate the results. Tractography can provide macroscopic neuroanatomical information of WM structure. Specifically, it can parcel the WM into fiber structures that contain bundles of axonal tracts that are running largely in the same orientation. Given the current resolution of DTI, on the order of 1–5 mm per dimension, it is presently not possible to resolve WM tracts into individual axons whose diameter is typically less than 10 m. There is accumulated knowledge about WM anatomy based on slice-by-slice examination using histology (Crosby et al., 1962; Carpenter, 1976; Nieuwenhuys et al., 1983). Using a proper preparation, the tract structure can be appreciated and, for some fibers, the trajectory can be visually followed over many slices. More direct information about axonal connectivity can be obtained from animal lesion studies, in which degenerating axons with specific stains are visualized after placing lesions. The most elaborated information is obtained using tracttracing methods based on chemical tracers, in which chemicals such as radioactively labeled amino acids are injected and their destinations confirmed histological analysis. Obviously, these tract-tracing techniques cannot be applied to human beings, where, most information has come from postmortem data on stroke patients. For connectivity studies, the

chemical tracer techniques are considered one of the gold standards. Although it is in principle possible to compare the results of this technique with the DTI-based tractography in animal models, there are several difficulties in such a validation approach. First, the chemical tracer techniques reveal connectivity at the cellular level. The axon of interest may merge into a WM tract and again leave it from some points. The result of chemical tracing therefore represents only a tiny portion of axons in a WM tract and, thus, it would not be surprising if the two results would not match. Second, the real advantage of the DTI-based tractography is its ability to quickly characterize the macroscopic WM structures. It is virtually impossible to generate similar datasets (there are 1011 neurons inside a brain) using the single-cell level chemical tracer techniques. Considering these factors, one possible way to validate the results is to observe only the core of major WM tracts using the tractography and compare them with anatomical knowledge, because trajectories and locations of the body of these tracts are fairly well known. Once the tracking data leaves the core and approaches the target GM regions, we do not have information to validate the results, especially for human beings. Therefore, great care has to be taken. Qualitative validation has been reported in animals (Mori et al., 1999, 2000; Xue et al., 1999) and human beings (Conturo et al., 1999; Lazar et al., 2000; Parker, 2000; Poupon et al., 2000; Tuch et al., 2000, 2001; Stieltjes et al., 2001), and the results are very encouraging. Comparison studies between a manganese-based tracking technique and DTI was also reported recently (Lin et al., 2001). The study showed that the DTI could correctly determine the fiber orientation with less than 10% of deviation for a signal-to-noise ratio (SNR) of 40 or greater. Probably the ideal validation study may require a phantom, in which such uncertainties as the effect of motions and PVE structures can be removed, although such a phantom is currently not available.

Summary and conclusions In this chapter, several DTI tractography techniques are discussed that are presently being used for WM

95

96

S. Mori and P. van Zijl

tract tracing. As the field of tract tracing by DTI is rather young, it is expected that many new technologies will be developed in the near future. Recent results, however, have shown that even the simple methodologies reviewed here are already able to visualize major WM connections in situ in animals and human beings. For instance, the brainstem fibers and several of the cortico-cortical association fibers have recently been reported, including the corpus callosum, thalamic radiations, corticospinal tracts, association fibers, and cerebellar peduncles (Conturo et al., 1999; Jones et al., 1999; Mori et al., 1999; Xue et al., 1999; Basser et al., 2000; Lazar et al., 2000; Mori et al., 2000; Parker, 2000; Poupon et al., 2000; Tuch et al., 2000; Wedeen et al., 2000; Werring et al., 2000; Stieltjes et al., 2001; Mori et al., 2002b). Upon using these exciting data to investigate specific neuroanatomical questions, it is very important to keep in mind the limitations of the DTI method used to acquire them. First of all, this technique can be used only for macroscopic analysis of WM architecture, but not to address connectivity questions at the cellular level. One particularly limiting problem related to this macroscopic character of DTI is the mixing of axonal tracts with different orientations within a pixel. DTI may be able to locate where these problematic pixels are, but it is difficult to decipher axonal information in such pixels. On the other hand, there are some approaches that circumvent this issue under favorable conditions, such as the use of reference ROI placement based on prior knowledge. The most important conclusion that can be drawn in this initial phase of the field is that DTI tractography can indeed delineate the core of large WM tracts as judged from the encouraging results from initial validation studies. At present, there are no other non-invasive techniques that can provide equivalent information and, as a consequence, DTI tractography is expected to be a powerful technique to investigate WM anatomy and disease in situ in human beings.

REFERENCES Ahrens ET, Laidlaw DH, Readhead C, Brosnan CF, Fraser SE, Jacobs RE. 1998. MR microscopy of transgenic mice

that spontaneously acquire experimental allergic encephalomyelitis. Magn Reson Med 40: 119–132. Albayram S, Melhem ER, Mori S, Zinreich SJ, Barkovich AJ, Kinsman SL. 2002. Holoprosencephaly in children: diffusion tensor MR imaging of white matter tracts of the brainstem-initial experience. Radiology 223: 645–651. Basser JB. 1998. In Proceeding of International Society for Magnetic Resonance in Medicine, Vol. 2, Sydney, pp. 1226. Basser PJ, Jones DK. 2002. Diffusion-tensor MRI: theory, experimental design and data analysis – a technical review. NMR Biomed 15: 456–467. Basser PJ, Mattiello J, Le Bihan D. 1994a. MR diffusion tensor spectroscopy and imaging. Biophys J 66: 259–267. Basser PJ, Mattiello J, LeBihan D. 1994b. Estimation of the effective self-diffusion tensor from the NMR spin echo. J Magn Reson B 103: 247–254. Basser PJ, Pajevic S, Pierpaoli C, Duda J, Aldroubi A. 2000. In vitro fiber tractography using DT-MRI data. Magn Reson Med 44: 625–632. Basser PJ, Pierpaoli C. 1996. Microstructural features measured using diffusion tensor imaging. J Magn Reson B 111: 209–219. Beaulieu C. 2002. The basis of anisotropic water diffusion in the nervous system – a technical review. NMR Biomed 15: 435–455. Beaulieu C, Allen PS. 1994. Determinants of anisotropic water diffusion in nerves. Magn Reson Med 31: 394–400. Carpenter M. 1976. Human Neuroanatomy, Williams & Wilkins, Baltimore. Clark C, Werring D, Miller D. 2000. Diffusion imaging of the spinal cord in vivo: estimation of the principal diffusion and application to multiple sclerosis. Magn Reson Med 43: 133–138. Conturo TE, Lori NF, Cull TS, Akbudak E, Snyder AZ, Shimony JS, McKinstry RC, Burton H, Raichle ME. 1999. Tracking neuronal fiber pathways in the living human brain. Proc Natl Acad Sci USA 96: 10422–10427. Crosby E, Humphrey T, Lauer E. 1962. Correlative Anatomy of the Nervous System, MacMillan, New York. Douek PRT, Pekar J, Patronas N, Le Bihan D. 1991. MR colour mapping of myelin fiber orientation. J Comput Assist Tomogr 15: 923–929. Ellis C, Simmons A, Jones D, Bland J, Dawson J, Horsfield M, Williams S, Leigh P. 1999. Diffusion tensor MRI assesses corticospinal tract damages in ALS. Neurology 22: 1051–1058. Frank LR. 2001. Anisotropy in high angular resolution diffusion-weighted MRI. Magn Reson Med 45: 935–939. Henkelman R, Stanisz G, Kim J, Bronskill M. 1994. Anisotropy of NMR properties of tissues. Magn Reson Med 32: 592–601. Holodny AI, Ollenschleger MD, Liu WC, Schulder M, Kalnin AJ. 2001. Identification of the corticospinal tracts achieved

MR tractography using diffusion tensor MR imaging

using blood-oxygen-level-dependent and diffusion functional MR imaging in patients with brain tumors. AJNR Am J Neuroradiol 22: 83–88. Hoon Jr AH, Lawrie Jr WT, Melhem ER, Reinhardt EM, Van Zijl PC, Solaiyappan M, Jiang H, Johnston MV, Mori S. 2002. Diffusion tensor imaging of periventricular leukomalacia shows affected sensory cortex white matter pathways. Neurology 59: 752–756. Horsfield MA, Jones DK. 2002. Applications of diffusionweighed and diffusion tensor MRI to white matter diseases. NMR Biomed 15: 570–577. Jones DK, Simmons A, Williams SC, Horsfield MA. 1999. Noninvasive assessment of axonal fiber connectivity in the human brain via diffusion tensor MRI. Magn Reson Med 42: 37–41. Klingberg T, Hedehus M, Temple E, Salz T, Gabrieli J, Moseley M, Poldrack R. 2000. Microstructure of temporo-parietal white matter as a basis for reading ability: evidence from diffusion tensor magnetic resonance imaging. Neuron 25: 493–500. Lazar M, Alexander AL. 2001. In Proceeding of International Society of Magnetic Resonance in Medicine, Glasgow, UK, pp. 506. Lazar M, Weinstein D, Hasan K, Alexander AL. 2000. In Proceeding of International Society of Magnetic Resonance in Medicine, Denver, CO, pp. 482. Lazar M, Weinstein DM, Tsuruda JS, Hasan KM, Arfanakis K, Meyerand ME, Badie B, Rowley HA, Haughton V, Field A, Alexander AL. 2003. White matter tractography using diffusion tensor deflection. Hum Brain Mapp 18: 306–321. Lim KO, Hedehus M, Moseley M, de Crespigny A, Sullivan EV, Pfefferbaum A. 1999. Compromised white matter tract integrity in schizophrenia inferred from diffusion tensor imaging. Arch Gen Psychiatry 56: 367–374. Lin CP, Tseng WY, Cheng HC, Chen JH. 2001. Validation of diffusion tensor magnetic resonance axonal fiber imaging with registered manganese-enhanced optic tracts. Neuroimage 14: 1035–1047. Lori NF, Akbuda E, Snyder AZ, Shimony JS, Conturo TE. 1999. In International Society of Magnetic Resonance in Medicine, Denver, CO, pp. 775. Lori NF, Akbudak JS, Shimony TS, Snyder RK, Conturo TE. 2002. Diffusion tensor fiber tracking of brain connectivity: reliability analysis and biological results. NMR Biomed 15: 494–515. Makris N, Worth AJ, Sorensen AG, Papadimitriou GM, Reese TG, Wedeen VJ, Davis TL, Stakes JW, Caviness VS, Kaplan E, Rosen BR, Pandya DN, Kennedy DN. 1997. Morphometry of in vivo human white matter association pathways with diffusion weighted magnetic resonance imaging. Ann Neurol 42: 951–962.

Mori S, Crain BJ, Chacko VP, van Zijl PCM. 1999. Three dimensional tracking of axonal projections in the brain by magnetic resonance imaging. Ann Neurol 45: 265–269. Mori S, Crain BJ, van Zijl PC. 1998. In Proceeding of International Conference on Functional Mapping of the Human Brain Montreal. Mori S, Fredericksen K, van Zijl PC, Stieltjes B, kraut AK, Solaiyappan M, Pomper MD. 2002a. Brain white matter anatomy of tumor patients using diffusion tensor imaging. Ann Neurol 51: 377–380. Mori S, Itoh R, Zhang J, Kaufmann WE, van Zijl PCM, Solaiyappan M, Yarowsky P. 2001. Diffusion tensor imaging of the developing mouse brain. Magn Reson Med 46: 18–23. Mori S, Kaufmann WE, Davatzikos C, Stieltjes B, Amodei L, Fredericksen K, Pearlson GD, Melhem ER, Solaiyappan M, Raymond GV, Moser HW, van Zijl PCM. 2002b. Imaging cortical association tracts in human brain. Magn Reson Med 47: 215–223. Mori S, Kaufmann WK, Pearlson GD, Crain BJ, Stieltjes B, Solaiyappan M, van Zijl PCM. 2000. In vivo visualization of human neural pathways by MRI. Ann Neurol 47: 412–414. Mori S, van Zijl PC. 2002. Fiber tracking: principles and strategies – a technical review. NMR Biomed 15: 468–480. Moseley ME, Cohen Y, Kucharczyk J, Mintorovitch J, Asgari HS, Wendland MF, Tsuruda J, Norman D. 1990. Diffusionweighted MR imaging of anisotropic water diffusion in cat central nervous system. Radiology 176: 439–445. Mukherjee P, Bahn M, McKinstry R, Shimony J, Cull T, Akudak E, Snyder A, TE C. 2000a. Difference between gray matter and white matter water diffusion in stroke: diffusion tensor MR imaging in 12 patients. Radiology 215: 211–220. Mukherjee P, Bahn MM, McKinstry RC, Shimony JS, Cull TS, Akbudak E, Snyder AZ, Conturo TE. 2000b. Differences between gray matter and white matter water diffusion in stroke: diffusion-tensor MR imaging in 12 patients. Radiology 215: 211–220. Nakada T, Matsuzawa H. 1995. Three-dimensional anisotropy contrast magnetic resonance imaging of the rat nervous system: MR axonography. Neurosc Res 22: 389–398. Neil J, Shiran S, McKinstry R, Schefft G, Snyder A, Almli C, Akbudak E, Arnovitz J, Miller J, Lee B, Conturo T. 1998. Normal brain in human newborns: apparent diffusion coefficient and diffusion anisotropy measured by using diffusion tensor MR imaging. Radiology 209: 57–66. Neil JM, Mukherjee P, Huppi PS. Diffusion tensor imaging of normal and injured developing human brain. NMR Biomed in press. Nieuwenhuys R, Voogd J, van Huijzen C. 1983. The Human Central Nervous System, Springer–Verlag. Pajevic S, Pierpaoli C. 1999. Colour schemes to represent the orientation of anisotropic tissues from diffusion tensor

97

98

S. Mori and P. van Zijl

data: application to white matter fiber tract mapping in the human brain. Magn Reson Med 42: 526–540. Parker GJ. 2000. In International Society of Magnetic Resonance, Denver, CO, pp. 85. Parker GJ, Stephan KE, Barker GJ, Rowe JB, MacManus DG, Wheeler-Kingshott CA, Ciccarelli O, Passingham RE, Spinks RL, Lemon RN, Turner R. 2002. Initial demonstration of in vivo tracing of axonal projections in the macaque brain and comparison with the human brain using diffusion tensor imaging and fast marching tractography. Neuroimage 15: 797–809. Pierpaoli C, Basser PJ. 1996. Toward a quantitative assessment of diffusion anisotropy. Magn Reson Med 36: 893–906. Poupon C, Clark CA, Frouin V, Regis J, Bloch L, Le Bihan D, Mangin JF. 2000. Regularization of diffusion-based direction maps for the tracking of brain white matter fascicules. NeuroImage 12: 184–195. Sakuma H, Nomura Y, Takeda K, Tagami T, Nakagawa T, Tamagawa Y, Ishii Y, Tshukamoto T. 1991. Adult and neonatal human brain: diffusional anisotropy and myelination with diffusion-weighted MR imaging. Radiology 180: 229–233. Scollan DF, Holmes A, Winslow R, Forder J. 1998. Histological validation of myocardial microstructure obtained from diffusion tensor magnetic resonance imaging. Am J Physiol 275: H2308–H2318. Sotak CH. 2002. The role of diffusion tensor imaging in the evaluation of ischemic brain injury. NMR Biomed 15: 561–569. Stanisz GJ, Szafer A, Wright GA, Henkelman RM. 1997. An analytical model of restricted diffusion in bovine optic nerve. Magn Reson Med 37: 103–111. Stejskal E. 1965. Use of spin echoes in a pulsed magnetic-field gradient to study restricted diffusion and flow. J Chem Physics 43: 3597–3603. Stieltjes B, Kaufmann WE, van Zijl PCM, Fredericksen K, Pearlson GD, Mori S. 2001. Diffusion tensor imaging and axonal tracking in the human brainstem. NeuroImage 14: 723–735.

Tanner JE. 1979. Self diffusion of water in frog muscle. Biophys J 28: 107–116. Tievsky A, Ptak T, Farkas J. 1999. Investigation of apparent diffusion coefficient and diffusion tensor anisotrophy in acute and chronic multiple sclerosis lesions. AJNR 20: 1491–1499. Tuch DS, Belliveau JW, Wedeen V. 2000. In Proceeding of International Society of Magnetic Resonance in Medicine, Denver, CO, pp. 791. Tuch DS, Wiegell MR, Reese TG, Belliveau JW, Wedeen V. 2001. In Proceeding of International Society of Magnetic Resonance in Medicine, Glasgow, UK, pp. 502. Ulug A, van Zijl PCM. 1999. Orientation-independent diffusion imaging without tensor diagonalization: anisotropy definitions based on physical attributes of the diffusion ellipsoid. J Magn Reson Imaging 9: 804–813. van Gelderen P, DesPres D, van Zijl PCM, Moonen CTW. 1994. Evaluation of restricted diffusion in cylinders. Phosphocreatine in rat muscle. J Magn Reson B 103: 247–254. Virta A, Barnett A, Pierpaoli C. 1999. Visualizing and characterizing white matter fiber structure and architecture in the human pyramidal tract using diffusion tensor MRI. Magn Reson Imaging 17: 1121–1133. Wedeen V, Reese TG, Tuch DS, Weigel MR, Dou JG, Weiskoff RM, Chessler D. 2000. In Proceeding of International Society of Magnetic Resonance in Medicine, Denvor, CO, pp. 82. Werring DJ, Toosy AT, Clark CA, Parker GJ, Barker GJ, Miller DH, Thompson AJ. 2000. Diffusion tensor imaging can detect and quantify corticospinal tract degeneration after stroke. J Neurol Neurosurg Psychiatry 69: 269–272. Wiegell M, Larsson H, Wedeen V. 2000. Fiber crossing in human brain depicted with diffusion tensor MR imaging. Radiology 217: 897–903. Xue R, van Zijl PCM, Crain BJ, Solaiyappan M, Mori S. 1999. In vivo three-dimensional reconstruction of rat brain axonal projections by diffusion tensor imaging. Magn Reson Med 42: 1123–1127.

6

Artifacts and pitfalls in diffusion MR imaging Martin A. Koch1 and David G. Norris2 1

Universitätsklinikum Hamburg-Eppendorf, Klinik und Poliklinik für Neurologie, Hamburg, Germany FC Donders Centre for Cognitive Neuroimaging, Nijmegen, The Netherlands

2

Key points • Artifacts may be related to the MR equipment itself, e.g. eddy currents and non-linear gradients. • Artifacts may also be due to properties of the subject being imaged, e.g. T2 effects, as in T2 shine-through, and anisotropy.

Introduction Although some diffusion-weighted imaging (DWI) techniques have entered the stage of clinical routine application, particularly in the detection of cerebral infarction, obtaining and interpreting diffusion imaging results is not always straightforward. This is both due to the numerous technical difficulties and also to the sensitivity of diffusion imaging experiments to phenomena other than diffusion (cf. Conturo et al., 1995; Norris, 2001a, 2001b). Further complications arise from the sheer number of diffusion parameters that can be derived from the measurement in biological tissue, such as eigenvectors, eigenvalues, anisotropy, and trace of the diffusion tensor, diffusion coefficients for a given direction, etc. This chapter aims at providing an overview of the most important difficulties encountered in MR diffusion imaging. The parameters derived from DWI can be affected by a number of sources of error. These error sources may be divided into two groups, according to whether they arise from properties of the measurement apparatus or from properties of the measured object itself.

Object-related sources of error T2 effect The time required for diffusion sensitization entails long echo times (TE), because of the duration of the gradient pulses required for significant diffusion weighting. The signal intensity in a diffusion-weighted image therefore depends on T2, i.e. these images are both T2 and diffusion weighted. Hence on these images tissue with long T2 values, e.g. due to an earlier ischemic insult, can appear as bright as tissue with a low-diffusion coefficient, e.g. due to a recent stroke. This can hamper attempts to delineate an infarct, and is a phenomenon often called “T2 shinethrough”. However, at large b values the image contrast is dominated by diffusion. In addition, the degree of T2 weighting is constant and independent of the b value such that calculated maps of diffusion coefficients or the trace of the diffusion tensor are independent of T2. Therefore, the T2 effect is not a problem for stroke detection when these maps are used. Also, with the development of clinical scanners with stronger gradient values, it should become possible in the future to record high b-value diffusion images with shorter TE and hence less T2 weighting. Anisotropy of white matter as an artifact The diffusion characteristics of water in tissue depend on a number of parameters, such as membrane permeability, size and shape of cells, volume ratio of different compartments, and intra- and extracellular diffusion coefficients. In particular, the diffusion 99

100

Martin A. Koch and David G. Norris

attenuation in an MR imaging (MRI) measurement of human brain white matter (WM) depends on the fiber orientation in the measured volume element. These influences imply a number of difficulties in the interpretation of results from MR diffusion imaging studies, the most relevant of which will be mentioned here. Anisotropy can mask lesion borders in diffusion-weighted images The anisotropy of diffusion, which can be quantified by means of the fractional anisotropy (FA) index (Basser, 1995), can be a significant confound for lesion delineation in clinical investigations (van Gelderen et al., 1994). However, the trace of the diffusion tensor, Trace(D), is independent of the fiber orientation. Hence, ischemic regions should be assessed on maps of the trace (or on trace-weighted images). The trace can be measured by averaging the results of apparent diffusion coefficient (ADC) measurements in any three directions that are orthogonal to each other. Since it is invariant under rotation, the orientation of the set of orthogonal directions in space can be chosen at will. Trace(D) can also be calculated from the full diffusion tensor, or from a “trace-weighted” experiment (Mori and van Zijl, 1995; Wong et al., 1995). Partial volume effects can modify anisotropy In a situation with different fiber orientations in a voxel the measured tensor represents a weighted “average” over these directions (Wiegell et al., 2000; Alexander et al., 2001). A voxel will exhibit zero anisotropy if the distribution of fiber orientations is purely random. Likewise, the anisotropy can be significantly reduced in a voxel containing two distinct fiber orientations. This partial volume effect (PVE) can lead to reduced measured anisotropy in the region between neighboring fiber tracts, or in regions of crossing or merging fibers (cf. Figure 6.1). Different fiber orientations can even occur in very small voxels. It should also be noted that PVEs between gray matter (GM) and white matter (WM) can also modify the results of group comparisons of anisotropy in subcortical WM. Although diffusion anisotropy also exists in GM (Shimony et al., 1999), it is much less than in WM and more difficult to interpret in terms of tissue structure. Reliable diffusion tensor imaging (DTI) measurements of diffusion anisotropy in

(a)

(b)

Fig. 6.1 Reduced anisotropy at fiber crossings in human brain WM. (a) T1-weighted image of a human brain with eigenvector orientations for the largest diffusion tensor eigenvalue (red), in voxels with FA  0.2 only. The arrow marks a region of approximately 3-mm diameter where the anisotropy is below the threshold, due to averaging over different fiber directions. The dark diagonal line is the central sulcus (slice orientation is axial). (b) Corresponding map of FA with eigenvector orientations for the largest (red) and the smallest (blue) eigenvalue, in voxels with FA  0.1 only. All orientations are displayed as projections of the vectors onto the imaging plane.

Artifacts and pitfalls in diffusion MR imaging

0

0.1 With IVIM

0.2 ln (S/S0)

cortical GM tissue structure have not been achieved to date. Unfortunately, the details of tissue alteration underlying observed changes in the diffusion characteristics have often not been investigated in detail. In particular, a difference in anisotropy can represent different distributions of the fiber directions in a voxel (Wiegell et al., 2000; Pierpaoli et al., 2001) instead of a difference in myelination.

0.3

0.4

0.5

0.6

Partial volume effects can lead to a wrong estimated fiber direction Not only the degree of diffusion anisotropy but also the direction of the eigenvector corresponding to the largest tensor eigenvalue depends on the distribution of fiber directions in the voxel (Wiegell et al., 2000; Pierpaoli et al., 2001). Strictly speaking, in some tissue regions the diffusion characteristics of water cannot be described by a 3 3 tensor. For example, if two perpendicular fiber orientations are present in a voxel then the displacement probability function is no longer the same for all possible starting points of a molecule within that voxel. In a voxel containing equal fractions of two perpendicular fiber orientations, the measured tensor ellipsoid is a flat circular disk, suggesting that the root mean square diffusion displacement is equal for all directions in the plane of the disk. This is, however, not true, since only two fiber orientations are present in the voxel. All directions in the plane of the disk are eigenvectors of the diffusion tensor. In the presence of noise any of these directions can come out as the eigenvector with the largest eigenvalue, which is usually taken to be the predominant fiber direction. However, in this situation the anisotropy is lower than in a voxel containing a single-fiber direction. Hence, the degree of diffusion anisotropy provides an estimate of how reliably the eigenvector corresponding to the largest eigenvalue can be taken as the fiber direction in WM. The problem can be circumvented using DWI with high angular resolution, e.g. three-dimensional q-space imaging (Tuch et al., 1999; Wedeen et al., 2000; Tuch et al., 2002). This technique comprises the acquisition of images with diffusion weighting in more than six directions. It aims at a direct measurement of the distribution of diffusion displacements over the directions in space. The iso-surface of the

0.7

0

100

200

300

400

500

600

b value (10 6 s mm 2) Fig. 6.2 Perfusion contribution. Schematic of the signal attenuation in a diffusion-weighted image in vivo (solid line). By taking the natural logarithm, ln (S/S0), of the ratio between signal intensity with and without diffusion weighting the exponential dependence on b is displayed as a linear relationship. In the case of pure diffusion (dashed straight line) the attenuation depends linearly on the b value, which is proportional to the square of the diffusion gradient amplitude. The deviation from linear behavior arises from perfusion contributions, i.e. intra-voxel incoherent motion (IVIM). At b values above approx. 200 s mm 2 this contribution is negligible. Figure after Conturo et al. (1995).

resulting function is not necessarily ellipsoidal, in contrast to a diffusion tensor surface. Influence of perfusion For b values less than approximately 100–200 s mm 2, the signal attenuation may deviate from the usual exponential dependence on the b value (Le Bihan et al., 1986; Conturo et al., 1995). This behavior is depicted schematically in Figure 6.2. The effect is considered to arise from signal attenuation due to the motion of blood water in the randomly oriented vessels of the capillary network. It has been termed intra-voxel incoherent motion (IVIM) or pseudodiffusion (Le Bihan et al., 1986, 1988). Due to the technical difficulties IVIM has not been established as a standard method for the assessment of perfusion. As in most applications the diffusion weighting is above 200 s mm 2, pseudo-diffusion can be neglected in the majority of diffusion measurements.

101

102

Martin A. Koch and David G. Norris

Motion and flow Measurements of diffusion are designed to detect particle displacements in the micrometer range. The influence of object motion on the measurement is therefore a major problem in DWI. In the following we disregard effects arising from mis-registration due to the gross displacement since these are not specific to diffusion imaging studies. Commonly two field gradient pulses are used to introduce diffusion weighting into an imaging sequence (Stejskal and Tanner, 1965). Gradient pulses modify the magnetic field for a short period of time. The first gradient pulse imposes a phase shift on the nuclei in the object which depends linearly on the nucleus position measured along the gradient direction. After a waiting period a second gradient pulse is used to rewind the phases to their original values. Any spin motion during this procedure, e.g. between the gradient pulses, leads to incomplete rewinding and hence to a residual phase shift. Since diffusion is a random motion these phase shifts result in an effective reduction of the signal arising from a given volume element. However, residual phase shifts also occur if the object position during the first and second diffusion weighting gradient is not the same. Coherent movement of the whole sample leads to a net phase shift, which may vary across the object. The effects of such phase shifts are discussed in the following. Most moving fluids in the human body flow in such a way that the particle velocity is not the same at all positions within an imaging voxel of common size. This means that flow in the presence of diffusion weighting usually leads to signal attenuation through intravoxel dephasing. Since this dephasing depends on the strength of diffusion weighting the attenuation cannot be distinguished from dephasing due to diffusion. For example, pulsatile motion of cerebrospinal fluid (CSF) in the lateral ventricles artificially increases the value of the measured diffusion coefficient in the flow direction. Hence, diffusion in the isotropic CSF inside lateral ventricles appears to be anisotropic (cf. Figure 6.3). Similar measurement confounds can also arise from other kinds of non-rigid body motion, such as pulsatile deformation related to the cardiac or respiratory cycle.

Fig. 6.3 Flow contribution. Fiber orientation map from a DTI experiment overlaid on an anatomical T1-weighted image of the human brain (axial slice). The straight lines indicate the in-plane components of the calculated fiber direction. The eigenvector orientations in the lateral ventricles (arrow) reflect predominant posterior–anterior oriented flow. Fiber orientations in voxels with low anisotropy (FA  0.1) are suppressed.

The situation with rigid body motion is somewhat different. As soon as motion leads not only to intravoxel dephasing but (also) to net phase errors, severe image artifacts can occur, which can compromise the diffusion measurement or render the image completely unusable. Any displacement of a rigid body may be described completely in terms of translation and rotation. Coherent translation of the whole sample between the diffusion-sensitizing gradient pulses leads to a global net phase shift of all spins in the sample. If the sample is rotated then the particle displacement is proportional to the distance from the rotation axis. This means that after the second diffusion gradient pulse the phase varies linearly along a direction which is perpendicular both to the diffusion gradient direction and to the axis of rotation (Anderson and Gore, 1994; Ordidge et al., 1994; Butts et al., 1996), which is illustrated in Figure 6.4. The effects of motion-induced phase errors depend on the specific imaging sequence applied. In the following discussion we concentrate on the echo

Artifacts and pitfalls in diffusion MR imaging

y

G

y 

grad f

a b c

v x

B0 f

x Fig. 6.4 Effect of rigid-body object rotation in DWI. The rotation of the object about an axis  (which is perpendicular to the paper plane) in the presence of a diffusion gradient, G, produces a linear phase ramp, grad , in a direction perpendicular to G and  (axis of rotation perpendicular to diffusion gradient for simplicity). The velocity (v) components along the field gradient, a, b, and c are equal because this velocity component depends on the x-coordinate only. By definition, the gradient vector, G, points in the direction of increasing magnetic field strength, B0, while the phase gradient, grad , is a vector pointing in the direction of increasing spin phase.

planar imaging (EPI) method (Mansfield, 1977) as this is, due to its speed and high signal-to-noise ratio (SNR), the most widely used sequence in diffusion imaging. A more detailed discussion of the motion sensitivity of different diffusion imaging methods can be found elsewhere (Norris, 2001b). With EPI, pure translation of the object as a whole does not induce artifacts in the magnitude images unless the acquired data points are unequally affected. The latter is the case in segmented (i.e. multiple excitation) data acquisition (Cho et al., 1987). The effects of a rotation-induced linear phase variation on the image depend on the direction of this phase gradient. 1. A phase gradient parallel to the phase-encoding gradient induces a temporal shift of the echo with the largest amplitude. This leads to a modified effective TE and consequently to a modified degree of T2* weighting. 2. If the phase gradient is parallel to the readout direction, alternate echoes are shifted in opposite

directions in time, which is identical to the effect of an incorrectly adjusted readout dephase gradient. The resulting ghosting artifacts (displaced by half the field-of-view, FOV) can be removed by a first-order phase correction (cf. Schmitt et al., 1998). Such a correction is also used for most EPI implementations in order to suppress Nyquist ghosting due to the fast-decaying eddy-current effects. However, it must be performed separately for each image to remove the motion-induced artifacts in diffusion weighting. 3. A rotation-induced phase gradient perpendicular to the measured slice results in intravoxel dephasing and consequential signal loss. Hence, although single-shot EPI is characterized by a readout train length of less than 100 ms which is short enough to freeze out most motion in regular in vivo imaging, the incorporation of diffusion sensitization renders the sequence susceptible to subject motion. While these effects are usually small, the situation is much worse if the image data are acquired in a segmented fashion. Due to its low-frequency bandwidth in the phaseencoding direction, echo planar images suffer from susceptibility artifacts that occur predominantly in ventral brain structures. The method of k-space segmentation (Cho et al., 1987; McKinnon, 1993; Ries et al., 2000; Stieltjes et al., 2001) is used to reduce these. It can also be used to increase the spatial resolution. However, in a segmented acquisition scheme the phase errors due to translation or rotation in the presence of diffusion weighting are in general not the same for all k-space segments. Hence they lead to severe blurring and ghosting artifacts in diffusionweighted segmented EPI unless some kind of correction is employed. The most widely established correction technique is the use of navigator echoes (Anderson and Gore, 1994; Ordidge et al., 1994). These approaches are based on the idea of acquiring an additional line of non-phase encoded data with each excitation. As these data contain the motion-induced phase shifts but not the phase information which encodes spatial position, the imaging echoes can be corrected accordingly. To correct for phase gradients in the phase-encode direction two-dimensional navigator echoes can be used (Anderson and Gore, 1995; Fu et al., 1995; Butts et al., 1996, 1997). These

103

104

Martin A. Koch and David G. Norris

approaches usually require image post-processing (Atkinson et al., 2000). In many cases scan rejection for cases of very severe motion as well as electrocardiogram (ECG) gating are also required for efficient artifact suppression. Even with these measures in effect, subject motion can still prevent the reconstruction of artifact-free images when strong diffusion weighting is employed. There are other diffusion imaging sequences that are less motion sensitive than EPI (cf. Norris, 2001b). However, these methods are characterized by other disadvantages which so far have prevented them from being widely used in DWI studies. Most of them are much slower than single-shot EPI, in particular in multi-slice imaging. For a reasonable choice between these sequences for a given application one has to consider the required number of slices and spatial resolution, the main field strength, the expected degree of subject motion, and the location of the anatomical structures to be imaged. If a low number of slices through the ventral brain is to be imaged at high magnetic field strength, sequences other than EPI may well be considered. Diffusion effects of susceptibility gradients The degrading effects of background gradients on the image quality aside (Mansfield, 1977), the magnetic field inhomogeneity arising from differences in the magnetic properties of tissues has an influence on the MR measurement of diffusion coefficients (Kärger et al., 1988; Neeman et al., 1990). However, if the geometric mean (i.e. the square root of the product) of two measurements with opposite diffusion gradient direction is used as input for the calculation of the diffusion coefficient or tensor, the effects of background gradients are cancelled to a considerable degree (Neeman et al., 1991; Conturo et al., 1995). In general, diffusion effects from susceptibility gradients are neglected in in vivo studies. Diffusion time dependence Care has to be taken when comparing results from different diffusion experiments. The diffusion characteristics of tissue depend on the diffusion time. Hence, duration and temporal separation of the diffusion gradients affect the measured diffusion

coefficients. This behavior of tissue is based on the fact that cell membranes hinder or restrict the diffusion of water molecules. Consequently, the root mean square displacement of a particle cannot increase beyond the cell diameter (in the case of perfectly impermeable membranes). This effect is also considered to be the reason for the anisotropy of diffusion in brain WM (arguments reviewed in Norris, 2001a).

Measurement system-related sources of error Imaging gradients The magnetic field gradients used for image formation also represent a source of diffusion weighting (Mattiello et al., 1994, 1997). In sequences other than EPI it can be very difficult to calculate this contribution. Generally, linear regression analysis with the diffusion gradient amplitude as the independent variable is used to calculate the diffusion coefficients or components of the diffusion tensor (Stejskal and Tanner, 1965; Basser et al., 1994). However, this does not completely rule out the influence of imaging gradients. Hence, the slope of the signal decrease with increasing diffusion gradient amplitude (in a semi-logarithmic plot as in Figure 6.2) also depends on the amplitude and duration of other gradient pulses used in the imaging sequence. It is important not to confuse these “cross terms” (Neeman et al., 1990) with the off-diagonal elements of the b matrix. However, the contribution of imaging gradients is mostly small (Mattiello et al., 1994, 1997). As a general rule, diffusion effects from imaging gradients can be minimized if gradients are refocused as soon as possible (Mattiello et al., 1997). Eddy currents A diffusion measurement usually relies on the use of strong magnetic field gradients which are rapidly switched on and off. The fast change of the magnetic field during the switching period induces an electric voltage in any surrounding material, which in turn generates electrical current with a resulting magnetic field. These eddy currents are particularly strong in the relatively cold, conducting metal of the magnet

Artifacts and pitfalls in diffusion MR imaging

cryostat which is filled with liquid helium. The magnetic fields produced by it can corrupt the MR measurement. Diffusion tensor measurements are particularly vulnerable to eddy currents produced by the diffusion gradient pulses because the degree and nature of eddy-current-induced artifacts will typically vary with amplitude and orientation of the gradient pulses applied (Koch and Norris, 2000). Any dependence of image intensity on the diffusion gradient direction introduces artifactual anisotropy in the diffusion tensor. The effect of these gradients on the image is analogous to that of susceptibility gradients. The error in the calculated diffusion coefficient can often be neglected in ADC maps but in DTI it can dramatically change the direction corresponding to the largest calculated principal diffusivity. This can lead to an incorrect determination of the fiber direction in the affected voxels while it affects measurements of the tensor trace usually to a lower degree. We concentrate here on the use of EPI-based sequences, although various other pulse sequences have been applied to DWI (cf. e.g. Norris, 2001b). EPI is particularly sensitive to the presence of eddy currents arising from switching strong diffusion gradients. Different approaches to the eddy-current problem in DWI and DTI have been proposed. One possibility is to apply some form of correction, which often involves the acquisition of additional data to obtain the correction parameters (Haselgrove and Moore, 1996; Jezzard et al., 1998; Bastin, 1999; Horsfield, 1999). The nature of this correction depends on the imaging sequence employed. In echo planar images eddy currents lead to shifting, shearing, and scaling of the image. Some of these effects are illustrated in Figure 6.5. As these geometric alterations differ between diffusion gradient directions, high artifactual anisotropy can be induced, particularly at tissue borders. The shift artifact is caused by a residual offset of the main magnetic field. This can be explained as follows. For simplicity let us assume that the eddy-current effects do not change considerably during the duration of the echo train. Through the relatively long time between the acquisition of successive echoes this offset results in k-space in a phase ramp in the phase-encoding direction. After the Fourier transform (FT) such a phase gradient

(a)

(b)

Fig. 6.5 Eddy-current effects. (a) Difference between two diffusion-weighted echo planar images with different diffusion gradient directions. Eddy-current-related differences in image position and size measured along the phase-encoding direction (vertical in the image) appear here as a dark edge (arrows). This translates directly into high FA in these regions and at other tissue borders. Dark and bright regions inside the brain represent true differences due to tissue anisotropy. (b) Difference image between high (b  800 s mm 2) and low (b  50 s mm 2) diffusion weighting in the same direction, for a spherical, isotropic phantom (agarose gel). Before subtraction the high b-value image was averaged over multiple repetitions to achieve the same SNR as in the low b-value image. Ideally the whole image should be gray without any structure: medium gray-scale value indicates zero difference. The dark and bright outer edges represent an eddy-currentinduced image shift as in (a), while the edges inside the object arise from an image “ghost”, displaced by half the FOV, whose intensity depends on the eddy currents due to the diffusion gradient applied.

105

106

Martin A. Koch and David G. Norris

appears as a shift in image space. The shear artifact can be explained in an analogous way: a residual field gradient in readout direction leads to an image shift that depends linearly on the spatial position along the readout direction. A phase gradient along the phase-encode axis has the same effect as adding a constant offset to the phase-encoding gradient function. This is equivalent to an increase in the time integral of each phase-encoding blip. Since the k-space increment between echoes is inversely proportional to the FOV, the image data are reconstructed with an incorrect assumption on the FOV. Thus, on the final image, the object appears shrunken or dilated along the phase-encode direction. Another possibility for the reduction of eddycurrent effects is to modify the diffusion weighting experiment in such a way that less eddy currents are produced (Gibbs and Johnson, 1991; Wider et al., 1994; Alexander et al., 1997; Smart et al., 1999). This can efficiently be achieved by using gradient pulses of alternating polarity (Wider et al., 1994; Heid, 2000). This approach is often applied in diffusionweighted EPI sequences (Reese et al., 1998). The diffusion gradient parameters can also influence the efficiency of additional “spoiler” gradients, which are used to suppress transverse magnetization created by the 180° radio-frequency (RF) pulse in Stejskal–Tanner (Stejskal and Tanner, 1965) type diffusion weighting. This can either be due to the eddy currents produced by the diffusion gradient pulses that modify the spoiler gradients, or due to the additional spoiling effect of the diffusion gradients themselves. Insufficient spoiler gradients typically lead to dark and bright stripes in the images and in the parameter maps calculated from these images. These effects are not removed by common eddy-current correction schemes. They have to be considered in particular if the direction of the diffusion gradients is varied, such as in DTI. However, the problem is suppressed if the spoiler gradients are sufficiently strong. Even image artifacts which are hardly visible in the diffusion-weighted images can influence the measured diffusion anisotropy if they depend on the diffusion gradient. A typical example is the presence of ghosting. The degree of ghosting often depends on the direction of the eddy-current-producing diffusion gradient pulses (Koch and Norris, 2000). This

results in increased anisotropy or incorrect eigenvector directions in the regions affected by the ghost. These effects can sometimes be reduced by increasing the FOV such that the ghost does not overlap with the object. However, this usually compromises spatial resolution. Non-linearity of gradient fields The gradient coils are designed to add a contribution to the main magnetic field which increases linearly along one of three orthogonal directions. However, the region where this linearity assumption holds is limited (Wald et al., 2001), due to the fact that the coil is not infinitely large. In fact, the field gradient decreases when moving more than a few centimeters away from the iso-center of a whole body MR system. In anatomical imaging this effect becomes apparent in distortions in the outer regions of the image. Additional but usually negligible deviations from the ideal behavior arise from so-called Maxwell gradients (Norris and Hutchison, 1990). The nonlinear behavior of the gradient fields also affects the measurement of diffusion coefficients. Algorithms that attempt to correct image distortions due to inhomogeneity of field gradients (cf. Bakker et al., 1992) generally do not account for an incorrect position or variable thickness of slices, and the effect on diffusion measurements is also not removed. Systematic error due to noise Since DWI relies on signal attenuation it benefits considerably from averaging. But in addition to this noise can also introduce systematic errors: the anisotropy of diffusion depends on the SNR in the diffusion-weighted images (Pierpaoli and Basser, 1996). This is rarely a problem as long as sufficient averaging is performed. However, when comparing anisotropy values in high and low signal regions of an object this dependence should be considered. Registration and normalization of tensor data sets Studies which compare diffusion tensor derived tissue properties between patient and control groups encounter difficulties arising from normalization

Artifacts and pitfalls in diffusion MR imaging

problems. In addition to common problems that occur when a pathologically altered brain is deformed (normalized) to match a standard normal brain, care has to be taken to preserve the tensor information in the normalization process, since the local deformations involved in normalization can contain a rotation component (Alexander and Gee, 2000). If the imaging data are rotated in the normalization process or during co-registration to account for subject motion during the scan, the gradient vectors used in the tensor calculation have to be rotated accordingly. This is relatively easy to perform for co-registration while it can be virtually impossible in normalization algorithms. However, the error related to the local rotation part in normalization is often negligible. REFERENCES Alexander AL, Tsuruda JS, Parker DL. 1997. Elimination of eddy current artifacts in diffusion-weighted echo-planar images: the use of bipolar gradients. Magn Reson Med 38: 1016–1021. Alexander AL, Hasan KM, Lazar M, Tsuruda JS, Parker DL. 2001. Analysis of partial volume effects in diffusion-tensor MRI. Magn Reson Med 45: 770–780. Alexander DC, Gee JC. 2000. Elastic matching of diffusion tensor images. Comput Vis Image Underst 77: 233–250. Anderson AW, Gore JC. 1994. Analysis and correction of motion artifacts in diffusion weighted imaging. Magn Reson Med 32: 379–387. Anderson AW, Gore JC. 1995. Using spiral navigator echoes to correct for motion in diffusion weighted imaging. In: Proceedings of the SMR/ESMRMB, 3rd/12th Annual Meeting, Nice, p. 743. Atkinson D, Porter DA, Hill DLG, Calamante F, Connelly A. 2000. Sampling and reconstruction effects due to motion in diffusion-weighted interleaved echo planar imaging. Magn Reson Med 44: 101–109. Bakker CJG, Moerland MA, Bhagwandien R, Beersma R. 1992. Analysis of machine-dependent and object-induced geometric distortion in 2DFT imaging. Magn Reson Imaging 10: 597–608. Basser PJ. 1995. Inferring microstructural features and the physiological state of tissues from diffusion-weighted images. NMR Biomed 8: 333–344. Basser PJ, Mattiello J, LeBihan D. 1994. Estimation of the effective self-diffusion tensor from the NMR spin echo. J Magn Reson B 103: 247–254.

Bastin ME. 1999. Correction of eddy current-induced artefacts in diffusion tensor imaging using iterative cross-correlation. Magn Reson Imaging 17: 1011–1024. Butts K, de Crespigny A, Pauly JM, Moseley M. 1996. Diffusionweighted interleaved echo-planar imaging with a pair of orthogonal navigator echoes. Magn Reson Med 35: 763–770. Butts K, Pauly J, Crespigny Ad, Moseley M. 1997. Isotropic diffusion-weighted and spiral-navigated interleaved EPI for routine imaging of acute stroke. Magn Reson Med 38: 741–749. Cho ZH, Ahn CB, Kim JH, Lee YE, Mun CW. 1987. Phase error corrected interlaced echo planar imaging. In: Proceedings of the SMRM, 6th Annual Meeting, New York, p. 912. Conturo TE, McKinstry RC, Aronovitz JA, Neil JJ. 1995. Diffusion MRI: precision, accuracy and flow effects. NMR Biomed 8: 307–332. Fu ZW, Wang Y, Grimm RC, Rossman PJ, Felmlee JP, Riederer SJ, Ehman RL. 1995. Orbital navigator echoes for motion measurements in magnetic resonance imaging. Magn Reson Med 34: 746–753. Gibbs SJ, Johnson CS, Jr. 1991. A PFG NMR experiment for accurate diffusion and flow studies in the presence of eddy currents. J Magn Reson 93: 395–402. Haselgrove JC, Moore JR. 1996. Correction for distortion of echo-planar images used to calculate the apparent diffusion coefficient. Magn Reson Med 36: 960–964. Heid O. 2000. Eddy current-nulled diffusion weighting. In: Proceedings of the ISMRM, 8th Annual Meeting, Denver, p. 799. Horsfield MA. 1999. Mapping eddy current induced fields for the correction of diffusion-weighted echo planar images. Magn Reson Imaging 17: 1335–1345. Jezzard P, Barnett AS, Pierpaoli C. 1998. Characterization of and correction for eddy current artifacts in echo planar diffusion imaging. Magn Reson Med 39: 801–812. Kärger J, Pfeifer H, Heink W. 1988. Principles and application of self-diffusion measurements by nuclear magnetic resonance. Adv Magn Reson 12: 1–89. Koch M, Norris DG. 2000. An assessment of eddy current sensitivity and correction in single-shot diffusion-weighted imaging. Phys Med Biol 45: 3821–3832. Le Bihan D, Breton E, Lallemand D, Grenier P, Cabanis E, Laval-Jeantet M. 1986. MR imaging of intravoxel incoherent motions: application to diffusion and perfusion in neurologic disorders. Radiology 161: 401–407. Le Bihan D, Breton E, Lallemand D, Aubin M-L, Vignaud J, Laval-Jeantet M. 1988. Separation of diffusion and perfusion in intravoxel incoherent motion MR imaging. Radiology 168: 497–505. Mansfield P. 1977. Multi-planar image formation using NMR spin echoes. J Phys C 10: L55–L58.

107

108

Martin A. Koch and David G. Norris

Mattiello J, Basser PJ, Le Bihan D. 1994. Analytical expressions for the b matrix in NMR diffusion imaging and spectroscopy. J Magn Reson A 108: 131–141. Mattiello J, Basser PJ, Le Bihan D. 1997. The b matrix in diffusion tensor echo-planar imaging. Magn Reson Med 37: 292–300. McKinnon GC. 1993. Ultrafast interleaved gradient-echo– planar imaging on a standard scanner. Magn Reson Med 30: 609–616. Mori S, van Zijl PCM. 1995. Diffusion weighting by the trace of the diffusion tensor within a single scan. Magn Reson Med 33: 41–52. Neeman M, Freyer JP, Sillerud LO. 1990. Pulsed-gradient spinecho diffusion studies in NMR imaging. Effects of the imaging gradients on the determination of diffusion coefficients. J Magn Reson 90: 303–312. Neeman M, Freyer JP, Sillerud LO. 1991. A simple method for obtaining cross-term-free images for diffusion anisotropy studies in NMR microimaging. Magn Reson Med 21: 138–143. Norris DG. 2001a. The effects of microscopic tissue parameters on the diffusion weighted magnetic resonance imaging experiment. NMR Biomed 14: 77–93. Norris DG. 2001b. Implications of bulk motion for diffusionweighted imaging experiments: effects, mechanisms, and solutions. J Magn Reson Imaging 13: 486–495. Norris DG, Hutchison JMS. 1990. Concomitant magnetic field gradients and their effects on imaging at low magnetic field strengths. Magn Reson Imaging 8: 33–37. Ordidge RJ, Helpern JA, Qing ZX, Knight RA, Nagesh V. 1994. Correction of motional artifacts in diffusion-weighted MR images using navigator echoes. Magn Reson Imaging 12: 455–460. Pierpaoli C, Barnett A, Pajevic S, Chen R, Penix L, Virta A, Basser P. 2001. Water diffusion changes in Wallerian degeneration and their dependence on white matter architecture. Neuroimage 13: 1174–1185. Pierpaoli C, Basser PJ. 1996. Toward a quantitative assessment of diffusion anisotropy. Magn Reson Med 36: 893–906 (Erratum in Magn Reson Med 37: 972 (1997)). Reese TG, Weisskoff RM, Wedeen VJ. 1998. Diffusion NMR facilitated by a refocused eddy-current EPI pulse sequence. In: Proceedings of the ISMRM, 6th Annual Meeting, Sydney, p. 663. Ries M, Jones RA, Dousset V, Moonen CTW. 2000. Diffusion tensor MRI of the spinal cord. Magn Reson Med 44: 884–892.

Schmitt F, Stehling MK, Turner R. 1998. Echo-Planar Imaging. Theory, Technique and Application. Springer, Berlin. Shimony JS, McKinstry RC, Akbudak E, Aronovitz JA, Snyder AZ, Lori NF, Cull TS, Conturo TE. 1999. Quantitative diffusiontensor anisotropy brain imaging: normative human data and anatomic analysis. Radiology 212: 770–784. Smart SC, Porter DA, Calamante F, Hall-Craggs MA, Connelly A. 1999. Eddy current compensation in diffusion-weighted, stimulated echo EPI. In: Proceedings of the ISMRM, 7th Annual Meeting, Philadelphia, pp. 1832. Stejskal EO, Tanner JE. 1965. Spin diffusion measurements: spin echoes in the presence of a time-dependent field gradient. J Chem Phys 42: 288–292. Stieltjes B, Kaufmann WE, van Zijl PCM, Fredericksen K, Pearlson GD, Solaiyappan M, Mori S. 2001. Diffusion tensor imaging and axonal tracking in the human brainstem. Neuroimage 14: 723–735. Tuch DS, Reese TG, Wiegell MR, Makris N, Belliveau JW, Wedeen VJ. 2002. High angular resolution diffusion imaging reveals intravoxel white matter fiber heterogeneity. Magn Reson Med 48: 577–582. Tuch DS, Weisskoff RM, Belliveau JW, Wedeen VJ. 1999. High angular resolution diffusion imaging of the human brain. In: Proceedings of the ISMRM, 7th Annual Meeting, Philadelphia, p. 321. van Gelderen P, de Vleeschouwer MH, DesPres D, Pekar J, van Zijl PCM, Moonen CT. 1994. Water diffusion and acute stroke. Magn Reson Med 31: 154–163. Wald L, Schmitt F, Dale A. 2001. Systematic spatial distortion in MRI due to gradient non-linearities. Neuroimage 13: S50. Wedeen VJ, Reese TG, Tuch DS, Weigel MR, Dou J-G, Weiskoff RM, Chessler D. 2000. Mapping fiber orientation spectra in cerebral white matter with Fourier-transform diffusion MRI. In: Proceedings of the ISMRM, 8th Annual Meeting, Denver, p. 82. Wider G, Dötsch V, Wüthrich K. 1994. Self-compensating pulsed magnetic-field gradients for short recovery times. J Magn Reson A 108: 255–258. Wiegell MR, Larsson HBW, Wedeen VJ. 2000. Fiber crossing in human brain depicted with diffusion tensor MR imaging. Radiology 217: 897–903. Wong EC, Cox RW, Song AW. 1995. Optimized isotropic diffusion weighting. Magn Reson Med 34: 139–143.

7

Cerebral perfusion imaging by exogenous contrast agents Leif Østergaard Department of Neuroradiology, Center for Functionally Integrative Neuroscience, Århus University Hospital, Århus, Denmark

Key points • Exogenous contrast material can give measures of cerebral blood flow, cerebral blood volume and mean transit time. • Spin-echo sequences may require higher concentrations of contrast material compared with gradient-echo (GE) techniques but are more sensitive to contrast material in small vessels as they are less sensitive to flow in larger vessels. • GE sequences are sensitive to changes in all vessels. • Modeling is more complex in the presence of blood–brain barrier breakdown and can benefit from pre-loading.

Contrast mechanisms To derive haemodynamic parameters from dynamic MR images by tracer kinetic analysis, the contrast agent concentrations in various tissue compartments must be known. For a given pulse sequence (e.g. spin-echo, SE or gradient-echo, GE EPI) the relation between observed signal changes during the contrast agent bolus passage and the corresponding concentration is hence of central importance. The study of contrast mechanisms in biological tissue is complex and continues to be an active field of research. In the following, findings of importance to the application and understanding of perfusion imaging are described in some detail.

Susceptibility contrast Introduction Perfusion measurements by dynamic susceptibility contrast imaging (DSCI) MR imaging (MRI) utilizes very rapid imaging (most commonly by echo planar imaging, EPI) to capture the first pass of intravenously injected paramagnetic contrast agent. By kinetic analysis of these data, haemodynamic indices, namely cerebral blood flow (CBF), cerebral blood volume (CBV) and mean transit time (MTT) can be derived. This provides an important supplement to structural images in a number of diseases, and a non-invasive, radiation-free alternative to classical techniques to determine tissue perfusion such as positron emission tomography (PET), single photon emission computed tomography (SPECT) and stable xenon-enhanced computed tomography (Xe-CT).

Cerebral perfusion imaging is most commonly carried out using DSCI, tracking the passage of a rapidly injected paramagnetic gadolinium (Gd)-based chelate by a T2 SE or T2*-weighted GE sequence (often EPI). In the brain, the first-pass extraction of contrast agent is zero when the blood–brain barrier (BBB) is reasonably intact, and the intravascular compartmentalization of contrast agent creates large, microscopic susceptibility gradients, and the dephasing of spins as they diffuse among these result in signal loss in T2- and T2*-weighted images, as first described by Villringer et al. (1988). The microscopic susceptibility gradients caused by the contrast agent cause signal loss in a manner that depends on the type and the timing of the applied pulse sequence. Whereas pulse sequences without full refocusing of 109

Leif Østergaard

GE, TE  60

SE, TE  100

8.0

4.0 Gd: 0.2 mmol/kg

Gd: 0.2 mmol/kg 3.0

6.0

4.0

Gd: 0.1 mmol/kg

ln(SA)/TE

ln(SA)/TE

110

2.0 Gd: 0.1 mmol/kg 1.0

2.0 Deoxyhemoglobin 0.0 1.0

10.0

Deoxyhemoglobin 0.0 1.0

10.0 Vessel radius (m)

Vessel radius (m)

Fig. 7.1 Change in R*2 and R2 for GE (echo times, TE 60 ms) and SE (TE 100 ms) sequences, respectively, as a function of vessel size. Curves are shown for typical Gd doses used in cerebral perfusion imaging (0.1–0.2 mmol/kg) and typical cerebral vascular volume fraction. Also shown are corresponding curves for the weakly paramagnetic deoxyhemoglobin, used in functional MRI. Whereas R2 peaks for small vessel sizes, R*2 reaches a plateau and remains equally sensitive to all vessel sizes. Also note R*2 exceed R2 for all vessel sizes (Boxerman et al., 1995).

static field inhomogeneities (as a GE sequence) will experience a general signal loss due to the presence of microscopic field perturbers largely irrespective of the vessel in which the contrast agent is compartmentalized, the signal loss is smaller (or less) for SE, where dephasing due to static field inhomogeneities is fully refocused SE. For the SE sequence, signal loss is observed at long echo times (TE), as time is sufficient for water to diffuse through areas of different magnetic fields (contrast filled vessels) and thereby dephase. Also, as signal loss is most pronounced if spins diffuse across susceptibility gradients, contrast agents contribute more to the signal loss when localized to vessels with a size that allow the most spins to diffuse across the field inhomogeneities they cause (mostly small vessels – cf. below). This diffusion-related signal loss is therefore a complex function of TE, the density of distribution of vessel sizes, and the concentration and magnetic properties of the contrast agent. Weisskoff, Boxerman and co-workers performed a detailed analysis of these effects in the context of cerebral physiology, using Monte Carlo modeling as well as experimental data (Fisel et al., 1991; Weisskoff et al., 1994b; Boxerman et al., 1995). Two results are particularly important in terms of interpreting perfusion imaging.

Firstly, due to the properties of susceptibility contrast mechanisms, SE measurements are mainly sensitive to vessel sizes comparable to the water diffusion length during the time-of-echo (⬃10 m), whereas GE measurements are equally sensitive to all vessel sizes (cf. Figure 7.1). Experimentally and clinically, this has the effect that to create similar signal losses during first pass, twice the amount of contrast agent (usually double dose of standard Gd chelate, 0.2 mmol/kg) must be injected if imaging is performed using SE EPI, relative to imaging with GE EPI (where 0.1 mmol/kg is generally injected). In return for this, the SE theoretically yields higher sensitivity in detecting changes in small vessel density. Preliminary studies suggest that in the brain, the microvascular CBV “visible” by SE EPI is roughly 45% of the “total” CBV as observed by PET (Østergaard et al., 1998b) or GE EPI (Simonsen et al., 1999). Secondly, an approximate linear relationship exists between tissue contrast agent concentration and change in T2 relaxation rate: R2 (t ) ∝ C t (t )

(7.1)

For GE and SE EPI sequences, signal intensity depends on an exponential fashion upon the

Cerebral perfusion imaging by exogenous contrast agents

transverse and longitudinal relaxation rate changes, R2 and R1:

(

(

)) (

)

S (t )  S (t 0 ) 1 exp TR ⋅ R1 (t ) exp TE ⋅ R2 (t )

(7.2) Assuming that R1 remains constant (i.e. the small enhancement due to shortening of blood T1 by the contrast agent yields the relation between concentration and signal intensity: ⎛ S (t ) ⎞ C t (t )  k ⋅ log ⎜ ⎟ TE ⎝ S (t 0 ) ⎠

(7.3)

The assumption of linearity in Eq. (7.1) has been confirmed by the simulations above and by indirect measurements in vivo (Simonsen et al., 1999) and is widely used in perfusion measurements, where S(t0) is determined from the baseline signal in the images prior to the contrast bolus arrival. In a recent simulation study, Kiselev et al. found that due to the complex physics of MR signal formation in perfused tissues (Kiselev and Posse 1998, 1999) the linearity in Eq. (7.3) may not hold for all ranges of contrast agent concentrations or tissues. This non-linearity may cause overestimation of perfusion estimates (Kiselev, 2001)

Bolus tracking: cerebral blood volume measurements Rosen and co-workers in the early 1990s (Rosen et al., 1990, 1991a, 1991b; Belliveau et al., 1991) derived maps of relative CBV by kinetic analysis of the concentration time curves (cf. above) while dynamically tracking the passage of a bolus high-susceptibility contrast agent. Below, the analysis of bolus tracking experiments for determining first CBV and, in a subsequent section, CBF and MTT, will be reviewed. The vascular volume fraction can be assessed by dynamically imaging the passage of a intravascular tracer, so-called bolus tracking, irrespective of modality (dynamic CT is also suited for this purpose) with high-temporal resolution (upon a standard 5 ml/s injection into an antecubital vein, the tissue bolus passage duration is of the order of 12–20 s in

adults). With EPI, a typical choice is repetition time (TR)  1.5 s or faster. With current high-performance gradient systems, this allows acquisition of 10–15 slices (typically with a spatial resolution of roughly 1.5 mm in-plane, 5–6 mm slice thickness) imaged for every TR, providing good brain coverage (cf. Figure 7.2). By detecting the arterial as well as the total tissue concentration as a function of time during a single transit, the CBV can be determined from the ratio of the areas under the tissue and arterial concentration time curves, respectively (Meier and Zierler, 1954; Zierler, 1962, 1965; Stewart, 1894). ∞

∫ C t (  )d CBV  ∞∞ ∫ ∞Ca (  )d

(7.4)

with concentrations determined from Eq. (7.3). As absolute arterial concentration measurements (due to limited spatial resolution) are not readily quantifiable, relative CBV values are usually reported, simply integrating the area under the concentration time curve (Rosen et al., 1990, 1991a, 1991b), occasionally by the use of a gamma variate function to correct for tracer recirculation (Thompson et al., 1964). In a recent report, Perkiö et al. (2002) concluded that numerically integrating the area of the tissue curve (over the full time range for which it was imaged) or integrating the area of the deconvolved tissue impulse response curve (cf. below) represent the most accurate methods of determining relative CBV.

The residue function: cerebral blood flow The analysis of residue data (i.e. the tracer concentration in tissue after a venous injection has reached the tissue through the feeding artery) is most easily understood by first considering a simple experiment where tracer is injected directly into the feeding artery of a tissue element. To describe the tissue retention of tracer, the so-called residue function is introduced: It measures the fraction of tracer present in the vasculature at time t after injection. So the residue is a decreasing function of time, R(0)  1

111

Leif Østergaard

(a)

(d)

MTT

Raw EPI

Signal (c)

CBF

CBV

GE EPI TR/TE 1500/45 ms 0.1 mmol/kg Gadovist at 5 ml/s

(b)

Concentration

112

Tissue Artery

(e)

Artery Tissue

Time

MTT CBF SE EPI TR/TE 1500/75 ms 0.2 mmol/kg Gadovist at 5 ml/s

CBV

Fig. 7.2 Typical bolus tracking experiment. In the raw EPI images (a), large signal losses are observed upon intravenous contrast injection, as the contrast agent appears in large vessels and tissue (b). Tissue signal loss varies according to regional blood volume, lowest in white matter (WM) and much higher in gray matter (GM) and major vessels. By converting signal curves into concentration time curves by Eq. (7.3). (c), maps of relative CBV can be generated (far right in rows d and e) by determining the area under the concentration time curve. CBF is determined in each pixel by deconvolution of the tissue concentration time curve by the arterial curve. The study compares SE and GE EPI-based perfusion maps. Note the difference in contrast agent dose and the less pronounced major vessels in the SE EPI images (arrows). Also note that SE EPI is less susceptible to susceptibility effects at air-tissue interfaces, clearly seen near the frontal sinuses (parallel arrows). Data courtesy of Professor Gyldensted, Århus University Hospital.

and (assuming the tracer is not bound to the vessel walls and thereby “trapped”) R()  0. For an infinitely short lasting injection giving rise to an arterial concentration Ca at time 0, the tissue concentration Ct(t) as a function of time is: C t (t ) = CBF ⋅Ca ⋅ R(t )

In real experiments, the arterial input function (AIF) Ca(t) is distributed in time and the tissue concentration time curves becomes the convolution (“sum” of individual, very short arterial “impulses” above) of the impulse response and the shape of the AIF:

(7.5)

The proportionality with CBF is intuitively clear, as the concentration of contrast agent present in the tissue at a given time is proportional to the amount of blood (with tracer concentration Ca) passing through the tissue element per unit time. CBF R(t) is called the impulse response, as it is the tissue concentration as a result of the aforementioned “impulse” (infinitely short) input.

C t (t ) = CBF ⋅Ca (t ) ⊗ R(t )

(7.6)

In order to derive CBF from this equation, the impulse response has to be determined by deconvolution, essentially fitting CBF R(t) from the experimentally measured arterial and tissue concentration time curves. As R(0)  1, CBF is determined as the initial height of the impulse response function.

Cerebral perfusion imaging by exogenous contrast agents

A number of difficulties arise when solving Eq. (7.6). Due to the experimental noise, the deconvolution is said to be ill posed, meaning that wildly different solutions for the impulse response can result in similar fits to the experimentally determined tissue concentration time curve. The approaches to solve the equation to regionally determine CBF can be divided into two main categories. By modeldependent techniques, specific analytical expressions are chosen to describe the shape of R(t). In the second, model-independent approach, deconvolution is performed in every image pixel, solving Eq. (7.6) for FtR(t) – the rationale being that the vascular retention of tracer described cannot be known a priori with any certainty. Model-dependent approaches have been explored in some depth. Simple models have been shown to provide poor accuracy of CBF estimates if the assumed R(t) does not accurately reflect the actual vascular retention (Østergaard et al., 1996c). There is some evidence that an exponential residue function (equivalent to assuming that the vasculature behaves as a well-mixed compartment) provides CBF estimates in good agreement with the more general, model-independent deconvolution approaches described below (Østergaard et al., 1996b; Marstrand et al., 2001). The model-dependent approaches have the advantage that by choosing sufficiently general residue models, they allow separate modeling of vascular delay (Vonken et al., 1999) and to some extent dispersion that otherwise bias flow estimates by deconvolution approaches (Østergaard et al., 1999). Model-independent approaches usages have, however, become the methods of choice due to their generality (no assumptions regarding vascular retention characteristics) and stability. These deconvolution methods and their shortcomings are therefore shortly reviewed below.

Determining cerebral blood flow and the residue function: deconvolution In this approach, Eq. (7.6) is solved for CBF  R(t) by standard mathematical deconvolution techniques, typically a transform approach, or by a linear algebraic approach.

In the Fourier transform (FT) approach, the convolution theorem of the FT is utilized, namely that the transform of two convolved function equals the product of their individual transforms. Hence, Eq. (7.6) can be solved (Gobbel and Fike, 1994; Rempp et al., 1994):

{

}

{ } ⎧ F {C (t )} ⎫ ⎪ ⎪ ⎬ ⎨ F C t ( ) { } ⎭⎪ ⎩⎪

F CBF ⋅ R(t ) ⊗ Ca (t )  F C t (t ) ⇒ CBF ⋅ R(t ) = F 1

t

(7.7)

a

where F and F 1 denote the discrete and inverse discrete FT, respectively. In the linear algebraic approach, Eq. (7.6) is rewritten into a matrix equation as follows (Valentinuzzi and Montaldo, 1975). Assuming that tissue and arterial concentrations are measured at equidistant time points t1, t2  t1  t, …, tN, the tissue concentration C(tj) at time tj can be be reformulated as a matrix equation by noting: tj

C t (t j )  CBF ∫ C a (  )R(t j  )d 0

≈ CBF ⋅ t ∑i0 C a (ti )R(t j ti ) j

equivalent to ⎛ C a (t 1 ) ( )⎞ ( ) ⎟⎟  CBF t ⎜⎜ C a (t 2 )

⎛ C t t1 ⎜ ⎜ Ct t2 ⎜ .. ⎜C t ⎝ t N

⎟

( )⎟⎠

L

0

( )

Ca t1 ⎜ M M ⎜C (t ) C t a N 1 ⎝ a N

(

)

0

L 0 M M L Ca t1

( )⎞ ( ) ⎟⎟

⎞⎛ R t1 ⎟⎜ ⎟⎜ R t 2 ⎟⎜ M ⎟⎜R t N ⎠⎝

⎟

( ) ( )⎟⎠ (7.8)

which is a standard matrix equation that can, theoretically, be inverted to yield CBF. R(t). Stable solutions to Eqs. (7.7) and (7.8) can only be obtained by applying techniques to suppress experimental noise. For the FT, this is achieved by applying a filter to the higher frequencies in the frequency (transformed) domain, assuming this can be done without loosing physiological information. In the case of matrix equations such as Eq. (7.11), noise is often suppressed by regularization (forcing the solution to satisfy a-priori, user-defined conditions, or

113

114

Leif Østergaard

otherwise be well behaved) (Bronikowski et al., 1983; Rempp et al., 1994) or by singular value decomposition (SVD) (Van Huffel et al., 1987; Østergaard et al., 1996c). (cf. also Liu et al., 1999 for details on noise suppression by SVD). The optimal choice of some transform and linear algebraic approaches was studied by Østergaard et al. using Monte Carlo simulations (Østergaard et al., 1996c). Despite some theoretical drawbacks, the FT approach has the attraction of, theoretically, being insensitive to delays between the AIF and the tissue, as may be observed in cerebrovascular disease. In normal volunteers, the FT approach provides similar CBF estimates to those obtained by the SVD approach (below) (Wirestam et al., 2000). Deconvolution by SVD has been showed to provide reasonably accurate CBF estimates with a remarkable independence upon vascular structure and CBV, even at the signal-to-noise ratio (SNR) of pixel-by-pixel of clinical EPI measurements (Østergaard et al., 1996c). This technique is now widely used, just as some of its shortcomings have been studied. The major disadvantage of this linear algebraic approach is a tendency to underestimate flow when the shape of the AIF is either delayed or dispersed (caused by stenoses or by passage through collaterals due to vascular occlusions) downstream of where it is non-invasively measured (Østergaard et al., 1996b, 1998a; Calamante et al., 2000, 2002). Techniques to circumvent this will be addressed further below. Even though a straight delay of tracer arrival can in theory be accounted for, model-independent approaches cannot distinguish tracer dispersion in feeding vessels from the tracer retention in the capillary bed. Large vessel dispersion will be interpreted as a low flow, although actual tissue flow is normal (Østergaard et al., 1996c). This is a more fundamental limitation that cannot be circumvented unless a specific model of major vessel dispersion is assumed (Østergaard et al., 1999). The deconvolution techniques make no assumptions regarding the vascular structure. Instead, regional vascular transit-time characteristics can be determined along with tissue flow by studying the residue function. The latter itself seems of some significance, as tracer retention may change in some diseases (Østergaard et al., 1999, 2000).

The mean transit time As pointed out by Weisskoff et al., the distinction between MTT and the first moment of the tissue concentration time curve is crucial in attempts to measure transit times using intravascular tracers (Weisskoff et al., 1993). The calculation of MTT cannot be determined without the determination of R(t) (the area under the R(t) curve can be shown to equal MTT) or CBF described above as, by the central volume theorem (Stewart, 1894): MTT 

CBV CBF

(7.9)

Pitfalls Quantification The formalism above produces absolute values for CBF and CBV provided arterial and tissue concentrations are experimentally determined in identical units. This, however, represents a number of practical problems in actual clinical applications. Firstly, the conversion of signal intensities into tracer concentrations represents a problem as mentioned above, partly as the constant in Eq. (7.3) may differ among blood and various tissue types, and that the assumption of linearity may indeed not hold (Kiselev, 2001). Secondly, due to the inherently limited resolution of MRI relative to vessel sizes, the concentrations time curves single vessels without partial volume effects (PVEs) are difficult to obtain from image data. Studies have, however, obtained absolute values in good agreement with accepted flow rates (Schreiber et al., 1998; Wirestam et al., 2000). The problem of obtaining the true amplitude of the arterial input curve in the face of PVEs of small arteries has been addressed in various ways. In one attempt, it was assumed that the injected contrast agent dose and the AIF area were proportional, subsequently using water clearance PET as a calibration method (Østergaard et al., 1998a, 1998b). This approach may, however, be too crude to allow general use in patients with severe cardiac or cerebrovascular disease. In another report,

Cerebral perfusion imaging by exogenous contrast agents

utilizing the fact that the arterial input curve and venous output curve (easily determined in the much larger venous sinuses) must have identical concentration time curve areas, a promising approach to eliminate PVEs and obtain absolute CBF values was introduced by Lin and co-workers (2001). Finally, normalizing flow rates to structures with constant, ageindependent CBF (WM, cerebellum) may prove useful for intersubject comparison purposes, as is customary, e.g. SPECT. Delay and dispersion From the place where the arterial input is recorded and downstream to the actual tissue element (pixel), where CBF is to be measured, the shape of the input may be delayed (due to the finite speed of blood) and dispersed (spread out in time) due to the laminar – and in cases of vascular disease sometimes turbulent – nature of flow between the upstream and tissue arterial inputs. Also, the passage of blood through collateral pathways due to vascular occlusions (chronic or acute as in acute stroke) may result in these two fundamental processes, delay and dispersion that affect the accuracy of most MR perfusion methods. These effects may cause underestimation of CBF (Østergaard et al., 1996b, 1998a; Calamante et al., 2000) and lead to a false impression of prolonged MTT in clinical evaluation of patients with collateral supply in, e.g. acute stroke (Calamante et al., 2002). Refining deconvolution approaches to be less sensitive to tracer delays has therefore become an active field of research, and promising modifications of the algebraic approach have been reported, e.g. a modification of the SVD by Wu et al. (2003) and a (computationally demanding) Gaussian process approach by Andersen et al. (2002). The appropriateness of using a single AIF, despite obvious dispersion of the arterial input curve shape by collateral circulation and stenosis in some regions (e.g. in acute thrombosis) has recently been questioned. Alsop et al. elegantly proposed using the high-spatial resolution of the image data to localize multiple arterial branches throughout the brain (Alsop et al., 2002). Using a computer algorithm, each imaging voxel was subsequently assigning an

arterial input curve based on the signal time course of nearby arterial branches. Although the approach has several practical problems, this may improve the precision of perfusion techniques.

Bolus tracking with a leaky blood–brain barrier Determination of CBV by DSC MRI has proven a useful tool in demonstrating early neoplastic changes in cerebral tumors (Aronen et al., 1994). For high-grade central nervous system (CNS) tumors and in diseases such as multiple sclerosis (MS), the BBB breaks down, and vessels become permeable to standard Gd chelates. The compartmentalization necessary to perform DSC studies partly break down as the T1-shortening properties of contrast agents become prominent (cf. Eq. (7.2) and cf. Figure 7.3). Although the T1 effects may be small in heavily T2*weighted sequences, a pre-dose of Gd-based contrast agent can be applied, saturating the interstitial space and thereby minimizing first-pass clearance during the subsequent bolus passage. Another approach, modeling the simultaneous T1 and T2 enhancement, has been pursued, yielding important information of perfusion as well as BBB integrity. Weisskoff et al. hence introduced a technique for simultaneous mapping of CBV and the BBB permeability to Gd chelates in first-pass studies (Weisskoff et al., 1994a). Due to the complex modeling involved in CBF mapping, simultaneous modeling of BBB permeability, CBV and CBF is less straightforward (Østergaard et al., 1996a).

Other haemodynamic indices The derivation of flow and transit time from bolus tracking requires derivation of arterial input tracer levels. In some cases, this may not be practical, just as the inherent complexity of deconvolution approaches may preclude the use of these techniques in some situations. Parameters derivable directly from the tissue concentration time curves involve time-to-peak (TTP, time from injection to maximum concentration is reached), arrival time (AT, arrival time of tracer in the pixel), full width at half maximum (FWHM) of the tissue bolus shape, and first moment of the peak, as

115

Leif Østergaard

0s

20 s

50 s

(a)

(d) Permeability

Normal tissue

Image number

Image intensity

Image intensity

116

(b)

Tumor

Image number

(c)

Fig. 7.3 Bolus tracking in tissue with BBB breakdown. Upper row (a) shows the dynamic raw images at different phases of the bolus passage. Notice the signal loss in normal tissue (b) due to the T2-shortening effects of vascular contrast agent. In tissue with leaky vessels, this effect is only barely visible, as the T1 effects of leaking contrast agent dominates (c). The effects can be modeled to generate maps of BBB permeability to the contrast agent, as seen in panel (d) (yellow colors correspond to high leakage in the tumor. In areas with no leakage, the map is made transparent to show anatomy by the corresponding T2-weighted image). Data courtesy of Dr. Jim Rabinov.

well. Although the dependence of these parameters on MTT and CBF depends strongly on the vascular structure and the AIF (Weisskoff et al., 1993), they often suffice to delineate pathological changes and provide important qualitative information in many diseases. It appears, however, that the derivation of CBF, CBV and MTT from kinetic principles somewhat improves specificity and sensitivity of clinical studies, facilitating inter- and intrasubject comparisons (Yamada et al., 2002).

REFERENCES Alsop DC, Wedmid A, Schlaug G. 2002. Defining a local input function for perfusion quantification with bolus contrast MRI. In Proc Int Soc Magn Reson Med (ISMRM), New York, p. 659.

Andersen IK, Szymkowiak A, Rasmussen CE, Hanson LG, Marstrand JR, Larsson HB, Hansen LK. 2002. Perfusion quantification using Gaussian process deconvolution. Magn Reson Med 48(2): 351–361. Aronen HJ, Gazit IE, Louis DN, Buchbinder BR, Pardo FS, Weisskoff RM, Harsh GR, Cosgrove GR, Halpern EF, Hochberg FH. 1994. Cerebral blood volume maps of gliomas: comparison with tumor grade and histologic findings. Radiology 191(1): 41–51. Belliveau JW, Kennedy-DNJ, McKinstry RC, Buchbinder BR, Weisskoff RM, Cohen MS, Vevea JM, Brady TJ, Rosen BR. 1991. Functional mapping of the human visual cortex by magnetic resonance imaging. Science 254(5032): 716–719. Boxerman JL, Hamberg LM, Rosen BR, Weisskoff RM. 1995. MR contrast due to intravascular magnetic susceptibility perturbations. Magn Reson Med 34(4): 555–566. Bronikowski TA, Dawson CA, Linehan JH. 1983. Model-free deconvolution techniques for estimating vascular transport functions. Int J Biomed Comput 14(5): 411–429.

Cerebral perfusion imaging by exogenous contrast agents

Calamante F, Gadian DG, Connely A. 2000. Delay and dispersion effects in dynamic susceptibility contrast MRI: simulations using singular value decomposition. Magn Reson Med 466–473. Calamante F, Gadian DG, Connely A. 2002. Quantification of perfusion using bolus tracking magnetic resonance imaging in stroke: assumptions, limitations, and potential implications for clinical use. Stroke 33: 1146–1151. Fisel CR, Ackerman JL, Buxton RB, Garrido L, Belliveau JW, Rosen BR, Brady TJ. 1991. MR contrast due to microscopically heterogeneous magnetic susceptibility: numerical simulations and applications to cerebral physiology. Magn Reson Med 17(2): 336–347. Gobbel GT, Fike JR. 1994. A Deconvolution method for evaluating indicator-dilution curves. Phys Med Biol 39: 1833–1854. Kiselev VG. 2001. On the theoretical basis of perfusion measurements by dynamic susceptibility contrast MRI. Magn Reson Med 46: 1113–1122. Kiselev VG, Posse S. 1998. Analytical theory of susceptibility induced NMR signal dephasing in a cerebrovascular network. Phys Rev Lett 81: 5696–5699. Kiselev VG, Posse S. 1999. Analytical model of susceptibilityinduced MR signal dephasing: effect of diffusion in a microvascular network. Magn Reson Med 41: 499–509. Lin W, Celik A, Derdeyn C, An H, Lee Y, Videen T, Østergaard L, Powers WJ. 2001. Quantitative measurements of cerebral blood flow in patients with unilateral carotid artery occlusion: a PET and MR study. J Magn Reson Imaging 14: 659–667. Liu HL, Pu Y, Liu Y, Nickerson L, Andrews T, Fox PT, Gao JH. 1999. Cerebral blood flow measurement by dynamic contrast MRI using singular value decomposition with an adaptive threshold. Magn Reson Med 167–172. Marstrand JR, Rostrup E, Rosenbaum S, Garde E, Larsson HB. 2001. Cerebral hemodynamic changes measured by gradientecho or spin-echo bolus tracking and its correlation to changes in ICA blood flow measured by phase-mapping MRI. J Magn Reson Imaging 14(4): 391–400. Meier P. Zierler KL. 1954. On the theory of the indicatordilution method for measurement of blood flow and volume. J Appl Phys 6: 731–744. Østergaard L, Chesler D, Weisskoff RM, Sorensen AG, Rosen BR. 1999. Modeling cerebral blood flow and flow heterogeneity from magnetic resonance residue detection. J Cereb Blood Flow Metab 19: 690–699. Østergaard L, Johannsen P, Høst-Poulsen P, VestergaardPoulsen P, Asboe H, Gee AD, Hansen SB, Cold GE, Gjedde A, Gyldensted C. 1998a. Cerebral blood flow measurements by MRI bolus tracking: comparison with [15O]H2O PET in humans. J Cereb Blood Flow Metab 18: 935–940.

Østergaard L, Rabinov JD, Rosen BR, Gyldensted C. 1996a. Simultaneous mapping of cerebral blood flow, cerebral blood volume and blood brain barrier permeability using Gd-based contrast agents. In Proc Int Soc Magn Reson Med (ISMRM), New York, p. 1307. Østergaard L, Smith DF, Vestergaard-Poulsen P, Hansen SB, Gee A, Gjedde A, Gyldensted C. 1998b. Absolute cerebral blood flow and blood volume measured by MRI bolus tracking: comparison with PET Values. J Cereb Blood Flow Metab 18: 425–432. Østergaard L, Sorensen AG, Chesler D, Weisskoff RM, Koroshetz WJ, Wu O, Gyldensted C, Rosen BR. 2000. Combined diffusion-weighted and perfusion-weighted flow heterogeneity magnetic resonance imaging in acute stroke. Stroke 31: 1097–1103. Østergaard L, Sorensen AG, Kwong KK, Weisskoff RM, Gyldensted C, Rosen BR. 1996b. High resolution measurement of cerebral blood flow using intravascular tracer bolus passages. Part II: Experimental comparison and preliminary results. Magn Reson Med 36(5) 726–736. Østergaard L, Weisskoff RM, Chesler DA, Gyldensted C, Rosen BR. 1996c. High resolution measurement of cerebral blood flow using intravascular tracer bolus passages. Part I: Mathematical approach and statistical analysis. Magn Reson Med 36(5): 715–725. Perkiö JP, Aronen HJ, Kangasmaki A, Liu Y, Karonen JO, Savolainen S, Østergaard L. 2002. Evaluation of four postprocessing methods for determination of cerebral blood volume and mean transit time by dynamic susceptibility contrast imaging. Magn Reson Med 47: 973–981. Rempp KA, Brix G, Wenz F, Becker CR, Guckel F, Lorenz WJ 1994. Quantification of regional cerebral blood flow and volume with dynamic susceptibility contrast-enhanced MR imaging. Radiology 193(3): 637–641. Rosen BR, Belliveau JW, Aronen HJ, Kennedy D, Buchbinder BR, Fischman A, Gruber M, Glas J, Weisskoff RM, Cohen MS, Brady TJ. 1991a. Susceptibility contrast imaging of cerebral blood volume: human experience. Magn Reson Med 22(2): 293–299. Rosen BR, Belliveau JW, Buchbinder BR, McKinstry RC, Porkka LM, Kennedy DN, Neuder MS, Fisel CR, Aronen HJ, Kwong KK, Brady TJ. 1991b. Contrast agents and cerebral hemodynamics. Magn Reson Med 19:(2) 285–292. Rosen BR, Belliveau JW, Vevea JM, Brady TJ. 1990. Perfusion imaging with NMR contrast agents. Magn Reson Med 14(2): 249–265. Schreiber WG, Guckel F, Stritzke P, Schmiedek P, Schwartz A, Brix G. 1998. Cerebral blood flow and cerebrovascular reserve capacity: estimation by dynamic magnetic resonance imaging. J Cereb Blood Flow Metab 18: 1143–1156.

117

118

Leif Østergaard

Simonsen CZ, Østergaard L, Vestergaard-Poulsen P, Røhl L, Bjørnerud A, Gyldensted C. 1999. CBF and CBV measurements by USPIO bolus tracking: reproducibility and comparison with Gd-based values. J Magn Reson Imaging 9: 342–347. Stewart GN. 1894. Researches on the circulation time in organs and on the influences which affect it. Parts I–III. J Physiol (London) 15: 1–89. Thompson HK, Starmer F, Whalen RE, McIntosh HD. 1964. Indicator transit time considered as a gamma variate. Circ Res 14: 502–515. Valentinuzzi ME, Montaldo VE. 1975. Discrete deconvolution. Med Biol Eng 13(1): 123–125. Van Huffel S, Vandewalle J, De Roo MC, Willems JL. 1987. Reliable and efficient deconvolution technique based on total linear least squares for calculating the renal retention function. Med Biol Eng Comput 25(1): 26–33. Villringer A, Rosen BR, Belliveau JW, Ackerman JL, Lauffer RB, Buxton RB, Chao YS, Wedeen VJ, Brady TJ. 1988. Dynamic imaging with lanthanide chelates in normal brain: contrast due to magnetic susceptibility effects. Magn Reson Med 6(2): 164–174. Vonken EP, Beekman FJ, Bakker CJ, Viergever MA. 1999. Maximum likelihood estimation of cerebral blood flow in dynamic susceptibility contrast MRI. Magn Reson Med 41: 343–350. Weisskoff RM, Boxerman JL, Sorensen AG, Kulke SF, Campbell TA, Rosen BR. 1994a. Simultaneous blood volume and permeability mapping using a single Gd-based contrast

injection. Society of Magnetic Resonance, San Francisco, p. 279. Weisskoff RM, Chesler D, Boxerman JL, Rosen BR. 1993. Pitfalls in MR measurement of tissue blood flow with intravascular tracers: which mean transit time? Magn Reson Med 29(4): 553–558. Weisskoff RM, Zuo CS, Boxerman JL, Rosen BR. 1994b. Microscopic susceptibility variation and transverse relaxation: theory and experiment. Magn Reson Med 31(6): 601–610. Wirestam R, Andersson L, Østergaard L, Bolling M, Aunola JP, Lindgren A, Geijer B, Holtås S, Ståhlberg F. 2000. Assessment of regional cerebral blood flow by dynamic susceptibility contrast MRI using different deconvolution techniques. Magn Reson Med 43: 691–700. Wu O, Østergaard L, Weisskoff RM, Benner T, Rosen BR, Sorensen AG. 2003. Tracer arrival timing-insensitive technique or estimating flow in MR perfusion-weighted imaging using singular value decomposition with a block-circulant deconvolution matrix. Magn Reson Med 50: 100–110. Yamada K, Wu O, Gonzalez RG, Bakker D, Østergaard L, Copen WA, Weisskoff RM, Rosen BR, Yagi K, Nishimura T, Sorensen AG. 2002. Magnetic resonance perfusion-weighted imaging of acute cerebral infarction: effect of the calculation methods and underlying vasculopathy. Stroke 33: 87–94. Zierler KL. 1962. Theoretical basis of indicator-dilution methods for measuring flow and volume. Circ Res 10: 393–407. Zierler KL. 1965. Equations for measuring blood flow by external monitoring of radioisotopes. Circ Res 16: 309–321.

8

MRI detection of regional blood flow using arterial spin labeling Alan P. Koretsky, S. Lalith Talagala, Shella Keilholz and Afonso C. Silva Laboratory of Functional and Molecular Imaging and NIH MRI Research Facility, National Institute of Neurological Disorders and Stroke, National Institutes of Health, Bethesda, USA

Introduction By the early 1980s MR imaging (MRI) was well on its way to establishing itself as a useful tool for diagnosis of a number of disorders, especially of the central nervous system (CNS) (Atlas, 2002). The reason for the rapid and spectacular success of MRI is the superb soft tissue contrast that imaging water distribution and relaxation times afford. In addition, the non-invasive nature of the technology makes it possible to readily test efficacy. By the end of the 1980s, the great success in generating anatomical images of normal and diseased tissue led a number of groups to seek ways to get functional information from MRI. Indeed, by the early 1990s techniques had been developed that enabled various aspects of tissue function to be assessed. Most important has been blood oxygen level dependent (BOLD) contrast, which enables detection of changes in hemoglobin oxygenation during regional activation of the brain (Kwong et al., 1992; Ogawa et al., 1992). BOLD-based MRI has rapidly grown into a technique that readily enables brain mapping during complex cognitive tasks (Moonen and Bandettini, 1999). Another class of MRI techniques sensitizes images to changes in diffusion of water in tissue (Wesbey et al., 1984) as well as quantifying preferred diffusion directions (Basser et al., 1994). Diffusion-weighted MRI has grown into an important tool for monitoring tissue damage due to ischemia in brain (Warach, 2002), increasing MRI sensitivity to white matter (WM) disorders (Ahrens et al., 1998) and for mapping fiber

orientation in the brain (Mori et al., 2000). Another important way to add functional information to MRI is the class of techniques that enable regional measurement of tissue blood flow or perfusion. The motivation to measure regional perfusion is well established in physiology and medicine. Techniques to measure regional blood flow in animal models such as microspheres (Heyman et al., 1977) and radio-labeled tracers (Reivich et al., 1969) have had a major impact in our understanding of the regulation of microcirculation in normal tissue and changes that occur during a variety of disease processes. A major limitation of these techniques is that, for the most part, they require sacrificing the animal after only one or a few independent measurements of blood flow. Techniques to measure regional blood flow in humans have relied on the wash-in/wash-out kinetics of tracers that can be detected by radiological imaging techniques. Most important have been the use of radio-labeled water detected by positron emission tomography (PET) (Herscovitch, 1989) and regional distribution of inhaled xenon detected by X-ray computerized tomography (Gur et al., 1982). The results from these techniques show a wide range of problems that perfusion imaging can address from functional mapping of active brain regions during cognitive task activation to attempts to detect the development of Alzheimer’s disease. These techniques are limited by low spatial resolution compared to MRI and the inability to make numerous serial measurements due to radiation dose issues. All of these approaches 119

120

Alan P. Koretsky, S. Lalith Talagala, Shella Keilholz and Afonso C. Silva

have been inspirational, offering theoretical frameworks and practical motivation to develop MRI techniques to measure regional perfusion. The goal has been to take advantage of the non-invasive nature of MRI and the very high resolution that can be obtained to make maps of tissue blood flow. Early approaches to measure regional blood flow by MR techniques relied on adapting the welldeveloped class of techniques that measure tissue specific wash-in and wash-out of tracers. Tracers such as deuterium oxide (Ackerman et al., 1987; Kim and Ackerman, 1990) or fluorinated inhalants (Eleff et al., 1988; Detre et al., 1990a) were first detected using MR spectroscopy (MRS) from specified regions and later images were made that enabled estimates of blood flow in brain (Detre et al., 1990b; Pekar et al., 1991) and tumors (Eskey et al., 1992). A major drawback with this class of MR techniques that relied on directly detecting tracers was the low spatial resolution that could be obtained compared to normal MRI. A solution to this problem was to follow the tracer kinetics of MRI contrast agents indirectly through their effects on tissue water relaxation (Rosen et al., 1989; Østergaard et al., 1996). After a rapid bolus of gadolinium chelates, the change in contrast in a tissue can be used to calculate regional blood volume and blood flow at the resolution of standard MRI. This approach has become an important technique for assessing hemodynamics during ischemia in heart and brain (Calamante et al., 1999), and is further described in another chapter in this volume. Another line of thinking led to the notion that endogenous water flowing in blood into a tissue voxel in MRI could be used to measure regional perfusion. MRI relies on being able to distinguish water in different regions of the body using slice selection, frequency encoding or phase encoding. Slice selection techniques and phase encoding techniques have been used to perform MRI angiography (Haacke and Frahm, 1991). These techniques enable the flow of blood in vessels large enough to be resolved to be detected and velocities can be quantified. While MRI angiography can be used to visualize large vessels and quantify blood velocity in large vessels, this class of angiography techniques does not allow measurements of regional blood flow or perfusion, i.e. the bulk delivery of water to a tissue volume. This

is because the smaller vessels delivering blood to specific areas cannot yet be resolved and the spatial pattern of flow is quite complicated in space making it difficult to use MRI angiography techniques. However, MRI angiography techniques demonstrated and developed a number of ways to distinguish flowing water in blood and to manipulate its magnetization or spin label the blood water with respect to non-flowing tissue water. It is well known from highresolution MR that chemical exchange between molecules could be measured if the magnetization of the exchanging pools could be separately manipulated and if the exchange process was on the same time scale as the T1 relaxation time. Magnetization transfer (MT) techniques to measure chemical exchange have been used to measure a number of enzyme reaction rates both in vitro and in vivo and a large variety of strategies have been developed to manipulate magnetization and quantify exchange rates (Koretsky and Weiner, 1984). Water in blood entering a tissue readily exchanges with tissue water making the regional flow of water an exchange process analogous to chemical exchange. As mentioned above, MRI angiographic techniques demonstrated that blood water magnetization could be manipulated separately from tissue water. All that is required to use the thinking from the welldeveloped class of chemical exchange MT techniques to measure regional blood flow is for the exchange time of water from blood to tissue to be in the order of T1. Typically, T1 values in the brain range from 800 ms at 1.5 T to over 2 s at 11.7 T. In brain, measurements of regional blood flows in gray matter (GM) range from 0.50 ml/min/g tissue to as high as 10 ml/min/g tissue, if fully dilated. Assuming a regional perfusion of 1 ml/min/g tissue and assuming all water exchanges with tissue, then approximately 0.017 ml of blood water exchanges per gram of tissue every second. Thus, about 1.7% of the water exchanges on the time scale of T1. It is this small perturbation on the MRI signal which is used to quantify perfusion with arterial spin-labeling (ASL) techniques. While this is a small signal change, stability of MRI systems are now typically well below 1% and spin-labeling strategies are amenable to alternating control images with spin labeled images to account for any instrumental instability, making it possible to

MRI detection of regional blood flow using arterial spin labeling

(a)

Rat brain (b)

Human brain Fig. 8.1 Regional blood flow in the (a) rat brain and (b) human brain measured with continuous arterial spin labeling (CASL). The rat brain image (a) was acquired on an 11.7 T MRI at a resolution of 0.2 0.2 1 mm3. The human brain image is a axial slice from a 3D data set acquired at a resolution of 3 3 5 mm3 on a 3 T MRI. High perfusion in GM leads to higher signal intensity than lower WM regions which have lower flow.

measure these small signal changes and generate perfusion images from a number of tissues. ASL techniques were first demonstrated in the rat brain (Detre et al., 1992; Williams et al., 1992) and were rapidly adopted for the use in the human brain (Edelman et al., 1994; Roberts et al., 1994). In addition to brain, ASL has been used to measure regional perfusion in many other tissues. Figure 8.1 shows examples of perfusion images from a coronal slice of a rat brain and from an axial slice of a human brain, illustrating the quality with which regional blood flow can be measured using ASL. Indeed, while the intensities are related to regional blood flow the

large difference in flow between GM and WM enable a high degree of anatomical information to be obtained from these images. A rich variety of strategies has been developed to label magnetization in blood with respect to tissue water or tissue water with respect to blood. In addition, strategies have been developed to characterize and minimize many of the problems associated with ASL techniques. As is typical with MRI, many of these complicating issues open up possibilities for generating new types of contrast. It is a very exciting time for ASL techniques in MRI. The past decade has seen extensive development of robust strategies to quantify regional perfusion with ASL. Most of the complicating issues that can lead to artifacts in the measurements have been identified and strategies developed to minimize or measure the important parameters to account for these complications. There have been a large number of studies that have demonstrated the potential usefulness of ASL techniques in both human and animal models to visualize a number of disease processes, and study a number of physiological processes. The major reason for our excitement at this stage is that specific ASL MRI techniques are becoming available as applications on commercial MRI scanners. Our expectation is that the next few years will demonstrate whether the ability to measure regional blood flow, repeatedly, at the resolution of MRI, without the need for contrast, will impact the understanding of the regulation of microcirculation in normal and disease states. There have been a number of reviews written about regional blood flow measurements using ASL techniques (Detre et al., 1994; Calamante et al., 1999; Barbier et al., 2001a). Our intention is not to be exhaustive, but to give an overview of the field at this stage with some comments on what we think will be the most promising directions for the future. By far the most applications for ASL have been in the brain. There are a growing number of examples of generating perfusion images of other tissues and these will be briefly mentioned. Finally, some concluding remarks will be made about the future with predictions about the resolution of perfusion images that should be achievable as the next generation of high field magnets come on line for both human and animal work.

121

122

Alan P. Koretsky, S. Lalith Talagala, Shella Keilholz and Afonso C. Silva

Theoretical basis of ASL

CBF • Ma(t)

CBF • Mv(t)

Basic model of ASL The theoretical modeling of MR measurements of perfusion with ASL have been extensively described and reviewed in great detail (Detre et al., 1994; Calamante et al., 1999; Wong et al., 1999; Barbier et al., 2001a). Therefore, only a general description is offered here. Figure 8.2 shows a schematic representation of a typical fraction of tissue and its associated vasculature for the purposes of establishing the ASL model. Blood flowing through the arterial vessels with flow rate cerebral blood flow (CBF) (in units of ml g 1 min 1) reaches the capillary bed, where a fraction E of water exchanges with tissue water in the extravascular space. The remaining fraction (1 E) of arterial water flows to the venous side of the capillary bed without exchanging with tissue water. Also represented in the model is the exchange of tissue water with tissue macromolecules. According to this model, the Bloch equations for the longitudinal magnetization of brain tissue water and macromolecular spins can be written as: dM (t ) M eq M (t )   E f [ M a (t ) M v (t )] dt T1

kfor M (t )  krev M m(t ) dM m(t ) M eq M m(t )  m  kfor M (t ) krev M m(t ) (8.1) dt T1m

where M(t), Mm(t)  tissue water and macromolecular magnetization per gram of brain tissue; Meq, eq Mm  tissue water and macromolecular equilibrium magnetization per gram of brain tissue; Ma(t), Mv(t)  arterial and venous water magnetization per ml of arterial and venous blood, respectively; T1, T1m  tissue water and macromolecular longitudinal relaxation time constants, in s; kfor, krev  MT rate constants between tissue water and macromolecular protons, in s 1; E  water extraction fraction; f  blood flow (perfusion), in ml s 1 g 1. The labeled arterial magnetization, Ma(t), is conveniently described by defining a parameter called eq degree of labeling, , using  (Meq a M a)/2M a ,

CBF • Ma(t)

Arterial H2O

(1 E)·CBF • Ma(t)

Venous H2O

CBF • Mv(t)

E • CBF • Ma(t) exchange kfor Tissue Tissue macromolecules H2O krev Mm(t) M t(t) E • CBF • Mt(t)/λ Fig. 8.2 Schematic representation of an MR image voxel for the purposes of establishing the ASL model. Blood flowing through the arterial vessels with a certain flow rate (in units of ml g 1 min 1) reach the capillary bed, where a fraction E of water exchanges with tissue water in the extravascular space. The remaining fraction (1 E) of arterial water flows to the venous side of the capillary bed without exchanging with tissue water. Also represented in the model is the exchange of tissue water with tissue macromolecules. (Adapted from Silva et al., 1997b.)

where Meq a is the arterial water equilibrium magnetization. When arterial magnetization is saturated, Ma  0 and  0.5. For inversion, M a  Meq a and  1. During transit of blood from the labeling site to tissue, degree of labeling decreases due to relaxation according to  0 exp( /T1a), where 0 is the degree of labeling at the labeling site,  is the blood transit time, and T1a is the longitudinal relaxation time of arterial blood. Assuming water is freely diffusible between blood and tissue, the venous magnetization, Mv(t), can be expressed in terms of tissue magnetization, M(t), as Mv(t)  M(t)/ where  is the tissue–blood partition coefficient for water in ml g 1. Further, at equilibrium, arterial inflow equals venous outflow, and the equilibrium arterial magnetization, Meq a , eq eq / can be expressed as, M a M .. In ASL perfusion methods, the basic strategy is to differentiate inflowing arterial water from water in

MRI detection of regional blood flow using arterial spin labeling

the tissue of interest, or vice versa. For this, the arterial magnetization, Ma(t), is manipulated in different ways to cause changes in the tissue signal, M(t). Exchange between tissue water and arterial water causes a change in magnetization which is proportional to the amount of arterial water which exchanged with tissue, i.e. proportional to perfusion. Eq. (8.1) can be solved for the blood flow according to many different ASL approaches. In the literature, implementations of ASL have been separated according to two main labeling strategies: continuous labeling or pulsed labeling. This separation is arbitrary in the sense that all rely on a similar underlying model. Furthermore, some of the continuous labeling strategies have been used in a pulsed manner and some of the pulsed strategies have been used in a continuous manner. However, it is still useful to divide ASL strategies into these two classes. Continuous arterial spin labeling The continuous arterial spin labeling (CASL) strategy consists of continuously labeling the arterial spins proximally to the region of interest (ROI). The continuous inflow of labeled water leads to the development of a tissue magnetization steady state that is dependent upon perfusion, T1 and the degree of labeling. Continuous labeling of arterial water can be accomplished either by applying a train of proximal radio frequency (RF) pulses (Detre et al., 1992), or by flowdriven adiabatic-fast passage (AFP) (Williams et al., 1992). Since the arterial water magnetization is inverted, the latter approach produces the largest change in tissue magnetization due to inflow of labeled blood. Therefore, CASL is normally performed using flow-driven AFP. Flow-driven AFP involves application of off-resonance RF in the presence of a gradient in the direction of flow (Dixon et al., 1986). The frequency of RF is set to correspond to a proximal location to the ROI. There have been a number of studies that have modeled the detailed process of AFP (Maccotta et al., 1997; Gach et al., 2002; Utting et al., 2003). In CASL, the labeled arterial blood magnetization that exchanges with tissue, Ma(t), is given by: M a(t )  M aeq; 0  t   M a(t )  (1 2 0 exp(  /T1a ))M aeq;   t  (    ) (8.2)

where  is the AFP labeling duration. The magnetization of labeled arterial blood that exchanges with tissue during  and (  ) is time invariant. CASL can be simply implemented with a single volume RF coil that covers not only the region to be imaged, but also the proximal area containing the feeding arteries. This scheme is shown in Figure 8.3(a), in which a labeling plane is defined to contain the main arterial supply to the brain. For example, in rodents, the labeling plane is situated in the neck, while in humans it is situated below the circle of Willis to include both the internal carotid as well as the vertebral arteries. The off-resonance RF radiation used to label the arterial spins also saturate macromolecules, which causes a strong decrease in tissue signal and longitudinal relaxation time due to MT effects. Since MT effects are symmetric in frequency, a control for MT can be achieved if the same offresonance RF radiation is applied to a plane placed distally from the imaging site at an equal distance with respect to the labeling plane (Figure 8.3(a)). A number of studies have investigated the extent of symmetry for control and developed strategies to control for any asymmetries in MT effects (Pekar et al., 1996). When the off-resonance RF is applied distal to the imaging plane, arterial blood is unaffected (  0). The solution of Eq. (8.1) for t   when macromolecules are saturated (Mm(t)  0) gives: ⎧1 f M (t )  T1appM eq ⎨   k f exp( t /T1app )  ⎩T1 ⎫ 2 0 exp( /T1a ) f

(1 exp( [t ]/T1 app ))⎬  ⎭

(8.3) where

1 f 1    kfor .  T1app T1

T1app is the longitudinal relaxation time of tissue in the presence of macromolecular saturation. In Eq. (8.3), labeled blood is assumed to be fully extracted (E  1). Eq. (8.3) shows the reduction in tissue magnetization due to MT and spin labeling as well as reduction in longitudinal relaxations time of tissue to a shorter apparent constant T1app, due to saturation of macromolecules. From Eq. (8.3), blood flow, f, can be measured by subtracting the tissue signal obtained

123

124

Alan P. Koretsky, S. Lalith Talagala, Shella Keilholz and Afonso C. Silva

(a)

Control plane

(b)

Imaging slab

Imaging slice Labeling plane Imaging RF coil

Labeling/ imaging RF coil

Labeling RF coil

Labeling plane

Fig. 8.3 RF coil arrangement for CASL experiments: (a) A single RF coil is used for labeling the arterial spins at a plane proximal to the slice of interest. In this case, off-resonance labeling of arterial blood induces MT effects, which causes a strong decrease in tissue signal and T1, but which can be controlled for in a separate experiment where the off-resonance RF is applied in a plane located symmetrically distally from the slice of interest. (b) MT effects can be avoided with a two-coil approach, which uses a small surface coil to label the carotid arteries. This labeling coil is decoupled from the imaging coil, thus avoiding MT effects. Thus, with the two-coil system, multi-slice or 3D acquisition can be performed without subtraction artifacts, and the control experiment is simply acquired without RF power applied to the labeling coil.

with spin labeling (  0) from a control signal measured without labeling (  0). f

 ⎛ M ⎞ 1 1 ⎜ ⎟ T1app ⎝ M eq ⎠ 2 0 exp( /T1a ) (1 exp( [   ]/T1app ))

(8.4) where M is the difference in tissue magnetization between control and arterial spin labeled conditions after a labeling duration of  (i.e. at t    ). Eq. (8.4) is applicable when the signals are acquired immediately after the labeling duration. In this case, the blood flow values are highly sensitive to the value of blood transit time  and the intra-luminal signal in the larger arteries. These issues can be f

 T1o

minimized by introducing a delay (w) between the labeling period and signal acquisition (Alsop and Detre, 1996). For w  , the expression for f is given by redefining M as the difference between control and label conditions at t      w and multiplying the right hand side of Eq. (8.4) by a factor /( ), where   1 exp( [ ]/Tlapp),   exp( [w ]/T1o ) and   T1o/Tlapp (1 exp ( /T1o))   exp( /T1o) with 1/T1o  1/T1  f/. T1o is the longitudinal relaxation time of tissue in the absence of macromolecular saturation. The effect of incorporating a delay to reduce the sensitivity to transit time is more clearly seen if one assumes that MT effects are absent (cf. below). In this case Tlapp  T1o and f is given by the expression:

⎛ M ⎞ 1 1 1 ⎜ eq ⎟ ⎝ M ⎠ 2 0 exp( [1/T1a 1/T1o ]) exp( w /T1o ) (1 exp(  /T1o ))

Since, in Eq. (8.5),  is multiplied by the difference between the blood and issue relaxation rates, the dependence of flow on transit time is significantly reduced. As mentioned above, one-coil implementation of CASL saturates the macromolecules, causes a

(8.5)

reduction in both signal amplitude as well as in T1, effectively reducing the signal-to-noise ratio (SNR) of the flow measurement. Furthermore, the control plane, illustrated in Figure 8.3(a), is effective only for a single slice parallel to the labeling plane. Clever approaches to allow multi-slice imaging have been

MRI detection of regional blood flow using arterial spin labeling

proposed (Alsop and Detre, 1998; Talagala et al., 1998). These strategies to obtain whole brain perfusion measurements require specialized pulse sequences but can be implemented readily on commercial MRI scanners. However, they do not avoid saturation of tissue macromolecules. A different approach to obtain full tissue coverage and avoid saturation of tissue macromolecules, is to use separate labeling and imaging coils (Silva et al., 1995; Zhang et al., 1995; Zaharchuk et al., 1999; Mildner et al., 2003; Talagala et al., 2004). A schematic of the two-coil CASL setup is shown in Figure 8.3(b). A small surfacecoil is placed over the neck region to label the carotid arteries. As the RF field generated by the small labeling coil does not reach the brain, tissue macromolecules are not saturated, eliminating the reduction in signal intensity and shortening of T1 due to MT effects. Therefore, multi-slice or 3D acquisition can be performed without subtraction artifacts. Since macromolecules are not saturated when using the two-coil system, flow can be calculated using the formulations of Zhang et al. (1995) or, more simply, by using Eq. (8.5). In addition to allowing flow measurements with more extensive coverage than the one-coil implementation of CASL, use of a separate labeling coil significantly reduces RF power deposition, which is a critical consideration for application of CASL in humans at high magnetic fields. In addition, the two-coil approach can also be used to probe more fundamental aspects of water exchange in the brain (Silva et al., 1997a, 1997b). A drawback of the two-coil approach is that it requires specialized hardware not yet widely available. Pulsed arterial spin labeling The pulsed approach to ASL consists of labeling a slab of blood upstream from the ROI with a slabselective RF pulse, and waiting a certain time to allow the labeled blood to mix with tissue prior to acquiring the image. In pulsed arterial spin labeling (PASL), the labeled arterial blood magnetization that exchanges with tissue, Ma(t), is given by: M a(t )  M aeq; 0  t   M a(t )  (1 2 0 exp( t /T1a )) M aeq;   t

(8.6)

In this case, unlike in CASL (Eq. (8.2)), after a time , the magnetization of labeled arterial blood that exchanges with tissue is time dependent. The tissue magnetization after proximal labeling of blood is given by: 2 0 f M eq exp( /T1a )  [exp( (t )/T1a ) exp( (t )/T1o )] (8.7) 1/T1o 1/T1a

M (t )  M eq

Eqs (8.6) and (8.7) are valid if at the time of proximal blood inversion (t  0), the tissue slab is left at equilibrium. Different implementations of PASL correspond to different initial conditions of tissue magnetization at t  0 (Kwong et al., 1992) and all depend on the notion that the measured T1 in tissue has a perfusion component (Detre et al., 1992). At t  0, the tissue in the ROI is saturated in EPI-signal tagging with alternating RF (EPISTAR) (Edelman et al., 1994) and quantitative imaging of perfusion using a single subtraction (QUIPSS) (Wong et al., 1997), inverted in flow-sensitive alternating inversion recovery (FAIR) (Kim, 1995; Kwong et al., 1995; Schwarzbauer et al., 1996) and flow-sensitive alternating inversion recovery with an extra radiofrequency pulse (FAIRER) (Mai et al., 1999a) and left at equilibrium in perfusion imaging by un-inverted flow-sensitive alternating inversion recovery (UNFAIR) (Helpern et al., 1997). More importantly, the PASL implementations differ in the methods used to acquire the control signal without arterial blood labeling. From Eq. (8.7), blood flow can be determined by subtracting a signal acquired with proximal labeling (  0) from a control signal (  0): ⎛ M ⎞ 1 f  ⎜ eq ⎟ ⎝ M b ⎠ 2 0 exp( /T1a ) 1/T1b 1/T1a (8.8) [exp( (TI )/T1a ) exp( (TI )/T1b )] where M is the difference in tissue magnetization between control and arterial spin labeled images acquired at time TI after proximal labeling. Eq. (8.8) is independent of the initial state of the tissue magnetization, and therefore, it is valid for all PASL

125

126

Alan P. Koretsky, S. Lalith Talagala, Shella Keilholz and Afonso C. Silva

implementations. Eq. (8.8) is analogous to Eq. (8.4) obtained for continuous labeling and the calculated flow values are sensitive to the value of blood transit time . In PASL, sensitivity to blood transit time can be minimized by inserting a proximal arterial saturation pulse in between the initial proximal inversion pulse and signal acquisition (Wong et al., 1997). In this case the blood flow is given by: ⎛ M ⎞ 1 f  ⎜ eq ⎟  2 exp(

1 ( /T1a 1/T1b )) M ⎝ b ⎠ 0 1 1/T1b 1/T1a (8.9) exp( w /T1b ) [exp(  /T1a ) exp(  /T1b )] where  is the time between the proximal inversion and saturation pulses and w is the time from proximal saturation to signal acquisition. Similar to Eq. (8.5), (8.9) shows reduced sensitivity to transit time, . The above ASL formulations were derived by using modified Bloch differential equations incorporating arterial inflow, venous outflow, and tissue water interaction with macromolecules similar to tracer kinetic models. These formulations can also be derived using equations involving convolution between inflow of labeled blood and the tissue residue function which incorporates the relaxation and venous outflow effects (Buxton et al., 1998). In addition, other models to incorporate effects of incomplete extraction of labeled water from blood have been proposed (Silva et al., 1997a, 1997b; Hiller et al., 1998; St Lawrence et al., 2000; Ewing et al., 2001; Zhou et al., 2001; Parkes and Tofts, 2002). An issue for PASL has been the effect on quantifying perfusion due to ill defined slice profiles on the pulses used and the advantages of using well-defined RF pulses with sharp profiles has been demonstrated (Yongbi et al., 1999; Schepers et al., 2002). There has not been much work directly comparing PASL and CASL techniques. A disadvantage of PASL is, that if a similar labeling geometry is used, the signal change associated with PASL techniques should be about half those with CASL. The advantage of PASL techniques are the relative ease with which they can be implemented on commercial MRI scanners. A study that directly compared PASL and CASL in human brain demonstrated equal

sensitivity, however, the labeling planes were not in the same place (Wong et al., 1998). More work comparing PASL and CASL is needed. Both CASL and PASL have in common the challenge of detecting a very small signal change. To help eliminate some of the physiological noise associated with the subject, techniques to suppress the control signal intensity have been developed. These background suppressed techniques make use of extra inversion pulses in ASL sequences that leave either the control or spin labeled image intensity close to zero so that there is little residual signal for motion to affect. Background suppressed techniques have been demonstrated for both CASL and PASL (Alsop and Detre, 1999; Ye et al., 2000b; Duyn et al., 2001; Talagala et al., 2004). In summary, a great number of different strategies have been developed to use ASL MRI to measure regional blood flow. Indeed, in the above discussion only the few most popular have been discussed. All ASL techniques rely on a similar model and the fact that tissue perfusion delivers water on the order of T1. The flexibility that MRI offers in terms of manipulating magnetization make it likely that new approaches to ASL will continue to be developed.

Applications of ASL to the brain Development of ASL has occurred primarily in the brain The first demonstration of ASL for measuring regional blood flow was in the rodent brain (Detre et al., 1992; Williams et al., 1992). In this early work, both saturation and inversion were used to label arterial blood flowing to the brain at the level of the neck. Theoretical issues were developed as were experimental approaches to determining factors that affect quantification of regional blood flow such as the effect of transit time from the labeling plane to the detection voxel (Zhang et al., 1992), contribution of label in vessels (Alsop and Detre, 1996), and complex relaxation in tissues due to MT (Zhang et al., 1995; Silva et al., 1997a). All of these factors can affect quantification of regional blood flow. Shortly after demonstration in rodents, ASL was applied to the human brain (Edelman et al., 1994; Roberts et al., 1994;

MRI detection of regional blood flow using arterial spin labeling

Kwong et al., 1995). Below is a summary of the most important issues for ASL that are being experimentally addressed in both animal and human brains. Comparison to other techniques that measure regional blood flow Considering the large amount of work performed with ASL MRI techniques, there are few comparisons to other techniques that measure regional blood flow. This is important to fully establish MRI techniques for measuring blood flow. The few cases where comparisons have been made demonstrate that MRI-based ASL techniques are making quantitatively accurate measurements of regional blood flow. An early paper demonstrated excellent agreement between microspheres and CASL MRI measurement of perfusion in the normal rat brain (Walsh et al., 1994). At low values of regional blood flow induced by ischemia, ASL tended to underestimate flow compared to microspheres. This was probably due to long transit time delays to the ischemic regions that were not fully characterized and taken into account. The underestimation of blood flow is a disadvantage for quantifying perfusion, it may be an advantage for generating larger MRI contrast to detect low flow areas. In other studies, good agreement was found between PASL as compared to iodoantipyrine autoradiographic measures of CBF in rats (Tsekos et al., 1998; Ewing et al., 2003). Two comparisons of regional blood flow measured either by CASL or PASL with PET have been made in human brain (Zaini et al., 1999; Ye et al., 2000a). In both studies there was excellent agreement. A recent study in gerbils compared PASL to hydrogen clearance measurements (Pell et al., 2003). Results agreed quite well, the larger random error of the PASL technique was cited as a drawback. In all cases to date, ASL MRI measurements of regional blood flow agree quite well with other modalities under normal flow conditions. To verify the accuracy of ASL MRI measurements of regional blood flow, there is interest in the precision and reliability of ASL. In one recent study five subjects were studied and within subject coefficients of variation were 5.8% for global flow and 13% for individual regions, if measurements were made within 1 h, and 13% and 14% respectively, for measurements

made over 1 week (Floyd et al., 2003). This indicates excellent reproducibility on ASL MRI even at 1.5 T. Accounting for transit times, extraction fraction and relaxation effects A number of the complicated issues that can affect quantification of regional CBF have been discussed in the theory section. Experiments have been performed in both animal and human brain addressing these factors. As discussed above ASL sequences have been developed that clearly define the transit time interval or minimize the effects of transit time differences (Alsop and Detre, 1996; Luh et al., 1999). A number of strategies have been developed to measure the transit time as part of an ASL study (Zhang et al., 1992; Alsop and Detre, 1996; Barbier et al., 1999; Luh et al., 1999; Wang et al., 2003). Indeed it has been shown that changes in transit times measured using ASL can be used to monitor regions of the brain where flow increases due to task activation (Gonzalez-At et al., 2000). Early in the development of ASL techniques it was appreciated that the amount of label extracted could influence the quantification of regional blood flow especially in the brain. In the rat, the extraction fraction of labeled water was estimated by measuring the apparent diffusion coefficient (ADC) of the labeled water (Silva et al., 1997a) and by measuring relaxation properties of the labeled water with and without magnetization transfer contrast (MTC) (Silva et al., 1997b). A model that incorporates capillary permeability has been proposed for ASL for the human brain (Parkes and Tofts, 2002), however, experiments to directly measure this parameter have yet to be extended to the human brain. This is worth pursuing since measures of extraction fraction or capillary permeability have the potential for assessing the integrity of the blood–brain barrier (BBB) (Silva et al., 1997b). Given that the time it takes for blood to traverse the microcirculation is from 2 to 5 s, most of the label can relax during this time. Therefore, very little labeled water can pass through a brain region. This insight has led to work that demonstrates that, independent of the extraction fraction of water, regional blood flow can be quantified with ASL (St Lawrence et al., 2000; Ewing et al., 2001). At high field, T1 will begin

127

128

Alan P. Koretsky, S. Lalith Talagala, Shella Keilholz and Afonso C. Silva

to approach transit times and thus it may be possible to quantify spin label in veins (Barbier et al., 2001b). Another complication for quantitative interpretation of ASL is measuring and interpreting the right T1 relaxation time. The problem is that many labeling strategies can cause signal loss due to MTC (Wolff and Balaban, 1989; Zhang et al., 1992). Application of off resonance RF, as is required for CASL, causes MTC effects which alter signal intensity and measured T1. Work in the rat brain addressed these issues using a Bloch equation approach in the context of CASL (Zhang et al., 1993). In general, PASL schemes do not cause significant MTC effects and so this problem is usually neglected. This may not be true as PASL is extended to whole brain coverage and experiments are performed at higher magnetic field strengths. There have been a number of studies in the brain addressing how to properly control or eliminate MTC effects with CASL using either special pulse sequences or specialized hardware (Silva et al., 1995; Alsop and Detre, 1998; Talagala et al., 1998, 2004; Zaharchuk et al., 1999; Mildner et al., 2003). A final issue for T1 determination is the effect of partial volume. In general studies in humans are at a resolution where a voxel contains GM, WM and cerebral spinal fluid (CSF) all of which have different T1. A solution to this problem is to use the time course of the small signal decrease that occurs during spin labeling to derive the relevant T1 (Zhang et al., 1992; Barbier et al., 1999). Whole brain coverage For robust clinical application it is important that ASL have whole brain coverage. To date none of the PASL strategies have led to whole brain coverage primarily for technical reasons. The widespread availability of whole body transmit RF coils and head receive coils may make it easier to perform PASL to cover the entire brain. CASL approaches have been demonstrated to give whole brain coverage in both rats and humans (Silva et al., 1995; Alsop and Detre, 1998; Talagala et al., 1998, 2004; Zaharchuk et al., 1999; Mildner et al., 2003). The issue that must be overcome is controlling for MTC effects as discussed above. Two solutions have been applied to the brain. One relies on a separate labeling coil that can be placed at the neck and label arterial blood flowing

into the brain without causing MTC effects (Silva et al., 1995; Zaharchuk et al., 1999; Mildner et al., 2003; Talagala et al., 2004). An example of perfusion images of the human brain obtained from a threedimensional (3D) CASL experiment is shown in Figure 8.4 (Talagala et al., 2004). Control and spin labeled images were acquired using a 3D, background suppressed MRI sequence. A separate coil placed over the neck was used to invert blood in the carotid and vertebral arteries in a manner that avoided MT effects and enabled coverage of the whole brain. The second solution uses specific labeling strategies to properly control for MTC effects (Alsop and Detre, 1998; Talagala et al., 1998). It is not clear which approach will become most widely used. The first requires specialized equipment and the second requires specialized pulses and may not lead to maximal signal to noise. Vessel territory mapping ASL techniques rely on altering the magnetization of blood in arteries feeding the brain. In general this is accomplished by labeling all of the feeding blood supply. However, it is possible to selectively label specific vessels and image the territory that the specific vessel feeds. This idea of territory mapping with ASL was first demonstrated on the rat by selectively labeling one or the other of the carotid arteries and imaging perfusion to each hemisphere (Detre et al., 1994). Regional perfusion that was dependent on the circle of Willis could be readily imaged when one carotid was made ischemic. Extension of the idea to image the territory perfused by a single carotid artery was demonstrated in humans (Zaharchuk et al., 1999). Work aimed at measuring changes in regional blood flow in the visual cortex (VC) relied on selectively labeling the vessels that feed this region of the brain (Talagala and Noll, 1998). Recently, ASL techniques have been used to map the regions perfused by a number of vessels using spatially selective labeling pulses (Davies and Jezzard, 2003). Vessel territory mapping is an exciting and under explored application of ASL. Little is known about the detailed areas fed from major vessels, furthermore, in conditions known to cause regional ischemia it will be interesting to determine if there are changes in

MRI detection of regional blood flow using arterial spin labeling

Fig. 8.4 Perfusion image demonstrating whole brain coverage from an individual obtained using CASL MRI. Ten slices from two different views (axial, top two rows; and sagittal, bottom two rows) from a 3D data set are shown. Control and spin labeled images were acquired using a 3D, background suppressed MRI sequence. A separate coil placed over the neck was used to invert blood in the carotid and vertebral arteries in a manner that avoided MT effects and enabled coverage of the whole brain. Resolution is 3 3 5 mm3 and the total acquisition time for the control and spin labeled images was 10 min. (Adapted from Talagala et al., 2004.)

perfusion territories. ASL MRI makes this class of measurement possible in humans and thus shows great potential for supplying unique information. Imaging blood volume using ASL During the development of ASL to measure regional perfusion of the brain contributions of blood vessels was considered a problem. One way to overcome this problem in the brain is to introduce a post-labeling delay to allow label to leave arteries (Alsop and Detre, 1996). However, as is often the case in MRI, one person’s problem is another’s source of contrast! Recent work is beginning to use the ideas that have been developed in ASL to measure blood volume. In rat brain, the complete time course from the onset of

labeling arterial blood to steady state and return to non-labeled control conditions has been proposed as a potential way to separate arterial, tissue and venous contributions in ASL (Barbier et al., 1999). Using dynamic ASL, voxels in the brain that labeled (and unlabeled) fast were identified as arteries and voxels that were slow to label were identified as veins. Using these reference arterial and venous time courses enabled reasonable arterial and venous blood volumes to be estimated from all voxels (Barbier et al., 2001b). Another approach to measuring blood volume in the brain used PASL to generate signal only from blood (Lu et al., 2003). This was accomplished by selecting appropriate intervals after a 180° pulse to null the tissue signal. Using this method, reasonable changes in blood volume during the activation of

129

130

Alan P. Koretsky, S. Lalith Talagala, Shella Keilholz and Afonso C. Silva

(a)

Left

T2 FSE

(b)

Blocked design

(c)

1 trial (1 activity)

Right

Fig. 8.5 Perfusion-based functional MRI of the human motor cortex. (a) T2-weighted MRI showing the slice from which functional MRI data was obtained. (b) Block averaged perfusion functional MRI during a finger apposition task showing the increase in blood flow in the motor cortex. (c) Perfusion functional MRI from the same task but from a single, 1 s stimulation. Data was obtained from a background suppressed PASL technique on a 1.5 T MRI (Duyn, et al., 2001; Ye, et al., 2000b). Images courtesy of Jeff H. Duyn, NINDS, NIH.

VC were measured. It would be very exciting if both regional blood flow and blood volume changes could be measured using ASL. These initial attempts indicate that this is a promising area of development. ASL-based functional MRI of the brain The ability to measure regional changes in brain hemodynamics as an indicator of altered neural activity has brought major success to MRI. The first study in humans relied on measuring changes in regional blood volume using an MRI contrast agent (Belliveau et al., 1990). This was quickly followed by studies that showed that BOLD functional MRI (fMRI) could detect regional changes in the oxygenation of hemoglobin due to changes in neural activity (Kwong et al., 1992; Ogawa et al., 1992). In their seminal paper, Kwong et al. verified that the BOLD signals represented changes in regional hemodynamics by using ASL to show a change in blood flow in the same areas of the VC that showed a BOLD response (Kwong et al., 1992). This study relied upon the prediction that changes in perfusion caused changes in apparent T1 (Williams et al., 1992). ASL measures of blood flow have been used to monitor regional brain activation for as long as BOLD fMRI. A number of studies have now shown that ASL can be used to measure changes in regional blood flow caused by changes in

neural activity even during tasks with a subtle cognitive component (Detre and Wang, 2002; Silva and Kim, 2003). Figure 8.5 shows perfusion increases in the motor cortex that were detected with a background suppressed PASL technique (Duyn et al., 2001; Ye et al., 2000b). Changes in blood flow are readily and sensitively detected even on the 1.5-T MRI used for this study. In general, the signal change detected by ASL is roughly half that measured with BOLD and therefore, the large majority of brain mapping studies rely on BOLD fMRI. There are a number of issues related to BOLDbased fMRI that indicate there will be a continuing role for perfusion fMRI using ASL. Most important is the issue of properly localizing the area of activation. BOLD fMRI detects oxygenation of venous blood and active regions can be assigned to draining veins. ASL fMRI measures tissue perfusion and so should localize to the area of activation. A recent study supported the better localization of ASL fMRI as compared to BOLD by comparing the T1 of the voxels that were measured to activate by BOLD or perfusion fMRI using PASL (Luh et al., 2000). The voxels that activated with BOLD tended to have long T1 indicative of CSF, presumably surrounding large veins. The voxels that activated with perfusion fMRI tended to have T1 closer to GM, indicating that perfusion fMRI activation sites localized better to GM than BOLD fMRI.

MRI detection of regional blood flow using arterial spin labeling

The issue of properly localizing the blood flow changes that occur with changes in neuronal activity is particularly important at very high resolution. Recently, a number of groups have demonstrated mapping ocular dominance columns in the human VC using BOLD fMRI (Menon et al., 1997; Cheng et al., 2001; Goodyear and Menon, 2001). Similarly, it has been shown that orientation columns could be mapped in the cat VC using BOLD fMRI (Kim et al., 2000). However, the orientation columns could only be properly mapped if the early and small BOLD fMRI changes were used. Steady-state BOLD fMRI responses did not make accurate maps of orientation columns most likely due to signal changes associated with draining veins. There are results in rodent whisker barrel and olfactory bulb that indicate steady-state BOLD responses do make accurate maps of columnar neuronal units (Yang et al., 1996; Kida et al., 2002). Thus, it is an open issue of how accurate BOLD fMRI can be at high resolution. Optical imaging of intrinsic signals of hemoglobin oxygenation/blood volume have also been claimed to be specific, or not, depending on the time after stimulation and the neural system studied (Malonek et al., 1997; Belluscio et al., 2002). There are a number of studies, primarily in the rodent, using radioactive or fluorescent tracers that show that regional CBF is specific to columnar neuronal elements (Cox et al., 1993; Chaigneau et al., 2003). Consistent with this idea, steady-state perfusion responses were shown to map orientation columns properly in the cat using PASL fMRI (Duong et al., 2001). It will be interesting to see if perfusionbased fMRI can map columnar units in human and other animal models. Since steady-state perfusion signals can be used it may be that perfusion fMRI for mapping columns is more sensitive than using early changes in BOLD fMRI. Another area where perfusion fMRI may contribute is in better defining the temporal properties of fMRI responses. BOLD fMRI signals tend to have a delay of a few seconds and take many seconds to peak after the onset of neural activity. BOLD fMRI is primarily sensitive to changes in venous oxygenation; these delays probably represent transit times for blood to move through the microcirculation after activation, and are probably not due to delays in the increase in CBF. There is growing evidence that

perfusion-based fMRI techniques may be able to follow neural activity with a faster response. In the rat it was demonstrated that blood flow as measured with CASL increased by 500 ms after forepaw stimulation, significantly before changes in BOLD fMRI (Silva et al., 2000). Similarly in humans, changes in blood flow have been reported to be earlier than changes in BOLD fMRI signals. Indeed in one study blood flow increases were detected by 1 s, the first time point examined, after a brief visual stimulation (Liu et al., 2002). It will be interesting to see how fast perfusion starts after neural stimulation. The faster time to the onset of perfusion fMRI as compared to BOLD fMRI may become more important as work aimed at using fMRI to order the sequence of events in the brain continues to grow. A third way in which ASL studies of regional blood flow are complementing BOLD fMRI studies are in those cases where a measurement of blood flow complements information from BOLD fMRI. There has been much interest in attempting to calibrate BOLD fMRI changes to quantify regional oxygen consumption during increases in neuronal activity (Kim and Ugurbil, 1997; Hoge et al., 1999). An absolute measure of blood flow using ASL has proven useful in these calibrations. There are cases where it has been difficult to interpret BOLD fMRI changes, and in these cases, blood flow measurements can help with interpretation of BOLD fMRI. For example, there has been much interest in negative BOLD fMRI signals associated with specific tasks. A recent study used ASL to show that the negative BOLD signals were associated with decreases in perfusion, increasing confidence in the conclusion that neural activity had decreased (Shmuel et al., 2002). It is well known that BOLD fMRI time courses are quantitatively affected by vascular effects such as changes in blood volume. Perfusion-based fMRI may be less susceptible to these effects. Recently, time courses of BOLD fMRI and perfusion fMRI using PASL were compared in primary and supplementary motor areas in the human brain (Obata et al., 2004). Responses in the primary motor area were quite similar, however, in secondary areas an initial overshoot and poststimulus undershoot in BOLD time course were absent in the perfusion time course. The overshoot and undershoot in BOLD time course was interpreted to be due to effects of changing blood volume rather

131

132

Alan P. Koretsky, S. Lalith Talagala, Shella Keilholz and Afonso C. Silva

ADC

CBF

Control

1.5 h after embolism

2 h after thrombolysis

5 h after thrombolysis

Fig. 8.6 Coronal images of a rat brain that received thrombolysis treatment after embolism induced occlusion of the right middle cerebral artery. The top row shows diffusion-weighted MRI ADC, while the bottom row shows CBF images acquired with the CASL method. Control images obtained prior to occlusion of the MCA are shown on the left column. Following embolism, a large region of signal hypo-intensity quickly develops on both the diffusion and the CBF maps, indicating ischemia of the right hemisphere. Upon thrombolytic treatment, a significant improvement in ADC and perfusion could be observed. Images courtesy of Professor Mathias Hoehn, Max-Planck Institute for Neurological Research, Cologne, Germany. (Adapted from Franke et al., 2000.)

than due to changes in neural activity because they were absent in the perfusion time courses. ASL MRI assessment of stroke, brain trauma and neurological disorders Most disorders of the brain are associated with changes in regional blood flow. The ability to make perfusion images completely non-invasively at the resolution of MRI has ASL techniques poised to make a major contribution to diagnosing and following the etiology of a number of disorders. Preliminary studies in animals and humans have demonstrated this potential of ASL to contribute to the study of diseases of the brain. There have been quite a number of papers that have demonstrated the usefulness of ASL to characterize pathological changes in blood flow in animal models. For example, regional perfusion during stroke (Franke et al., 2000; Lythgoe et al., 2000; Zaharchuk et al., 2000), traumatic brain injury (TBI) (Forbes et al., 1997), and global ischemia and hypoxia (Xu et al., 2002) have been measured using ASL techniques in

rats. The ability of MRI to measure regional perfusion, define tissue that is at risk with diffusionweighted MRI, and measure permanent damage with T2-weighted MRI is a powerful tool-kit to characterize the etiology of stroke, brain injury or global ischemia/hypoxia. Figure 8.6 shows an example of diffusion weighted and ASL images of the rat brain after regional ischemia and treatment with thrombolytic agents (Franke et al., 2000). The increase in blood flow after treatment is clearly evident. This work illustrates the usefulness of ASL to follow therapeutic interventions in animal models. An interesting recent application of ASL was to define ischemic regions in the rat brain to guide histology. Histology was performed to examine if stem cells added to the peripheral circulation were homing to the stroke (Zhang et al., 2002). A major advantage of ASL to measure regional blood flow is that repeated measurements can be made continuously with temporal resolution as high as 100 ms (Silva and Kim, 1999) and quantitative studies have been performed in the same animals for up to a year (Kochanek et al., 2002). A serious issue for studies

MRI detection of regional blood flow using arterial spin labeling

using ASL to quantify blood flow during stroke, brain injury or global ischemia/hypoxia is that all of the issues that can effect quantification must be considered especially changes in transit time and changes in relaxation properties. For example, it has been shown during studies of TBI that T1 changes, most likely caused by edema during the injury, had to be properly quantified to get accurate measures of blood flow (Hendrich et al., 1999). So far there has been very little work using ASL to characterize stroke or TBI in humans (Siewert et al., 1997; Detre, 2001). There is great interest in having a measure of regional blood flow after a stroke. The majority of studies have used Gadolinium-based contrast for bolus tracking (Østergaard et al., 1996). An advantage of bolus tracking is the speed at which cerebral hemodynamics can be probed. A mismatch in regions that have low flow compared to decreased diffusion in ischemic regions during stroke is proving to be quite useful for deciding on whether thrombolytics should be used (Chalela et al., 2003). The success of the bolus tracking techniques may have slowed the application of ASL to stroke. There has been work comparing contrast bolus with ASL (Wolf et al., 2003). Advantages of ASL that may make it useful in the future for studies of stroke in humans is that, in principle, higher resolution can be obtained. The need is to acquire images rapidly to follow the bolus of contrast limits resolution as compared to ASL where longer acquisition times can be used to get higher resolution. Another advantage of ASL is that repeated measurements can be made indefinitely, as opposed to contrast bolus tracking where dose constraints limit the number of measurements. ASL has been applied to characterize a number of neurological disorders in humans primarily due to the work of Detre and co-workers. They have shown hypo-perfusion in the medial temporal lobe in patients with temporal lobe epilepsy (TLE) when there was no seizure activity, demonstrating a role for ASL in MRI of epilepsy (Wolf et al., 2001). The asymmetric hypo-perfusion agreed with PET measures of hypo-metabolism using deoxyglucose. Recently, ASL has been extended to pediatric populations and the potential usefulness demonstrated for a number of pathologies of the brain (Wang et al., 2003). Figure 8.7 shows ASL MRI perfusion images of

133

400

ml/100 g/min

0 Fig. 8.7 Regional perfusion measured in a 14-year-old patient with epilepsy partialis continua demonstrating the usefulness of ASL in pediatric populations. The arrow points to the region of increased flow due to the continuous seizure activity. A CASL MRI technique was used to acquire the data on a 1.5 T MRI (Wang et al., 2003). Image courtesy of John A. Detre, University of Pennsylvania.

a 14-year-old child with epilepsy partialis continua showing right frontal hyper-perfusion due to the continuous seizures associated with this disorder. An exciting application for ASL may be to Alzheimer’s disease. It has been well established that amyloid plaques can affect blood flow regulation (Iadecola, 2003). Consistent with this, decreases in blood flow in Alzheimer’s patients were detected in a number of regions of the brain that normally accumulate high levels of amyloid using ASL (Alsop et al., 2000). Recent results from Alsop and co-workers are illustrated in Figure 8.8. They used a background suppressed 3D CASL sequence on a 3 T MRI. Figure 8.8(a) shows the average perfusion through sixteen slices in 13 healthy volunteers and 17 mild Alzheimer’s age-matched patients. Figure 8.8(b) shows the areas that show significant decreases in regional blood flow in the patient group, demonstrating that ASL is sensitive to mild Alzheimer’s.

134

Alan P. Koretsky, S. Lalith Talagala, Shella Keilholz and Afonso C. Silva

(a)

(b)

Fig. 8.8 Assessment of regional blood flow changes due to mild Alzheimer’s disease using ASL MRI. (a) Top two rows show regional blood flow from sixteen slices averaged from 13 age-matched healthy controls and the bottom two rows show regional blood flow averaged from 17 age-matched mild Alzheimer’s patients. (b) 3D rendering of human brain with red regions illustrating those areas that showed significant declines in regional blood flow in the Alzheimer’s patients. Data was obtained on a 3 T MRI using a 3D, background suppressed CASL sequence. Images courtesy of D. Alsop, M. Casement, T. Fong, DZ. Press, Beth Israel Deaconess Medical Center.

Conclusions There is continued enthusiasm for measuring regional blood flow at the resolution of MRI. A large variety of ASL techniques has been developed over the past decade. The issues affecting quantification and the problems preventing robust application have been overcome. It is clear that the technology and the applications of ASL are constantly expanding. Many studies of pulmonary and renal perfusion have been performed. ASL has also been used to image perfusion in he heart (Poncelet et al., 1999), skeletal muscle (Raynaud et al., 2001) and tumors (Warmuth et al., 2003) in humans. As with brain perfusion studies with ASL, this work in other tissues has relied on earlier studies in animals. New target tissues for application of ASL continue to be explored, as recent work in ovary (Tempel and Neeman, 1999) and breast (Zhu and Buonocore, 2003) demonstrate. New technological approaches and more detailed models to better represent ASL are also being actively pursued. For example, recent ideas to use velocity encoding to enable ASL to be performed at vessels within the voxel being imaged show great promise for reducing the effects of transit time

(Norris and Schwarzbauer, 1999; Duhamel et al., 2003). There are ideas for development of specialized gradients to optimize ASL (Trampel et al., 2002) and for utilization of advanced visualization strategies (Zhu and Buonocore, 2003). It is remarkable how varied ASL MRI has grown, both in terms of the number of tissues that can be studied, and in the number of strategies that can be used to perform this class of perfusion MRI. A key issue for the future success of ASL is increasing the resolution of ASL MRI to ⬃1 mm3 in humans to match the resolution routinely obtainable by anatomical MRI. At this resolution partial volume problems will become significantly less and subregions of tissues will be more robustly defined. Indeed, at this resolution GM and WM can be separated in the brain. Two developments make us very optimistic that this can be achieved. First, is the development of very sensitive parallel MRI detectors (De Zwart et al., 2002). Applications that speed up MRI acquisition are the primary motivation for developing these coils, however, a major benefit for ASL is the increased sensitivity. The second development is the advent of very high field MRI systems for human use from 7 to 9 T. There has been work

MRI detection of regional blood flow using arterial spin labeling

showing the advantage of going from 1.5 to 3–4 T for ASL (Yongbi et al., 2001; Wang et al., 2002). ASL MRI has been demonstrated at 7 T using PASL on the human brain (Pfeuffer et al., 2002). The advantages of these high fields for ASL are the increased sensitivity obtained and the increased T1. This leads to an increase in signal change associated with ASL due to less loss of magnetization during transit times and longer tissue T1 for label to accumulate. We are optimistic that the combination of better MRI detectors and higher magnetic fields will enable ASL perfusion images to be obtained at resolutions approaching 1 mm3, at least in the brain. ASL MRI has established itself as an important research tool to understanding a wide variety of questions about the physiology and pathophysiology of blood flow in humans and animal models. It is unique among techniques to measure regional blood flow in that ASL MRI can be performed at the resolution of any MRI pulse sequence and it is totally non-invasive and so can be repeated as often as required. The research use of ASL is expected to continue to expand. However, the true promise of new MRI techniques are to have a routine and widespread impact on clinical practice. Robust ASL sequences on clinical scanners are just now becoming commercially available. This should set the stage for a broader assessment of the clinical utility of ASL MRI and possibly a very fruitful second decade for ASL.

REFERENCES Ackerman JJH, Ewy CS, Becker NN, Shalwitz RA. 1987. Deuterium nuclear magnetic resonance measurements of blood flow and tissue perfusion employing 2H2O as a freely diffusible tracer. Proc Natl Acad Sci USA 84(12): 4099–4102. Ahrens ET, Laidlaw DH, Readhead C, Brosnan CF, Fraser SE, Jacobs RE. 1998. MR microscopy of transgenic mice that spontaneously acquire experimental allergic encephalomyelitis. Magn Reson Med 40(1): 119–132. Alsop DC, Detre JA. 1996. Reduced transit-time sensitivity in noninvasive magnetic resonance imaging of human cerebral blood flow. J Cereb Blood Flow Metab 16(6): 1236–1249. Alsop DC, Detre JA. 1998. Multisection cerebral blood flow MR imaging with continuous arterial spin labeling. Radiology 208(2): 410–416.

Alsop DC, Detre JA. 1999. Background suppressed 3D RARE arterial spin labeling perfusion MRI. In Proceedings of the International Society of Magnetic Resonance in Medicine: Seventh Scientific Meeting and Exhibition, Philadelphia, Pennsylvania, p. 601. Alsop DC, Detre JA, Grossman M. 2000. Assessment of cerebral blood flow in Alzheimer’s disease by spin-labeled magnetic resonance imaging. Ann Neurol 47(1): 93–100. Atlas SW. 2002. Magnetic resonance imaging of the brain and spine. (Ed. Atlas III, SW), Williams and Wilkins, Philadelphia, Pennsylvania. Barbier EL, Lamalle L, Decorps M. 2001a. Methodology of brain perfusion imaging J Magn Reson Imaging 13(4): 496–520. Barbier EL, Silva AC, Kim HJ, Williams DS, Koretsky AP. 1999. Perfusion analysis using dynamic arterial spin labeling (DASL). Magn Reson Med 41(2): 299–308. Barbier EL, Silva AC, Kim SG, Koretsky AP. 2001b. Perfusion imaging using dynamic arterial spin labeling (DASL). Magn Reson Med 45(6): 1021–1029. Basser PJ, Mattiello J, LeBihan D. 1994. MR diffusion tensor spectroscopy and imaging. Biophys J 66(1): 259–267. Belliveau JW, Rosen BR, Kantor HL, Rzedzian RR, Kennedy Jr DN, McKinstry RC, Vevea JM, Cohen MS, Pykett IL, Brady TJ. 1990. Functional cerebral imaging by susceptibility-contrast NMR. Magn Reson Med 14(3): 538–546. Belluscio L, Lodovichi C, Feinstein P, Mombaerts P, Katz LC. 2002. Odorant receptors instruct functional circuitry in the mouse olfactory bulb. Nature 419(6904): 296–300. Buxton RB, Frank LR, Wong EC, Siewert B, Warach S, Edelman RR. 1998. A general kinetic model for quantitative perfusion imaging with arterial spin labeling. Magn Reson Med 40(3): 383–396. Calamante F, Thomas DL, Pell GS, Wiersma J, Turner R. 1999. Measuring cerebral blood flow using magnetic resonance imaging techniques. J Cereb Blood Flow Metab 19(7): 701–735. Chaigneau E, Oheim M, Audinat E, Charpak S. 2003. Two-photon imaging of capillary blood flow in olfactory bulb glomeruli. Proc Natl Acad Sci USA 100(22): 13081–13086. Chalela JA, Ezzeddine M, Latour L, Warach S. 2003. Reversal of perfusion and diffusion abnormalities after intravenous thrombolysis for a lacunar infarction. J Neuroimaging 13(2): 152–154. Cheng K, Waggoner RA, Tanaka K. 2001. Human ocular dominance columns as revealed by high-field functional magnetic resonance imaging. Neuron 32(2): 359–374. Cox SB, Woolsey TA, Rovainen CM. 1993. Localized dynamic changes in cortical blood flow with whisker stimulation corresponds to matched vascular and neuronal architecture of rat barrels. J Cereb Blood Flow Metab 13(6): 899–913.

135

136

Alan P. Koretsky, S. Lalith Talagala, Shella Keilholz and Afonso C. Silva

Davies NP, Jezzard P. 2003. Selective arterial spin labeling (SASL): perfusion territory mapping of selected feeding arteries tagged using two-dimensional radiofrequency pulses. Magn Reson Med 49(6): 1133–1142. De Zwart JA, Ledden PJ, Kellman P, van Gelderen P, Duyn JH. 2002. Design of a SENSE-optimized high-sensitivity MRI receive coil for brain imaging Magn Reson Med 47(6): 1218–1227. Detre JA. 2001. MR perfusion imaging of hyperacute stroke. Am J Neuroradiol 22(5): 806–807. Detre JA, Eskey CJ, Koretsky AP. 1990a. Measurement of cerebral blood flow in rat brain by 19F-NMR detection of trifluoromethane washout [published erratum appears In Magn Reson Med 1990 Oct; 16(1): 179] Magn Reson Med 15(1): 45–57. Detre JA, Subramanian VH, Mitchell MD, Smith DS, Kobayashi A, Zaman A, Leigh Jr JS. 1990b. Measurement of regional cerebral blood flow in cat brain using intracarotid 2H2O and 2H NMR imaging. Magn Reson Med 14: 389–395. Detre JA, Leigh Jr JS, Williams DS, Koretsky AP. 1992. Perfusion imaging. Magn Reson Med 23(1): 37–45. Detre JA, Wang J. 2002. Technical aspects and utility of fMRI using BOLD and ASL. Clin Neurophysiol 113(5): 621–634. Detre JA, Zhang W, Roberts DA, Silva AC, Williams DS, Grandis DJ, Koretsky AP, Leigh JS. 1994. Tissue specific perfusion imaging using arterial spin labeling. NMR Biomed 7(1–2): 75–82. Dixon WT, Du LN, Faul DD, Gado MH, Rossnick S. 1986. Projection angiograms of blood labeled by adiabatic fast passage. Magn Reson Med 3(3): 454–462. Duhamel G, de Bazelaire C, Alsop DC. 2003. Evaluation of systematic quantification errors in velocity-selective arterial spin labeling of the brain. Magn Reson Med 50(1): 145–153. Duong TQ, Kim DS, Ugurbil K, Kim SG. 2001. Localized cerebral blood flow response at submillimeter columnar resolution. Proc Natl Acad Sci USA 98(19): 10904–10909. Duyn JH, Tan CX, van Gelderen P, Yongbi MN. 2001. Highsensitivity single-shot perfusion-weighted fMRI. Magn Reson Med 46(1): 88–94. Edelman RR, Siewert B, Darby DG, Thangaraj V, Nobre AC, Mesulam MM, Warach S. 1994. Qualitative mapping of cerebral blood flow and functional localization with echoplanar MR imaging and signal targeting with alternating radio frequency. Radiology 192(2): 513–520. Eleff SM, Schnall MD, Ligeti L, Osbakken M, Subramanian VH, Chance B, Leigh Jr JS. 1988. Concurrent measurement of cerebral blood flow, sodium, lactate, and high-energy phosphate metabolism using 19F, 23Na, 1H and 31P nuclear magnetic resonance spectroscopy. Magn Reson Med 7: 412–424. Eskey CJ, Koretsky AP, Domach MM, Jain RK. 1992. 2H-nuclear magnetic resonance imaging of tumor blood flow: spatial

and temporal heterogeneity in a tissue-isolated mammary adenocarcinoma. Cancer Res 52(21): 6010–6019. Ewing JR, Cao Y, Fenstermacher J. 2001. Single-coil arterial spin-tagging for estimating cerebral blood flow as viewed from the capillary: relative contributions of intra- and extravascular signal. Magn Reson Med 46(3): 465–475. Ewing JR, Wei L, Knight RA, Pawa S, Nagaraja TN, Brusca T, Divine GW, Fenstermacher JD. 2003. Direct comparison of local cerebral blood flow rates measured by MRI arterial spintagging and quantitative autoradiography in a rat model of experimental cerebral ischemia. J Cereb Blood Flow Metab 23(2): 198–209. Floyd TF, Ratcliffe SJ, Wang J, Resch B, Detre JA. 2003. Precision of the CASL-perfusion MRI technique for the measurement of cerebral blood flow in whole brain and vascular territories. J Magn Reson Imaging 18(6): 649–655. Forbes ML, Hendrich KS, Kochanek PM, Williams DS, Schiding JK, Wisniewski SR, Kelsey SF, DeKosky ST, Graham SH, Marion DW, Ho C. 1997. Assessment of cerebral blood flow and CO2 reactivity after controlled cortical impact by perfusion magnetic resonance imaging using arterial spinlabeling in rats., J Cereb Blood Flow Metab 17(8): 865–874. Franke C, Brinker G, Pillekamp F, Hoehn M. 2000. Probability of metabolic tissue recovery after thrombolytic treatment of experimental stroke: a magnetic resonance spectroscopic imaging study in rat brain. J Cereb Blood Flow Metab 20(3): 583–591. Gach HM, Kam AW, Reid ED, Talagala SL. 2002. Quantitative analysis of adiabatic fast passage for steady laminar and turbulent flows. Magn Reson Med 47(4): 709–719. Gonzalez-At JB, Alsop DC, Detre JA. 2000. Cerebral perfusion and arterial transit time changes during task activation determined with continuous arterial spin labeling. Magn Reson Med 43(5): 739–746. Goodyear BG, Menon RS. 2001. Brief visual stimulation allows mapping of ocular dominance in visual cortex using fMRI. Hum Brain Mapp 14(4): 210–217. Gur D, Good WF, Wolfson Jr SK, Yonas H, Shabason L. 1982. In vivo mapping of local cerebral blood flow by xenonenhanced computed tomography. Science 215: 1267–1268. Haacke EM, Frahm J. 1991. A guide to understanding key aspects of fast gradient-echo imaging [editorial]. J Magn Reson Imaging 1(6): 621–624. Helpern JA, Branch CA, Yongbi MN, Huang NC. 1997. Perfusion imaging by un-inverted flow-sensitive alternating inversion recovery (UNFAIR). Magn Reson Imaging 15(2): 135–139. Hendrich KS, Kochanek PM, Williams DS, Schiding JK, Marion DW, Ho C. 1999. Early perfusion after controlled cortical impact in rats: quantification by arterial spinlabeled MRI and the influence of spin-lattice relaxation time heterogeneity. Magn Reson Med 42(4): 673–681.

MRI detection of regional blood flow using arterial spin labeling

Herscovitch P. 1989. Cerebral blood flow and metabolism measured with oxygen-15 radiotracers. J Neuropsychiatr Clin Neurosci 1(1): S19–S29. Heyman MA, Payne BD, Hoffman JI. 1977. Blood flow measurements with radionuclide-labeled particles. Prog Cardiovasc Dis 20: 55–79. Hiller KH, Bock M, Wacker CM, Schad LR, Waller C, Haase A, van Kaick G, Ertl G, Bauer WR. 1998. MR-perfusion measurements: basic methodology and current status. MAGMA 6(2–3): 98–99. Hoge RD, Atkinson J, Gill B, Crelier GR, Marrett S, Pike GB. 1999. Linear coupling between cerebral blood flow and oxygen consumption in activated human cortex. Proc Natl Acad Sci USA 96(16): 9403–9408. Iadecola C. 2003. Cerebrovascular effects of amyloid-beta peptides: mechanisms and implications for Alzheimer’s dementia. Cell Mol Neurobiol 23(4–5): 681–689. Kida I, Xu F, Shulman RG, Hyder F. 2002. Mapping at glomerular resolution: fMRI of rat olfactory bulb. Magn Reson Med 48(3): 570–576. Kim DS, Duong TQ, Kim SG. 2000. High-resolution mapping of iso-orientation columns by fMRI. Nat Neurosci 3(2): 164–169. Kim SG. 1995. Quantification of relative cerebral blood flow change by flow-sensitive alternating inversion recovery (FAIR) technique: application to functional mapping. Magn Reson Med 34(3): 293–301. Kim SG, Ugurbil K. 1997. Comparison of blood oxygenation and cerebral blood flow effects in fMRI: estimation of relative oxygen consumption change. Magn Reson Med 38(1): 59–65. Kim SG, Ackerman JJH. 1990. Quantification of regional blood flow by monitoring of exogenous tracer via nuclear magnetic resonance spectroscopy. Magn Reson Med 14: 266–282. Kochanek PM, Hendrich KS, Dixon CE, Schiding JK, Williams DS, Ho C. 2002. Cerebral blood flow at one year after controlled cortical impact in rats: assessment by magnetic resonance imaging. J Neurotrauma 19(9): 1029–1037. Koretsky AP, Weiner MW. 1984. 31P NMR magnetization transfer measurement of phosphorus reactions, In Biomedical Magnetic Resonance, 1st edn. (Eds. Margulis A, James TL), Radiology Research and Education Foundation, San Francisco, pp. 209–230. Kwong KK, Belliveau JW, Chesler DA, Goldberg IE, Weisskoff RM, Poncelet BP, Kennedy DN, Hoppel BE, Cohen MS, Turner R. 1992. Dynamic magnetic resonance imaging of human brain activity during primary sensory stimulation. Proc Natl Acad Sci USA 89(12): 5675–5679. Kwong KK, Chesler DA, Weisskoff RM, Donahue KM, Davis TL, Ostergaard L, Campbell TA, Rosen BR. 1995. MR perfusion studies with T1-weighted echo planar imaging. Magn Reson Med 34(6): 878–887.

Liu TT, Wong EC, Frank LR, Buxton RB. 2002. Analysis and design of perfusion-based event-related fMRI experiments. Neuroimage 16(1): 269–282. Lu H, Golay X, Pekar JJ, van Zijl PC. 2003. Functional magnetic resonance imaging based on changes in vascular space occupancy. Magn Reson Med 50(2): 263–274. Luh WM, Wong EC, Bandettini PA, Hyde JS. 1999. QUIPSS II with thin-slice TI1 periodic saturation: a method for improving accuracy of quantitative perfusion imaging using pulsed arterial spin labeling. Magn Reson Med 41(6): 1246–1254. Luh WM, Wong EC, Bandettini PA, Ward BD, Hyde JS. 2000. Comparison of simultaneously measured perfusion and BOLD signal increases during brain activation with T(1)based tissue identification. Magn Reson Med 44(1): 137–143. Lythgoe MF, Thomas DL, Calamante F, Pell GS, King MD, Busza AL, Sotak CH, Williams SR, Ordidge RJ, Gadian DG. 2000. Acute changes in MRI diffusion, perfusion, T(1), and T(2) in a rat model of oligemia produced by partial occlusion of the middle cerebral artery. Magn Reson Med 44(5): 706–712. Maccotta L, Detre JA, Alsop DC. 1997. The efficiency of adiabatic inversion for perfusion imaging by arterial spin labeling. NMR Biomed 10(4–5): 216–221. Mai VM, Hagspiel KD, Christopher JM, Do HM, Altes T, KnightScott J, Stith AL, Maier T, Berr SS. 1999a. Perfusion imaging of the human lung using flow-sensitive alternating inversion recovery with an extra radiofrequency pulse (FAIRER). Magn Reson Imaging 17(3): 355–361. Malonek D, Dirnagl U, Lindauer U, Yamada K, Kanno I, Grinvald A. 1997. Vascular imprints of neuronal activity: relationships between the dynamics of cortical blood flow, oxygenation, and volume changes following sensory stimulation. Proc Natl Acad Sci USA 94(26): 14826–14831. Menon RS, Ogawa S, Strupp JP, Ugurbil K. 1997. Ocular dominance in human V1 demonstrated by functional magnetic resonance imaging. J Neurophysiol 77(5): 2780–2787. Mildner T, Trampel R, Moller HE, Schafer A, Wiggins CJ, Norris DG. 2003. Functional perfusion imaging using continuous arterial spin labeling with separate labeling and imaging coils at 3 T. Magn Reson Med 49(5): 791–795. Moonen CT, Bandettini PA. 1999. Functional MRI; 1st. Springer-Verlag, Berlin; Baert AL, Heuck FHW, Youker JE. Medical Radiology. Diagnostic imaging. Mori S, Kaufmann WE, Pearlson GD, Crain BJ, Stieltjes B, Solaiyappan M, van Zijl PC. 2000. In vivo visualization of human neural pathways by magnetic resonance imaging. Ann Neurol 47(3): 412–414. Norris DG, Schwarzbauer C. 1999. Velocity selective radiofrequency pulse trains. J Magn Reson 137(1): 231–236. Obata T, Liu TT, Miller KL, Luh WM, Wong EC, Frank LR, Buxton RB. 2004. Discrepancies between BOLD and flow

137

138

Alan P. Koretsky, S. Lalith Talagala, Shella Keilholz and Afonso C. Silva

dynamics in primary and supplementary motor areas: application of the balloon model to the interpretation of BOLD transients. Neuroimage 21(1): 144–153. Ogawa S, Tank DW, Menon RS, Ellermann JM, Kim S-G, Merkle H, Ugurbil K. 1992. Intrinsic signal changes accompanying sensory stimulation: functional brain mapping with magnetic resonance imaging. Proc Natl Acad Sci USA 89(13): 5951–5955. Ostergaard L, Sorensen AG, Kwong KK, Weisskoff RM, Gyldensted C, Rosen BR. 1996. High resolution measurement of cerebral blood flow using intravascular tracer bolus passages. Part II: Experimental comparison and preliminary results. Magn Reson Med 36(5): 726–736. Parkes LM, Tofts PS. 2002. Improved accuracy of human cerebral blood perfusion measurements using arterial spin labeling: accounting for capillary water permeability. Magn Reson Med 48(1): 27–41. Pekar J, Jezzard P, Roberts DA, Leigh JSJ, Frank JA, McLaughlin AC. 1996. Perfusion imaging with compensation for asymmetric magnetization transfer effects. Magn Reson Med 35(1): 70–79. Pekar J, Ligeti L, Ruttner Z, Lyon RC, Sinnwell TM, van GP, Fiat D, Moonen CTW, McLaughlin AC. 1991. In vivo measurement of cerebral oxygen consumption and blood flow using 17O magnetic resonance imaging. Magn Reson Med 21(2): 313–319. Pell GS, King MD, Proctor E, Thomas DL, Lythgoe MF, Gadian DG, Ordidge RJ. 2003. Comparative study of the FAIR technique of perfusion quantification with the hydrogen clearance method. J Cereb Blood Flow Metab 23(6): 689–699. Pfeuffer J, Adriany G, Shmuel A, Yacoub E, Van De Moortele PF, Hu X, Ugurbil, K. 2002, Perfusion-based high-resolution functional imaging in the human brain at 7 Tesla. Magn Reson Med 47(5): 903–911. Poncelet BP, Koelling TM, Schmidt CJ, Kwong KK, Resse TG, Ledden P, Kantor HL, Brady TJ, Weisskoff RM. 1999. Measurement of human myocardial perfusion by doublegated flow alternating inversion recovery EPI. Magn Reson Med 41(3): 510–519. Raynaud JS, Duteil S, Vaughan JT, Hennel F, Wang C, LeroyWillig A, Carlier PG. 2001. Determination of skeletal muscle perfusion using arterial spin labeling NMRI: validation by comparison with venous occlusion plethysmography. Magn Reson Med 46(2): 305–311. Reivich M, Jehle J, Sokoloff L, Kety SS. 1969. Measurement of regional cerebral blood flow with antipyrine-14C in awake cats. J Appl Physiol 27(2): 296–300. Roberts DA, Detre JA, Bolinger L, Insko EK, Leigh Jr JS. 1994. Quantitative magnetic resonance imaging of human brain perfusion at 1.5 T using steady-state inversion of arterial water. Proc Natl Acad Sci USA 91(1): 33–37. Rosen BR, Belliveau JW, Chien D. 1989. Perfusion imaging by nuclear magnetic resonance. Magn Reson Q 5(4): 263–281.

Schepers J, Garwood M, van der SB, Nicolay K. 2002. Improved subtraction by adiabatic FAIR perfusion imaging. Magn Reson Med 47(2): 330–336. Schwarzbauer C, Morrissey SP, Haase A. 1996. Quantitative magnetic resonance imaging of perfusion using magnetic labeling of water proton spins within the detection slice. Magn Reson Med 35(4): 540–546. Shmuel A, Yacoub E, Pfeuffer J, Van De Moortele PF, Adriany G, Hu X, Ugurbil K. 2002. Sustained negative BOLD, blood flow and oxygen consumption response and its coupling to the positive response in the human brain. Neuron 36(6): 1195–1210. Siewert B, Schlaug G, Edelman RR, Warach S. 1997. Comparison of EPISTAR and T2*-weighted gadoliniumenhanced perfusion imaging in patients with acute cerebral ischemia. Neurology 48(3): 673–679. Silva AC, Kim SG. 1999. Pseudo-continuous arterial spin labeling technique for measuring CBF dynamics with high temporal resolution. Magn Reson Med 42(3): 425–429. Silva AC, Kim SG. 2003. Perfusion-based functional magnetic resonance imaging. Concept Magn Reson Part A 16A(1): 16–27. Silva AC, Lee SP, Iadecola C, Kim SG. 2000. Early temporal characteristics of cerebral blood flow and deoxyhemoglobin changes during somatosensory stimulation. J Cereb Blood Flow Metab 20(1): 201–206. Silva AC, Zhang W, Williams DS, Koretsky AP. 1995. Multi-slice MRI of rat brain perfusion during amphetamine stimulation using arterial spin labeling. Magn Reson Med 33(2): 209–214. Silva AC, Williams DS, Koretsky AP. 1997a. Evidence for the exchange of arterial spin-labeled water with tissue water in rat brain from diffusion-sensitized measurements of perfusion. Magn Reson Med 38(2): 232–237. Silva AC, Zhang W, Williams DS, Koretsky AP. 1997b. Estimation of water extraction fractions in rat brain using magnetic resonance measurement of perfusion with arterial spin labeling. Magn Reson Med 37(1): 58–68. St Lawrence KS, Frank JA, McLaughlin AC. 2000. Effect of restricted water exchange on cerebral blood flow values calculated with arterial spin tagging: a theoretical investigation. Magn Reson Med 44(3): 440–449. Talagala SL, Barbier EL, Williams DS, Silva AC, Koretsky AP. 1998. Multi-slice perfusion MRI using continuous arterial water labeling controlling for MT effects with simultaneous proximal and distal RF irradiation In Proceedings of the International Society for Magnetic Resonance in Medicine: Sixth Scientific Meeting and Exhibition, Sydney, Australia, p. 381. Talagala SL, Noll DC. 1998. Functional MRI using steady-state arterial water labeling. Magn Reson Med 39(2): 179–183. Talagala SL, Ye FQ, Ledden PJ, Chesnick S. 2004. Whole brain 3D perfusion MRI at 3 .0 T using CASL with a separate labeling coil. Magn Reson Med 52(1): 131–140.

MRI detection of regional blood flow using arterial spin labeling

Tempel C, Neeman M. Perfusion of the rat ovary: application of pulsed arterial spin labeling MRI. Magn Reson Med 1999; 41(1): 113–123. Trampel R, Mildner T, Goerke U, Schaefer A, Driesel W, Norris DG. 2002. Continuous arterial spin labeling using a local magnetic field gradient coil. Magn Reson Med 48(3): 543–546. Tsekos NV, Zhang F, Merkle H, Nagayama M, Iadecola C, Kim SG. 1998.Quantitative measurements of cerebral blood flow in rats using the FAIR technique: correlation with previous iodoantipyrine autoradiographic studies. Magn Reson Med 39(4): 564–573. Utting JF, Thomas DL, Gadian DG, Ordidge RJ. 2003. Velocitydriven adiabatic fast passage for arterial spin labeling: results from a computer model. Magn Reson Med 49(2): 398–401. Walsh EG, Minematsu K, Leppo J, Moore SC. 1994. Radioactive microsphere validation of a volume localized continuous saturation perfusion measurement. Magn Reson Med 31: 147–153. Wang J, Alsop DC, Li L, Listerud J, Gonzalez-At JB, Schnall MD, Detre JA. 2002. Comparison of quantitative perfusion imaging using arterial spin labeling at 1.5 and 4.0 Tesla. Magn Reson Med 48(2): 242–254. Wang J, Alsop DC, Song HK, Maldjian JA, Tang K, Salvucci AE, Detre JA. 2003. Arterial transit time imaging with flow encoding arterial spin tagging (FEAST). Magn Reson Med 50(3): 599–607. Warach S. 2002. Thrombolysis in stroke beyond three hours: Targeting patients with diffusion and perfusion MRI. Ann Neurol 51(1): 11–13. Wesbey GE, Moseley ME, Ehman RL. 1984. Translational molecular self-diffusion in magnetic resonance imaging. II. Measurement of the self-diffusion coefficient. Invest Radiol 19(6): 491–498. Williams DS, Detre JA, Leigh JS, Koretsky AP. 1992. Magnetic resonance imaging of perfusion using spin inversion of arterial water. Proc Natl Acad Sci USA 89(1): 212–216. Wolf RL, Alsop DC, Levy-Reis I, Meyer PT, Maldjian JA, Gonzalez-Atavales J, French JA, Alavi A, Detre JA. 2001, Detection of mesial temporal lobe hypoperfusion in patients with temporal lobe epilepsy by use of arterial spin labeled perfusion MR imaging. Am J Neuroradiol 22(7): 1334–1341. Wolf RL, Alsop DC, McGarvey ML, Maldjian JA, Wang J, Detre JA. 2003. Susceptibility contrast and arterial spin labeled perfusion MRI in cerebrovascular disease. J Neuroimaging 13(1): 17–27. Wolff SD, Balaban RS. 1989. Magnetization transfer contrast (MTC) and tissue water proton relaxation in vivo. Magn Reson Med 10(1): 135–144. Wong EC, Buxton RB, Frank LR. 1997. Implementation of quantitative perfusion imaging techniques for functional

brain mapping using pulsed arterial spin labeling. NMR Biomed 10(4,5): 237–249. Wong EC, Buxton RB, Frank LR. 1998. A theoretical and experimental comparison of continuous and pulsed arterial spin labeling techniques for quantitative perfusion imaging. Magn Reson Med 40(3): 348–355. Wong EC, Buxton RB, Frank LR. 1999. Quantitative perfusion imaging using arterial spin labeling. Neuroimaging Clin N Am 9(2): 333–342. Xu Y, Liachenko S, Tang P. 2002. Dependence of early cerebral reperfusion and long-term outcome on resuscitation efficiency after cardiac arrest in rats. Stroke 33(3): 837–843. Yang X, Hyder F, Shulman RG. 1996. Activation of single whisker barrel in rat brain localized by functional magnetic resonance imaging. Proc Natl Acad Sci USA 93(1): 475–478. Ye FQ, Berman KF, Ellmore T, Esposito G, van Horn JD, Yang Y, Duyn J, Smith AM, Frank JA, Weinberger DR, McLaughlin AC. 2000a. H(2)(15)O PET validation of steady-state arterial spin tagging cerebral blood flow measurements in humans. Magn Reson Med 44(3): 450–456. Ye FQ, Frank JA, Weinberger DR, McLaughlin AC. 2000b. Noise reduction in 3D perfusion imaging by attenuating the static signal in arterial spin tagging (ASSIST). Magn Reson Med 44(1): 92–100. Yongbi MN, Fera F, Mattay VS, Frank JA, Duyn JH. 2001. Simultaneous BOLD/perfusion measurement using dualecho FAIR and UNFAIR: sequence comparison at 1.5 T and 3.0 T. Magn Reson Imaging 19(9): 1159–1165. Yongbi MN, Yang Y, Frank JA, Duyn JH. 1999. Multislice perfusion imaging in human brain using the C-FOCI inversion pulse: comparison with hyperbolic secant. Magn Reson Med 42(6): 1098–1105. Zaharchuk G, Ledden PJ, Kwong KK, Reese TG, Rosen BR, Wald LL. 1999. Multislice perfusion and perfusion territory imaging in humans with separate label and image coils. Magn Reson Med 41(6): 1093–1098. Zaharchuk G, Yamada M, Sasamata M, Jenkins BG, Moskowitz MA, Rosen BR. 2000. Is all perfusion-weighted magnetic resonance imaging for stroke equal? The temporal evolution of multiple hemodynamic parameters after focal ischemia in rats correlated with evidence of infarction. J Cereb Blood Flow Metab 20(9): 1341–1351. Zaini MR, Strother SC, Anderson JR, Liow JS, Kjems U, Tegeler C, Kim SG. 1999. Comparison of matched BOLD and FAIR 4.0 T-fMRI with [15O] water PET brain volumes. Med Phys 26(8): 1559–1567. Zhang W, Silva AC, Williams DS, Koretsky AP. 1995. NMR measurement of perfusion using arterial spin labeling without saturation of macromolecular spins. Magn Reson Med 33(3): 370–376.

139

140

Alan P. Koretsky, S. Lalith Talagala, Shella Keilholz and Afonso C. Silva

Zhang W, Williams DS, Detre JA, Koretsky AP. 1992. Measurement of brain perfusion by volume-localized NMR spectroscopy using inversion of arterial water spins: accounting for transit time and cross-relaxation. Magn Reson Med 25(2): 362–371. Zhang W, Williams DS, Koretsky AP. 1993. Measurement of rat brain perfusion by NMR using spin labeling of arterial water: in vivo determination of the degree of spin labeling. Magn Reson Med 29(3): 416–421. Zhang ZG, Zhang L, Jiang Q, Chopp M. 2002. Bone marrowderived endothelial progenitor cells participate in cerebral

neovascularization after focal cerebral ischemia in the adult mouse. Circ Res 90(3): 284–288. Zhou J, Wilson DA, Ulatowski JA, Traystman RJ, van Zijl PC. 2001. Two-compartment exchange model for perfusion quantification using arterial spin tagging. J Cereb Blood Flow Metab 21(4): 440–455. Zhu DC, Buonocore MH. 2003. Breast tissue differentiation using arterial spin tagging. Magn Reson Med 50(5): 966–967.

9

Artifacts and pitfalls in perfusion MR imaging Fernando Calamante Radiology and Physics Unit, Institute of Child Health, University College London, London, UK

Key points Dynamic susceptibility contrast perfusion imaging • Absolute perfusion quantification is difficult due to complex relationship between signal and tracer concentration, and unknown constants. • Error in calculation of perfusion parameters arises from: delay in bolus arrival time (BAT), bolus dispersion, blood–brain barrier breakdown, and voxel shift/motion. • Summary parameters (e.g. time to peak, maximum peak concentration, area under the peak and BAT) are easy to measure but prone to systematic errors. • Deconvolution of arterial input function (AIF) reduces errors. • AIF errors include partial volume and saturation effects. • Bolus delay and dispersion may depend on the type of vascular pathology being studied and affects appropriate site for defining the AIF. • Dynamic susceptibility contrast perfusion imaging is most robust used semi-quantitatively. Arterial spin labeling • Arterial spin labeling is a noise-limited technique and requires relatively long acquisition times. • Transit time effects cause signal loss due to T1 relaxation: more severe in continuous than pulsed-labeled methods.

• Transit time effects are more marked in the presence of pathology. • Results also depend on labeling efficiency, inflow time, magnetization transfer effects and subtraction errors. • Models for quantification contain assumptions which are not necessarily valid.

Introduction As described in the previous two chapters, MR imaging (MRI) provides two alternative approaches to measure perfusion: using a bolus of an exogenous paramagnetic contrast agent (dynamic susceptibility contrast imaging (DSCI); see Chapter 7 for details); or by means of magnetically labeled blood as an endogenous tracer (arterial spin labeling (ASL); see Chapter 8 for details). Both techniques have been used for over a decade, and important improvements have made them more accurate and robust. However, some issues must be considered whenever implementing, using, and/or interpreting the images generated by either technique. This is because both methods rely on certain assumptions that cannot always be satisfied, and they can be prone to artifacts. This chapter discusses the most important caveats relating to absolute quantification of perfusion MRI, with particular emphasis on the potential implications for absolute cerebral blood flow (CBF) measurements in clinical use. 141

142

Fernando Calamante

Artifacts and limitations DSCI Although DSCI can provide, in principle, absolute measurements of CBF, cerebral blood volume (CBV), and mean transit time (MTT), there are a number of issues that can affect the accuracy of these measurements. The most common potential sources of error are described below. Relationship between changes in R2/R2* and concentration As discussed in Chapter 7, the contrast agent concentration is not directly measured using MRI, but it is indirectly determined from the changes in relaxation rates R2 (or R*2)1. These changes have been shown empirically (Rosen et al., 1990) and using numerical simulations (Weisskoff et al., 1994b) to be linearly proportional to the concentration of contrast agent: C t (t )  t R2(t )

(9.1)

where Ct(t) is the concentration at time t, and the proportionality constant t depends on the tissue type, the contrast agent, the field strength, and the pulse sequence. However, recent studies have suggested that this linear relationship may not always be valid (Kiselev, 2001; van Osch et al., 2003). Using numerical simulations, Kiselev has shown a nonlinear relationship for large contrast concentrations. Therefore, although the assumption of a linear relationship may be valid in the tissue being studied, it may be a significant source of error in the arterial input function (AIF) (due to the higher concentration found in large arteries). As suggested by Akbudak and Conturo (1996) and van Osch et al. (2003) one possible solution is the use of the phase information of the MR images: while a quadratic dependency was observed with the changes in R2*, a linear relationship was measured for the phase velocity ( / TE, where  is the phase of the MR signal and TE is the echo time) (van Osch et al., 1

For the remainder of this chapter, all the statements referring to T2 (and R2  1/T2) are also applicable to T2* (and R *2  1/T2*), unless otherwise stated.

2003). Furthermore, the quadratic relationship for R2 was dependent on the hematocrit levels, and therefore it could vary between subjects and even between different areas of the brain. Therefore, although a simple linear relationship is generally assumed in practice, this can lead to significant errors, particularly at high contrast agent concentrations. Unknown constants Even if the relationship between the contrast agent concentration and the change on the transverse relaxation rates were indeed linear for the doses commonly used in clinical studies, the value of this proportionality constant is required for absolute measurement of CBF. Since this value cannot be determined in vivo, a fixed value throughout the brain tissue is commonly assumed (Rempp et al., 1994; Schreiber et al., 1998; Vonken et al., 1999; Smith et al., 2000; Grandin et al., 2001). Furthermore, the same value is also assumed for the proportionality constant in the AIF. However, some studies have shown that the proportionality constant may vary between tissue types (Johnson et al., 2000) and between different subjects (Hedehus et al., 1997). Furthermore, recent numerical simulations (Kiselev, 2001) have also suggested that the relationship is different in the tissue and in the reference artery (AIF). Therefore, the assumption of a uniform proportionality constant across different tissues and in the arteries may lead to significant errors in the absolute quantification of CBF and CBV. It is interesting to note that assuming a universal value for the constant has been shown to produce CBF values (in ml/100 g/min) in healthy volunteers that are consistent with literature values measured using other techniques (e.g. Rempp et al., 1994; Schreiber et al., 1998). However, this agreement might have been fortuitous. Furthermore, it should be noted that these proportionality constants might also be expected to change in the presence of pathology, and therefore the extrapolation of this approach to patients can be criticized. The proportionality constants described above are not the only unknown constants required for absolute measurement of perfusion. The fundamental relationship to determine CBF involves a

Artifacts and pitfalls in perfusion MR imaging

further constant , which depends on the brain tissue density (to provide the correct units), and the hematocrit levels in capillaries and large vessels (because only the plasma volume is accessible to the contrast agent) (Calamante et al., 1999): C t (t )  CBFI(C a (t ) 䊟 R(t ))

(9.2)

where Ca(t) is the AIF and R(t), the tissue residue function (see Chapter 7). The quantities determining the constant are not easily measured, and therefore, fixed values are also assumed (Rempp et al., 1994; Schreiber et al., 1998; Vonken et al., 1999; Smith et al., 2000; Grandin et al., 2001). Changes in these values during pathology have been reported (Loufti et al., 1987; Yamamuchi et al., 1998), which can lead to significant errors in perfusion quantification (Calamante et al., 2002). An alternative approach to fixing these constants is to obtain an empirical conversion factor (to absolute units) by cross-calibration to a “gold standard” technique (e.g. positron emission tomography, (PET)) (Østergaard et al., 1998a, 1998b). However, the validity of a single conversion factor under various physiological conditions remains to be shown. In a study of CBF measurements (using MRI and PET) in patients with unilateral carotid artery occlusion, Lin et al. proposed an extra correction factor based on measurements of the venous output function in the superior sagittal sinus (Lin et al., 2001). This correction factor was shown to improve the correlation between MR and PET measurements, although the accuracy at low flows is still limited due to a significant non-zero y-intercept of the regression line (Lin et al., 2001). AIF: bolus delay and dispersion Inaccuracy in the AIF is one of the major potential sources of error in perfusion quantification (Calamante et al., 2002). Quantification of CBF using deconvolution requires knowledge of the AIF (see Eq. (9.2)). However, this function cannot be determined accurately at present, and it is in practice estimated from the signal changes in a major artery (e.g. the middle cerebral artery (MCA) or the internal carotid artery (ICA)). Since it is assumed that this estimated AIF represents the exact input to the tissue, any

delays and dispersion of the bolus that are introduced during its passage from the site of AIF estimation to the tissue of interest will introduce an error in the quantification of CBF (Calamante et al., 2000). Furthermore, this error could well vary from one region to another because of differences in the amount of delay and dispersion between these different regions. Numerical simulations have been used to assess the magnitude of the error introduced in quantification of DSCI data (using the commonly used singular value decomposition (SVD) approach) by various degrees of delay and/or dispersion. It was found that delays of 1–2 s (similar to the typical time resolution used in DSCI studies) could introduce large errors (⬃40% underestimation of CBF and ⬃60% overestimation of MTT). These delays are not uncommon in patients with cerebrovascular disease (see e.g. Neumann-Haefelin et al., 2000; Calamante et al., 2001), and the associated errors are further increased if there is also dispersion of the bolus (Calamante et al., 2000). Various methods to correct for errors due to bolus delay have been proposed, such as by shifting the curves to a common time origin, or by using a deconvolution method that can account for delays (Vonken et al., 1999; Smith et al., 2000; Wu et al., 2003). However, correction of bolus dispersion is not straightforward since it requires modeling of the vascular effects (Østergaard et al., 1999; Calamante et al., 2000; Calamante et al., 2003), and the vascular operator that properly characterizes the bolus dispersion is unknown. The problem with bolus dispersion is that the kinetic model used cannot differentiate between the vascular contribution to dispersion in the transit of the bolus to the tissue of interest, and the true intravoxel dispersion upon which the theory is based. Since the model cannot differentiate between these two contributions, both are assigned to the intravoxel dispersion and errors are introduced in MTT and CBF quantification (Calamante et al., 2000). Therefore, the presence of bolus delay and dispersion can lead to very misleading information in stroke patients where there may commonly be vessel occlusion, stenosis, or collateral circulation, and the perfusion maps should be interpreted with caution in the presence of significant delay and/or dispersion.

143

144

Fernando Calamante

The above is illustrated by the two cases shown in Figure 9.1. In both cases, perfusion imaging provides important information about the tissue pathophysiology. However, in the second of these cases, there is considerable scope for misinterpretation of the nature of the abnormality. The data in Figure 9.1(a) are from a 5-year-old boy with sickle cell disease and bilateral moyamoya syndrome, who had an acute right hemiparesis 5 weeks before the MRI examination. The T2/diffusion images show a chronic left occipital infarct. However, the perfusion abnormality extends beyond that area, with reduced CBF and prolonged MTT in a non-infarcted region in the right hemisphere (arrows). The CBF ratio (right side to left side) was ⬃0.50. The gamma-variate fitting to the regional concentration time course indicated that the bolus arrived at both regions at approximately the same time (⬃0.01 s apart). Therefore, the ⬃0.50 ratio is likely to reflect (in the absence of significant dispersion) a true ⬃50% reduction in the right CBF compared to the left side. Figures 9.1(b) and 9.1(c) show the DSCI data from a second patient, a 6-year-old boy with right ICA stenosis. Although there is only a small mature infarct in the right frontal white matter (WM) (with no other abnormality on diffusion or structural MRI), there is a very extensive perfusion abnormality throughout the right hemisphere (Figure 9.1(b)). The CBF ratio (right side to left side) estimated from the deconvolution analysis was ⬃0.55, which is similar to the value obtained for the first patient. However, in contrast to the first patient, a regional analysis of the concentration time course for the two regions showed that the bolus arrival was delayed by ⬃1.75 s on the right side compared to the left side (Figure 9.1(c)). This delay would introduce an underestimation in the calculated CBF of ⬃40% and a MTT overestimation of ⬃60% in the right region (Calamante et al., 2000). Therefore, the perfusion maps in Figure 9.1(b) are misleading in the sense that the bolus delay can almost completely account for the apparently low CBF value observed in the map; the actual CBF is probably close to normal. As discussed previously (Calamante et al., 2000), although the effect of delay can, in principle, be accounted for by shifting back the peak, the effect of dispersion (which to a varying degree is likely

to be associated with any delay) cannot easily be corrected. Such correction would require modeling of the vascular bed and generalization of the kinetic model to include dispersion effects (Østergaard et al., 1999; Calamante et al., 2000, 2003). AIF: partial volume effects Absolute quantification of CBF requires an absolute measurement of the AIF, and this leads to further problems. For example, there is a potential problem associated with partial volume effects (PVEs) as a result of the relatively low spatial resolution of the images commonly used in DSCI (typical voxel size ⬃2 2 5 mm3). Some corrections have recently been suggested (Wirestam et al., 2000; Lin et al., 2001; van Osch et al., 2001) and, in general, they involve scaling of the estimated AIF. However, the effect in some cases might be more complicated, since the partial volume could be with perfused tissue, which would also influence the shape of the AIF as well as it’s amplitude. The effect of partial volume can be further complicated by the angular dependency of the phase of the MR signal since this angular dependency is different for the extravascular and the intravascular contributions (Boxerman et al., 1995), and the total signal in the pixel (weighted average of the intravascular and extravascular signal) will be dependent on the orientation of the vessel. In any case, an uncorrected partial volume will introduce an underestimation of the AIF, with a corresponding overestimation of CBF (see Eq. (9.2)). AIF peak saturation The larger concentration of contrast agent in the arteries can produce a “saturation” in the AIF peak: due to a relatively long TE commonly used in DSCI studies, the signal intensity measured in arterial pixels during the first pass of the contrast agent can fall to the background noise levels. This effect leads to an underestimation of the AIF, and a corresponding overestimation of CBF (Ellinger et al., 2000). Two approaches have been proposed to minimize or eliminate this source of error. Firstly, a shorter TE can be used to acquire the slice where the AIF is measured. This method has been shown to be

(c)

(a)

Concentration [a.u.]

0.2

0.04

0

0.04

0.08

0.12

0.16

R

0

R

CBF

6

L

L

T2-weighted TSE

12

Time [s]

18

L

MTT

DWI

24

R

30

CBV

ADCAV

36

42

(b)

CBF

L

MTT

DWI

CBV

ADCAV

Fig. 9.1 Data from two children illustrating a potential problem of using deconvolution analysis in the presence of cerebrovascular disease. (a) Five-year-old boy with bilateral moyamoya syndrome. Although the T2/diffusion images show a chronic left occipital infarct (top row), there is also an extensive CBF abnormality in the right hemisphere (bottom row, see arrows), with increased MTT in equivalent regions. (b) Six-year-old boy with right ICA stenosis where the deconvolution analysis gives misleading information. Although there is only a small mature infarct in the right frontal WM (top row), there is a very extensive perfusion abnormality throughout the right hemisphere (bottom row). (c) The graph shows the concentration–time course for two regions (see inset). There is a delay of ⬃1.75 s in the arrival of the bolus to the region in the right. This delay introduces an underestimation in the calculated CBF making the CBF maps unreliable. See text for more information. (Figure previously published in reference (Calamante et al., 2002), © 2002, American Heart Association, Inc.) ADCAV: average apparent diffusion coefficient; DWI: diffusion-weighted imaging.

R

T2-weighted EPI

146

Fernando Calamante

effective, although at the expense of reducing the sensitivity in the tissue in that slice. This option may not be available in clinical standard pulse sequences, and may require pulse-programming modifications. Alternatively, the measurement points whose intensities are below a given threshold (dependent on the background noise level) can be excluded from the analysis (Ellinger et al., 2000). This method has been shown to improve the perfusion quantification in the presence of peak saturation, although it requires that the number of remaining measurement points are enough to characterize the shape of the peak properly. Contrast agent re-circulation – gamma-variate fitting The original kinetic model for perfusion quantification was developed assuming that once the contrast agent leaves the tissue of interest, it does not re-enter (Zierler, 1965). Therefore, only the first passage of the bolus should be considered. However, a second (overlapping) smaller, delayed peak is commonly observed, usually referred to as bolus re-circulation. To reduce this effect, two alternative methods are commonly used: (i) the analysis is only performed in the part of the peak corresponding to the first pass, or (ii) the first pass is fitted to an assumed bolus shape (e.g. gamma-variate function (Thompson et al., 1964)) and the resulting function extended to longer times. Both methods can introduce errors: while the first method always underestimates the area under the peak (AUP) (with a corresponding CBV underestimation), the second method can give erroneous measures if the assumed model is inaccurate (Levin et al., 1995; Kassner et al., 2000). A further potential source of error with the latter method is due to inaccuracies in the fitting (the non-linear fitting to the model is very sensitive to noise (Boxerman et al., 1997)). Contrast agent residual effects Although the signal intensity in a DSCI experiment returns to the baseline value a few minutes after the injection, a significant residual effect has been shown when a second bolus is injected. Levin et al. have investigated the effect of successive injections on the peak shape and they found a residual effect even 2 h after the first injection (Levin et al., 1995).

This residual effect must be taken into account when studies with multiple injections are required, such as for the assessment of cerebrovascular reactivity (measurements performed before and after vascular challenge). If the residual effect is not taken into account, an overestimation in the second measurement will be obtained. Although a full understanding of the cause of this residual effect is still unclear (Levin et al., 1998), a common approach to reduce it is by injecting a small pre-dose of contrast agent a few minutes before the study (Levin et al., 1995; Sorensen and Reimer, 2000). This approach has been shown to be effective in minimizing the T1 enhancement associated with contrast leakage (Kassner et al., 2000). Blood–brain barrier breakdown The kinetic model described in Chapter 7 is based on the assumption that the contrast agent remains intravascular. If this is not the case (e.g. when the blood–brain barrier (BBB) is disrupted), then the distribution of the contrast agent outside the vascular compartment decreases the T2 effects, as well as increases the (usually neglected) T1 effects during the passage of the bolus. If these effects are not minimized (Sorensen and Reimer, 2000) or taken into account (Vonken et al., 2000), significant errors can be introduced in quantification of DSCI data. Although the use of a dual echo sequence (to calculate T2) has been shown to eliminate the confounding effects of T1 enhancement (Vonken et al., 2000), this is done at the expense of reducing the maximum number of slices available. In order to account for the T1 effects, Weisskoff et al. (1994a) modeled the MR signal in terms of the combined T1 and T*2 contributions. In such a way, they proposed a method to quantify CBV in the presence of contrast leakage, as well as an estimation of vascular permeability (Weisskoff et al., 1994a; Donahue et al., 2000). In a more recent study, Vonken et al. (2000) extended the kinetic model (cf. Eq. (9.2) above) to quantify not only CBV and a measure of permeability, but also CBF. Since the effects of contrast leakage are included, it should provide a more accurate estimation of perfusion when the BBB is disrupted (Figure 9.2), although a full validation of this modified model remains to be done. Finally, a further

Artifacts and pitfalls in perfusion MR imaging

150 Simulated CBV Corrected Uncorrected

60

CBF (ml/100 g/min)

CBV (ml/100 g)

80

40

20

0 0

0.2

0.4 0.6 Extraction ratio

0.8

1

Simulated CBF Corrected Uncorrected

125

100

75

50 0

0.2

0.4 0.6 Extraction ratio

0.8

1

Fig. 9.2 Simulations results with an exponential residue function, with CBV: 10 ml/100 g, CBF: 100 ml/100 g/min and signal-to-noise ratio (SNR): 10. The plots show CBV (left) and CBF (right) as a function of the extraction ratio, with and without application of the contrast extravasation correction. The dotted lines show the true values for reference. As can be seen in the figure, more accurate results are obtained using the correction method. (Figure kindly provided by Dr. Evert-jan Ph. A. Vonken, and previously published in reference (Vonken et al., 2000), © 2000, John Wiley & Sons, Inc.)

issue that must be taken into account when there is contrast leakage is related to the bolus re-circulation: it has been pointed out that modeling the first passage as a gamma-variate function may be inaccurate in these cases (Levin et al., 1995; Kassner et al., 2000), and a different model may be necessary. “Voxel shift” (vessel misregistration) Due to the high temporal resolution required for the accurate characterization of the bolus passage, echo planar imaging (EPI) is currently the most common imaging technique used to measure cerebral perfusion. However, EPI suffers from a series of image artifacts (Fischer and Ladebeck, 1998), which must be considered when interpreting the calculated maps. One of the main sources of artifact in EPI is related to the very low bandwidth per pixel in the phase encoding direction (⬃10 Hz/pixel), responsible for artifacts such as the chemical shift artifact, and geometric image distortion near the interface between two materials with different susceptibility properties (e.g. air/tissue and bone/tissue) (Fischer and Ladebeck, 1998). These are artifacts not particular to DSCI, but common to any EPI application. However, there is a further EPI artifact that is introduced by the presence of the bolus of contrast agent itself. Due to the paramagnetic properties of the contrast agent, local field inhomogeneities are

created during the passage of the bolus, introducing vessel misregistration, and distorting the signal time curve in the voxels near the boundaries of a vessel (Hou et al., 1999); this is illustrated in Figure 9.3. As can be seen in the figure, this type of artifact can produce some pixels with an erroneous negative CBV (due to an artificial negative contrast concentration during the first pass). Therefore, it should be kept in mind that when EPI is used, the signal close to big vessels may be artifactual. It should also be kept in mind that this source of artifact can influence the measurement of the AIF, and any automatic criteria to select pixels for the AIF should avoid the erroneous selection of these artifactual pixels. Summary parameters The use of summary parameters to quantify DSCI is still a common technique (e.g. Sunshine et al., 1999; Latchaw et al., 2003). These are parameters that can be calculated directly from the shape of the first pass of the bolus of contrast agent, such as time to peak (TTP), maximum peak concentration (MPC), AUP and bolus arrival time (BAT). The use of these summary parameters is quite popular because they do not require measurement of the AIF (cf. deconvolution method), and can be generated in a more straightforward way than CBF and MTT. However, early studies have pointed out that none of

147

148

Fernando Calamante

Fig. 9.3 Vessel misregistration artifact introduced by the passage of a bolus of contrast agent. The top left figure shows a CBV map. The bottom left image shows a zoomed-version of a section of the CBV map (area indicated with the dashed rectangle in the top figure). The graphs on the right correspond to the signal intensity–time courses in three pixels (indicated by the arrows in the zoomed-image). As can be seen, the time course data is distorted in the top and middle graphs, with an unrealistic shape (note that the graphs are plots of signal intensity and not contrast concentration). In particular, the signal intensity–time course in the middle graph has a shape of a “peak”, corresponding to a negative CBV in the map.

the parameters provides direct measures of perfusion (Weisskoff et al., 1993) since the shape of the first pass is influenced not only by CBF, but also by the AIF and the tissue residue function (see Eq. (9.2)). Therefore, changes in the summary parameters can represent not only perfusion changes, but also changes in the injection conditions (volume injected, injection rate, and cannula size), in the vascular structure, and/or in the patient cardiac output

(Perthen et al., 2002). It is therefore not always straightforward to interpret the observed summary parameters in terms of patient physiology. In spite of these problems, summary parameters are commonly used as physiological markers (e.g. TTP and maximum peak concentration (MPC) are used as a measure of CBF). However, recent numerical simulations have shown that this can lead to very large errors, and that these parameters can vary over a

Artifacts and pitfalls in perfusion MR imaging

BAT

TTP (s)

TTP

10 8 6 4 R

2

MPC

0

L AUP

TTPAIF

MPCAIF Fig. 9.4 Results obtained using numerical simulations for the dependency of the tissue TTP on a range of AIF (characterized by their MPC and TTP, labeled MPCAIF and TTPAIF in the figure). The figure shows a 3D plot of tissue TTP as a function of AIF for normal GM (i.e. the CBF, CBV, and MTT were kept constant, and only the shape of the AIF was varied in the simulations). It can be seen that despite a fixed perfusion value, the AIF can greatly influence the actual tissue TTP value, which covers a wide range varying approximately between 4 and 10 s. Therefore, intersubject comparison of TTP values, or tissue classification based on TTP thresholds should be done with caution. Note that tissue TTP has only been calculated for a realistic range of AIFs, and outside this range its value has been set to zero for display purposes. (Figure kindly provided by Dr. Joanna E. Perthen, and previously published in reference (Perthen et al., 2002), © 2002, John Wiley & Sons, Inc.)

large range without any perfusion change (Perthen et al., 2002) (e.g. Figure 9.4). This variability was reduced, although not eliminated, when relative values (by calculating ratios or differences to a reference value) of summary parameters were calculated. Therefore, although summary parameter maps have been shown to provide very relevant clinical information, it is essential that they are interpreted with caution. Figure 9.5 shows an example of data from a patient with right internal carotid stenosis (same patient as that in Figure 9.1(b and c)). The figures show an extensive abnormality throughout the right hemisphere in various summary parameters

Fig. 9.5 Summary parameter maps (BAT, TTP, MPC, and AUP) calculated from the data acquired on a patient with right internal carotid stenosis (same patient as that in Figure 9.1(b) and (c)). An extensive area of prolonged BAT and TTP, and increase AUP can be seen in the right hemisphere. However, the MPC map is symmetric.

(prolonged BAT and TTP, and increase AUP). However, the MPC map is symmetric indicating that there is no reduction in the peak height. These maps indicate that the peak in the right side is delayed (both in arrival and in the time to its maximum value), but that it has normal height and increased AUP. Note that this is consistent with the graph in Figure 9.1(c), where the concentration time curves for two region of interest (ROI) in the WM are shown. These data suggest that the right side is in the autoregulation stage, with vasodilatation (increased AUP), prolonged MTT (wider peak), and possibly normal CBF (normal MPC). Although these are speculations based on the summary parameter maps (which illustrate the difficulty in interpreting these maps), the conclusions are consistent with those found in Figure 9.1: the bolus delay to the right side was found to account almost completely for the CBF reduction observed using the deconvolution analysis.

149

150

Fernando Calamante

Movement Subject movement can be a significant source of artifact in DSCI. Since this technique involves the rapid injection of a bolus of contrast agent, there is an increased chance of having subject motion at the time of the injection. Depending on the size of the motion, the artifact might be so severe as to render the data unusable. However, this is not always the case; if the movement (even a large displacement) only occurs during the bolus injection, the data can still be used: since the bolus takes several seconds to reach the brain, good quality data can be obtained during the passage of the bolus. However, one should not use the initial images (before the movement) to calculate the baseline image since, in general, they would not correspond anatomically to the position after the movement. Although image registration approaches can be used to correct for subject movement, they are not always a solution to the problem. This is because limited spatial coverage is usually acquired (due to the demand on temporal resolution), and it is common to have relatively large gaps between the slices acquired. Therefore, although in plane motion could be accurately corrected for, through plane motion correction could be difficult.

ASL There are a number of issues that can potentially introduce errors in perfusion quantification using ASL. Most of these issues affect both continuous ASL (CASL) and pulsed ASL (PASL), although to varying degrees. The main potential sources of errors are described below.

Transit time The transit time () of the labeled blood from the site of labeling to the exchange site decreases the sensitivity of the ASL techniques. This results from loss of magnetization labeling due to T1 relaxation (see Chapter 8). This effect is more important in CASL techniques due to the much longer distance the blood has to travel between the labeling and exchange sites. However, it has been shown that the transit time in PASL approaches, in general, cannot

be neglected (Wong et al., 1997; Yang et al., 2000). Therefore, transit times can be a significant source of error for both ASL methodologies. The presence of the transit time itself does not introduce errors since the model used for perfusion quantification can be easily modified to take account of a finite  (Alsop and Detre, 1996; Buxton et al., 1998; Pell et al., 1999). The main problem with the transit times is that they can be very heterogeneous throughout the brain, particularly in the presence of abnormalities (Detre et al., 1998) or during brain activity (Yang et al., 2000). Therefore, to avoid errors due to the presence of transit times, not only must the model account for transit time effects, but also the transit times should (ideally) be determined on a pixel-by-pixel basis. The low signal-to-noise ratio (SNR) of the difference images ( M(t)) acquired using ASL makes the determination of  on a pixelby-pixel basis very unreliable, and although, in principle, enough averages could be acquired to provide an accurate fitting, this is not feasible in practice. There are various approaches to the transit time problem. First, the transit time can be neglected. Although erroneous, this approach is common in PASL because it simplifies the quantification of perfusion (Kim, 1995; Kwong et al., 1995). With this assumption, measurements performed at a single inversion time (TI) can be used to quantify CBF (Kim, 1995). However, significant errors are introduced, and the measurement is dependent on the chosen TI. The second approach consists of including the transit time in the model, but assuming it homogeneous over the whole brain (Yen et al., 2002), or over relatively large regions (e.g. a single value for gray matter (GM) (Yang et al., 1998)). In this way, a pixelby-pixel fitting for  is not required. However, it has been shown that the unaccounted for distribution of transit times in the ROI can lead to an underestimation of perfusion (Figueiredo et al., 2002). The third approach to the transit time problem is the use of a sequence that is less sensitive to transit times. Sequence modifications have been proposed for CASL and PASL to such effect. For example, Alsop and Detre (1996) introduced a post-tagging delay (between the end of the labeling period and the

Artifacts and pitfalls in perfusion MR imaging

Fig. 9.6 Delayed arterial transit effects in a patient with right MCA stenosis. Top row: Multi-slice CASL perfusion MR images acquired acutely shows focal hypoperfusion in the right MCA territory as well as cortical delayed arterial transit (thick arrows). A 1.5 s post-labeling delay was used. Bottom row: Diffusion-weighted images acquired acutely shows hyperintensity confined to subcortical MCA territory (thin arrows). Although the effects of CASL on brain tissue are small, the effects in the intravascular space can be relatively large. This effect can lead to bright intraluminal signal in patients with delayed arterial transit, as seen in this figure. However, these bright areas are believed to represent the presence of collateral circulation and, although they introduce inaccuracies in CBF quantification, they can be used to assess collateral flow to hypoperfused areas. (Figure kindly provided by Dr. John A. Detre, and previously published in reference (Chalela et al., 2000), © 2000, Amercian Heart Association, Inc.)

image acquisition) in the conventional CASL sequence to reduce the sensitivity to variations in transit time (provided the delay is greater than the longest arterial transit time across the image). For the case of PASL, Wong et al. (1998) modified the conventional PASL sequences by introducing saturation pulses either in the imaging slice (quantitative imaging of perfusion using a single subtraction (QUIPSS)) or in the labeling site (QUIPSS II), which were shown to reduce the sensitivity to transit times. Both approaches are only effective if the timing sequence parameters satisfy certain conditions (Alsop and Detre, 1996; Wong et al., 1998), and these may become impractical if the transit time is very long, such as in patients with cerebrovascular abnormalities (e.g. Figure 9.6). The drawback of these techniques is a reduction of the perfusion signal due to T1 decay. Furthermore, the reduced sensitivity is primarily valid in GM, while WM retains significant sensitivity to transit times (e.g. Figure 9.2 in Alsop and Detre, 1996).

A further alternative approach to reduce the sensitivity to transit times in PASL is to reduce the distance the blood has to travel to the exchange site, for example by using pulses with better slice profiles (Yongbi et al., 1998), or by allowing for an interaction between the slice profiles of the labeling and imaging pulses, but correcting for this effect (Sidaros et al., 2001). It should be noted that the measurement of transit time involves the acquisition of a considerable amount of data, and therefore can be very time consuming. For PASL approaches, the conventional way to measure transit time is to acquire M images at various TI times, and to fit the data with a model that includes the transit time as a variable (Buxton et al., 1998; Pell et al., 1999). Measurement of transit time in CASL can be done by acquiring data with various post-tagging delays, and fitting for  with the corresponding model (Gonzalez-At et al., 2000). More recently, a method to estimate the arterial transit time from images acquired in the presence

151

152

Fernando Calamante

and absence of appropriate vascular crusher gradients has been proposed (Wang et al., 2003). Inflow time The inflow time () is defined in a PASL study as the time it takes for un-labeled blood spins beyond the labeling region to reach the slice. The PASL techniques involve the use of either a wide labeling (typically an inversion) slab positioned close to the imaging slice (e.g. in EPI-signal tagging with alternating radio frequency (RF), EPISTAR (Edelman et al., 1994)), or the use of a non-selective inversion pulse (e.g. in flow-sensitive alternating inversion recovery, FAIR (Kim, 1995)). Due to the finite length of the labeling slab, there is a time  at which the distant edge of the slab reaches the imaging slice. After that time , the un-labeled blood exchanges with the tissue magnetization. Although a non-selective inversion is used for FAIR-type sequences, a similar inflow time can be observed in practice due to the finite length of the coils commonly used in clinical studies (e.g. head coils). Due to the presence of unlabeled blood, the difference between the control and labeled images (the perfusion-weighted signal M) is reduced compared to the ideal case of an infinitely wide labeled slab (  ). If this effect is not taken into account in the quantification model, an underestimation of perfusion can be obtained (Calamante et al., 1996; Buxton et al., 1998). Therefore, the TI used in the PASL measurements should be shorter than , or alternatively this parameter must be included in the analysis (either as an extra variable to be determined, or as a fixed value, if known). There is a further issue related to inflow times that must be taken into consideration when quantifying PASL data. Due to the long scan times required to compensate for the low SNR of PASL, the delay time between multiple acquisitions is usually decreased to a minimum. However, if this delay is short enough such that the blood water has not fully relaxed (or fully replaced by un-labeled blood) by the time of the subsequent labeling period, the blood spin history will confound the labeling state for the subsequent images (Pell et al., 1999). Once again, if this effect is not taken into account in the PASL modeling, errors in perfusion quantification

will be introduced (Pell et al., 1999). An alternative approach is to modify the PASL sequence such that the time length of the labeling bolus is well defined. This approach is the basis of the QUIPSS sequences (Wong et al., 1998). As mentioned before, a saturation pulse is applied either to the imaging slice (QUIPSS) or to the labeling region (QUIPSS II) such that a labeled bolus with well-defined temporal width is obtained (Wong et al., 1998).

Intravascular signal The single-compartment model for CBF quantification assumes the signal comes exclusively from tissue water (see Chapter 8). However, labeled water still in the blood will contribute to the ASL signal. Therefore, the blood signal must be eliminated. As discussed in Chapter 8, this can be achieved by including “diffusion” pulse gradients (typically bipolar gradients) in the ASL sequence, which can selectively crush out signal from moving water (e.g. Figure 9.7) (Ye et al., 1997). Alternatively, some of the sequence modifications introduced to minimize transit time sensitivity (the use of a post-labeling delay in CASL and QUIPSS II) have also reduced sensitivity to intravascular signal: the delay period in these sequences allows most of the labeled blood to exchange/flow through slice (Alsop and Detre, 1996; Wong et al., 1998). This is illustrated in Figure 9.8 for the CASL case. Finally, an alternative approach to address the effect of intravascular signal is to modify the quantification model to include the intravascular compartment (see below).

Labeling efficiency The size of the perfusion-weighted signal ( M ) depends on the labeling efficiency ( ) of the ASL sequence. The labeling efficiency (also known as degree of inversion) reflects the amount of labeling, and it is equal to 1 for perfect inversion and 0.5 for perfect saturation (see Chapter 8 for more details). The labeling efficiency obtained in practice is very different for PASL and CASL sequences. Values close to the ideal  1 are often obtained using adiabatic RF pulses in PASL. Therefore, after careful calibration of the required power of the adiabatic pulses

Artifacts and pitfalls in perfusion MR imaging

(a)

(b)

(c)

(d)

(e)

(f)

Fig. 9.7 Images of M acquired using CASL. Images were calculated for six different values of the amplitude of the bipolar crusher gradient (G/cm): (a) 0, (b) 0.24, (c) 0.48, (d) 0.96, (e) 1.43, and (f) 1.7. The width of each pulse gradient was 6 ms, and the separation between the pulses was 2 ms. As can be seen in the figure, a substantial fraction of the signal from tagged arterial water spins has been “crushed” at the high bipolar gradient amplitudes. (Figure kindly provided by Dr. Frank Q. Ye, and previously published in reference (Ye et al., 1997), © 1997, John Wiley & Sons, Inc.)

100

300

600

900

1200

Fig. 9.8 Perfusion images acquired using CASL with various post-labeling delays (indicated in ms under each image). With short delays, the majority of spin-labeled blood resides in the vasculature. As the delay increases, spin-labeled blood distributes throughout the brain tissue in a manner directly related to CBF. (Figure kindly provided by Dr. David C. Alsop, and previously published in reference (Thomas et al., 2000), © 2000, IOP Publishing Ltd.)

153

154

Fernando Calamante

(Frank et al., 1997), the errors due to a miscalculated labeling efficiency are usually small in PASL. On the other hand, the labeling efficiency in CASL studies is much smaller, and it is highly affected by many factors, such as the blood relaxation rates, the blood velocity, the flow pulsatility, etc. (Utting et al., 2003). In clinical studies, the efficiency is further reduced due to hardware limitations and power deposition restrictions: the continuous off-resonance RF is often replaced by a train of short RF pulses (Ye et al., 1997; Utting et al., 2003). The situation is further complicated for multi-slice CASL, since not only can the efficiency of the labeling image be far from ideal, but also the efficiency of the control image. For example, the amplitude-modulated control for multi-slice CASL proposed by Alsop and Detre has been shown to have reduced efficiency, and to be highly dependent on the particular experimental conditions (Alsop and Detre, 1998). It should be noted that an accurate determination of the labeling efficiency is much more complicated in CASL than in PASL (Zhang et al., 1993), and an assumed value or a value measured on a different subject is commonly used. Therefore, an erroneous labeling efficiency value can lead to significant errors in the quantification of CBF using CASL. Magnetization transfer As described in Chapter 8, the RF pulses used for labeling can introduce a decrease in the signal intensity in the imaging slice through the indirect effect of magnetization transfer (MT). This effect is particularly important in CASL due to the very long (several seconds) off-resonance pulses used for labeling. Therefore, almost all ASL approaches also involve the acquisition of a “control” image in which the blood is not labeled, but the same off-resonance effects are present. This image is subtracted from the labeled image to cancel out the MT effects, leaving a perfusion-weighted image (see Chapter 8 for further details). If a complete cancellation of the MT effects is not achieved, errors in the quantification of perfusion will be introduced because the remaining MT signal would be considered to be perfusion related. A potential source of this incomplete cancellation is the asymmetric characteristics of the MT effects (Pekar et al., 1996). The control image can

be generated by changing the sign of the frequency offset of the labeling pulse, by reversing the polarity of the gradient used during labeling, or by a fourstep protocol in which both frequency and gradient polarity are alternated (Pekar et al., 1996). The first method assumes that the MT effect produced by an off-resonance pulse with a frequency offset   is the same as that produced with a frequency offset . Therefore, it will only produce complete cancellation when the MT effects are symmetric. Another potential source of incomplete saturation is in multi-slice acquisition. If any of the three methods to obtain the control image mentioned above is used, cancellation of the MT effect will only take place in one slice, since the “control” plane must be symmetrically opposite the inversion plane with respect to the imaging slice. Therefore, these schemes for acquiring the control image are not commonly used in a multi-slice set-up, and a different approach must be taken (see Chapter 8 for more details). Even when complete cancellation of the MT effect is achieved, MT can still introduce errors in the measurement of perfusion. This error is related to the use of an erroneous model for perfusion quantification. For example, to simplify the model it is commonly assumed that complete saturation of the macromolecular magnetization is achieved during the labeling and control periods. However, due to hardware limitations, power deposition constraints, or the particular choice of pulse-sequence parameters, only partial saturation of the macromolecules can be achieved. In this case, if this effect is not taken into account in the model, the measured values of perfusion will be erroneous (McLaughlin et al., 1997). Although the MT effects are much less important in PASL (due to relatively short pulses used for labeling), a complete cancellation of these effects must still be ensured to avoid errors in the quantification of perfusion. Quantification model The most common ASL model used for cerebral perfusion quantification is that of a single brain compartment. In this model, blood water is assumed to be a freely diffusible tracer that

Artifacts and pitfalls in perfusion MR imaging

exchanges completely with brain tissue water (Detre et al., 1992; Williams et al., 1992; Calamante et al., 1999; Barbier et al., 2001). The Bloch equation for longitudinal relaxation can be modified to incorporate the effects of perfusion and MT. This equation is then solved with the appropriate initial conditions and magnetization state for the blood water, and an expression for M (the perfusion-weighted signal) in terms of CBF and other parameters (e.g. T1, , ) is obtained (see Chapter 8 for a detailed description). However, some further assumptions are sometimes included to simplify the equations for perfusion quantification. For example, some PASL studies neglect the difference between the longitudinal relaxation rate of blood (T1a) and of brain tissue (T1) (e.g. Kim, 1995). This assumption simplifies the relationship between M and CBF, but it has been shown to lead to errors (overestimation) in perfusion quantification, particularly in WM where the error can be greater than 100% (Calamante et al., 1996). Although the single-compartment model is still the most popular model for perfusion quantification using ASL, it has been long recognized that it has some limitations, particularly at high flow values where the complete exchange assumption is known to be invalid (Silva et al., 1997; St. Lawrence et al., 2000). Recent studies have proposed a modified model to include a two-compartment exchange model, with an intravascular compartment and an extravascular compartment (Zhou et al., 2001; Parkes and Tofts, 2002). Although the two-compartment models provide a more complete description of tissue perfusion, they have not found widespread use yet since they introduce several further unknown parameters that must be determined or assumed from literature values. Furthermore, there are some disagreements regarding the effects predicted by these models for clinical data obtained at typical whole-body field strength (1.5–3 T). For example, while the results from Zhou et al. suggested that the single compartment is appropriate in these cases, Parkes and Tofts have predicted that such a model would significantly overestimate perfusion, particularly in WM. The possible reasons for this disagreement include different assumed values for the blood T1 relaxation, and different consideration of the

venous outflow in the model (Parkes and Tofts, 2002). Subtraction errors Both ASL techniques can be prone to static tissue subtraction errors. In theory, the M signal in the absence of perfusion (e.g. a static phantom sample) should be zero. However, this is not always the case in practice. For example, an offset can be observed in CASL due to incomplete cancellation of MT effects (see above), and a similar offset can be observed in PASL due to the interaction between the slice profiles of the labeling and imaging pulses (Frank et al., 1997; Sidaros et al., 2001). Since the perfusion signal is very small, this offset can introduce large errors in perfusion quantification. To avoid this source of error, a careful calibration of the ASL pulse sequence should be performed in a nonflowing phantom. Alternatively, to reproduce more closely the in vivo conditions, the calibration can be performed in humans, following the injection of a bolus of gadolinium dimeglumine gadopentetate (Gd-DTPA) (Yongbi et al., 2000). Due to the paramagnetic properties of the contrast agent, the T1 of blood water is significantly decreased, effectively eliminating the labeling (Figure 9.9). Yongbi et al. have suggested this method as a calibration protocol for the ASL sequences (Yongbi et al., 2000). When extending ASL sequences to multi-slice imaging, special care must be taken to ensure that these subtraction errors are cancelled out for all the slices in order to avoid errors when quantifying CBF. Partial volume effects To alleviate the low SNR of the perfusion-weighted images, the voxel dimensions are commonly very large in clinical studies (voxels of ⬃3 3 10 mm3 are not uncommon). Therefore, significant PVEs are present in perfusion images obtained using ASL. While the effect of partial volume between GM and WM is to reduce the difference between the two (underestimation of GM perfusion and overestimation of WM perfusion), there can be also partial volume with CSF (introducing an underestimation of perfusion) and with arterial contributions (introducing an overestimation of perfusion). All

155

156

Fernando Calamante

i

ii

iii

iv

Fig. 9.9 Multi-slice FAIR perfusion images acquired pre-contrast using selective labeling inversion widths of (i) 50 mm, and (ii) 80 mm. Corresponding FAIR images acquired post-Gd-DTPA using (iii) 50 mm, and (iv) 80 mm. Note the artifactually increased signal in the outermost two slices for the 50 mm case, reflecting contributions from non-perfused tissue. (Figure kindly provided by Dr. Jeff H. Duyn, and previously published in reference (Yongbi et al., 2000), © 2000, John Wiley & Sons, Inc.)

these effects will compete with each other, and although the voxel can be considered as a weighted average of the individual components, the effect on perfusion is not straightforward due to the complex relationship between perfusion, T1, and M(TI) (Kwong et al., 1995). Blood spin–lattice relaxation time The magnetic labeling in the blood water is lost by spin–lattice relaxation (T1a). Therefore, to account for this effect properly, a measure of T1a is required. However, an in vivo measurement of this parameter is not straightforward, and an assumed value, based on previous studies or published literature values, is usually used. This can be a potential source of error in perfusion quantification since it has been shown that T1a can be affected by the hematocrit levels and oxygen saturation (Silvennoinen et al., 2003).

Blood–brain partition coefficient It is common practice to assume a uniform value (typically 0.9 ml/g) for the blood–brain partition coefficient () throughout the brain when quantifying perfusion using ASL techniques. However, it has

been shown that  can vary between tissue types and for varying hematocrit levels (Roberts et al., 1996). Furthermore,  can be influenced by evolving pathophysiological mechanisms, for example, in the presence of oedema. Values of 0.93–1.04 and 0.77–0.89 ml/g have been reported for healthy human GM and WM, respectively (Roberts et al., 1996). If a uniform value of 0.9 ml/g were assumed, an underestimation of GM perfusion and an overestimation of WM perfusion would be introduced. Since the CBF/ ratio is obtained using ASL (Detre et al., 1992), the size of this under or overestimation is directly related to the error in . Movement Subject movement can be a significant source of artifact in ASL. However, the source of this artifact is different from that in DSCI. In ASL, the movement can arise from the very long scan time required to acquire enough data to quantify perfusion accurately. Due to the low SNR of the perfusion-weighted images in ASL, many averages (typically ⬃40) are required. PASL may also require the acquisition of multiple images at various TI to characterize properly the various parameters for perfusion

Artifacts and pitfalls in perfusion MR imaging

quantification (CBF, transit time, inflow time, etc.). Furthermore, a map of the longitudinal relaxation time T1 is required for perfusion quantification. Therefore, scan times of ⬃30 min are common, which can be very susceptible to subject motion. Image registration approaches can be used to reduce the motion effects, however the correction of through plane motion may be problematic due to the small number of slices commonly used.

Conclusion As discussed in this chapter, several factors can affect the accuracy of CBF quantification using MRI. However, the potential problems described here should not hinder the use of DSCI and ASL to assess the perfusion status in clinical and research studies. Both methodologies are very powerful techniques that provide unique information regarding cerebral hemodynamics. They have been extensively used for the assessment and management of patients, as well as being an invaluable tool in experimental studies. They have played, and will continue to play, a key role in the MR evaluation of patients. Solutions to many of the issues discussed in this chapter are the subject of current research. In the interim, these issues should be taken into consideration whenever MR is used to measure perfusion, and the users of these techniques should be aware of the potential problems to avoid misinterpretation of the findings, and make the most of the invaluable physiological information provided by MR perfusion.

REFERENCES Akbudak E, Conturo TE. 1996. Arterial input functions from MR phase imaging. Magn Reson Med 36: 809–815. Alsop DC, Detre JA. 1996. Reduced transit-time sensitivity in noninvasive magnetic resonance imaging of human cerebral blood flow. J Cereb Blood Flow Metab 16: 1236–1249. Alsop DC, Detre JA. 1998. Multisection cerebral blood flow MR imaging with continuous arterial spin labeling. Radiology 208: 410–416.

Barbier EL, Lamalle L, Décorps M. 2001. Methodology of brain perfusion imaging. J Magn Reson Imaging 13: 496–520. Boxerman JL, Hamberg LM, Rosen BR, Weisskoff RM. 1995. MR contrast due to intravascular magnetic-susceptibility perturbations. Magn Reson Med 34: 555–566. Boxerman JL, Rosen BR, Weisskoff RM. 1997. Signal-to-noise analysis of cerebral blood volume maps from dynamic NMR imaging studies. J Magn Reson Imaging 7: 528–537. Buxton RB, Frank LR, Wong EC, Siewert B, Warach S, Edelman RR. 1998. A general kinetic model for quantitative perfusion imaging with arterial spin labeling. Magn Reson Med 40: 383–396. Calamante F, Gadian DG, Connelly A. 2000. Delay and dispersion effects in dynamic susceptibility contrast MRI: simulations using singular value decomposition. Magn Reson Med 44: 466–473. Calamante F, Gadian DG, Connelly A. 2002. Quantification of perfusion using bolus tracking MRI in stroke. Assumptions, limitations, and potential implications for clinical use. Stroke 33: 1146–1151. Calamante F, Ganesan V, Kirkham FJ, Jan W, Chong WK, Gadian DG, Connelly A. 2001. MR perfusion imaging in moyamoya syndrome. Potential implications for clinical evaluation of occlusive cerebrovascular disease. Stroke 32: 2810–2816. Calamante F, Thomas DL, Pell GS, Wiersma J, Turner R. 1999. Measuring cerebral blood flow using magnetic resonance techniques. J Cereb Blood Flow Metab 19: 701–735. Calamante F, Williams SR, van Bruggen N, Kwong KK, Turner R. 1996. A model for quantification of perfusion in pulsed labelling techniques. NMR Biomed 8: 79–83. Calamante F, Yim PJ, Cebral JR. 2003. Estimation of bolus dispersion effects in perfusion MRI using image-based computational fluid dynamics. Neuroimage 19: 341–353. Chalela JA, Alsop DC, Gonzalez-Atavales JB, Maldjian JA, Kasner SE, Detre JA. 2000. Magnetic resonance perfusion imaging in acute ischemic stroke using continuous arterial spin labeling. Stroke 31(3): 680–687. Detre JA, Alsop DC, Vives LR, Maccotta L, Teener JW, Raps EC. 1998. Noninvasive MRI evaluation of cerebral blood flow in cerebrovascular disease. Neurology 50: 633–641. Detre JA, Leigh JS, Williams DS, Koretsky AP. 1992. Perfusion imaging. Magn Reson Med 23: 37–45. Donahue KM, Krouwer HGJ, Rand SD, Pathak AP, Marszalkowski CS, Censky SC, Prost RW. 2000. Utility of simultaneously acquired gradient-echo and spin-echo cerebral blood volume and morphology maps in brain tumor patients. Magn Reson Med 43: 845–853. Edelman RR, Siewert B, Darby DG, Thangaraj V, Nobre AC, Mesulam MM, Warach S. 1994. Qualitative mapping of cerebral blood-flow and functional localization with echo-planar MR-imaging and signal targeting with alternating radio-frequency. Radiology 192: 513–520.

157

158

Fernando Calamante

Ellinger R, Kremser C, Schocke MFH, Kolbitsch C, Griebel J, Felber SR, Aichner FT. 2000. The impact of peak saturation of the arterial input function on quantitative evaluation of dynamic susceptibility contrast enhanced MR studies. J Comput Assist Tomogr 24: 942–948. Figueiredo P, Clare S, Jezzard P. 2002. Issues in quantitative perfusion and arterial transit time mapping using pulsed AST. In Proceeding of the 10th International Society for Magnetic Resonance in Medicine, Honolulu, USA, p. 623. Fischer H, Ladebeck R. 1998. Echo-planar imaging image artifacts. In Echo-planar Imaging. Theory, Technique and Application. (Eds. Schmitt F, Stehling MK, Turner R) Berlin, Springer, pp. 179–200. Frank LR, Wong EC, Buxton RB. 1997. Slice profile effects in adiabatic inversion: application to multi-slice perfusion imaging. Magn Reson Med 38: 558–564. Gonzalez-At JB, Alsop DC, Detre JA. 2000. Cerebral perfusion and transit time changes during task activation determined with continuous arterial spin labelling. Magn Reson Med 43: 739–746. Grandin CB, Duprez TP, Smith AM, Mataigne F, Peeters A, Oppenheim C, Cosnard G. 2001. Usefulness of magnetic resonance-derived quantitative measurements of cerebral blood flow and volume in prediction of infarct growth in hyperacute stroke. Stroke 32: 1147–1153. Hedehus M, Steensgaard, Rostrup E, Larsson EBW. 1997. Investigation of the linear relation between R *2 and gadolinium concentration in vivo. In Proceedings of the 5th International Society for Magnetic Resonance in Medicine, Vancouver, Canada, p. 1792. Hou L, Yang Y, Mattay VS, Frank JA, Duyn JH. 1999. Optimization of fast acquisition methods for whole-brain relative cerebral blood volume (rCBV) mapping with susceptibility contrast agents. J Magn Reson Imaging 9: 233–239. Johnson KM, Tao JZT, Kennan RP, Gore JC. 2000. Intravascular susceptibility agent effects on tissue transverse relaxation rates in vivo. Magn Reson Med 44: 909–914. Kassner A, Annesley DJ, Zhu XP, Li KL, Kamaly-Asl ID, Watson Y, Jackson A. 2000. Abnormalities of the contrast re-circulation phase in cerebral tumors demonstrated using dynamic susceptibility contrast-enhanced imaging: a possible marker of vascular tortuosity. J Magn Reson Imaging 11: 103–113. Kim SG. 1995. Quantification of relative cerebral blood flow change by flow-sensitive alternating inversion recovery (FAIR) technique: application to functional mapping. Magn Reson Med 34: 293–301. Kiselev VG. 2001. On the theoretical basis of perfusion measurements by dynamic susceptibility contrast MRI. Magn Reson Med 46: 1113–1122.

Kwong KK, Chesler DA, Weisskoff RM, Donahue KM, Davis TL, Østergaard L, Campbell TA, Rosen BR. 1995. MR perfusion studies with T1-weighted echo-planar imaging. Magn Reson Med 34: 878–887. Latchaw RE, Yonas H, Hunter GJ, Yuh WTC, Ueda T, Sorensen AG, Sunshine JL, Biller J, Wechsler L, Higashida R, Hademenos G. 2003. Guidelines and recommendations for perfusion imaging in cerebral ischemia. A scientific statement for healthcare professionals by the Writing Group on perfusion imaging, from the Council on cardiovascular Radiology of the American Heart Association. Stroke 34: 1084–1104. Levin JM, Kaufman MJ, Ross MJ, Mendelson JH, Maas LC, Cohen M, Renshaw PF. 1995. Sequential dynamic susceptibility contrast MR experiments in human brain: residual contrast agent effect, steady state, and hemodynamic perturbation. Magn Reson Med 34: 655–663. Levin JM, Wald LL, Kaufman MJ, Ross MJ, Maas LC, Renshaw PF. 1998. T1 effects in sequential dynamic susceptibility contrast experiments. J Magn Reson 130: 292–295. Lin W, Celik A, Derdeyn C, An H, Lee Y, Videen T, Østergaard L, Powers WJ. 2001. Quantitative measurements of cerebral blood flow in patients with unilateral carotid artery occlusion: a PET and MR study. J Magn Reson Imaging 14: 659–667. Loufti I, Frackowiak RS, Myers MJ, Lavender JP. 1987. Regional brain hematocrit in stroke by single photon emission computer tomography imaging. Am J Physiol Imaging 2: 10–16. Mclaughlin AC, Ye FQ, Pekar JJ, Santha AKS, Frank JA. 1997. Effect of magnetization transfer on the measurement of cerebral blood flow using steady-state arterial spin tagging approaches: a theoretical investigation. Magn Reson Med 37: 501–510. Neumann-Haefelin T, Wittsack H-J, Fink GR, Wenserski F, Li T-Q, Seitz RJ, Siebler M, Mödder U, Freund H-J. 2000. Diffusion- and perfusion-weighted MRI. Influence of severe carotid artery stenosis on the DWI/PWI mismatch in acute stroke. Stroke 31: 1311–1317. Østergaard L, Chesler DA, Weisskoff RM, Sorensen AG, Rosen BR. 1999. Modeling cerebral blood flow and flow heterogeneity from magnetic resonance residue data. J Cereb Blood Flow Metab 19: 690–699. Østergaard L, Johannsen P, Poulsen PH, Vestergaard-Poulsen P, Asboe H, Gee AD, Hansen SB, Cold GE, Gjedde A, Gyldensted C. 1998a. Cerebral blood flow measurements by magnetic resonance imaging bolus tracking: comparison with [O-15] H2O positron emission tomography in humans. J Cereb Blood Flow Metab18: 935–940. Østergaard L, Smith DF, Vestergaard-Poulsen P, Hansen SB, Gee AD, Gjedde A, Gyldensted C. 1998b. Absolute cerebral blood flow and blood volume measured by magnetic

Artifacts and pitfalls in perfusion MR imaging

resonance imaging bolus tracking: comparison with positron emission tomography values. J Cereb Blood Flow Metab 18: 425–432. Parkes LM, Tofts PS. 2002. Improved accuracy of human cerebral blood perfusion measurements using arterial spin labeling: accounting for capillary water permeability. Magn Reson Med 48: 27–41. Pekar J, Jezzard P, Roberts DA, Leigh Jr JS, Frank JA, McLaughlin AC. 1996. Perfusion imaging with compensation for asymmetric magnetization transfer effects. Magn Reson Med 35: 70–79. Pell GS, Thomas DL, Lythgoe MF, Calamante F, Howseman AM, Gadian DG, Ordidge RJ. 1999. The implementation of quantitative FAIR perfusion imaging with a short repetition time in time-course studies. Magn Reson Med 41: 829–840. Perthen JE, Calamante F, Gadian DG, Connelly A. 2002. Is quantification of bolus tracking MRI reliable without deconvolution? Magn Reson Med 47: 61–67. Rempp KA, Brix G, Wenz F, Becker CR, Guckel F, Lorenz WJ. 1994. Quantification of regional cerebral blood flow and volume with dynamic susceptibility contrast-enhanced MR imaging. Radiology 193: 637–641. Roberts DA, Rizi R, Lenkinski RE, Leigh JS. 1996. Magnetic resonance imaging of the brain: blood partition coefficient for water: application to spin-tagging measurement of perfusion. J Magn Reson Imaging 6: 363–366. Rosen BR, Belliveau JW, Vevea JM, Brady TJ. 1990. Perfusion imaging with NMR contrast agents. Magn Reson Med 14: 249–265. Schreiber WG, Gückel F, Stritzke P, Schmiedek P, Schwartz A, Brix G. 1998. Cerebral blood flow and cerebrovascular reserve capacity: estimation by dynamic magnetic resonance imaging. J Cereb Blood Flow Metab 18: 1143–1156. Sidaros K, Andersen IK, Gesmar H, Rostrup E, Larsson HB. 2001. Improved perfusion quantification in FAIR imaging by offset correction. Magn Reson Med 46: 193–197. Silva AC, Zhang WG, Williams DS, Koretsky AP. 1997. Estimation of water extraction fractions in rat brain using magnetic resonance measurement of perfusion with arterial spin labeling. Magn Reson Med 37: 58–68. Silvennoinen MJ, Kettunen MI, Kauppinen RA. 2003. Effects of hematocrit and oxygen saturation level on blood spin-lattice relaxation. Magn Reson Med 49: 568–571. Smith AM, Grandin CB, Duprez T, Mataigne F, Cosnar G. 2000. Whole brain quantitative CBF and CBV measurements using MRI bolus tracking: comparison of methodolgies. Magn Reson Med 43: 559–654. Sorensen AG, Reimer P. 2000. Cerebral MR Perfusion Imaging. Principles and Current Applications. Georg Thieme Verlag, Stuttgart, Germany.

St. Lawrence KS, Frank JA, Mclaughlin AC. 2000. Effect of restricted water exchange on cerebral blood flow values calculated with arterial spin tagging: a theoretical investigation. Magn Reson Med 44: 440–449. Sunshine JL, Tarr RW, Lanzieri CF, Landis DMD, Selman WR, Lewin JS. 1999. Hyperacute stroke: ultrafast MR imaging to triage patients prior to therapy. Radiology 212: 325–332. Thomas DL, Lythgoe MF, Pell GS, Calamante F, Ordidge RJ. 2000. The measurement of diffusion and perfusion in biological systems using magnetic resonance imaging. Phys Med Biol 45: R97–R138. Thompson HK, Starmer F, Whalen RE, McIntosh HD. 1964. Indicator transit time considered as a gamma variate. Circ Res 14: 502–515. Utting JF, Thomas DL, Gadian DG, Ordidge RJ. 2003. Velocitydriven adiabatic fast passage for arterial spin labeling: results from a computer model. Magn Reson Med 49: 398–401. van Osch MJP, Vonken EPA, Bakker CJG, Viergever MA. 2001. Correcting partial volume artifacts of the arterial input function in quantitative cerebral perfusion MRI. Magn Reson Med 45: 477–485. van Osch MJP, Vonken EPA, Viergever MA, van der Grond J, Bakker CJG. 2003. Measuring the arterial input function with gradient echo sequences. Magn Reson Med 49: 1067–1076. Vonken EPA, van Osch MJP, Baker CJG, Viergever MA. 1999. Measurement of cerebral perfusion with dual-echo multislice quantitative dynamic susceptibility contrast MRI. J Magn Reson Imaging 10: 109–117. Vonken EPA, van Osch MJP, Baker CJG, Viergever MA. 2000. Simultaneous qualitative cerebral perfusion and Gd-DTPA extravasation measurements with dual-echo dynamic susceptibility contrast MRI. Magn Reson Med 43: 820–827. Wang J, Alsop DC, Song HK, Maldjian JA, Tang K, Salvucci AE, Detre JA. 2003. Arterial transit time imaging with flow encoding arterial spin tagging (FEAST). Magn Reson Med 50: 599–607. Weisskoff RM, Boxerman JL, Sorensen AG, Kulke SM, Campbell TA, Rosen BR. 1994a. Simultaneous blood volume and permeability mapping using a single Gd-based contrast injection. In Proceedings of the 2nd Annual Meeting of Soceity for Magnet Resonance in Medicine, San Francisco, USA, p. 279. Weisskoff RM, Chesler D, Boxerman JL, Rosen BR. 1993. Pitfalls in MR measurement of tissue blood flow with intravascular tracers: Which mean transit-time? Magn Reson Med 29: 553–559. Weisskoff RM, Zuo CS, Boxerman JL, Rosen BR. 1994b. Microscopic susceptibility variation and transverse relaxation. Theory and experiment. Magn Reson Med 31: 601–610.

159

160

Fernando Calamante

Williams DS, Detre JA, Leigh JS, Koretsky AP. 1992. Magnetic resonance imaging of perfusion using spin inversion of arterial water. Proc Natl Acad Sci USA 89: 212–216. Wirestam R, Ryding E, Lindgren A, Geijer B, Holtas S, Stahlberg F. 2000. Absolute cerebral blood flow measured by dynamic susceptibility contrast MRI: a direct comparison with Xe-133 SPECT. MAGMA 11: 96–103. Wong EC, Buxton RB, Frank LR. 1997. Implementation of quantitative perfusion imaging techniques for functional brain mapping using pulsed arterial spin labeling. NMR Biomed 10: 237–249. Wong EC, Buxton RB, Frank LR. 1998. Quantitative imaging of perfusion using a single subtraction (QUIPSS and QUIPSS II). Magn Reson Med 39: 702–708. Wu O, Østergaard L, Weisskoff RM, Benner T, Rosen BR, Sorensen AG. 2003. Tracer arrival timing-insensitive technique for estimating flow in MR perfusion-weighted imaging using singular value decomposition with a block-circulant deconvolution matrix. Magn Reson Med 50: 164–174. Yamamuchi H, Fukuyama H, Nagahama Y, Katsumi Y, Okazawa H. 1998. Cerebral hematocrit decreases with hemodynamic compromise in carotid artery occlusion: a PET study. Stroke 29: 98–103. Yang Y, Engelien W, Xu S, Gu H, Silbersweig DA, Stern E. 2000. Transit time, trailing time, and cerebral blood flow during brain activation: measurement using multislice, pulsed spin-labeling perfusion imaging. Magn Reson Med 44: 680–685.

Yang Y, Frank JA, Hou L, Ye FQ, McLaughlin AC, Duyn JH. 1998. Multi-slice imaging of quantitative cerebral perfusion with pulsed arterial spin labeling. Magn Reson Med 39: 825–832. Ye FQ, Mattay VS, Jezzard P, Frank JA, Weinberger DR, Mclaughlin AC. 1997. Correction for vascular artifacts in cerebral blood flow values measured by using arterial spin tagging techniques. Magn Reson Med 37: 226–235. Yen YF, Field AS, Martin EM, Ari N, Burdette JH, Moody DM, Takahashi AM. 2002. Test–retest reproducibility of quantitative CBF measurements using FAIR perfusion MRI and acetazolamide challenge. Magn Reson Med 47: 921–928. Yongbi MN, Branch CA, Helpern JA. 1998. Perfusion imaging using FOCI RF pulses. Magn Reson Med 40: 938–943. Yongbi MN, Tan CX, Frank JA, Duyn JH. 2000. A protocol for assessing subtraction errors of arterial spin-tagging perfusion techniques in human brain. Magn Reson Med 43: 896–900. Zhang W, Williams DS, Koretsky AP. 1993. Measurement of rat brain perfusion by NMR using spin labeling of arterial water: in vivo determination of the degree of spin labeling. Magn Reson Med 29: 416–421. Zhou J, Wilson DA, Ulatowski JA, Traystman RJ, van Zijl PC. 2001. Two-compartment exchange model for perfusion quantification using arterial spin tagging. J Cereb Blood Flow Metab 21: 440–455. Zierler KL. 1965. Equations for measuring blood flow by external monitoring of radioisotopes. Circ Res 16: 309–321.

Section 2 Cerebrovascular disease

10

Cerebrovascular disease: overview Brian M. Tress Department of Radiology, Royal Melbourne Hospital, Parkville, Australia

Background Stroke is the third biggest cause of death in Western countries. As the commonest form of long-term adult disability it is a major consumer of health spending. Eighty percent of strokes are due to ischemic infarcts and 20% due to hemorrhages. Despite the magnitude of the problem successful active treatment of ischemic stroke was almost nonexistent until the mid-1990s. Medical management was predominantly prophylactic, particularly concentrated upon the control of hypertension. The role of diagnostic tests was limited to confirmation of stroke diagnosis, differentiation of infarct from hemorrhage and identification of those entities, which may present with stroke like symptoms and signs (“pseudostroke”). Chronic subdural hematoma, slow growing tumors, demyelination and arteriovenous malformations (AVM) are all potentially treatable lesions which may present as pseudostroke. As many as 20% of stroke like presentations can be due to entities other than stroke (Libman et al., 1995).

Thrombolysis in ischemic stroke The catalyst for enormously renewed interest in ischemic stroke was the publication in 1995 of the first trial suggesting patient outcome benefit following active thrombolytic treatment in patients treated within 3 h of the onset of symptoms (National Institute of Neurological Disorders and Stroke rtPA Stroke Study Group, 1995). In 1999 a prospective randomised trial of intra-arterial pro-urokinase in

patients with occluded middle cerebral arteries (MCA) showed patient outcome benefit when treatment was initiated between 3 and 6 h after the onset of the stroke symptoms (Furlan et al., 1999). Many animal trials indicated benefit from both thrombolytic and neuroprotective drugs, but the positive results for neuroprotective drugs have not been reproduced in human subjects. It is argued that better selection criteria are required if neuroprotective drugs are to be successful in humans.

Limitations of basic stroke screening methods The apparent success of thrombolytic drugs immediately highlighted the need for tests which were highly sensitive and specific for diagnosis, that provided information on which group of patients was most likely to respond to therapy and were sufficiently non-invasive to serially measure the effect of therapy. Current diagnostic tools for the investigation of stroke have become increasingly sophisticated and non-invasive. The basic principle has been that when investigation has been clinically justified the brain and the carotid vessels in the neck need to be imaged. The imaging modalities used vary according to those available at the imaging center. Computed tomography (CT) of the brain and duplex Doppler ultrasound of the carotid arteries was until recent years the mainstay of investigation. MR imaging (MRI) and MR angiography (MRA), or CT and CT angiography (CTA) performed with high-resolution multidetector CT machines are progressively replacing single detector plain CT and ultrasound. Digital subtraction angiography (DSA) and its associated 163

164

Brian M. Tress

cost and morbidity are progressively being used only as a conduit to interventional procedures. Although both CT and MRI have been revolutionary in their ability to directly depict the normal and abnormal brain, they each have significant limitations in hyperacute stroke. CT has close to 100% sensitivity in diagnosing hemorrhage, but even with complete MCA occlusion, only approximately two thirds of patients will have CT evidence of infarct at 2 h after the onset of symptoms (von Kummer et al., 1994). The sensitivity for diagnosis of acute brain stem and lacunar infarcts (LACI) is considerably less. Conventional T1- and T2-weighted MRI sequences are surprisingly insensitive in the detection of early infarcts. This is because of the relative insensitivity of T2-weighted sequences to cytotoxic edema, which is the predominant form of edema during the first 6 h. It is only when the tight junctions between endothelial cells break down, allowing fluid to leak into the extravascular space (vasogenic edema) that the infarct becomes easily detectable. Subtle brain swelling may be detected on T1-weighted sequences, but conventional noncontrast MRI studies are very unreliable in the first 6 to 8 h of an evolving infarct.

The ischemic penumbra Apart from the actual identification of infarcted tissue, there clearly is a need to identify those patients who would most likely benefit from thrombolytic therapy. The concept of the ischemic penumbra as a zone of functionally inert ischemic brain capable of recovery following reperfusion in a location adjacent to the irreversibly damaged ischemic core is not new. It was described in a baboon model in 1988 as a region of brain with a cerebral blood flow (CBF) of less than 20 ml/100 g/min sufficient to eliminate evoked potentials but capable of recovery if the occluding device was released (Symon et al., 1977; Astrup et al., 1981). The identification of perfusion deficits has been dependent upon well-established blood flow measuring techniques, such as positron emission tomography (PET), single photon emission CT (SPECT) and xenon CT. These techniques are currently available at only

limited numbers of well-equipped research centers and are not suitable for routine clinical use. Standard MRI and CT studies are unable to provide detailed perfusion information and are unable to definitely differentiate between ischemic brain tissue capable of recovery and that destined to undergo infarction.

Imaging the ischemic penumbra – diffusion and perfusion imaging and MRS Technical advances in magnet technology and the application of ultrafast echo planar imaging (EPI) sequences at high field strength (1.5 -T or more) have combined to provide the tools needed for the identification of the ischemic penumbra. Diffusionweighted sequences depict irrevocably infarcted tissue as an area of restricted diffusion with high sensitivity and specificity. In one study of patients with a clinical diagnosis of stroke diagnosed within 6 h of onset, sensitivity and specificity of 100% were reported (Barber et al., 1999), but negative diffusionweighted cases are well documented (Lovblad et al., 1998; Wang et al., 1999). Sensitivity of 88% and specificity of 95% were reported in patients studied within 24 h of stroke onset (Lovblad et al., 1998). Perfusion-weighted sequences demonstrate areas of relative hypoperfusion. The difference between the area of hypoperfusion and the area of restricted diffusion is considered to represent the ischemic penumbra. Perfusion–diffusion mismatches have been shown in 75% of stroke patients imaged within 6 h of symptom onset, diminishing to 44% of patients imaged at 18 h after symptom onset, presumably as a result of spontaneous thrombolysis and reperfusion (Darby et al., 1999). The presence of a perfusion–diffusion mismatch strongly implies major vessel occlusion (Rordorf et al., 1998). Conventional diffusion-weighted imaging (DWI) depicts isotropic diffusion, which basically is the average of diffusion in all directions. By performing diffusion sequences with gradients applied in at least six directions Pierpaoli was able to perform full tensor diffusion imaging, which depicts anisotropic, or directionally limited diffusion (Pierpaoli et al., 1996). The degree of directionally limited diffusion was

Cerebrovascular disease: overview

calculated on a pixel-by-pixel basis as a numerical entity, known as fractional anisotropy (FA). In the brain anisotropic diffusion is most prominent in axonal bundles, as diffusion perpendicular to the direction of an axon is markedly limited by the myelin sheath. FA measured acutely has been reported to correlate with clinical outcome (Yang et al., 1999). Acutely elevated FA in the absence of a depressed apparent diffusion coefficient (ADC) appeared to be associated with good outcome (Yang et al., 1999), but other studies suggested those findings might be artifactual, due to mathematical calculation artifact (Green et al., 2002). FA measurement can be used by connectivity algorithms to calculate axonal direction and integrity. Wallerian degeneration has been detected 10 weeks before it became evident on T2-weighted algorithms (Watanebe et al., 2001), aiding in prognostication. MR spectroscopy (MRS) provides information about the biochemical substrate of ischemia. Detection of a lactate (Lac) peak is an indicator of anaerobic metabolism in the underlying tissue and the height of the N-acetyl aspartate (NAA) peak is a direct indicator of neuronal integrity. In vivo MRS has been in existence since the mid-1980s, but was limited in its practicability, because it was limited to the interrogation of one large voxel at a time. Technical developments have resulted in the rapid examination of whole tissue slices or even volumes and voxel sizes have been reduced from the order of 8 –1 cc. The ability of MRS to provide metabolic information before cellular structural damage becomes visible promises to shed more light on the ischemic penumbra. Animal data suggests that Lac appears at a CBF threshold of 20 ml/100 g/min and that it appears within minutes of vessel occlusion (Sorensen et al., 1996). There is case material to suggest that elevated Lac and no change in NAA level may indicate reversibility (Barker et al., 1994), offering another MRI method of identifying penumbric tissue and monitoring therapy. MRS has also been used to prognosticate. Low NAA levels combined with infarct volume have helped predict mortality and poor functional outcome (Pereira et al., 1999), as have Lac levels combined with infarct volume measured by DWI (Parsons et al., 2000).

CT perfusion imaging vs. MRI diffusion and perfusion imaging CT technical development has entered a new, rapid phase of development. The combination of spiral techniques with multidetector technology has drastically reduced imaging time, increased spatial resolution and facilitated perfusion studies. Quantitative CBF, cerebral blood volume (CBV) and mean transit times (MTT) can be computed following the injection of iodinated contrast water-soluble medium (Wintermark et al., 2001a). The technique has been validated against xenon CT (Wintermark et al., 2001b) and the technique is limited to four adjacent slices currently, but the impending release of 64 detector machines and more in the future will largely overcome that limitation. The radiation dose involved becomes less of a factor in view of the seriousness of the condition for which the modality is being used and the relatively advanced age of many of the subjects imaged with CT. The perfusion and CTA studies can be performed as an addendum to brain CT more rapidly than can be achieved with MRI. Most significantly, CT machines are far more widely distributed within the community and are particularly conveniently placed in emergency departments. The major deficiency compared with MRI is the inability to identify definitely infarcted brain in a manner comparable to diffusion-weighted sequences. However, it has been hypothesized that reduced CBF in the presence of normal or increased CBV implies viable brain and that reduced CBV is the equivalent of the diffusion defect in MRI (Wintermark et al., 2002a), a theory that was confirmed in a small number of subjects in a comparative study of CT and MRI (Wintermark et al., 2002b), but this has yet to be satisfactorily proven in prospective, comparative studies of substantial subject numbers.

Transient ischemic attacks – role of diffusion imaging and MRS A transient ischemic attack (TIA) has been considered a warning stroke, defined as “an acute disturbance of cerebral function of vascular origin,

165

166

Brian M. Tress

with disability lasting less than 24 h” (Capildeo et al., 1978). All subsequent definitions include the 24-h time limitation. The implication is that no permanent brain damage (infarction) has taken place. Diffusion-weighted MRI studies have demonstrated foci of restricted diffusion in up to 40% of patients imaged within 12 h of the onset of their symptoms (Kidwell et al., 1999). TIAs as defined are therefore a mixture of true ischemic episodes without infarction and probable small infarcts. Most TIAs are thought to be the result of emboli from ulcerated internal carotid artery (ICA) atheromatous plaques, or from the heart, particularly in patients with atrial fibrillation. More controversial is the role of hemodynamic factors. Although a statistically significant improvement in outcome for patients treated by endarterectomy vs. medical therapy has been demonstrated in symptomatic patients with internal carotid stenoses greater than or equal to 70% (National Institute of Neurological Disorders and Stroke, Stroke and Trauma Division, 1991), the same supportive evidence has not been elicited for asymptomatic patients. Furthermore, not all symptomatic patients with hemodynamically significant stenoses treated with endarterectomy or angioplasty and stenting improve. The routine preoperative workup in most centers consists only of one or a combination of the tests listed above; tests which depict disordered anatomy only. No information is provided regarding the physiological state and vascular reserve of the brain parenchyma. CBF and MRS studies show promising signs of addressing these deficiencies. Symptomatic patients with hemodynamically significant carotid stenoses have been shown to have reduction in NAA/choline (Cho) ratio in the ipsilateral centrum semiovale (Van der Grand et al., 1996), but other studies have not reproduced these results. Even asymptomatic patients with hemodynamically significant carotid stenoses have been shown to have significantly improved NAA/creatine (Cr) and Cho/Cr ratios after carotid endarterectomy (CEA) (Kim et al., 2002). Clearly, there have been major advances in the understanding of the underlying pathophysiological basis of many stroke syndromes as a direct result of the development of MRI methods of looking at brain function. The following seven chapters cover in

detail these exciting advances in physiological and biochemical MRI imaging methods made possible by constant advances in magnet and software technology.

REFERENCES Astrup J, Siesjo BK, Symon L. 1981. Thresholds in cerebral ischemia – the ischemic penumbra. Stroke 12: 723–725. Barber PA, Davis SM, Darby DG, et al. 1999. Screening for thrombolytic therapy in acute stroke: diffusion-weighted imaging vs. computed tomography. [Abstract]. Cerebrovasc Dis 9: 73. Barker PB, Gillard JH, van Zijl PC, et al. 1994. Acute stroke: evaluation with serial proton MR spectroscopic imaging. Radiology 192: 723–732. Capildeo R, Haberman S, Rose FC. 1978. The definition and classification of stroke. A new approach. Q J Med 47: 177–196. Darby DG, Barber PA, Gerraty RP, et al. 1999. Pathophysiological topography of acute ischemia by combined diffusion-weighted and perfusion MRI. Stroke 30: 2043–2052. Furlan A, Higashida R, Weschler L, et al. 1999. For the PROACT investigators. Intra-arterial prourokinase for acute ischemic stroke. The PROACT II study: a randomised controlled trial. J Am Med Assoc 282: 2003–2011. Green HA, Pena A, Price CJ, et al. 2002. Increased anisotropy in acute stroke: a possible explanation. Stroke 33: 1517–1521. Kidwell CS, Alger JR, Di Salle F, et al. 1999. Diffusion MR in patients with transient ischemic attacks. Stroke 30: 1174–1180. Kim GE, Lee JH, Cho YP. 2002. Can carotid endarterectomy improve metabolic status in patients with asymptomatic internal carotid artery flow lesion? Studies with localized in vivo proton magnetic resonance spectroscopy. J Vasc Surg 36: 559–564. Libman RB, Wirkowski E, Alvir J, Rao TH. 1995. Conditions that mimic stroke in the emergency department. Implications for acute stroke trials. Arch Neurol 52: 1119–1122. Lovblad KO, Laubach HJ, Baird AE, et al. 1998. Clinical experience with diffusion-weighted MR in patients with acute stroke. Am J Neuroradiol 19: 1061–1066. National Institute of Neurological Disorders and Stroke, Stroke and Trauma Division. 1991. North American symptomatic carotid arterectomy trial (NASCET) Investigators. Clinical alert: benefit of carotid endarterectomy for patients with high grade stenosis of the internal carotid artery. Stroke 22: 816–817.

Cerebrovascular disease: overview

National Institute of Neurological Disorders and Stroke rtPA Stroke Study Group. 1995. Tissue plasminogen activator for acute ischemic stroke. New Eng J Med 333: 1581–1589. Parsons MW, Li T, Barber PA, et al. 2000. Combined (1) H MR spectroscopy and diffusion-weighted MR improves the prediction of stroke outcome. Neurology 55: 485–505. Pereira AC, Saunders DA, Doyle VL, et al. 1999. Measurement of initial N-acetyl aspartate concentration by magnetic resonance spectroscopy and initial infarct volume by MRI predicts outcome in patients with middle cerebral artery territory infarction. Stroke 30: 1577–1582. Pierpaoli C, Jezzard P, Basser PJ, Barnett A, Di Chiro G. 1996. Diffusion tensor MR imaging of the human brain. Radiology 201: 637–648. Rordorf G, Koroshetz WJ, Copen WA, et al. 1998. Regional ischemia and ischemic injury in patients with acute middle cerebral artery stroke as defined by early diffusionweighted and perfusion-weighted MRI. Stroke 29: 939–943. Sorensen AG, Buonanno FS, Gonzalez RG, et al. 1996. Hyperacute stroke: evaluation with combined multisection diffusion-weighted and hemodynamically weighted echoplanar imaging. Radiology 199: 391–401. Symon L, Branston NM, Strong AJ, Hope JD. 1977. The concepts of thresholds of ischemia in relation to brain structure and function. J Clin Pathol Suppl (R Coll Pathol)11: 149–154. Van der Grand J, Eikelboom BC, Mali WPTM. 1996. Flowrelated anaerobic metabolic changes in patients with severe stenosis of the internal carotid artery. Stroke 27: 2026–2032.

von Kummer R, Meyding-Lamade U, Forsting M, et al. 1994. Sensitivity and prognostic value of early computed tomography in middle cerebral artery trunl occlusion. Am J Neuroradiol 15: 9–15. Wang PY, Barker PB, Wiutyk RJ, Ulug AM, van Zijl PC, Beauchamp NJ. 1999. Diffusion-negative stroke: a report of two cases. Am J Neuroradiol 20: 1876–1880. Watanebe T, Honda Y, Fujii Y, Koyama M, Matsuzawa H, Tanaka R. 2001. Three-dimensional anisotropy contrast magnetic resonance axonography to predict the motor function in patients suffering from stroke. J Neurosurg 94: 955–960. Wintermark M, Maeder P, Thiran J-Ph, et al. 2001a. Quantitative assessment of regional cerebral blood flows by perfusion CT studies at low injection rates: a critical review of the underlying theoretical models. Eur Radiol 11: 1220–1230. Wintermark M, Maeder P, Thiran J-Ph, et al. 2001b. Simultaneous measurements of regional cerebral blood flows by perfusion-CT and stable xenon-CT: a validation study. Am J Neuroradiol 22: 905–914. Wintermark M, Reichhart M, Thiran J-P, et al. 2002a. Prognostic value of cerebral blood flow measurement by perfusion computed tomography, at the time of emergency room admission, in acute stroke patients. Ann Neurol 51: 417–432. Wintermark M, Reichhart M, Cuisenaire O, et al. 2002b. Comparison of admission perfusion computed tomography and qualitative diffusion- and perfusion-weighted magnetic resonance imaging in acute stroke patients. Stroke 33: 2025–2031. Yang Q, Tress BM, Barber PA, et al. 1999. Serial study of apparent diffusion coefficient and anisotropy in patients with acute stroke. Stroke 30: 2382–2390.

167

11

MR spectroscopy in stroke Peter B. Barker1 and Jonathan H. Gillard2 1 2

The Russell H. Morgan Department of Radiology and Radiological Science, The Johns Hopkins University School of Medicine, Baltimore, USA University Department of Radiology, Addenbrooke’s Hospital, Cambridge, UK

Key points • Localized decreased N-acetyl aspartate (NAA) is seen within a few hours of ischemia onset, and is typically very low or absent in chronic infarcts. • Lactate (Lac) is elevated in acute stroke due to anaerobic glycolysis in ischemic brain. • Lac in sub-acute or chronic stroke may be due to the presence of macrophages. • Creatine and choline may change in acute and chronic stroke. Lipid and other macromolecule resonances may reflect necrosis. • Added clinical value of MR spectroscopy (MRS) to diffusion and perfusion MR in ischemic stroke is uncertain. • MRS yields information useful for research and possibly for therapeutic trials in stroke.

Proton MR spectroscopy (MRS) of the human brain was first demonstrated in the mid-1980s (Bottomley et al., 1985; Luyten and den Hollander, 1986; Hanstock et al., 1988; Frahm et al., 1989), and shortly thereafter the first reports of its application to the study of human stroke appeared (Berkelbach van der Sprenkel et al., 1988, Bruhn et al., 1989). Although there have been reports of 31P (Helpern et al., 1993), 23Na (Thulborn et al., 1999) and 13C (Rothman et al., 1991) spectroscopy in human stroke, the majority of studies to date have utilized the proton nucleus, because of its high sensitivity and logistical convenience 168

(i.e. proton MRS can be readily combined with conventional MR imaging (MRI). Most studies of proton MRS of the human brain have focused on the signals from N-acetyl aspartate (NAA) and lactate (Lac), as potential surrogate markers of neuronal integrity and ischemia, respectively, although there are also often changes in the other metabolite signals, particularly in the chronic stages of brain infarction. The significance of changes in brain metabolites in the context of ischemia and infarction is briefly discussed below (for a more detailed review of these compounds and their significance to other pathologies, please see Chapter 1).

NAA The evidence for and against NAA (2.02 ppm) as a neuronal marker is discussed in Chapter 1. Overall, there is reasonably good evidence that NAA, at least in some pathologies, can be used as a surrogate marker for neuronal function and integrity. Therefore, there is reason to hope that MRS measurements of NAA in patients with cerebrovascular disease may provide some information on the neuronal integrity of the brain, and perhaps the degree of stroke progression. Information on the metabolic changes that occur during ischemia (and in particular changes in NAA) comes in large part from animal models. Various studies have looked at the timecourse of NAA changes following the induction of

MR spectroscopy in stroke

ischemia (Monsein et al., 1993; Sager et al., 1995, 1999; Higuchi et al., 1997). In animal models of focal ischemia (middle cerebral artery (MCA) occlusion), it has consistently been shown that NAA declines quite slowly over a time scale of hours after the induction of ischemia. Typically, NAA may be reduced to 50% of its pre-ischemic value after approximately 10–12 h (Monsein et al., 1993), although this time-course is probably highly sensitive to the degree of cerebral blood flow (CBF) reduction. For instance, NAA reductions were found to be greater in the core of the ischemic region compared to the periphery in the study by Higuchi et al. (1996) – Figure 11.1. Several papers have also described an initial rapid decrease in NAA of about 10% within the first few minutes (van Zijl and Moonen, 1993; Sager et al., 1995; Higuchi et al., 1996) followed by a slower decrease in NAA with a time constant of hours. The origin of this is uncertain, but it has been suggested that this may be due to the presence of more than one pool of NAA, or perhaps changes in other molecules (e.g. glutamate (Glu), glutamine (Gln), -amino-butyric acid (GABA) etc.) which overlap with the spectral resonance of NAA. It might also be due to the dilutional effect of cytotoxic edema. Since it is generally thought that irreversible brain damage in most of these models occurs over the first 1–3 h of ischemia, the reduction in NAA may occur more slowly than the loss of tissue viability. In infarcted brain (e.g. weeks to months after stroke onset), either in animal models or in humans, NAA is generally either very low or completely absent, consistent with the complete loss of neuronal tissue from these regions (Mathews et al., 1995; Gillard et al., 1996; Berkelbach van der Sprenkel et al., 1988). Figure 11.2 shows an example of a hemispheric loss of NAA in a patient with a tandem occlusion of the left internal carotid artery (ICA) and MCA who was imaged 24 h after stroke onset. In addition to loss of NAA, this case also exhibited elevated Lac and choline (Cho) signals (cf. below).

Creatine The creatine (Cr) signal (“Cr”, 3.02 ppm) is actually the sum of both Cr and phosphocreatine (PCr),

which are in exchange via the Cr kinase reaction. Because induction of ischemia causes PCr to be converted to Cr, but no net change in the total Cr this has lead some authors to suggest that the Cr signal would be a good reference signal in the spectrum. However, more recent studies have suggested that Cr may change in both acute and chronic infarction; it is probably unwise to always assume Cr levels to be normal in human stroke.

Choline The “Cho” signal (3.24 ppm) arises from the trimethylamine groups of glycerophosphocholine (GPC), phosphocholine (PC), and a small amount of free Cho (Barker et al., 1994a), compounds which are involved in membrane synthesis and degradation. Cho has been observed to either be increased in chronic human stroke (Duijn et al., 1992; Barker et al., 1994b). Increases in Cho in stroke may be the result of gliosis or ischemic damage to myelin, while decreases are probably the result of edema, necrosis and cell loss.

Lactate In normal human brain, Lac (1.33 ppm) is below (or at the limit of) detectibility in most spectra. When the brain becomes ischemic or hypoxic, the lack of oxygen results in an inability for glucose to be metabolized through the tricarboxylic acid (TCA) cycle (its normal pathway), and the alternative, less efficient non-aerobic pathway from pyruvate to Lac is employed instead. Hence, ischemia in the brain causes an elevation in brain Lac (Rehncrona et al., 1981). Lac may also be detected in chronic brain infarction, where its presence may be explained by the glycolytic metabolism of macrophages, rather than to on-going chronic ischemia (Petroff et al., 1992). Lac may also accumulate in the extracellular space, cysts and in cerebrospinal fluid. It should also be recognized that ischemia and infarction are not the only causes of increased brain Lac. For instance, lac is quite frequently observed

169

170

Peter B. Barker and Jonathan H. Gillard

(a)

NAA

1.0 0.8

0.8

0.6

0.6

0.4

0.4

0.2

NAA

1.0

Central

0.2

Peripheral 0.0

0.0 5

10

15

20

25 (h)

Lac

1.0 0.8 0.6 0.4 0.2 0.0 2

4

6

8

10 30

50 70 (h)

(b)

Cho Cr

NAA Lac

Cho Cr

NAA Lac

2.25

12

0.25

9

Pre

4.5 5

4

3 ppm

2

1

5

4

3 2 ppm

1

Fig. 11.1 NAA and Lac measured as a function of time after the onset of ischemia in a variety of animal models. (a) Left – NAA and lac in the core and periphery of a focal ischemic lesion in rat. Lac is higher in the core than the periphery, and increases steadily over the first 8 h. NAA decreases more slowly, with large changes only seen at greater than 30 h of ischemia. NAA is reduced more in the core than the periphery. Right – NAA determined biochemically in a model of complete global ischemia in rat. After an inital 10% decline (also seen in a), NAA decreases gradually over a 24 h period. Adapted from Higuchi et al. (1996) and Sager et al. (1995). (b) Spectra recorded as a function of time in the basal ganglia of the baboon after permanent occlusion of the MCA. Lac is seen to increase and NAA decrease over a time period of 12 h.

MR spectroscopy in stroke

T1

Cho

Left

Lac Cho Cr

NAA

NAA

Lac

Right

ppm

4.0

3.0

2.0

1.0

Fig. 11.2 Proton (MR spectroscopic imaging) MRSI in a patient with tandem occlusion of the left ICA and MCA, imaged 24 h after symptom onset. T1-weighted localized image shows swelling of the left hemisphere but no signal abnormality. Spectroscopic images show elevated Lac and near absent NAA throughout the left hemisphere, consistent with severe ischemic injury. Cho is also focally elevated, particularly in the white matter (WM) regions of the left hemisphere. The patient died 2 days after the imaging was performed.

in brain tumors (Alger et al., 1990), mitochondrial diseases (Mathews et al., 1993) and other conditions such as inflammatory demyelinating lesions (Kruse et al., 1994). Small elevations of Lac have also been reported in the visual cortex (VC) during photic stimulation (Prichard et al., 1991), believed to be due to increased non-oxidative glycolysis, but this effect does not appear to be particularly reproducible (Merboldt et al., 1992).

Macromolecules/lipids In spectra recorded at short echo times (TE), a number of broad signals originating from macromolecules with short T2 can be detected which underlie the sharper, metabolite resonances. Discrimination of these macromolecular signals from metabolites

can be enhanced using a inversion recovery pulse sequence, based on the shorter T1 relation times of the macromolecules (Behar et al., 1994). The assignment of the macromolecules has been the source of some debate, possible contributors being cytosolic proteins or lipids. Although these signals are normally fairly small, pathological processes such as necrosis or apoptosis can cause substantial increases to occur. When these signals resonate between 0.9 and 1.3 ppm, their most likely assignment is from the methylene and methyl groups of mobile saturated lipids, probably accumulating as a result of cell membrane degradation. Increase of lipid signals have been observed in both sub-acute and chronic stroke (Saunders et al., 1995; Hwang et al., 1996; Saunders et al., 1997). Again, these are not specific to ischemic damage, and may be seen in necrotic regions of tumors and infective or inflammatory lesions.

171

172

Peter B. Barker and Jonathan H. Gillard

Metabolic changes during cerebral ischemia: relationship to blood flow The two signals in the proton MR spectrum which have received the most attention in animal models of ischemia are NAA and Lac, on the basis of their potential roles as indicators of neuronal viability and ischemia, respectively. MRS has the advantage over conventional analytical biochemical techniques (for these compounds and others) in that it is noninvasive, so that repetitive serial measurements can be made in the same animal, thus allowing the time-course of ischemic changes to be mapped without possible errors to due tissue sampling, preparation or inter-animal variations. As CBF decreases, various processes related to cerebral homeostasis gradually fail (Hossman, 1994). Although it can be difficult to ascribe specific thresholds for individual processes, it appears that once CBF has decreased below 15–20 ml/100 g/min (Crockard et al., 1987), the brain becomes ischemic, with the cessation of electrical function, and the transfer of energy metabolism from aerobic to anaerobic glycolysis with accumulation of Lac. Reported CBF thresholds may vary depending on the animal model used, the type of anesthesia, the type and duration of ischemia, arterial oxygenation and hematocrit, and the method used to measure CBF. However, in complete, global ischemia induced by cardiac arrest, Lac levels rise abruptly (Petroff et al., 1988) and reach a steady state within 10 min of cessation of blood flow. As Lac accumulates, the tissue may become acidotic (Crockard et al., 1987). The “final’’ Lac concentration depends on a number of factors, but in particular on the pre-ischemic blood glucose and brain glycogen stores (Petroff et al., 1988). Under normoglycemic conditions, Lac may typically reach 10–12 mM (Rehncrona et al., 1981). Pre-ischemic hyperglycemia may increase final Lac concentrations, and worsen eventual clinical outcome. If ischemia is incomplete, or reperfusion occurs, blood flow continues to supply glucose to the tissue, which, if sufficiently damaged, is unable to metabolize it aerobically, and extremely high Lac concentrations may result (Rehncrona et al., 1981).

In models of focal ischemia (where presumably CBF reductions are more moderate because of collateral circulation), the increase in Lac may often be significantly slower, increasing over a period of hours (Monsein et al., 1993; Sager et al., 1995; Higuchi et al., 1996; Dreher et al., 1998; Norris et al., 1998). For instance, in a permanent MCA occlusion model, Lac was observed to increase steadily up to 12 h after induction of ischemia (Monsein et al., 1993). In one report, it was also suggested that transient Lac elevations coincided with burst of cortical spreading depression (CSD) in peri-infarct tissue, which has been postulated to be a mechanism for infarct enlargement into surrounding tissues (Norris et al., 1998). In addition to the increase in Lac, NAA is observed to decrease following the onset of ischemia. It would appear that the rate of NAA decrease (like that of the Lac increase) is dependent on the degree of blood flow reduction to the ischemic tissue, but it is likely that the CBF thresholds for these processes are different, and that they also have different time constants. For instance, in both animal models of ischemia and in human stroke, elevated Lac in periinfarct regions with near normal NAA levels has been reported (Barker et al., 1994b, Dreher et al., 1998). It is tempting to speculate that tissue may represent an ischemic penumbra of dysfunctional tissue, with relative neuronal preservation (coined a “metabolic penumbra” (Gillard et al., 1996)), although at present this concept is untested. If the duration and severity of ischemia is short enough (e.g. no more than a few minutes in the case of complete ischemia), then most of the metabolic alterations described above are reversible, i.e. establishment of reperfusion will result in restoration of normal metabolite levels and function (Nagatomo et al., 1995). Reperfusion after a longer period of ischemia may result in initial restoration of metabolite levels, only to be followed by secondary energy failure over the subsequent 24–48 h (60, 66). As this secondary energy failure continues, or in the case of permanent ischemia, irreversible changes occur and the tissue will progress to neuronal loss, infarction and gliosis. These longer term changes can also be detected with MRS. In the first two

MR spectroscopy in stroke

papers reporting MRS of human brain infarction, NAA was completely absent from both infarcted tissue at 4 days (Bruhn et al., 1989) and at 10 months post stroke onset (Berkelbach van der Sprenkel et al., 1988). As described above, other metabolic changes have also been reported in the chronic stage of stroke; these include increases of Cho containing compounds (Barker et al., 1994b) and mobile lipid signals (Saunders et al., 1995).

Application of proton spectroscopy to human stroke The earliest studies of 1H MRS of human stroke used single-voxel localization techniques (Berkelbach van der Sprenkel et al., 1988; Bruhn et al., 1989). Using single voxel (SV) MRS, it was found that elevated Lac and decreased NAA levels could be detected in cases of acute (24 h) (Felber et al., 1992; Gideon et al., 1992; Henriksen et al., 1992), sub-acute (24 h 7 days) (Bruhn et al., 1989; Fenstermacher and Narayana, 1990; Henriksen et al., 1992) and chronic (7 days) (Fenstermacher and Narayana, 1990; Felber et al., 1992; Sappey-Marinier et al., 1992; Gideon et al., 1994; Berkelbach van der Sprenkel et al., 1988) stroke. Single voxel techniques, however, do not provide information regarding the spatial distribution and extent of metabolic abnormalities, and require that the location of the ischemic or infarcted region be already known or visible on MRI studies. To address these issues, it is possible to use MR spectroscopic imaging (MRSI) methods for the study of cerebral ischemia, either in one (Graham et al., 1992, 1993; Petroff et al., 1992) or two spatial dimensions (Duijn et al., 1992, Hugg et al., 1992), or using multi-slice 2D MRSI (Barker et al., 1994b; Gillard et al., 1996). A case study of an acute stroke patient was reported (Barker et al., 1994b) using multi-slice MRSI; (cf. Case Study 11.1) since this case nicely represents the evolving metabolic changes detectable by proton MRSI, it will be described here. The patient presented with a dense left hemiparesis as the result of a complete occlusion of the right ICA,

and low flow in the right MCA. Conventional T2weighted MR images were normal. However, proton MRSI performed 24 h after symptom onset revealed elevated Lac throughout much of the right MCA territory, with the highest concentration in the basal ganglia (Figure 11.3(a)). At this stage, much of this tissue may still have been salvageable, since no signs of infarction (T2 hyperintensity and severe reduction of NAA) were visible. The patient was treated with hypervolemic hypertensive therapy and improved clinically; follow-up imaging performed 1 week later (Figure 11.3(b)) found that the much of the previously ischemic lateral cortex had resolved (with no elevation of Lac or other abnormality), but that the basal ganglia had progressed to infarct (T2 hyperintensity), with almost complete depletion of NAA. Also of note is the high Cho signal in the peri-infarct white matter (WM). Follow-up imaging at 5 months (Figure 11.3(c)) showed a new, small infarct adjacent to the anterior horn of the right lateral ventricle, and proton MRSI revealed extensive reductions of NAA throughout much of the anterior right hemisphere, and high Cho in WM. No Lac was detected at this stage. The widespread NAA reductions were consistent with the patient’s clinical status, which had deteriorated at this stage, with the return of the dense left hemiparesis. This case shows that, as expected, in the earliest stages of stroke, the only spectroscopic abnormality is an increase in Lac. If no other abnormalities are detected (e.g. signs of infarction in conventional MR images), then tissue with this characteristic may represent ischemic tissue at risk of infarction, i.e. a penumbra. Identification of tissue of this type may be a primary condition for selecting patients for therapeutic intervention. It is interesting to note that some of the tissue which was ischemic on day 1 progressed to infarction (basal ganglia), while the rest (lateral cortex) apparently resolved. Therefore, the detection of Lac does not necessarily predict infarction, since the eventual outcome depends on a number of factors, particularly the temporal evolution of regional blood flow. The basal ganglia are an end-arterial region which does not have extensive collateral circulation, which could explain its progression to infarction in this case.

173

174

Peter B. Barker and Jonathan H. Gillard

(a) T 2

Cho

(A) Right basal ganglia NAA Cho Cr

A

Lac

B

Lac

NAA

(B) Left basal ganglia

ppm (b) T2

Cho

4.0

3.0

2.0

1.0

(A) Right basal ganglia Cho Lac

A

NAA

Cr

B B

Lac

NAA

(B) Left basal ganglia

ppm

4.0

3.0

2.0

1.0

Fig. 11.3 (a) T2 MRI, and proton spectroscopic imaging data in an acute stroke patient with occlusion of the right ICA and low flow in the right MCA 24 h after symptom onset. No changes are visible in the conventional T2 weighted MRI; however, spectroscopic images show an elevation of lactate through much of the right MCA territory, indicating ischemia. NAA is relatively preserved in the right hemisphere. (b) Follow-up imaging performed 1 week later shows development of infarction in the right basal ganglia (T2 hyperintensity, absence of NAA). Lac is present only in the region of infarction. High Cho is seen in the peri-infarct WM. (c) Follow-up imaging performed 5 months after symptom onset. A new, small infarct has appeared anterior to the right frontal horn of the lateral ventricle, and proton MRSI now reveals extensive reductions of NAA throughout much of the anterior right hemisphere, and high Cho in WM. No Lac is detected at this stage, although a small signal attributable to mobile lipid is seen in the infarcted right basal ganglia.

MR spectroscopy in stroke

(c) T 2

Cho

(A) Right basal ganglia Cho NAA

A

B

Cr Lipid

NAA

Lac

(B) Left basal ganglia

ppm

4.0

3.0

2.0

1.0

Fig. 11.3 (cont.)

Potential clinical applications Staging of stroke progression: selection of patients for thrombolysis The primary role for imaging of acute stroke is to diagnose stroke type and etiology, in order to stratify patients to the appropriate treatment protocol. Currently CT (or MRI, if available) is used to identify which patients would be ineligible for thrombolysis, either on the basis of the presence of hemorrhage or early signs of large infarction. Ideally, selection for thrombolysis should be based on a positive diagnosis of stroke, with the demonstration of ischemic but viable tissue that could be salvaged by the establishment of reperfusion. Several imaging techniques, including positron emission tomography (PET), and diffusion- and perfusion-weighted MRI (diffusion-weighted imaging (DWI)/perfusion weighted imaging (PWI) (Sorensen et al., 1996) show great promise for this purpose. However, the use of PET in acute stroke appears to be extremely difficult

except at a few dedicated PET centers, because of its lack of availability, particularly on an emergent basis. On the other hand, the integration of MRSI with DWI/PWI has only just begun to be explored, but promises to add important information, particularly in patients who may have a larger perfusion deficit than diffusion (i.e. a PWI/DWI mismatch). In the PWI–DWI mismatch region (now often thought to be a “target” therapeutic regions) MRSI measurements of Lac and NAA might provide additional information of the status of the tissue in terms of level of ischemia and neuronal viability. Ultimately, the decision to thrombolyze or not (or to initiate other acute stroke treatments such as neuroprotection) could be based on imaging criteria rather than on time post onset, which might allow a greater of number of patients to be eligible for these interventions. However, it should be recognized that the integration of MRS or MRSI into the rapid imaging protocols required for acute stroke is extremely difficult, and will require the routine implementation of fast MRS or MRSI techniques that are currently available only on research systems.

175

176

Peter B. Barker and Jonathan H. Gillard

Another obstacle to performing successful MRS or MRSI in acute stroke is head motion due to lack of patient cooperation-this can severely degrade spectral quality and make the spectrum virtually uninterpretable. In particular, head motion can cause lipid contamination from the scalp which can easily obscure smaller Lac signals in ischemic brain regions. For all of these reasons, there is a paucity of studies currently of MRS or MRSI in acute stroke.

measurements have been shown to correlate well with disability indices in other disorders, such as multiple sclerosis (MS) (De Stefano et al., 2001). It seems likely that such relationships also exist in cerebrovascular disease, and in particular measurements of NAA in peri-infarct or watershed regions may be useful for evaluating the functional significance of selective neuronal loss.

Prognostic indicator Selection of patients for other therapies Since MRSI is currently difficult to perform in acute stroke, it is logical that its more immediate application may occur in evaluating patients with subacute stroke, or evaluating patients who may be eligible for other less urgent treatments such carotid endarterectomy (CEA) or hypervolemic hypertensive therapy (Rutgers et al., 2000). For instance, in one study (Klijn et al., 2000), a low ipsilateral ratio of NAA/Cho in patients who had symptomatic carotid artery occlusion was found to be predictive of recurrent or further ischemic events, suggesting that this finding may be useful in deciding who should have CEA. In comparing patients before and after CEA, it was found that the NAA/Cho ratio improved post CEA, but only in patients who did not have Lac (in non-infarcted tissue ipsilateral to the occlusion) prior to CEA (van der Grond et al., 1996). These results and others indicate that proton MRS may be useful for both the selection of patients and monitoring of sub-acute stroke treatments, and in particular CEA.

Outcome measures Another role where MRSI may play a role in cerebrovascular disease is as an outcome measure for therapeutic trials. In one study of chronic stroke, SV measurements of NAA in the internal capsule correlated with performance on motor tasks, suggesting that it may be a good surrogate marker for axonal preservation in the internal capsule and corticalspinal tract (Pendlebury et al., 1999). Although there have been only isolated reports of functional and metabolic correlations in stroke, global NAA

There have only been a few studies of the prognostic value of MRS in acute stroke (Federico et al., 1996; Pereira et al., 1999; Parsons et al., 2000). Pereira et al. (1999) found significantly lower NAA in patients who ultimately died or were dependent on others for their daily living activities, as opposed to those who were able to live independently, and the prognostic value was enhanced by combining NAA levels with acute infarct volumes on T2 MR (Pereira et al., 1999). Similar results have also been found by Federico et al. (1996), while Parsons et al. (2000) found good predictive value based on acute measurements of Lac levels and lesion volume as detected by DWI.

Summary: future developments and research applications The role that MRS and MRSI have to play in the evaluation of human cerebrovascular disease still remains to be determined. Proton spectroscopy provides metabolic information, which potentially could be of great value in optimizing management of ischemic stroke patients. Proton MRS and MRSI also promises to be widely available (compared to other metabolic imaging techniques such as PET) in the near future, since it is readily implemented on most existing 1.5 or 3.0 T MRI machines. A further advantage is that it can be combined with current MRI stroke protocols which include sequences for the detection of chronic infarction (e.g. T2weighted fast spin echo (FSE) and fluid attenuated inversion recovery (FLAIR)), early ischemic changes (DWI), cerebral perfusion (rapid imaging during

MR spectroscopy in stroke

bolus injection of GdDTPA contrast agent (PWI)), hemorrhage (gradient-echo, T2*-weighted MRI) and MR angiography (MRA) for the detection of occlusion or stenosis of major vessels. This combined MRI and MRSI exam should become routinely possible with the development of appropriate fast MRSI techniques, and promises to provide a complete evaluation of structure, metabolism and hemodynamics in acute stroke. DWI and PWI are already having a significant impact on stroke evaluation and management in many medical centers, and an important question which remains to be determined is the “added value” that MRSI might provide in addition to these sequences. Should all these techniques, and MRS, be performed in evaluating an acute stroke patient? Time constraints, particularly if the patient is to be enrolled in a treatment protocol, probably forbid all of these different sequences to be performed. However, potential examples where MRSI may be helpful include the ability to define the metabolic status of the PWI/ DWI “mismatch” region, and to determine reversible vs. irreversible tissue damage based on NAA levels. For instance, a recent study (Parsons et al., 2002) elegantly showed that MRS measurements of Lac in the DWI/PWI mismatch region correlated with both blood glucose levels and poor final outcome, suggesting that lactic acidosis of the penumbral region could indeed hasten the conversion of penumbral tissue to infarction. MRSI measurements of NAA, Cho and lipids in chronic stroke (not necessarily in regions with MRI visible infarction) could be important outcome measures (surrogate markers) in trials of new stroke therapies. Finally, it seems likely that research applications of proton spectroscopy will continue to be developed. For instance, one interesting recent example used diffusion-weighted MRS to probe the mechanisms controlling changes in diffusion constants in acute ischemic tissue (Dreher et al., 2001). It was found that the % reduction in metabolic diffusion constants (NAA, Cr, Cho: approximately 67–77% of the contralateral hemisphere in a unilateral MCA occlusion model) was almost identical to that of water. Since these metabolites are known to be exclusively located in the intracellular space, these results indicate a generalized reduction of diffusion of all

molecules in the intracellular environment, and do not indicate any special mechanism involving selective motion of water. Such information is useful in understanding the mechanisms of ischemic signal changes in DWI. In conclusion, MRS and MRSI will continue to be important research tools for the investigation of mechanisms of ischemic brain damage. With continued technique development, clinical applications in cerebrovascular disease should also emerge, in particular for selection and monitoring the effects of treatments for acute and sub-acute ischemic stroke.

REFERENCES Alger JR, Frank JA, Bizzi A, Fulham MJ, DeSouza BX, Duhaney MO, Inscoe SW, Black JL, van ZP, Moonen CT, et al. 1990. Metabolism of human gliomas: assessment with H-1 MR spectroscopy and F-18 fluorodeoxyglucose PET [see comments]. Radiology 177: 633–641. Barker P, Breiter S, Soher B, Chatham J, Forder J, Samphilipo M, Magee C, Anderson J. 1994a. Quantitative proton spectroscopy of canine brain: in vivo and in vitro correlations. Magn Reson Med 32: 157–163. Barker PB, Gillard JH, van Zijl PCM, Soher BJ, Hanley DF, Agildere AM, Oppenheimer SM, Bryan RN. 1994b. Acute stroke: evaluation with serial proton magnetic resonance spectroscopic imaging. Radiology 192: 723–732. Behar KL, Rothman DL, Spencer DD, Petroff OA. 1994. Analysis of macromolecule resonances in 1H NMR spectra of human brain. Magn Reson Med 32: 294–302. Berkelbach van der Sprenkel JW, Luyten PR, van Rijen PC, Tulleken CA, den Hollander JA. 1988. Cerebral lactate detected by regional proton magnetic resonance spectroscopy in a patient with cerebral infarction. Stroke 19: 1556–1560. Bottomley PA, Edelstein WA, Foster TH, Adams WA. 1985. In vivo solvent-suppressed localized hydrogen nuclear magnetic resonance spectroscopy: a window to metabolism? Proc Natl Acad Sci USA 82: 2148–2152. Bruhn H, Frahm J, Gyngell ML, Merboldt KD, Hanicke W, Sauter R. 1989. Cerebral metabolism in man after acute stroke: new observations using localized proton NMR spectroscopy. Magn Reson Med 9:126–131. Crockard HA, Gadian DG, Frackowiak RS, Proctor E, Allen K, Williams SR, Russell RW. 1987. Acute cerebral ischaemia: concurrent changes in cerebral blood flow, energy

177

178

Peter B. Barker and Jonathan H. Gillard

metabolites, pH, and lactate measured with hydrogen clearance and 31P and 1H nuclear magnetic resonance spectroscopy II. Changes during ischaemia. J Cereb Blood Flow Metab 7: 394–402. De Stefano N, Narayanan S, Francis GS, Arnaoutelis R, Tartaglia MC, Antel JP, Matthews PM, Arnold DL. 2001. Evidence of axonal damage in the early stages of multiple sclerosis and its relevance to disability. Arch Neurol 58: 65–70. Dreher W, Busch E, Leibfritz D. 2001. Changes in apparent diffusion coefficients of metabolites in rat brain after middle cerebral artery occlusion measured by proton magnetic resonance spectroscopy. Magn Reson Med 45: 383–389. Dreher W, Kuhn B, Gyngell ML, Busch E, Niendorf T, Hossmann KA, Leibfritz D. 1998. Temporal and regional changes during focal ischemia in rat brain studied by proton spectroscopic imaging and quantitative diffusion NMR imaging. Magn Reson Med 39: 878–888. Duijn JH, Matson GB, Maudsley AA, Hugg JW, Weiner MW. 1992. Human brain infarction: proton MR spectroscopy. Radiology 183: 711–718. Federico F, Simone IL, Conte C, Lucivero V, Giannini P, Liguori M, Picciola E, Tortorella C. 1996. Prognostic significance of metabolic changes detected by proton magnetic resonance spectroscopy in ischaemic stroke. J Neurol 243: 241–247. Felber SR, Aichner FT, Sauter R, Gerstenbrand F. 1992. Combined magnetic resonance imaging and proton magnetic resonance spectroscopy of patients with acute stroke. Stroke 23: 1106–1110. Fenstermacher MJ, Narayana PA. 1990. Serial proton magnetic resonance spectroscopy of ischemic brain injury in humans. Invest Radiol 25: 1034–1039. Frahm J, Bruhn H, Gyngell ML, Merboldt KD, Hanicke W, Sauter R. 1989. Localized high-resolution proton NMR spectroscopy using stimulated echoes: initial applications to human brain in vivo. Magn Reson Med 9: 79–93. Gideon P, Henriksen O, Sperling B, Christiansen P, Olsen TS, Jorgensen HS, Arlien-Soborg P. 1992. Early time course of N-acetylaspartate, creatine and phosphocreatine, and compounds containing choline in the brain after acute stroke. Stroke 23: 1566–1572. Gideon P, Sperling B, Arlien-Soborg P, Olsen TS, Henriksen O. 1994. Long-term follow-up of cerebral infarction patients with proton magnetic resonance spectroscopy. Stroke 25: 967–973. Gillard JH, Barker PB, van Zijl PCM, Bryan RN, Oppenheimer SM. 1996. Proton MR spectroscopic imaging in acute middle cerebral artery stroke Am J Neuroradiol 17: 873–886. Graham G, Blamire A, Rothman D, Brass L, Fayad P, Petroff O, Prochard J. 1993. Early temporal variation of cerebral metabolites after human stroke. Stroke 24: 1891–1896.

Graham GD, Blamire AM, Howseman AM, Rothman DL, Fayad PB, Brass LM, Petroff OA, Shulman RG, Prichard JW. 1992. Proton magnetic resonance spectroscopy of cerebral lactate and other metabolites in stroke patients Stroke 23: 333–340. Hanstock CC, Rothman DL, Prichard JW, Jue T, Shulman RG. 1988. Spatially localized 1H NMR spectra of metabolites in the human brain. Proc Natl Acad Sci USA 85: 1821–1825. Helpern JA, Vande Linde AMQ, Welch KMA, Levine SR, Schultz LR, Oridge RJ, Halvorson HR, Hugg JW. 1993. Acute elevation and recovery of intracellular [Mg2+] following human focal cerebral ischemia. Neurology 43: 1577–1581. Henriksen O, Gideon P, Sperling B, Olsen TS, Jorgensen HS, Arlien-Soborg P. 1992. Cerebral lactate production and blood flow in acute stroke. J Magn Reson Imaging 2: 511–517. Higuchi T, Fernandez EJ, Maudsley AA, Shimizu H, Weiner MW, Weinstein PR. 1996. Mapping of lactate and N-acetyl-L-aspartate predicts infarction during acute focal ischemia: in vivo 1H magnetic resonance spectroscopy in rats. Neurosurgery 38: 121–129; discussion 129–130. Higuchi T, Graham SH, Fernandez EJ, Rooney WD, Gaspary HL, Weiner MW, Maudsley AA. 1997. Effects of severe global ischemia on N-acetylaspartate and other metabolites in the rat brain. Magn Reson Med 37: 851–857. Hossman K-A. 1994. Viability thresholds and the penumbra of focal ischemia. Ann Neurol 36: 557–565. Hugg JW, Duijn JH, Matson GB, Maudsley AA, Tsuruda JS, Gelinas DF, Weiner MW. 1992. Elevated lactate and alkalosis in chronic human brain infarction observed by 1H and 31P MR spectroscopic imaging. J Cereb Blood Flow Metab 12: 734–744. Hwang JH, Graham GD, Behar KL, Alger JR, Prichard JW, Rothman DL. 1996. Short echo time proton magnetic resonance spectroscopic imaging of macromolecule and metabolite signal intensities in the human brain. Magn Reson Med 35: 633–639. Klijn CJ, Kappelle LJ, van der Grond J, Algra A, Tulleken CA, van Gijn J. 2000. Magnetic resonance techniques for the identification of patients with symptomatic carotid artery occlusion at high risk of cerebral ischemic events. Stroke 31: 3001–3007. Kruse B, Barker PB, van Zijl PCM, Duyn JH, Moonen CTW, Moser HW. 1994. Multislice proton MR spectroscopic imaging in X-linked adrenoleukodystrophy. Ann Neurol 36: 595–608. Luyten PR, den Hollander JA. 1986. Observation of metabolites in the human brain by MR spectroscopy. Radiology 161: 795–758.

MR spectroscopy in stroke

Mathews PM, Andermann F, Silver K, Karpati G, Arnold DL. 1993. Proton MR spectroscopic characterization of differences in regional brain metabolic abnormalities in mitochondrial encephalomyopathies. Neurology 43: 2484–2490. Mathews VP, Barker PB, Blackband SJ, Chatham JC, Bryan RN. 1995. Cerebral metabolites in patients with acute and subacute strokes: concentrations determined by quantitative proton MR spectroscopy. Am J Roentegenol 165: 633–638. Merboldt K-D, Bruhn H, Hanicke W, Michaelis T, Frahm J. 1992. Decrease of glucose in the human visual cortex during photic stimulation. Magn Reson Med 25: 187–194. Monsein LH, Mathews VP, Barker PB, Pardo CA, Blackband SJ, Whitlow WD, Wong DF, Bryan RN. 1993. Irreversible regional cerebral ischemia: serial MR imaging and proton MR spectroscopy in a nonhuman primate model. Am J Neuroradiol 14: 963–970. Nagatomo Y, Wick M, Prielmeier F, Frahm J. 1995. Dynamic monitoring of cerebral metabolites during and after transient global ischemia in rats by quantitative proton NMR spectroscopy in vivo. NMR Biomed 8: 265–270. Norris DG, Hoehn-Berlage M, Dreher W, Kohno K, Busch E, Schmitz B. 1998. Characterization of middle cerebral artery occlusion infarct development in the rat using fast nuclear magnetic resonance proton spectroscopic imaging and diffusion-weighted imaging. J Cereb Blood Flow Metab 18: 749–757. Parsons MW, Barber PA, Desmond PM, Baird TA, Darby DG, Byrnes G, Tress BM, Davis SM. 2002. Acute hyperglycemia adversely affects stroke outcome: a magnetic resonance imaging and spectroscopy study. Ann Neurol 52: 20–28. Parsons MW, Li T, Barber PA, Yang Q, Darby DG, Desmond PM, Gerraty RP, Tress BM, Davis SM. 2000. Combined (1)H MR spectroscopy and diffusion-weighted MRI improves the prediction of stroke outcome. Neurology 55: 498–505. Pendlebury ST, Blamire AM, Lee MA, Styles P, Matthews PM. 1999. Axonal injury in the internal capsule correlates with motor impairment after stroke. Stroke 30: 956–962. Pereira AC, Saunders DE, Doyle VL, Bland JM, Howe FA, Griffiths JR, Brown MM. 1999. Measurement of initial Nacetyl aspartate concentration by magnetic resonance spectroscopy and initial infarct volume by MRI predicts outcome in patients with middle cerebral artery territory infarction. Stroke 30: 1577–1582. Petroff OA, Prichard JW, Ogino T, Shulman RG. 1988. Proton magnetic resonance spectroscopic studies of agonal carbohydrate metabolism in rabbit brain. Neurology 38: 1569–1574. Petroff OAC, Graham GD, Blamire AM, al Rayess M, Rothman DL, Fayad PB, Brass LM, Shulman RG, Prichard JW. 1992. Spectroscopic imaging of stroke in humans: histopathology correlates of spectral changes. Neurology 42: 1349–1354.

Prichard J, Rothman D, Novotny E, Petroff O, Kuwabara T, Avison M, Howseman A, Hanstock C, Shulman R. 1991. Lactate rise detected by 1H NMR in human visual cortex during physiologic stimulation. Proc Natl Acad Sci USA 88: 5829–5831. Rehncrona S, Rosen I, Siesjo BK. 1981. Brain lactic acidosis and ischemic cell damage: 1. Biochemistry and neurophysiology. J Cereb Blood Flow Metab 1: 297–311. Rothman DL, Howseman AM, Graham GD, Petroff OA, Lantos G, Fayad PB, Brass LM, Shulman GI, Shulman RG, Prichard JW. 1991. Localized proton NMR observation of [3–13C] lactate in stroke after [1–13C]glucose infusion Magn Reson Med 21: 302–307. Rutgers DR, Klijn CJ, Kappelle LJ, van der Grond J. 2000. Cerebral metabolic changes in patients with a symptomatic occlusion of the internal carotid artery: a longitudinal 1H magnetic resonance spectroscopy study. J Magn Reson Imaging 11: 279–286. Sager TN, Laursen H, Hansen AJ. 1995. Changes in N-acetyl-aspartate content during focal and global brain ischemia of the rat. J Cereb Blood Flow Metab 15: 639–646. Sager TN, Laursen H, Fink-Jensen A, Topp S, Stensgaard A, Hedehus M, Rosenbaum S, Valsborg JS, Hansen AJ. 1999. N-Acetylaspartate distribution in rat brain striatum during acute brain ischemia. J Cereb Blood Flow Metab 19: 164–172. Sappey-Marinier D, Calabrese G, Hetherington HP, Fisher SN, Deicken R, Van DC, Fein G, Weiner MW. 1992. Proton magnetic resonance spectroscopy of human brain: applications to normal white matter, chronic infarction, and MRI white matter signal hyperintensities. Magn Reson Med 26: 313–327. Saunders DE, Howe FA, van den Boogart A, McLean MA, Griffiths JR, Brown MM. 1995. Continuing ischemic damage after acute middle cerebral artery infarction in humans demonstrated by short-echo time proton spectroscopy. Stroke 26: 1007–1013. Saunders DE, Howe FA, van den Boogaart A, Griffiths JR, Brown MM. 1997. Discrimination of metabolite from lipid and macromolecule resonances in cerebral infarction in humans using short echo proton spectroscopy. J Magn Reson Imaging 7: 1116–1121. Sorensen AG, Buonanno FS, Gonzalez RG, Schwamm LH, Lev MH, Huang-Hellinger FR, Reese TG, Weisskoff RM, Davis TL, Suwanwela N, Can U, Moreira JA, Copen WA, Look RB, Finklestein SP, Rosen BR, Koroshetz WJ. 1996. Hyperacute stroke: evaluation with combined multisection diffusion- weighted and hemodynamically weighted echoplanar MR imaging. Radiology 199: 391–401.

179

180

Peter B. Barker and Jonathan H. Gillard

Thulborn KR, Gindin TS, Davis D, Erb P. 1999. Comprehensive MR imaging protocol for stroke management: tissue sodium concentration as a measure of tissue viability in nonhuman primate studies and in clinical studies. Radiology 213: 156–166. van der Grond J, Balm R, Klijn CJ, Kapelle LJ, Eikelboom BC, Mali WP. 1996. Cerebral metabolism of patients with stenosis

of the internal carotid artery before and after endarterectomy. J Cereb Blood Flow Metab 16: 320–326. van Zijl PCM, Moonen CTW. 1993. In situ changes in purine nucleotide and N–acetyl concentrations upon inducing global ischemia in cat brain. Magn Reson Med 29: 381–385.

MR spectroscopy in stroke

Case Study 11.1 MRSI in acute brain ischemia Peter Barker, D.Phil., Jonathan Gillard MD, FRCR, Johns Hopkins University School of Medicine, Baltimore, MD, USA History 56-year-old male with acute onset left sided hemiparesis.

T2

NAA

Lac

Right

Conventional MRI and multi-slice proton MRSI at long TE (272 ms).

Imaging findings At presentation, T2 MRI was normal, while MRSI demonstrated elevation of lactate and mild reduction of NAA in the right MCA territory. At 1 week follow-up, following hypervolemic– hypertensive therapy, the patient was asymptomatic. Consistent with clinical recovery, the cortical Lac signal had resolved, although an infarct was apparent in the deep WM, with increased Lac and almost absent NAA.

Acute (24 h)

Technique

Cho Cr

NAA Lac

Left

ppm 4.0

T2

NAA

Lac

Right

3.0

2.0

1.0

MRSI has the ability to detect ischemic brain regions based on the elevation of lactate. Brain injury in these regions may be reversible if there are no other imaging signs of infarction (Barker et al., 1994b). The relationship of MRSI to DWI and PWI remains to be determined. However, the development of higher field magnets and rapid spectroscopic imaging techniques may allow routine MRSI evaluation of stroke.

Key points

Follow-up (1 week)

Discussion

Cho Lac

Cr

NAA

3.0

2.0

Left

Lac is increased in brain ischemia. Infarcts contain low levels of NAA.

ppm 4.0

Reference Barker PB, Gillard JH, van Zijl PCM, Soher BJ, Hanley DF, Agildere AM, Oppenheimer SM, Bryan RN. 1994b. Acute stroke: evaluation with serial proton MR spectroscopic imaging. Radiology 192: 723–732.

1.0

181

12

Diffusion and perfusion MR in stroke Joanna M. Wardlaw Department of Clinical Neurosciences, Western General Hospital, Crewe Road, Edinburgh, Scotland

Key points • Patients with acute stroke are more likely to be confused, uncooperative and so are more difficult to image with MR safely. This may preclude imaging in up to a third of cases. • Diffusion weighted imaging (DWI) helps particularly in cases of mild stroke because even small ischemic lesions show up clearly when conventional computed tomography or MR may appear normal. • Hyperintense areas on DWI early after stroke can initially reverse following thrombolysis. • Although diffusion/perfusion mismatch can give an indication of the volume of brain “at risk of infarction” the ideal perfusion parameters have yet to be confirmed and may be related to the duration from stroke ictus.

History Diffusion weighted imaging (DWI) rapidly captured the imagination of the stroke community (Chien et al., 1992; Fisher and Sotak, 1992; Warach et al., 1992, 1995). A rapid diagnostic test was needed that would not just exclude primary intracerebral hemorrhage, but would reliably identify signs of ischemia very early after stroke (Yoneda et al., 1999). Introduction of perfusion weighted imaging (PWI) increased enthusiasm further, with the idea that the DWI/PWI mismatch would identify the ischemic penumbra (Fisher et al., 1995). Perfusion imaging might identify regions of brain with blood flow below the level of ischemia but still above the level of 182

permanent damage, and so enable treatments like thrombolysis to be targeted more effectively. These techniques are increasingly widely used, but questions remain unanswered, and they are yet to replace computed tomography (CT) as the truly routine imaging method in acute stroke, although developments in multi-channel phased array coils appear to redress these differences.

Technology It is not the purpose of this chapter to detail the technology required, or the finer points of the various sequences and approaches which can be used to acquire DWI or PWI data. Chapters 4 and 7 deal with these points. Rather this will highlight the most salient points for practicing stroke physicians and radiologists to be aware of when reading the literature or implementing these techniques in clinical practice. DWI Can be undertaken with echo-planar MR equipment, or using non-echo-planar sequences. The echoplanar sequences are very fast and reduce the risk that head motion will result in blurred images. On the other hand, echo-planar imaging is very prone to air/bone/brain interface artifacts which can obscure detail in some parts of the brain (inferior frontal and anterior temporal lobes, and around the midbrain). Also, even with echo-planar imaging, head movement can still impair the final images. Non-echo-planar technology takes a little longer to acquire and so is more prone to movement artifact. This can be overcome with some image realignment

Diffusion and perfusion MR in stroke

techniques, like navigator echo (de Crespigny et al., 1995), single-shot fast spin-echo technique (Lovblad et al., 1998a), or proprietary MR manufacturer’s tools like periodically rotated overlapping parellEL line with enhanced reconstruction (PROPELLER™) (from GE Medical Systems) or prospective acquisition and corrEction (PACE™) (from Siemens Medical). Nonecho-planar DWI has the distinct advantage that it is less susceptible to artifacts at air/bone/brain interfaces, so lesions in the anterior and medial temporal and inferior frontal lobes, and anterior brain stem may be clearly seen. It is important to have DWI acquisition in a minimum of three different gradient directions at 90° to each other (Ulug et al., 1997; Chong et al., 1998). While three gradient directions would be routine nowadays, it was not when DWI was first introduced, and some of the debate concerning which lesion DWI or apparent diffusion coefficient (ADC) values were indicative of tissue damage may have arisen from inclusion of some data acquired in only one gradient direction (Lovblad et al., 1998; Nagesh et al., 1998). PWI The most frequently used technique is to give a rapid intravenous injection of a bolus of gadolinium and image very rapidly through the brain during the first pass of the contrast. This requires echo-planar imaging. The contrast medium changes the signal intensity as it passes through the brain vascular compartments from which a signal/time curve is derived. That is the easy part. The measurement of cerebral perfusion is a difficult problem and it is worth noting that people have been trying to turn this signal/time curve into a numerical value of perfusion since the 1800s, originally with dilution and later with isotope techniques (Stewart, 1894). The problem with MR is how the “perfusion” information should be extracted from the signal/time curve. Various interdependent parameters – cerebral blood flow, cerebral blood volume, and mean transit time (CBF, CBV, and MTT, respectively) – can be obtained from measuring different aspects of the curve (Teng et al., 2001; Grandin, 2002; Mukherjee et al., 2003; Latchaw et al., 2003). These generally assume that the signal is directly proportional to the

concentration of contrast agent, but the relationship is non-linear (Smith et al., 2003). There are many aspects of the curve which can be (and have been) analyzed: time to reach peak concentration (TTP); first moment of the curve and full width at half maximum (both proportional to the relative MTT); the maximum slope; area under the curve (proportional to the CBV); bolus arrival time; and peak height, all giving relative perfusion values only (Grandin et al., 2002). More complex methods including deconvolution of the curve and comparing the curve for brain tissue with that from an artery (arterial input function, AIF) have also been developed (Østergaard et al., 1996a, 1996b) but are more complex to analyze and it is not clear that they do actually yield reliable quantitative absolute values (Calamante et al., 2002; Mukherjee et al., 2003). There are some assumptions made in this technique which mean that the perfusion values produced are not “absolute” (Calamante et al., 2002; Latchaw et al., 2003). For example, the mathematical calculation assumes that flow in white matter (WM) is constant and does not alter with age (this is clearly incorrect, particularly in patients with stroke who tend to be older (Mukherjee et al., 2003); it does not deal with areas of the brain where there is no flow (e.g. in the centre of an infarct supplied by a totally occluded artery with poor collaterals (Armitage et al., 2003); it assumes that the signal is directly proportional to the contrast medium concentration and the relationship is actually non-linear and complex (Carpenter et al., 2003). Other approaches have also been suggested, but meantime, many resort to simple visual inspection of the TTP map of relative perfusion abnormalities. A further technique which uses endogenous contrast from flowing blood–arterial spin labeling or ASL, as a method of obtaining an AIF and so absolute measure of CBF, is described in the following chapter.

General feasibility of MR in stroke In general, many patients with stroke can have an MR examination which yields useful diagnostic information. This is in spite of many stroke patients being elderly and many patients (regardless of age) find MR a little daunting. It is more difficult to image patients in

183

184

Joanna M. Wardlaw

the hyperacute phase than in the subacute or chronic phases after stroke. Thus, in one recent study in 234 patients presenting to a neurovascular outpatient clinic with subacute stroke, only four patients referred were unable to undergo MR and in one this was because the patient was simply too large (Wardlaw et al., 2003). However, patients with acute stroke are more likely to be confused, uncooperative and those with right-hemisphere lesions may not even be aware that they have had a stroke, so are more difficult to image with MR safely. Some reports have been optimistic and suggested that “most” acute stroke patients can undergo MR as first-line investigation. However, many of these reports do not mention the number of patients that were admitted to hospital with suspected stroke during the time of the study (the “denominator”) who did not have MR. In the largest single centre study on “feasibility” of DWI in acute stroke, (Mullins et al., 2002) 691 patients were imaged with MR and CT soon after admission to hospital. However, the study was retrospective and relied on extraction of information from radiology department request cards and case notes, so it was impossible to assess the “denominator”. Furthermore, in 75% of patients it was not possible to determine the time of stroke (and so the time of hospital admission was used instead), there was no blinding (as the final neurological diagnosis was made in full knowledge of the information obtained from imaging), and only about two-thirds of patients whose radiology records were traced and were included in the study actually had MR (the rest had CT). In another retrospective study, Buckley et al. audited feasibility of MR with DWI as first-line investigation for acute stroke in 124 patients; of these 119 were “MR safe” but only 88 (70%) had MR as first-line investigation (Buckley et al., 2003). These, and similar, studies may be optimistic because they are retrospective and fail to catch the true “denominator”. In other words, if all patients who arrived in the emergency room with a diagnosis of suspected stroke (after family doctor or paramedic assessment), in how many would MR be feasible, and in how many would it actually make a positive contribution to the diagnostic process by providing information not available from other techniques? In our own prospective study of MR feasibility (Hand, 2002), of 138 patients admitted to hospital with suspected acute stroke during the study period, 53 (38%) could not

undergo MR. Compared with those who were imaged, patients not imaged were significantly older (median age 77 years vs. 74 years, P  0.03), had more severe stroke (median National Institute of Health Stroke Scale, NIHSS 11 vs. 5, P  0.03), were drowsy (57% vs. 32%, P  0.01) and less likely to have an exact time of onset of symptoms (64% vs. 86%, P  0.01). The commonest reasons for not performing MR were that the patient was medically unstable or had a metallic contraindication. Some patients were unable to complete an MR study because of anxiety, confusion, or because the medical staff were concerned about the patient’s medical state. Many MR machines are becoming more “open” or at least accessible for the patient, which may reduce the problems of patient anxiety and difficulties in monitoring and ensuring adequate life support during scanning, but even with this (and when more MR facilities are available), there will still be a proportion of acute stroke patients who cannot have MR. Instead of a policy of “MR for all”, it would be much more sensible to focus the use of MR in patients where the information which can best be gained from DWI is most useful (cf. below). MR raises safety issues different to those of CT. It is unsafe to sedate for MR without full anesthetic cover, but this adds to the time taken to get the diagnosis so may increase ischemic brain injury, the one thing that MR is being used to try to reduce. While it is possible to exclude the presence of a pacemaker box in the dysphasic patient with reasonable certainty on clinical examination, it is not possible to exclude retained pericardial pacing wires which might induce arrhythmias in MR (one would hope a good history of recent cardiac surgery would be available but this may not always be the case). Patients unable to give any history and without a reliable relative may harbour intraocular or intracranial foreign bodies and these are not detectable on clinical examination – a plain film of the skull and orbits is required unless a CT of the brain is already available with complete orbit coverage. Because of dysphasia or aphasia and lack of a reliable history from friend or relative, 11 patients (13%) required an orbital radiograph to exclude a metallic foreign body before entering the scanner. This added 20–30 min delay to the time to diagnosis. This further delays the time to treatment. A high proportion (up to 50% in the acute phase) of acute stroke patients are unable to protect their airway and are

Diffusion and perfusion MR in stroke

T2 MRI

DWI

Fig. 12.1 DWI is helpful in patients with mild stroke presenting late. A 76-year-old lady with mild left hemiparesis 6 weeks previously. On T2 (left) there are WMH obscuring any new infarct. On DWI (right), the hyperintense recent infarct is clearly seen.

vulnerable to aspiration and hypoxia when placed supine (Rowat et al., 2001). We attempted to monitor blood oxygen saturation during MR, and 11/61 (18%) monitored had at least one episode of hypoxia (SaO2  90% for at least 1 min) during the MR study (Rowat et al., 2002). The lowest recorded SaO2 was 74%. It was not possible to predict which patients were most likely to become hypoxic because all patients started out with good O2 saturation in the emergency room, and were not judged to be at “high risk” before MR. Patients with severe stroke are most at risk of hypoxia and aspiration when placed supine, often the very group examined in studies of DWI/PWI mismatch and outcome prediction.

elevated diffusion constants. Therefore, the diffusion characteristics of a lesion may help establish its age. In humans, much emphasis so far in the literature has been placed on the use of DWI in patients with symptoms of large hyperacute middle cerebral artery (MCA) territory infarction (Schellinger et al., 2001). These studies have highlighted that many ischemic lesions become visible very quickly, increasing confidence that the diagnosis is stroke, rather than some other form of “brain attack” (Lutsep et al., 1997; Lovblad et al., 1998b). This emphasis has rather overlooked other diagnostic benefits of DWI.

Diffusion imaging – when is it particularly useful?

Mild strokes usually result from small lesions which are often hard to see or not visible on CT or T2weighted imaging. DWI helps particularly in these cases because even small ischemic lesions show up clearly on DWI, and can be visible on DWI for several weeks after stroke, not just in the first few days. Thus patients with milder strokes who may present late to hospital (i.e. several weeks later because their symptoms were so mild) may still have a visible lesion on DWI (Figure 12.1) (Schulz et al., 2003). This may

Experimental studies in animals have shown that the brain water diffusion constant drops rapidly after the induction of ischemia, leading to hyperintensity in DWI. The evolution of the diffusion constant with time is a complex phenomenon which will be discussed in more detail later in this chapter, but it appears that most chronic strokes showed

Patients with mild- or late-presenting strokes

185

Joanna M. Wardlaw

60

DWI only

worsening of their residual neurological signs – sometimes it can be difficult to be certain that the patient has had another stroke and that their worsened neurological symptoms are not simply the result of some intercurrent illness (Figure 12.4) (Fitzek et al., 1998; Albers et al., 2000; Wardlaw et al., 2000).

No lesion

Diagnosing cardioembolic stroke

DW is most helpful in minor strokes and for weeks after onset

50

Visibility of lesion T2 and DWI

40 Count

186

30 20 10 0 TACI

PACI

LACI

POCI

TIA/non stroke Clinical syndrome (OCSP)

Fig. 12.2 Graph illustrating which types of stroke patient (Oxfordshire Community Stroke Project Classification, OCSP) had the greatest diagnostic yield from DWI compared with T2. TACI, total anterior circulation infarct; PACI, partial anterior circulation infarct; LACI, lacunar infarct; and POCI, posterior circulation infarct.

help to increase confidence that the symptoms were due to stroke, as it is often the milder strokes that are harder to differentiate from some non-vascular cause of “brain attack” especially several weeks after the event (Figure 12.2). In this respect, DWI is more helpful than T2-weighted imaging or CT, on which it is harder to age some infarcts (Albers et al., 2000; Wardlaw et al., 2000; Mullins et al., 2002). This particularly applies to lacunar stroke, as patients often have white matter hyperintensities (WMH) on T2 and fluid attenuated inversion recovery (FLAIR) imaging (and hypodensities on CT) making it very difficult to identify a recent infarct on the background of WM disease. However DWI is particularly useful as it will identify the recent subcortical infarct in a large proportion of cases (Figure 12.3) (Schonewille et al., 1999; OliveiraFilho et al., 2000; Yonemura et al., 2002).

Identifying a recurrent infarct DWI is also very helpful to identify or exclude a new infarct in patients with a previous stroke and recent

DWI can aid diagnosis of cardioembolic stroke in patients with symptoms which suggest an infarct in only one part of the brain, but on DWI multiple lesions are visible in different arterial territories (Figure 12.5) (Albers et al., 2000; Roh et al., 2000). These lesions are more clearly visible on DWI than on T2-weighted imaging or CT, and occur in lacunar stroke as well as cortical infarction (Ay et al., 1999b).

Does DWI always show ischemic lesions? No. Although the proportion of patients with ischemic stroke and a normal DWI study is less than the proportion with a normal T2 or CT, there are still occasional patients, even ones with quite severe strokes, that can have no lesion visible on DWI (Figure 12.6) (Ay et al., 1999a; Wang et al., 1999; Lefkowitz et al., 2000; Wang et al., 2003). Absence of a visible lesion is more likely in milder strokes than severe, rather as is the case with CT and T2 (Ay et al., 1999a). A lesion may become visible if the DWI is repeated hours or days later.

Does DWI differentiate TIA from stroke? No. Patients who make a rapid early recovery from their “brain attack”, and are therefore classed as transient ischaemic attacks (TIAs), may have a lesion on DWI in the appropriate area of the brain for their symptoms (Kidwell et al., 1999; Crisostomo et al., 2003). Amongst those imaged within 12 h of onset of symptoms, about 40% of patients who subsequently turned out to have had a TIA (i.e. their symptoms have completely resolved by 24 h) had an area of increased signal in the appropriate part of the brain for their symptoms (Kidwell et al., 1999). The DWI lesions in TIA patients tend to be smaller and less hyperintense than patients scanned within the same

Diffusion and perfusion MR in stroke

CT

T2

DWI

Fig. 12.3 DWI is helpful in mild stroke. A 46-year-old patient with a left-hemisphere lacunar syndrome and multiple lesions on CT and T2. The recent infarct is only clearly distinguished on DWI. Imaged 2 days after stroke.

Fig. 12.4 DWI is helpful in determining whether a patient with a previous stroke and worsening neurological symptoms, has had a recurrent stroke or not. DWI identifies the new lesion next to the old lesion.

time lapse after “brain attack” who are later classed as a stroke (Ay et al., 2002). Nonetheless, the presence of a lesion on DWI does not exclude TIA. The absence of a DWI (or other) lesion in a patient with a diagnosis of TIA may suggest that other (non-ischemic) etiologies of brain injury should be investigated.

Is DWI specific for ischemic injury? If DWI is to be the “electrocardiogram (ECG)” for stroke, then it needs to be as specific as possible for ischemic stroke. While in general bright lesions on DWI are ischemic, there have been reports of increased signal on DWI in multiple sclerosis (MS)

187

188

Joanna M. Wardlaw

10h

4 days

lesions, after an epileptic fit (which could be ischemic, but thrombolysis would not be appropriate), in encephalitis, and a number of other conditions (Wang et al., 1998). In general, if imaged within a few days of the symptom onset and the lesion was an ischemic stroke, then the ADC image should show a dark area corresponding with the bright area on DWI. If not ischemic, then the ADC lesion should not be dark even if the DWI lesion is bright. However, if imaged later, then the ADC lesion might be isointense or bright even if the lesion were ischemic. The imaging therefore needs to be interpreted in light of the clinical picture (Figure 12.7).

Does a visible lesion on DWI indicate permanent brain damage? 12 days

Fig. 12.5 Multiple recent infarcts in different arterial territories in a 65-year-old patient in atrial fibrillation and cardioembolic stroke. Only the right parietal lesion was symptomatic. Note the development of the new right frontal infarct by 12 days. This was asymptomatic (the patient was being scanned sequentially for research otherwise this new lesion would not have been detected).

DWI

DWI

In general, hyperintense lesions on DWI are likely to represent areas of permanent cerebral damage. However, recent reports have suggested that hyperintense areas on DWI early after stroke can reverse to isointense following thrombolysis, but may later become hyperintense at about 7 days (Kidwell et al., 2000, 2002; Schaefer et al., 2002). This has been interpreted as indicating “late secondary ischemic injury”, a phenomenon well described in animal models. Patients whose lesions showed early reversal and no secondary decline had the least abnormal

MRA

Fig. 12.6 Sometimes even patients with severe stroke have a negative DWI. This patient presented with a TACS equivalent to a large MCA territory stroke, but had no definite lesion visible on DWI at 8 h. The MRA showed a right MCA occlusion. She died before the imaging could be repeated.

Diffusion and perfusion MR in stroke

DWI

T2 Fig. 12.7 DWI is not specific for ischemia. This patient has MS, but one of the lesions looks hyperintense on DWI and could be mistaken for an infarct. She presented several weeks after onset of right-sided sensory symptoms at a time when the ADC image would be expected to have normalized anyway.

ADC values in their stroke lesion on early imaging. Those with no reversal had the lowest ADC values initially, and those with early reversal and secondary decline had ADC values in between the two. This implies that the duration or depth of the initial ischemic injury was least in those with apparent recovery and worst in those with permanent DWI lesions. In this paper (Kidwell et al., 2002) the patients were not examined later than 7 days after the stroke and so the relationship to late T2 appearance or functional outcome is uncertain. However these findings concur with other data (Figure 12.8) showing that the ADC values are lowest in patients

with severe MCA territory strokes (total anterior circulation infarct, TACI) and less abnormal in milder cortical or lacunar strokes (partial anterior circulation infarction, PACI and lacunar anterior circulation infarction, LACI, respectively) (Wardlaw et al., 2002). It is not certain whether areas that are of increased signal on DWI can recover completely, that is not progress at all to a T2 visible infarct. More data are needed from thrombolysis studies with serial imaging and coregistration of images to be sure of matching the same areas of brain, to see whether some “bright” lesions can recover completely and what the determinants are.

189

Joanna M. Wardlaw

1.8

108

1.6

How long does the increased signal last on DWI?

1.4 ADC ratio

190

1.2 1.0

79

0.8 0.6 0.4 0.2 0.0 n

n  120 patients 12 TACI

52 24 14 PACI LACI POCI Clinical syndrome

Fig. 12.8 The relationship between ADC and clinical stroke severity according to the Oxfordshire Community Stroke Project Classification in a prospective study of 102 patients. Note mild strokes (LACI, PACI, and POCI) have less abnormal ADC ratios of infarct/normal values than TACI.

Can DWI identify hemorrhage? If DWI is to be used in acute stroke prior to treatments like thrombolysis (or any antithrombotic treatment), then one would need to be confident of excluding intracerebral hemorrhage. Although recent studies have suggested high sensitivity for detection of cerebral hemorrhage using MR, hyperacute intracerebral hemorrhage can be difficult to diagnose confidently within the first few hours of stroke, even with a gradient echo (GE) T 2* sequence. Hemorrhage can look like a mass lesion such as a tumour, especially to observers not used to looking at hyperacute stroke images (Figure 12.9). It is possible to recognize hemorrhage, or at least be suspicious of the presence of hemorrhage, on DWI if there are dark bands around the bright lesion. The echo-planar images recorded without diffusion weighting (b  0 s/mm2) may be helpful because they are quite heavily susceptibility weighted. In the study by Lam et al. (2003) no hematomas or old hemorrhages, and few hemorrhagic infarcts were missed on the b  0 s/mm2 image compared with CT and the GE image. Therefore, it is probably wise to include a GE (T 2*) sequence routinely in acute stroke.

Increased signal on DWI can last up to 6–8 weeks, sometimes more, after stroke (in our experience sometimes many months, Figure 12.10) (Augustin et al., 2000; Geijer et al., 2001; Schulz et al., 2003). The brightness of the signal on DWI is composed of the reduction in ADC and T2. Early after stroke, most of the bright signal in the DWI is due to the reduced ADC and later it is mostly due to “T2 shine-through” (Eastwood et al., 2003). Various studies measuring the ADC averaged across the entire infarct have shown that the ADC reduction can last much longer than the 24–72 h as originally thought (Schlaug et al., 1997; Lansberg et al., 2001). It is possible that the DWI characteristics of different types of infarcts evolve in different ways, depending on several factors such as blood supply and the proportion of gray matter (GM) and WM in the lesion (Bastin et al., 2000; Huang et al., 2001). However, sequential DWI of the same patients over several weeks to months after stroke shows that some infarcts retain some hyperintense signal in some parts of the infarct over many weeks whereas others become totally hypointense after 2 weeks (Figure 12.11). It is not clear why some lesions become hypointense on DWI rapidly and others show some persistent increased signal. Perhaps the latter have areas of ongoing ischemia or a different repair process in the infarct, but it is likely that there is some pathophysiological explanation for the observation which may be related to patterns of tissue and functional recovery. Further study will be needed to determine this.

Lesion development on diffusion imaging and perfusion imaging Numerous studies have suggested that in patients imaged very early after acute stroke, the visible hyperintense lesion on DWI was smaller than on scans repeated at several days after stroke (Baird et al., 1997; Schwamm et al., 1998; Beaulieu et al., 1999). This implied that more tissue was recruited into the infarct from a penumbral region as time lapsed. If that were correct, the perfusion imaging should show an area of

T2

GRE

DWI

Fig. 12.9 Acute intracerebral hemorrhage on T2, DWI and GRE. Note the black serpiginous bands around the hematoma which are the clue that this is a hemorrhage.

Stroke 1 year ago

Stroke 3 weeks ago

DWI

T2

ADC

Fig. 12.10 Increased signal on DWI at 1 year after stroke. One year ago, this 76-year-old lady had a left striatocapsular infarct. She represented with nystagmus and dizziness and was scanned 3 weeks after the recurrence. The old lesion (top row, arrow) is still partly hyperintense on DWI and the ADC image shows some corresponding hypointensity, arrow, (so it is not just “T2 shinethrough”). The midbrain hyperintensity (bottom row) is more recent, but the ADC has already normalized.

192

Joanna M. Wardlaw

12 h

1 month

3 months

Fig. 12.11 DWI images at 12 hours, 1 month and 3 months after stroke showing that some of the lesion is still hyperintense while other areas are not on DWI. Presumably this reflects some aspect of the pathophysiological process, but what? (The figure was prepared by Ms. C. Rivers.)

reduced perfusion extending beyond the diffusion abnormality and corresponding with the area into which the diffusion abnormality would “grow” over time. This suggested that the ischemic penumbra or “tissue at risk” (Marchal et al., 1999) could be identified as the difference between the diffusion lesion volume and the perfusion lesion volume early after stroke (Schlaug et al., 1999). In theory, prompt intervention, say with a thrombolytic, might improve the abnormal perfusion and prevent the “growth” of the DWI lesion. A word of caution – some infarct “growth” is definite extension of the infarct into areas of the brain that were not ischemic/infarcted initially, but much “growth” within the first few days is actually simply swelling of the infarct and not a true extension into more tissue (Figure 12.12). Indeed, further studies with diffusion and perfusion imaging have shown a variety of diffusion/perfusion patterns consistent with that theory, (NeumannHaefelin et al., 1999; Schlaug et al., 1999; Grandin et al., 2001; Parsons et al., 2001, 2002; Rohl et al., 2001; Thijs et al., 2001) but these have so far largely been observational studies of smallish sample size and restricted patient selection. Several patterns of diffusion and perfusion lesions are illustrated in Figures 12.13–12.15. The natural history of ischemic stroke is complex and highly varied. It is only possible to

prove that “tissue at risk” can be rescued in a randomized trial where half the patients get a placebo and the other half an active perfusion-improving (or other “tissue rescuing”) agent. These are only now underway, and include several trials of thrombolytic agents. The problem is not in the fundamental concept of the “ischemic penumbra” (which is quite reasonable and reflects the results of years of research in animal models and stroke patients), but rather in the detail. What DWI and PWI characteristics should be used to define salvageable tissue? What degree of abnormality indicates that the tissue has gone beyond the point of “no return”. How does this alter as time passes after the stroke? Studies with positron emission tomography (PET) suggest values of absolute CBF that correspond with permanent tissue damage, (Baron, 1999) but so far MR PWI cannot provide reliable absolute CBF readings. Therefore which relative perfusion parameter should be used? It is clear that the MTT image usually shows larger lesions than the CBF image, and it is probable that the MTT lesion includes tissue which has reduced flow but is not at all “at risk” (Figures 12.13–12.15) (Grandin et al., 2002; Hand, 2002). Lesions are less often seen (and are harder to see) on CBF images, but the CBF lesion may more closely reflect “tissue at risk”. PET studies

Diffusion and perfusion MR in stroke

(a)

(b)

Fig. 12.12 DWI and T2 images demonstrate infarct “growth”. In (a) this is predominantly swelling without any true lesion extension into previously unaffected areas. In (b) the infarct extends further into the occipital region (arrow), i.e. it is not just swollen. Studies of infarct evolution need to distinguish between these two processes, and not simply assume that all volume change is growth in extent. A lot of the “growth” within the first few days in many lesions is actually just swelling. (The figure was prepared by Ms. S.M. Maniega.)

indicate that the CBF threshold of permanent damage rises as time lapses after stroke, that is tissue can only withstand perfusion at critical levels for a limited time before progressing to infarction (Baron et al., 1995). In other words, tissue at a CBF abnormality of 10 ml/100 g/min might only be salvageable for a few minutes, whereas tissue at 15 ml/100 g/min might be salvageable for a few hours but would eventually infarct even if perfusion could be improved towards the end of the 3 h (Baron, 1999). Although many studies have tried to identify individual DWI or PWI parameters, or combinations of the two, few studies have tried to incorporate the duration of ischemia (Lin et al., 2003). Estimates of DWI parameters associated with recoverable vs. permanent damage vary. Oppenheim et al. (2001) found an ADC ratio of 0.8  0.07 in the infarct core and a ratio of 0.97  0.08 for ischemic to normal brain was the best discriminator between tissue at risk and mere oligemia. However, final follow-up images (to determine final infarct extent) were only obtained at 4 days, too early to reliably identify the true

final infarct. Desmond et al. (2001) found much more abnormal ADC ratios – 0.65  0.11 in the infarct core, 0.83  0.05 in penumbra that went on to infarct, and 1.01  0.07 in penumbra that did not infarct. However, Fiehler et al. (2002) found areas with ADC ratios of 0.6 apparently not progressing to infarction according to a day 7 T2 image. Clearly such wide variation in estimates of what ADC ratio indicates permanent damage are not of much use in clinical practice. Literature studies are confounded by such issues as variable imaging time relative to stroke onset and also variations in blood flow (e.g. reperfusion status before and after imaging). More information from larger studies to account for case mix and incorporating information on time to imaging and the effect of thrombolysis is required to sort this out. Estimates of PWI parameters indicating recoverable vs. permanent damage are somewhat variable too (Lin et al., 2003) found a CBF threshold of 21 ml/ 100 g/min corresponded with a sharp drop in ADC value, but patients imaged within 4 h had a lower

193

194

Joanna M. Wardlaw

CBF

MTT

DWI

CBF

MTT

DWI

8h

6 days

Fig. 12.13 DWI/PWI patterns. Severe right-hemisphere stroke. MTT mismatch without CBF mismatch but the infarct grows into the MTT lesion.

CBF threshold (15 ml/100 g/min) than those imaged at 4.5–6.5 h (24 ml/100 g/min). Grandin et al. (2002) favoured a relative peak height or TTP threshold (54% or 5.2 s, respectively) or an absolute CBF of 35 ml/100 g/min to identify tissue proceeding to infarction. Shih et al. (2003) found time to maximum peak (Tmax) of 6 s identified infarcted tissue. Schaefer et al. (2003) found rCBF values of 0.32  0.11 in the infarct core, 0.46  0.13 in the penumbra progressing to infarction, and 0.58  0.12 in penumbra not progressing to infarction. These are some examples of the wide variation in perfusion parameters which may or may not identify the penumbra. As with DWI, more studies are needed to sort out which parameter, if any, most reliably and practically identifies tissue at risk in the clinical environment. What about just looking at the DWI/PWI image and “eyeballing” the mismatch. A single parameter would be ideal for use in clinical practice where speed is

of the essence in acute stroke, and complex image processing on off-line workstations is not appropriate. This might give as much practical information as detailed measurements of ADC and perfusion parameters. Several patterns of diffusion and perfusion lesions are illustrated in Figures 12.13–12.15. None of these patients received thrombolysis, but note that the patients with mismatch do not necessarily progress to large infarcts, the mismatch can be in a different part of the brain to the site of infarct growth and the MTT is larger and more visible than the CBF abnormality. In our own series, we found that “eyeballing” the lesion gives a significantly different estimate of the percentage mismatch than does formal measurement of lesion volume, because a volume difference is hard to estimate from an image of lesion area. (Hand, 2002). Coutts et al. (2003) found that “eyeballing” was not reliable between observers either.

Diffusion and perfusion MR in stroke

CBF

MTT

DWI

4h

CBF

MTT

DWI

6 days

Fig. 12.14 DWI/PWI patterns. Severe left-hemisphere stroke. MTT mismatch without CBF mismatch and no lesion growth but a new CBF deficit appears within part of the infarct area at follow-up.

Day 1

T2

DWI

ADC

FA

rCBF

rMTT

Day 7

Fig. 12.15 DWI/PWI patterns. MTT and CBF mismatch, but the MTT lesion is larger than the CBF lesion. No perfusion lesion at follow-up and no infarct growth.

195

Joanna M. Wardlaw

Did DWI/PWI mismatch predict outcome? 160

300 250

Size of penumbra (ml)

Size of penumbra (ml)

196

200 150 100 50

120

80

40

0

0 Poor

Good

Outcome MTT maps P  0.25

Poor

Good Outcome CBF maps P  0.33

Fig. 12.16 DWI/PWI mismatch and outcome. In 68 patients, there was no statistically significant relationship between mismatch (penumbra) and outcome.

How does thrombolysis affect these patterns? A case series of patients with DWI/PWI mismatch showed less infarct growth with thrombolysis treatment than would have been expected from historical controls (Parsons et al., 2002). In another case series of 29 patients (most with an MCA occlusion on MR angiography, MRA) treated with recombinant tissue plasminogen activator (rt-PA), although the DWI/ PWI mismatch was larger on admission than 24 h later, the DWI/PWI mismatch did not correlate with NIHSS score at baseline or late follow-up (Nighoghossian et al., 2003). Instead, the initial DWI volume and early recanalization were the only independent correlates with 60 day NIHSS score. In our own observational study of patients (n  68) not treated with rt-PA (but some of whom had spontaneous reperfusion) we found that the DWI/PWI mismatch measured from neither the CBF nor the MTT image predicted outcome (Figure 12.16) (Hand, 2002).

Is it worth measuring the DWI lesion? DWI lesion volume might provide a surrogate outcome measure for clinical trials to help reduce

sample size, or to improve patient selection for focussed studies (Lovblad et al., 1997; Warach et al., 2000). Some studies have found a relationship between larger DWI lesion volume and poor functional outcome at late follow-up (Beaulieu et al., 1999; Thijs et al., 2000; Warach et al., 2000; Baird et al., 2001; Engelter et al., 2003). Lesion volume was usually determined by tracing round the visible lesion on a workstation or MR console on each slice on which it was visible, and then summing the slices. Some used automated threshold values for DWI parameters like the ADC (Nagesh et al., 1998). One study generated an outcome prediction equation using DWI lesion volume, increasing time from onset and NIHSS (Baird et al., 2001). However, some of these studies had relatively small sample sizes with quite selected patients with a narrow range of characteristics, and some were retrospective. This means that it may be difficult to correct for biases and confounding factors (e.g. patients with severe strokes are likely to have large DWI lesions and poor outcomes), and it is difficult to extrapolate the results to broader groups of patients (Powers and Zivin, 1998; Keir and Wardlaw, 2000). Thus two other prospective studies, with broader selections of patients (i.e. mild and severe stroke, and including

Diffusion and perfusion MR in stroke

lacunar stroke) have not found an independent relationship between DWI lesion volume and functional outcome (Hand et al., 2002; Wardlaw et al., 2002). While they may have found a univariate association between DWI lesion volume and outcome, it was not independent because the relationship between clinical stroke severity (e.g. on the NIHSS score) was much stronger. Why the different results? All these studies were relatively small (even 60 patients is not a lot for a disease as heterogeneous as stroke), and there may be differences by chance. Casemix is important, as for example, patients with right-hemisphere lesions may appear to have a less severe stroke on the NIHSS score (despite large DWI lesions) than patients with left-hemisphere lesions (Fink et al., 2002). Most studies that found an independent association between DWI lesion volume and outcome had selected mostly severe strokes (in some cases retrospectively) and there were questions of blinding of analysis. The most likely explanation of the apparent discrepancy between studies is the strong relationship between clinical stroke severity, for example as measured by the NIHSS Score and outcome, most elegantly demonstrated by the Trial of Org 10172 in Acute Stroke Treatment (TOAST) data (Adams et al., 2003). Figure 12.17 shows the steep relationship found in the TOAST trial between NIHSS score at entry and the probability of a good functional outcome. Various MR studies examining the relationship between DWI lesion volume and outcome, and which measured stroke severity using the NIHSS score, have been plotted on it, with the position of the vertical line indicating the mean or median stroke severity for the patients in that study and the study sample size by the number next to the study name. Most of the studies that did find a relationship between DWI lesion volume and outcome are smaller and lie further towards the more severe end of the stroke scale, whereas the one that did not lies near the milder end of the NIHSS score. Note that the relationship between severity and outcome is not linear, but is steeper in the milder range and flatter in the severe range. Mild strokes fall on the steep part of the curve and it would be very difficult for another variable to add independently to this already near-perpendicular relationship. However for more severe strokes, the curve is flatter, and then it would be

possible for other factors, which “capture” some information about severity, to add independent predictive value. Thus studies with mainly severe strokes find an independent relationship between lesion volume and outcome and those with a broader mix do not.

Is DWI lesion volume easy to measure? Various methods are used for this, from tracing round the hyperintense lesion on each slice on which it is visible and then summing the slices, to setting a particular ADC threshold and measuring the amount of tissue contained within that using an automated image analysis program. The former are prone to operator variation. In general, the larger the lesion, the greater the potential for error, because most of the volume of a lesion is in its outer rim (e.g. a 9 cm diameter orange is 30% smaller than a 10 cm diameter orange) (Rana et al., 2003). The latter method, while attractive as it reduces operator input, is prone to missing out tissue which is permanently damaged but just did not reach the required ADC threshold. In one study, a high proportion of patients with milder strokes would have had their lesions missed altogether if an automated method had been used with an ADC threshold set at 60% of normal despite the fact that they had definitely had a stroke, had a visible lesion on T2, and had a residual neurological deficit at 6 months (Figure 12.8) (Wardlaw et al., 2002). Unless the observer reliability can be improved, it is unlikely that DWI could provide a surrogate outcome marker to reduce sample size in stroke treatment trials, because to overcome the effect of observer reliability would need nearly as large a sample size as a trial with a functional outcome measure (Rana et al., 2003). More studies of what aspect of a DWI lesion to measure (volume, ADC, etc.) and how best to measure it (manual tracing, automated methods) are required if DWI is to be used in stroke trials to assess treatment effects or to predict outcome. Virtually no studies have addressed the reproducibility of perfusion lesion volume measurement, but it is likely to be no better than DWI lesion volume as the perfusion lesion edges are often less clear. “Eyeballing” is not as good as formal volume measurement (Hand, 2002; Coutts et al., 2003).

197

Joanna M. Wardlaw

1.0

0.9

Edinburgh, median (82)

Stroke subtype

Non-lacunar at 7 days Lacunar at 7 days Non-lacunar at 3 months Lacunar at 3 months

0.8

Stanford, median (63) Probablity of excellent outcome

0.7

Boston, mean (66)

0.6

Melbourne, mean (63) 0.5

Heidelberg, median (51) 0.4

0.3

Admission NIHSS vs. Barthel Index at 7 days and 3 months

0.2

0.1

0.0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42

198

Baseline NIH Stroke Scale Score Fig. 12.17 The relationship between stroke severity according to the NIHSS score and functional outcome, (Adams et al., 2003) with studies which examined DWI lesion volume and outcome superimposed at the mean or median value of their patient cohort. Studies with an average of more severe stroke patients more often found a positive relationship between DWI lesion volume and outcome (Stanford, Boston, Melbourne, Heidelberg), than did the study with a broader range of severities which did not find such a relationship (Edinburgh).

Is MR DWI “better” than CT in hyperacute acute stroke? Randomized comparisons Bright white lesions on black backgrounds are easier to see than dark lesions on light backgrounds, so there is a general belief that DWI must be more sensitive and specific for hyperacute ischemic stroke than CT. Several studies have compared the two but not in random order (Barber et al., 1999; Lansberg et al., 2000; Kucinski et al., 2002; Mullins et al., 2002). It is only possible to obtain an objective assessment of which is best if the order of CT and MR are

randomized, otherwise it is likely that whichever technique is performed second (i.e. later, even by 30 min) will always show more ischemic lesions. The one study comparing DWI and CT in hyperacute stroke in random order so far (Fiebach et al., 2001) found DWI to be more sensitive and specific than CT for large infarcts. However, until it is clear that seeing a positive infarct (instead of merely the absence of hemorrhage) is important, the practical and safety advantages of CT listed above mean that CT will likely remain the “workhorse” for hyperacute stroke for some time to come.

Diffusion and perfusion MR in stroke

Where to next? DWI (and PWI) are potentially useful techniques, but the above discussion has highlighted many areas where more information is needed to guide their use in clinical practice. While the evidence that DWI is helpful in patients with mild stroke, late presenters, possible new infarct in addition to an old one, and to determine the etiology of the stroke is good, the information on use in hyperacute moderate to severe stroke is more patchy. There are several reasons for this, but the main problems are small studies with limited case mix, far too early follow-up to identify the final infarct, lack of clinical data, the need for tissue recovery to be properly assessed, and for treatment effects to be tested in randomized controlled trials. In order to achieve some of these aims, it is necessary for centres to join together and share data and some steps are already being taken in this direction. The big question for perfusion imaging is how to quantify it. One problem is that individual studies tend to use some aspects of the signal/time curve information and ignore other aspects. Studies may also measure lesions differently and not define in enough detail what was actually done, making comparisons between studies difficult. Some standardization, once more information is available, would help, but in the meantime, careful documentation of clinical features, late outcome, and imaging parameters is very important.

REFERENCES Adams HP, Davis PH, Leira EC, Chang KC, Bendixen BH, Clarke WR, Woolson RF, Hansen MD. 2003. Baseline NIH Stroke Scale score strongly predicts outcome after stroke. A report of the Trial of Org 10172 in Acute Stroke Treatment (TOAST). Neurology 53: 126–135. Albers GW, Lansberg MG, Norbash AM, Tong DC, O’Brien MW, Woolfenden AR, Marks MP, Moseley ME. 2000. Yield of diffusion-weighted MRI for detection of potentially relevant findings in stroke patients. Neurology 54(8): 1562–1567. Armitage PA, Rivers CS, Carpenter TK, Bastin ME, Hand PJ, Wardlaw JM. 2003. MR perfusion imaging: problems resulting from a lack of contrast agent in infarcted regions. Cerebrovasc Dis 16(suppl 4): 92. Augustin M, Bammer R, Simbrunner J, Stollberger R, Hartung HP, Fazekas F. 2000. Diffusion-weighted imaging of

patients with subacute cerebral ischemia: comparison with conventional and contrast-enhanced MR imaging. [see comments]. Am J Neuroradiol 21(9): 1596–1602. Ay H, Buonanno FS, Rordorf G, Schaefer PW, Schwamm LH, Wu O, Gonzalez RG, Yamada K, Sorensen GA, Koroshetz WJ. 1999a. Normal diffusion-weighted MRI during stroke-like deficits. Neurology 52(9): 1784–1792. Ay H, Oliveira-Filho J, Buonanno FS, Ezzeddine M, Schaefer PW, Rordorf G, Schwamm LH, Gonzalez RG, Koroshetz WJ. 1999b. Diffusion-weighted imaging identifies a subset of lacunar infarction associated with embolic source. Stroke 30(12): 2644–2650. Ay H, Oliveira-Filho J, Buonanno FS, Schaefer PW, Furie KL, Chang YC, Rordorf G, Schwamm LH, Gonzalez RG, Koroshetz WJ. 2002. “Footprints” of transient ischemic attacks: a diffusion-weighted MRI study. Cerebrovasc Dis 14(3–4): 177–186. Baird AE, Benfield A, Schlaug G, Siewert B, Lovblad KO, Edelman RR. 1997. Enlargement of human cerebral ischemic lesion volumes measured by diffusion-weighted magnetic resonance imaging. Ann Neurol 41: 581–589. Baird AE, Dambrosia J, Janket S, Eichbaum Q, Chaves C, Silver B, Barber PA, Parsons M, Darby D, Davis S, Caplan LR, Edelman RE, Warach S. 2001. A three-item scale for the early prediction of stroke recovery. Lancet 357(9274): 2095–2099. Barber PA, Darby DG, Desmond PM, Gerraty RP, Yang Q, Li T, Jolley D, Donnan GA, Tress BM, Davis SM. 1999. Identification of major ischemic change: diffusion-weighted imaging versus computed tomography. Stroke, 30(10): 2059–2065. Baron JC. 1999. Mapping the ischaemic penumbra with PET: implications for acute stroke treatment. Cerebrovasc Dis 9: 193–201. Baron JC, von Kummer R, del Zoppo GJ. 1995. Treatment of acute stroke. Challenging the concept of a rigid and universal time window. Stroke 26: 2219–2221. Bastin ME, Rana AK, Wardlaw JM, Armitage PA, Keir SL. 2000. A study of apparent diffusion coefficient of grey and white matter in human ischaemic stroke. Neuroreport 11: 2867–2874. Beaulieu C, de Crespigny A, Tong DC, Moseley ME, Albers GW, Marks MP. 1999. Longitudinal magnetic resonance imaging study of perfusion and diffusion in stroke: evolution of lesion volume and correlation with clinical outcome. Ann Neurol 46(4): 568–578. Buckley BT, Wainwright A, Meagher T, Briley D. 2003. Audit of a policy of magnetic resonance imaging with diffusion weighted imaging as first-line neuroimaging for in-patients with clinically suspected acute stroke. Clin Radiol 58: 234–237. Calamante F, Gadian DG, Connelly A. 2002. Quantification of perfusion using bolus tracking magnetic resonance imaging in stroke: assumptions, limitations, and potential implications for clinical use. Stroke 33(4): 1146–1151.

199

200

Joanna M. Wardlaw

Carpenter TK, Armitage PA, Bastin ME, Wardlaw JM. 2003. Artefacts in MR perfusion parametric images in ischaemic stroke-effects of variation in bolus arrival time. Cerebrovasc Dis 16(suppl 4): 92. Chien D, Kwong KK, Gress DR, Buonanno FS, Buxton RB, Rosen BR. 1992. MR diffusion imaging of cerebral infarction in humans. Am J Neuroradiol 13: 1097–1102. Chong J, Lu D, Aragao F, Singer MB, Schonewille WJ, Silvers A, Tuhrim S, Atlas SW. 1998. Diffusion-weighted MR of acute cerebral infarction: comparison of data processing methods. Am J Neuroradiol 19(9): 1733–1739. Coutts SB, Simon JE, Tomanek AI, Barber PA, Chan J, Hudon ME, Mitchell R, Frayne R, Eliasziw M, Buchan AM, Demchuk, AM. 2003. Reliability of assessing percentage of diffusion-perfusion mismatch. Stroke 34: 1681–1685. Crisostomo RA, Garcia MM, Tong DC. 2003. Detection of diffusion-weighted MRI abnormalities in patients with transient ischemic attack: correlation with clinical characteristics. Stroke 34(4): 932–937. de Crespigny AJ, Marks MP, Enzmann DR, Moseley ME. 1995. Navigated diffusion imaging of normal and ischemic human brain. Magn Reson Med 33(5): 720–728. Desmond PM, Lovell AC, Rawlinson AA, Parsons MW, Barber PA, Yang Q, Li T, Darby DG, Gerraty RP, Davis SM, Tress BM. 2001. The value of apparent diffusion coefficient maps in early cerebral ischemia. Am J Neuroradiol 22: 1260–1267. Eastwood JD, Engelter ST, MacFall JF, DeLong DM, Provenzale JM. 2003. Quantitative assessment of the time course of infarct signal intensity on diffusion-weighted images. Am J Neuroradiol 24: 680–687. Engelter ST, Provenzale JM, Petrella JR, DeLong DM, Alberts MJ. 2003. Infarct volume on apparent diffusion coefficient maps correlates with length of stay and outcome after middle cerebral artery stroke. Cerebrovasc Dis 15(3): 188–191. Fiebach J, Jansen O, Schellinger P, Knauth M, Hartmann M, Heiland S, Ryssel H, Pohlers O, Hacke W, Sartor K. 2001. Comparison of CT with diffusion-weighted MRI in patients with hyperacute stroke. Neuroradiology 43(8): 628–632. Fiehler J, Foth M, Kucinski T, Knab R, von Bezold M, Weiller C, Zeumer H, Rother J. 2002. Severe ADC decreases do not predict irreversible tissue damage in humans. Stroke 33(1): 79–86. Fink JN, Selim MH, Kumar S, Silver B, Linfante I, Caplan LR, Schlaug, G. 2002. Is the association of National Institutes of Health Stroke Scale scores and acute magnetic resonance imaging stroke volume equal for patients with right- and left-hemisphere ischemic stroke? Stroke 33(4): 954–958. Fisher M, Sotak CH. 1992. Diffusion-weighted MR imaging and ischemic stroke. Am J Neuroradiol 13: 1103–1105.

Fisher M, Prichard JW, Warach S. 1995. New magnetic resonance techniques for acute ischemic stroke. J Am Med Assoc 274(11): 908–911. Fitzek C, Tintera J, Muller-Forell W, Urban P, Thomke F, Fitzek S, Hopf HC, Stoeter P. 1998. Differentiation of recent and old cerebral infarcts by diffusion-weighted MRI. Neuroradiology 40(12): 778–782. Geijer B, Lindgren A, Brockstedt S, Stahlberg F, Holtas S. 2001. Persistent high signal on diffusional-weighted MRI in the late stages of small cortical and lacunar ischaemic lesions. Neuroradiology 43(2): 115–122. Grandin CB, Duprez TP, Smith AM, Mataigne F, Peeters A, Oppenheim C, Cosnard G. 2001. Usefulness of magnetic resonance-derived quantitative measurements of cerebral blood flow and volume in prediction of infarct growth in hyperacute stroke. Stroke 32: 1147–1153. Grandin CB, Duprez TP, Smith AM, Oppenheim C, Peeters A, Robert AR, Cosnard G. 2002. Which MR-derived perfusion parameters are the best predictors of infarct growth in hyperacute stroke? Comparative study between relative and quantitative measurements. Radiology 223: 361–370. Hand PJ. 2002. “Brain Attack” a New Approach to Stroke and Transient Ischaemic Attack, The University of Edinburgh, MD. Hand PJ, Rivers CS, Rowat AM, Bastin ME, Dennis MS, Wardlaw JM. 2002. Does DWI lesion volume predict outcome after stroke? Cerebrovasc Dis 13(3): 57. Huang IJ, Chen CY, Chung HW, Chang DC, Lee CC, Chin SC, Liou M. 2001. Time course of cerebral infarction in the middle cerebral arterial territory: deep watershed versus territorial subtypes on diffusion-weighted MR images. Radiology 221(1): 35–42. Keir SL, Wardlaw J. 2000. A systematic review of diffusion and perfusion imaging in acute ischaemic stroke. Stroke 31(2723): 2731. Kidwell CS, Alger JR, Di Salle F, Starkman S, Villablanca P, Bentson J, Saver JL. 1999. Diffusion MRI in patients with transient ischemic attacks. Stroke 30(6): 1174–1180. Kidwell CS, Saver JL, Mattiello J, Starkman S, Vinuela F, Duckwiler YP, G. G., Jahan, R, Vespa P, Kalafut M, Alger JR. 2000. Thrombolytic reversal of acute human cerebral ischemic injury shown by diffusion/perfusion magnetic resonance imaging. Ann Neurol 47(4): 462–469. Please provide full name. Kidwell CS, Saver JL, Starkman S, Duckwiler G, Jahan R, Vespa P, Villablanca JP, Liebeskind DS, Gobin YP, Vinuela F, Alger JR. 2002. Late secondary ischemic injury in patients receiving intraarterial thrombolysis. Ann Neurol 52: 698–703. Kucinski T, Vaterlein O, Glauche V, Fiehler J, Klotz E, Eckert B, Koch C, Rother J, Zeumer H. 2002. Correlation of apparent diffusion coefficient and computed tomography density in acute ischemic stroke. Stroke 33: 1786–1791.

Diffusion and perfusion MR in stroke

Lam WW, So NM, Wong KS, Rainer T. 2003. B0 images obtained from diffusion-weighted echo planar sequences for the detection of intracerebral bleeds. J Neuroimaging 13(2): 99–105. Lansberg MG, Albers GW, Beaulieu C, Marks MP. 2000. Comparison of diffusion-weighted MRI and CT in acute stroke. Neurology 54(8): 1557–1561. Lansberg MG, Thijs VN, O’Brien MW, Ali JO, de Crespigny AJ, Tong DC, Moseley ME, Albers GW. 2001. Evolution of apparent diffusion coefficient diffusion-weighted, and T2-weighted signal intensity of acute stroke. Am J Neuroradiol 22: 637–644. Latchaw RE, Yonas H, Hunter GJ, Yuh WT, Ueda T, Sorensen AG, Sunshine JL, Biller J, Wechsler L, Higashida R, Hademenos G. 2003. Guidelines and recommendations for perfusion imaging in cerebral ischemia: a scientific statement for healthcare professionals by the writing group on perfusion imaging, from the Council on Cardiovascular Radiology of the American Heart Association. Stroke 34(4): 1084–1104. Lefkowitz D, LaBenz M, Nudo SR, Steg RE, Bertoni JM. 2000. Hyperacute ischaemic stroke missed by diffusion-weighted imaging. Am J Neuroradiol 20: 1871–1875. Lin W, Lee JM, Lee YZ, Vo KD, Pilgram T, Hsu CY. 2003. Temporal relationship between apparent diffusion coefficient and absolute measurements of cerebral blood flow in acute stroke patients. Stroke 34(1): 64–70. Lovblad KO, Baird AE, Schlaug G, Benfield A, Siewert B, Voetsch B, Connor A, Burzynski C, Edelman RR, Warach S. 1997. Ischemic lesion volumes in acute stroke by diffusion-weighted magnetic resonance imaging correlate with clinical outcome. Ann Neurol 42(2): 164–170. Lovblad KO, Jakob PM, Chen Q, Baird AE, Schlaug G, Warach S, Edelman RR. 1998a. Turbo spin-echo diffusion-weighted MR of ischemic stroke. Am J Neuroradiol 19(2): 201–208. Lovblad KO, Laubach HJ, Baird AE, Curtin F, Schlaug G, Edelman RR, Warach S. 1998b. Clinical experience with diffusion-weighted MR in patients with acute stroke. Am J Neuroradiol 19(6): 1061–1066. Lutsep HL, Albers GW, DeCrespigny A, Kamat GN, Marks MP, Moseley ME. 1997. Clinical utility of diffusion-weighted magnetic resonance imaging in the assessment of ischemic stroke. Ann Neurol 41(5): 574–580. Marchal G, Benali K, Iglesias S, Viader F, Derlon JM, Baron JC. 1999. Voxel-based mapping of irreversible ischaemic damage with PET in acute stroke. Brain 123: 2387–2400. Mukherjee P, Kang HC, Videen TO, McKinstry RC, Powers WJ, Derdeyn CP. 2003. Measurement of cerebral blood flow in chronic carotid occlusive disease: comparison of dynamic susceptibility contrast perfusion MR imaging with positron emission tomography. Am J Neuroradiol 24(5): 862–871. Mullins ME, Schaefer PW, Sorensen AG, Halpern EF, Ay H, He J, Koroshetz WJ, Gonzalez RG. 2002. CT and conventional and

diffusion-weighted MR imaging in acute stroke: Study in 691 patients at presentation to the Emergency Department. Neuroradiology 224: 353–360. Nagesh V, Welch KM, Windham JP, Patel S, Levine SR, Hearshen D, Peck D, Robbins K, D’Olhaberriague L, Soltanian-Zadeh H, Boska MD. 1998. Time course of ADCw changes in ischemic stroke: beyond the human eye! Stroke 29(9): 1778–1782. Neumann-Haefelin T, Wittsack H-J, Wenserski F, Siebler M, Seitz RJ. 1999. Diffusion- and perfusion-weighted MRI. The DWI/PWI mismatch region in acute stroke. Stroke 30: 1591–1597. Nighoghossian N, Hermier M, Adeleine P, Derex L, Dugor JF, Philippeau F, Ylmaz H, Honnorat J, Dardel P, Berthezene Y, Froment JC, Trouillas P. 2003. Baseline magnetic resonance imaging parameters and stoke outcome in patients treated by intravenous tissue plasminogen activator. Stroke 34: 458–463. Oliveira-Filho J, Ay H, Schaefer PW, Buonanno FS, Chang Y, Gonzalez ER, Koroshetz WJRG. 2000. Diffusion-weighted magnetic resonance imaging identifies the “clinically relevant” small-penetrator infarcts” Arch Neurol 57(7): 1009–1014. Oppenheim C, Grandin C, Samson Y, Smith A, Duprez T, Marsault C, Cosnard G. 2001. Is there an apparent diffusion coefficient threshold in predicting tissue viability in hyperacute stroke? Stroke 32(11): 2486–2491. Østergaard L, Sorensen AG, Kwong KK, Weisskoff RM, Gyldensted C, Rosen BR. 1996a. High resolution measurement of cerebral blood flow using intravascular tracer bolus passages. Part II: experimental comparison and preliminary results. Magn Reson Med 36(5): 726–736. Østergaard L, Weisskoff RM, Chesler D, Gyldensted C, Rosen BR. 1996b. High resolution measurement of cerebral blood flow using intravascular tracer bolus passages. Part I: mathematical approach and statistical analysis. Magn Reson Med 36: 715–725. Parsons MW, Barber PA, Chalk J, Darby DG, Rose S, Desmond PM, Gerraty RP, Tress BM, Wright PM, Donnan GA, Davis SM. 2002. Diffusion- and perfusion-weighted MRI response to thrombolysis in stroke. Ann Neurol 51(1): 28–37. Parsons MW, Yang Q, Barber PA, Darby DG, Desmond PM, Gerraty RP, Tress BM, Davis SM. 2001. Perfusion magnetic resonance imaging maps in hyperacute stroke: relative cerebral blood flow most accurately identifies tissue destined to infarct. Stroke 32(7): 1581–1587. Powers WJ, Zivin J. 1998 Magnetic resonance imaging in acute stroke: not ready for prime time. Neurology 50(4): 842–843. Rana AK, Wardlaw JM, Armitage PA, Bastin ME. 2003. Apparent diffusion coefficient (ADC) measurements may be more reliable and reproducible than lesion volume on

201

202

Joanna M. Wardlaw

diffusion-weighted images from patients with acute ischaemic stroke-implications for study design. Magn Reson Imaging. In Press. Roh JK, Kang DW, Lee SH, Yoon BW, Chang KH. 2000. Significance of acute multiple brain infarction on diffusionweighted imaging. Stroke 31(3): 688–694. Rohl L, Ostergaard L, Simonsen CZ, Vestergaard-Poulsen P, Andersen G, Sakoh M, Le Bihan D, Gyldensted C. 2001. Viability thresholds of ischemic penumbra of hyperacute stroke defined by perfusion-weighted MRI and apparent diffusion coefficient. Stroke 32(5): 1140–1146. Rowat AM, Hand PJ, Janneke H, Wardlaw JM. 2002. Hypoxia in the acute phase of stroke during MR brain imaging. Stroke 33(1): 383 (No. P119). Rowat AM, Wardlaw JM, Dennis MS, Warlow CP. 2001. Patient positioning influences oxygen saturation in the acute phase of stroke. Cerebrovasc Dis 12: 66–72. Schaefer PW, Hassankhani A, Koroshetz CRW, Rordorf G, Schwamm LH, Buonanno F, Gonzalez RG. 2002. Partial reversal of DWI abnormalities in stroke patients undergoing thrombolysis: evidence of DWI and ADC thresholds. Stroke 33(1): 357 (84). Schaefer PW, Ozsunar Y, He J, Hamberg LM, Hunter GJ, Sorensen AG, Koroshetz WJ, Gonzalez RG. 2003. Assessing tissue viability with MR diffusion and perfusion imaging. Am J Neuroradiol 24(3): 436–443. Schellinger PD, Fiebach JB, Jansen O, Ringleb PA, Mohr A, Steiner T, Heiland S, Schwab S, Pohlers O, Ryssel H, Orakcioglu B, Sartor K, Hacke W. 2001. Stroke magnetic resonance imaging within 6 h after onset of hyperacute cerebral ischemia. Ann Neurol 49(4): 460–469. Schlaug G, Benfield A, Baird AE, Siewert B, Lovblad KO, Parker RA, Edelman RR, Warach S. 1999. The ischemic penumbra: operationally defined by diffusion and perfusion MRI. Neurology 53(7): 1528–1537. Schlaug G, Siewert B, Benfield A, Edelman RR, Warach S. 1997. Time course of the apparent diffusion coefficient (ADC) abnormality in human stroke. Neurology 49(1): 113–119. Schonewille WJ, Tuhrim S, Singer MB, Atlas SW. 1999. Diffusionweighted MRI in acute lacunar syndromes: a clinicalradiological correlation study. Stroke 30(10): 2066–2069. Schulz UGR, Briley D, Meagher T, Molyneux A, Rothwell PM. 2003. Abnormalities on diffusion weighted magnetic resonance imaging performed several weeks after a minor stroke or transient ischaemic attack. J Neurol Neurosurg Psychiatry 74: 734–738. Schwamm LH, Koroshetz WJ, Sorensen G, Wang B, Copen WA. 1998. Time course of lesion development in patients with acute stroke. Serial diffusion-and hemodynamic-weighted magnetic resonance imaging. Stroke 29: 2268–2276.

Shih LC, Saver JL, Alger JR, Starkman S, Leary MC, Vinuela F, Duckwiler G, Gobin YP, Jahan R, Villablanca JP, Vespa PM, Kidwell CS. 2003. Perfusion-weighted magnetic resonance imaging thresholds identifying core, irreversibly infarcted tissue. Stroke 34: 1425–1430. Smith MR, Lu H, Frayne R. 2003. Signal-to-noise ratio effects in quantitative cerebral perfusion using dynamic susceptibility contrast agents. Magn Reson Med 49: 122–128. Stewart GN. 1894. Researches on the circulation time in organs and on the influences which affect it. J Physiol 15: 1–89. Tanne D, Kasner SE, Demchuk AM, Koren-Morag N, Hanson S, Grond M, Levine SR and the Multicentre rt-PA Stroke Survey Group 2002. Markers of increased risk of intracerebral hemorrhage after intravenous recombinant tissue plasminogen activator therapy for acute ischemic stroke in clinical practice. The Multicentre rt-PA Acute Stroke Survey. Circulation 1679–1685. Teng MM, Cheng HC, Kao YH, Hsu LC, Yeh TC, Hung CS, Wong WJ, Hu HH, Chiang JH, Chang CY. 2001. MR perfusion studies of brain for patients with unilateral carotid stenosis or occlusion: evaluation of maps of “time to peak” and “percentage of baseline at peak”. J Comput Assist Tomogr 25(1): 121–125. Thijs VN, Adami A, Neumann-Haefelin T, Moseley ME, Marks MP, Albers GW. 2001. Relationship between severity of MR perfusion deficit and DWI lesion evolution. Neurology 57(7): 1205–1211. Thijs VN, Lansberg MG, Beaulieu C, Marks MP, Moseley ME, Albers GW. 2000. Is early ischemic lesion volume on diffusionweighted imaging an independent predictor of stroke outcome? A multivariable analysis. Stroke (Online) 31(11): 2597–2602. Ulug AM, Beauchamp Jr N, Bryan RN, van Zijl PC. 1997. Absolute quantitation of diffusion constants in human stroke. Stroke 28(3): 483–490. Wang AM, Shetty AN, Woo H, Rao SK, Manzione JV, Moore JR. 1998. Diffusion weighted MR imaging in evaluation of CNS disease. Rivista di Neuroradiologia 11(suppl 2): 109–112. Wang PY, Barker PB, Wityk RJ, Ulug AM. 1999. Diffusion-negative stroke: a report of two cases. Am J Neuroradiol 20: 1876–1880. Wang W, Goldstein S, Scheuer ML, Branstetter BF. 2003. Acute stroke syndrome with fixed neurological deficit and falsenegative diffusion-weighted imaging. J Neuroimaging 13(2): 158–161. Warach S, Chien D, Li W, Ronthal M, Edelman RR. 1992. Fast magnetic resonance diffusion-weighted imaging of acute human stroke. Neurology 42: 1717–1723. Warach S, Gaa J, Siewert B, Wielopolski P, Edelman RR. 1995. Acute human stroke studied by whole brain echo planar

Diffusion and perfusion MR in stroke

diffusion-weighted magnetic resonance imaging. Ann Neurol 37: 231–241. Warach S, Pettigrew LC, Dashe JF, Pullicino P, Lefkowitz DM, Sabounjian L, Harnett K, Schwiderski U, Gammans R. 2000. Effect of citicoline on ischemic lesions as measured by diffusion-weighted magnetic resonance imaging. Citicoline 010 Investigators [In Process Citation]. Ann Neurol 48(5): 713–722. Wardlaw JM, Armitage P, Dennis MS, Lewis S, Marshall I, Sellar R. 2000. The use of diffusion-weighted magnetic resonance imaging to identify infarctions in patients with minor strokes. J Stroke Cerebrovasc Dis 9: 70–75. Wardlaw JM, Keir SL, Bastin ME, Armitage PA, Rana AK. 2002. Is diffusion imaging appearance an independent predictor

of outcome after ischaemic stroke? Neurology 59(9): 1381–1387. Wardlaw JM, Keir SL, Dennis MS. 2003. The impact of delays in computed tomography of the brain on the accuracy of diagnosis and subsequent management in patients with minor stroke, J Neurol Neurosurg Psychiatry 74(1): 77–81. Yoneda Y, Tokui K, Hanihara T, Kitagaki H, Tabuchi M, Mori E. 1999. Diffusion-weighted magnetic resonance imaging: detection of ischemic injury 39 minutes after onset in a stroke patient. Ann Neurol 45(6): 794–797. Yonemura K, Kimura K, Minematsu K, Uchino M, Yamaguci T. 2002. Small centrum ovale infarcts on diffusion-weighted magnetic resonance imaging. Stroke 33: 1541–1544.

203

204

Joanna M. Wardlaw

Case Study 12.1 Diffusion tensor imaging and tissue anisotropy Alonso Peña, Ph.D., Hadrian AL Green, MB, ChB, Martin Graves, M.Sc., Jonathan H Gillard, B.Sc., M.D., FRCR, University Department of Radiology, University of Cambridge. History

Fig. 1

The most common measures of tissue anisotropy in MR diffusion tensor imaging (MR DTI) are fractional anisotropy (FA) and relative anisotropy (RA). However, there are many other measures including: the isotropic component of diffusion (p), anisotropic component of diffusion (q), and the total diffusion (L). What anisotropy measure should one use in assessing tissue damage?

A

qo Lo

po Technique

Discussion When these Region of interests (ROIs) were plotted in the p:q plane they form clearly segregated clusters Figure 2. Thus offering the analyst aconcise and easy-to-use representation of all the tensor measures (FA, RA, q, p, D, L). Conclusion There are many mathematically valid measures of tissue anisotropy. It is unknown which is the best measure to quantify pathological changes in stroke. Key point There are many ways of measuring tissue anisotropy in MRI DTI: p:q diagrams provide an easy-to-use visualization methodology.

Fig. 2 1–noise

1.5 q: anisotropic component of diffusion 10 3 mm2/sec

We can easily visualize all these measures by using a methodology known as “p:q diagrams”. For this, take a Cartesian plane in which the x axis corresponds to p and the y axis to q. In this plane each voxel corresponds to a point, such as A in Figure 1. The relationship between RA, FA and p, q, L is simple: RA  q/p and FA  sqrt(3/2) q/L.

Do

2–cerebrospinal fluid 3– internal capsule

2 1

4– corpus callosum 5– visual cortex

3 0.5 1

4 5

0

0

0.5

1

1.5 2 2.5 3 3.5 p: isotropic component of diffusion 10 3 mm2/sec

4

4.5

5

As a clinical illustration, MR DTI was performed on a healthy 27 year-old volunteer on a 3 T (Bruker Medspec S300). FA maps of the volunteer (Fig. 2, inset) were used to select square anatomical ROIs. These were placed in the corpus callosum (2), occipital cortex (4), cerebrospinal fluid (5), internal capsule (3) and noise regions (1).

References Basser PJ.1995. Inferring microstructural features and the physiological state of tissues from diffusion-weighted images. NMR Biomed 8(7–8): 333–344. Green HA, et al. 2002. Increased anisotropy in acute stroke: a possible explanation. Stroke 33(6): 1517–1521.

Diffusion and perfusion MR in stroke

Case Study 12.2 Reversal of diffusion lesion in acute stroke Chelsea Kidwell, M.D., Jeffrey Saver, M.D., Jeffry Alger, Ph.D., University of California, Los Angeles History 85-year-old female presenting with sudden onset of left hemiparesis and neglect.

Pre-treatment DWI

Pre-treatment PWI

Post-treatment DWI

Post-treatment PWI

Technique Diffusion-weighted (DWI) and perfusion-weighted (PWI) MRI obtained pre- and post-treatment with intra-arterial thrombolysis, initiated within 6 h of symptom onset.

Imaging findings Pre-treatment diffusion images demonstrate a hyperintense lesion in the right temporal lobe indicative of acute ischemia. A much larger perfusion lesion is shown on the colour-coded time to peak (TTP) of residue function. Following vessel recanalization, both the PWI and the DWI lesions have almost completely resolved, with no evidence of right temporal infarction on any sequence. Images copyright of UCLA Stroke Center Discussion In acute stroke, hyperintense lesions on DWI correspond to regions of tissue bioenergetic compromise and often are indicative of core, irreversibly injured brain. However, in some cases, when blood flow is restored, these lesions may reverse. Therefore, the ischemic penumbra, as depicted by MRI, includes not only regions of diffusion–perfusion mismatch, but also portions of the initial DWI abnormality.

Key points Diffusion lesions can be reversible in patients with acute ischemic stroke. Some patients without diffusion–perfusion mismatch may still benefit from reperfusion therapy.

References Kidwell CS, Alger JR, Saver JL. 2003. Beyond mismatch: evolving paradigms in imaging the ischemic penumbra with multimodal magnetic resonance imaging. Stroke 34: 2729–2735. Kidwell CS, Saver JL, Mattiello J, Starkman S, Vinuela F, Duckwiler G, Gobin P, Jahan R, Vespa P, Kalafut M, Alger JR. 2000. Thrombolytic reversal of acute human cerebral ischemic injury shown by diffusion/perfusion magnetic resonance imaging. Ann Neurol 47: 462–469.

205

206

Joanna M. Wardlaw

Case Study 12.3 Diffusion and perfusion MR in subarachnoid hemorrhage Adam Waldman M.D. Ph.D., FRCR, Hammersmith Hospitals and Institute of Neurology, London History 60-year-old woman with grade 1 subarachnoid hemorrhage at presentation. Evolving headache and confusion 6 days after ictus. MRI performed acutely. Further CT 6 days later. Subsequently died.

DSA

T2

CT

Technique X-ray CT, catheter digital subtraction cerebral arteriography (DSA), conventional MRI, DWI, DSC-PWI.

ADC

Findings CT was normal. DSA showed an anterior communicating artery (ACA) aneurysm, while T2-MRI showed minor mesial frontal cortical swelling and hyperintensity. ADC was more extensively reduced in mesial frontal lobes. There was negligible perfusion in the low ADC region, and markedly increased MTT and reduced rCBV throughout the ACA territories. MCA perfusion normal. Day 6 CT showed bilateral ACA infarction.

rCBV

MTT

Discussion This is an extreme case of territorial ischemia due to delayed vasospasm, which is a major cause of morbidity and mortality after SAH. A perfusion–diffusion mismatch similar to that in occlusive arterial stroke was identified in this case.

6 days later

CT

Key points DWI identifies infarcted brain perfusion abnormality helps identify potentially salvageable brain at risk of infarction. DWI and PI may allow early identification of patients with vasospasm, who may benefit from endovascular or pharmacological treatment. References Griffiths PD, Wilkinson ID, Mitchell P, et al. 2002. Multimodality MR imaging depiction of hemodynamic changes and cerebral ischemia in subarachnoid hemorrhage. Am J Neuroradiol 22: 1690–1697. Waldman AD, Kitchen N, Jager HR Cox TC. 2001. MR perfusion imaging predicts anterior cerebral artery territory infarct following acute aneurysmal subarachnoid haemorrhage. Proc Int Soc Mag Res Med 9.

13

Arterial spin labeling perfusion MRI in stroke Jiongjiong Wang and John A. Detre Departments of Neurology and Radiology, University of Pennsylvania, Philadelphia, USA

Key points • Arterial spin labeling (ASL) perfusion MR imaging can directly measure cerebral blood flow (CBF) using magnetically labeled arterial blood water as endogenous tracer. • ASL can be repeated as often as required in the same imaging session without accumulative effects, thus allowing for the assessment of cerebrovascular reserve. • ASL benefits from higher magnetic field strengths, not only because of the increased signal-to-noise ratio for detecting subtle effects of ASL, but also because T1 values are prolonged at higher fields, which increases the amount of labeling obtained. • Gray matter CBF measures agree well between ASL and positron emission tomography (PET). There is less agreement for white matter. • Pediatric ASL can provide improved contrastto-noise due to the normally increased blood flow in children.

Introduction Assessment of regional cerebral perfusion provides highly desirable information for the diagnosis and management of cerebrovascular disease and acute stroke. Over the past several decades, a variety of approaches have been used to image regional perfusion in cerebrovascular disease, including positron emission tomography (PET), single photon emission Acknowledgment: Supported by a Neuroscience Neuroimaging NIH Grant (P30 NS045839).

computed tomography (SPECT), xenon-enhanced X-ray computed tomography (XeCT), and MR imaging (MRI). The majority of these methods utilize an exogenous tracer that is administered intravenously or by inhalation. Most of the existing MRI studies of cerebral hemodynamics in cerebrovascular disease and stroke have also relied on dynamic tracking of susceptibilityrelated signal changes accompanying the passage of an exogenous bolus of intravenous contrast agent such as gadolinium DTPA. This dynamic susceptibility contrast imaging (DSCI) approach primarily measures blood volume and transit times (Guckel et al., 1994), but cerebral blood flow (CBF) can be estimated from these parameters based on the central volume principal (Østergaard et al., 1996, 1998; Smith et al., 2000). Arterial spin labeling (ASL) perfusion MRI is an emerging technology to directly measure CBF using magnetically labeled arterial blood water as endogenous tracer (Detre and Alsop, 1999a; Wong, 1999). The methodological scheme of ASL is analogous to that used in the steady-state PET or SPECT method (Detre et al., 1994). Arterial blood water is magnetically labeled proximal to the tissue of interest, and perfusion can be determined by pair-wise comparison with separate images acquired without labeling. The tracer of arterial blood water has a decay rate of T1, which is sufficiently long to allow perfusion of the microvasculature and tissue to be detected, but short enough to allow dynamic changes to be monitored. As ASL does not require administration of contrast agents or radioactive tracers, it may be more convenient than other approaches, and the perfusion measurement using ASL can be repeated as often as required in the same imaging session without accumulative effects. This allows ASL perfusion contrast to be used to monitor changes in CBF in 207

208

Jiongjiong Wang and John A. Detre

response to pharmacological manipulation or task activation. Furthermore, ASL can provide quantitative tissue-specific perfusion values in classical units of ml/g/min. Perfusion data can potentially contribute to stroke management in several ways. The combination of perfusion and diffusion imaging can be used to quantify the initial extent of the stroke and the volume of tissue with compromised blood flow (Baird et al., 1997; Fisher and Bogousslavsky, 1998; Grandin et al., 2001; Fiehler et al., 2002). This information is used to confirm the diagnosis of stroke, to establish a baseline against which stroke therapies can be assessed, and to contribute to prognosis. In the management of patients who are eligible for thrombolytic therapy, perfusion imaging might be used to improve the risk-to-benefit ratio of this therapy. For example, the presence of profound hypoperfusion in the ischemic core has been associated with an increased risk of hemorrhagic complications, whereas the absence of hypoperfusion in acute stroke may signify spontaneous reperfusion (Firlik et al., 1998; Rubin et al., 2000), indicating that thrombolytic therapy and its associated risk is not required. Hypoperfusion is also recognized as an important primary etiology for ischemic stroke in patients with cerebrovascular disease (Derdeyn et al., 1999), and hypoperfusion was found to be predictive of recurrent stroke in patients with stroke or transient ischemic attack (TIA) (Bogousslavsky et al., 1990; Webster et al., 1995; Gur et al., 1996). Reduced blood flow is considered the primary cause of so-called “borderzone” and white matter (WM) infarctions at the terminal distributions of vessels (Weiller et al., 1991; Pantoni et al., 1996; Mull et al., 1997), though reduced “washout” of microemboli may also be a contributing factor (Caplan and Hennerici, 1998). Building on technical developments in ASL such as labeling strategies with improved efficiency (Yongbi et al., 1999; Alsop, 2001) and controlling for the off-resonance effects due to magnetization transfer (MT) (Kim, 1995; Kwong et al., 1995; Alsop and Detre, 1998; Wong et al., 1998a), modern ASL methods are capable of multi-slice perfusion imaging with considerable accuracy in humans. Over the past decade, evidence has accumulated that ASL methods can provide reliable CBF measurements,

and can visualize clinically significant abnormalities in patients with cerebrovascular disease. This approach will particularly benefit from higher magnetic field strengths, not only because of the increased signal-tonoise ratio (SNR) for detecting subtle effects of ASL, but also because T1 values are prolonged at higher field, which increases the amount of labeling obtained. In this chapter, we review existing data and concepts concerning the validity of ASL perfusion MRI and its applications in cerebrovascular disease, including acute stroke, chronic cerebrovascular disease, pharmacological testing of hemodynamic reserve, and pediatric applications.

Validity and accuracy of ASL perfusion MRI As described in Chapter 8, ASL techniques are capable of quantifying CBF in classical physiological units of ml/100 g/min, based on the standard diffusible tracer model (Kety and Schmidt, 1945). During the past decade, theoretical and experimental studies have been carried out to improve the accuracy of CBF quantification in ASL, by taking into account multiple parameters such as arterial transit time (Alsop and Detre, 1996; Ye et al., 1997; Wong et al., 1998a; Gonzalez-At et al., 2000; Yang et al., 2000), MT effect (Alsop and Detre, 1996; McLaughlin et al., 1997; Alsop and Detre, 1998), T1 (Wong et al., 1998b; Wang et al., 2002), labeling efficiency (Roberts et al., 1994; Maccotta et al., 1997; Utting et al., 2003), and capillary water permeability (Buxton et al., 1998; Ewing et al., 2001; Parkes and Tofts, 2002). ASL perfusion MRI is prone to transitrelated effects, because in humans the tracer decay rate (blood T1) is comparable with the arterial transit time for the labeled blood to flow from the labeling region to the tissue of interest. This may lead to loss of labeling effects and potential errors in perfusion quantification due to uncertainties in arterial transit time. These transit-related errors are largely eliminated by introducing an appropriate delay between arterial labeling and image acquisition (Alsop and Detre, 1996; Wong et al., 1998a), to allow the labeled blood to enter the image slices by the time image is acquired. The precise time for the labeled blood to exchange from the capillary into

Arterial spin labeling perfusion MRI in stroke

brain tissue (exchange time) is also an important yet unsolved issue in ASL modeling, because the magnetic tracer will relax with the T1 of brain tissue instead of the T1 of blood from that time point. This uncertainty of exchange time can be handled by using uniform blood T1 to represent the overall tracer decay rate (Chalela et al., 2000), which is justified by the similarity between the T1s of blood and brain tissue, especially in the gray matter (GM) (Ye et al., 1997). Recent modeling work incorporating capillary water permeability also suggested that the labeled blood stays longer in the blood compartment before exchange, rendering blood T1 a more important parameter in determining the accuracy of perfusion measurements as compared to the T1 of brain tissue (Parkes and Tofts, 2002). CBF measurements using a continuous arterial spin labeling (CASL) technique have been compared with 15O-PET results from the same cohort of healthy subjects (Ye et al., 2000a). The overall and GM CBF measures were found to match with each other in the ASL and PET approaches, whereas discrepancy was observed in the WM. These data supported the general validity of perfusion quantification using ASL methods, but indicated that model-dependent perfusion measurement may not be very accurate in brain regions with less well-characterized parameters such as transit time and T1. Another important feature of ASL perfusion MRI is its stability and reproducibility over relatively long time scales, rendering itself appealing for longitudinal studies on the development of cerebrovascular disease. In ASL, perfusion-weighted images are typically produced by pair-wise subtraction of temporally adjacent label and control acquisitions, followed by calibration with a reference image (e.g. M0) to obtain the absolute CBF values. Based on these processing steps, drift effects in the baseline of the MR machine are minimized in ASL methods. CBF measurements obtained using ASL have been shown to be reproducible over sequential imaging sessions with intervals from a few minutes to a few days (Floyd et al., 2001; Parkes and Tofts, 2001; Yen et al., 2002), with precision and reproducibility comparable to invasive perfusion imaging methods. A recent functional MR imaging (fMRI) study has also demonstrated that sensorimotor activation,

separated from the resting state as far as 1 day, can be reliably detected using ASL contrast (Wang et al., 2003a).

Comparison of ASL and DSC perfusion MRI As noted above, the most common technique in use for perfusion MRI in cerebrovascular disease is DSCI approach which provides hemodynamic images of high sensitivity within a short period of time, building on the relatively large fractional signal change induced by the contrast agent. However, absolute quantification of cerebral blood volume (CBV) and CBF remains challenging in contrast agent methods due to the difficulties in removing the temporal spread of the intravenous bolus as well as in determining tissue-specific relaxation of susceptibility contrast agents (Østergaard et al., 1996, 1998; Johnson et al., 2000; Calamante et al., 2002). DSCI and ASL perfusion measurements are both prone to systematic errors when the vascular parameters deviate significantly from normal, and the effect of transit delay tends to be greater for ASL methods. Although the SNR is generally lower in the existing ASL methods, potential advantages of ASL perfusion MRI as compared to contrast agent approaches include no requirements of injection, capability for successive measurements, and reduced sensitivity to motion effects due to pairwise subtraction in generating perfusion images. A few studies have compared the performance of these two perfusion MR methods in patients with cerebrovascular disease. In an initial study, a singleslice pulsed arterial spin labelling (PASL) technique (echo planar imaging-(EPI) signal tagging with alternating radio frequency (RF), EPISTAR) was compared to DSCI method in 18 patients with diagnosis of acute stroke, and agreement between the two techniques was found for most patients (Siewert et al., 1997). A more recent study compared DSCI and CASL techniques in 11 patients with known cerebrovascular disease (Wolf et al., 2003). The results demonstrated that time-to-peak (TTP) maps obtained using DSC approach correlated best with CASL perfusion measurements when all subjects studied were considered (cf. Figure 13.1). If subjects with a major transit delay were excluded, however,

209

210

Jiongjiong Wang and John A. Detre

CASL CBF

DSCI rCBF

DSCI TTP

Fig. 13.1 Comparison of CASL and DSCI methods in a patient with left hemispheric TIA symptoms secondary to a left internal carotid artery (ICA) dissection. CASL CBF maps show apparent perfusion deficit involving left middle cerebral artery (MCA) and bilateral anterior cerebral artery (ACA) distributions. DSCI rCBF maps reveal no clear perfusion deficit. DSCI TTP maps reveal a delay in left MCA and bilateral ACA distributions. Adapted from Wolf et al. (2003).

relative CBF (rCBF) images obtained using DSCI approach correlated best with CASL CBF measurements. These data demonstrated the validity and potential of ASL perfusion MRI in the setting of cerebrovascular disease, as well as the importance for characterizing transit-related effects in perfusion quantification. The comparative pros and cons of ASL vs. DSCI methods are illustrated in Table 13.1.

Transit artifact in cerebrovascular ASL Hypoperfusion produces a reduced magnitude of the ASL signal. The prolonged arterial transit time which typically accompanies hypoperfusion may

also result in vascular artifact. This artifact is attributed to a high concentration of labeled spins in arteries that have not exchanged with the microvasculature or tissue, and appears as a bright intra-luminal signal. While over a very large region of interest (ROI) such as an entire vascular distribution CBF values should not be affected by the precise location of the label at the time the image is acquired, locally there is artificially increased apparent perfusion in the region of feeding arterioles as well as underestimation of distal tissue perfusion. The majority of studies on cerebrovascular disease have utilized the CASL approach to provide the maximum perfusion contrast (Wong et al., 1998b; Wang et al., 2002), although the feasibility for PASL methods even at low magnetic field strength (0.5 T) has also been demonstrated (Tsuchiya et al., 2000). The use of a post-labeling delay between the application of ASL and image acquisition can greatly reduce the effects of variations in arterial transit times on perfusion quantification (Alsop and Detre, 1996). A longer post-labeling delay time is generally required in cerebrovascular patients to counteract the effect of delayed transit time. For a representative CASL technique with eight-slice (1 cm thick) coverage (Alsop and Detre, 1998), a delay time of 1.5 s was found to be necessary in most patients with cerebrovascular disease as opposed to an adequate delay of 1.0 s in healthy subjects (Detre and Alsop, 1999). Even using a long post-labeling delay, in many instances, bright linear features were present in CASL perfusion MR images, suggesting effects of delayed arterial transit. While ASL arterial transit effects are likely most problematic in the acute setting, it remains possible to identify areas of hemodynamic compromise and transit effects themselves may provide diagnostically relevant information. Presumably regions of delayed arterial transit are supplied by collateral sources of blood flow. Figure 13.2 illustrates the presence of transit artifact in the left middle cerebral artery (MCA) distribution in a patient with intracranial stenosis. Conventional angiography in this patient demonstrates that this region is supplied primarily by the right carotid artery. As noted above, arterial transit time has been a major confounding factor in ASL perfusion imaging. Most existing techniques to assess arterial transit

Arterial spin labeling perfusion MRI in stroke

Table 13.1. Comparison between DSCI and ASL perfusion MRI methods

Invasiveness Cost Repeatability Speed Slice coverage

Quantification

Motion sensitivity

Contrast-to-noise Artifacts

DSCI

ASL

Requires intravenous access Requires contrast agent Limited due to contrast medium accumulation Faster; no signal averaging Somewhat limited since time course of contrast passage must be sampled at high temporal resolution Relative CBV can be easily quantified; relative CBF and mean transit time (MTT) can be quantified with arterial input and model assumptions Susceptible to motion during injection and contrast bolus passage (⬃1 min)

Noninvasive No contrast agent required Unlimited Slower; requires signal averaging Unlimited with signal averaging

Relatively high because of large effect of contrast agent on brain MRI signal Underestimates flow in regions with delayed arterial transit due to contrast dispersion; may also be inaccurate in high flow regions due to limits in bolus width by intravenous injection

time in ASL rely on multiple measurements of perfusion signals at different post-labeling delay times (Wong et al., 1997; Ye et al., 1997; Yang et al., 1998, 2000; Gonzalez-At et al., 2000; Figueiredo et al., 2002). In normal subjects, shortening of arterial transit time has been found to accompany augmentations in CBF during brain activation (Gonzalez-At et al., 2000; Yang et al., 2000). However, the prohibitively long imaging time to acquire multi-slice arterial transit time images at different delay times limits their potential use in clinical settings and makes the data prone to motion artifacts (Detre et al., 1998; Calamante et al., 2002). Recently, a method termed flow encoding arterial spin tagging (FEAST) has been introduced to measure tissue transit time, determined as the time for the labeled blood to enter microvasculature (capillaries) (Wang et al., 2003b). This technique utilizes appropriate flow encoding bipolar gradients to differentiate the ASL signals in the vascular and microvascular compartments. The ratio of these two quantities yields an estimation of tissue transit time, as more labeled blood water enters the microvascular compartment with shorter

CBF in classical units of ml/100 g/min can be quantified CBV and hemodynamic information (e.g. MTT, TTP) not available with standard implementation Interleaving of ASL and control studies (4–8 s) and background suppression can reduce motion artifacts Relatively low because label is only a few percent of brain MRI signal Underestimates CBF and has intra-luminal signal (transit effect) in regions with delayed arterial transit; may underestimate CBF in very high flow regions due to reduced extraction fraction of blood water

transit time. The significance of transit time measurement is that it can not only detect (correct) transitrelated artifact in ASL perfusion images, but may also provide additional information regarding collateral blood supply. The combination of transit time and perfusion measurements provides the potential to grade the status of cerebral hemodynamic impairment into two stages of “hemodynamic compromise” and “misery perfusion” (Derdeyn et al., 1999). Figure 13.3 illustrates perfusion and transit time mapping in a patient with left MCA and ICA stenosis. The spatial extents of the deficits in perfusion and transit time images match well with clinical results based on MR angiography (MRA).

ASL in acute stroke CASL perfusion MRI has been used to measure perfusion in acute stroke patients at 1.5 T (Chalela et al., 2000). Interpretable data were reliably obtained from all patients and showed focal perfusion deficits corresponding to the vascular distribution of the patients’

211

212

Jiongjiong Wang and John A. Detre

(a)

Before angioplasty

(b)

(c)

After angioplasty

Fig. 13.2 Multi-slice perfusion maps obtained using CASL (a and c) compared with conventional angiography (b) in a patient with left MCA stenosis before and after angioplasty. Only four of eight perfusion MRI slices are shown. (a) Before angioplasty there are focal hyperintensities throughout the left MCA distribution (arrows). (b) Digital subtraction angiogram showing an anteroposterior projection 5 s following a 1 s contrast bolus into the right ICA. Both hemispheres show contrast filling, indicating collaterallization of right internal carotid blood flow to the left hemisphere. At this delay, most of the contrast in the right hemisphere and parasagittally in the left hemisphere is in a capillary blush phase. However, contrast in the left MCA leptomeningeal distribution remains intra-luminal indicating markedly delayed arterial transit time to this region. (c) Following angioplasty, focal hyperintensities in the left MCA distribution are much less evident.

symptoms ranging from small focal deficits of a similar size to diffusion lesions, to large deficits extending well beyond the diffusion lesion. CBF measurements in the affected vascular territories correlated with the severity of stroke indicated by National Institute of Health Stroke Scale (NIHSS) and Rantion Scale scores. Examples of ASL perfusion images and corresponding diffusion-weighted imaging (DWI) are shown in Figure 13.4, illustrating the variability observed. These range from a large perfusion deficit without significant diffusion abnormality in Patient 14, to a

large diffusion abnormality with recovered perfusion in much of the affected territory in Patient 15. ASL methods have also been used in animal models to study the pathophysiology of stroke. ASL imaging in animals provides improved perfusion contrast as compared to humans, because of the increased blood flow, reduced transit time, and higher magnetic field for imaging. Although CBF quantification is less accurate in animals due to limited efforts in refining the corresponding ASL perfusion model, the accuracy and stability of the current techniques are

Arterial spin labeling perfusion MRI in stroke

CBF

150 ml/100 g/min

MRA

0 Tissue transit time

3000 ms

500 ms Fig. 13.3 CASL CBF, tissue transit time and MRA images obtained from a patient with left MCA and ICA stenosis. The MRA results clearly indicate abnormal narrowing (stenosis) of the left MCA, ICA branches as well as blockage of the left communicating artery (arrows). The CASL CBF images show corresponding widespread focal intravascular signals caused by delayed arterial transit time in the left hemisphere. However, average perfusion measured in the left hemisphere (65.4 ml/100 g/min) is even higher than perfusion in the right hemisphere (57.2 ml/100 g/min). The measured tissue transit time using the FEAST technique is prolonged in the whole affected left hemisphere (1644 ms) compared to the normal right side (1168 ms). The tissue transit time results not only explain the transit effects observed in the CASL CBF images, but the spatial extent of the deficits also suggests a lesion in the large proximal arteries which is consistent with the clinical MRA results. Reprinted from Wang et al. (2003b) with permission from John Wiley & Sons.

Patient 1

Patient 5

Patient 11

Patient 15

DWI

Perfusion

Patient 14

Fig. 13.4 Representative examples of various degrees of perfusion (top) and diffusion (bottom) mismatching. Only a single slice is shown for each patient. The extent of perfusion/diffusion mismatching ranges from a large perfusion deficit with no diffusion lesion in Patient 14 (far left) to a large diffusion lesion with a much smaller perfusion deficit in Patient 15 (far right). Reprinted from Chalela et al. (2000) with permission from Lippincott, Williams & Wilkins.

213

214

Jiongjiong Wang and John A. Detre

(a)

CASL perfusion (ml/100 g/min)

(b)

FLAIR (T2-weighted)

150

0 Fig. 13.5 (a) CASL perfusion MRI in a patient with recurrent right hemiparesis showing chronic hypoperfusion in a left MCA distribution along with transit artifact in the left MCA. (b) Concurrent T2-weighted MRI shows only a few scattered hyperintensities in this vascular distribution (FLAIR: fluid attenuated inversion recovery).

adequate to track the CBF change following arterial occlusion (Calamante et al., 1999; Lythgoe et al., 2000; Zaharchuk et al., 2000) or reduced arterial pressure (Zaharchuk et al., 1999b). In rodent ischemic stroke model, time resolved perfusion information provides detailed spatiotemporal pattern of hemodynamic changes following complete or partial occlusion of the MCA. Previous studies have successfully demonstrated that ASL perfusion MRI is able to distinguish between infarct region and tissue at risk in the occluded MCA territory, in conjunction with diffusion MR (Calamante et al., 1999; Lythgoe et al., 2000; Zaharchuk et al., 2000). These data from animal models are in line with the ischemic penumbra concept in human strokes (Touzani et al., 2001), and support a role for perfusion MRI in stroke management.

ASL in chronic cerebrovascular disease While embolism rather than primary hypoperfusion has been considered to be the cause of most cerebrovascular symptoms, patients presenting with stroke, TIA, or severe carotid stenosis may have clinical features suggesting hypoperfusion. An early study using CASL perfusion MRI in such a cohort suggested that resting perfusion abnormalities are

indeed prevalent (Detre et al., 1998), particularly in patients with high-grade stenotic lesions of the cerebral vasculature. ASL perfusion images revealed both focal and hemispheric hypoperfusion, and the localization of CBF abnormalities agreed with the lateralization of most significant stenosis for the anterior circulation distribution. An example of chronic hypoperfusion in a patient with recurrent TIA is shown in Figure 13.5. These findings are consistent with other reports correlating cerebral perfusion or perfusion reserve with the presence of extracranial stenoses of the carotid arteries (Leblanc et al., 1987; Powers et al., 1989; Carpenter et al., 1990; Nighoghossian et al., 1994), yet most studies have failed to clearly implicate primary hypoperfusion as a cause of large vessel stroke. This discrepancy has begun to be reconciled through the hypothesis that hypoperfusion may influence the outcome from cerebral embolization (Caplan and Hennerici, 1998). This hypothesis, if true, suggests an increasingly important role for perfusion imaging in predicting cerebrovascular ischemia, particularly in situations of increased embolization such as cardiovascular surgery. Some patients with carotid stenosis have multiple regions of atherosclerotic narrowing, so-called “tandem lesions”. There is controversy regarding the management of such patients, as the effects of surgically

Arterial spin labeling perfusion MRI in stroke

treating one lesion alone are unknown. As ASL perfusion MRI measures the ultimate effects of stenosis on tissue perfusion, it provides an opportunity both to identify significant perfusion decrements which occur as a result of tandem lesions (e.g. the carotid lesion may only be moderate), and of assessing the response to surgical or endovascular therapy. ASL perfusion MRI has also been applied in other cerebrovascular disorders. For example, relatively normal CBF was observed in a small series of patients with postpartum vasculopathy (Chalela et al., 2001), consistent with the generally benign prognosis for this disorder. The identification of primary hypoperfusion as an etiology for progressive cerebral ischemic injury has important implications for patient management, in particular with regard to antihypertensive therapy. Management of hypertension has been widely recognized as the most important intervention in stroke prevention (Bronner et al., 1995). While clear benefits have been demonstrated for even modest reductions in systolic or diastolic blood pressure, the lower bounds for such reductions have not been established. Evidence now suggests that antihypertensive therapy may be associated with nocturnal hypotension and cerebral hypoperfusion (Watanabe et al., 1996). Antihypertensive therapy induced infarcts occur frequently following acute stroke when cerebral autoregulation is impaired, and the duration of this impairment may be highly variable (Widder et al., 1994). Furthermore, antihypertensive agents may have differing effects on cerebral perfusion, so the choice of specific agents could be influenced by results of perfusion testing. By providing a safe and inexpensive means of assessing cerebral perfusion either before or in response to drug intervention, ASL perfusion MRI might improve the management of antihypertensive therapy in patients at risk for or with known cerebrovascular disease.

Cerebrovascular reserve testing using ASL CBF is thought to be maintained over a broad range of perfusion pressures by a property of the cerebrovascular system termed “autoregulation”. Due to the autoregulatory mechanism to maintain blood

flow in the presence of reduced arterial pressure, perfusion measurement alone may be inadequate to assess the status of hemodynamic compromise. While resting reductions in perfusion are clearly abnormal, alterations in hemodynamic reserve are also significant because they suggest that the autoregulatory capacity of the cerebral vasculature may be exhausted. Cerebrovascular reserve is tested by measuring the increase in CBF induced by carbon dioxide inhalation or acetazolamide administration (“cerebrovascular reactivity”). Several recent studies have demonstrated the utility of measuring vascular reserve in predicting subsequent stroke (Vernieri et al., 1999; Markus and Cullinane, 2001; Imaizumi et al., 2002), although the general validity of this approach has not been established (Derdeyn et al., 1999). Most of the studies assessing hemodynamic impairment used radioactive approaches such as PET, SPECT, or XeCT, and macroscopic blood flow (velocity) was evaluated using transcranial Doppler ultrasonography. The application of MR in studying the status of cerebral circulation has been very limited, and existing CBF measurement based on DSCI is not ideally suited for vascular reactivity testing which requires repeated studies. ASL perfusion MRI provides a convenient method for quantitatively measuring the effects of pharmacological augmentation throughout the brain, because it is noninvasive and can be repeated in the same study. CBF responses to acetazolamide have been studied in patients with cerebrovascular disease using CASL at 1.5 T (Detre et al., 1999). Various patterns of perfusion augmentation failure were observed including diffuse, patchy, or focal deficits corresponding to the vascular distributions of proximal stenoses. Examples of normal and abnormal patterns of pharmacological perfusion augmentation measured using CASL are demonstrated in Figure 13.6. The use of ASL perfusion MRI to monitor pharmacological augmentation of CBF is indicative of its general utility in monitoring CBF changes over various time scales. This may be useful in the management of patients with cerebrovascular disease. Figure 13.7 shows sequential CASL perfusion MRI data obtained from a patient with severe left carotid stenosis before and after urgent carotid endarterectomy (CEA) for recurrent TIA. These

215

Jiongjiong Wang and John A. Detre

100

100

0

(b) Focal deficit

100

200

0

(c) Hyper-augmentation

% change

0

(a) Normal augmentation

% change

100

% change

216

200

Fig. 13.6 Cerebrovascular reserve testing using CASL perfusion MRI before and after acetazolamide administration. Examples from three patients with symptomatic right MCA stenosis are shown. Displayed images show percent CBF change from baseline 10 min following 1 mg intravenous acetazolamide. (a) Normal augmentation pattern showing about 50% increase in CBF in both hemispheres, not suggestive of hemodynamic impairment. (b) Focal augmentation deficit in a right MCA distribution, suggesting impaired hemodynamic reserve. (c) Hyper-augmentation pattern showing a global marked increase in CBF which is attributed to excessive antihypertensive therapy, reducing systemic blood pressure below the autoregulatory range. Adapted from Detre et al. (1999c).

CBF ml/100 g/min 150

Before endarterectomy

After endarterectomy

0 Fig. 13.7 ASL perfusion MRI before (top) and on day after left CEA (bottom) in a patient with bilateral carotid stenosis presenting with recurrent aphasia. Before endarterectomy, CBF is globally reduced to 18 ml/100 g/min with transit artifact present in the left MCA distribution. Following left CEA, global CBF increased to 81 ml/100 g/min with CBF values exceeding 100 ml/100 g/min in the left hemisphere. This patient went on to develop a hyperperfusion syndrome with confusion and seizures.

images demonstrate a marked increase in CBF from ischemic levels prior to CEA to hyperperfusion following CEA. This patient went on to develop a transient hyperperfusion syndrome with seizures and confusion (Hosoda et al., 2001), suggesting the possibility that perfusion MRI could be used to screen for this potential complication.

ASL in pediatric cerebrovascular disease Pediatric stroke is now recognized as a common childhood brain disorder with incidence as high as one in 4000 term birth during the perinatal period (between 28 weeks gestation and 7 days of age) (Lynch and Nelson, 2001; Lynch et al., 2002).

Arterial spin labeling perfusion MRI in stroke

100 ml/ 100 g/min CBF (48 h) 0

CBF (72 h)

DWI (48 h)

Fig. 13.8 PASL CBF and DWI of a 6-year old boy with ischemic stroke. The perfusion images were acquired on two occasions at 48 and 72 h after symptom onset. Hypoperfusion and delayed transit effects (bright focal signals) are present in the left MCA territory, consistent with the infarct location detected in the diffusion images. A mismatch between the perfusion and diffusion deficits is also observed, akin to the ischemic penumbra pattern in adult stroke. Reprinted from Wang et al. (2003c) with permission from John Wiley & Sons.

However, there is no well-documented standard for care and evaluation of pediatric stroke patients, primarily due to the diversity of risk factors and the lack of suitable evaluation methodology (Lanthier et al., 2000; Carlin and Chanmugam, 2002). Existing methods to assess perfusion in adults cannot be easily applied in children because of the safety concerns and technical difficulties associated with the use of radioisotopes and contrast agents in the pediatric population. ASL perfusion MRI may become an ideal and practical approach for pediatric perfusion imaging, and provides unique advantages compared to application of ASL in adult population. While the widespread application of ASL in adult (especially aged) population has been hampered by the relatively small fractional perfusion signal as compared with contrast agent methods, pediatric ASL can provide improved contrast-to-noise ratio due to the normally increased blood flow in children (Chiron et al., 1992). Previous evidence also suggested the water

content of brain is higher in children than adults (Dobbing and Sands, 1973), resulting in increased equilibrium MR signal and spin–lattice, spin–spin relaxation time (T1, T2), thereby may further improve pediatric ASL signal through increased tracer concentration and life time. A recent study has shown a 70% SNR increase of the perfusion images in neurologically normal children as compared to healthy adults, which was attributed to integrated effects of increased CBF (30%), T1, and proton density in the pediatric population (Wang et al., 2003c). An example of perfusion and diffusion images of a pediatric patient with ischemic stroke (6-year old boy) is given in Figure 13.8. Regions of hypoperfusion accompanied by delayed transit effects (bright focal signal) were present in the left MCA territory, consistent with the infarct location detected in the DWI. The mismatch between the perfusion and diffusion deficits resembles the ischemic penumbra pattern in adult stroke. The two CBF measurements with 1 day interval (48 and 72 h

217

218

Jiongjiong Wang and John A. Detre

after symptom onset) were reproducible, which demonstrated the stability of the technique as well as its potential use in longitudinal studies to track the development of pediatric stroke.

Improving ASL for clinical applications One major limitation of ASL perfusion MRI, as compared to contrast agent approaches, is the relatively low SNR due to the small fractional signal of the labeled blood (1% raw MR signal). ASL perfusion images might not be of adequate image quality even after lengthy signal averaging, resulting in reduced diagnostic reliability and sensitivity to brain activity. One natural solution to this problem is to implement the ASL technique at high field. High field ASL would provide not only increased SNR which is proportional to the main field strength, but also an important advantage for labeling. Due to the increased relaxation time T1 at high field, the loss of spin labeling during the transit time is much less than that at standard field, producing greater perfusion signals in brain tissue while reducing the arterial transitrelated artifacts and quantification errors. This T1 effect, in combination with the increased SNR, could yield major improvements in the perfusion signal allowing increased spatial and temporal resolution as well as wider imaging coverage. A 2-fold SNR gain is readily achievable by performing PASL at 4.0 T as compared to ASL methods at 1.5 T (Wang et al., 2002). The technical challenge to implement CASL at high field is the high level of RF deposition, which can be overcome using special multi-coil approach (Zaharchuk et al., 1999a) or single-coil approach with simultaneous modulation of the RF and gradient for labeling (Alsop, 2001). Image acquisition schemes in ASL techniques are also evolving rapidly to improve the spatial and temporal resolution, as well as to reduce the susceptibility effects in perfusion imaging. To date, most of the ASL perfusion methods have used fast gradient-echo (GE) techniques such as EPI for image acquisition because of the speed. Susceptibility contrast is manifested in the resulting perfusion images, and causes signal loss or distortion at tissue– air and tissue–bone interfaces, especially at high magnetic field. In order

to circumvent this problem, susceptibility-resistant techniques, especially spin-echo (SE) approaches, will be highly preferable for ASL perfusion imaging. Several approaches have been carried out in ASL, including fast spin-echo (FSE) methods with repeated RF refocusing (Chen et al., 1997; Crelier et al., 1999; Liu et al., 2001), spin-echo EPI (Wang et al., 2004), reversed spiral (Yang et al., 2002), and multi-shot three-dimensional (3D) imaging based on background suppression (Alsop and Detre, 1999; Ye et al., 2000b). The opportunity to obtain high quality perfusion images of the whole brain in a few minutes is fast at hand.

Conclusions The past decade has seen the development of ASL perfusion MRI from feasibility studies into practical clinical research applications, at least at academic medical centers where this methodology is available. Perfusion information is achievable in patients with cerebrovascular disease using existing ASL methods. The use of high magnetic field strengths and transit time imaging may further improve ASL image quality and reliability. These developments may allow ASL to become a realistic clinical tool in the diagnosis and prognosis of cerebrovascular disease, especially for children.

REFERENCES Alsop DC. 2001. Improved efficiency for multi-slice continuous arterial spin labeling using time varying gradients. Proc Inter Soc Magn Reson Med 9: 1562. Alsop DC, Detre JA. 1996. Reduced transit-time sensitivity in noninvasive magnetic resonance imaging of human cerebral blood flow. J Cereb Blood Flow Metab 16: 1236–1249. Alsop DC, Detre JA. 1998. Multisection cerebral blood flow MR imaging with continuous arterial spin labeling. Radiology 208: 410–416. Alsop DC, Detre JA. 1999. Background suppressed 3D RARE arterial spin labeled perfusion MRI. Proc Intl Soc Magn Reson Med 7: 601. Baird AE, Benfield A, Schlaug G, Siewert B, Lovblad KO, Edelman RR, Warach S. 1997. Enlargement of human

Arterial spin labeling perfusion MRI in stroke

cerebral ischemic lesion volumes measured by diffusionweighted magnetic resonance imaging. Ann Neurol 41(5): 581–589. Bogousslavsky J, Delaloye-Bischof A, Regli F, Delaloye B. 1990. Prolonged hypoperfusion and early stroke after transient ischemic attack. Stroke 21: 40–46. Bronner LL, Kanter DS, Manson JE. 1995. Primary prevention of stroke (review). N Engl J Med 333: 1392–1400. Buxton RB, Frank LR, Wong EC, Siewert B, Warach S, Edelman RR. 1998. A general kinetic model for quantitative perfusion imaging with arterial spin labeling. Magn Reson Med 40(3): 383–396. Calamante F, Gadian DG, Connelly A. 2002. Quantification of perfusion using bolus tracking magnetic resonance imaging in stroke: assumptions, limitations, and potential implications for clinical use. Stroke 33(4): 1146–1151. Calamante F, Lythgoe MF, Pell GS, Thomas DL, King MD, Busza AL, Sotak CH, Williams SR, Ordidge RJ, Gadian DG. 1999. Early changes in water diffusion, perfusion, T1, and T2 during focal cerebral ischemia in the rat studied at 8.5 T. Magn Reson Med 41(3): 479–485. Caplan LR, Hennerici M. 1998. Impaired clearance of emboli (washout) is an important link between hypoperfusion, embolism, and ischemic stroke. Arch Neurol 55: 1475–1482. Carlin TM, Chanmugam A. 2002. Stroke in children. Emerg Med Clin North Am 20(3): 671–685. Carpenter DA, Grubb Jr RL, Powers WJ. 1990. Borderzone hemodynamics in cerebrovascular disease. Neurology 40: 1587–1592. Chalela JA, Alsop DC, Gonzalez-Atavalez JB, Maldjian JA, Kasner SE, Detre JA. 2000. Magnetic resonance perfusion imaging in acute ischemic stroke using continuous arterial spin labeling. Stroke 31: 680–687. Chalela JA, Kasner SE, McGarvey M, Alsop DC, Detre JA. 2001. Continuous arterial spin labeling perfusion magnetic resonance imaging findings in postpartum vasculopathy. J Neuroimaging 11: 444–446. Chen Q, Siewert B, Bly BM, Warach S, Edelman RR. 1997. STAR-HASTE: perfusion imaging without magnetic susceptibility artifact. Magn Reson Med 38: 404–408. Chiron C, Raynaud C, Maziere B, Zilbovicius M, Laflamme L, Masure MC, Dulac O, Bourguignon M, Syrota A. 1992. Changes in regional cerebral blood flow during brain maturation in children and adolescents. J Nucl Med 33(5): 696–703. Crelier GR, Hoge RD, Munger P, Pike GB. 1999. Perfusion-based functional magnetic resonance imaging with single-shot RARE and GRASE acquisitions. Magn Reson Med 41: 132–136. Derdeyn CP, Grubb Jr RL, Powers WJ. 1999. Cerebral hemodynamic impairment: methods of measurement and association with stroke risk. Neurology 53(2): 251–259.

Detre JA, Alsop DC. 1999a. Perfusion fMRI with arterial spin labeling (ASL). In Functional MRI (Eds., Moonen CTW, Bandettini PA), Springer-Verlag, Heidelberg, pp. 47–62. Detre JA, Alsop DC. 1999b. Perfusion MRI with continuous arterial spin labeling: methods and clinical applications in the nervous system. Eur J Radiol 30: 115–124. Detre JA, Alsop DC, Vives LR, Maccotta L, Teener JW, Raps EC. 1998. Noninvasive MRI evaluation of cerebral blood flow in cerebrovascular disease. Neurology 50: 633–641. Detre JA, Samuels OB, Alsop DC, Gonzalez-At JB, Kasner SE, Raps EC. 1999c. Noninvasive MRI evaluation of CBF with acetazolamide challenge in patients with cerebrovascular stenosis. J Magn Reson Imaging 10: 870–875. Detre JA, Zhang W, Roberts DA, Silva AC, Williams DS, Grandis DJ, Koretsky AP, Leigh JS. 1994. Tissue specific perfusion imaging using arterial spin labeling. NMR Biomed 7: 75–82. Dobbing J, Sands J. 1973. Quantitative growth and development of human brain. Arch Dis Child 48(10): 757–767. Ewing JR, Cao Y, Fenstermacher J. 2001. Single-coil arterial spin-tagging for estimating cerebral blood flow as viewed from the capillary: relative contributions of intra- and extravascular signal. Magn Reson Med 46: 465–475. Fiehler J, von Bezold M, Kucinski T, Knab R, Eckert B, Wittkugel O, Zeumer H, Rother J. 2002. Cerebral blood flow predicts lesion growth in acute stroke patients. Stroke 33(10): 2421–2425. Figueiredo P, Clare S, Jezzard P. 2002. Issues in quantitative perfusion and arterial transit time mapping using pulsed ASL. Proc Intl Soc Magn Reson Med 10: 623. Firlik AD, Rubin G, Yonas H, Wechsler LR. 1998. Relation between cerebral blood flow and neurologic deficit resolution in acute ischemic stroke. Neurology 51(1): 177–182. Fisher M, Bogousslavsky J. 1998. Further evolution toward effective therapy for acute ischemic stroke. J Am Med Assoc 279: 1298–1303. Floyd TF, Maldjian JA, Gonzalez-Atavales JB, Detre JA. 2001. Test–retest stability with continuous arterial spin labeled (CASL) perfusion MRI in regional measurement of cerebral blood flow. Proc Intl Soc Magn Reson Med 9: 1569. Gonzalez-At JB, Alsop DC, Detre JA. 2000. Perfusion and transit time changes during task activation determined with steady-state arterial spin labeling. Magn Reson Med 43: 739–746. Grandin CB, Duprez TP, Smith AM, Mataigne F, Peeters A, Oppenheim C, Cosnard G. 2001. Usefulness of magnetic resonance-derived quantitative measurements of cerebral blood flow and volume in prediction of infarct growth in hyperacute stroke. Stroke 32(5): 1147–1153. Guckel F, Brix G, Rempp K, Deimling M, Rother J, Georgi M. 1994. Assessment of cerebral blood volume with dynamic

219

220

Jiongjiong Wang and John A. Detre

susceptibility contrast enhanced gradient-echo imaging. J Comput Assist Tomogr 18(3): 344–351. Gur AY, Bova I, Bornstein NM. 1996. Is impaired cerebral vasomotor reactivity a predictive factor of stroke in asymptomatic patients? Stroke 27(12): 2188–2190. Hosoda K, Kawaguchi T, Shibata Y, Kamei M, Kidoguchi K, Koyama J, Fujita S, Tamaki N. 2001. Cerebral vasoreactivity and internal carotid artery flow help to identify patients at risk for hyperperfusion after carotid endarterectomy. Stroke 32(7): 1567–1573. Imaizumi M, Kitagawa K, Hashikawa K, Oku N, Teratani T, Takasawa M, Yoshikawa T, Rishu P, Ohtsuki T, Hori M, Matsumoto M, Nishimura T. 2002. Detection of misery perfusion with split-dose 123I-iodoamphetamine single-photon emission computed tomography in patients with carotid occlusive diseases. Stroke 33(9): 2217–2223. Johnson KM, Tao JZ, Kennan RP, Gore JC. 2000. Intravascular susceptibility agent effects on tissue transverse relocation rates in vivo. Magn Reson Med 44: 909–914. Kety SS, Schmidt CF. 1945. The determination of cerebral blood flow in man by the use of nitrous oxide in low concentrations. Am J Physiol 143: 53–66. Kim SG. 1995. Quantification of relative cerebral blood flow change by flow-sensitive alternating inversion recovery (FAIR) technique: application to functional mapping. Magn Reson Med 34: 293–301. Kwong KK, Chesler DA, Weisskoff RM, Donahue KM, Davis TL, Ostergaard L, Campbell TA, Rosen BR. 1995. MR perfusion studies with T1-weighted echo planar imaging. Magn Reson Med 34: 878–887. Lanthier S, Carmant L, David M, Larbrisseau A, de Veber G. 2000. Stroke in children: the coexistence of multiple risk factors predicts poor outcome. Neurology 54(2): 371–378. Leblanc R, Yamamoto YL, Tyler JL, Diksic M, Hakim A. 1987. Borderzone ischemia. Ann Neurol 22: 707–713. Liu HL, Kochunov P, Hou J, Pu Y, Mahankali S, Feng CM, Yee SH, Wan YL, Fox PT, Gao JH. 2001. Perfusion-weighted imaging of interictal hypoperfusion in temporal lobe epilepsy using FAIR-HASTE: comparison with H(2)(15)O PET measurements. Magn Reson Med 45(3): 431–435. Lynch JK, Nelson KB. 2001. Epidemiology of perinatal stroke. Curr Opin Pediatr 13(6): 499–505. Lynch JK, Hirtz DG, DeVeber G, Nelson KB. 2002. Report of the National Institute of Neurological Disorders and Stroke workshop on perinatal and childhood stroke. Pediatrics 109(1): 116–123. Lythgoe MF, Thomas DL, Calamante F, Pell GS, King MD, Busza AL, Sotak CH, Williams SR, Ordidge RJ, Gadian DG. 2000. Acute changes in MRI diffusion, perfusion, T(1), and T(2) in a rat model of oligemia produced by partial occlusion of the middle cerebral artery. Magn Reson Med 44(5) 706–712.

Maccotta L, Detre JA, Alsop DC. 1997. The efficiency of adiabatic inversion for perfusion imaging by arterial spin labeling. NMR Biomed 10: 216–221. Markus H, Cullinane M. 2001. Severely impaired cerebrovascular reactivity predicts stroke and TIA risk in patients with carotid artery stenosis and occlusion. Brain 124(Pt 3): 457–467. McLaughlin AC, Ye FQ, Pekar JJ, Santha AK, Frank JA. 1997. Effect of magnetization transfer on the measurement of cerebral blood flow using steady-state arterial spin tagging approaches: a theoretical investigation. Magn Reson Med 37(4): 501–510. Mull M, Schwarz M, Thron A. 1997. Cerebral hemispheric lowflow infarcts in arterial occlusive disease; lesion patterns and angiomorphological conditions. Stroke 28: 118–123. Nighoghossian N, Trouillas P, Philippon B, Itti R, Adeleine P. 1994. Cerebral blood flow reserve assessment in symptomatic versus asymptomatic high-grade internal carotid artery stenosis. Stroke 25(5): 1010–1013. Østergaard L, Smith DF, Vestergaard-Poulsen P, Hansen SB, Gee AD, Gjedde A, Gyldensted C. 1998. Absolute cerebral blood flow and blood volume measured by magnetic resonance imaging bolus tracking: comparison with positron emission tomography values. J Cereb Blood Flow Metab 18(4): 425–432. Østergaard L, Weisskoff RM, Chesler DA, Gyldensted C, Rosen BR. 1996. High resolution measurement of cerebral blood flow using intravascular tracer bolus passages. Part I: mathematical approach and statistical analysis. Magn Reson Med 36(5): 715–725. Pantoni L, Garcia JH, Gutierrez JA. 1996. Cerebral white matter is highly vulnerable to ischemia. Stroke 27: 1641–1647. Parkes LM, Tofts PS. 2001. Normal cerebral perfusion changes over time measured using arterial spin labeling. Proc Inter Soc Magn Reson Med 9: 1571. Parkes LM, Tofts PS. 2002. Improved accuracy of human cerebral blood perfusion measurements using arterial spin labeling: accounting for capillary water permeability. Magn Reson Med 48(1): 27–41. Powers WJ, Tempel LW, Grubb Jr RL. 1989. Influence of cerebral hemodynamics on stroke risk: one-year follow-up of 30 medically treated patients. Ann Neurol 25: 325–330. Roberts DA, Detre JA, Bolinger L, Insko EK, Leigh Jr JS. 1994. Quantitative magnetic resonance imaging of human brain perfusion at 1.5 T using steady-state inversion of arterial water. Proc Natl Acad Sci USA 91: 33–37. Rubin G, Firlik AD, Levy EI, Pindzola RR, Yonas H. 2000. Relationship between cerebral blood flow and clinical outcome in acute stroke. Cerebrovasc Dis 10(4): 298–306. Siewert B, Schlaug G, Edelman RR, Warach S. 1997. Comparison of EPISTAR and T2*-weighted gadolinium-enhanced

Arterial spin labeling perfusion MRI in stroke

perfusion imaging in patients with acute cerebral ischemia. Neurology 48: 673–679. Smith AM, Grandin CB, Duprez T, Mataigne F, Cosnard G. 2000. Whole brain quantitative CBF, CBV, and MTT measurements using MRI bolus tracking: implementation and application to data acquired from hyperacute stroke patients. J Magn Reson Imaging 12(3): 400–410. Touzani O, Roussel S, MacKenzie ET. 2001. The ischaemic penumbra. Curr Opin Neurol 14(1): 83–88. Tsuchiya K, Katase S, Hachiya J, Kimura T, Yodo K. 2000. Cerebral perfusion MRI with arterial spin labeling technique at 0.5 Tesla. J Comput Assist Tomogr 24(1): 124–127. Utting JF, Thomas DL, Gadian DG, Ordidge RJ. 2003. Velocitydriven adiabatic fast passage for arterial spin labeling: results from a computer model. Magn Reson Med 49(2): 398–401. Vernieri F, Pasqualetti P, Passarelli F, Rossini PM, Silvestrini M. 1999. Outcome of carotid artery occlusion is predicted by cerebrovascular reactivity. Stroke 30: 593–598. Wang J, Aguirre GK, Kimberg DY, Roc AC, Li L, Detre JA. 2003a. Arterial spin labeling perfusion FMRI with very low task frequency. Magn Reson Med 49: 796–802. Wang J, Alsop DC, Li L, Listerud J, Gonzalez-At JB, Schnall MD, Detre JA. 2002. Comparison of quantitative perfusion imaging using arterial spin labeling at 1.5 and 4.0 Tesla. Magn Reson Med 48(2): 242–254. Wang J, Alsop DC, Song HK, Maldjian JA, Tang K, Schnall MD, Detre JA. 2003b. Transit time imaging with flow encoding arterial spin tagging (FEAST). Magn Reson Med 50: 599–607. Wang J, Li L, Roc AC, Alsop DC, Tang K, Butler N, Schnall MD, Detre JA. 2004. Reduced susceptibility effect in perfusion fMRI using single-shot spin-echo EPI acquisitions. J Magn Reson Imaging 22: 1–7. Wang J, Licht DJ, Liu CS, Jahng GH, Haselgrove JC, Zimmerman RA, Detre JA. 2003c. Pediatric perfusion imaging using pulsed arterial spin labeling. J Magn Reson Imaging 18: 404–413. Watanabe N, Imai Y, Nagai K, Tsuji I, Satoh H, Sakuma M, Sakuma H, Kato J, Onodera-Kikuchi N, Yamada M, Abe F, Hisamichi S, Abe K. 1996. Nocturnal blood pressure and silent cerebrovascular lesions in elderly Japanese. Stroke 27: 1319–1327. Webster MW, Makaroun MS, Steed DL, Smith HA, Johnson DW, Yonas H. 1995. Compromised cerebral blood flow reactivity is a predictor of stroke in patients with symptomatic carotid artery occlusive disease. J Vasc Surg 21(2): 338–344. Weiller C, Ringelstein E, Reiche W, Buell U. 1991. Clinical and hemodynamic aspects of low-flow infarcts. Stroke 22: 1117–1123.

Widder B, Kleiser B, Krapf H. 1994. Course of cerebrovascular reactivity in patients with carotid artery occlusions. Stroke 25(10): 1963–1967. Wolf RL, Alsop DC, McGarvey ML, Maldjian JA, Wang J, Detre JA. 2003. Susceptibility contrast and arterial spin labeled perfusion MRI in cerebrovascular disease. J Neuroimaging 13(1): 17–27. Wong EC. 1999. Potential and pitfalls of arterial spin labeling based perfusion imaging techniques for MRI. In Functional MRI (Eds., Moonen CTW, Bandettini PA), Springer-Verlag, Heidelberg, pp. 63–69. Wong EC, Buxton RB, Frank LR. 1997. Implementation of quantitative perfusion imaging techniques for functional brain mapping using pulsed arterial spin labeling. NMR Biomed 10(4–5): pp. 237–249. Wong EC, Buxton RB, Frank LR. 1998a. Quantitative imaging of perfusion using a single subtraction (QUIPSS and QUIPSS II). Magn Reson Med 39(5): 702–708. Wong EC, Buxton RB, Frank LR. 1998b. A theoretical and experimental comparison of continuous and pulsed arterial spin labeling techniques for quantitative perfusion imaging. Magn Reson Med 40(3): 348–355. Yang Y, Engelien W, Xu S, Gu H, Silbersweig DA, Stern E. 2000. Transit time, trailing time, and cerebral blood flow during brain activation: measurement using multislice, pulsed spin-labeling perfusion imaging. Magn Reson Med 44(5): 680–685. Yang Y, Frank JA, Hou L, Ye FQ, McLaughlin AC, Duyn JH. 1998. Multislice imaging of quantitative cerebral perfusion with pulsed arterial spin labeling. Magn Reson Med 39(5); 825–832. Yang Y, Gu H, Zhan W, Xu S, Silbersweig DA, Stern E. 2002. Simultaneous perfusion and BOLD imaging using reverse spiral scanning at 3T: characterization of functional contrast and susceptibility artifacts. Magn Reson Med 48(2): 278–289. Ye FQ, Berman KF, Ellmore T, Esposito G, van Horn JD, Yang Y, Duyn J, Smith AM, Frank JA, Weinberger DR, McLaughlin AC. 2000a. H(2)(15)O PET validation of steady-state arterial spin tagging cerebral blood flow measurements in humans. Magn Reson Med 44(3): 450–456. Ye FQ, Mattay VS, Jezzard P, Frank JA, Weinberger DR, McLaughlin AC. 1997. Correction for vascular artifacts in cerebral blood flow values measured by using arterial spin tagging techniques. Magn Reson Med 37(2): 226–235. Ye FQ, Frank JA, Weinberger DR, McLaughlin AC. 2000b. Noise reduction in 3D perfusion imaging by attenuating the static signal in arterial spin tagging (ASSIST). Magn Reson Med 44(1): 92–100. Yen YF, Field AS, Martin EM, Ari N, Burdette JH, Moody DM, Takahashi

AM.

2002.

Test–retest

reproducibility

of

221

222

Jiongjiong Wang and John A. Detre

quantitative CBF measurements using FAIR perfusion MRI and acetazolamide challenge. Magn Reson Med 47(5): 921–928. Yongbi MN, Yang Y, Frank JA, Duyn JH. 1999. Multislice perfusion imaging in human brain using the C-FOCI inversion pulse: comparison with hyperbolic secant. Magn Reson Med 42(6): 1098–1105. Zaharchuk G, Ledden PJ, Kwong KK, Reese TG, Rosen BR, Wald LL. 1999a. Multislice perfusion and perfusion territory imaging in humans with separate label and image coils. Magn Reson Med 41(6): 1093–1098.

Zaharchuk G, Mandeville JB, Bogdanov Jr AA, Weissleder R, Rosen BR, Marota JJ. 1999b. Cerebrovascular dynamics of autoregulation and hypoperfusion. An MRI study of CBF and changes in total and microvascular cerebral blood volume during hemorrhagic hypotension. Stroke 30(10): 2197–2204. Zaharchuk G, Yamada M, Sasamata M, Jenkins BG, Moskowitz MA, Rosen BR. 2000. Is all perfusion-weighted magnetic resonance imaging for stroke equal? The temporal evolution of multiple hemodynamic parameters after focal ischemia in rats correlated with evidence of infarction. J Cereb Blood Flow Metab 20(9): 1341–1351.

14

MR diffusion-tensor imaging in stroke Pamela W. Schaefer, Luca Roccatagliata and R. Gilberto Gonzalez Department of Radiology, Massachusetts General Hospital, Boston, USA

Key points • Similar to diffusion-weighted imaging, diffusiontensor imaging (DTI) can be used to detect reduced diffusion in acute cerebral ischemia. • Fractional anisotropy (FA) is variable in acute stroke, it may be slightly increased or reduced. FA decreases with time after stroke. • Decreased FA (and fiber tracking) may be helpful for detecting Wallerian degeneration. • DTI may be helpful in distinguishing stroke mimics (with increased average diffusivity) from acute cerebral ischemia (reduced average diffusivity).

Background Diffusion MR imaging (MRI) has greatly improved diagnostic accuracy in acute stroke as documented elsewhere in this book. To date, most reported diffusion data is based on the calculation of the apparent diffusion coefficient (ADC) along the direction of a single gradient or along three orthogonal gradients with averaging to minimize the effects of anisotropy. Diffusion anisotropy refers to the principle that in tissues, water diffusion is different in different directions due to tissue structural geometry. White matter (WM) has relatively high anisotropy; diffusion is much greater parallel than perpendicular to major WM tracts. Gray matter (GM) has relatively low anisotropy. MR diffusion-tensor imaging (DTI) permits measurements that are related

to tissue diffusion anisotropy, and may provide additional diagnostic information in ischemic stroke. The purpose of this chapter is to report progress in the development and applications of DTI in stroke. To derive information related to the anisotropic diffusion of water in the brain, investigators sample the full diffusion tensor (Pierpaoli et al., 1996). DTI allows the calculation of three groups of parameters. 1. The trace of the diffusion tensor [Tr(ADC)] or the average diffusivity, D (D  (1  2  3)/3 where 1, 2, and 3 are the eigenvalues of the diffusion tensor). These allow the calculation of the overall diffusion in a region of tissue, independent of direction (Le Bihan et al., 2001). 2. Indices of diffusion anisotropy such as fractional anisotropy (FA) or the lattice index (LI). These allow the calculation of the degree of differences in diffusion in different directions (Basser and Pierpaoli, 1996; Shimony et al., 1999). 3. Fiber orientation mapping. This provides information on structure and integrity of WM tracts and the connectivity between them (Makris et al., 1997; Conturo et al., 1999; Bammer et al., 2003). All of these parameters have been measured in ischemia and show promise as diagnostic aids in this disease.

Average diffusivity in acute stroke Measurement of diffusion (average diffusivity) in acute stroke with full DTI has provided information similar to that obtained with the averaged orthogonal 223

224

Pamela W. Schaefer, Luca Roccatagliata and R. Gilberto Gonzalez

approach (Zelaya et al., 1999). That is, the measured drop in diffusion associated with acute ischemia and the change in diffusion over time are similar to those based on the averaged orthogonal approach. However, measurement of D has also provided new information on differences between GM and WM diffusion that was not appreciable with the measurement of diffusion along a single gradient direction or along three orthogonal directions (Sorensen, 1999; Mukherjee et al., 2000). These differences are likely detected with DTI because DTI has a much higher signal to noise ratio. Yang et al. (1999) in a study of 26 patients, imaged at less than at 24 h, at 3–5 days and at 3 months with DTI, demonstrated that overall diffusion decreases were greater in WM than GM in the acute and subacute periods. At outcome, diffusion increases were much higher in WM than in GM. In a study of 12 patients with unilateral middle cerebral artery (MCA) territory infarctions, imaged at 17 h to 5 days after stroke onset, Mukherjee et al. (2000) found that in the infarcted regions, D values were less in WM than in GM in every patient. On average, D values were reduced by 46% in WM but only by 31% in GM. Furthermore, D images detected regions of reduced WM diffusion that were not visualized on diffusion-weighted imaging (DWI) images. Mukerjee et al. (2000) point out that while GM has been considered to be more vulnerable than WM in acute stroke, recent animal experiments have demonstrated that histopathological changes occur in WM as early as 30 min after acute stroke onset and can be severe. One study conducted in cats, following 3 h of MCA occlusion, demonstrated astrocytic swelling in GM and WM as well as hydropic swelling of oligodendrocytes in WM (Kuroiwa et al., 1998). Also, the authors point out that biophysical mechanisms of diffusion reduction, such as reduced bulk water motion from cytoskeletal collapse and disruption of fast axonal transport, that do not exist in GM, may occur in WM. By contrast, Sorensen et al. (1999) in a study of 50 patients imaged at less than 24 h, found no significant differences between GM and WM diffusion with DTI. It is clear that additional DTI studies are needed to delineate the nature of WM and GM diffusion changes in acute stroke.

FA in acute stroke Biophysical basis of water diffusion anisotropy in normal tissues Diffusion anisotropy refers to the principle that water diffusion is different in different directions due to tissue structural geometry (Reese et al., 1995; Pierpaoli et al., 1996). GM has relatively low diffusion anisotropy. WM has relatively high anisotropy in part due to highly organized WM tract bundles. Diffusion is much greater parallel than perpendicular to WM tract bundles (Moseley et al., 1990, 1991; Pierpaoli et al., 1996; Le Bihan et al., 2001). The myelin sheath is considered an essential component for anisotropic diffusion since lipid rich myelin has limited permeability to water and contributes to hindering diffusion across fibers (Rutherford et al., 1991; Sakuma et al., 1991). However, this model is too simplistic since diffusion anisotropy is present in the normal, intact, non-myelinated olfactory nerve of the garfish (Beaulieu and Allen, 1994; Beaulieu, 2002) and in rat brain, significant anisotropy develops before myelin can be detected by histochemistry and electron microscopy (Prayer et al., 2001). Diffusion anisotropy in unmyelinated WM may result from a significant increase in oligodendrocyte number around axon fascicles and/or the development of functional ionic channels immediately before myelination (Waxman and Ritchie, 1993; Prayer et al., 2001). It may be that these factors also contribute to diffusion anisotropy in myelinated WM. Fast axonal transport may also contribute to diffusion anisotropy. However, diffusion anisotropy is preserved in experiments where excised myelinated and non-myelinated nerves of the garfish are treated with vinblastine that depolymerize microtubules and inhibit fast axonal transport (Beaulieu and Allen, 1994). Furthermore, it has been proposed that the intracellular compartment is more anisotropic while the extracellular compartment is less anisotropic (Le Bihan and Van Zijl, 2002). Sources of anisotropy in the intracellular compartment could be the presence of microtubules, organelles and intact membranes (Beaulieu, 2002).

MR diffusion-tensor imaging in stroke

(a)

(b)

(c)

(d)

(e)

(f)

(g)

(h)

(i)

Fig. 14.1 Evolution of FA changes in ischemic stroke. A 50-years-old male with left hemiparesis was examined at less than 6 h (Figure 14.1(a)–(c)), 3 days (Figure 14.1(d)–(f)), and 3 months (Figure 14.1(g)–(i)) after onset. At less than 6 h (row 1), the right corona radiata/caudate body stroke is hyperintense on FA images (a), hyperintense on isotropic DWI images (b), and hypointense on ADC images (c). These findings are consistent with the first stage of FA changes in stroke described by Yang et al. (1999). After 3 days (row 2), the lesion is hypointense on FA images (d), hyperintense on DWI images (e), and hypointense on ADC images (f). These findings are consistent with second stage of FA changes in stroke described by Yang et al. (1999). At 3 months (row 3), the lesion is hypointense on FA images (g), hypointense on DWI images (h), and hyperintense on ADC images (i). These findings are consistent with third stage of FA changes in stroke described by Yang et al. (1999).

Time course of FA changes in acute stroke and correlation with ADC and T2 changes In the setting of acute stroke, FA is variable, but correlates with time of stroke onset (Figures 14.1 and 14.2). In general, FA is elevated (FA ratio greater than one) or slightly reduced in the hyperacute and early

acute periods, and progressively decreases over time (Yang et al., 1999; Zelaya et al., 1999). Yang et al. (1999) demonstrated that there is heterogeneity between lesions and within lesions of FA evolution. That is, there are different temporal rates of stroke progression. They described three patterns of FA

225

226

Pamela W. Schaefer, Luca Roccatagliata and R. Gilberto Gonzalez

(a)

(b)

(c)

(d)

(e)

(f)

Fig. 14.2 Correlation of FA changes with T2 changes in ischemic stroke. A 50-year-old male with left hemiparesis was examined at less than 6 h (Figure 14.1(a) and (b)), 3 days (Figure 14.1(c) and (d)), and 3 months (Figure 14.1(e) and (f)) after onset. At less than 6 h, the right putamen stroke is hyperintense on FA images (1a) and is not visualized on T2-weighted images (1b). After 3 days, the lesion is hypointense on FA images (1c) and hyperintense on T2-weighted images (1d). At 3 months, the lesion remains hypointense on FA images (1e) and hyperintense on T2-weighted images (1f).

MR diffusion-tensor imaging in stroke

evolution: (1) elevated FA acutely and subacutely, (2) elevated FA acutely and reduced FA subacutely, and (3) reduced FA acutely and subacutely. In that study, small and large lesions had similar FA evolution except that no deep GM lesions had elevated FA acutely or subacutely. Similarly, Zelaya et al. (1999) studied serial anisotropy measurements in six patients with acute stroke. In two, there was a progressive decrease in FA over time, in two there were initial elevations in FA followed by decreases and later increases. In two there were early progressive decreases in FA over time, followed by later elevations of FA. In the patients with initial elevation, the patients had better functional improvement in the first 80 h and relatively small perfusion deficits compared with the other patients. Differences in temporal rates of stroke progression may relate to differences in timing of reperfusion and differences in collateralization. They may also relate to structural differences between GM and WM, ischemia induced spreading depression, differences in the amount of glial scar tissue (this may explain the late elevations of FA in two patients) and apoptosis in which delayed cell death may occur in mildly ischemic tissue for days after the initial infarction. Also, Green et al. (2002) performed DTI in 10 patients with acute stroke and decomposed the total magnitude of the tensor (L) into the p (isotropic) and q (anisotropic) components where FA is proportional to q/L. They found that there were significant reductions in p, q, and L but that FA was elevated in 5 of 10 patients due to a larger reduction in L than q. Yang et al. (1999) also described three temporally related different phases in the relationship between FA and ADC. Increased FA and reduced ADC characterize the first phase, reduced FA and reduced ADC characterize the second phase and reduced FA with elevated ADC characterizes the third phase. Furthermore, FA inversely correlates with T2 signal change (Ozsunar et al., 2001). These changes can be explained as follows. As cytotoxic edema develops, there is shift of water from the extracellular to the intracellular space, but cell membranes remain intact and there is not yet an overall increase in tissue water. This would explain elevated FA, reduced ADC and normal T2. As the ischemic insult

progresses, there is an overall increase in tissue water, predominantly in the extracellular space as cells lyse, the glial reaction occurs and there is disruption of the blood brain barrier. This would explain reduced FA, elevated ADC and elevated T2. Reduced FA, reduced ADC and elevated T2 may occur when there is an overall increase in tissue water, the intracellular fraction is still high enough to cause reduced ADC and the extracellular portion is high enough to cause reduced FA. Other factors which may contribute to decreases in FA over time include loss of axonal transport, loss of cellular integrity and decreases in interstitial fluid flow.

FA: differences between GM and WM in acute stroke Sorensen et al. (1999) in a study of 50 patients with acute stroke, demonstrated that the FA of ischemic WM significantly decreased compared to normal contralateral brain tissue while the FA of ischemic GM did not significantly change compared to normal contralateral brain tissue. Furthermore, Yang et al. (1999) found that both cortical GM and WM could have elevated or decreased FA in the acute period and showed progressive decreases in FA over time. However, WM showed significantly greater decreases in FA at 3 months compared with cortical GM. Also, deep GM had slightly reduced FA in the acute period and did not show significant change at 3 months. The reason for these differences likely relates to the structural differences between GM and WM. In the WM extracellular space, there are dense arrays of parallel WM tracts, and there is much greater diffusion along than perpendicular to the WM tracts. With acute ischemia, the diffusion decrease is much greater in 1, the eigenvalue that coincides with the long axis of the WM fiber tracts, compared with the other eigenvalues. In the GM extracellular space, there is a meshwork. With acute ischemia, the diffusion decrease is more similar between eigenvalues.

FA: correlation with clinical outcome Yang et al. (1999) demonstrated statistically significant correlations between FA ratios measured

227

228

Pamela W. Schaefer, Luca Roccatagliata and R. Gilberto Gonzalez

within 12 h in 26 acute infarctions and the acute Canadian Neurological Scale (at 12 h, r  0.46), the subacute Canadian Neurological Scale (at 3–5 days, r  0.55), the outcome Barthel index (at 3 months, r  0.62), and the Rankin Scale (at 3 months, r 

0.77). By contrast, no significant correlation was found between ADC and any of the clinical scores. The authors suggest that early FA changes reflect the severity of ischemic injury and may predict stroke outcome. FA ratios measured in the subacute and chronic stages did not correlate with any of the clinical scales. At later time points, extracellular space swelling with vasogenic edema and membrane fragmentation may lead to more heterogeneity of FA measurements within ischemic tissue.

FA in predicting tissue viability In rare cases of acute stroke, when there is reversal of the DWI abnormality (acute depression in ADC without T2 abnormality on follow-up imaging), increased FA has been reported (Yang et al., 1999). However, increased FA is also frequently observed in the core of acute infarctions. Also, in one study, FA ratios were not significantly different in penumbra that progresses to infarction compared with penumbra that does not progress to infarction (Schaefer et al., 2003). It may be that tissue with increased FA is potentially salvageable with early intervention, but it is clear that more work is needed in this area.

FA in diffuse ischemic diseases A number of studies have addressed measurements of FA and diffusion in diffuse ischemic diseases such as leukoariosis, cerebral autosomal dominant arteriopathy with subcortical infarcts and leukoencephalopathy (CADASIL) and multiple lacunar infarction (LACI) (Ishihara et al., 1999; Jones et al., 1999; Molko et al., 2001; Pierpaoli et al., 2001). Jones et al. (1999) measured FA and mean diffusivity in the forceps major and minor in patients with ischemic leukoariosis (radiological leukoariosis and clinical lacunar stroke). They found a characteristic abnormal pattern with decreased FA and increased mean diffusivity. In addition, there was a strong negative

correlation between mean diffusivity and FA (p  0.92).They concluded that these changes are consistent with axonal loss and proliferation of glial cells characteristic of ischemic leukoariosis and may be important in improving diagnosis, and monitoring disease progression. Molko et al. (2001) assessed mean diffusivity and FA in the thalami and putamina in patients with CADASIL. They found a significant increase in mean Tr(D) in the putamina and thalami with or without small infarcts and a significant decrease in FA in the putamina and thalami with small infarcts and in the thalami without small infarcts. The right/left indices of Tr(D) in the thalami positively correlated with the right/left indices of Tr(D) values as well with the right/left indices of FA in the WM. The diffusion increase in the thalami correlated positively with increase in diffusion and lesion load in small deep WM infarcts and negatively with the mini-mental state examination (MMSE). They concluded that microstructural tissue alterations are present in the thalami and putamina outside of the typical lacunar infarcts associated with CADASIL and those thalamic changes may result from degeneration of thalamocortical pathways secondary to ischemic WM injury. Ishihara et al. (1999) found a positive correlation between active frontal cortical volume and degree of FA in the anterior corpus callosum of nine patients with multiple lacunar infarctions and vascular dementia and six volunteers. They concluded that there may be a relationship between anisotropy in the anterior corpus callosum and frontal associative function.

Wallerian degeneration and fiber orientation mapping Investigators are beginning to evaluate how strokes affect adjacent WM tracts. Higano et al. (2001) measured FA changes in the corona radiata and internal capsule ipsilateral to acute infarctions in 16 patients. They found that the FA was significantly decreased in the internal capsules and corona radiata in patients with moderate to severe hemiparesis compared with patients with no or mild hemiparesis. None of these

MR diffusion-tensor imaging in stroke

patients had detectable Wallerian degeneration on T2-weighted images. In addition, Watanabe et al. (2001) performed three-dimensional anisotropy contrast MR axonography in 16 stroke patients at 2, 3, and 10 weeks following stroke onset. They found a significant reduction in the eigenvalues (blue) perpendicular to the axial imaging plane at 2–3 weeks in eight patients with poor recovery following acute stroke, but not in eight patients with good recovery. Furthermore, in the eight patients with poor recovery, the Wallerian degeneration was not detected until 10 weeks on T2-weighted images. The authors concluded that axonography can detect Wallerian degeneration prior to conventional images and that it may be useful in predicting motor function prognosis. Other investigators have assessed Wallerian degeneration at later timepoints. Werring et al. (2000) in a study of five patients at 2–6 months following MCA infarction, demonstrated loss of coherence of eigenvectors in the corticospinal tract ispsilateral to the stroke. This study and another study assessing DTI in Wallerian degeneration in the rostral pons and cerebral peduncle in seven patients with internal capsule strokes at least 1 year following injury (Werring et al., 2000; Pierpaoli et al., 2001) demonstrated that, in general, the primary stroke had reduced FA and elevated mean diffusivity while the corticospinal tract had reduced FA but preserved or only slightly elevated mean diffusivity. However, in the pons, where there are crossing fibers, there is little change in FA. Both authors concluded that DTI can therefore distinguish between the primary lesion and the region of Wallerian degeneration. Pierpaoli et al. (2001) also demonstrated that diffusion-tensor imaging (DTI) based fiber tractography, using an algorithm that traces a path along which the local diffusivity is the maximum, may lead to erroneous calculations of brain connectivity. For example, they demonstrated that where there is Wallerian degeneration, reconstructed trajectories cross over to the contralateral side at the pontine level and continue along contralateral motor pathways. DTI in stroke mimics Syndromes with potentially reversible vasogenic edema include eclampsia (Schaefer et al., 1997;

Koch et al., 2001), hypertensive encephalopathy (Schwartz et al., 1998) cyclosporin toxicity, other posterior leukoencephalopathies (Hinchey et al., 1996), venous thrombosis (Chu et al., 2001; Ducreux et al., 2001), HIV encephalopathy, and hyperperfusion syndrome following carotid endarterectomy (CEA) (Schaefer et al., 2000, 2001). Patients with these syndromes frequently present with neurological deficits, which raise the question of acute ischemic stroke, or with neurological deficits such as headache or seizure, which suggest vasogenic edema, but ischemic stroke is still a strong diagnostic consideration. Furthermore, conventional imaging cannot always differentiate cytotoxic from vasogenic edema. Both types of edema produce T2 hyperintensity in GM and/or WM. DTI, however, can reliably distinguish vasogenic from cytotoxic edema (Schaefer et al., 1997). Vasogenic edema is characterized by elevated diffusion due to a relative increase in water in the extracellular compartment where water is more mobile and cytotoxic edema is characterized by restricted diffusion (Ito et al., 1996; Barzo et al., 1997). Vasogenic edema is characterized on the diffusion MR images by hypointense to slightly hyperintense signal because these images have both T2 and diffusion contributions. When vasogenic edema is hyperintense on the diffusion MR images, it can mimic hyperacute or subacute infarction. On mean diffusivity images cytotoxic edema due to ischemia is always hypointense for 1–2 weeks and vasogenic edema is always hyperintense. Therefore, comparison of the mean diffusivity images with diffusion MR images is mandatory for accurate diagnosis. In posterior reversible leukoencephalopathy, the FA in affected WM regions is decreased and correlates inversely with the mean diffusivity which is increased (Figure 14.3). While studies of diffuse WM ischemic diseases have suggested that reductions in anisotropy can be used as a marker for irreversible WM injury, decreases in anisotropy associated with posterior reversible leukoencephalopathy are reversible. Furthermore, in a case in which posterior reversible leukoencephalopathy was complicated by acute ischemia, the degree of anisotropy loss was much greater in the region of vasogenic edema compared with cytotoxic edema. So FA may provide adjunctive

229

230

Pamela W. Schaefer, Luca Roccatagliata and R. Gilberto Gonzalez

(a)

(b)

(d)

(e)

(c)

Fig. 14.3 Reversible posterior leukoencephalopathy in a 64-year-old female who presented with altered mental status. Fluidattenuated inversion recovery (FLAIR) images (a) show confluent bilateral hyperintense lesions predominantly in the parietooccipital subcortical WM (arrows). The lesions are not evident on isotropic DWI images (b), hyperintense on ADC images (c), hypointense on exponential images (d), and hypointense on FA images (e), consistent with vasogenic edema with an increase of water in the extracellular space.

information in differentiating cytotoxic from vasogenic edema (Mukherjee and McKinstry, 2001). Differentiating vasogenic from cytotoxic edema affects patient management and outcome. The incorrect diagnosis of a vasogenic edema syndrome as acute ischemia could lead to unnecessary and potentially dangerous use of thrombolytics, antiplatelet agents, anticoagulants and vasoactive agents. Furthermore, failure to correct relative hypertension could result in increased cerebral edema, hemorrhage, seizures or death. Misinterpretation of acute ischemic infarction as a vasogenic edema syndrome would discourage anticoagulation, evaluation for an

embolic source and liberal blood pressure control which could increase the risk of recurrent brain infarction.

Conclusion DTI is beginning to provide new insights into the pathophysiology of acute stroke. The time course of FA and mean diffusivity changes in combination with ADC and T2 have been delineated. This may be important in determining stroke onset time and in triaging patients for thrombolysis or other potential

MR diffusion-tensor imaging in stroke

Fig. 14.4 Wallerian degeneration. FA images obtained 3 months after a stroke in the right MCA territory demonstrates decreased anisotropy (low signal) along the right corticospinal tract (arrows), consistent with Wallerian degeneration.

stroke therapies. FA ratios correlate with clinical neurological scales and may prove to be important in determining tissue viability and patient outcome. FA values are different in GM and WM strokes and may help delineate differences in GM and WM postischemic tissue responses. DTI can detect Wallerian degeneration prior to conventional images and it may be useful in predicting motor function prognosis (Figure 14.4). DTI is providing new insights into the neural pathways affected by diffuse ischemic diseases such as leukoariosis and CADASIL. Further investigation will undoubtedly yield new information that will improve treatment of acute stroke patients.

REFERENCES Bammer R, Acar B, Moseley ME. 2003. In vivo MR tractography using diffusion imaging. Eur J Radiol 45: 223–234. Barzo P, Marmarou A, Fatouros P, Hayasaki K, Corwin F. 1997. Contribution of vasogenic and cellular oedema to traumatic brain swelling measured by diffusion-weighted imaging. J Neurosurg 87: 900–907. Basser PJ, Pierpaoli C. 1996. Microstructural and physiological features of tissues elucidated by quantitativediffusion-tensor MRI. J Magn Reson B 111: 209–219. Beaulieu C. 2002. The basis of anisotropic water diffusion in the nervous system – a technical review. NMR Biomed 15: 435–455.

231

232

Pamela W. Schaefer, Luca Roccatagliata and R. Gilberto Gonzalez

Beaulieu C, Allen PS. 1994. Determinants of anisotropic water diffusion in nerves. Magn Reson Med 31: 394–400. Chu K, Kang DW, Yoon BW, Roh JK. 2001. Diffusion-weighted magnetic resonance in cerebral venous thrombosis. Arch Neurol 58: 1569–1576. Conturo TE, Lori NF, Cull TS, et al. 1999. Tracking neuronal fibre pathways in the living human brain. Proc Natl Acad Sci USA 96: 10422–10427. Ducreux D, Oppenheim C, Vandamme X, et al. 2001. Diffusion-weighted imaging patterns of brain damage associated with cerebral venous thrombosis. Am J Neuroradiol 22: 261–268. Green HA, Pena A, Price CJ, et al. 2002. Increased anisotropy in acute stroke: a possible explanation. Stroke 33: 1517–1521. Higano S, Zhong J, Shrier DA, et al. 2001. Diffusion anisotropy of the internal capsule and the corona radiata in association with stroke and tumors as measured by diffusion-weighted MR imaging. Am J Neuroradiol 22:456–463. Hinchey J, Chaves C, Appignani B, et al. 1996. A reversible posterior leukoencephalopathy syndrome. New Engl J Med 334: 494–500. Ishihara M, Kumita S, Hayashi H, Kumazaki T. 1999. Loss of interhemispheric connectivity in patients with lacunar infarction reflected by diffusion-weighted MR imaging and single-photon emission CT. Am J Neuroradiol 20: 991–998. Ito J, Marmarou A, Barzo P, Fatouros P, Corwin F. 1996. Characterization of oedema by diffusion-weighted imaging in experimental traumatic brain injury. J Neurosurg 84: 97–103. Jones DK, Lythgoe D, Horsfield MA, Simmons A, Williams SC, Markus HS. 1999. Characterization of white matter damage in ischaemic leukoaraiosis with diffusion tensor MRI. Stroke 30: 393–397. Koch S, Rabinstein A, Falcone S, Forteza A. 2001. Diffusionweighted imaging shows cytotoxic and vasogenic oedema in eclampsia. Am J Neuroradiol 22: 1068–1070. Kuroiwa T, Nagaoka T, Ueki M, Yamada I, Miyasaka N, Akimoto H. 1998. Different apparent diffusion coefficient: water content correlations of gray and white matter during early ischaemia. Stroke 29: 859–865. Le Bihan D, Van Zijl P. 2002. From the diffusion coefficient to the diffusion tensor. NMR Biomed 15: 431–434. Le Bihan D, Mangin JF, Poupon C, et al. 2001. Diffusion tensor imaging: concepts and applications. J Magn Reson Imaging 13: 534–546. Makris N, Worth AJ, Sorensen AG, et al. 1997. Morphometry of in vivo human white matter association pathways with diffusion-weighted magnetic resonance imaging. Ann Neurol 42: 951–962.

Molko N, Pappata S, Mangin JF, et al. 2001. Diffusion tensor imaging study of subcortical gray matter in cadasil. Stroke 32: 2049–2054. Moseley ME, Cohen Y, Kucharczyk J, et al. 1990. Diffusionweighted MR imaging of anisotropic water diffusion in cat central nervous system. Radiology 176: 439–445. Moseley ME, Kucharczyk J, Asgari HS, Norman D. 1991. Anisotropy in diffusion-weighted MRI. Magn Reson Med 19: 321–326. Mukherjee P, McKinstry RC. 2001. Reversible posterior leukoencephalopathy syndrome: evaluation with diffusiontensor MR imaging. Radiology 219: 756–765. Mukherjee P, Bahn MM, McKinstry RC, et al. 2000. Differences between gray matter and white matter water diffusion in stroke: diffusion-tensor MR imaging in 12 patients. Radiology 215: 211–220. Ozsunar Y GE, Huisman TA, Wu O, Sorensen AG, Gonzalez G. 2001. Evolution of water diffusion anisotropy in hyperacute ischaemia: correlation between fractional anisotropy and T2. In: Book of Abstracts of the 87th Scientific Assembly and Annual Meeting of the Radiological Society of North America, Chicago, November 25–30, 2001, vol. 221. Pierpaoli C, Jezzard P, Basser PJ, Barnett A, Di Chiro G. 1996. Diffusion tensor MR imaging of the human brain. Radiology 201: 637–648. Pierpaoli C, Barnett A, Pajevic S, et al. 2001. Water diffusion changes in Wallerian degeneration and their dependence on white matter architecture. Neuroimage 13: 1174–1185. Prayer D, Barkovich AJ, Kirschner DA, et al. 2001. Visualization of nonstructural changes in early white matter development on diffusion-weighted MR images: evidence supporting premyelination anisotropy. Am J Neuroradiol 22: 1572–1576. Reese TG, Weisskoff RM, Smith RN, Rosen BR, Dinsmore RE, Wedeen VJ. 1995. Imaging myocardial fibre architecture in vivo with magnetic resonance. Magn Reson Med 34: 786–791. Rutherford MA, Cowan FM, Manzur AY, et al. 1991. MR imaging of anisotropically restricted diffusion in the brain of neonates and infants. J Comput Assist Tomogr 15: 188–198. Sakuma H, Nomura Y, Takeda K, et al. 1991. Adult and neonatal human brain: diffusional anisotropy and myelination with diffusion-weighted MR imaging. Radiology 180: 229–233. Schaefer PW, Buonanno FS, Gonzalez RG, Schwamm LH. 1997. Diffusion-weighted imaging discriminates between cytotoxic and vasogenic oedema in a patient with eclampsia. Stroke 28: 1082–1085. Schaefer PW, Grant PE, Gonzalez RG. 2000. Diffusionweighted MR imaging of the brain. Radiology 217: 331–345. Schaefer PW, Gonzalez RG, Hunter G, Wang B, Koroshetz WJ, Schwamm LH. 2001. Diagnostic value of apparent diffusion

MR diffusion-tensor imaging in stroke

coefficient hyperintensity in selected patients with acute neurologic deficits. J Neuroimaging 11: 369–380. Schaefer PW, Ozsunar Y, He J, et al. 2003. Assessing tissue viability with MR diffusion and perfusion imaging. Am J Neuroradiol 24: 436–443. Schwartz RB, Mulkern RV, Gudbjartsson H, Jolesz F. 1998. Diffusion-weighted MR imaging in hypertensive encephalopathy: clues to pathogenesis. Am J Neuroradiol 19: 859–862. Shimony JS, McKinstry RC, Akbudak E, et al. 1999. Quantitative diffusion-tensor anisotropy brain MR imaging: normative human data and anatomic analysis. Radiology 212: 770–784. Sorensen AG, Wu O, Copen WA, et al. 1999. Human acute cerebral ischaemia: detection of changes in water diffusion anisotropy by using MR imaging. Radiology 212: 785–792.

Watanabe T, Honda Y, Fujii Y, Koyama M, Matsuzawa H, Tanaka R. 2001. Three-dimensional anisotropy contrast magnetic resonance axonography to predict the prognosis for motor function in patients suffering from stroke. J Neurosurg 94: 955–960. Waxman SG, Ritchie JM. 1993. Molecular dissection of the myelinated axon. Ann Neurol 33: 121–136. Werring DJ, Toosy AT, Clark CA, et al. 2000. Diffusion tensor imaging can detect and quantify corticospinal tract degeneration after stroke. J Neurol Neurosurg Psychiat 69: 269–272. Yang Q, Tress BM, Barber PA, et al. 1999. Serial study of apparent diffusion coefficient and anisotropy in patients with acute stroke. Stroke 30: 2382–2390. Zelaya F, Flood N, Chalk JB, et al. 1999. An evaluation of the time dependence of the anisotropy of the water diffusion tensor in acute human ischaemia. Magn Reson Imaging 17: 331–348.

233

15

MR spectroscopy in severe obstructive carotid artery disease Jeroen van der Grond and Dirk R. Rutgers Department of Radiology, University Medical Center Utrecht, Heidelberglaan CX Utrecht, The Netherlands

Key points • MR spectroscopy reveals abnormalities related to neuronal loss (N-acetyl aspartate (NAA) loss) and anaerobic metabolism (lactate elevation) in regions with impaired blood flow. • Metabolic abnormalities correlate better with clinical features than with degree of arterial stenosis. • No clear relationship between metabolism and blood flow; collateral flow patterns or cerebrovascular reactivity has been found in chronic lesions. • Decreased NAA/choline in symptomatic internal carotid artery occlusion is a predictor of future ischemic events (an association not yet investigated in stenosis or asymptomatic cases). • Metabolite measures may provide surrogate markers of therapeutic response.

stimulated by a reduction of the peripheral resistance due to vasodilatation of the peripheral brain arteries. Under normal circumstances this mechanism is adequate and a small decrease in the cerebral perfusion pressure (CPP) has little effect on the cerebral blood flow (CBF). However, when the CPP continues to fall this mechanism may be insufficient to maintain a normal CBF. Three stages of hemodynamic failure have been identified (Powers et al., 1987; Derdeyn et al., 2002; Nemoto et al., 2003). Stage 1 of hemodynamic compensation is identified as an increase in cerebral blood volume (CBV) in the hemisphere distal to the occlusive lesion, with normal CBF, oxygen Hemodynamic impairment Stage 1

Stage 2

Stage 3

CBF

CBV

Hemodynamic changes in severe obstructive carotid artery disease OEF

Severe stenosis or occlusion of the internal carotid artery (ICA) causes a reduction in arterial pressure distal to the stenosis or occlusion. This activates several regulatory mechanisms in the brain to maintain cellular function. The primary physiological changes are recruitment of collateral channels, i.e. collateral flow via the circle of Willis, via the ophthalmic artery (OphtA) or via leptomeningeal vessels. The recruitment of blood flow via these alternative channels is 234

CMRO2

120

90 60 30 Perfusion pressure (mm Hg)

0

Fig. 15.1 Compensatory cerebral response mechanisms to reduced CPP.

MR spectroscopy in severe obstructive carotid artery disease

extraction fraction (OEF) and the cerebral metabolic rate of oxygen metabolism (CMRO2) (Figure 15.1). When the mechanisms for collateral flow are limited, e.g. poor collateral flow in the circle of Willis or poorly developed leptomeningeal vessels, or when the capacity for compensatory vasodilatation has been exceeded, autoregulation fails, and a decrease in the CPP results in a deceased CBF (Leblanc et al., 1987; Powers, 1991; Yamauchi et al., 1992, 1996). To preserve cellular integrity CMRO2 the OEF increases (Leblanc et al., 1987). This situation is known as Stage 2 of hemodynamic failure. In addition, it was recently recognized that when the arterial pressure continues to fall, CBF, CBV, OEF and CMRO2 are all likely to decrease (Stage 3 of hemodynamic failure) (Nemoto et al., 2003). In patients with chronic cerebral hemodynamic compromise, it has been suggested that Stage 3 is important in the pathophysiology of so-called lowflow infarcts (Nemoto et al., 2003). These low-flow infarcts, or border-zone infarcts, are located in the most distal part of the perfusion territory of the main cerebral arteries, and are the first areas to suffer ischemic damage when blood flow decreases (Challa et al., 1990; Pappata et al., 1993; Garcia et al., 1996). The entire concept of hemodynamical staging is based on positron emission tomography (PET) techniques in patients with severe atherosclerotic carotid artery stenosis or occlusion, and have been widely applied in the study of human cerebrovascular disease (Hirano et al., 1994; Yamauchi et al., 1996).

Why perform MR spectroscopy in patients with severe obstructive carotid artery disease? The underlying assumption of the use of MR spectroscopy (MRS) in ICA occlusive disease is that changes in CMRO2 could also be monitored by changes in MRS-visible brain metabolites. Changes in cerebral metabolism can be monitored with MRS, which determines the concentrations of N-acetylaspartate (NAA), choline (Cho), creatine (Cr) and lactate (Lac) in the brain (cf. Chapter 1). In vivo proton MRS of the human brain has been applied to physiological and pathological conditions, including stroke. Preliminary studies in the late 1980s have shown alterations in the proton spectrum of

acute cerebral infarcts (Berkelbach et al., 1988; Bruhn et al., 1989; Gideon et al., 1992). Brain infarction results in a marked decrease of cellular density: dead neurons and dead glial cells are removed by macrophages, while edema, astrocytes and new glial cells partially fill the infarcted regions, resulting in a large decrease in NAA and also in a decrease in the Cho and Cr concentration. After the onset of cerebral ischemia, deterioration of the energy status and accumulation of Lac are observed, generally as a result of the change from aerobic metabolism into anaerobic glycolysis in the infarcted brain area. The metabolic changes in patients with ICA obstructive disease are less predictable since, in general, little information is available on their exact stage of hemodynamic impairment. This dilemma is worsened by the fact that no straightforward relation exists between symptomatology and stage of hemodynamic impairment. Moreover, the precise etiology of infarction in patients with severe ICA lesions is still controversial. It has been reported that hemodynamic impairment due to obstructive disease of the ICA is a cause for infarction (Bogousslavsky and Regli, 1986a; Baumgartner and Regard, 1994). However, other studies did not find an association between infarction and hemodynamic injury and reported that arterial embolism is responsible for ischemic lesions in territories that are presumably at risk for low-flow infarctions (Torvik, 1984; Graeber et al., 1992; Hennerici et al., 1998; Derdeyn et al., 2001). Recently, it was suggested that an impaired clearance of (micro-) emboli due to hypoperfusion in the border-zones territories could also cause ischemic lesions (Caplan and Hennerici, 1998). Information on neuronal damage and/or anaerobic glycolysis could offer practical information to determine the hemodynamic status of the brain, which is beneficial for the clinical management of the individual patient. This information could be especially useful in identifying those patients most likely to suffer infarcts, and in whom the risk-benefit of intervention is most favorable. In particular, the decision whether or not to perform carotid endarterectomy (CEA) in patients with a less than 70% ICA stenosis, or in asymptomatic patients with an ICA stenosis. In patients with ICA occlusion, such information is also valuable in the decision whether or not to perform CEA on the contralateral ICA or to perform bypass surgery.

235

236

Jeroen van der Grond and Dirk R. Rutgers

A

H A A P

A

A

L

H

P

Fig. 15.2 Typical location of the MRS-Volumes of Interest (VOI) in the centrum semi-ovale.

MR spectra of symptomatic and asymptomatic patients with severe obstructive carotid artery disease Patients with ICA stenosis or occlusion have a high prevalence of border-zone lesions (Del Sette et al., 2000; Szabo et al., 2001). These border-zone lesions can generally be divided into lesions in the external and internal border zones. External border-zone infarcts are located between the supply territories of the middle cerebral artery (MCA) and anterior cerebral artery (ACA) or between the MCA and posterior cerebral artery (PCA), whereas internal border-zone infarcts are located between the superficial and deep branches of the MCA (Torvik, 1984; Bogousslavsky and Regli, 1986a, 1986b). These regions are located in the most distal part of the perfusion territory of the main cerebral arteries and receive the lowest CBF (Moody et al., 1990). These regions are expected to be the first areas to suffer ischemic damage when blood flow decreases (Challa et al., 1990). Since the location of external border-zone infarcts is unclear due to the large variation in the vascular territories, especially between

the PCA and MCA (Henniciri et al., 1998; van der Zwan et al., 1992), many MRS studies have chosen to study cerebral metabolic changes in internal borderzone regions or in the flow territory between the ACA and MCA (anterior external border-zone region). Figure 15.2 shows a typical MRS-volume of interest (VOI) located in the latter region (centrum semiovale) of a healthy control subject. Figure 15.3 shows the corresponding long echo (repetition time (TR)/echo time (TE): 2000/144 ms) MR spectrum, showing clearly identifiable Cho, Cr and NAA resonances. Figure 15.4 shows an MR spectrum measured in the centrum semi-ovale on the symptomatic side of a 62-year-old male patient with a severe (70%) ICA stenosis who suffered from a transient ischemic attack (TIA). Figure 15.5 shows a similar MR spectrum measured on the symptomatic side of a 67year-old male patient with an ICA occlusion who suffered from a TIA. The latter spectra are characterized by a slightly reduced NAA concentration, increased Cho concentration and have a relatively higher prevalence of Lac, compared with control subjects. Similar findings have also been reported previously (van der Grond et al., 1996c; Tsuchida et al., 2000).

MR spectroscopy in severe obstructive carotid artery disease

NAA

NAA Cho

Cr Cho Cr

Lac

3.5

3.0

2.5 2.0 1.5 1.0 Chemical shift (ppm)

0.5

0.0

Fig. 15.3 The corresponding (Figure 15.2) long echo (TR/TE: 2000/144 ms) MR spectrum.

NAA

Cho Cr

Lac

3.5

3.0

2.5 2.0 1.5 1.0 0.5 Chemical shift (ppm)

0.0

Fig. 15.4 The 1H-MR spectrum on the symptomatic side of a 62-year-old male patient with a severe (70%) ICA stenosis who suffered from TIA.

Tsuchida et al. (2000) have shown that the NAA/Cr ratio correlated with the CMRO2 in patients with ICA occlusive disease. Although a (complex and indirect) relation exists between the presence of severe ICA lesions and deteriorated cerebral metabolism, the clinical symptomatology of the individual patients may also affect the cerebral metabolism. In this respect, differences in cerebral metabolism between the symptomatic and asymptomatic hemispheres in

3.5

3.0

2.5 2.0 1.5 1.0 Chemical shift (ppm)

0.5

0.0

Fig. 15.5 The 1H-MR spectrum on the symptomatic side of a 67-year-old male patient with an ICA occlusion who suffered from TIA.

patients with bilateral ICA occlusion or bilateral severe stenosis have been reported (van Everdingen et al., 2000a). In addition, metabolic differences between patients with hemispheric (TIA, stroke) and non-hemispheric (transient monocular blindness, chronic ocular ischemia) neurological deficit, with both ICA stenosis and occlusion, have been found. The NAA/Cho ratio in patients with hemispherical complaints have been found to be significantly lower than in similar patients with non-hemispheric complaints (Rutgers et al., 2000a; van Everdingen et al., 2000b). Moreover, the NAA/Cho ratio in patients with non-hemispheric complaints did not significantly differ from that obtained in healthy control subjects (Rutgers et al., 2000a; van Everdingen et al., 2000b). These findings are partly confirmed in a recent study which included many asymptomatic patients with unilateral stenosis or occlusion and showed no significant metabolic changes between patients and control subjects (Lythgoe et al., 2001). In patients with severe ICA stenosis, large randomization studies have shown that surgical intervention by CEA is beneficial (European Carotid Surgery Trialists’ Collaborative Group, 1991; North American Symptomatic Carotid Endarterectomy Trial (NASCET), 1991a; Rothwell et al., 2003). Still, it remains unclear the degree to which beneficial effect of CEA is caused by removal of a possible source of emboli, or

237

238

Jeroen van der Grond and Dirk R. Rutgers

(a)

(b)

(a)

(b)

(c)

(d)

(c)

(d)

Fig. 15.6 MR spectroscopic imaging (MRSI) study of a 74year-old man with a severe stenosis (70% reduction) of the left ICA. (a) Cho metabolic map, (b) NAA metabolic map, (c) Lac metabolic map and (d) corresponding MR imaging (MRI). The white box within the brain parenchyma indicates the VOI selected for 1H MRSI. On each metabolic map the high-pass filtered T2-weighted MRI (Figure 15.1(d)) was superimposed.

by restoration of the arterial perfusion pressure. Although collateral flow is often sufficient to keep the CBF at a level at which the brain function is not affected, as mentioned previously, in some cases it is not sufficient to accommodate metabolic demands, resulting in cerebral infarcts. Regions surrounding these infarcts or regions with low flow, that are not infarcted yet, may contain cells that still receive sufficient oxygen and glucose to survive, but insufficient to maintain function. It has been shown previously that re-establishment of the CBF by CEA may improve cellular function (van der Grond et al., 1996a; Kim et al., 2001, 2002). Figures 15.6 and 15.7 show the results of a MR spectroscopic imaging (MRSI) study of a 74-year-old man with severe stenosis (70% reduction) of the left ICA before and after CEA. Before CEA relatively large Lac hyperintensities can be observed in the symptomatic hemisphere, which disappears after CEA. The high signal intensities outside the VOI in the Lac image are caused by contamination of lipid resonances. The finding of a decreased prevalence of Lac in the symptomatic hemisphere after CEA was recently confirmed in a study in which CEA of the contralateral ICA (asymptomatic side) was

Fig. 15.7 MRSI study of the same patient as shown in Figure 15.6 but after CEA. (a) Cho metabolic map, (b) NAA metabolic map, (c) Lac metabolic map and (d) corresponding MRI.

performed in a group of patients with a symptomatic ICA occlusion (Rutgers et al., 2001). Lac is an important marker of ischemic cellular damage (Graham et al., 1995). Previous studies have shown that the Lac concentration correlates with both the degree of acute neurological impairment and the degree of permanent cerebral injury (Graham et al., 1995). These results are in agreement with studies showing that the presence of cerebral Lac is related to an increased amount of cerebral damage after the restoration of the CPP (Wagner et al., 1992; Chopp et al., 1987).

The association of MRS and hemodynamical measures Although the degree of ICA stenosis is somehow associated with the degree of hemodynamic impairment, it has been shown that no clear correlation between the two exists (Powers, 1991). To identify the stage of hemodynamic failure in the individual patient, examination of other parameters is necessary. It has been shown that determination of the level of hemodynamic impairment by measuring the CBV, CBF, OEF and CMRO2 with PET is useful. Additionally, other more available methods are being used to identify the level of hemodynamic impairment. With the widely available MR technique the presence of low-flow

MR spectroscopy in severe obstructive carotid artery disease

infarctions can be studied, indicating hemodynamic failure. Moreover, MR imaging (MRI) can also be used to determine the CBV and CBF or to investigate the anatomy and function of the circle of Willis (cf. Chapter 16). Another commonly used measure of hemodynamic failure is the determination of the cerebrovascular reserve (CVR) capacity, reflecting the ability of brain arterioles to dilate when the CPP decreases.

(a)

ICA

BA

Flow and MRS in patients with severe obstructive carotid artery disease Studies in acute stroke patients have shown that severe reductions in cerebral perfusion can cause spectroscopic alterations in the brain (Berkelbach et al., 1988; Bruhn et al., 1989; Gideon et al., 1992). In patients with severe obstructive carotid artery disease, who may generally be characterized by a more chronic development of perfusion abnormalities than acute stroke patients, such a relation may be less obvious. Several studies have investigated the relation between flow and MRS in patients with severe carotid stenosis or occlusion. For example, van der Grond et al. (1996b) studied 56 symptomatic patients (TIA or minor stroke) with a 70% ICA stenosis. The patients were investigated within 3 months after the onset of symptoms. On the symptomatic side, a positive correlation was found between volume flow in the carotid artery (common as well as internal) and the NAA/Cho ratio in the centrum semi-ovale. Figure 15.8(a) and (b) illustrates how volume flow through the ICAs, basilar artery and MCAs can be measured with phase contrast (PC) MR angiography (MRA). Uno et al. (2001) investigated 17 patients with severe ICA stenosis (5 TIAs, 12 asymptomatic) and reported a positive relation between the NAA concentration and regional CBF as measured with single photon emission computed tomography (SPECT). No distinction was made between symptomatic and asymptomatic patients. Tsuchida et al. (2000) investigated 11 symptomatic patients (TIA or minor stroke) with severe obstructive carotid artery disease within 3 months after the onset of symptoms. Their results showed a positive correlation between the NAA/Cr ratio and regional CBF in the centrum semi-ovale. In contrast to these studies, others found no correlation between flow

(b) MCA

Fig. 15.8 (a) Sagittal localizer MRA scan shows the typical positioning of a transverse two-dimensional (2D) phase contrast MRA slice through the ICAs and the basilar artery to measure volume flow in these arteries.(b) Maximum intensity projection of a transverse three-dimensional (3D) time-offlight MRA scan of the circle of Willis shows the typical positioning of an 2D phase contrast MRA slice through the left MCA to measure volume flow in this artery.

and metabolism. Van Everdingen et al. (2000a) investigated 170 patients with severe ICA stenosis or occlusion within 6 months after symptoms occurred. No relation was found between the NAA/Cho ratio in the centrum semi-ovale and volume flow in the ICAs, the basilar artery and the MCAs. Rutgers et al. (2003) studied 19 patients with a symptomatic ICA occlusion (TIA, minor stroke or retinal ischemia) within 6 months after the onset of symptoms. Regional quantitative perfusion was measured with dynamic susceptibility contrast imaging (DSCI) MRI

239

240

Jeroen van der Grond and Dirk R. Rutgers

Fig 15.9 CBF image through the brain at the level of the centrum semi-ovale, obtained by DSCI in a patient with carotid artery occlusion.

(cf. Figure 15.9). No correlation was found between the NAA, Cho and Cr concentration in the centrum semi-ovale and regional quantitative perfusion. Although these studies illustrate that the relationship between flow and MRS in severe obstructive carotid artery disease has not been investigated extensively, some general remarks can be made. In a chronic clinical stage, i.e. several months after the symptoms occurred, no relation between flow and metabolism has been observed. As both cerebral metabolism (Rutgers et al., 2000b) and CBF (Derdeyn et al., 1999) can normalize over time, although not necessarily simultaneously, such a relation may be present in an earlier clinical stage. However, an important confounder of an observed relation between flow and metabolism may be the type of symptoms the patient has experienced. MRS and primary and secondary collateral flow patterns in patients with severe obstructive carotid artery disease In patients with a severe unilateral ICA stenosis CEA decreases the risk of embolic stroke by removal of the atheromatous plaque (NASCET, 1991a; The European Carotid Surgery Trialists Collaborative Group, 1998).

Studies on carotid plaque morphology, distal microembolism and infarct location (i.e. large artery strokes) have shown that the carotid plaque itself on the one hand can be considered as a source of thrombo-embolisms (Valton et al., 1998; Molloy et al., 1999; Barnett et al., 2000; Liapis et al., 2001; Tegos et al., 2001). However, it has also been shown that in patients with symptomatic ICA occlusive disease, absence of collateral flow is associated with increased stroke risk (Henderson et al., 2000; Silvestrini et al., 2000; Markus et al., 2001; Vernieri et al., 2001). Generally two types of collateral pathways can be considered: primary and secondary collaterals. The major collateral pathway in the brain is the circle of Willis (primary collaterals), which is able to respond quickly to a low perfusion by providing blood from the posterior circulation via the posterior communicating artery (PComA) or by providing blood from the contralateral hemisphere via the anterior communicating artery (AComA) to the affected hemisphere (Powers, 1991). When the primary collateral pathways fall short, other collateral pathways, such as retrograde flow in the OphtA and/or leptomeningeal collateral flow, are presumed to be recruited. The role of these secondary pathways is still unclear, but their presence is regarded as a sign of a deteriorated hemodynamic condition of the brain (Powers, 1991; Smith et al., 1994; Muller et al., 1996; Mull et al., 1997). An example of recruited collateral flow via the AComA and PComA is shown in Figure 15.10. Van Everdingen et al. (1998) demonstrated that the NAA/Cho ratio in patients with an ICA occlusion having both collateral flow via the AComA and PComA was significantly higher than in patients with collateral flow via the AComA only, the PComA only or no primary collateral flow at all. Moreover, no metabolic differences were found between patients with only AComA and only PComA flow, whereas patients without primary collateral flow demonstrated the lowest NAA/Cho ratios. The authors suggested that the association between the number of primary collaterals and the decrease in the NAA/Cho ratio demonstrated the importance of the circle of Willis to preserve cerebral metabolism. In contrast to the association of primary collaterals and the NAA/Cho ratio in these patients, no association between the presence of secondary collaterals (i.e. reversed flow in the ipsilateral OphtA) and the NAA/Cho ratio was found.

MR spectroscopy in severe obstructive carotid artery disease

ACoA

ICA occlusion

ICA occlusion

PCoA

3D MRA TOF

2D PC

Fig. 15.10 Collateral pathways in the circle of Willis. Upper row shows directional flow images phase encoded in the left–right direction showing cross flow via the AComA. Bottom row shows the directional flow images phase encoded in the anterior– posterior direction, showing collateral flow via the PComA.

Cerebrovascular reactivity and MRS in patients with severe obstructive carotid artery disease As mentioned above, reduced CPP can be compensated by a decrease of the peripheral resistance in the cerebral arterial vasculature. This is achieved by dilatation of cerebral arterioles. Arterioles may demonstrate a range of compensatory vasodilatation. The remaining capacity of vasodilatation in these arterioles can be demonstrated with a vasodilatory stimulus, e.g. CO2 inhalation or intravenous (i.v.) acetazolamide. The change in CBF or flow in brain feeding arteries in response to the vasodilatory stimulus can be measured with transcranial Doppler sonography (TCD), PET or SPECT and is supposed to represent the remaining capacity of vasodilatation. Therefore, absent cerebrovascular reactivity after a vasodilatory stimulus indicates a state of already maximal vasodilatation in the cerebral arterioles. Lythgoe et al. (2001) studied 21 patients with unilateral severe ICA stenosis or ICA occlusion. The majority of their patients had been asymptomatic for at least 2 years. The symptomatic patients had experienced symptoms on average approximately half a year before investigation. Spectroscopic voxels were located in the border zone between the MCAs and ACAs. No cor-

relation was found between the concentration of Cho, Cr or NAA in these voxels and cerebrovascular reactivity in the MCA territory, as measured with TCD. Rutgers et al. (2003) studied 19 patients with a symptomatic ICA occlusion who had symptoms on average 4 months before investigation. They found no correlation between the NAA concentration in the centrum semi-ovale and the cerebrovascular reactivity in the MCA territory measured with TCD. Visser et al. (1999) studied 66 patients with severe ICA stenosis, of whom 10 were asymptomatic (Visser et al., 1999). They found a modest correlation in both hemispheres between cerebrovascular reactivity in the MCA territory, measured with TCD, and the NAA/Cho ratio in the centrum semi-ovale. Mihara et al. (2000) studied the correlation between the Lac/NAA ratio and cerebrovascular reactivity measured with PET in 11 patients with chronic cerebrovascular disease who had experienced cerebral infarction. Strokes had occurred between 50 days and 7 years before investigation. Five patients had an ICA occlusion, two an ICA stenosis and four an MCA occlusion. Spectroscopic voxels were placed in normal appearing brain outside infarcted regions. In this somewhat heterogeneous patient group a negative correlation was found between the Lac/NAA ratio and regional cerebrovascular reactivity. From these studies it can be concluded that no association has been demonstrated between reduced metabolite concentration and impaired cerebrovascular reactivity. However, associations have been demonstrated between cerebrovascular reactivity and NAA/Cho and Lac/NAA ratios. This discrepancy is hard to explain but may be related to differences in patient selection in the various studies. MRS and hemodynamic failure indicated by border-zone infarcts In Stage 3 of hemodynamic failure it is expected that CBF and compensatory mechanisms are unable to respond adequately to the decrease in CPP. In these cases, the CBF is not sufficient to accommodate the metabolic demands, causing cerebral infarcts. The so-called border-zone regions in the brain, located in the most distal part of the perfusion territory of the main cerebral arteries, or between the deep and superficial supply area of the MCA, are the first to suffer ischemic damage when blood flow decreases

241

242

Jeroen van der Grond and Dirk R. Rutgers

(a)

(b)

(c)

Fig. 15.11 Three types of infarct: a territorial infarct (a), an external borderzone infarct located between the flow territory of the ACA and MCA (b) and an internal borderzone infarct located in the deep white matter (WM) (c).

(Challa et al., 1990; Moody et al., 1990). Regions surrounding these infarcts, and also regions with low flow that are not infarcted, may contain cells that still receive sufficient oxygen and glucose to survive, but insufficient to maintain function. On the other hand, patients having territorial infarcts, which are believed to have a thrombo-embolic origin rather than a hemodynamic origin, are likely not suffering from hemodynamic failure. Figure 15.11 shows a territorial infarct (Figure 15.11(a)), an external borderzone infarct located between the flow territory of the ACA and MCA (Figure 15.11(b)) and an internal borderzone infarct located in the deep white matter (WM) (Figure 15.11(c)). It has been shown previously that patients with severe stenosis of the ICA and borderzone infarcts (critical hypoperfusion) demonstrated significantly lower NAA/Cho ratios (van der Grond et al., 1996c) and an increased prevalence of Lac (van der Grond et al., 1999) in the centrum semi-ovale compared with patients having territorial infarcts or those without infarcts (both can be considered having non-critical hypoperfusion). These results suggest that the NAA/Cho ratio is most reduced in the patient category which is likely to have critically reduced cerebral perfusion resulting in anaerobic glycolysis (increased prevalence of Lac).

Clinical relevance The previous paragraphs have shown that several spectroscopic abnormalities can be found in severe obstructive carotid artery disease. These

abnormalities include decrease of the NAA concentration, decrease of the NAA/Cho ratio and increase of Lac. Increase of the Cho/Cr ratio has also been reported; if Cr concentration is assumed to be stable, this suggests an increase of Cho concentration. These spectroscopic changes have been observed in non-infarcted cerebral WM. Cerebral gray matter (GM) has not been investigated previously. Reduced NAA concentration has been demonstrated in patients with symptomatic ICA occlusion (Rutgers et al., 2003). In patients with symptomatic ICA stenosis, the presence of NAA loss has been inferred from decrease of the NAA/Cho ratio (van Everdingen et al., 2000a). Reduction of the NAA concentration is an important observation because the amount of NAA may be related to cognitive function (Valenzuela et al., 2000). It may be suggested that NAA loss causes cognitive impairment in patients with a symptomatic carotid artery obstruction (Bakker et al., 2000). However, in a subgroup of patients with ICA occlusion and TIAs no relation between cognition and NAA/Cr ratio could be demonstrated (Bakker et al., 2003). Cognitive impairment did show a relation with the presence of Lac in these patients. As Lac is a product of anaerobic glycolysis, this may indicate that cognition is related to hemodynamic compromise. Cho increase that may be found in patients with severe obstructive carotid artery disease reflects gliosis, increased phospholipid breakdown or reduced phospholipid biosynthesis (Saunders et al., 1995). The clinical relevance of Cho changes still has to be investigated.

MR spectroscopy in severe obstructive carotid artery disease

Studies on the relation between spectroscopic abnormalities and clinical prognosis are scarce. One prospective study in patients with symptomatic ICA occlusion has shown that a decrease of the NAA/Cho ratio in non-infarcted cerebral WM predicts the occurrence of future cerebral ischemic events (Klijn et al., 2000). If the NAA/Cho ratio is lower than the normal range found in healthy control subjects, the annual risk of recurrent cerebral ischemic events may increase approximately four times. Most likely, the decrease of the NAA/Cho ratio in these patients is caused by a decrease of the NAA concentration (Rutgers et al., 2003) although also an increase in Cho concentration may contribute. In patients with symptomatic ICA stenosis or asymptomatic ICA stenosis or occlusion, no studies have investigated the relation between spectroscopic changes and future cerebral ischemic events. In conclusion, from a clinical point of view 1H MRS may be an important investigation in patients with severe obstructive carotid artery disease because it can demonstrate cerebral damage while MRI shows no abnormalities. Moreover, it can identify patients who are at a risk of future stroke or TIA. Possibly, it may serve as a surrogate marker to monitor the effect of treatment such as surgical revascularization procedures. REFERENCES Bakker FC, Klijn CJ, Jennekens-Schinkel A, Kappelle LJ. 2000. Cognitive disorders in patients with occlusive disease of the carotid artery: a systematic review of the literature. J Neurol 247: 669–676. Bakker FC, Klijn CJ, Jennekens-Schinkel A, van der Twell I, van der Grond J, van Huffelen AC, et al. 2003. Cognitive impairment is related to cerebral lactate in patients with carotid artery occlusion and ipsilateral transient ischemic attacks. Stroke 34(6): 1419–1424. Epub 2003 Apr. 24. Barnett HJM, Gunton RW, Eliasziw M, Fleming L, Sharpe B, Gates P, et al. 2000. Causes and severity of ischemic stroke in patients with internal carotid artery stenosis. J Am Med Assoc 283: 1429–1436. Baumgartner RW, Regard M. 1994. Role of impaired CO2 reactivity in the diagnosis of cerebral low flow infarcts. J Neurol Neurosurg Psychiatr 57: 814–817. Berkelbach van der Sprenkel JW, Luyten PR, van Rijen PC, Tulleken CA, den Hollander JA. 1988. Cerebral lactate detected

by regional proton magnetic resonance spectroscopy in a patient with cerebral infarction. Stroke 19: 1556–1560. Bogousslavsky J, Regli F. 1986a. Borderzone infarctions distal to internal carotid artery occlusion: prognostic implications. Ann Neurol 20: 346–350. Bogousslavsky J, Regli F. 1986b. Unilateral watershed cerebral infarcts. Neurology 36: 373–377. Bruhn H, Frahm J, Gyngell ML, Merboldt KD, Hanicke W, Sauter R. 1989. Cerebral metabolism in man after acute stroke: new observations using localized proton NMR spectroscopy. Magn Reson Med 9: 126–131. Caplan LR, Hennerici M. 1998. Impaired clearance of emboli (washout) is an important link between hypoperfusion, embolism, and ischemic stroke. Arch Neurol 55: 1475–1482. Challa VR, Bell MA, Moody DM. 1990. A combined H & E alkaline phosphatase and high resolution microradiography study of lacunes. Clin Neuropathol (Berl) 9: 196–204. Chopp M, Frinak SF, Walton DR, Smith MB, Welch KM. 1987. Intracellular acidosis during and after cerebral ischemia: in vivo nuclear magnetic resonance study of hyperglycemia in cats. Stroke 18: 919–923. Del Sette M, Eliasziw M, Streifler JY, Hachinski VC, Fox AJ, Barnett HJ. 2000. Internal borderzone infarction: a marker for severe stenosis in patients with symptomatic internal carotid artery disease. For the North American Symptomatic Carotid Endarterectomy (NASCET) Group. Stroke 31: 631–636. Derdeyn CP, Khosla A, Videen TO, Fritsch SM, Carpenter DL, Grubb Jr. RL, et al. 2001. Severe hemodynamic impairment and border zone-region infarction. Radiology 220: 195–201. Derdeyn CP, Videen TO, Fritsch SM, Carpenter DA, Grubb RL, Powers WJ. 1999. Compensatory mechanisms for chronic cerebral hypoperfusion in patients with carotid occlusion. Stroke 30: 1019–1024. Derdeyn CP, Videen TO, Yundt KD, Fritsch SM, Carpenter DA, Grubb RL, et al. 2002. Variability of cerebral blood volume and oxygen extraction: stages of cerebral haemodynamic impairment revisited. Brain 125: 595–607. European Carotid Surgery Trialists Collaborative Group. 1998. Randomised trial of endarterectomy for recently symptomatic carotid stenosis: final results of the MRC European Carotid Surgery Trial (ECST). Lancet 351: 1379–1387. European Carotid Surgery Trialists’ Collaborative Group. 1991. MRC European Carotid Surgery Trial: interim results for symptomatic patients with severe (70–99%) or with mild (0–29%) carotid stenosis. Lancet 337: 1235–1243. Garcia JH, Lassen NA, Weiller C, Sperling B, Nakagawara J. 1996. Ischemic stroke and incomplete infarction. Stroke 27: 761–765. Gideon P, Henriksen O, Sperling B, Christiansen P, Olsen TS, Jorgensen HS, et al. 1992. Early time course of N-acetylaspartate, creatine and phosphocreatine, and compounds

243

244

Jeroen van der Grond and Dirk R. Rutgers

containing choline in the brain after acute stroke. A proton magnetic resonance spectroscopy study. Stroke 23: 1566–1572. Graeber MC, Jordan JE, Mishra SK, Nadeau SE. 1992. Watershed infarction on computed tomographic scan. An unreliable sign of hemodynamic stroke. Arch Neurol 49: 311–313. Graham GD, Kalvach P, Blamire AM, Brass LM, Fayad PB, Prichard JW. 1995. Clinical correlates of proton magnetic resonance spectroscopy findings after acute cerebral infarction. Stroke 26: 225–229. Henderson RD, Eliasziw M, Fox AJ, Rothwell PM, Barnett HJM. 2000. Angiographically defined collateral circulation and risk of stroke in patients with severe carotid artery stenosis. Stroke 31: 128–132. Hennerici M, Daffertshofer M, Jakobs L. 1998. Failure to identify cerebral infarct mechanisms from topography of vascular territory lesions. Am J Neuroradiol 19: 1067–1074. Hirano T, Minematsu K, Hasegawa Y, Tanaka Y, Hayashida K, Yamaguchi T. 1994. Acetazolamide reactivity on 123I-IMP single photon emission computed tomography in patients with major cerebral artery occlusive disease: correlation with positron emission tomography parameters. J Cereb Blood Flow Metab 14: 763–770. Kim GE, Lee JH, Cho YP, Kim ST. 2001. Metabolic changes in the ischemic penumbra after carotid endarterectomy in stroke patients by localized in vivo proton magnetic resonance spectroscopy (1H-MRS). Cardiovasc Surg 9: 345–355. Kim GE, Lee JH, Cho YP. 2002. Can carotid endarterectomy improve metabolic status in patients with asymptomatic internal carotid artery flow lesion? Studies with localized in vivo proton magnetic resonance spectroscopy. J Vasc Surg 36: 559–564. Klijn CJ, Kappelle LJ, Van der Grond J, Algra A, Tulleken CA, van Gijn J. 2000. Magnetic resonance techniques for the identification of patients with symptomatic carotid artery occlusion at high risk of cerebral ischemic events. Stroke 31: 3001–3007. Leblanc R, Yamamoto YL, Tyler JL, Diksic M, Hakim A. 1987. Borderzone ischemia. Ann Neurol 22: 707–713. Liapis CD, Kakisis JD, Kostakis AG. 2001.Carotid stenosis: factors affecting symptomatology. Stroke 32: 2782–2786. Lythgoe D, Simmons A, Pereira A, Cullinane M, Williams S, Markus HS. 2001. Magnetic resonance markers of ischaemia: their correlation with vasodilatory reserve in patients with carotid artery stenosis and occlusion. J Neurol Neurosurg Psychiatr 71: 58–62. Markus H, Cullinane M. 2001. Severely impaired cerebrovascular reactivity predicts stroke and TIA risk in patients with carotid artery stenosis and occlusion. Brain 124: 457–467. Mihara F, Kuwabara Y, Yoshida T, Yoshiura T, Sasaki M, Masuda K, et al. 2000. Correlation between proton magnetic resonance spectroscopic lactate measurements and vascular reactivity in chronic occlusive cerebrovascular disease: a comparison

with positron emission tomography. Magn Reson Imaging 18: 1167–1174. Molloy J, Markus HS. 1999. Asymptomatic embolization predicts stroke and TIA risk in patients with carotid artery stenosis. Stroke 30: 1440–1443. Moody DM, Bell MA, Challa VR. 1990. Features of the cerebral vascular pattern that predict vulnerability to perfusion or oxygenation deficiency: an anatomic study. Am J Neuroradiol 11: 431–439. Mull M, Schwarz M, Thron A. 1997. Cerebral hemispheric lowflow infarcts in arterial occlusive disease. Lesion patterns and angiomorphological conditions. Stroke 28: 118–123. Muller M, Schimrigk K. 1996. Vasomotor reactivity and pattern of collateral blood flow in severe occlusive carotid artery disease. Stroke 27: 296–299. Nemoto EM, Yonas H, Chang Y. 2003. Stages and thresholds of hemodynamic failure. Stroke 34: 2–3. North American Symptomatic Carotid Endarterectomy Trial Collaborators. 1991a. Beneficial effect of carotid endarterectomy in symptomatic patients with high-grade carotid stenosis. New Engl J Med 325: 445–453. North American Symptomatic Carotid Endarterectomy Trial (NASCET) Investigators. 1991b. Clinical alert: benefit of carotid endarterectomy for patients with high-grade stenosis of the internal carotid artery. National Institute of Neurological Disorders and Stroke, Stroke and Trauma Division. Stroke 22: 816–817. Pappata S, Fiorelli M, Rommel T, Hartmann A, Dettmers C, Yamaguchi T, et al. 1993. PET study of changes in local brain hemodynamics and oxygen metabolism after unilateral middle cerebral artery occlusion in baboons. J Cereb Blood Flow Metab 13: 416–424. Powers WJ. 1991. Cerebral hemodynamics in ischemic cerebrovascular disease. Ann Neurol 29: 231–240. Powers WJ, Press GA, Grubb Jr. RL. 1987. The effect of hemodynamically significant carotid artery disease on the hemodynamic status of the cerebral circulation. Ann Intern Med 106: 27–35. Rothwell PM, Gutnikov SA, Warlow CP. 2003. Reanalysis of the final results of the European carotid surgery trial. Stroke 34: 514–523. Rutgers DR, Donders RC, Vriens EM, Kappelle LJ, Van der Grond J. 2000a. A comparison of cerebral hemodynamic parameters between transient monocular blindness patients, transient ischemic attack patients and control subjects. Cerebrovasc Dis 10: 307–314. Rutgers DR, Klijn CJM, Kappelle LJ, van der Grond J. 2000b. Cerebral metabolic changes in patients with a symptomatic occlusion of the internal carotid artery: a longitudinal 1H magnetic resonance spectroscopy study. J Magn Reson Imaging 11: 279–286.

MR spectroscopy in severe obstructive carotid artery disease

Rutgers DR, Klijn CJ, Kappelle LJ, Eikelboom BC, van Huffelen AC, van Der G, et al. 2001. Sustained bilateral hemodynamic benefit of contralateral carotid endarterectomy in patients with symptomatic internal carotid artery occlusion. Stroke 32: 728–734. Rutgers DR, van Osch MJ, Kappelle LJ, Mali WP, van der Grond J. 2003. Cerebral hemodynamics and metabolism in patients with symptomatic occlusion of the internal carotid artery. Stroke 34: 648–652. Saunders DE, Howe FA, van den Boogaart A, McLean MA, Griffiths JR, Brown MM. 1995. Continuing ischemic damage after acute middle cerebral artery infarction in humans demonstrated by short-echo proton spectroscopy. Stroke 26: 1007–1013. Silvestrini M, Vernieri F, Pasqualetti P, Matteis M, Passarelli F, Troisi E, et al. 2000. Impaired cerebral vasoreactivity and risk of stroke in patients with asymptomatic carotid artery stenosis. J Am Med Assoc 283: 2122–2127. Smith HA, Thompson Dobkin J, Yonas H, Flint E. 1994. Correlation of xenon-enhanced computed tomographydefined cerebral blood flow reactivity and collateral flow patterns. Stroke 25: 1784–1787. Szabo K, Kern R, Gass A, Hirsch J, Hennerici M. 2001. Acute stroke patterns in patients with internal carotid artery disease: a diffusion-weighted magnetic resonance imaging study. Stroke 32: 1323–1329. Tegos TJ, Kalomiris KJ, Sabetai MM, Kalodiki E, Nicolaides AN. 2001. Significance of sonographic tissue and surface characteristics of carotid plaques. Am J Neuroradiol 22: 1605–1612. Torvik A. 1984. The pathogenesis of watershed infarcts in the brain. Stroke 15: 221–223. Tsuchida C, Kimura H, Sadato N, Tsuchida T, Tokuriki Y, Yonekura Y. 2000. Evaluation of brain metabolism in steno-occlusive carotid artery disease by proton MR spectroscopy: a correlative study with oxygen metabolism by PET. J Nucl Med 41: 1357–1362. Uno M, Harada M, Nagahiro S. 2001. Quantitative evaluation of cerebral metabolites and cerebral blood flow in patients with carotid stenosis. Neurol Res 23: 573–580. Valenzuela MJ, Sachdev PS, Wen W, Shnier R, Brodaty H, Gillies D. 2000. Dual voxel proton magnetic resonance spectroscopy in the healthy elderly: subcortical–frontal axonal N-acetylaspartate levels are correlated with fluid cognitive abilities independent of structural brain changes. Neuroimage 12: 747–756. Valton L, Larrue V, le Traon AP, Massabuau P, Geraud G. 1998. Microembolic signals and risk of early recurrence in patients with stroke or transient ischemic attack. Stroke 29: 2125–2128. van der Grond J, Balm R, Kappelle LJ, Eikelboom BC, Mali WPTM. 1995. Cerebral metabolism of patients with stenosis or occlusion of the internal carotid artery. A 1H-MR spectroscopic imaging study. Stroke 25: 822–828.

van der Grond J, Balm R, Klijn CJM, Kappelle LJ, Eikelboom BC, Mali WPTM. 1996a. Cerebral metabolism of patients with stenosis of the internal carotid artery before and after endarterectomy. J Cereb Blood Flow Metab 16: 320–326. van der Grond J, Eikelboom BC, Mali WPTM. 1996b. Flowrelated anaerobic metabolic changes in patients with severe stenosis of the internal carotid artery. Stroke 27: 2026–2032. van der Grond J, Ramos LMP, Eikelboom BC, Mali WPTM. 1996c. Cerebral metabolic differences between the severe and critical hypoperfused brain. Neurology 47: 399–404. van der Grond J, van Everdingen KJ, Eikelboom BC, Kenez J, Mali WPThM. 1999. Assessment of borderzone ischemia with a combined MR imaging–MR angiography–MR spectroscopy protocol. J Magn Reson Imaging 9: 1–9. van der Zwan A, Hillen B, Tulleken CAF, Dujovny M, Dragovic L. 1992. Variability of the territories of the major cerebral arteries. J Neurosurg 77: 927–940. van Everdingen KJ, Kappelle LJ, Klijn CJ, Mali WP, Eikelboom BC, Van der Grond J. 2000a. Cerebral ischaemic changes in association with the severity of ICA lesions and cerebropetal flow. Eur J Vasc Endovasc Surg 20: 528–535. van Everdingen KJ, Kappelle LJ, Klijn CJ, Mali WP, van Der Grond. 2000b. Clinical features associated with internal carotid artery occlusion do not correlate with MRA cerebropetal flow measurements. J Neurol Neurosurg Psychiatr 70: 333–339. van Everdingen KJ, Visser GH, Klijn CJM, Kappelle LJ, Van der Grond J. 1998. Role of collateral flow on cerebral hemodynamics in patients with unilateral internal carotid artery occlusion. Ann Neurol 44: 167–176. Vernieri F, Pasqualetti P, Matteis M, Passarelli F, Troisi E, Rossini PM, et al. 2001. Effect of collateral blood flow and cerebral vasomotor reactivity on the outcome of carotid artery occlusion. Stroke 32: 1552–1558. Visser GH, van der Grond J, van Huffelen AC, Wieneke GH, Eikelboom BC. 1999. Decreased transcranial Doppler carbon dioxide reactivity is associated with disordered cerebral metabolism in patients with internal carotid artery stenosis. J Vasc Surg 30(2): 252–260. Wagner KR, Kleinholz M, de Courten Myers GM, Meyers RE. 1992. Hyperglycemic versus normoglycemic stroke: topography of brain metabolites, intracellular pH, and infarct size. J Cereb Blood Flow Metab 12: 213–222. Yamauchi H, Fukuyama H, Fujimoto N, Nabatame H, Kimura J. 1992. Significance of low perfusion with increased oxygen extraction fraction in a case of internal carotid artery stenosis. Stroke 23: 431–432. Yamauchi H, Fukuyama H, Nagahama Y, Nabatame H, Nakamura K, Yamamoto YL, et al. 1996. Evidence for misery perfusion and risk for recurrent stroke in major cerebral arterial occlusive diseases from PET. J Neurol Neurosurg Psychiatr 61: 18–25.

245

16

Perfusion and diffusion imaging in chronic carotid disease Iain D. Wilkinson Academic Radiology, University of Sheffield and Hon Consultant Clinical Scientist, Sheffield Teaching Hospitals NHS Trust, Sheffield, UK

Key points • Atherosclerotic disease of the internal carotid arteries is common. It is often asymptomatic because of compensatory mechanisms (cerebral vasodilation and provision of collateral arterial supply). • Patients with carotid disease may present with stroke, transient ischemic attacks or amaurosis fugax. • MR perfusion images (mean transit times or time-to-peak) showing greater than 3.5–4 s increase suggest severe hemodynamic impairment and risk of ischemia. • Inadequate cerebrovascular reserve (CVR) is also a risk factor for hemodynamic injury; a pharmacological challenge (e.g. acetazolamide) in conjunction with MR perfusion can measure CVR. • Diffusion MR imaging (MRI) can show characteristic patterns of embolic damage due to either carotid or cardiac sources. • Diffusion, perfusion and contrast-enhanced MRI (particular gadolinium-enhanced fluid attenuated inversion recovery) can be used to monitor both early and long-term effects of carotid surgical or endoluminal intervention (hyperemia, leptomeningeal enhancement, peri-procedural ischemic events and perfusion normalization). • Perfusion MRI can also be helpful in the evaluation of various other vascular diseases such as moyamoya, Takayasu arteritis, vascular malformations, aneurysms, etc. 246

Diseases affecting the arterial supply to and distribution within the brain often interfere with central nervous system (CNS) metabolism. Such interference, which frequently results from internal carotid artery (ICA) pathology, can lead to the onset of clinical symptoms, signaling the need for investigation of the brain via an appropriate imaging modality. Imaging has historically been used to exclude the presence of pathology such as hemorrhage, neoplasm or infection and latterly to investigate hemodynamic cause and status. The nature of metabolic change and the onset of associated symptoms can be broadly classified as being either acute or chronic. In the acute case (“brain attack”), an abrupt alteration in brain function due to changes in vascular supply, which is nontransient, is termed stroke. The ability of a health-care system to respond quickly and effectively to the presentation of acute stroke may well depend heavily on the provision of imaging technology: the applications of diffusion and perfusion MR imaging (MRI) in this context are detailed elsewhere (Chapter 12). The current chapter concentrates on the utility of perfusion and diffusion in the context of chronic ICA disease. The majority of carotid disease is atherosclerotic and so emphasis will be placed on this pathology.

Alterations to the brain’s vascular supply, hemodynamic failure and the search for clinically relevant physiological indicators Patients with chronic cerebrovascular blood supply deficits often present with a history of symptoms such as those associated with transient ischemic attacks (TIAs), amaurosis fugax or having experienced an

Perfusion and diffusion imaging in chronic carotid disease

episode of minor or non-disabling stroke. Sometimes they have experienced more than one such ischemic event. They may be at high risk of having a significant disabling stroke and although previous symptoms may have had sudden onset, for the purposes of this chapter, their speedy resolution classifies their overall disease state as being chronic. In the clinical setting, the overall aim of imaging in this patient group is to aid the prevention of recurrent ischemia thereby reducing the risk of disabling stroke. It is apparent that, where blood supply to and within the brain is abnormal or sub-optimal, metabolism and hence overall parenchymal function may not always be deleteriously affected. The brain can cope with alterations in the efficiency of oxyhemoglobin transport. With regard to short-term fluctuations, provision of this ability may be linked with the apparent “overkill” in the brain’s hemodynamic reaction to individual tasks: neuroscience has yet to explain why the blood oxygen level dependent (BOLD) response to a set of neuronal events is so dramatically large compared to the amount of oxygen actually required. With regard to long-term fluctuations, it is not uncommon for occlusion of one (or more rarely both) ICA to develop without neurological symptoms, especially if the disease process is slow. In this situation the brain appears to cope with changes to the dynamic range of its arterial input. Both the short- and long-term cerebral metabolic “buffers” suggest that it is only in relatively extreme cases of perfusion deficit that patients become symptomatic. However, the shear abundance of chronic as well as acute cerebrovascular disease burden indicates that the ability of the cerebral vessels to sustain flow and respond to increased flow demand is nonetheless vital. In describing the hemodynamic status of the brain, several parameters are often used. On the vascular supply side, arterial cerebral blood volume (CBV) plus knowledge of blood transit time (time to pass through the vascular bed) are used to indicate cerebral blood flow (CBF). These can be modulated by the cerebrovascular reserve (CVR) which enables autoregulated localized vasodilation. The relative amount of oxygen extracted from the blood is termed the oxygen extraction fraction (OEF) and the amount of oxygen used in cellular respiration is the cerebral metabolic rate of oxygen (CMRO2). One of the models of hemodynamic failure

involves three stages (Powers, 1991) which occur sequentially as the arterial input function (AIF) (or cerebral perfusion pressure (CPP)) decreases: stage I when vasodilation leads to an increase in CBV (maintaining constant CBF, OEF and, most importantly, CMRO2); stage II when CBF starts to fall and OEF increases in an attempt to keep CMRO2 constant and finally stage III when CBF drops further, leading to a decrease in CBV, OEF and CMRO2. Continued energy demand then leads to anaerobic metabolism and ischemia. The length of time for which ischemic tissue is in stage III failure influences whether cell death and permanent loss of brain parenchyma ensues. This three-stage model has been partially verified by in vivo estimates of CBF, CBV, OEF and CMRO2, the majority of which have been made using positron emission tomography (PET). However, the interplay between the parameters appears to be highly complex and other factors (hematocrit level (Yamauchi et al., 1998), variations in individual vascular anatomy, etc.) are likely to require incorporation into any clinically realistic prognostic model relating to the consequences of hemodynamic failure, particularly on an individual patient basis. In addition to PET, imaging modalities such as dynamic contrastenhanced computed tomography (CT), xenon-CT (Xe-CT) and single photon emission computed tomography (SPECT) can indicate cerebral perfusion status. Each of these techniques has certain drawbacks to their entering widespread clinical use in cerebrovascular disease, such as limited availability, anatomical coverage and resolution, radiopharmaceutical hyperfixation (Sperling and Lassen, 1993) and the use of ionizing radiation. MR perfusion imaging is a technique which enables certain aspects of the hemodynamic status of the brain to be determined, its availability is ever-increasing and it provides good anatomical coverage/spatial resolution. The lack of ionizing radiation burden makes it amenable to use in follow-up studies of chronic disease.

Cellular effects of ischemia: the imaging of water diffusivity The ability of MR diffusion imaging to differentiate between cytotoxic and vasogenic edema is of great

247

248

Iain D. Wilkinson

(a)

(b)

(c)

(d)

Fig. 16.1 Different stages of infarction on (a) T2-weighted, (b) fluid attenuated inversion recovery (FLAIR), (c) diffusion-weighted imaging (DWI) and (d) calculated apparent diffusion coefficient (ADC) maps. Areas of chronic infarction, predominantly within the left hemisphere (green arrows), show different characteristics than the small areas of acute infarction within the right frontal lobe (orange arrows).

importance in the context of chronic ICA disease. New infarcts show change (hypointensity on apparent diffusion coefficient (ADC) maps or hyperintensity on diffusion-weighted imaging (DWI)) relative to non-infarcted tissue whilst areas of old infarction (several days, extending to weeks after onset of ischemia) appear isointense with surrounding non-infarcted parenchyma (Warach et al., 1995). In chronic ischemia, the presence of old infarcts can be inferred from hyperintensity on T2-weighted or fluid attenuated inversion recovery (FLAIR) imaging concomitant with normal ADC (Figure 16.1). Additional information can be gained from standard imaging: the lack of apparent mass effect (localized swelling) or changes in blood–brain barrier (BBB) permeability (damaged capillary endothelial lining) demonstrated by enhancement on T1-weighted images, sometimes observed within the first week or two following infarction; or from the presence of chronic calcification or blood breakdown products, detectable on T*-weighted images. Diffusion imaging is the most 2 sensitive technique available for the depiction of cytotoxic edema, end-stage damage and reversible ischemic insult (cf. below). It is thought that it can provide an indicator of the time scale over which endstage infarction occurs (Burdette et al., 1998). In cerebrovascular disease, the clinician may wish to assess gross anatomy, blood–brain barrier (BBB) permeability, macrovascular topology and function, microvascular flow (perfusion), cytoxic swelling, CVR, neuronal activation and axonal tract connectivity. MR can provide information regarding all of these, potentially during a single imaging episode.

Imaging techniques on offer The mapping of both parenchymal perfusion and diffusion can of course be performed via traditional or by echo-planar encoding techniques. The latter has been the most commonly implemented technique for both perfusion and diffusion imaging of patients with severe carotid disease. For perfusion assessment, the ability to cover all of the desired vascular territories at well-defined time intervals with sufficient temporal resolution to adequately sample the perfusion curve is a necessity. Practically all published in vivo studies have used the exogenous contrast or dynamic susceptibility mapping method. It is worth noting that a recent study investigating the effects of injecting two boluses of gadolinium (Gd) chelate concluded that two perfusion assessments can be performed during the same patient imaging episode (Pandya et al., 2003). This has particular importance in the context of the assessment of CVR via, say, a pharmacological challenge where perfusion assessments can be made before and after instigation of vascular stress using exogenous contrast technique (Nighoghossian et al., 1997). For consistency, the use of a power injector is advisable when performing intravenous contrast-based studies. Such studies are invasive and the development and further clinical implementation of endogenous contrast or arterial spinlabeling (ASL) methods are highly relevant to studies in the current context where repeat perfusion assessment and a move towards quantitative measurement of CBF and CBV are desirable.

Perfusion and diffusion imaging in chronic carotid disease

As detailed elsewhere in this book, weighting an image’s contrast to reflect water diffusivity can be achieved using a number of “conventional” nonecho planar imaging (EPI) image acquisition techniques, for example via spin-echo (Le Bihan et al., 1986) or turbo-stimulated echo acquisition mode (STEAM) sequences (Merboldt et al., 1992). However, the presence of pulsatile physiological or involuntary bulk motion of the cranium, brain parenchyma or cerebrospinal fluid (CSF) can lead to unwanted degradation in image quality. The various techniques, electrocardiogram (ECG) gating or use of navigator echoes (Ordidge et al., 1994) that have been implemented to reduce the influence of bulk motion, carry with them some time penalties. For this reason and as the technology is available on most clinical systems (with the exception of some very low field strength designs) EPI-based diffusion imaging is most common. As with diffusion imaging in all clinical situations, the user should be wary of diffusion anisotropy during image interpretation, remembering that it is easier for water to diffuse along white matter (WM) tracts than at right angles to them. Ensure that data is obtained with diffusion-encoding gradients placed along at least the three-encoding directions. As with diffusion imaging in other pathologies, increased accuracy in absolute quantification of ADC, determination of fractional anisotropy (FA) and visualization of connective pathways (or, more correctly, pathways of maximum water diffusivity) via diffusion tractography may yield useful information regarding the parenchymal sequelae of carotid disease. However, at present these necessitate an increase in the number of diffusion-encoding directions and hence imaging time.

Carotid artery disease: types, MR findings and clinical utility Atherosclerosis Background and methods of impairment Atherosclerosis of the carotid arteries is a leading cause of ischemic brain injury in the western world. It is often systemic in nature, and also influences cardiac function when significant coronary artery

disease is present. Its prevalence is linked to various risk factors including tobacco use, hypertension, hypercholesterolemia, hyperinsulinemia (the metabolic syndrome) and socioeconomic status. Disease mechanisms involve the deposition of plaque, most commonly at the origin of the ICA. The build-up of plaque leads to vessel stenosis (or occlusion). In such cases there is an increased risk of cerebral damage due to: (i) reduced flow causing hypoperfusion in territory supplied by the diseased vessel, (ii) the shedding of emboli from the plaque causing embolization at a site distal to the stenosis or (iii) a combination of the two. It is thought that in the majority of cases the threat is embolic in nature rather than hemodynamic. However, tissue which is already hemodynamically compromised may be at greater risk of damage from plaque embolization (Caplan and Hennerici, 1998). The ICAs supply the retina (via the ophthalmic artery (OphtA)), the parietal, frontal and superior temporal lobes (via the middle cerebral artery (MCA)) plus the anterior and superior medial frontal lobe (via the anterior cerebral artery (ACA)). A list of typical symptoms that result from ischemic damage to these areas can be found elsewhere (Jäger and Saunder, 2001). Where chronic sub-optimal brain perfusion is suspected, any underlying vascular disease process needs to be investigated. Clinical workup typically includes ultrasound investigation of the carotid bifurcation which, if a severe stenosis is detected, is followed by further vascular assessment either by MR angiography (MRA) (Figure 16.2) or invasive conventional catheter angiography. If intervention is being considered (either surgical or endovascular), a full examination of the vascular supply from the aortic arch to the circle of Willis is performed. This is to assess the presence of any tandem disease that may influence patient management (such as significant narrowing of the aortic arch vessels that may prevent endoluminal access or significant intracranial involvement that may hamper surgical procedures). Collateral macrovascular flow One of the mechanisms responsible for the provision of sustained CBF to parenchyma distal to a significant stenosis or occlusion of a major vessel is the provision of alternative sources of arterial supply via

249

250

Iain D. Wilkinson

Right

Left

Fig. 16.2 Time-of-flight MR angiograms of both carotid bifurcations of a patient who presented with TIA. Some flow is shown distal to the tight stenosis of the left ICA indicating that the vessel is not occluded.

(a)

pre

(b)

KL

collateral pathways. In terms of redistribution to MCA territory, the primary collateral supply source is from the MCA of the opposite hemisphere via the circle of Willis. It is likely our understanding of collateral flow will be enhanced by the correlation of data regarding macro- and micro-hemodynamic flow (Liebeskind, 2003). One study that used MRA to depict the vascular anatomy of the circle of Willis reported that a high number of patients with ICA stenosis or occlusion who had minor neurological deficits were found to have a complete circle of Willis (Hartkamp et al., 1999) when compared to normal controls. The authors concluded that the anatomic and functional configuration reflects the degree of ICA obstruction. This may imply that a need for collateral flow results in changes to the apparent vascular anatomy of the circle of Willis. In addition, they identified that unilateral occlusion showed increased anterior whilst bilateral occlusion showed increased posterior collateral flow. Hypoplasia or aplasia of the relevant communicating segment(s) (perhaps due to congenital variability) can lead to insufficient primary collateral flow (Figure 16.3) in which case secondary sources such as the leptomeningeal and/or ophthalmic vessels may be recruited. The primary role of the circle of Willis in collateral provision is highlighted by the finding that poor cerebral hemodynamic status (as measured by CO2 reactivity) is found when collateral flow increases via the leptomeningeal or ophthalmic (c)

post

KL

Fig. 16.3 The functional anatomy of the circle of Willis can influence the hemodynamic consequences of intervention: perfusion timing maps in a patient with 90% and 70% ICA stenoses on their symptomatic and asymptomatic sides, respectively, indicate (a) longer transit time (0.9 s) in the symptomatic hemisphere prior to stenting. An incomplete circle of Willis (no anterior communicating artery (AComA) flow observed on TOF MRA (b)) in combination with contralateral disease leads to (c) timing-asymmetry reversal rather than complete resolution ( 1.0 s) 1 month after intervention.

Perfusion and diffusion imaging in chronic carotid disease

vessels in addition to the circle of Willis (Hofmeijer et al., 2002). The direct assessment of secondary, leptomeningeal collateral flow via delayed contrast filling on MR perfusion has recently been reported (Hermier et al., 2003). Perfusion: parameters and patterns of microvascular compromise Most published work in this area reports perfusion timing in terms of differences between bolus arrival time (BAT), time-to-peak (TTP) or mean transit times (MTTs – for mathematical purists this is often calculated as the first moment (TTFM) of the inverted concentration–time curve). The differences are calculated between the asymptomatic and symptomatic hemispheres (or contralateral and ipsilateral to the side with the most severe stenosis). The idea of comparing territories between the two hemispheres fits rather well in the context of carotid disease, thanks to the bilateral nature of normal cerebral arterial input. Different centers tend to use different timing indices, or more than one. One comparative study reported no differences in appearance between TTP and MTT in a group of 11 patients with unilateral occlusion or high-grade stenosis (Teng et al., 2001). Timing differences greater than 3.5–4 s seem to imply severe hemodynamic impairment/risk of failure (Nasel et al., 2001; Kajimoto et al., 2003). Relative blood volumes or interhemispheric relative CBV (rCBV) ratios are also usually reported. The timing and volume are related to each other by the central volume principle: CBF  CBV/MTT (for a review of these concepts cf. Chapter 7 and Griffiths et al. (2001)). In general, patterns have more often been identified in transit time rather than CBV, but this is not always the case. Parenchymal hypoperfusion due to severe stenosis or occlusion at or near the level of the carotid bifurcation tends to occur at the distal boundaries between supply territories (Figure 16.4) where perfusion pressure would be expected to be least. Infarcts which develop in these areas are termed border-zone infarcts. Abnormal anterior and posterior borderzone perfusion timing concomitant with normal central vascular territory timing has been demonstrated in severe stenotic and/or occlusive disease (Nasel et al., 2001). A knowledge of such border-zone

Fig. 16.4 Relative hypoperfusion can often be observed at the border zones between different supply territories. High signal corresponding to long transit time can be seen between MCA– posterior cerebral artery (PCA) and MCA–ACA territories at the four axial levels shown in this patient with type-2 diabetes. An infarct is also depicted within the right posterior WM.

perfusion insufficiency may be of use in assessing the risk of perfusion-related ischemic stroke secondary to hypotensive events (such as those that may result from general anesthesia). The majority of studies have found that severe unilateral ICA stenosis or occlusion leads to interhemispheric asymmetry in MTT within MCA territory (Nighoghossian et al., 1996; Gillard et al., 1999; Maeda et al., 1999; Wilkinson et al., 2003) (Figure 16.5). This is not always the case (Bozao et al., 2002), perhaps reflecting differing cohort characteristics, most notably the degrees of stenosis in both symptomatic and asymptomatic supply arteries. One study reported that below 80% stenosis there was no significant alteration in transit time, but above this cut-off, there was (Doerfler et al., 2001), which is not

251

252

Iain D. Wilkinson

(a)

Pre-stress

(b)

Post-stress

Fig. 16.5 Perfusion timing maps (first-moment MTT or TTFM) acquired (a) before and (b) after vascular stress testing in a patient with an occluded left ICA demonstrate impaired CVR. The relative lack of vasodilation within the left MCA territory leads to an increase in interhemispheric asymmetry from 2.6 to 12.1 s following administration of acetazolamide.

too heterogeneous from cut-off between moderate and severe used in part to warrant intervention (NASCET, 1991). Interhemispheric asymmetry is generally not seen within ACA and posterior cerebral artery (PCA) territories, as they are not directly supplied by the ICAs. There is greater disparity between CBV results: some studies report similar significant differences in interhemispheric MCA territory blood volumes (Kluytmans et al., 1998; Lythgoe et al., 2000; Doerfler, 2001) whilst others do not (Nighoghossian et al., 1996; Gillard et al., 1999; Maeda et al., 1999; Wilkinson et al., 2003). Quite often the standard deviations in measured rCBV within the groups are large. Further increased variation in rCBV within ACA territory (Wilkinson et al., 2003), may reflect the role that this artery may play in primary collateral redistribution when circle of Willis anatomy permits. Vascular reserve In the presence of very slow flow past the carotid bifurcation plus the absence of, or poor, collateral flow, the brain possesses another mechanism that may avert hypoxic damage: that of a capacity for vasodilation or CVR. Vasodilatory capability and capacity can be assessed by stress testing the vasculature via administration of a carbonic anhydrase inhibitor such as acetazolamide. Studies using Xe-CT or transcranial Doppler (TCD) have shown the

importance of CVR as, when compromised in the presence of low flow, a high risk of stroke may ensue (Yonas et al., 1993; Markus and Cullinane, 2001). Another study which uses TCD suggests that there is a link between the risk of ipsilateral ischemic events and impaired CVR in asymptomatic severe ICA stenosis (Silvestrini et al., 2000). Severely compromised CVR has been documented using exogenous contrastbased MR perfusion and acetazolamide stress in a group of symptomatic patients with occlusive carotid disease (Guckel et al., 1996). Continuous arterial spin labeling (CASL) perfusion methodology can also be used to assess CVR in patients with ICA stenosis (Detre et al., 1999). Recent work demonstrates how MR can facilitate an integrated examination, assessing vascular anatomy from the carotid bifurcation to the circle of Willis, standard imaging and perfusion via the common exogenous contrast method before and after acetazolamide challenge (Griffiths et al., 2003) (Figure 16.5). A multi-modal MR examination enables the “one-stop” assessment of each of the three areas thought to be major risk factors for hemodynamic failure: decrease in arterial input to the brain (degree of bifurcation stenosis), provision of primary collateral supply (circle of Willis functional anatomy) and CVR (perfusion change on stress). Diffusion and ischemic damage The sensitivity of diffusion imaging in stroke is not 100% and is temporally variant. There are cases where no lesion can be observed (Lovblad et al., 1998), or are undetected in the hyperacute state but detected on follow-up (Lefkowitz et al., 1999). A volume of infarcted tissue, depicted by low ADC, may be associated with an initial clinical neurological deficit that resolves completely in the short term (Fiehler et al., 2002) (i.e. normal function returns within 24 h) and is thus classified as TIA rather than stroke. These observations are perhaps unsurprising given the complex nature of hemodynamic failure, the possible existence of collateral neuronal/functional pathways and that stroke is a clinical syndrome; moreover, the differential classification of stroke and chronic carotid disease is partly based on a qualitative definition of neurological function. The anatomical location and distribution of infarcts which result from carotid disease, be they old

Perfusion and diffusion imaging in chronic carotid disease

(detected on T2-weighted/FLAIR imaging) or recent (detected on diffusion imaging), can be quite complex. However, certain patterns appear to be present which may correlate with the degree of stenosis when large patient numbers are considered. These patterns have previously been split into five categories (Szabo et al., 2001): • Pattern 1: Large territorial ischemia including cortex (partial MCA infarction due to distal MCA branch occlusion, large MCA infarction if occlusion has occurred at the source of the MCA and no suitable collateral supply is available or complete MCA and ACA territory infarction resulting from a distal ICA embolism). • Pattern 2: Subcortical lesion with or without additional smaller lesion (occlusion of the MCA with collateral flow). • Pattern 3: Large territorial ischemia including cortex plus additional smaller lesions (as in pattern 1 with the addition of the effects of fragments from the embolus). • Pattern 4: Several disseminated small lesions in distal MCA territory due to multiple small emboli or fragments of a large embolus. • Pattern 5: Small lesions in the border-zone areas (between the ACA and MCA, PCA and MCA or the deep and superficial arterial systems). If diffusion imaging is performed several weeks after onset of symptoms, a substantial number of patients with minor stroke but few with TIA will demonstrate clinically appropriate lesions (Schulz et al., 2003). Close to an event (produced by intervention, e.g. cf. below), clinically silent lesions can be demonstrated. The degree of diffusion abnormality seems to depend upon severity of event and event–scan interval. Diffusion imaging may have a role during the work-up to intervention in carotid disease, yielding patterns of ischemic change consistent with suitability for intervention as opposed to patterns resulting from, say, cardiac emboli (Kastrup et al., 2002). Interventional techniques in carotid stenosis Treatment for patients with atherosclerotic disease of the proximal ICA can be medical (control of blood pressure, reduction in serum cholesterol and administration of antiplatelet agents), surgical (carotid endarterectomy (CEA) (NASCET, 1991; ECST, 1998))

or endovascular (percutanious transluminal angioplasty (PTA) (CAVATAS, 2001) and/or stenting (Yadav et al., 1997)). Surgical intervention has been shown to be beneficial and superior to best medical therapy in the treatment of symptomatic patients (with severe bifurcation stenoses 70% as defined by NASCET criteria (NASCET, 1991)). In asymptomatic carotid stenosis, evidence gathered to date implies that intervention is at best barely significant. However, there are nonetheless some patients with moderate carotid artery disease who have symptoms which develop over the short term and may thus benefit from direct intervention. The “one-stop” MR assessment approach may help to identify this potential subgroup in the future, although this is not possible at present given our current level of understanding of cerebral hemodynamics and vascular risk. Altered hemodynamics and carotid intervention Knowledge of MR perfusion and diffusion characteristics before and after treatment may further our understanding of the hemodynamic sequelae of revascularization. It may also provide information regarding the relative hemodynamic efficacy of the different techniques available. There have been a variety of published studies relating to MR perfusion assessment performed before and after CEA (Gillard et al., 1998, 1999; Kluytmans et al., 1998; Doerfler et al., 2001; Wilkinson et al., 2001; Soinne et al., 2003) (Figure 16.6) and PTA/stent insertion (Macdonald et al., 2002b; Wilkinson et al., 2003) (Figure 16.7). The timing of these MR assessments varied from less than 3 h to 12 months following intervention. On the whole, normalization in perfusion followed CEA, although the time taken varied from within hours to 6–12 months. Similarly, data obtained before and after carotid stenting demonstrated a significant reduction in mean TTFM interhemispheric asymmetry (48% and 61% at the two MCA levels studied) but within 3 h of stent placement (Wilkinson et al., 2003). Although not encountered during the studies which have looked at acute (3 h) effects, cerebral perfusion characteristics plus standard imaging following intervention may provide information regarding the nature of or perhaps a prognostic marker for the hyperemia syndrome (Powers and Smith, 1990; Gillard et al.,

253

254

Iain D. Wilkinson

(a) Pre

R

TTP

rCBV

Inv. Intensity vs. Time 55

ROI Area Peak 1. 292.3 38.1 2. 358.2 35.6

0 0.00 13.2 26.5 39.7 52.9 66.2 79.4 92.6 s

(b) Post

Inv. Intensity vs. Time 56

ROI Area Peak 1. 219.9 32.2 2. 258.2 32.2

0 0.00 13.2 26.5 39.7 52.9 66.2 79.4 92.6 s

Fig. 16.6 Timing maps, relative blood volume maps and concentration–time curves in a patient with a 90% stenosis of the left ICA obtained (a) before and (b) 2.5 h after surgical endarterectomy. The initial interhemispheric MCA territory timing asymmetry resolves soon after the procedure.

1999). Symptoms are usually non-specific whereas imaging shows increased perfusion and diffuse hypointensity on T1-weighted plus hyperintensity on T2-weighted imaging. These appearances are typical of edema located within the MCA and ACA territories ipsilateral to the treated side and tend to resolve in line with clinical symptoms following anti-hypertensive medication (Schwartz, 2002). One interesting acute consequence of intervention is the reported demonstration of unilateral leptomeningeal enhancement on FLAIR imaging caused by collection of Gd during or shortly after both the PTA/stenting and CEA procedure (Wilkinson et al., 2000, 2001) (Figure 16.8). This phenomenon, due to its anatomical distribution within symptomatic MCA territory is thought to result from sudden changes in arterial inflow allowed by intervention. The underlying mechanisms remain

unclear – perhaps it results from leaky BBB junctions in previously hypoperfused areas, in which case it may be a sub-clinical precursor to hyperemia, or is a marker for leptomeningeal collateral involvement or results from disruptions to the mechanism that underpins vasodilation. Intervention aims to restore not only flow but also impaired CVR. MR perfusion and acetazolamide have been used to study the effects of intravascular stenting and CEA on CVR. Normalization in the provision of CVR can be demonstrated (Figure 16.9). However, a recent study in a group of patients with high-grade unilateral stenosis (Wiart et al., 2000) demonstrated no significant differences between ipsi- and contralateral border-zone reserve before intervention and no changes in differences in reserve following intervention, probably indicating

Perfusion and diffusion imaging in chronic carotid disease

(a) Pre

R

rCBV

TTP

Inv. Intensity vs. Time 44

255

ROI Area Peak 1. 260.5 34.3 2. 224.2 32.7

0 0.00 12.8 25.6 38.3 51.1 63.9 76.7 89.5 s Inv. Intensity vs. Time

(b) Post

65

ROI Area Peak 1. 415.1 35.2 2. 381.0 35.1

0 0.00 12.8 25.6 38.3 51.1 63.9 76.7 89.5 s

Fig. 16.7 Conventional catheter angiograms demonstrate increased flow at the level of the bifurcation following PTA and the deployment of an intravascular stent. The initial interhemispheric asymmetry in TTP reflects the 90% ICA stenosis and this asymmetry is seen to resolve following intervention. No such asymmetry or changes in rCBV can be seen.

that the patient population enjoyed good collateral supply and thus did not suffer from significantly hypoperfused parenchyma or deficiency in CVR. Ischemia and carotid intervention As is apparent, the interventional procedures outlined above are not without risk: one of these being peri-procedural clinical events due to disturbance of plaque and associated release of embolic material to the distal circulation. In addition to symptomatic procedure-related embolization, silent ischemic events can be detected on diffusion imaging. These have been noted following PTA, PTA plus stenting (Jaeger et al., 2001) and CEA. There is quite a range in their incidence following CEA: not detected (Forbes et al., 2001), detecting low incidence (4%) (Barth et al., 2000; Feiwell et al., 2001); or high incidence – 31% (Tomczak et al., 2001) or 75% (Muller et al., 2000). These discrepancies may reflect varying interventional technique, sample characteristics or imaging technique. If silent events do occur with

(a)

Pre-stent

(b)

Post-stent

Fig. 16.8 Acute enhancement of the leptomeninges (pia and/or arrachnoid mater) observed on FLAIR images following carotid stenting for a right-sided severe (90%) ICA stenosis. The image obtained (a) before intervention was acquired after the administration of contrast, whilst the image obtained (b) after intervention was acquired before a second injection of contrast, implying that the collection of Gd-chelate occurred during or shortly after (3 h) stent placement.

Iain D. Wilkinson

Pre-stress

Post-stress (b)

Pre-stent

(a)

(c)

(d)

Post-stent

256

Fig. 16.9 Changes in CVR following intervention can be inferred from the perfusion timing (TTP) maps of this patient. A severe stenosis of the left ICA resulted in longer TTP within the left MCA territory (a). This timing asymmetry was enhanced following acetazolamide (b), indicating a lack of vasodilatory response on the symptomatic side. After the stenting procedure, the timing asymmetry resolved (c) and remained symmetrical following further stress testing (d), indicating improved CVR.

greater frequency than clinical events, the use of imaging as a surrogate marker may reduce the numbers of patients required to participate in riskreduction studies. Such studies are needed when new interventional procedures or safety devices are developed (Macdonald and Gaines, 2003). Examples of devices being assessed for use in the angioplasty/stenting procedure are an intravascular filter system (Macdonald et al., 2002a) and a balloon protection catheter (Terada et al., 2003). The filter, which is shaped rather like an umbrella when deployed, is placed distal to the plaque and opened prior to initial vessel ballooning. The purpose of these devices is to reduce procedure-related embolization by capturing any plaque fragments before they can enter the intracranial vascular system (Figure 16.10). Diffusion imaging is being used as part of the

assessment of overall efficacy. One group has reported that a quarter of their neuroprotected cohort demonstrate silent ischemic lesions, suggesting that manipulation of endoluminal equipment within the supra-aortic vessels in itself introduces a major embolic risk factor (Schluter et al., 2003). Influence of cardiac output and heart disease Fluctuations in cardiac output can influence the assessment of cerebral perfusion and, to a first approximation, these effects can be mitigated when evaluation is made using interhemispheric ratios. Systemic hypoperfusion and hence a generalized reduction in CPP often results from heart disease and this may influence the susceptibility of cerebral parenchyma to embolic events. In addition, atrial fibrillation and myocardial infarction can lead to the release of emboli. There is a need to differentiate between the effects of cardiac and ICA emboli. On diffusion, FLAIR or T2-weighted imaging (depending on the event to imaging time interval, cf. above), cardiac emboli frequently yield a pattern of thromboembolic ischemic lesions in both hemispheres and both the anterior and posterior circulation territories which can be used as a marker of probable embolic source (Figure 16.11). Although beyond the remit of this chapter, it is worth mentioning that diffusion and perhaps perfusion imaging may be useful for monitoring the cerebral effects of coronary artery bypass grafting (Wityk et al., 2001; Restrepo et al., 2002). Moyamoya Moyamoya is a chronic cerebrovascular occlusive disease, characterized by progressive vascular occlusion at the peripheral ICA and the development of abnormal collateral circulation at the cerebral basal region. This collateral supply via a myriad of small vessels gives rise to the distinctive “puff of smoke” blush sometimes observed on conventional angiography between the stenotic ICA and relatively normal, re-appearing, distal vasculature. Such a blush can even be observed on time-of-flight magnetic resonance angiography at 3 T (Figure 16.12). Although abnormal thrombogenesis, inflammation and autoimmune processes may be involved in the etiology, its genetic pathogenesis is uncertain. The perfusion

Perfusion and diffusion imaging in chronic carotid disease

DWI

Emboli ADC

Fig. 16.10 The presence of embolic material captured in this intravascular umbrella-shaped filter during stenting highlights one of the potential complications of direct intervention in atherosclerotic ICA disease. DWI and coefficient maps (ADC) obtained 24 h after intravascular stenting of a patient who was being treated for severe right-sided stenosis highlight the occurrence of a clinically silent ischemic event (courtessy of Dr. S. Macdonald). The procedure was undertaken without placement of a protective filter.

(a) DWI

(b) DWI

TTP

Fig. 16.11 Differences in distribution of embolic stroke depicted on DWI from (a) a cardiac source and (b) an ICA source. In (a) multiple hyperintense lesions are present in both hemispheres, while in (b) lesions are only present in the left hemisphere. (b) also includes an MR perfusion image (TTP) showing a characteristic pattern of increased TTP (corresponding to hypoperfusion) of the entire left ICA territory. Figure courtesy of Dr. Peter Barker, Johns Hopkins University, School of Medicine, Baltimore, Maryland, USA.

257

258

Iain D. Wilkinson

Fig. 16.12 Time-of-flight angiographic and T2-weighted imaging appearances of moyamoya disease obtained at 3 T. The “puff” of contrast within small collaterals can be seen providing blood transport from the occluded right MCA to distal vasculature.

appearances can also be characteristic (Wityk et al., 2002). When endogenous collateralization fails to provide sufficient arterial supply, the main treatment is surgical encephaloduroarteriosynangiosis (EDAS). The effects of the EDAS procedure on cerebral perfusion have been assessed in a group of 13 children (Lee et al., 2003). This study demonstrated postoperative changes, concluding that a shortening of TTP in the MCA territory of the hemisphere operated on is a marker for collateralization from the external to the ICAs. The role of MR perfusion as an adjunct to clinical findings in moyamoya disease (or in moyamoya-like syndromes associated with, e.g. neurofibromatosis (NF) type-1 (El-Koussy et al., 2002)) may include an assessment of disease severity (Wityk et al., 2002) and when EDAS is warranted, although this has yet to be assessed. Takayasu arteritis Takayasu arteritis is also a chronic inflammatory disease of the arterial system and is most common in the Far East. Patients present with similar neurological symptoms such as headaches, visual disturbances, seizures plus others attributable to TIA, stroke and intracerebral hemorrhage (Cantu et al., 2000). Multiple areas of cerebral hypoperfusion have been demonstrated using SPECT, and imply that an assessment of vascular perfusion may provide a good prognostic indicator (Hoffmann et al., 2000). MR may also provide this together with information required for accurate diagnosis such as arterial wall anatomy or differentiation between plaque type/function before morphological changes are apparent. This may facilitate

the detection of disease activity at a more treatable stage (Kissin and Merkel, 2004). A recent study involving 60 patients identified 98% with stenotic lesions and 27% with aneurysms of the circulatory system; restenosis following bypass or angioplasty was found to be common. Medical therapy comprised immunosuppressive treatment with glucocorticoids alone or in combination with a cytotoxic agent, although this failed in 25% of cases and relapse occurred in 50% (Kerr et al., 1994). These findings suggest that serial perfusion assessment with MR could prove useful. Perfusion assessment may also have a role following bypass therapy in individuals who have hemodynamically significant lesions in all four cervical arteries, and who may be at risk of hyperemia. Vascular abnormalities Intracranial vascular abnormalities such as aneurysms, arteriovenous malformations (AVM) or angiomas can also cause symptomatic perfusion deficits. Perhaps more importantly in such patient groups, a significant risk of hemorrhage is associated with high morbidity and mortality. The management of such “sleeping” pathology can be difficult and information regarding vascular and parenchymal function may be of high value in determining prognosis and therapeutic approach. Giant aneurysms of the intracranial ICA cause regional increases in MTT (Figure 16.13), most likely resulting from flow disturbance or mass effect of the sack and following bypass surgery, normalization of perfusion parameters has been observed (Caramia et al., 2001). Perfusion and diffusion have also been used to highlight oligemia

Perfusion and diffusion imaging in chronic carotid disease

(a)

(b)

(c)

Fig. 16.13 Complex disturbances to parenchymal perfusion timing caused by the presence of the giant MCA aneurysm (a) depicted on time-of-flight MRA. Reduction in transit time (low signal) is observed (b) superior to and increase in transit time (high signal) is observed (c) at the same axial level and adjacent to the aneurysm.

and developing ischemia adjacent to the MCA in vasospasm secondary to subarachnoid hemorrhage. Normalization of perfusion and diffusion has been reported following appropriate endovascular treatment, concordant with normalization in clinical symptoms and signs (Bracard et al., 2001).

Summary and future possibilities Both perfusion and diffusion are useful and quite different markers of disease in chronic ICA pathology. The implementation of MR perfusion is evolving and it is expected that future studies will move towards the use of quantitative (Yoneda et al., 2003) rather than qualitative assessment techniques. It is hoped that, in combination with other functional MR modalities, the production of task-specific CMRO2 maps will be possible in the clinical setting. Increased knowledge of other factors of influence such as normal variations in macrovascular topography and its relationship to spin history or whether brain vascular regulation is closely coupled with cardiovascular status may further our understanding of brain perfusion and its relationship to the risk of ischemia. Diffusivity assessment is also developing rapidly, both in our understanding of silent ischemia and in measurement technique and applications. For example, its role may be extended to functional activation studies where measurement of ADC contrast, which is thought be closely coupled to neuronal activation (thus reducing the point-spread function associated with BOLD contrast), may aid

high-resolution fiber tracking/activation studies (Song et al., 2003). Whilst perfusion and diffusion imaging provide surrogate prognostic information, the clinical goal of providing a “one-stop” battery of tests capable of quantifying individual patient cerebrovascular risk is unfortunately still some way off. Diffusion and perfusion imaging are helping us to assess new interventional techniques and devices, providing independent means of monitoring changes in function which directly effect quality of life. As with all imaging investigations, data interpretation must be made with reference to the overall clinical picture.

REFERENCES Barth A, Remonda L, Lovblad KO, Schroth G, Seiler RW. 2000. Silent cerebral ischemia detected by diffusion-weighted MRI after carotid endarterectomy. Stroke 31(8): 1824–1828. Bozzao A, Floris R, Gaudiello F, Finocchi V, Fantozzi LM, Simonetti G. 2002. Hemodynamic modifications in patients with symptomatic unilateral stenosis of the internal carotid artery: evaluation with MR imaging perfusion sequences. Am J Neuroradiol 23(8): 1342–1345. Bracard S, Anxionnat R, Auliac S, Melo Neto J, Lebendinsky A, Audibert G, Picard L. 2001. Relevance of diffusion and perfusion weighted MRI for endovascular treatment of vasospasm in subarachnoid hemorrhage. J Neuroradiol 28(1): 27–32. Burdette JH, Ricci PE, Petitti N, Elster AD. 1998. Cerebral infarction: time course of signal intensity changes on diffusion-weighted MR images. Am J Roentgenol 171(3): 791–795.

259

260

Iain D. Wilkinson

Cantu C, Pineda C, Barinagarrementeria F, Salgado P, Gurza A, Paola de Pablo, Espinosa R, Martinez-Lavin M. 2000. Noninvasive cerebrovascular assessment of Takayasu arteritis. Stroke 31(9): 2197–2202. Caplan LR, Hennerici M. 1998. Impaired clearance of emboli (washout) is an important link between hypoperfusion, embolism, and ischemic stroke. Arch Neurol 55(11): 1475–1482. Caramia F, Santoro A, Pantano P, Passacantilli E, Guidetti G, Pierallini A, Fantozzi LM, Cantore GP, Bozzao L. 2001. Cerebral hemodynamics on MR perfusion images before and after bypass surgery in patients with giant intracranial aneurysms. Am J Neuroradiol 22(9): 1704–1710. CAVATAS investigators. 2001. Endovascular versus surgical treatment in patients with carotid stenosis in the carotid and vertebral artery transluminal angioplasty study (CAVATAS): a randomized study. Lancet 357: 1729–1737. Detre JA, Samuels OB, Alsop DC, Gonzalez-At JB, Kasner SE, Raps EC. 1999. Noninvasive magnetic resonance imaging evaluation of cerebral blood flow with acetazolamide challenge in patients with cerebrovascular stenosis. J Magn Reson Imaging 10(5): 870–875. Doerfler A, Eckstein HH, Eichbaum M, Heiland S, Benner T, Allenberg JR, Forsting M. 2001. Perfusion-weighted magnetic resonance imaging in patients with carotid artery disease before and after carotid endarterectomy. J Vasc Surg 34: 587–593. El-Koussy M, Lovblad KO, Steinlin M, Kiefer C, Schroth G. 2002. Perfusion MRI abnormalities in the absence of diffusion changes in a case of moyamoya-like syndrome in neurofibromatosis type 1. Neuroradiology 44(11): 938–941. European Carotid Surgery Trials Collaborative Group. 1998. Randomised trial of endarterectomy for recently symptomatic carotid stenosis: final results of the MRC European Carotid Surgery Trial (ECST). Lancet 351: 1379–1387. Feiwell RJ, Besmertis L, Sarkar R, Saloner DA, Rapp JH. 2001. Detection of clinically silent infarcts after carotid endarterectomy by use of diffusion-weighted imaging. Am J Neuroradiol 22(4): 646–649. Fiehler J, Foth M, Kucinski T, Knab R, von Bezold M, Weiller C, Zeumer H, Rother J. 2002. Severe ADC decreases do not predict irreversible tissue damage in humans. Stroke 33(1): 79–86. Forbes KP, Shill HA, Britt PM, Zabramski JM, Spetzler RF, Heiserman JE. 2001. Assessment of silent embolism from carotid endarterectomy by use of diffusion-weighted imaging: work in progress. Am J Neuroradiol 22(4): 650–653. Gillard JH, Hardingham CR, Kirkpatrick PJ, Antoun NM, Freer CEL, Griffiths PD. 1998. Evaluation of carotid endarterectomy with sequential MR perfusion imaging: a preliminary report. Am J Neuroradiol 19: 1747–1752.

Gillard JH, Hardingham CR, Antoun NM, Freer CE, Kirkpatrick PJ. 1999. Evaluation of carotid endarterectomy with sequential MR perfusion imaging: a preliminary 12-month follow up. Clin Radiol 54: 798–803. Griffiths PD, Hoggard N, Dannells W, Wilkinson ID. 2001. In vivo measurement of cerebral blood flow: a review of methods and applications. Vasc Med 6(1): 51–60. Griffiths PD, Salam S, Gaines P, Cleveland T, Beard J, Venables G, Wilkinson ID. 2003. Assessment of cerebral haemodynamics and vascular reserve in patients with symptomatic carotid artery occlusion: an integrated MR method. Annual meeting of the British Society of Neuroradiologists, KNAW, Amsterdam, 10–11th October, 2003. Guckel FJ, Brix G, Schmiedek P, Piepgras Z, Becker G, Kopke J, Gross H, Georgi M. 1996. Cerebrovascular reserve capacity in patients with occlusive cerebrovascular disease: assessment with dynamic susceptibility contrast-enhanced MR imaging and the acetazolamide stimulation test. Radiology 201(2): 405–412. Hartkamp MJ, van Der Grond J, van Everdingen KJ, Hillen B, Mali WP. 1999. Circle of Willis collateral flow investigated by magnetic resonance angiography. Stroke 30(12): 2671–2678. Hermier M, Ibrahim AS, Wiart M, Adeleine P, Cotton F, Dardel P, Derex L, Berthezene Y, Nighoghossian N, Froment JC. 2003. The delayed perfusion sign at MRI. J Neuroradiol 30(3): 172–179. Hoffmann M, Corr P, Robbs J. 2000. Cerebrovascular findings in Takayasu disease. J Neuroimaging 10(2): 84–90. Hofmeijer J, Klijn CJ, Kappelle LJ, Van Huffelen AC, Van Gijn J. 2002. Collateral circulation via the ophthalmic artery or leptomeningeal vessels is associated with impaired cerebral vasoreactivity in patients with symptomatic carotid artery occlusion. Cerebrovasc Dis 14(1): 22–26. Jaeger HJ, Mathias KD, Drescher R, Hauth E, Bockisch G, Demirel E, Gissler HM. 2001. Diffusion-weighted MR imaging after angioplasty or angioplasty plus stenting of arteries supplying the brain. Am J Neuroradiol 22(7): 1251–1259. Jäger R, Saunder D. 2001. Cranial and intracranial pathology (2): cerebrovascular disease and non-traumatic intracranial haemorrhage. In Diagnostic Radiology: A Textbook of Medical Imaging, 4th edn. (Eds., Grainger RS, Allison DJ, Adam A, Dixon AK), Harcourt Publishers Ltd, London. Kajimoto K, Moriwaki H, Yamada N, Hayashida K, Kobayashi J, Miyashita K, Naritomi H. 2003. Cerebral hemodynamic evaluation using perfusion-weighted magnetic resonance imaging: comparison with positron emission tomography values in chronic occlusive carotid disease. Stroke 34(7): 1662–1666. Kastrup A, Schulz JB, Mader I, Dichgans J, Kuker W. 2002. Diffusion-weighted MRI in patients with symptomatic internal carotid artery disease. J Neurol 249(9): 1168–1174.

Perfusion and diffusion imaging in chronic carotid disease

Kerr GS, Hallahan CW, Giordano J, Leavitt RY, Fauci AS, Rottem M, Hoffman GS. 1994. Takayasu arteritis. Ann Intern Med 120(11): 919–929. Kim JH, Lee SJ, Shin T, Kang KH, Choi PY, Gong JC, Choi NC, Lim BH. 2000. Correlative assessment of hemodynamic parameters obtained with T2*-weighted perfusion MR imaging and SPECT in symptomatic carotid artery occlusion. Am J Neuroradiol 21: 1450–1456. Kissin EY, Merkel PA. 2004. Diagnostic imaging in Takayasu arteritis. Curr Opin Rheumatol 16(1): 31–37. Kluytmans M, van der Grond J, Eikelboom BC, Viergever MA. 1998. Long-term hemodynamic effects of carotid endarterectomy. Stroke 29: 1567–1572. Le Bihan D, Breton E, Lallemand D, Grenier P, Cabanis E, Laval-Jeantet M. 1986. MR imaging of intravoxel incoherent motions: application to diffusion and perfusion in neurologic disorders. Radiology 161(2): 401–407. Lee SK, Kim DI, Jeong EK, Kim SY, Kim SH, In YK, Kim DS, Choi JU. 2003. Postoperative evaluation of moyamoya disease with perfusion-weighted MR imaging: initial experience. Am J Neuroradiol 24(4): 741–747. Lefkowitz D, LaBenz M, Nudo SR, Steg RE, Bertoni JM. 1999. Hyperacute ischemic stroke missed by diffusion-weighted imaging. Am J Neuroradiol 20(10): 1871–1875. Liebeskind DS. 2003. Collateral circulation. Stroke 34(9): 2279–2284; Epub 2003 July 24. Lovblad KO, Laubach HJ, Baird AE, Curtin F, Schlaug G, Edelman RR, Warach S. 1998. Clinical experience with diffusion-weighted MR in patients with acute stroke. Am J Neuroradiol 19(6): 1061–1066. Lythgoe DJ, Ostergaard L, William SC, Cluckie A, BuxtonThomas M, Simmons A, Markus HS. 2000. Quantitative perfusion imaging in carotid artery stenosis using dynamic susceptibility contrast-enhanced magnetic resonance imaging. Magn Reson Imaging 18(1): 1–11. Macdonald S, Gaines PA. 2003. Current concepts of mechanical cerebral protection during precutaneous carotid intervention. Vasc Med 8(1): 25–32. Macdonald S, McKevitt F, Venables GS, Cleveland TJ, Gaines PA. 2002a. Neurological outcomes after carotid stenting protected with the NeuroShield filter compared to unprotected stenting. J Endovasc Ther 9(6): 777–785. Macdonald S, Wilkinson ID, Gaines PA, Cleveland TJ, Frost K, Venables GS, Griffiths PD. 2002b. Changes in cerebral hemodynamics following carotid stenting by magnetic resonance (MR) perfusion imaging. Proceedings of the Cardiovascular and Interventional Radiological Society of Europe, Lucerne, Switzerland. Maeda M, Yuh WT, Ueda T, Maley JE, Crosby DL, Zhu MW, Magnotta VA. 1999. Severe occlusive carotid artery disease: hemodynamic assessment by MR perfusion

imaging in symptomatic patients. Am J Neuroradiol 20(1): 43–51. Markus H, Cullinane M. 2001. Severely impaired cerebrovascular reactivity predicts stroke and TIA risk in patients with carotid artery stenosis and occlusion. Brain 124(Pt 3): 457–467. Merboldt KD, Hanicke W, Bruhn H, Gyngell ML, Frahm J. 1992. Diffusion imaging of the human brain in vivo using highspeed STEAM MRI. Magn Reson Med 23(1): 179–192. Muller M, Reiche W, Langenscheidt P, Hassfeld J, Hagen T. 2000. Ischemia after carotid endarterectomy: comparison between transcranial Doppler sonography and diffusionweighted MR imaging. Am J Neuroradiol 21(1): 47–54. Nasel C, Azizi A, Wilfort A, Mallek R, Schindler E. 2001. Measurement of time-to-peak parameter by use of a new standardization method in patients with stenotic or occlusive disease of the carotid artery. Am J Neuroradiol 22: 1056–1061. Nighoghossian N, Berthezene Y, Phillippon B, Adeleine P, Froment JC, Trouillas P. 1996. Hemodynamic parameter assessment with dynamic susceptibility contrast magnetic resonance imaging in unilateral symp-tomatic internal carotid artery occlusion. Stroke 27: 474–479. Nighoghossian N, Berthezene Y, Meyer R, Cinotti L, Adeleine P, Philippon B, Froment JC, Trouillas P. 1997. Assessment of cerebrovascular reactivity by dynamic susceptibility contrast-enhanced MR imaging. J Neurol Sci 149(2): 171–176. North American Symptomatic Carotid Endarterectomy Trial Collaborators (NASCET). 1991. Beneficial effects of carotid endarterectomy in symptomatic patients with high grade carotid stenosis. N Eng J Med 325: 445–453. Ordidge RJ, Helpern JA, Qing ZX, Knight RA, Nagesh V. 1994. Correction of motional artifacts in diffusion-weighted MR images using navigator echoes. Magn Reson Imaging 12(3): 455–460. Pandya H, Wilkinson ID, Griffiths PD. 2003. Sequential dynamic gadolinium MR perfusion weighted imaging: effects on transit time and cerebral blood volume measurements. Annual Meeting of the British Society of Neuroradiologists, KNAW, Amsterdam, 10–11th October, 2003. Powers AD, Smith RR. 1990. Hyperperfusion syndrome after carotid endarterectomy: a transcranial doppler evaluation. Neurosurgery 26: 56–59. Powers WJ. 1991. Cerebral hemodynamics in ischemic cerebrovascular disease. Ann Neurol 29(3): 231–240. Restrepo L, Wityk RJ, Grega MA, Borowicz Jr L, Barker PB, Jacobs MA, Beauchamp NJ, Hillis AE, McKhann GM. 2002. Diffusion- and perfusion-weighted magnetic resonance imaging of the brain before and after coronary artery bypass grafting surgery. Stroke 33(12): 2909–2915. Rodda RA. 1986. The arterial patterns associated with internal carotid disease and cerebral infarcts. Stroke 17(1): 69–75.

261

262

Iain D. Wilkinson

Schluter M, Tubler T, Steffens JC, Mathey DG, Schofer J. 2003. Focal ischemia of the brain after neuroprotected carotid artery stenting. J Am Coll Cardiol 42(6): 1007–1013. Schulz UG, Briley D, Meagher T, Molyneux A, Rothwell PM. 2003. Abnormalities on diffusion weighted magnetic resonance imaging performed several weeks after a minor stroke or transient ischaemic attack. J Neurol Neurosurg Psychiatry 74(6): 734–738. Schwartz RB. 2002. Hyperperfusion encephalopathies: hypertensive encephalopathy and related conditions. Neurology 8(1): 22–34. Silvestrini M, Vernieri F, Pasqualetti P, Matteis M, Passarelli F, Troisi E, Caltagirone C. 2000. Impaired cerebral vasoreactivity and risk of stroke in patients with asymptomatic carotid artery stenosis. J Am Med Assoc 283(16): 2122–2127. Soinne L, Helenius J, Tatlisumak T, Saimanen E, Salonen O, Lindsberg PJ, Kaste M. 2003. Cerebral hemodynamics in asymptomatic and symptomatic patients with high-grade carotid stenosis undergoing carotid endarterectomy. Stroke 34(7): 1655–1661. Song AW, Harshbarger T, Li T, Kim KH, Ugurbil K, Mori S, Kim DS. 2003. Functional activation using apparent diffusion coefficient-dependent contrast allows better spatial localization to the neuronal activity: evidence using diffusion tensor imaging and fiber tracking. Neuroimage 20(2): 955–961. Sperling B, Lassen NA. 1993. Hyperfixation of HMPAO in subacute ischemic stroke leading to spuriously high estimates of cerebral blood flow by SPECT. Stroke 24(2): 193–194. Szabo K, Kern R, Gass A, Hirsch J, Hennerici M. 2001. Acute stroke patterns in patients with internal carotid artery disease: a diffusion-weighted magnetic resonance imaging study. Stroke 32(6): 1323–1329. Teng MM, Cheng HC, Kao YH, Hsu LC, Yeh TC, Hung CS, Wong WJ, Hu HH, Chiang JH, Chang CY. 2001. MR perfusion studies of brain for patients with unilateral carotid stenosis or occlusion: evaluation of maps of “time to peak” and “percentage of baseline at peak”. J Comput Assist Tomogr 25(1): 121–125. Terada T, Tsuura M, Matsumoto H, Masuo O, Yamaga H, Tsumoto T, Itakura T. 2003. Results of endovascular treatment of internal carotid artery stenoses with a newly developed balloon protection catheter. Neurosurgery 53(3): 617–623. Tomczak R, Wunderlich A, Liewald F, Stuber G, Gorich J. 2001. Diffusion-weighted MRI: detection of cerebral ischemia before and after carotid thromboendarterectomy. J Comput Assist Tomogr 25(2): 247–250.

Warach S, Gaa J, Siewert B, Wielopolski P, Edelman RR. 1995. Acute human stroke studied by whole brain echo planar diffusion-weighted magnetic resonance imaging. Ann Neurol 37(2): 231–241. Wiart M, Berthezene Y, Adeleine P, Feugier P, Trouillas P, Froment JC, Nighoghossian N. 2000. Vasodilatory response of border zones to acetazolamide before and after endarterectomy: an echo planar imaging-dynamic susceptibility contrastenhanced MRI study in patients with high-grade unilateral internal carotid artery stenosis. Stroke 31(7): 1561–1565. Wilkinson ID, Beard JD, Hoggard N, Griffiths PD, Venables GS. 2001. Short-term haemodynamic consequences of carotid endarterectomy. Proceedings of International Society of Magnetic Resonance of Medicine, p. 1440. Wilkinson ID, Griffiths PD, Hoggard N, Cleveland TJ, Gaines PA, Venables GS. 2000. Unilateral leptomeningeal enhancement after carotid stent insertion detected by magnetic resonance imaging. Stroke 31: 848–851. Wilkinson ID, Griffiths PD, Hoggard N, Cleveland T, Gaines P, Macdonald S, McKevitt F, Venables GS. 2003. Short-term changes in cerebral micro-hemodynamics following carotid stenting. Am J Neuroradiol 24: 1501–1507. Wityk RJ, Goldsborough MA, Hillis A, Beauchamp N, Barker PB, Borowicz Jr LM, McKhann GM. 2001. Diffusionand perfusion-weighted brain magnetic resonance imaging in patients with neurologic complications after cardiac surgery. Arch Neurol 58(4): 571–576. Wityk RJ, Hillis A, Beauchamp N, Barker PB, Rigamonti D. 2002. Perfusion-weighted magnetic resonance imaging in adult moyamoya syndrome: characteristic patterns and change after surgical intervention: case report. Neurosurgery 51(6): 1499–505. Yadav JS, Roubin GS, Iyer S, Vitek J, King P, Jordan WD, Fisher WS. 1997. Elective stenting of the extracranial carotid arteries. Circulation 95: 376–381. Yamauchi H, Fukuyama H, Nagahama Y, Katsumi Y, Okazawa H. 1998. Cerebral hematocrit decreases with hemodynamic compromise in carotid artery occlusion: a PET study. Stroke 29(1): 98–103. Yonas H, Smith HA, Durham SR, Pentheny SL, Johnson DW. 1993. Increased stroke risk predicted by compromised cerebral blood flow reactivity. J Neurosurg 79(4): 483–489. Yoneda K, Harada M, Morita N, Nishitani H, Uno M, Matsuda T. 2003. Comparison of FAIR technique with different inversion times and post-contrast dynamic perfusion MRI in chronic occlusive cerebrovascular disease. Magn Reson Imaging 21(7): 701–705.

17

Imaging migraine pathogenesis K. Michael Welch* Professor of Neurology, Rosalind Franklin University of Medicine and Science, Chicago, IL, USA

Key points • Physiological imaging techniques are improving the understanding of the pathogenesis of migraine and headache. • Subtle abnormalities in phosphorus-31 MR spectroscopy (MRS) (decreased phosphocreatine, calculated magnesium levels) may be detectable interictally in patients with migraine, and in patients with severe migraine (e.g. hemiplegic migraine). • Proton MRS during visual stimulation may show differences (lower occipital N-acetyl aspartate) between migraine patients and controls. • Diffusion, perfusion or blood oxygen level dependent MR imaging (MRI) may observe changes associated with the spreading depression of Leao during the aura preceding a migraine. • Physiological MRI and positron emission tomography suggest a role for periaqueductal gray matter in headache pathogenesis.

Abstract Although the pathogenesis of migraine is incompletely understood, non-invasive brain imaging techniques have heralded major advances in understanding pathogenesis of migraine because they can be applied safely to a population that is healthy *

Disclosures: K. Michael Welch, MD has been a consultant/scientific advisor for AstraZeneca Pharmaceuticals, LP; GlaxoSmithKline, Allergan, Ortho-McNeill and Pharmacia.

between attacks. Recent imaging studies reviewed in this chapter have shed light on the neuronal events mediating both the aura and the headache phases of migraine, identifying a cerebral cortical origin of migraine aura, susceptibility to attacks based on cortical hyper-excitability, and headache originating in the trigeminovascular system and its central projections. Abnormal modulation of brain nociceptive systems, at first transient but becoming permanent with continuing illness, may explain prolonged headache of the migraine attack and shift of the migraine phenotype from episodic to chronic headache.

Introduction The last quarter of a century has seen rapid advances in understanding the mechanisms and management of primary headaches, most of all migraine. Of all the primary headaches migraine exacts the major toll on disability in the population, being a highly prevalent chronic episodic illness accompanied by severe pain in two-thirds of cases. Non-invasive brain imaging techniques heralded major advances in understanding pathogenesis of migraine because they could be applied safely to a population that is healthy, at least between attacks. This chapter will highlight aspects of migraine pathogenesis that have been explored and elucidated by brain imaging, while reviewing current concepts in general. Such concepts include (a) neuronal hyper-excitablity during the interictal phase, (b) cortical spreading depression (CSD) as the basis of aura, (c) trigeminal neurovascular activation at a peripheral or central (brainstem) origin that accounts for headache, and (d) the provocative 263

264

K. Michael Welch

18.2 13.8

8.1 3.1 0 min Activation

Suppression of activation

Fig. 17.1 Functional MR imaging–blood oxygen level dependent (FMRI-BOLD) study of migraine with aura patient who had visually triggered headache and aura. Activation of primary and associated visual cortex is seen in the left panel, and suppression of the prior activation in the right. In the suppression map, the time that the suppression was initiated was color coded. It can be seen that suppression commenced at primary visual cortex then spread laterally, anteriorly, and bilaterally. Suppression was also triggered in regions remote from the primary visual cortex (cf. Cao et al., 1999).

concept that progressive damage to the periaqueductal gray matter (GM) may explain some aspects of central sensitization or change in phenotypic expression of the disorder.

The interictal status in migraine The concept that migraine attacks originate in brain and can be triggered under various conditions argues in favor of a threshold that governs the incidence of attacks. The nature of the final common pathway with which these factors interact probably constitutes the true cause of migraine. In this regard, transient or persistently exaggerated excitability of neurons in the cerebral cortex, especially occipital, has received strong consideration (Welch et al., 1990). Transcranial magnetic stimulation (TMS) of the occipital cortex required to produce phosphene generation akin to the scintillating visual experiences of migraine aura was significantly lower in patients with migraine with aura between their headaches than it was in normal controls (Aurora et al., 1998). Using the same technology, but with different paradigms, other studies have added consistent data that support cortical hyper-excitability (Batelli et al., 2002; Mulleners et al., 2002; Young et al., 2004), also indicating that hyper-excitability of the visual cortex (VC) in migraine goes beyond visual area V1. Observing phosphenes is a subjective

experience, however; one drawback of these studies leading to understandable controversy (Afra et al., 1998). Nevertheless, Huang et al. used functional MR imaging-blood oxygen level dependent (fMRIBOLD) to document a hyper-excitable neuronal response in terms of peak magnitude of BOLD signal and visual illusions and distortions when migraine with aura patients viewed square wave gratings at different spatial frequencies (Huang et al., 2003). When subjects with a natural and reproducible history of migraine attacks induced by visual stress were studied, the success rates for experimentally induced attacks using checkerboard visual stimulation were high (Cao et al., 1999; Bowyer et al., 2001). Visual activation monitored by magneto-encephalography and fMRI-BOLD confirmed abnormal excitability of widespread regions of the occipital, occipito-temporal and occipito-parietal cortex, with consequent triggering of the neuro-electric accompaniments of aura symptoms (Cao et al., 1999; Bowyer et al., 2001). In an fMRI-BOLD study of migraine patients (Cao et al., 1999), visual stimulation designed to activate the total occipital cortex initiated spreading suppression of initial activation at rates compatible with cortical spreading depression (CSD) (Figure 17.1). Multiple events were evoked bilaterally from different regions of the occipital, occipital– parietal, and occipital–temporal cortex. Magnetoencephalography results, despite using a stimulus designed to activate primary visual cortex alone, confirmed the multi-focality of neuronal excitation

Imaging migraine pathogenesis

DC MEG data segment 6000 Amplitude (fT)

Mean cortical source amplitudes 200 150 100

4000 2000 0

2000

4000 850

50 0 250

900 Time (s)

950

Mean MEG data contour

200

Back of head

150 100 50 0 300

250

200

150

100

50

0 Right side

Fig. 17.2 Magneto-encephalography (MEG) study collected during visually induced migraine aura, 150 s after beginning stimulus. The MEG images were generated using two-dimensional inverse imaging. This technique can produce whole-brain images of both focal and extended source structures that may be simultaneously active. For these studies the two-dimensional inverse imaging technique used approximately 3000 cortical source locations to model the continuum of cortical GM. The figure shows the average cortical source amplitude locations for 200 s, displayed on the patient’s MRI. Areas of activation are seen in the primary visual, right temporal, and left occipital cortices. DC measurements and contour maps are shown on the right panels (cf. Bowyer et al., 2001).

throughout occipital cortex and the DC shifts that arose from these sites (Bowyer et al., 2001) (Figure 17.2). Excitability of cell membranes of neurons, especially in the occipital cortex, therefore, seems fundamental to the migraine brain’s susceptibility to attacks (Welch et al., 1990). Factors that increase or decrease neuronal excitability in turn may modulate the threshold for triggering attacks. Different cellular mechanisms may underlie increased central neuronal excitability in migraine. Primary disorders of brain mitochondria, for example mitochondrial encephalopathy with stroke like episodes, are associated with symptomatic migraine attacks (Welch and Ramadan, 1995). Presumably, impaired energy metabolism causes cellular ionic inhomeostasis, membrane instability, and readily depolarizable neurons when subjected to triggering stimuli, culminating in spreading cortical depression (cf. later). Previously, localized phosphorus spectroscopy performed in migraine with aura yielded data on brain

energy status that had suggested dysfunction of brain mitochondria (Welch et al., 1989); abnormal muscle energetics in the same patients raised the possibility of the disorder being generalized (Barbirolli et al., 1992). These single-voxel studies were extended to include multiple brain regions and larger numbers of patients by multislice phosphorus-31 MR spectroscopic imaging (MRSI) (Boska et al., 2002). Migraine with aura, migraine without aura, and hemiplegic migraine patients were studied between attacks. In migraine without aura, consistent increases in phosphodiester concentration (PDE) were measured in most brain regions, with a trend towards increase in Mg 2 in posterior brain. In migraine with aura, phosphocreatine concentration (PCr) was decreased to a minor degree in anterior brain regions and a trend towards decreased Mg 2 was observed in posterior regions, but no consistent changes were found in phosphomonoester concentration (PME), PDE, inorganic phosphate concentration (Pi), or pH.

265

K. Michael Welch

In hemiplegic migraine patients, PCr had a tendency to be lower, and Mg 2 was significantly lower than in the posterior brain regions of control subjects. Trend analysis showed a significant decrease of brain Mg 2 and PDE in posterior brain regions with increasing severity of neurological symptoms. Overall, the results showed no substantial or consistent abnormalities of energy metabolism, but there were trends towards abnormality in posterior brain regions in severe forms of migraine as previously shown by single-voxel studies reviewed above. Methods of measurement may explain differences between studies, but also because interictal measurements are only a single snapshot in time they may miss shifts in brain metabolism that permit triggering of attacks to explain differences in energy status recorded between series of patients separately studied. In further support, Figure 17.3 shows results of an experiment wherein migraine patients and controls were subjected to visual stress with a continuous flashing black and white checkerboard between attacks during serial single-voxel proton spectroscopy of the primary visual cortex (Brooks et al., 2003). Levels of N-acetyl aspartate (NAA) fell during this time compared to normal controls and migraine without aura. Inasmuch as NAA is indicative of mitochondrial function of neurons, these findings would support the importance of stressing neuronal function and energy status is to unmask mitochondrial abnormality. On the other hand, magnesium imaging by means of phosphorus spectroscopy, revealed consistent and profound changes in posterior brain regions of patients severely compromised with hemiplegic migraine (Boska et al., 2002). In a rare autosomal dominant subtype of migraine, familial hemiplegic migraine, mutation of a gene involved in production of a brain-specific P/Q-type calcium channel was identified in about 50% of families (Ophoff et al., 1996). Possibly a gain of function mutation, this channelopathy may result in increased release of excitatory neurotransmitters with consequent neuronal hyper-excitability. Figure 17.4 shows low magnesium in phosphorus-31 spectroscopic images of members of a family with hemiplegic migraine. The changes could reflect attempts by the brain to maintain homeostasis and counteract hyper-excitability by

Migraine aura Migraine aura Normal controls

0 Change of NAA (%)

266

2

4

6

8

10

off 2

on 4

6 8 Acquisition

off 10

Fig. 17.3 Single-voxel proton spectroscopy of the primary visual cortex performed during visual activation (constant 8 Hz red light through goggles). Acquisition time per spectrum was 6 min. NAA fell in all groups, although more rapidly in migraine with aura. NAA fell 9% (P  0.02) in migraine with aura, and 4% (P  0.04) in migraine without aura and in normal controls (cf. Brookes et al., 2003).

magnesium fixation in cell membranes and by gating excitatory receptors. In fact migraine patients without aura showed compensatory changes in Mg2 and membrane phospholipids, indicating perhaps that neurological symptoms only occur in migraine susceptible individuals when the brain fails to adjust its function to maintain homeostasis. Low magnesium certainly is a fundamental mechanism of neuronal excitability, but fits with a general as opposed to localized hyper-excitability of specific structures as indicated by the regional changes shown in our study with otherwise normal values. Curiously, migraine patients have lower circulating magnesium levels than normal, which may stress brain capacity to effectively maintain regional magnesium levels (cf. Welch and Ramadan, 1995). Supplementing magnesium to prevent migraine attacks makes sense under these circumstances and indeed has proven modestly successful (Peikert et al., 1996).

Mechanism underlying aura Research into aura mechanisms has fascinated investigators for several centuries. Up to 20% of

Imaging migraine pathogenesis

Control 700 ␮M

Hemiplegic migraine Right aura

Right/left aura

100 ␮M Right aura

Fig. 17.4 Magnesium maps were obtained from three members of a family with familial hemiplegic migraine and a normal control. Attention to the color coding reveals regional reductions in all three family members, most consistently in posterior brain regions. The arrows represent the hemisphere in which the aura originated, suggesting an association between low magnesium and initiation of aura (cf. Boska et al., 2002).

migraine sufferers experience aura prior to headache onset (Stewart et al., 1994; Russell et al., 1996), which is predominantly visual (Russell and Olesen, 1996). Evidence for spreading depression as the mechanism of aura has accumulated very slowly, dependent on the availability of non-invasive methods of studying brain function and the opportunity to study the onset of spontaneous aura. The advent of non-invasive techniques has given impetus to research on the problem, and the basis of aura has become more and more established over time as a neuro-electric event similar to the CSD of Leao et al. (1944a, 1944b). Advances in brain imaging, including regional cerebral blood flow (rCBF) measured by Xe-133, diffusion/perfusion MRI, MRI-BOLD,

and magneto-encephalography have allowed investigators to observe CSD during migraine aura (Olesen et al., 1981; Olesen, 1991; Woods et al., 1994; Cutrer et al., 1998; Cao et al., 1999; Bowyer et al., 2001; Hadjikhani et al., 2001). Cortical neuronal excitation followed by depression of normal neuronal activity spreads slowly from the site of initiation at rates between 2 and 6 mm/min. Not respecting arterial boundaries, pial arterial, and venous dilatation occurs simultaneously with activated neural activity (Lauritzen et al., 1982). Cao and colleagues using fMRI to study migraine attacks induced by visual stimulation referred to previously, found that transient activation followed by spreading neuronal suppression was accompanied by

267

268

K. Michael Welch

MEG during spontaneous aura Occipital

Images; 25–56

Images; 57–112

Images; 113–168

P  0.001 per pixel Fig. 17.5 Serial T*-weighted imaging (image acquisition time 2 3 s) collapsed into three frames obtained from a migraine patient during an aura of progressing and subsiding left quadrantanopsia. The study commenced between 7 and 10 min of spontaneous onset of the aura. The red depicts regions of increased signal compared to baseline immediately after aura, interpreted as hyperemia and hyperoxia during the initial depolarization phase of spreading depression. Though limited to GM, as expected for changes secondary to spreading depression, the activity is widespread throughout both occipital cortices. How this can be explained when the neurological symptom was localized to the primary visual cortex remains to be determined. Note also activation of brainstem structures.

vasodilatation and hyper-oxygenation (Cao et al., 1999) (Figure 17.5). This brief increase in rCBF is followed by decreased rCBF to oligemic values lasting approximately 1 h after the wave of neuronal inhibition (Lauritzen et al., 1984). This oligemia too does not follow vascular boundaries. Using highfield functional MRI with near-continuous recording during visual aura in three subjects, BOLD signal changes were observed by (Hadjikhani et al., 2001) that demonstrated at least eight characteristics of CSD, time-locked to perception of the onset of aura. The timing of the spread of neuronal suppression as that which characterizes CSD was aided in these fMRI-BOLD studies by computerized “flattening” of the occipital cortex. This also permitted retinotopic correlations of symptoms and their progression with cortical loci. Nevertheless, in the past, sceptics among neurophysiologists have been unconvinced that cortical spreading depression occurs in the highly convoluted

Fig. 17.6 DC-magneto-encephalography (MEG) recording of spontaneous aura involving the right occipital cortex in a migraine patient with left visual aura. Recording was approximately 3 min in duration. Note the slow wave negative and positive shifts in the right posterior region compared to no deflections in the left uninvolved cortex (cf. Bowyer et al., 2001).

human cortex. Measuring the DC waveforms of CSD by standard electro encephalograph (EEG) techniques proved technically impossible, requiring the development of magneto-encephalography. Accordingly, direct measurements of DC brain activity that characterizes spreading depression have been obtained only rarely. In the DC-magneto-encephalography study published by Bowyer et al. referred to above (Bowyer et al., 2001), direct current fields were measured during spontaneous onset of migraine auras in four migraine patients, and compared with recordings from eight migraine with aura patients and six normal controls during visual stimulation of the occipital cortex. Complex DC-magneto-encephalography shifts, similar in waveform, were observed in spontaneous and visually induced migraine patients (Figure 17.6). Two-dimensional inverse imaging showed multiple cortical areas activated in spontaneous and visually induced migraine aura patients (cf. Figure 17.2), whereas in normal subjects, activation was only observed in the primary visual cortex. These results supported a spreading, depression-like neuro-electric event occurring during migraine aura arising spontaneously or being visually triggered in widespread regions of hyper-excitable occipital

Imaging migraine pathogenesis

cortex. Although the waveforms in the magnetoencephalography studies were typical of SD, their complexity ruled out effective timing of the spread of activity, emphasizing the value of the fMRI-BOLD studies referred to above that were able to achieve this, and the complimentary nature of the magnetoencephalography and fMRI techniques. Thus new imaging methods have added to the progressively strengthening case that an SD-like neuro-electric event is responsible for the migraine aura.

Headache mechanisms The mechanisms by which aura transduces the headache of migraine remain to be determined definitively. Spreading depression activates trigeminal nucleus caudalis (Moskowitz et al., 1993; Choudhuri et al., 2002), part of the central pathway mediating migraine pain. The wave of spreading depression may invade cortical trigeminal terminals setting up a cascade of events leading to inflammation as a persistent source of trigeminal stimulation and headache. Recent experimental evidence has implicated both the trigeminal and parasympathetic systems through brainstem connections accounting for vasodilatation and inflammatory change of the extra cerebral circulation, particularly meningeal vessels, during and after CSD (Bolay et al., 2002). This mechanism explains extra cerebral vasodilatation during headache at a time when rCBF reaches oligemic values after SD has traversed the tissues. It is unclear, however, whether meningeal inflammation occurs in human beings during migraine; no imaging studies to date have documented convincingly that this happens in migraine sufferers, a point of controversy. An entirely separate mechanism might be recruitment of cortical–subcortical connections to nociceptive centers directly by the wave of cortical excitation/suppression invading the cortical components of these networks. Magneto-encephalography studies referred to earlier showed that visual stress activated abnormal DC slow wave shifts in widespread regions of occipital, occipito-parietal, and occipito-temporal cortex (Bowyer et al., 2001). Should such activation invade the cortical component of

cortical–subcortical connections to nociceptive centers (Aicardi et al., 1988), this might prove an alternative mechanism for trigeminal activation. On the other hand, migraine aura without headache that often occurs in children, that is failure to activate the trigeminal nociceptive system, may be an example where CSD does not spread into cortical– subcortical connections to brainstem nociceptive networks. Initiation of migraine headache not preceded by aura that is migraine without aura, remains a mechanistic dilemma. Nevertheless, some evidence points to cortical events similar to those underlying aura. Woods et al., (1994) observed profound and bilateral spreading oligemia beginning occipitally and advancing into anterior cortical regions in association with spontaneous headache in a patient with migraine but no aura. Cao and colleagues reported a spreading suppression of neural activity in the occipital cortex prior to headache in migraine patients with or without aura (Cao et al., 1999). Also, this might explain why others have not observed changes in blood flow in migraine without aura (Olesen, 1991). These studies suggest that the same primary neuronal and secondary hemodynamic events may precede the initiation of headache in all patients, remaining clinically silent in migraine without aura. Based on a positron emission tomography (PET) study of acute migraine without aura, central brainstem structures were recently proposed as loci for a primary “generator” role for migraine headache. Although the technology could not define precisely these structures, an area of the rostral brainstem covering the dorsal pons and midbrain, particularly the periaqueductal GM, dorsal raphe nucleus, and the locus ceruleus, was activated in patients with right hemi-cranial headache (Weiller et al., 1995) Figure 17.7. Although the observations were seminal in first demonstrating convincing brainstem involvement in migraine attacks, the activated regions were contra-lateral to the side of head pain, provoking an alternative explanation that these centers were not directly responsible for pain, but responsible instead for modulating the flow of pain impulses. Reasoning along these same lines, in conscious rats, when patterns of brain activity were investigated

269

270

K. Michael Welch

25 mm

23 mm

20 mm

18 mm

16 mm

14 mm

11 mm

9 mm

Fig. 17.7 PET of rCBF (O15) (composite measurements) obtained from nine patients during an attack of migraine without aura. Note activation in the left mesencephlon overlying the periaqueductal GM, the dorsal raphe nucleus and the locus ceruleus (top three right frames). (Reprinted from Nature, Medicine with permission, Weiller et al., 1995.)

after noxious trigeminal stimulation with capsaicin, increased Fos immuno-reactivity, indicative of neuronal activity, was found in the trigeminal nucleus caudalis, but also in many other brainstem structures (Ter Horst et al., 2001). Since many of the regions expressing Fos activity were anti-nociceptive, it seems difficult to determine confidently that a number of the same structures identified in the human PET study “generate” a migraine attack. On the other hand, (Cao et al., 2002) used fMRI-BOLD to study the red nucleus (RN) and substantia nigra (SN) function during visually induced migraine (Cao et al., 2002). These structures had appeared activated in a spontaneous migraine attack studied with the BOLD technique, reported earlier (Figure 17.8). In 75% of subjects who developed migraine symptoms the RN and SN were activated before visual symptoms or change in occipital BOLD signal occurred. In all, it does seem likely that a network of cortical–subcortical structures with modulatory nociceptive and anti-nociceptive function become abnormally activated in a migraine attack, or even may be abnormal between attacks. Thus episodic dysfunction of certain brainstem centers may play a

key role in migraine pain, either through aberrant activation or modulation of impulse flow in the trigeminal system. Consistent with this reasoning, abnormally high iron levels were discovered recently in the midbrain periaqueductal GM of patients with episodic migraine between attacks and chronic migraine (Welch et al., 2001) (Figure 17.9). Further, iron levels accumulated over time with the burden of illness and repeated migraine attacks. (Figure 17.10). The periaqueductal GM is the center of the most powerful anti-nociceptive neuromodulator system in the brain. The orbitofrontal-periaqueductal GM system is key to behavioral responses to threat and stress, switching attention away from pain when essential for survival. Migraine-like headache was produced when periaqueductal GM was electrically stimulated in human beings (Raskin et al., 1987; Veloso et al., 1998) or triggered by certain structural pathologies involving periaqueductal GM (Haas et al., 1993; Goadsby, 2002). If periaqueductal GM function is altered so that it fails to switch on appropriately or is inappropriately activated, then this might account for migraine headache as imaging

Imaging migraine pathogenesis

Patient with aura in L visual field L

BVR SN

RN

RN

SN

Fig. 17.8 Serial T2*-weighted imaging (image acquisition time 3 s) collapsed into three frames obtained from a migraine patient during an aura of progressing and subsiding left quadrantanopsia. (The figure is an enlargement from Figure 17.5, cf. legend for details.) The red depicts regions of increased signal compared to baseline immediately after aura, interpreted as hyperemia and hyperoxia. The basal vein of Rosenthal (BVR), carries hyper-oxygenated blood consistent with its occipital cortex drainage territory, shown also to be hyperoxic in Figure 17.5. This figure further illustrates that brainstem centers are activated during a migraine attack, although the exact sequence or role of each structure remains to be determined.

studies reviewed above suggested. Permanent periaqueductal GM dysfunction between attacks might also explain increased risk for pain from other body structures, reported in some epidemiological studies of migraine patients. The periaqueductal GM itself is a site of triptan binding and an alternative or additional site for the action of these drugs (Goadsby and Hoskin, 1996). Of interest, periaqueductal GM modulation of trigeminovascular nociceptive afferent signals in experimental animals (Knight and Goadsby, 2001) is blocked after local blockade of the P/Q voltage-gated calcium channel with agatoxin-IVA (Knight et al., 2002). As noted previously, mis-sense mutations in the CACNA.1A subunit of this channel were found in families with hemiplegic migraine (Ophoff et al., 1996). One possibility to explain iron accumulation in periaqueductal GM, a structure normally with high metabolic activity and high iron turnover, is that

repeated hyperoxia accompanying activation might result in progressive free-radical mediated cellular damage. In support, we recently identified the vigorous upregulation of cortical anti-oxidant genes using CSD as a model of migraine aura in the mouse, supporting a protective response of the brain against free-radical damage during migraine attacks (Choudhuri et al., 2002). Cellular dysfunction or damage by free radicals may be detected by iron deposition. Iron homeostasis is tightly maintained in brain, and wherever levels are abnormal then function is disturbed in some way. The strong associations of duration of illness, from which frequent repeated migraine attacks may be inferred, with iron deposition and accordingly periaqueductal GM damage, might explain why episodic migraine becomes chronic over time in some patients. This being the case, rapid and effective abortive treatment and aggressive preventive measures seem essential in those patients

271

K. Michael Welch

Iron deposition (R2 ) in PAGM Normal control Chronic migraine SN R2 6.0 R2 6.9

RN

PAGM R2 4.2–4.4

Fig. 17.9 Three MRI views of the mesencephalon are shown. The right upper and lower panels show image segmentation using ISODATA to identify the periaqueductal gray matter (PAGM), the SN, and the RN. R2 is an index of iron content which can be calculated from differences in relaxation rates measured from spin-echo (R2) and gradient-echo (R2*) MRI. The left MRI is from a normal patient and the right images are from a patient with chronic migraine. Note an approximate 50% increase in iron content in the migraine patient [cf. Welch et al., 2001].

R2' of PAGM (1/ms)

272

8

y  0.0713x  5.0657 r  0.77, P  0.00347

7.5 7 6.5 6 5.5 5 0

5

10 15 20 25 30 Years of (EM  CDH)

35

40

Fig. 17.10 Twelve patients were studied in whom the years of episodic migraine (EM) and chronic migraine (CM) were recorded. PAGM iron content (R2) was measured during the stage of chronic migraine. Note the highly statistically significant increase in PAGM iron levels with the duration, and assumedly burden of attacks [cf. Welch et al., 2001].

with frequent migraine attacks. The potential for even subtle but cumulative free-radical damage during attacks, also might suggest the importance of combining treatment with a free-radical scavenger, for example vitamin E. Despite the importance of caution in interpreting a migraine generator role from the PET studies

reviewed above (Weiller et al., 1995), if periaqueductal GM function is permanently abnormal, failing to switch on appropriately or being inappropriately activated, then this indeed might offer a primary mechanism for migraine headache. Observations from electrode stimulation and structural disease involving periaqueductal GM add to this notion (Raskin et al., 1987; Haas et al., 1993; Veloso et al., 1998; Goadsby, 2002). Iron measurements provided some, but not conclusive, evidence that periaqueductal GM levels might be abnormally high at the outset of the illness (Welch et al., 2001). Although not directly implicating abnormal iron levels in periaqueductal GM, recent epidemiological and genotyping studies conducted in Scandinavia have linked hemochromatosis with an increased risk of headache and possibly migraine (Hagen et al., 2002). Such connected evidence of periaqueductal GM dysfunction evolving from separate studies of very different approach seems to warrant further investigation centered on structure and function of systems that modulate nociception as the origin of migraine headache.

Imaging migraine pathogenesis

REFERENCES Afra J, Mascia A, Gerard P, Maertens de Noordhout A, Schoenen J. 1998. Interictal cortical excitability in migraine: a study using transcranial magnetic stimulation of motor and visual cortices. Ann Neurol 44: 209–215. Aicardi G, Giuffrida R, Rapisarda C, Albe-Fessard D. 1988. Effects of cortical spreading depression on spontaneous activity of red nucleus cells in the guinea pig. Arch Ital Biol 126: 199–203. Aurora SK, Ahmad BK, Welch KMA, Bhardhwaj P, Ramadan NM. 1998. Transcranial magnetic stimulation confirms hyperexcitability of occipital cortex in migraine. Neurology 50: 1111–1114. Barbirolli B, Montagna P, Cortelli P, Fanicello R, Iotti S, Munari L, Pierangeli G, Zaniol P, Lugaresi E. 1992. Abnormal brain and muscle energy metabolism shown by 31P magnetic resonance spectroscopy in patients affected by migraine with aura. Neurology 42: 1209–1214. Batelli L, Black KR, Wray SH. 2002. Transcranial magnetic stimulation of visual area V5 in migraine. Neurology 58(7): 1066–1069. Bolay H, Reuter U, Dunn AK, Huang Z, Boas DA, Moskowitz MA. 2002. Intrinsic brain activity triggers trigeminal meningeal afferents in a migraine model. Nat Med 8(2): 136–4270. Boska MD, Welch KM, Barker PB, Nelson JA, Schultz L. 2002. Contrasts in cortical magnesium, phospholipid and energy metabolism between migraine syndromes. Neurology 58(8): 1227–1233. Bowyer SM, Aurora SK, Moran JE, Tepley N, Welch KMA. 2001. Magnetoencephalographic fields from patients with spontaneous and induced migraine aura. Ann Neurol 50(5): 582–587. Brooks WM, Welch KMA, Jung RE, Friedman SD, Stidley CA, Rozell I CL. 2003. H-MRS evidence of a mitochondrial disorder in migraine. Cephalagia; Proceeding IHC 2003. Cao Y, Aurora SK, Nagesh V, Patel SC, Welch KMA. 2002. Functional MRI-BOLD of brainstem structures during visually triggered migraine. Neurology 59: 72–78. Cao Y, Welch KM, Aurora S, Vikingstad EM. 1999. Functional MRI-BOLD of visually triggered headache in patients with migraine. Arch Neurol 56(5): 548–554. Choudhuri R, Cui L, Yong C, Bowyer S, Klein RM, Welch KM, Berman NE. 2002. Cortical spreading depression and gene regulation: relevance to migraine. Ann Neurol 51(4): 499–506. Cutrer FM, Sorensen AG, Weisskoff RM, et al. 1998. Perfusionweighted imaging defects during spontaneous migrainous aura. Ann Neurol 43: 25–31. Goadsby PJ. 2002. Neurovascular headache and a midbrain vascular malformation-evidence for a role of the brainstem in chronic migraine. Cephalalgia 22.

Goadsby PJ, Hoskin KL. 1996. Inhibition of trigeminal neurons by intravenous administration of the serotonin (5HT)1B/D receptor agonist zolmitriptan (311C90): are brain stem sites a therapeutic target in migraine? Pain 67: 355–359. Haas DC, Kent PF, Friedman DI. 1993. Headache caused by a single lesion of multiple sclerosis in the periaqueductal gray area. Headache 33: 452–455. Hadjikhani N, Sanchez del Rio M, Wu O, et al. 2001. Mechanisms of migraine aura revealed by functional MRI in human visual cortex. Proc Natl Acad Sci USA 98: 4687–4692. Hagen K, Stovner LJ, Asberg A, Thorstensen K, Bjerve KS, Hveem K. 2002. High headache prevalence among women with hemachromatosis: the Nord-Trondelag Health Study. Ann Neurol 51: 786–789. Huang J, Cooper TG, Satana B, Kaufman DI, Cao Y. 2003. Visual distortion provoked by a stimulus in migraine associated with hyperneuronal activity. Headache 43: 664–671. Knight YE, Bartsch T, Kaube H, Goadsby PJ. 2002. P/Q-type calcium channel blockade in the PAG facilitates trigeminal nociception: a functional genetic link for migraine? J Neurosci (in press). Knight YE, Goadsby PJ. 2001. The periaqueductal gray matter modulates trigeminovascular input: a role in migraine? Neuroscience 106: 793–800. Lauritzen M. 1984. Long-lasting reduction of cortical blood flow of the rat brain after spreading depression with preserved autoregulation and impaired CO2 response. J Cereb Blood Flow Metab 4: 546–554. Lauritzen M, Jorgensen MB, Diemer NH, Gjedde A, Hansen AJ. 1982. Persistent oligaemia of rat cerebral cortex in the wake of spreading depression. Ann Neurol 12: 469–474. Leao AAP. 1944a. Spreading depression of activity in cerebral cortex. J Neurophysiol 7: 359–390. Leao AAP. 1944b. Pial circulation and spreading activity in the cerebral cortex. J Neurophysiol 7: 391–396. Moskowitz MA, Nozaki K, Kraig RP. 1993. Neocortical spreading depression provokes the expression of C-fos protein-like immunoreactivity within the trigeminal nucleus caudalis via trigeminovascular mechanisms. J Neurosci 13: 1167–1177. Mulleners WM, Chronicle EP, Vredeveld JW, Koehler PJ. 2002. Visual cortex excitability in migraine before and after valproate prophylaxis: a pilot study using TMS. Eur J Neurol 9(1): 35–40. Olesen J. 1991. Cerebral and extracranial circulatory disturbances in migraine: pathophysiological implications. Cerebrovasc Brain Metab Rev 3: 1–28. Olesen J, Larsen B, Lauritzen M. 1981. Focal hyperemia followed by spreading oligemia and impaired activation of rCBF in classic migraine. Ann Neurol 9: 344–352. Ophoff RA, Terwindt GM, Vergouwe MN, et al. 1996. Familial hemiplegic migraine and episodic ataxia type-2 are caused

273

274

K. Michael Welch

by mutations in the Ca2+ channel gene CACNLA4. Cell 87: 543–552. Peikert A, Wilimzig C, Kohne-Volland R. 1996. Prophylaxis of migraine with oral magnesium: results from a prospective multi-center, placebo-controlled and double blind randomized study. Cephalalgia 1996;16: 257–263. Raskin NH, Hosobuchi Y, Lamb S. 1987. Headache may arise from perturbation of brain. Headache 27: 416–420. Russell MB, Olesen J. 1996. A nosographic analysis of the migraine aura in a general population. Brain 119: 355–361. Russell MB, Rassmussen BK, Fenger K, Olesen J. 1996. Migraine without aura and migraine with aura are distinct clinical entities: a study of four hundred and eighty-four male and female migraineurs from the general population. Cephalalgia 16: 239–245. Stewart WF, Sechter A, Rasmussen BK. 1994. Migraine prevalence. A review of population-based studies. Neurology 44: S17–S23. Ter Horst GJ, Meijler WJ, Korf J, Kemper RH. 2001. Trigeminal nociception-induced cerebral Fos expression in the conscious rat. Cephalalgia 21(10): 963–975. Veloso F, Kumar K, Toth C. 1998. Headache secondary to deep brain implantation. Headache 38: 507–515.

Weiller C, May A, Limmroth V, et al. 1995. Brain stem activation in spontaneous human migraine attacks. Nat Med 1: 658–660. Welch KM, Ramadan NM. 1995. Mitochondria, magnesium and migraine. J Neurol Sci 134: 9–14. Welch KMA, D’Andrea G, Tepley N, Barkeley GL, Ramadan NM. 1990. The concept of migraine as a state of central neuronal hyperexcitability. Headache 8: 817–828. Welch KMA, Levine SR, D’Andrea G, Schultz LR, Helpern JA. 1989. Preliminary observations on brain energy metabolism in migraine studied by in vivo phosphorus 31 NMR spectroscopy. Neurology 39: 538–554. Welch KMA, Nagesh V, Aurora SK, Gelman N. 2001. Periaqueductal gray matter dysfunction in migraine: cause or the burden of illness? Headache 41: 629–637. Woods RP, Iacoboni M, Mazziotta JC. 1994. Bilateral spreading cerebral hypoperfusion during spontaneous migraine headache. N Engl J Med 331: 1689–1692. Young WB, Oshinsky ML, Shechter AL, Gebeline-Myers C, Bradley KC, Wassermann EM. 2004. Consecutive transcranial magnetic stimulation: phosphene thresholds in migraineurs and controls. Headache 44(2): 131–135.

Imaging migraine pathogenesis

Case Study 17.1 Stroke or migraine? An MR perfusion study Doris Lin, M.D. Ph.D., Philippe Gailloud, M.D. and Peter Barker, D.Phil, Johns Hopkins University School of Medicine, Baltimore History 46-year-old male with episodic right homonymous hemianopia, diplopia, vertigo, nausea, and headache.

T2

Technique Conventional, diffusion- and perfusion-weighted MRI, and digital subtraction angiography (DSA). Imaging findings T2-MRI shows decreased signal in the left occipital subcortical white matter. Diffusion weighted imaging (DWI) is normal. Perfusion weighted imaging (PWI) shows slightly shortened MTT and increased regional CBV, indicating hyperperfusion, confirmed by DSA which showed slightly early opacification of the parietal-occipital and calcarine arteries, and a prominent capillary blush in the left occipital region. There was also mild meningeal and parenchymal enhancement in the same region (not shown). Follow-up MRI was normal.

At presentation DWI

Follow-up (6 weeks)

CBV

MTT

Discussion 30 ΔR2 (s)

Vertebrobasilar insufficiency or vertebral artery dissection were suspected clinically – DWI and PWI ruled out these diagnoses. PWI showed mild occipital hyperperfusion, more consistent with migraine. CBF changes in migraine are complex (Oleson, Sanchez del Rio). Initial left occipital T2-hypointensity is presumably due to increased deoxyhemoglobin and/or CBV.

*

20 10 0

10

20 30 Time (s)

40

50

Key points DWI and PWI may help distinguish stroke from migraine. In migraine, perfusion may be increased or decreased depending on the timing of the scan relative to symptom onset. L CCA References Olesen J, Friberg L, Olsen TS, Iversen HK, Lassen NA, Andersen AR, Karle A. 1990. Timing and topography of cerebral blood flow, aura, and headache during migraine attacks. Ann Neurol 28(6): 791–798. Sanchez del Rio M, Bakker D, Wu O, et al. 1999. Perfusion weighted imaging during migraine: spontaneous visual aura and headache. Cephalalgia 19(8): 701–707.

275

Section 3 Adult neoplasia

18

Adult neoplasia: overview Tom Mikkelsen Hermelin Brain Tumor Center, Henry Ford Hospital, Detroit, MI, USA

Brain tumor incidence and outcome Malignant gliomas, including the anaplastic astrocytoma (AA) and glioblastoma multiforme (GBM), are the most common primary brain tumors, and occur at a rate of approximately 6.08/100,000 individuals annually within the US, an annual incidence of 17,500 cases (CBTRUS, 2000). Current treatment options include surgery, radiation therapy (RT) and chemotherapy. Unfortunately, prognosis remains extremely poor and the median survival of 12 months for GBM has not changed appreciably over the last several decades (Walker et al., 1980). Limitations to therapy include both the infiltrative nature and prominent angiogenesis of AA and GBM.

Pathological patterns of infiltration of peritumoral brain Gliomas in general, and gliomas that are more anaplastic, in particular, infiltrate and spread great distances in the brain (Mikkelsen and Edvardsen, 1995; Mikkelsen and Rosenblum, 1995). Regional infiltration during tumor progression has been most strikingly shown in the whole-mount studies of Scherer and Burger (Scherer, 1940; Giangasporo and Berger, 1983; Burger and Kleihues, 1989), where glioblastomas have a central area of necrosis, a highly vascularized cellular rim of tumor and a peripheral zone of infiltrating cells. Infiltration occurs along white matter (WM) tracts, around nerve cells, beneath the pia and, prominently, along angiogenic blood

vessels. Studies have shown that tumor cells have migrated from the primary site of malignant gliomas, resulting in the almost inevitable local recurrence and tumor progression seen clinically (Burger et al., 1983; Daumas-Duport et al., 1987). Recurrence of human gliomas following surgery and radiation is most commonly seen in the margin adjacent to the initial tumor where leaking tumor neovasculature is permeable to imaging contrast agents, but may also be remote (Hochberg and Pruitt, 1980; Bashir et al., 1988). Angiogenesis is quantitatively most prominent in glioblastoma compared to malignancies elsewhere in the body (Brem et al., 1972), and the patterns of growth of invading glioma and angio/vasculogenesis suggest that these processes are fundamentally related. Angiogenesis is new capillary formation that sprouts from pre-existing blood vessels (Carmeliet, 2000). It is an essential component of tumor progression in which neovasculature facilitates tumor expansion beyond 2 mm (Folkman, 1990; Mikkelsen and Rosenblum, 1995). Events in angiogenesis include the initial phase of basement membrane degradation, cell migration, matrix invasion, endothelial cell proliferation, penetration of stroma and sprouting, and capillary branch and lumen formation. Following this, migration ceases, cell division is inhibited, the basement membrane is reconstituted and periocytes attach. Soluble factors that include stimulators and inhibitors are produced by both tumor and normal cells (Folkman, 1995). Stimulators play critical roles in the activation phase and include vascular endothelial growth factor (VEGF), basic fibroblast growth factor 279

280

Tom Mikkelsen

(bFGF, FGF-2 (Klagsburn, 1989)), and others (Koch et al., 1992). Inhibitors counterbalance the stimulator function to suppress the angiogenic response. These include angiostatin, endostatin, interferon, and thrombospondins 1 and 2 (Brower, 2000). Each of these inhibitors acts through receptors and a downstream signal transduction pathway. Initiation and maintenance of tumor neovascularization presumably requires that there be an overbalance of stimulator signal.

Imaging endpoints in brain tumor trials MR imaging (MRI) measurements of brain tumor response to therapy have not typically correlated well with survival. Measuring the product of the largest perpendicular diameters of the gadolinium dimeglumine gadopentetate (Gd-DTPA)-enhancing lesion is the standard approach currently used in most clinical trials (MacDonald et al., 1990). Tumor response is based on the percentage change in the visual metric and is categorized as complete response, partial response, stable disease, or progressive disease; however, the visual metric has a large variation between observers, particularly for difficult cases where the tumor margins are not well delineated. In one study, no consensus was found for the tumor response measurement in 30% of the cases examined (Clarke et al., 1998). Automated methods defining tumor response compare well to ground truth techniques with a correlation coefficient of 0.96 (Vaidyanathan et al., 1997; Velthuizen et al., 1999). Part of the difficulty in the assessment of imaging studies lies in the fact that brain tumors may have no distinct border on MRI or other imaging modalities, making the process of tumor segmentation extremely difficult. Conventional MRI techniques (including Gd contrastenhanced studies) are not specific for the differentiation of tumor, necrosis and edema. In particular, lack of contrast enhancement does not necessarily imply an absence of tumor cells. It is expected that physiological imaging methods that are sensitive to tissue characteristics, such as metabolism and perfusion (e.g. microvasculature), may improve the assessment of these lesions in response to therapy. They may also allow dose-titration if the molecular target

can actually be visualized in the course of the clinical therapy.

Assessing response to angiogenic therapy for glioma In the case of targeted molecular therapies, host–responder characteristics are not typically well understood. Conventional dose-finding in Phase 1 trials relies on assessment of toxicity to define the maximum tolerated dose, the dose at which efficacy trials typically begin, a model that may fit cytotoxic agents better than those aimed towards specific molecular targets. Furthermore, the ideal schedule of administration of this broad class of agents has also not been well established, and may differ dramatically between agents that seek to induce apoptosis of endothelial cells and agents that attempt to prevent their invasion of tumors in the course of forming new vessels, for example. It follows that dose-finding and administration schedules should aim at adequately blocking the molecular target, but, unfortunately, no adequate measure of biological efficacy of such agents has yet emerged. These issues will require that new standards for monitoring therapeutic response be developed, implemented, and validated. This makes the endpoint of conventional Phase 2 trials, that of objective response, problematic. Phase 2 endpoints, such as time-to-tumor progression and disease stabilization, have been proposed as alternatives in novel clinical trial designs. The other biological fact that must be addressed in drug development and clinical trial design, is the intricate network of signaling pathways with layers of functional redundancy, such as is seen in the angiogenic cascade, that has the potential to bypass specific molecular therapeutic blockade. While multiple pathway targeting and combination strategies may be expected to enhance clinical outcome, this approach does tend to challenge our precepts of clinical trial design in that it becomes difficult to define which is responsible for response. From a regulatory point of view in drug development, this is problematic. Alternative strategies intended to identify predictors of ultimate outcomes like survival will be critical in the assessment of agents developed as

Adult neoplasia: overview

clinical therapeutics. Surrogate endpoints will be required for the definition of biologically effective concentrations and schedules of drug administration (dose-finding) in addition to the assessment of efficacy, presumably on the vasculature of the tumor. The characteristics of an ideal surrogate marker of drug effectiveness were described by Prentice in 1989 as a response variable for which a test of the null hypothesis of no relationship to the treatment group under comparison is also a valid test of the corresponding null hypothesis based on the true endpoint (i.e. survival). Attempts to identify meaningful correlative surrogate endpoints in the recent clinical trials of endostatin (Herbst et al., 2002) failed to validate any of the numerous variables examined, including growth factor levels, such as VEGF. Alternatives to biological endpoints are imaging endpoints, which have been considered to be definitive as surrogates for long-term efficacy in glioma trials. MRI measurements of brain tumor response to therapy, however, have not typically correlated well with survival. Measuring the product of the largest perpendicular diameters of the Gd-DTPA-enhancing lesion is the standard approach currently used in most clinical trials (MacDonald et al., 1990). However, as discussed above, regions of contrast enhancement often correlate poorly with actual tumor growth, and so alternative imaging approaches based on function and metabolism may be more promising for correlation with long-term outcome measures. Three imaging approaches to developing endpoints for angiogenesis inhibitors can be defined. Anatomic imaging, as currently employed has attempted to define tumor margins for the assessment of tumor shrinkage, albeit with limited predictive ability as discussed above. A more direct approach to monitoring treatments targeting specific molecules is by the imaging of these same specific molecules. For example, targeted imaging of integrin v3 (an angiogenesis receptor highly expressed in glioblastomas) has been demonstrated in animals using several techniques. More directly relevant for tumor angiogenesis assessment, physiological imaging uses several techniques to assess the state of the vasculature and allows temporal resolution of this dynamic process. Treatment approaches which target angiogenesis have been described as “normalizing”

tumor vasculature (Jain, 2001) using such imaging techniques and are strongly correlated with efficacy (Gossmann et al., 2002), at least in animal studies. As mentioned above, the blood vessels of a tumor are abnormal in the way that they form and so their physiological characteristics are not identical to normal vessels. As a result, tumor microcirculation differs profoundly from that of normal brain in three ways: (a) flow characteristics and blood volume of the microvasculature; (b) microvascular permeability; and (c) increased fractional volume of extravascular/extracellular space (EES) or interstitial volume fraction (IVF). Tumor neovessels have been shown to deviate markedly from normal hierarchical branching patterns and to contain gaps in which tumor cells lack close contact with perfusing vessels (Carmeliet, 2000). These characteristics lead to blood flow that is both spatially and temporally more heterogeneous than normal. Tumors also differ markedly from the surrounding brain in the permeability (permeability/ surface (PS) area or PS product) of their capillaries, which in turn changes the rules governing the transfer of compounds between blood and tumor tissue. Another abnormality, the marked alteration in relative volumes of major tissue compartments (vascular, intracellular, and EES/IVF), specifically affects the trapping and clearance of agents in tumors. Gliomas differ from normal tissue in the blood flow and volume characteristics of their microvessels, in microvascular permeability, and in increased fractional volume of the EES/IVF compartment. Methods that measure blood volume and flow are sensitive to the first of these, but methods that measure uptake or clearance of contrast media from tumor are sensitive to all three. Clinical studies using MR methods for measuring the uptake and/or clearance of contrast agent in tumors have demonstrated the utility of these parameters for predicting or assessing response to therapy in tumors of bladder, breast, brain, cervix, and bone (Barentsz et al., 1999; Hawighorst et al., 1999; Knopp et al., 1999; Mayr et al., 1999). PS product appears to be a validated measure of tumor responsiveness to antivascular agents (Beauregard et al., 2001). Relative cerebral blood volumes (rCBV) and relative cerebral blood flow (rCBF) can be estimated from dynamic contrast-enhancement MRI from the first

281

282

Tom Mikkelsen

pass of a bolus of contrast agent through the microcirculation. The techniques for performing and quantifying perfusion MRI have been discussed in detail in Chapters 7–9. Usually, rCBV is the parameter most often chosen as an index related to tumor angiogenesis. Several studies have generally shown increasing levels of rCBV with increasing tumor grade. It should be remembered that in regions of blood–brain barrier (BBB) breakdown, standard analysis techniques will not accurately measure rCBV from Gd bolus data, so appropriate alternative modeling or experimental approaches need to be used. These are described in Chapters 7–9. Alternative techniques combine this approach with arterial spin-labeled (ASL) contrast, where PaCO2 reactivity of microvasculature is assessed as a measure of vascular maturation. Another approach is to use large molecular weight contrast agents that remain within the blood pool, for instance superparamagnetic, ultra small particulates of iron oxides (USPIO). Pilot studies in primary and metastatic brain tumors showed readily detectable changes in signal intensity on T2-weighted spin-echo (SE) images. Unlike the pattern of enhancement with a Gd chelate, which occurs immediately and decreases within hours, the pattern of enhancement with the USPIO occurs gradually, with a peak at 24 h, owing to its prolonged intravascular half-life. Several comparative trials in animals have supported its use in dynamic contrast-enhanced MRI, and Phase 2 clinical trials have demonstrated safety and feasibility. Whether these improved agents and techniques, which are intended to accurately reflect the neovascular tumor burden, are actually predictive of response and therefore good surrogates of long-term outcomes, such as survival, will depend on validation imaging trials, which must be carried out in parallel to therapy. A single clinical study described the utility of rCBV mapping in the routine follow-up of brain tumors (Wong et al., 1998). Tumor progression was detected by rCBV map earlier than on conventional MRI in 32% of the studies ( 4.5 months; P  0.01); earlier than 201Tl-single photon emission computed tomography (SPECT) in 63% ( 4.5 months; P  0.01), and earlier than clin-ical assessment in 55% ( 6 months; P  0.01). rCBV mapping proved very sensitive to small regional changes, unlike functional imaging, such as positron emission tomography (PET) or

SPECT. High CBV can be found in non-Gd-enhancing tumors and heterogeneity is typically seen in highgrade tumors. Using a normalized rCBV cutoff ratio of 1.5, biopsy studies have shown high confidence for detecting high-grade tumor and earlier evidence of tumor progression (Cha et al., 2000).

Use of imaging endpoints in drug development Jain has published an intriguing concept relevant for the action of angiogenesis inhibiting agents. He describes animal models using window preparations of tumor vascularization and their effects in normalizing tumor vasculature. Tumor vasculature is described as chaotic and irregular in its organization. The initial sign of drug efficacy on this target vascular bed is, in fact, an improved perfusion as neovessels are pruned and the efficiency of the vascular bed improves to resemble a normal vascular bed rather than tumor (Jain, 2001). This may, paradoxically, ultimately result in improved efficacy against tumor as the improved perfusion improves the delivery of angio-inhibitors, which then further restrict vessels and eventually restrict tumor growth. The characterization of vascular parameters by non-invasive MRI techniques, then, may not simply show reduced perfusion, as one might imagine the end result to be, but may show paradoxical improvements in perfusion parameters with therapy. Ultimately, though, one might hypothesize that this would be followed by a reduction in the tumor vascular bed with efficacious agents.

Proton MR spectroscopic imaging Several correlative histopathological studies have shown that the contrast-enhancing margin is not representative of the lesion boundary (Burger et al., 1983; Daumas-Duport et al., 1987). Therefore, there has been interest in using other physiological imaging techniques, such as proton MR spectroscopic imaging (MRSI) to evaluate tumor borders and disease burden (Croteau et al., 2001). Using directed surgical biopsies, one study sought to determine a correlation between different proton MRSI metabolic ratios and the degree

Adult neoplasia: overview

of tumor infiltration in a series of 31 untreated patients with low- and high-grade gliomas. With 247 tissue samples and 307 observations, choline (Cho)containing compounds using contralateral creatine (Cr) and Cho for normalization or ipsilateral Nacetylaspartate (NAA) appeared to correlate best with the degree of tumor infiltration, regardless of tumor histological grade. Our preliminary results show that MRSI seems more accurate than conventional MRI in defining indistinct tumor boundaries and quantifying the degree of tumor infiltration; this suggests that MRSI could play a role in the selection of stereotactic biopsy sites, especially in non-enhancing tumors, but should not be used as a substitute for histopathological diagnosis. In addition, it may be possible to plan surgical resection using this technique. This assessment of residual disease after surgery is another important potential application, particularly for lowgrade glioma or non-enhancing tumor components in high-grade glioma, since studies using other imaging parameters have suggested improved prognosis in patients who have gross total resections or imagingcomplete resections, no matter how they are defined (Keles et al., 2001). Contrast enhancement may not be the optimal parameter for tumor response assessment in the era of cytostatic agents interfering with proliferation, invasion, angiogenesis, and differentiation processes. Serial monitoring using MRSI appears to be a promising tool for tumor response assessment with these novel non-cytotoxic agents and needs to be compared with conventional contrast-enhancement measurements. Lastly, MRSI would theoretically be a more appropriate tool for target delineation for glioma radiation treatment planning; however, several series using modern radiation techniques with 3D conformal planning with or without dose-escalation have shown that the majority of recurrences still occur in-field and not remotely or at the margin (Hochberg and Pruitt, 1980; Klagsburn, 1989; Brower, 2000). In this setting, MRSI appears unlikely to improve local control or change the pattern of recurrence of fractionated RT unless it proves to be more specific than conventional MRI. This technique would be particularly useful if higher doses appeared to be better tolerated by minimizing inclusion of normal brain parenchyma. One study also investigated the use of MRSI in focal radiation planning, such as for

stereotactic radiosurgery or brachytherapy (Mikkelsen and Edvardsen, 1995). These results support the conclusion that 1H-MRSI accurately reflects glioma tumor burden and microscopic anatomy. This has important diagnostic and therapeutic implications for more accurately assessing the burden of disease in glioma as well as for planning and assessing response to therapy. The second major application of proton MRSI, apart from the assessment of the burden of disease, is in grading and diagnosis. Several investigators have used statistical correlation with primary diagnosis and predictive modeling to suggest that MRSI is able to define lesion type and grade (Preul et al., 1996). Malignant degeneration in low-grade neoplasms of the brain has also been detected using MRSI (Tedeschi, 1997). Recently, an intraoperative biopsy correlation method was used to define diagnostic imaging parameters capable of discriminating recurrent tumor from radiation necrosis (Rock et al., 2002). Twenty-seven patients were studied who had been treated previously with surgery, radiotherapy, and chemotherapy and re-operated for clinical and/or radiographical signs that caused suspicion for recurrent disease. Tissues were categorized into four groups: normal, pure tumor, mixed tumor with necrosis, and pure radiation necrosis. Analysis was performed on 99 proton MRSI observations to determine whether metabolite ratios varied according to tissue category. Several ratios (in particular the Cho signal normalized to the contralateral hemisphere) were found to distinguish pure tumor from pure necrosis. However, no values suggested that mixed specimens could be distinguished in a statistically significant way from either pure tumor or pure necrosis. Therefore, further work is needed to improve the discrimination of mixed cases (which are the most commonly encountered clinically) from pure necrosis or pure recurrence. A third use for MRSI is in the choice of site within a lesion for biopsy and use in image-guided therapy, including definition of radiation ports. The suggestion from these early data is that maximal efficacy from local field RT will not be achieved unless the spectroscopic abnormality, which frequently extends beyond the conventional contrast-enhancing

283

284

Tom Mikkelsen

lesion (Nelson, 2001), is included within the treatment ports.

Diffusion MRI Diffusion MRI has recently been suggested as being an early predictor of tumor response (Chenevert, 2000). It appears that there is an inverse correlation between glioma cellular density and apparent diffusion coefficient (ADC), and that early response to therapy may be associated with an increase in ADC. These observations suggest that early changes in diffusion parameters predict long-term response to therapy. While compelling, these results are observed in relatively few patients, and further studies are needed to confirm the value of ADC in brain tumor diagnosis and response to therapy. In addition, the newly developed technique of brain fiber tracking based on diffusion-tensor imaging (DTI) promises to add important information for the pre-surgical evaluation of brain tumors. For instance, information on key white matter tract locations may help reduce morbidity associated with tumor resection, and also provide prognostic information (e.g. if the tumor encompasses or displaces particular tracts). Information of this type from physiological MR techniques could also be combined in the future with data from functional MRI and/or PET to assist in diagnosis of and pre-surgical planning for brain neoplasms. In summary, the implementation of imaging endpoints in clinical drug development, which are directed as closely as possible to the mechanism of action of the agent under investigation, ideally to the target molecule in question, are desirable. Indeed, without such surrogate markers, the process of dose-finding and efficacy assessment has proven difficult, especially with a class of cytostatic agents, such as angiogenesis antagonists, for which conventional dose-escalation schemes, which seek to identify maximal tolerated doses, are irrelevant. While significant progress has been made in our understanding of the molecular cascade of events involved in the process of glioma angiogenesis, progress must be made in the development of surrogate markers for therapeutic efficacy. Only with validated surrogates will it be possible to critically and

objectively measure endpoints for clinical trial design, including dosing and scheduling, as well as for efficacy testing. Molecular therapy trials can only capitalize on the growing knowledge of molecular tumor biology when deployed together with biological markers and biological imaging studies.

REFERENCES Aronen HJ, Gazit IE, Louis DN, Buchbinder BR, Pardo FS, Weisskoff RM, Harsh GR, Cosgrove GR, Halpern EF, Hochberg FH. 1994. Cerebral blood volume maps of gliomas: comparison with tumor grade and histologic findings. Radiology 191: 41–51. Barentsz JO, Engelbrecht M, Jager GJ, Witjes JA, de LaRosette J, van der Sanden BPJ, Huisman H-J, Heerschap. 1999. Fast dynamic gadolinium enhanced MR imaging of urinary bladder and prostate cancer. J Magn Reson Imaging 10: 295–304. Bashir R, Hochberg F, Oot R. 1988. Regrowth patterns of glioblastoma multiforme related to planning of interstitial brachytherapy radiation fields. Neurosurgery 23: 27–30. Beauregard DA, Hill SA, Chaplin DJ, Brindle KM. 2001. The susceptibility of tumors to the antivascular drug combretastatin A4 phosphate correlates with vascular permeability. Cancer Res 61: 6811–6815. Bello L, Francolini M, Marthyn M, Zhang P, Carroll J, Nikas RS, Strasser DC, Villani JF, Cheresh DA, Black P. 2001. Alpha(v)beta3 and alpha(v)beta5 integrin expression in glioma periphery. Neurosurgery 49: 380–390. Brem S, Cotran R, Folkman J. 1972. Tumor angiogenesis: a quantitative method for histologic grading. J Natl Cancer Inst 48: 335–347. Brooks PC, Clark RA, Cheresh DA. 1994. Requirement of vascular integrin v3 for angiogenesis. Science 264: 569–571. Brooks PC, Montgomery AM, Rosenfeld M, Reisfeld RA, Hu T, Klier G, Cheresh DA. 1994. Integrin v3 antagonists promote tumor regression by inducing apoptosis of angiogenic blood vessels. Cell 79: 1157–1164. Brooks PC, Stromblad S, Sanders LC, Von Schalschat L, Aimes RT, Stetler-Stevenson WG, Quigley JP, Cheresh DA. 1996. Localization of matrix metalloproteinase MMP-2 to the surface of invasive cells by interaction with integrin v3. Cell 85: 683–693. Brower V. 2000. Tumor angiogenesis – new drugs on the block. Nat Biotechnol 17: 963–968. Burger PC, Dubois PJ, Schold SC, Smith KR, Odom GL, Crafts DC, Giangaspero F. 1983. Computerized tomographic

Adult neoplasia: overview

and pathologic studies of the untreated, quiescent, and recurrent glioblastoma multiforme. J Neurosurg 58: 159–169. Burger PC, Kleihues P. 1989. Cytologic composition of the untreated glioblastoma with implications for evaluation and needle biopsies. Cancer 63: 2014–2023. Carmeliet P. 2000. Mechanisms of angiogenesis and arteriogenesis. Nat Med 4: 389–395. CBTRUS Year 2000 Standard Statistical Report. 1999. Central Brain Tumor Registry of the US, p. 23. Cha S, Knopp EA, Johnson G, Litt A, Glass J, Gruber ML, Lu S, Zagzag D. 2000. Dynamic contrast-enhanced T2-weighted MR imaging of recurrent malignant gliomas treated with thalidomide and carboplatin. Am J Neuroradiol 21(5): 881–890. Chenevert TL, Stegman LD, Taylor JMG, Robertson PL, Greenberg HS, Rehemtulla A, Ross BD. 2000. Diffusion magnetic resonance imaging: an early surrogate marker of therapeutic efficacy in brain tumors. J Natl Cancer Inst 92: 2029–2036. Clarke LP, Velthuizen RP, Clark M, Gaviria J, Hall L, Goldgof D, Murtagh R, Phuphanich S, Brem S. 1998. MRI measurement of brain tumor response: comparison of visual metric and automatic segmentation. Magn Reson Imaging 16: 271–279. Croteau D, Scarpace L, Hearshen D, Gutiérrez J, Rock J, Rosenblum M, Fisher J, Mikkelsen T. 2001. Correlation between magnetic resonance spectroscopy imaging and image-guided biopsies: semi-quantitative and qualitative histo-pathologic analysis of patients with untreated glioma. Neurosurgery 49: 823–829. Daumas-Duport C, Scheithauer BW, Kelly PJ. 1987. A histologic and cytologic method for the spatial definition of gliomas. Mayo Clin Proc 62: 435–449. Dennie J, Mandeville JB, Boxerman JL, Packard SD, Rosen BR, Weisskoff RM. 1998. NMR imaging of changes in vascular morphology due to tumor angiogenesis. Magn Reson Med 40: 793–799. Drake CJ, Cheresh DA, Little CD. 1995. An antagonist of integrin v3 prevents maturation of blood vessels during embryonic neo-vascularization. J Cell Sci 108: 2655–2661. Edelman RR, Runge VM, Outwater EK, Morris M, Lucas M. 1999. Safety profile of ultrasmall superparamagnetic iron oxide ferumoxtran-10: phase II clinical trial data. J Magn Reson Imaging 9: 291–294. Eliceiri BP, Cheresh DA. 1999. The role of av integrins during angiogenesis: insights into potential mechanisms of action and clinical development. J Clin Invest 103: 1227–1230. Enenstein J, Kramer RH. 1994. Confocal microscopic analysis of integrin expression on the microvasculature and its sprouts in the neonatal foreskin. J Invest Dermatol 103, 381–386. Enochs WS, Harsh G, Hochberg F, Weissleder R. 1999. Improved delineation of human brain tumors on MR

images using a long-circulating, superparamagnetic iron oxide agent. J Magn Reson Imaging 9: 228–232. Folkman J. 1990. What is the evidence that tumors are angiogenesis dependent? J Natl Cancer Inst 82: 4–6. Folkman J. 1995. Angiogenesis inhibitors generated by tumors. Mol Med 1: 120–122. Giangaspero F, Burger PC. 1983. Correlations between cytologic composition and biologic behavior in the glioblastoma multiforme: a postmortem study of 50 cases. Cancer 52: 2320–2333. Gossmann A, Helbich TH, Kuriyama N, Ostrowitzki S, Roberts TP, Shames DM, van Bruggen N, Wendland MF, Israel MA, Brasch RC. 2002. Dynamic contrast-enhanced magnetic resonance imaging as a surrogate marker of tumor response to anti-angiogenic therapy in a xenograft model of glioblastoma multiforme. J Magn Reson Imaging 15: 233–240. Graves EE, Pirzkall A, Nelson SJ, Larson D, Verhey L. 2001. Registration of magnetic resonance spectroscopic imaging to computed tomography for radiotherapy treatment planning. Med Phys 28: 2489–2496. Gupta RK, Cloughesy TF, Sinha U, Garakian J, Lazzareff J, Rubino G, Rubino L, Becker DP, Vinters HV, Alger JR. 2001. Relationships between choline magnetic resonance spectroscopy, apparent diffusion coefficient and quantitative histopathology in human glioma. J Neurooncol 50: 215–226. Gupta RK, Sinha U, Cloughesy TF, Alger JR. 1999. Inverse correlation between choline magnetic resonance spectroscopy signal intensity and the apparent diffusion coefficient in human glioma. Magn Reson Med 41: 2–7. Hawighorst H, Libicher M, Knopp MV, Moehler T, Kauffmann GW, van Kaick G. 1999. Evaluation of angiogenesis and perfusion of bone marrow lesions: role of semiquantitative and quantitative dynamic MRI. J Magn Reson Imaging 10: 286–294. Herbst RS, Mullani NA, Davis DW, Hess KR, McConkey DJ, Charnsangavej C, O’Reilly MS, Kim HW, Baker C, Roach J, Ellis LM, Rashid A, Pluda J, Bucana C, Madden TL, Tran HT, Abbruzzese JL. 2002. Development of biologic markers of response and assessment of antiangiogenic activity in a clinical trial of human recombinant endostatin. J Clin Oncol 20(18): 3804–3814. Hochberg FH, Pruitt A. 1980. Assumptions in the radiotherapy of glioblastoma. Neurology 30: 907–911. Jain, RK. 2001. Normalizing tumor vasculature with antiangiogenic therapy: a new paradigm for combination therapy. Nat Med 7: 987–989. Keles GE, Lamborn KR, Berger MS. 2001. Low-grade hemispheric gliomas in adults: a critical review of extent of resection as a factor influencing outcome. J Neurosurg 95: 735–745.

285

286

Tom Mikkelsen

Klagsburn M. 1989. The fibroblast growth factor family: structural and biological properties. Prog Growth Factor Res 1: 207–235. Knopp MV, Weiss E, Sinn HP, Matter NJ, Junkermann H, Radeleff J, Magener A, Brix G, Mayr A, Hawighorst H, Yuh WTC, Essig M, Magnotta VA, Knopp MV. 1999. MR microcirculation assessment in cervical cancer: correlations with histomorphological tumor markers and clinical outcome. J Magn Reson Imaging 10: 267–276. Koch AE, Polverini PF, Kunkel SL, Harlow LA, Dipietro LA, Elner VM, Elner SG, Strieter RM. 1992. Interleukin-8 as a macrophage-derived mediator of angiogenesis. Science 258: 1798–1801. Koshimoto Y, Yamada H, Kimura H, Maeda M, Tsuchida C, Kawamura Y, Ishii Y. 1999. Quantitative analysis of cerebral microvascular hemodynamics with T2 weighted dynamic MR imaging. J Magn Reson Imaging 9: 462–467. Lang FF, Gilbert MR, Puduvalli VK, Weinberg J, Levin VA, Yung WKA, Sawaya R, Fuller GN, Conrad CA. 2002. Toward better early-phase brain tumor clinical trials: a reappraisal of current methods and proposals for future strategies. Neurooncology 4: 268–277. Lev MH, Hochberg F. 1998. Perfusion magnetic resonance imaging to assess brain tumor responses to new therapies. Cancer Contr 5: 115–123. Lev MH, Rosen BR. 1999. Clinical applications of intracranial perfusion MR imaging. Neuroimag Clin N Am 9: 309–331. Li KL, Zhu XP, Jayson G, Carrington B, Jones A, Lawrance J, Waterton JC, Checkley D, Tessier JJL, Jackson A. 2000. Quantitative dynamic contrast enhanced MRI in tumors. A reproducible technique in the head? A reproducible technique in the breast? Proc Int Soc Magn Reson Med 8: 724. Loubeyre P, De Jaegere T, Bosmans H, Miao Y, Ni Y, Landuyt W, Marchal, G. 1999. Comparison of iron oxide particles (AMI 227) with a gadolinium complex (Gd-DOTA) in dynamic susceptibility contrast MR imagings (FLASH and EPI) for both phantom and rat brain at 1.5 Tesla. J Magn Reson Imaging 9: 447–453. Loubeyre P, De Jaegere T, Miao Y, Landuyt W, Marchal G. 1999. Assessment of iron oxide particles (AMI 227) and a gadolinium complex (Gd-DOTA) in dynamic susceptibility contrast MR imaging (FLASH and EPI) in a tumor model implanted in rats. Magn Reson Imaging 17: 627–631. Loubeyre P, Zhao S, Canet E, Abidi H, Benderbous S, Revel D. 1997. Ultrasmall superparamagnetic iron oxide particles (AMI 227) as a blood pool contrast agent for MR angiography: experimental study in rabbits. J Magn Reson Imaging 7: 958–962. Luker GD. 2002. Special conference of the American Association for Cancer Research on molecular imaging in cancer: linking biology, function, and clinical applications in vivo. Cancer Res 62: 2195–2198.

MacDonald DR, Cascino TL, Schold Jr SC, Cairncross JG. 1990. Response criteria for phase I studies of supratentorial malignant glioma. J Clin Oncol 8: 1277–1280. Mardor Y, Pfeffer R, Spiegelmann R, Roth Y, Maier SE, Nissim O, Berger R, Glicksman A, Baram J, Orenstein A, Cohen JS, Tichler T. 2003. Early detection of response to RT in patients with brain malignancies using conventional and high b-value diffusion-weighted magnetic resonance imaging. J Clin Oncol 21: 1094–1100. Mayr NA, Hawighorst H, Yuh WT, Essig M, Magnotta VA, Knopp MV. 1999. MR microcirculation assessment in cervical cancer: correlations with histomorphological tumor markers and clinical outcome. J Magn Reson Imaging 10(3): 267–276. Mcgirt MJ, Bulsara KR, Cummings TJ, New KC, Little KM, Friedman HS, Friedman AH. 2003. Prognostic value of magnetic resonance imaging-guided stereotactic biopsy in the evaluation of recurrent malignant astrocytoma compared with a lesion due to radiation effect. J Neurosurg 98: 14–20. Mikkelsen T, Edvardsen K. 1995. Invasiveness in nervous system tumors. In Cancer of the Nervous System (Eds, Black P, Loeffler JS), Blackwell Scientific Publications, Cambridge MA. Mikkelsen T, Rosenblum ML. 1995. Tumor invasiveness. In Genetics/Basic Science. Section Textbook of Gliomas, (Eds, Berger MS, Wilson CB), W.B. Saunders Company, Cambridge MA. Ostergaard L, Sorensen AG, Kwong KK, Weisskoff RM, Gyldensted C, Rosen BR. 1996. High resolution measurement of cerebral blood flow using intravascular tracer bolus passages. Part II: Experimental comparison and preliminary results. Magn Reson Med 36: 726–736. Ostergaard L, Weisskoff RM, Chesler DA, Gyldensted C, Rosen BR. 1996. High resolution measurement of cerebral blood flow using intravascular tracer bolus passages. Part I: Mathematical approach and statistical analysis. Magn Reson Med 36: 715–725. Prentice RL. 1989. Surrogate endpoints in clinical trials: definition and operational criteria. Stat Med 8(4): 431–440. Preul MC, Carmanos Z, Collins DL, Villemure JG, Leblanc R, Olivier A, Pokrupa R, Arnold DL. 1996. Accurate, noninvasive diagnosis of human brain tumors by using proton magnetic resonance spectroscopy. Nat Med 2: 323–325. Roberts HC, Roberts TPL, Bollen AW, Ley S, Brasch RC, Dillon WP. 2001. Correlation of microvascular permeability derived from dynamic contrast enhanced MR imaging with histologic grade and tumor labeling index: a study in human brain tumors. Acad Radiol 8(5): 384–391. Rock J, Scarpace L, Hearshen D, Croteau D, Gutiérrez J, Rosenblum M, Fisher J, Mikkelsen T. 2002. Correlations between magnetic resonance spectroscopy and image-guided

Adult neoplasia: overview

histopathology with special attention to radiation necrosis. Neurosurgery 51: 1–9. Rosen BR, Belliveau JW, Vevea JM, Brady TJ. 1990. Perfusion imaging NMR contrast agents. Magn Reson Med 14: 249–265. Rothenberg M, Carbone D, Johnson DH. 2003. Improving the evaluation of new cancer treatments: challenges and opportunities. Nat Rev 3: 303–309. Sanden van der BPJ, Rozijn TH, Rijken PFJW, Peters HPW, Heerschap A, Kogel van der AJ, Bovée WMMJ. 1999. GdDTPA uptake rates are linearly correlated to the perfused microvessel density and surface area in 9L-glioma rat models. Proceedings of the ISMRM, Vol. 1, pp. 22–28, ISSN: 1524–6965. Scherer HJ. 1940. The forms of growth in gliomas and their practical significance. Brain 63: 1–35. Sharma R, Saini S, Ros PR, Hahn PF, Small WC, de Lange EE, Stillman AE, Simonsen CZ, Ostergaard L, VestergaardPoulsen P, Rohl L, Bjornerud A, Gyldensted C. 1999. CBF and CBV measurements by USPIO bolus tracking: reproducibility and comparison with Gd-based values. J Magn Reson Imaging 9: 342–347. Siegal T, Rubinstein R, Tzuk-Shina T, Gomori JM. 1997. Utility of relative cerebral blood volume mapping derived from perfusion magnetic resonance imaging in the routine follow up of brain tumors. J Neurosurg 86: 22–27. Taylor JS, Tofts PS, Port R, Evelhoch JL, Knopp M, Reddick WE, Runge VM, Mayr N. 1999. MR imaging of tumor microcirculation: promise for the new millenium. J Magn Reson Imaging 10(6): 903–907. Tedeschi G, Lundbom N, Raman R, Bonavita S, Duyn JH, Alger JR, Di Chiro G. 1997. Increased choline signal coinciding

with malignant degeneration of cerebral gliomas: a serial proton magnetic resonance spectroscopy imaging study. J Neurosurg 87: 516–524. Tynninen O, Aronen HJ, Ruhala M, Paetau A, Von Boguslawski K, Salonen O, Jaaskelainen J, Paavonen T. 1999. MRI enhancement and microvascular density in gliomas. Correlation with tumor cell proliferation. Invest Radiol 34: 427–434. Vaidyanathan M, Clarke LP, Hall LO, Heidtman C, Velthuizen R, Gosche K, Phuphanich S, Wagner H, Greenberg H, Silbiger ML. 1997. Monitoring brain tumor response to therapy using MRI segmentation. Magn Reson Imaging 15: 323–334. Velthuizen RP, Hall LO, Clarke LP. 1999. Feature extraction for MRI segmentation. J Neuroimaging 9: 85–90. Walker MD, Green SB, Byar DP, Alexander Jr E, Batzdorf U, Brooks WH, Hunt WE, McCarty CS, Mahaley Jr MS, Mealey Jr J, Owens G, Ransohoff J, Robertson JT, Shapiro WR, Smith Jr KR, Wilson CB, Strike TA. 1980. Randomized comparisons of radiotherapy and nitrosoureas for the treatment of malignant glioma after surgery. N Engl J Med 303: 1323–1329. Wong ET, Jackson EF, Hess KR, Schomer DF, Hazle JD, Kyritisis AP, Jaeckle KA, Yung WK, Levin VA, Leeds NE. 1998. Correlation between dynamic MRI and outcome in patients with malignant gliomas. Neurology 50(3): 777–781. Zaharchuk G, Bogdanov Jr AA, Marota JJ, Shimizu-Sasamata M, Weisskoff RM, Kwong KK, Jenkins BG, Weissleder R, Rosen BR. 1998. Continuous assessment of perfusion by tagging including volume and water extraction (CAPTIVE): a steady-state contrast agent technique for measuring blood flow, relative blood volume fraction, and the water extraction fraction. Magn Res Med 40: 666–678.

287

19

MR spectroscopy of brain tumors in adults Jeffry R. Alger Department of Radiological Sciences, Ahmanson-Lovelace Brain Mapping Center, Brain Research Institute, Jonsson Comprehensive Cancer Center, David Geffen School of Medicine at UCLA, University of California, Los Angeles, USA

Key points • Neoplastic masses typically produce elevated choline (Cho) and creatine (Cr) signals. • Elevated Cho in neoplasms is attributed to accelerated metabolism of choline-containing compounds and appears to act as a surrogate marker of altered membrane metabolism. • Increased Cho resonances are associated with higher grade tumors, e.g. astrocytomas and anaplastic astrocytomas than lower-grade tumors. • Although Cr signal is sometimes assumed to be constant, there are significant variations in different regions of the brain and different tumors. • Lactate and mobile lipid signals increase in necrotic tumors and Cho and Cr decrease. • No study has demonstrated indisputably that proton MR spectroscopy can definitely classify tumor type or distinguish tumors from nonneoplastic lesions in individual patients. • MRS may help differentiate tumor recurrence (increased Cho) from radiation necrosis and therapeutic response (reduced Cho and Nacetyl aspartate). • MRS can be helpful for planning surgical or radiation treatment of benign tumors.

neoplasms, but provide limited prognostic and diagnostic information in individual patients. There is currently considerable interest in functional imaging techniques, including MR spectroscopy (MRS), that can be used as an adjunct to structural imaging for probing tumor physiologically, and to predict biological behavior. MRS research studies of human brain tumors began in the late 1980s (Bruhn et al., 1989; Langkowski et al., 1989). Research during the past decade has been the basis for a growing acceptance among neuroradiologists, neurosurgeons and neurooncologists that MRS is a useful clinical tool. This chapter reviews MRS studies of adult brain tumors, with an emphasis on clinical utility. Reference will be made to MRS investigations of animal brain tumor models and tumor cells when there is relevance to the clinical use of MRS in humans. This chapter will focus on proton MRS (1H-MRS) which has been used for most studies, and may be performed at the same time as structural imaging. The application of 31P-MRS to brain tumors has been explored (Arnold et al., 1990; Negendank, 1992; Barker et al., 1993), but has not developed as fully due to its limited availability and spatial resolution.

Background Introduction

288

Brain tumors are a major cause of morbidity and mortality in adults. Clinically, structural brain imaging by MR imaging (MRI) and computed tomography (CT) is widely used to evaluate intracranial

General 1H-MRS features of brain tumors Proton MRS allows major brain metabolites to be measured in vivo in defined regions of the brain (Chapters 1–3). The clinical use of 1H-MRS for brain tumor evaluation, therefore, depends on the

MR spectroscopy of brain tumors in adults

existence of a relationship between MRS-detectable metabolites and important clinical features including histological category, grade, location, malignancy, chemosensitivity, or recurrence potential. In large part, in vivo 1H-MRS studies of human brain tumors began when suitable 1H-MRS equipment and techniques became available, without clear guidance from previous brain tumor biochemistry research that demonstrated a direct relationship between the tumor metabolism and important clinical features. Much brain tumor MRS research has sought to define clinical utility, 1H-MRS continues to be used in the absence of detailed background knowledge that clearly defines how and why metabolite levels are altered in brain tumors. The majority of brain tumor 1H-MRS studies have focused on the major signals of the in vivo spectrum at echo times (TE) of 135 ms or greater, which include choline (Cho)-containing compounds, creatine (Cr) and phosphocreatine, N-acetyl aspartate (NAA), alanine and lactate (Lac). This is due to the widespread availability of single voxel localized 1H-MRS procedures that function well at longer TE, and the somewhat more limited availability of spectroscopic imaging (SI) procedures that use long TE of 60–288 ms. The more recent introduction of short TE single voxel spectroscopy and SI has allowed preliminary study of additional metabolites in human brain tumors including myoinositol, glycine, glutamate (Glu), glutamine (Gln), and various macromolecules and mobile lipids. Typical proton MR spectroscopic imaging (1HMRSI) and single volume 1H-MRS illustrations of the appearance of these signals in intracranial tumors are provided in Figures 19.1–19.3. These figures qualitatively illustrate that the 1H-MRS features produced by brain tumors differ substantially from those produced by normal brain tissue. NAA signals are not produced by brain neoplasms, although the neoplastic masses do typically produce Cho and Cr signals. The Cho signal from a neoplasm is frequently elevated relative to the Cr signal (or the NAA signal). Cho signal is also typically elevated relative to that produced by adjacent normal tissues. In many instances, a Lac signal is also detected, although it can be difficult to distinguish Lac from mobile lipid resonances, which appear in the same region of the spectrum.

289

Cho signal Cr signal NAA signal Lac  lipid signal

3

2

1 ppm

Fig. 19.1 Selected spectra (right panel) taken from a TE  272 ms MRSI study of a glioblastoma multiforme case. The location from which each spectrum shown in the right column is taken is identified by the white square on the contrast enhanced T1-weighted image shown to the left. The spectra illustrate the characteristic spectroscopic patterns shown by high grade glial tumors at long TE. Also illustrated is the variability shown by the spectra taken from various locations around the tumor’s periphery.

0

(a) Reference MRI

T2-weighted

T1-weighted post-contrast

(b) Spectroscopic images

Cho

Cr

NAA

Lac  lipids

Fig. 19.2 SI data from the same case shown in Figure 19.1. Reference MRI data are provided in (a). The relative levels of specific signals are shown in color. Red denotes the most elevated signal and blue the lowest signal. The Cho image shows the a ring shaped elevation similar to that shown in the contrast-enhanced MRI. NAA signal is decreased in the tumor and Lac is slightly elevated above baseline.

(a) Voxel location

(b) Localized spectrum Cho

Glu  Gln Lac  lipid

Cr

4.0

3.0

2.0 ppm

1.0

0.0

Fig. 19.3 Short TE (TE  25 ms) single voxel spectrum of a recurrent glioblastoma multiforme. The spectrum illustrates an elevation of Cho signal by approximately two-fold relative Cr. NAA signal is not detected. The absence of the NAA signal unmasks the wide signal having contributions from Glu and Gln. An elevation of Lac and or lipid signal relative to baseline is apparent.

MR spectroscopy of brain tumors in adults

Cho

NAA

mIG

Astrocytoma II

Parietal white matter

Cr Cho Cr mIG

NAA

Cr GIx + MM

Lac L2

GIx L1

6

5

4

3

2

L2

1

0

6

5

4

5

GIx

4

AIa GIx + MM

3

2

L2

1

mIG

Anaplastic astrocytoma

0

6

5

Cr

NAA

GIx + MM

4

3

Lac + L1

2

1

0

Lac + L1

GBM

Metastases

mIG

MM4

5

0

L2

Lac + L1

6

1

Lac + L1

Cr

6

2

Cho

Cho Meningioma

3

4

Cho Cr

3

MM

L2

mIG MM4

2

1

0

6

5

4

MM2 Cho MM1 MM3 Cr

3

2

L2

1

0

Fig. 19.4 Mean quantitative spectra obtained from various intracranial tumors at TE  30 ms. Vertical lines denote standard deviations. The figure illustrates that each of the tumors produces mean spectra having patterns that differ greatly from normal tissue, but that some tumors (e.g. metastases and glioblastoma multiforme, GBM) cannot be distinguished. The figure also illustrates that even tumors within a single diagnostic category have larger standard deviations than does normal brain tissue. Reprinted with permission from Howe et al. (2003).

Brain tumor spectra are highly variable, even in tumors of the same type and histological grade, as illustrated in Figures 19.4 and 19.5, which are taken from a recent study (Howe et al., 2003). A recent case report (Londono et al., 2003) illustrated that some astrocytomas do not show an elevated Cho signal. The underlying cause for variability in brain tumor 1 H-MRS is an area of active research. SI studies (Figures 19.1, 19.2 and Case Study 19.1) show that different regions of a single tumor produce variable signal levels. It must also be recognized that many of the cases seen in everyday clinical situations will already have been treated in some manner; treatment may also influence spectroscopic patterns.

Cho 1H-MRS signal Many clinical 1H-MRS studies of brain tumors have concentrated on Cho signals as a surrogate marker of membrane turnover. Ex vivo studies of perchloric acid extracts of tumor tissue indicate that Cho signal from intracranial tumor represents mostly free Cho, phosphocholine (PC) and glycerophosocholine (GPC) molecules (Chang et al., 1995; Miller et al., 1996). The individual signals from these three compounds cannot be resolved in vivo 1H-MRS at clinical field strengths, and are seen as a single resonance, often referred to as the “total Cho signal” or just the “Cho signal”. It is not likely that phospholipid molecules (phosphatidylcholine and

291

292

Jeffry R. Alger

NAA

Parietal white matter

Cho

Astrocytoma II

Cr NAA

Cho Cr

GIy

Cr

Lac

GIx

6

5

4

3

2

1

0

Cho

6

5

4

3

2

1

0

Cho Anaplastic astrocytoma

Meningioma

Cr GIy

GIx

6

5

4

GIx

3

2

1

0

6

5

4

Lac  L1

3

2

1

0

Cho L1

GBM

Cr

L2

AIa

Cho

Metastases

Lac

Cr

NAA

L2

L2 Lac 6

5

4

3

2

1

0

6

5

4

3

2

1

0

Fig. 19.5 Mean quantitative spectra obtained from various intracranial tumors at TE  136 ms. See Figure 19.4 for further details. Reprinted with permission from Howe et al. (2003).

phosphatidylethanolamine) make an appreciable contribution to the total Cho 1H-MRS signal because the majority of these molecules are associated with membranes and are not sufficiently mobile to permit detection. Moreover, the above studies (Chang et al., 1995; Miller et al., 1996) demonstrated that the total Cho signal measured in vivo was not directly correlated with levels of phosphatidylcholine extracted from biopsy material. A simplified rendering of the relevant Cho metabolism based on the studies by Gillies et al. (Gillies et al., 1994; Aiken and Gillies, 1996) is provided in Figure 19.6. Free Cho is transported from the blood into the cell and then phosphorylated to form PC, which can then be utilized in further anabolic reactions to synthesize phosphatidylcholine for ultimate incorporation into lipid membranes. Balancing catabolic

pathways convert the membrane constituents, phosphatidylcholine and phosphatidylethanolamine, to form PC and GPC. Phosphatidylcholine can also be formed by methyl transfer to phosphatidylethanolamine, so that PC can also be formed catabolically from phosphatidylethanolamine. In certain glioma cells the methyl transfer route accounts for very little of the total PC synthesis, suggesting that the majority of the Cho compounds within tumors composed of these cell types is obtained from free Cho via transport or is derived catabolically from phosphatidylcholine (Gillies et al., 1994). Possible causes for the elevated Cho signal seen in brain tumors include: • Extracellular and/or blood Cho may be elevated as a result of hyperperfusion and/or systemic metabolic alterations.

MR spectroscopy of brain tumors in adults

Lipid membranes

Lipid membranes

Methyl transferases Phosphatidylethanolamine

Phosphatidylcholine Phospholipases A and B

Phospholipase D

Phospholipase C

CTP-mediated synthetic enzymes

Phospholipase C

PC

CTP-mediated synthetic enzymes Phosphoethanolamine

Cho kinase

GPC

Chointracellular Cho transporters Choextracellular

Fig. 19.6 Key elements of biochemical pathways relevant to the Cho signal seen in 1H-MRS. The figure is based on the work of Gillies and colleagues (Gillies et al., 1994; Aiken and Gillies, 1996). Anabolic pathways are represented by upward-pointing arrows. Catabolic pathways are represented by downward pointing arrows. See text for further discussion.

• Intracellular Cho may be elevated as a result of enhanced transport. • Intracellular PC may be elevated as a result of accelerated phosphorylation and transport. • Intracellular Cho, PC and GPC may be elevated as a result of enhanced catabolic activity (i.e membrane break down). Hence, elevated total Cho 1H-MRS signal may be the result of either elevated anabolic activity or elevated catabolic activity of phospholipid metabolism, although it is not possible to establish which process dominates without resolution of the individual metabolites. Non-specific language such as “increased membrane turnover” or “altered membrane or phospholipid metabolism” is therefore frequently used to explain elevated Cho in cancer. In vitro 1HMRS analyses of biopsy material from brain tumors (Sabatier et al., 1999; Cheng et al., 2000; Tzika et al., 2002), have suggested that the predominant cause for the Cho signal elevation seen with in vivo 1HMRS is increased intracellular PC. Extract studies

have also showed that the PC elevation is related to malignancy; it is present at a two-fold greater level in high grade glioma compared to lower grades of glioma and to normal tissue (Usenius et al., 1994a, 1994b). This is in agreement with results from cell culture evaluations (Gillies et al., 1994; Aiken and Gillies, 1996) that show PC increases with cell proliferation rate and that glioma cells have enhanced Cho transport and phosphorylation during their growth phase. Therefore, upregulation of the anabolic pathway to PC is probably more important than acceleration of catabolic pathways.

Cr 1H-MRS signal Cr 1H-MRS signal represents Cr and phosphocreatine. This “total Cr signal” has been assumed to be constant between different cell types and to be invariant over differing conditions of oxygenation and perfusion. It is therefore, often used as an internal calibration

293

294

Jeffry R. Alger

standard, and other metabolites represented as ratios to Cr; e.g. Cho/Cr signal ratios in brain tumor. Reliable absolute quantification methods that allow tissue concentrations of individual metabolites to be derived independently are now more widely available, and more recent studies (Meyerand et al., 1999; Howe et al., 2003) report that a number of brain tumors have total Cr levels that are significantly less than normal brain tissue. Furthermore, there appears to be substantial variation in Cr within individual tumors of a specific type, and Cr levels are known to be different between different brain regions (e.g. supratentorial vs. infratentorial brain, or gray matter (GM) vs. white matter (WM)). Hence, caution should be exercised in interpreting signal ratio measures when Cr signal is used as the internal standard. NAA 1H-MRS signal The NAA signal originates predominantly from NAA, which is present only within neurons. The reduced NAA signal in brain tumor spectra is typically attributed to the displacement and destruction of neuronal tissue that accompanies neoplastic infiltration. Urenjak et al. (1993) demonstrated that cultured oligoastrocyte progenitor cells do in fact contain NAA, suggesting the possibility that some primary brain tumor tissue may express some NAA. The vast majority of in vivo 1H-MRS studies have demonstrated that NAA signal is markedly decreased in brain tumor spectra, although it is not unusual to detect a small residual NAA. When NAA is present in a brain tumor spectrum, it is difficult to exclude the possibility that a small amount of NAA-containing (normal) brain tissue has been included in the sampled volume. Lac 1H-MRS signal Classical biochemical studies of tumors have revealed derangements of glucose metabolism, leading to the prediction that tumors should display elevated Lac levels on 1H-MRS (Alger et al., 1990; Herholz et al., 1992; Kurhanewicz et al., 2000; Ricci et al., 2000). Glucose is broken down to form pyruvate by glycolysis (an anaerobic process). Normally, pyruvate is further metabolized (by oxidative

phosphorylation (OXPHOS)) to carbon dioxide and water, but in the absence of sufficient oxygen or due to disruption of OXPHOS, pyruvate and Lac accumulate, OXPHOS is frequently inefficient in tumors, which cause pyruvate and hence Lac to accumulate. The reasons for this may include: • Inadequate delivery of oxygen to sustain the aerobic pathway, due to microvascular insufficiency. • Loss of feedback regulation between the aerobic and anaerobic pathways, causing the anaerobic pathway to produce pyruvate and hence Lac at a high uncontrolled rate despite adequate oxygenation. • Mitochondrial derangements that lead to inefficient operation of the aerobic pathway. Brain tumor Lac signals were studied extensively in the early 1990s (Alger et al., 1990; Arnold et al., 1990; Segebarth et al., 1990; Demaerel et al., 1991; Herholz et al., 1992). These studies reported abnormally elevated Lac signals in many brain tumors. There can, however, be considerable variability between tumors and within tumor, which is frequently interpreted as a reflection of heterogeneous vascularization.

Mobile lipid 1H-MRS signals Lipid signals are most sensitively detected in brain tumor 1H-MRS spectra collected using (short) TE (e.g. 35 ms or less). Longer TE studies show less prominent lipid signals because their transverse relaxation times are relatively short. Mobile lipid signals have been associated with the degree of malignancy, with metastatic potential and the presence of necrosis (cf. below). In a study of intracranial tumor extracts from rats, Remy et al. (1997) demonstrated that 1H-MRS lipid signals most likely originate from lipid droplets rather than fluid microdomains within biological membranes, supporting the association between lipid 1H-MRS signals and necrosis.

Metabolism vs. necrosis Derangements of metabolism within tumor cells is not necessarily the only cause for altered spectra.

MR spectroscopy of brain tumors in adults

Many brain tumors, either untreated or following radio- or chemo-therapy exhibit developed regions of microscopic and macroscopic necrosis. Generally, necrotic tissue shows an absence or very low levels of all metabolites. Spectra obtained from regions of interest which contain both necrosis and viable tumor will show intermediate levels of metabolites reflecting the relative contributions from each tissue component. There is compelling empirical support for the concept that necrosis influences MRS. Miller et al. (1996) and Chang et al. (1995) have confirmed that tumor Cho levels measured in vivo correlated with cellular density, in addition to correlating with the water soluble extracted levels of Cho-containing compounds. Through the combination of MRS and diffusion MRI, it was demonstrated that (Gupta et al., 1999) variation in Cho signal levels is most likely closely related to local cellularity. In a follow up study, it was demonstrated that (Gupta et al., 2000) there was also a direct correlation between Cho and cell density as measured by quantitative histology. Necrosis may be responsible for lower Cho levels in the most malignant form of glioma (Negendank et al., 1996), because high grade gliomas typically have a larger necrotic fraction than do lower grade tumors. The finding of reduced Cr levels in brain tumors (cf. above) is also likely to be due to necrosis. Lac and mobile lipid signals, in contrast, appear to be associated with necrotic areas. Howe et al. (2003) demonstrated an inverse correlation between Cho signal and Lac plus lipid signal. Such a correlation is consistent with Cho being intracellular and Lac plus lipid being extracellular, so that the latter is increased in necrotic areas while the former is decreased. Whether necrosis has an influence on 1H-MRS signal depends on how the signal levels are expressed. If an absolute measure of signal intensity is used, the level of necrosis will likely have a large influence on the results. However, if the results are expressed with ratios where the internal standard (i.e. the denominator) signal (e.g. Cr) is measured in an identical volume of interest, necrosis will have less influence. However, even if ratio measures are used, necrosis can have a statistical influence. As the volume of interest becomes more necrotic, the two signal measures that are used to compute the ratio will become increasingly inaccurate due to signal loss and a constant noise

level. Hence, the more necrotic portions of tumors are expected to give highly variable values for ratio measures such as the Cho/Cr ratio.

Diagnostic uses of 1H-MRS Neoplastic vs. non-neoplastic differentiation The differentiation of neoplastic from non-neoplastic masses in the brain can be difficult using conventional structural neuroimaging techniques; therefore, there has been interest in the role of 1H-MRS for this purpose. Although spectra from focal brain lesions are often clearly abnormal, it has been difficult to distinguish brain neoplasia from other lesions using 1 H-MRS. The frequency with which non-neoplastic lesions are identified tends to be much smaller than for neoplasms in certain clinical environments (e.g. tertiary tumor referral centers, where obvious cases of infectious, inflammatory or demyelinating disease have already been excluded). Rather large studies are therefore needed to obtain meaningful numbers of non-neoplastic lesions. Furthermore, it tends to be difficult to confirm non-neoplastic lesions except by follow up neuroimaging if no biopsy is performed. Hence, verification of the presumed diagnosis may be difficult, e.g. a relatively stable mass lesion may in fact be a slow growing low grade neoplasm. At least some of the spectroscopic characteristics of neoplasia, particularly Cho signal elevation, have been attributed to cellular proliferation (Shimizu et al., 2000; Herminghaus et al., 2002). Neoplasms are not unique in showing proliferative activity; for instance, infiltration and proliferation of macrophages or microglia may occur in non-neoplastic lesions. Furthermore, in active infective lesions, proliferation of the causitive organism occurs. These proliferation processes can cause Cho signal changes similar to those caused by true neoplastic proliferation (Gupta et al., 2002). The extent of Cho signal elevation may be important. Neoplastic lesions would be expected to show substantially more elevated Cho levels compared to more benign processes. However, there is no clear definition of what level of Cho signal elevation unequivocally represents tumor.

295

296

Jeffry R. Alger

Butzen et al. (2000) evaluated a variety of data analysis methods for distinguishing neoplastic from non-neoplastic lesions with 1H-MRS. They classified lesions with a Cho/NAA ratio of greater than 1.0 were classified as neoplasm. They also developed a logistic regression (LR) model with 10 input variables from the 1H-MRS spectrum. Sensitivity and specificity for the LR model was 87% and 85% respectively, which was higher than (but was not significantly different from) the 79% and 77% achieved for the Cho/NAA ratio method. The statistical power of this study was limited by a relatively low number (13) in the non-neoplastic group. Saindane et al. (2002) evaluated the use of 1H-MRS for distinguishing tumefactive demyelinating lesions (TDL) from neoplastic lesions. They found the spectra from these two lesion types to be quite similar in appearance. Quantitatively, no significant differences in Cho/Cr ratio between the two lesion types was found. However, NAA/Cr signal ratio (although lower than normal brain) was significantly higher in the demyelinating lesions. Overall, however, the differences between the two groups was slight. Case Study 19.2 provides an additional example. The differential diagnosis of intracranial infections from neoplastic disease with 1H-MRS has been a significant area of activity during the past few years. This problem is being encountered with increasing frequency, because patients in the developing world are beginning to have access to modern MRI and CT, yet incidences of intracranial infections are still relatively high. Patients who have acquired immune deficiency are at risk of developing intracranial lesions that can be either opportunistic infections or neoplasms (e.g. lymphoma). Spectroscopic characteristics of lesions in immune compromised patients is discussed in Chapter 27. The MRS characteristics of intracranial infectious masses are distinctly different from neoplasms. Neoplastic lesions produce prominent Cho signals, while abscesses usually do not (Gupta et al., 1993; Yamagata et al., 1994; Remy et al., 1995). Furthermore, abscesses tend to produce prominent 1H-MRS signals from a variety of amino acids (Grand et al., 1999) that are not typically seen in neoplasms, most likely the products of bacterial metabolism. Hence, 1H-MRS has substantial practical utility for addressing the

diagnostic problem of intracranial neoplasm vs. infection. 1H-MRS characteristics of intracranial infections are discussed in Chapter 23. Metastatic brain tumors A solitary brain metastasis, high grade glioma or cerebral abscess may be difficult to distinguish with conventional MRI or CT. A reliable non-invasive diagnostic technique which differentiated between these, and ideally identified the type of metastasis is particularly desirable for guiding clinical management in cases where there is no known primary malignancy. Although less numerous than studies of primary brain neoplasms, there have been a number of 1H-MRS studies of brain metastases. One of these (Sijens et al., 1995) was a moderately large multicenter study, which confirmed earlier observations (Kugel et al., 1992) that metastatic lesions typically display reduced NAA/Cho signal ratios compared to normal brain tissue. However, the data did not support the use of 1H-MRS to distinguish different types of metastatic lesions, nor to distinguish metastatic lesions from primary lesions. However, it was suggested that the study supports the concept of using 1 H-MRS for excluding metastatic tumor as cause of solitary focal brain disorders that are sometimes hard to diagnose with conventional MRI or CT. A follow up study of a larger patient group (Sijens et al., 1996) also failed to show distinguishing 1H-MRS characteristics for the various types of metastatic disease. Through comparison of 1H-MRS with contrast enhanced MRI appearances, the authors developed an interpretive model in which the early stage metastatic infiltration is characterized by relatively elevated Cho signal, intermediate stage metastatic infiltration is characterized by the presence of lipid signal together with elevated Cho signal, and late stage metastatic infiltration is characterized by Lac signal and lipid signals with less intense Cho signal. These are similar to MRS characteristics of gliomas. Differentiation of primary neoplastic lesions A number of studies have investigated using 1H-MRS as a diagnostic tool for evaluating primary brain tumors prior to biopsy or therapy. This was the focus

MR spectroscopy of brain tumors in adults

of the first two MRS studies of human intracranial tumors (Bruhn et al., 1989; Langkowski et al., 1989). Interest in using 1H-MRS as a preoperative diagnostic tool continues to the present day. The intervening years have seen publication of successively larger studies and the use of more sophisticated statistical analysis methodology. Many studies have sought to associate the levels of various compounds accessible to in vivo 1H-MRS (notably Cho, NAA, Lac and lipid) with diagnostic categories of tumors. Early studies were relatively inconclusive. Gill et al. (1990) demonstrated that the Cho/Cr signal ratio differs significantly in grade IV astrocytoma (glioblastoma multiforme) from its value in the other grades. Demaerel et al. (1991) concluded that all tumors were distinct in their spectroscopic characteristics compared to normal tissues but were unable to assign specific 1H-MRS patterns to specific tumor types. Ott et al. (1993) arrived at similar conclusions regarding the non-specificity of 1 H-MRS findings across primary brain tumors, as did other studies done in the early 1990s (Kugel et al., 1992; Usenius et al., 1994a; McBride et al., 1995). A cooperative, multicenter study of untreated glioma with localized 1H-MRS involving 15 institutions that was published in 1996 (Negendank et al., 1996) provided the following findings: Glial tumors have significantly elevated Cho signals, decreased Cr signals, and decreased NAA signal compared to brain. Cho signal intensities were highest in astrocytomas and anaplastic astrocytomas (AA), and Cr signal intensities were lowest in glioblastomas. Lac signal occurred infrequently in all grades. Mobile lipids occurred in just under half of the high grade tumors with higher mean being produced by GBM. As with the earlier studies, there were large variations in 1H-MRS patterns within each glioma subtype. Tien et al. (1996) obtained findings that were generally consistent with those of Negandank et al. (1996) However, Tien et al. (1996) detected Lac signals more frequently. In a recent study involving the correlation between 1H-MRS findings and stereotactic biopsy results, Burtscher and Holtas (2001) found no significant correlation between different tumor types and 1H-MRS signal ratios. Meyerand et al. (1999) and Howe et al. (2003) addressed categorization of glial tumors with

1

H-MRS using absolute signal quantification. Howe et al. (2003) propose a classification scheme based on the sum of the absolute concentrations of alanine, Lac and mobile lipid signals and the myoinositol to Cho signal ratio. This scheme appears to distinguish tumors having different levels of malignancy, although it does not distinguish metastatic lesions from primary high grade tumors (GBM), nor does it distinguish AA from lower grade astrocytomas. Sophisticated pattern classification techniques have also been employed in hopes of reaching more accurate predictive correlations between tumor diagnosis and 1H-MRS features (Confort-Gouny et al., 1993; Hagberg et al., 1995; Preul et al., 1996; Roser et al., 1997, 1998a; Poptani et al., 1999). The largest of these studies (Preul et al., 1996) found that a linear discriminant analyses using 1H-MRS signal intensities correctly classified 104 of 105 cases. Follow up studies (De Stefano et al., 1998; Preul et al., 1998a) confirmed these results. However, this classification system appears not to have been used subsequently for the prospective evaluation of individual patients. Two recently published studies substantiate the use of the linear discriminant analysis procedure. Tate et al. (2003) used a linear discriminant analysis procedure to correctly classify 133 out of 144 cases, and further showed the method was robust to changes in acquisition parameters across centers. However, this study used only three rather broad classification categories, meningioma, low grade astrocytoma, and “aggressive tumors” which included GBM and metastatic tumors. Using a linear discriminant analysis procedure, Herminghaus et al. (2003) had an overall success rate of 86% in distinguishing primary astrocytic and oligodendroglial tumors having the four main World Health Organization histological grades and a 95% success rate in distinguishing low grade from high grade tumors. Many of the studies described above were performed using localized spectroscopy with relatively large voxel dimensions compared to the macroscopic features of tumor heterogeneity identifiable with MRI. Accordingly, many of the studies were concerned about the extent to which partial volume averaging of necrotic and cellular tumor domains influenced the results. Ricci et al. (2000) specifically addressed this issue. Their retrospective study of 1H-MRS diagnostic accuracy, emphasized that the diagnostic

297

298

Jeffry R. Alger

accuracy of 1H-MRS is dependent on voxel position, and that spectra obtained from the enhancing edge of a tumor more accurately reflect histopathology. The use of MRSI simplifies this problem because it collects voxel spectra from entire slices of tissue so that the reader may evaluate the spectroscopic characteristics at multiple characteristic locations. It is therefore anticipated that MRSI will eventually become the method of choice for the evaluation of such lesions as it becomes more widely available. One noteworthy study (De Edelenyi et al., 2000) has made explicit use of 1H-MRSI to differentially categorize individual portions of tumors. The study shows that the topographical heterogeneity within an individual tumor can be interpreted in terms of malignancy. Clinical significance of 1H-MRS tumor diagnosis The clinical rationale for using 1H-MRS as a noninvasive diagnostic tool for brain tumor has been that accurate non-invasive diagnostic categorization would obviate potentially dangerous neurosurgical procedures. Such a procedure would allow the patient to be referred directly for the appropriate external beam radiation therapy (RT) or chemotherapy without undergoing intracranial biopsy or cytoreductive neurosurgery. The foregoing discussion illustrates that 1H-MRS show a reasonable level of promise for obtaining diagnostic information without surgery, particularly if more sophisticated pattern recognition techniques are used. However, it must be borne in mind that after more than ten years of effort there is still no study that indisputably demonstrates 1H-MRS can definitely classify tumor type and distinguish tumors from non-neoplastic lesions, especially when only individual signal analysis is used. Hence, at least at the present time, 1H-MRS cannot supplant neurosurgical biopsy and conventional histopathology. This point is illustrated by the study of Adamson et al. (1998) who examined 78 patients suspected of having neoplastic mass lesions with 1 H-MRS. Forty-nine of the patients had positive findings on 1H-MRS for neoplastic lesions, yet only eight of them were referred for treatment without biopsy. Twenty-nine of the patients had negative findings on 1H-MRS for tumor yet only 15 of these patients were not treated without biopsy confirmation. This

study aptly illustrates that few clinicians are willing to proceed to therapy using a presumptive 1H-MRS diagnosis alone. Moreover, it can be argued the need for accurate non-invasive diagnosis by 1H-MRS (or any other technique) is now much diminished compared to only a few years ago. Intraoperative neuronavigation has improved substantially. Preoperative and intraoperative functional imaging and diffusion imaging techniques are available for identifying key language and motor centers within the cortex and the associated WM fibers. Imaging techniques, including 1 H-MRS, can more accurately define tumor margin than was possible only a few years ago as is described below. Together these technical advances in neurosurgery permit safer biopsies and resections. Surgical cytoreduction is the most effective means of prolonging survival for most brain tumor patients. The availability of a non-surgical 1H-MRS diagnostic procedure may therefore serve to deprive patients of one of the most effective treatments available. Accordingly, even if 1H-MRS proves to provide an accurate noninvasive diagnosis without neurosurgery, caution should be exercised in its routine application.

1

H-MRS guidance of (therapeutic) procedures

An alternative view is that rather than attempting to obviate neurosurgery, 1H-MRS should be used as a presurgical planning tool to further enhance surgical effectiveness. This potential clinical use of 1H-MRS is particularly relevant to primary brain tumors. Some astrocytomas and virtually all oligodendrogliomas do not contrast enhance, and even the high grade astrocytomas that do contrast enhance frequently show enhancement in only a portion of the tumor. Such tumors have abnormal transverse relaxation time (T2) that is difficult to distinguish from surrounding edematous but otherwise normal brain tissue. Accordingly, neither postcontrast MRI nor T2-weighted MRI is as helpful as might be wished for defining the surgical margins of such tumors. The foregoing discussion emphasizes that 1H-MRS signal patterns produced by tumors are markedly different from normal brain tissue. This leads to the hypothesis that 1H-MRS may augment conventional MRI and CT for defining tumor margin. This goal is best

MR spectroscopy of brain tumors in adults

2.00 Normal appearing MRI

1.80 1.60 1.40 NAA/nNAA

approached through the use of 1H-MRSI, although some investigators (Frahm et al., 1991) have argued that the availability of short TE high signal-to-noise ratio (SNR) localized 1H-MRS techniques may allow systematic high spatial resolution sampling of brain tumor tissue. There is now much support for the use of 1H-MRS as a preoperative planning tool. Studies illustrate that Cho signal is elevated in cellular portions of the tumor beyond the contrast enhancing margin. Burtscher and Holtas (2001) demonstrated that gliomas and lymphomas showed abnormal spectra outside the area of contrast enhancement while non-astrocytic circumscribed tumors (meningioma, pineocytoma, metastasis and germinoma) showed no pathological spectra outside the region of enhancement. Preul et al. (1998b) used 1H-MRSI for guiding stereotactic biopsy and resection. The latter study emphasized the utility of Cho signal elevation for defining recurrent tumor following RT where contrast enhancement is even less effective. Dowling et al. (2001) performed a study in which metabolite levels measured with preoperative 1H-MRSI were compared to biopsy findings obtained during image-guided resections. Representative results are provided in Figure 19.7. Biopsy locations that showed abnormally increased Cho and decreased NAA signals had histological findings consistent with tumor. On the other hand, below normal Cho and NAA signals were found to correlate with a mixed pattern of necrosis, gliosis, macrophage infiltration and tumor. These findings illustrate that 1H-MRSI can be used as a preoperative guidance tool. It may be used to identify biopsy locations that will give the greatest diagnostic yield or for identifying the location and extent of cellular tumor so that cytoreduction can be maximized in open resections. The definition of the target volume for external beam RT is analogous to surgical margin definition. Accordingly, it has also been recognized that 1H-MRSI may become a tool in external beam radiation treatment planning (Pirzkall et al., 2001). Findings supporting the use of 1H-MRS and particularly 1H-MRSI as a preoperative neurosurgical planning tool have prompted the development of methods for performing intraoperative 1H-MRS and 1H-MRSI using newly available 1.5 T MRI units that are placed with

299

Biopsies without any tumor (n  28) Biopsies with 90% tumor (n  19)

1.20 1.00 0.80 0.60 0.40 0.20 0.00 0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80 2.00 2.20 2.40 Cho/nCho

Fig. 19.7 Cho and NAA signal levels determined by 1H-MRSI prior to stereotactic biopsy. Biospy location NAA levels are expressed as a ratio relative to normal brain tissue NAA (vertical axis). Biopsy location Cho levels are expressed as a ratio relative to normal brain tissue Cho (horizontal axis). The shades in the ellipsoids show biopsy results for the sampled tissue. The figure illustrates that below normal NAA signal is present in all biopsies, but that only biopsies with substantial tumor burden show above normal Cho signal. The study illustrated that higher than normal Cho signal on spectroscopic imaging can be used to guide stereotactic biopsy procedures to locations having substantial tumor burden. Reprinted from Dowling et al. (2001) with permission.

neurosurgical operating suites (Hall et al., 2001; Liu et al., 2001).

Assessment of response to therapy Several investigations have focused on evaluating the response of intracranial tumors to therapy. Heesters et al. (1993) described 1H-MRS changes that accompanied the response of 11 inoperable gliomas to external beam RT. NAA signals were decreased within tumor and treatment did not lead to reappearance of this signal. In some but not all of the tumors, the initially elevated Cho signal decreased with treatment and this was accompanied by tumor volume reduction on MRI. Findings regarding the Lac response to therapy were similar to those for Cho. In a recent study, Tarnawski

300

Jeffry R. Alger

et al. (2002) demonstrated that survival for patients having tumors with Lac to NAA ratios of greater than 2 following RT was significantly shorter than for patients having tumors with Lac to NAA ratios of less than 2. External beam RT exposes the normal brain to sublethal, but possibly damaging, radiation doses and the clinical effects of radiation exposure may not appear for an extended period of time following exposure. Therefore, Usenius et al. (1995) sought to determine whether sublethal radiation doses could produce an NAA signal loss in brain tissue not infiltrated by tumor. They found a mean NAA signal reduction of about 35% in brain tissue that had been exposed to radiation compared to normal brain and concluded that irradiation causes neuronal damage, which is reflected in NAA signal reduction. Their finding suggests that 1HMRS could provide a means for clinically evaluating patients for adverse effects induced by radiation exposure. A number of additional studies (Kinoshita et al., 1994; Esteve et al., 1998; Waldrop et al., 1998; Walecki et al., 1999) further substantiate the concept of using 1 H-MRS signals to assess the effect of various forms of treatment on otherwise normal brain tissue. The response of intracranial glioma to chemotherapy is considerably more variable than is the case for RT. Accordingly, there is significance in developing 1H-MRS procedures to assess chemotherapuetic response in the tumor, as this would provide a objective outcome measure that could be used by the neuro-oncologist to tailor treatment. A recent study by Preul et al. (2000) assessed the sensitivity of 1H-MRS as a chemotherapy response indicator in recurrent glioma. They studied 16 patients who underwent high dose orally administered tamoxifen for recurrent glioma. They found that linear discriminant analysis using the prominent 1H-MRS signal intensities obtained at baseline prior to initiation of tamoxifen therapy could predict whether a particular patient would respond. Interestingly, they also found relatively little spectroscopic change during therapy.

Malignant transformation of low grade primary brain tumors The progression of relatively benign primary brain tumors (e.g. astrocytoma) to more malignant grades

(AA and GBM) is a well-known phenomenon. One of the goals of follow up imaging in patients who have been diagnosed with low grade primary brain tumors is to document progression. Tedeschi et al. (1997) suggests that 1H-MRS may be useful for this purpose. Their study of 27 patients who underwent multiple 1H-MRSI examinations demonstrated that the 11 cases that had clinically documented progression showed a between-study Cho signal increase of more than 45%, whereas the 16 stable cases showed a Cho signal change of less than 35%, no change, or even a decreased signal. They concluded that increased Cho signal levels coincide with malignant degeneration of gliomas and therefore may possibly be used as a supportive indicator of progression. Here it is important to emphasize that Tedeschi et al. (1997) demonstrated that a change in the Cho signal level between two subsequent 1H-MRSI studies was an indicator of progression. They do not specifically comment on the extent to which the Cho signal measured in a single study is helpful in making this determination. This underscores the importance of performing regular 1H-MRS evaluations in order to establish baseline Cho signal levels prior to a progression event. It is further important to emphasize that Tedeschi’s et al. (1997) analysis mixed cases showing low grade to high grade progression with cases of high grade recurrence. The extent to which there are differences in the spectroscopic signal behavior between these two types of malignant degeneration remains unclear.

Radiation necrosis vs. high grade neoplastic recurrence Brain tumors that have been treated by fractionated external beam RT typically show prolonged quiescence prior to a clinically defined recurrence. Tumor recurrence and radiation-induced necrosis have similar contrast-enhancing properties. They are difficult to distinguish with contrast enhanced MRI or CT. Hence, there has been interest in using 1H-MRS, for distinguishing tumor recurrence from radiation necrosis. There have been a number of relevant studies related to the differentiation of radiation necrosis from tumor recurrence. Taylor et al. (1996) followed a group of patients who had clinical and

MR spectroscopy of brain tumors in adults

MRI changes suggestive of either delayed radiation necrosis or progressive/recurrent brain tumor prior to biopsy, resection, or other confirmatory histological procedure. They hypothesized that a marked reduction of Cho, Cr and NAA signals was indicative of delayed radiation necrosis, while median-to-high Cho signals with easily visible Cr signals were indicative of progressive/recurrent tumor. They then compared 1H-MRS readings based on these hypotheses with histopathological findings. The hypothesized 1 H-MRS criteria prospectively identified five out of seven patients with active tumor, and four out of five patients with delayed radiation necrosis. Discriminant analysis suggested that the primary diagnostic information for differentiating delayed radiation necrosis from tumor lay in the Cho and Cr signals. Wald et al. (1997) performed a serial 1H-MRSI study of 12 patients with GBM following brachytherapy. The 1H-MRSI data demonstrated significant differences between suspected residual or recurrent tumor and contrast-enhancing radiation-induced necrosis. Furthermore, patients, who demonstrated subsequent clinical progression, had increased Cho signal in regions that previously appeared either normal or necrotic. Preul et al. (1998b) used 1H-MRSI in two newly symptomatic patients that had previously undergone resection and external beam RT. They compared 1H-MRSI results to histological findings obtained at subsequent surgery. They found that a high Cho signal coincided with areas of histologically proven dense tumor recurrence, while low Cho signal was present where radiation changes predominated. Lin et al. (1999) addressed the question of whether 1H-MRS could be used as an alternative or adjunct to brain biopsy in 15 patients having likely recurrent neoplastic lesions. MRS suggested the presence of recurrent tumor in seven cases, all of which were subsequently confirmed by surgical biopsy or by disease progression on MRI. 1H-MRS suggested the presence of necrosis in three patients, all of whom remained radiographically stable during the follow up period. Henry et al. (2000) used the combination of 1H-MRS and cerebral blood volume (CBV) MRI to evaluate previously treated gliomas. They found that, presumably necrotic, hypovascular regions showed low 1H-MRS signal levels of all types, while hypervascular regions that were presumably

infiltrated with tumor, showed spectra having high Cho to NAA signal ratios. Their results pointed to the combined use of rCBV imaging and MRS to identify the presence of cellular neoplastic lesions in previously treated glioma patients. Rabinov et al. (2002) used 3 T 1H-MRSI to address the issue in a group of 14 glioma patients who had been previously treated with RT. Histological evaluation of biopsy material obtained subsequent to the 1H-MRSI study provided the diagnoses in each case. They found that relatively modest Cho signal increases of 30% greater than Cr measured in normal tissue to be consistent with recurrent tumor. Each of the aforementioned studies indicate that 1 H-MRS has a particularly powerful clinical utility in evaluating whether previously treated brain tumor patients who develop new contrast enhancing lesions are experiencing neoplastic progression/recurrence or therapeutically induced necrosis. While the specific diagnosis of recurrence of tumor following RT is of value, it is equally noteworthy that the study by Tarnawski et al. (2002) indicates 1H-MRS has a clear prognostic value for predicting survival whether or not it is possible to clearly diagnosis recurrence following therapy.

Summary The body of existing literature indicates that 1H-MRS has an important clinical role to play in the management of intracranial neoplasia. It has distinct value in the following areas: (1) diagnosis without neurosurgery, (2) guidance of neurosurgical procedures, (3) targeting of focal ablative therapies, (4) providing an assessment of response to therapy and (5) detecting recurrent or progressing disease. However, 1 H-MRS is not the only technique that can be helpful in these clinically relevant areas. Other techniques such as diffusion MRI, perfusion MRI and certain nuclear medicine procedures can be valuable as well. The sole focus of 1H-MRS in this chapter is not meant to advocate that only 1H-MRS will be used in the future to address these clinical needs. The fact is that it is more likely that synergisms between the various available techniques for addressing specific clinical problems will be developed.

301

302

Jeffry R. Alger

REFERENCES Adamson AJ, Rand SD, Prost RW, Kim TA, Schultz C, Haughton VM. 1998. Focal brain lesions: effect of singlevoxel proton MR spectroscopic findings on treatment decisions. Radiology 209: 73–78. Aiken NR, Gillies RJ. 1996. Phosphomonoester metabolism as a function of cell proliferative status and exogenous precursors. Anticancer Res 16: 1393–1397. Alger JR, Frank JA, Bizzi A, Fulham MJ, DeSouza BX, Duhaney MO, Inscoe SW, Black JL, van Zijl PC, Moonen CT. 1990. Metabolism of human gliomas: assessment with H-1 MR spectroscopy and F-18 fluorodeoxyglucose PET. Radiology 177: 633–641. Arnold DL, Shoubridge EA, Villemure JG, Feindel W. 1990. Proton and phosphorus magnetic resonance spectroscopy of human astrocytomas in vivo. Preliminary observations on tumor grading. NMR Biomed 3: 184–189. Barker PB, Glickson JD, Bryan RN. 1993. In vivo magnetic resonance spectroscopy of human brain tumors. Top Magn Reson Imaging 5: 32–45. Bruhn H, Frahm J, Gyngell ML, Merboldt KD, Hanicke W, Sauter R, Hamburger C. 1989. Noninvasive differentiation of tumors with use of localized H-1 MR spectroscopy in vivo: initial experience in patients with cerebral tumors. Radiology 172: 541–548. Burtscher IM, Holtas S. 2001. Proton magnetic resonance spectroscopy in brain tumours: clinical applications. Neuroradiology 43: 345–352. Butzen J, Prost R, Chetty V, Donahue K, Neppl R, Bowen W, Li SJ, Haughton V, Mark L, Kim T, Mueller W, Meyer G, Krouwer H, Rand S. 2000. Discrimination between neoplastic and nonneoplastic brain lesions by use of proton MR spectroscopy: the limits of accuracy with a logistic regression model. Am J Neuroradiol 21: 1213–1219. Chang L, McBride D, Miller BL, Cornford M, Booth RA, Buchthal SD, Ernst TM, Jenden D. 1995. Localized in vivo 1H magnetic resonance spectroscopy and in vitro analyses of heterogeneous brain tumors. J Neuroimaging 5: 157–163. Cheng LL, Anthony DC, Comite AR, Black PM, Tzika AA, Gonzalez RG. 2000. Quantification of microheterogeneity in glioblastoma multiforme with ex vivo high-resolution magic-angle spinning (HRMAS) proton magnetic resonance spectroscopy. Neurooncol 2: 87–95. Confort-Gouny S, Vion-Dury J, Nicoli F, Dano P, Donnet A, Grazziani N, Gastaut JL, Grisoli F, Cozzone PJ. 1993. A multiparametric data analysis showing the potential of localized proton MR spectroscopy of the brain in the metabolic characterization of neurological diseases. J Neurol Sci 118: 123–133.

Demaerel P, Johannik K, Van Hecke P, Van Ongeval C, Verellen S, Marchal G, Wilms G, Plets C, Goffin J, Van Calenbergh F. 1991. Localized 1H NMR spectroscopy in fifty cases of newly diagnosed intracranial tumors. J Comput Assist Tomogr 15: 67–76. De Edelenyi FS, Rubin C, Esteve F, Grand S, Decorps M, Lefournier V, Le Bas JF, Remy C. 2000. A new approach for analyzing proton magnetic resonance spectroscopic images of brain tumors: nosologic images. Nat Med 6: 1287–1289. De Stefano N, Caramanos Z, Preul MC, Francis G, Antel JP, Arnold DL. 1998. In vivo differentiation of astrocytic brain tumors and isolated demyelinating lesions of the type seen in multiple sclerosis using 1H magnetic resonance spectroscopic imaging. Ann Neurol 44: 273–278. Dowling C, Bollen AW, Noworolski SM, McDermott MW, Barbaro NM, Day MR, Henry RG, Chang SM, Dillon WP, Nelson SJ, Vigneron DB. 2001. Preoperative proton MR spectroscopic imaging of brain tumors: correlation with histopathologic analysis of resection specimens. [Comment In: Am J Neuroradiol 2001 April 22(4): 597–8 UI: 21185634.] Am J Neuroradiol 22: 604–612. Esteve F, Rubin C, Grand S, Kolodie H, Le Bas JF. 1998. Transient metabolic changes observed with proton MR spectroscopy in normal human brain after radiation therapy. Int J Radiat Oncol Biol Phys 40: 279–286. Frahm J, Bruhn H, Hanicke W, Merboldt KD, Mursch K, Markakis E. 1991. Localized proton NMR spectroscopy of brain tumors using short-echo time STEAM sequences. J Comput Assist Tomogr 15: 915–922. Gill SS, Thomas DG, Van Bruggen N, Gadian DG, Peden CJ, Bell JD, Cox IJ, Menon DK, Iles RA, Bryant DJ. 1990. Proton MR spectroscopy of intracranial tumours: in vivo and in vitro studies. J Comput Assist Tomogr 14: 497–504. Gillies RJ, Barry JA, Ross BD. 1994. In vitro and in vivo 13C and 31 P NMR analyses of phosphocholine metabolism in rat glioma cells. Magn Reson Med 32: 310–318. Grand S, Passaro G, Ziegler A, Esteve F, Boujet C, Hoffmann D, Rubin C, Segebarth C, Decorps M, Le Bas JF, Remy C. 1999. Necrotic tumor versus brain abscess: importance of amino acids detected at 1H MR spectroscopy – initial results. Radiology 213: 785–793. Gupta RK, Cloughesy TF, Sinha U, Garakian J, Lazareff J, Rubino G, Rubino L, Becker DP, Vinters HV, Alger JR. 2000. Relationships between choline magnetic resonance spectroscopy, apparent diffusion coefficient and quantitative histopathology in human glioma. J Neurooncol 50: 215–226. Gupta RK, Husain M, Vatsal DK, Kumar R, Chawla S, Husain N. 2002. Comparative evaluation of magnetization transfer MR imaging and in-vivo proton MR spectroscopy in brain tuberculomas. Magn Reson Imaging 20: 375–381.

MR spectroscopy of brain tumors in adults

Gupta RK, Pandey R, Khan EM, Mittal P, Gujral RB, Chhabra DK. 1993. Intracranial tuberculomas: MRI signal intensity correlation with histopathology and localised proton spectroscopy. Magn Reson Imaging 11: 443–449. Gupta RK, Sinha U, Cloughesy TF, Alger JR. 1999. Inverse correlation between choline magnetic resonance spectroscopy signal intensity and the apparent diffusion coefficient in human glioma. Magn Reson Med 41: 2–7. Hagberg G, Burlina AP, Mader I, Roser W, Radue EW, Seelig J. 1995. In vivo proton MR spectroscopy of human gliomas: definition of metabolic coordinates for multi-dimensional classification. Magn Reson Med 34: 242–252. Hall WA, Martin A, Liu H, Truwit CL. 2001. Improving diagnostic yield in brain biopsy: coupling spectroscopic targeting with real-time needle placement. J Magn Reson Imaging 13: 12–15. Heesters MA, Kamman RL, Mooyaart EL, Go KG. 1993. Localized proton spectroscopy of inoperable brain gliomas. Response to radiation therapy. J Neurooncol 17: 27–35. Henry RG, Vigneron DB, Fischbein NJ, Grant PE, Day MR, Noworolski SM, Star-Lack JM, Wald LL, Dillon WP, Chang SM, Nelson SJ. 2000. Comparison of relative cerebral blood volume and proton spectroscopy in patients with treated gliomas. Am J Neuroradiol 21: 357–366. Herholz K, Heindel W, Luyten PR, denHollander JA, Pietrzyk U, Voges J, Kugel H, Friedmann G, Heiss WD. 1992. In vivo imaging of glucose consumption and lactate concentration in human gliomas. Ann Neurol 31: 319–327. Herminghaus S, Dierks T, Pilatus U, Moller-Hartmann W, Wittsack J, Marquardt G, Labisch C, Lanfermann H, Schlote W, Zanella FE. 2003. Determination of histopathological tumor grade in neuroepithelial brain tumors by using spectral pattern analysis of in vivo spectroscopic data. J Neurosurg 98: 74–81. Herminghaus S, Pilatus U, Moller-Hartmann W, Raab P, Lanfermann H, Schlote W, Zanella FE. 2002. Increased choline levels coincide with enhanced proliferative activity of human neuroepithelial brain tumors. NMR Biomed 15: 385–392. Howe FA, Barton SJ, Cudlip SA, Stubbs M, Saunders DE, Murphy M, Wilkins P, Opstad KS, Doyle VL, McLean MA, Bell BA, Griffiths JR. 2003. Metabolic profiles of human brain tumors using quantitative in vivo 1H magnetic resonance spectroscopy. Magn Reson Med 49: 223–232. Kinoshita Y, Kajiwara H, Yokota A, Koga Y. 1994. Proton magnetic resonance spectroscopy of brain tumors: an in vitro study. Neurosurgery 35: 606–613. Kugel H, Heindel W, Ernestus RI, Bunke J, du MR, Friedmann G. 1992. Human brain tumors: spectral patterns detected with localized H-1 MR spectroscopy. Radiology 183: 701–709. Kurhanewicz J, Vigneron DB, Nelson SJ. 2000. Threedimensional magnetic resonance spectroscopic imaging of brain and prostate cancer. Neoplasia 2: 166–189.

Langkowski JH, Wieland J, Bomsdorf H, Leibfritz D, Westphal M, Offermann W, Maas R. 1989. Pre-operative localized in vivo proton spectroscopy in cerebral tumors at 4.0 Tesla – first results. Magn Reson Imaging 7: 547–555. Lin A, Bluml S, Mamelak AN. 1999. Efficacy of proton magnetic resonance spectroscopy in clinical decision making for patients with suspected malignant brain tumors. J Neurooncol 45: 69–81. Liu H, Hall WA, Martin AJ, Truwit CL. 2001. An efficient chemical shift imaging scheme for magnetic resonance-guided neurosurgery. J Magn Reson Imaging 14: 1–7. Londono A, Castillo M, Armao D, Kwock L, Suzuki K. 2003. Unusual MR spectroscopic imaging pattern of an astrocytoma: lack of elevated choline and high myo-inositol and glycine levels. Am J Neuroradiol 24: 942–945. McBride DQ, Miller BL, Nikas DL, Buchthal S, Chang L, Chiang F, Booth RA. 1995. Analysis of brain tumors using 1H magnetic resonance spectroscopy. Surg Neurol 44: 137–144. Meyerand ME, Pipas JM, Mamourian A, Tosteson TD, Dunn JF. 1999. Classification of biopsy-confirmed brain tumors using single-voxel MR spectroscopy. Am J Neuroradiol 20: 117–123. Miller BL, Chang L, Booth R, Ernst T, Cornford M, Nikas D, McBride D, Jenden DJ. 1996. In vivo 1H MRS choline: correlation with in vitro chemistry/histology. Life Sci 58: 1929–1935. Negendank W. 1992. Studies of human tumors by MRS: a review. NMR Biomed 5: 303–324. Negendank WG, Sauter R, Brown TR, Evelhoch JL, Falini A, Gotsis ED, Heerschap A, Kamada K, Lee BC, Mengeot MM, Moser E, Padavic-Shaller KA, Sanders JA, Spraggins TA, Stillman AE, Terwey B, Vogl TJ, Wicklow K, Zimmerman RA. 1996. Proton magnetic resonance spectroscopy in patients with glial tumors: a multicenter study. J Neurosurg 84: 449–458. Ott D, Hennig J, Ernst T. 1993. Human brain tumors: assessment with in vivo proton MR spectroscopy. Radiology 186: 745–752. Pirzkall A, McKnight TR, Graves EE, Carol MP, Sneed PK, Wara WW, Nelson SJ, Verhey LJ, Larson DA. 2001. MRspectroscopy guided target delineation for high-grade gliomas. Int J Radiat Oncol Biol Phys 50: 915–928. Poptani H, Kaartinen J, Gupta RK, Niemitz M, Hiltunen Y, Kauppinen RA. 1999. Diagnostic assessment of brain tumours and non-neoplastic brain disorders in vivo using proton nuclear magnetic resonance spectroscopy and artificial neural networks. J Cancer Res Clin Oncol 125: 343–349. Preul MC, Caramanos Z, Collins DL, Villemure JG, Leblanc R, Olivier A, Pokrupa R, Arnold DL. 1996. Accurate, noninvasive diagnosis of human brain tumors by using proton magnetic resonance spectroscopy. Nat Med 2: 323–325.

303

304

Jeffry R. Alger

Preul MC, Caramanos Z, Leblanc R, Villemure JG, Arnold DL. 1998a. Using pattern analysis of in vivo proton MRSI data to improve the diagnosis and surgical management of patients with brain tumors. NMR Biomed 11: 192–200. Preul MC, Leblanc R, Caramanos Z, Kasrai R, Narayanan S, Arnold DL. 1998b. Magnetic resonance spectroscopy guided brain tumor resection: differentiation between recurrent glioma and radiation change in two diagnostically difficult cases. Can J Neurol Sci 25: 13–22. Preul MC, Caramanos Z, Villemure JG, Shenouda G, Leblanc R, Langleben A, Arnold DL. 2000. Using proton magnetic resonance spectroscopic imaging to predict in vivo the response of recurrent malignant gliomas to tamoxifen chemotherapy. Neurosurgery 46: 306–318. Rabinov JD, Lee PL, Barker FG, Louis DN, Harsh GR, Cosgrove GR, Chiocca EA, Thornton AF, Loeffler JS, Henson JW, Gonzalez RG. 2002. In vivo 3-T MR spectroscopy in the distinction of recurrent glioma versus radiation effects: initial experience. Radiology 225: 871–879. Remy C, Fouilhe N, Barba I, Sam-Lai E, Lahrech H, Cucurella MG, Izquierdo M, Moreno A, Ziegler A, Massarelli R, Decorps M, Arus C. 1997. Evidence that mobile lipids detected in rat brain glioma by 1H nuclear magnetic resonance correspond to lipid droplets. Cancer Res 57: 407–414. Remy C, Grand S, Lai ES, Belle V, Hoffmann D, Berger F, Esteve F, Ziegler A, Le Bas JF, Benabid AL. 1995. 1H MRS of human brain abscesses in vivo and in vitro. Magn Reson Med 34: 508–514. Ricci PE, Pitt A, Keller PJ, Coons SW, Heiserman JE. 2000. Effect of voxel position on single-voxel MR spectroscopy findings. Am J Neuroradiol 21: 367–374. Roser W, Hagberg G, Mader I, Dellas S, Seelig J, Radue EW, Steinbrich W. 1997. Assignment of glial brain tumors in humans by in vivo 1H-magnetic resonance spectroscopy and multidimensional metabolic classification. MAGMA 5: 179–183. Sabatier J, Gilard V, Malet-Martino M, Ranjeva JP, Terral C, Breil S, Delisle MB, Manelfe C, Tremoulet M, Berry I. 1999. Characterization of choline compounds with in vitro 1H magnetic resonance spectroscopy for the discrimination of primary brain tumors. Invest Radiol 34: 230–235. Saindane AM, Cha S, Law M, Xue X, Knopp EA, Zagzag D. 2002. Proton MR spectroscopy of tumefactive demyelinating lesions. Am J Neuroradiol 23: 1378–1386. Segebarth CM, Baleriaux DF, Luyten PR, den Hollander JA. 1990. Detection of metabolic heterogeneity of human intracranial tumors in vivo by 1H NMR spectroscopic imaging. Magn Reson Med 13: 62–76. Shimizu H, Kumabe T, Shirane R, Yoshimoto T. 2000. Correlation between choline level measured by proton MR spectroscopy and Ki-67 labeling index in gliomas. Am J Neuroradiol 21: 659–665.

Sijens PE, Knopp MV, Brunetti A, Wicklow K, Alfano B, Bachert P, Sanders JA, Stillman AE, Kett H, Sauter R. 1995. 1H MR spectroscopy in patients with metastatic brain tumors: a multicenter study. Magn Reson Med 33: 818–826. Sijens PE, Levendag PC, Vecht CJ, van Dijk P, Oudkerk M. 1996. 1H MR spectroscopy detection of lipids and lactate in metastatic brain tumors. NMR Biomed 9: 65–71. Tarnawski R, Sokol M, Pieniazek P, Maciejewski B, Walecki J, Miszczyk L, Krupska T. 2002. 1H-MRS in vivo predicts the early treatment outcome of postoperative radiotherapy for malignant gliomas. Int J Radiat Oncol Biol Phys 52: 1271–1276. Tate AR, Majos C, Moreno A, Howe FA, Griffiths JR, Arus C. 2003. Automated classification of short echo time in in vivo 1H brain tumor spectra: a multicenter study. Magn Reson Med 49: 29–36. Taylor JS, Langston JW, Reddick WE, Kingsley PB, Ogg RJ, Pui MH, Kun LE, Jenkins III JJ, Chen G, Ochs JJ, Sanford RA, Heideman RL. 1996. Clinical value of proton magnetic resonance spectroscopy for differentiating recurrent or residual brain tumor from delayed cerebral necrosis. Int J Radiat Oncol Biol Phys 36: 1251–1261. Tedeschi G, Lundbom N, Raman R, Bonavita S, Duyn JH, Alger JR, Di Chiro G. 1997. Increased choline signal coinciding with malignant degeneration of cerebral gliomas: a serial proton magnetic resonance spectroscopy imaging study. J Neurosurg 87: 516–524. Tien RD, Lai PH, Smith JS, Lazeyras F. 1996. Single-voxel proton brain spectroscopy exam (PROBE/SV) in patients with primary brain tumors. Am J Roentgenol 167: 201–209. Tzika AA, Cheng LL, Goumnerova L, Madsen JR, Zurakowski D, Astrakas LG, Zarifi MK, Scott RM, Anthony DC, Gonzalez RG, Black PM. 2002. Biochemical characterization of pediatric brain tumors by using in vivo and ex vivo magnetic resonance spectroscopy. J Neurosurg 96: 1023–1031. Urenjak J, Williams SR, Gadian DG, Noble M. 1993. Proton nuclear magnetic resonance spectroscopy unambiguously identifies different neural cell types. J Neurosci 13: 981–989. Usenius JP, Kauppinen RA, Vainio PA, Hernesniemi JA, Vapalahti MP, Paljarvi LA, Soimakallio S. 1994a. Quantitative metabolite patterns of human brain tumors: detection by 1H NMR spectroscopy in vivo and in vitro. J Comput Assist Tomogr 18: 705–713. Usenius T, Usenius JP, Tenhunen M, Vainio P, Johansson R, Soimakallio S, Kauppinen R. 1995. Radiation-induced changes in human brain metabolites as studied by 1H nuclear magnetic resonance spectroscopy in vivo. Int J Radiat Oncol Biol Phys 33: 719–724. Usenius JP, Vainio P, Hernesniemi J, Kauppinen RA. 1994b. Choline-containing compounds in human astrocytomas studied by 1H NMR spectroscopy in vivo and in vitro. J Neurochem 63: 1538–1543.

MR spectroscopy of brain tumors in adults

Wald LL, Nelson SJ, Day MR, Noworolski SE, Henry RG, Huhn SL, Chang S, Prados MD, Sneed PK, Larson DA, Wara WM, McDermott M, Dillon WP, Gutin PH, Vigneron DB. 1997. Serial proton magnetic resonance spectroscopy imaging of glioblastoma multiforme after brachytherapy. J Neurosurg 87: 525–534. Waldrop SM, Davis PC, Padgett CA, Shapiro MB, Morris R. 1998. Treatment of brain tumors in children is associated with abnormal MR spectroscopic ratios in brain tissue remote from the tumor site. Am J Neuroradiol 19: 963–970.

Walecki J, Sokol M, Pieniazek P, Maciejewski B, Tarnawski R, Krupska T, Wydmanski J, Brzezinski J, Grieb P. 1999. Role of short TE 1H-MR spectroscopy in monitoring of postoperation irradiated patients. Eur J Radiol 30: 154–161. Yamagata NT, Miller BL, McBride D, Chang L, Chiang F, Nikas D, Osborne D,Buchthal SD. 1994. In vivo proton spectroscopy of intracranial infections and neoplasms. J Neuroimaging 4: 23–28.

305

306

Jeffry R. Alger

Case Study 19.1 Metabolic heterogeneity of glioma Alberto Bizzi, M.D., Ugo Danesi, Ph.D. and Bianca Pollo, M.D., Istituto Nazionale Neurologico Carlo Besta, Milan, Italy History 57-year-old male presenting with a single episode of loss of consciousness. Neurological examination was normal.

1A

Technique Conventional MRI and MRSI (PRESS: TR/TE  1500/136; 32 32). Imaging findings At presentation, T 2-weighted MRI showed a small hyperintense lesion, with ring enhancement after injection of Gadolinium (Gd), in the left temporo-occipital region (Figure 1A). Twenty days later a second MRI showed a fast growing ring-enhancing mass with vasogenic edema in the surrounding tissue (Figure 1B). Preoperative MRSI (Figure 2) showed a metabolic heterogeneous mass with markedly high Cho signal in the posterolateral rim of the tumour, mild Cho signal in the anterior and medial rim, and low Cho in the core of the mass, where high mobile lipids signal were also present. In the T2 hyperintense surrounding tissue Cho was not elevated. NAA was very low in the mass and slightly decreased in the edematous area. Cr signal was very low within the mass. Neuropathological studies Surgical specimens were correlated with MR imaging data intra operatively using a neuronavigational device. Voxels with high lipids in the center of the mass correlated with areas of low cell density and large pseudopalisading necrosis (N in Figure 3A) on HE stains. Areas with markedly high Cho signal correlated with areas of tumors with high cell density and high vascular proliferation (VP in Figure 3B) on HE stains. A diagnosis of GBM was made.

1B

Key points High Cho signal corresponded to high tumor cell density. Low Cho and elevation of lipids was found in the core of the mass which histopathology confirmed was due to the presence of necrosis. Areas which consisted predominantly of vasogenic edema showed normal Cho and slightly decreased NAA. Cho was highest in the posterolateral segment of the enhancing rim, suggesting that this may be the faster growing side of the tumor.

MR spectroscopy of brain tumors in adults

2A

19 14

20 14

21 14

22 14

23 14

24 14

25 14

26 14

19 15

20 15

21 15

22 15

23 15

24 15

25 15

26 15

19 16

20 16

21 16

22 16

23 16

24 16

25 16

26 16

19 17

20 17

21 17

22 17

23 17

24 17

25 17

26 17

19 18

20 18

21 18

22 18

23 18

24 18

25 18

26 18

19 19

20 19

21 19

22 19

23 19

24 19

25 19

26 19

19 20

20 20

21 20

22 20

23 20

24 20

25 20

26 20

19 21

20 21

21 21

22 21

23 21

24 21

25 21

26 21

19 22

20 22

21 22

22 22

23 22

24 22

25 22

26 22

2B

3A

3B

N VP

References Fulham MJ, Bizzi A, Dietz MJ, Shih HH, Raman R, Sobering GS, Frank JA, Dwyer AJ, Alger JR, Di Chiro G. 1992. Mapping of brain tumor metabolites with proton MR spectroscopic imaging: clinical relevance. Radiology 185: 675–686. Law M, Yang S, Wang H, Babb JS, Johnson G, Cha S, Knopp EA, Zagzag D. 2003. Glioma grading: sensitivity, specificity and predictive values of perfusion MR imaging and proton MR spectroscopic imaging compared with conventional MR imaging. Am J Neuroradiol 24: 1989–1998. Li X, Lu Y, Pirzkall A, McKnight T, Nelson SJ. 2002. Analysis of the spatial characteristics of metabolic abnormalities in newly diagnosed glioma patients. J Magn Reson Imaging 16: 229–237.

307

308

Jeffry R. Alger

Case Study 19.2 Tumefactive multiple sclerosis – MRSI Peter Barker, D.Phil. and Martin Pomper, M.D. Ph.D., Johns Hopkins University School of Medicine, Baltimore, MD, USA History 41-year-old male presenting with a focal seizure.

FLAIR

Cho

NAA

Lac

Technique Conventional MRI and multi-slice MRSI (TE 280 ms).

Imaging findings FLAIR images demonstrate a hyperintense lesion in the right frontal lobe. On MRSI the lesion shows elevated Cho, low NAA and increased Lac compared to the contralateral hemisphere.

Discussion Fulminant, active demyelination shows similar spectral patterns to gliomas – high Cho, low NAA and often the presence of lac (Bitsch, 1999). Therefore, it is difficult to distinguish tumor from demyelination using lesion MRS alone. However, MRSI in MS may sometimes show diffuse or remote metabolic abnormalities (red arrow) not usually seen in untreated primary brain tumors. Also, CBV measured by MR perfusion imaging may be helpful (Cha, 2001) (cf. Case Study 26.1).

Cho

Key points

NAA

Spectra of actively demyelinating lesions may be indistinguishable from those of gliomas.

Lac

MR perfusion may be helpful. ppm

4.0

3.0

2.0

1.0

References Bitsch A, Bruhn H, Vougioukas V, Stringaris A, Lassmann H, Frahm J, Bruck W. 1999. Inflammatory CNS demyelination: histopathologic correlation with in vivo quantitative proton MR spectroscopy. Am J Neuroradiol 20(9): 1619–1627. Cha S, Pierce S, Knopp EA, Johnson G, Yang C, Ton A, Litt AW, Zagzag D. 2001. Dynamic contrast-enhanced T* 2 -weighted MR imaging of tumefactive demyelinating lesions. Am J Neuroradiol 22:1109–1116.

MR spectroscopy of brain tumors in adults

Case Study 19.3 MRS in meningioma Adam D., Waldman, M.D., Ph.D. and Dawn Saunders, M.D., Hammersmith Hospitals, Institute of Neurology and Great Ormond Street Hospital, London, UK History 43-year-old woman with progressive focal neurological deficit. Technique Short TE (35ms) single voxel MRS of right hemisphere mass and conventional MRI. Imaging findings Large heterogeneous signal right hemisphere mass, with surrounding edema and marked mass effect. Pathological enhancement post-Gddimeglumine gadopentetate (DTPA). MRS findings

Cho Short TE MRS

The spectrum showed an alanine (Ala) doublet at 1.4 ppm, as well as an elevation of Cho and the presence of Lac (overlapping doublet at 1.3 ppm). NAA is absent.

Ala Lac

Discussion Ala is expressed in tumors of meningeal origin, and may be a discriminant metabolite compared to other tumor types. Ala has been reported in 30–40% of meningioma cases. NAA is not present in tissue of non-neural origin. Elevated Cho is in keeping with rapid membrane turnover of the neoplasm. Although not present in this case, some meningiomas may show a prominent methylene resonance (i.e. a broad, singlet resonance at 1.3 ppm) that is also associated with malignant behavior. Lac and Ala can be distinguished from lipid by using long TE (i.e. TE 140 (inverted) or 280ms).

ppm

3.0

2.0

1.0

Key points Alanine is characteristic of mengingeal tumors, but is not always present. Mobile lipid and high Cho are associated with aggressive lesions.

References Florian CL, Preece NE, Bhakoo KK, Williams SR, Noble M. 1995. Characteristic metabolic profiles revealed by 1H NMR spectroscopy for three types of human brain and nervous system tumours. NMR Biomed 8(6) : 253–264. Shino A, Nakasu S, Matsuda M, Handa J, Morikawa S, Inubushi T. 1999. Noninvasive evaluation of the malignant potential of intracranial meningiomas performed using proton magnetic resonance spectroscopy. J Neurosurg 91(6): 928–934.

309

310

Jeffry R. Alger

Case Study 19.4a Recurrent astrocytoma David Hearshen, Ph.D., Suresh Patel, M.D. and Tom Mikkelsen, M.D., Henry Ford Health System, Detroit, MI, USA History 22-year-old male, astrocytoma previously treated with surgery, chemotherapy and radiation.

NAA

T1 Post-Gd

Technique

1

3.0 T multi-slice MRSI (TR/ TE  2250/140 ms) and MRI, registered with image-guided, directed biopsy.

B

2 3

4

Imaging findings Post-Gd image demonstrates enhancement in the left parietal lobe. Edema and mass effect were increased compared to a previous study. Cho was elevated both coincident and medial and lateral to the region of enhancement (A). Lesion NAA deficit was widespread (B), while Lac (C) was only elevated in the necrotic core.

Cho

Lac

A

C

Discussion Conventional imaging is consistent with recurrent astrocytoma, confirmed by biopsy. MRSI shows infiltrative (2), solid (3), and early necrotic (4) tumor, all with high Cho (1.7 normal Cr (Croteau et al., 2001)), but different NAA and Lac. The presence of a small Lac signal without lipid may be early indicator of transformation to high grade tumor.

NAA

Cho

1

Key points Lac

High Cho, very low NAA and no lipid is characteristic of solid tumor.

2 3

Heterogeneous spectral features reflect tumor function/morphology.

Reference

4 ppm

3.5

3.0

2.5

2.0

1.5

1.0

0.5

Croteau D, Scarpace, L, Hearshen D, et al. 2001. Correlation between MRSI and image-guided biopsies: semiquantitative and qualitative histopathologic analyses of patients with untreated glioma. Neurosurgery 49: 823–829.

MR spectroscopy of brain tumors in adults

Case Study 19.4b Radiation necrosis David Hearshen, Ph.D., Suresh Patel, M.D. and Tom Mikkelsen, M.D., Henry Ford Health System, Detroit, MI, USA History 51-year-old male, GBM previously treated with radiation.

NAA

T1Post-Gd 2

B

3

Technique 1

3.0 T multi-slice MRSI (TR/ TE  2250/140 ms) and MRI, registered with image-guided, directed biopsy.

Imaging findings

Cho

Lac

A

Post-Gd image demonstrates left frontal lobe enhancement and surrounding areas of low signal. Compression of the lateral ventricle was noted. Cho (A) and NAA (B) were reduced throughout the lesion. A small amount of lipid was detected (3).

Discussion Although the compressed lateral ventricle on conventional imaging suggested recurrent tumor, MRSI shows Cho, NAA, and Cr (2,3) reduced relative to normal appearing WM (NAWM) in the contralateral hemsiphere (1), suggesting radiation necrosis (Rock et al., 2002), consistent with all biopsy findings. The presence of a small lipid signal and Cho/NAA ratio approximately 1 (2) also suggest necrosis.

Cho

Cr

Lip/Lac

NAA

3

2

Key point Reduced Cho, NAA and Cr relative to normal brain indicates necrosis.

1 ppm

3.5

3.0

2.5

2.0

1.5

1.0

0.5

Reference Rock JP, Hearshen D, Scarpace L, Croteau D, Gutierrez J, Fisher JL, Rosenblum ML, Mikkelsen T. 2002. Correlations between magnetic resonance spectroscopy and image-guided histopathology, with special attention to radiation necrosis. Neurosurgery 51(4): 912–919.

311

20

Diffusion MR imaging in adult neoplasia Bradford A. Moffat, Thomas L. Chenevert and Brian D. Ross Departments of Radiology and Biological Chemistry, University of Michigan, Ann Arbor, USA

Key points • Apparent diffusion coefficient (ADC) is inversely proportional to cellular density, and reflects tumor cellularity. • Overlap between cellularity in different grades of glioma limits grading using ADC alone. • ADC maps give spatial information about components in heterogenous tumors (e.g. cystic, cellular and necrotic) which may help to guide biopsy. • Diffusion parameters reflect regional response to therapy within tumors. • When used with structural imaging ADC helps to distinguish malignant from non-malignant lesions, e.g. tumor from radionecrosis. • Diffusion-tensor imaging allows distinction of tumor invasion from displacement of adjacent fiber tracts in the brain, and hence can aid surgical planning.

Increasingly, diffusion-weighted imaging (DWI) is being incorporated into standard clinical evaluation of brain tumors. Driven by compelling results from experimental animal models, an increasing number of clinical research groups have begun to evaluate the potential of DWI to improve clinical management of brain tumors. This chapter contains an overview of the current developments in DWI as applied to adult neuro-oncology. Specific examples have been drawn from the literature to demonstrate how DWI can provide unique and valuable information with regard to diagnosis, treatment planning 312

Table 20.1. A summary of DWI as applied to the clinical management of adult brain neoplasia Diagnosis (apparent diffusion coefficient (ADC) imaging)

Surgical treatment planning diffusion tensor imaging

Monitoring of therapeutic efficacy (ADC imaging)

(DTI) Non-invasive assessment of tumor cellularity and heterogeneity

Location of vital white matter (WM) fiber relative to tumor

Early detection of therapeutic efficacy

Distinguish solid tumor from cysts, abscess and radiation necrosis

Determine tumor dislocation or tumor infiltration of fiber tracts

Visualization and detection of heterogenous response to therapy

Stereotactic information for improved biopsy of tumors

and therapeutic monitoring of brain neoplasia (Table 20.1).

Introduction Despite considerable development in new therapies for brain tumors over the last 20 years, only modest improvements in 5-year survival rates have been achieved (Legler et al., 1999). The failure of new chemotherapy to improve patient survival may be because less than 50% of tumors will respond to a given drug regimen (Greenberg et al., 1999). A rational strategy is to tailor drug treatment based on

Diffusion MR imaging in adult neoplasia

(a)

(b)

(c) 0.7

0.7

2.6 T2-weighted

Diffusion-weighted

ADC (10 3 mm2/s)

Fig. 20.1 Comparison of a T2-weighted, diffusion-weighted and quantitative ADC image of a pilocytic astrocytoma.

lesion-specific properties beyond the particular tumor type and grade. For these approaches to be successful, it will become increasingly important for radiological assessments to provide information on tumor morphology, physiology and heterogeneity, both at the time of initial diagnosis and during therapy. This information could potentially provide valuable guidance as to therapeutic effectiveness or indeed futility of a particular treatment regimen. The application of DWI is under active investigation with the aims of improving tumor characterization, treatment planning and therapeutic monitoring. The current clinical value of conventional MR imaging (MRI) resides in its ability to demonstrate gross tumor morphology and temporal changes non-invasively. Conventional MRI exploits a variety of tissue properties that allow the neuro-oncologist/ neuro-radiologist to assess gross tumor extent on the resultant contrasts, such as “T2-weighted” (Figure 20.1(a)), “gadolinium (Gd)-enhanced T1weighted” and “diffusion-weighted” (Figure 20.1(b)) images. The typical radiological assessment is somewhat interpretive and based on the spatial extent and location of abnormal tissue contrast. The actual image contrast values are rarely quantified, as these are usually scaled arbitrarily, and do not have a

simple relationship to microscopic tissue properties. There is significant clinical potential for MR techniques which provide additional quantitative or semiquantitative functional, structural, or molecular information related to tumor biology and physiology. Such information may be derived from MR signals which reflect perfusion dynamics, oxygenation levels, biochemistry/metabolism, cellularity and levels of gene expression. The focus of this chapter is the application of MRI to provide information related to the microscopic cellular environment in solid tumors from regional apparent diffusion coefficient (ADC) of tissue water (Figure 20.1(c)). As discussed below, the ADC image as acquired by diffusion-weighted MRI (Figure 20.2(a)) is a quantitative representation of water molecular mobility within the tissues being imaged (Figure 20.2(b)). Of critical importance to its application to oncology is that studies have shown that ADC is, in certain circumstances, inversely related to tissue cellularity, which impedes the translational motion of water (Figure 20.3). For this reason, there has been growing interest in the use of ADC in conjunction with other radiological techniques for tumor diagnosis, treatment planning and quantitative assessment of therapeutic response. Since

313

Bradford A. Moffat, Thomas L. Chenevert and Brian D. Ross

spatial information is retained, regional heterogeneity in tumor cellularity is available. The use of water diffusion to probe tissue cellularity is reasonable, since this parameter is strongly affected by membrane permeability between intra- and extracellular compartments, active transport and flow, and directionality of tissue/cellular structures that impede water mobility. Thus, DWI can be applied for a variety of purposes including the straightforward distinction of solid from cystic regions (Figure 20.4), as well as detection of treatment response manifesting as a change in tumor cellularity. Technical aspects of DWI sequences are discussed in Chapter 4. It is worthy of note that diffusion sequences incorporate an additional pair of magnetic field gradient pulses (Figure 20.2(a)) to render an MR signal intensity that is dependent on the mobility of the signal source, i.e. water molecules (Stejskal and Tanner, 1965). Conceptually, the first of these two gradient pulses imparts a phase shift to the

180° RF

(a) 90° RF G

G

Time (Δ) (b)

Distance traveled (low-ADC value to high)

314

Unrestricted (high ADC) Restricted (low ADC)

Time (Δ) Fig. 20.2 (a) Schematic representation of a generalized (G) DWI pulse sequence used to acquire ADC images. (b) Diagrammatic illustration of the effect of restricted diffusion (by cell membranes) on the ADC as measured by the MR pulse sequence. RF: radio frequency.

(a)

(b)

Primitive neuroectodermal tumor ADC  0.6 10 3 mm2/s

Chordoma ADC  2.3 10 3 mm2/s

y  0.021x  1.2749 r  0.77

ADC (10 3 mm2/s)

ADC (10 3 mm2/s) 1.5 1.4 1.3 1.2 1.1 1 0.9 0.8 0.7 0.6

Glioblastoma Low-grade astrocytoma

1.2 1.1 1 0.9 0.8 0.7 0.6 0.5

0

(c)

y  0.017x  0.9687 r  0.673

5

10

15

20

25

Tumor cellularity (%)

30

35

0

(d)

4

8

12

16

20

24

28

Tumor cellularity (%)

Fig. 20.3 Histopathology results for (a) a chordoma tumor with a relatively high ADC (2.3 10 3 mm2/s) and low cell density, and (b) a PNET with a relatively low ADC (0.6 10 3 mm2/s) and high cell density. A linear correlation of ADC with tumor cellularity for (c) astrocytic tumors (r  0.77) and (d) meningiomas (r  0.67) was observed (figure courtesy of Kono et al., 2001).

Diffusion MR imaging in adult neoplasia

Solid/cellular

Water/acellular

Cyst Necrotic tumor Typical tumor (e.g. GBM) Edema Normal brain Dense tumor (e.g. PNET) Radiation necrosis Abscess 0.5

1

1.5

2 2.5 3 2

ADC (10 mm /s)

3

Fig. 20.4 Schematic representation of the ADC range for some common brain lesions and normal neural tissue.

MR signal proportional to the initial location of the water molecules. The second gradient pulse will totally unwind this phase shift (“rephase”) if the water molecules remain at their original location. Any molecular movement between first and second pulses, however, leads to incomplete rephasing (dephasing). The diffusion of water molecules due to “Brownian motion” produces a net dephasing, or signal loss. The amount of signal loss is a direct reflection of water mobility, i.e. the greater signal loss implies greater molecular mobility and thus a greater diffusion coefficient. In routine clinical applications, one is less interested in the physics of water mobility than its interpretation in relation to tissue structure. Thus, consider if the time interval between gradient pulses is sufficient to allow water molecules to migrate distances comparable to the size of and spacing between cells, then the apparent mobility or ADC will be reduced by the impediments of cellular membranes and tortuosity of the extracellular space (Figure 20.2(b)). In addition, the directionality of cellular structures can be probed by controlling the direction of the applied diffusion pulses (Basser and Pierpaoli, 1996). In highly directional tissues such as in white matter (WM) fiber tracts, water mobility is roughly 3fold higher when measured parallel to the fiber axis compared to perpendicular to the fibers. The study of

diffusion directionality or “anisotropy” is itself a significant area of investigation (Chenevert et al., 1990; Moseley et al., 1990; Brunberg et al., 1995; Conturo et al., 1995; Basser and Pierpaoli, 1996). For many applications in oncology, however, one seeks to simply quantify mobility and avoid the complexities related to the orientation of the tissue relative to the imaging system. Toward this end, ADC measurements are made directionally independent by calculating the mean of three ADC measurements using three orthogonal diffusion gradients (Chenevert et al., 1990).

Cellular density DWI has proven to be a sensitive technique for identifying regions of ischemic tissue damage in animal models of stroke and in human patients (Moseley et al., 1990; Warach et al., 1995; Sorensen et al., 1996). Theoretical Monte Carlo simulations suggest that changes in tissue water diffusion following tissue damage are predominantly attributable to alterations in the volume and tortuosity of the extracellular space (Norris et al., 1994; Sykova et al., 1994; Szafer et al., 1995). These properties of the extracellular space are primarily a function of cell density, and recent work has shown that tumor water diffusion is associated with tumor cellularity (Figure 20.3)

315

316

Bradford A. Moffat, Thomas L. Chenevert and Brian D. Ross

Abscess

T1-Gd-DTPA

ADC

Radiation necrosis

Epidermoid tumor

Arachnoid cyst

T2-weighted

T2-weighted

T2-weighted

ADC

ADC

ADC

Fig. 20.5 DWI application to diagnosis of different brain lesions. ADC maps in conjunction with T2- or T1-weighted images are able to differentiate a brain abscess (figure courtesy of Chang et al., 2002), radiation necrosis (figure courtesy of Biouse et al., 2003) and cysts from brain tumors.

(Gupta et al., 1999; Lyng et al., 2000; Kono et al., 2001; Yang et al., 2002). Moreover, the gradation of increasing diffusion from cellular/solid lesions such as a primative neuro-ectodermal tumor (PNET) to normal brain tissue, to edematous brain, and to viable and necrotic human gliomas is illustrated in Figure 20.4 and has been noted by others (Chenevert et al., 1990, 2000; Eis et al., 1994; Kono et al., 2001; Bastin et al., 2002; Bitzer et al., 2002; Chang et al., 2002). DWI offers quantification that can be used in conjunction with other MR techniques to distinguish tumors from other lesions for which treatment is quite different (Figure 20.5). Moreover, it also suggests that the temporal evolution from viable tumor to treatment-induced tumor necrosis should be documented by an increasing diffusion coefficient.

Comparison of ADC values from individual tumors with biopsy-derived histological sections provides important validation of this approach for non-invasive assessment of tumor cellularity (Figure 20.3). For example (Figure 20.3(a)), consider the following two extremes. A chordoma was found to contain a range of neoplastic cells (from spindle fibrosarcoma-like cells to epithelial-like elements) arranged in aggregates floating within the mucus. This is an especially fluid-like tumor which was found to have a very high ADC value of 2.3 10 3 mm2/s. In contrast (Figure 20.3(b)), a biopsied PNET contained a very high neoplastic cell density which, when probed using DWI, was found to have a relatively low ADC value of 0.6 10 3 mm2/s. Quantitative data from Kono et al. (2001) (Figure 20.3(c)) supports these histological findings; a significant inverse relationship between

Diffusion MR imaging in adult neoplasia

ADC and tumor cellularity in both astrocytic tumors and meningiomas was observed. These data suggest that DWI can detect differences in tumor cellular density based upon the relative mobility of water contained within the tumor tissue. However, the difference in the two correlations (Figure 20.3) suggests that ADC values alone cannot predict tumor cellularity.

Application of DWI to tumor diagnosis Other studies (Lam et al., 2002; Yang et al., 2002) have also shown that ADC maps are insufficient on their own to predict tumor cellularity, tumor type and tumor grade, but require prior knowledge from biopsy or other radiological scans. What then, are the potential advantages and reasons for acquiring ADC maps during a routine radiological assessment of a patient with a “potential” brain tumor? Firstly, specific tumor classification based only upon information obtained from anatomical images (Figure 20.5) is often difficult. In such cases, the enhancement on T1-weighted images by Gd-dimeglumine gadopentetate (DTPA) or hyperintensity on T2 dependent sequences may indicate a tumor. ADC maps, as an indication of cellularity, provide a potentially more accurate assessment of the lesion than is available from conventional structural images. The examples given in Figure 20.5 show that quantitative ADC maps in conjunction with more traditional MR imaging differentiates tumors from cysts (Tsuruda et al., 1990), abscesses (Chang et al., 2002) and radiation-induced necrosis (Biousse et al., 2003) of normal brain tissue. In these examples, an arachnoid cyst, radiation necrosis, epidermoid tumor and an abscess were presented as hyperintense lesions on T2-weighted or Gd-DTPA contrast-enhanced MRI in different patients. However, DWI revealed an arachnoid cyst as a lesion with an extremely high ADC value (3.0 10 3 mm2/s) while the epidermoid tumor had a lower ADC value (0.9 10 3 mm2/s). Interestingly, the abscess and radiation necrosis exhibit an ADC that is even lower than normal brain. While there is a continuum of diffusion environments across tissues with substantial overlap in ADC values between lesion types, these examples nonetheless reveal that DWI can provide valuable

information reflecting the cellularity of a lesion within the central nervous system (CNS) which can aid clinical diagnosis. ADC maps may aid biopsy guidance in heterogeneous tumors. The accuracy of biopsy results greatly depends on sampling the most malignant part of the tumor. Since high-grade tumors are often heterogeneous, sampling an area of low malignancy often jeopardizes the accuracy of the biopsy. A recent study (Yang et al., 2002) showed that areas of minimum ADC within tumors were associated with histological grade, suggesting that in the case of a highly heterogeneous tumor the use of ADC maps to guide sample location could improve the biopsy yield. Further, since grading of tumors greatly influences the prescribed therapeutic regimen such an application of DWI may help to improve therapeutic outcome. Stereotactic guidance of tumor biopsy based on regions of pathological enhancement has been used previously (Lam et al., 2002) to obtain samples from areas within the tumor that were more likely to represent the highest-grade. Thirdly, DWI may be very important for diagnosing the extent and type of tumor-related edema (Chenevert et al., 1990; Steen, 1992; Eis et al., 1994; Krabbe et al., 1997; Pronin et al., 2000; Bastin et al., 2002; Bitzer et al., 2002). In the study by Pronin et al. (2000) vasogenic, ischemic and interstitial edemas were found to have significantly different (P  0.05) ADC values (1.3 10 3, 1.0 10 3 and 1.9 10 3, respectively). ADC ranges, however, for the different types of edema overlap, and edema ADC values also overlap with those of tumors. If the tumor margins can be delineated by other imaging techniques, however, this type of information can be helpful for evaluating the effects of tumors on the surrounding neuronal tissue.

Diffusion-tensor imaging for surgical treatment planning Surgical resection remains one of the most common methods for treating brain tumors. An important aspect of surgical planning is location of the WM tracts in the area of the tumor (Witwer et al., 2002). Whether or not a tumor has infiltrated or has shifted,

317

318

Bradford A. Moffat, Thomas L. Chenevert and Brian D. Ross

the location of the WM tracts can influence the surgical approach and extent of a resection. One of the more recent applications of DWI to brain tumors is the use of diffusion-tensor imaging (DTI) to locate WM tracts in close proximity to the tumor and evaluate the degree of tumor infiltration and/or dislocation (Wieshmann et al., 2000; Mori et al., 2002; Witwer et al., 2002). Water diffusion in WM is less restricted in the direction parallel to the axonal fibers than the diffusion in the orthogonal direction (Doran et al., 1990; Moseley et al., 1990; Hajnal et al., 1991; Le Bihan et al., 1993; Conturo et al., 1995). The diffusion of water in this environment is anisotropic and cannot be fully described by a scalar quantity such as ADC. To describe the water diffusion for such tissue completely it becomes necessary to measure the “diffusion tensor” (Basser and Pierpaoli, 1996), which is a three-by-three matrix. While the acquisition and analysis of diffusion-tensor data is more time consuming and complex than for the calculation of ADC, anisotropic diffusion provides insight into the three-dimensional (3D) fiber patterns and their connectivity across regions of the brain. Anisotropic diffusion is commonly visualized via 2D or 3D maps of “fractional anisotropy” (FA). The FA is itself a scalar quantity that ranges from 0 (purely isotropic) to 1 (highly anisotropic) and provides a sense of the density of uniformly directional tissue structures (Basser and Pierpaoli, 1996). The actual alignment direction of anisotropic tissues is further specified by the eigenvectors of the diffusion tensor. These quantities are displayed directly (often in color, Figure 20.6(d–g)) or serve as inputs to “fiber-tracking” algorithms that aim to deduce continuity and connection between distant regions of the brain. There are several fiber-tracking algorithms (Mori et al., 1999; Gossl et al., 2002; Mori and Van Zijl, 2002; Bammer et al., 2003) that can be employed to visualize the fiber tracts. These vary in levels of complexity and robustness. Directionally color-coded FA maps (Figure 20.6), as used by Mori and co-workers (2002) are a good compromise between simplicity, accuracy and robustness. The FA can be considered as a quantitative measure of how anisotropic the diffusion of water is within a tissue. The calculation of FA is relatively simple

R

(a)

L

R

L

(b)

R

L

(c)

Cingulum Corona radiata (d)

Corpus callosum Cingulum Slf

(e) Cingulum Slf (f)

(g) Fig. 20.6 DTI of tumor growth effect on surrounding myelinated neuronal cells compared to a healthy volunteer. (a) T2-weighted image, (b) FA map, (c) ADC map and (d)–(g) color-coded FA maps (figure courtesy of Mori et al., 2002).

provided a complete DTI data set is acquired (Basser and Pierpaoli, 1996). Since gray matter (GM) and tumor tissue have FA values close to 0 they appear as hypointense on FA maps (Figure 20.6(b)) while unidirectional myelinated WM appears hyperintense. The color (red, green or blue) (Figure 20.6(d–g)) indicates the direction of the WM tracts relative to the plane of the image. An example of the way in which a tumor can displace WM tracts was shown by Mori et al. (2002) (Figure 20.6) who compared an FA map of a healthy volunteer to that of a patient with an anaplastic

Diffusion MR imaging in adult neoplasia

astrocytoma (AA) in the frontal lobe. It was observed that the tumor grew discretely, compressing and displacing the corona radiata medially and the superior longitudinal fasciculus superiorly. The fiber tracts were reconstructed and rendered in 3D (Mori et al., 1999) for improved visualization (Figure 20.7(a)). A second patient with an astrocytoma was also evaluated using this method. In contrast to the first case, the patient’s tumor did not appear to induce adjacent WM tract deformation. This is shown by the 3D reconstructions (Figure 20.7(b)) where the fibers do not diverge and travel around the tumor as in patient 1 but are terminated at the tumor boundary. It should be noted that this is not necessarily where the fiber tracts are terminated but where the FA drops to the same level as the tumor periphery. FA maps or DTI data cannot at this stage determine the level of WM destruction or tumor infiltration along the tracts without supporting histological data. Although research of DTI for use in treatment planning of the resection of brain tumors is limited to a few case reports (Wieshmann et al., 2000; Coenen et al., 2001; Mori et al., 2002; Witwer et al., 2002), the data are promising. This approach has the potential to reduce the morbidity resulting from brain tumor resection or better evaluate possible surgical risks.

Therapeutic monitoring Sequential MRI is routinely used for monitoring the response of CNS tumors to therapy. This is accomplished by assessing changes in tumor dimensions and volumes (Sorensen et al., 2001) and Gd-enhancement. Comparisons of tumor burden are usually made between pre-treatment imaging and those obtained weeks to months following the conclusion of a therapeutic protocol (Therasse et al., 2000), and depend largely on changes in bulk tumor size. Methods of assessing treatment response that are not dependent on relatively slow changes in tumor volume may be capable of providing earlier indications of therapeutic outcome since molecular and cellular changes typically precede observable macroscopic changes in gross tumor size. The use of quantitative MR surrogate markers (i.e. water diffusion) to determine therapeutic-induced

(a)

(b)

Fig. 20.7 Volume rendered WM fiber tracts (yellow) and tumor mass (red). (a) An astrocytoma patient (same patient as Figure 20.6) where the tumor mass has clearly dislocated the WM fiber tracts. This is contrasted to a second patient (b) where the WM appears terminated at the tumor boundary (figure courtesy of Mori et al., 2002).

changes in the tumor cellular matrix is an area of active research (Chenevert et al., 2000; Kauppinen, 2002; Mardor et al., 2003). Successful treatment of a tumor with cytotoxic agents results in significant damage and/or killing of cells, altering membrane integrity and the degree of local cellularity. This has a net effect of increasing the fractional volume of the interstitial space due to cell loss resulting in an increase in the mobility (diffusion) of water within the damaged tumor tissue (Figure 20.8). The sensitivity of DWI for detection of therapeutic changes depends upon the overall dynamic range, which may be observed by ADC measurements. For example, relatively solid tissue such as normal adult brain has an ADC value of 0.6–0.8 10 3 mm2/s while cerebrospinal fluid (CSF) is about 3.0 10 3 mm2/s (Figure 20.4); this represents the range of ADC values typically observed in the CNS. Enthusiasm for the use of DWI for therapy assessment stems from previous animal studies, which have reported that this approach can be used to monitor early events in tumor treatment

319

Bradford A. Moffat, Thomas L. Chenevert and Brian D. Ross

Distribution

Necrosis

Cellular dense

Pure water

0 1 2 3 Water mobility or ADC

0

1 2 3 Water mobility or ADC

0

1 2 3 Water mobility or ADC

Distribution

Microscopic/ cellular necrosis

0

0

Effective therapy

Solid/cellular tumor

Distribution

320

0

Fig. 20.8 Schematic representation of the relationship between change in tissue cellularity and molecular water mobility measured as an “ADC”. At left is the evolution toward necrosis following an effective therapy. The increase in extracellular space and membrane permeability allows greater water mobility as illustrated by distributions of diffusion on the right.

in a variety of tumor models (Ross et al., 1994; Zhao et al., 1996; Galons et al., 1999; Chinnaiyan et al., 2000; Stegman et al., 2000) along with a preliminary application to patients with primary CNS tumors (Chenevert et al., 1990; Chinnaiyan et al., 2000). Studies exploring the potential for using DWI for the detection of early therapeutic-induced changes in tumors are ongoing. The goal of these studies is to determine if DWI can provide early evidence of cancer treatment efficacy in an individual patient prior to the completion of the therapeutic regimen (Chenevert et al., 2000). Figure 20.9 shows an example of how DWI can be used clinically. The radiological assessment at week 3 revealed an increase in size of the contrast-enhancing area which is used to reflect the extent or boundaries of the primary tumor mass. The fact that the tumor diffusion

value did not increase as displayed in the adjacent ADC histograms also suggested that the treatment produced no significant positive therapeutic benefit over this interval. In fact, the mean diffusion value decreased slightly throughout the treatment protocol, which could be interpreted as a lack of cellkilling effect during this treatment protocol. Following completion of the fractionated radiation dosage schedule at week 7, a follow-up set of images was acquired at week 8, which revealed progressive disease. Thus, the lack of change in tumor diffusion values in this patient appears consistent with the lack of therapeutic benefit. In contrast to the non-responsive patient shown in Figure 20.9, a patient with a solid tumor, which responded to a fractionated therapeutic regimen, would be anticipated to show an increase in tumor

Diffusion MR imaging in adult neoplasia

ADC map

Post-treatment week 8

Post-treatment week 3

Pre-treatment

(c) Quantity of pixels

(b)

Quantity of pixels

(a)

Histogram of tumor ADC

Quantity of pixels

T1-Gd MRI

0

1 ADC

2 (10 3

3

mm2/s)

Fig. 20.9 Clinical example of a 63-year-old woman treated for an astrocytoma by standard fractionated radiotherapy (total dose 70 Gy in 35 fractions). T1-weighted, Gd-enhanced images are shown along with the corresponding ADC maps pre-radiotherapy, 3 weeks into therapy and finally at 8 weeks following initiation of therapy which ended on week 7. All images were spatially registered to the pre-treatment T1-weighted image for comparison over time (Meyer et al., 1997). Rapid disease progression on the contrast-enhanced T1-weighted MR (a) is apparent through week 8 following initiation of therapy. ADC maps (b) and diffusion histograms (c) suggest the cellularity of the tumor was not significantly altered over this period. This patient went on to receive chemotherapy, but survived for only 18 weeks from initiation of radiation treatment.

ADC values over time during the therapeutic regimen. For example, a 56-year-old female diagnosed with an anaplastic oligoastrocytoma who had failed chemotherapy was entered into a DWI study. This patient was subsequently treated with radiotherapy (week 0, 70 Gy in 35 fractions), which was completed 7 weeks later (Figure 20.10). The clinical/ radiological assessment revealed that radiotherapy provided a substantial benefit to this particular patient. This positive therapeutic effect was also detected at 3 weeks as a substantial increase in the tumor diffusion values (Figure 20.10). The early increase in observed diffusion values for the tumor

mass of this patient is consistent with the subsequent clinical diagnosis of a partial therapeutic response. The data obtained from this patient indicated that DWI appears to be sensitive enough to detect tumor cell kill. It is possible that even a stronger diffusion shift might be anticipated in patients more responsive to treatment and perhaps at earlier time points such as within the first few weeks of fractionated therapy. These results demonstrate that tumor diffusion values can be measured during treatment and appear to reflect dynamic therapeutic-induced changes (or lack thereof) in tissue cytoarchitecture. A non-responsive

321

Bradford A. Moffat, Thomas L. Chenevert and Brian D. Ross

T1-Gd MRI

Post-treatment week 6

Post-treatment week 3

Pre-treatment

(c) Quantitiy of pixels

(b)

Quantitiy of pixels

(a)

Histogram of tumor ADC

ADC map

Quantitiy of pixels

322

0

1 ADC

(10 3

2

3

mm2/s)

Fig. 20.10 A 56-year-old woman was treated for an anaplastic oligodendroglioma by standard fractionated radiotherapy (total dose 70 Gy). T1-weighted, Gd-enhanced and ADC images from pre-radiotherapy, week 3 and week 6 are shown. All images are co-registered to the pre-treatment T1-weighted image. There was a reduction in contrast enhancement on the T1-weighted MR as shown on the left (a) (times are relative to start of radiotherapy). ADC maps (b) and diffusion histograms (c) indicate greater water mobility, which suggests tumor necrosis over this period. This patient went on to receive chemotherapy and survived for 73 weeks from the start of radiation therapy.

tumor showed no significant increase in diffusion values throughout the treatment protocol (Figure 20.9). This supports the hypothesis that the magnitude of change in tumor water mobility, as assessed using DWI, may be related to the fraction of cells killed and hence therapeutic efficacy. These data are supported by the work of Mardor et al., 2003 (Figure 20.11), where 10 patients with brain tumors had DWI before and 1-week after radiosurgery. There was a statistically significant (P  0.006) difference in the change in mean ADC between responders and non-responders to therapy as well as linear correlation between the relative

change in ADC and the normalized change in tumor volume. Tumors are known to be highly heterogeneous in terms of cell viability, perfusion and oxygenation levels. Since these biophysical properties are factors that can modulate efficacy of chemo- and radiotherapies, one could reasonably expect that therapyinduced changes may be heterogeneous within a given tumor. Since ADC images are quantitative they can be used to map therapy response regionally. Such information has the potential to be valuable for guiding spatially directed therapies such as gamma knife radiosurgery or intratumoral injection

Change in ADC conventional 8 days post-treatment

Diffusion MR imaging in adult neoplasia

1.5

Response

 0.63 P  0.006 Slope  0.32 r2

1

No response 0.5

100

50 0 50 Change in tumor size 48 days post-treatment (%)

Fig. 20.11 The correlation between the relative change in ADC, from 1 week after initiation of radiosurgical treatment, and the normalized value of tumor volume, measured 7 weeks later, in 10 lesions. The correlation is considered extremely significant (P  .006) (figure courtesy of Mardor et al., 2003).

of agents. Since tissue “change” is of key interest, temporal shifts in diffusion coefficients are measurable by select region-of-interests defined on ADC images. Alternatively, an “ADC difference” map may provide an efficient means to survey regional tissue alterations by emphasizing heterogeneity of tissue changes within and near the tumor, where the unaffected brain is unchanged.

Discussion DWI has been shown to be extremely valuable in differentiating malignant lesions from other common brain lesions. Specifically, ADC maps when used in conjunction with more routine MR images, can greatly aid in differentiating malignancy from radiation-induced necrosis, cysts and abscesses. It is also becoming apparent that lower ADC values are associated with higher tumor grade (Yang et al., 2002). Although this is at best an association and is unlikely to replace tissue biopsy in tumor grading, the spatial information with regards to tumor heterogeneity has the potential to improve stereotactic guidance of biopsy to a region within the tumor that is more likely to reflect the site of highest tumor grade. Since therapy may depend on tumor grading, any improvement in the quality of biopsy results could favorably influence patient outcome. Finally, DWI

appears to differentiate different types of edema. This may be of significance to the neuro-oncologist in the assessment of the damage to normal brain tissue surrounding the tumor. Surgical resection remains the preferred treatment for many intracranial tumors. Unfortunately, such procedures are associated with significant risk and some tumors may be erroneously classified as “inoperable” using conventional MR sequences. The location and condition of WM fiber tracts in and around the tumor may be important. Determining tumor and edema margins can be particularly challenging. Although the application of DTI for improved visualization of WM for surgical planning is still in its infancy, the high quality of the results thus far is reason for optimism. Improved margin detection could lead to better surgical accuracy and assessment of risk. The use of DWI and DTI also has the potential for monitoring early changes in tumor cellularity possibly reflecting treatment response. This could lead to improved tailoring of treatments for individual patients and allow for alternative therapies to be attempted in a more timely fashion if a tumor is found to be resistant. This approach also provides the significant potential of assessing the regional/ spatial heterogeneity of therapeutic response within a tumor. The heterogeneity of response may be accentuated in applications involving direct intratumoral administration of the therapeutic agent as is done in certain therapeutic protocols involving cancer gene therapy.

Conclusions Although DWI has the potential to be a surrogate marker for tumor cell density, further research is still required to determine more precisely how diffusion relates to tumor grade within tumor types. In addition, more results are needed to understand the relationship between the different types of tissue edema and ADC. DTI shows great promise in its ability to show visually the effect of tumors on the surrounding WM tracts. The current results seem compelling; however, the conclusions made from these are yet to be fully confirmed by histopathology.

323

324

Bradford A. Moffat, Thomas L. Chenevert and Brian D. Ross

As a surrogate marker for therapeutic efficacy more clinical data are needed to determine whether the observed changes in tumor diffusion as a result of therapy are a universal response to the successful killing of tumor cells and improved statistical analysis is needed to delineate more fully the prognostic ability of DWI.

ACKNOWLEDGEMENTS

The authors of this chapter were supported in part by research grants from the Charles A. Dana Foundation, the NIH (5R24CA83099, 5P20CA86442, 1PO1CA85878 and 1P50CA93990) and the University of Michigan Clinical Research Partnership Fund.

REFERENCES Bammer R, Acar B, et al. 2003. In vivo MR tractography using diffusion imaging. Eur J Radiol 45(3): 223–234. Basser PJ, Pierpaoli C. 1996. Microstructural and physiological features of tissues elucidated by quantitative-diffusiontensor MRI. J Magn Reson B 111(3): 209–219. Bastin ME, Sinha S, et al. 2002. Measurements of water diffusion and T1 values in peritumoural oedematous brain. Neuroreport 13(10): 1335–1340. Biousse V, Newman NJ, et al. 2003. Diffusion weighted imaging in radiation necrosis. J Neurol Neurosurg Psychiatr 74(3): 382–384. Bitzer M, Klose U, et al. 2002. Alterations in diffusion and perfusion in the pathogenesis of peritumoral brain edema in meningiomas. Eur Radiol 12(8): 2062–2076. Brunberg JA, Chenevert TL, et al. 1995. In vivo MR determination of water diffusion coefficients and diffusion anisotropy: correlation with structural alteration in gliomas of the cerebral hemispheres. Am J Neuroradiol 16(2): 361–371. Chang SC, Lai PH, et al. 2002. Diffusion-weighted MRI features of brain abscess and cystic or necrotic brain tumors: comparison with conventional MRI. Clin Imaging 26(4): 227–236. Chenevert TL, Brunberg JA, et al. 1990. Anisotropic diffusion in human white matter: demonstration with MR techniques in vivo. Radiology 177(2): 401–405. Chenevert TL, Stegman LD, et al. 2000. Diffusion magnetic resonance imaging: an early surrogate marker of therapeutic

efficacy in brain tumors. J Natl Cancer Inst 92(24): 2029–2036. Chinnaiyan AM, Prasad U, et al. 2000. Combined effect of tumor necrosis factor-related apoptosis-inducing ligand and ionizing radiation in breast cancer therapy. Proc Natl Acad Sci USA 97(4): 1754–1759. Coenen VA, Krings T, et al. 2001. Three-dimensional visualization of the pyramidal tract in a neuronavigation system during brain tumor surgery: first experiences and technical note. Neurosurgery 49(1): 86–92; discussion 92–93. Conturo TE, McKinstry RC, et al. 1995. Diffusion MRI: precision, accuracy and flow effects. NMR Biomed 8(7–8): 307–332. Doran M, Hajnal JV, et al. 1990. Normal and abnormal white matter tracts shown by MR imaging using directional diffusion weighted sequences. J Comput Assist Tomogr 14(6): 865–873. Eis M, Els T, et al. 1994. Quantitative diffusion MR imaging of cerebral tumor and edema. Acta Neurochir Suppl (Wien) 60: 344–346. Galons JP, Altbach MI, et al. 1999. Early increases in breast tumor xenograft water mobility in response to paclitaxel therapy detected by non-invasive diffusion magnetic resonance imaging. Neoplasia 1(2): 113–117. Gossl C, Fahrmeir L, et al. 2002. Fiber tracking from DTI using linear state space models: detectability of the pyramidal tract. Neuroimage 16(2): 378–388. Greenberg H, Chandler W, et al. 1999. Brain Tumors. Oxford University Press, New York. Gupta RK, Sinha U, et al. 1999. Inverse correlation between choline magnetic resonance spectroscopy signal intensity and the apparent diffusion coefficient in human glioma. Magn Reson Med 41(1): 2–7. Hajnal JV, Doran M, et al. 1991. MR imaging of anisotropically restricted diffusion of water in the nervous system: technical, anatomic, and pathologic considerations. J Comput Assist Tomogr 15(1): 1–18. Kauppinen RA. 2002. Monitoring cytotoxic tumour treatment response by diffusion magnetic resonance imaging and proton spectroscopy. NMR Biomed 15(1): 6–17. Kono K, Inoue Y, et al. 2001. The role of diffusion-weighted imaging in patients with brain tumors. AJNR Am J Neuroradiol 22(6): 1081–1088. Krabbe K, Gideon P, et al. 1997. MR diffusion imaging of human intracranial tumours. Neuroradiology 39(7): 483–489. Lam WW, Poon WS, et al. 2002. Diffusion MR imaging in glioma: does it have any role in the pre-operation determination of grading of glioma? Clin Radiol 57(3): 219–225. Le Bihan D, Douek P, et al. 1993. Diffusion and perfusion magnetic resonance imaging in brain tumors. Top Magn Reson Imaging 5(1): 25–31.

Diffusion MR imaging in adult neoplasia

Legler JM, Ries LA, et al. 1999. Cancer surveillance series [corrected]: brain and other central nervous system cancers: recent trends in incidence and mortality. J Natl Cancer Inst 91(16): 1382–1390. Lyng H, Haraldseth O, et al. 2000. Measurement of cell density and necrotic fraction in human melanoma xenografts by diffusion weighted magnetic resonance imaging. Magn Reson Med 43(6): 828–836. Mardor Y, Pfeffer R, et al. 2003. Early detection of response to radiation therapy in patients with brain malignancies using conventional and high b-value diffusionweighted magnetic resonance imaging. J Clin Oncol 21(6): 1094–1100. Meyer CR, Boes JL, et al. 1997. Demonstration of accuracy and clinical versatility of mutual information for automatic multimodality image fusion using affine and thin-plate spline warped geometric deformations. Med Image Anal 1(3): 195–206. Mori S, Crain BJ, et al. 1999. Three dimensional tracking of axonal projections in the brain by magnetic resonance imaging. Ann Neurol 45: 265–269. Mori S, Frederiksen K, et al. 2002. Brain white matter anatomy of tumor patients evaluated with diffusion tensor imaging. Ann Neurol 51(3): 377–380. Mori S, Van Zijl PC. 2002. Fiber tracking: principles and strategies – a technical review. NMR Biomed 15(7–8): 468–480. Moseley ME, Cohen Y, et al. 1990. Diffusion-weighted MR imaging of anisotropic water diffusion in cat central nervous system. Radiology 176(2): 439–445. Norris DG, Niendorf T, et al. 1994. Health and infarcted brain tissues studied at short diffusion times: the origins of apparent restriction and the reduction in apparent diffusion coefficient. NMR Biomed 7(7): 304–310. Pronin IN, Kornienko VN, et al. 2000. Diffusion-weighted image in the study of brain tumors and peritumoral edema. Zh Vopr Neirokhir Im N N Burdenko 3: 14–6; discussion 17. Ross BD, Chenevert TL, et al. 1994. Magnetic resonance imaging and spectroscopy: application to experimental neurooncology. Quart Magn Reson Biol Med 1: 89–106. Sorensen AG, Buonanno FS, et al. 1996. Hyperacute stroke: evaluation with combined multisection diffusion-weighted and hemodynamically weighted echo-planar MR imaging. Radiology 199(2): 391–401.

Sorensen AG, Patel S, et al. 2001. Comparison of diameter and perimeter methods for tumor volume calculation. J Clin Oncol 19: 551–557. Steen RG. 1992. Edema and tumor perfusion: characterization by quantitative 1H MR imaging. Am J Roentgenol 158(2): 259–264. Stegman LD, Rehemtulla A, et al. 2000. Diffusion MRI detects early events in the response of a glioma model to the yeast cytosine deaminase gene therapy strategy. Gene Ther 7(12): 1005–1010. Stejskal EO, Tanner JE. 1965. Spin diffusion measurements – spin echoes in presence of a time-dependent field gradient. J Chem Phys 42(1): 288–292. Sykova E, Svoboda J, et al. 1994. Extracellular volume fraction and diffusion characteristics during progressive ischemia and terminal anoxia in the spinal cord of the rat. J Cereb Blood Flow Metab 14(2): 301–311. Szafer A, Zhong J, et al. 1995. Theoretical model for water diffusion in tissues. Magn Reson Med 33(5): 697–712. Therasse P, Arbuck SG, et al. 2000. New guidelines to evaluate the response to treatment in solid tumors. European Organization for Research and Treatment of Cancer, National Cancer Institute of the United States, National Cancer Institute of Canada. J Natl Cancer Inst 92(3): 205–216. Tsuruda JS, Chew WM, et al. 1990. Diffusion-weighted MR imaging of the brain: value of differentiating between extraaxial cysts and epidermoid tumors. Am J Neuroradiol 11(5): 925–931; discussion 932–934. Warach S, Gaa J, et al. 1995. Acute human stroke studied by whole brain echo planar diffusion-weighted magnetic resonance imaging. Ann Neurol 37(2): 231–241. Wieshmann UC, Symms MR, et al. 2000. Diffusion tensor imaging demonstrates deviation of fibres in normal appearing white matter adjacent to a brain tumour. J Neurol Neurosurg Psychiatr 68(4): 501–503. Witwer BP, Moftakhar R, et al. 2002. Diffusion-tensor imaging of white matter tracts in patients with cerebral neoplasm. J Neurosurg 97(3): 568–575. Yang D, Korogi Y, et al. 2002. Cerebral gliomas: prospective comparison of multivoxel 2D chemical-shift imaging proton MR spectroscopy, echoplanar perfusion and diffusionweighted MRI. Neuroradiology 44(8): 656–666. Zhao M, Pipe JG, et al. 1996. Early detection of treatment response by diffusion-weighted 1H-NMR spectroscopy in a murine tumour in vivo. Br J Cancer 73(1): 61–64.

325

326

Bradford A. Moffat, Thomas L. Chenevert and Brian D. Ross

Case Study 20.1 DWI of epidermoids and arachnoid cysts Adam Waldman, M.D., Ph.D., FRCR, Charing Cross Hospital and Institute of Neurology, London, and Doris Lin, M.D., Ph.D., Johns Hopkins University School of Medicine, Baltimore History

DWI

T2

ADC

(a) A 36-year-old male with dizziness and rightsided cerebellar signs and (b) A 43-year-old male with headaches. Technique Conventional MRI and DWI (b  1000 s/mm2), ADC map. Imaging findings •

Epidermoid. Non-enhancing right cerebellopontine angle lesion almost isointense to CSF on both T1 and T2 MRI, distorting brainstem and right cerebellar peduncle. ADC approximately isotense to brain, DWI is hyperintense.

•

Large left temporal arachnoid cyst, isointense to CSF on T1, T2 and ADC.

(a)

(b) Discussion Epidermoids typically show internal structure, usually as a “marbled” signal heterogeneity on T1-weighted sequences and may have irregular margins; when these features are not present, they may be difficult to distinguish from other cystic lesions. The internal structure of an epidermoid results in restricted water diffusion, with ADC similar to that of normal brain, while cysts have free diffusion (ADC similar to that of CSF). Since DW-EPI sequences are prone to susceptibility artifact near bone–tissue–air interfaces, DWI and ADC should be interpreted carefully in these locations.

Key points DWI may be helpful for: Distinguishing epidermoids and other extraaxial cystic lesions, e.g, arachnoid cysts. Identifying small lesions which are not conspicuous on other sequences. Post-operative assessment; defining extent of the lesion and residual or recurrent disease.

References Bergui M, Zhong J, Bradac GB, Sales S. 2001. Diffusion-weighted images of intracranial cyst-like lesions. Neuroradiology 43: 824–829. Laing AD, Mitchell PJ, Wallace D. 1999. Diffusion-weighted magnetic resonance imaging of intracranial epidermoid tumors. Australas Radiol 43: 16–19.

Diffusion MR imaging in adult neoplasia

Case Study 20.2 Diffusion tensor imaging of glioma infiltration Stephen J. Price, FRCS, Alonso Peña, Ph.D. and Jonathan Gillard, MD. Addenbrooke’s Hospital, Cambridge, CB2 2QQ History Sixty year old female presenting with expressive dysphasia 1 year after resection of a right frontal glioblastoma (WHO Grade IV).

T1Post-Gd

T2

FA

Follow-up CT

Technique Conventional and DTI.

Imaging findings Conventional images demonstrate a left frontal recurrence without involvement of the corpus allosum. DTI imaging (FA map) shows abnormalities in the corpus callosum (arrowed). Plotting maps of p (the isotropic component) and q (the anisotropic component) for the corpus callosum reveals abnormalities. Follow up CT imaging 6 weeks later revealed tumor in the corpus callosum.

Discussion

Key points Gliomas diffusely infiltrate WM tracts. DTI can be sensitive to this infiltration.

1.5 pq diagram q (10 3 mm2s)

Glioma cells can infiltrate beyond the abnormalities seen on conventional imaging (Kelly et al., 1987). Since DTI is sensitive to the diffusion of water it can identify WM disruption due to glioma infiltration (Price et al., 2003).

1

0.5

0

Normal genu corpus callosum Splenium corpus callosum Infiltrated genu corpus callosum Tumor

0

0.5

1 p

(10 3

1.5

2

2.5

mm2/s)

References Kelly PJ, Daumas-Duport C, Kispert DB, Kall BA, Scheithauer BW, Illig JJ.1987. Imaging-based stereotaxic serial biopsies in untreated intracranial glial neoplasms. J Neurosurg 66: 865–874. Price SJ, Burnet NG, Donovan T, et al. 2003. Diffusion tensor imaging of brain tumours at 3T: a potential tool for assessing white matter tract invasion? Clin Radiol 58: 455–462.

327

328

Bradford A. Moffat, Thomas L. Chenevert and Brian D. Ross

Case Study 20.3 Differentiating gliomas from metastases with DTI Stephen J. Price, FRCS, Alonso Peña, Ph.D. and Jonathan Gillard, M.D. Addenbrooke’s Hospital, Cambridge, CB2 2QQ History A 74-year-old female presented with a right homonymous hemianopia. Biopsies of the lesion revealed it to be a WHO Grade IV glioblastoma. A 42-year-old male with known metastatic melanoma presented with headaches and a right upper quadrantanopia.

T2 Glioblastoma

T2 Metastasis

Technique Conventional and DTI.

DTI Glioblastoma

DTI Metastasis

Imaging findings

Discussion Whilst gliomas infiltrate WM tracts and disrupt them, metastases spread along vascular planes and leave WM intact (Nelson et al., 2000).

Key point Gliomas can be differentiated from metastases as they infiltrate WM tracts.

q (anistropic component) (10 3 mm2/s)

Tumors were identified in the left posterior temporal lobe in both patients. In the glioblastoma patient the DTI reveals a loss of anisotropy in the left visual radiation. The DTI in the metastasis patient reveals normal WM tracts below the obvious abnormality (arrowed). Measurements of pq show marked abnormalities for the glioblastoma patient but normal for metastasis patient.

1 Metastasis

0.5

Glioblastoma 0

0

0.5 1 1.5 p (isotropic component) (10 3 mm2/s)

Reference Nelson JS, von Deimling A, Petersen I, Janzer RC, 2000. Metastatic tumours of the CNS. In Pathology and Genetics of Tumours of the Nervous System. (Eds., Kleihues P, Cavenee WK), IARC Press, pp. 250–253.

2

21

Perfusion MR imaging in adult neoplasia Alan Jackson Professor of Neuroradiology, University of Manchester, UK

Key points • Angiogenesis, and increased vascular permeability, are characteristic of cerebral neoplasms; these processes can be imaged using perfusion MR imaging. • Most commonly, tumor perfusion is measured using rapid gradient T 2*-weighted imaging during bolus injection of gadolinium dimeglumine gadopentetate. • Care has to be taken to avoid blood–brain barrier leakage (itself of interest) affecting perfusion results. • Pharmacokinetic models are available for estimation of blood–brain permeability. • Cerebral blood volume (CBV) increases with tumor grade, and maybe helpful in identifying tumor recurrence, and peri-tumoral edema, and distinguishing malignant from benign lesions. • Extra-axial masses typically have higher CBV than intra-axial masses.

Introduction All normal and pathological tissues depend on their blood supply for an adequate supply of nutrients and removal of waste metabolic materials. As tissues develop, an adequate and appropriately structured vascular supply must also develop at the same time. This process, known as angiogenesis, is a common feature of many pathological tissues including tumors and may also be seen in inflammatory disease

(Folkman, 1990; Padhani and Husband, 2001). The angiogenic process is complex and can be stimulated by any one of several mechanisms. Typically, growth of tissue which has outstripped its local blood supply results in regional hypoxia and hypoglycemia which stimulates the release of local chemical messengers from the cells of the tissue itself. The best known of these messengers is the cytokine, vascular endothelial growth factor (VEGF). VEGF is a common and potent angiogenic stimulator which is found in many pathological tissues. It is released in response to local hypoglycemia and/or hypoxia and has several effects, each of which will improve metabolic supply. In the short term, VEGF will act directly on local capillaries to increase endothelial permeability resulting in an immediate increase in the supply of nutrients (Shweiki et al., 1995). This increase in permeability is also believed to form an important part of the metastatic mechanism, allowing passage of tumor cells into the circulation. In the medium-to-long term, VEGF will stimulate mitosis in endothelial cells from local blood vessels so that they divide and develop a new vascular infrastructure to supply the tumor. The angiogenic mechanism is also responsible for breakdown of local connective tissues which allows in-growth of new blood vessels. Where the angiogenic process fails, tissue development and growth cannot occur; novel antiangiogenic therapies are being developed which exploit this feature for the treatment of a wide range of cancers and other pathologies (Lund et al., 1998; Zhu et al., 2000; Jayson et al., 2002). The increasing understanding of the role of the angiogenic process in disease progression has led to increasing interest in methods for documenting the presence and 329

Alan Jackson

(a)

(b)

Fig. 21.1 Parametric maps of blood volume (a) and relative recirculation (rR) (b) in a grade IV glioma. The maps are derived from T2*-weighted data. rR indicates areas of irregular flow and poor perfusion and elevated values are shown in red (Jackson et al., 2002b).

3500 3000 rCBV (AU)

330

2500 2000 1500 1000 500 0

II

III Grade

IV

Fig. 21.2 Scattergram showing the relationship between blood volume and tumor grade in glioma.

activity of angiogenesis (Tofts and Kermode, 1991; Tofts et al., 1999; Zhu et al., 2000). In particular, there has been considerable interest in the development of reliable quantitative methods which can provide independent indicators of the status of, and changes in, microvascular structure (Tofts and Kermode, 1991; Tofts et al., 1999). In pathological tissues, the angiogenic process is often abnormal, leading to the development of distorted vascular beds, characterized by an excessive proportion of blood vessels and blood vessels with abnormal morphology and flow characteristics (Figures 21.1 and 21.2) (Kassner et al., 2000). Central

areas of a rapidly growing tumor will commonly exhibit inadequate blood flow due to the reduced local perfusion pressure resulting from a combination of inadequate vascularization and increased interstitial tumor pressure (Figure 21.1). Finally, the angiogenic neovasculature will exhibit increase endothelial permeability to medium- and large-sized molecules (Knopp et al., 1999a; Uematsu et al., 2000). It is important to realize that the increase in vascular permeability seen in neoangiogenic tissue is most commonly a direct effect of cytokine stimulation and can be rapidly reversed by inhibition of the active cytokine (Jayson et al., 2002). In contrast, vascular density, vascular tortuosity and other abnormalities of vascular structure represent the cumulative effects of the angiogenic process that has occurred to date. Due to this, the biological information contained in these two groups of measurements, perfusion and permeability, offers distinct and separate insight into the status of the angiogenic process. Attempts to quantify the features of the microvasculature using histopathological techniques are essentially unsatisfactory for a number of reasons. Pathological assessment relies on the acquisition of tissue samples which is clearly invasive and can be repeated only infrequently. Furthermore, many tumors and other pathological tissues demonstrate considerable heterogeneity in microvascular structure so that isolated regional biopsies may give

Perfusion MR imaging in adult neoplasia

misleading information. These limitations have stimulated the development of imaging-based methods for quantification of microvascular structure. These methods are commonly based on dynamic contrast enhanced imaging techniques using MR (DCE-MRI) or computed tomography (DCE-CT) data collection combined with analytic algorithms to calculate descriptive parameters related to microvascular structure. MR techniques may be based on T2-(susceptibility) or T1-(relaxivity) based-contrast mechanisms. This chapter will describe both DCE-MRI techniques using both susceptibility (DSCI-MRI) and relaxivity (DRCI-MRI) based techniques and the range of image analysis approaches available. Much of the basic theory of DSCE-MRI, and particularly its application for quantitative flow imaging have been discussed at length in Chapters 7 and 9. We will therefore limit the discussion of DSCE-MRI to problems specific to their application in enhancing tissues such as tumors. The chapter is designed to enable typical clinical radiologists to understand the possible applications of DCE-MRI and the benefits and disadvantages of specific approaches without recourse to mathematical theory, for those wishing to grasp the mathematical concepts and to develop a deeper understanding of the methodology a wide inclusive bibliography is supplied.

Microvascular features of pathological tissues It is worth considering the structural features of the microvasculature which will affect the behavior of injected media (Tofts and Kermode, 1991). Following an intravascular injection of a contrast medium, the bolus will pass through the vascular bed: entering through arterial vessels, passing through the capillary bed and draining into the venous system. The amount of contrast that passes into the vascular system will depend on the blood flow rate through the vessels and the contrast dose injected. Within any given voxel, the amount of intravascular contrast will depend on the proportion of the voxel formed by blood vessels. As contrast passes through the capillary bed, it will leak into the extravascular extracellular space (EES). The rate at which this leakage

occurs will reflect the difference in contrast concentration between the blood plasma and the EES. For any given concentration ratio, the amount of contrast that leaks will also be restricted by the permeability and surface area of the endothelial membrane. As contrast leaks into the EES, it will diffuse through the space so that the concentration of contrast will also depend on the size of the EES. Eventually, as the contrast concentration within the vascular space decreases, due to leakage into other tissues and renal excretion, contrast will begin to pass back from the EES into the vessels. It can be seen that the behavior of contrast material within any given voxel will be related to the concentration time course of contrast entering the arterial vessels, the regional blood flow, the local blood volume, the endothelial permeability, the surface area of the endothelium and the size of the EES. Analysis schemes for DCE-MRI are designed to identify surrogate markers which represent one of, or combinations of these biological features (Tofts and Kermode, 1991; Jayson et al., 2002; Li et al., 2003).

Collecting DCE-MRI data Data acquisition for DCE-MRI is a relatively straightforward process. The imaging technique must collect a time course series of spatially registered images with sufficient temporal resolution to allow accurate analysis and without significant slice-to-slice movement. For many analysis techniques, the imaging volume must include an appropriate large vessel which can be used to provide a surrogate input function to describe the time course of contrast concentration changes in the vessels supplying the tumor. The typical imaging strategy for T*2 data (cf. also Chapters 7 and 8) is to collect data using a fast-imaging technique such as single or multi-shot echo planar imaging (EPI) to produce a temporal resolution of approximately 2 s. During this 2 s acquisition window, it is usually possible to acquire in the region of 5–25 slices at a resolution of 128 128, depending on scanner specifications. The imaging sequence can be a gradient echo (GE), which will maximize T*2 weighting or alternatively a spin echo (SE) approach can be used, which will

331

332

Alan Jackson

minimize the signal contribution from large vessels. Some authors prefer the latter approach since it produces signal changes which predominantly reflect the passage of contrast through the capillary bed, although this is at the expense of temporal resolution. The data collection period needs to address only the first passage of the contrast bolus through the tissues so that the total length of the acquisition seldom exceeds 1.5–2 min. For T1-based techniques the data acquisition is typically based around a 2D or more commonly a 3D GE sequence. The temporal resolution required depends entirely on the analysis algorithms to be applied and may need to be in the order of 5 s or less for techniques which analyze the first passage of the contrast bolus (Li et al., 2003) whilst temporal resolutions of 20 s or more may be acceptable for standard analysis methods (Tofts et al., 1995, 1999). The length of the acquisition for first-pass techniques may be as little as 40 s (Jackson et al., 2002a) and can support breath-holding acquisitions but more conventional analysis approaches need considerably longer acquisitions in the region of 5–10 min. For all methods the patient should be comfortably positioned, with adequate cushioning to reduce movement and light restraining straps should be used in the same way as they would be used for normal MRI. This level of restraint, combined with the relatively short acquisition times for T2* and firstpass-based T1 techniques usually results in data with sufficiently little movement that data co-registration is seldom required. Longer T1-based acquisitions in the range of minutes are profoundly affected by normal physiological motion, however for the imaging of intracranial neoplasms adequate motion-free data collection is usually possible. A series of at least five pre-contrast images should be collected prior to the passage of the bolus and many centers will collect for up to 1 min to provide a large number of precontrast images to improve the estimation of the signal intensity baseline during analysis. The contrast agent is administered by intravenous injection and the injection technique must be carefully standardized. Some T1-based analysis techniques require slow infusion of contrast but most currently used techniques for both T1- and T2*-based methods

require injection as a fast bolus. A standard contrast dose (0.1 mmol kg 1) is adequate in most cases although some centers use double this dose in order to improve signal-to-noise ratio (SNR). The use of an automated pressure injector is recommended and should be programmed to deliver the contrast over approximately 4 s followed by a saline flush of at least 25 ml delivered at the same rate. A careful manual injection technique can also produce acceptable and reproducible results. The injection should be given through a large cannula, preferably introduced into a large antecubital vein; the cannula should be at least 18G for manual injection to reduce the resistance to flow.

DSCI-MRI The use of T*2 dynamic imaging acquisitions is commonly performed for the calculation of perfusion and local blood volume in brain tissue (cf. Chapters 7 and 9). The presence of an intact blood–brain barrier (BBB) in normal brain means that no contrast leakage into the EES will be seen and the data can be treated as a purely intravascular or blood pool marker. For perfusion calculations, this is beneficial since T*-weighted sequences display changes due to 2 the effect of contrast on both blood vessels and the surrounding tissues (Barbier et al., 2001). In capillary beds where the cerebral blood volume (CBV) is low the signal changes on T*-weighted images will 2 be proportionally larger than that which results from contrast in large vessels. The signal changes observed during the passage of a contrast bolus through the vessels can be transformed to contrast concentration maps and these can subsequently be used to derive quantitative images representing physiological parameters. A number of analysis techniques have been described, and a common feature of many is the use of a gamma-variate fitting procedure to define the shape and position of the firstpass bolus (Figure 21.3). The need for this step arises from the signal changes which occur after passage of the contrast bolus as contrast recirculates into the cerebral circulation from the periphery. As a result of this recirculation the contrast concentration fails to

Perfusion MR imaging in adult neoplasia

100 80

C (t )

60 40 20 0

20

40

20

0

20 40 Time (s)

60

80

Fig. 21.3 Contrast concentration time course data from a T2*-weighted acquisition in normal brain tissue. The raw data can be seen as individual points and the fitted gammavariant curve as a continuous line. Note that the contrast concentration data does not return to sleep preenhancement levels following passage of the contrast bolus.

return to pre-enhancement levels. The use of a curve-fitting technique allows us to estimate what the first-pass bolus would have looked like if no recirculation had occurred. Parametric images of regional CBV (rCBV) can be derived from the area under the contrast concentration time course curve and indicators of bolus arrival time can also be accurately derived. Measurements of absolute blood flow are more complicated and subject to errors due to effects such as bolus dispersion and arrival time delays (Calamante et al., 2000). Another interesting parameter provides an estimate of abnormality during the recirculation phase of the contrast passage. This “relative recirculation” (rR) parameter quantifies any abnormal elevation of the contrast concentration in the period immediately after the passage of the contrast bolus (Figure 21.1) (Kassner et al., 1999). In theory the rR will be increased by local vascular factors such as absolute flow rate, flow rate heterogeneity and therefore by local perfusion pressure. Contrast leakage and tissue enhancement The analysis of DSCI-MRI contrast bolus studies assumes that the signal change observed results

entirely from contrast within the blood vessels. However, leakage of contrast into the interstitial space will cause additional signal changes, principally by relaxivity mechanisms (Figure 21.4). Susceptibilitybased imaging methods offer the opportunity to separate these relaxivity and susceptibility-based effects and to produce images in which the effect of contrast leakage is eliminated or minimized. In theory, the use of a true intravascular contrast media, or a contrast with negligible relaxivity effects, such as iron oxide, would allow pure susceptibility measurements. In fact, the restrictions of standard contrast media force us to use methods to separate as far as possible the susceptibility and relaxivity effects. The use of techniques with reduced T1 sensitivity such as low flip angle GE-based sequences has become a common technique (Aronen et al., 1994; Kuhl et al., 1997a; Maeda et al., 1997; Kassner et al., 1999). This technique effectively removes relaxivity effects although some workers have still observed residual effects in rapidly enhancing tumors (Maeda et al., 1997). The major problem with this method is the loss of SNR produced by the reduction in flip angle (Kassner et al., 1999). However, this can be partially compensated by increased contrast doses (Aronen et al., 1994). Another approach to reducing T1 sensitivity is to use a dual echo technique in which the T1-weighted first echo is used to correct the predominantly T2-weighted second echo (Miyati et al., 1997a). This method is a simple and effective way to remove relaxivity effects. Unfortunately, the requirement for two echoes places considerable demands on the sampling time and inevitably restricts the number of samples, and therefore slices which can be obtained. In addition, the use of a calculated parameter of this type derived from single pixels in noisy data sets produces a mathematical coupling effect on background noise which adversely affects the SNR (Kassner et al., 1999). The use of pre-enhancement techniques is based on the signal changes that occur in response to changing concentrations of contrast media. The change in signal intensity with contrast medium concentration resulting from T1 shortening is bi-exponential so that for any given sequence, there exists a plateau phase during which signal intensity

333

0

25

50 Time (s)

75

1000

1800

Intensity 2000 2200

Intensity 1200 1400

2400

1600

Alan Jackson

0

25

50 Time (s)

75

Fig. 21.4 Contrast concentration time course data from a T2*-weighted acquisition in a large meningioma. The data on the left shows the signal change observed following a standard injection of contrast. Note the initial dip in contrast concentration due to the T2*-susceptibility effects of contrast in the vascular space and the almost immediate increase in signal due to T1 mediated contrast mechanisms following leakage of contrast into the extracellular space.

14000 12000 10000 8000 SI

334

6000 4000 2000 0

35° 15°

25° 10°

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 Gd DTPA-BMA

Fig. 21.5 The signal intensity (SI) changes with increasing concentrations of contrast in a phantom experiment. The four lines represent GE acquisitions with differing flip angles. Note that with large flip angles there is a marked enhancement effect due to T1 shine-through but that this is attenuated and reaches a plateau phase even with the flip angle of 35°. With smaller flip angles enhancement effect is far less but note that signal intensity is also proportionately reduced.

remains relatively constant (Figure 21.5). The position and length of this plateau phase will vary with the sequence. The effect of this response curve is that pre-enhancement of tumors will reduce the relaxivity-based signal intensity responses to subsequent contrast doses. The major problem with this

approach is that the efficiency of the technique is dependent on the interstitial contrast concentration at the time of the bolus passage. Since tumors show differing contrast diffusion rates (Bullock et al., 1991; Gowland et al., 1992) this concentration cannot be accurately predicted, although it can be measured (Kassner et al., 1999). Elimination of T1 “shinethrough” requires an interstitial contrast concentration gadolinium dimeglumine gadopentetate (Gd-DTPA-BMA) greater that 0.4 mmol l 1 and this figure will differ slightly depending on the relaxivity of the contrast agent (Kassner et al., 1999). Another problem with pre-enhancement techniques is the residual contrast effects seen in sequential dynamic susceptibility contrast (DSC) experiments (Levin et al., 1995). The use of sequential injections demonstrates a change in the signal response curve so that it no longer conforms to a gamma-variate pattern and demonstrates elevation of R2 during the recirculation phase. The cause of these effects is unclear, although they do not appear to result from relaxivity changes (Levin et al., 1995; Kuhl et al., 1997a). The effect of these changes is that analysis of the perfusion data using standard gamma-variate fitting techniques becomes less accurate and may result in significant levels of fitting error (Kuhl et al., 1997b). Each of these methods has specific advantages and disadvantages. The choice of technique must

Perfusion MR imaging in adult neoplasia

DRCE-MRI As we have seen, the use of DSCI-MRI will allow us to measure a number of parameters which describe the angiogenic process, most notably rCBV and rR. Measurement of endothelial permeability, endothelial surface area and the size of the EES requires quantitative analysis of contrast leakage from the vascular space into the EES. Using T2- or T2*-weighted imaging to quantify contrast leakage is difficult for a number of reasons and, although such techniques have been described, they suffer from problems related to the non-linearity of the signal changes observed and the antagonistic effects of relaxivity and susceptibility mechanisms on the MR signal. For these reasons, T1-weighted imaging sequences are most commonly used for data collection. These T1-weighted data can be used to generate contrast concentration maps which can then be analyzed using standard pharmacokinetic models (Tofts et al., 1999). The choice of imaging sequence for DRCE-MRI is driven by a number of considerations. Since we are interested in the changes in contrast concentration over time the temporal resolution of the imaging sequence will be constrained by the analysis technique to be employed (vide infra). The main consideration when choosing an appropriate temporal

3 Gd-DTPA (mmol kg 1)

be made by considering the requirements of the individual study. Dual echo methods will reliably eliminate relaxivity effects but suffer from poor SNR and limited sampling volume. Low flip angle methods provide reliable elimination of relaxivity effects as long as repetition times (TR) are adequate. These methods are fast, but also suffer from poor SNR in normal tissue. Pre-enhancement techniques allow the use of sequences which are sensitive to relaxivity changes and consequently provide good SNR. However, the effect of pre-enhancement on T1 shine-through will be highly dependent on tumor contrast concentration which will vary from tumor to tumor. In addition, pre-enhancement causes residual bolus effects, which affect the dynamics of signal change during subsequent bolus passage in normal tissue.

335

First-pass peak

2.5

Cp – concentration in capillary plasma Ce – concentration in extra-cellular space Ct – concentration in whole tissue

2 1.5 1 0.5 0

0

60

120

180

240 300 Time (s)

360

420

Fig. 21.6 Contrast concentration time course data in tumor capillary bed, extravascular space and in the voxel as a whole following injection of a bolus of contrast.

resolution is the shape of contrast concentration curve with time in plasma, which is commonly called the arterial input function (AIF) (Figure 21.6). If a bolus injection of contrast is employed, accurate quantification of the initial passage of contrast bolus will require a high temporal resolution usually of 5 s or less. This time constraint will limit the spatial resolution of the imaging sequence and the amount of tissue that can be encompassed in any given imaging volume. Most centers use a simple multi-slice GE sequence, which on current clinical scanners can provide an imaging matrix in the region of 256 256 25 within the 5-s period. The imaging sequences currently used as our center are shown in Table 21.1. In order to provide adequate data to allow clear separation of the effects of contrast diffusion through the EES, the dynamic data collection will typically need to continue in excess of 5 min. Two other constraints on the imaging sequences must also be considered if pharmacokinetic analysis of the data is to be performed. The pharmacokinetic analysis will require contrast concentration data rather than signal change data (Figure 21.7). The relationship between signal change and contrast concentration is non-linear and will depend on the baseline T1 value of the voxel. It is therefore necessary to produce quantitative T1 images of the imaging volume prior to contrast injection. This can be time consuming and complex; most centers have developed quantification methods using multiple

480

336

Alan Jackson

Table 21.1. Typical imaging protocol for DCE-MRI Scan

Sequences

TR/TE/

Matrix

t(s)

Tdy(min)

FE (2°) FE (10°) FE (35°) T1dy

3D T1W-FE 3D T1W-FE 3D T1W-FE 3D T1W-FE

4.3–7/1.1–1.6/2° 4.3–7/1.1–1.6/10° 4.3–7/1.1–1.6/35° 4.3–7/1.1–1.6/35°

128 128 25 128 128 25 128 128 25 128 128 25

– – – 5.1⬃8.7

– – – 11⬃12

FE: field echo; t: temporal resolution; Tdy: length of dynamic acquisition.

Fig. 21.7 Top: DCE images in a patient with glioma using the imaging sequence shown in Table 21.1. Bottom: calculated contrast concentration maps from the images shown above.

T1-weighted images acquired with varying flip angles to allow T1 calculations (Zhu et al., 1999, 2000). Pharmacokinetic analyses also require measurement of the contrast concentration changes with time in the vessels supplying the tissue. Identification of an appropriate AIF can be difficult and is further complicated by the additional signal changes produced by inflow effects on most imaging sequences (Li et al., 1999, 2000; Zhu et al., 2000). Using a multi-slice GE acquisition, inflow effects are usually negligible in all slices except for those at the edges of the volume so that an AIF can be acquired from an appropriate vessel on one of the remaining slices.

Data analysis The extraction of appropriate parameters to describe the microvasculature from DRCE-MRI data is extremely complex and many analysis approaches are available. The choice of analysis approach will depend on the expected changes in the tissue and the features of the microvasculature that are of particular interest. The simplest approach relies on quantification of the signal changes observed which avoids the complexity of calculating contrast concentrations from the baseline data. A number of such metrics have been described, and most characterize the shape of the signal time course

Perfusion MR imaging in adult neoplasia

Table 21.2. List of abbreviations and standard terms used in pharmacokinetic analyses Symbol

Definition

Units

Ktrans Kep ve vp Cp C(t) Cv

Volume transfer constant between blood plasma and EES Rate constant between EES and plasma Volume of the EES per unit volume of tissue Blood plasma volume Tracer concentration in arterial blood plasma Concentration of contrast medium in the voxel at time t Intravascular component of C(t), which equals tracer concentration in plasma multiplied by vp Extravascular extracellular component of C(t), which equals tracer concentration in EES multiplied by ve Total permeability of the capillary wall Permeability surface area product per unit mass of tissue Extravascular extracellular space

min 1 min 1 None ml mM mM mM

Ce P PS EES

curve based on the maximal amplitude of the signal and the time taken to reach this value. Other metrics compare the signal intensity achieved in any given period of time. Typical metrics include T90 (Stack et al., 1990) which is the time taken to reach 90% of the maximal enhancement value and the maximal intensity change per time interval ratio (MITR) (Flickinger et al., 1993) which measures the maximal rate of enhancement. These simple metrics suffer from problems which have limited their application and led many authors to adopt the more complex pharmacokinetic analysis techniques. Firstly, metrics are based entirely on signal change characteristics which will reflect both intra- and extravascular contrast concentrations; separation of signal change effects due to blood flow and contrast leakage is therefore impossible. More worryingly, these measurements will be unpredictably affected by variations in imaging protocol, including those, such as receiver gain, that may change from examination to examination. Despite these limitations, simple measurements based on signal change alone can be useful diagnostically and have been used in a number of clinical applications (Knopp et al., 1999b). Pharmacokinetic analyses of T1-weighted DCEMRI data have a number of theoretical advantages (Tofts and Kermode, 1991; Tofts, 1997; Tofts et al.,

mM cm min 1 ml min 1 g 1 None

1999). The use of pharmacokinetic models leads to the derivation of parameters which are independent of the scanning acquisition protocol or any features associated with it. In theory, such parameters should reflect only tissue characteristics, supporting the use of these measurements in multi-center studies employing varying image acquisition protocols and equipment (Padhani and Husband, 2001; Jayson et al., 2002). Table 21.2 shows a synopsis of the terms commonly used to describe specific features of microvasculature in pharmacokinetic analyses (Tofts et al., 1999). In practice, pharmacokinetic analysis is complex and the choice of pharmacokinetic model controls the range of parameters that can be extracted. Each of the pharmacokinetic analysis approaches uses curve-fitting techniques to characterize the AIF and the tissue contrast concentration curve. These two functions are then used to derive the parameters which control the relationship between AIF and tissue contrast content. The simplest of the pharmacokinetic models, such as that described by Tofts and Kermode (1991), use a single AIF and time course data describing contrast concentration from individual voxels to calculate the size of the EES (ve) and the bulk transfer coefficient Ktrans (Figure 21.8). The transfer constant Ktrans is simply a mathematical function which describes the relationship between the AIF and contrast

337

338

Alan Jackson

R1

K trans

ve

Fig. 21.8 Maps of initial 1/T1 (R1), Ktrans and ve in a patient with glioma calculated using the model of Tofts and Kermode (1991).

concentration changes occurring in the voxel. The measurements of Ktrans will be affected by blood flow, blood volume, endothelial permeability and endothelial surface area. Changes in any of these variables can produce observable changes in Ktrans and the specific contribution of the individual components cannot be identified. This simple model, however, assumes that the signal changes within the measurement voxels will result entirely from extravasated contrast medium within the EES. This gives rise to significant errors, manifest as artificially elevated measured values of Ktrans, in voxels which contain intravascular contrast agent. Despite these shortcomings, the model described by Tofts and Kermode (1991) is widely used (Parker et al., 1997). Many workers have attempted to refine the pharmacokinetic analysis to provide more accurate estimates of individual microvascular parameters, particularly permeability surface area product and blood flow. One reason for the impetus to refine of analysis techniques is that they are increasingly used in drug development and discovery and, particularly, for the study of new anti antiangiogenic therapies (Jayson et al., 2002). As we have described, the angiogenic cytokine VEGF has a specific action in promoting endothelial permeability; measurements of endothelial permeability, uncontaminated by other factors, are therefore highly desirable. The basic pharmacokinetic model described by Tofts

and Kermode can be modified to model specifically the signal contribution produced by contrast medium within the plasma (Tofts et al., 1999). This reduces errors due to the so-called “pseudopermeability” effect where intravascular contrast gives rise to falsely elevated values of Ktrans (Figure 21.9). The Ktrans values from this modified model will differ significantly from those obtained with the classic model and will more accurately reflect changes in permeability surface area product, although they will still be dependent on adequate blood flow to the tissue to support contrast leakage. It is important to realize that the exact meaning of the Ktrans variable depends on the method of analysis used. More complex pharmacokinetic models such as those described by St. Laurence and Lee (1998) allow direct estimation of local tissue blood flow (F) in addition to ve, vb and Ktrans (Figure 21.10). In this model, the Ktrans will effectively represent the permeability surface area product. It seems clear that models such as these are more desirable than simpler approaches to the analysis, however, separate identification of the extra-fitting parameters requires more accurate and reliable curve fitting and is associated with increased variability and susceptibility to noise. The choice of analysis techniques is therefore not straightforward and must be based on the likely quality of the data to be obtained and the specific biological question to be answered.

Perfusion MR imaging in adult neoplasia

Fig. 21.9 Maps of Ktrans in a patient with high-grade glioma. The map on the left was calculated using the model of Tofts and Kermode (1991). Note apparent areas of high Ktrans in the region of blood vessels. The map on the right represents Ktrans calculated using the first-pass pharmacokinetic model described by Li et al. (2000). Note that no “pseudo-permeability” effects are now seen.

0.5

F PS vb ve

0.45 Gd-DTPA (mM)

0.4 0.35

(b)

0.22 ml min 1 ml 1 0.03 ml min 1 ml 1 0.02 ml ml 1 0.12 ml ml 1

0.5 0.45 0.4 Gd-DTPA (mM)

(a)

0.3 0.25 0.2 0.15

0.3 0.25 0.2 0.15 0.1

0.05

0.05 0

1

2

3 4 Time (min)

5

6

7

0.47 ml min 1 ml 1 0.12 ml min 1 ml 1 0.04 ml ml 1 0.20 ml ml 1

0.35

0.1 0

F PS vb ve

0 0

0.5

1

1.5

2 2.5 3 Time (min)

3.5

4

4.5

Fig. 21.10 The results of curve fitting and analysis of DCE-MRI data in patients with grade III (a) and grade IV (b) gliomas. Blue curves show the AIF, red shows the data and curve fit to the tumor data. The analysis has been performed on a region of interest (ROI) using the model described by St. Lawrence and Lee (1998). The analysis has been performed on ROI data because of the requirements for high SNR but has enabled separate calculation of flow, permeability surface area product (PS), ve and vb (courtesy of Dr. D. Buckley, University of Manchester, UK).

Clinical applications of angiogenesis imaging in cerebral tumors DSCI -MRI Gliomas Despite a large number of studies describing quantitative imaging angiogenesis in brain tumors, none

of these techniques has yet passed into routine clinical use. There is, however, considerable evidence that such quantitative techniques can provide valuable clinical data concerning tumor type, tumor grade and therapeutic response. In gliomas, tumor capillary blood volumes measured by DSCI-MRI have been shown to correlate with and predict tumor grade (Figures 21.2 and 21.11–21.15(a and b)

339

Alan Jackson

1.4

Normal Tumor I Tumor II

Gd (mmol l 1)

1.2 1.0 0.8 0.6 0.4 0.2 0.0 0

10

20 Time (s)

30

40

Fig. 21.11 Illustration of breath-hold dynamic data in the liver of a patient with metastatic disease. Parametric images illustrate the contrast arrival time (top right), Ktrans (bottom left) and blood volume (bottom right).

2.0

Fitted K trans (min 1)

340

1.5

1.0

0.5

0.0 0.0

0.4

0.8

1.2

0.0

0.4

0.8

Actual

K trans

(min 1)

1.2

0.0

0.4

0.8

1.2

Fig. 21.12 Results from a Monte Carlo modeling experiment to examine the relationship between true and estimated values of Ktrans using the first past technique. From left to right the data represent increasing signal to noise ratio in the imaging data. Notice that the analysis technique is relatively insensitive to increasing levels of signal to noise ratio but consistently underestimates Ktrans at high values.

Perfusion MR imaging in adult neoplasia

0.8

0.040 Day 2 Kep means (min 1)

0.6

0.4

0.2

0.0

Scan1 Scan2 Glioma

GM

0.030

0.020

r  0.997

0.010

Scan1 Scan2 Breast

Fig. 21.13 Reproducibility measurments of Ktrans in patients with cerebral glioma and primary breast tumors. Scans were performed 48 hours apart and the analysis uses standard method described by Tofts and Kermode.

(Aronen et al., 1994, 1995, 2000; Knopp et al., 1999a). More importantly rCBV maps identify areas of malignant transformation or dedifferentiation, allowing more accurate targeting of stereotaxic biopsies and therefore more accurate estimation of tumor grade (Knopp et al., 1999a). Histological comparisons show close relationships between rCBV values within tumors and histological features indicative of tumor aggression including mitotic activity and vascularity (Aronen et al., 1995; Sugahara et al., 1998). Direct comparison between rCBV mapping and other indicators of malignancy such as flourodeoxyglucose positron emission tomography (PET) shows close agreement between local rCBV values and glucose uptake and moderate significant correlation between maximal glucose uptake and rCBV (n  21; r  0.572; P  0.023) (Aronen et al., 2000). Similar comparisons with thallium-201 single photon emission CT (SPECT) show greater sensitivity to glioma grade (Lam et al., 2001) and also higher sensitivity for demonstrating early tumor recurrence after therapy. In addition to potential uses in identifying grade and histological heterogeneity it has also been shown that susceptibility-based methods may be helpful in differentiating between primary gliomas and solitary cerebral metastasis on the basis of difference in peri-tumoral rCBV measurements (Cha et al., 2000, 2002). In metastatic tumors, peri-tumoral edema represents pure

0.000 0.000

Day 2 Kep 97.5% percentiles (min 1)

K trans (min 1)

GIII

0.010 0.020 0.030 Day 0 Kep means (min 1)

0.040

0.120 GIII

GM

0.100 0.080 0.060 0.040 r  0.984 0.020 0.000 0.000 0.020 0.040 0.060 0.080 0.100 0.120 Day 0 Kep 97.5% percentiles (min 1)

Fig. 21.14 Reproducibility of Ktrans measurements using the first past leakage profile model. Notice the excellent reproducibility not only in mean values but also in the values of the upper 2.5% of the data sets. This is of particular importance in the study of angiogenic tissues where the angiogenic process can be expected to stimulate increased values within the distribution. (r is correlation coefficient)

vasogenic edema caused by increased interstitial water due to microvascular extravasation of plasma fluid and proteins through the inter-endothelial space (Strugar et al., 1994, 1995). In high-grade gliomas the peri-tumoral region represents a variable combination of vasogenic edema and tumor cells infiltrating along the perivascular space (Strugar et al., 1995). Jackson et al. (2000b) have

341

Alan Jackson

(a) 500

Astrocytoma grade III

0.005

0.01 0.015 K trans

0.02

0.025

Number of pixels

0

300 200

100

Glioblastoma multiforme

0.013 0.083 0.113 0.183 0.213 0.283 0.313 0.383 0.413 0.483 0.513 0.583 0.613 0.683 0.713 0.783

0

rR GII (b) 300 0.005

0.01 0.015 trans K

0.02

0.025

Fig. 21.15 The distribution of individual pixel values in a grade III (a) and grade IV (b) glioma. Each point represents one pixel. Notice the increases in pixels with high values in the high-grade tumor.

described a correlation between the skewness of the distribution of rR values in glioma and tumor grade (Figures 21.16 and 21.17). Unlike rCBV and rCBF measurements the rR parameter is affected only by areas of local ischemia, decreased perfusion pressure or vascular tortuosity. Changes are therefore typically seen at the boundary between well-vascularized peripheral growing tissue and central tumor necrosis in high-grade gliomas (Figure 21.1). Abnormalities of rR affected only a small subpopulation of the tumor pixels and presumably represent areas of incipient ischemia where the ischemic drive for production of angiogenic cytokines such as VEGF is high. The limitation of abnormalities of rR to small subgroups of pixels within the tumor explains the lack of any relationship between mean values and tumor grade in the presence of a strong correlation between skewness of the distribution and grade. Other tumor types Susceptibility contrast MRI has also been reported to have some clinical benefit in a number of other

Number of pixels

0

Skewness  1.22

200

100

0

0.013 0.083 0.113 0.183 0.213 0.283 0.313 0.383 0.413 0.483 0.513 0.583 0.613 0.683 0.713 0.783

(b) 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0

Skewness  0.45

400

vl

(a) 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0

vl

342

rR GM Fig. 21.16 The distribution of rR values in a grade III (a) and grade IV (GII) (b) glioma (GM). Notice the normal distribution of rR values in the lower-grade tumor and the increased scheme in the distribution in a high-grade tumor.

tumor types. Comparisons of rCBV values between common extra- and inter-axial tumors shows that extra-axial masses typically have higher values. This may be helpful in differentiating between intra- and extra-axial masses where other features are equivocal (Sugahara et al., 1999a). Measurements of rCBV can also help to differentiate between meningiomas, schwannomas and neurinomas. Meningiomas having higher values of rCBV

Perfusion MR imaging in adult neoplasia

Skewness of rR

4.50 4.00 3.50 3.00 2.50 2.00 1.50 1.00 0.50 0.00

0.50

II

III

IV

Grade Fig. 21.17 The relationship between histologic grade and the skewness of rR measurements in a group of patients with glioma.

than those seen in neurinomas or schwannomas (Maeda et al., 1994; Miyati et al., 1997; Sugahara et al., 1999a; Uematsu et al., 2000). rCBV can provide supplementary information to differentiate between malignant lymphoma and glioma because the absence of tumor neovascularization of malignant lymphoma leads to low rCBV, which is in contrast to those of malignant gliomas (Sugahara et al., 1999b). Cha et al. in their series of 19 consecutive patients (Cha et al., 2002) with primary cerebral lymphoma found the maximum rCBV ranged from 0.42 to 3.41 (mean, 1.44  0.67) compared to mean rCBV of 5.5  4.5 in 51 patient with Glioblastoma multiforme (GBM) (P  0.01) (Frazzini et al., 1999). Cha et al. (2002) have also identified DSCI-MRI as a useful diagnostic tool in differentiating tumefactive demyelinating lesions (TDL) from intracranial neoplasms. The rCBV values of TDLs ranged from 0.22 to 1.79 (n  12), with a mean of 0.88  0.46 (SD) compared to rCBV values of 1.55–19.20 (n  11), with a mean of 6.47  6.52 in intracranial neoplasms (P  0.009). DRCE-MRI Gliomas As early as 1992, simplistic metrics based on estimations of the time taken to reach maximal enhancement or some proportion of maximal enhancement was shown to relate to tumor type (Fujii et al., 1992;

Nagele et al., 1993). Studies comparing different tumors found significant differences in the early enhancement pattern which were believed to relate both to tumor vascularity and to the rates of contrast leakage into the interstitial space. Later studies concentrated on the use of derived physiological parameters using pharmacokinetic analysis (Andersen and Jensen, 1998) in an attempt to separate variations in enhancement pattern resulting from local variations in blood volume from those resulting from variations in blood flow and endothelial permeability. In an early study of this type Andersen and Jensen (1998) examine differences in tumor capillary permeability, Ktrans and extracellular distribution volume (ve) in human intracranial tumors based on a two compartment pharmacokinetic model (Ohno et al., 1978; Paulson and Hertz, 1983; Larsson et al., 1990; Tofts and Kermode, 1991). Results from 17 brain tumor patients (seven glioblastoma, four metastasis, and six meningioma) showed that Ktrans was higher in meningiomas compared to glioblastomas and metastasis. Studies comparing tumor grade and estimates of Ktrans show a close correlation in gliomas with higher values seen in grade III and grade IV tumors and in low-grade minimally enhancing masses (Roberts et al., 2000, 2001). The studies also showed strong correlation between Ktrans and mitotic indices derived from histological samples and that the correlation between Ktrans and grade was higher than the correlation between mitotic indices or fractional blood volume measurements and grade suggesting high sensitivity for aggressive behavior in glial tumors.

Other tumor types Zhu et al. (2000) calculated Ktrans and ve maps in patients with meningioma, glioma and acoustic neuroma. Ktrans in acoustic neuromas was found to be consistently lower than those observed in meningioma and glioma, however, and observation of interest was that measurements of ve (the size of the EES) were higher in meningioma than in glioma and were consistently highest in the acoustic neuroma (P  0.001) (Figure 21.18). These results agree with those of previous investigators (Long, 1973) who demonstrated very large extracellular spaces in

343

Alan Jackson

0.7

Meningiomas

Gliomas

Acoustics

0.6 0.5 vl

344

Transitional

0.4 Fibroblastic

0.3

GIII GIII

0.2

Angioblastic GIII

Glioblastoma

0.1 0

Glioblastoma

0

0.1

0.2

0.3

0.4

0.5

K trans Fig. 21.18 The relationship between EES (vl) and transfer coefficient (Ktrans) in a group of patients with acoustic neuroma (triangles), meningioma (diamonds) and glioma (squares). Notice the increased transfer coefficient is in aggressive gliomas and meningiomas and the high values of vl in acoustic neuromas.

schwannomas using fluorescence and electron microscopy. Several workers have shown relationship between the severity of peri-tumoral or edema in meningioma, VEGF expression and measure values of Ktrans (Bitzer et al., 1998). Therapeutic monitoring Several workers have documented radiationinduced changes in normal brain and tumors including gliomas and meningioma (Wenz et al., 1996; Fuss et al., 2000). These studies have shown short-term increases and medium-to-long-term decreases in Ktrans in response to radiation therapy (RT) in both normal brain and tumor. More importantly some studies have shown that DRCE-MRI data can differentiate between patients who show subsequent local tumor control and those who do not (Hawighorst et al., 1998). Other workers have also suggested that DCE techniques may be useful in differentiating between tumor recurrence, characterized by high CBV and Ktrans, and radiation necrosis characterized by low values (Song et al., 1984; Frahm et al., 1986; Reinhold and Endrich, 1986; Alavi et al., 1988; Rosen et al., 1991, 1993; Maeda et al., 1993; Boxerman et al., 1995). However, radiation necrosis represents a heterogeneous process, with features resembling inflammation. Immature vessels may grow into previously necrotic areas (Gobbel et al., 1992), and viable tumor cells

may still be found in the areas with decreased blood volume. Therefore, when this technique is used to monitor irradiated areas, the risk of overlooking the active tumor sites cannot be excluded. Further studies are necessary (Sugahara et al., 1999a). One of the main proposed roles for DRCE-MRI is to detect and quantify responses to novel antiangiogenic therapies. Measurements of Ktrans and regional blood volume have been used successfully in a number of studies of peripheral tumors (Figures 21.19 and 21.20). Due to the anxiety that anti-angiogenic therapy may cause cerebral hemorrhage, no systematic trials of specific antiangiogenic therapies in cerebral tumors have been conducted. In animal models (Gossmann et al., 2002), the effects of a neutralizing anti-VEGF antibody on tumor microvascular permeability have been evaluated; and demonstrated significant inhibition of tumor microvascular permeability (6.1  3.6 ml min 1/100 cc min 1), compared to the control, saline-treated tumors (28.6  8.6 ml min 1/ 100 cc min 1) together with significant suppression of tumor growth (P  0.05).

Conclusions Quantitative characterization of microvascular structure using DCE-MRI is a powerful tool capable of providing valuable information for clinical purposes and for therapeutic trials. The data acquisition and image analysis protocols must be independently selected for individual studies following consideration of which parameters are of interest, the tissue characteristics and the spatial and temporal resolution required in the imaging sequence. Although pharmacokinetic analyses are complex, they yield biological parameters of direct relevance to tissue characterization in a way that is reproducible and independent of imaging parameterrelated variables. Continued improvements in the design of imaging equipment and analysis algorithms are progressively improving the specificity of biological parameters which can be calculated, allowing detailed quantitative characterization of microvascular structure in a wide range of pathological tissues.

Perfusion MR imaging in adult neoplasia

(a)

(b)

K trans

rBV trans

Fig. 21.19 Parametric images of K (a) and blood volume (BV) (b) in a patient with metastatic disease before (top) and after (bottom) treatment with an anti-VEGF antibody (Jayson et al., 2002). Note the changes in tumor Ktrans and BV following treatment.

% Maximum K trans

100

0.3 mg kg 1

80 60

1–10 mg kg 1

40 20 0 Day 0

Day 2

Day 35

Fig. 21.20 Changes in Ktrans occurring in a group of patients with various endothelial cancers 2 and 35 days after treatment with anti-VEGF antibody (Jayson et al., 2002).

REFERENCES Alavi JB, Alavi A, et al. 1988. Positron emission tomography in patients with glioma. A predictor of prognosis. Cancer 62(6): 1074–1078. Andersen C, Jensen FT. 1998. Differences in blood–tumourbarrier leakage of human intracranial tumours: quantitative

monitoring of vasogenic oedema and its response to glucocorticoid treatment. Acta Neurochir (Wien.) 140(9): 919–924. Aronen HJ, Gazit IE, et al. 1994. Cerebral blood volume maps of gliomas: comparison with tumor grade and histologic findings. Radiology 191(1): 41–51. Aronen HJ, Glass J, et al. 1995. Echo-planar MR cerebral blood volume mapping of gliomas. Clinical utility. Acta Radiol 36(5): 520–528. Aronen HJ, Pardo FS, et al. 2000. High microvascular blood volume is associated with high glucose uptake and tumor angiogenesis in human gliomas. Clin Cancer Res 6(6): 2189–2200. Barbier EL, Lamalle L, et al. 2001. Methodology of brain perfusion imaging. J Magn Reson Imaging 13(4): 496–520. Bitzer M, Opitz H, et al. 1998. Angiogenesis and brain oedema in intracranial meningiomas: influence of vascular endothelial growth factor. Acta Neurochir (Wien.) 140(4): 333–340. Boxerman JL, Hamberg LM, et al. 1995. MR contrast due to intravascular magnetic susceptibility perturbations. Magn Reson Med 34(4): 555–566. Bullock PR, Mansfield P, et al. 1991. Dynamic imaging of contrast enhancement in brain tumors. Magn Reson Med 19(2): 293–298.

345

346

Alan Jackson

Calamante F, Gadian DG, et al. 2000. Delay and dispersion effects in dynamic susceptibility contrast MRI: simulations using vas. Magn Reson Med 44(3): 466–473. Cha S, Knopp EA, et al. 2002. Intracranial mass lesions: dynamic contrast-enhanced susceptibility-weighted echoplanar perfusion MR imaging. Radiology 223(1): 11–29. Cha S, Law M, et al. 2000. Peritumoral region: differentiation between primary high-grade neoplasma and solitary metastasis using dynamic contrast-enhanced T2*-weighted echoplanar perfusion MR imaging. In Proceeding of the 38th Annual Meeting of the American Society of Neuroradiology. American Society of Neuroradiology, Atlanta, GA, p. 22. Flickinger F, Allison J, et al. 1993. Differentiation of benign from malignant breast masses by time–intensity evaluation of contrast enhanced MRI. Mag Reson Imaging 11: 617–620. Folkman J. 1990. What is the evidence that tumours are angiogenesis dependent? J Natl Cancer Inst 82: 4–6. Frahm J, Haase A, et al. 1986. Rapid three-dimensional MR imaging using the FLASH technique. J Comput Assist Tomogr 10(2): 363–368. Frazzini VI, Cha S, et al. 1999. Dynamic contrast enhanced T*2 weighted echo-planar perfusion MR imaging of primary CNS lymphoma and glioblastoma multiforme. In Proceedings of the 37th Annual Meeting of the American Society of Neuroradiology. American Society of Neuroradiology, San Diego, California, p. 185. Fujii K, Fujita N, et al. 1992. Neuromas and meningiomas: evaluation of early enhancement with dynamic MR imaging. Am J Neuroradiol 13(4): 1215–1220. Fuss M, Wenz F, et al. 2000. Radiation-induced regional cerebral blood volume (rCBV) changes in normal brain and lowgrade astrocytomas: quantification and time and dose-dependent occurrence. Int J Radiat Oncol Biol Phys 48(1): 53–58. Gobbel GT, Seilhan TM, et al. 1992. Cerebrovascular response after interstitial irradiation. Radiat Res 130(2): 236–240. Gossmann A, Helbich TH, et al. 2002. Dynamic contrastenhanced magnetic resonance imaging as a surrogate marker of tumor response to anti-angiogenic therapy in a xenograft model of glioblastoma multiforme. J Magn Reson Imaging 15(3): 233–240. Gowland P, Mansfield P, et al. 1992. Dynamic studies of gadolinium uptake in brain tumors using inversion-recovery echo-planar imaging. Magn Reson Med 26(2): 241–258. Hawighorst H, Knopp MV, et al. 1998. Pharmacokinetic MRI for assessment of malignant glioma response to stereotactic radiotherapy: initial results. J Magn Reson Imaging 8(4): 783–788. Jackson A, Haroon H, et al. 2002a. Breath-hold perfusion and permeability mapping of hepatic malignancies using

magnetic resonance imaging and a first-pass leakage profile model. NMR Biomed 15(2): 164–173. Jackson A, Kassner A, et al. 2002b. Abnormalities in the recirculation phase of contrast agent bolus passage in cerebral gliomas: comparison with relative blood volume and tumor grade. Am J Neuroradiol 23(1): 7–14. Jayson GC, Zweit J, et al. 2002. Molecular imaging and biological evaluation of HuMV833 anti-VEGF antibody: implications for trial design of antiangiogenic antibodies. J Natl Cancer Inst 94(19): 1484–1493. Kassner A, Annesley D, et al. 1999. Abnormalities of the contrast re-circulation phase in cerebral tumours demonstrated using dynamic susceptibility contrast-enhanced MR imaging: A possible marker of vascular tortuosity. Proceedings of the 7th scientific meeting of the International Society of Magnetic Resonance in Medicine, Philadelphia, p. 151. Kassner A, Annesley D, et al. 2000. Abnormalities of the contrast re-circulation phase in cerebral tumours demonstrated using dynamic susceptibility contrast-enhanced MR imaging: a possible marker of vascular tortuosity. J Magn Reson Imaging 11: 103–113. Knopp EA, Cha S, et al. 1999a. Glial neoplasms: dynamic contrast-enhanced T*-weighted MR imaging. Radiology 211(3): 2 791–798. Knopp MV, Weiss E, et al. 1999b. Pathophysiologic basis of contrast enhancement in breast tumors. J Magn Reson Imaging 10(3): 260–266. Kuhl CK, Bieling HB, et al. 1997a. Breast neoplasms: T*2 susceptibility-contrast, first-pass perfusion MR imaging [see comments]. Radiology 202(1): 87–95. Kuhl CK, Bieling HB, et al. 1997b. Healthy premenopausal breast parenchyma in dynamic contrast-enhanced MR imaging of the breast: normal contrast medium enhancement and cyclical-phase dependency. Radiology 203(1): 137–144. Lam WW, Chan KW, et al. 2001. Pre-operative grading of intracranial glioma. Acta Radiol 42(6): 548–554. Larsson HB, Stubgaard M, et al. 1990. Quantitation of blood–brain barrier defect by magnetic resonance imaging and gadolinium-DTPA in patients with multiple sclerosis and brain tumors. Magn Reson Med 16(1): 117–131. Levin JM, Kaufman MJ, et al. 1995. Sequential dynamic susceptibility contrast MR experiments in human brain: residual contrast agent effect, steady state, and hemodynamic perturbation. Magn Reson Med 34: 655–663. Li KL, Tessier JJL, et al. 1999. Accurate measurement of arterial input function (AIF) using a 3D T1 gradient echo imaging method. In Proceedings of the 7th Scientific Meeting of the International Society of Magnetic Resonance in Medicine, Philadelphia, p. 573.

Perfusion MR imaging in adult neoplasia

Li K, Zhu X, et al. 2000. Improved 3D quantitative mapping of blood volume and endothelial permeability in brain tumours. J Magn Reson Imaging 12: 347–357. Li KL, Zhu XP, Checkley DR, Tessier JJ, Hillier VF, Waterton JC, Jackson A. 2003. Simultaneous mapping of blood volume and endothelial permeability surface area product in gliomas using iterative analysis of first-pass dynamic contrast enhanced MRI data. Br J Radiol 76(901): 39–50. Long DM. 1973. Vascular ultrastructure in human meningiomas and schwannomas. J Neurosurg 38(4): 409–419. Lund EL, Spang-Thomsen M, et al. 1998. Tumor angiogenesis – a new therapeutic target in gliomas. Acta Neurol Scand 97(1): 52–62. Maeda M, Itoh S, et al. 1993. Tumor vascularity in the brain: evaluation with dynamic susceptibility-contrast MR imaging. Radiology 189(1): 233–238. Maeda M, Itoh S, et al. 1994. Vascularity of meningiomas and neuromas: assessment with dynamic susceptibilitycontrast MRF imaging. Am J Roentgenol 163(1): 181–186. Maeda M, Maley JE, et al. 1997. Application of contrast agents in the evaluation of stroke: conventional MR and echoplanar MR imaging. J Magn Reson Med 7: 723–728. Miyati T, Banno T, et al. 1997. Dual dynamic contrastenhanced MR imaging. J Magn Reson Imaging 7(1):230–235. Nagele T, Petersen D, et al. 1993. Dynamic contrast enhancement of intracranial tumors with snapshot-FLASH MR imaging. Am J Neuroradiol 14(1): 89–98. Ohno K, Pettigrew KD, et al. 1978. Lower limits of cerebrovascular permeability to nonelectrolytes in the conscious rat. Am J Physiol 235: H299–H307. Padhani AR, Husband JE. 2001. Dynamic contrast-enhanced MRI studies in oncology with an emphasis on quantification, validation and human studies. Clin Radiol 56(8): 607–620. Parker GJ, Suckling J, et al. 1997. Probing tumor microvascularity by measurement, analysis and display of contrast agent uptake kinetics. J Magn Reson Imaging 7(3): 564–574. Paulson OB, Hertz MM. 1983. Tracer kinetics and physiologic modelling. Theory to Practice. In Lecture Notes in Biomathematics. (Ed. Rescigno A), Springer, Berlin Heidelberg, New York, Tokyo, pp. 428–444. Reinhold HS, Endrich B. 1986. Tumour microcirculation as a target for hyperthermia. Int J Hyperthermia 2(2): 111–137. Roberts HC, Roberts TP, et al. 2000. Quantitative measurement of microvascular permeability in human brain tumors achieved using dynamic contrast-enhanced MR imaging: correlation with histologic grade. Am J Neuroradiol 21(5): 891–899. Roberts HC, Roberts TP, et al. 2001. Correlation of microvascular permeability derived from dynamic contrast-enhanced MR

imaging with histologic grade and tumor labeling index: a study in human brain tumors. Acad Radiol 8(5): 384–391. Rosen BR, Aronen HJ, et al. 1993. Advances in clinical neuroimaging: functional MR imaging techniques. Radiographics 13(4): 889–896. Rosen BR, Belliveau JW, et al. 1991. Susceptibility contrast imaging of cerebral blood volume: human experience. Magn Reson Med 22(2): 293–299. Shweiki D, Neeman M, et al. 1995. Induction of vascular endothelial growth factor expression by hypoxia and by glucose deficiency in multicell spheroids: implications for tumor angiogenesis. Proc Natl Acad Sci USA 92(3): 768–772. Song CW, Lokshina A, et al. 1984. Implication of blood flow in hyperthermic treatment of tumors. IEEE Trans Biomed Eng 31(1): 9–16. St. Lawrence K, Lee T. 1998. An adiabatic approximation to the tissue homogeneity model for water exchange in the brain: I. Theoretical derivation. J Cereb Blood Flow Metab 18: 1365–1377. Stack J, Redmond O, et al. 1990. Breast disease: tissue characterization with Gd-DTPA enhancement profiles. Radiology 174: 491–494. Strugar JG, Criscuolo GR, et al. 1995. Vascular endothelial growth/permeability factor expression in human glioma specimens: correlation with vasogenic brain edema and tumor-associated cysts. J Neurosurg 83(4): 682–689. Strugar J, Rothbart D, et al. 1994. Vascular permeability factor in brain metastases: correlation with vasogenic brain edema and tumor angiogenesis. J Neurosurg 81(4): 560–566. Sugahara T, Korogi Y, et al. 1998. Correlation of MR imagingdetermined cerebral blood volume maps with histologic and angiographic determination of vascularity of gliomas. Am J Roentgenol 171(6): 1479–1486. Sugahara T, Korogi Y, et al. 1999a. Value of dynamic susceptibility contrast magnetic resonance imaging in the evaluation of intracranial tumors. Top Magn Reson Imaging 10(2): 114–124. Sugahara T, Korogi Y, et al. 1999b. Perfusion-sensitive MRI of cerebral lymphomas: a preliminary report. J Comput Assist Tomogr 23(2): 232–237. Tofts PS. 1997. Modeling tracer kinetics in dynamic Gd-DTPA MR imaging. J Magn Reson Imaging 7: 91–101. Tofts PS, Kermode AG. 1991. Measurement of the blood brain barrier permeability and leakage space using dynamic MR imaging: Fundemental concepts. Mag Res Med 17: 357–367. Tofts P, Berkowitz B, et al. 1995. Quantitative analysis of dynamic Gd-DTPA enhancement in breast tumours using a permeability model. Mag Res Med 33: 564–568.

347

348

Alan Jackson

Tofts PS, Brix G, et al. 1999. Estimating kinetic parameters from dynamic contrast-enhanced T(1)-weighted MRI of a diffusable tracer: standardized quantities and symbols. J Magn Reson Imaging 10(3): 223–232. Uematsu H, Maeda M, et al. 2000. Vascular permeability: quantitative measurement with double-echo dynamic MR imaging – theory and clinical application. Radiology 214(3): 912–917. Wenz F, Rempp K, et al. 1996. Effect of radiation on blood volume in low-grade astrocytomas and normal brain tissue:

quantification with dynamic susceptibility contrast MR imaging. Am J Roentgenol 166(1): 187–193. Zhu X, Li KL, et al. 1999. 3D T1 mapping by means of fast field echo technique. Proceedings of the 7th scientific meeting of the International Society of Magnetic Resonance in Medicine, Philadelphia, p. 2143. Zhu XP, Li KL, et al. 2000. Quantification of endothelial permeability, leakage space, and blood volume in brain tumors using combined T1 and T*2 contrast-enhanced dynamic MR imaging. J Magn Reson Imaging 11(6): 575–585.

Perfusion MR imaging in adult neoplasia

Case Study 21.1 Anaplastic oligodendroglioma: ASL MR perfusion Ronald L. Wolf M.D., Ph.D., Jiongjiong Wang, Ph.D., University of Pennsylvania Medical Center, Philadelphia History

FLAIR

41-year-old female, new onset seizure.

Technique FLAIR, T1-weighted SE post-contrast at 1.5 T, continuous arterial spin labeled perfusion (CASL) MRI at 3.0 T.

Imaging findings FLAIR images demonstrate a hyperintense mass in the right frontal and temporal lobes, which shows no enhancement. CASL perfusion MRI shows clearly increased CBF in portions of the tumor (more than twice that in contralateral mirror region).

T1Post-Gd

Discussion Conventional imaging is suggestive of glioma. Grade is uncertain, but grade II or III glioma is most likely. CBF increase mass favors high-grade glioma. Absolute, as well as relative increases in CBF as measured by non-invasive arterial spin labeled (ASL) perfusion MRI, may indicate higher-grade neoplasms (Warmuth et al., 2003). Pathology showed WHO grade III glioma (anaplastic oligodendroglioma).

CBF

rCBV

Key points Elevated CBF, as well as rCBV, suggests higher grade for gliomas. ASL perfusion MR is non-invasive and quantitative.

References Wang J, Wolf RL, Zhang Y, Roc AC, Alsop DC, Detre JA. 2004. Amplitude modulated continuous arterial spin labeling perfusion MR with single coil at 3.0 Tesla. Radiology [in press]. Warmuth C, Günther M, Zimmer C. 2003. Quantification of blood flow in brain tumors: comparison of arterial spin labeling and dynamic susceptibility-weighted contrast-enhanced MR imaging. Radiology 228: 523–532.

349

350

Alan Jackson

Case Study 21.2 Radiation necrosis vs. recurrence Ronald L. Wolf, M.D., Ph.D., University of Pennsylvania Medical Center, Philadelphia History GBM status post-RT, recurrence vs. radiation necrosis?

FLAIR

T1Post-Gd

2 mo f/u FLAIR

T1Post-Gd

Technique FLAIR, T 1 -weighted SE post-contrast, DSC perfusion-weighted MRI.

Imaging findings FLAIR reveals marked increase in surrounding edema and/or infiltrating tumor over 2-month period. Nodular enhancement around prior resection site also shows progression, associated with increasing mass effect. The rCBV was decreased (ratio 1 compared to left side) in enhancing regions and regions with abnormal intensity on FLAIR.

Discussion Conventional imaging showed increasing nodular enhancement, edema/infiltrating tumor, and mass effect. The decreased rCBV argued in favor of radiation necrosis. Surgery was performed due to mass effect, and pathology revealed primarily radiation necrosis (5% tumor).

2 mo f/u rCBV

Key points Radiation necrosis can appear identical to recurrent neoplasm. Decreased rCBV argues against recurrent high-grade tumor.

References Cha S, Knopp EA, Johnson G, Wetzel SG, Litt AW, Zagzag D. 2002. Intracranial mass lesions: dynamic contrastenhanced susceptibility-weighted echo-planar perfusion MR imaging. Radiology 223: 11–29. Wong J, Provenzale JM, Petrella JR. 2000. Perfusion MR imaging of brain neoplasms. Am J Roentgenol 174: 1147–1157.

Section 4 Infection, inflammation and demyelination

22

Physiological imaging in infection, inflammation and demyelination: overview Robert D. Zimmerman Professor of Radiology, Weill Medical College of Cornell University; Director of Diagnostic Imaging, New York Preslyterian Hospital, Cornell, New York, USA

Introduction This chapter will offer an overview of physiological imaging of infectious inflammatory and demyelinating diseases. It will describe routine imaging findings and the current and potential roles of a variety of techniques. Subsequent chapters will provide more detailed information of specific techniques as they apply to a variety of disease processes. MR imaging (MRI) has had a profound effect on detection, management and outcome of central nervous system (CNS) infectious, inflammatory and demyelinating diseases. While this statement seems to apply uniformly to virtually all CNS disorders, there is a difference. Despite recent therapeutic advances, outcomes for many disorders (e.g. stroke, neoplasm and neurodegenerative diseases) have improved only modestly because effective treatments are not yet available. In these disorders, the MRI tools discussed in this book will play an important role in developing and monitoring new therapies. In infectious and inflammatory diseases, effective treatments are available for many disorders but success is dependent upon early institution of the correct therapeutic regimen. MR has led to a dramatic improvement in outcome by allowing for early and accurate detection of many of these disorders. The presence of characteristic MRI features often allow for accurate diagnosis, at least in the hands of experienced neuroradiologists. In these disorders, physiological techniques such as diffusion weighted imaging (DWI) (Tsuchiya et al., 1999), diffusion tensor imaging (DTI) (Ulug et al., 1999), perfusion

weighted imaging (PWI) and MR spectroscopy (MRS) (Barker et al., 1995; Cecil and Kenkinski, 1998) are useful additions to “routine” MR since they may allow for earlier detection (e.g. DWI in herpes simplex type I (HSV1) encephalitis – HSV1 (Sener, 2001)) or more accurate characterization of lesions (e.g. DWI (Ebisu et al., 1996) and MRS in brain abscesses (Remy et al., 1995; Kim et al., 1998)). There is a tendency to assume that these new tools will be used most often and prove most beneficial to experts but this is not the case. Experts typically recognize critical but subtle differences between different lesions while less experienced readers may not always appreciate these differences. Findings on physiological techniques can therefore improve diagnostic accuracy by providing simple unambiguous information suggestive of the correct diagnosis. Physiological imaging may also prove helpful in selecting and monitoring treatment. MRI in general and physiological imaging techniques, in particular, have become increasingly important in the development of novel therapies by providing quantitative outcome measures (e.g. multiple sclerosis (MS)). There are a few disorders that produce very little abnormality on routine MRI studies, at least in their early stages. In these disorders, the physiological tools described in subsequent chapters may play the primary role in detection. Diffuse disease processes that affect most of the brain can be particularly difficult to detect in their early stages. In these processes, physiological imaging techniques can be used to assess the “whole brain” value of a specific parameter (e.g. a metabolite on MRS apparent diffusion coefficient 353

354

Robert D. Zimmerman

(ADC), fractional anisotropy (FA) (Filippi et al., 2001a)). Whole brain measures can serve as markers of disease load and can be used to assess treatment response. DTI may prove to be particularly valuable in the assessment of diffuse disorders that extend along or have a predilection for white matter (WM) tracks – e.g. MS, human immunodeficiency virus (HIV) encephalitis (Filippi et al., 2001a; Ragin et al., 2004), progressive multifocal leukoencephalopathy (PML).

Pyogenic infections Bacteria (gram negative and positive cocci and bacilli) cause these acute fulminate infections. The host response is mediated by polymorphonuclear neutrophils (PMN), leading to tissue necrosis and liquefaction. Infections may involve one or more intracranial compartments producing brain abscesses, meningitis and/or extra-cerebral empyemas and they may be diffuse (meningitis) or focal (abscess). Hematogenous spread is the most common route of entry followed by direct spread from the paranasal sinuses or through a pathological opening in the dura. In about one third of cases an obvious extra-cerebral source of infection cannot be found. Patients are usually immunocompetent. Systemic signs of infection (e.g. fever and leukocytosis) are present in the majority of patients and focal or diffuse CNS dysfunction usually evolves rapidly over a few days to a week. Unless quickly treated these diseases have a high morbidity and mortality. Even after the development of modern antibiotics and neurosurgical techniques, mortality was high (50%). The introduction of computed tomography (CT) improved outcome. MR provides even earlier and potentially more accurate diagnosis. Mortality is now less than 5% (Alvard and Shaw, 1977). Imaging features allowing for detection of pyogenic processes and differentiation from other processes are a reflection of host response. With routine morphological imaging techniques it is not possible to differentiate the most defining feature that characterizes pyogenic processes – pus – from other forms of necrotic debris. The introduction of physiological imaging techniques has changed this. It turns out that pus is unique after all. On DWI, purulent

material is hyperintense due to restricted water motion (decreased ADC) (Ebisu et al., 1996; Desprechins et al., 1999) while necrotic tumor debris is typically hypointense on DWI with increased ADCs. In addition, MRS reveals the presence of amino acids from extra-cellular proteolysis and bacterial metabolism (fermentation products) including succinate, acetate, leucine, valine and alanine that are not seen in necrotic neoplasms (Remy et al., 1995; Kim et al., 1998). Brain abscess Abscesses are focal infections that evolve over 10 days to 2 weeks. They begin as regions of edema with active bacterial growth and small areas of noncoalescent necrosis. Patients do not usually present during this early cerebritis stage and imaging features are non-specific (vasogenic edema with illdefined enhancement). After about 5–7 days, there is coalescence of the central necrotic material surrounded by a layer of reactive macrophages that enter the brain via an open blood brain barrier. Peripheral to this, there is reactive edema. Over the next few days blood borne fibroblasts migrate into the reactive zone and lay down collagen. The presence of collagen in the developing capsule marks the change from late cerebritis to early abscess (10–14 days). If untreated, the abscess grows rapidly over the next week. The capsule will thicken and loculations may develop. During this phase the abscess tends to grow away from the surface of the brain (where the blood supply and therefore host response is greatest) and if unchecked may rupture into the ventricle, causing ventriculitis (Enzmann et al., 1983). The features of the abscess wall are usually sufficient for differentiation from neoplastic masses (Figure 22.1). The abscess capsule is typically hypointense on T2 weighted (T2W) images due to presence of paramagnetic atomic oxygen generated by the macrophages and it enhances intensely. The thickness of the capsule is variable and it may be loculated but there is no nodularity or internal irregularity, findings that distinguishes abscesses from metastases, the lesions they most closely resemble. Both abscesses and metastases are focal foreign lesions with well–defined external borders. Gliomas

Physiological imaging in infection, inflammation and demyelination: overview

(a)

(b)

(d)

(e)

(c)

Fig. 22.1 Brain abscess. Focal lesion noted in the WM of the right frontal lobe. There is peripheral edema and central necrosis that are hypointense on T1W image (a) and hyperintense on T2W image (b). The abscess capsule is minimally hyperintense on T1W image (a) and hypointense on T2W image (b). It undergoes uniform enhancement (c). Note the discrete smooth lateral margins of the capsule and the absence of wall nodularity on the interior or exterior edge. The central necrotic material reveals subtle internal partial rings on both T1W and T2W images, an inconsistent but characteristic feature of abscesses. On DW1 (d) the central necrotic material is uniformly and markedly hyperintense while on ADC (e) the central material markedly hypointense due to restricted diffusion. Note that the surrounding vasogenic edema is isointense on DWI due to competing effects of T2 and diffusion.

are infiltrative intrinsic lesions with ill-defined margins and a more irregular central cavity, features that allow for easy differentiation from abscess (Haimes et al., 1989; Zimmerman and Weingarten, 1991). Use of DWI has made the diagnosis of abscess easier. The central necrotic material is extremely hyperintense on DWI and hypointense on ADC

maps due to marked water motion restriction. Most necrotic tumors are hypointense on DWI and hyperintense on ADC maps due to increased water diffusion (Ebisu et al., 1996; Desprechins et al., 1999). Some authors have questioned the utility of DWI pointing out that some necrotic tumors are hyperintense on DWI and some chronic abscesses are not

355

356

Robert D. Zimmerman

Fig. 22.2 Brain abscess. Long repetition time (TR) point resolved spectroscopy (PRESS) single voxel MRS reveals absence of normal brain metabolite peaks. Several peaks are identified that correspond to fermentation products of anaerobic bacterial metabolism. Courtesy of Lawrence Tannenbaum MD.

(Hartmann et al., 2001; Tung et al., 2001). While this may be true these exceptions do not really cause problems. When evaluated in conjunction with the routine MR images the diagnosis of abscess and differentiation from metastases is easily made (Figure 22.1). MRS has also proven to be useful in evaluation of abscesses (Remy et al., 1995; Kim et al., 1998; Burtscher and Holtas, 1999). Metabolite peaks specific to infection including amino acids that are the result of extra-cellular proteolysis and fermentation products of the bacteria themselves produce spectra not seen in neoplasms (Figure 22.2). Proponents of MRS occasionally overstate its value by stating that the routine imaging findings in abscesses are non-specific. We do not routinely perform MRS because the diagnosis can usually be made with confidence on MRI with DWI. MRS is useful in rare cases where clinical or imaging findings lead to diagnostic uncertainty. Aerobic and anaerobic bacteria produce somewhat different spectral patterns and therefore MRS may be prove to be most useful in choosing the most appropriate antibiotic (Garg et al., 2004) (cf. Chapter 24). Meningitis Bacterial meningitis is an acute disease with well-established clinical features. The diagnosis is

confirmed by spinal tap (lumbar puncture) and imaging does not play a primary role in the detection or treatment of this disorder. MR can detect abnormalities in uncomplicated bacterial and viral meningitis but they are of little prognostic significance (Chang et al., 1990; Harris and Edwards, 1991; Whiteman et al., 1996). T1 weighted (T1W) and T2W images are normal but diffuse subarachnoid hyperintensity most marked over the convexities may be seen on fluid attenuated inversion recovery (FLAIR) secondary to T1 shortening and less commonly on DWI due to restricted water diffusion. Leptomeningeal enhancement may be seen (Whiteman et al., 1996). MRI is used to detect the complications of meningitis including hydrocephalus (rare) and vascular occlusion and include dural venous sinus thrombosis and arterial spasm. These processes are best assessed with MRI and MR angiography (MRA)/venography (Harris and Edwards, 1991). Vascular compromise can lead to infarction and edema. DWI is useful in the detection of hyperacute arterial infarction (restricted water motion) and its differentiation from vasogenic edema produced by venous occlusion (increased water motion). Perfusion imaging can be used to identify tissue at risk for infarction due to decreased blood flow. These techniques could therefore prove useful in identifying patients who could benefit from intervention and assessing the effects of treatment. Post-meningitic subdural effusions often occur in children less than 18 months of age with acute hemophilus influenza or streptococcus pneumonia meningitis. They are the consequence of the relative immaturity of the meningeal linings of the brain. Large collections develop often while the child improves clinically. The subdural effusions are sterile; however, the protein content and therefore the MR intensity characteristics may be intermediate between cerebrospinal fluid (CSF) and frank empyema (cf. below) and meningeal enhancement may be extensive. Subdural effusions have been reported to be hypointense on DWI, aiding in differentiation from empyemas. Clinical factors such as severity and progression of illness are more accurate indicators of the need for surgical evacuation than imaging findings (Centeno et al., 1983; Sze and Zimmerman, 1988; Harris and Edwards, 1991).

Physiological imaging in infection, inflammation and demyelination: overview

Epidural and subdural empyema The role of physiological imaging in empyemas is similar to that of brain abscesses. These focal extracerebral purulent collections are the result of direct spread of infection from the paranasal sinuses or less commonly the mastoid air cells. The dura is the periosteum of the inner table of the skull and therefore epidural empyemas (EDE) are intracranial subperiosteal extensions of osteomyelitis. When pus accumulates in the epidural space, the dura prevents direct spread to deeper spaces. The bacteria can, however, gain access to the intradural space by retrograde thrombophlebitis. The cortical veins pass through the subdural space to enter the dural sinuses which also communicate with the intradiploic veins. Once the bacteria pass through the dura they quickly produce a thin purulent discharge that spreads over the brain throughout the subdural space (subdural empyema (SDE)). This process can occur with or without concomitant osteomyelitis and therefore SDE may be seen in association with EDE or as an isolated infection (Zimmerman et al., 1984; Weingarten et al., 1989; Sze and Zimmerman, 1988). Subsequent spread from the subdural space to the adjacent subarachnoid space and brain can lead to meningitis or brain abscess unless there is rapid surgical intervention. Retrograde venous thrombosis leads to cortical venous stasis with marked cortical swelling and eventual infarction. Empyemas most commonly occur in adolescents or young adults. An episode of acute sinusitis or mastoiditis (often incompletely treated with antibiotics) precedes the onset of neurological abnormalities. Isolated EDE are slowly enlarging lesions that eventually produce signs of focal intracranial mass. Systemic signs of infection are often absent because the dura effectively isolates the infection. Successful treatment usually requires surgical intervention and a course of antibiotics but because the empyemas grow slowly this typically is not an emergency. On MR, a lentiform fluid collection is seen with the same intensity characteristics as brain abscesses and as in abscesses, the pus is hyperintense on DWI and MRS reveals the presence of unique amino acids. The displaced inflamed dura is always visible as a thick T2 hypointense-enhancing rim interposed between the empyema and the brain.

By contrast, SDEs produce an acute, rapidly progressive syndrome characterized by fever and leukocytosis, rapid development of neurological abnormalities (e.g. seizure and hemiparesis) and depression of mental status. SDEs require emergent surgical drainage because antibiotics do not penetrate the subdural space. An extensive craniotomy is required since more limited procedures (e.g. burr holes) will not allow for complete evacuation of these thin diffuse collections. If incorrectly treated, the disease will inevitably progress. Venous stasis leads to cerebral infarction and spread of the bacteria to the brain produces cerebral abscesses. Loculations develop within the SDE that will eventually require separate surgical drainage. In short, if emergent aggressive surgical intervention is not undertaken, multiple surgical procedures will be required in the weeks that follow and morbidity and mortality will increase. The requirement for aggressive surgical treatment makes early diagnosis imperative but these lesions present diagnostic challenges. There is always evidence of sinusitis or mastoiditis with or without osteomyelitis but the SDE itself is difficult to detect. The key finding is the presence of a small innocuous appearing subdural collection with disproportionate mass on the adjacent brain (Zimmerman et al., 1984). The collection is typically narrow but extends over most of the convexity and/or within the interhemispheric fissure as a thin fluid collection. The empyema is most conspicuous on FLAIR where it is hyperintense to adjacent brain, bone and spinal fluid (Weingarten et al., 1989). It is also hyperintense on DWI, a finding that can help alert the radiologist and the clinician to the true nature of the lesion. There is thin linear enhancement of the deep and superficial layers of the dura. Leptomeningeal enhancement implies the presence of secondary meningitis and/or vascular stasis. The cortical gray matter (GM) and subcortical WM have normal intensity unless cerebritis or venous infarctions have occurred. If surgical intervention is delayed or inadequate, subsequent scans will reveal rapid enlargement and loculations of the empyema. Brain abscesses and/or areas of venous infarction may develop (Figure 22.3).

357

358

Robert D. Zimmerman

(a)

(b)

(d)

(e)

(c)

Fig. 22.3 SDE. A 16-year old male with history of frontal sinusitis presents with obtundation and seizures. CT scan at another institution was read as normal. Axial T1W (a) T2W (b) and FLAIR (c) images reveal a small left interhemispheric collection. On T1W and T2W images it is isointense to CSF but on FLAIR it is hyperintense. Axial and coronal contrast enhanced T1W images (d and e) scans reveal enhancement of the dura and arachnoid. Note that the empyema is bilateral anteriorly and that it extends inferiorly and posteriorly. Craniotomy is required to insure adequate drainage of the entire collection.

Granulomatous disease Granulomatous infections are typically more indolent than pyogenic infections. They develop more slowly and are more difficult to treat. These chronic infections persist for months to years with or without treatment often with a waxing and waning

course. Granulomatous diseases are caused by a large and diverse group of pathogens including bacteria fungi and parasites (Bazan et al., 1991; Castro and Hesselink, 1991). Prolonged exposure is typically required to produce infection and the diseases therefore tend to occur in endemic regions where poor living conditions mean that much of the

Physiological imaging in infection, inflammation and demyelination: overview

population is infected. Debilitated and immunocompromised patients are susceptible and therefore these diseases have become more common because of the Acquired Immuno deficiency syndrome (AIDS) epidemic (Alvard and Shaw, 1977; Theur et al., 1990). Sarcoidosis is an idiopathic granulomatous disease that most commonly affects young, otherwise healthy adult patients (Ulmer and Elster, 1991). Macrophages, lymphocytes, plasma cells and giant cells mediate the host response. The characteristic lesion is a granuloma, a cellular mass without liquefied necrotic debris. Caseous (“cheesy”) necrosis is typical of tuberculosis (TB). As these lesions mature they become less cellular, extensive collagen is laid down at the periphery and calcification may occur. CNS granulomatous infections usually result from hematogenous spread. An extra-cranial source of infection is found in most cases (usually the lung) although isolated CNS involvement may occur in younger patients and it is seen in up to 10% of patients with sarcoidosis (Ulmer and Elster, 1991). CNS involvement may occur at the time of initial or primary infection (children) or at the time of reactivation (adults). In debilitated (e.g. diabetic) and immunocompromised patients, direct spread of fungal disease may occur from the paranasal sinuses (aspergillosis) or temporal bones (mucormycosis) with an often fatal outcome (Bazan et al., 1991). The initial lesion most often develops in the leptomeninges or at the cortico-medullary junction with subsequent rupture into the subarachnoid space. Spread to the basal meninges and development of parenchymal granuloma may follow. It is, therefore, typical for granulomatous disease to involve both the meninges and brain parenchyma, although one component will predominate. Meningitis Two major consequences of granulomatous disease may dominate the clinical and imaging findings. First, basal meningitis often produces obstructive hydrocephalus. Many patients will require shunting since regression of hydrocephalus may take many months despite adequate anti-microbial treatment. Second, there is often compromise of the vascular system with secondary infarction and/or hemorrhage. The infarcts

are most commonly seen at the base of the brain in the deep GM or brainstem. The combination of hydrocephalus and deep infarction in a young adult should therefore always raise the suspicion of granulomatous meningitis (Figure 22.4). The imaging features of granulomatous disease reflect the complex and variable histopathology of this group of disorders. Pyogenic infections produce a few relatively stereotypical patterns while granulomatous diseases have more protean and less characteristic findings. The findings on DWI are variable and less definitive. MRS has been reported to show the presence of unique lipid peaks in tuberculous granulomas (Gupta et al., 1993). Physiological imaging may prove most useful in identifying and monitoring outcome (with or without treatment) of hydrocephalous and infarction. Meningeal inflammation is the most common and easily recognized manifestation of CNS granulomatous disease (Sze and Zimmerman, 1988; Bazan et al., 1991; Castro and Hesselink, 1991; Whiteman et al., 1996). The leptomeninges at the base of the brain are the most frequently involved. On gross inspection, gelatinous material is seen in the basal cisterns surrounding the arteries at the base of the brain. The process is diffuse but often asymmetric. On imaging, there may be subtle increased signal on T1W images. Subarachnoid hyperintensity is present on FLAIR. Intense enhancement within the cisterns at the base of the brain (supra- and infratentorial) is the hallmark MR finding of granulomatous meningitis. In pyogenic and viral meningitis, the enhancement is thinner, more symmetric and most marked over the cerebral convexities. Carcinomatous meningitis is also thinner and it has a predilection for the retrocerebellar cisterns. Sarcoidosis has a predilection for the suprasellar cistern often producing thickening of the pituitary stalk. It tends to coat the surface of the brain stem and upper spinal cord (“sugar frosting”) with linear extension into the brain along the course of the perivascular spaces. Enhancement of cranial nerves is characteristic of sarcoidosis (Ulmer and Elster, 1991) but it can also be seen in lymphoma. Sarcoidosis (and to a lesser extent TB) may also involve the pachymeninges producing focal extra-axial masses that extend along dural surfaces. These intensely enhancing lesions are hypointense on T2W images

359

360

Robert D. Zimmerman

(a)

(b)

(c)

(d)

(e)

(f)

(g)

(h)

(i)

Fig. 22.4 Tuberculous meningitis: axial FLAIR (a–c) images reveal hydrocephalus. Note subtle hyperintensity on the anterior surface of the pons (a) and hyperintensity within the right cerebral peduncle (b) and basal ganglia bilaterally (c). DWI (d–f) images reveal that the brain stem and ganglionic lesions are hyperintense. These lesions are acute infarcts secondary to meningitis induced vasculitis. Contrast enhanced T1W (g–i) images reveal enhancement along the surfaces of the brain stem, chiasm frontal and temporal lobes indicative of meningitis. There is no abnormal signal in the basal meninges on FLAIR or DWI, a finding that helps to distinguish this process from pyogenic meningitis.

Physiological imaging in infection, inflammation and demyelination: overview

(a)

(b)

Fig. 22.5 Pachymeningeal sarcoidosis: contrast enhanced T1W image (a) reveals thick en plaque enhancement that follows the right frontal convexity and bilateral parafalcine dura. T2W (b) image reveals that the enhancing material is hypointense. Note extensive edema.

(Figure 22.5). They may be mistaken for enplaque meningiomas (Sze and Zimmerman, 1988). Irregular borders, associated leptomeningeal or parenchymal lesions, the age of the patient (usually young adults) and the presence of extra-cranial granulomatous disease usually suggest the correct diagnosis. Physiological imaging studies (DWI, MRS and PWI) could conceivably play a role in differentiation between inflammation and benign neoplasm. Cryptococcal meningitis is seen most commonly in AIDS patients. There is usually very little reaction to the presence of the fungus and therefore leptomeningeal enhancement and hydrocephalus are uncommon. If the infection is severe, the organisms fill and dilate the perivascular spaces in the anterior perforated substances where they produce bilateral cyst-like masses with little or no enhancement (gelatinous pseudo-cysts) (Mathews et al., 1992). The lesions are iso- to mildly hyperintense to spinal fluid on T1W images and FLAIR and isointense to CSF on DWI (Figure 22.6). Further detailed description of physiological imaging in AIDS is found in Chapter 28.

multiple solid or discrete ring-enhancing lesions with surrounding edema at the cortico-medullary junction. These lesions have variable intensity on T1W, T2W and DWI images (Figure 22.7). Due to this lack of specificity, they may mimic the appearance of metastatic lesions. Granulomas are usually more uniform in size and appearance than metastases but over time, multiple small lesions may coalesce into large conglomerate masses that present as thick walled or multi-nodular lesions with small areas of necrosis which can mimic the appearance of malignant tumors (Figure 22.8) (Gupta et al., 1988; Castro and Hesselink, 1991). Aspergillosis and mucormycosis typically produce hemorrhagic masses because of their propensity to grow within and erode the walls of arteries (Figure 22.10). MRS studies have demonstrated lipid peaks that are said to be the result of large amounts of fat present in the tuberculous bacillus (Gupta et al., 1993). With healing the granuloma may resolve completely or produce small areas of calcification. Vascular disease

Parenchymal disease Parenchymal lesions vary in configuration and size. Granulomas are most commonly small single or

Granulomatous diseases may also affect intracranial vessels. Vascular involvement may be the result of direct spread from the leptomeninges along the

361

362

Robert D. Zimmerman

(a)

(b)

(d)

(e)

(c)

Fig. 22.6 Cryptococcal meningitis – gelatinous pseudo-cysts: axial T2W images (a–c) reveal the presence of spinal fluid intensity foci in the medial cerebellum (a), the mesencephalon (b) and the inferior portions of the basal ganglia and anterior perforated substance (c). Foci are predominantly isointense to CSF on T1W image (d) and FLAIR (e) as well but a few foci are mildly hyperintense on FLAIR in particular in the left basal ganglia. These foci represent crowding and expansion of normal perivascular spaces with fungi. The low virulence of the organism accounts for absence of edema mass effect enhancement and hydrocephalus.

perivascular spaces (e.g. sarcoidosis) or spread to the vessel wall invasion from parenchymal granuloma producing an endarteritis obliterans. Direct invasion and growth within the lumen of the vessel occurs with angiophylic fungal infections (aspergillosis and mucormycosis) (Alvard and Shaw, 1977; Sze and

Zimmerman, 1988; Whiteman et al., 1996). Infarcts most often occur in the deep GM and WM secondary to involvement of perforating arteries. When larger vessels are involved (e.g. internal carotid or middle cerebral artery (MCA)) the patient may present clinically and radiographically with findings of acute infarction.

Physiological imaging in infection, inflammation and demyelination: overview

(a)

(b)

(c)

Fig. 22.7 Tuberculoma: contrast enhanced T1W image (a) reveals a discrete 2 cm mass in the brain stem. There is minimal central necrosis and a peripherally located cystic or necrotic component along the posterior right margin. T2W (b) image reveals that the mass is mildly hyperintense with areas of marked hyperintensity corresponding to the cystic/necrotic components. DWI (c) reveals that the mass is isointense and that the cystic regions have decreased intensity due to unrestricted water motion. Imaging findings are thus non-specific.

Abnormalities on DWI and MRS may therefore be those of infarction rather than infection. Serial MR perfusion examinations may prove helpful in assessing vascular compromise, and its response to therapy. Leptomeningeal enhancement, hydrocephalus or parenchymal masses (granuloma) provide clues to the correct diagnosis. MRA may be used to assess the intracranial vessels although small vessel abnormalities may be visible only on catheter (Castro and Hesselink, 1991; Harris and Edwards, 1991).

Parasitic infections CNS parasitic infections share many features with granulomatous diseases (Alvard and Shaw, 1977; Chang et al., 1991). They are common in poor areas with inadequate hygiene and they are rare in most parts of the United States and other developed countries. They produce a variety of inflammatory responses and therefore may have protean clinical and imaging features. At the time of initial infestation there is often an acute episode of encephalitis. Subsequently, the infection enters an indolent phase in which typical granulomatous changes are seen.

Imaging features including leptomeningeal enhancement, hydrocephalus, parenchymal granulomas and infarction secondary to vasculitis (Chang et al., 1991). There are three unique imaging features of parasitic diseases. First, since many of the organisms are macroscopic rather than microscopic they may be directly visualized in imaging studies. Organisms such as cysticercosis have readily identifiable features when alive. Second, the parasite may be mobile and capable of changing positions over time. Third, the organism has a finite lifespan and typically does not reproduce in the human brain. Symptoms often arise only after the death of the parasite when host response occurs. A comprehensive review of imaging features of CNS parasitic disease is beyond the scope of this chapter. We will concentrate on the two parasitic diseases that have significant prevalence in the developed world: cysticercosis and toxoplasmosis. Cysticercosis Cysticercosis is the most common intracranial parasitic disease (Esocbar, 1983; Chang et al., 1991; Davis and Kornfield, 1991). It is caused by ingestion of

363

364

Robert D. Zimmerman

(a)

(b)

(c)

(d)

(e)

(f)

Fig. 22.8 Aspergillosis in patient with lymphoma. Axial FLAIR (a–c) reveals multiple lesions in both hemispheres. Lesions have mild surrounding edema and a central ill-defined solid or ring-like hypointensity. On T2W echo planar imaging (EPI) image (d) (the b0 image from a DWI sequence) the lesions are more hypointense indicative of strong susceptibility effect of hemorrhage; on DWI (e), the lesions are hyperintense. Other lesions (not shown) were iso- to hypointense. Contrast enhanced image (f) reveals very little enhancement. Minimal edema and enhancement reflect the paucity of host response due to immunosupression.

undercooked pork. The larvae of the pork tapeworm (Taenia solium) enter the intestinal wall where they develop into secondary larvae (cysticerci), translucent cysts containing a scolex (the parasite), to enter the vascular system and lodge in the brain, the subarachnoid space or the ventricles. An initial episode

of encephalitis may occur but symptoms are transient and this stage of the infestation is rarely imaged. Once the scolex is established, it makes itself immunologically invisible to the host and therefore incites no inflammatory reaction (Chang et al., 1991; Davis and Kornfield, 1991).

Physiological imaging in infection, inflammation and demyelination: overview

Live cysts are uniform in size (4–15 mm, averaging about 10 mm), isointense to CSF on all pulse sequences (including DWI), with a slightly hypointense rim which makes them visible when they reside in the CSF. The scolex is seen as a 2–4 mm mural nodule in the cyst wall that is hyperintense on T1W images and FLAIR. There is no enhancement or edema while the organism is alive. Cysts may occur in the parenchyma, the subarachnoid spaces or within the ventricles. Involvement of more than one space is common (Teitelbaum et al., 1989). Ventricular and subarachnoid cysts may float within the CSF and lodge at narrow ventricular outlets leading to acute obstructive hydrocephalus. When the organism dies (the larvae can live from 1 to 6 years), it becomes immunologically “visible” leading to an inflammatory response. The cyst fluid becomes hyperintense to CSF on T2W images and FLAIR, the wall enhances and there is vasogenic edema. The scolex will no longer be visualized. With healing the inflammatory response resolves and the dead cysts may calcify. Punctate foci of hypointensity may be identified on T2W images or susceptibility weighted sequences. Physiological techniques do not play a large roll in the diagnosis of cysticercosis. The live organisms have a highly characteristic appearance on routine MRI. Due to the small size of the cysts and their predilection for surface of the brain MRS has been difficult to perform and findings are relatively nonspecific. DTI and perfusion MR would appear to have little value in the assessment of this disorder. Toxoplasmosis Toxoplasma gondii is an obligate intracellular protozoan parasite with a worldwide distribution. Over 20% of the population is seropositive. The organism has low virulence and causes intracranial infection only in immunocompromised patients. Rarely seen prior to the onset of the AIDS epidemic, toxoplasmosis has become quite common. When toxoplasmosis invades the brain, it causes acute encephalitis. Focal mass lesions of variable size are seen with central areas of necrosis. Although grossly similar to an abscess, the lesion is not encapsulated and therefore histologically classified as encephalitis rather than

abscess or granuloma. In the majority of cases multiple mass lesions are present and they may be located anywhere within the brain (Post et al., 1983; Dina, 1991; Whiteman et al., 1996; Smirniotopoulos et al., 1997) The imaging findings in toxoplasmosis are a reflection of these histopathological features. Multiple lesions are usually present. The intensity and enhancement patterns are variable much as in granulomatous lesions. The central necrosis is typically hyperintense on FLAIR and T2W images. The intermediate reactive inflammatory zone has heterogeneous intensity but it is usually relatively hypointense to the central necrosis and peripheral edema on T2W images and FLAIR. DWI reveals heterogeneous intensity. Occasionally the central necrosis is hyperintense (as in abscesses) and in other lesions (sometimes in the same patient) the intermediate zone is hyperintense. Ring enhancement is common and a small central focus of enhancement within the necrotic cavity (“target sign”) is characteristic for toxoplasmosis (Figure 22.9). Enhancement may vary from shaggy, multilocular rings to discrete solid masses. Hemorrhage is not present at the time of initial diagnosis (Post et al., 1983; Dina, 1991; Smirniotopoulos et al., 1997). Toxoplasmosis can mimic the appearance of a wide variety of inflammatory and neoplastic processes. Most patients who present with mass lesions are known to have AIDS, simplifying the diagnosis. The major goal in these patients is the differentiation between toxoplasmosis and lymphoma, the two most common causes of mass lesions in AIDS patients. Lymphoma produces a more consistent pattern of brain involvement. Lesions are usually single and confined to the deep GM and WM (basal ganglia and corpus callosum). Lymphoma is often hypointense on T2W images and may be mildly hyperintense on T1W images (Dina, 1991; IglesiasRozas et al., 1991; Smirniotopoulos et al., 1997). There is mild edema and uniform enhancement. Lymphoma is often hyperintense on DWI and hypointense on ADC maps most likely due to the high nuclear to cytoplasmic ratio of these small cell tumors (Camacho et al., 2003). When imaging features typical of toxoplasmosis are encountered it is appropriate to treat empirically

365

366

Robert D. Zimmerman

(a)

(b)

(c)

Fig. 22.9 Toxoplasmosis: 35-year old man with AIDS. Axial T2W (a) and contrast and enhanced T1W images (b) reveal non-specific bilateral ill-defined rim-enhancing masses with surrounding edema. On DWI (c) the left-sided lesion is centrally hyperintense while the right-sided lesion has hyperintensity in the capsule. In the setting of AIDS toxoplasmosis is the most likely diagnosis. Patient will be treated empirically and then rescanned in 7–10 days. If lesions are decreasing in size treatment is continued but if there is no change or an increase in the lesions MRS, MR perfusion or SPECT thallium could be performed to evaluate for lymphoma. Biopsy would be reserved for cases in which non-invasive studies fail to reveal a definitive diagnosis.

with anti-toxoplasmosis therapy (Whiteman et al., 1996). Serial studies should reveal partial resolution (decreased lesion size, mass effect and enhancement) of all lesions and lesions should be followed to resolution. When a diagnosis of lymphoma is suspected single photon emission CT (SPECT) thallium can be used to confirm the diagnosis prior to therapy. Inflammatory lesions including toxoplasmosis are negative on SPECT while lymphoma is positive (O’Malley et al., 1994). False negative SPECT thallium may occur if patients are receiving steroids. MRS profiles differ between neoplastic and inflammatory processes and may prove helpful in ambiguous cases (Simone et al., 1998) (cf. Chapter 28). MR perfusion studies may also prove helpful since lymphoma typically produces hyperperfusion while toxoplasmosis produces hypoperfusion. When the diagnosis cannot be established non-invasive biopsy may prove necessary. It is important to remember that AIDS patients can simultaneously harbor lesions with different etiologies and therefore lesions that do not completely resolve with anti-toxoplasmosis therapy should be evaluated. With healing, toxoplasmosis foci may become hyperintense on T1W images. These hyperintense foci can be confused with persistent enhancement and therefore it is important to compare the

pre- and post-enhanced scans carefully. This hyperintensity is probably the result of calcification (about half of these foci are also dense on CT) hemorrhage of laminar necrosis.

Encephalitis All unencapsulated inflammatory lesions of the brain belong to the broad histopathological category of encephalitis. This group is extremely heterogeneous with few characteristic pathological or imaging features (Alvard and Shaw, 1977; Jordon and Enzmann, 1991; Whiteman et al., 1996; Sze and Zimmerman, 1988). Encephalitis may be diffuse or localized. While it is most often an acute and selflimiting process, there are several chronic progressive diseases as well (HIV, PML and subacute sclerosing panencephalitis (SSPE)). Most of these disorders result from viral infections but they may also be seen on autoimmune basis (acute disseminated encephalomyelitis (ADEM)) or secondary to prion disease (Creutzfeldt–Jakob). CNS viral infections can be caused by a wide variety of agents. These diseases often involve the meninges and spinal cord as well as the brain. Some of these diseases have characteristic and distinctive

Physiological imaging in infection, inflammation and demyelination: overview

appearances that allow for differentiation from each other and from non-inflammatory lesions. However, experience with the imaging features of many is limited because patients are either too well to need imaging or too ill to undergo it. Epidemic viral illnesses often occur in parts of the world where there is no easy access to MR scanners. Many of these disorders produce subtle and/or diffuse changes that are difficult to identify on CT and MR in particular in the early course of the disease. The advent of FLAIR and physiological techniques such as DWI and MRS has dramatically improved our ability to detect and characterize these processes. Areas of acute encephalitis are usually isointense on T1W images. Diffuse, ill-defined hyperintensity is seen on FLAIR and, to a lesser extent, T2W images. In many cases subtle changes are not visible on low quality images and/or are easily overlooked. Hyperintensity becomes more apparent as the disease progresses. Mild mass effect is often present in acute encephalitis while enhancement is absent unless there is associated meningitis. In the subacute phase diffuse ill-defined GM or WM enhancement may be seen that usually resolves in chronic encephalitis (Jordon and Enzmann, 1991). DWI is a valuable tool in the diagnosis of viral encephalitis. Many types of encephalitis have been reported to be hyperintense on DWI due to restricted water movement (decreased ADC) (Tsuchiya et al., 1999). This is probably the result of change in the intracellular environment. Cell damage and death lead to increased intracellular viscosity that restricts water motion in a process similar to that seen in ischemic disease. As with infarction, DWI may reveal signal abnormality prior to the development of hyperintensity on FLAIR and hyperintensity on DWI resolves in the subacute and chronic phases as the dead cells are removed and replaced by gliosis and encephalomalacia (Tsuchiya et al., 1999). However, findings of DWI in encephalitis are more variable and less predictable than in infarction. Some cases of acute encephalitis may not demonstrate hyperintensity on DWI. Whether this is a characteristic feature of some forms of encephalitis or an individual variation is not known. Absence of DWI hyperintensity may provide valuable clinical information. In one example of a patient with West Nile encephalitis

and an episode of hypoxia, routine images revealed deep GM hyperintensity on FLAIR compatible with either infarction or encephalitis. Absence of hyperintensity on DWI ruled out infarction. There have been only a few MRS examinations in patients with viral encephalitis. The most common finding is a non-specific reduction in N-acetyl aspartate (NAA) and in some instances increased lactate (Lac) (Gupta and Lufkin, 2001). It is unlikely that MRS will be a useful tool in the detection and characterization of acute encephalitis, but it may prove helpful in assessing the extent of injury and therefore prognosis. The extent of regional and/or whole brain reduction in NAA may prove to be a marker of the extent of brain damage in chronic and/or diffuse processes such as HIV encephalitis, SSPE, progressive multifocal leukoencephalits (PML) and Creutzfeldt– Jakob disease (CJD). In these disorders, there is progressive diffuse or multifocal brain involvement that may not be detectable on routine imaging. DTI and MR perfusion studies may also provide information on the extent of damage to WM and cerebral blood flow (CBF) respectively in these diffuse chronic infections. While the pattern of intensity changes are relatively similar for all types of encephalitis, the distribution of lesions and their natural history allows for accurate diagnosis in most diseases. The pathological distribution of these lesions is dependent upon a number of factors. Viral agents most commonly gain access to and spread within the CNS hematogenously but direct spread from the adjacent sinuses, and retrograde perineural spread along cranial nerves also occurs (Alvard and Shaw, 1977; Jordon and Enzmann, 1991; Whiteman et al., 1996). Once the organism gains access to the CNS it can extend along the cranial nerves, WM tracks, the perivascular spaces or within the vessel walls. The virus may infect different cell types within the CNS. Viral agents vary in virulence. The important imaging features of common causes of encephalitis will be discussed briefly. HSV1 This is the most common cause of sporadic viral encephalitis with a mortality of over 50% if untreated

367

368

Robert D. Zimmerman

but acyclovir is effective if given early (3 days). HSV1 is a common pathogen of the upper respiratory tract and infection may spread directly via olfactory nerves through the cribiform plate. More commonly, encephalitis results from reactivation of latent viral infection of the Gassarian ganglion. The virus spreads along the meningeal branches of the fifth nerve and directly invades adjacent brain where it produces a hemorrhagic necrotizing meningo-encephalitis (Carey and Spear, 1986a, 1986b; Sener, 2001). This mode of spread accounts for its predilection for the inferior anterior surfaces of the temporal lobes and the inferior surface of the frontal lobes. The process affects contiguous GM and WM extending medially across the sylvian fissure to the lateral margin of the lentiform nucleus and posteriorly along the hippocampus. Initially the infection may appear unilateral but over time involvement of the contra-lateral temporal and frontal lobe will become apparent. As the disease progresses it is common to see diffuse involvement of the supratentorial WM. Early changes (3 days) are difficult to detect on CT and routine spin echo (SE) MR sequences usually show only subtle hyperintensity on T2W sequences confined to one temporal lobe (Tien et al., 1983; Carey and Spear, 1986a, 1986b). The acute phase of HSV1 encephalitis is much more easily appreciated with FLAIR and DWI (Tsuchiya et al., 1999). Hyperintensity is encountered in the expected locations and the bilateral nature of the process is usually evident. Gradient echo (GE) scans may reveal small foci of hypointensity, reflecting areas of hemorrhage. DWI has facilitated the early diagnosis (3 days) of herpes encephalitis but this may not have a major effect on outcome since acyclovir should be given whenever clinical findings of suggestive HSV1 are present. Imaging should be performed after the institution of therapy to confirm the diagnosis and determine the extent of involvement. Superficial leptomeningeal and/or cortical enhancement is encountered after 5–7 days. Serial imaging usually reveals gliosis and atrophy at the sites of brain destruction. Generalized atrophy is encountered in severe cases. MRS reveals a decreased NAA/creatine (Cr) ratio, a non-specific finding indicative of neuronal loss (Menon et al., 1990) (Figure 22.10).

Other herpes viruses Epstein–Barr virus (EBV) has been linked to diverse entities such as chronic fatigue syndrome, Guillaine–Barrre syndrome and lymphoma in AIDS patients. Lymphocytes of patients with AIDS are often infected with the virus and polymerase chain reaction (PCR) studies reveal evidence of EBV in the CSF in all AIDS patients with lymphoma. It is thought that over time the chronic EBV infection incites malignant transformation in infected lymphocytes. It is the agent that causes infectious mononucleosis. About 5% of patients with mononucleosis develop an acute self-limited encephalomyelitis. This disorder has a striking predilection for the deep GM and central GM of the spinal cord. These structures are swollen and hyperintense in FLAIR and T2W images. Rapid resolution of clinical findings and imaging abnormalities usually occurs (Tolly et al., 1989; Donovan and Zimmerman, 1996). Herpes Zoster produces chicken pox in children and superficial neuritis in adults. In immunocompromised patients acute infection may lead to a hematogenous spread and diffuse encephalitis or there may be perineural extension from the face to the cavernous sinus. When this occurs the virus may spread along cranial arteries producing an infarctlike pattern of acute edema (Tenser, 1984).

Enterovirus These viruses include poliovirus, coxsackie and echoviruses. The infections are most common in infants and young children. Epidemics occur but the disease is usually mild and self-limiting (e.g. hand, foot and mouth disease). The epidemics associated with the poliovirus were a tragic exception to this rule. Encephalitis has a characteristic pattern of involvement regardless of its severity. The disease affects the ventral horns of the spinal cord and central portion of the medulla pons and mesencephalon, the dentate nucleus of cerebellum and occasionally the thalamus. These structures appear swollen and hyperintense on T2W images. In mild cases these changes resolve but in severe cases (such as occurred in the polio epidemic) permanent damage to these areas

(a)

(b)

(c)

(d)

(e)

(f)

(g)

(h)

(i)

(j)

(k)

(l)

(m)

(n)

(o)

Fig. 22.10 HSV1 encephalitis: 30-year old man with 3 days of viral symptoms and 1 day of obtundation and seizures. CT scan (a and b) is normal. FLAIR (c and d) performed 1 h after CT reveal subtle hyperintensity in the left uncus hippocampus, anterior temporal lobe and insula. On DWI (e and f) the hyperintensity is more apparent and extensive. ADC maps (g and h) reveal hypointensity in these regions due to restricted water motion. Repeat examination 10 days later reveals increase in the extent of edema and mass effect on FLAIR (i and j). Note involvement of the anterior medial frontal lobe (i) and the contra-lateral insula (j). Hyperintensity on DWI (k and l) has resolved indicating that signal changes on FLAIR and mass effect are due to vasogenic edema. There is no contrast enhancement at the time the original examination (m) while examination on day 10 (n) reveals extensive cortical and leptomeningeal enhancement. Long TR PRESS single voxel MRS (o) reveals mild elevation in choline (Cho), marked depression of NAA and presence of Lac.

370

Robert D. Zimmerman

(a)

(d)

(b)

(c)

(e)

(f)

Fig. 22.11 PML: 41-year old patient with lymphoma and anterior frontal meningioma (not shown). Axial FLAIR (a–c) images reveal the presence of WM hyperintensity in both parieto-occipital lobes and the right side of the genu of corpus callosum. No mass effect or enhancement was present. Axial DWI (d–f) reveals hyperintensity in the margins of the lesions.

produce gliosis and encephalomalacia. The cause of this striking anatomic predilection is not known (Wasserstrom et al., 1992; Shen et al., 1999).

features. MRI in one of these patients revealed bilateral deep GM edema similar to that seen in EBV encephalitis (cf. Case Study 24.1).

Arbovirus

PML

Viral encephalitis may result from spread of a structurally diverse group of viruses from animal hosts via insect vectors. Relatively non-specific appearing areas of WM edema have been reported (Jordon and Enzmann, 1991). In the summer of 1999, a mosquito borne disease initially thought to be St. Louis encephalitis but subsequently shown to be West Nile encephalitis occurred in New York. MRI in one of these patients revealed bilateral deep GM edema similar to that seen in EBV encephalitis. Over the past few years many cases of West Nile Encephalitis, some fatal have been identified with similar imaging

PML is cause by a papovavirus (the JC virus). This virus is ubiquitous but it causes encephalitis only in immunocompromised patients. The virus infects the myelin-producing oligodendroglia resulting in demyelination with little inflammatory reaction (Levy et al., 1986; Richardson, 1988; Whiteman et al., 1993). There is patchy ill-defined hyperintensity in the WM on T2W images and FLAIR without mass effect or enhancement. DWI studies reveal heterogeneous intensity with hyperintensity occasionally encountered at the advancing edge of the lesions (Figure 22.11). Absence of diffuse hyperintensity can

Physiological imaging in infection, inflammation and demyelination: overview

(a)

(b)

Fig. 22.12 HIV encephalitis: axial FLAIR images reveals symmetric bilateral hyperintensity on the periventricular WM and upper brain stem. There is no mass effect or enhancement.

help to differentiate subcortical lesions from acute infarction. There is a predilection for the parietooccipital WM. As the name implies there is steady increase in size and number of lesions. There is no direct treatment and patients usually die within 6 months of onset; however anti-AIDS drugs may improve survival by strengthening the immune response. Brain stem and cerebellar lesions occur but less than 10% of PML is solely infratentorial. GM structures including cortex and deep nuclei may be involved. HIV encephalitis AIDS is caused by the human RNA retrovirus HIV-1. The brain is virtually always affected which acts as viral reservoir. The immunosupression caused by AIDS leads to numerous secondary CNS infections and primary CNS lymphoma (there is probably an increase in gliomas as well). The HIV virus causes subacute progressive encephalitis. The organism replicates within multi-nuclear giant cells and macrophages in the WM and later spreads to the GM. Atrophy, micro-glial nodules and gliosis are

seen. Despite the primary involvement of the WM, imaging evidence of demyelination is seen only in advanced stages of the disease. Acute inflammatory changes are typically absent (Levy et al., 1986; Post et al., 1988; Sze and Zimmerman, 1988; Theur et al., 1990; Whiteman et al., 1993, 1996; Barker et al., 1995; Simone et al., 1998). The most common finding in imaging studies is generalized atrophy without focal CT density or MR intensity abnormalities. In severe cases, diffuse symmetric hyperintensity is seen in the supratentorial WM predominantly in the periventricular region. Mass effect and enhancement are absent. In the past, rapid progression of atrophy occurred and death usually ensued within a year. The introduction of effective anti-AIDS drugs has allowed for treatment of HIV encephalitis. Regression of WM hyperintensity can occur and in some patients atrophy may appear to improve as well (Figure 22.12). The imaging findings of early HIV encephalitis (mild atrophy) are minimal and non-specific. Normal variation in ventricular and sulcal size is present in the general population and patients with AIDS may develop some degree of non-atrophic brain

371

372

Robert D. Zimmerman

shrinkage due to systemic effects of the disease. Clinical and neuropsychiatric evidence of AIDS dementia complex typically precedes the development of abnormalities on CT and MR. MRS studies have revealed extensive abnormalities in these patients (Chong et al., 1993; Jarvik et al., 1993). For instance, the NAA/Cr ratio is decreased reflecting neuronal loss while Cho/Cr ratios is increased. Whole brain metabolite measurement could prove extremely useful in determining the extent of disease and treatment response. DTI also holds promise in assessing the extent of WM abnormality (Ragin et al., 2004). CJD The characteristic pathological finding of CJD is spongiform degeneration of neurons in the cortex and subcortical nuclei. Several human and animal diseases produce this distinctive pattern including Kuru, Bovine Spongiform Encephalopathy (“Mad Cow Disease”) and Scrapie (sheep). These diseases are caused by prions, unique non-viral, non-nucleic acid transmissible agents. Prions are normal protein constituents of the neuronal cell membranes but they can exist in an isoform that is insoluble. Over time these pathological proteins can accumulate in cells leading to vacuolization and cell death. Conversion to the pathological isoform can be spontaneous but the proteins have the unique ability to cause normal prions to flip into the pathological conformation and therefore act as infectious agents. Once introduced into the brain, there is progressive conversion or spread leading to diffuse degeneration. Prions are smaller than viruses and therefore pass through devices capable of filtering viral agents from blood. Infection may result from blood transfusion, corneal transplantation or other contact with contaminated tissue. Ceremonial ingestion of infected human brains by hunter–gatherer tribes in New Guinea is the cause of Kuru. Cases of variant CJD are attributed to cross species transmission of humans with bovine prion protein from infected beef. To make matter more complex, some cases have a familial incidence suggesting that the disease has a heritable form (Prusiner, 1987). CJD presents late in life (50 years of age) with rapid onset of dementia and myocclonic jerks. Most

patients are dead within a year of onset of symptoms. Imaging reveals evidence of atrophy that progresses rapidly on serial studies. Symmetric basal ganglia hyperintensity on T2W images have been identified in many patients. Some of the initial reports of the incidence of this finding were contradictory, but it is likely that ganglionic hyperintensity is present in about 75% of patients with CJD (Milton et al., 1991; Falcone et al., 1992). The hyperintensity is often mild and difficult to detect on T2W images. Hyperintensity is more apparent on FLAIR. Cortical hyperintensity may also be seen on FLAIR, reflecting the known pathological distribution of the disease. DWI has been shown to be more sensitive than FLAIR in detecting areas of CJD encephalitis (Bahn and Parchi, 1999). This probably reflects changes in the intracellular environment of the affected neurons rather than shift of fluid from the extra-cellular to the intracellular space (cytotoxic edema), and it has been shown that this can change over the course of the disease. Most imaging abnormalities in CJD are bilateral but these are frequently asymmetrical and the disease may appear unilateral. Infarct and CJD can be differentiated on a clinical basis in most cases (Figure 22.13). MRS has shown varying distributions of metabolite abnormality; notably widespread non-specific NAA loss, but at short TE increased myo-inositol (mI) is also seen.

Demyelinating diseases Perhaps the most dramatic contribution of MR in clinical imaging has been in the detection of demyelinating diseases. Even in its infancy the clear superiority of MR over CT in the detection of MS made the potential of this new modality apparent (Jackson et al., 1985). The sensitivity of MR to changes in tissue water made it ideal for the detection of absence of hydrophobic myelin. With each advance in MR, the ability to detect and assess demyelinating disease has improved. Fast spin echo (FSE) scanning techniques, FLAIR and fat suppression made it possible to detect lesions near spinal fluid spaces, within the optic nerves and the spinal cord (Figure 22.14). Routine imaging techniques are usually sufficient to make the correct diagnosis but physiological

Physiological imaging in infection, inflammation and demyelination: overview

(a)

(b)

(c)

(d)

(e)

(f)

Fig. 22.13 CJD: axial FLAIR (a–c) reveals subtle hyperintensity in the cortical GM and basal ganglia without mass effect. Hyperintensity is more pronounced and extensive on DWI (d–f). Note that the disease is bilateral but asymmetric.

imaging will likely play an important roll in at least two ways. Acute foci of demyelination may present clinically and in imaging studies as mass lesions that may be mistaken, at least by the unwary, for malignant gliomas or infectious lesions. These cases of tumifactive demyelination may occur in the patients with MS or in ADEM (Figure 22.15). In addition, physiological imaging studies, in particular those that measure whole brain values of various parameters (e.g. NAA, FA, ADC) may allow for a more accurate assessment of the extent of disease and its progression. These techniques have demonstrated that MS is a diffuse rather than or in addition to a multifocal disease (Allen and McKeown,

1979). Normal appearing WM (on routine MRI) will often be abnormal on MRS DTI and perfusion studies. Demyelinating diseases were also in the forefront of another use for MR, as a primary outcome measure for treatment trials. The waxing and waning nature of the illness, its inter-patient variability and chronicity make it difficult to use clinical status as an outcome measure. A variety of findings on MR have been utilized to assess lesion load and treatment response including number and volume of lesions on T2W images (including proton density and FLAIR), T1W images and contrast enhanced images. None of these measures have proven wholly satisfactory in

373

374

Robert D. Zimmerman

(b)

(a)

(d)

(h)

(c)

(e)

(i)

(f)

(j)

(g)

(k)

Fig. 22.14 MS: axial (a–c) and sagittal (d and e) FLAIR images reveal typical location of MS plaques in the periventricular WM (c and e) (Dawson’s fingers), the under surface of the corpus callosum (d and e), the temporal lobes adjacent to the temporal horns (a and b) and the brain stem (a). Multiple ring and nodular-enhancing foci (f) noted in the periventricular WM in patient with chronic MS and confluent demyelination (g). Incomplete ring enhancement (h) noted in patient with single large acute focus of demyelination (i) with mild mass effect. Chronic demyelination produces focal “black holes” on T1W images (j). Subtle hyperintensity noted in the rims of some of the lesions. On FLAIR (k) lesions are confluent and irregular.

part because they do not allow for assessment of diffuse disease in normal appearing WM. MRS and DTI have shown much promise as tools for assessing lesion load (Arnold et al., 1990; Grossman et al., 1992). Whole brain NAA measures allow for assessment of neuronal integrity. DTI seems almost ideally suited to assessing the extent of WM damage (Filippi et al., 2001b). Diffusion in normal WM is highly ordered

(anisotropic) with water molecules moving along not across WM tracts. Processes that damage WM lead to less ordered more isotropic water diffusion. Local or diffuse decrease in anisotropy is therefore a feature of all demyelinating processes. MS has several characteristic features on MR. Lesions are typically round or oval and measure less than a centimeter. Foci of ischemic demyelination are

Physiological imaging in infection, inflammation and demyelination: overview

(a)

(b)

(e)

(f)

(i)

(j)

(c) (a)

(d)

(g)

(k)

(h)

(l)

Fig. 22.15 Tumifactive demyelination – ADEM: 42-year old man with AIDS, normal CD4 count and negative toxoplasmosis titers. Axial FLAIR (a and b) reveals a large hyperintense lesion in the left frontal WM. A second small focus is noted just anterior to the large lesion (b). There is a central hypointense rim in the lesion (b). The lesion is hyperintense on DWI with a central core that is hypointense (c and d). Contrast enhanced axial (e and f) and coronal (g and h) T1W images scans reveal rim enhancement. Note that on coronal image (h) there are subtle concentric rings. Also note additional foci of enhancement in the right hemisphere WM (e) and the left frontal subcortical region (f). These regions appeared normal on FLAIR. Repeat examination 1 week later reveals increase in the size of the large lesion in the left periventricular region on FLAIR (i and j). Hyperintensity is now noted in the region of the small enhancing lesions and additional foci of hyperintensity are present. Contrast enhanced saggital (k) and axial (l) images reveal increase in the extent of enhancement and number of enhancing lesions. Note concentric rings in large lesion on saggital FLAIR (k).

on the other hand more irregular in contour and size. Both MS and ischemia have a predilection for the periventricular region but MS plaques have a long axis that is perpendicular to the ventricular lining. These orientation of these lesions, termed Dawson’s fingers, is a reflection of their perivenous location. MS has a predilection for several locations not seen in ischemic disease including the under surface of the corpus callosum (Gean-Marton et al., 1991), the WM adjacent to the temporal horns, the brachium pontus and medulla (Nusbaum et al., 2002).

The intensity of the MS plaque and presence or absence of enhancement is dependent upon the age of the lesion. The initial event in the formation of a plaque is disruption of the blood brain barrier. Therefore in rare cases, enhancement may occur in the absence of any intensity change. Edema develops as a consequence of the barrier disruption producing a focus of hyperintensity on any long repetition time (TR) (T2W) sequence. There may be a subtle rim of hypointensity near the margin of the plaque (Powell et al., 1992). This finding is similar to that seen in brain

375

376

Robert D. Zimmerman

abscesses and may have the same etiology, the presence of atomic oxygen generated by macrophages. Plaques are typically isointense on T1W images at this time although a subtle focus of hypointensity with a subtle hyperintense rim may be encountered. Enhancement of small plaques is typically smooth and ring like although small plaques may also reveal nodular enhancement (Grossman et al., 1986). After a few weeks the acute inflammatory edema is replaced by demyelination. The plaque no longer enhances. If demyelination persists the lesion often becomes discretely hypointense on T1W images (black hole). Over time the plaque may diminish and disappear due to remyelination or it may persist indefinitely. The ability to age plaques is complicated by the fact that remyelination and repeat demyelination can produce findings that do not correlate with clinical findings. Physiological examinations may aid in aging plaques. MRS spectral patterns change while plaques evolve but the clinical value of this information is not clear (Grossman et al., 1992). MS plaques have variable signal on DWI (Nusbaum et al., 2000a; 2000b). Most often they are iso- to hypointense. Hyperintensity may be seen secondary to T2 shine through or infrequently restricted water motion. Plaques with restricted diffusion are more likely to be acute than chronic but most acute plaques are not hyperintense and hyperintensity may be seen in the absence of other MR or clinical evidence activity. Thus in clinical practice DWI adds little (except confusion) to the assessment of patients with MS and we longer use it in our routine imaging of these patients. Occasionally foci of demyelination may present as rapidly enlarging mass lesions with central necrosis rim enhancement and extensive edema (Atlas et al., 1986; Kepes, 1993; Zagzag et al., 1993). These lesions may be mistaken for malignant neoplasms (hence the name). If the correct diagnosis is not suspected surgical biopsy or resection may be performed. To make matters more complex, the lesions often contain bizarre astrocytes that may be mistaken for neoplastic cells. If the incorrect diagnosis is made radiation may be given with devastating effect since MS renders the brain extremely sensitive to radiation. Correct interpretation of the imaging studies is therefore critical to patient care. There are several features that help to distinguish tumifactive demyelination from neoplasm.

The presence of additional WM lesions typical for MS is a good indication of the correct diagnosis but it must be remembered that patients with MS may also develop gliomas. Tumifactive demyelination tends to enlarge more rapidly than neoplastic masses often doubling in size over 10–14 days. Malignant gliomas usually require at least a month to appreciably enlarge. Two patterns of enhancement are typical of demyelination. First, the enhancing rim is often incomplete (C-shaped) along its deep margin. Second, concentric rings of enhancement may be present presumably due to recurrent episodes of inflammation. This pattern may correspond to the pathological variant of MS, Balo’s concentric sclerosis. When either of these patterns is identified tumifactive demyelination should be suspected. Although MS and gliomas are typically typically described as having different spectral patterns, the use of MRS in this differential diagnosis should be treated with caution, as spectra of the two conditions are sometimes indistinguishable. Perfusion imaging may be more helpful in this regard (cf. Case Study 19.4 and Case Study 26.1).

REFERENCES Allen IV, McKeown SR. 1979. A histological, histochemical and biochemical study of the macroscopically normal white matter in multiple sclerosis. J Neurol Sci 41: 81–91. Alvard EC, Shaw CM. 1977. Infectious allergic and demyelinating diseases of the central nervous system. In Radiology of the Skull and Brain: Anatomy and Pathology (Eds., Newton TH, Potts DG), St. Louis Mo. C.V. Mosby Co, 3088–3172. Arnold DL, Mathews PM, Francis G, et al. 1990. Proton magnetic resonance spectroscopy of human brain in in vivo the evaluation of multiple sclerosis: assessment of the load of the disease. Magn Reson Med 14: 151–159. Atlas SW, Grossman RI, Goldberg HI, et al. 1986. MR diagnosis of acute disseminated encephalomyelitis. J Comput Assist Tomogr 10: 798–801. Bahn MM, Parchi P. 1999. Abnormal diffusion-weighted magnetic resonance imaging images in Creutzfeldt–Jakob disease. Arch Neurol 56: 577–583. Barker PB, Lee RR, McArthur JC. 1995. AIDS dementia complex: evaluation with proton MR spectroscopic imaging. Radiology 195: 58–64.

Physiological imaging in infection, inflammation and demyelination: overview

Bazan C, Rinaldi MG, Rauch RR, Jinkins JR. 1991. Fungal infections of the brain. Neuroimaging Clin N Am 1: 57–88. Burtscher IM, Holtas M. 1999. In vivo proton MR spectroscopy of untreated and treated brain abscesses. Am J Neuroradiol 20: 1049–1053. Camacho DLA, Smith JK, Castillo M. 2003. Differentiation of toxoplasmosis and lymphoma in AIDS patients by using apparent diffusion coefficients. Am J Neuroradiol 24: 633–637. Carey L, Spear PG. 1986a. Infection with herpes simplex viruses: Part 1. N Engl J Med 314: 680–691. Carey L, Spear PG. 1986b. Infection with herpes simplex viruses: Part 2. N Engl J Med 314: 749–757. Castro CC, Hesselink JA. 1991. Tuberculosis. Neuroimaging Clin N Am 1: 119–140. Cecil KM, Kenkinski RE. 1998. Proton MR spectroscopy in inflammatory and infectious brain disorders. Neuroimaging Clin N Am 8: 863–880. Centeno RS, Winter J, Bentson JR, et al. 1983. CT in the evaluation of Haemophilus influenzae meningitis with clinical and pathologic corrrelation. Comput Radiol 7: 243–249 Chang KH, Cho YS, Hesselink JR, Han MH, Han MC. 1991. Parasitic diseases of the central nervous system. Neuroimaging Clin N Am 1: 159–178. Chang KH, Han MH, Roh JK, et al. 1990. GTPA-enhanced imaging of the brain in patients with meningitis: comparison with CT. Am J Neuroradiol 11: 69–76. Chong WK, Sweeney B, Wilkinson L, et al. 1993. Proton spectroscopy of the brain in HIV infection: correlation with clinical immunologic and MR imaging findings. Radiology 188: 119–124. Davis LE, Kornfield M. 1991. Neurocysticercosis; neurologic, pathogenic, diagnostic and therapeutic aspects. Eur Neurol 31: 229–240. Desprechins B, Stadnik T, Koerts G, Shabana W, Breucq C, Osteaux M. 1999. Use of diffusion-weighted MR imaging in differential diagnosis between intracerebral necrotic tumors and cerebral abscesses. Am J Neuroradiol 20: 1252–1257. Dina TS. 1991. Primary central nervous system lymphoma versus toxoplasmosis in AIDS. Radiology 179: 823–828. Donovan W, Zimmerman RD. 1996. MRI finding in severe Epstein–Barr virus encephalomyelitis. J Comput Assist Tomogr 1010–1011. Ebisu T, Tanaka C, Umeda M, et al. 1996. Discrimination of brain abscess from necrotic or cystic tumors by diffusionweighted echo planar imaging. Magn Reson Imaging 14: 1113–1116. Enzmann DR, Britt RH, Placone R. 1983. Staging of human brain abscess by computed tomography. Radiology 146: 703–708. Esocbar A. 1983. The pathology of Cysticercosis. In Cysticercosis of the Central Nervous System (Eds., Palaiose E,

Rodriquez-Cabajal J, Taveras JM), Springfield, Ill, Charles C. Thomas 27–59. Falcone S, Quencer RM, Bowen BC, Bruce JH, Nadich TP. 1992. Creutzfeldt–Jakob disease: focal symmetric cortical involvement demonstrated by MR imaging. Am J Neuroradiol 13: 403–406. Filippi C, Ulug A, Ryan E, et al. 2001a. Diffusion tensor imaging of patients with HIV and normal-appearing white matter on MR images of the brain. Am J Neuroradiol 22(2): 277–283. Filippi M, Cercignani M, Inglese M, Horsfield MA, Comi G. 2001b. Diffusion tensor magnetic resonance imaging in multiple sclerosis. Neurology 56: 304–311. Garg M, Rakesh K, Gupta MH, et al. 2004 Brain abscesses: etiologic categorization with in vivo proton MR spectroscopy. Radiology 230: 519–527. Gean-Marton AD, Venzina LG, Marton KI, et al. 1991. Abnormal corpus callosum; a sensitive and specific indicator of multiple sclerosis. Radiology 180: 215–221. Grossman RI, Gonzalez-Scarano F, Atlas SW, et al. 1986. Multiple sclerosis; gadolinium enhancement in MR imaging. Radiology 161: 721–725. Grossman RI, Lenkinski RF, Ramer KN, et al. 1992. MR proton spectroscopy in multiple sclerosis. Am J Neuroradiol 13: 1535–1543. Gupta RK, Jena A, Sharma A, Guha DK, Khushu S, Gupta AK. 1988. MR imaging of intracranial tuberculomas. J Comput Assist Tomogr 12(2): 280–285. Gupta RK, Lufkin RB. 2001. Viral infections. In MR Imaging and Spectroscopy of Central Nervous System infection (Eds., Gupta RK, Lufkin RB), Kluwer Academic/ Plenum Publishers, New York, pp. 147–175. Gupta RV, Pandey R, Khan EM, et al. 1993. Intracranial tuberculoma: MRI signal intensity correlation with histopathology and localized MR spectroscopy. Magn Reson Imaging 11: 443–449. Haimes AB, Zimmerman RD, Morgello S, Weingarten K, Becker RD, Jennis R, Deck MDF. 1989. MR imaging of brain abscesses. Am J Neuroradiol 10: 279–291; Am J Roentgenol 152: 615–622. Harris TM, Edwards MK. 1991. Meningitis. Neuroimaging Clin N Am 1: 39–56. Hartmann M, Jansen O, Heiland S, Sommer C, Munkel K, Sartor K. 2001. Restricted diffusion within ring enhancement is not pathognomonic for brain abscess. Am J Neuroradiol 22: 1738–1742. Iglesias-Rozas JR, Bantz B. Adler T. 1991. Cerebral lymphoma in AIDS: clinical, radiological, neuropathological and immunopathological study. Clin Neuropathol 10: 65–72. Jackson JA, Leake DR, Schneiders NJ, et al. 1985. Magnetic resonance imaging in multiple sclerosis; results in 32 cases. Am J Neuroradiol 6: 171–176.

377

378

Robert D. Zimmerman

Jarvik JG, Lenkinski RE, Grossman RI, et al. 1993. Proton MR spectroscopy of HIV-infected patients: characterization of abnormalities with imaging and clinical correlation. Radiology 186: 739–744. Jordon J, Enzmann DR. 1991. Encephalitis. Neuroimaging Clin N Am 1: 17–38. Kepes JJ. 1993. Large focal tumor-like demyelinating lesion of the brain; intermediate entity between multiple sclerosis and acute disseminated encephalomyelitis? A study of 31 patients. Ann Neurol 33: 18–27. Kim SH, Chang KH, Song IC, Han MH, Kim KC, Kang HS, Han MS. 1998. Brain abscess and brain tumor: discrimination with in vivo H-1 MR spectroscopy. Radiology 204: 239–245. Levy RM, Rosenbloom S, Perrett LV. 1986. Neuroradiologic findings in AIDS: a review of 200 cases. Am J Roentgenol 147: 977–983. Mathews VP, Alo PL, Glass JD, et al. 1992. AIDS related CNS cryptococcosis: radiologic–pathologic correlation. Am J Neuroradiol 13: 1477–1486. Menon DK, Sargentoni J, Peden CJ, et al. 1990. Proton MR spectroscopy in herpes simplex encephalitis: assessment of neuronal loss. J Comput Assist Tomogr 14: 449–452. Milton WJ, Atlas SW, Lai E, Mollman JE. 1991. MRI findings in Creutzfeldt–Jakob disease. Ann Neurol 29: 439–440. Nusbaum AO, Kar-Ming F, Atlas SW. 2002. White matter and inherited metabolic disorders in magnetic resonance imaging of the brain and spine. 3rd edn. Lippincott Williams and Wilkins, Philadelphia PA, pp. 457–563. Nusbaum AO, Lu D, Tang CY, Atlas SW. 2000a. Quantitative diffusion measurements in focal multiple sclerosis lesions: correlations with appearance on T1-weighted MR images. Am J Roentgenol 175: 821–825. Nusbaum AO, Tang CY, Wei TC, Buchsbaum MS, Atlas SW. 2000b. Whole-brain diffusion MR histograms differ between MS subtypes. Neurology 54: 1421–1426. O’Malley JP, Ziessman HA, Kumar PM, et al. 1994. Diagnosis of intracranial lymphoma in patients with AIDS. Value of 201-T single photon emission tomography. Am J Roentgenol 163: 417–421. Post MJD, Chen JC, Hensley GT. 1983. Toxoplasma encephalitis in Haitian adults with aquired immune deficiency syndrome: a clinical, pathologic–CT correlation. Am J Roentgenol 140: 8861–8868. Post MJD, Tate LG, Quencer RM, et al. 1988. CT, MR and pathology in HIV encephalitis and meningitis. Am J Neuroradiol 9: 469–476. Powell T, Sussman JG, Davies-Jones GAB. 1992. MR imaging in acute multiple sclerosis; ring-like appearance in plaques suggesting the presence of paramagnetic free radicals. Am J Neuroradiol 13: 1469–1475.

Prusiner SB. 1997. Prions and neurodegenerative disease. N Eng J Med 317: 1571–1581. Ragin AB, Storey P, Cohen BP, et al. 2004. Whole brain diffusion tensor imaging in HIV-associated cognitive impairment. Am J Neuroradiol 25: 195–200. Remy C, Grand S, Lai ES, et al. 1995. 1H MRS of human abscesses in vivo and in vitro. Magn Reson Med 508–514. Richardson EP. 1988. Progressive multifocal leukoencephalopathy. 30 years experience. N Engl J Med 318: 315–317. Sener RN. 2001. Herpes simplex encephalitis: diffusion MR imaging findings. Comput Med Imaging Graph 25: 391–397. Shen WC, Chice HH, Chiu KC, Tsai CH. 1999. MR imaging findings in Enterovirus encephalo-myelitis: an outbreak in Taiwan. Am J Neuroradiol 20: 1889–1895. Simone I, Federico F, Tortorella C, et al. 1998. Localised 1H-MR spectroscopy for metabolic characterisation of diffuse and focal brain lesions in patients infected with HIV. J Neurol Neurosur Psychiatr 64: 516–523. Smirniotopoulos JG, Koeller KK, Nelson AM, Murphy FM. 1997. Neuro-imaging – autopsy correlation in AIDS. Neuroimaging Clin N Am 7: 615–637. Sze G, Zimmerman RD. 1988. The magnetic resonance imaging of infections and inflammatory diseases. Radiol Clin N Am 26(4): 839–859. Teitelbaum GP, Otto RJ, Lin M, et al. 1989. MR imaging of neurocysticercosis. Am J Neuroradiol 10: 709–718. Tenser RB. 1984. Herpes simplex and Herpes Zoster: nervous system involvement. Neurol Clin N Am 2: 215–240. Theur CP, Hopewell PC, Elias D, et al. 1990. Human immune deficiency virus infection in tuberculosis patients. J Infec Dis 162: 8–12. Tien RD, Feldberg GJ, Osumi AK. 1983. Herpes virus infections of the CNS: MR findings. Am J Roentgenol 161: 167–176. Tolly TL, Wells RG, Sty JR. 1989. MR features of fleeting CNS lesions in association with Epstein–Barr infection. J Comput Tomogr 13: 665–668. Tsuchiya K, Katase S, Yoshino A, Hachiya J. 1999. Diffusionweighted MR imaging of encephalitis. Am J Rontgenol 173: 1097–1099. Tung GA, Evangelista P, Rogg JM, Duncan JA. 2001. Diffusionweighted MR imaging of rim-enhancing brain masses: is markedly decreased water diffusion specific for brain abscess? Am J Roentgenol 177: 709–712. Ulmer JL, Elster AD. 1991. Sarcoidosis of the central nervous system. Neuroimaging Clin N Am 1: 141–150. Ulug AM, Moore DF, Bojko AS, Zimmerman RD. 1999. Clinical use of diffusion tensor imaging for diseases causing neuronal and axonal damage. Am J Neuroradiol 20: 1044–1048.

Physiological imaging in infection, inflammation and demyelination: overview

Wasserstrom R, Mamourian AC, McGary CT, Miller G. 1992. Bulbar poliomyelitis: MRI findings with pathologic correlation. Am J Roentgenol 13: 371–373. Weingarten K, Zimmerman RD, Becker RD, Heier LA, Haimes AB, Deck MDF. 1989. Magnetic resonance imaging of subdural and epidural empyemas. Am J Neuroradiol 10: 81–87; Am J Roentgenol 152: 615–622. Whiteman MLH, Post MJD, Berger JR, Tate LG, Bell MD, Limonte LP. 1993. Progressive multifocal leukoencephalopathy in 417 HIV-seropositive pateints: neuroimaging with clinical and pathologic correlation. Radiology 187: 233–240.

Whiteman ML, Bowen BC, Post MJD, Bell MD. 1996. Intracranial infection. In Magnetic Resonance Imaging of the Brain and Spine, 2nd edn. (Ed., Atlas SW), LippincottRaven, Philadelphia PA, 707–772. Zagzag D, Miller DC, Kleinman GM, et al. 1993. Demyelinating disease versus tumor in surgical neuropathology. Clues to a correct pathologic diagnosis. Ann J Surg Pathol 17: 537–545. Zimmerman RD, Leeds NE, Danziger A. 1984. Subdural empyema: CT findings. Radiology 150: 417–422. Zimmerman RD, Weingarten KW. 1991. Neuroimaging of cerebral abscess. Neuroimaging Clin N Am 1: 1–16.

379

23

MR spectroscopy in intracranial infection Monika Garg and Rakesh K. Gupta SGPGIMS Campus, Lucknow, India

Key points • The true utility of MR spectroscopy (MRS) in intracranial infection is unknown, mainly due to the limited available literature. • The presence of amino acid signal (0.9 ppm) (consisting of cytosolic amino acids namely valine, leucine, and isoleucine) aids the differentiation of pyogenic abscesses from tumor. • Tuberculous abscesses show only lactate and lipid signals (at 0.9 and 1.3 ppm), without any evidence of cytosolic amino acids (present in pyogenic abscesses).

Introduction Infection of the central nervous system (CNS) can be life threatening and is a consequence of an encounter of potentially pathogenic microorganism with a susceptible human host (Evans, 1990). Early diagnosis is necessary for optimal treatment. Routine diagnostic techniques involve culture of various clinical samples and immunological tests, which may be invasive, time consuming and may delay definitive management. Non-invasive imaging modalities such as CT and MR imaging (MRI) have established themselves in the diagnosis of various CNS diseases; MRI offering greater inherent sensitivity, specificity, and multiplanar capability. MRS provides additive metabolic information as an adjunct to MRI. Although MRS has established itself as a noninvasive diagnostic technique for investigating intracranial lesions, its applications in different types of intracranial infections are less well known. This is 380

due to the fact that there are very few studies available on intracranial infections where MRS is correlated with MRI. Here we review studies of the role of MRS in intracranial infections.

Pyogenic infections Intracranial bacterial infection may manifest as a parenchymal abscess, meningitis and/or an extra axial empyema. These infections may result from direct trauma, surgery or by hematogenous spread from an extracranial source of infection. Meningitis Pyogenic meningitis has been defined as an inflammatory response to bacterial infection of the pia– arachnoid matter and the cerebrospinal fluid (CSF) of the subarachnoid space. Two to three decades ago, Hemophilus influenzae was considered as the chief offending agent involved in bacterial meningitis. However, following widespread H. influenzae type b vaccination, Streptococcus pneumoniae meningitis has taken its place. Other causative bacteria include Neisseria meningitidis, various streptococcal, staphylococcal, and Listeria species (Durand et al., 1993). The diagnosis of meningitis is usually made by CSF analysis using cytology, biochemistry, and culture. MR is generally superior to CT in demonstrating the distention of the subarachnoid space, which is reported to be an early finding in severe meningitis (Zimmerman et al., 1987). Runge et al. while working on experimental animals, have mentioned that

MR spectroscopy in intracranial infection

(a)

(b)

(c)

2

5 4

1

3

8

6

4 ppm

2

0

Fig. 23.1 Pyogenic meningitis (a–c). T2-weighted axial image (a) at the level of lateral ventricles shows small periventricular hyperintensities. Post-contrast T1-weighted magnetization transfer (MT) image (b) reveals the meningeal enhancement. Ex vivo proton MRS of CSF from the meningitis patient shows large amount of amino acids (1), lactate (Lac) (2), alanine (3), acetate (4), and acetoacetate (5).

the inflamed meninges may remain imperceptible on pre-contrast T1-weighted images (Runge et al., 1995). However, abnormal meningeal enhancement on triple dose post-contrast T1-weighted images appears to correlate with the number of inflammatory cells in CSF T1-weighted magnetization transfer (MT). MRI appears better than conventional T1-weighted images; similar enhancement being achieved with a single dose of contrast material. The main complications associated with meningitis including stroke, subdural collections, hydrocephalus, cerebritis, edema, and ventriculitis, all of which can be identified on MRI. Fluid-attenuated inversion recovery (FLAIR) imaging does not appear to provide significant additional diagnostic information in these patients (Tsuchiya et al., 1997). MRS has not yet been shown to be of much diagnostic value in patients with meningitis. Earlier, in vivo phosphorus-31 (31P) spectroscopy has been performed in one case of pyogenic meningitis (Matthews et al., 1989). This study reported normal phosphate metabolites concentrations and intracellular pH despite increased lactate (Lac) concentration. In vivo proton MRS in meningitis has reported normal levels of brain N-acetyl aspartate (NAA), creatine (Cr), choline (Cho), and inositol levels with mild levation of Lac (Shawl, 1995). Ex vivo MRS of CSF performed in proven cases of pyogenic meningitis has

reported the signals of cytosolic amino acids (0.9 ppm), Lac (1.33 ppm), alanine (1.47 ppm), acetate (1.92 ppm), and acetoacetate (2.24 ppm) along with reduced glucose levels (Venkatesh and Gupta, 2001) (Figure 23.1). Anaerobic bacterial metabolism is likely to explain increased glucose consumption and consequently lactic acidosis in the CSF. Abscess Brain abscesses are characteristically defined as a focal suppurative process within the brain parenchyma (Habib and Mozaffar, 2001). The incidence of brain abscess has remained relatively stable in the antibiotic era and its frequency varies from 1% to 2% and 8% of total intracranial lesions in developed and developing countries, respectively (Osenbach and Loftus, 1992). The abscesses result from the progression of cerebritis, the preliminary focus of infection. The early cerebritis stage (1–3 days) evolves into the late cerebritis stage (4–9 days), which is followed by early encapsulation (10–13 days) and late capsule stage (days 14 onwards), the well-developed abscess (Britt et al., 1981). The causative organisms are quite variable, and may consist of mixed cultures: aerobes, anaerobes, facultative anaerobes, and facultative anaerobes in combination with aerobes/anaerobes. Pus culture is considered

381

382

Monika Garg and Rakesh K. Gupta

the reference standard for identification of its causative agents. The fully developed mature abscess with central liquefactive necrosis appears as hypointense on T1weighted images and hyperintense on T2-weighted images. The abscess rim appears iso- or slightly hyperintense on T1-weighted and hypointense on T2-weighted images, and shows rim enhancement on post-contrast T1-weighted images (Venkatesh and Gupta, 2001). It is not always possible to differentiate pyogenic abscesses from other cystic intracranial mass lesions including tuberculous abscesses and neoplasms solely on the basis of MR features. In the absence of signs or symptoms of an extracranial infective focus, glioblastoma multiforme (GBM) is always considered in the differential diagnosis of brain abscess (Kim et al., 1997). Combined use of MT-MRI along with spectroscopy helps in discrimination of pyogenic from tuberculous abscesses (Gupta et al., 2001a). Recently, FLAIR has been reported to be better than the conventional T2-weighted images in depicting the hypointense capsule. However, post-contrast T1-weighted spin echo (SE) images show the enhancing capsule better than FLAIR itself (Tsuchiya et al., 1997). In the presence of hemorrhage in the abscess wall, a feature more commonly observed in cystic neoplasms, its differentiation on imaging gets complicated further (Sudhakar et al., 2001). In such situation, in vivo MRS has been found to be helpful in delineation the pathology more precisely. Recently, diffusionweighted imaging (DWI) has been found to be promising in differentiation of tumor from abscess as the latter shows restricted diffusion (low apparent diffusion coefficient (ADC)) while the former shows no restriction of diffusion (Ebisu et al., 1996; Holtas et al., 2000). However, later reports have shown restricted diffusion in cystic metastases and high-diffusion values in brain abscess that are being followed on treatment (Holtas et al., 2000; Mikami et al., 2002). In a recent study, in vivo MRS has helped in making the definitive diagnosis of pyogenic abscesses among the various lesions with comparable imaging features (Shukla-Dave et al., 2001). The spectral pattern permits the differentiation of pyogenic abscesses from tumors due to the presence of amino acid signal (0.9 ppm) consisting of cytosolic amino

acids namely valine, leucine, and isoleucine in the former (Kim et al., 1997). The amino acids persist in varied concentration even or while on medical therapy, and when sterile. The appearance of cytosolic amino acids in MR spectrum has been explained by the fact that pyogenic abscesses contain large amounts of neutrophils and proteins. The cytolysis of neutrophils results in the release of proteolytic enzymes that hydrolyze proteins into amino acids and thus contribute to MR spectrum (Gupta, 2001). Mature abscesses not undergoing antibiotic treatment present a variety of spectral pattern depending upon the type of bacteria residing in the abscess cavity (Garg et al., 2004). The abscesses with obligate aerobes (Nocardia asteroides and Pseudomonas aeruginosa) show peaks of cytosolic amino acids, Lac, alanine, and glycine (3.56 ppm) along with mobile lipid resonances at various chemical shifts (0.9, 1.3, 1.54, 2.02, and 2.24 ppm) (Figure 23.2). The in vivo spectra of abscesses containing only anaerobic bacteria (Bacteroides fragilis group and Peptostreptococcus sp.) are comparatively metabolically richer. In addition to the above metabolites, the spectra show acetate (1.92 ppm) with or without succinate (2.4 ppm) (Figure 23.3). When, both signals are present, acetate seems to be present always in higher concentration than succinate. The spectra taken from the abscesses containing facultative anaerobes (such as Staphylococcus aureus, Escherichia coli, Streptococcus mirabilis, Klebsiella pneumoniae, Enterococcus fecalis, Proteus mirabilis, Streptococcus intermedius, etc.) show metabolite patterns similar to those of aerobic abscesses. Facultative anaerobes when mixed with aerobes in abscess give a spectral pattern typical of aerobic bacteria. However, these in combination with anaerobes, give the metabolite pattern of anaerobic abscesses. The metabolism of facultative anaerobes is by definition flexible, i.e. they are not restricted to their mode of metabolism. In the presence of oxygen, they follow the aerobic pathways for energy similar to aerobic bacteria; however, in oxygen deficient conditions, they pursue anaerobic pathways. The glycolytic pathway, whereby glucose is broken down to two molecules of pyruvate, is universal to all groups of bacteria whether aerobic or anaerobic. Pyruvate formed during glycolysis may then

MR spectroscopy in intracranial infection

(a)

383

(c)

(b)

AA Ala Lac

4

3.5

3

2.5

2

1.5

1

Frequency (ppm)

(d)

TSP

Gly Glx

Ala

Lac

AA

Lac

Gly Ile

Lac  L Val

AA  L

3.65 3.60 ppm Lac Gly Lys L TT

4.0

3.5

3.0

L L

2.5

2.0 ppm

Ala L

1.5

1.0

0.5

TSP

Fig. 23.2 MR image and MRS of the patient with frontal lobe abscess secondary to infection of obligate aerobe P. aeruginosa (a–d). T2-weighted axial MRI (a) shows a well-defined large abscess in the right frontal lobe as a hyperintense core with peripheral hypointense rim, perifocal edema, and mass effect. This image also shows the location of the rectangular voxel placed for MRS. Post-intravenous contrast T1-weighted image (b) shows rim enhancement. In vivo MRS obtained from spin echo (SE) (135 ms) sequence (c) shows inversion of the cytosolic amino acids (AA, 0.9 ppm), Lac, 1.33 ppm, and alanine (Ala, 1.47 ppm) peaks. Ex vivo 1D MRS (d) from the pus using single pulse experiment (lower) shows TSP as a reference, AA, Lac, Ala, lipid (Lip) peaks (L (0.9, 1.3, 1.54, 2.02, 2.24, and 2.8 ppm)), glutamate (Glu)  glutamine (Gln)(Glx, 2.09–2.36 ppm), lysine (Lys, 3.01), taurine (T, 3.26 and 3.42 ppm), and glycine (Gly, 3.56 ppm) signals. Inset illustrates the expanded region highlighting valine (Val, 3.62 ppm) and isoleucine (Ile, 3.67 ppm) signals. The upper SE 160 spectrum shows inversion of AA, Lac, and Ala peaks along with the suppression of Lip components.

0.5

Monika Garg and Rakesh K. Gupta

S

(b)

(a)

Ace

Gly TSP

Ace

Glx

T T

S Lac

AA Ala Ace

Lac

Ala AA Ile Val

Lac

Gly S

3.7

Lac S

4

Lac  L AA  L

3.6 ppm

Ace

Ala AA

3.5 3 2.5 2 1.5 1 Frequency (ppm)

Lac

0.5

4.0

Gly Lys TT 3.5

3.0

Glx

2.5 2.0 ppm

TSP

1.5

1.0

0.5

(c)

TSP

0.5

L

1.0

L

1.5 Ile Leu

2.0 ppm

384

Val

2.5

Glx Lys

3.0 3.5 4.0

T

Ala Lac

4.5 5.0 5.0 4.5 4.0 3.5 3.0 2.5 2.0 1.5 1.0 0.5 ppm

Fig. 23.3 MRI and proton MRS of the patient with cerebellar abscess secondary to obligate anaerobe B. fragilis group infection (a–c). In vivo proton MRS (a) using STEAM (lower) demonstrates AA, Lac, acetate (Ace, 1.92 ppm), and succinate (S, 2.4 ppm) signals. On SE 135 (upper), AA, Lac, and Ala signals show phase reversal suggestive of J-coupling. Ex vivo one-dimensional (1D) proton MRS (b) with single pulse experiment of the pus (lower) shows trimethyl lysyl sodium propionate (TSP) (0.0 ppm) as an external reference, L (0.9 and 1.3 ppm), Glx, Lys, T, and Gly signals in addition to in vivo signals. Inset shows the expanded region from 3.5 to 3.75 ppm highlighting the signals of Val and Ile. On SE 160 spectrum (upper) AA, Lac, and Ala signals have inverted. Leucine (Leu, 1.7 ppm) along with aforesaid metabolites seen in 1D spectra are clearly assigned on ex vivo two-dimensional COSY of the pus (c).

MR spectroscopy in intracranial infection

follow a variety of metabolic pathways. Part of the pyruvate is reduced to Lac. In aerobic metabolism, the two molecules of pyruvate formed during glycolysis enter the tricarboxylic acid (TCA) cycle and, ultimately form carbon dioxide, water, and stored energy in the form of adenosine triphosphate (ATP). Although succinate is an intermediate of TCA cycle, its steady-state concentration is negligible and it is not, therefore, not seen in aerobic abscesses (Kim et al., 1997). In contrast, in anaerobic bacterial abscesses, the pyruvate undergoes anaerobic fermentation, and is a carboxylated reaction to oxaloacetate and eventually transforms to malate. The malate is metabolized in mitochondria to form acetate and succinate. MRS helps to distinguish infective from noninfective lesions and gives some information on the type of infective agent which can guide initial antibiotic therapy, before diagnostic cultures are available.

Tuberculous infections Tuberculosis (TB) of the CNS is always secondary to TB elsewhere in the body and is caused by Mycobacterium tuberculosis (Tandon and Pathak, 1973). The incidence of CNS TB has increased due to its association with the acquired immuno-deficiency syndrome (AIDS) (Hopewell, 1994). TB causes a granulomatous inflammatory reaction, which may involve the meninges causing TB meningitis and or brain parenchyma causing tuberculoma or tuberculous abscesses. Meningitis Tuberculous meningitis is the most frequent manifestation of TB in the CNS. The incidence of tuberculous meningitis in a given community is proportional to the prevalence of tuberculous infection in general, which in turn depends largely on the socio-economic factors (Tandon and Pathak, 1973). The clinical diagnosis of TB meningitis may be difficult. Diagnosis is dependent on CSF cytology and biochemistry, detection of acid-fast bacilli (AFB) in smear and culture; however only 8–30% of cases

show positive result on smear and culture. Moreover, because of the relatively low sensitivity of laboratory tests, non-invasive imaging plays an important role in diagnosis. MRI is significantly more sensitive than CT in the detection of the meningeal disease and evaluation of its complications (Gupta, 2002). The imaging features in tuberculous meningitis largely depend upon the stage at which MRI is performed. MRI in tuberculous meningitis may be normal in the early stages of disease especially on non-contrast studies. Gupta et al. have reported that MT is superior to conventional SE sequences for imaging the abnormal meninges, which are seen as hyperintense on precontrast T1-weighted MT images and enhance further on post-contrast T1-weighted MT images (Gupta et al., 1999). In addition, quantification of MT ratio (MTR) helps in predicting the meningitis with exact etiology that is reported around 19.49  1.22 (Gupta et al., 1999; Gupta, 2002). In our experience, visibility of the inflamed meninges on pre-contrast T1weighted MT images with low MTR is very specific of tuberculous meningitis and differentiates it from other non-tuberculous chronic meningitis. The tuberculous bacteria remain laden with high lipid content that is probably responsible for the low MTR in tubercular meningitis. As with pyogenic meningitis, FLAIR imaging does not appear to provide additional diagnostic information (Tsuchiya et al., 1997). Though there is no published study of in vivo MRS in tuberculous meningitis, ex vivo spectroscopy of CSF has been attempted in this context (Gupta, 2001). High-resolution ex vivo MRS of CSF showed the Lac, acetate, and sugars along with the signals from cyclopropyl rings ( 0.5 to  0.5 ppm) and phenolic glycolipids (7.1 and 7.4 ppm); these have not been observed with pyogenic meningitis (Figure 23.4). The combination of ex vivo MRS with MRI (possibly MT imaging (MTI)) may be of value in the diagnosis of tuberculous meningitis. Tuberculoma Tuberculoma is a space-occupying mass of granulomatous tissue (Tandon and Pathak, 1973). It arises from hematogenous spread from the primary infectious site; the commonest primary infection sites are

385

386

Monika Garg and Rakesh K. Gupta

(a)

(b)

(c) 1

2

9

8

7

6

5

4 3 ppm

2

1

0

Fig. 23.4 Tuberculous meningitis (a–c). T2-weighted axial image (a) shows dilated ventricular system. Post-contrast T1-weighted MT image (b) shows thick basal meningeal enhancement. Ex vivo proton MR spectrum (c) of CSF shows large resonances from Lac (1), acetate (2), and sugars. Note the resonances at 0.1 ppm (arrowhead) assigned to cyclopropyl rings characteristic of tuberculous meningitis and at 7.1 ppm assigned to phenolic glycolipids. These resonances are characteristically observed in CSF obtained from patients with tuberculous meningitis.

lungs and lymph nodes. The definitive diagnosis is necessary, as the majority of tuberculomas respond to medical management. MRI has greater sensitivity and specificity than CT in this regard, but both are still limited. On MRI, a tuberculoma’s appearance varies depending upon its stage of maturation, i.e. whether non-caseating, caseating with solid center, or caseating with a liquid center (Gupta et al., 1993). The non-caseating tuberculoma usually appears as hyperintense on T2-weighted and slightly hypointense on T1-weighted images; metastases, lymphoma, and other infective granulomas also have similar imaging features (Gupta et al., 2001b). On MTI, the cellular components of the lesions appear brighter and relatively specific for the disease. In addition, lesion conspicuity is greater on T1-weighted MTI compared to conventional SE imaging, and thus may help to assess disease load. The solid caseating tuberculoma appears iso- to hypointense on both T1- and T2-weighted images. This T2 hypointense appearing solid caseation often overlaps with imaging features of lymphoma, glioblastoma, fungal, and cysticercus granulomas. On T1-weighted MT images, the solid center appears hypointense with hyperintense rim. The calculated MTR from the rim and core are 23.8  1.76 and 24.2  3.1, respectively (Gupta et al., 1999; Gupta, 2002). The

significantly lower MTR of the T2 hypointense tuberculoma than cysticercus granuloma helps in its differential diagnosis. When the solid center of caseating lesion liquefies, the center appears hyperintense with a hypointense rim on T2-weighted images. On T1- and T2-weighted MT images, the rim appears hyperintense and undergoes contrast enhancement on post-contrast study. The lower MTR in different stages of tuberculoma is because of high lipid content present in tuberculous bacteria. In vivo MRS may help in differentiation of the tuberculomas with solid caseation from other nontuberculous lesions. In vivo, ex vivo, and in vitro MRS has been performed to fingerprint the metabolites of M. tuberculosis in tuberculomas (Gupta et al., 1996). In vivo MRS with stimulated echo acquisition mode (STEAM) sequence shows only lipid resonances at 0.9, 1.3, 2.0, 2.8, and 3.7 ppm corresponding to terminal methyl group [ˆ(CH3)], methylene group ˆ(CH2)n, CH2¨CH of fatty acyl chain, ¨CHˆCH2CH¨ of fatty acyl chain, and phosphoserine, respectively (Figure 23.5). On SE sequences, 0.9 and 1.3 ppm peaks show mark reductions in peak intensities while rest of the lipid signals are barely visible. Ex vivo MRS of the excised tuberculomas confirms the resonances seen in vivo. In addition to the signals

MR spectroscopy in intracranial infection

(b)

(c)

(a) 2 AA 9 2

1

1 11 5

4

10

3

3.5 2.5 1.5 0.5 0.5 Chemical shift (ppm) (d)

6

7

8

5

4 ppm

5

6

3

4

3

2

1

(e)

1 CS 2

Ala Thr

4

ppm

3 SS

5

Ace AA

6 Gua

6

5

4 ppm

3

2

1

Lys

Ly Glx B Leu/Val 7

6

5

4 3 ppm

2

1

Fig. 23.5 In vivo, ex vivo, and in vitro spectroscopy of an intracranial tuberculoma (a–e). T2-weighted axial image (a) of the brain shows hypointense well-defined mass in the right occipital region with hyperintense perifocal edema. Inset shows the volume of interest where the in vivo proton MR spectrum was obtained. In vivo proton MR spectrum (b) using STEAM sequence from the tuberculoma shows different lipid resonances (1, ˆCH3; 2, ˆ(CH2)n; 3, H2C¨CH; 4, CHˆCH2CH¨; 5, phosphoserine). Ex vivo (c) single pulse spectrum (lower) shows lipid resonances (6, OCˆCH2CH2; 7, OCˆCH2; 8, Cho; 9, phospholipids; 10, triglyceride backbone; 11, olefinic group). Ex vivo SE spectrum (upper) shows acetoacetate (AA) at 2.29 ppm along with Lac and serine (Ser). Note the marked reduction or disappearance of various lipid resonances seen in (c). 2D COSY (d) spectrum shows the signals of alanine (Ala), threonine (Thr), and cholesterol (CS). In vitro (perchloric acid extract) (e) by single pulse (lower) and SE (upper) spectra of tuberculoma show leucine/valine (Leu/Val), -hydroxy butyrate (B), acetate (Ace), Glu  Gln (Glx), lysine (Lys), guanine (Gua), and adenine (arrowhead) signals.

387

388

Monika Garg and Rakesh K. Gupta

seen in vivo, lipids peaks at 1.58, 2.24, 3.22, 4.1, 4.29, and 5.3 ppm are also seen, corresponding to OCˆCH2ˆCH2 of fatty acyl chain, COCH2 of fatty acyl chain, ˆN(CH3)3 of Cho, and glycerol backbone of phospholipids and olefinic groups of lipids, respectively. On ex vivo and in vitro (lipid extract) spectroscopy, caseating tuberculomas returned signals attributed to cyclopropane rings (0.5 and 0.1 ppm) and phenolic glycolipids (7.1–7.4 ppm). These signals have also been reported from the lipid extracts of pure strain of M. tuberculosis. Presence of the phenolic lipids represents the biochemical fingerprint of M. tuberculosis in a granuloma. The phenolic glycolipids remain present in the virulent as well as non-virulent strains of tuberculous bacteria. One of the characteristic features of mycobacterium is the presence of a lipid rich cell wall that contributes lipid signals in tuberculomas and in spectra obtained from the pure cultures of mycobacterium. On perchloric acid extract of tuberculoma as well as cultured M. tuberculosis, serine could be demonstrated. Though in vivo spectroscopy is known to show only lipid in T2 hypointense tuberculoma, the lesion with variegated appearance shows Cho at 3.22 ppm along with lipid (Figure 23.6). Recently it has been shown that T1-weighted MTI is superior to MRS in characterization of such lesions. As these lesions show large amount of cellularity and minimal solid caseation, the cellular regions appear brighter on MTI and show large Cho on spectroscopy. Predominance of cellularity in such variegated tuberculoma is responsible for prominent Cho resonance and may cause difficulty in differentiation from neoplastic lesions. Tuberculous abscess Tuberculous brain abscesses are the result of infection of M. tuberculosis; these are relatively rare and form about 4–7% of the total CNS TB in developing countries. True tuberculous abscesses, according to the criteria of Whitener, show macroscopic evidence of abscess formation within the brain parenchyma; histological confirmation that the abscess wall is composed of vascular granulation tissue containing acute and chronic inflammatory cells and bacteriological proof of the tuberculous origin (Whitener, 1978).

Tuberculous abscesses on MRI generally appear as non-specific large, solitary, and frequently multiloculated ring-enhancing lesions with surrounding edema and mass effect (Farrar et al., 1997). Recently, calculation of MTR from the rim of the abscess has helped in the differential diagnosis of tuberculous abscess from the pyogenic abscesses (Gupta et al., 2001a). The rim of tuberculous abscesses show significantly lower MTR values (19.89  1.55) than the rim of pyogenic abscesses (24.81  0.03) as the latter remain rich in protein content as compared to the former which is loaded with high lipid containing M. tuberculosis bacilli. In vivo MRS has also been used for the noninvasive differentiation of tuberculous abscesses from other lesions such as pyogenic abscesses, and fungal lesions, which may appear similar on structural MRI. The in vivo proton spectra from tuberculous abscesses show only Lac and lipid signals (at 0.9 and 1.3 ppm), without any evidence of cytosolic amino acids (Figure 23.7). On ex vivo spectroscopy, more lipid peaks, as observed in cases of tuberculoma, are also apparent. Amino acid signals therefore appear to discriminate pyogenic from tuberculous abscess (Gupta et al., 2001a).

Viral infections Viruses may cause CNS disease by causing primary encephalitis due to direct viral attack on neural tissue, and para-infectious or post-infectious encephalitis due to systemic viral infection, i.e. without evidence of direct viral invasion in the CNS (Cassady and Whitley, 1997). Viruses cause meningitis, encephalitis, meningoencephalitis, meningoencephalomyelitis, and myelitis. Patients with viral infection may remain clinically asymptomatic due to slow viral growth and latency.

Herpes simplex encephalitis Herpes simplex encephalitis (HSE), the most common type of encephalitis, is caused by both herpes simplex virus-1 and -2 (HSV-1 and -2) types. However, HSV-1 accounts for 95% of all the HSE cases while

MR spectroscopy in intracranial infection

(a)

(d)

(b)

(c)

(f)

(e)

Cho

Lip

Cr

3.5

3.0

2.5

2.0

1.5

1.0

0.5

Fig. 23.6 Large tuberculoma simulating neoplasm (a–f). T2-weighted axial image (a) shows a large left periventricular mass with perifocal edema and mass effect. It appears predominantly hyperintense with small hypointense foci, which is primarily hypointense with well-defined margin on T1-weighted image (b). The areas of T2 hyperintensity show peripheral rim-like subtle hyperintensity on T1-weighted image and demonstrate its pronouncement on T1-weighted MT image (c). Post-contrast T1-weighted MT image (d) shows heterogeneous enhancement of the mass. In vivo proton MRS (e) shows a high Cho and Lip resonance suggestive of a neoplastic lesion. Histopathology shows presence of giant cells, endothelial cells with small areas of caseation (f). Zeihl–Nelsen stain showed AFB consistent with M. tuberculosis within the lesion.

the majority (80–90%) of the neonatal encephalitis is due to the HSV-2 type (Tien et al., 1993). The clinical presentations though variable are non-specific to the disease. Furthermore, it is hard to culture HSVs in the CSF. Enzyme-linked immunosorbent assay (ELISA) tests used for antibody detection from serum and CSF do not always give reliable results, and may be positive for non-HSE cases also (Lakeman and Whitley,

1995). Biopsy of the brain tissue for herpes virus is 96% sensitive and 99% specific (Morawetz et al., 1983). Herpes infection characteristically involves the medial temporal and frontal lobes. MRI has been the modality of choice for the diagnosis of HSE and is characterized as the high signal intensity in the temporal and frontal lobe on conventional T2-weighted images (Gupta and Lufkin,

389

(a)

(b)

(c)

Lip

(e)

(d)

(f)

Lip/L Cho Lip

Lip

3.8 3.4 3.0 2.6 2.2 1.8 1.4 1.0 0.6 0.2 Chemical shift (ppm)

Lip

Lip

3.8 3.4 3.0 2.6 2.2 1.8 1.4 1.0 0.6 0.2 Chemical shift (ppm)

(g)

Lip/L

Lip

4.5 4.0 3.5 3.0 2.5 2.0 1.5 1.0 ppm Fig. 23.7 Multiple tuberculous abscesses (a–g). T2-weighted axial MRI (a) shows abscesses in the left temporal and parietal region along with mass effect on ventricular system and extensive edema. The lesions show hyperintense core and peripheral hypointense rim along with perifocal edema. On T1-weighted image (b), these lesions show hypointense core with hyperintense rim, more clearly visible on T1-weighted MT image (c). Post-contrast T1-weighted MT image (d) shows enhancement of the peripheral rim. In vivo proton MR spectrum (e) using STEAM with a voxel of 1.5 cm3 only Lip and Lac at 1.3 ppm are shown. On SE 135 spectrum (f), phase reversal along with reduction in signal is seen. The presence of Cho in this spectrum results from the large size of the voxel (2.0 cm3) selected on SE sequences that included the wall of the lesion. Ex vivo (g) single pulse (lower) and SE (upper) spectrum confirm the presence of Lip/Lac with no evidence of Cho.

MR spectroscopy in intracranial infection

(a)

(b)

Lac Cho Cr

3.8 3.4 3.0 2.6 2.2 1.8 1.4 1.0 0.8 0.4 Chemical shift (ppm) Fig. 23.8 HSE (a and b). T2-weighted axial image (a) shows hyperintensity in the right temporal lobe extending into the frontal region. In vivo proton MRS from the right temporal lobe (b) shows presence of large resonance of Lac along with altered (Cho/Cr) ratio.

2001), hemorrhagic changes may also be apparent. T1-weighted MT images show more extensive involvement of the brain parenchyma than shown by conventional imaging. In addition, post-contrast T1-weighted MT images help in delineating the meningeal component of the meningoencephalitis. DWI is found to be superior to conventional SE imaging in patients with encephalitis with respect to delineation of the lesion (Tsuchiya et al., 1999). FLAIR imaging has also been applied in the patients with HSE with better results than conventional imaging (Tsuchiya, 1997). There are few studies in the literature of MRS in HSE patients. Demaerel et al. and Menon et al. reported a decrease in NAA signal resulting in lowering of NAA/Cr ratio (Menon et al., 1990; Demaerel et al., 1992). The depletion of NAA content in the brain has been ascribed to neuronal loss associated with the disease process. Brain Lac has also been observed following herpes infection (Gupta and Lufkin, 2001) (Figure 23.8). Presence of Lac indicates the impairment of oxidative metabolism and/or macrophage activity. A longitudinal study performed in a case of HSE revealed that imaging returns to normal more rapidly than metabolites. The findings suggest that the imaging abnormalities

due to interstitial edema regress early while neuronal dysfunction improves over a longer period (Takanashi et al., 1997). The spectroscopic abnormalities are non-specific, and contribute little to the diagnostic accuracy of the MR examination; whether MRS adds prognostic value in established cases is not yet clear. Epstein–Barr virus infection Epstein–Barr virus (EBV), a member of the herpes family, is recognized as a causative agent of infectious mononucleosis (Tien et al., 1993). Only 5% of the patients with infectious mononucleosis show CNS complications (Silverstein et al., 1972). Laboratory investigation of CSF in patients with CNS complications usually shows increased lymphocytes and protein. The definite diagnosis is usually based on fourfold rise in EBV-specific IgM antibodies against viral capsid antigen in serum and positive polymerase chain reaction (PCR) for EBV DNA in CSF (Ross and Cohen, 1997). MRI shows symmetrical low- and high-intensity lesions of both basal ganglia (predominantly in the putamen) on T1- and T2-weighted images, respectively. Post-intravenous contrast medium, the lesions

391

392

Monika Garg and Rakesh K. Gupta

do not show enhancement (Gupta and Lufkin, 2001). FLAIR imaging performed in a case of EBV encephalitis also gives high signal in the involved basal ganglia. The cerebellum is also frequently involved. Nonspecific clinical presentation and the variable occurrence of the lesions in different brain locations makes the imaging diagnosis of EBV encephalitis difficult (Cecil et al., 2000). There is one study in which MRS has also been applied to study the biochemical characteristic of basal ganglia with EBV (Cecil et al., 2000). Reduced NAA to Cr ratio with concomitant elevation in excitatory amino acids, macromolecules, and myoinositol levels have been reported. However, the composite Cho to Cr ratio remains normal. Such non-specific spectroscopic findings have also been observed in other viral encephalitides. Subacute sclerosing panencephalitis Subacute sclerosing panencephalitis (SSPE) is a slowly progressing fatal inflammatory disease of the CNS resultant from infection of measles virus of genus Morbillivirus, a subgroup of paramyxoviruses (Parameshwaran and Radhakrishnan, 2002). SSPE is one of the various manifestations produced by the measles viruses due to direct attack of the virus in the setup of an altered immune status. Of the total patients having SSPE, 65–70% of the patients had the history of previous measles infection. In recent years, SSPE has shown marked reduction in its occurrence in the developed countries due to the adoption of proper vaccination strategies; however, it is still a significant problem for the developing countries due to oblivious activities towards vaccination. The diagnosis for SSPE is generally made by high antibody titer against measles virus along with mildly raised protein concentration and markedly high gamma globulin level in CSF and blood. CT and MRI findings in SSPE are usually non-specific. The lesions on conventional MRI appear hyperintense on T2-weighted images and hypointense on T1weighted images (Gupta and Lufkin, 2001). The lesions are usually located in the posterior parietal, temporal, and occipital regions, corona radiata, and subcortical and deep white matter (WM). MRI sometimes remains inconclusive and shows normal

imaging pattern even when the disease is at advanced stage on clinical examination. There is only a single published study of in vivo MRS in SSPE (Salvan et al., 1999). Spectra were obtained from imaging abnormal and normal appearing brain regions. The spectrum from the abnormal appearing brain region showed markedly decreased NAA, with slight increases in inositol and Cho, and Lac; Cr appeared normal. The spectrum from the region with normal imaging also showed increased inositol and Cho, but with normal NAA levels. Depletion of NAA suggests neuronal dysfunction in SSPE. The increase of Cho might be due to demyelination or due to inflammation, and increased inositol has been presumed to represent active gliosis. Though, the spectroscopic findings are non-specific, it gives additional information regarding the extent of the disease in regions which appear normal on standard imaging sequences. Japanese encephalitis Japanese encephalitis is the leading cause of acute encephalitis affecting children and adolescents in the tropics. Japanese encephalitis virus is transmitted in a zoonotic cycle among mosquitoes and vertebrates. Diagnosis of Japanese encephalitis is confirmed by showing a fourfold rise in the antibody titer in the blood samples (Ishii, 1967). Neuroimaging studies have revealed bilateral thalamic lesions along with the involvement of brain stem and basal ganglia. It is important to distinguish Japanese encephalitis from other types of encephalitis, particularly HSE, because antiviral therapy for HSE is very effective in the acute stage (Kumar et al., 1997). However, a patient presenting with encephalitic illness from an endemic zone of Japanese encephalitis with demonstration of lesions in the basal ganglia, thalami, and brainstem should help in differentiating Japanese encephalitis from atypical HSE. On MRI, the lesions are uniformly hyperintense on T2-weighted and iso- to hypointense on T1weighted images. The post-intravenous contrast study may show meningeal enhancement especially on T1-weighted MT images and appear more

MR spectroscopy in intracranial infection

(a)

(b)

3.8 3.4 3.0 2.6 2.2 1.8 1.4 1.0 0.6 0.2 Chemical shift (ppm) Fig. 23.9 Japanese encephalitis (a and b). T2-weighted axial image (a) shows evidence of hyperintensity in the medial parts of bilateral thalami, caudate nuclei, and frontal lobes. Singlevoxel in vivo proton MRS (b) from the frontal region using SE sequence shows presence of Lac at 1.33 ppm, Cr at 3.02 ppm, and Cho at 3.22 ppm with no visible NAA resonance.

pronounced than T1-weighted images. However, there is no relationship between the imaging and the patient’s prognosis. We have in vivo proton MRS data from patients with Japanese encephalitis (Gupta and Lufkin, 2001). It shows presence of Lac along with reduced Cr, Cho, and NAA signals (Figure 23.9). Such a spectral pattern is non-specific to Japanese encephalitis and can be observed in other infective and noninfective intracranial conditions. Para-infectious encephalopathy The two main forms of para-infectious encephalopathies are acute disseminated encephalomyelitis (ADEM) and acute necrotizing encephalopathy (ANE) (Harada et al., 2000). The most common virus associated with ADEM is measles, followed by rubella, chicken pox, EBV infection, and mumps. The changes of ADEM remain localized mainly in the WM, and the inflammation of myelin and associated demyelination are the major findings of the disease. MR of ADEM patients’ demonstrates disseminated T2-weighted hyperintensity in the deep and

subcortical WM of the cerebral hemispheres as well as in the cerebellar WM, midbrain, and brainstem (Caldemeyer et al., 1994; Harada et al., 2000; Gupta and Lufkin, 2001). The post-intravenous contrast images show variable enhancement of the lesion. On MRI, ANE lesions are characterized by hyperintensity in the bilateral thalamic regions on conventional T2-weighted images and DWI. The calculated ADC values from ADEM and ANE abnormal regions are much higher and slightly lower than that of normal matter, respectively. DWI in ADEM and ANE cases shows lesion regression on consecutives studies following therapy (Harada et al., 2000). It has been observed that ADEM patients recover faster on steroids with the ADC values falling rapidly to the normal values as compared to ANE patients. There is single study in which MRS was performed in patients with ADEM and ANE (Harada et al., 2000). Lac is present in both cases; however, ADEM shows (Figure 23.10) higher Lac level than ANE. Similarly, more marked reduction in NAA/Cr has been observed in ADEM as compared to ANE. The spectral findings in ADEM and ANE are nonspecific, and spectroscopy is not useful for their diagnosis. MRS may, however, prove useful for

393

394

Monika Garg and Rakesh K. Gupta

(a)

(b)

(c)

NAA

Cho Cr

5

4

3 2 Frequency (ppm)

Lac 1

0

Fig. 23.10 ADEM (a–c). T2-weighted axial MRI (a) shows bilateral multiple hyperintense lesions. DWI (b) shows the lesions with hypointense core with hyperintense rim. In vivo proton MR spectrum (c) using SE shows normal NAA, Cr, and Cho peaks along with Lac.

monitoring treatment response in ADEM and ANE patients reflected in Lac/Cr ratio.

Fungal infections Fungal infections of CNS are relatively rare in immunocompetent individuals, where they tend to be caused by pathogenic species such as Cryptococcus. Immunocompromised patients, however, are susceptible to invasive infection with a wider range of opportunist fungal pathogens, such as Aspergillus and Candida (Satishchandra and Sharma, 2002). Infection in the context of HIV is covered in Chapter 27. Fungal abscess Both Aspergillus and Candida species can produce cerebral abscesses; however in the current chapter we will discuss only the fungal abscess resulting from Aspergillus flavus. Aspergillus abscess The Aspergillus infection has been observed in immunocompromised patients. Two general patterns of infection are seen: (i) direct extension from the paranasal sinuses, eye, or middle ear leading to abscesses in frontal or temporal lobe and (ii) hematogenous spread causing formation of multiple small

abscesses at the junction of gray and WM (Sepkowitz and Armstrong, 1997). The CSF is usually non-specifically abnormal. Elevated opening pressure, decreased glucose, and elevated protein levels are generally observed. The presence of red blood cells in the CSF of a patient with brain abscess is suggestive of possible intracranial aspergillosis (Meyer et al., 1973). On MRI, the fungal abscess resembles pyogenic abscesses. The abscess core appears hyperintense with hypointense rim on T2-weighted images and hypointense core with iso- to minimally hyperintense rim on T1-weighted images. Post-intravenous contrast, the abscess rim shows contrast enhancement (Kathuria and Gupta, 2001). There are to date no published studies of in vivo proton MRS in fungal abscesses. Our own experience is of a case of A. flavus abscess in a patient with non-Hodgkin’s lymphoma. The in vivo spectrum showed cytosolic amino acids (valine, leucine, and isoleucine) and Lac along with multiple signals between 3.6 and 4.0 ppm (Figure 23.11). These signals have been assigned to trehalose sugar present in the fungal wall, confirmed with ex vivo highresolution spectroscopy of aspirated material. Such sugar signals have also been reported in cryptococcomas (Dzendrowskyi et al., 2000). The presence of characteristic sugar signals from the fungal wall on in vivo spectrum may help in diagnosis in such cases.

MR spectroscopy in intracranial infection

(b)

(a)

(c)

(d)

3.6 3.8 ppm

*

L/Lip

AA

Lac

AA/Lip *

3.6 3.8 ppm

3.8 3.4 3.0 2.6 2.2 1.8 1.4 1.0 0.6 0.2 Chemical shift (ppm)

6

5

4

3 ppm

2

1

0

Fig. 23.11 Fungal abscess secondary to the infection of A. flavus (a–d). T2-weighted coronal MRI (a) shows hyperintense core with hypointense rim along with perifocal edema in the left frontal lobe. Post-contrast T1-weighted MT image (b) shows rim enhancement of the abscess wall. In vivo SE 135 proton MR spectrum (c) from the core of the abscess shows the inverted signals of amino acids (AA) and Lac peaks. Besides this, multiple signal (*) present in the region of 3.6–3.8 ppm correspond to trehalose. Ex vivo high-resolution proton spectrum (d) from the pus using single pulse (lower) sequence reconfirms the amino acids (AA), Lac, and Lip signals along with the trehalose (*) signal. The upper SE spectrum shows the inverted signals of AA, Lac, and trehalose. Inset in the lower as well as in the upper spectrum shows trehalose more precisely (expansion from 3.6 to 4.0 ppm).

Mucormycosis Mucormycosis or phycomycosis is a fatal fungal infection of immunocompromised patients caused by Mucor species. Predisposing factors include diabetes mellitus with ketoacidosis, naso-orbital necrotizing infection, and meningoencephalitis (Sepkowitz and

Armstrong, 1997). CSF studies typically show the elevated opening pressure, elevated protein, and pleocytosis with at least 50% polymorphonuclear cells. MRI shows parenchymal abnormalities in the basal ganglia, thalamus, and midbrain. The lesions appear hypointense on T1-, T2-weighted and FLAIR

395

396

Monika Garg and Rakesh K. Gupta

(a)

(b)

(c) Cho

Cr

Lac

3.5 3.0 2.5 2.0 1.5 1.0 Chemical shift (ppm)

0.5

Fig. 23.12 Aspergillus granuloma (a–c). T2-weighted axial MRI (a) at the level of ventricles showing a heterogeneous intensity lesion with focal hypointense areas in the left parieto-occipital region. Post-contrast T1-weighted image (b) shows intense enhancement of the mass lesion. In vivo proton MR spectrum SE 135 (c) shows raised Cho, low Cr along with presence of Lac and marked decrease in NAA.

images with surrounding edema and contrast enhance on post-gadolinium studies. In vivo MRS performed in mucormycosis shows Lac, alanine, acetate, succinate, and lipid signals along with the signals of NAA, Cr, and Cho (Siegal et al., 2000). The appearance of normal brain metabolites in the spectrum is suggestive of brain parenchyma within the voxel. Amino acid signals at 0.9 ppm were not described in the paper, although the spectrum shown has amino acid signals and the spectrum resembles those of anaerobic pyogenic abscesses. An unassigned resonance at 3.8 ppm was also present; ex vivo high-resolution studies may allow this to be characterized, and provide a biochemical fingerprint for mucormycosis. Fungal granuloma Fungal granulomas are the space-occupying lesions caused by Aspergillus and Cryptococcus species and have been referred to as Aspergillus granuloma and A. cryptoccoma, respectively. Aspergillus granuloma On T2-weighted MRI, the aspergillus granuloma gives heterogenous pattern of intensity, i.e. hyperintense lesions mixed with hypointense foci. On T1-weighted images, the lesion is hypointense. The T1-weighted

MTI reveals the hyperintense lesion that undergo enhancement on post-contrast T1-weighted MT images. There is a single study available in the literature in which in vivo spectroscopy has been performed (Kathuria and Gupta, 2001). On spectroscopy, it showed high Cho, low Cr and Lac with no NAA (Figure 23.12). Since the spectrum is completely nonspecific for the aspergillus granuloma, it does not help in its differentiation from other neoplastic and non-neoplastic mass lesions. Cryptococcoma Cryptococcal brain lesions (cf. Figure 23.13) are covered in Chapter 27.

Parasitic infections In the following section we will discuss the role of MRI and MRS (in vivo and ex vivo) in the diagnosis of neurocysticercosis, echinococcosis (hydatid disease), and malaria. Neurocysticercosis Neurocysticercosis, caused by the metacestodes of tapeworm Taenia solium, is common in developing countries, and its frequency is also increasing in

MR spectroscopy in intracranial infection

(c)

(b)

4

2 ppm

0

1

5

Lipid

3

1

Cho

2 ppm

Cr

3

Tre/Lip

4

Tre

(a)

4

3 ppm

2

1

Fig. 23.13 Cryptococcoma (a–c). T2-weighted axial image (a) showing two hyperintense areas in the right frontal (inset) and occipital regions. In vivo proton MRS (b) at echo time (TE)  30 ms (upper) shows broad resonances at 1.3, 2.1, and 3.4 ppm. At TE  270 ms (lower), most of the resonances seen at TE  30 ms have disappeared suggesting metabolites with short T2 values, probably lipids. (Courtesy Drs. L. Chang and T. Ernst, Harbor University–University of California Los Angeles Medical Center, UCLA School of Medicine.) Ex vivo proton spectrum (c) of cryptococcomas tissue sample from rat model shows Lip peaks, Cr, and Cho signals along with signals of trehalose (Tre) at different chemical shift positions. (Courtesy Dr. Uwe Himmelerich, University of Sydney, Australia.)

developed countries, due to increased immigration and more frequent travel to endemic regions (Garg and Kar, 2002). The natural history of neurocysticercosis and its clinical course are poorly understood. It is thought that a high percentage of the population harboring cysticercus cysts in the brain is asymptomatic (Carpio and Hauser, 2002). Clinical diagnosis of neurocysticercosis is most frequently based on ELISA and enzyme-linked immunoelectrotransfer blot (EITB). The EITB is highly sensitive in patients with multiple, enhancing intracranial lesions; however, the sensitivity of this assay has been reported to be lower in cases with single lesion or calcified lesions (Garcia et al., 1997). For ELISA, some studies have demonstrated sensitivity of 87% with the specificity of 95% in CSF specimens and in addition to neurocysticercosis, it often gives false-positive results with sera from patients with other cestode infections, caused by Taenia saginata, Echinococcus granulosus, and Hymenolepis nana (Wilkins et al., 2002). Cysticercus cysts may remain viable for many years. Locations of cysts vary; they may be intracranial

(parenchymal, ventricular, subarachnoid (cisternal)) or spinal. During the course of development, cysticercus cysts in the brain go through a cystic or vesicular stage, the viable stage; colloidal or granular stage, the degenerating stage; and the chronic calcified stage. Symptoms depend on the number of cysts, their location in the brain, stage of development of cyst, and the intensity of host immune-inflammatory response (Gupta and Chang, 2001). CT is relatively insensitive and not highly specific (Sharma and Gupta, 1993). MR appearances vary with the different stages of degeneration of the cysts (Sharda et al., 2002). Cysts in the viable vesicular stage appear hyperintense on T2-weighted and hypointense on proton density (PD) and T1-weighted images; while the scolex is sometimes seen as an eccentrically placed nodule, hypointense on T2-weighted and hyperintense on PD and T1-weighted images without perifocal edema and contrast enhancement. The colloidal stage cysts appear as hyperintense with iso- to hypointense rim on T2-weighted images, and with perifocal edema. Often this stage undergoes rim enhancement on

397

398

Monika Garg and Rakesh K. Gupta

post-contrast study. On T1-weighted images, the cyst appears as a hypointense center with isointense periphery. In nodular granular stage, the imaging feature simulates with tuberculoma, small abscesses, and metastatic tumors and it appears isointense to the normal parenchyma on T1-weighted images and iso- to hypointense with or without surrounding edema. The lesion in nodular calcified stage appear iso- to hypointense on T1-weighted images and hypointense on T2-weighted images. The calcification on imaging can be studied best by the use of gradient echo (GE) sequence with corrected phase imaging (Chawla et al., 2002). On phase corrected GE, the calcification appears bright. MTR has also been calculated from different stages of neurocysticercosis (Kathuria et al., 1998). The visibility of a lesion on T1-weighted MT sequence depends on its MTR and its location in the cerebral hemisphere. Different stages of neurocysticercosis have different MTR values; maximum MTR has been calculated from healing lesions (mean  SD  31.0  2.8) and core of SE invisible lesions (30.0  5.1). FLAIR imaging performed in neurocysticercosis clearly depicts that this sequence does not provide additional information than conventional SE imaging (Tsuchiya et al., 1997). Quite often the diagnosis of neurocysticercosis becomes difficult in the absence of characteristic scolex. In vivo and ex vivo MRS has been performed in few studies of neurocysticercosis (Gupta and Chang, 2001; Pandit et al., 2001). Pandit et al. performed in vivo spectroscopy in neurocysticercosis and reported signals of cytosolic amino acids (valine, leucine, and isoleucine), Lac, alanine, succinate, NAA, Cr, and Cho (Pandit et al., 2001). The peaks of NAA, Cr, and Cho are assumed to be due to normal brain parenchyma within the voxel. However, Garg et al. using ex vivo MRS of aspirated fluid detected Cr in addition to the other metabolites. The presence of Cr in cysticercus cysts from humans and pigs helps in differentiating this from intracranial hydatid cysts (Garg et al., 2002a). We have observed that the fluid of degenerating cysticercus cysts (whether from humans or animals) are devoid of Cr (Figures 23.14 and 23.15), whereas that from viable vesicular cysts contain Cr along with the other metabolites (Figure 23.15). Histopathology has

proved the occurrence of muscle fibers in the bladder wall and scolex of cysticercus cyst and that these may contribute to Cr in the fluid; the degenerated cysts showed minimal musculature as compared to the vesicular cyst that had prominent musculature in the cyst wall and the scolex. In vivo spectroscopy may be of value in the large cysticercus cyst without visible scolex, where there is a large range of differential diagnosis including brain abscess and cystic metastases. In vivo MRS shows resonances of acetate, succinate (succinate  acetate), and Lac, the presence of Cr depending on whether the lesion is in the vesicular or colloid stage.

Echinococcosis (hydatid disease) Echinococcosis, caused by metacestodes of genus Echinococcus, is a zoonosis of worldwide distribution. The two common species E. granulosus and E. multilocularis cause cystic and alveolar echinococcosis, respectively. The endemic regions for cystic echinococcosis are the Middle East, India, South America, and Australia. The alveolar echinococcosis though relatively rare poses serious problem in colder areas namely Alaska, Central Europe, Turkey, Russia, and China. In the following section, we will discuss only the cystic echinococcosis caused be E. granulosus. In E. granulosus, the definitive hosts (with adult worms) are dogs, cats, and other canine species, while the intermediate hosts (with lesions due to larvae) are man, sheep, camels, and other domestic animals. The cysts can be localized in any organ of the body, such as the liver, lung, kidney, and brain, but the commonest site is the liver. Intracranial involvement is seen in only about 2% of all the hydatid cases reported even in endemic areas (Rudman and Khaffai, 1988). Almost all the symptoms of the hydatid disease are due to the pressure effect on surrounding structures, resulting from distension, obstruction, erosion, or infection. The primary hydatid cysts, formed by the direct implantation of larvae into neural parenchyma are usually fertile and contain brood capsules and scoleces; while secondary cysts, resulting from spontaneous or surgical rupture of primary cysts are sterile and lack brood capsules (Önal et al., 1997).

MR spectroscopy in intracranial infection

(c)

(a)

S Cho * Gly

(B)

Glx Ac AA

a-glu b-glu Ala

L

S

(b)

Suc

NAA Cho Gly

a-glu Lac

(A)

3.8

L

Cho Cr

3.4

3.0

2.6

2.2

1.8

1.4

1.0

0.6

0.2

Chemical shift (ppm)

9

8

7

6

5

4

Lip

* Ala AA Lip Lip Glx Ace Lip

3

2

1

ppm

Fig. 23.14 Human brain with degenerated cysticercus cysts (a–c). T2-weighted axial MRI (a) at the supraventricular region shows the degenerated cysticercus cyst as a hyperintense core with hypointense rim along with perifocal edema and has given the similar imaging feature as of abscess. In vivo proton MR spectrum SE 135 (b) shows the usual brain metabolites namely NAA, Cr, and Cho along with the Lac and succinate (Suc) peaks. Ex vivo high-resolution proton MR (c) single pulse (A) and SE (B) spectra show the signals of amino acids, Lip, Lac, alanine (Ala), acetate (Ace), Glu  Gln (Glx), Suc, Cho, glycine (Gly), -glucose (a-glu), and -glucose (b-glu). The asterisk (*) at 2.24 ppm is the contaminant due to acetone.

On MRI, it appears as well-defined cystic mass, hypointense on T1-weighted, hyperintense on T2-weighted images with or without edema, and may shows rim enhancement on post-contrast Gd-DTPA studies. It has been observed that MRI is superior to CT in detecting pericystic enhancement and edema (Nurchi et al., 1992). Where there is edema and enhancement, the hydatid cyst may resemble a pyogenic abscess.

In vivo MRS from intracranial hydatid cysts shows Lac, alanine, acetate, succinate, and glycine resonances (Kohli et al., 1995; Gupta and Chang, 2001) (Figure 23.16). 31P spectroscopy has also been performed in hydatid cyst (Novak et al., 1992). Ex vivo high-resolution MRS of the aspirated fluid from the cyst confirms the in vivo assignments. The other peaks observed were of cytosolic amino acids, -hydroxy butyrate (1.2 ppm), lipid peaks (0.9 and

399

400

Monika Garg and Rakesh K. Gupta

(a)

(b)

(c)

5

13 14

3

14

13

4

3 5

4

1

24

1

9 24 11 5

23

14

7

11 12

13 2

2

7

14 10

2

8

28

12

3

2 6

5.0 4.5 4.0 3.5 3.0 2.5 2.0 1.5 1.0 0.5 ppm

2 13 2

2

25

9

5.0 4.5 4.0 3.5 3.0 2.5 2.0 1.5 1.0 0.5 ppm

Fig. 23.15 Swine brain with live and degenerated cysticercus cysts (a–c). Conventional SE T2-weighted ex vivo image (a) of the swine brain demonstrates a hyperintense cyst with hypointense scolex in the right temporal lobe without perilesional edema (live cyst) and another cyst (arrow) in the left periventricular region along with perilesional edema (degenerated cyst). Ex vivo high-resolution (b) single pulse proton MR spectrum (lower) of fluid from the right temporal lobe cyst showing TSP (1), lipid (2), Amino acids (valine (Val), isoleucine (Ile), and leucine (Leu)) (3), Lac (4), alanine (5), lysine (6), acetate (7), Glu/Gln (8), succinate (9), Cr (10), Cho containing compounds (11), glycine (12), -glucose (13), -glucose (14). Ex vivo SE 160 spectrum (upper) shows phase reversal of 3, 4, 5, 13, and 14 peaks. Ex vivo (c) single pulse (lower) of cyst fluid from the left periventricular region of the swine’s brain shows 1, 2, 3, 4, 5, 6, 7, 8, 9, 11, 12, 13, and 14 peaks. Note the absence of Cr at 3.03 ppm. The SE (upper) spectrum shows phase reversal of 3, 4, 5, 13, and 14 peaks.

MR spectroscopy in intracranial infection

(b)

(a)

(c)

(d)

Suc

STEAM 20

TSP

Suc

SE 135

Lac Suc

Lac

Ace Ace

Ace Ala

Ala Lac 3.5 3.0 2.5 2.0 1.5 1.0 0.5 Chemical shift (ppm)

3.5 3.0 2.5 2.0 1.5 1.0 0.5 Chemical shift (ppm)

4.0 3.5 3.0 2.5 2.0 1.5 1.0 0.5 0.0 ppm

Fig. 23.16 Cerebral hydatid cyst (a–d). T2-weighted axial image (a) shows evidence of a well-defined, rounded hyperintense area in the right parieto-occipital region with mass effect. On post-intravenous contrast T1-weighted imaging (b) the cyst does not enhance and shows signal intensity similar to CSF. In vivo proton MRS (c) with STEAM 20 (left) from the inset in (a) shows Lac at 1.33 ppm, acetate (Ace) at 1.92 ppm, and succinate (Suc) at 2.4 ppm; at SE 135 (right) Lac and alanine (Ala) at 1.5 ppm show phase reversal while Ace and Suc show normal phase. Ex vivo study (d) of the cyst fluid confirms the assignments seen in vivo.

1.33 ppm), Cho, -glucose (5.23 ppm) and -glucose (4.64 ppm). In vivo, brain abscesses have prominent cytosolic amino acid resonances, and the acetate (A) to succinate (S) ratio always more than one (A/S  1); succinate may be absent. In contrast, in hydatid cysts

cytosolic amino acids generally present at lower concentrations (barely detectable in vivo) and A/S ratio always less than one (A/S  1). Besides the in vivo identification of hydatid cysts, advanced knowledge of whether a cyst is fertile is also valuable for planning surgery as the accidental

401

402

Monika Garg and Rakesh K. Gupta

(a)

NAA βCH2

Gly γCH2

Glx γCH2 GABA CH2

Asp βCH2

2.8

2.6

2.4

(b)

(c)

Cr N–CH3 NAA CH3

GPC N–(CH3)3

Ace CH3

PC

3.0

2.5

2.0

Lac CH3 GABA CH2

Ala CH3 1.5

ppm Fig. 23.17 In vitro (perchloric acid extract) proton MR spectrum of malaria from the murine model at 600.13 MHz. Proton spectrum from control mouse (a) without malaria, mouse with cerebral malaria (b), and from mouse with non-cerebral malaria (c). The spectrum from cerebral malaria mouse shows elevated signals of Lac and alanine (Ala) peak as compared to the spectra from control and non-cerebral malarial mouse. The remaining signals are of NAA, Glu/Gln (Glx), -aminobutyric acid (GABA), aspartate (Asp), phosphocholine (PC), glycerophosphocholine (GPC), and Cr. (Courtesy Dr. Caroline Rae, Department of Biochemistry, University of Sydney, Australia.)

MR spectroscopy in intracranial infection

rupture of fertile cysts during surgery results in disease recurrence. Previously, this could only be determined through histopathology. High-resolution MRS facilitates the differentiation of fertile and sterile hydatid cysts (Garg et al., 2002b). Garg et al. observed two additional metabolites: malate (4.3 ppm) and fumarate (6.52 ppm) in the microscopically proven fertile hydatid cysts of humans and sheep; these metabolites are not found in sterile cysts. If future technological developments allow these additional metabolites to be detected in vivo MRS will allow non-invasive therapeutic monitoring, define the dosage and duration of therapy, and refine the selection of patients for percutaneous therapy in future. Malaria Cerebral malaria is a life-threatening complication in 2% of cases of Plasmodium falciparum infection (Marsden and Bruce-Chwatt, 1975). Diffuse involvement of the brain in cerebral malaria patients lead to the non-specific neurological presentations. The prognosis of human cerebral malaria is based on the continual lactic acidosis of brain and CSF; however, lumbar puncture may be hazardous because of cerebral swelling. To date no MRS studies in cerebral malaria patients have been published. However, in the last 3 years, one group has performed in vitro proton MRS on cerebral malaria in murine models inoculated with Plasmodium berghei (Sanni et al., 2001). Increased levels of brain Lac and alanine have been detected (Figure 23.17). Raised intracranial pressure (ICP) due to cerebral edema predisposes the tissue to ischemia, and adherence of parasitized erythrocytes to the cerebral microvascular endothelium leading to microvascular obstruction and regional hypoxia provide a plausible explanation for the elevated levels of Lac. Alanine is considered to be a better marker than Lac for the hypoxia as Lac level may also rise with increase in tumor necrosis factor- and anemia. Moreover, elevation in essential amino acids, such as valine, leucine, and isoleucine has also been observed in experimental cerebral malaria. The concentration of usual brain metabolites NAA and Cho decreases significantly in the murine brain with

cerebral malaria. The significant linear correlation between the time elapsed after infection and small progressive decrease in cell density/cell viability markers glycerophosphocholine (GPC) and NAA has been reported. The metabolite information from in vivo spectroscopy is non-specific for diagnostic purposes, however, it is possible that metabolites may in future provide non-invasive prognostic markers in known disease.

Conclusion MRS should be considered as a part of the imaging protocol for suspected intracranial infections as it may provide valuable information that may be of significance for the diagnosis and management of such cases.

REFERENCES Britt RH, Enzmann DR, Yeager AS. 1981. Neuro-pathological and computerized tomographic findings in experimental brain abscess. J Neurosurg 55: 590–603. Caldemeyer KS, Smith RR, Harris TM, Edwards MK. 1994. MRI in acute disseminated encephalomyelitis. Neuroradiology 36: 216–220. Carpio A, Hauser WA. 2002. Neurocysticercosis and epilepsy. In Taenia solium Cysticercosis from Basic to Clinical Science (Eds., Singh G, Prabhakar S), CAB International, New York, pp. 211–220. Cassady KA, Whitley RJ. 1997. Pathogenesis and pathophysiology of viral infections of the central nervous system. In Infections of the Central Nervous System (Eds, Scheld WM, Whitley RJ, Durack DT), Lippincott-Raven Press, Philadelphia, pp. 7–22. Cecil KM, Jones BV, Williams S, Hedlund GL. 2000. CT, MRI and MRS of Epstein–Barr virus infection: case report. Neuroradiology 42: 619–622. Chawla S, Gupta RK, Kumar R, Garg M, Pradhan S, Pal L, Husain N, Gupta A, Rathore RKS. 2002. Demonstration of scolex in calcified cysticercus lesion using gradient echo with or without corrected phase imaging and its clinical implications. Clin Radiol 57: 826–834. Demaerel P, Wilms G, Robberecht W, Johannik K, Van Hecke P, Carton H, Baert AL. 1992. MRI and herpes simplex encephalitis. Neuroradiology 32: 490–493.

403

404

Monika Garg and Rakesh K. Gupta

Durand ML, Calderwood SB, Weber DJ, Miller SI, Southwick FS, Caviness VS, Swartz MN. 1993. Acute bacterial meningitis in adults. New Engl J Med 328: 21–28. Dzendrowskyi T, Himmelreich U, Malik R, Dowd S, Mountford C, Sorrell T. 2000. Distinction between cerebral Cryptococcomas, Staphylococcus aureus infections and tumours in an animal model. Proc Eight Int Soc Magn Reson Med 8: 173. Ebisu T, Tanaka C, Umeda M, Kitamura M, Naruse S, Higuchi T, Ueda S, Sato H. 1996. Discrimination of brain abscess from necrotic or cystic tumors by diffusion-weighted echo planar imaging. Magn Reson Imaging 14: 1113–1116. Evans AS. 1990. Epidemiological concepts. In Bacterial Infections of Humans (Eds, Evans AS, Brachman PS), Plenum Publishing Corporation, New York, pp. 3–57. Farrar DJ, Flanigan TP, Gordon NM, Gold RL, Rich JD. 1997. Tuberculous brain abscess in a patient with HIV infection: case report and review. Am J Med 102: 297–301. Garcia HH, Gilman RH, Catacora M, Verastegui M, Gonzalez AF, Tsang VC. 1997. Serologic evolution of neurocysticercosis patients after antiparasitic therapy. Cysticercosis working group in Peru. J Infect Dis 175: 486–489. Garg M, Chawla S, Prasad KN, Roy R, Sikora SS, Kumar R, Husain M, Khetrapal CL, Gupta RK. 2002a. Differentiation of hydatid cyst from cysticercus cyst by proton MR spectroscopy. NMR Biomed 15: 320–326. Garg M, Gupta RK, Prasad KN, Sikora SS, Pal L, Chawla S, Kumar R, Husain M, Saxena S, Husain N, Roy R. 2002b. Fertility assessment of hydatid cyst by proton MR spectroscopy. J Surg Res 106: 196–201. Garg M, Gupta RK, Husain M, Chawla S, Chawla J, Kumar R, Rao SB, Misra MK, Prasad KN. 2004. Etiological categorization of brain abscesses with in vivo proton MR spectroscopy. Radiology 230: 893–899. Garg RK, Kar AM. 2002. Neurocysticercosis: diagnosis and treatment in special situations. In Taenia solium Cysticercosis from Basic to Clinical Science (Eds., Singh G, Prabhakar S), CAB International, New York, pp. 281–287. Gupta RK. 2001. Tuberculosis and other non-tuberculous bacterial graunulomatous infections. In MR Imaging and Spectroscopy of Central Nervous System Infection (Eds, Gupta RK, Lufkin RB), Kluwer Academic/Plenum Publishers, New York, pp. 95–145. Gupta R. 2002. Magnetization transfer MR imaging in central nervous system infections. Indian J Radiol Imaging 12: 51–58. Gupta RK, Chang KH. 2001. Parasitic infections. In MR Imaging and Spectroscopy of Central Nervous System Infection (Eds, Gupta RK, Lufkin RB), Kluwer Academic/Plenum Publishers, New York, pp. 205–239. Gupta RK, Husain N, Kathuria MK, Datta S, Rathore RKS, Husain M. 2001b. Magnetization transfer MR imaging

correlation with histopathology in intracranial tuberculomas. Clin Radiol 56: 656–663. Gupta RK, Lufkin RB. 2001. Viral infections. In MR Imaging and Spectroscopy of Central Nervous System infection (Eds., Gupta RK, Lufkin RB), Kluwer Academic/Plenum Publishers, New York, pp. 147–175. Gupta RK, Kathuria MK, Pradhan S. 1999. Magnetization transfer MR imaging in CNS tuberculosis. Am J Neuroradiol 20: 867–875. Gupta RK, Pandey R, Khan EM, Mittal P, Gujral RB, Chhabra DK. 1993. Intracranial tuberculomas: MRI signal intensity correlation with histopathology and localized proton spectroscopy. Magn Reson Imaging 11: 443–449. Gupta RK, Roy R, Dev R, Husain M, Poptani H, Pandey R, Kishore J, Bhaduri AP. 1996. Finger printing of Mycobacterium tuberculosis in patients with intracranial tuberculomas by using in vivo, ex vivo, and in vitro magnetic resonance spectroscopy. Magn Reson Med 36: 829–833. Gupta RK, Vatsal DK, Husain N, Chawla S, Prasad KN, Roy R, Kumar R, Jha D, Husain M. 2001a. Differentiation of tuberculous from pyogenic brain abscesses with in vivo proton MR spectroscopy and magnetization transfer MR imaging. Am J Neuroradiol 22: 1503–1509. Habib AA, Mozaffar T. 2001. Brain abscess. Arch Neurol 58: 1302–1304. Harada M, Hisaoka S, Mori K, Yoneda K, Noda S, Nishitani H. 2000. Difference in water diffusion and lactate production in two different types of postinfectious encephalopathy. J Magn Reson Imaging 11: 559–563. Holtas S, Geijer B, Stromblad LG, Maly-Sundgren P, Burtscher IM. 2000. A ring-enhancing metastasis with cerebral high signal on diffusion-weighted imaging and low apparent diffusion coefficients. Neuroradiology 42: 824–827. Hopewell PC. 1994. Overview of clinical tuberculosis. In Tuberculosis: Pathogenesis, Protection, and Control (Ed., Bloom BR), American Society of Microbiology, Washington DC, pp. 25–46. Ishii K. 1967. Virological and serological diagnosis of Japanese encephalitis. Shinkei Kenkyu No Shimpo 11: 300–311. Kathuria MK, Gupta RK. 2001. Fungal infections. In MR Imaging and Spectroscopy of Central Nervous System Infection (Eds, Gupta RK, Lufkin RB), Kluwer Academic/ Plenum Publishers, New York, pp. 177–203. Kathuria MK, Gupta RK, Roy R, Gaur V, Husain N, Pradhan S. 1998. Measurement of magnetization transfer in different stages of neurocysticercosis. J Magn Reson Imaging 8: 473–479. Kim SH, Chang KH, Song IC, Han MH, Kim HC, Kang HS, Han MC. 1997. Brain abscess and brain tumor: discrimination with in vivo H-1 MR spectroscopy. Radiology 204: 239–245.

MR spectroscopy in intracranial infection

Kohli A, Gupta RK, Poptani H, Roy R. 1995. In vivo proton magnetic resonance spectroscopy in a case of intracranial hydatid cyst. Neurology 45: 562–564. Kumar S, Misra UK, Kalita J, Sawlani V, Gupta RK, Gujral R. 1997. MRI in Japanese encephalitis. Neuroradiology 39: 180–184. Lakeman FD, Whitley RJ. 1995. Diagnosis of herpes simplex encephalitis: application of polymerase chain reaction to cerebrospinal fluid from brain-biopsied patients and correlation with disease. J Infect Dis 171: 857–863. Marsden PD, Bruce-Chwatt IJ. 1975. Cerebral malaria. In Tropics on Tropical Neurology (Ed., Hornabrook RW), Devis, Philadelphia, pp. 29–44. Matthews PM, Shoubridge E, Arnold DL. 1989. Brain phosphorus magnetic resonance spectroscopy in acute bacterial meningitis. Arch Neurol 46: 994–996. Menon DK, Sargentoni J, Peden CJ, Bell JD, Cox IJ, Coutts GA, Baudouin C, Newman CG. 1990. Proton MR spectroscopy in herpes simplex encephalitis: assessment of neuronal loss. J Comput Assist Tomogr 14: 449–452. Meyer RD, Young LS, Armstrong D, Yu B. 1973. Aspergillosis complicating neoplastic disease. Am J Med 54: 6–15. Mikami T, Saito K, Kato T, Irie S, Yoshikawa J, Kondo S. 2002. Detection and characterization of the evaluation of cerebral abscesses with diffusion-weighted magnetic resonance imaging – two case reports. Neurol Med Chir 42: 86–90. Morawetz RB, Whitley RJ, Murphy DM. 1983. Experience with brain biopsy for suspected herpes encephalitis: review of forty consecutive cases. Neurosurgery 12: 654–657. Novak M, Hameed N, Buist R, Blackburn BJ. 1992. Metabolites of alveolar Echinococcus as determined by 31-P and 1H nuclear magnetic resonance spectroscopy. Parasitol Res 78: 665–670. Nurchi G, Floris F, Montaldo C, Mastio F, Peltz T, Corradu M. 1992. Multiple cerebral hydatid disease: case report with magnetic resonance imaging study. Neurosurgery 30: 436–438. Önal Ç, Barlas O, Orakdögen M, Hepgül K, Izgi N, Ünal F. 1997. Three unusual cases of intracranial hydatid cyst in the pediatric age group. Pediatr Neurosurg 26: 208–213. Osenbach RK, Loftus CM. 1992. Diagnosis and management of brain abscess. Neurosurg Clin N Am 3: 403–420. Pandit S, Lin A, Gahbauer H, Libertin CR, Erdogan B. 2001. MR spectroscopy in neurocysticercosis. J Comput Assist Tomogr 25: 950–952. Parameshwaran K, Radhakrishnan K. 2002. Subacute sclerosing panencephalitis. In Reviews in Tropical Neurology (Eds, Kar AM, Shukla R, Agarwal A, Verma R), Shivam Arts, Lucknow, pp. 30–40. Ross JP, Cohen JI. 1997. Epstein–Barr virus. In Infections of the Central Nervous System (Eds., Scheld WM, Whitley RJ, Durack DT), Lippincott-Raven Press, Philadelphia, pp. 117–127.

Rudman MA, Khaffai S. 1988. CT of cerebral hydatid disease. Neuroradiology 30: 496–499. Runge VM, Welle JW, Williams NM, Lee C, Timoney JF, Young AB. 1995. Detectability of early brain meningitis with magnetic resonance imaging. Invest Radiol 30: 484–495. Salvan AM, Confort-Gouny S, Cozzone PJ, Vion-Dury J, Chabrol B, Mancini J. 1999. In vivo cerebral proton MRS in a case of subacute sclerosing panencephalitis. J Neurol Neurosurg Psychiatr 66: 547–555. Sanni LA, Rae C, Maitiland A, Stocker R, Hunt NH. 2001. Is ischemia involved in the pathogenesis of murine cerebral malaria? Am J Pathol 159: 1105–1112. Satishchandra P, Sharma GRK. 2002. Fungal infections of the nervous system. In Reviews in Tropical Neurology (Eds, Garg RK, Kar AM, Agarwal A, Shukla R, Verma R), Shivam Arts, Lucknow, pp. 111–124. Sepkowitz K, Armstrong D. 1997. Space-occupying fungal lesions. In Infections of the Central Nervous System (Eds, Scheld WM, Whitley RJ, Durack DT), Lippincott-Raven Press, Philadelphia, pp. 741–762. Sharda D, Chawla S, Gupta RK. 2002. Imaging and spectroscopy of neurocysticercosis. In Taenia solium Cysticercosis from Basic to Clinical Science (Eds., Singh G, Prabhakar S), CAB International, New York, pp. 311–327. Sharma K, Gupta RK. 1993. Scan-negative neurocysticercosis. Pediatr Neurosurg 19: 206–208. Shawl S. 1995. Neurologic evaluation of patient with acute bacterial meningitis. Neurology Clinic 13: 549–577. Shukla-Dave A, Gupta RK, Roy R, Husain N, Paul L, Venkatesh SK, Rashid MR, Chhabra DK, Husain M. 2001. Prospective evaluation of in vivo proton MR spectroscopy in differentiation of similar appearing intracranial cystic lesions. Magn Reson Imaging 19: 103–110. Siegal JA, Cacayorin ED, Nassif AS, Rizk D, Galambos C, Levy B, Kennedy D, Visconti J, Perman W. 2000. Cerebral mucormycosis: proton MR spectroscopy and MR imaging. Magn Reson Imaging 18: 915–920. Silverstein A, Steinberg G, Nathanson M. 1972. Nervous system involvement in infectious mononucleosis: the heralding and-or major manifestation. Arch Neurol 26: 353–358. Sudhakar KV, Agarwal S, Rashid MR, Hussain N, Hussain M, Gupta RK. 2001. MRI demonstration of haemorrhage in the wall of a brain abscess: possible implications for diagnosis and management. Neuroradiology 43: 218–222. Takanashi J, Sugita K, Ishii M, Aoyagi M, Niimi H. 1997. Longitudinal MR imaging and proton MR spectroscopy in herpes simplex encephalitis. J Neurol Sci 149: 99–102. Tandon PN, Pathak SN. 1973. Tuberculosis of the central nervous system. In Tropical Neurology (Ed., Spillane JD), Oxford University Press, New York, pp. 37–62.

405

406

Monika Garg and Rakesh K. Gupta

Tien RD, Felsberg GJ, Osumi AK. 1993. Herpesvirus infections of the CNS: MR findings. Am J Roentgenol 161: 167–176. Tsuchiya K, Inaoka S, Mizutani Y, Hachiya J. 1997. Fast fluidattenuated inversion-recovery MR of intracranial infections. Am J Neuroradiol 18: 909–913. Tsuchiya K, Katase S, Yoshino A, Hachiya J. 1999. Diffusionweighted MR imaging of encephalitis. Am J Roentgenol 173: 1097–1099. Venkatesh SK, Gupta RK. 2001. Pyogenic infections. In MR Imaging and Spectroscopy of Central Nervous System Infection (Eds, Gupta RK, Lufkin RB), Kluwer Academic/Plenum Publishers, New York, pp. 57–93.

Whitener DR. 1978. Tuberculous brain abscess. Arch Neurol 35: 148–155. Wilkins PP, Wilson M, Allan JC, Tsang VCW. 2002. Taenia solium cysticercosis: immunodiagnosis of neurocysticercosis and taeniasis. In Taenia solium Cysticercosis from Basic to Clinical Science (Eds, Singh G, Prabhakar S), CAB International, New York, pp. 329–341. Zimmerman RA, Bilaniuk LT, Sze G. 1987. Intracranial infection. In Magnetic Resonance Imaging of the Central Nervous System (Eds, Brant-Zawadzki M, Norman D), Raven Press, New York, pp. 235–257.

MR spectroscopy in intracranial infection

407

Case Study 23.1 MRS in variant Creutzfeldt–Jakob Disease Adam D. Waldman, M.D., Ph.D., Hammersmith Hospitals and Institute of Neurology, London, UK History mI

Three patients with moderately advanced probable variant Creutzfeldt–Jakob Disease (vCJD) (male 19 years, female 31 years, male 38 years) as diagnosed from tonsillar biopsy.

vCJD NAA

Technique Quantitative short TE single voxel MRS (PRESS TE  30 ms, TR  2000 ms) of left pulvinar (thalamus). 4

3 2 1 Frequency (ppm)

0

Imaging findings NAA

Pulvinar sign was demonstrated in all three cases on T2-weighted MRI.

Controls (n  8)

vCJD

Control

3.5 3

Markedly elevated myo-inositol (mI) and decreased NAA was shown compared to normal control subjects (mI/NAA: z  38, 30, 81).

mI/NAA

MRS findings

mI

2.5 2 1.5 1 0.5

Discussion Decreased pulvinar NAA and increased mI are in keeping with severe neuronal loss and intense gliosis which are known to be features of vCJD. Metabolite changes are similar to those in parietal cortex in Alzheimer’s disease, but much more severe. The magnitude of mI/NAA increase in vCJD suggests that MRS changes may precede visible imaging abnormalities in early disease. Less marked metabolite abnormalities are seen in basal ganglia and frontal white matter in vCJD. Other Creutzfeldt Jakob Disease (CJD) types show different regional distributions, and elevated mI but with normal NAA may be seen in familial prion gene carriers.

0 0.4

0.6

0.8

1

1.2

4

3 2 1 Frequency (ppm)

Key points MRS at short TE allows mI abnormality to be detected. mI/NAA is a potential diagnostic index for vCJD when MRI findings are negative or equivocal. Regional distribution of metabolite abnormalities differ between prion diseases.

Reference Cordery R, MacManus D, Rossor MN, Collinge J, Waldman AD. 2003. Short TE proton spectroscopy in variant and familial Creutzfeldt Jakob Disease. Proc Int Soc Mag Res Med 11: 438.

0

24

The role of diffusion-weighted imaging in intracranial infection Christopher G. Fillipi Fletcher Allen Health Care-University of Vermont, Burlington, Vermont, USA

Key points • A brain abscess typically has markedly hyperintense diffusion-weighted imaging (DWI) and hypointense apparent diffusion coefficient (ADC) maps in the central cavity. • Patients with necrotic or cystic primary brain tumors or metastatic disease typically demonstrate hypointensity on the DWI and marked hyperintensity on the ADC map. • Radiation necrosis may demonstrate hyperintensity on the DWI and hypointense ADC (similar to abscess). • Necrotic or cystic primary or secondary tumors with intratumoral hemorrhage may appear identical to an abscess (increased DWI and hypointense ADC maps). • The mean regional cerebral blood volume of tumor walls is generally higher than that of a pyogenic abscess wall.

Diffusion-weighted imaging (DWI) is now part of the routine brain MR imaging (MRI) protocol at many institutions. The principles and techniques for DWI are covered in detail in Chapters 4–6. Diffusionweighted sequences are sensitive to the microscopic motion of water molecules. MR diffusion imaging uses the incoherent motion of water molecules as tissue contrast (Ulug et al., 1999). Alterations in the degree of diffusion reflect alterations in the microscopic environment of these water molecules. It is 408

reasonable to infer that the changes in diffusion reflect changes at the scale of cellular and extracellular structures of the brain (Chun et al., 2000). The high sensitivity and specificity of echo-planar DWI in the diagnosis of acute cerebral infarction is widely known (Benveniste et al., 1992; Hossman et al., 1994; Lovblad et al., 1998; Stadnick et al., 2001). Reduced diffusion that is observed during an acute infarct is thought to represent cytotoxic edema and contraction of the extracellular space (Benveniste et al., 1992; Hossman et al., 1994; Lovblad et al., 1998; Stadnick et al., 2001). The translational motion of water in brain tissue is the basis of clinical diffusion-weighted MRI (LeBihan, 1991; Horsfield and Jones, 2002). Diffusing molecules within brain tissue will be impeded or influenced by the interaction with cell membranes and other intracellular and extracellular structures (Horsfield and Jones, 2002). On diffusion-weighted MRI, one is measuring the average signal within a voxel or the volume-averaged propagation of the diffusing molecules as they interact with the cellular structures present within the voxel (Horsfield and Jones, 2002). One can quantify this value for the average diffusion within a specified region-of-interest from the apparent diffusion coefficient (ADC) map, which is generated during DWI, and this is termed the diffusion constant or Dav. DWI may provide insight into the nature and degree of pathological damage that occurs in diseases of the central nervous system (CNS) when cellular structures are damaged or disrupted as part of the pathological process (Ulug et al., 1999; Horsfield

The role of diffusion-weighted imaging in intracranial infection

and Jones, 2002). Furthermore, when cellular structures are highly ordered, such as axonal fibers and white matter (WM) fiber tracts, diffusion-tensor imaging (DTI) may provide unique information, because the directional nature of the diffusionsensitizing gradients can encode properties that vary with direction (Ulug et al., 1999; Horsfield and Jones, 2002). By acquiring diffusion-sensitive (diffusiontensor) images in which this directional information is measured, one can determine the anisotropy of WM fiber tracts, which are highly ordered and have distinct directions. From this information, one can make inferences about the integrity of the WM microstructure (Ulug et al., 1999; Filippi et al., 2001). DWI has been used to study intra-axial neoplasms, extra-axial meningiomas, and demyelinating lesions. DWI allows for the easy differentiation of epidermoids from arachnoid cysts within the extra-axial compartment of the brain (Tsuruda et al., 1990; Maeda et al., 1992). DWI provides unique information on the diffusion properties of water in diseased brain, and this chapter reviews the role of DWI and diffusion-tensor MRI in the diagnosis of intracranial infections.

Brain abscess The diagnosis of brain abscess, a potentially fatal lesion, remains a diagnostic challenge to both clinicians and radiologists, because the presenting signs and symptoms as well as the neuroimaging findings on computed tomography (CT) and MR are often nonspecific (Kim et al., 1998). Only 40–50% of patients are febrile on presentation to the hospital (Kim et al., 1998). The most common manifestations of intracerebral abscess are those of any expanding mass lesion such as headache, change in mental status, nausea, emesis, seizure, or focal sensorimotor or neurological deficit (Chun et al., 1986; Wispelwey et al., 1991; Kim et al., 1998). Typically, on MRI, an abscess presents as a ring-enhancing lesion on postcontrast images which has focal mass effect and incites surrounding vasogenic edema. The differential diagnosis for such a lesion includes primary brain neoplasm, metastatic disease, abscess, radiation

necrosis, demyelinating disease, and less commonly subacute infarct or resolving contusion. The development of a brain abscess is the result of an acute, fulminant bacterial illness. The route of infection is from either hematogenous spread and/ or direct invasion. An acute abscess forms after an initial period of cerebritis, which typically lasts from 6 to 12 days. During the initial cerebritis stage, an ill-defined subcortical area of hyperintensity on T2-weighted images is associated with iso- to mildly hypointense regions on non-contrast T1-weighted images, representing an edema pattern (Osborn, 1994; Desprechins et al., 1999). As a collagenous abscess capsule develops, on routine MRI, there is peripheral or surrounding vasogenic edema, which is hyperintense on the long repetition time (TR) sequences and hypointense on the short TR sequences with focal mass effect. (Figure 24.1) There is a central area of necrosis which shows hypointensity relative to the brain parenchyma on T1-weighted images, and sometimes concentric rings of subtle hypointensity can sometimes be seen on the short TR sequences. The abscess forms a distinctive capsule, which is typically hypointense on the long TR sequences (Figure 24.2). The area of central necrosis is hyperintense on the long TR sequences, and the hyperintense, surrounding vasogenic edema allows for easy delineation of this abscess capsule, then the abscess capsules intensely enhance (Figure 24.2). The wall can have variable thickness, but both the internal and external walls of the abscess capsule are usually smooth and non-nodular. Even if the abscess appears lobulated, those lobulations are typically smooth in appearance. Although the marked hypointensity of the abscess capsule on the T2weighted images is a characteristic that has been used to differentiate infection from neoplasm, it is not always seen in all abscess cases, and it may be seen in some cases of tumor (Haimes et al., 1989; Kim et al., 1998). Typically, in a brain abscess, the central cavity is markedly hyperintense on DWI, and on the ADC maps, the center of the abscess is markedly hypointense relative to brain parenchyma and cerebrospinal fluid (CSF) (Figure 24.3). In a small study by Kim et al., five consecutive patients with proven brain abscess were compared to four patients with

409

410

Christopher G. Fillipi

P120

P120

Fig. 24.1 On the T1-weighted image, a hypointense lesion within the right posterior frontal lobe is noted with focal mass effect on the body of the right lateral ventricle. Surrounding vasogenic edema is more apparent on the T2-weighted image. On the long TR sequences, the hypointense rim of the abscess capsule is seen.

Fig. 24.2 On the T2-weighted image, a well-marginated lesion is noted in the right occipital lobe which has central hyperintensity, surrounding vasogenic edema, and the characteristic thin hypointense rim of an abscess, which demonstrates marked ring enhancement on the corresponding T1-weighted image to the left, taken post-contrast.

The role of diffusion-weighted imaging in intracranial infection

Fig. 24.3 In this 84-year-old patient, who developed an abscess following a dental procedure, the center of the abscess demonstrates marked hyperintensity on the trace image and hypointensity on the ADC map, indicative of marked restriction to water diffusion.

cystic or necrotic brain tumors. All the patients with brain abscess on DWI showed markedly hyperintense signal on the DWI images, whereas the patients with brain tumor showed marked hypointensity. In this study, on the DWI, the abscess capsule was less well defined and hypointense than it was on the long TR sequences. The hypointensity of the abscess capsule on the T2-weighted spin-echo MR images is thought to reflect a susceptibility artifact from the presence of free radicals (Haimes et al., 1989; Kim et al., 1998). Susceptibility artifacts are more pronounced on DWI, which is an echo-planar pulse sequence. One explanation for this apparent discrepancy may be that the markedly hyperintense appearance of the central abscess cavity masks the low diffusional properties of the capsule (Kim et al., 1998). The marked hyperintensity within the abscess capsule on the diffusion-weighted images and corresponding hypointensity on the ADC maps is a consequence of the physical and biochemical

components of the abscess cavity. Pus within an abscess cavity is thick, mucoid, material consisting of inflammatory cells, necrotic tissue, bacteria, and proteinaceous fluid, which has high viscosity (Winn and Kissane, 1996; Kim et al., 1998). In an environment of high viscosity, the diffusion of water molecules will be markedly restricted. Water molecules in this environment are bound to carboxyl, hydroxyl, and amino acid groups on the surface of molecules (Castillo and Mukherji, 2000). All of these characteristics of an abscess cavity likely account for the changes in signal intensity that are observed within an abscess cavity on DWI. This association was first suggested in a study by Ebisu et al. (1996) in which DWI of pus aspirate from a brain abscess was performed in vitro. This aspirate demonstrated marked hyperintensity on the diffusion-weighted images and a markedly reduced ADC. In a study by Desprechins et al. (1999) two cases of cerebral abscess were compared to eight cases of cystic or necrotic gliomas and two cases of necrotic

411

412

Christopher G. Fillipi

Fig. 24.4 A right parietal ring-enhancing lesion is seen, which demonstrates more eccentric and nodular enhancement medially, which mitigates against the diagnosis of abscess and suggests a metastatic lesion in this patient with a primary lung carcinoma.

metastatic disease. Both of the patients with cerebral abscess had the expected appearance of a central hyperintense cavity on DWI that is hypointense on the ADC map. The calculated value of the diffusion constant from the ADC map was markedly reduced compared to the value of normal brain parenchyma, which suggested that the markedly reduced value of the ADC may be helpful in the differentiation between abscess and neoplasm. Patients with necrotic or cystic primary brain tumors or metastatic disease typically demonstrate hypointensity on the diffusion-weighted image, marked hyperintensity on the ADC map, and an elevation of the diffusion constant indicating less restriction to the translational motion of water molecules within these lesions (Tien et al., 1994; Krabbe et al., 1997; Tung et al., 2001) (Figures 24.4 and 24.5). The observation that a ring-enhancing lesion causes marked restriction to water diffusion is not specific for cerebral abscess. In a retrospective study by Tung et al., two ring-enhancing lesions on post-contrast MRI displayed marked hyperintensity

on DWI, suggesting brain abscess, but one of these lesions represented radiation necrosis and the other was metastatic squamous cell carcinoma (Tung et al., 2001). Although DWI remains an important MR pulse sequence for the diagnosis of intracerebral abscess, DWI is not pathognomonic for brain abscess (Hartmann et al., 2001; Tsui et al., 2002). In a prospective study of ring-enhancing lesions on MR, Hartmann et al. examined three abscesses, six glioblastoma, and eight metastatic lesions in which one of the metastatic lesions demonstrated marked central restriction to water diffusion (Hartmann et al., 2001). Necrotic or cystic areas in primary or metastatic neoplasms are usually more hemorrhagic, less cellular, and lower in viscosity compared to purulent collections or pus (Hartmann et al., 2001). These lesions are typically hypointense on DWI and have marked elevations in the average diffusion constant quantified from the ADC map (Tien et al., 1994; Tsuchiya et al., 1998; Noguchi et al., 1999; Monbati et al., 2000; Hartmann et al., 2001; Guzman et al., 2002; Tsui et al.,

The role of diffusion-weighted imaging in intracranial infection

Fig. 24.5 In this same patient, the lesion is markedly hypointense on the trace image and hyperintense on the ADC map, which is in contradistinction to the typical appearance of an abscess capsule on DWI. The marked hyperintensity on the ADC map indicates little restriction to the translational motion of water molecules. The appearance of this lesion on DWI is typical of a cystic or necrotic metastatic lesion.

2002). In these reports of metastatic disease and radiation necrosis demonstrating hyperintensity on DWI and decreased ADC values, the authors have hypothesized that sterile liquefaction necrosis may be responsible for the observed imaging characteristics (Tien et al., 1994; Tsuchiya et al., 1998; Noguchi et al., 1999; Monbati et al., 2000; Hartmann et al., 2001; Guzman et al., 2002; Tsui et al., 2002). Radiation necrosis causes vasculopathy and necrosis in treated tissue, which can lead to areas of sterile liquefaction necrosis containing pus-like material and polymorphonuclear leukocytes (Tsuchiya et al., 1998). Another hypothesis for the restriction of water diffusion within some metastatic lesions would be the presence of mucin-containing material, which is highly viscous (Tsuchiya et al., 1998). Cystadenocarcinomas, in particular, are often mucin containing. Difficulty in the diagnosis of intracerebral abscess occurs when a cystic or necrotic primary brain neoplasm or metastatic lesion develops intratumoral

hemorrhage or superimposed infection. The signal characteristics of these lesions will appear identical on DWI. Recent investigators have used a combination of perfusion MRI and DWI to differentiate pyogenic cerebral abscess from infected brain tumor (Chan et al., 2002). MR perfusion imaging acquires rapid, dynamic images during a bolus injection of gadolinium (Gd) contrast, a paramagnetic agent. The principles underlying MR perfusion imaging during bolus injection of Gd contrast is described in detail in Chapter 4. During the bolus injection of Gd contrast, the contrast agent enters the intravascular compartment of the brain which causes a drop in signal intensity on either a T*2 or T2-weighted image (Caramia et al., 1995). By tracking the first pass of the contrast through the brain, regional cerebral blood volume (rCBV) maps can be generated (Caramia et al., 1995). In a study by Chan et al., the tumor wall of various types of cystic or necrotic brain tumors had significantly lower ADCs relative to those of the abscess

413

414

Christopher G. Fillipi

Fig. 24.6 In this image, courtesy of Chan et al., a 56-year-old woman with metastatic adenocarcinoma was imaged using DWI (superior image) and perfusion MRI (inferior image). On DWI, there is hyperintensity within the wall of this metastatic tumor lesion due to relatively restricted diffusion. On the axial rCBV map, there is signal hyperintensity within the tumor wall, indicating that the tumor possesses a high rCBV value.

1997). The peripheral portion or wall of the cystic or necrotic brain tumor has relatively restricted diffusion due to the malignant tumor cells which are closely packed in the tumor wall resulting in high cellularity, which would be expected to generate hyperintensity on the DWI trace image (Chenevert et al., 1997). Vascular proliferation and angiogenesis are important factors in the determination of histopathological grade of primary brain neoplasms and are essential for tumor viability, growth, and propagation (Brem et al., 1972; Burger, 1986). There is a strong correlation between tumor grade and rCBV in which higher measurements of rCBV correlate with higher-grade tumors (Aronen et al., 1994; Knopp et al., 1999). In the study by Chan et al., the cystic or necrotic tumors had large rCBV values for the tumor wall and were hyperperfused relative to normal WM, which likely relates to tumor angiogenesis, whereas the abscess capsule had low rCBV measurements (Chan et al., 2002). Typically, an abscess capsule is poorly vascularized and collagenous, which likely accounts for this observation (Wenz et al., 1996).

Meningitis

Fig. 24.7 In this image, courtesy of Chan et al., a 62-year-old man who suffered from multiple pyogenic abscesses was imaged using DWI (superior image) and MR perfusion (inferior image). On the axial DWI, there is hypointensity within the abscess wall, indicating relatively free diffusion. The axial rCBV map shows signal hypointensity in the abscess wall, indicating that the abscess wall has a low rCBV value.

wall and appeared relatively hyperintense of the diffusion-weighted image (Chan et al., 2002). The rCBV ratio relative to normal WM of the peripheral tumor wall of various types of cystic or necrotic brain tumor were significantly larger than the mean rCBV of the pyogenic cerebral abscess wall (Chan et al., 2002). (Figures 24.6 and 24.7) The hypointensity of the abscess capsule that is observed on both the long TR sequences and DWI may relate to an increase in the extracellular fluid in the capsular wall as a result of inflammation (Chenevert et al.,

MRI has a limited role in the evaluation of acute meningitis, which is typically diagnosed by a combination of clinical findings on neurological examination and CSF results from lumbar puncture. Leptomeningeal enhancement is typically seen in the subarachnoid spaces in patients with acute meningitis on the post-contrast T1-weighted images (Hajnal et al., 1992; Singer et al., 1998). Rarely, abnormal signal intensity can be detected within the CSF on intermediate or proton density-weighted images and fluid attenuated inversion recovery (FLAIR) sequences (Hajnal et al., 1992; Singer et al., 1998). The role of MR is to detect the complications from acute meningitis such as venous sinus thrombosis, hydrocephalus, arterial occlusion, infarction, or extension into the subdural compartment or ventricles. In our experience, patients with acute meningitis who have been imaged with DWI, often show a hyperintense signal within the subarachnoid spaces, corresponding to the areas of abnormal enhancement on the post-contrast

The role of diffusion-weighted imaging in intracranial infection

Fig. 24.8 In this case of a young adult male with meningitis, three axial and one coronal postcontrast T1-weighted image show marked enhancement within the leptomeninges, sulci, and basal cisterns, all of which are consistent with acute meningitis. In addition, there is loss of the normal CSF signal within the atrial of the lateral ventricles, which suggests the presence of debris and ventriculitis, a complication of acute meningitis.

images. It is likely that this hyperintense signal on MR reflects the presence of inflammatory cells or pus within this compartment, both of which would be expected to restrict water diffusion (Figure 24.8). The appearance on DWI of intraventricular extension or ventriculitis as a complication of meningitis has been recently described (Pezzullo et al., 2003). It is easily diagnosed on MR by a fluid–fluid level on DWI in which the dependent portion is markedly

hyperintense, indicative of restricted diffusion, which may be on the basis of the presence of pus and/or hemorrhage (Figure 24.9).

Subdural empyema Subdural empyema (SDE) is a life-threatening condition and neurosurgical emergency. Although it may

415

416

Christopher G. Fillipi

Fig. 24.9 In the same case as preceding figure, corresponding images from DWI show marked hyperintensity within the basal cisterns and leptomeninges, indicative of restricted diffusion. Hyperintense fluid–fluid levels are noted within the lateral ventricles, consistent with ventriculitis. This may be infectious and/or hemorrhagic material within the ventricles which markedly restricts the translational motion of water and accounts for the marked hyperintensity seen.

occur as a result of meningitis, SDE is more frequently a complication of mastoiditis and sinusitis, and it typically occurs in adolescents or young adults. On imaging, there is a subdural collection, which frequently displays a disproportionate amount of mass effect, is more apparent on MR than CT, is accompanied by marked dural enhancement, and is usually markedly hyperintense on both FLAIR and DWI. Like the pus within an intracerebral abscess, the infected fluid within a subdural collection causes marked restriction to water diffusion, which accounts for the hyperintense signal on DWI (Figure 24.10(a–g)).

Encephalitis Encephalitis refers to a non-localized infection of the brain, which is usually due to a viral agent, but parasites, fungi, and prions are implicated as other etiological agents for encephalitis. Perivascular lymphocytic infiltration characterizes the histopathological process. On MRI, the pattern of encephalitis is determined principally by the mode of spread. Herpes simplex virus (HSV) encephalitis type I (HSV I), which is the most common sporadic meningoencephalitis, results from reactivation of the virus in the Gasserian ganglion. It is direct perineural or

The role of diffusion-weighted imaging in intracranial infection

(a)

(b)

(c)

(d)

Fig. 24.10 (a) A crescentic-shaped, enhancing, extra-axial fluid collection is noted in the subdural compartment of the left frontotemporal region consistent with SDE, which is immediately subjacent to an area of left frontal sinusitis and marked dural enhancement, suggesting that this represents direct contiguous spread of sinus infection intracranially in this 38-year-old woman with chronic sinusitis presenting with headache. In Figure 24.10(b), more superiorly, a discrete abscess is seen within the left frontal lobe. On FLAIR sequences, Figures 24.10(c) and (d), these lesions demonstrate marked hyperintensity. There is abnormal hyperintense signal noted within the left Sylvian fissure indicative of leptomeningeal infection. On DWI, Figures 24.10(e ) and (f), the left frontotemporal SDE and left frontal abscess demonstrate marked hyperintensity indicative of restricted diffusion, likely due to the presence of pus. These areas show corresponding hypointensity on the ADC map, Figure 24.10(g).

axoplasmic spread along the meningeal branches of the trigeminal nerve to the base of brain, which causes this encephalitis. Since there is high morbidity and mortality from HSV I, prompt diagnosis and treatment is of the utmost importance. If the diagnosis is suspected, treatment is often empirically started using intravenous acyclovir until the results from the polymerase chain reaction (PCR) tests of CSF are determined (Skoldenberg, 1996). MRI is often used to confirm the diagnosis, to assess the extent of

damage, and to determine whether there is unilateral or bilateral involvement. On MR, an edema pattern within the anteromedial temporal lobe and insular cortex is typically seen. Hypointense swollen gyri are seen on the short TR sequences, which are hyperintense on the long TR sequences. A hemorrhagic component is often present, and enhancement may not be seen initially (Demaerl et al., 1992; Tien et al., 1993; Ashikaga et al., 1996). Since DWI is sensitive to environments of restricted water diffusion

417

418

Christopher G. Fillipi

(e)

(f)

(g)

Fig. 24.10 (cont.)

on a microscopic level, DWI may offer the best hope for prompt detection of signal abnormality in the anteromedial temporal lobe and insular cortex for which HSV I has an exquisite predilection. The findings on DWI in HSV I have been described in a couple of papers (Sener, 2001; Heiner and Demaerel, 2003). (Figure 24.11(a–c)) In one of these reports (Sener, 2001), the hyperintense signal abnormality in the anteromedial temporal lobe, which likely represented cytotoxic edema, occurred in patients who had a more fulminant disease course and poorer outcome. More investigative work will be needed to determine if the pattern of signal abnormality on DWI in HSV I can predict disease course or outcome. Epstein–Barr virus (EBV) is a member of the herpes virus family. Fewer than 5% of patients with infectious mononucleosis will develop encephalomyelitis (Cecil et al., 2000). Epstein–Barr encephalomyelitis is typically an acute, self-limited illness. There are a handful of case reports in recent

neuroradiological literature in which the MR findings include bilateral deep gray nuclei hyperintensity on T2-weighted imaging and FLAIR. The brainstem and spinal cord gray matter (GM) can sometimes be involved. Enhancement is not usually a prominent feature (Cecil et al., 2000). In a recent case at our institution, DWI was performed on a patient with Epstein–Barr encephalomyelitis (Figure 24.12(a–b)). The thalami and basal ganglia demonstrated symmetric signal abnormality and a swollen appearance. The hyperintensity that was observed on DWI likely reflected the T2 shine-through phenomena.

Human immunodeficiency virus Recent studies have examined the MR diffusion properties of infectious abscesses and neoplastic lesions in patients with acquired immunodeficiency

The role of diffusion-weighted imaging in intracranial infection

(b)

(a)

(c)

(a)

Fig. 24.11 On the FLAIR images, Figure 24.11(a), in this patient with HSV I, the classic appearance of unilateral anteromedial temporal lobe swelling is demonstrated on the right side in this patient. The uncus and anteromedial temporal lobe are hyperintense, and there is extension of this abnormal signal into inferior right frontal lobe via the uncinate fasciculus. These areas do not demonstrate significantenhancement post-contrast, Figure 24.11(b), an expected finding in the early stages of this disease process. Marked hyperintense signal is noted within the right anteromedial temporal lobe, right inferior frontal lobe, and right parahippocampal gyrus on DWI, Figure 24.11(c), indicating restricted diffusion, which has been reported in patients with HSV I.

(b)

Fig. 24.12 In this child with chronic EBV infection, on the long TR sequences, Figure 24.12(a), the thalami, basal ganglia, and caudate nuclei demonstrate bilateral symmetric hyperintensity and appear swollen or expanded. On the diffusion-weighted sequence, Figure 24.12(b), these areas demonstrate some hyperintense signal that does not appear as prominent as the signal observed on the trace image, and this likely represented T2 shine-through. This was confirmed on ADC maps in which calculation of the diffusion constant was within normal limits within these areas of abnormal signal intensity.

419

420

Christopher G. Fillipi

syndrome (AIDS) (Filippi et al., 2001; Camacho et al., 2003). The differential diagnosis of focal intracranial lesions in AIDS patients is long and extensive, but the two most common lesions include toxoplasmosis and lymphoma (Dina, 1991; Chang et al., 1995; Smirniotopoulos et al., 1997). On MRI, both toxoplasmosis and CNS lymphoma can have similar appearances frequently presenting as ring-enhancing lesions. The differentiation between these two etiological possibilities is important. If AIDS patients are treated presumptively for toxoplasmosis but have lymphoma, these patients suffer the potential toxicities of the anti-toxoplasmosis medications and frequently deteriorate neurologically (Dina, 1991; Chang et al., 1995; Ramsey and Gean, 1997; Smirniotopoulos et al., 1997; Mamidi et al., 2002; Camacho et al., 2003). If an AIDS patient has a lesion, which is lymphoma, but the patient is treated with anti-toxoplasmosis medications, the patient’s condition will deteriorate, which may lead to biopsy with its inherent risks (Dina, 1991; Chang et al., 1995; Ramsey and Gean, 1997; Smirniotopoulos et al., 1997; Mamidi et al., 2002; Camacho et al., 2003). Further, if steroids are given during treatment for presumed toxoplasmosis, this therapy may mask a hidden lymphoma (Dina, 1991; Chang et al., 1995; Ramsey and Gean, 1997; Smirniotopoulos et al., 1997; Mamidi et al., 2002; Camacho et al., 2003). In a study by Camacho et al., 13 brain toxoplasmosis and 8 brain lymphoma lesions were studied with DWI. Toxoplasmosis lesions demonstrated significantly greater diffusion than that of lymphoma lesions (Chang et al., 1995). If the ADC ratio exceeded 1.6, only toxoplasmosis lesions fell into that category, but nearly half of the toxoplasmosis lesions were between 1.0 and 1.6 for the ADC ratio, which overlapped with the intracranial lymphoma group (Chang et al., 1995). Thus, the utility of the DWI sequence, and, in particular, calculating the diffusion constant form the ADC has yet to be determined. However, in many cases, the imaging features of toxoplasmosis on DWI are distinct from CNS lymphoma. Typically, the toxoplasomosis lesions are hypointense on the short TR sequences, hyperintense on the long TR sequences, ring-enhancement post-contrast, and markedly hyperintense on DWI (Figure 24.13) whereas CNS lymphoma is often hypointense on DWI (Figure 24.13).

Perfusion MRI may also prove to be a useful diagnostic tool to distinguish cerebral toxoplasmosis from intracranial lymphoma. In a study by Ernst et al., albeit a small sample of 13 patients, an elevated rCBV or hyperperfusion was typically seen in CNS lymphoma whilst toxoplasmosis was consistently characterized by hypoperfusion or reduced rCBV values (Ernst et al., 1998). Early in the course of infection, human immunodeficiency virus (HIV) enters the CNS and produces neuropsychiatric impairment throughout the course of the illness for most patients with AIDS, which ranges from HIV-associated mild cognitive motor disorder to HIV-associated dementia (Heaton et al., 1995; Simpson and Berger, 1996; Bencherif and Rottenberg, 1998; Filippi et al., 1998). These disorders combine cognitive, motor, and behavioral symptoms, which suggests that the virus has a predilection for subcortical WM (Kure et al., 1991; Brew et al., 1997; Bencherif and Rottenberg, 1998; van Gorp et al., 1999a, 1999b). Neuropathological studies have shown histopathological changes in the subcortical WM, including multinucleated giant cells, astrogliosis, and myelin pallor (Kure et al., 1991; Brew et al., 1997; van Gorp et al., 1999). Elevations in the CNS viral load levels have been associated with HIV-related neurocognitive disorders (Brew et al., 1997; Ellis et al., 1997; McArthur et al., 1997; DiStefano et al., 1998). Recent evidence supports the concept that the CNS acts as a “reservoir” for HIV (Price and Strapans, 1997). With the advent of protease inhibitor therapy, in combination with azidothymidine (AZT), reductions in CSF viral load levels (Gisslen et al., 1997, 1998), reversal of WM lesions on MR (Filippi et al., 2001), and improvements in neuropsychological test performance (Ferrando et al., 1998) have all been observed. In a recent study, investigators examined a small cohort of HIV-positive patients using both DWI and diffusion-tensor MRI (Filippi et al., 2001). Quantitative differences in both the average diffusion constant and anisotropy were detected in the corpus callosum and subcortical WM, which seemed to correlate with viral load levels despite all patients having normal-appearing brain MR studies (Filippi et al., 2001). Further investigation will be needed to determine if these quantitative

The role of diffusion-weighted imaging in intracranial infection

(a)

(b)

(c) Fig. 24.13 In this immunocompromised patient with toxoplasmosis, in Figure 24.13(a), on two axial FLAIR images, multiple lesions are noted bilaterally within the frontal and parietal lobes with the dominant lesion in the left periatrial WM region demonstrating central hyperintensity and a peripheral rim of hypointensity, which is characteristic of an abscess. In Figure 24.13(b), this dominant lesion is hypointense on the T1-weighted sequence and markedly hyperintense on the DWI trace image. In Figure 24.13(c), this dominant lesion demonstrates faint peripheral enhancement, and marked central hypointensity on the ADC map indicative of restricted diffusion.

diffusion MR techniques can be used as a neuroimaging marker of CNS disease in HIV, which may provide a neuroimaging outcome marker of the efficacy of highly active antiretroviral therapy (HAART). Anisotropy changes in HIV were observed in a similar study, although the mean diffusivity and T2 values were normal (Pomara et al., 2001), which may indicate that anisotropy and diffusion-tensor MRI may be a more sensitive indicator of abnormality in HIV disease.

Creutzfeldt–Jakob disease Sporadic Creutzfeldt–Jakob disease (sCJD) is an acute spongiform encephalopathy that causes a rare, fatal, dementing illness that is believed to be transmitted by a prion agent composed of protease-resistant protein (Johnson and Gibbs, 1998; Mao-Drayer et al., 2002). The clinical course of CJD is characterized

by the onset of rapidly progressive dementia, myoclonus, and periodic synchronous discharge (PSD) on electroencephalograms (EEGs) (Brown et al., 1986; Mao-Drayer et al., 2002). In the early stage of CJD, this clinical triad may not be present, which makes the diagnosis of CJD difficult. Several CSF markers of neurodegeneration have been used to diagnose CJD including neuron-specific enolase, S-100, and/or 14-3-3 protein (Hsich et al., 1996; Zerr et al., 1998; Beaudry et al., 1999). These markers have good sensitivity and specificity but are not always present (Hsich et al., 1996; Zerr et al., 1998; Beaudry et al., 1999). Periodic sharp-wave complex discharges have been described on EEG in up to two-thirds of patients with CJD, but these changes may not develop until late in the disease course (Levy et al., 1986; Steinhoff et al., 1996) and have been reported to disappear as the disease progresses (Aguglia et al., 1997). Routine MRI has not been shown to be useful in the diagnosis of sCJD. Some studies have shown

421

422

Christopher G. Fillipi

(a)

(b)

Fig. 24.14 In this patient with CJD, on the DWI trace image, Figure 24.14(a), hyperintense signal abnormality is seen within the thalami, left greater than right, and right caudate. In Figure 24.14(b), ribbon-like, cortical hyperintense signal abnormality is noted within the right frontal and parietal gyri, which is characteristic of CJD infection. This patient subsequently went to autopsy, which confirmed the presumed clinical diagnosis of CJD.

hyperintense signal abnormality on the long TR sequences within the basal ganglia and cerebral cortex, but often the MR studies are normal or demonstrate non-specific atrophy (Gertz et al., 1988; Yoon et al., 1995; Finkenstaedt et al., 1996; MaoDrayer et al., 2002). Even when these abnormalities are detected, they occur late in the disease course, which limits the diagnostic value of routine MR (cf. Case Study 24.2). Recent reports in the neuroradiological literature have demonstrated increased signal intensity within the basal ganglia and/or cerebral cortex on DWI indicative of restricted diffusion, which have been reported as early as 4 months from the onset of behavioral symptoms (Mao-Drayer et al., 2002; Murata et al., 2002). These progressive hyperintense signal abnormalities within the cerebral cortex and striatum bilaterally on DWI are now thought to represent characteristic imaging finding in patients with CJD (Bahn et al., 1997; Demaerel et al., 1997; Daemaerel et al., 1999; Na et al., 1999; Yee et al., 1999; Murata et al., 2002) (Figure 24.14) In one recent case report, distinctive ribbon-like hyperintensity within the cortex was noted on DWI prior to elevations in CSF 14-3-3 protein levels, which suggests that DWI may be the most sensitive study for the earliest possible detection of CJD (Mao-Drayer et al., 2002). In another recent study, the lesion conspicuity on DWI was either superior to or as good as the FLAIR

sequence, especially for the lesions within the cerebral cortex (Murata et al., 2002). In a recent study by Murata et al. (2002), chronological changes in CJD at DWI were reported. All eight patients who were imaged twice or more had progressive or constant lesion distribution in the striatum and cerebral cortex. In particular, striatal involvement became more bilateral and symmetric with the passage of time, and lesion signal intensity tended to start within anteroinferior portions of the putamen and spread posteriorly to involve the entire putamen (Murata et al., 2002). Findings from experimental studies on prion disease have postulated axonal transsynaptic spread of disease (Fraser and Dickinson, 1985; Heye and Cervos-Navarro, 1992; Taraboulos et al., 1992), which may explain this observation. In the study by Murata et al., the calculated mean diffusion constant on ADC maps was reduced in all lesions measured, and this abnormality in the diffusion constant persisted for greater than 2 weeks, unlike the abnormalities in diffusion that have been reported with acute infarct, which tend to subside within this 2-week timeframe. However, there are other investigations which have reported ADC mapping abnormalities in CJD at 1–2 months from the onset of symptoms in which values have been increased or decreased, so it is not clear whether ADC mapping abnormalities correlate to the stage of clinical disease or disease severity at the

The role of diffusion-weighted imaging in intracranial infection

current time (Bahn and Parchi, 1999; Mao-Drayer et al., 2002).

Conclusions DWI is becoming a routine part of MR imaging (MRI) of the CNS. Qualitative reports, which describe the appearance of intracranial infections on DWI, are well documented in the neuroradiological literature. Many investigators have reported the mean diffusivity values or average diffusion constant of intracranial, infectious lesions from ADC maps, which provide useful diagnostic and quantitative information. The ability to derive quantitative information from an MR pulse sequence marks a significant transition for neuroradiologists, researchers, and the clinicians who order such studies. The diffusion constant or mean diffusivity is not just a quantitative measurement but a physical property of the brain tissue that is being measured and imaged unlike values for T1 and T2, and magnetization transfer (MT), which are MRI-specific properties (Horsfield and Jones, 2002). Diffusion-tensor MRI is an emerging MR modality that allows for the measurement of the amount of anisotropy of water diffusion in tissues and to assess the degree to which directionally ordered tissueslike axons have lost their normal integrity (Ulug et al., 1999). From diffusion-tensor MRI, inferences about the microscopic structural integrity of tissue can be made. This modality has only recently been used to explore HIV in vivo. Diffusion-tensor MR offers great promise in the study of infectious disease in the CNS. The tensor MR sequence allows for a quantitative determination of how infectious agents affect the anisotropy of WM fiber tracts during the course of illness. This information, in turn, may offer clinicians ways to predict patient outcome or monitor disease progression. MR diffusion tractography or axonal fiber tracking, a technique derived from DTI data (Horsfield and Jones, 2002; Mori and van Zijl, 2002), may offer the most promise to researchers and clinicians who study how infectious diseases affect the brain and spinal cord. Sites of axonal injury and the axonal pathways that they subtend may possibly be identified

with the use of MR diffusion tractography. As DWI becomes a part of mainstream diagnostic imaging, the quantitative information that this technique provides from the calculation of the mean diffusivity from ADC maps and the anisotropy from the tensor data sets will be increasing relied upon to diagnose and monitor intracranial disease processes.

REFERENCES Aguglia U, Gambardella A, LePiane E, et al. 1997. Disappearance of periodic sharp wave complexes in Creutzfeldt–Jakob disease. Neurophysiol Clin 27: 277–282. Aronen HJ, Gazit IE, Loius DN, et al. 1994. Cerebral blood volume maps of gliomas: comparison with tumor grade and histologic findings. Radiology 191: 41–51. Ashikaga R, Araki Y, Ishida O. 1996. MR flair imaging of herpes simplex encephalitis. Radiat Med 14: 349–352. Bahn NM, Parchi P. 1999. Abnormal diffusion-weighted magnetic resonance images in Creutzfeldt–Jakob disease. Arch Neurol 56: 577–583. Bahn MM, Kido DK, Lin W, Pearlman AL.1997. Brain magnetic resonance diffusion abnormalities in Creutzfeldt–Jakob disease. Arch Neurol 54: 1411–1415. Beaudry P, Cohen P, Brandel JP, et al. 1999. 14-3-3 Protein, neurospecific enolase, and S-100 protein cerebrospinal fluid of patients with Creutzfeldt–Jakob disease. Dement Geriatr Cogn Disord 10: 40–46. Bencherif B, Rottenberg DA.1998. Neuroimaging and AIDS dementia complex. AIDS 12: 233–244. Benveniste H, Hedlund LW, Johnson GA. 1992. Mechanism of detection of acute cerebral ischemia in rats by diffusionweighted magnetic resonance microscopy. Stroke 23: 746–754. Brem S, Cotran R, Folkman J. 1972. Tumor angiogenesis: a quantitative method for histologic grading. J Natl Cancer Inst 48: 347–356. Brew BJ, Pemberton L, Cunningham P, Law M. 1997. Levels of human immunodeficiency virus type I RNA in cerebrospinal fluid correlate with AIDS dementia stage. J Infect Dis 175: 963–966. Brown P, Cathala F, Castaigne P, Gajdusek DC. 1986. Creutzfeldt– Jakob disease: clinical analysis of a consecutive series of 230 neuropathologically verified cases. Ann Neurol 20: 597–602. Burger P. 1986. Malignant astrocytic neoplasms: classification, pathology, anatomy, and response to therapy. Semin Oncol 13: 16–20. Camacho DLA, Smith JK, Castillo M. 2003. Differentiation of toxoplasmosis and lymphoma in AIDS patients by using apparent diffusion coefficients. Am J Neuroradiol 24: 633–637.

423

424

Christopher G. Fillipi

Caramia F, Aronen HJ, Sorensen G, Belliveau JW, Gonzalez RG, Rosen BR. 1995. Perfusion MR imaging with exogenous contrast agents. In Diffusion and Perfusion Magnetic Resonance Imaging: Applications to Functional MRI (Ed., LeBihan D), Raven Press, New York, pp. 255–265. Castillo M, Mukherji SK. 2000. Diffusion-weighted imaging in the evaluation of intracranial lesions. Sem Ultrasound CT MRI 21: 405–416. Cecil KM, Jones BV, Williams S, Hedlund GL. 2000. CT, MRI, and MRS of Epstein–Barr virus infection: case report. Neuroradiology 8: 619–622. Chan JHM, Tsui EYK, Chau LF, et al. 2002. Discrimination of an infected brain tumor from a cerebral abscess by combined MR perfusion and diffusion imaging. Comput Med Imag Graph 26: 19–23. Chang L, Cornford ME, Chiang FL, Ernst TM, Sun NC, Miller BL. 1995. Radiologic–pathologic correlation: cerebral toxoplasmosis and lymphoma in AIDS. Am J Neuroradiol 16: 1653–1663. Chenevert TL, McKeever PE, Ross BD. 1997. Monitoring early response of experimental brain tumors to therapy using diffusion magnetic resonance imaging. Clin Cancer Res 3: 1457–1466. Chun CH, Johnson JD, Hofstetter M, Raff MJ. 1986. Brain abscess: a study of 45 consecutive cases. Medicine 65: 415–431. Chun T, Filippi CG, Zimmerman RD, Ulug AM. 2000. Diffusion changes in the aging brain. Am J Neuroradiol 21: 1078–1083. Demaerel P, Heiner L, Robberecht W, Sciot R, Wilms G. 1999. Diffusion-weighted MRI in sporadic Creutzfeldt–Jakob disease. Neurology 52: 205–208. Demaerel P, Baert AL, Vanopdenbosch L, Robberecht W, Dom R. 1997. Diffusion-weighted magnetic resonance imaging in Creutzfeldt–Jakob disease. Lancet 349: 847–848. Demaerl P, Wilms G, Robberecht W, et al. 1992. MRI of herpes simplex encephalitis. Neuroradiol 34: 490–493. Desprechins B, Stadnik T, Koerts G, Shabana W, Breucq C, Osteaux M. 1999. Use of diffusion-weighted MR imaging in differential diagnosis between intracerebral necrotic tumors and cerebral abscesses. Am J Neuroradiol 20: 1251–1257. Dina TS. 1991. Primary central nervous system lymphoma versus toxoplasmosis in AIDS. Radiology 179: 823–828. DiStefano M, Monno L, Fiore JR, et al. 1998. Neurological disorders during HIV-1 infection correlate with viral load in cerebrospinal fluid but not with virus phenotype. AIDS 12: 737–743. Ebisu T, Tanaka C, Umeda M, et al. 1996. Discrimination of brain abscess from necrotic or cystic tumors by

diffusion-weighted echo-planar imaging. Magn Reson Imaging 14: 1113–1116. Ellis RJ, Hsia K, Spector SA, et al. 1997. Cerebrospinal fluid human immunodeficiency virus type 1 RNA levels are elevated in neurocognitively impaired individuals with acquired immunodeficiency syndrome. Ann Neurol 42: 679–688. Ernst TM, Chang L, Witt MD, et al. 1998. Cerebral toxoplasmosis and lymphoma in AIDS: perfusion MR imaging experience in 13 patients. Radiology 208: 663–669. Ferrando SJ, van Gorp WG, McElhiney M, Goggin K, Sewell M, Rabkin J. 1998. Highly active antiretroviral treatment (HAART) in HIV infection: benefits for neuropsychological function. AIDS 12: F65–F70. Filippi CG, Sze G, Farber SG, Shamanesh M, Selwyn P. 1998. Regression of HIV encephalopathy and basal ganglia signal intensity abnormality at MR imaging in patients with AIDS after initiation of protease inhibitor therapy. Radiology 206: 491–499. Filippi CG, Ulug AM, Ryan E, Ferrando SJ, vanGorp WG. 2001. Diffusion tensor imaging of patients with HIV and normalappearing white matter on MR images of the brain. Am J Neuroradiol 22: 277–283. Finkenstaedt M, Szudra A, Zerr I, et al. 1996. MR imaging of Creutzfeldt–Jakob disease. Radiology 199: 793–798. Fraser H, Dickinson AG. 1985. Targeting of scrapie lesions and spread via the retino-tectal projection. Brain Res 346: 32–41. Gertz HJ, Henkes H, Cervos-Navarro J. 1988. Creutzfeldt– Jakob disease: correlation of MRI and neuropathologic findings. Neurology 38: 1481–1482. Gisslen M, Hagberg L, Svennerholm B, Norkrans G. 1997. HIV-1 RNA is not detectable in the cerebrospinal fluid during antiretroviral combination therapy. AIDS 11: 1194. Gisslen M, Norkrans G, Svennerholm B, Hagberg L. 1998. HIV-1 RNA detectable with ultrasensitive quantitative polymerase chain reaction in plasma but not in cerebrospinal fluid during combination treatment with zidovudine, lamivudine, and indinavir. AIDS 12: 114–116. Guzman R, Barth A, Lovblad KO, et al. 2002. Use of diffusionweighted magnetic resonance imaging in differentiating prurulent brain processes from cystic brain tumors. J Neurosurg 97: 1101–1107. Haimes AB, Zimmerman RD, Morgello S, et al. 1989. MR imaging of brain abscess. Am J Roentgenol 152: 1073–1085. Hajnal JV, Bryant DJ, Kasuboski L, et al. 1992. Use of fluid attenuated inversion recovery (FLAIR) pulse sequences in MRI of the brain. J Comput Assist Tomogr 16: 841–844. Hartmann M, Jansen O, Heiland S, Sommer C, Munkel K, Sartor K. 2001. Restricted diffusion within ring enhancement is not pathognomonic for brain abscess. Am J Neuroradiol 22: 1738–1742.

The role of diffusion-weighted imaging in intracranial infection

Heaton RK, Grant I, Butters N, et al. 1995. The HNRC 500: neuropsychology of HIV infection at different disease stages. J Int Neuropsychol Soc 3: 231–251. Heiner L, Demaerel P. 2003. Diffusion-weighted MR imaging findings in a patient with herpes simplex encephalitis. Eur J Radiol 45: 195–198. Heye N, Cervos-Navarro J. 1992. Focal involvement and lateralization in Creutzfeldt–Jakob disease: correlation of clinical, electroencephalographic, and neuropathological findings. Eur Neurol 32: 289–292. Horsfield MA, Jones DK. 2002. Applications of diffusionweighted and diffusion tensor MRI to white matter diseases – a review. NMR Biomed 15: 570–577. Hossman KA, Fischer M, Bockhorst K, Heohn-Berlage M. 1994. NMR imaging of the apparent diffusion coefficient (ADC) for the evaluation of metabolic suppression and recovery after prolonged cerebral ischemia. J Cereb Blood Flow Metab 14: 723–731. Hsich G, Kenney K, Gibbs Jr CJ, Lee KH, Harrington MG. 1996. The 14-3-3 brain protein in cerebrospinal fluid as a marker for transmissible spongiform encephalopathies. New Engl J Med 335: 924–930. Johnson RT, Gibbs CJ. 1998. Creutzfeldt–Jakob disease and related transmissible spongiform encephalopathies. New Engl J Med 339: 1994–2004. Kim YJ, Chang K, Song IC, et al. 1998. Brain abscess and necrotic or cystic brain tumor: discrimination with signal intensity on diffusion-weighted MR imaging. Am J Roentgenol 171: 1487–1490. Knopp EA, Cha S, Johnson G, et al. 1999. Glial neoplasms: dynamic contrast-enhanced T 2*-weighted MR imaging. Radiology 211: 791–798. Krabbe K, Gideon P, Waga P, Hansen U, Thomsen C, Madsen F. 1997. MR diffusion imaging of human intracranial tumors. Neuroradiology 39: 483–489. Kure K, Llena JF, Lyman WD, et al. 1991. Human immunodeficiency virus-I infection of the nervous system: an autopsy study of 268 adult, pediatric, and fetal brains. Hum Pathol 22: 700–710. LeBihan D. 1991. Molecular diffusion nuclear magnetic resonance imaging. Magn Reson 7: 1–30. Levy SR, Chiappa KH, Burke CJ, Young RR. 1986. Early evolution and incidence of electroencephalographic abnormalities in Creutzfeldt–Jakob disease. J Clin Neurophysiol 3: 1–21. Lovblad KO, Laubach HJ, Baird AE, et al. 1998. Clinical experience with diffusion-weighted MR in patients with acute stroke. Am J Neuroradiol 19: 1061–1066. Maeda M, Kawamura, Tamagawa Y, et al. 1992. Intra voxel incoherent motion (IVIM) MRI in intracranial, extracranial tumors and cysts. J Comput Assist Tomogr 16: 514–518.

Mamidi A, DeSimone JA, Pomerantz RJ. 2002. Central nervous system infections in individuals with HIV-1 infection. J Neurovirol 8: 158–167. Mao-Drayer Y, Braff SP, Nagle KJ, Pendlebury W, Penar PL, Shapiro RE. 2002. Emerging patterns of diffusion-weighted MR imaging in Creutzfeldt–Jakob disease: case report and review of the literature. Am J Neuroradiol 23: 550–556. McArthur JC, McClernon DR, Cronin MF, et al. 1997. Relationship between human immunodeficiency virusassociated dementia and viral load in cerebrospinal fluid and brain. Ann Neurol 42: 689–698. Monbati A, Kumar P, Kamkarpour A. 2000. Intraoperative cytodiagnosis of metastatic brain tumors confused clinically with brain abscess: a report of three cases. Acta Cytol 44: 437–441. Mori S, van Zijl PCM. 2002. Fiber tracking: priniciples and strategies – a technical review. NMR Biomed 15: 468–480. Murata T, Shiga Y, Higano S, Takahashi S, Mugikura S. 2002. Conspicuity and evolution of lesions in Creutzfeldt–Jakob disease at diffusion-weighted imaging. Am J Neuroradiol 23: 1164–1172. Na DL, Suh CK, Choi SH, et al. 1999. Diffusion-weighted magnetic resonance imaging in probable Creutzfeldt–Jakob disease: a clinical-anatomic correlation. Arch Neurol 56: 951–957. Noguchi K, Watanabe N, Nagayoshi T, et al. 1999. Role of diffusion-weighted echo-planar MRI in distinguishing between brain abscess and tumor: a preliminary report. Neuroradiology 41: 171–174. Osborn AG. 1994. Pyogenic parenchymal infections. In Diagnostic Neuroradiology (Ed., Osborn AG), Mosby Yearbook, St. Louis, MO, pp. 688–692. Pezzullo JA, Tung GA, Mudigonda S, Rogg JM. 2003. Diffusionweighted MR imaging of pyogenic ventriculitis. Am J Roentgenol 180: 71–75. Pomara N, Crandall DT, Choi SJ, Johnson G, Lim KO. 2001. White matter abnormalities in HIV-1 infection: a diffusion tensor imaging study. Psychiat Res Neuroimag 106: 15–24. Price RW, Strapans S. 1997. Measuring the viral load in cerebrospinal fluid in human immunodeficiency virus infection: window into brain infection. Ann Neurol 42: 675–678. Ramsey RG, Gean AD. 1997. Central nervous system toxoplasmosis. Neuroimaging Clin N Am 7: 171–186. Sener RN. 2001. Herpes simplex encephalitis: diffusion MR imaging findings. Comput Med Imaging Graph 25: 391–397. Simpson DM, Berger JR. 1996. Neurological manifestations of HIV infection. Med Clin North Am 80: 1363–1394. Singer MB, Atlas SW, Drayer BP. 1998. Subarachnoid space disease: diagnosis with fluid-attenuated inversion-recovery MR imaging and comparison with gadolinium-enhanced spin-echo MR imaging-blinded reader study. Am J Neuroradiol 208: 417–422.

425

426

Christopher G. Fillipi

Skoldenberg B. 1996. Herpes simplex encephalitis. Scand J Infect Dis Suppl 100: 8–13. Smirniotopoulos JG, Koeller KK, Nelson AM, Murphy FM. 1997. Neuropathology-autopsy correlations in AIDS. Neuroimaging Clinics N Am 7: 615–637. Stadnick TW, Chaskis C, Michotte A, et al. 2001. Diffusionweighted MR imaging of intracerebral masses: comparison with conventional MR imaging and histologic findings. Am J Neuroradiol 22: 969–976. Steinhoff BJ, Racker S, Herrendorf G, et al. 1996. Accuracy and reliability of periodic sharp wave complexes in Creutzfeldt– Jakob disease. Arch Neurol 53: 162–166. Taraboulos A, Jendroska K, Serban D, Yang SL, DeArmond SJ, Prusnier SB. 1992. Regional mapping of prion proteins in brain. Proc Nat Acad Sci USA 7620–7624. Tien RD, Felsberg GJ, Osumi AK. 1993. Herpes virus infection of the CNS. Am J Roentgenol 161: 167–176. Tien RD, Felsberg GJ, Friedman H, Brown M, MacFall J. 1994. MR imaging of high-grade cerebral gliomas: value of diffusionweighted echoplanar pulse sequences. Am J Roentgenol 162: 671–677. Tsuchiya K, Yamakami N, Hachiya J, Saito I, Kobayashi H. 1998. Multiple brain abscesses: differentiation from cerebral metastases by diffusion-weighted magnetic resonance imaging. Int J Neuroradiol 4: 258–262. Tsui EYK, Chan JH, Cheung YK, Lai KF, Fong D, Ng SH. 2002. Evaluation of cerebral abscesses by diffusion-weighted MR imaging and MR spectroscopy. Comput Med Imag Grap 26: 347–351. Tsuruda JS, Chew WM, Moseley ME, Norman D. 1990. Diffusion-weighted MR imaging of the brain: value of differentiating between extraaxial cysts and epidermoid tumors. Am J Neuroradiol 11: 925–934. Tung GA, Evangelista P, Rogg JM, Duncan JA. 2001. Diffusionweighted MR imaging of rim-enhancing brain masses: is

markedly decreased water diffusion specific for brain abscess? Am J Roentgenol 177: 709–712. Ulug AM, Moore DF, Bojko AS, Zimmerman RD. 1999. Clinical use of diffusion tensor imaging for diseases causing neuronal and axonal damage. Am J Neuroradiol 20: 1044–1048. van Gorp WG, Baerwald JP, Ferrando SJ, McElhiney MC, Rabkin JG. 1999a. The relationship between employment and neuropsychological impairment in HIV infection. J Int Neuropsychol Soc 5: 534–539. van Gorp WG, Mandelkern MA, Gee M, et al. 1999b. Cerebral metabolic dysfunction in AIDS: findings in a sample with dementia and without dementia. J Neuropsychiatry Clin Neurosci 4: 280–287. Wenz F, Rempp K, Hess T, et al. 1996. Effect of radiation on blood volume in low-grade astrocytomas and normal brain tissue quantification with dynamic susceptibility contrast MR imaging. 166: 187–193. Winn Jr WC, Kissane JM. 1996. Bacterial disease. In Anderson’s Pathology 10th edn. (Eds., Damjanov I, Linder J), St. Louis, Mosby, pp. 747–842. Wispelwey B, Decay Jr RG, Scheld WM. 1991. Brain abscess. In Infection of the Central Nervous System (Eds., Scheld WM, Whitley RJ, Durack DT), Raven, New York, pp. 457–458. Yee AS, Simon JH, Anderson CA, Sze CI, Filley CM. 1999. Diffusion-weighted MRI of right-hemisphere dysfunction in Creutzfeldt–Jakob disease. Neurology 52: 1514–1515. Yoon SS, Chan S, Chin S, Lee K, Goodman RR. 1995. MRI of Creutzfeldt–Jakob disease: asymmetric high signal intensity of the basal ganglia. Neurology 45: 1932–1933. Zerr I, Bodemer M, Gefeller O, et al. 1998. Detection of 14-3-3 protein in the cerebrospinal fluid supports the diagnosis of Creutzfeldt–Jakob disease. Ann Neurol 43: 32–40.

The role of diffusion-weighted imaging in intracranial infection

Case Study 24.1 West Nile encephalitis Deepak Takhtani, M.D., Johns Hopkins University School of Medicine, Baltimore History A 55-year-old female, a recent transplant recipient, with chills and gradually worsening neurological symptoms.

FLAIR

Technique Conventional MRI and diffusion-weighted MRI, including calculation of ADC map.

Imaging findings FLAIR images show involvement of the basal ganglia, thalami and the cerebellum. Diffusion and ADC map at the level of the midbrain shows restricted diffusion in the red nuclei.

DWI

ADC

Discussion West nile virus (WNV) infection is an arthropodborne contagious disease. MRI in WNV infection shows progressive involvement of caudate nuclei and thalami to the brain stem and cerebellar hemispheres. The significance of substantia nigra and red nucleus involvement is uncertain but could be a clue to support this diagnosis.

Encephalitis causes a cytotoxic as well as vasogenic edema. The differential diagnosis based on MRI findings would include Japanese encephalitis, acute disseminated encephalomyelitis (ADEM), or Wernicke encephalopathy. Japanese encephalitis or Eastern equine encephalitis can have identical MRI findings as in this case. Wernicke encephalopathy involves mamillary bodies and the thalami. ADEM is more asymmetric and has characteristic monophasic lesions in the WM. For definitive diagnosis, serum or CSF laboratory tests may be required to help distinguish WNV from other encephalitides. In this patient transplant was the possible source of the infection.

Key points MRI findings in encephalitis are non-specific. The lesions may show restricted diffusion. Reference Humberto Rosas, Franz J, Wippold II. 2003. West nile virus: case report with MR imaging findings. Am J Neuroradiol 24: 1376–1378.

427

428

Christopher G. Fillipi

Case Study 24.2 Creutzfeldt–Jakob disease: DWI Adam Waldman, M.D., Ph.D. and Peter Barker, D.Phil. Hammersmith Hospitals and Institute of Neurology, London and Johns Hopkins University School of Medicine, Baltimore History

• A 58-year-old male with sCJD. • A 38-year-old male with variant CJD (vCJD).

FLAIR

DWI

ADC

Technique Conventional MRI and DWI (b  1000 s/mm2).

(a) sCJD Imaging findings (a) sCJD, FLAIR and DWI showed symmetric bilateral hyperintensity in the caudate nuclei, putamen and less marked in the thalami (pulvinar). ADC was reduced in these regions. (b) vCJD, FLAIR images were degraded by motion artifact. DWI showed high signal in the thalami (pulvinar and diencephalon), and asymmetrically in the caudate, putamen. ADC (not shown) was restricted in the caudate and putamen but increased in the pulvinar.

(b) vCJD Key points

Discussion Signal abnormality more prominent in the thalami than basal ganglia is the “pulvinar sign”. DWI abnormalities may be dominated by T2 effects, so examination of both DWI and ADC maps can be helpful; ADC may change with disease progression and may be higher or lower than normal. EPI sequences used for DWI allow rapid image acquisition, which minimizes image degradation due to motion.

References

DWI is more sensitive than FLAIR and T 2-weighted imaging in CJD diagnosis. The “pulvinar sign” is characteristic of variant CJD. ADC may be high or low. Rapid EPI sequences are helpful for agitated or confused patients. Cortical DWI hyperintensity may be seen in sCJD and vCJD.

Collie DA, Summers DM, Sellar RJ, Ironside, et al. 2003. Diagnosing variant Creutzfeldt–Jakob disease with the pulvinar sign: MR imaging findings in 86 neuropathologically confirmed cases. Am J Neuroradiol 24: 1560–1569. Demaerel P, Heiner L, Robberecht W, Sciot R, Wilms G. 1999. Diffusion-weighted MRI in sporadic Creutzfeldt–Jakob disease. Neurology 52: 205–208. Waldman AD, Jarman P, Merry RTG. 2003. Rapid echoplanar diffusion imaging in a case of variant CJD; where speed is of the essence. Neuroradiology 4: 528–531.

25

MR spectroscopy in demyelination and inflammation Gioacchino Tedeschi and Simona Bonavita Second Department of Neurology, Second University of Naples, Naples, Italy

Key points • MR spectroscopy (MRS) is a sensitive technique for evaluating axonal damage (decreased N-acetylaspartate, NAA) and demyelination (increased Choline (Cho), myo-inositol (mI)) in multiple sclerosis (MS). • Acute MS plaques usually show decreased NAA (and decreased Creatine (Cr) in large plaques) and increased Cho and lactate (Lac) and, in short echo time spectra, increased mI and lipids. • Spectra from acute MS plaques (“tumefactive demyelianting lesions”) may be similar to those of neoplasms (elevated Cho, Lac, decreased NAA). • As plaques resolve, Cr and Lac return quickly to normal, while Cho and lipids need months to return to normal. NAA may or may not recover to normal. • Metabolic changes (reduced NAA) can be seen in white matter (WM) with normal conventional T2 MRI appearance (NAWM); Reductions of NAA in NAWM correlate with clinicial disability. • NAA may also be reduced in gray matter in MS. • Monophasic acute disseminated encephalomyelitis with good clinical outcome generally shows mild, reversible NAA reductions without changes in other metabolites. • MRS may be helpful in the diagnosis of other WM diseases (mucolipidosis type IV, cerebral autosomal dominant arteriopathy

with subcortical infarcts and leukoencephalopathy, cerebrotendinous xanthomatosis, Sjogren–Larsson syndrome, Salla disease, leukodystrophy with ovarian dysgenesis).

Introduction In the first section of this book, the reader has already been introduced to the ways that MR spectroscopy (MRS) furnishes in vivo metabolic information in central nervous system (CNS) diseases. This chapter addresses the application of proton MRS to the study of acquired demyelinating and inflammatory disorders of the CNS. Pediatric disorders associated with inborn errors of metabolism and other dysmyelinating and white matter (WM) diseases are largely covered in Chapters 44 and 45, but some specific disorders are also discussed in this chapter. Although the technical and methodological principles of MRS have already been covered in previous chapters, we believe that the reader will appreciate a brief comment on the specific spectroscopy issues in the study of demyelinating and inflammatory disorders of the CNS. These are related to the features of these disorders, as well as the principles and limitations of MRS. First of all, we should consider how disease related metabolic abnormalities may be explained by in vivo metabolic information obtainable with MRS. As we will see, even multiple sclerosis (MS), the prototype 429

430

Gioacchino Tedeschi and Simona Bonavita

of demyelinating disorders, in addition to the typical demyelination, is characterized by other histopathological features like neuronal damage, remyelination, inflammation, and gliosis (Peterson et al., 2001). As we already know, choline (Cho) signal is generated by a group of molecules, mainly glyceropho Cho and phosphocholine (PC), involved with phospholipid and membrane biochemistry, which may be increased in myelin breakdown (demyelination), as well as in increased turnover of membrane (remyelination). A second matter is the inherent spatial resolution of MRS; the lower practical limit of approximately a 1 cm3 voxel size may often be too large to assess tiny lesions or small CNS structures. The choice of single voxel or MR spectroscopic imaging (MRSI) technique depends on the disease one is dealing with, and on the questions one is trying to answer. Both MRS and MRSI have advantages and disadvantages. The first one is easier to perform in a quantitative way, thus providing a challenge for higher sensitivity to study single-focal lesions, whereas it may have limited value for diffuse CNS disorders. On the contrary, MRSI is mainly based on the use of metabolites ratios, and may have less biochemical sensitivity, whereas it allows a more exhaustive approach to diffuse CNS disorders, as well as to large and more heterogeneous lesions. In the following part of this chapter we will review the evidence for the role of MRS in the study of demyelinating and inflammatory disorders of the CNS. The major contribution of MRS has been mainly focused on the understanding of disease mechanisms, and has opened new perspective in the management of the diseases. So far, MRS has been largely appreciated in clinical research settings, while the direct clinical applications of MRS are still limited. We anticipate that the ongoing technical improvements and developments in therapy will soon increase its impact on clinical practise.

Multiple sclerosis MS is a complex CNS disorder, whose pathological hallmark has long been thought to be the occurrence

of multiple foci of inflammation and demyelination within the WM of the CNS. Although evidence of axonal pathology in and around the lesions has been recognized since Charcot’s time (Charcot, 1968), only recently, histopathological studies suggested that in lesions destructive or degenerative changes with axonal loss are relatively common, and that the extent of neuronal pathology in both WM and gray matter (GM) is potentially substantial. According to the disease course MS has been codified in four types (Lublin and Reingold, 1996). The relapsing–remitting (RR) type is characterized by disease relapses with fully recovery or with sequelae and residual deficit upon recovery and no progression between relapses. The primary-progressive (PP) type is characterized by disease progression from onset with no acute exacerbations, occasional plateaus and possible temporary but minor improvements. The secondary-progressive (SP) type is characterized by steady disease progression with or without occasional superimposed relapses, minor remissions and plateaus following period of clearly defined RR disease, and between relapses steady progression of disability is seen. The progressive relapsing type is characterized by disease progression from onset but with clear relapses with or without full recovery during the disease course. Moreover according to the severity of the disease two types of MS have been defined (Lublin and Reingold, 1996): benign MS, with patients remaining fully functional after 15 years from disease onset; and the malignant MS, with rapid progressive course leading to significant disability or death within a relatively short time after disease onset. Although MR imaging (MRI) is the most sensitive tool to support the diagnosis of MS, it has limitations in distinguishing the histopathological heterogeneity of MS plaques, and in detecting WM involvement beyond them. In the last few years, MRS, due to its greater pathological specificity, has played a very important role in the study of MS, and has led the scientific community to become familiar with the now well-accepted concepts of axonal pathology, diffuse WM involvement, lesion heterogeneity, and possible GM involvement. MS has been one of the major clinical research applications for MRS and MRSI, and it would have been impossible to report in details even a short selection of the many relevant

MR spectroscopy in demyelination and inflammation

papers that have appeared in the last years. For the sake of clarity and brevity, we will summarize the major findings and we will show a few images to depict them. MRS can complement conventional MRI in the assessment of patients with MS by simultaneously defining biochemical correlates of the pathological changes occurring within and outside the lesions. Changes in metabolites have been described in MS plaques, both in acute and chronic stage, as well as in normal appearing white matter (NAWM) on conventional MRI and GM. In acute lesions, MRS can reveal an increase of Cho, which may reflect pathological processes characterized by increased membrane breakdown (demyelination), and lactate (Lac), reflecting macrophage infiltration, resulting from the inflammatory phase or mitochondrial dysfunction (Kapeller et al., 2001). In large acute demyelinating lesions a decrease of creatine (Cr) signal can also occur. As Cr is mainly concentrated in glial cells, a reduction of its signal in the phase of demyelination can reflect an impairment in olygodendrocytes metabolism (Davie et al., 1994; De Stefano et al., 1995a). Short echo time (TE) studies have reported evidences for transient increase in lipids and myo-inositol (mI) peaks, possibly released during myelin breakdown (Davie et al., 1994; Narayana et al., 1998). All these changes are usually followed by a decrease (Kapeller et al., 2001) of the neuronal marker N-acetyl-aspartate (NAA), which is presumably due to secondary axonal dysfunction (Figure 25.1). The heterogeneity of the metabolic changes is in good agreement with the histopathological heterogeneity of the MS lesions, showing the possible coexistence of inflammation, demyelination, remyelination, axonal damage, and macrophage infiltration. It is important to note that the spectroscopic changes seen in acute MS plaques are often very similar to the spectra observed in high grade brain tumors (high Cho, low NAA, increased Lac, etc.), and therefore this should be kept in mind when evaluating spectra from patients with undiagnosed brain lesions. In particular, a patient with a tumefactive MS plaque can easily be misdiagnosed as astrocytoma or glioblastoma multiforme (GBM) on the basis of the spectral pattern of the lesion.

After the acute phase, usually over a period ranging from days to weeks, Cr and Lac signals tend to return to normal levels, while the Cho and lipids signals need months to return to basal levels. At the same time, the NAA signal can increase and reach almost normal levels or can remain decreased (Davie et al., 1994; Arnold et al., 1992; De Stefano et al., 1995b). The reversible decrease in NAA was initially thought to reflect axonal loss from Wallerian degeneration, while it is now clear that it is related to reversible axonal dysfunction and/or damage that may occur in MS. Moreover a number of evidences have shown that the reversible axonal damage and/or dysfunction is correlated with reversible functional impairment. Possible explanations for the initial fall in NAA levels, followed by a subsequent rise, include the resolution of edema, a transient impairment of mitochondrial NAA production, the migration to the lesion or local proliferation of adult oligodendrocyte type II astrocyte progenitor cells, which also contain NAA. A number of studies correlating MRS and immunopathological findings from brain MS lesions have confirmed that the decrease of NAA signal correlates with axonal loss, while the increase of Cho signal correlates with active demyelination and/or gliosis (Bitsch et al., 1999). The NAA signal is a reliable marker of neuron and/or axonal pathology, in fact it is particularly reduced in T1 hypointense lesions which are known to be plaques with severe axonal loss (van Walderveen et al., 1998, 1999). Another major contribution of MRS has been in the study of the so-called NAWM (Arnold, 1999). A decrease of NAA signal has been found in NAWM adjacent to or distant from lesions (Figure 25.2). The possible explanations for this finding may be either the dysfunction of axons transversing macroscopic MS lesions, as shown by the finding of reversible changes of NAA in the NAWM of the hemisphere contra lateral to solitary acute MS lesion (Figure 25.3), or microscopic focal abnormalities which may fall below the detection power of conventional MRI (De Stefano et al., 1999). This can be the explanation of the reduced NAA levels in NAWM of patients with PP MS despite the paucity of T2 lesion load (Leary et al., 1999). Spectroscopic abnormalities in the NAWM, may also include an increase of Cr signal, which has been interpreted as suggestive of ongoing

431

(a) NAA

Cho mI

Cr

3

WM

2

1

0

Frequency (ppm)

NAA Cr Cho mI

3

Cortical GM

2

1

0

Frequency (ppm)

NAA

(b)

Cho mI

Cr

3

Cortical GM

2

1

0

Frequency (ppm)

NAA Cho Cr mI

3

2

1

0

Frequency (ppm)

Normal appearing WM Cho

NAA

mI Cr Lipids

3

2

1

0

Frequency (ppm)

Lesion

Fig. 25.1 (a) T2-weighted image with overlaid short TE MRSI grid in a normal control subject and examples of a WM and GM spectrum. (b) T2-weighted image of an MS patient with overlaid MRSI grid and examples of spectra from NAWM, lesion and cortical GM. Compared to the normal spectra showed in (a), the decrease of NAA is visible in all three MS spectra. The MI peak is more prominent in the NAWM and lesion voxel of the MS patient than in the control WM spectrum. (From Kapeller et al. (2001), with permission.)

MR spectroscopy in demyelination and inflammation

NAA

Cho Cr

3.0

Signal intensity (arbitrary unit)

(b)

Signal intensity (arbitrary unit)

(a)

(1) WM

2.0

1.0

0.0

Chemical shift (ppm)

NAA

Cho Cr

(1) NAWM

(2) Lesion

3.0

2.0

1.0

0.0

Chemical shift (ppm)

Fig. 25.2 T2-weighted MRI (a) and proton spectra (b) from a normal control subject (left) and an MS patient (right). The spectra were selected from the voxels indicated in the MRI (1 in the normal control, 1 and 2 in the MS patient). The NAA/Cr ratio is lower in voxels within the NAWM of MS patients than in those from normal WM of the control subjects, and even lower in voxels from lesions in the patients. (From Fu et al. (1998), with permission.)

gliosis, possibly reactive to inflammatory sub-clinical processes (Tedeschi et al., 2002). On the other hand, serial MRSI studies have identified spectroscopic changes in NAWM (regional increases of Cho and mobile lipids and decreases of NAA), suggestive of ongoing demyelination and axonal dysfunction, that can precede by several months the appearance of MRI visible abnormalities (Narayana et al., 1998; Tartaglia et al., 2002). A number of papers have reported metabolite changes in cortical GM of MS patients (Sharma et al., 2001; Sarchielli et al., 2002), even in the early phases of the disease. A reduction of NAA has also been found in the thalamus (Figure 25.4) of MS patients (Cifelli et al., 2002). In patients with SP MS there is an inverse correlation between NAA levels and the development of

brain atrophy (Coles et al., 1999), which has not been found in patients with RR clinical forms. It is therefore clear that MS affects the CNS in a more diffuse way, and in particular that GM is involved in the disease. This has a direct clinical impact, as it supports the frequent finding of cognitive deficits in MS patients. A number of studies focused on the relationship between disability and MRS data (De Stefano et al., 1998). The first relevant, and somehow expected, finding is that disability is correlated with the reduction of NAA signal. The second, perhaps less expected finding, is that disability is more dependent on the reduction of NAA in NAWM than in the lesions visible on conventional MRI sequences. It is possible that, because NAWM represents the greatest bulk of WM, although axonal loss and/or damage are less severe in

433

434

Gioacchino Tedeschi and Simona Bonavita

NAA

NAA NAA

Cho

3.78

Cr

ppm

Cho

Cho Cr

0.75 3.78

ppm

0.75

3.78

Cr

ppm

0.75

Fig. 25.3 MRI and MRSI of a MS patient performed during the acute phase (left), 1 month later (center) and 6 months later (right). MRIs show a large, solitary, demyelinating lesion that decreases in size over time (top panels). Spectroscopic volume of interests (VOIs) are shown by the dotted line in each transverse MRI. Averaged spectra from voxels located in NAWM contralateral and homologous to the demyelinating lesion (small squares in top panels) are shown in the bottom panels. Note the significant decrease in the NAA/Cr resonance intensity 1 month after the acute phase of the disease (bottom center) and its complete recovery by 6 months (bottom right). (From De Stefano et al. (1999), with permission.)

NAWM than in individual lesions, axonal loss and/or damage in the NAWM may be proportionally more significant for clinical disability. Furthermore, there is now consensus on the hypothesis that the progression of the disease, up to a non-return stage, is dependent on the cumulative effect of the axonal damage, which may ultimately result in an MRI visible brain atrophy. Another important finding is the correlation between MRS data and the clinical form of the disease (Brex et al., 1999; Tourbah et al., 1999; Sarchielli et al., 1999). The reduction of NAA has shown significant group differences between RR MS and SP MS patients, with the latter having more neuronal damage (Figure 25.5(a) and (b)). Once again, it appears that the metabolic abnormalities of the NAWM are more

relevant than the ones of the lesions, suggesting that the accumulation of microscopic pathology in the NAWM is of particular relevance for determining MS clinical evolution. In fact, in NAWM, the reduction of NAA varies from a substantial normality of NAA signal in patients with a clinically isolated syndrome to a severe decrease of NAA levels in the chronic progressive forms. The new approach of the whole brain spectroscopy, measures total NAA level in the whole brain, and the correlation of global axonal pathology with the clinical status (including disability and disease course) may become easier to detect. All this evidence suggests that NAA measurements may be useful, especially in a research setting, to monitor disease evolution and the effectiveness of therapy.

MR spectroscopy in demyelination and inflammation

(a)

(c)

(b)

NAA Cr Cho

A

B

Fig. 25. 4 Top (a) Typical VOIs selected within thalamic region (using coronal plane), and frontal WM, in the axial plane (b) and in the coronal plane (c). Bottom: spectra from thalamic regions of a control subject (A) and a patient with RR MS (B). The NAA peak is lower in the patient than in the control, while no differences can be appreciated in Cr and Cho. (From Wylezinska et al. (2003), with permission.)

(a) A

NAA

C

NAA

B 1.57875

2.02446

Cho Cr Cho

Cr mI GLX Lip  aa

0.00160928

0.0806237

2 ppm

0

2

0 ppm

Fig. 25.5 (a) Normal spectra at TE of 136 ms (A), 18 ms (B) from a VOI (C) in the left centrum semiovale. Glx: glutamine  glutamate; Lip  aa: lipids and amino acids; mI: myoinositol. (b) FLAIR sequence (A) showing the volume of interest (VOI) in a periventricular posterior unenhanced lesion on a T1-weighted image (B). Average lesion spectra (each point of an average spectrum is the average of the values of each individual normalized spectrum, where the Cho amplitude equals unity) at both TE in patients with RR (C and D) and SP (E and F) disease. Compared with control spectra: at TE of 18 ms, resonances at 0.9 and 1.3 ppm are increased in patients with RR MS (C) and SP MS (E), while mI is increased only in SP patients (E); at TE of 136 ms, NAA is decreased and Cho is increased (D and F), more severely in patients with SP MS (F). (From Tourbah et al. (1999), with permission.)

435

436

Gioacchino Tedeschi and Simona Bonavita

(b)

A

B

TE  136 ms

TE  15 ms C

D

ppm

ppm F

E

ppm

ppm

Fig. 25.5(b) (cont.)

A number of MRS studies showed that axonal damage may happen early in the course of the disease. Indirect evidence of early axonal damage derives from the observation that the reduction of NAA seems to correlate with the patients’ clinical status, as the higher is the level of disability, the faster

the drop in NAA (De Stefano et al., 2001a). The discovery that metabolic abnormalities can be found early in the course of the disease has had a relevant impact in the clinical field, and it is modifying the therapeutic approach to MS patients. The advent of effective disease-modifying therapies lend greater

MR spectroscopy in demyelination and inflammation

impetus to the accurate early diagnosis of MS, and surrogate markers for therapeutic monitoring. In summary, MRS and MRSI have redefined MS as a diffuse condition of the WM and GM of the CNS, in which metabolic abnormalities may be present early in the course of the disease.

Acute disseminated encephalomyelitis Acute disseminated encephalomyelitis (ADEM) is an inflammatory demyelinating disease that is often characterized by multiple, diffuse lesions through the CNS on T2 MRI. Although the differential diagnosis is large, ADEM is most frequently compared to MS. The two are usually distinguished from one another by multiple factors, including age of onset, presence or absence of a preceding infection or immunization and evidence for intrathecal immunoglobulin synthesis. No MRI feature can distinguish the two reliably (Kesselring et al., 1990). Early differentiation of the disorders is important, as prognosis and treatment may differ. A diagnosis of ADEM generally predicts that the disease will be monophasic, or, if multiphasic, any relapses will be limited in number and eventually remit, whereas a diagnosis of MS generally carries a poor long-term prognosis. There have been relatively few reports of MRS in ADEM. Recently, proton MRSI in a case (cf. Case Study 25.1) of ADEM (Bizzi et al., 2001) a pattern of reduced-NAA and normal levels of other metabolites. On follow-up the lesions resolved and NAA returned to normal. In recurrent ADEM, however, with more severe involvement, additional changes such as increased Cho and Lac may also be observed, and the lesions may not be completely reversible. Other conditions involving demyelination or dysmyelination The majority of these pathologies are covered in Chapters 44 and 45, however, some specific disorders are described below. Mucolipidosis type IV Mucolipidosis type IV is an autosomal-recessive disorder characterized by abnormalities involving mainly the eye and brain. CNS involvement is

characterized by mental retardation, with severe motor and speech impairment. Conventional MRI shows dysmyelination with thinning of the corpus callosum and cerebellar atrophy. Spectroscopic findings show a diffuse decrease of the NAA correlating with neurological status since NAA signal was lower in more motor impaired patients (Bonavita et al., 2003). Moreover, the relative sparing of mainly sensory areas, such as thalamus and parietal GM, is in good agreement with the sparing of sensory function in these patients. CADASIL Cerebral autosomal-dominant arteriopathy with subcortical infarcts and leukoencephalopathy (CADASIL) is a hereditary syndrome. Its radiological hallmarks are subcortical WM hyperintensities (WMH) and small cystic lesions. MRS showed a diffuse decrease in cerebral NAA, indicating the presence of widespread axonal damage and suggesting that axonal damage may be an early finding of the disease. MRSI studies in CADASIL patients showed a significant reduction of NAA, Cho and Cr in WMH and NAWM (Auer et al., 2001). Metabolite changes were more prominent in more severely affected subjects. These findings suggest axonal injury, myelin loss, and gliosis. The neuroaxonal abnormalities may reflect structural damage or functional neuronal impairment secondary to WM pathology. NAA reductions were correlated with clinical disability, emphasizing the clinicopathological relevance of axonal injury in CADASIL. Cerebrotendinous xanthomatosis Cerebrotendinous xanthomatosis (CTX) is a rare disorder due to inherited defect in the metabolic pathway of cholesterol. Early diagnosis of the disease is particularly important as patients benefit from therapy with chenodeoxycholic acid and HMG-CoA reductase inhibitors (statins). Although the disease is characterized by the concomitant presence of tendon xanthomas, juvenile cataracts, and progressive neurological impairment, these clinical features may vary greatly. Neuroradiological studies have suggested that the bilateral abnormality of the dentate nuclei could be typical of this disease. However, this finding has been seen inconsistently on conventional MRI.

437

438

Gioacchino Tedeschi and Simona Bonavita

(a)

(c)

(b)

NAA NAA

Cho Cho

Cr

Cr

La La

3.6 3.2 2.8 2.4 2.0 1.6 1.2 0.8 ppm

3.6 3.2 2.8 2.4 2.0 1.6 1.2 0.8 ppm

Fig. 25.6 Conventional MRI scan of a patient with CTX in sagittal and transverse orientation, illustrating the VOI used for spectroscopy (a) and the spectrum relative to the VOI (b). The proton spectrum of a normal control (originating from a similar VOI) is shown for comparison (c). Note the decrease in the NAA/Cr ratio and the increase in Lac/Cr in the patient spectrum. From De Stefano et al. 2001b. Brain 124: 121–131, with permission.

The CNS pathology is complex, and whether demyelination or axonopathy has primary importance in the pathogenesis is not known. On conventional MRIs, bilateral hyperintensities of the dentate nuclei, cerebellar and cerebral WM and pyramidal tracts are evident using a FLAIR or T2 sequence. MRS studies showed significant decrease in NAA and increase in Lac signals in large regions of interests (ROIs) (Figure 25.6) localized above the lateral brain ventricles and in the cerebellar hemispheres (De Stefano et al., 2001b). Moreover, cerebral values of NAA intensities show close correlation with patients’ disability. MRS data suggest widespread axonal damage (as revealed by the decrease in NAA) and diffuse brain mitochondrial dysfunction (as revealed by the increase in Lac). The close correlation between values of the putative axonal marker NAA and patients’ disability scores, suggests that MRS can provide a useful measure of disease outcome.

Sjogren–Larsson syndrome Sjogren–Larsson syndrome (SLS) is an autosomal recessively inherited neurocutaneous disorder caused by a deficiency of the microsomal enzyme fatty aldehyde dehydrogenase. The genetic mutation on chromosome 17, responsible for this deficiency, has been identified. SLS was originally defined as a clinical triad consisting of ichthyosis, spastic di- or tetraplegia, and mental retardation. MRI may show an arrest of myelination, periventricular signal abnormalities of WM and mild ventricular enlargement. MRS reveals a characteristic, abnormal lipid peak even before the stage of MRI visible dysmyelination (Willemsen et al., 2001). MRS findings suggest an accumulation of long-chain fatty alcohol intermediates, resulting in retarded myelination and dysmyelination. The degree of WM abnormality in the MRIs, and the spectroscopic abnormalities do not correlate with the severity of the neurological status.

MR spectroscopy in demyelination and inflammation

(b)

(a)

(c)

(d)

NAA

NAA

Cho

Cr

Cr

Cho

3.5

3.0 2.5 2.0 Chemical shift (ppm)

1.5

3.5

3.0 2.5 2.0 Chemical shift (ppm)

1.5

Fig. 25.7 (a) T2-weighted MRI of a 6-year-old patient with Salla disease showing homogeneously abnormally high signal intensity in the WM, indicating defective myelination. The box indicates the VOI for the quantitative 1H MRS in the left parietal WM. (b) The location of the VOI in the left basal ganglia of the same patient. From Varho et al. 1999. Neurology 52: 1668–1672, with permission.

Salla disease In Salla disease (also described in Chapter 46), an autosomal-recessive free sialic acid storage disorder, N-acetylneuraminic acid (NANA), accumulates in lysosomes of brain tissue. MRS from parietal WM revealed increased NAA and Cr concentrations, and decreased Cho concentration. In the basal ganglia the Cr concentration was increased (Figure 25.7). NAA is considered to be a neuronal marker that, except for Canavan’s disease, has been found or assumed to be either stable or reduced. However, in Salla disease the high NAA signal may have a contribution from accumulated lysosomal NANA, which offsets the possible loss of NAA (Varho et al., 1999). The high Cr is in line with the increased glucose

uptake found in 2-fluoro-2-deoxy-D-glucose-PET studies, reflecting increased energy demand. Leukodystrophy in patients with ovarian dysgenesis Only four patients have been described with the above-mentioned clinical features. In this pathological condition MRSI has been particularly helpful in understanding the pathophysiology of this disease. It has shown a decrease of the Cho signal in frontal WM, which was not always abnormal on conventional MRI, and a decrease of the NAA signal in the same ROI in all but one patient. The spectroscopic results were in agreement with biopsy findings that showed hypomyelination. A follow-up

439

440

Gioacchino Tedeschi and Simona Bonavita

study in one case showed an increase of the Cho signal and the appearance of Lac (Schiffmann et al., 1997), thus suggesting that, in parallel with a clinical worsening, there was ongoing active demyelination and possibly astrocytosis. The association of hypomyelination and ovarian dysgenesis suggested that a hypothetic trophic factor could sustain both myelination and ovarian development.

REFERENCES Arnold DL. 1999. Magnetic resonance spectroscopy: imaging axonal damage in MS. J Neuroimmunol 98: 2–6. Arnold DL, Matthews PM, Francis GS, O’Connor J, Antel JP. 1992. Proton magnetic resonance spectroscopic imaging for metabolic characterization of demyelinating plaques. Ann Neurol 31: 235–241. Auer DP, Schirmer T, Heidenreich JO, et al. 2001. Altered white and gray matter metabolism in CADASIL: a proton MR spectroscopy and MRSI study. Neurology 56: 635–642. Bitsch A, Bruhn AA, Vougioukas V, et al. 1999. Inflammatory CNS demyelination: histopathologic correlation with in vivo quantitative proton MR spectroscopy. Am J Neuroradiol 20: 1619–1627. Bizzi A, Ulug AM, Crawford T, Passe T, Bugiani M, Bryan RN, Barker PB. 2001. Quantitative proton MR spectroscopic imaging in acute disseminated encephalomyelitis. Am J Neuroradiol 22: 1125–1130. Bonavita S, Virta A, Jeffries N, et al. 2003. Diffuse neuroaxonal involvement in mucolipidosis IV as assessed by proton magnetic resonance spectroscopic imaging. J Child Neurol 18: 443–449. Brex PA, Gomez-Anson B, Parker GJ, et al. 1999. Proton MR spectroscopy in clinically isolated syndromes suggestive of multiple sclerosis. J Neurol Sci 166: 16–22. Charcot JM. 1968. Gaz Hôp Paris 41: 554–555, 557–558, 566. Cifelli A, Arridge M, Jezzard P, et al. 2002. Thalamic neurodegeneration in multiple sclerosis. Ann Neurol: 52: 650–653. Coles AJ, Wing MG, Molyneux P, et al. 1999. Monoclonal antibody treatment exposes three mechanisms underlying the clinical course of multiple sclerosis. Ann Neurol 46: 296–304. Davie CA, Hawkins CP, Barker GJ, et al. 1994. Serial proton magnetic resonance spectroscopy in acute multiple sclerosis lesions. Brain 117: 49–58. De Stefano N, Dotti MT, Mortilla M, et al. 2001b. Magnetic resonance imaging and spectroscopic changes in brains of patients with cerebrotendinous xanthomatosis. Brain 124: 121–131.

De Stefano N, Matthews PM, Antel JP, et al. 1995a. Chemical pathology of acute demyelinating lesions and its correlation with disability. Ann Neurol 38: 901–909. De Stefano N, Matthews PM, Arnold DL. 1995b. Reversible decreases in N-acetylaspartate after acute brain injury. Magnet Reson Med 34: 721–727. De Stefano N, Matthews PM, Fu L, et al. 1998. Axonal damage correlates with disability in patients with relapsing remitting multiple sclerosis. Results of a longitudinal magnetic resonance spectroscopy study. Brain 121: 1469–1477. De Stefano N, Narayanan S, Matthews PM, et al. 1999. In vivo evidence for axonal dysfunction remote from focal cerebral demyelination of the type seen in multiple sclerosis. Brain 122: 1933–1939. De Stefano N, Narayanan S, Francis GS, et al. 2001a. Evidence of axonal damage in the early stages of multiple sclerosis and its relevance to disability. Arch Neurol-Chicago 58: 65–70. Fu L, Matthews P, De Stefano N, Worsley K, Narayanan S, Francis G, Antel J, Wolfson C, Arnold D. 1998. Imaging axonal damage of normal-appearing white matter in multiple sclerosis. Brain 121(1): 103–113. Kapeller P, McLean MA, Griffin CM, et al. 2001. Preliminary evidence for neuronal damage in cortical grey matter and normal appearing white matter in short duration relapsing remitting multiple sclerosis: a quantitative MR spectroscopic imaging study. J Neurol 248: 131–138. Kesselring J, Miller DH, Robb SA, Kendall BE, Moseley IF, Kingsley D, du Boulay EP, McDonald WI. 1990. Acute disseminated encephalomyelitis. MRI findings and the distinction from multiple sclerosis. Brain 113: 291–302. Leary SM, Davie CA, Parker GJM, et al. 1999. Magnetic resonance spectroscopy of normal appearing white matter in primary progressive multiple sclerosis. J Neurol 246: 1023–1026. Lublin FD, Reingold SC. 1996. Defining the clinical course of multiple sclerosis: results of an international survey. National Multiple Sclerosis Society (USA) Advisory Committee on Clinical Trials of New Agents in Multiple Sclerosis. Neurology 46: 907–911. Narayana PA, Doyle TJ, Lai D, Wolinsky JS. 1998. Serial proton magnetic resonance spectroscopic imaging, contrastenhanced magnetic resonance imaging, and quantitative lesion volumetry in multiple sclerosis. Ann Neurol 43: 56–71. Peterson JW, Bo L, Mork S, Chang A, Trapp BD. 2001. Transected neurites, apoptotic neurons, and reduced inflammation in cortical multiple sclerosis lesions. Ann Neurol 50: 389–400. Sarchielli P, Presciutti O, Pelliccioli GP, et al. 1999. Absolute quantification of brain metabolites by proton magnetic resonance spectroscopy in normal-appearing white matter of multiple sclerosis patients. Brain 122: 513–521.

MR spectroscopy in demyelination and inflammation

Sarchielli P, Presciutti O, Tarducci R, et al. 2002. Localized (1)H magnetic resonance spectroscopy in mainly cortical gray matter of patients with multiple sclerosis. J Neurol 249: 902–910. Schiffmann R, Tedeschi G, Kinkel RP, et al. 1997. Leukodystrophy in patients with ovarian dysgenesis. Ann Neurol 41: 654–661. Sharma R, Narayana PA, Wolinsky JS. 2001. Grey matter abnormalities in multiple sclerosis: proton magnetic resonance spectroscopic imaging. Mult Scler 7: 221–226. Tartaglia MC, Narayanan S, De Stefano N, et al. 2002. Choline is increased in pre-lesional normal appearing white matter in multiple sclerosis. J Neurol 249: 1382–1390. Tedeschi G, Bonavita S, McFarland HF, et al. 2002. Proton MR spectroscopic imaging in multiple sclerosis. Neuroradiol 44: 37–42. Tourbah A, Stievenart JL, Gout O, et al. 1999. Localized proton magnetic resonance spectroscopy in relapsing remitting versus secondary progressive multiple sclerosis. Neurology 53: 1091–1097.

van Walderveen MAA, Barkhof F, Pouwels PJW, et al. 1999. Neuronal damage in T1-hypointense multiple sclerosis lesions demonstrated in vivo using proton magnetic resonance spectroscopy. Ann Neurol 46: 79–87. van Walderveen MAA, Kamphorst W, Scheltens P, et al. 1998. Histopathologic correlate of hypointense lesions on T1weighted spin-echo MRI in multiple sclerosis. Neurology 50: 1282–1288. Varho T, Komu M, Sonninen P, et al. 1999. A new metabolite contributing to N-acetyl signal in MRS of the brain in Salla disease. Neurology 53(5): 1162. Willemsen MA, IJlst L, Steijlen PM, et al. 2001. Clinical, biochemical and molecular genetic characteristics of 19 patients with the Sjogren–Larsson syndrome. Brain 124: 1426–1437. Wylezinska M, Cifelli A, Jezzard P, Palace J, Alecci M, Matthews PM. 2003. Thalamic neurodegeneration in relapsingremitting multiple sclerosis. Neurology 60(12): 1949–1954.

441

442

Gioacchino Tedeschi and Simona Bonavita

Case Study 25.1 Acute disseminated encephalomyelitis Peter Barker, D.Phil., Johns Hopkins University School of Medicine, Baltimore History 16-year-old female with headache, vomiting and altered level of consciousness, after a fever of unknown origin. Cerebrospinal fluid white blood cell count was elevated.

FLAIR

Right frontal lesion

Cho

NAA

Cr

Technique Conventional MRI and multi-slice proton MRSI at long TE (280 ms).

Imaging findings Multiple, asymmetric T2 hyperintense lesions were present in the sub-cortical WM of both hemispheres. The lesions were characterized by decreased levels of NAA (arrows), but an absence of any other metabolic abnormalities. The patient made a rapid recovery and on follow-up showed near normal MRI.

Left frontal (Normal)

NAA

NAA

Cho Cr

Discussion Monophasic ADEM with good outcome appears to show selective NAA reduction only, which can recover on follow-up scan. This may be helpful for distinction from pediatric MS or recurrent ADEM, which may show increased Cho.

Cho

Follow-up FLAIR

Key points Monophasic ADEM with good outcome is generally associated with low NAA only (no increase in Cho). NAA decreases can sometimes be reversible.

Reference Bizzi A, Ulug AM, Crawford T, Passe T, Bugiani M, Bryan RN, Barker PB. 2001. Quantitative proton MR spectroscopic imaging in acute disseminated encephalomyelitis. Am J Neuroradiol 22: 1125–1130.

MR spectroscopy in demyelination and inflammation

Case Study 25.2 Reversible posterior leukoencephalopathy: MRSI Peter Barker, D.Phil., Johns Hopkins University School of Medicine, Baltimore History

T2

Cho

36-year-old female with eclampsia, hypertension, pancreatitis, and seizures. Technique Conventional MRI and multi-slice MRSI (TE  280 ms).

b

Imaging findings T2 MRI showed hyperintensity in the occipitoparietal lobe. T2 hyperintense regions were characterized by low levels of all metabolites, consistent with vasogenic edema (a). Anterior WM regions with normal MRI appearance showed increased levels of Cho and decreased NAA (b). No Lac was detected. MRI and MRSI performed 2 months later were normal.

a

NAA

Cr

Discussion The pathology of reversible posterior leukoencephalopathy syndrome (RPLS) is unclear. However, the absence of lactate suggests that ischemia is unlikely, while widespread increase in Cho and vasogenic edema may indicate blood–brain barrier breakdown, and micro-glial proliferation (Eichler, 2002).

b

Cho

NAA

Key points Cr

RPLS lesions are usually characterized by low levels of all metabolites.

*

Metabolic abnormalities are widespread, and may be reversible.

a

MRSI is convenient for the evaluation of lesion and normal appearing brain regions simultaneously.

ppm

4.0

3.0

2.0

1.0

Reference Eichler FS, Wang P, Wityk RJ, Beauchamp Jr NJ, Barker PB. 2002. Diffuse metabolic abnormalities in reversible posterior leukoencephalopathy syndrome. Am J Neuroradiol 23(5): 833–837.

443

26

Diffusion imaging in demyelination and inflammation Marco Rovaris and Massimo Filippi Department of Neurology, Neuroimaging Research Unit, Scientific Institute and University Ospedale San Raffaele, Milan, Italy

Key points • MS plaques show increased apparent diffusion coefficient (ADC) and decreased fractional anisotropy (FA), with the largest changes in non-enhancing T1-hypointense lesions. • Average lesion ADC and FA seem to correlate with the overall lesion burden on T2 and T1 images in multiple sclerosis (MS). • ADC values may be decreased in the acute phase of central pontine myelinolysis (intracellular hypotonicity leads to cytotoxic edema). • ADC of normal-appearing white matter (WM) and gray matter is higher in MS patients than in controls, particularly in peri-plaque regions. • DWI studies of normal appearing WM suggest that, contrary to what happens in MS, brain tissue outside T2 visible lesions is spared by the pathological process in acute disseminated encephalomyelitis. • Increased ADC may precede the development of MS lesions visible on conventional MR imaging. • Diffusion tensor imaging (DTI) fiber-tracking shows promise for evaluating specific WM bundles for axonal damage and correlation with clinical disability. DTI of the spine and optic nerve in the future may also be helpful for the assessment of MS.

Introduction Conventional MR imaging (MRI) is very sensitive for the detection of demyelinating and inflammatory 444

lesions of the central nervous system (CNS), but it is not without some limitations. MRI findings are unable to differentiate the heterogeneous pathological substrates of individual lesions, since different tissue changes, including edema, inflammation, demyelination, remyelination, gliosis, and axonal loss, all lead to a similar appearance of increased signal intensity on T2-weighted images. Moreover, conventional MRI does not delineate tissue damage occurring beyond T2-visible lesions, i.e. in the normal-appearing white matter (NAWM) and gray matter (NAGM) (Allen et al., 1979; Kidd et al., 1999). These limitations are only partially overcome by the use of post-contrast T1-weighted imaging. Gadolinium-enhanced T1weighted images allow lesions with ongoing inflammation (Katz et al., 1993) to be distinguished from those without active inflammation, since enhancement occurs as a result of increased blood–brain barrier (BBB) permeability (Kermode et al., 1990). However, post-contrast T1-weighted MRI may not provide information on the characteristics of tissue damage associated with the presence of inflammation. Hypointense lesions on T1-weighted images correspond to areas where, in patients with multiple sclerosis (MS) and other demyelinating diseases, chronic severe tissue disruption has occurred (van Walderveen et al., 1998). Measuring the extent of T1-hypointense lesions may still not correspond to the severity of intrinsic lesion pathology and cannot provide any information about NAWM and NAGM damage. Diffusion-weighted imaging (DWI) may be a quantitative MR-based technique able to improve our understanding of the pathophysiology of

Diffusion imaging in demyelination and inflammation

inflammatory and demyelinating disorders of the CNS, by overcoming some of the shortcomings of conventional MRI. Diffusion is the microscopic random translational motion of molecules in a fluid system. In the CNS, diffusion is influenced by the microstructural components of tissue, including cell membranes and organelles (cf. Chapter 4). The diffusion coefficients of biological tissues (which can be measured in vivo by MRI) is, therefore, lower than the diffusion coefficient in free water and for this reason is named apparent diffusion coefficient (ADC) (Le Bihan et al., 1986). Pathological processes which modify tissue integrity, thus resulting in a loss or increased permeability of “restricting” barriers, can determine an increase of the ADC. The measurement of diffusion is also dependent on the direction in which diffusion is measured, since within the timescale of the diffusion experiment water may diffuse unequally in different directions, depending on the size and shape of the cellular structures within the brain. As a consequence, diffusion measurements can give information about the size, shape, integrity, and orientation of tissues (Le Bihan et al., 1991). A measure of diffusion which is independent of the orientation of structures is provided by the mean – diffusivity (D), the average of the ADCs measured in three orthogonal directions. A full characterization of diffusion can be obtained in terms of a tensor (Basser et al., 1994), a 3 3 matrix which accounts for the correlation existing between molecular displacement along orthogonal directions (cf. Chapter 5). From the tensor, it is possible to derive, equal to the one-third of its trace, and some other dimensionless indexes of anisotropy. One of the most used of these indices is named fractional anisotropy (FA) (Basser and Pierpaoli, 1996; Pierpaoli et al., 1996). Inflammation and demyelination have the potential to alter the permeability or geometry of structural barriers to water molecular diffusion in the brain, thus leading to DWI-detectable changes. The present chapter outlines the major contributions given by DWI to the study of inflammatory and demyelinating CNS diseases, with a particular focus on MS, which can be considered as a model of the complex interactions between inflammation and demyelination in the pathogenesis of irreversible tissue damage.

DWI in the study of MRI-visible inflammation and demyelination Multiple sclerosis Currently available DWI and diffusion tensor imaging (DTI) studies have shown highly variable ADC, – D, and FA values in MRI-visible MS lesions (Larsson et al., 1992; Christiansen et al., 1993; Horsfield et al., 1996; Droogan et al., 1999; Werring et al., 1999; Bammer et al., 2000a; Cercignani et al., 2000; Filippi et al., 2000; Nusbaum et al., 2000a; Rocca et al., 2000; Roychowdury et al., 2000; Werring et al., 2000; Cercignani et al., 2001a; Filippi et al., 2001b). This is consistent with the known pathological heterogeneity of MS lesions (Bruck et al., 1997; van Waesberghe et al., 1999). Nevertheless, water diffusion abnormalities do differ in different types of MS lesions. All – investigators have shown higher ADC or D and lower FA values in non-enhancing T1-hypointense than in non-enhancing T1-isointense lesions (Droogan et al., 1999; Werring et al., 1999; Bammer et al., 2000a; Filippi et al., 2000; Nusbaum et al., 2000a; Roychowdury et al., 2000; Filippi et al., 2001b; Castriota-Scanderbeg et al., 2003) (Figure 26.1). T1-hypointense lesions are those where severe tissue loss has occurred (van Walderveen et al., 1998) and their extent is correlated with disease progression in patients with secondary progressive MS (Truyen et al., 1996). Although post-mortem studies correlating histopathology and DWI changes are needed, this observation shows the potential for DWI to provide quantitative metrics for monitoring irreversible tissue damage in MS. Conflicting results have been – achieved when comparing ADC or D values in enhancing vs. non-enhancing lesions. Some studies – reported higher ADC or D values in non-enhancing than in enhancing lesions (Droogan et al., 1999; Werring et al., 1999; Roychowdury et al., 2000), but others, based on larger samples of patients and lesions, did not report any significant difference between the two populations of lesions (Bammer et al., 2000a; Filippi et al., 2000; Filippi et al., 2001b). This discrepancy confirms that variable degrees of tissue damage occur in new enhancing lesions of MS – and might reflect different lesion ages (D increases sharply at the time of enhancement onset and then

445

446

Marco Rovaris and Massimo Filippi

(a)

(b)

(c)

(d)

(e)

Fig. 26.1 Axial proton density-weighted (a), T2-weighted (b), and post-contrast (gadolinium DTPA, 0.1 mmol/kg) T1-weighted (c) MR images of the brain from a patient with MS. In (a) and (b), multiple hyperintense lesions are visible, suggestive of multifocal WM pathology. In (c), some of these lesions are T1-hypointense, indicating that marked tissue destruction has occurred. In the – corresponding D (d) map, diffusivity is increased within MS lesions, which appear hyperintense. T1-hypointense lesions are also visible as areas of reduced signal on FA map (e), indicating a pronounced local decrease of anisotropic diffusion.

decreases rapidly in the next few weeks (Rocca et al., 2000; Werring et al., 2000)). On the contrary, FA has always been found to be lower in enhancing than in non-enhancing lesions (Werring et al., 1999; Filippi et al., 2001b). Several studies have shown that water diffusivity is markedly increased in ring-enhancing lesions when compared to homogeneously-enhancing lesions (Bammer et al., 2000a; Nusbaum et al., 2000a; Roychowdury et al., 2000), or in the non-enhancing portions of enhancing lesions when compared with enhancing portions (Nusbaum et al., 2000a). In acute, Balò-like MS lesions, where ring-like enhancement corresponds to T2 hypointensity, water diffusion characteristics change across the different lesion layers and are the highest in this inner ring (Iannucci et al., 2000). Markedly reduced FA values have also been found in ring-enhancing lesions (Bammer et al., 2000a). A longitudinal study of large MS lesions which were followed-up for 1–3 months (Castriota-Scanderbeg et al., 2002) has shown that – D values were increased in all acute lesions, but they continued to increase during follow-up only in a subgroup of these lesions. All of this confirms that pronounced tissue destruction can occur in active MS lesions, but it also underpins that further longitudinal studies are warranted to address the important issue of how much of this tissue disorganization in enhancing lesions is permanent (i.e. related to axonal loss) and how much is transient (i.e. related to edema, demyelination, and remyelination). – D and FA of MS lesions were found to be strongly correlated with lesion burden on T2- and T1-weighted images (Filippi et al., 2001b). This provides support for the concept that, on average, the size of lesions

and the severity of the tissue damage within them run in parallel. Interestingly, two different studies of MS patients (Iannucci et al., 2001; De Stefano et al., 2002) did not find a significant correlation between – measures of brain atrophy and the average D value of T2-hyperintense lesions, suggesting that the destructive process underlying MS-related brain volume reduction does not merely depend upon the severity of focal, MRI-visible pathology. A significant – correlation between D and FA of macroscopic MS lesions has also been reported (Werring et al., 1999; Filippi et al., 2001b). This correlation was, however, far from being a strict relationship. Since tissue – damage alone would both increase D and decrease FA, this observation suggests the potential of serial DWI scans to monitor tissue repair. For example, – marked glial proliferation would decrease both D and FA in concert, thus reducing the magnitude of the correlation that would result from a preponderance of tissue damage over tissue repair. Other inflammatory/demyelinating diseases In patients with acute disseminated encephalomyelitis (ADEM), heterogeneous DWI changes are detectable within T2-hyperintense lesions soon after the clinical onset of the disease (Bernarding et al., 2002). A spatial heterogeneity of ADC values can be detected even within single lesions (Bernarding et al., 2002), reflecting the prevalence of inflammatory vs. demyelinating tissue changes in different portions of the same area of abnormal T2-signal intensity. On average, ADC values of ADEM lesions tend to increase in the post-acute phase of the disease

Diffusion imaging in demyelination and inflammation

(Harada et al., 2000; Bernarding et al., 2002) and, a few weeks after its onset, they may have the same magnitude as those seen in chronic MS lesions. This explains why, in patients with ADEM who underwent DWI 1–9 years after the onset of the neurological – manifestations, average lesion D was not significantly different from that of MS patients (Inglese et al., 2002). These findings suggest that DWI may be useful for the characterization of different phases of the inflammatory process in demyelinating conditions. DWI has also been preliminarily used to study “pure” demyelinating diseases of the CNS in case reports of patients with central pontine myelinolysis (Cramer et al., 2001) and progressive multifocal leukoencephalopathy (PML) (Ohta et al., 2001). DWI findings in two patients with central pontine myelinolysis (Cramer et al., 2001) indicate that, within the MRI-visible lesions, ADC values are decreased in the acute phase of the disease, but they may normalize a few weeks later. This may help to differentiate central pontine myelinolysis from “mixed” inflammatory and demyelinating disorders, such as ADEM and MS. DWI findings in central pontine myelinolysis also support the notion that relative intracellular hypotonicity, leading to cytotoxic edema and, as a consequence, restricted water diffusion, is a key event in the pathogenesis of the disease. In PML, demyelination is known to result from the lytic infection of oligodendrocytes by the JC virus. A longitudinal DWI study of a single patient with PML (Ohta et al., 2001) has described ADC patterns consistent with a loss of myelin in affected brain regions, with an initial decrease followed by a progressive increase of ADC values. Interestingly, in this patient DWI was able to detect initial signs of cytotoxic edema (leading to decreased ADC values) with a greater sensitivity than T2-weighted MRI, suggesting the presence of JC virus infection in brain areas where demyelination has not occurred yet.

DWI in the study of conventional MRIundetectable white and gray matter damage Multiple sclerosis Numerous studies of water diffusion in MS have – consistently shown that ADC or D values of NAWM

are higher than those of corresponding WM from controls, but lower than those measured in T2-visible lesions (Larsson et al., 1992; Christiansen et al., 1993; Horsfield et al., 1996; Droogan et al., 1999; Werring et al., 1999; Bammer et al., 2000a; Cercignani et al., 2000; Filippi et al., 2000; Nusbaum et al., 2000a; Rocca et al., 2000; Roychowdury et al., 2000; Werring et al., 2000; Cercignani et al., 2001a; Ciccarelli et al., 2001, 2003; Filippi et al., 2001b; Cercignani et al., 2002; Castriota-Scanderbeg et al., 2003). DTI studies also showed that FA values of NAWM are lower than those of corresponding WM from controls and higher than those of T2-visible lesions (Werring et al., 1999; Bammer et al., 2000a; Filippi et al., 2000; Filippi et al., 2001b; Cercignani et al., 2001a; Ciccarelli et al., 2001; Castriota-Scanderbeg et al., 2003). An exception is the study by Griffin et al. (2001), which was conducted in a group of 28 patients with early relapsing–remitting MS and did not find any signifi– cant difference in D and FA values of NAWM regions between patients and controls. A preliminary study of patients at presentation with clinically isolated syndrome suggestive of MS (Caramia et al., 2002) also failed to detect significant differences of NAWM ADC between patients and healthy controls. The vast majority of DWI studies are, therefore, consistent with other pathological and quantitative MRI studies showing that tissue damage does occur outside T2-visible lesions in MS (Filippi et al., 1999), but they suggest that this damage might become permanent a few years after the clinical onset of the disease. Although DWI results indicate a net loss and disorganization of structural barriers to water molecular motion in the NAWM, the possible pathological substrates of these findings are not defined yet. Subtle changes are known to occur in the NAWM from patients with MS, including diffuse astrocytic hyperplasia, patchy edema, perivascular infiltration, gliosis, abnormally thin myelin, and axonal loss (Allen et al., 1979). While all these pathological substrates might reduce FA, myelin and axonal loss should lead to increased water diffusivity. As a consequence, they are the most likely contributors to the increased water diffusivity and decreased FA observed in the NAWM. DWI changes in the NAWM have been reported to be correlated with the burden of MRI-visible lesions (Cercignani et al., 2000, 2001a,

447

448

Marco Rovaris and Massimo Filippi

2001b; Filippi et al., 2000, 2001b; Iannucci et al., 2001; Ciccarelli et al., 2003), but the average magnitude of the observed correlations was far from being strict. This suggests that these subtle NAWM changes are not merely the result of Wallerian degeneration of axons traversing T2-visible lesions, but they also may represent small focal abnormalities beyond the resolution of conventional scanning and independent of large lesions. That NAWM damage is widespread in the MS brain has also been shown in a recent – study (Cercignani et al., 2001a), where average D was measured in a large portion of brain NAWM and found to be significantly higher than that of healthy controls. Several studies assessed DWI changes in different brain regions and showed that NAWM abnormalities in MS are widespread, but tend to be more severe in peri-plaque regions (Guo et al., 2001, 2002) and in sites where MRI-visible MS lesions are usually located (Bammer et al., 2000a; Filippi et al., 2000, 2001b; Cercignani et al., 2001a). Intriguingly, significant DWI changes have been found in the internal capsule and corpus callosum from patients with primary progressive MS (Filippi et al., 2001b), which might, at least partially, explain the locomotor disability and cognitive impairment of these patients (Thompson et al., 1997). Two longitudinal – DWI studies detected increased D values in NAWM areas subsequently involved by new MS lesions (Rocca et al., 2000; Werring et al., 2000), suggesting that focal vasogenic edema and demyelination beyond the resolution of conventional MRI can play a part in the NAWM changes preceding new lesion formation in MS. In the normal brain, GM and WM have different water diffusion characteristics. In GM, diffusion is isotropic when averaged on a voxel scale, while it is extremely anisotropic in WM due to the directionality of the myelin fiber tracts (Hajnal et al., 1991). Using a segmentation technique based on FA thresholds and histogram analysis, Cercignani et al. – (2001a) showed that the average D of brain NAGM from MS patients is higher than that of brain GM from sex- and age-matched healthy volunteers, thus indicating that NAGM is not spared by the MS pathological process. Previous post-mortem studies (Kidd et al., 1999; Brownell and Hughes, 1962)

showed that lesions are relatively frequent in the cerebral GM of patients with MS. Studies with positron emission tomography (PET) (Blinkenberg et al., 2000) and quantitative MRI techniques (Chard et al., 2002; Dehmeshki et al., 2003) have consistently shown functional and structural abnormalities in the NAGM of MS patients with various disease phenotypes. In a post-mortem study (Peterson et al., 2001), it has been found that GM MS lesions show a reduced amount of inflammatory components when compared to those located in the WM of the same patients, thus supporting the hypothesis that MS pathology might follow different patterns in these two tissue compartments. As a consequence, one explanation of DWI findings in the study by Cercignani et al. (2001a) might be the presence of a certain amount of discrete MS lesions in the NAGM of MS patients, which go undetected on conventional T2-weighted imaging because they are usually small (Brownell and Hughes, 1962), have relaxation characteristics which result in poor contrast between them and normal GM and, in case of cortical lesions, because of partial volume effects with surrounding cerebrospinal fluid. Demyelinated regions of the cerebral cortex from MS patients were found to harbor transected dendrites, transected axons, and apoptotic neurons (Peterson et al., 2001). This suggests that T2-undetectable cortical lesions might also provoke a significant increase of NAGM – D. An alternative, but not mutually exclusive, explanation of the observed changes in NAGM might be the retrograde degeneration of GM neurons secondary to the damage of fibers traversing MS WM lesions (Evangelou et al., 2000). That retrograde degeneration may have a role in explaining NAGM changes in MS is supported by the correlations found – between T2-lesion volume and NAGM D (Cercignani et al., 2001a; Iannucci et al., 2001). Using DWI, Bozzali et al. (2002) have recently demonstrated that NAGM changes are more pronounced in patients with secondary progressive and primary progressive MS than in patients with the relapsing–remitting course of the disease, whereas Rovaris et al. (2002a) showed that NAWM and NAGM pathology is more severe in patients with secondary progressive MS than in patients with primary progressive MS (Figure 26.2). These findings suggest that NAGM

Diffusion imaging in demyelination and inflammation

(a) 160

120

Normalized pixel count

140 Normalized pixel count

(b) 70

Controls PPMS SPMS

100 80 60 40 20 0 0.0

Controls PPMS SPMS

60 50 40 30 20 10 0

0.6

1.2

1.8

2.4

3.0

0.0

0.6

D ( 10 3 mm2 s 1)

1.2 1.8 2.4 D ( 10 3 mm2 s 1)

3.0

–

Fig. 26.2 Average D histograms of the NAWM (a) and NAGM (b) from 44 healthy controls (dotted line), 96 primary progressive MS (grey line), and 47 secondary progressive MS (black line) patient groups. In comparison with healthy controls, a reduction of the histogram peak height (which reflects the amount of “truly” normal tissue) can be observed for both primary and secondary progressive MS patients and is more pronounced in the latter group of patients (reproduced from Rovaris et al., 2002a.)

(b) 1.30 1.20 1.10 1.00 0.90 30

35 40

45

50 55 60 65 SDMT Score

70

75

80

Average NAGM diffusivity

Average NAGM diffusivity

(a)

1.30 1.20 1.10 1.00 0.90 0

0.5

1

1.5

2

2.5 3 3.5 EDSS Score

4

4.5

5

– Fig. 26.3 Scatterplots of average NAGM D values vs. symbol digit modalities test (SDMT) (Figure 26.2(a)) and EDSS (Figure 26.2(b)) scores in 34 patients with relapsing–remitting MS. SDMT is widely used to measure sustained attention and concentration in the – assessment of MS-related cognitive impairment. A significant correlation was found between NAGM D and SDMT scores – (r value: 0.42, P  0.01), but not between NAGM D and the severity of neurological disability (r value: 0.27, P  0.11).

pathological abnormalities might yet be an additional factor contributing to the worsening of clinical disability in patients with progressive MS. The presence of NAGM damage also fits well with the frequent demonstration of cognitive impairment in patients with MS. In an exploratory study, Rovaris et al. (2002b) found a moderate correlation between – a global cognitive impairment index and D of the NAGM in 34 mildly-disabled relapsing–remitting MS patients. Moderate correlations were also found between several individual neuropsychological test scores, MRI-visible lesion burden and average

lesion, NAWM and NAGM values (Figure 26.3). The results of this study suggest that the extent and the intrinsic nature of the macroscopic lesions as well as the damage of the NAWM and NAGM all contribute to the neuropsychological deficits of MS patients. DWI has also been used to investigate whether subtle structural changes can be detected in the basal ganglia of patients with MS (Ciccarelli et al., 2001; Filippi et al., 2001a). In a study of 31 patients with MS (Filippi et al., 2001a), no DWI changes were found in any of the basal ganglia regions studied. Interestingly,

449

450

Marco Rovaris and Massimo Filippi

Ciccarelli et al. (2001) found significantly increased anisotropy in the caudate nucleus, putamen, and thalamus of 39 MS patients compared to healthy sub– jects, whereas D values in the same regions did not differ between patients and controls. The latter findings might be explained by intrinsic basal ganglia connection reorganization, as a consequence of MSrelated damage to cortical–subcortical WM projections. These findings confirm that basal ganglia damage is modest in MS and suggest that the PET (Roelcke et al., 1997) and functional MRI (fMRI) (Filippi et al., 2002a) changes detected in the basal ganglia of MS patients are more likely to be secondary to diaschisis phenomena than to the presence of intrinsic lesions. Although DWI has the potential to substantially improve our ability to delineate the full extent of the pathological changes occurring in the brain of MS patients, it remains to be established whether DWI is also reliable enough to monitor longitudinal diseaserelated changes. A 1-year longitudinal study of DWI changes in the NAWM of clinically isolated syndrome patients (Caramia et al., 2002) has found that ADC values become significantly higher than those of healthy controls only at follow-up scans and are significantly correlated with T2-hyperintense lesion load at the same time-point. Considering that the vast majority (84%) of these patients had also shown MRI evidence of spatial and temporal disease dissemination, thus fulfilling criteria for MS diagnosis (McDonald et al., 2001), these findings indicate that DWI is sensitive to the appearance of NAWM changes since the earliest phases of the disease. In a study (Oreja-Guevara et al., 2003) of 26 untreated patients with relapsing–remitting MS, who underwent conventional MRI and DTI scans every 3 months for 18 months, a significant decrease of average NAGM FA and a significant increase of aver– age NAGM D and brain T2-hyperintense lesion load over the study period were observed. No mutual correlations were found between the changes of aver– age NAGM D and those of GM volume. These findings seem to indicate that DTI is sensitive to subtle changes occurring in the NAGM of untreated patients with established MS over a short-term period and that such changes do not depend upon the concomitant development of GM atrophy.

Other inflammatory/demyelinating diseases In patients with ADEM, conventional MRI patterns of brain lesions may resemble those of patients with MS, although ADEM lesions tend to partially or completely resolve at follow-up and new lesion formation occurs rarely. In a recent study by Inglese et al. (2002), no significant abnormalities of water diffusion were detected in the normal appearing brain tissue (NABT) of eight ADEM patients, who were scanned after an average interval of 2 years from the clinical onset of the disease. These data cannot rule out that reversible changes may occur in the NAWM and NAGM during the acute phase of ADEM, but they show that, differently from what happens in MS, brain tissue outside residual MRI-visible lesions seems to be spared by the pathological process. On the other hand, however, the – average D values of the basal ganglia were found to be significantly higher in ADEM than in MS patients (Holtmannspotter et al., 2003), consistently with the notion that deep GM lesions on T2-weighted images are a relatively frequent finding of ADEM. Because of their high inter-subject variability, DWI changes are unlikely to be used in isolation in the diagnostic workup of individual patients with ADEM. Nevertheless, these results indicate that the in-vivo quantification of NAWM and basal ganglia pathology using DWI might represent an additional piece of information in the context of a differential diagnosis between ADEM and MS and call for longitudinal studies of patients at presentation with multifocal CNS demyelinating damage to assess the diagnostic and prognostic role of DWI findings.

DWI in the study of overall brain tissue damage DWI changes can be assessed using histogram analysis, thus obtaining quantities which encompass MRI-visible and MRI-occult lesion burden in different brain tissues. The first step in the creation of the histogram is a preliminary manual or semi-automated image segmentation aimed at excluding all the extraparenchymal tissues. Then, the data set of values from – D or FA maps is displayed as a histogram, which is usually normalized to the total number of pixels to allow comparisons between subjects with different

Diffusion imaging in demyelination and inflammation

brain sizes. For each histogram, the height and posi– tion of the histogram peak and the average D or FA are usually computed. – Cercignani et al. (2000) measured D in a large portion of the central brain from 35 patients with mildly-disabling relapsing–remitting MS and compared it with that of 24 age- and sex-matched – healthy volunteers. Significantly higher brain D and lower histogram peak height were found in patients. Similar results were found by two other studies (Nusbaum et al., 2000b; Cercignani et al., 2001b). In the former study by Cercignani et al. (2000), magnetization transfer ratio (MTR) histograms were also produced and no correlation was found for average – MTR and D taken from the histograms. The lack of – correlation between MTR and D in the brain tissue is likely to be the result of the complex relationship between destructive (inflammation, demyelination, and axonal loss) and reparative (remyelination and gliosis) mechanisms and their variable effects on – MTR and D values. Using histogram analysis, Cercignani et al. (2001b) also showed significant changes of FA histogram-derived metrics in a large cohort of MS patients with various disease phenotypes. This study also showed a moderate correla– tion between brain D and FA, as previously described for T2-visible lesions (Werring et al., 1999; Filippi et al., 2001b). Whole brain tissue histograms are influenced by the presence of macroscopic MS lesions, as well as by changes in NAWM and NAGM. Several studies have assessed water diffusivity changes in NABT in isolation, by excluding from the histogram analysis those pixels belonging to T2-visible lesions (Iannucci et al., 2001; Rovaris et al., 2002b). No significant correlation was found between MT- and DWI-derived metrics of the NABT (Iannucci et al., 2001), thus confirming previous results assessing correlation between these quantities taken from histograms of brain tissue containing T2-visible lesions (Cercignani et al., 2000). Interestingly, in a study of 20 MS – patients, average brain D was found to be significantly correlated with normalized brain volume, whereas this was not the case for either the average – D of T2-hyperintense lesions or the burden of MRIvisible MS lesions (De Stefano et al., 2002). In the same study, N-acetyl aspartate (NAA) concentration

in a large portion of brain tissue, together with – average brain D , entered a composite MR score that yielded a strong correlation with NBV (r value: 0.70). This suggests that the destructive process underlying MS-related brain atrophy might be mostly driven by diffuse and subtle changes in the normal-appearing tissues. Clearly, using histogram analysis, information related to the status of specific brain structures is inevitably lost. However, in the context of MS clini– cal trials, it may be unfeasible to measure D and FA changes from several different brain regions and tissues, whereas a quantitative measure reflecting overall lesion burden might be a desirable outcome measure. In addition, compared with a region-ofinterest-based analysis, a histogram-based analysis has the advantage of being less affected by observer variability. Preliminary data (Cercignani et al., 2003a) suggest that the inter-sequence and scan-rescan reproducibility of DWI histogram-derived quantities are relatively high. Additional studies are now – needed to validate the use of D and FA histogramderived quantities as regards their sensitivity to biological changes over time.

Correlations between DWI and clinical findings Significant correlations between DWI findings and MS clinical manifestations or disability were not found in several studies (Horsfield et al., 1996; Droogan et al., 1999; Cercignani et al., 2000; Filippi et al., 2000; Griffin et al., 2001), perhaps because of the relatively small samples studied (Horsfield et al., 1996), the limited brain coverage (Horsfield et al., 1996; Droogan et al., 1999), or the narrow range of disabilities that was considered (Cercignani et al., 2000; Filippi et al., 2000; Griffin et al., 2001). With improved DWI technology and increased numbers of patients studied, correlations between DWI findings and MS clinical manifestations or disability are now emerging (Castriota-Scanderbeg et al., 2000; Nusbaum et al., 2000b; Cercignani et al., 2001b; 2002; Filippi et al., 2001b; Bozzali et al., 2002; Rovaris et al., 2002b). – Average lesion D, but not average lesion FA, was found to be significantly correlated, albeit moderately,

451

452

Marco Rovaris and Massimo Filippi

with clinical disability in a study of 78 patients with MS (Filippi et al., 2001b). The lack of a correlation between disability and FA indicates that the loss of overall impediment to diffusional motion is more important than the loss of tissue anisotropy in determining patients’ clinical status. This fits well with the concept that loss of anisotropy might also result from reparative mechanisms. Interestingly, in patients with secondary progressive MS a moderate and significant correlation was found between average – lesion D or FA and disability, whereas no significant correlation was found between disability and T2-lesion volume. On the contrary, a significant correlation between disability and T2-lesion volume was found in patients with relapsing–remitting MS, where, in turn, – there was no correlation between average lesion D or FA and disability. These findings suggest that mechanisms leading to disability are likely to be different in patients with relapsing–remitting and secondary progressive MS. Although caution must be exercised, one might speculate that new lesion formation is a relevant pathological aspect in relapsing–remitting MS, whereas tissue loss in pre-existing lesions is one of the pathological hallmarks of secondary progressive MS. As a consequence, these results indicate DWI measures as promising MR markers to be used in addition to conventional MRI to monitor the evolution of secondary progressive MS. Consistent with these observations, it has been shown that water diffusivity in the whole brain tissue (Nusbaum et al., 2000b), NAGM (Bozzali et al., 2002) and T2-visible lesions (Castriota-Scanderbeg et al., 2000) is significantly increased in patients with secondary progressive MS when compared to those with relapsing–remitting MS. Castriota-Scanderbeg et al. (2000) also found strong correlations between average lesion diffusivity, disability, and disease duration. Cercignani et al. (2001b) found a strong correlation between disability and FA histogram peak position in patients with secondary progressive MS. A recent multiparametric MR study has shown that a composite MR score, based on the T1-hypointense lesion volume, brain – NAA to creatine (Cr) ratio, and brain D, is strictly correlated with the level of MS-related disability (Mainero et al., 2001). DWI findings were also found to correlate with cognitive impairment in patients with relapsing–remitting MS (Rovaris et al., 2002b). Two recent studies (Codella et al., 2002a, 2002b)

correlated DWI findings with the presence and severity of fatigue in 28 MS patients. Average NABT, NAGM – and lesion D or FA were not significantly different between fatigued and non-fatigued MS patients, nor was there a significant correlation between any of these quantities and the fatigue severity scale scores. These preliminary data suggest that the severity of structural brain and GM pathology, as assessed by DWI, does not seem to be a critical factor contributing to the pathogenesis of fatigue in MS. More recently, in patients with relapsing–remitting MS (Rocca et al., 2002a) and primary progressive MS (Filippi et al., 2002b; Rocca et al., 2002b), moderate to strong correlations have been found between the severity of structural changes of the NABT (as measured using DWI) and the relative activations of several cortical areas located in a widespread network for sensory-motor and multimodal integration, measured using f MRI. Similar correlations have also been found in patients with clinically definite MS and non-specific findings on conventional MRI scans of the brain (Rocca et al., 2003). This suggests that not only macroscopic MS lesions, but also subtle NABT changes can cause adaptive cortical reorganization with the potential to limit the functional consequences of MS-related structural damage.

Future perspectives Diffusion tensor tractography and newer DWI acquisition schemes DTI also provides a method by which it may become possible to obtain in vivo human connectivity data both rapidly and non-invasively. Due to the different tissue anisotropies, DTI is able not only to distinguish WM from GM, but also to elucidate the macroscopic volume-averaged orientation of the microstructure therein. The anisotropy provides the basis of DT tractography methods, developed to determine the pathways of anatomical CNS connections through the WM in vivo. There are several methodical limitations to DTI that degrade the information about the spatial orientation of fibre tracts, which is contained within the DT. These include the poor resolution of DTI data in comparison with the size of fibres tracts, the presence of crossing fibres on a voxel scale, and the

Diffusion imaging in demyelination and inflammation

low signal-to-noise ratio of the data from which the DT is estimated, which introduces errors in the directional information derived from it. Chapters 4 and 5 discuss these issues in detail. Recently, several authors have developed different tractography techniques that utilize the directional and anisotropy information contained in a given voxel to connect neighbouring voxels and to visualize WM tracts in vivo (Conturo et al., 1999; Mori et al., 1999, 2002) (Figure 26.4). The development of these methods to identify and segment individual brain tracts is important because this would allow a detailed assessment of their intrinsic damage. Tench and co-workers (2002) used this technique to map the corpus callosum and the pyramidal tracts from 14 MS patients and 10 controls and to measure the corresponding ADC values within these tracts. The ADC was significantly higher in patients than in controls, and it was higher in the corpus callosum than in the pyramidal tracts for both groups. The application of tractography methods to segment the different WM structures in MS patients might allow the strength of the correlations between clinical and MRI findings to be improved. Wilson et al. (2003) produced maps of pyramidal tracts from DTI scans of 25 patients with relapsing–remitting MS and measured the relative anisotropy along these pathways. Significant correlations were found between the latter parameter and patients’ expanded disability status scale (EDSS) and pyramidal functional system scores, whereas T2 lesion burden and diffusion histogram parameters did not correlate with clinical findings. Vaithianathar et al. (2002) used DTI to map the pyramidal tracts in 25 patients with relapsing–remitting MS. Then, they sampled T1 relaxation time values along the corresponding trajectories on co-registered whole-brain T1 maps. Pyramidal tract T1, but not total WM T1 values, correlated significantly with the severity of patients’ neurological disability. The results of these preliminary studies confirm that in vivo WM tract mapping using DTI is feasible and may increase the sensitivity and specificity of MR-based techniques in the assessment of clinically relevant MS pathology. This technique may also be applicable to other inflammatory and demyelinating diseases, to provide objective measures for monitoring the progression of clinical disability. It is also worth mentioning newer acquisition schemes

453

Parietal fibres Frontal fibres Occipital fibres

Cerebral peduncle

Temporal fibres

Cerebellar fibres

Internal capsule

Fig. 26.4 3D projection of DT tracking of the cortico-spinal tracts in a normal subject (Courtesy of Dr. Derek Jones, National Institute of Health, Bethesda, MD, USA).

for DWI, such as high b-value images (Assaf et al., 2002), which seem to have the potential to increase further the sensitivity of “conventional” low b-value DWI in the detection of NAWM abnormalities. Investigation of optic nerve and spinal cord damage Although DWI of the optic nerve and the spinal cord would be desirable to achieve a more complete picture of how inflammatory and demyelinating diseases can affect clinically eloquent sites in the CNS, DWI in these regions presents considerable technical challenge (Clark and Werring, 2002). Nevertheless, successful DWI of the optic nerve (Iwasawa et al., 1997; Wheeler-Kingshott et al., 2002b) and spinal cord (Bammer et al., 2000b, 2002; Clark et al., 2000; Ries et al., 2000; Robertson et al., 2000; Wheeler-Kingshott et al., 2002a; Cercignani et al., 2003b) has been recently obtained from normal controls and patients with different neurological conditions. At present, only one study assessed water diffusion in the optic nerve of patients with optic neuritis (Iwasawa et al., 1997), demonstrating significant different optic nerve ADC

454

Marco Rovaris and Massimo Filippi

(a)

(b)

(c)

(d)

– Fig. 26.5 Sagittal T2-weighted images (a and c) and D (b and d) maps of the cervical cord from two patients with MS and mild neurological disability. DTI scans were obtained using a sensitivity-encoded (SENSE) single-shot echo planar imaging (EPI) pulse sequence. The advantage of SENSE over standard EPI is the reduction of off-resonance artifacts, due to an increased bandwidth in the phase-encoding direction, which decreases the sensitivity to susceptibility variations. In (a), no T2-hyperintense cord lesions – are visible and there are no visible abnormalities on the corresponding D map (b). On the other hand, although a lesion is seen at – C2–C3 level in (c), it appears isointense on the corresponding D map (d). This suggests a relative sparing of the cervical cord tissue microstructural integrity in both patients.

values between controls and patients. This study also showed that ADC differs between acute and chronic optic neuritis cases: ADC was found to be decreased in the acute (inflammatory) stage of optic neuritis, and increased in the chronic phase. Another study assessed water diffusion in seven cord lesions of three MS patients with locomotor disability (Clark et al., 2000). They found that MS cord lesions had – higher D values than the cord tissue from healthy volunteers, thus confirming that DWI may quantify the severity of tissue damage. The clinical application of spinal cord DWI holds promise for the assessment of functional recovery after acute inflammatory or demyelinating pathologies, since it may enable to investigate the residual integrity of important sensory and motor pathways (Figure 26.5). In addition, because the cord contains uniformly oriented fiber tracts, some caveats of interpretation of anisotropy measures in the brain (e.g. those related to the presence of crossing fibers on a voxel scale) may be obviated.

Conclusions Conventional MRI is limited by its lack of specificity to the heterogeneous pathological substrates of inflammatory and demyelinating diseases of the CNS. DWI is one of the most promising MR techniques for overcoming, at least partially, these limitations. DWI allows to quantify the amount of tissue damage of T2-visible lesions and to detect more subtle changes occurring in NAWM and NAGM. Although DWI changes reflect a net loss of structural organization, we can only speculate on their possible pathological substrates. The investigation of the relationship between DWI metrics and other MR quantities derived from MT imaging, MR spectroscopy and fMRI might increase our understanding of this important issue. Post-mortem studies correlating DWI findings with histopathology are also warranted, as well as longitudinal studies to elucidate the correlation between evolving diffusion abnormalities and the accumulation of irreversible disability.

Diffusion imaging in demyelination and inflammation

REFERENCES Allen IV, McKeown SR. 1979. A histological, histochemical and biochemical study of the macroscopically normal white matter in multiple sclerosis. J Neurol Sci 41: 81–91. Assaf Y, Ben-Bashat D, Chapman J, Peled S, Biton IE, Kafri M, Segev Y, Hendler T, Korczyn AD, Graif M, Cohen Y. 2002. High b-value q-space analyzed diffusion-weighted MRI: application to multiple sclerosis. Magn Reson Med 47: 115–126. Bammer R, Augustin M, Strasser-Fuchs S, Seifert T, Kapeller P, Stollberger R, Ebner F, Hartung HP, Fazekas F. 2000a. Magnetic resonance diffusion tensor imaging for characterizing diffuse and focal white matter abnormalities in multiple sclerosis. Magn Reson Med 44: 583–59. Bammer R, Augustin M, Prokesch RW, Stollberger R, Fazekas F. 2002. Diffusion-weighted imaging of the spinal cord: interleaved echo-planar imaging is superior to fast spinecho. J Magn Reson Imaging 15: 364–373. Bammer R, Fazekas F, Augustin M, Simbrunner J, StrasserFuchs S, Seifert T, Stollberger R, Hartung HP. 2000b. Diffusion-weighted MR imaging of the spinal cord. Am J Neuroradiol 21: 587–591. Basser PJ, Mattiello J, Le Bihan D. 1994. Estimation of the effective self-diffusion tensor from the NMR spin-echo. J Magn Reson B 103: 247–254. Basser PJ, Pierpaoli C. 1996. Microstructural features measured using diffusion tensor imaging. J Magn Reson B 111: 209–219. Bernarding J, Braun J, Koennecke HC. 2002. Diffusion- and perfusion-weighted MR imaging in a patient with acute demyelinating encephalomyelitis (ADEM). J Magn Reson Imaging 15: 96–100. Blinkenberg M, Rune K, Jensen CV, Ravnborg M, Kyllingsbaek S, Holm S, Paulson OB, Sorensen PS. 2000. Cortical cerebral metabolism correlates with MRI lesion load and cognitive dysfunction in MS. Neurology 54: 558–564. Bozzali M, Cercignani M, Sormani MP, Comi G, Filippi M. 2002 Quantification of brain gray matter damage in different MS phenotypes by use of diffusion tensor MR imaging. Am J Neuroradiol 23: 985–988. Brownell B, Hughes JT. 1962. The distribution of plaques in the cerebrum in multiple sclerosis. J Neurol Neurosurg Psychiatry 25: 315–320. Brück W, Bitsch A, Kolenda H, Brück Y, Stiefel M, Lassmann H. 1997. Inflammatory central nervous system demyelination: correlation of magnetic resonance imaging findings with lesion pathology. Ann Neurol 42: 783–793. Caramia F, Pantano P, Di Legge S, Piattella MC, Lenzi D, Paolillo A, Nucciarelli W, Lenzi GL, Bozzao L, Pozzilli C. 2002. A longitudinal study of MR diffusion changes in normal appearing white matter of patients with early multiple sclerosis. Magn Reson Imaging 20: 383–388.

Castriota-Scanderbeg A, Fasano F, Hagberg G, Nocentini U, – Filippi M, Caltagirone C. 2003. Coefficient D is more sensitive than fractional anisotropy in monitoring the progression of the irreversible tissue damage in focal non-active multiple sclerosis lesions. Am J Neuroradiol 24: 663–670. Castriota-Scanderbeg A, Sabatini U, Fasano F, Floris R, Fraracci L, Mario MD, Nocentini U, Caltagirone C. 2002. Diffusion of water in large demyelinating lesions: a followup study. Neuroradiology 44: 764–767. Castriota-Scanderbeg A, Tomaiuolo F, Sabatini U, Nocentini U, Grasso MG, Caltagirone C. 2000. Demyelinating plaques in relapsing–remitting and secondary-progressive multiple sclerosis: assessment with diffusion MR imaging. Am J Neuroradiol 21: 862–868. Cercignani M, Iannucci G, Rocca MA, Comi G, Horsfield MA, Filippi M. 2000. Pathologic damage in MS assessed by diffusion-weighted and magnetization transfer MRI. Neurology 54: 1139–1144. Cercignani M, Bammer R, Sormani MP, Fazekas F, Filippi M. 2003a. Inter-sequence and inter-imaging unit variability of diffusion tensor MR imaging histogram-derived metrics of the brain in healthy volunteers. Am J Neuroradiol 24: 638–643. Cercignani M, Bozzali M, Iannucci G, Comi G, Filippi M. 2001a. Magnetisation transfer ratio and mean diffusivity of normal appearing white and grey matter from patients with multiple sclerosis. J Neurol Neurosurg Psychiatry 70: 311–317. Cercignani M, Bozzali M, Iannucci G, Comi G, Filippi M. 2002. Intra-voxel and inter-voxel coherence in patients with multiple sclerosis assessed using diffusion tensor MRI. J Neurol 249: 875–883. Cercignani M, Horsfield MA, Agosta F, Filippi M. 2003b. Sensitivity-encoded diffusion tensor MR imaging of the cervical cord. Am J Neuroradiol (in press). Cercignani M, Inglese M, Pagani E, Comi G, Filippi M. 2001b. Mean diffusivity and fractional anisotropy histograms in patients with multiple sclerosis. Am J Neuroradiol 22: 952–958. Chard DT, Griffin CM, McLean MA, Kapeller P, Kapoor R, Thompson AJ, Miller DH. 2002. Brain metabolite changes in cortical grey and normal-appearing white matter in clinically early relapsing–remitting multiple sclerosis. Brain 125: 2342–2352. Christiansen P, Gideon P, Thomsen C, Stubgaard M, Henriksen O, Larsson HBW. 1993. Increased water self-diffusion in chronic plaques and in apparently normal white matter in patients with multiple sclerosis. Acta Neurol Scand 87: 195–199. Ciccarelli O, Werring DJ, Wheeler-Kingshott CA, Barker GJ, Parker GJ, Thompson AJ, Miller DH. 2001. Investigation of MS normal-appearing brain using diffusion tensor MRI with clinical correlations. Neurology 56: 926–933. Ciccarelli O, Werring DJ, Barker GJ, Griffin CM, WheelerKingshott CA, Miller DH, Thompson AJ. 2003. A study of the

455

456

Marco Rovaris and Massimo Filippi

mechanisms of normal-appearing white matter damage in multiple sclerosis using diffusion tensor imaging. Evidence of wallerian degeneration. J Neurol 250: 287–292. Clark CA, Werring DJ. 2002. Diffusion tensor imaging in spinal cord: methods and applications – a review. NMR Biomed 15: 578–586. Clark CA, Werring DJ, Miller DH. 2000. Diffusion imaging of the spinal cord in vivo: estimation of the principal diffusivities and application to multiple sclerosis. Magn Reson Med 43: 133–138. Codella M, Rocca MA, Colombo B, Martinelli-Boneschi F, Comi G, Filippi M. 2002a. Cerebral grey matter pathology and fatigue in patients with multiple sclerosis: a preliminary study. J Neurol Sci 194: 71–74. Codella M, Rocca MA, Colombo B, Rossi P, Comi G, Filippi M. 2002b. A preliminary study of magnetization transfer and diffusion tensor MRI of multiple sclerosis patients with fatigue. J Neurol 249: 535–537. Conturo TE, Lori NF, Cull TS, Akbudak E, Snyder AZ, Shimony JS, McKinstry RC, Burton H, Raichle ME. 1999. Tracking neuronal fiber pathways in the living human brain. Proc Natl Acad Sci USA 96: 10422–10427. Cramer SC, Stegbauer KC, Schneider A, Mukai J, Maravilla KR. 2001. Decreased diffusion in central pontine myelinolisis. Am J Neuroradiol 22: 1476–1479. Dehmeshki J, Chard DT, Leary SM, Watt HC, Silver NC, Tofts PS, Thompson AJ, Miller DH. 2003. The normal appearing grey matter in primary progressive multiple sclerosis. A magnetisation transfer imaging study. J Neurol 250: 67–74. De Stefano N, Iannucci G, Sormani MP, Guidi L, Bartolozzi ML, Comi G, Federico A, Filippi M. 2002. MR correlates of cerebral atrophy in patients with multiple sclerosis. J Neurol 249: 1072–1077. Droogan AG, Clark CA, Werring DJ, Barker GJ, McDonald WI, Miller DH. 1999. Comparison of multiple sclerosis clinical subgroups using navigated spin echo diffusion-weighted imaging. Magn Reson Imaging 17: 653–661. Evangelou N, Konz D, Esiri MM, Smith S, Palace J, Matthews PM. 2000. Regional axonal loss in the corpus callosum correlates with cerebral white matter lesion volume and distribution in multiple sclerosis. Brain 123: 1845–1849. Filippi M, Bozzali M, Comi G. 2001a. Magnetization transfer and diffusion tensor MR imaging of basal ganglia from patients with multiple sclerosis. J Neurol Sci 183: 69–72. Filippi M, Cercignani M, Inglese M, Horsfield MA, Comi G. 2001b. Diffusion tensor magnetic resonance imaging in multiple sclerosis. Neurology 56: 304–311. Filippi M, Iannucci G, Cercignani M, Rocca MA, Pratesi A, Comi G. 2000. A quantitative study of water diffusion in multiple sclerosis lesions and normal-appearing white matter using echo-planar imaging. Arch Neurol 57: 1017–1021.

Filippi M, Rocca MA, Colombo B, Falini A, Codella M, Scotti G, Comi G. 2002a. Functional magnetic resonance imaging correlates of fatigue in multiple sclerosis. Neuroimage 15: 559–567. Filippi M, Rocca MA, Falini A, Caputo D, Ghezzi A, Colombo B, Scotti G, Comi G. 2002b. Correlations between structural CNS damage and functional MRI changes in primary progressive MS. NeuroImage 15: 537–546. Filippi M, Tortorella C, Bozzali M. 1999. Normal-appearingwhite-matter changes in multiple sclerosis: the contribution of magnetic resonance techniques. Mult Scler 5: 273–282. Griffin CM, Chard DT, Ciccarelli O, Kapoor B, Barker GJ, Thompson AJ, Miller DH. 2001. Diffusion tensor imaging in early relapsing–remitting multiple sclerosis. Mult Scler 7: 290–297. Guo AC, Jewells VL, Provenzale JM. 2001. Analysis of normal-appearing white matter in multiple sclerosis: comparison of diffusion tensor MR imaging and magnetization transfer imaging. Am J Neuroradiol 22: 1893–1900. Guo AC, MacFall JR, Provenzale JM. 2002. Multiple sclerosis: diffusion tensor MR imaging for evaluation of normal-appearing white matter. Radiology 222: 729–736. Hajnal JV, Doran M, Hall AS, Collins AG, Oatridge A, Pennock JM, Young IR, Bydder GM. 1991. MR imaging of anisotropically restricted diffusion of water in system: technical, anatomic, and pathologic considerations. J Comput Assist Tomogr 15: 1–18. Holtmannspotter M, Inglese M, Rovaris M, Rocca MA, Codella M, Filippi M. 2003. A diffusion tensor MRI study of basal ganglia from patients with ADEM. J Neurol Sci 206: 27–30. Horsfield MA, Lai M, Webb SL, Barker GJ, Tofts PS, Turner R, Rudge P, Miller DH. 1996. Apparent diffusion coefficients in benign and secondary progressive multiple sclerosis by nuclear magnetic resonance. Magn Reson Med 36: 393–400. Iannucci G, Mascalchi M, Salvi F, Filippi M. 2000. Vanishing Balò-like lesions in multiple sclerosis. J Neurol Neurosurg Psychiatry 69: 399–400. Iannucci G, Rovaris M, Giacomotti L, Comi G, Filippi M. 2001. Correlations between measures of multiple sclerosis pathology derived from T2, T1, magnetization transfer and diffusion tensor MR imaging. Am J Neuroradiol 22: 1462–1467. Inglese M, Salvi F, Iannucci G, Mancardi GL, Mascalchi M, Filippi M. 2002. Magnetization transfer and diffusion tensor MR imaging of acute disseminated encephalomyelitis. Am J Neuroradiol 23: 267–272. Iwasawa T, Matoba H, Ogi A, Kurihara H, Saito K, Yoshida T, Matsubara S, Nozaki A. 1997. Diffusion-weighted imaging of the human optic nerve: a new approach to evaluate optic neuritis in multiple sclerosis. Magn Reson Med 38: 484–491. Katz D, Taubenberger JK, Cannella B, McFarlin DE, Raine CS, McFarland HF. 1993. Correlation between magnetic

Diffusion imaging in demyelination and inflammation

resonance imaging findings and lesion development in multiple sclerosis. Ann Neurol 34: 661–669. Kermode AG, Tofts P, Thompson AJ, MacManus DG, Rudge P, Kendall BE, Kingsley DP, Moseley IF, du Boulay EP, McDonald WI. 1990. Heterogeneity of blood–brain barrier changes in multiple sclerosis: an MRI study with gadoliniumDTPA enhancement. Neurology 40: 229–235. Kidd D, Barkhof F, McConnel R, Algra PR, Allen IV, Revesz T. 1999. Cortical lesions in multiple sclerosis. Brain 122: 17–26. Larsson HBW, Thomsen C, Frederiksen J, Stubgaard M, Henriksen O. 1992. In vivo magnetic resonance diffusion measurement in the brain of patients with multiple sclerosis. Magn Reson Imaging 10: 7–12. Le Bihan D, Breton E, Lallemand D, Grenier P, Cabanis E, Laval-Jeanter M. 1986. MR imaging of intravoxel incoherent motions: application to diffusion and perfusion in neurologic disorders. Radiology 161: 401–407. Le Bihan D, Turner R, Pekar J, Moonen CTW. 1991. Diffusion and perfusion imaging by gradient sensitization: design, strategy and significance. J Magn Reson Imaging 1: 7–8. Mainero C, De Stefano N, Iannucci G, Sormani MP, Guidi L, Federico A, Bartolozzi ML, Comi G, Filippi M. 2001. Correlates of MS disability assessed in-vivo using aggregates of MR quantities. Neurology 56: 1331–1334. McDonald WI, Compston A, Edan G, Goodkin D, Hartung HP, Lublin FD, McFarland HF, Paty DW, Polman CH, Reingold SC, Sandberg-Wollheim M, Sibley W, Thompson A, van den Noort S, Weinshenker BY, Wolinsky JS. 2001. Recommended diagnostic criteria for multiple sclerosis: guidelines from the international panel on the diagnosis of multiple sclerosis. Ann Neurol 50: 121–127. Mori S, Crain BJ, Chacko VP, van Zijl PC. 1999. Threedimensional tracking of axonal projections in the brain by magnetic resonance imaging. Ann Neurol 45: 265–269. Mori S, Kaufmann WE, Davatzikos C, Stieltjes B, Amodei L, Fredericksen K, Pearlson GD, Melhem ER, Solaiyappan M, Raymond GV, Moser HW, van Zijl PC. 2002. Imaging cortical association tracts in the human brain using diffusion-tensor-based axonal tracking. Magn Reson Med 47: 215–223. Nusbaum AO, Lu D, Tang CY, Atlas SW. 2000a. Quantitative diffusion measurements in focal multiple sclerosis lesions: correlations with appearance on T1-weighted MR images. Am J Roentgenol 175: 821–825. Nusbaum AO, Tang CY, Wei TC, Buchsbaum MS, Atlas SW. 2000b. Whole-brain diffusion MR histograms differ between MS subtypes. Neurology 54: 1421–1426. Ohta K, Obara K, Sakauchi M, Obara K, Takane H, Yogo Y. 2001. Lesion extension detected by diffusion-weighted magnetic resonance imaging in progressive multifocal leukoencephalopathy. J Neurol 248: 809–811.

Oreja-Guevara C, Rovaris M, Caputo D, Cavarretta R, Sormani MP, Ferrante P, Comi G, Filippi M. 2003. Changes in cortical gray matter in untreated relapsing–remitting MS patients: a follow-up study (Abstract). Neurology 60 (suppl. 1): A297 Peterson JW, Bo L, Mork S, Chang A, Trapp BD. 2001. Transected neurites, apoptotic neurons and reduced inflammation in cortical multiple sclerosis lesions. Ann Neurol 50: 389–400. Pierpaoli C, Jezzard P, Basser PJ, Blarnett A, Di Chiro G. 1996. Diffusion tensor MR imaging of the human brain. Radiology 201: 637–648. Ries M, Jones RA, Dousset V, Moonen CTW. 2000. Diffusion tensor MRI of the spinal cord. Magn Reson Med 44: 884–892. Robertson RL, Maier SE, Mulkern RV, Vajapayam S, Robson CD, Barnes PD. 2000. MR line-scan diffusion imaging of the spinal cord in children. Am J Neuroradiol 21: 1344–1348. Rocca MA, Cercignani M, Iannucci G, Comi G, Filippi M. 2000. Weekly diffusion-weighted imaging of normal-appearing white matter in MS. Neurology 55: 882–884. Rocca MA, Falini A, Colombo B, Scotti G, Comi G, Filippi M. 2002a. Adaptive functional changes in the cerebral cortex of patients with nondisabling multiple sclerosis correlate with the extent of brain structural damage. Ann Neurol 51: 330–339. Rocca MA, Matthews PM, Caputo D, Ghezzi A, Falini A, Scotti G, Comi G, Filippi M. 2002b. Evidence for widespread movement-associated functional MRI changes in patients with PPMS. Neurology 58: 866–872. Rocca MA, Pagani E, Ghezzi A, Falini A, Zaffaroni M, Colombo B, Scotti G, Comi G, Filippi M. 2003. Functional cortical changes in patients with MS and nonspecific findings on conventional MRI scans of the brain. NeuroImage 79: 826–836. Roelcke U, Kappos L, Lechner-Scott J, Brunnschweiler H, Huber S, Ammann W, Plohmann A, Dellas S, Maguire RP, Missimer J, Radu EW, Steck A, Leenders KL. 1997. Reduced glucose metabolism in the frontal cortex and basal ganglia of multiple sclerosis patients with fatigue: a 18 F-fluorodeoxyglucose positron emission tomography study. Neurology 48: 1566–1571. Rovaris M, Bozzali M, Iannucci G, Ghezzi A, Caputo D, Montanari E, Bertolotto A, Bergamaschi R, Capra R, Mancardi GL, Martinelli V, Comi G, Filippi M. 2002a. Assessment of normal-appearing white and gray matter in patients with primary progressive multiple sclerosis. Arch Neurol 59: 1406–1412. Rovaris M, Iannucci G, Falautano M, Possa F, Martinelli V, Comi G, Filippi M. 2002b. Cognitive dysfunction in patients with mildly disabling relapsing–remitting multiple sclerosis: an exploratory study with diffusion tensor MR imaging. J Neurol Sci 195: 103–109.

457

458

Marco Rovaris and Massimo Filippi

Roychowdhury S, Maldijan JA, Grossman RI. 2000. Multiple sclerosis: comparison of trace apparent diffusion coefficients with MR enhancement pattern of lesions. Am J Neuroradiol 21: 869–874. Tench CR, Morgan PS, Wilson M, Blumhardt LD. 2002. White matter mapping using diffusion tensor MRI. Magn Reson Med 47: 967–972. Thompson AJ, Polman CH, Miller DH, McDonald WI, Brochet B, Filippi M, Montalban X, De Sa J. 1997. Primary progressive multiple sclerosis. Brain 120: 1085–1096. Truyen L, van Waesberghe JHTM, van Walderveen MAA, van Oosten BW, Polman CH, Hommes OR, Ader HJ, Barkhof F. 1996. Accumulation of hypointense lesions (“black holes”) on T1 spin-echo MRI correlates with disease progression in multiple sclerosis. Neurology 47: 1469–1476. Vaithianathar L, Tench CR, Morgan PS, Wilson M, Blumhardt LD. 2002. T1 relaxation time mapping of white matter tracts in multiple sclerosis defined by diffusion tensor imaging. J Neurol 249: 1272–1278. van Waesberghe JHTM, Kamphorst W, De Groot CJA, van Walderveen MA, Castelijns JA, Ravid R, Lycklama a Nijeholt GJ, van der Valk P, Polman CH, Thompson AJ, Barkhof F. 1999. Axonal loss in MS lesions: MR insights into substrates of disability. Ann Neurol 46: 747–754. van Walderveen MAA, Kamphorst W, Scheltens P, van Waesberghe JH, Ravid R, Valk J, Polman CH, Barkhof F. 1998.

Histopathologic correlate of hypointense lesions on T1-weighted spin-echo MRI in multiple sclerosis. Neurology 50: 1282–1288. Werring DJ, Brassat D, Droogan AG, Clark CA, Symms MR, Barker GJ, MacManus DG, Thompson AJ, Miller DH. 2000. The pathogenesis of lesions and normal-appearing white matter changes in multiple sclerosis. A serial diffusion MRI study. Brain 123: 1667–1676. Werring DJ, Clark CA, Barker GJ, Thompson AJ, Miller DH. 1999. Diffusion tensor imaging of lesions and normalappearing white matter in multiple sclerosis. Neurology 52: 1626–1632. Wheeler-Kingshott CA, Hickman SJ, Parker GJ, Ciccarelli O, Symms MR, Miller DH, Barker GJ. 2002a. Investigating cervical spinal cord structure using axial diffusion tensor imaging. NeuroImage 16: 93–102. Wheeler-Kingshott CA, Parker GJM, Symms MR, Hickman SJ, Tofts PS, Miller DH, Barker GJ. 2002b. ADC mapping of the human optc nerve: increased resolution, coverage, and reliability with CSF-suppressed ZOOM-EPI. Magn Reson Med 47: 24–31. Wilson M, Tench CR, Morgan PS, Blumhardt LD. 2003. Pyramidal tract mapping by diffusion tensor magnetic resonance imaging in multiple sclerosis: improving correlations with disability. J Neurol Neurosurg Psychiatry 74: 203–207.

Diffusion imaging in demyelination and inflammation

Case Study 26.1 Tumefactive MS: MR perfusion Peter Barker, D.Phil., Johns Hopkins University School of Medicine, Baltimore History 41-year-old male presenting with a focal seizure.

FLAIR

T1Post-Gd

DWI

ADC

CBV

Follow-up FLAIR

Technique Conventional, diffusion- and gadolinium bolus perfusion-weighted MRI.

Imaging findings FLAIR images demonstrate a hyperintense lesion in the right frontal lobe which shows incomplete peripheral enhancement. DWI is mildly hyperintense; the high ADC indicates that this is due to T2 shine-through. PWI shows reduced cerebral blood volume CBV in the lesion compared to contralateral WM.

Discussion Conventional imaging and DWI in this case could not rule out a primary glioma. However, reduced CBV is unusual in neoplasia, particularly of high grade (Cha et al., 2002; Hacklander et al., 1996). This observation, coupled with the incomplete rim enhancement pattern, made a tumefactive MS plaque more likely. The lesion regressed on follow-up imaging without surgical intervention.

Key points CBV is usually not increased in MS plaques. Tumor CBV tends to be increased in high grade tumors.

References Cha S, Pierce S, Knopp EA, Johnson G, Yang C, Ton A, Litt AW, Zagzag D. 2000. Dynamic contrast-enhanced T*2- weighted MR imaging of tumefactive demyelinating lesion. Am J Neuroradiol 22: 1109–1116. Hacklander T, Reichenbach JR, Hofer M, Modder U.1996. Measurement of cerebral blood volume via the relaxing effect of low-dose gadopentetate dimeglumine during bolus transit. Am J Neuroradiol 17: 821–830.

459

27

Physiological MR to evaluate HIV-associated brain disorders Linda Chang and Thomas Ernst Department of Medicine, University of Hawaii, John A. Burns School of Medicine, Honolulu, Hawaii

Key points • Abnormal metabolites are seen in the brains of neurologically asymptomatic human immunodeficiency virus (HIV) patients; notably elevated choline and myoinositol. • Decreased N-acetyl aspartate (NAA) and NAA/creatine are associated with neurological symptoms. • Metabolite abnormalities reverse with antiretroviral therapy, and may provide a surrogate marker of treatment response. • Both increased and decreased perfusion have been detected in different regions of the brain; the magnitude of some correlate with virological and cognitive indices. • Diffusion studies show increased apparent diffusion coefficient and decreased anisotropy; particularly in frontal white matter and basal ganglia. • MR spectroscopy (MRS) is helpful for distinguishing toxoplasma from lymphoma lesions. • MRS, perfusion and diffusion imaging are useful for characterizing opportunist brain infections in HIV patients.

Introduction Physiological MR imaging (MRI) and MR spectroscopy (MRS) are highly sensitive, objective, noninvasive techniques to monitor the severity of brain injury as well as the effects of treatments. MR techniques have no radiation and thus are ideal for 460

monitoring progression of disease or treatment effects when repeat measurements are needed. Several recent MR techniques, including proton 1 H MRS, functional MRI (fMRI), and physiological MRI techniques such as diffusion MRI, magnetization MRI, and perfusion MRI, have all been applied to evaluate brain injury and opportunistic infections or tumors associated with human immunodeficiency virus (HIV) infection and acquired immunodeficiency syndrome (AIDS). Since most of these techniques are available on commercial MR machines, they are particularly suitable for diagnostic purposes and for monitoring treatment effects. This chapter delineates salient features of HIV associated brain injury (both in HIV dementia and HIV neurologically asymptomatic individuals) on MRS and physiological MRI, as well as opportunistic infections associated with HIV. Future directions for the applications of these studies to evaluate the pathophysiology of HIV associated brain injury and for treatment monitoring will be discussed.

HIV associated brain injury or HIV dementia HIV patients with normal-appearing structural brain imaging may demonstrate abnormalities on various physiological and neurochemical measurements. In vitro, preclinical and in vivo human studies strongly indicate that a major pathway for neuropathogenesis of HIV dementia is central nervous system (CNS) invasion by HIV infected monocytic cells (macrophage and microglia). Monocytes release or transport neurotoxic proteins (Tat, gp120,

Physiological MR to evaluate HIV-associated brain disorders

and cytokines), which may lead to neuronal dysfunction or loss, and glial activation. Therefore, MRS, which can assess metabolites related to neuronal and glial cell populations, has played an important role in assessing the severity of brain injury in HIV patients. 1

H MRS in HIV patients

Although both phosphorus and proton MRS have been applied to study HIV dementia, 1H MRS has higher sensitivity and is much more common in the clinical setting. Therefore, it has been applied much more extensively. Since 1H MRS provides in vivo non-invasive evaluation of several metabolites that reflect neuronal integrity and glial responses, it may be used to assess the severity of HIV associated brain injury, and to monitor the effects of antiretroviral treatments in both adult and pediatric populations. Localized MRS studies employing long echo times (TE) (Menon et al., 1992; Chong et al., 1993; Paley et al., 1996; Salvan et al., 1997b) as well as short TE (19–35 ms) (Jarvik et al., 1993; Paley et al., 1995; Laubenberger et al., 1996; Tracey et al., 1996; English et al., 1997; Chang et al., 1999a) found that HIV dementia is associated with decreased N-acetyl aspartate (NAA) or NAA relative to total creatine (Cr) (NAA/Cr), suggesting neuronal injury or dysfunction, especially in those with more severe dementia. However, neurologically asymptomatic HIV patients have minimal or no change in NAA/Cr (Menon et al., 1992; Chong et al., 1994; Laubenberger et al., 1996; Paley et al., 1996; Tracey et al., 1996; Salvan et al., 1997a; Chang et al., 1999a) or NAA (Barker et al., 1995; Chang et al., 1999a; Meyerhoff et al., 1999). Consistent with neuropathological findings of increased number of macrophages and microglial cells in the brains of AIDS patients, elevated levels of choline-containing compounds (Cho) or Cho/Cr suggesting such inflammation are often observed on MRS in HIV patients (Jarvik et al., 1993; Tracey et al., 1996; English et al., 1997; Chang et al., 1999a). Alternatively, higher Cho compounds might be related to increased cell membrane break down and release of soluble Cho compounds due to direct or indirect effects of HIV infection. Elevated Cho is

also found in patients with minor cognitive motor disorder (MCMD), but more so in those with moderate to severe dementia (Figure 27.1) (Barker et al., 1995; Chang et al., 1999a; Meyerhoff et al., 1999). More recent studies, however, reported little or no change in Cho during the early stages of HIV dementia (Marcus et al., 1998; Suwanwelaa et al., 2000; von Giesen et al., 2001; Chang et al., 2002), which might reflect the changing disease pattern of HIV dementia or the antiretroviral medication-naïve status of these subjects (Suwanwelaa et al., 2000; Chang et al., 2002). Elevated myoinositol (mI) or mI/Cr, which is observable only at short TE, has been reported at various stages of HIV dementia (Laubenberger et al., 1996; Lopez-Villegas et al., 1997; Chang et al., 1999a; von Giesen et al., 2001). Since mI is present primarily in glial cells (Brand et al., 1993) and has the putative function of regulating the cellular osmotic environment and maintaining the cell volume (Graf et al., 1993), glial activation or hypertrophy induced by HIV-mediated cytokines and chemokines would be associated with elevated cytoplasmic mI. HIV patients indeed show elevated mI (Chang et al., 2002) during the neurologically asymptomatic stage and further elevation with more severe dementia (Figure 27.1), especially in the frontal white matter (WM) where glial activation is often observed in neuropathological studies (Power et al., 1993). Total Cr, commonly used as an internal reference on MRS, may also change depending on HIV disease severity, dementia stage, and brain region, as well as age of the patient. Thus, when used as an internal reference, HIV patients with decreased NAA/Cr may actually have either normal or mildly decreased [NAA], along with elevated [Cr] (Chang et al., 1999a, 1999b). In the later stages of AIDS dementia complex, [Cr] is significantly elevated in the frontal lobe, but decreased in the basal ganglia, along with decreased [NAA] (Chang et al., 2002). The elevated [Cr] may also obscure concomitant elevations in Cho and mI. Techniques for quantitative MRS are discussed in Chapter 2, and are generally preferable to metabolite ratio measurements in most circumstances, but particularly when changes in Cr (or all metabolites) may be occurring.

461

462

Linda Chang and Thomas Ernst

Cho ADC 3

Severe dementia

ADC 2 Frontal WM

Moderate dementia

ADC 1 Mild dementia

mI ADC 0.5 MCMD

NAA Cr 4

3

2 ppm

1

0

Seronegative control

Fig. 27.1 Compared to seronegative control subjects, the HIV patients show elevated mI early in the course of dementia apparent diffusion coefficient (ADC, 0.5), and additional mild elevation of Cho and further elevation of mI with disease progression. Note that NAA is decreased only during moderate and severe dementia stages. MR spectra were acquired with a point resolved spectroscopy (PRESS) sequence (repetition time (TR)/TE  3000/30 ms, 64 averages) in the frontal WM (see voxel location on axial MRI). Figure modified from Chang et al. (1999).

Regional variations in metabolite abnormalities Technical advances, such as adjustments of the slice order (Ernst and Chang, 1996), have made it possible to evaluate the frontal lobe and subcortical brain regions, including the basal ganglia and the thalamus. Elevated Cho and mI are observed in the frontal WM of neurologically asymptomatic HIV patients and those with mild dementia (English et al., 1997), and extend to the frontal gray matter (GM) and basal ganglia in those with more severe dementia (Chang et al., 1999a; von Giesen et al., 2001). Regional variations can be more efficiently evaluated with spectroscopic imaging (SI) techniques (Meyerhoff et al., 1993, 1994, 1996, 1999; Barker et al., 1995; Lopez-Villegas et al., 1997; Marcus et al., 1998; Moller et al., 1999), especially multi-slice MR spectroscopic imaging (MRSI) (Figure 27.2). A multi-slice MRSI study found higher Cho and lower NAA in the frontal WM of patients with AIDS dementia complex and lower CD4 count, but primarily higher Cho without decreased NAA in the frontal WM in patients with

only MCMD and higher CD4 count (Meyerhoff et al., 1999).

Reversal of metabolite abnormalities after antiretroviral treatment Several longitudinal MRS studies evaluated the effects of antiretroviral treatment in HIV patients, and most found improvements in metabolite ratios or concentrations in HIV patients after successful antiretroviral treatment. Two studies evaluated HIV patients before and after zidovudine (ZDV) treatment, and found that NAA/Cr or NAA/Cho improved (increased) after treatment along with clinical improvement (Wilkinson et al., 1997), especially in those with decreased NAA/Cr at baseline (Salvan et al., 1997). Since these studies relied on metabolite ratios, changes in NAA/Cr or NAA/Cho might be due to decreases in Cr and Cho after treatment. Highly active antiretroviral therapy (HAART) may reverse brain metabolite abnormalities in patients

Physiological MR to evaluate HIV-associated brain disorders

T2

Cho

NAA

MCMD (ADC stage 0.5)

HIV Dementia (Severe dementia)

Fig. 27.2 SI from two patients with HIV. Top row: 58-year-old HIV patient with mild cognitive motor disorder (MCMD) and CD4 count of 300/mm3; note mildly elevated Cho compounds in the frontal WM region with relatively normal NAA. Bottom row: 35-year-old HIV patient with AIDS dementia complex and CD4 count of 143/mm3 show further elevation of Cho compounds and decreased NAA in the frontal brain region. Spectroscopic images were acquired on a 1.5 T MR machine, PRESS, TR/TE  2300/272 ms (courtesy of Dr. Peter Barker, modified from data presented in Barker et al., 1995).

with mild dementia. Normalization of the initially increased mI concentration (prior to HAART), and to a lesser degree normalization of Cho concentration, in the frontal WM and basal ganglia, may occur after 9 months (Figure 27.3) (Chang et al., 1999b). Another study of HIV patients also found that those with cognitive impairment, and lower NAA/Cr and higher mI/Cr at baseline showed improvement on the metabolite abnormalities (Stankoff et al., 2001). However, metabolite abnormalities (elevated Cho and mI) may persist after 3 months of HAART (Chang et al., 2000b), and improve only after 6–9 months (Chang et al., 2001b), although viral loads and CD4 counts usually rebound dramatically and rapidly within 1–3 months after onset of HAART. Therefore, monitoring treatment with MRS should include a

baseline assessment with follow-up evaluation at 6–9 months after treatment. MRS abnormalities in pediatric HIV patients The incidence for vertical transmission of HIV has declined by two-third since the introduction of ZDV prophylaxis during the perinatal period (Fiscus et al., 1996). However, the incidence of HIV encephalopathy in infants born to HIV positive mothers in developing countries remains high. MRS studies of brain abnormalities in pediatric HIV patients (Cortey et al., 1994; Lu et al., 1996; Pavlakis et al., 1998; Salvan et al., 1998) found that children with encephalopathy had decreased NAA and elevated mI, as well as increased lipid signals in the centrum semiovale (Salvan et al., 1998).

463

464

Linda Chang and Thomas Ernst

ADC Stage 1 (before treatment) NAA

ADC Stage 0 (9 months after treatment)

Cho Cr mI

4

3

2 ppm

1

0

4

3

2 ppm

1

0

Fig. 27.3 Coronal T2WI showing the voxel location where both spectra were acquired. Compared to the MR spectrum before treatment (left), the spectrum obtained after HAART shows reversal of MRS metabolite abnormalities (right, normalization of the initially elevated Cho and mI) in an AIDS patient. The patient also showed clinical improvement. (Modified from Chang et al., 1999b).

HIV-infected children who were neurologically asymptomatic, however, showed normal NAA/Cr but lower Cho/Cr in the basal ganglia region (Lu et al., 1996). Relative to structural MRI or immunological testing, MRS appears to be more sensitive for detecting brain abnormalities. Two children with AIDS encephalopathy showed normalization of NAA/Cr and disappearance of lactate (Lac) in the basal ganglia 4–8 months after treatment with ZDV (Pavlakis et al., 1998). These studies demonstrate that MRS might be useful for monitoring treatment effects in HIV-infected children. Perfusion MRI in HIV dementia Perfusion MRI using the dynamic susceptibility contrast imaging (DSCI) bolus-tracking technique has been performed in small groups of HIV patients. One study reported increased relative regional cerebral blood volume (rCBV) in the deep GM and cortical GM of HIV patients (Tracey et al., 1998); patients with mild to moderate dementia had higher rCBV compared to those without dementia. These findings were thought to reflect subcortical inflammatory changes, and were in general agreement with prior positron emission tomography (PET) findings of increased subcortical glucose metabolism during the early stages of HIV dementia (Rottenberg et al., 1987). However, previous PET and single-photon emission computed tomography (SPECT) studies in

HIV dementia patients also found decreased cortical metabolism or perfusion (Rottenberg et al., 1987; Pohl et al., 1988; Masdeu et al., 1991; Holman et al., 1992; Harris et al., 1994; Schwartz et al., 1994; Rosci et al., 1996; Rottenberg et al., 1996), which has been shown in another perfusion MRI study that evaluated regional cerebral blood flow (rCBF) in the entire brain, using voxel-by-voxel comparisons. Decreased rCBF was observed bilaterally in the inferior lateral frontal cortices and in an inferior medial parietal region, while increased rCBF was present bilaterally in the posterior inferior parietal WM of HIV patients compared to control subjects (Chang et al., 2000a) (Figure 27.4). The rCBF abnormalities correlated with CD4 count, plasma viral load, Karnofsky score and HIV dementia scale. Therefore, both studies observed increased perfusion (rCBV or rCBF), suggesting a reactive inflammatory processes or glial proliferation in patients with HIV brain injury. rCBF is related to regional metabolism and neuronal activity (Raichle, 1987), while the relationship between rCBV and neuronal activity is less clear. Diffusion MRI in HIV dementia Diffusion MRI is based on the microscopic, random movement of molecules (“Brownian motion”). Diffusion is characterized by the apparent diffusion coefficient (ADC) or diffusivity, which is related to

Physiological MR to evaluate HIV-associated brain disorders

Seronegative control

L

SPM

HIV-CMC

L

R

R

Fig. 27.4 Perfusion MRI in a seronegative control subject (left) and in an HIV patient (ADC stage 2, right). Bottom row: Group comparison (voxel-by-voxel using statistical parametric mapping, SPM) between seronegative controls and HIV patients (data also presented in Chang et al., 2000a). Green regions on the SPM map indicate significantly decreased rCBF while yellow regions on the SPM map denote elevated rCBF in the HIV patients compared to controls. Hence, rCBF may be abnormally decreased or elevated depending on the pathology and brain regions.

the average distance traveled by a molecule in a given time. As diffusion in tissues, but not in pure fluids, is restricted by cell membranes and other structures, the ADC values of tissues are typically smaller than that of pure water. In pure fluids and many body tissues, diffusion is independent of spatial orientation, or “isotropic”. However, the ADC may also be directional (“anisotropic”), for instance, in WM fiber tracts in the brain. In case of anisotropic diffusion, an “anisotropy” index can be calculated that reflects the amount of directionality. The diffusivity and anisotropy values of tissue are very sensitive to changes in the microscopic environment, such as changes in water content and cell membrane destruction.

Several studies have applied diffusion MRI to evaluate the microstructural integrity of brain tissues in HIV patients. The typical findings include increased diffusivity and decreased fractional anisotropy (FA) in normal appearing WM (NAWM) and the corpus callosum, especially in those with higher viral load and lower CD4 counts (Filippi et al., 2001; Pomara et al., 2001). A recent study also found increased diffusivity in the basal ganglia region that correlated with psychomotor slowing in patients with mild HIV dementia (Cloak et al., 2003). Since diffusion MRI appears to be more sensitive than structural MRI, it has the potential to detect microscopic or mild brain injury. One study even suggested

465

466

Linda Chang and Thomas Ernst

that diffusion tensor imaging (DTI) might indicate whether a patient is at risk for developing dementia (Schaefer et al., 2001). No study has yet evaluated whether these diffusion abnormalities may improve with antiretroviral treatment. Diffusion imaging in HIV-related dementia is also discussed in Chapter 33.

Opportunistic focal brain lesions in AIDS HIV preferentially infects the T-lymphocyte CD4 cells and thereby compromises the immune system. Numerous opportunistic organisms are able to penetrate and flourish in the CNS of HIV patients, especially when the CD4 cell count declines below 200/mm3. Other factors, such as substance abuse (Siddiqui et al., 1993; Chiappelli et al., 1994), poor nutrition, or other systemic illnesses may further suppress the immune system and hasten the opportunistic infections. The most common opportunistic focal brain lesions are cerebral toxoplasmosis, progressive multi-focal leukoencephalopathy (PML), and primary brain lymphoma. Cryptococcoma are occasionally seen with cryptococcal meningitis. Toxoplasma abscess and CNS lymphoma lesions are particularly difficult to differentiate clinically and with conventional neuroimaging procedures, including computerized tomography (CT) and routine MRI. In addition, the two diseases have been documented to coexist, which further complicates diagnosis (Navia et al., 1986; Levy and Bredesen, 1988; Rodesch et al., 1989; Iglesias-Rozas et al., 1991; Cornford et al., 1992; Gray and Sharer, 1993; Chang et al., 1995c). Differential diagnoses of focal brain lesions in AIDS also include tuberculoma, syphilitic gumma, other bacterial abscesses, and focal encephalitic lesions of cytomegalovirus (cf. Case Study 27.1). Physiological imaging may help with early diagnosis so that biopsies may be avoided in some patients (Chang et al., 1995c). Since AIDS patients without prior knowledge of their HIV status may present with focal opportunistic brain lesions as their index diagnosis, rapid and accurate diagnosis using MRS and physiological MRI techniques are particularly important in the management of these lesions.

Toxoplasmosis Toxoplasmosis is caused by opportunistic infection with the obligate intracellular protozoan Toxoplasma gondii. Approximately 70% of cerebral toxoplasmosis lesions, or abscesses, are multi-focal (Weisberg et al., 1988; Dina, 1991). Neurological presentations include subacute headaches, fever, seizures, focal neurological signs (Navia et al., 1986; Rodesch et al., 1989), or progressive dementia (Arendt et al., 1991). Serology for toxoplasma is frequently positive but specificity is low; only one-third of cases show a rise in the titer of immunoglobulin G (IgG) antibody (Derouin et al., 1991) and only 50% show intrathecal production of antibodies to T. gondii (Luft et al., 1984). Equally disappointing are the recent studies in polymerase chain reaction (PCR) for T. gondii in plasma and cerebrospinal fluid (CSF), which showed low sensitivity (Weiss et al., 1991; Cristina et al., 1993) and occasional false positive results (Burg et al., 1989). As a result, the clinical response to anti-toxoplasma therapy has been the main criterion for diagnosis. Toxoplasmosis lesions are most commonly located in the cerebral WM and subcortical GM, such as thalamus and basal ganglia (Navia et al., 1986; Ramsey and Geremia, 1988). CT characteristically shows multiple ring-enhancing lesions (Navia et al., 1986; Ramsey and Geremia, 1988), although solitary lesion or hypodense non-enhancing lesions have been reported (Weisberg et al., 1988). MRI demonstrates multiple discrete hyperintense foci on T2-weighted images that are mostly heterogeneous, have well-circumscribed margins, and are enhanced by gadolinium (Kaissar et al., 1991). Edema and hemorrhages are commonly associated with these lesions (Trenkwalder et al., 1992). MRS of toxoplasmosis lesions show characteristic features, which are helpful in differentiating toxoplasmosis from lymphoma in the majority of the cases (Confort-Gouny et al., 1993; Yamagata et al., 1994; Chang et al., 1995a) (Figure 27.5). In a toxoplasmosis lesion, Lac and lipids are markedly elevated while all other normal brain metabolites are virtually absent (Chang et al., 1995a; Chinn et al., 1995; Chang and Ernst, 1997). This MRS pattern reflects the anerobic acellular environment within an abscess, and the inflammatory response (including

Physiological MR to evaluate HIV-associated brain disorders

Lipids/Lac

TE  30 ms

Lac

TE  270 ms 4

2 ppm

3

Toxoplasmosis

1

0

Lipids/Lac

Cho

NAA Cr

mI

TE  30 ms

TE  270 ms

4

3

Lymphoma

2 ppm

1

0

Fig. 27.5 Axial post-gadolinium T1WI showing ring-enhancement by gadolinium contrast agent in toxoplasmosis and lymphoma lesion. Note thinner rim of toxoplasmosis lesion (top) compared to the lymphoma lesion (bottom). Characteristic MR spectra show marked elevation of lipids in the toxoplasmosis lesion at 30 ms and the Lac doublet at 270 ms. Bottom spectra are characteristic for lymphoma lesions with the prominent Cho peak and presence of some lipids and Lac as well as small amounts of NAA, Cr and mI due to partial volume effect (PVE) of the surrounding brain tissue.

macrophages) surrounding the abscess. This spectral profile, however, may resemble that observed in the necrotic center of brain tumors in non-AIDS patients (Chang et al., 1995b) or the necrotic center of a rapidly growing lymphoma (Chinn et al., 1995). Other abscesses, such as tuberculoma or fungal abscesses, may also show similar patterns of increased lipids and Lac. In addition, dynamic susceptibility MRI (or perfusion MRI) and thallium-201 SPECT demonstrate decreased perfusion or metabolism in toxoplasmosis or other abscesses. This helps to differentiate a neoplastic lesion from one that is infectious since

neoplastic lesions will show increased vascularity, while infectious lesions will show decreased rCBF throughout the lesion and in the surrounding edema (Figure 27.6). Magnetization transfer (MT) MRI and diffusion MRI can further assess the physiological changes associated with these lesions. For example, the ADC on diffusion MRI is typically increased in the surrounding edematous or inflammatory brain regions (Figure 27.6); this increase is typically more fulminate in toxoplasmosis than that observed with lymphoma lesions (Figure 27.6). Treatment of toxoplasmosis requires antibiotics (pyrimethamine and sulfadiazine or clindamycin); if

467

468

Linda Chang and Thomas Ernst

Post-Gd T1

FLAIR

MTR Map

Post-Gd T1

Regional blood volume map

ADC map

Toxoplasmosis

CNS lymphoma

Regional blood volume map

ADC map

Fig. 27.6 Structural and physiological images of toxoplasmosis (top row) and lymphoma (bottom row) in two AIDS patients. Both lesions types show gadolinium contrast enhancement on T1WI. However, significantly decreased rCBV on the DSC perfusion MRI and increased diffusion on the ADC map are observed in the toxoplosmosis lesion (top row). In contrast, the lymphoma lesions show moderately decreased magnetisation transfer ratio (MTR), markedly elevated rCBV at the rim of the ring-shaped lesions on the perfusion map and increased diffusion in the surrounding edema of the lesions on the ADC maps.

diagnosed early and no cerebral infarction has occurred due to the mass effect, clinical improvement is apparent within 1–2 weeks although the lesion(s) on MRI may persist for longer periods of time. The use of physiological MRI to monitor disease progression or treatment effects has not yet been reported.

CNS Lymphoma Primary CNS lymphoma in AIDS is nearly always of high-grade B-cell type. Since most cells contain the Epstein–Barr virus (EBV) (Rosenberg et al., 1986; Levine, 1992), it has been hypothesized that CNS lymphoma in AIDS may arise from EBV-infected B cells (Bashir et al., 1993). Between 19% and 71% of primary brain lymphoma present as a solitary lesion on neuroimaging studies (Ciricillo and Rosenblum, 1990; Dina, 1991), but may become multi-centric

rapidly, as reported in 80–100% of autopsies of AIDS patients (Morgello et al., 1990). Typical clinical presentation consists of progressive neurological deterioration with encephalopathy, focal signs, and seizures leading to death within 4–8 months despite radiation therapy (RT) or steroid treatment. Rarely, patients may go into clinical remission and may live for many years. CSF cytology is rarely diagnostic and brain biopsy is generally required for diagnosis (Levy et al., 1990). However, CSF PCR for EBV deoxyribonucleic acid (DNA) appears to be sensitive and specific for confirmation of CNS lymphoma in the setting of characteristic clinical and imaging findings. In the non-AIDS population, primary brain lymphoma usually shows a solid pattern of contrast enhancement on CT and MRI. Subependymal spread of lymphoma encasing the ventricles is highly characteristic when present (Dina, 1991). The solid hypercellular periphery of lymphoma lesions

Physiological MR to evaluate HIV-associated brain disorders

are typically wider than the inflammatory rim around toxoplasmosis lesions (Eisenberg et al., 1990; Dina, 1991). These lesions on average are larger and fewer than toxoplasma lesions. However, in the setting of AIDS, lymphoma is often multicentric and can grow rapidly, more than doubling in size within weeks, with associated edema and midline shifts (So et al., 1986; Cordoliani et al., 1992; Chiang et al., 1996). Therefore, on MRI, these lesions typically show ring-enhancement with gadolinium; the lesions are hypointense on T1-weighted images (WI), iso- to hyperintense on T2-WI, and are often ring-enhancing (Dina, 1991; Cordoliani et al., 1992) (Figures 27.5 and 27.6). Spontaneous hemorrhage is uncommon but may occur following therapy with steroids or radiation (Dina, 1991; Cordoliani et al., 1992). MRS of CNS lymphoma in AIDS shows mild to moderately increased Lac and lipids along with a prominent Cho peak, minimal NAA, Cr, and mI, depending on the amount of surrounding brain tissue in the voxel (Chang et al., 1995c; Chang and Ernst, 1997; Simone et al., 1998) (Figure 27.5). This reflects increased cell membrane turnover and breakdown (increased Cho and lipids), and necrotic portions of the tumor (Lac and lipids). Placement of voxel at the “cellular rim”, growing edge, of the lesion is essential; otherwise, the central necrotic region, with a large Lac/lipid peak, may resemble the spectrum of a toxoplasmosis lesion (Chinn et al., 1995). This is analogous to sampling errors during brain biopsy. In this regard, SI may be more useful in the evaluation of focal lesions. Lymphoma lesions in AIDS patients often show accelerated outward growth of the tumor which may divert the vascular supply to the growing edge of the tumor, with central areas of necrosis resulting from thrombosis and deterioration of the vessels in the oldest parts of the lesions (So et al., 1986; Cordoliani et al., 1992). Therefore, DSCI demonstrates increased vascularity and perfusion, especially at the hypercellular rim of the lesion(s) (Figure 27.6). Thallium (Tl) 201 SPECT also can demonstrate increased tracer uptake in the lymphoma lesions compared to toxoplasma lesions, although false positive and false negatives may occur (Skiest et al., 2000). Serological studies (serum toxoplasma IgG) or PCR for EBV combined with imaging techniques

might provide sufficient diagnostic accuracy to spare the patient a brain biopsy. PML PML is caused by infection of oligodendrocytes by JC virus, and produces characteristic demyelinating lesions on neuropathology and MRI (Berger et al., 1987; Sze et al., 1987; Mark and Atlas, 1989; Gillespie et al., 1991; Major et al., 1992; Wheeler et al., 1993; Whiteman et al., 1993; Newton et al., 1995). Reactivation of latent JCV is postulated to occur in patients with CD4 cell count 100/mm3 (Simpson and Tagliati, 1994). Neurological symptoms and signs vary depending upon the location(s) of the lesion(s). Common clinical presentations may include hemiparesis, cognitive impairment, gait imbalance, dysarthria, and hemi- or quadrantanopia (VonEinsiedel et al., 1993). Despite typical clinical and neuroimaging characteristics, diagnosis often requires brain biopsy. Improved techniques to aid in the diagnosis will be even more important when experimental therapy becomes available. MRS of PML lesions show a characteristic metabolite profile of moderately increased lipid  Lac, normal to increased Cho, and typically some NAA and Cr due to partial volume of surrounding brain tissue. However, mI may be highly variable (low initially and high during the repair and remission phase) (Chang et al., 1995c; Chang and Ernst, 1997; Chang et al., 1997; Simone et al., 1998; Iranzo et al., 1999). This reflects loss of normal tissue (reduced NAA and Cr), increased cell membrane turnover and demyelination (increased Cho), and glial activation at later stages of PML (increased mI) (Figure 27.7). Small amounts of Lac and lipid signals, much less than those observed in toxoplasmosis, are consistently observed during the acute phase of the infection (Iranzo et al., 1999), and likely reflect the presence of macrophages and the breakdown products of myelin. The contralateral or surrounding normal-appearing brain regions in AIDS patients with PML lesions typically show MRS patterns consistent with HIV-associated brain injury. Recent studies using MT MRI demonstrate a high degree of specificity for differentiating demyelinating PML lesions from those of non-PML associated gliosis (Ernst et al., 1999) (Figure 27.7), or HIV

469

470

Linda Chang and Thomas Ernst

Post-Gd T1

FLAIR

FLAIR

MTR map

Regional blood volume map

MTR

ADC map

ADC

Fig. 27.7 Structural and physiological MR images of a 43-year-old male AIDS patient with PML (top row), and a 38-year-old AIDS patient with non-specific (and non-PML) WM lesions (bottom row). Rainbow scale showing low values at the blue–green range (bottom) and higher values in orange and red (top). Note that both the PML lesions (top row) and the non-specific WM lesions (bottom row) show similar and large confluent hyperintense lesions on the FLAIR images. However, MTR map of the PML lesions shows significantly lower (MTR) (green and blue), especially in the center of the lesion (dark blue), suggesting demyelination within the lesions. The non-specific WM lesions show only minimally decreased MTR. ADC maps show marked increases in diffusion in the PML lesions and the surrounding edema, and only moderately increased diffusion in the non-PML lesions. Normal or minimal rCBV abnormalities are observed on the DSCI perfusion maps.

encephalitis (Dousset et al., 1997), despite the similar appearance of hyperintense lesions on T2WI. Compared to other WM lesions, PML lesions typically show significantly lower magnetization transfer ratio (MTR) on MTR maps (which illustrate the ratio between T1 studies performed with and without magnetisation transfer pulses. The lower MTR is probably due to loss of myelin in the lesions since the lower MTR reflects increased free water and decreased water bound to macromolecules (e.g. myelin). Typical duration of survival in patients diagnosed with PML was 9 months prior to HAART. The introduction of HAART has led to prolonged survival (46 weeks) in some patients (Clifford et al., 1999); however, some patients develop PML despite the suppression of HIV replication by HAART

(Tantisiriwat et al., 1999), and few patients actually deteriorated during reconstitution of their immune system (Cinque et al., 2001; Safdar et al., 2002). Physiological imaging, including MRS and MT can provide useful surrogate markers to detect early signs of demyelination (with elevated Cho compounds on MRS and regional decreases in MTR) or to monitor the effects of treatment. Since mI reflects glial activity, increased mI in PML lesions may indicate a stronger immune response and ongoing repair, hence more favorable outcome or clinical remission (Chang et al., 1997). Cryptococcoma Cryptococcus neoforman had caused one of the commonest opportunistic infection in AIDS

Physiological MR to evaluate HIV-associated brain disorders

patients during the first two decades of the epidemic in the US. Cryptococcosis is a diverse disease histopathologically. It can produce subacute chronic meningitis, causing progressive dilatation of the subarachnoid spaces and the formation of gelatinous pseudocysts (encapsulated masses of organisms with little inflammatory reaction, soap bubble lesions), or a fungal abscess (cryptococcoma). Headaches and fever are common. Focal neurological signs and seizures may occur depending on the location(s) of the lesion(s). CT and MRI of patients with documented intracranial cryptococcal infection found a variable pattern of imaging characteristics, but MRI is superior to CT for the assessment of these lesions (Tien et al., 1991). Four patterns have been reported: (1) parenchymal cryptococcoma; (2) numerous clustered tiny foci that are hyperintense on T2-weighted images and non-enhancing on post-contrast T1WI, located relatively symmetrically in the basal ganglia and the midbrain, representing dilated VirchowRobin space; (3) multiple miliary enhancing parenchymal and leptomeningeal nodules; and (4) a mixed pattern of dilated Virchow-Robin spaces with mixed lesions such as cryptococcoma and miliary nodules. Cryptococcoma also have been reported to occur in the dentate nuclei (Ruiz et al., 1997), in the pituitary (Yu et al., 1995), and in the spinal cord (Lai et al., 2001). MRS of cryptococcoma, gelatinous pseudocyst type (round cystic lesions with no edema and minimal mass effect) typically show decreases in the major cerebral metabolites (NAA, Cr, Cho, and mI) and moderate amounts of lipids compared to contralateral normal appearing brain tissue (NABT). Lac is rarely observed in cryptococcoma (ConfortGouny et al., 1993; Yamagata et al., 1994; Chang et al., 1995c). Compared to toxoplasmosis or lymphoma, cryptococcoma demonstrate lesser amounts of lipids and Lac. A proton MRS study of cryptococcal neoforman culture and cerebral cryptococcomas from rat brains showed primarily resonances from the cytosolic disaccharide alpha, alpha-trehalose, which appeared to be specific for cryptococcus infection (Himmelreich et al., 2001). Diffusion and MT findings of cryptococcama have not been reported.

Tuberculous abscess, tuberculomas, and miliary TB HIV/AIDS has dramatically increased the incidence of tuberculosis (TB), especially in countries where antiretroviral medications are not yet available. In sub-Saharan Africa, up to 60% of TB patients are coinfected with HIV and each year 200,000 TB deaths are attributable to HIV co-infection (Williams and Dye, 2003). Tuberculous infection of the brain can present in a variety of forms, and may occur throughout the CNS. These lesions show low attenuation on CT scans, have thin enhancing rims and considerable surrounding edema, and can be identical in appearance to pyogenic abscesses. Highly variable patterns of MRI characteristics (Boukobza et al., 1999) include miliary lesions (smaller than 2 mm in diameter), large hemispheric tuberculomas, infarction, abscesses, hydrocephalus, and pachymeningitis. Tuberculomas are granulomas that may be non-caseating, caseating with a solid center, or caseating with a liquid center. The noncaseating granuloma usually is hypointense relative to brain on T1WI and hyperintense on T2WI, with homogenous post-contrast enhancement (Gupta et al., 1993). The caseating granulomas with solid center are hypointense or isointense on T1WI and hypointense on T2WI. The granulomas with central liquefaction are hypointense at the center on T1WI but hyperintense on T2WI; the periphery of the lesions is hypointense on T2WI. Recent case reports of diffusion-weighted MRI showed bright signal intensity in the core of all lesions in two patients, which would differentiate these lesions from malignant gliomas and metastatic brain tumors (Kaminogo et al., 2002). MT imaging also shows lower MTR in tuberculous abscess wall compared to that of the pyogenic abscess, and may be used for differential diagnosis from caseating tuberculomas (Gupta, 2002). MTI, especially with gadolinium enhancement, can further enhance the hyperintense rim on post-contrast T1WI. MTI also significantly improves image contrast for miliary TB on MT T1WI; the lesions are further enhanced in the post-contrast T1WI. MRS in tuberculous and other intracranial infection is also discussed in Chapter 23. The largest series of MRS studies of tuberculoma in AIDS

471

472

Linda Chang and Thomas Ernst

patients are from India. Tuberculoma in patients with or without AIDS showed a similar characteristic MRS pattern (Gupta et al., 1993, 1995). The lipid peaks consist of a large resonance at 1.3 ppm and a small resonance at 0.9 ppm, which can be assigned to methylene-(CH2)n and terminal methyl group (CH3) of free fatty acids, respectively. These signals have been attributed to the high lipid content of caseous materials within tuberculomas. Unlike the toxoplasmosis abscess, Lac was typically not observed at the long TE spectra of the tuberculomas. In addition, amino acids that are visible in pyogenic brain abscesses are not visible in tuberculous abscesses. Recent reports also demonstrated the presence of Cho signal in a small proportion of tuberculoma lesions (Gupta et al., 2002), as well as other fungal lesions (Venkatesh et al., 2001). It is unclear whether the presence of Cho signals in these lesions might represent underlying metabolic disorder, such as diabetes or HIV encephalitis, since Cho might be elevated in many metabolic conditions. Nevertheless, when the Cho signal is present in an infectious abscess, the differential diagnosis of neoplasm will need to be considered. Therefore, the large lipid signal is diagnostic of an infectious lesion only when Cho is absent, and may be more characteristic for chronic infection, such as cryptococcoma or tuberculoma, when Lac is absent as well.

Focal brain lesions of combined etiologies Combined or mixed lesions of lymphoma with toxoplasmosis, or with PML, CMV, or candida have been reported in AIDS (Navia et al., 1986; Rodesch et al., 1989; Cornford et al., 1992; Gray and Sharer, 1993). Correlation of MRI, MRS, and neuropathology of such combined toxoplasmosis and lymphomas have been reported (Chang et al., 1995a). On MRI, the combined lesions show a hypointense central region on T2WI with a relatively wide margin of hypointense rim (not the typical iso- or hyperintense signals in lymphoma), which enhances with contrast. On short TE 1H MRS, the mixed lesions demonstrate characteristics of both the spectra seen in toxoplasmosis and in the lymphoma. Elevated Lac with lipids (typically observed in toxoplasma lesions)

combined with elevated Cho (more common in lymphoma) are observed. The MRS pattern of combined lesions suggests that MRS may be highly sensitive in reflecting the chemical changes associated with the underlying pathological processes. If the biochemical profile appears intermediate between two lesion types, co-existing lesions should be considered (e.g. toxoplasmosis  lymphoma) (Chang et al., 1995a; Issakhanian et al., 2001).

Multi-modality considerations Multi-modality imaging will become more prevalent clinically, since MRS, structural and physiological MRI, SPECT, PET, and fMRI, could yield data regarding different aspects of the pathophysiology or chemistry. For example, diagnostic confidence would increase if a suspected toxoplasmosis lesion shows marked elevation of lipids and Lac on MRS and decreased rCBV throughout the lesion on perfusion MRI. To assess possible contributing factors to abnormal brain function in HIV patients, one study found total Cr on MRS correlated with brain activation on BOLD-fMRI, which suggests an up-regulated oxidative metabolic state may contribute to the increased prefrontal activation (Ernst et al., 2003). Another study used co-registered quantitative 1H MRS with quantitative 133Xenoncalibrated 99mTcHMPAO SPECT to detect brain injury patients with mild dementia, and found that MRS is more sensitive than SPECT for detecting early brain injury (Ernst et al., 2000). Future studies using more than one neuroimaging modality in the same subjects are needed to determine the optimal method, and whether they may provide complementary information to further elucidate the pathogenesis of HIV brain injury.

Summary, conclusions and future directions Physiological imaging techniques, including MRS, can assess the severity of HIV-associated brain injury in patients with HIV dementia, and show particular promise in early differential diagnoses of focal brain lesions.

Physiological MR to evaluate HIV-associated brain disorders

In HIV-brain injury, the glial marker mI is elevated during the early disease stage (neurologically asymptomatic or mild dementia), whereas the neuronal marker NAA is decreased at later stages of HIV dementia, along with further increases in Cho and mI. These progressive metabolite changes with early glial activation and subsequent neuronal damage or loss are consistent with findings of neuropathology studies. Several small MRS studies also evaluated the effects of antiretroviral treatments in HIV patients, and found that metabolite abnormalities may improve after treatment, and thus may serve as objective surrogate markers. Future clinical trials may benefit from the additional use of MRS and physiological MRI to the current approach of neuropsychological assessments. MRS and the combination of physiological imaging will be important for the differential diagnoses of focal brain lesions. Perfusion MRI typically shows significantly elevated perfusion in neoplasms (e.g. lymphoma) but markedly decreased perfusion in infectious lesions. MT imaging (MTI) is particularly useful for confirming and assessing demyelination (e.g. PML lesions). Continued technical advances, such as betteroptimized short TE localized MRS and SI sequences, or two-dimensional MRS techniques, will allow better assessment of regional changes and additional metabolites of interest (e.g. glutamate (Glu), -aminobutyric acid (GABA), etc.). BOLD-fMRI appears to be exquisitely sensitive for evaluating subtle changes in brain physiology in HIV patients (Chang et al., 2001a), even in those who are neurologically asymptomatic (Ernst et al., 2002). Future research will compare the sensitivity of these techniques for detecting early brain injury so that neuroprotective treatments might be implemented at the earliest stages. With further methodological improvements, these techniques will ultimately become more “user-friendly” and can be applied in routine clinical settings.

ACKNOWLEDGEMENTS

L.C. and T.E. are supported by the NIH (K24- DA016170; K02-DA16991; R01-NS38834; R01-MH61427) and the Department of Energy (OBER).

REFERENCES Arendt G, Hefter H, Figge C. 1991. Two cases of cerebral toxoplasmosis in AIDS patients mimicking HIV-related dementia. J Neurol 238: 439–442. Barker PB, Lee RR, McArthur JC. 1995. AIDS dementia complex: evaluation with proton MR spectroscopic imaging. Radiology 195: 58–64. Bashir R, Luka J, Cheloha K, et al. 1993. Expression of Epstein–Barr virus proteins in primary CNS lymphoma in AIDS patients. Neurology 43: 2358–2362. Berger JR, Kashovitz B, Donovan-Post JD, et al. 1987. Progressive multifocal leukoencephalopathy associated with human immunodeficiency virus infection: a review of the literature and report of sixteen cases. Ann Int Med 107: 78–87. Boukobza M, Tamer I, Guichard J, et al. 1999. Tuberculosis of the central nervous system. MRI features and clinical course in 12 cases. J Neuroradiol 26(3): 172–181. Brand A, Richter-Landsberg C, Leibfritz, D. 1993. Multinuclear NMR studies on the energy metabolism of glial and neuronal cells. Dev Neurosci 15: 289–298. Burg JL, Grover CM, Pouletty P, et al. 1989. Direct and sensitive detection of a pathogenic protozoan, Toxoplasma gondii, by polymerase chain reaction. J Clin Microbiol 27: 1787–1792. Chang L, Ernst, T. 1997. Proton magnetic resonance spectroscopy and diffusion-weighted MRI in focal AIDS brain lesions. Neuroimaging in AIDS II, a Special Issue of Neuroimaging Clinics of North America, Vol. 7 (Ed, Post J), W.B. Saunders Company, pp. 409–425. Chang L, Cornford ME, Chiang, FL, et al. 1995a. Radiologic–pathologic correlation: cerebral toxoplasmosis and lymphoma in AIDS. Am J Neuroradiol 16: 1653–1663. Chang L, McBride D, Miller B, et al. 1995b. Localized in vivo 1H magnetic resonance spectroscopy and in vitro analyses of heterogeneous brain tumors. J Neuroimaging 5(3): 157–163. Chang L, Miller BL, McBride D, et al. 1995c. Brain lesions in patients with AIDS: H-1 MR spectroscopy. Radiology 197: 527–531. Chang L, Ernst T, Tornatore C, et al. 1997. Metabolite abnormalities in progressive multifocal leukoencephalopathy by proton magnetic resonance spectroscopy. Neurology 48: 836–845. Chang L, Ernst T, Leonido-Yee M, et al. 1999a. Cerebral metabolite abnormalities correlate with clinical severity of HIV-cognitive motor complex. Neurology 52(1): 100–108. Chang L, Ernst T, Leonido-Yee M, et al. 1999b. Highly active antiretroviral therapy reverses brain metabolite abnormalities in mild HIV dementia. Neurology 53: 782–789. Chang L, Ernst T, Leonido-Yee M, et al. 2000a. Perfusion MRI Detects rCBF abnormalities in early stages of HIV-cognitive motor complex. Neurology 54: 389–396.

473

474

Linda Chang and Thomas Ernst

Chang L, Ernst T, Witt M, et al. 2000b. Cerebral metabolite abnormalities in antiretroviral-Naïve HIV patients before and after HAART. Neurology 54: S47.002. Chang L, Speck O, Miller E, et al. 2001a. Neural correlates of attention and working memory deficits in HIV patients. Neurology 57: 1001–1007. Chang L, Witt M, Miller E, et al. 2001b. Cerebral metabolite changes during the first nine months of HAART. Neurology 56: S63.001. Chang L, Ernst T, Witt M, et al. 2002. Relationships among cerebral metabolites, cognitive function and viral loads in antiretroviral-Naïve HIV patients. NeuroImage 17(3): 1638–1648. Chiang F, Miller B, Chang L, et al. 1996. Fulminant cerebral lymphoma in AIDS. Am J Neuroradiol 17: 157–160. Chiappelli F, Frost P, Manfrini E, et al. 1994. Cocaine blunts human CD4+ cell activation. Immunopharmacology 28: 233–240. Chinn RJS, Wilkinson ID, Hall-Craggs MA, et al. 1995. Toxoplasmosis and primary central nervous system lymphoma in HIV Infection: diagnosis with MR Spectroscopy. Radiology 197: 649–654. Chong WK, Paley M, Wilkinson ID, et al. 1994. Localized cerebral proton MR spectroscopy in HIV infection and AIDS. Am J Neuroradiol 15: 21–25. Chong WK, Sweeney B, Wilkinson ID, et al. 1993. Proton spectroscopy of the brain in HIV infection: correlation with clinical, immunologic and MR imaging findings. Radiology 188: 119–124. Cinque P, Pierotti C, Vigano M, et al. 2001. The good and evil of HAART in HIV-related progressive multifocal leukoencephalopathy. J Neurovirol 7(4): 358–363. Ciricillo SF, Rosenblum ML 1990. Use of CT and MR imaging to distinguish intracranial lesions and to define the need for biopsy in AIDS patients. J Neurosurg 73: 720–724. Clifford D, Yiannoutsos C, Glicksman M, et al. 1999. HAART improves prognosis in HIV-associated progressive multifocal leukoencephalopathy. Neurology 523: 623–625. Cloak C, Chang L, Miller E, et al. 2003. Increased Subcortical Diffusion is Associated with Poorer Cognitive Scores in HIV. ISMRM, Eleventh Annual Meeting, Toronto, Canada. Confort-Gouny S, Vion-Dury J, Nicoli F, et al. 1993. A multiparametric data analysis showing the potential of localized proton MR spectroscopy of the brain in the metabolic characterization of neurological diseases. J Neurol Sci 118: 123–133. Cordoliani Y, Derosier C, Pharaboz C, et al. 1992. Primary cerebral lymphoma in patients with AIDS: MR findings in 17 cases. Am J Roentgenol 159: 841–847. Cornford ME, Holden JK, Boyd MC, et al. 1992. Neuropathology of the acquired immune deficiency syndrome (AIDS): report of 39 autopsies from Vancouver, British Columbia. Can J Neurol Sci 19: 442–452.

Cortey A, Jarvik JG, Lenkinski RE, et al. 1994. Proton MR spectroscopy of brain abnormalities in neonates born to HIVpositive mothers. Am J Neuroradiol 15: 1853–1859. Cristina N, Pelloux H, Goulhot C, et al. 1993. Detection of toxoplasma gondii in AIDS patients by the polymerase chain reaction. Infection 21(3): 150–153. Derouin F, Thulliez P, Garin YJF. 1991. Toxoplasma serology in HIV-infected patients: value and limitations. Pathologie Biologie 39: 255–259. Dina TS. 1991. Primary central nervous system lymphoma versus toxoplasmosis in AIDS. Radiology 179: 823–828. Dousset V, Armand JP, Lacoste D, et al. 1997. Magnetization transfer study of HIV encephalitis and progressive multifocal leukoencephalopathy. Am J Neuroradiol 18(5): 895–901. Eisenberg AD, Mani JR, Norman D. 1990. Differentiation of toxoplasmosis and lymphoma in HIV-positive patients, utilizing gadolinium-enhanced MRI. Radiology 177: 231. English C, Kaufman M, Worth J, et al. 1997. Elevated frontal lobe cytosolic choline levels in minimal or mild AIDS dementia complex patients: a proton magnetic resonance spectroscopy study. Biol Psychiat 41(41): 500–502. Ernst T, Chang L. 1996. Elimination of artifacts in short echo time 1H MR spectroscopy of the frontal lobe. Magn Reson Med 36: 462–468. Ernst T, Chang L, Arnold S. 2003. Increased glial markers predict increased working memory network activation in HIV patients. Neuroimage 19(4): 1686–1693. Ernst T, Chang L, Jovicich J, et al. 2002. Abnormal brain activation on functional MRI in cognitively asymptomatic HIV patients. Neurology 59(9): 1343–1349. Ernst T, Chang L, Witt MD, et al. 1999. Magnetization transfer imaging of progressive multifocal leukoencephalopathy and HIV-associated white matter lesions in AIDS. Radiology. Ernst T, Itti E, Itti L, et al. 2000. Changes in cerebral metabolism are detected prior to perfusion changes in early HIVCMC: a coregistered 1H MRS and SPECT study. J Magn Reson Imaging 12(6): 859–865. Filippi C, Ulug A, Ryan E, et al. 2001. Diffusion tensor imaging of patients with HIV and normal-appearing white matter on MR images of the brain. Am J Neuroradiol 22(2): 277–283. Fiscus S, Adimora A, Schoenbach V, et al. 1996. Perinatal HIV infection and the effect of zidovudine therapy on transmission in rural and urban counties. J Am Med Assoc 275(19): 1483–1488. Gillespie SM, Chang Y, Lemp G, et al. 1991. Progressive multifocal leukoencephalopathy in persons infected with human immunodeficiency virus, San Francisco, 1981–1989. Ann Neurol 30(4): 597–604. Graf J, Guggino W, Turnheim K. 1993. Volume regulation in transporting epithelia. Interactions in Cell Volume and Cell

Physiological MR to evaluate HIV-associated brain disorders

Function, (Eds, Lang F, Häussinger D), Springer-Verlag, Heidelberg, pp. 67–1117. Gray F, Sharer LR. 1993. Combined pathologies. Atlas of the Neuropathology of HIV infection, (Ed, Gray F), Oxford Science Publishers, Oxford, pp. 162–165. Gupta R. 2002. Magnetization transfer MR imaging in central nervous system infections. Indian J Radiol Imaging 12(1): 51–58. Gupta R, Husain M, Vatsal D, et al. 2002. Comparative evaluation of magnetization transfer MR imaging and in-vivo proton MR spectroscopy in brain tuberculomas. Magn Reson Imaging 20(5): 375–381. Gupta RK, Pandey R, Khan EM, et al. 1993. Intracranial tuberculomas: MRI signal intensity correlation with histopathology and localised proton spectroscopy. Magn Reson Imaging 11: 443–449. Gupta RK, Poptani H, Kohli A, et al. 1995. In vivo localized proton magnetic resonance spectroscopy of intracranial tuberculomas. Indian J Med Res 101: 19–24. Harris GJ, Pearlson GD, McArthur JC, et al. 1994. Altered cortical blood flow in HIV-seropositive individuals with and without dementia: a single photon emission computed tomography study. AIDS 8: 495–499. Himmelreich U, Dzendrowskyj T, Allen C, et al. 2001. Cryptococcomas distinguished from gliomas with MR spectroscopy: an experimental rat and cell culture study. Radiology 220: 1. Holman BL, Garada B, Johnson KA, et al. 1992. A comparison of brain perfusion SPECT in cocaine abuse and AIDS dementia complex. J Nucl Med 33: 1312–1315. Iglesias-Rozas JR, Bantz B, Adler T. 1991. Cerebral lymphoma in AIDS: clinical, radiological, neuropathological and immunopathological study. Clin Neuropathol 10: 65–72. Iranzo A, Moreno A, Pujol J, et al. 1999. Proton magnetic resonance spectroscopy pattern of progressive multifocal leukoencephalopathy in AIDS. J Neurol Neurosug Psychiat 66(4): 520–523. Issakhanian M, Chang L, Cornford M, et al. 2001. HIV-2 Infection with Cerebral toxoplasmosis and lymphomatoid granulomatosis. J Neuroimaging 11: 212–216. Jarvik JG, Lenkinski RE, Grossman RI, et al. 1993. Proton MR spectroscopy of HIV-infected patients: characterization of abnormalities with imaging and clinical correlation. Radiology 186: 739–744. Kaissar G, Edwards M, Smith R. 1991. Neuroimaging of AIDS. Indiana Med 84(7): 470–474. Kaminogo M, Ishimaru H, Morikawa M, et al. 2002. Proton MR spectroscopy and diffusion-weighted MR imaging for the diagnosis of intracranial tuberculomas. Report of two cases. Neurol Res 24(6): 537–543.

Lai P, Wang J, Chen W, et al. 2001. Intramedullary spinal cryptococcoma: a case report. J Formosa Med Assoc 100(11): 776–778. Laubenberger J, Haussinger D, Bayer S, et al. 1996. HIV-related metabolic abnormalities in the brain: depiction with proton MR spectroscopy with short echo times. Radiology 199: 805–810. Levine AM. 1992. Acquired immunodeficiency syndromerelated lymphoma. Blood 80: 8–20. Levy R, Bredesen DE. 1988. Central nervous system dysfunction in AIDS. In AIDS and the Nervous System, (Eds, Rosenblum ML, Levy RM, Bredesen DE), Raven Press, New York, pp. 29–63. Levy RM, Mills CM, Posin JP, et al. 1990. The efficacy and clinical impact of brain imaging in neurologically symptomatic AIDS patients: a prospective CT/MRI study. J Acq Immun Def Synd 3: 461–471. Lopez-Villegas D, Lenkinski RE, Frank I. 1997. Biochemical changes in the frontal lobe of HIV-infected individuals detected by magnetic resonance spectroscopy. Proc Nat Acad Sci USA 94(18): 9854–9859. Lu D, Pavlakis SG, Frank Y, et al. 1996. Proton MR spectroscopy of the basal ganglia in healthy children and children with AIDS. Radiology 199(2): 423–428. Luft BJ, Brooks RG, Conley FK, et al. 1984. Toxoplasmic encephalitis in patients with acquired immunodeficiency syndrome. J Am Med Assoc 252: 913–917. Major EO, Amemiya K, Tornatore CS, et al. 1992. Pathogenesis and molecular biology of progressive multifocal leukoencephalopathy, the JC virus-induced demyelinating disease of the human brain. Clin Microbiol Rev 5(1): 49–173. Marcus C, Taylor-Robinson S, Sargentoni J, et al. 1998. 1H MR spectroscopy of the brain in HIV-1 seropositive subjects evidence for diffuse metabolic abnormalities. Metab Brain Disord 13(2): 123–36. Mark AS, Atlas SW. 1989. Progressive multifocal leukoencephalopathy in patients with AIDS: appearance on MR images. Radiology 173: 517–520. Masdeu JC, Yudd A, Van Heertun RL, et al. 1991. Single photon emission computed tomography in human immunodeficiency virus encephalopathy: a preliminary report. J Nucl Med 32: 1471–1475. Menon DK, Ainsworth JG, Cox IJ. 1992. Proton MR spectroscopy of the brain in AIDS dementia complex. J Comput Assist Tomo 16: 538–542. Meyerhoff D, Bloomer C, Cardenas V, et al. 1999. Elevated subcortical choline metabolites in cognitively and clinically asymptomatic HIV  patients. Neurology 52(5): 995–1003. Meyerhoff D, MacKay S, Poole N, et al. 1994. N-acetyl aspartate reductions measured by 1H MRSI in cognitively impaired HIVseropositive individuals. Magn Reson Imaging 12: 653–659.

475

476

Linda Chang and Thomas Ernst

Meyerhoff D, Weiner M, Fein, G. 1996. Deep gray matter structures in HIV infection: a proton MR spectroscopic study. Am J Neuroradiol 17: 973–978. Meyerhoff DJ, MacKay S, Bachman L, et al. 1993. Reduced brain N-acetyl aspartate suggests neuronal loss in cognitively impaired immunodeficiency virus-seropositive individuals: In vivo 1H magnetic resonance spectroscopic imaging. Neurology 43: 509–515. Moller H, Vermathen P, Lentschig M, et al. 1999. Metabolic characterization of AIDS dementia complex by spectroscopic imaging. J Magn Reson Imaging 9(1): 10–8. Morgello S, Petito CK, Mouradian JA. 1990. Central nervous system lymphoma in the acquired immunodeficiency syndrome. Clin Neuropathol 9: 205–215. Navia BA, Petito CK, Gold, JWM, et al. 1986. Cerebral toxoplasmosis complicating the acquired immune deficiency syndrome: clinical and neuropathological findings in 27 patients. Ann Neurol 19: 224–238. Newton HB, Makley M, Slivka AP, et al. 1995. Progressive multifocal leukoencephalopathy presenting as multiple enhancing lesions on MRI: case report and literature review. J Neuroimaging 5: 125–128. Paley M, Cozzone P, Alonso J, et al. 1996. A multicenter proton magnetic spectroscopy study of neurological complications of AIDS. AIDS Res Hum Retrov 12(3): 213–222. Paley M, Wilkinson ID, Hall-Craggs MA, et al. 1995. Short echo time proton spectroscopy of the brain in HIV infection/AIDS. Magn Reson Imaging 13(6): 871–875. Pavlakis SG, Lu D, Frank Y, et al. 1998. Brain lactate and N-acetyl aspartate in pediatric AIDS encephalopathy. Am J Neuroradiol 19(2): 383–385. Pohl P, Vogl G, Fill H, et al. 1988. Single photon emission computed tomography in AIDS dementia complex. J Nucl Med 29: 1382–1386. Pomara N, Crandall D, Choi S, et al. 2001. White matter abnormalities in HIV-1 infection: a diffusion tensor imaging study. Psychiat Res 106(1): 15–24. Power C, Kong PA, Crawford TO, et al. 1993. Cerebral white matter changes in acquired immunodeficiency syndrome dementia: alterations of the blood–brain barrier. Ann Neurol 34: 339–350. Raichle M. 1987. Circulatory and metabolic correlates of brain function in normal humans. Handbook of Physiology – The Nervous System, (Eds, Mountcastle V, Plum F, Geiger S), American Physiological Society, pp. 643–674. Ramsey RG, Geremia GK. 1988. CNS complications of AIDS: CT and MR findings. Am J Radiol 151: 449–454. Rodesch G, Parizel PM, Farber CM. 1989. Nervous system manifestations and neuroradiologic findings in acquired immunodeficiency syndromes (AIDS). Neuroradiology 31: 33–39.

Rosci MA, Pignorini F, Bernabei A, et al. 1996. Methods for detecting early signs of AIDS dementia complex in asymptomatic subjects: a quantitative tomography study of 18 cases. AIDS 6(11): 1309–1316. Rosenberg NL, Hochberg FH, Miller G, et al. 1986. Primary central nervous system lymphoma related to Epstein–Barr virus in a patient with acquired immune deficiency syndrome. Ann Neurol 20: 98–102. Rottenberg DA, Moeller JR, Strother SC, et al. 1987. The metabolic pathology of the AIDS dementia complex. Ann Neurol 22: 700–706. Rottenberg DA, Sidtis JJ, Strother SC, et al. 1996. Abnormal cerebral glucose metabolism in HIV-1 seropositive subjects with and without dementia. J Nucl Med 37(7): 1133–1141. Ruiz A, Post M, Bundschu C. 1997. Dentate nuclei involvement in AIDS patients with CNS cryptococcosis: imaging findings with pathologic correlation. J Comput Assist Tomo 21(2): 175–182. Safdar A, Rubocki R, Horvath J, et al. 2002. Fatal immune restoration disease in human immunodeficiency virus type 1-infected patients with progressive multifocal leukoencephalopathy: impact of antiretroviral therapy-associated immune reconstitution. Clin Infect Dis 35(10): 1250–1257. Salvan A, Vion-Dury J, Confort-Gouny S, et al. 1997a. Brain Proton magnetic resonance spectroscopy in HIV-related encephalopathy: identification of evolving metabolic patterns in relation to dementia and therapy. AIDS Res Human Retrov 13: 1055–1066. Salvan A, Vion-Dury J, Confort-Gouny S, et al. 1997b. Cerebral metabolic alterations in human immunodeficiency virus related encephalopathy detected by proton magnetic resonance spectroscopy. Comparison using short and long echo times. Invest Radiol 32(8): 485–495. Salvan A-M, Lamoureux S, Michel G, et al. 1998. Localized proton magnetic resonance spectroscopy of the brain in children infected with human immunodeficiency virus with and without encephalopathy. Ped Res 44(5): 755–762. Schaefer P, Gonzalez R, Hunter G, et al. 2001. Diagnostic value of apparent diffusion coefficient hyperintensity in selected patients with acute neurologic deficits. J Neuroimaging 11: 4. Schwartz RB, Komaroff AL, Garada BM, et al. 1994. SPECT imaging of the brain: comparison of findings in patients with chronic fatigue syndrome, AIDS dementia complex, and major unipolar depression. Am J Roentgenol 162: 943–951. Siddiqui NS, Brown LS, Makuch RW. 1993. Shot-term declines in CD4 levels associated with cocaine use in HIV-1 seropositive, minority injecting drug users. J Nat Med Assoc 85: 293–296. Simone I, Federico F, Tortorella C, et al. 1998. Localised 1H-MR spectroscopy for metabolic characterisation of diffuse and

Physiological MR to evaluate HIV-associated brain disorders

focal brain lesions in patients infected with HIV. J Neurol Neurosurg Psychiat 64: 516–523. Simpson DM, Tagliati M. 1994. Neurologic manifestations of HIV infection. Ann Intern Med 121: 769–785. Skiest D, Erdman W, Chang W, et al. 2000. SPECT thallium-201 combined with Toxoplasma serology for the presumptive diagnosis of focal central nervous system mass lesions in patients with AIDS. J Infect 40(3): 274–281. So YT, Beckstead JH, Davis RL. 1986. Primary central nervous system lymphoma in acquired immune deficiency syndrome: clinical and pathologic study. Ann Neurol 20: 566–572. Stankoff B, Tourbah A, Suarez S, et al. 2001. Clinical and spectroscopic improvement in HIV-associated cognitive impairment. Neurology 56(1): 112–115. Suwanwelaa N, Phanuphak P, Phanthumchinda K, et al. 2000. Magnetic resonance spectroscopy of the brain in neurologically asymptomatic HIV-infected patients. Magn Reson Imaging 18(7): 859–865. Sze G, Brant-Zawadzki MN, Normal D, et al. 1987. The neuroradiology of AIDS. Semin Roentgenol 22: 42–53. Tantisiriwat W, Tebas P, Clifford D, et al. 1999. Progressive multifocal leukoencephalopathy in patients with AIDS receiving highly active antiretroviral therapy. Clin Infect Dis 28(5): 1152–1154. Tien R, Chu P, Hesselink J, et al. 1991. Intracranial cryptococcosis in immunocompromised patients: CT and MR findings in 29 cases. Am J Neuroradiol 12(2): 283–289. Tracey I, Carr CA, Guimaraes AR, et al. 1996. Brain cholinecontaining compounds are elevated in HIV-positive patients before the onset of AIDS dementia complex: a proton magnetic resonance spectroscopic study. Neurology 46: 783–788. Tracey I, Hamberg LM, Guimaraes AR, et al. 1998. Increased cerebral blood volume in HIV-positive patients detected by functional MRI. Neurology 50(6): 1821–1826. Trenkwalder P, Trenkwalder C, Feiden W, et al. 1992. Toxoplasmosis with early intracerebral hemorrhage in a patient with the acquired immunodeficiency syndrome. Neurology 42: 436–438.

Venkatesh S, Gupta R, Pal L, et al. 2001. Spectroscopic increase in choline signal is a nonspecific marker for differentiation of infective/inflammatory from neoplastic lesions of the brain. J Magn Reson Imaging 14(1): 8–15. von Giesen H, Wittsack H, Wenserski F, et al. 2001. Basal ganglia metabolite abnormalities in minor motor disorder associated with human immunodeficiency virus type 1. Archiv Neurol 58(8): 1281–1286. VonEinsiedel RW, Fife TD, Aksamit AJ, et al. 1993. Progressive multifocal leukoencephalopathy in AIDS: a clinicopathologic study and review of the literature. J Neurol 240: 391–406. Weisberg LA, Greenberg J, Stazio A. 1988. Computed tomographic findings in cerebral toxoplasmosis in adults. Comput Med Imaging Graph 12(6): 379–383. Weiss LM, Udem S, Salgo M, et al. 1991. Sensitive and specific detection of toxoplasma DNA in an experimental murine model: use of toxoplasma gondii-specific cDNA and the polymerase chain reaction. J Infect Dis 163: 180–186. Wheeler AL, Truwit CL, Kleinschmidt-DeMasters BK, et al. 1993. Progressive multifocal leukoencephalopathy: contrast enhancement on CT scans and MR images. Am J Roentol 161: 1049–1051. Whiteman MLH, Post MJD, Berger JR, et al. 1993. Progressive multifocal leukoencephalopathy in 47 HIV-seropositive patients: Neuroimaging with clinical and pathologic correlation. Radiology 187: 233–240. Wilkinson ID, Lunn S, Miszkiel KA, et al. 1997. Proton MRS and quantitative MRI assessment of the short term neurological response to antiretroviral therapy in AIDS. J Neurol, Neurosurg Psychiat 63(4): 477–482. Williams B, Dye C. 2003. Antiretroviral drugs for tuberculosis control in the era of HIV/AIDS. Science 301(5639): 1535–1537. Yamagata NT, Miller BL, McBride D, et al. 1994. In vivo proton spectroscopy of intracranial infections and neoplasms. J Neuroimaging 4: 23–28. Yu Y, Jiang X, Gao Y. 1995. MRI of a pituitary cryptococcoma simulating an adenoma. Neuroradiology 37(6): 449–450.

477

478

Linda Chang and Thomas Ernst

Case Study 27.1 Progressive multifocal leukoencephalophathy (PML) Peter Barker D.Phil. and Martin Pomper M.D. Ph.D., Johns Hopkins University School of Medicine, Baltimore History 50-year-old male, HIV positive, unsteady gait

T2

Cho

and multiple brain lesions.

Technique Conventional MRI and multi-slice MRSI (TE 280 ms).

a

b

Imaging findings T2 MRI shows a hyperintense lesion in the left occipital-parietal lobe involving the corpus callosum. On MRSI, the lesion has low NAA and high Cho signal, which extends across the splenium of the corpus callosum. Only a small Lac/lipid is detected. The dark regions in the frontal lobe on MRSI are due to magnetic susceptibility artifacts.

PET

NAA

Discussion The spectral patterns in this case could be consistent with either PML or lymphoma, but were considered less likely for toxoplasmosis (Pomper, 2002). Fluoro-2-deoxyglucose (FDG)-PET showed no uptake, suggestive of PML rather than lymphoma.

(a)

(b)

Cho

NAA Cho

Key points

Cr

Cr

MRSI may be helpful in diagnosing HIVrelated brain lesions

Lac/Lip NAA

PML has high Cho and low NAA, similar to other demyelinating conditions, and also similar to brain tumours (e.g. lymphoma) ppm

3.0

2.0

1.0

ppm

3.0

2.0

1.0

Reference Pomper MG, Constantinides CD, Barker PB, Bizzi A, Dobgan AS, Yokoi F, McArthur JC, Wong DF. 2000. Quantitative MR spectroscopic imaging of brain lesions in patients with AIDS: correlation with [11C-methyl] thymidine PET and thallium-201 SPECT. Acad Radiol 9(4): 398–409.

Section 5 Seizure disorders

28

Seizure disorders: overview Thomas R. Henry Emory Epilepsy Center, Department of Neurology, Emory University School of Medicine, Atlanta, GA, USA

Summary Physiological MR modalities are highly useful for pathophysiological research in the human epilepsies and in experimental epilepsies. Physiological MR is most effectively interpreted in parallel analysis with structural MR imaging (MRI) and with other functional imaging modalities, in defining human epileptic pathophysiology. For example, new hippocampal lesions incurred during complex partial status epilepticus are most sensitively detected with diffusionweighted imaging (DWI), both in humans and in the rodent kainate model, and delayed imaging with structural MRI confirms the lesion as hippocampal sclerosis (HS). Physiological MR does not currently have a diagnostic role on general epilepsy care. Physiological MR and other functional imaging modalities serve as adjuncts to electrophysiological studies and structural MRI in planning epilepsy surgery.

Seizures and epilepsies Epilepsy is the most common of disabling neurological conditions, and seizures are among the most common of neurological symptoms (Engel, 1989). Seizures are paroxysmal, transitory events that alter consciousness or other cortical function, due to episodic neurological, psychiatric, or extracerebral (particularly cardiovascular) dysfunction. Epileptic seizures are distinguished from other such events by their abnormally synchronized electrical discharges in localized or widely distributed groups of cerebral neurons; such hypersynchronous discharges do not occur during organic or psychogenic non-epileptic

seizures, certain of which may produce behavior closely resembling that of epileptic seizures. Many individuals experience a single generalized tonicclonic seizure some time in life, which can be caused by electrolyte disturbances, hypoglycaemia or other extracerebral conditions. Epilepsy is diagnosed only when persisting cerebral dysfunction causes recurring epileptic seizures. Approximately 5% of the general population has one or more epileptic seizure during their lifetimes. At any point in time 1–2% of the population has epilepsy; cumulative lifetime incidence exceeds 3% (Engel, 1989). Seizures are refractory to control with antiepileptic drugs (AEDs) in more than 30% of all epilepsies, but the incidence of drug refractoriness varies considerably across the wide range of epileptic syndromes (Kwan, 2001). Seizure phenomenology Seizure manifestations are remarkably heterogeneous across individuals. While a single patient often experiences two or more different types of seizures at various times, each seizure type within a single individual usually is quite stereotyped in terms of negative phenomena (such as impaired speech or consciousness, or loss of postural tone) and positive phenomena (such subjective sensory hallucinations or psychic illusions, or focal clonus). Partial-onset seizures begin electrophysiologically in one cerebral region before ceasing at that site, or propagating to other sites. Simple partial seizures can cause various sensory, motor and psychic phenomena, but do not cause globally impaired awareness. Some degree of bihemispheric propagation occurs during complex 481

482

Thomas R. Henry

partial seizures, so as to cause globally altered consciousness (which may or may not be associated with simple or complex movements, termed “automatisms”). The further spread of ictal discharges over both cerebral hemispheres and the brainstem can cause a generalized tonic-clonic (convulsive, or grand mal) seizure. Generalized tonic-clonic seizures also can begin without any preceding behavioral or electroencephalographical (EEG) changes that suggest a focal cortical onset. Such generalized-onset seizures probably begin with pathological, synchronous discharges of thalamocortical neurons bilaterally, which “pace’’ the simultaneous onset of ictal discharges over the entire cortex (Steriade, 1993). Absence seizures also are generalized-onset seizures which from onset involve thalamus and cortex bilaterally, but during absences only a minority of neurons in each area actually have ictal discharges, so as to cause behavioral and cognitive arrest, without convulsive activity. Many types of seizures are followed by transitory postictal cerebral dysfunction, which lasts for minutes or hours and is more severe than or different to the individual’s interictal dysfunction. Status epilepticus may be viewed as the extreme seizure state, in which generalized tonic-clonic or complex partial seizures occur repetitively, without full clearing of the postictal state before the next seizure. Risk of death and permanent brain injury distinguishes status epilepticus from isolated seizures (Wasterlain, 1974). Seizure pathophysiology The various types of epileptic seizures have in common the features of pathologically repetitive neuronal action potentials, termed “paroxysmal depolarization shifts” (PDSs), and excessive interneuronal synchronization of postsynaptic potentials (Steriade, 1993). Interneuronal hypersynchrony and PDSs underlie both interictal epileptiform discharges (or “spikes”) and electrographical seizures, which can be recorded with EEG. The pathophysiological correlates of interneuronal hypersynchrony are variable, however. Structural changes that give rise to increased interneuronal connectivity, such as mossy fiber sprouting in HS, presumably predispose to increased interneuronal synchrony of postsynaptic

potentials, but such changes probably do not occur in primary generalized epilepsies (Chang and Lowenstein, 2003). Partial-onset seizures often are associated with inadequacy of inhibitory -amino-butyric acid (GABA) ergic neurotransmission and increased excitatory glutamatergic neurotransmission, while absence seizures are associated with disordered thalamic calcium channel function (Chang and Lowenstein, 2003). Epilepsy classification Epilepsy is diagnosed only when persisting cerebral dysfunction causes recurring epileptic seizures. The epilepsies are remarkably heterogeneous. Functional imaging research and clinical application require full classification of the patients’ seizures and epilepsies. Epilepsies are classified in two domains, as localization-related (with partial-onset seizures) vs. generalized (with generalized-onset seizures), and as primary vs. secondary in etiology. In primary epilepsies, seizures are the only clinical manifestations of the cerebral dysfunction, epidemiological evidence points to variably penetrant autosomal inheritance, and no post-conceptional cerebral insult appears necessary for initiation of epileptogenesis. Secondary (or symptomatic) epilepsies are acquired secondary to cerebral insult, although the precise nature of the insult cannot always be determined. Particular genetic polymorphisms may cluster with environmental injuries in certain syndromes of symptomatic epilepsy. Frequently an epileptogenic insult causes interictal cerebral dysfunction, including the mental retardation usually seen in symptomatic generalized epilepsies, and the deficits of delayed recall that typify the interictal hippocampal dysfunction of limbic temporal lobe epilepsy (TLE).

Brain imaging in the initial diagnosis of human epilepsies Electroclinical syndromes of epilepsy are fundamental to human epilepsy therapy and research (Engel 1989). Careful description of observed behavior and subjective phenomena during and following seizures, and analysis of hypersynchronized interictal electrocerebral activities (“spikes”) on

Seizure disorders: overview

scalp EEG, generally support accurate diagnosis of electroclinical syndrome in an individual patient. In some cases the initial diagnosis requires ictal videoEEG recording, to permit detailed analysis of behaviors and EEG during actual seizures. Ictal video-EEG can, for example, determine whether staring spells are complex partial seizures, atypical absence seizures, or simply daydreaming with uncooperativeness, and whether spells with bizarre behavior are epileptic or psychogenic in nature, among many other problems of clinical seizure diagnosis. In nearly all epileptic syndromes, structural imaging with brain MRI also is widely considered a central diagnostic tool. Functional imaging, including physiological MR, is not currently used in the initial epilepsy diagnosis. Structural MRI detects lesions that may in themselves require urgent neurosurgical therapy, in patients for whom a seizure is the initial clinical sign of the lesion. Alien-tissue lesions that are highly associated with partial epilepsies, and readily detected with MRI, include gliomas and other primary brain tumors, intracranial metastases, meningiomas and other intracranial-extracerebral neoplasia, cavernous angiomata, and arteriovenous malformations (AVM) (Kuzniecky and Jackson, 1995). More often MRI reveals lesions that cause epilepsy and give insight into likely therapeutic outcome, but the lesions are not in themselves associated with death or progressive disability. Ablative lesions, such as HS, post-traumatic and postinfarctional encephalomalacia, and acute or chronic, focal or diffuse infectious processes, are strongly associated with partial epilepsies (Cascino, 1996). Such lesions also are strongly associated with symptomatic generalized epilepsies, when bihemispheric and of early childhood onset. Malformations of cortical development, including focal cortical dysplasias and neuronal heterotopias, similarly are associated with symptomatic partial and symptomatic generalized epilepsies (Duncan, 1997). Some lesions are found incidentally in seizure evaluations, such as leukoariosis, arachnoid cysts and venous angiomata, but are not associated with epileptogenesis. Many epilepsy patients have normal MRI. High-quality brain MRI must precede functional imaging in any clinical application, and also is essential for epilepsy classification and anatomical correlation

with functional imaging abnormalities in research applications.

Brain imaging in the presurgical evaluation of human epilepsies Focal cerebral resection can fully control seizures, without clinically detectable loss of cortical function, in many medically refractory partial epilepsies (Engel, 1996). The prototypical syndrome of surgically amenable, drug-resistant epilepsy is limbic TLE. Limbic TLE is the single most common epilepsy of older children and adults, and often is associated with HS, less often with foreign-tissue or ablative lesions, and sometimes with normal MRI (Kuzniecky and Jackson, 1995). When HS is detected on MRI at initial evaluation of limbic TLE, the prognosis for achieving full seizure control with medication is only about 50% (poorer than average across the epilepsies), but for patients with proven drug-resistant seizures who have all EEG-recorded ictal onsets ipsilateral to HS, the probability of full post-resective seizure control may exceed 80% (better than average for most other surgically treated epileptic syndromes) (Wiebe, 2001). Efficacy of resection generally is greater in TLE than in partial epilepsies with extratemporal ictal onset zones, although the presence of certain lesions such as cavernous malformations appears to greatly increase the likelihood of success no matter their location (Engel, 1996). Many patients will require neurosurgical implantation of intracerebral or subdural electrodes in order to permit ictal EEG recordings with sufficient localizing value to support definitive resection. Interictal 2-[18F]fluoro-2-deoxyglucose (FDG) positron emission tomography (PET) can detect hypometabolic zones that are highly associated with the electrophysiologically defined ictal onset zones, and are routinely used as an adjunct to MRI and EEG in presurgical evaluations (Spencer and Bautista, 2000). Similarly, paired ictal-interictal [99mTc]technetiumhexamethylpropyleneamine oxime (HMPAO) single photon emission computed tomography (SPECT) scans, with subtraction and coregistration to MRI, are routinely used as an adjunct to MRI and EEG in presurgical evaluations (Spencer and Bautista, 2000).

483

484

Thomas R. Henry

During the interictal state of unilateral TLE, glucose metabolic dysfunction is seen in a widespread area over the affected temporal lobe, less severely over the contralateral temporal lobe, usually over the ipsilateral thalamus, and often over other ipsilateral sites (Henry et al., 1990). These same sites tend to be involved in the ictal hypermetabolism detectable with FDG and PET, and the ictal hyperperfusion detectable with subtraction ictal-interictal SPECT (Berkovic, 2000; Engel et al., 1982; Lee et al., 1988; Van Paesschen et al., 2003). A rapid “postictal switch” phenomenon occurs, in which structures with ictal hyperperfusion then transiently have lower blood flow postictally than they do in the persisting interictal state (Newton, 1992). Interictal hypoperfusion is frequently observed but often does not include the ictal onset zone, in contrast to the high association of interictally hypometabolic regions with ictal onset zones; it now is known that the interictal state of TLE is characterized by uncoupling of blood flow and glucose metabolism (Gaillard, 1995). The evaluation of MR perfusion techniques as an alternative for SPECT and PET imaging is just beginning, but is potentially promising because of the greater availability and lower cost of MRI. However, ictal MRI is unlikely to be practical in all but a few cases. Several types of functional imaging abnormalities may prove useful in presurgical evaluation, following further validation. Pathologically decreased densities of GABAA receptor complexes can be mapped with [11C]flumazenil and PET; the sites of decreased [11C]flumazenil binding are smaller than those mapped with FDG PET and with ictal SPECT, and current data suggests high association with the ictal onset zone (Duncan, 2000; Henry et al., 1993; Savic et al., 1995). In tuberous sclerosis (TS) there sometimes is a single active epileptogenic tuber causing refractory partial seizures, but functional imaging, MRI and EEG do not detect the epileptogenic tuber amongst a multitude of lesions; in such cases PET with [11C]alpha-methyl-L-tryptophan may show increased signal (increased serotonergic synthetic capacity) at the epileptogenic tuber and decreased radioligand activity at the other tubers (Juhasz et al., 2003), leading directly to efficacious resection. In the future, N-acetyl aspartate (NAA) mapping with proton MR Spectroscopy, (MRS) and perhaps other physiological

MR techniques, may also prove clinically useful in multimodality mapping of the ictal onset zone.

Brain imaging in pathophysiological investigations of experimental and human epilepsies Structural and functional imaging permit noninvasive, anatomical determination of lesional and physiological substrates of epileptogenesis, of transitory dysfunctions during ictal and postictal states, and of persisting interictal cerebral dysfunctions. Animal models have been developed for some human epileptic syndromes. Human imaging studies may parallel experimental findings with combined imaging and subsequent tissue-destructive (histopathological, ultrastructural, autoradiographical, biochemical and other) studies. Concordance of abnormalities on parallel experimental and human studies can advance pathophysiological knowledge more than do single-technique approaches, in these epileptic conditions that more often seem to feature interacting genetic and environmental factors over any single etiological factor. Pathophysiology of limbic TLE Limbic TLE with HS is likely to be a multifactorial condition arising after initial brain development, with multiple possible pathways to this electroclinicalhistopathological syndrome. The rodent kainate model and other animal models of TLE generally begin with an acute phase of complex partial status epilepticus, with recovery to a chronic phase during which brief partial-onset seizures occasionally interrupt prolonged interictal states. The status epilepticus phase in these models has been extensively studied with imaging techniques, which demonstrate characteristic progression of regional limbic hyperperfusion (often with relative hypoperfusion of extra-limbic sites), to early decreases (and later increases) of water diffusability on DWI, then vasogenic (and later cytotoxic) edema on T2-weighted MRI, early lactate (Lac) elevation and later NAA decrease (among other changes) on MRS, and various other imaging abnormalities (Ebisu et al., 1994; Jackson and Opdam, 2000;

Seizure disorders: overview

Meric et al., 1994; Najm et al., 1997; 1998; Nakasu et al., 1995; Wall et al., 2000; Wang 1996). Essentially identical changes also occur in humans during complex partial status epilepticus (Diehl, 2001; Flacke, 2000; Lansberg et al., 1999; Warach et al., 1994)! After cessation of status epilepticus, during a prolonged phase of intermittent partial-onset seizures, both the animal subjects and human patients have MRI-detectable HS (Perez et al., 2000; Pirttila et al., 2001; Salmenpera et al., 2000). Many patients develop limbic TLE-HS without preceding status epilepticus (Briellmann et al., 2002; O’Brien et al., 1999), and a few such patients have familial TLE (Kobayashi et al., 2003). Development of future therapy and even prevention of this highly disabling syndrome likely will rely on physiological MR and other functional imaging techniques. Imaging studies of limbic TLE and animal models of TLE illustrate the important and as yet incomplete contributions of these techniques. Space does not permit review here of numerous other important imaging-based investigations of epileptic pathophysiology relevant to other epileptic syndromes.

Brain imaging in therapeutics research of human epilepsies Functional neuroimaging has clarified some therapeutic mechanisms, and also mechanisms of toxicity of AEDs and other epilepsy therapies (Henry 2000). Regional cerebral drug distributions of intravenously administered phenytoin and valproate have been studied in human subjects, using carbon11 labeling of the drug and PET imaging (Baron et al., 1983). Regional [15O]H2O mapping of synaptic activation and de-activation during vagus nerve stimulation, a non-pharmacological therapy used in both partial and generalized epilepsies, showed that bilateral thalamic activation was associated with seizure reduction (Henry et al., 1999). Presurgical baseline and chronic post-resective comparisons in refractory TLE demonstrated that in patients whose seizures ceased after surgery, the contralateral (unresected) hippocampus sustained increases in resting FDG activity and in relative NAA peaks (Hajek et al., 1994; Hugg et al., 1996). The results of these PET and MRS studies correlate well with

psychometric evidence of improved postoperative memory function in these patients. The ketogenic diet, in which carbohydrate intake is severely restricted, can reduce seizure frequency in symptomatic generalized epilepsies of childhood. Diet-induced ketosis was associated with a small but significant increase in the phosphocreatine/ gamma-adenosine triphosphate (ATP) ratio, which was measured at baseline (regular diet) and again after the ketogenic diet was used chronically (Pan, 1999). Acetone peaks in occipital gray matter (GM) on proton MRS were higher than the normal values in most but not all of epilepsy patients who had significant improvement in seizure control on the ketogenic diet (Seymour et al., 1999). Pharmacodynamic effects of AEDs that can be studied with functional imaging include whole-brain and regional effects on synaptic activity, effects on neurotransmitter concentrations, and effects on receptor density and occupancy. Human AED pharmacodynamic imaging with FDG PET, at baseline and after chronic AED use, showed that phenobarbital causes greater generalized cerebral glucose metabolic decreases than do other agents, consistent with relative AED effects on interictal alertness and cognition (Theodore, 1988). Patients with the partial epilepsy TLE and those with juvenile myoclonic epilepsy, a primary generalized epilepsy, have lower MRSmeasured occipital lobe GABA concentration than do healthy subjects (Petroff et al., 2000). In the reported MRS studies the epilepsy patients were all taking carbamazepine, valproate or other AEDs and the healthy subjects were not using AEDs, however. Brain GABA concentration is increased by acute or chronic administration of topiramate, gabapentin and vigabatrin, but not by administration of tiagabine, on comparison of occipital lobe spectroscopic measurements made during periods with and without AED exposure, in healthy subjects who were not using other medications and epileptic subjects who also were chronically receiving other “older’’ AEDs (Petroff et al., 2000). Chronic valproate therapy was associated with generalized cerebral decreases in [11C]flumazenil binding, but interictal [11C]flumazenil activity was normal in patients with absence seizures who were receiving other AEDs (Prevett et al., 1995). This suggests that valproate may act to increase cerebral

485

486

Thomas R. Henry

endozepine concentration, as higher central benzodiazepine receptor occupancy by endozepines would reduce flumazenil binding; valproate actions to reduce neuronal expression of this receptor or to reduce affinity of the receptor for flumazenil are less likely but not entirely excluded as alternative explanations for this observation.

REFERENCES Baron JC, Roeda D, Munari C, et al. 1983. Brain regional pharmacokinetics of 11C-labeled diphenylhydantoin positron emission tomography in humans. Neurology 33: 580–585. Berkovic SF. 2000. The neurobiology of ictal SPECT. In Functional Imaging in the Epilepsies (Eds, Henry TR, Berkovic SF, Duncan JS), Lippincott Williams & Wilkins, Philadelphia, pp. 103–110. Briellmann RS, Berkovic SF, Syngeniotis A, King MA, Jackson GD. 2002. Seizure-associated hippocampal volume loss: a longitudinal magnetic resonance study of temporal lobe epilepsy. Ann Neurol 51: 641–644. Cascino GD, Jack CR. Eds. 1996. Neuroimaging in Epilepsy: Principles and Practice, Butterworth-Heinemann, Philadelphia, 289 pp. Chang BS, Lowenstein DH. 2003. Mechanisms of disease: epilepsy. New Engl J Med 349: 1257–1266. Diehl B, Najm I, Ruggieri P, et al. 2001. Postictal diffusionweighted imaging for the localization of focal epileptic areas in temporal lobe epilepsy. Epilepsia 42: 21–28. Duncan JS. 1997. Imaging and epilepsy. Brain 120: 339–377. Duncan JS. 2000. [11C]Flumazenil PET in partial epilepsies. In Functional Imaging in the Epilepsies (Eds. Henry TR, Berkovic SF, Duncan JS), Lippincott Williams & Wilkins, Philadelphia, pp. 204–211. Ebisu T, Rooney WD, Graham SH, Weiner MW, Maudsley AA. 1994. N-acetylaspartate as an in vivo marker of neuronal viability in kainate-induced status epilepticus: 1H magnetic resonance spectroscopic imaging. J Cereb Blood Flow Metab 14: 373–382. Engel Jr J, Kuhl DE, Phelps ME. 1982. Patterns of human local cerebral glucose metabolism during epileptic seizures. Science 218: 64–66. Engel Jr J. 1989. Seizures and Epilepsy. FA Davis, Philadelphia, pp. 536. Engel Jr J. 1996. Surgery for seizures. New Engl J Med 334: 647–652. Flacke S, Wullner U, Keller E, Hamzei F, Urbach H. 2000. Reversible changes in echo planar perfusion- and diffusionweighted MRI in status epilepticus. Neuroradiology 42: 92–95.

Gaillard WD, Fazilat S, White S, et al. 1995. Interictal metabolism and blood flow are uncoupled in temporal cortex of patients with complex partial epilepsy. Neurology 45: 1841–1847. Hajek M, Wieser H-G, Khan N, et al. 1994. Preoperative and postoperative glucose consumption in mesiobasal and lateral temporal lobe epilepsy. Neurology 44: 2125–2132. Henry TR, Mazziotta JC, Engel Jr J, Christenson PD, Zhang JX, Phelps ME, Kuhl DE. 1990. Quantifying interictal metabolic activity in human temporal lobe epilepsy. Journal of Cerebral Blood Flow and Metabolism 10: 748–757. Henry TR, Frey KA, Sackellares JC, Gilman S, Koeppe R, Brunberg J, Ross D, Berent S, Young AB, Kuhl DE. 1993. In vivo cerebral metabolism and central benzodiazepine-receptor binding in temporal lobe epilepsy. Neurology 43: 1998–2006. Henry TR, Votaw JR, Pennell PB, Epstein CM, Bakay RAE, Faber TL, Grafton ST, Hoffman JM. 1999. Acute blood flow changes and efficacy of vagus nerve stimulation in partial epilepsy. Neurology 52: 1166–1173. Henry TR. 2000. Functional imaging studies of epilepsy therapies. In Functional Imaging in the Epilepsies (Eds, Henry TR, Berkovic SF, Duncan JS), Lippincott Williams & Wilkins, Philadelphia, pp. 305–317. Hugg JW, Kuzniecky RI, Gilliam FG, Morawetz RB, Faught RE, Hetherington HP. 1996. Normalization of contralateral metabolic function following temporal lobectomy demonstrated by 1H magnetic resonance spectroscopic imaging. Ann Neurol 40: 236–239. Jackson GD, Opdam HI. Ictal fMRI: methods and models. 2000. In Functional Imaging in the Epilepsies (Eds, Henry TR, Duncan JS, Berkovic SF), Lippincott Williams & Wilkins, Philadelphia, 203–211. Juhasz C, Chugani DC, Muzik O, et al. 2003. Alpha-methyl-Ltryptophan PET detects epileptogenic cortex in children with intractable epilepsy. Neurology 60: 960–968. Kobayashi E, D’Agostino MD, Lopes-Cendes I, Berkovic SF, Li ML, Andermann E, Andermann F, Cendes F. 2003. Hippocampal atrophy and T2-weighted signal changes in familial mesial temporal lobe epilepsy. Neurology 60(3): 405–409. Kuzniecky RI, Jackson GD. Eds. 1995. Magnetic Resonance in Epilepsy, New York, Raven Press. 345 pp. Kwan P, Brodie MJ. 2001. Effectiveness of first antiepileptic drug. Epilepsia 42: 1255–1260. Lansberg MG, O-Brien MW, Norbash AM, Moseley ME, Morrell M, Albers GW. 1999. MRI abnormalities associated with partial status epilepticus. Neurology 52: 1021–1027. Lee BI, Markand ON, Wellman HN, Siddiqui AR, Park HM, Mock B, et al. 1988. HIPDM-SPECT in patients with medically intractable complex partial seizures. Arch Neurol 45: 397–412. Meric B, Barrere B, Peres M, et al. 1994. Effects of kainateinduced seizures on brain metabolism: a combined 1H and 31P NMR study in rat. Brain Res 638: 53–60.

Seizure disorders: overview

Najm I, Wang Y, Hong SC, Lüders HO, Ng T, Comair Y. 1997. Temporal changes in proton MRS metabolites after kainic acid-induced seizures in rat brain. Epilepsia 38: 87–94. Najm IM, Wang Y, Shedid D, Lüders HO, Ng T, Comair Y. 1998. MRS metabolic markers of seizures and seizure-induced neuronal damage. Epilepsia 244–250. Nakasu Y, Nakasu S, Morikawa S, Uemura S, Inubushi T, Handa J. 1995. Diffusion-weighted MR in experimental sustained seizures elicited with kainic acid. Am J Neuroradiol 16: 1185–1192. Newton MR, Berkovic SF, Austin MC, Reutens DC, McKay WJ, Bladin PF. 1992. Dystonia, clinical lateralization, and regional blood flow changes in temporal lobe seizures. Neurology 42: 371–377. O’Brien TJ, So EL, Meyer FB, Parisi JE, Jack CR. 1999. Progressive hippocampal atrophy in chronic intractable temporal lobe epilepsy. Ann Neurol 45: 526–529. Pan JW, Bebin EM, Chu WJ, Hetherington HP. 1999. Ketosis and epilepsy: 31P spectroscopic imaging at 4.1 T. Epilepsia 40: 703–707. Perez ER, Maeder P, Villemure KM, Vischer VC, Villemure JG, Deonna T. 2000. Acquired hippocampal damage after temporal lobe seizures in 2 infants. Ann Neurol 48: 384–387. Petroff OA, Mattson RH, Rothman DL. 2000. Proton MRS: GABA and glutamate. In Functional Imaging in the Epilepsies (Eds, Henry TR, Berkovic SF, Duncan JS), Lippincott Williams & Wilkins, Philadelphia, pp. 261–271. Pirttila TR, Pitkanen A, Tuunanen J, Kauppinen RA. 2001. Ex vivo MR microimaging of neuronal damage after kainateinduced status epilepticus in rat: correlation with quantitative histology. Magn Res Med 46: 946–954. Prevett MC, Lammertsma AA, Brooks DJ, et al. 1995. Benzodiazepine-GABAA receptors in idiopathic generalized epilepsy measured with [11C]flumazenil and positron emission tomography. Epilepsia 36: 113–121. Salmenpera T, Kalviainen R, Partanen K, Mervaala E, Pitkanen A. 2000. MRI volumetry of the hippocampus, amygdala,

entorhinal cortex, and perirhinal cortex after status epilepticus. Epilepsy Res 40: 155–170. Savic I, Thorell JO, Roland P. 1995. [11C]flumazenil positron emission tomography visualizes frontal epileptogenic regions. Epilepsia 36: 1225–1232. Seymour KJ, Bluml S, Sutherling J, Sutherling W, Ross BD. 1999. Identification of cerebral acetone by 1H-MRS in patients with epilepsy controlled by ketogenic diet. Magma 8: 33–42. Spencer SS, Bautista RED. 2000. Functional neuroimaging in localization of the ictal onset zone. In Functional Imaging in the Epilepsies, (Eds, Henry TR, Berkovic SF, Duncan JS), Lippincott Williams & Wilkins, Philadelphia, pp. 285–296. Steriade M. 1993. Cellular substrates of brain rhythms. In Electroencephalography: Basic Principles, Clinical Applications, and Related Fields, 3rd (Eds., Niedermeyer E, Lopes da Silva F), Williams & Wilkins, Baltimore, pp. 27–62. Theodore WH. 1988. Antiepileptic drugs and cerebral glucose metabolism. Epilepsia 29 (suppl 2): S48–S55. Van Paesschen W, Dupont P, Van Driel G, Van Billoen H, Maes A. 2003. SPECT perfusion changes during complex partial seizures in patients with hippocampal sclerosis. Brain 126: 1103–1111. Wall CJ, Kendall EJ, Obenaus A. 2000. Rapid alterations in diffusion-weighted images with anatomic correlates in a rodent model of status epilepticus. Am J Neuroradiol 21: 1841–1852. Wang Y, Majors A, Najm I, et al. 1996. Postictal alteration of sodium content and apparent diffusion coefficient in epileptic rat brain induced by kainic acid. Epilepsia 37: 1000–1006. Warach S, Levin JM, Schomer DL, Holman BL, Edelman RR. 1994. Hyperperfusion of ictal seizure focus demonstrated by MR perfusion imaging. Am J Neuroradiol 15: 965–968. Wasterlain CG. 1974. Mortality and morbidity from serial seizures. Epilepsia 15: 155–176. Wiebe S, Blume WT, Girvin JP, Eliasziw M. 2001. A randomized, controlled trial of surgery for temporal-lobe epilepsy. New Engl J Med 345: 311–318.

487

29

MR spectroscopy in seizure disorders Regula S. Briellmann1,2, R. Mark Wellard1 and Graeme D. Jackson1,2 1

Brain Research Institute, Austin Health, Heidelberg West, Victoria, Australia Department of Neurology, The University of Melbourne, Australia

2

Key points • The clinical aims of MR spectroscopy (MRS) in seizure disorders are to help identify, localize and characterize epileptogenic foci. • Lateralizing MRS abnormalities in temporal lobe epilepsy (TLE) may be used clinically in combination with structural and T2 measurements. • Characteristic metabolite abnormalities are decreased N-acetylaspartate (NAA) with increased choline (Cho) and myoinositol (mI) (short-echo time). • Contralateral metabolite abnormalities are frequently seen in TLE, but are of uncertain significance. • In extra-temporal epilepsy, metabolite abnormalities may be seen where MR imaging (MRI) is normal; but may not be sufficiently localized to be useful clinically. • MRS may help to characterize epileptogenic lesions visible on MRI (aggressive vs. indolent neoplastic, dysplasia). • Spectral editing techniques are required to evaluate specific epilepsy-relevant metabolites (e.g. -aminobutyric acid (GABA)) which may be useful in drug development and evaluation. • MRS with phosphorus (31P) and other nuclei probe metabolism of epilepsy, but are less useful clinically.

Background: brain metabolism and epilepsy The official International League Against Epilepsy (ILAE) classification divides epilepsy into generalized 488

and partial (focal or localization related) seizures (1989). In generalized epilepsy (accounting for approximately 40% of cases) the epileptic discharge begins simultaneously over both cerebral hemispheres, presumed to reflect an underlying diffuse abnormality. In focal epilepsies (accounting for the majority of other cases), the discharge begins in a localized region, reflecting a lesion or other focal abnormality.

Brain metabolism in genetic and acquired causes of seizures Generalized seizures appear to be largely inherited, whereas partial seizures are principally acquired. While this is broadly true, focal epilepsies may also have a genetic background, and generalized epilepsies may also have coexisting developmental abnormalities. Recently there has been progress in identifying specific inherited epilepsies and finding genetic linkages and genetic defects (Berkovic et al., 1998; Roll and Szepetowski, 2002). The first gene found was a miss-sense mutation in 2 subunit of the neuronal nicotinic acetylcholine receptor. It was discovered in patients with autosomal dominant nocturnal frontal lobe epilepsy (ADNFLE), a focal epilepsy first described in 1994 (Scheffer et al., 1994; Steinlein et al., 1995). Other known epilepsy genes include mutations in ion channels, such as potassium channels (KCNQ2 and KCNQ3) (Biervert et al., 1998), and sodium channels (SCN1B) (Scheffer and Berkovic, 1997; Wallace et al., 1998) or the -aminobutyric acid (GABA)-A receptor (Wallace et al., 2001). The mutation in the GABA-A receptor 2 subunit is particularly

MR spectroscopy in seizure disorders

interesting, as it was found in a family with childhood absence epilepsy and febrile convulsions. For the common inherited epilepsies, such as childhood absence epilepsy, the inheritance is complex, and other factors influence the expression of disease. On the other hand, partial epilepsies are generally related to an acquired lesion. Several brain lesions are known to be epileptogenic, including tumours, large dysplastic lesions and brain damage related to trauma, stroke or infection (Engel, 1989). These lesions involve a change in brain structural organization, usually resulting in reduced, displaced or malfunctioning neurons and gliotic reactions. However, in the majority of patients with focal epilepsy no underlying structural abnormalities can be detected. Many of these patients have mild epilepsy with rare seizures, and may suffer from an age- related (Lerman, 1997) or genetic epilepsy syndrome (Berkovic et al., 1996). Some 30% of patients with refractory, severe epilepsy also show no obvious structural abnormalities. Their management is one of the major current challenges in epilepsy centers.

MR investigation of epilepsy The major rationale for the MR investigation of epilepsy patients is to detect and characterize the epileptogenic seizure focus. In the clinical context, MR is mainly used for patients with recent onset or refractory focal seizures. Patients with refractory focal epilepsy may become seizure free after surgical removal of the seizure focus and may return to a normal working life. Therefore, major effort is made to localize the seizure focus in patients where the initial structural MR investigation demonstrates no obvious abnormalities. The pre-surgical investigation of a patient with severe epilepsy typically includes not only MR, but also various other investigations, such as video-electroencephalograms (EEG) telemetry, positron emission tomography (PET), single photon emission computed tomography (SPECT) and neuropsychological assessment (Rayboud et al., 2001). These investigations usually involve inpatient monitoring over several days to characterize habitual seizures. Several reviews evaluate the optimal

clinical imaging protocol for an adult or child with intractable epilepsy (Rayboud et al., 2001; Ruggieri and Najm, 2001; Wright, 2001). A typical clinical scanning protocol for a patient with refractory epilepsy may include T1-weighted imaging, T2weighted imaging, fluid-attenuated inversion recovery (FLAIR) imaging, and three-dimensional (3D) volume acquisition sequences (Jackson et al., 1993; Bradley and Shey, 2000; Rayboud et al., 2001). A 3D T1-weighted volumetric acquisition can be used for tissue segmentation of each voxel into gray matter (GM), white matter (WM) and cerebrospinal fluid (CSF). Common 3D acquisition sequences include magnetization prepared rapid acquisition gradient echo (MPRAGE) and 3D fast spoiled gradient recalled echo acquisition at steady state (3D-SPGR), (Ruggieri and Najm, 2001). When MR abnormalities are concordant with both EEG and clinical findings, the confidence of localization of the seizure focus is improved, and the patient is more likely to be a candidate for surgical treatment (Wright, 2001). In a patient whose MR imaging (MRI) study is normal to visual inspection, abnormalities may sometimes be detected when quantitative analysis methods are used, such as measurements of hippocampal volumes or T2 relaxation times (Bernasconi et al., 2000). However, these methods have not yet found widespread clinical acceptance. Figure 29.1 shows the MR investigation of a patient with refractory focal epilepsy. High-resolution T1-weighted images demonstrate a small lesion with blurred GM–WM interface. Histopathological examination of the resected tissue confirmed the radiological suggestion of focal cortical dysplasia. Apart from finding the cause of the seizures, e.g. defining the epileptogenic lesion, MR is also being used to monitor and understand the disease process, and assess the consequences of seizures, e.g. seizure associated brain damage, and to predict postoperative outcome (Berkovic et al., 1995). These further implications usually involve MR techniques, currently used mainly in research applications. Temporal lobe epilepsy on conventional MR Patients with temporal lobe epilepsy (TLE) have a well-defined clinical pattern, and if they are refractory to medication, a focal structural abnormality can

489

490

Regula S. Briellmann, R. Mark Wellard and Graeme D. Jackson

(a)

(b)

Fig. 29.1 Subtle focal cortical dysplasia. This 30-year-old man presented with focal epilepsy, consisting of complex partial seizures only. MR imaging performed at 3 T showed subtle cortical thickening and blurred GM–WM interface in the post-central gyrus (1A, arrow) suggestive of a focal cortical dysplasia. Other investigations localized the seizure focus to the same area. The patient underwent resective surgery under local anesthesia to minimize the risk of functional impairment due to the excision. The postoperative anatomical deficit is shown in 1B (arrowhead). The patient remained seizure free after surgery.

often be detected. The most common single pathology is hippocampal sclerosis (HS), with typical MR (Jackson et al., 1990) and histopathological features (Graham and Lantos, 1997). Temporal lobectomy is the treatment of choice, and renders many of these patients seizure free (Engel, 1987). The homogeneity and availability of resected brain tissue has lead to a large number of studies in this patient population. Histopathological assessment of the resected sclerotic hippocampus reveals a typical pattern with neuron cell loss in the areas CA1 and CA3, and increased glial cell numbers (Graham and Lantos, 1997). On structural MRI, the affected hippocampus typically shows reduced volume, increased T2weighted signal, and disturbed internal architecture (Jackson et al., 1993). Figure 29.2 shows a typical example of unilateral HS. The volume loss and signal increase can be quantified by volumetric techniques, and T2-relaxometry respectively. Many studies have reported reduced hippocampal volumes (Jack, 1994) and increased T2-relaxometry values (Jackson et al., 1993; Briellmann et al., 1999). These techniques have been widely used for research applications, and it has been shown that volumes and signal inversely correlate with each other (Kälviäinen et al., 1997;

Fig. 29.2 Hippocampal sclerosis (HS). Oblique coronal T2-weighted image recorded on a 3 T machine. The slice shown is at the level of the hippocampus. This patient is a 24-year-old woman, with a past history of prolonged febrile convulsions in early childhood. Her epilepsy started at the age of 10 years as complex partial seizures with oral automatism and infrequent generalization. Her MR shows the typical features of left-sided HS (white arrow). Note the reduced left hippocampal volume, compared to the normal contralateral hippocampus (black arrow), the disturbed internal architecture and the increased signal on these T2-weighted images. Figure taken from Briellman et al. 2002. Neurology 58: 265–271.

MR spectroscopy in seizure disorders

Briellmann et al., 1998). While it is well established that reduced volume reflects neuron cell loss, the pathogenetic background of increased signal has been less clear. In a recent study, we demonstrated a correlation between T2-relaxometry and glial cell counts in the hippocampus (Briellmann et al., 2002), confirming theoretical implications that the increased signal may reflect increased glial cell count.

MR spectroscopy in the investigation of seizure disorders As a non-invasive technique for investigating brain metabolites, MR spectroscopy (MRS) enables the study of metabolic abnormalities in patients with epilepsy. The biochemical information available from MRS complements the detailed structural information that is provided by conventional MR. The value of MRS lies in its role as a means for the diagnosis of the epileptogenic lesion, where it may help to define the extent of a surgical resection, or help to predict post-operative outcome (Eberhardt et al., 2000; Suhy et al., 2002). MRS may give insights into the mechanisms of seizure generation and termination, and into the pharmacodynamics of antiepileptic drugs (AEDs), and may thus possibly help in deciding which patient may tolerate particular drugs. With its broad application, MRS has been used in the clinical context to solve the riddle of seizure focus localization in patients without obvious MR abnormalities (“MR-negative” epilepsy). Over more than two decades MRS has developed a prominent role in research applications where the basic seizure disorder process is investigated. Early studies using phosphorus (31P) MRS to study a rabbit model of status epilepticus were carried out in the 1980s by Prichard and colleagues in Yale (Prichard et al., 1983; Petroff et al., 1984b) showing changes in high-energy phosphates and pH. Later it was shown that lactate (Lac) was elevated for a prolonged time following brief seizures (Prichard et al., 1987; Young et al., 1989), and 13C MRS documented that this elevation was the result of continuing turnover of the Lac pool (Petroff et al., 1992). In 1986, a 31P MRS study of infants (Younkin et al., 1986) showed, that the phosphocreatine (PCr)/inorganic

phosphate (Pi) ratio was decreased at the seizure focus, during seizures and returned to normal after the seizure ended. Elevated Lac in a region of chronic encephalitis in a patient with Rasmussen’s encephalitis was reported using proton MRS (Matthews et al., 1990). The first large study with proton MRS in TLE was reported by Connelly and co-workers in 1994 (Connelly et al., 1994). Figure 29.3 shows an MRS investigation in a patient that had a seizure of frontal lobe origin about 2 h before the scan. A Lac peak was evident at 1.3 ppm in both frontal lobes. Many subsequent studies have demonstrated metabolic changes associated with epilepsy; both associated with the seizure focus and in other areas, possibly related to seizure spread. These include studies of proton, phosphorous and carbon nuclei. Proton (1H) MRS in epilepsy Proton (1H) MRS provides information on several brain metabolites (cf. Chapter 1). Of major interest for the investigation of epilepsy patients are: Nacetylaspartate (NAA), Choline (Cho)-containing metabolites, creatine (Cr)/PCr, and Lac, all of which are measurable using short- or long-echo acquisitions. Spectra acquired with short-echo times (TE) provide information about these metabolites, and in addition myoinositol (mI), glutamate (Glu) and glutamine (Gln), aspartate and alanine, and several other metabolites (Danielsen and Ross, 1999), some of which require spectral editing techniques for detection (e.g. GABA). NAA, is commonly regarded as specific to neurons (Urenjak et al., 1993). A reduction in the NAA signal, or in its ratio to other metabolite signals, is commonly interpreted as indicative of neuronal loss or impaired neuronal function. It has been suggested that Cr and Cho, are concentrated in glial cells (Urenjak et al., 1993). Increased Cho suggests gliosis and elevated membrane turnover (Danielsen and Ross, 1999). MI is a putative marker of gliosis and an organic osmolyte (Nonaka et al., 1999) involved in cellular volume control (Gullans and Verbalis, 1993). Studies in animals have identified rapid reduction of mI as a mechanism of brain volume regulation following hyponatremia, with levels being slow to return to normal on normalization of the osmotic

491

492

Regula S. Briellmann, R. Mark Wellard and Graeme D. Jackson

TE 35 ms

TR 1500 TE 135 3.0 2.0 1.0 Chemical shift (ppm)

0.0

Fig. 29.3 Rasmussen’s encephalitis. MR investigation of a girl with Rasmussen’s encephalitis affecting the left hemisphere, particularly the mesial temporal and mesial frontal structures. This girl was healthy up to the age of 6 years, when she suddenly developed epilepsy with frequent focal seizures. Two days before the MR scan she had an episode of epilepsia partialis continua, a focal status epilepticus continuing for several hours. The MR investigation shown includes single voxel MRS with long TE (bottom spectrum) and short TE (top spectrum) acquisition. The localization of the voxel can be seen on the axial T2-weighted image on the left side (box). Note the increased T2-signal in the mesial temporal structures, seen both on the axial slice, and on the two coronal slices (right). The two coronal slices also demonstrate involvement of the left mesial frontal area. MRS examination of the high-intensity lesions seen in T2-weighted images shows reduced NAA, increased mI and the presence of lactate (Lac) together with possibly elevated mobile lipid signal at 1.0 ppm. The two spectra were recorded with a PRESS sequence and a TE/repetition time (TR) of 30/1500 ms (upper) and a TE/TR of 135/1500 ms (lower spectrum). The inverted signal at 1.3 ppm in the long TE measurement confirms the presence of Lac (spectra provided by M. Kean, Royal Children’s Hospital, Melbourne).

environment (Brand et al., 1999). Figure 29.4 shows an extreme example of NAA reduction and increase in lac, Cr and Cho. This young girl had a dysembryoplastic tumor. Spectral editing techniques (Hetherington et al., 1997) allow the differentiation of metabolites with signals that co-resonate with other metabolites, such as Gln, Glu and GABA. GABA concentrations have been implicated in the control of epilepsy and therefore, the tissue concentration of this metabolite is of interest,

particularly in developing an understanding of the mechanisms of action of medications that increase tissue GABA content (Weber et al., 1999). Proton MRS is of particular importance in patients with brain tumours who may present with seizures. The characteristic elevation of Cho makes MRS a valuable tool for the diagnosis of tumours and their differentiation from other lesions (Burtscher and Holtas, 2001). There is also good evidence that metabolite profile can differentiate between tumour

MR spectroscopy in seizure disorders

2.0 106 151,773

(b)

1.9 106 212,253

(a)

NA Cr

Cho Glx

0

ml

4.0 3.8 3.6 3.4 3.2 3.0 2.8 2.6 2.4 2.2 2.0 1.8 1.6 1.4 1.2

3.8 3.6 3.4 3.2 3.0 2.8 2.6 2.4 2.2 2.0 1.8 1.6 1.4 1.2

Chemical shift (ppm)

Chemical shift (ppm)

ml

Cho

Cr

NA

Glx

Right

3.1

4.4

7.7

3.2

0.8 0.7

3.1

Left

3.4

5.1

6.0

3.3 (0.3)

1.5 (0.1)

5.0 (0.6)

7.5 (0.6)

8.2 (1.2)

Control

Fig. 29.4 This figure shows the results of the MR investigation of a 9-year-old girl. Epilepsy started at age 5 years with complex partial seizures, occurring in clusters. Neuropsychological function was probably normal. MR investigation at 3 T showed a left temporal lesion, possibly a low-grade tumor (a). MRS investigation was performed to assess the metabolite profile of the lesion, and to examine the metabolic integrity of the contralateral hemisphere. MRS data were analyzed by linear combination model (LCModel). Spectra were acquired from four single voxels (TE/TR 30/3000 ms, voxel size 2 2 2 cm), comprising bilateral temporal (shown in figure) and bilateral frontal lobe positions. The quality of the LCModel fit is demonstrated by the residual plotted above each spectrum (b). The metabolite concentrations are given in a table below the spectra. Metabolites are expressed in institutional units, approximating mmol/l. Control values are based on a series of 20 healthy volunteers with average values and standard deviations (in brackets) given. Patient values outside the range of normal controls (2 SD from the mean of control values) are highlighted in red. The left-temporal voxel included parts of the lesion (for voxel placement cf. Figure 29.6). The temporal lobe spectra showed bilateral reduction of NAA. This suggests that the refractory epilepsy is also associated with impaired neuronal integrity in the hemisphere not affected by the lesion. Cho and Cr levels also appeared low bilaterally. Cho was not elevated in the lesion, consistent with a non-aggressive tumor. The frontal lobe spectra were normal.

493

494

Regula S. Briellmann, R. Mark Wellard and Graeme D. Jackson

Right

Left

NAA (3.0)

NAA (4.5)

ml (1.5)

ml (6.3)

NA

ml NA

ml

4.2 4.0 3.8 3.6 3.4 3.2 3.0 2.8 2.6 2.4 2.2 2.0 1.8 1.6 1.4 1.2 1.0

4.2 4.0 3.8 3.6 3.4 3.2 3.0 2.8 2.6 2.4 2.2 2.0 1.8 1.6 1.4 1.2 1.0

Chemical shift (ppm)

Chemical shift (ppm)

Fig. 29.5 Ganglioglioma. Proton MRS of a patient with left-temporal ganglioglioma and TLE performed at 3 T. Temporal lobe single voxel spectra where acquired. A coronal T1-weighted image (top of the figure) shows the lesion in the region of the left hippocampus, and the placement of the temporal lobe voxels. MRS data were analyzed by LC Model. The lower section of this figure shows the fitted spectra from the left and right side, and above, as an insert, the fitted components attributed to NAA (top) and mI (middle). Explanations of the abbreviations and the table are given in Figures 29.3 and 29.5. The individual metabolite curves show the elevated mI and reduced NAA on the left side, compared to the right side, which has normal metabolite concentrations. Concentrations for each metabolite are given in brackets. These findings suggest reduced neuron cell count or integrity and glial cell proliferation. Spectra were acquired using a PRESS sequence with parameters: TE/TR 30/3000 ms, 2048 data points over a spectral width of 5000 Hz from a 2 2 2 cm voxels. Note the peak in the in vivo spectrum is labeled NA because it includes a small contribution from NAAG, whereas the individual metabolite curve above the spectrum is for NAA. Figure taken from Briellmann RS. et al. 2003. Epilep Dis 5: 3–20.

types (Moller-Hartmann et al., 2002). Figure 29.5 shows the structural and metabolic profile of a girl with recent onset focal epilepsy. The large lesion showed normal Cho levels, a strong argument against the presence of a fast growing tumor. TLE and proton (1H) MRS Lateralizing MRS abnormalities were described in TLE patients in 1994 (Connelly et al., 1994). They have been replicated in adults (Cendes et al., 1997a; Najm et al., 1998; Meiners et al., 2000) and children (Cross et al., 1996; Holopainen et al., 1998). The abnormalities typically consist of reduced NAA

signal and increased Cho and mI signals, suggestive of gliosis (Najm et al., 1998). These MRS findings are consistent with the histopathological characteristics of reduced neuron cell counts and increased glial cell numbers. More recently, increased mI has been reported as a consequence of induction of Na/myoinositol cotransporter (SMIT) following seizure activity (Nonaka et al., 1999) associated with glial proliferation in the seizure focus. Figure 29.5 shows temporal lobe spectra in a patient with ganglioglioma on the left side. The mI was increased and NAA was reduced in the voxel covering his epileptogenic lesion.

MR spectroscopy in seizure disorders

Contralateral MRS abnormalities Bilateral temporal lobe MRS abnormalities have been found in up to 50% of TLE patients (Ende et al., 1997; Connelly et al., 1998). The significance of bilateral abnormalities is poorly understood, partly because pathological specimens are rarely available from the temporal lobe contralateral to the seizure focus. However, autopsy studies suggest a high prevalence of bilateral hippocampal abnormalities in TLE patients (Margerison and Corsellis, 1966; Babb, 1991). This is consistent with quantitative MRI studies reporting the frequent occurrence of bilateral abnormalities (Barr et al., 1997; Quigg et al., 1997). It has been shown that contralateral NAA abnormalities measured by MRS are reversible with time, suggestive of transient neuronal dysfunction (Serles et al., 2001). The presence of bilateral metabolic changes, has been associated with poor post-operative seizure outcome (Kuzniecky et al., 1999; Eberhardt et al., 2000). Widespread proton MRS abnormalities Whereas the presence of bilateral temporal lobe abnormalities may be consistent with the presence of independent bilateral foci, there is also the possibility that the contralateral changes may reflect abnormalities due to seizure spread. In severe TLE, functional abnormalities beyond the seizure focus have been documented by PET (Arnold et al., 1996) and SPECT (Rabinowicz et al., 1997). Metabolite changes in lobes other than that harboring the seizure focus suggest that seizure spread may cause changes

(A) 6.0 Frontal lobe mI (mM)

MRS may be helpful in the identification of the seizure focus in refractory focal epilepsy patients without obvious MR abnormalities. Some studies reported MRS metabolite abnormalities in patients with focal epilepsy, without obvious structural MR abnormalities (Connelly et al., 1998; Woermann et al., 1999; Li et al., 2000). They consisted of decreased NAA and increased Cho, lateralized to the seizure focus, similar to the patients with HS. However, the specificity of these abnormalities has been questioned. Metabolic abnormalities have been found not only in seizure foci (Najm et al., 1998), but also in areas distant to the seizure focus (Najm et al., 1998), such as the contralateral temporal lobe (Serles et al., 2001).

5.5 5.0 4.5

∗

4.0 ∗∗∗ 3.5 3.0 Control

LO-TLE

HS-TLE

Fig. 29.6 The mI changes in frontal lobe. A comparison of frontal lobe mI concentration, in control, late onset-TLE (no MRI abnormality) and HS-TLE groups is shown in Figure A. The mean and standard deviation is shown for each group by the solid square and error bars (compared with controls: *p  0.05; **p  0.01 and ***p  0.001; from Wellard et al. (2003) with permission). Figure B shows spectra from right frontal lobe (left side: (a), (b) and (c)) and right-temporal lobe (right side: (d), (e) and (f)). Note the higher spectral quality from the frontal lobe. Spectra from controls (top row: (a) and (d)), late-onset epilepsy (middle row: (b) and (e)) and HS (bottom row: (c) and (f)) are shown. The abbreviations are as indicated in Figure 29.3. The mI component of the LC Model-fitted spectrum is inset. The dotted line marks the mI signal in each spectrum. Note the reduced NAA in the HS temporal lobe (f) and reduced mI in HS frontal lobe (c) (from Wellard et al. (2003) with permission).

detectable by MRS. Widespread MRS abnormalities have been found in several studies (Cendes et al., 1997a; Kuzniecky et al., 1998; Meiners et al., 2000; Miller et al., 2000; Mueller et al., 2002; Wellard et al., 2003). In a recent study, we found frontal lobe reduction of mI concentrations in both TLE with HS and in mild TLE (Wellard et al., 2003), which may be due to either a short-term change following recent seizure activity or a cumulative effect of chronic seizures (Figure 29.6). A recent report has shown reduced frontal lobe NAA in severe refractory TLE (Suhy et al., 2002). When compared to volumetric changes, also known to occur in brain lobes distant to the seizure focus, MRS abnormalities were not associated with volumetric deficits in the frontal GM and

495

496

Regula S. Briellmann, R. Mark Wellard and Graeme D. Jackson

(B)

(a)

(d)

(b)

(e)

(c)

(f)

4.2 4.0 3.8 3.6 3.4 3.2 3.0 2.8 2.6 2.4 2.2 2.0 1.8 1.6 1.4 1.2

Chemical shift (ppm) Fig. 29.6 (cont.)

4.2 4.0 3.8 3.6 3.4 3.2 3.0 2.8 2.6 2.4 2.2 2.0 1.8 1.6

Chemical shift (ppm)

1.4 1.2

MR spectroscopy in seizure disorders

WM (Mueller et al., 2002). This suggests that the widespread MRS abnormalities in TLE may have a different origin from the widespread volumetric changes or that the metabolite changes may precede the structural changes. Proton MRS has proven to be a sensitive measure to detect metabolic pathologies in patients with TLE, however, it remains to be clarified whether this additional information adds to the overall management of the patients. With its high sensitivity, metabolite abnormalities in brain regions, distinct from the seizure focus, can be detected and it remains difficult to disentangle which abnormalities are due to causes or consequences of seizures. Some indication can be taken from a large study performed on 82 patients with refractory TLE. Ipsilateral and contralateral temporal lobe NAA/Cr ratios were negatively correlated with the duration of epilepsy (Bernasconi et al., 2002). Patients with frequent generalized tonic–clonic seizures had lower NAA/Cr than patients with no or rare generalized tonic–clonic seizures. This suggests that ongoing seizures may induce additional neuronal damage, which will progress in parallel to the duration of the epilepsy.

Extra-TLE and proton MRS In contrast to the numerous proton MRS studies of TLE patients, there are only few reports on other types of partial epilepsy (Stanley et al., 1998; Kikuchi et al., 2000). These studies suggest that the potential for correct seizure focus lateralization is less than in TLE. In one study comparing the coincidence rate between the seizure focus and the reduction of the NAA/Cr ratio, correct lateralization was present in 19 of the 21 TLE patients, but only in four of the seven frontal lobe epilepsy patients (Kikuchi et al., 2000). NAA reduction was observed to be greatest at the seizure focus in a study of 20 patients with epilepsy of frontal or central/post-central origin, suggesting that the MRS abnormalities in extra-temporal epilepsy might therefore not be localized enough to readily specify the seizure focus (Stanley et al., 1998). Figure 29.7 shows frontal lobe spectra in a patient with focal cortical dysplasia, in which the direction of the MRS abnormalities suggested a pathology on the left side.

In extra-TLE, the seizure focus is usually neocortical. In contrast, in patients with hypothalamic hamartoma, seizure onset is in the subcortical lesion (Berkovic et al., 1988; Cascino et al., 1993). Patients with hypothalamic hamartoma usually have frequent gelastic seizures starting early in life, and may show cognitive impairment and behavioral disturbances. Figure 29.8 shows an MRS spectrum recorded from a hypothalamic hamartoma, documenting reduced NAA and increased mI content, compared to healthy thalamic tissue.

Single voxel MRS and chemical shift imaging in epilepsy Various different MRS protocols have been developed for the evaluation of patients with temporal lobe or extra-temporal epilepsy. Generally, these protocols fall either into the classification of single voxel or MR spectroscopic imaging (MRSI) techniques (de Graaf, 1998; Pauli et al., 2000) (cf. Chapter 1 for full details). MRSI has been increasingly used in the investigation of epilepsy patients (Stanley et al., 1998; Eberhardt et al., 2000; Capizzano et al., 2002). One study reported no significant difference between hippocampal NAA/(Cho  Cr) ratios obtained by single voxel spectroscopy and chemical shift imaging (CSI) (Hsu et al., 1999). Further, abnormalities evident from MRSI measurements have been reported in areas of the brain remote from the clinically identified seizure focus (Capizzano et al., 2002).

Improvements in the analysis of MRS data Methods for the analysis and quantification of MRS data are presented in detail in Chapter 2. In epilepsy patients, reduction in NAA and increases in Cr and Cho are commonly observed. Therefore, results have often been presented as NAA/Cho or NAA/Cho  Cr ratio (Cross et al., 1993; Connelly et al., 1994). Ratios are dimensionless and can therefore be used to compare results between centers, without the need

497

105937 6688

Regula S. Briellmann, R. Mark Wellard and Graeme D. Jackson

0

0

498

4.0 3.8 3.6 3.4 3.2 3.0 2.8 2.6 2.4 2.2 2.0 1.8 1.6 1.4 1.2 1.0

4.0 3.8 3.6 3.4 3.2 3.0 2.8 2.6 2.4 2.2 2.0 1.8 1.6 1.4 1.2 1.0

Chemical shift (ppm)

Chemical shift (ppm)

Fig. 29.7 Focal cortical dysplasia. The figure shows the MR investigation of a young woman with epilepsy. She had complex partial seizures, which started at the age of 7 years. Her seizures involved bizarre movements, were sometimes very short, but at other times prolonged, lasting several hours. She was initially diagnosed as having non-epileptic attacks. MR investigation at the age of 29 years showed a lesion with blurred white gray matter (GM) distinction in the left frontal lobe. As this lesion involved the inferior frontal gyrus, extensive pre-surgical language assessment was performed, including functional MRI of language. The lesion was removed under local anesthesia and she showed excellent post-operative outcome, with no seizures, and complete recovery from initial language difficulties. MRI was performed on a 3 T GE LX Horizon scanner (Milwaukee, US). The axial T1-weighted image in the middle shows the voxel position (black box) in the right hemisphere. The area of dysplastic cortex is highlighted with arrows. Bilateral single voxel spectra where acquired using a PRESS sequence (TE/TR 30/3000 ms, voxel size 2 2 2 cm). MRS data were analyzed with LCModel. The figure shows bilateral frontal spectra, covering the area of the lesion on the left side, and a congruent anatomical area on the right. Relative to controls, the dysplastic region in the left frontal lobe showed significantly reduced NAA (30%), elevated Cho (35%) and mI (50%). This suggests reduced neuron cell counts, or impaired neuronal integrity in the region of the seizure onset. The increased Cho/Cr may indicate the presence of glial proliferation at this site. Explanation of abbreviations and the table are given in legends for Figures 19.3 and 19.5.

for a reference measurement. However, they only allow assessment of changes in a combination of metabolites, and give no information on the changes in single metabolites. Also, abnormalities will not be detected when all metabolites change in the same direction. An advance was made with the development of software enabling spectral deconvolution, providing information on the concentration of individual metabolites. Of the several published software programs used for spectral analysis (Maudsley et al., 1992; Provencher, 1993; Naressi et al., 2001), LCModel, developed by Provencher (1993), is of use for 1H and

13

C MRS analysis. It allows operator independent estimation of absolute metabolite concentrations by comparison of in vivo data with a basis set of reference spectra prepared from chemical solutions of known concentration. The fitting routine uses the complexity of the individual metabolite spectra to aid in the identification of individual metabolite signals (Provencher, 1993). An increasing number of reports maximize the information gained from spectra by reporting absolute metabolite concentrations (Savic et al., 2000). 31P spectra are more often analyzed by other methods, e.g. MR user interface (MRUI)

MR spectroscopy in seizure disorders

80,815 0 507,764

0 4.0 3.8 3.6 3.4 3.2 3.0 2.8 2.6 2.4 2.4 2.0 1.8 1.6 1.4 1.2 1.0

(Naressi et al., 2001), due to the reduced complexity of the recorded spectra. Other improvements in MRS data analysis have been gained with the use of methods accounting for the differences in metabolite concentration between GM and WM (Hetherington et al., 2002) and correction for magnetic field inhomogeneity across regions examined by CSI (McLean et al., 2000). Estimation of the metabolite concentrations in pure cortical GM or WM can be made following tissue segmentation and regression with the fractional GM, WM and CSF content within the assessed voxel (McLean et al., 2000).

Chemical shift (ppm) 31

P MRS in epilepsy

31

0.0050 0.0040 0.0030 0.0020 0.0010 0.0000

0.0010 4

3

2

1

0

Frequency (ppm) Fig. 29.8 Hypothalamic hamartoma. The middle panel shows a T1-weighted sagittal image. The hypothalamic hamartoma lies within the box, which indicates the MRS voxel position. The MRS was recorded at 1.5 T. The fitted spectrum is shown in the upper panel (using LCModel). This spectrum shows a significant reduction in NAA and elevated mI, compared to spectra recorded from thalamus in healthy subjects. The lower panel shows the same spectrum as presented by the scanner manufactures software. (Spectrum provided by M. Kean, Royal Children’s Hospital, Melbourne.)

P MRS allows evaluation of the energetic state of the brain, by providing a measure of nucleoside triphosphates (NTP) (predominantly adenosine triphosphate (ATP)), PCr, phosphomonoesters (PME), phosphodiesters (PDE) and Pi in brain tissue (Buchli et al., 1994). The information about the levels of high-energy phosphate metabolites is derived from ATP and PCr. Typically the PCr/ATP ratio is quoted as an indicator of the tissue energy status. While cerebral ATP is depleted only under severe metabolic conditions, changes in PCr have been observed with 31P MRS following neuronal activity (SappeyMarinier et al., 1992). The PME peak predominantly comprises phosphoethanolamine, phosphocholine (precursors of cell membranes (Bretscher, 1972)). The PDE resonance includes glycerophosphocholine (GPC), glycerophosphoethanolamine (GPE) and mobile phospholipids. The GPC and GPE represent membrane breakdown products (Bretscher, 1972) but the majority of the signal (80%) probably originates from intracellular mobile phospholipids. Additionally, the chemical shift of the Pi resonance provides an excellent measure of intracellular pHi (Gadian et al., 1982). Changes in the relative PME and PDE concentrations have been associated with membrane turnover (Bluml et al., 1999). Phosphoesters are of considerable interest because they represent precursors of membrane synthesis and breakdown products. It is possible to increase the sensitivity and improve the differentiation of phosphoesters by using a technique known as proton

499

500

Regula S. Briellmann, R. Mark Wellard and Graeme D. Jackson

decoupling, which involves a second radiofrequency (RF) transmitter channel which is used simultaneously with the detection of the 31P signal. Decoupling can increase the 31P signal for some compounds by more than 50% and also considerably improves the resolution of the phosphoester regions of the spectrum. An example of a 31P spectra in a volunteer is shown in Figure 29.9. TLE and 31P MRS In 31P MRS studies of epilepsy patients, changes in PCr can be expected following neuronal activity (Sappey-Marinier et al., 1992). During and shortly after a seizure, changes in pHi and high-energy phosphates may be found (Petroff et al., 1984a). Interestingly, these metabolites have been reported to be abnormal in relatively seizure-free patients (Chu et al., 1998a). 31P MRS also provides information on mobile PME and PDE associated with membranes (Gadian, 1995). 31P MRS has shown potential for lateralizing metabolic dysfunction (Chu et al., 1998a). Decreased PCr/Pi was observed in 65–75% of patients with TLE (Laxer et al., 1992a). This may relate to the timing of measurements relative to recent seizure activity. Altered 31P metabolites and pHi have been reported in the post-ictal period. An important feature of the findings reported to date is that abnormalities are present in regions that are normal to a range of other measurements, including MRI. They may therefore be unique markers for an early stage of processes by which normal neurons are recruited into the permanently epileptogenic neuronal population (Figure 29. 9). Several groups have investigated a potential reduction in pH in patients with TLE (Hugg et al., 1992; Kuzniecky et al., 1992; Laxer et al., 1992b; van der Grond et al., 1998). In one study (Kuzniecky et al., 1992) there were significant differences between patients and controls with respect to the PCr/Pi ratio. However, no significant differences were found between patients and controls with respect to pH. In contrast, others have found an elevation of pH, along with increased Pi and reduced PME in the ipsilateral temporal lobe (Hugg et al., 1992; van der Grond et al., 1998). Reduced ATP/Pi and PCr/Pi ratios in the temporal lobe were predictive for the side

of the seizure focus in more than 70% of patients studied (Chu et al., 1998b). Overall, the results of these studies are controversial, although it appears that some 31P metabolite abnormalities are indeed present in the epileptogenic temporal lobe. Variations between the reports may originate from a difference in timing of the MRS study, relative to seizures. Some of the limitations of 31P spectroscopy should also be recognized; because of its lower resonance frequency, and the relatively low concentrations of 31P-containing compounds, signal-to-noise ratios (SNR) are low in 31P MRS, and large voxel sizes are needed (typically 30 cm3 or larger). Therefore, the technique is relatively insensitive and cannot readily be used to study small focal lesions or seizure foci. It is also not commonly available because special hardware is required (dedicated 31P head coils and broadband transmitter and receiver electronics).

Other nuclei for MRS in epilepsy There are also other nuclei which can be examined by in vivo MRS, and some of these may be of potential interest for the study of seizure disorders. These nuclei include carbon (13C), sodium (23Na) and fluorine (19F). The low natural abundance of 13C can be used to advantage if enriched in exogenous metabolites. The metabolic fate of 13C-labeled substrates can be followed to determine flux through metabolic pathways with the appearance of newly labeled metabolites in sequential spectra. For example, administration of 13C-labeled glucose and acetate can be used to study the tricarboxylic acid (TCA) cycle and Gln–Glu cycling (Bluml et al., 2001). Metabolite flux in such studies can also be used to determine the kinetics of the metabolic pathways (Shen et al., 1999). In vitro 13C MRS has already been used to show reduced Gln–Glu cycling in surgically resected tissue from epileptogenic hippocampus (Petroff et al., 2002). This technique also been used to examine other metabolite pathways in animals (Patel et al., 2001), and it is likely that such studies eventually will be extended to humans. Studies including other nuclei are rare. There is only one 23Na study of epilepsy reported (Schnall et al., 1988). This is an area yet to be examined with current

MR spectroscopy in seizure disorders

(a)

(b)

(c)

4

4.0 104

2

3.0 104 3

1

7

2.0 104

6

5

1.0 104

0 10

0

10 Frequency (ppm)

20

Fig. 29.9 31P MRS. (a) 31P chemical shift image recorded from a normal volunteer at 3 T, using a flip angle of 60°, 1.1 s TR, 14 NEX, field of view (FOV) of 26 cm, 10 phase encoding steps, slice thickness of 3 cm and NEX of 18, yielding a nominal voxel size of 20 cm3. Total acquisition time was 33 min. The spectral data are overlayed onto a proton image to provide anatomical information about individual voxel positions. (b) The intensity of individual signals can be used to generate a metabolite map, as shown for PCr in this figure. The data were zero filled to 16 16 points prior to Fourier transformation. (c) Average 31P spectrum recorded from the region shown in (a) and (b). Peaks, labeled from right to left: ATP ,  and  phosphates (1, 2, 3, respectively), PCr (4), PDE (5), Pi (6) and PME (7).

high-field facilities. The nucleus 19F is not naturally present in vivo and has yet to be used for the study of epilepsy in humans. However, there are studies utilizing MRS of this nucleus in animal models of epilepsy (Eleff et al., 1988) and other diseases to examine the fate of 19 F-labeled compounds (Strauss et al., 1997).

As with 31P MRS, the clinical application of these other nuclei is limited by their low sensitivity and the need for special hardware. In addition, for 13C or other studies using exogenously labeled substrates, the cost of the labeled compounds can be prohibitively high, and long scan times are required.

501

502

Regula S. Briellmann, R. Mark Wellard and Graeme D. Jackson

High-field MRS and future direction

Conclusion

The future lies in the development of new, faster MRSI acquisition protocols (Li et al., 2001; Dreher and Leibfritz, 2002) as well as in the refinement of existing techniques. This should allow the use of techniques that would otherwise require acquisitions times greater than practicable for human studies. This includes techniques such as spectral editing of J-coupled resonances measured with subtraction techniques such as those used for GABA (Behar et al., 1994; Weber et al., 1999) and sequences that inherently detect only weak signals, such as those detected by multiple quantum techniques. The presence of specific mutations to ion channels in several inherited epilepsies suggests that changes in brain metabolism underlie seizure generation. These abnormalities may be detectable by MRS. For example, it is possible to detect elevated GABA levels following the administration of vigabatrin (Petroff et al., 1995; Weber et al., 1999), a drug used to prevent seizures that increases intracerebral GABA concentrations. Only a few studies have examined MRS changes in idiopathic epilepsy (Cendes et al., 1997b; Hill et al., 1999; Savic et al., 2000). Despite these epilepsies having normal neuroimaging features, one study reported a reduction in frontal NAA levels in patients with juvenile myoclonic epilepsy (Savic et al., 2000), suggesting neuronal abnormalities in these patients. Several studies have reported MRS abnormalities in epilepsy, assessed at very high-field strength. Reduced NAA/Cho ratios have been shown in patients with epilepsy due to malformations of cortical development (MCD501), using a magnetic field strength of 4.1 T (Kuzniecky et al., 1997). Future developments in MRS may enable the technique to benefit the investigation of seizure generation by correlating metabolism with interictal discharges (Maton et al., 2001; Park et al., 2002). MRS may also be of use in monitoring the progression of neuronal damage due to ongoing seizure activity (Bernasconi et al., 2002). There is also the possibility that MRS will help to evaluate anticonvulsant therapy in vivo (Braun et al., 2001).

There is no doubt that 1H and 31P spectroscopy detects relevant metabolite changes in patients with TLE. Numerous studies confirmed reduction in NAA and in the ratio of PCr/Pi (Hetherington et al., 2002). In his 1999 review, Kuzniecky concluded that proton MRS, using single-voxel or CSI, lateralizes TLE in 65–96% of cases, with bilateral changes seen in 35–45% of cases, while 31P MRS shows a lateralizing PCr/Pi ratio in 65–75% of the TLE patients (Kuzniecky, 1999). There are indications that these changes are reversible with seizure treatment. Improvements in MRS technology, such as the ability to calculate absolute concentrations, to account for differences between GM and WM and to achieve better spectral resolution by use of a higher magnetic field strength, will now allow more extensive use of this “old” MR technique for patients with epilepsy.

REFERENCES Antel SB, Li LM, Cendes F, Collins DL, Kearney RE, Shinghal R, et al. 2002. Predicting surgical outcome in temporal lobe epilepsy patients using MRI and MRSI. Neurology 58: 1505–1512. Arnold S, Schlaug G, Niemann H, Ebner A, Lüders H, Witte OW, et al. 1996. Topography of interictal glucose hypometabolism in unilateral mesiotemporal epilepsy. Neurology 46: 1422–1430. Babb TL. 1991. Bilateral pathological damage in temporal lobe epilepsy. Can J Neurol Sci 18: 645–648. Barr WB, Ashtari M, Schaul N. 1997. Bilateral reductions in hippocampal volume in adults with epilepsy and a history of febrile seizures. J Neurol Neurosurg Psychiatr 63: 461–467. Behar KL, Rothman DL, Spencer DD, Petroff OA. 1994. Analysis of macromolecule resonances in 1H NMR spectra of human brain. Magn Reson Med 32: 294–302. Berkovic SF, Andermann F, Melanson D, Ethier RE, Feindel W, Gloor P. 1988. Hypothalamic hamartomas and ictal laughter: evolution of a characteristic epileptic syndrome and diagnostic value of magnetic resonance imaging. Ann Neurol 23: 429–439. Berkovic SF, Howell RA, Hay DA, Hopper JL. 1988. Epilepsies in twins: genetics of the major epilepsy syndromes. Ann Neurol 43: 435–445.

MR spectroscopy in seizure disorders

Berkovic SF, McIntosh A, Howell RA, Mitchell A, Sheffield LJ, Hopper JL. 1996. Familial temporal lobe epilepsy: a common disorder identified in twins. Ann Neurol 40: 227–235. Berkovic SF, McIntosh AM, Kalnins RM, Jackson GD, Fabinyi GCA, Brazenor GA, et al. 1995. Preoperative MRI predicts outcome of temporal lobectomy: an actuarial analysis. Neurology 45: 1358–1363. Bernasconi A, Bernasconi N, Caramanos Z, Reutens DC, Andermann F, Dubeau F, et al. 2000. T2 relaxometry can lateralize mesial temporal lobe epilepsy in patients with normal MRI. NeuroImage 12: 739–746. Bernasconi A, Tasch E, Cendes F, Li LM, Arnold DL. 2002. Proton magnetic resonance spectroscopic imaging suggests progressive neuronal damage in human temporal lobe epilepsy. Prog Brain Res 135: 297–304. Biervert C, Schroeder BC, Kubisch C, Berkovic SF, Propping P, Jentsch TJ, et al. 1998. A potassium channel mutation in neonatal human epilepsy. Science 279: 403–406. Bluml S, Moreno A, Hwang JH. 2001. 1-(13)C glucose magnetic resonance spectroscopy of pediatric and adult brain disorders. NMR Biomed 14: 19–32. Bluml S, Tan J, Harris KJ, et al. 1999. Quantitative protondecoupled 31P MRS of the schizophrenic brain in vivo. Comput Assist Tomogr 23: 272–275. Bradley WG, Shey RB. 2000. MR imaging evaluation of seizures. Radiology 214: 651–656. Brand A, Leibfritz D, Richter-Landsberg C. 1999. Oxidative stress-induced metabolic alterations in rat brain astrocytes studied by multinuclear NMR spectroscopy. J Neurosci Res 58: 576–585. Braun J, Seyfert S, Bernarding J, Schilling A, Marx P, Tolxdorff T. 2001. Volume-selective proton MR spectroscopy for in-vitro quantification of anticonvulsants. Neuroradiology 43: 211–217. Bretscher MS. 1972. Asymmetrical lipid bilayer structure for biological membranes. Nature New Biol 236: 11–12. Briellmann RS, Jackson GD, Kalnins R, Berkovic SF. 1998. Hemicranial volume deficits in patients with temporal lobe epilepsy with and without hippocampal sclerosis. Epilepsia 39: 1174–1181. Briellmann RS, Jackson GD, Mitchell LA, Fitt GJ, Kim SE, Berkovic S. 1999. Occurrence of hippocampal sclerosis: is one hemisphere or gender more vulnerable? Epilepsia 40: 1816–1820. Briellmann RS, Kalnins RM, Berkovic SF, Jackson GD. 2002. Hippocampal pathology in refractory TLE: T2-weighted signal change reflects dentate gliosis. Neurology 58: 265–271. Briellmann RS, Pell GS, Wellard RM, Mitchell LA, Abbott DF, Jackson GD. 2003. MR imaging of epilepsy: State of the art at 1.5T and potential of 3T. Epilep Dis 5: 3–20.

Buchli R, Duc CO, Martin E, Boesiger P. 1994. Assessment of absolute metabolite concentrations in human tissue by 31P MRS in vivo. Part I: Cerebrum, cerebellum, cerebral gray and white matter. Mag Reson Med 32: 447–452. Burtscher IM, Holtas S. 2001. Proton magnetic resonance spectroscopy in brain tumours: clinical applications. Neuroradiology 43: 345–352. Capizzano AA, Vermathen P, Laxer KD, Matson GB, Maudsley AA, Soher BJ, et al. 2002. Multisection proton MR spectroscopy for mesial temporal lobe epilepsy. AJNR Am J Neuroradiol 23: 1359–1368. Cascino GD, Andermann F, Berkovic SF, Kauzniecky RI, Sharbrough FW, Keene DL, et al. 1993. Gelastic seizures and hypothalamic hamartomas: evaluation of patients undergoing chronic intracranial EEG monitoring and outcome of surgical treatment. Neurology 43: 747–750. Cendes F, Caramanos Z, Andermann F. 1997a. Proton magnetic resonance spectroscopic imaging and magnetic resonance imaging volumetry in the lateralization of temporal lobe epilepsy: a series of 100 patients. Ann Neurol 42: 737–746. Cendes F, Stanley JA, Dubeau F, Andermann F, Arnold DL. 1997b. Proton magnetic resonance spectroscopic imaging for discrimination of absence and complex partial seizures. Ann Neurol 41: 74–81. Chu WJ, Hetherington HP, Kuzniecky RI, Simor T, Mason GF, Elgavish GA. 1998. Lateralization of human temporal lobe epilepsy by 31P NMR spectroscopic imaging at 4.1 T. Neurology 51: 472–479. Connelly A, Jackson GD, Duncan JS, King MD, Gadian DG. 1994. Magnetic resonance spectroscopy in temporal lobe epilepsy. Neurology 44: 1411–1417. Connelly A, van Paesschen W, Proter DA, Johnson CL, Duncan JS, Gadian DG. 1998. Proton magnetic resonance spectroscopy in MRI-negative temporal lobe epilepsy. Neurology 51: 61–66. Cross JH, Connelly A, Jackson GD. 1996. Proton magnetic resonance spectroscopy in children with temporal lobe epilepsy. Ann Neurol 39: 107–113. Cross JH, Jackson GD, Neville BGR, Connelly A, Kirkham FJ, Boyd SG, et al. 1993. Early detection of abnormalities in partial epilepsy using magnetic resonance. Arch Dis Childhood 69: 104–109. Danielsen ER, Ross B. 1999. The clinical significance of metabolites. In Magnetic Resonance Spectroscopy of Neurological Diseases (Eds. Danielsen ER, Ross B), Marcel Dekker Inc, New York, pp. 23–42. de Graaf R, editor. 1998. In Vivo NMR Spectroscopy. John Wiley & Sons, New York. Dreher W, Leibfritz D. 2002. Fast proton spectroscopic imaging with high signal-to-noise ratio: Spectroscopic RARE. Magn Reson Med 47: 523–528.

503

504

Regula S. Briellmann, R. Mark Wellard and Graeme D. Jackson

Eberhardt KE, Stefan H, Buchfelder M, Pauli E, Hopp P, Huk W, et al. 2000. The significance of bilateral CSI changes for the postoperative outcome in temporal lobe epilepsy. J Comput Assist Tomogr 24: 919–926. Eleff SM, Schnall MD, Ligetti L, Osbakken M, Subramanian VH, Chance B, et al. 1988. Concurrent measurements of cerebral blood flow, sodium, lactate, and high-energy phosphate metabolism using 19F, 23Na, 1H, and 31P nuclear magnetic resonance spectroscopy. Magn Reson Med 7: 412–424. Ende GR, Laxer KD, Knowlton RC, Matson GB, Schuff N, Fein G, Weiner MW. 1997. Temporal lobe epilepsy: bilateral hippocampal metabolite changes revealed at proton MR spectroscopic imaging. Radiology 202: 809–817. Engel JJ. 1987. Outcome with respect to seizures. In Surgical Treatment of the Epilepsies (Ed. Engel JJ). Raven Press, New York, pp. 553–571. Engel JJ. 1989. Seizures and epilepsy. In Seizures and epilepsy (Ed. Engel JJ). FA Davies, Philadelphia, pp. 112–134. Gadian DG. 1995. NMR and its Application to Living Systems. Oxford University Press, New York. Gadian DG, Radda GK, Dawson MJ, Wilke DR. 1982. pH Measurements of Cardiac and Skeletal Muscle Using 31P NMR. Liss Inc, New York, pp. 65–77. Graham D, Lantos P. 1997. Greenfield’s neuropathology. In Graham D and Lantos P, editors. Vol. I. Arnold, London, pp. 936ff. Gullans SR, Verbalis JG. 1993. Control of brain volume during hyperosmolar and hypoosmolar conditions. Annu Rev Med 44: 289–301. Hetherington HP, Pan JW, Chu WJ, Mason GF, Newcomer BR. 1997. Biological and clinical MRS at ultra-high field. NMR In Biomedicine 10: 360–371. Hetherington HP, Pan JW, Spencer DD. 2002. 1H and 31P spectroscopy and bioenergetics in the lateralization of seizures in temporal lobe epilepsy. J Magn Reson Imaging 16: 477–483. Hill RA, Chiappa KH, Huang-Hellinger F, Jenkins BG. 1999. Hemodynamic and metabolic aspects of photosensitive epilepsy revealed by functional magnetic resonance imaging and magnetic resonance spectroscopy. Epilepsia 40: 912–920. Holopainen IE, Valtonen ME, Komu ME, Sonninen PH, Manner TE, Lundbom NM, et al. 1998. Proton spectroscopy in children with epilepsy and febrile convulsions. Pediatr Neurol 19: 93–99. Hsu YY, Chang C, Chang CN, Chu NS, Lim KE, Hsu JC. 1999. Proton MR spectroscopy in patients with complex partial seizures: single-voxel spectroscopy versus chemical-shift imaging. AJNR Am J Neuroradiol 20: 643–651. Hugg JW, Matson GB, Twieg DB, Maudsley AA, Sappey-Marinier D, Weiner MW. 1992. Phosphorus-31 MR

spectroscopic imaging (MRSI) of normal and pathological human brains. Mag Res Imag 10: 227–243. International League Against Epilepsy, Commission on Classification and Terminology. 1989. Proposal for revised classification of epilepsies and epileptic syndromes. Epilepsia 30: 389–399. Jack CR. 1994. MRI-based hippocampal volume measurements in epilepsy. Epilepsia 35: S21–S29. Jackson GD, Berkovic SF, Duncan JS, Connelly A. 1993. Optimizing the diagnosis of hippocampal sclerosis using magnetic resonance imaging. AJNR Am J Neuroradiol 14: 753–762. Jackson GD, Berkovic SF, Tress BM, Kalnins RM, Fabinyi GCA, Bladin PF. 1990. Hippocampal sclerosis can be reliably detected by magnetic resonance imaging. Neurology 40: 1869–1875. Kälviäinen R, Partanen K, Aeikiä M, Mervaala E, Vainio P, Riekkinen PJ, et al. 1997. MRI-based hippocampal volumetry and T2 relaxometry: correlation to verbal memory performance in newly diagnosed epilepsy patients with left-sided temporal lobe focus. Neurology 48: 286–287. Kikuchi S, Kubota F, Akata T, Shibata N, Hattori S, Oya N, et al. 2000. A study of the relationship between the seizure focus and 1H-MRS in temporal lobe epilepsy and frontal lobe epilepsy. Psychiatry Clin Neurosci 54: 455–459. Kuzniecky R. 1999. Magnetic resonance spectroscopy in focal epilepsy: 31P and 1H spectroscopy. Rev Neurol (Paris) 155: 495–498. Kuzniecky R, Burgard S, Faught E, Morawetz R, Bartolucci A. 1993. Predictive value of magnetic resonance imaging in temporal lobe epilepsy surgery. Arch Neurol 50: 65–69. Kuzniecky R, Elgavish GA, Hetherington HP, Evanochko WT, Pohost GM. 1992. In vivo 31P nuclear magnetic resonance spectroscopy of human temporal lobe epilepsy. Neurology 42: 1586–1590. Kuzniecky R, Hetherington H, Pan J. 1997. Proton spectroscopic imaging at 4.1 tesla in patients with malformations of cortical development and epilepsy. Neurology 48: 1018–1024. Kuzniecky R, Hugg J, Hetherington H, Martin R, Faught E, Morawetz R, et al. 1999. Predictive value of 1H MRSI for outcome in temporal lobectomy. Neurology 53: 694–698. Kuzniecky R, Hugg JW, Hetherington H. 1998. Relative utility of 1H spectroscopic imaging and hippocampal volumetry in the lateralization of mesial temporal lobe epilepsy. Neurology 51: 66–71. Kuzniecky RI, Jackson GD. 1995. Magnetic resonance spectroscopy in epilepsy. In Magnetic Resonance in Epilepsy (Eds. Kuzniecky RI, Jackson GD), Raven Press, New York, pp. 289–314.

MR spectroscopy in seizure disorders

Laxer KD, Hubesch B, Sappey-Marinier D, Weiner MW. 1992. Increased pH and inorganic phosphate in temporal seizure foci demonstrated by [31P]MRS. Epilepsia 33: 618–623. Lerman P. 1997. Benign childhhod epilepsy with centro temporal spikes. Epilepsy – a Comprehensive Textbook. Vol. III. Lippincott-Raven, Philadelphia, pp. 2307–2314. Li BS, Regal J, Gonen O. 2001. SNR versus resolution in 3D 1H MRS of the human brain at high magnetic fields. Magn Reson Med 46(6): 1049–1053. Li LM, Dubeau F, Andermann F, Arnold DL. 2000. Proton magnetic resonance spectroscopic imaging studies in patients with newly diagnosed partial epilepsy. Epilepsia 41: 825–831. Margerison JH, Corsellis JAN. 1966. Epilepsy and the temporal lobes. Brain 89: 499–530. Maton B, Gilliam F, Sawrie S, Faught E, Hugg J, Kuzniecky R. 2001. Correlation of scalp EEG and 1H-MRS metabolic abnormalities in temporal lobe epilepsy. Epilepsia 42: 417–422. Matthews PM, Andermann F, Arnold DL. 1990. A proton magnetic resonance spectroscopy study of focal epilepsy in humans. Neurology 40: 985–989. Maudsley AA, Lin E, Weiner MW. 1992. Spectroscopic imaging display and analysis. Magn Reson Imaging 10: 471–485. McLean MA, Woermann FG, Barker GJ, Duncan JS. 2000. Quantitative analysis of short echo time 1H-MRSI of cerebral gray and white matter. Magn Reson Med 44: 401–411. Meiners LC, van der Grond J, van Rijen PC, Springorum R, de Kort GAP, Jansen GH. 2000. Proton magnetic resonance spectroscopy of temporal lobe white matter in patients with histologically proven hippocampal sclerosis. J Magn Res Imag 11: 25–31. Miller SP, Li LM, Cendes F, Tasch E, Andermann F, Dubeau F, et al. 2000. Medial temporal lobe neuronal damage in temporal and extratemporal lesional epilepsy. Neurology 54: 1465–1470. Moller-Hartmann W, Herminghaus S, Krings T. 2002. Clinical application of proton magnetic resonance spectroscopy in the diagnosis of intracranial mass lesions. Neuroradiology 44: 371–381. Mueller SG, Suhy J, Laxer KD, Flenniken DL, Axelrad J, Capizzano AA, et al. 2002. Reduced extrahippocampal NAA in mesial temporal lobe epilepsy. Epilepsia 43: 1210–1216. Najm IM, Wang Y, Shedid D, Lueders HO, Ng TC, Comair YG. 1998. MRS metabolic markers of seizures and seizureinduced neuronal damage. Epilepsia 39: 244–250. Naressi A, Couturier C, Castang I, de Beer R, GraveronDemilly D. 2001. Java-based graphical user interface for MRUI, a software package for quantitation of in vivo/ medical magnetic resonance spectroscopy signals. Comput Biol Med 31: 269–286.

Nonaka M, Kohmura E, Yamashita T. 1999. Kainic acidinduced seizure upregulates Na()/myo-inositol cotransporter mRNA in rat brain. Mol Brain Res 70: 179–186. Park SA, Kim GS, Lee SK, Lim SR, Heo K, Park SC, et al. 2002. Interictal epileptiform discharges relate to 1H-MRSdetected metabolic abnormalities in mesial temporal lobe epilepsy. Epilepsia 43: 1385–1389. Patel AB, Rothman DL, Cline GW. 2001. Glutamine is the major precursor for GABA synthesis in rat neocortex in vivo following acute GABA-transaminase inhibition. Brain Res 919: 207–220. Pauli E, Eberhardt WE, Schafer I, Tomandl B, Huk WJ, Stefan H. 2000. Chemical shift imaging spectroscopy and memory function in temporal lobe epilepsy. Epilepsia 41: 282–289. Petroff OA, Errante LD, Rothman DL. 2002. Glutamateglutamine cycling in the epileptic human hippocampus. Epilepsia 43: 703–710. Petroff OA, Novotny EJ, Avison M, Rothman DL, Alger JR, Ogino T, et al. 1992. Cerebral lactate turnover after electroshock: in vivo measurements by 1H/13C magnetic resonance spectroscopy. J Cereb Blood Flow Metab 12: 1022–1029. Petroff OA, Prichard JW, Behar KL, Alger JR, Shulman RG. 1984. In vivo phosphorus nuclear magnetic resonance spectroscopy in status epilepticus. Ann Neurol 16: 169–177. Petroff OAC, Rothman DL, Behar KL, Mattson RH. 1995. Initial observations on effect of vigabatrin on in vivo 1H spectroscopic measurement of g-aminobutric acid, glutamate, and glutamine in human brain. Epilepsia 36: 457–464. Prichard JW, Alger JR, Behar KL, Petroff OA, Shulman RG. 1983. Cerebral metabolic studies in vivo by 31P NMR. Proc Natl Acad Sci USA 80: 2748–2751. Prichard JW, Petroff OA, Ogino T, Shulman RG. 1987. Cerebral lactate elevation by electroshock: a 1H magnetic resonance. Ann NY Acad Sci 508: 54–63. Provencher SW. 1993. Estimation of metabolite concentrations from localized in vivo proton NMR spectra. Magn Reson Med 30: 672–679. Quigg M, E.H. B, Jackson T, Laws E. 1997. Volumetric magnetic resonance imaging evidence of bilateral hippocampal atrophy in mesial temporal lobe epilepsy. Epilepsia 38: 588–594. Rabinowicz AL, Salas E, Beserra F, Leiguarda RC, Vazquez SE. 1997. Changes in regional cerebral blood flow beyond the temporal lobe in unilateral temporal lobe epilepsy. Epilepsia 38: 1011–1014. Rayboud C, Guye M, Mancini J, Girard N. 2001. Neuroimaging of epilepsy in children. Magn Reson Imaging Clin N Am 9: 121–147. Roll P, Szepetowski P. 2002. Epilepsy and ionic channels. Epilep Dis 4: 165–172.

505

506

Regula S. Briellmann, R. Mark Wellard and Graeme D. Jackson

Ruggieri PM, Najm IM. 2001. MR imaging in epilepsy. Neurol Clin 19: 477–489. Sappey-Marinier D, Calabrese G, Fein G. Hugg JW, Biggins C, Weiner MW. 1992. Effect of photic stimulation on human visual cortex lactate and phosphates using 1H and 31P magnetic resonance spectroscopy. J Cereb Blood Flow Metab 12: 584–592. Savic I, Lekvall A, Greitz D, Helms G. 2000. MR spectroscopy shows reduced frontal lobe concentrations of N-acetyl aspartate in patients with juvenile myoclonic epilepsy. Epilepsia 41: 290–296. Scheffer IE, Berkovic SF. 1997. Generalized epilepsy with febrile seizures plus: a genetic disorder with heterogeneous clinical phenotypes. Brain 120: 479–490. Scheffer IE, Bhatia KP, Lopes-Cendes I, Fish DR, Marsden CD, Andermann F, et al. 1994. Autosomal dominant frontal epilepsy misdiagnosed as sleep disorder. Lancet 343: 515–517. Schnall MD, Yoshizaki K, Chance B, Leigh JS Jr. 1988. Triple nuclear NMR studies of cerebral metabolism during generalized seizure. Magn Reson Med 6: 15–23. Serles W, Li LM, Antel SB, Cendes F, Gotman J, Olivier A, et al. 2001. Time course of postoperative recovery of N-AcetylAspartate in temporal lobe epilepsy. Epilepsia 42: 190–197. Shen J, Petersen KF, Behar KL. 1999. Determination of the rate of the glutamate/glutamine cycle in the human brain by in vivo 13C NMR. Proc Natl Acad Sci USA 96: 8235–8240. Stanley JA, Cendes F, Dubeau F, Andermann F, Arnold DL. 1998. Proton magnetic resonance spectroscopic imaging in patients with extratemporal epilepsy. Epilepsia 39: 267–273. Steinlein OK, Mulley JC, Propping P, Wallace RH, Phillips HA, Sutherland GR, et al. 1995. A missense mutation in the neuronal nicotinic acetylcholine receptor a4 subunit is associated with autosomal dominant nocturnal frontal lobe epilepsy. Nature Genet 11: 201–203. Strauss WL, Layton ME, Hayes CE, Dager SR. 1997. 19F magnetic resonance spectroscopy investigation in vivo of acute and steady-state brain fluvoxamine levels in obsessivecompulsive disorder. Am J Psychiatry 154: 516–522. Suhy J, Laxer KD, Capizzano AA, Vermathen P, Matson GB, Barbaro NM, et al. 2002. H MRSI predicts surgical outcome in MRI-negative temporal lobe epilepsy. Neurology 58: 821–823.

Tasch E, Cendes F, Li LM, Dubeau F, Andermann F, Arnold DL. 1999. Neuroimaging evidence of progressive neuronal loss and dysfunction in temporal lobe epilepsy. Ann Neurol 45: 568–576. Urenjak J, Williams SR, Gadian DG, Noble M. 1993. Proton nuclear magnetic resonance spectroscopy unambiguously identifies different neural cell types. J Neurosci 13: 981–989. van der Grond J, Gerson JR, Laxer KD, Hugg JW, Matson GB, Weiner MW. 1998. Regional distribution of interictal 31P metabolic changes in patients with temporal lobe epilepsy. Epilepsia 39: 527–536. Wallace RH, Marini C, Petrou S, Harkin LA, Bowser DN, Panchal RG, et al. 2001. Mutant GABAA receptor 2-subunit in childhood absence epilepsy and febrile seizures. Nature Genetics 28: 49–52. Wallace RH, Wang DW, Singh R, Scheffer IE, George AL, Phillips HA, et al. 1998. Febrile seizures and generalized epilepsy associated with a mutation in the Na-channel B1 subunit gene SCN1B. Nat Gen 19: 366–370. Weber OM, Verhagen A, Duc CO, Meier D, Leenders KL, Boesiger P. 1999. Effects of vigabatrin intake on brain GABA activity as monitored by spectrally edited magnetic resonance spectroscopy and positron emission tomography. Mag Res Imag 17: 417–425. Wellard RM, Briellmann RS, Prichard JW, Syngeniotis A, Jackson GD. 2003. Myoinositol abnormalities in temporal lobe epilepsy. Epilepsia 44: 815–821. Woermann FG, McLean MA, Bartlett PA, Parker GJ, Berker GJ, Duncan JS. 1999. Short echo time single-voxel 1H magnetic resonance spectroscopy in magentic resonance imagingnegative temporal lobe epilepsy: different biochemical profile compared with hippocampal sclerosis. Ann Neurol 45: 369–376. Wright NB. 2001. Imaging in epilepsy: a paediatric perspective. Br J Radiol 74: 575–589. Young RS, Chen B, Petroff OA, Gore JC, Cowan BE, Novotny EJ, Jr., et al. 1989. The effect of diazepam on neonatal seizure: in vivo 31P and 1H NMR study. Pediatr Res 25: 27–31. Younkin DP, Delivoria-Papadopoulos M, Maris J, Donlon E, Clancy R, Chance B. 1986. Cerebral metabolic effects of neonatal seizures measured with in vivo 31P NMR spectroscopy. Ann Neurol 20: 513–519.

MR spectroscopy in seizure disorders

Case Study 29.1 Rasmussen’s encephalitis: MRSI Peter Barker, D.Phil and Steven Breiter M.D., Johns Hopkins University, School of Medicine, Baltimore, MD, USA History 3-year-old female with 6 month history of focal left-sided seizures.

T T22

Cho

NAA NAA

Lac

Technique Conventional MRI and multi-slice MRSI (TE 280 ms).

Imaging findings T2 MRI shows right hemiatrophy. MRSI shows a hemispheric increase in Cho (particularly in the right frontal WM) and decrease in NAA.

Discussion Rasmussen’s encephalitis is a progressive childhood disease of unknown etiology, with intractable seizures, progressive hemiparesis and mental retardation. Usually the only effective means of seizure control is hemispherectomy (Kossoff, 2003). MRSI usually suggests hemispheric involvement with low NAA (neuronal loss or dysfunction) and high Cho (believed to be due to microglial proliferation) (Matthews, 1990). NAA Cho

Key points Rasmussen’s encephalitis shows low NAA and high Cho.

NAA Cr Cr Cho

Patterns of involvement may be hemispheric, or predominantly insular/frontal lobe. Lac may be present during seizures. ppm

3.0

2.0

1.0

ppm

3.0

2.0

References Kossoff EH, Vining EP, Pillas DJ, Pyzik PL, Avellino AM, Carson BS, Freeman JM. 2003. Hemispherectomy for intractable unihemispheric epilepsy etiology vs outcome. Neurology 61(7): 887–890. Matthews PM, Andermann F, Arnold DL. 1990. A proton magnetic resonance spectroscopy study of focal epilepsy in humans. Neurology 40(6): 985–989.

1.0

507

508

Regula S. Briellmann, R. Mark Wellard and Graeme D. Jackson

Case Study 29.2 Temporal lobe epilepsy: MRSI Peter Barker, D.Phil, Johns Hopkins University School of Medicine, Baltimore, MD, USA History 22-year-old female with focal, left sided seizures and right sphenoidal EEG electrodes.

Coronal T2

Coronal T1 (SPGR)

T1 Localizer for MRSI

Cho

Technique Conventional MRI and proton MRSI (TE 280 ms).

Imaging findings Conventional MR, including axial and coronal T1 and T2, was normal, with no hippocampal asymmetry or signal change. MRSI showed an elevated Cho signal in the right mesial temporal lobe (arrow), as well as slightly lower NAA.

Discussion The patient underwent a right anterior temporal lobectomy and has been seizure free. The case illustrates an example where MRSI provided lateralizing information while MRI was normal (Connolley, 1998). The case is unusual since the most common spectroscopic finding in mesial temporal sclerosis is reduced NAA, rather than increased Cho (Cendes, 1997).

NAA

Key points MRSI may provide lateralizing information in temporal lobe epilepsy, sometimes even when MRI is normal. Metabolic changes in seizure foci are usually low NAA, or (ictally) increased lac.

References Cendes F, Caramanos Z, Andermann F, Dubeau F, Arnold DL. 1997. Proton magnetic resonance spectroscopic imaging and magnetic resonance imaging volumetry in the lateralization of temporal lobe epilepsy: a series of 100 patients. Ann Neurol 42(5): 737–746. Connelly A, Van Paesschen W, Porter DA, Johnson CL, Duncan JS, Gadian DG. 1998. Proton magnetic resonance spectroscopy in MRI-negative temporal lobe epilepsy. Neurology 51(1): 61–66.

30

Diffusion and perfusion MR imaging in seizure disorders Konstantinos Arfanakis1 and Bruce P. Hermann2 1

Department of Biomedical Engineering, Illinois Institute of Technology, Chicago, IL, USA Department of Neurology, University of Wisconsin, Madison, WI, USA

2

Key points • Perfusion MR imaging (MRI) offers higher spatial resolution than radionuclide techniques such as positron emission tomography and single-photon emission computed tomography. • Ictal perfusion studies generally show increased perfusion around the seizure focus, but sometimes more widespread. • Interictally perfusion is decreased, sometimes well-localised to the seizure focus. • Experimental seizure models show decreases in cortical apparent diffusion coefficient (ADC), usually transient. • Human ictal diffusion weighted imaging studies show decreased cortical ADC, but increased ADC in adjacent white matter normalisation may be slow. • Interictal human studies show increased ADC and decreased anisotropy on diffusion tensor imaging (DTI). • Ictal ADC decrease may be caused by cell swelling due to osmotic changes under anaerobic conditions. • Chronic interictal ADC/DTI changes are attributed to increased extracellular space due to neuronal loss and gliosis.

Introduction Conventional, anatomical MR Imaging (MRI) has been widely used for the detection of brain tissue

volume changes caused by chronic seizures, and for diagnosis of brain lesions that result in seizure activity. However, seizures are often not associated with lesions or volume changes visible in conventional MRI. In contrast, diffusion and perfusion MRI are sensitive to the physiological changes that take place in brain tissue ictally and interictally. In this chapter, we provide an in-depth discussion of the application of both diffusion and perfusion MRI in seizure disorders. We describe the changes in perfusion (cerebral blood flow (CBF), cerebral blood volume (CBV)) and diffusion (apparent diffusion coefficient (ADC), fractional anisotropy (FA)) that occur in the epileptogenic regions, or globally in the brain, in the ictal and interictal phase. Also, we evaluate diffusion and perfusion MRI as methods to localize seizure focus. Finally, we discuss the mechanisms that may be responsible for the ictal and interictal changes in perfusion and diffusion MRI.

Ictal and interictal perfusion MRI Positron emission tomography (PET) (Engel, 1984; Franck et al., 1986) and single-photon emission computed tomography (SPECT) (Magistretti et al., 1982; Lee et al., 1986; Marks et al., 1992) have been used to identify focal changes in regional CBF in patients with epilepsy. However, the low spatial resolution of PET and SPECT, and the ionizing radiation emitted from the nuclear medicine tracers are major concerns. MR perfusion techniques have also been developed and offer higher spatial resolution without the use of ionizing radiation (Rosen et al., 1990; 509

510

Konstantinos Arfanakis and Bruce P. Hermann

Belliveau et al., 1990; Detre et al., 1992). MR perfusion has been applied in several studies of cerebral ischemia (Warach et al., 1996; Siewert et al., 1997), brain tumors (Aronen et al., 1994; Bruening et al., 1996), and functional brain mapping (Belliveau et al., 1991; Liu and Gao, 1999). MR perfusion techniques are based on exogenous or endogenous tracers. In the method based on exogenous tracers, a paramagnetic agent such as gadolinium dimeglumine gadopentetate (Gd-DTPA) is injected, and the resulting decrease and subsequent recovery of the MR signal is used to estimate perfusion (Østergaard et al., 1996a, 1996b; Liu et al., 1999) (cf. Chapter 7). In the method using endogenous tracers, the magnetization of the spins of arterial water are noninvasively labeled using radiofrequency (RF) pulses, and the regional accumulation of the label is measured in the tissues by comparison with an image acquired without labeling (Edelman et al., 1994; Kim, 1995; Kwong et al., 1995) (cf. Chapter 8). Perfusion MRI in the ictal period Perfusion MR has also been applied in seizure disorders. Hyperperfusion has been demonstrated during the ictal period. In an early study, using a type of arterial spin tagging technique, ictal perfusion MRI showed increased cerebral perfusion in the left frontal lobe of a patient diagnosed with dysphasia, spastic right hemiparesis, and right homonymous hemianopsia (Fish et al., 1988). The surface electroencephalogram (EEG) was characterized by diffuse left fronto-temporal abnormalities. In another case, using contrast-enhanced MRI, increased CBV was demonstrated in the right temporoparietal region of a patient during seizures (Warach et al., 1994). Since, CBF equals the ratio of CBV to mean transit time (MTT), and MTT was the same for the regions under study and the controlateral control regions, it was deduced that CBF was increased in the right temporoparietal region. This patient was presented with focal status epilepticus, characterized by left visual field hemianopsia and visual hallucinations with left body convulsions. EEG showed epileptic discharge in the right temporal region, and SPECT also demonstrated marked hyperperfusion of the right temporo-parietooccipital area. In an ictal patient

with unilateral clonic status epilepsy, increased relative CBV (rCBV) was detected in the vicinity of lesions apparent in T2-weighted images, using contrast-enhanced perfusion MRI (Wu et al., 1999). In a patient with acute onset of confusion and left-sided hemiparesis, contrast-enhanced perfusion MRI showed ictal elevation of CBV and CBF in the right hemisphere (El-Koussy et al., 2002). For the same patient, EEG also revealed focal “slow” activity in the right hemisphere. Contrast-enhanced perfusion MRI on an ictal subject with paraphasia and right hemiparesis showed an increase of CBV in the left temporoparietal cortex (Flacke et al., 2000). The same region was identified by EEG as the source of the epileptic discharges. Therefore, based on the studies mentioned above, as well as PET and SPECT research (Magistretti et al., 1982; Engel, 1984; Franck et al., 1986; Lee et al., 1986; Marks et al., 1992), seizure activity may lead to an increase of perfusion in the ictal phase. This increase may also be measured with perfusion MRI. Furthermore, the areas with increased perfusion, shown on MRI, were found to be in reasonably good agreement with the regions thought to be epileptogenic, based on clinical symptoms and other clinical exams (EEG, PET, SPECT). While these results are promising for the use of ictal perfusion MRI to identify the location of seizure foci, it should also be recognized that it is extremely challenging in practice to perform ictal MR studies in patients. It is probably only possible to perform ictal MRI in patients who are in status epilepticus or having seizures very frequently, and even in these cases there are concerns about head motion and potential safety issues. Perfusion MRI in the interictal period In the interictal period, several perfusion MRI studies have shown hypoperfusion (Figure 30.1). Interictal continuous arterial spin-labeling perfusion MRI on twelve temporal lobe epilepsy (TLE) patients and twelve normal controls revealed significantly reduced global CBF in the patients compared to the controls (Wolf et al., 2001). Also, although it was shown that mesial temporal CBF was asymmetric in the normal controls, with the left side being dominant (5% higher CBF), this asymmetry was more pronounced

Diffusion and perfusion MR imaging in seizure disorders

L

R

Fig. 30.1 This figure shows a case of interictal hypoperfusion in the right temporal lobe of a patient with TLE. The image on the left was acquired with PET and demonstrates reduced concentration of H215O in the right temporal lobe. The image on the right was obtained with dynamic susceptibility contrast (DSC) MR and shows reduced relative CBF in the same regions. Both images correspond to the same slice. The MR image is distorted due to local changes in magnetic susceptibility. (This figure is a kind contribution of Dr. HLA. Liu).

in the TLE patients. The ipsilateral mesial temporal CBF of the patients was significantly decreased compared to the contralateral mesial temporal CBF (ipsilateral being the side of the clinically diagnosed epileptogenic focus). The asymmetry in mesial temporal CBF, measured with perfusion MRI, was correlated with an asymmetry in metabolism in the same regions (r  0.79, P  0.004) measured with flouro2-deoxyglucose (FDG) PET (FDG-PET), an asymmetry in hippocampal volume (r  0.61, P  0.05), and clinical laterality (r  0.81, P  0.001). In 11 out of 12 patients the clinically diagnosed seizure side coincided with the side characterized by the lowest mesial temporal CBF. However, one concern that was expressed by the authors in this study was the fact that global CBF reduction in the patients might be a result of antiseizure medications (Theodore et al., 1989; Gaillard et al., 1996). On the other hand, even if medications had an effect on global CBF it should be similar for both hemispheres. Therefore, antiseizure treatment should not be a reason for the increased CBF asymmetry in patients. A second potential confounding factor was the fact that decreased mesial temporal CBF might be due to mesial temporal atrophy. However, the authors respond to this concern by pointing out that in several of their patients the data

suggests that laterality shown by volumetric MRI was opposite to asymmetry in perfusion and metabolism. Therefore, they concluded that structure and function appeared to be dissociated. In a different study on nine interictal TLE patients, using contrastenhanced MRI perfusion, hippocampal rCBV was reduced in the epileptogenic side (Wu et al., 1999). Hippocampal volume and metabolism, measured with PET, was also reduced in the same side. In a study introducing a novel arterial spin-labeling MRI method for perfusion measurements in areas of high magnetic susceptibility, interictal hypoperfusion was shown in TLE patients (Liu et al., 2001). The temporal lobes with reduced perfusion on MR studies were also characterized by reduced perfusion on PET imaging. In a paper discussing localization of epileptogenic foci in drug-resistant epilepsy, lesions suspected on SPECT of being epileptogenic showed mild hypoperfusion interictally, with contrastenhanced perfusion MRI (Heiniger et al., 2002) (Figure 30.2). Therefore, interictal hypoperfusion has been demonstrated in several studies with perfusion MRI, as well as PET and SPECT. In addition, the areas with hypoperfusion, shown on MRI, were found to be in reasonably good agreement with the regions thought to contain the seizure focus, based on

511

512

Konstantinos Arfanakis and Bruce P. Hermann

Perfusion MRI SPECT

CBV

ADC

MTT

DWI

Diffusion MRI

Fig. 30.2 Ictal SPECT in an axial slice of a patient shows areas of hyperperfusion. In contrast, in the same patient, interictal perfusion MRI shows regions of hypoperfusion. Furthermore, interictal diffusion MRI demonstrates regions of increased diffusivity in the left temporal lobe. (This figure is a kind contribution of Dr. KO. Lovblad).

clinical symptoms and other clinical exams (EEG, PET, SPECT). Finally, in most of the studies that were summarized above, both for the ictal and interictal conditions, perfusion MRI abnormalities were detected even in patients with no apparent lesions in conventional MRI. This demonstrates the sensitivity of perfusion MRI techniques to the physiological changes that take place in brain tissue ictally and interictally.

Ictal and interictal diffusion MRI Diffusion MR, is a non-invasive technique that can be used to probe in vivo the diffusion properties of water molecules in tissue (Stejskal and Tanner, 1964). One implementation of this method, diffusion weighted imaging (DWI), involves acquisition of diffusion weighted (DW) signals in three orthogonal directions (x, y, z) (Le Bihan, 1991). The diffusion information from the three axes is then combined to estimate the ADC, which is proportional to the mean diffusivity. DWI has been used extensively in

studies of stroke and cerebral ischemia, where, in the acute stage, ADC values of ischemic tissue decrease, days later return to normal, and in the chronic condition ADC values become higher than normal (Moseley et al., 1990; Kucharczyk et al., 1991; Busza et al., 1992; Warach et al., 1992; van Gelderen et al., 1994). A more recent implementation of diffusion MRI is diffusion tensor imaging (DTI) (Basser et al., 1994a, 1994b). Unlike DWI, where three DW images are used to calculate ADC, DTI requires diffusion measurements in at least six non-colinear orientations, and characterizes diffusive transport of water by an effective diffusion tensor D. This symmetric 3 3 matrix D is of great importance because not only does it allow estimation of ADC, but it also contains useful structural information about the tissue. The eigenvalues of D are the three principal diffusivities and the eigenvectors define the corresponding diffusional directions, which in the brain are thought to reflect the directions of the local fiber tracts (Basser et al., 1994b). Moreover, one can derive from D rotationally invariant scalar quantities that describe the intrinsic diffusion properties of the tissue. The most

Diffusion and perfusion MR imaging in seizure disorders

commonly used are: the trace of the tensor, which measures mean diffusivity and is equal to three times the ADC, and FA, which characterizes the anisotropy of the fiber structure (Basser et al., 1994b; Basser, 1995; Basser and Pierpaoli, 1996; Pierpaoli et al., 1996). FA is high when diffusion is more restricted in particular directions compared to others (e.g. diffusion is more restricted in directions perpendicular to white matter (WM) axons than parallel to them, due to the effect of axonal membranes, myelin, microtubules, etc.). In contrast, if diffusion occurs almost equally in any direction in 3-dimensional space, then the FA value tends to 0. DWI is equivalent to DTI when all the off-diagonal elements of D are equal to 0. However, this is not always valid and therefore FA cannot be estimated correctly with DWI. DTI has been applied in several disease conditions such as cerebral ischemia (de Crespigny et al., 1995; Lythgoe et al., 1997), acute stroke (Warach et al., 1995), multiple sclerosis (MS) (Bammer et al., 2000), traumatic brain injury (TBI) (Arfanakis et al., 2002a), and schizophrenia (Lim et al., 1999). Diffusion MRI (DWI or DTI) has also been applied in studies of seizure disorders. In the next sections, we discuss the effects of seizures on the diffusion properties of brain tissue, during the ictal and interictal period. Diffusion MRI in experimentally induced seizures Several different seizure models have been used in order to study the effects of seizure activity on the diffusion properties of brain tissue. Although there are differences between studies as to the timing of the changes in diffusion, and in some cases the type of changes, most of the diffusion MRI studies in experimentally induced seizures show an initial decrease of ADC, shortly after the onset of the seizures, followed by, in some cases, a return of ADC to normal values. In a DWI study on rats during status epilepticus induced by administration of bicuculline, ADC continuously decreased, starting immediately after seizure onset (Zhong et al.,1993). The maximum ADC reduction was 18%, and occurred at 40 min after status epilepticus was induced. In DWI studies on rats exhibiting prolonged complex partial seizures caused by systemic administration of kainic acid, ADC was reduced in the cortex, by 54%, and amygdala, by 36%,

24 h after the injection of kainic acid (Righini et al., 1994; Nakasu et al., 1995; Ebisu et al., 1996). The ADC values in the same regions returned to normal 9 days after the seizures were induced (Righini et al., 1994). In a rodent model of TLE, where kainic acid was injected in the left posterior hippocampus, only the ADC of the ipsilateral hippocampus was increased 14 days after seizure onset (Tokumitsu et al., 1997). In another epilepsy model, where seizures were induced in rats by administration of fluorothyl, ADC reduction started as soon as the exposure to fluorothyl (Zhong et al., 1995). The magnitude of the ADC reduction was well correlated with the duration of fluorothyl exposure. For rats with the longest fluorothyl exposure time, the reduction in ADC was 17% compared to the preseizure condition. When pentobarbital was injected in some of the rats in order to arrest seizure activity, ADC started to increase, returning to normal values. In a rodent model of status epilepticus induced by pilocarpine, ADC values were reduced by 48% in the cortex, 33% in the amygdala, and 37% in the retrosplenial cortex, 12 h after seizure onset (Wall et al., 2000). At 24 h, the ADC values from the cortex and amygdala slightly increased, but remained well below normal values, while the hippocampal ADC became 19% higher than the normal values. In a DWI experiment on soman-mediated seizures in rats, ADC significantly decreased in the hippocampus, by 33%, thalamus, by 31%, and retrosplenial cortex, by 20%, within the first 12 h after Soman treatment (Bhagat et al., 2001). At 24 h ADC values returned to normal, and 7 days later ADC values once again declined. However, in all these studies of diffusion MRI in experimentally induced seizures it is not possible to separate the effects of the convulsant drugs from those of seizure activity on the diffusion properties of brain tissue. For this purpose, an electroshock model was also used in rats to simulate seizure activity (Zhong et al., 1997). After a single shock train of 0.1 s in duration, consisting of 10 pulses of 150 V, ADC values were reduced in the regions affected by the shock (Figure 30.3). The mean maximum reduction for all rats was 4%. Approximately 3.15 min after the shock the ADC values tended to normalize to the pre-shock values. For longer shock train lengths, ADC was reduced more than 4% and over a larger region.

513

514

Konstantinos Arfanakis and Bruce P. Hermann

10%

0.1 sec

1 sec

10 sec

0

10%

Fig. 30.3 This figure shows changes in the ratio of diffusion weighted to non-diffusion weighted signal in rats following electroshock trains of 0.1 s, 1 s and 10 s in duration, superimposed onto gray-scale anatomical images. Increased ratio of diffusion weighted to non-diffusion weighted signal corresponds to reduced ADC. Therefore, this figure demonstrates a reduction of ADC shortly after application of electroshock. The size of the region of the brain with reduced ADC increased with the duration of the electroshock trains. (This figure is a kind contribution of Dr. J. Zhong).

The DWI studies in experimentally induced seizures that are discussed above cannot be easily compared to each other, because different mechanisms for inducing seizures were used in different models. Also, as it was already mentioned, most of the studies used convulsant drugs making it impossible to separate the effects of the drugs from the effects of the seizures on the diffusion properties of brain tissue. However, most of these studies demonstrated a decrease of ADC values acutely in relation to the onset of the seizures, and in some cases a return to normal values after a certain time period. Diffusion MRI in seizure disorders in humans MRI research on the effects of seizure disorders on the diffusion properties of human brain tissue has demonstrated a sequence of diffusion changes similar to that shown in cerebral ischemia. Several studies have measured a decrease of ADC in the ictal period or shortly after seizure onset, followed by a return to normal, and in some cases, higher than normal ADC values. Additionally, interictal DTI studies have demonstrated regions with reduced diffusion anisotropy. Ictal diffusion MRI in humans In a DWI study during status epilepticus ADC was decreased in the epileptogenic side of the motor cortex by 27% (Wieshmann et al., 1997). In contrast,

ictal ADC in the subcortical WM was increased by 31%. All DWI abnormalities were resolved after cessation of status epilepticus. In a DWI study immediately after a generalized seizure ADC values in the right frontal gray matter (GM) and WM were lower than normal (Sagiuchi et al., 2001). This change in ADC corresponded to hyperperfusion in the same regions, measured by SPECT. In a study where TLE patients were scanned within the first hours after seizures, ADC was reduced in the region of the seizure focus (Diehl et al., 2001). Follow-up exams 17 h later for one of the patients and 3 months later for another, showed a clear tendency of ADC values to normalization for the first patient, and normal ADC values for the second. In a group of patients diagnosed with generalized tonic-clonic seizures or status epilepticus, ictal ADC values were reduced in different WM and GM regions (Kim et al., 2001). Follow-up DWI studies, after adequate seizure control was achieved with the administration of antiepileptic drugs (AEDs), showed in some patients tendency for recovery of the ADC values, and in others complete resolution of the ADC changes. The follow-up DWI scans were performed between 9 days and 18 months after seizure onset. In a different experiment on three patients during status epilepticus, DWI scans demonstrated gyriform cortical hypointensity on ADC maps (Lansberg et al., 1999). ADC changes occurred throughout the hemisphere affected by seizures, as identified by EEG. The follow-up DWI study in one of the patients showed that the hypointensity previously seen on ADC had resolved 1 year later. The same was true for the second patient at the 2-month follow-up, with the addition of two small hyperintense lesions in the affected hemisphere. For these two patients conventional MRI studies during the follow-up showed some cerebral atrophy on the affected hemisphere. The third patient expired before follow-up. In another DWI study on a patient 6 days after status epilepticus, mean DW signal was increased (ADC was reduced) in the WM of the left hemisphere (Hisano et al., 2000). Seven days after the first scan, mean DW signal was also increased (ADC was reduced) in the GM of the occipital and parietal lobes. Twenty-three days after the second scan all DWI abnormalities were resolved. However, considerable brain atrophy was detected. In a DWI study on

Diffusion and perfusion MR imaging in seizure disorders

patients with nocturnal frontal lobe epilepsy, ADC was reduced throughout the brain and not only in the frontal lobe (Ferini-Strambi et al., 2000). The time between the last seizure and the DWI scan for each patient ranged from 6 to 210 days. In a DWI exam on a patient during status epilepticus ADC maps were hypointense primarily in the occipital lobes (Chu et al., 2001). The 5-month follow-up DWI scan showed higher than normal ADC values in the same regions, accompanied by cortical atrophy and ventricular enlargement. Interictal diffusion MRI in humans In an interictal DWI study on a group of patients with hippocampal sclerosis (HS), which is a very common structural abnormality in patients with refractory epilepsy, the sclerotic hippocampi had higher ADC and lower diffusion anisotropy than normal appearing hippocampi in patients and hippocampi in controls (Wieshmann et al., 1999). Hippocampal T2relaxation time and ADC were positively correlated (r  0.76, P  0.001) and hippocampal volume and ADC were negatively correlated (r  0.61, P  0.001) showing that ADC was a reasonably good indicator of sclerotic hippocampi. In an interictal DWI study in TLE patients with and without HS, ADC values were significantly higher on the ictogenic side compared to the non-ictogenic side and compared to normal controls (Leonhardt et al., 2002). During hyperventilation, which can be used to provoke epileptiform activity (Foerster et al., 1924), ADC in sclerotic hippocampi on the epileptogenic side was significantly decreased in comparison to the nonictogenic side and normal controls. In an interictal DWI study in patients with mesial TLE, ADC was elevated by a mean of 10% in the hippocampus of the epileptogenic side, as that was determined by clinical history and EEG recordings (Hugg et al., 1999). In another interictal DWI study on patients with TLE who subsequently underwent epilepsy surgery, hippocampal ADC was higher on the side of the temporal lobe that was operated in 80% of the cases, and temporal stem ADC in 65% of the cases (Kantarci et al., 2002). In a patient with chronic seizures, an interictal DTI study revealed increased ADC in the right orbitofrontal cortex (cf. Case Study 30.1) (Rugg-Gunn

et al., 2002). EEG also showed interictal right frontotemporal slow waves and ictal epileptiform activity arising from the right frontal lobe. In patients with epilepsy and malformations of cortical development or lesions, interictal DTI showed increased ADC and reduced diffusion anisotropy not only in the malformations of cortical development regions, but in other brain regions as well (Eriksson et al., 2001; Rugg-Gunn et al., 2001). Finally, an interictal DTI study that we performed in TLE patients focused only on large WM structures that were not part of the temporal lobes (Arfanakis et al., 2002b). In that study, a significant reduction of the diffusion anisotropy was detected in selected WM pathways. This revealed that the diffusion changes associated with seizures may be global. Also, for the posterior corpus callosum diffusion anisotropy was significantly correlated with the age at onset of seizures. This showed that the effects associated with seizure activity may depend on the age at onset of the seizures. Based on the diffusion MRI studies described above, it appears that the type and temporal behavior of the diffusion changes associated with seizure activity are very similar to the type and progression of the diffusion changes secondary to cerebral ischemia. Several studies on experimentally induced seizures, as well as studies on human subjects, demonstrated an initial decrease of ADC during, or immediately after the seizures. This decrease was sustained for minutes, hours, or days, depending on the study. Some researchers also report a return of the ADC to normal, or even to higher than normal values during the interictal stage. Another important finding of most of the diffusion MRI studies listed above is that diffusion changes were detected in regions which often appeared normal on conventional MRI, but were identified by clinical evaluation and EEG recordings as epileptogenic. Therefore, diffusion MRI appears to be sensitive to the physiological effects of seizures on brain tissue, and in some cases it could provide unique information on seizure localization, which is essential in surgical treatment. However, the limitations of diffusion MRI should also be recognized. For instance, diffusion MRI is often performed using single-shot echo-planar imaging (EPI) sequences that are sensitive to magnetic susceptibility artifacts, which commonly arise

515

516

Konstantinos Arfanakis and Bruce P. Hermann

in the temporal lobe (cf. Chapter 6). When specifically evaluating diffusion in patients with suspected TLE it may be advisable to choose pulse sequences which have reduced sensitivity to susceptibility effects, such as so-called “PROPELLER” (Pipe et al., 2002) or line-scan DWI (Maier et al., 1998). Also, it should be recognized that changes in DWI or ADC values may be small, which can make diagnosis in individual cases problematic.

Mechanisms responsible for changes in diffusion and perfusion MRI in seizure disorders Although the diffusion changes that were reported secondary to seizure activity may be similar to those following cerebral ischemia, they cannot be attributed to ischemia since CBF was shown to increase significantly during, and shortly after seizures. Several mechanisms have been proposed to explain the decrease in ADC following seizure activity. During seizures, glucose levels and oxygen metabolism increase resulting in increased blood flow (Kim et al., 2001). However, relative increase in metabolism after seizure onset may persist longer than increased blood flow (Men et al., 2000), or the increase in blood flow may not be sufficient (Kim et al., 2001). This causes anaerobic metabolism to take over, resulting in excess production of lactic acid and decrease of phosphocreatine. In addition, due to the increased metabolism during seizures there is increased energy expenditure. This in turn may cause energy failure of the Na/K ATPase pump (Wang et al., 1996) as well as increased membrane permeability, resulting in increased extracellular potassium and increased net uptake of calcium and sodium intracellularly (Diehl et al., 2001). This is followed by increased influx of water from the extracellular to the intracellular space and therefore cytotoxic edema (Schaefer et al., 1997) which leads to a reduction of ADC (Sevick et al., 1992). A reduction of 30% of the extracellular space has been reported in status epilepticus (Lux et al., 1986). Intracellular accumulation of calcium can cause cell death by activating calcium dependent enzymes such as proteases and phospholipases, which can result to cell

membrane breakdown (Wasterlain et al., 1993; Helpern and Huang, 1995; Kim et al., 2001). Another mechanism that may contribute to cellular damage in the region of the seizure focus might be excessive release of excitatory amino acids such as glutamate (Glu), leading to neurotoxicity mostly mediated through increased calcium influx (Olney, 1985; Choi et al., 1988; Diehl et al., 2001). Furthermore, neuronal cell death, which may begin as early as 3 h after the onset of seizures (Wall et al., 2000), leads to macrophage and astrocyte proliferation and hypertrophy which may also contribute to reduced water mobility and reduced ADC (Wall et al., 2000). After cell death has occurred and dead cells have been cleared away, increased extracellular space may cause an increase in ADC. Increased ADC is also commonly associated with vasogenic edema. A study combining DWI, histological analysis and MR spectroscopy (MRS) correlated increased interictal ADC with reduced N-acetyl aspartate (NAA), neuronal loss and gliosis (Tokomitsu et al., 1997). In a different DWI study, epileptogenic brain tissue with increased ADC was resected and histology revealed substantial diffuse gliosis. Therefore, neuronal cell death and gliosis may eventually lead to increased extracellular space, giving a less restrictive environment for diffusion of water molecules, and causing an increase in ADC. The same process may also lead to a decrease in diffusion anisotropy in WM tissue due to reduced restriction in directions perpendicular to the axons.

REFERENCES Arfanakis K, Haughton VM, Carew JD, Rogers BP, Dempsey RJ, Meyer ME. 2002a. Diffusion tensor MR imaging in diffuse axonal injury. Am J Neuroradiol 23(5): 794–802. Arfanakis K, Hermann BP, Rogers BP, Carew JD, Seidenberg M, Meyerand ME. 2002b. Diffusion tensor MRI in temporal lobe epilepsy. Magn Reson Imaging 20(7): 511–519. Aronen HJ, Gazit IE, Louis DN, Buchbinder BR, Pardo FS, Weisskoff RM, Harsh GR, Cosgrove GR, Halpern EF, Hochberg FH. 1994. Cerebral blood volume maps of gliomas: comparison with tumor grade and histologic findings. Radiology 191(1): 41–51. Bammer R, Augustin M, Strasser-Fuchs S, Seifert T, Kapeller P, Stollberger R, Ebner F, Hartung HP, Fazekas F. 2000.

Diffusion and perfusion MR imaging in seizure disorders

Magnetic resonance diffusion tensor imaging for characterizing diffuse and focal white matter abnormalities in multiple sclerosis. Magn Reson Med 44: 583–591. Basser PJ. 1995. Inferring microstructural features and the physiological state of tissues from diffusion-weighted images. NMR in Biomedicine 8: 333–344. Basser PJ, Mattiello J, Le Bihan D. 1994a. Estimation of the effective self-diffusion tensor from the NMR spin echo. J Magn Reson B 103: 247–254. Basser PJ, Mattiello J, Le Bihan D. 1994b. MR diffusion tensor spectroscopy and imaging. Biophys J 66: 259–267. Basser PJ, Pierpaoli C. 1996. Microstructural and physiological features of tissues elucidated by quantitative diffusiontensor MRI. J Magn Reson B 111: 209–219. Belliveau JW, Kennedy Jr DN, McKinstry RC, Buchbinder BR, Weisskoff RM, Cohen MS, Vevea JM, Brady TJ, Rosen BR. 1991. Functional mapping of the human visual cortex by magnetic resonance imaging. Science 254(5032): 716–719. Belliveau JW, Rosen BR, Kantor HL, Rzedzian RR, Kennedy DN, McKinstry RC, Vevea JM, Cohen MS, Pykett IL, Brady TJ. 1990. Functional cerebral imaging by susceptibility-contrast NMR. Magn Reson Med 14(3): 538–546. Bhagat YA, Obenaus A, Hamilton MG, Kendall EJ. 2001. Magnetic resonance imaging predicts neuropathology from soman-mediated seizures in the rodent. Neuroreport 12(7): 1481–1487. Bruening R, Kwong KK, Vevea MJ, Hochberg FH, Cher L, Harsh IVth, GR. Niemi PT, Weisskoff RM, Rosen BR. 1996. Echoplanar MR determination of relative cerebral blood volume in human brain tumors: T1 vs. T2 weighting. Am J Neuroradiol 17(5): 831–840. Busza AL, Allen KL, King MD, van Bruggen N, Williams SR, Gadian DG. 1992. Diffusion-weighted imaging studies of cerebral ischemia in gerbils. Potential relevance to energy failure. Stroke 23(11): 1602–1612. Choi DW, Koh JY, Peters S. 1988. Pharmacology of glutamate neurotoxicity in cortical cell culture: attenuation by NMDA antagonists. J Neurosci 8(1): 185–196. Chu K, Kang DW, Kim JY, Chang KH, Lee SK. 2001. Diffusionweighted magnetic resonance imaging in nonconvulsive status epilepticus. Arch Neurol 58(6): 993–998. de Crespigny AJ, Marks MP, Enzmann DR, Moseley ME. 1995. Navigated diffusion imaging of normal and ischemic human brain. Magn Reson Med 33: 720–728. Detre JA, Leigh JS, Williams DS, Koretsky AP. 1992. Perfusion imaging. Magn Reson Med 23(1): 37–45. Diehl B, Najm I, Ruggieri P, Tkach J, Mohamed A, Morris H, Wyllie E, Fisher E, Duda J, Lieber M, Bingaman W, Luders HO. 2001. Postictal diffusion-weighted imaging for the localization of focal epileptic areas in temporal lobe epilepsy. Epilepsia 42(1): 21–28.

Ebisu T, Rooney WD, Graham SH, Mancuso A, Weiner MW, Maudsley AA. 1996. MR spectroscopic imaging and diffusion-weighted MRI for early detection of kainate-induced status epilepticus in the rat. Magn Reson Med 36(6): 821–828. Edelman RR, Siewert B, Darby DG, Thangaraj V, Nobre AC, Mesulam MM, Warach S. 1994. Qualitative mapping of cerebral blood flow and functional localization with echoplanar MR imaging and signal targeting with alternating radio frequency. Radiology 192(2): 513–520. El-Koussy M, Mathis J, Lovblad KO, Stepper F, Kiefer C, Schroth G. 2002. Focal status epilepticus: follow-up by perfusion- and diffusion MRI. Eur Radiol 12(3): 568–574. Engel Jr J. 1984. The use of positron emission tomographic scanning in epilepsy. Ann Neurol (suppl 15): S180–S191. Eriksson SH, Rugg-Gunn FJ, Symms MR, Barker GJ, Duncan JS. 2001. Diffusion tensor imaging in patients with epilepsy and malformations of cortical development. Brain 124(Pt 3): 617–626. Ferini-Strambi L, Bozzali M, Cercignani M, Oldani A, Zucconi M, Filippi M. 2000. Magnetization transfer and diffusionweighted imaging in nocturnal frontal lobe epilepsy. Neurology 54(12): 2331–2333. Fish DR, Brooks DJ, Young IR, Bydder GM. 1988. Use of magnetic resonance imaging to identify changes in cerebral blood flow in epilepsia partialis continua. Magn Reson Med 8(2): 238–240. Flacke S, Wullner U, Keller E, Hamzei F, Urbach H. 2000. Reversible changes in echo planar perfusion- and diffusionweighted MRI in status epilepticus. Neuroradiology 42(2): 92–95. Foerster O. 1924. Hyperventilationsepilepsie. Deutch Z Nervenheilk 83: 347–356. Franck G, Sadzot B, Salman E, Depresseux JC, Grisar T, Peters JM, Guillaume M, Quaglia L, Delfiore G, Lamotte D. 1986. Regional cerebral blood flow and metabolic rates in human focal epilepsy and status epilepticus. Adv Neurol 44: 935–948. Gaillard WD, Zeffiro T, Fazilat S, DeCarli C, Theodore WH. 1996. Effect of valproate on cerebral metabolism and blood flow: an 18F-2-deoxyglucose and 15O water positron emission tomography study. Epilepsia 37(6): 515–521. Heiniger P, el-Koussy M, Schindler K, Lovblad KO, Kiefer C, Oswald H, Wissmeyer M, Mariani L, Donati F, Schroth G, Weder B. 2002. Diffusion and perfusion MRI for the localisation of epileptogenic foci in drug-resistant epilepsy. Neuroradiology 44(6): 475–480. Helpern JA, Huang N. 1995. Diffusion-weighted imaging in epilepsy. Magn Reson Imaging 13(8): 1227–1231. Hisano T, Ohno M, Egawa T, Takano T, Shimada M. 2000. Changes in diffusion-weighted MRI after status epilepticus. Pediatr Neurol 22(4): 327–329.

517

518

Konstantinos Arfanakis and Bruce P. Hermann

Hugg JW, Butterworth EJ, Kuzniecky RI. 1999. Diffusion mapping applied to mesial temporal lobe epilepsy: preliminary observations. Neurology 53(1): 173–176. Kantarci K, Shin C, Britton JW, So EL, Cascino GD, Jack Jr CR. 2002. Comparative diagnostic utility of 1H MRS and DWI in evaluation of temporal lobe epilepsy. Neurology 58(12): 1745–1753. Kim SG. 1995. Quantification of relative cerebral blood flow change by flow-sensitive alternating inversion recovery (FAIR) technique: application to functional mapping. Magn Reson Med 34(3): 293–301. Kim JA, Chung JI, Yoon PH, Kim DI, Chung TS, Kim EJ, Jeong EK. 2001. Transient MR signal changes in patients with generalized tonicoclonic seizure or status epilepticus: periictal diffusion-weighted imaging. Am J Neuroradiol 22(6): 1149–1160. Kucharczyk J, Mintorovitch J, Asgari H, Tsuura M, Moseley M. 1991. In vivo diffusion-perfusion magnetic resonance imaging of acute cerebral ischemia. Can J Physiol Pharmacol 69(11): 1719–1725. Kwong KK, Chesler DA, Weisskoff RM, Donahue KM, Davis TL, Ostergaard L, Campbell TA, Rosen BR. 1995. MR perfusion studies with T1-weighted echo planar imaging. Magn Reson Med 34(6): 878–887. Lansberg MG, O’Brien MW, Norbash AM, Moseley ME, Morrell M, Albers GW. 1999. MRI abnormalities associated with partial status epilepticus. Neurology 52(5): 1021–1027. Le Bihan D. 1991. Diffusion NMR imaging. Magn Reson Q 7: 1–30. Lee BI, Markand ON, Siddiqui AR, Park HM, Mock B, Wellman HH, Worth RM, Edwards MK. 1986. Single photon emission computed tomography (SPECT) brain imaging using N,N,N-trimethyl-N-(2 hydroxy-3-methyl-5-123I-iodobenzyl)-1,3-propanediamine 2 HCl (HIPDM): intractable complex partial seizures. Neurology 36: 1471–1477. Leonhardt G, de Greiff A, Marks S, Ludwig T, Doerfler A, Forsting M, Konermann S, Hufnagel A. 2002. Brain diffusion during hyperventilation: diffusion-weighted MR-monitoring in patients with temporal lobe epilepsy and in healthy volunteers. Epilepsy Res 51(3): 269–278. Lim KO, Hedehus M, Moseley M, de Crespigny A, Sullivan EV, Pfefferbaum A. 1999. Compromised white matter tract integrity in schizophrenia inferred from diffusion tensor imaging. Arch Gen Psychiatry 56: 367–374. Liu HL, Gao JH. 1999. Perfusion-based event-related functional MRI. Magn Reson Med 42(6): 1011–1013. Liu HL, Pu Y, Liu Y, Nickerson L, Andrews T, Fox PT, Gao JH. 1999. Cerebral blood flow measurement by dynamic contrast MRI using singular value decomposition with an adaptive threshold. Magn Reson Med 42(1): 167–172. Liu HL, Kochunov P, Hou J, Pu Y, Mahankali S, Feng CM, Yee SH, Wan YL, Fox PT, Gao JH. 2001. Perfusion-weighted imaging of

interictal hypoperfusion in temporal lobe epilepsy using FAIR-HASTE: comparison with H(2)(15)O PET measurements. Magn Reson Med 45(3): 431–435. Lux HD, Heinemann U, Dietzel I. 1986. Ionic changes and alterations in the size of the extracellular space during epileptic activity. Adv Neurol 44: 619–639. Lythgoe MF, Busza AL, Calamante F, Sotak CH, King MD, Bingham AC, Williams SR, Gadian DG. 1997. Effects of diffusion anisotropy on lesion delineation in a rat model of cerebral ischemia. Magn Reson Med 38: 662–668. Magistretti P, Uren R, Blume H, Schomer D, Royal H. 1982. Delineation of epileptic focus by single photon emission tomography. Eur J Nucl Med 7(10): 484–485. Maier SE, Gudbjartsson H, Patz S, Hsu L, Lovblad KO, Edelman RR, Warach S, Jolesz FA. 1998. Line scan diffusion imaging: characterization in healthy subjects and stroke patients. AJR Am J Roentgenol 171(1): 85–93. Marks DA, Katz A, Hoffer P, Spencer SS. 1992. Localization of extratemporal epileptic foci during ictal single photon emission computed tomography. Ann Neurol 31(3): 250–255. Men S, Lee DH, Barron JR, Munoz DG. 2000. Selective neuronal necrosis associated with status epilepticus: MR findings. Am J Neuroradiol 21(10): 1837–1840. Moseley ME, Kucharczyk J, Mintorovitch J, Cohen Y, Kurhanewicz J, Derugin N, Asgari H, Norman D. 1990. Diffusion-weighted MR imaging of acute stroke: correlation with T2-weighted and magnetic susceptibility-enhanced MR imaging in cats. Am J Neuroradiol 11(3): 423–429. Nakasu Y, Nakasu S, Kizuki H, Uemura S, Morikawa S, Inubushi T, Handa J. 1995. Changes in water diffusion of rat limbic system during status epilepticus elicited by kainate. Psychiatry Clin Neurosci 49(3): S228–S230. Olney JW. 1985. Excitatory transmitters and epilepsy-related brain damage. Int Rev Neurobiol 27: 337–362. Østergaard L, Sorensen AG, Kwong KK, Weisskoff RM, Gyldensted C, Rosen BR. 1996a. High resolution measurement of cerebral blood flow using intravascular tracer bolus passages. Part II: experimental comparison and preliminary results. Magn Reson Med 36(5):726–736. Østergaard L, Weisskoff RM, Chesler DA, Gyldensted C, Rosen BR. 1996b. High resolution measurement of cerebral blood flow using intravascular tracer bolus passages. Part I: mathematical approach and statistical analysis. Magn Reson Med 36(5): 715–725. Pierpaoli C, Jezzard P, Basser PJ, Barnett A, Di Chiro G. 1996. Diffusion tensor MR imaging of the human brain. Radiology 201: 637–648. Pipe JG, Farthing VG, Forbes KP. 2002. Multishot diffusionweighted FSE using PROPELLER MRI. Magn Reson Med 47: 42–52.

Diffusion and perfusion MR imaging in seizure disorders

Righini A, Pierpaoli C, Alger JR, Di Chiro G. 1994. Brain parenchyma apparent diffusion coefficient alterations associated with experimental complex partial status epilepticus. Magn Reson Imaging 12(6): 865–871. Rosen BR, Belliveau JW, Vevea JM, Brady TJ. 1990. Perfusion imaging with NMR contrast agents. Magn Reson Med 14(2): 249–265. Rugg-Gunn FJ, Eriksson SH, Symms MR, Barker GJ, Duncan JS. 2001. Diffusion tensor imaging of cryptogenic and acquired partial epilepsies. Brain 124(Pt 3): 627–636. Rugg-Gunn FJ, Eriksson SH, Symms MR, Barker GJ, Thom M, Harkness W, Duncan JS. 2002. Diffusion tensor imaging in refractory epilepsy. Lancet 359(9319): 1748–1751. Sagiuchi T, Ishii K, Asano Y, Aoki Y, Woodhams R, Yanaihara H, Kan S, Hayakawa K. 2001. Transient seizure activity demonstrated by Tc-99m HMPAO SPECT and diffusion-weighted MR imaging. Ann Nucl Med 15(3): 267–270. Schaefer PW, Buonanno FS, Gonzalez RG, Schwamm LH. 1997. Diffusion-weighted imaging discriminates between cytotoxic and vasogenic edema in a patient with eclampsia. Stroke 28(5): 1082–1085. Sevick RJ, Kanda F, Mintorovitch J, Arieff AI, Kucharczyk J, Tsuruda JS, Norman D, Moseley ME. 1992. Cytotoxic brain edema: assessment with diffusion-weighted MR imaging. Radiology 185(3): 687–690. Siewert B, Schlaug G, Edelman RR, Warach S. 1997. Comparison of EPISTAR and T2*-weighted gadoliniumenhanced perfusion imaging in patients with acute cerebral ischemia. Neurology 48(3): 673–679. Stejskal EO, Tanner JE. 1964. Spin diffusion measurements: Spin echoes in the presence of a time-dependent field gradient. J Chem Physics 42: 288–292. Theodore WH, Bromfield E, Onorati L. 1989. The effect of carbamazepine on cerebral glucose metabolism. Ann Neurol 25(5): 516–520. Tokumitsu T, Mancuso A, Weinstein PR, Weiner MW, Naruse S, Maudsley AA. 1997. Metabolic and pathological effects of temporal lobe epilepsy in rat brain detected by proton spectroscopy and imaging. Brain Res 744(1): 57–67. van Gelderen P, de Vleeschouwer MH, DesPres D, Pekar J, van Zijl PC, Moonen CT. 1994. Water diffusion and acute stroke. Magn Reson Med 31(2): 154–163. Wall CJ, Kendall EJ, Obenaus A. 2000. Rapid alterations in diffusion-weighted images with anatomic correlates in a rodent model of status epilepticus. Am J Neuroradiol 21(10): 1841–1852.

Wang Y, Majors A, Najm I, Xue M, Comair Y, Modic M, Ng TC. 1996. Postictal alteration of sodium content and apparent diffusion coefficient in epileptic rat brain induced by kainic acid. Epilepsia 37(10): 1000–1006. Warach S, Chien D, Li W, Ronthal M, Edelman RR. 1992. Fast magnetic resonance diffusion-weighted imaging of acute human stroke. Neurology 42(9):1717–1723. Warach S, Dashe JF, Edelman RR. 1996. Clinical outcome in ischemic stroke predicted by early diffusion-weighted and perfusion magnetic resonance imaging: a preliminary analysis. J Cereb Blood Flow Metab 16(1): 53–59. Warach S, Gaa J, Siewert B, Wielopolski P, Edelman RR. 1995. Acute human stroke studied by whole brain echo planar diffusion-weighted magnetic resonance imaging. Ann Neurol 37: 231–241. Warach S, Levin JM, Schomer DL, Holman BL, Edelman RR. 1994. Hyperperfusion of ictal seizure focus demonstrated by MR perfusion imaging. Am J Neuroradiol 15(5): 965–968. Wasterlain CG, Fujikawa DG, Penix L, Sankar R. 1993. Pathophysiological mechanisms of brain damage from status epilepticus. Epilepsia 34 (suppl 1): S37–S53. Wieshmann UC, Clark CA, Symms MR, Barker GJ, Birnie KD, Shorvon SD. 1999. Water diffusion in the human hippocampus in epilepsy. Magn Reson Imaging 17(1): 29–36. Wieshmann UC, Symms MR, Shorvon SD. 1997. Diffusion changes in status epilepticus. Lancet 350(9076): 493–494. Wolf RL, Alsop DC, Levy-Reis I, Meyer PT, Maldjian JA, Gonzalez-Atavales J, French JA, Alavi A, Detre JA. 2001. Detection of mesial temporal lobe hypoperfusion in patients with temporal lobe epilepsy by use of arterial spin labeled perfusion MR imaging. Am J Neuroradiol 22(7): 1334–1341. Wu RH, Bruening R, Noachtar S, Arnold S, Berchtenbreiter C, Bartenstein P, Drzezga A, Tatsch K, Reiser M. 1999. MR measurement of regional relative cerebral blood volume in epilepsy. J Magn Reson Imaging 9(3): 435–440. Zhong J, Petroff OA, Prichard JW, Gore JC. 1993. Changes in water diffusion and relaxation properties of rat cerebrum during status epilepticus. Magn Reson Med 30(2): 241–246. Zhong J, Petroff OA, Prichard JW, Gore JC. 1995. Barbituratereversible reduction of water diffusion coefficient in flurothyl-induced status epilepticus in rats. Magn Reson Med 33(2): 253–256. Zhong J, Petroff OA, Pleban LA, Gore JC, Prichard JW. 1997. Reversible, reproducible reduction of brain water apparent diffusion coefficient by cortical electroshocks. Magn Reson Med 37(1): 1–6.

519

520

Konstantinos Arfanakis and Bruce P. Hermann

Case Study 30.1 Identifying an epileptic focus with DTI John S Duncan FRCP, Department of Clinical and Experimental Epilepsy, Institute of Neurology UCL. National Hospital for Neurology and Neurosurgery, Queen Square WC1N 3BG. National Society for Epilepsy, Chalfont St Peter, SL9 0LR. History A 30 year old female had a two decade history of refractory frontal lobe epilepsy. Conventional MRI was unremarkable.

Technique DTI with calculation of mean diffusivity and voxel-based comparison of image with 30 normal controls, using statistical parametric mapping (SPM).

Imaging findings Area of increased mean diffusivity in right frontal lobe.

Focal area of increased mean diffusivity.

Discussion Imaging with DTI and analysis of mean diffusivity image with SPM revealed a focal abnormality that was not evident on conventional MRI and which presented a target for intra-cerebral EEG recording that demonstrated the site of seizure onset in the orbito-frontal gyrus that showed increased diffusivity. This area was then resected with a good result (Rugg-Gunn et al., 2002).

Key point Advanced MRI techniques may show abnormalities that underlie refractory focal epilepsy that are not evident on conventional MRI, and identify a target for surgical treatment.

Top: unremarkable pre-op conventional MRI Bottom: post resection MRI.

Reference Rugg-Gunn FJ, Eriksson SH, Symms MR, Barker GJ, Thom M, Harkness W, Duncan JS. 2002. Diffusion tensor imaging in refractory epilepsy – localization of focus and histopathological confirmation following resective surgery. Lancet 359: 1748–1751.

Section 6 Psychiatric and neurodegenerative diseases

31

Psychiatric and neurodegenerative disease: overview Adam D. Waldman Department of Imaging, Charing Cross Hospital, London, UK Imperial College of Science Technology and Medicine, Institute of Neurology, University College London, UK

Introduction The advent of neuroimaging techniques which yield physiological in addition to structural information are of particular interest in the scientific and clinical investigations of neurodegenerative and psychiatric disorders. These are groups of conditions where any structural changes which are evident on anatomical imaging sequences generally correlate poorly with clinical diagnostic categories, underlying pathophysiology and disease severity. T1- and T2-dependent MR sequences which are the mainstay of routine clinical neuroimaging are frequently insensitive to the underlying pathological processes in these diseases. Focal or global atrophy due to associated neuronal loss is also frequently subtle or absent, particularly early in the course of disease. As a result, clinical brain imaging using standard techniques is frequently normal, or non-specifically abnormal. Physiological imaging can be considered to have two main purposes in this context: The first is clinical; to provide diagnostic information which augments that available from clinical examination, laboratory tests and conventional structural brain imaging. The aim here is to increase the sensitivity and or specificity of the imaging examination as a whole, and improve diagnostic confidence, which will ultimately guide clinical management. In this context, the technique must provide a surrogate marker of disease, which is of predictive value in diagnosis or prognosis for an individual patient. In practice, this requires a distinctive imaging appearance or, in the

case of quantitative techniques, sufficient separation between parameters measured to allow an individual being examined to be placed confidently in a diagnostic or prognostic clinical group. In order to have widespread clinical impact, the technique must improve specificity, sensitivity, safety or cost-efficacy in securing a diagnosis. A related application is in the provision of surrogate markers of disease progression and therapeutic response. As this involves longitudinal measures, the stability of the technique and normal physiological variation in comparison to the magnitude of biological changes being examined are key issues. It is also worth commenting that statistically significant differences between clinical groups, even if there is overlap between them, may be valuable in evaluation of therapy, as therapeutic response is itself frequently only reflected in group analyses. The second main role is scientific, and aims toward better understanding of pathological changes in the brain, their anatomical distribution and how they manifest as clinical disease. Although the statistical demands in terms of separation of groups may be less, a more rigorous interpretation of the measured physiological parameters in terms of the underlying disease process is required. Details of the application of physiological techniques and references are covered in Chapters 32, 33, 34 and 35, and will not be repeated here. The remit of this section is an overview of some general issues concerning the application of diffusion, perfusion and MR spectroscopic (MRS) techniques in neurodegenerative and psychiatric disease, and their clinical impact in the context of 523

524

Adam D. Waldman

other available tests. Some areas, such as perfusion imaging in degenerative disease and physiological imaging in prion disease, where there has been insufficient published work to justify a separate chapter are also discussed here.

Issues in neurodegeneration and psychiatry Studies of neurodegenerative and psychiatric diseases present particular challenges, which are worthy of brief mention.

where recruitment is difficult for other reasons) to group patients with heterogenous pathologies to achieve statistical power for data analysis; in these circumstances, there is strong argument for multicenter studies using consistent diagnostic criteria and imaging protocols. Many neurodegenerative conditions are diseases of the elderly. It is important to distinguish changes in physiological imaging parameters due to the pathology of interest from those related to normal aging, comorbidity and intercurrent medication which are also very much more common in older people.

Patient factors and diagnostic validation Validation of diagnosis is a perennial difficulty in the study of patients with psychiatric and neurodegenerative diseases. In most psychiatric conditions, diagnosis is based on a constellation of clinical features and no laboratory test exists as a reference standard. Moreover the diagnostic criteria, and the way they are applied, may differ between countries (cf. Chapter 33). In organic neurodegerative diseases, definitive diagnosis relies on characteristic histopathological changes within brain parenchyma; however, the morbidity and mortality associated with brain biopsy is rarely justified in this clinical population. Any pathological examination is therefore usually post-mortem, which for chronic diseases may be several years later; hence, interpretation may be confounded by the normal aging process and other intercurrent pathology. Changes in public attitudes, clinical practice and, in some countries, laws governing the examination and disposal of human tissue also mean that autopsy and the use of pathological specimens in a research context will be performed less frequently. The situation is easier in those relatively rare conditions where a specific gene has been identified as a cause, although it must be borne in mind that phenotypic expression of the underlying genetic abnormality may be highly variable. It is also, bearing in mind the diagnostic difficulties outlined above, important to study the pure pathological groups where possible. It is perilous (although tempting, particularly with unusual disorders where subject numbers are limited, or

Technical and practical issues In addition to the technical considerations and limitations which are common to the application of physiological MR imaging (MRI) techniques, and are discussed in Chapters 1–9, there are a number of considerations particularly relevant to the study of neurodegenerative and psychiatric disease. Total examination time should be kept to a minimum (ideally less than 45 min), and maximum effort made to ensure subjects who are often elderly are as comfortable as possible in the magnet.Degenerative kyphosis of the neck can present particular problems with positioning within a headcoil. By the nature of neurodegenerative and psychiatric diseases, patients may be confused or agitated and may experience involuntary movements; safety in the scanner and problems with subject motion may therefore be significant issues. Elderly patients frequently have vascular disease, which can affect both cardiac output and transit times, and will influence processing of perfusion parameters using either bolus or arterial spin labeling (ASL) methods.

Physiological MRI in common dementias Alzheimer’s disease (AD) is the leading cause of dementia in the developed world. The increasing incidence of AD as populations become more aged and the advent of specific treatments, lends impetus to the development of reliable and accessible tests for early diagnosis and for monitoring disease progression. The

Psychiatric and neurodegenerative disease: overview

diagnosis of AD usually relies principally on clinical criteria and it is widely accepted that even for patients who fulfill the National Institute of Neurological and Communicative Disorders and Stroke–Alzheimer’s Disease and Related Disorders Association (NINCDS– ADRDA) diagnostic criteria (McKhann et al., 1984), this is in error in 10–20% of cases. There is also a growing appreciation of common risk factors for AD and vascular dementias (VaD), and that both pathologies may contribute to cognitive decline in an individual. Other primary degenerative dementias, such as fronto-temporal degeneration (FTD) may present atypically. Consequently there may be considerable overlap between clinical and imaging features in these conditions. MRS Although the earliest pathological changes in AD may take place in the hippocampi, the best current diagnostic MRS test appears to be in the posterior cingulate region, from which high-quality, quantifiable and reproducible, short-TE spectra can be acquired from a predominantly cortical voxel. It is a region of the brain in which there appears to be progressive pathological change throughout the course of AD (Scahill et al., 2002), and is hence also a suitable target for longitudinal studies. Characteristic metabolite changes are of reduced N-acetyl asparate (NAA) and increased myo-inositol (mI) compared to the normal individuals, and in most series the ratio NAA/mI shows good specificity and sensitivity for distinguishing AD patients from normal subjects, and from those with VaD. Whereas, metabolite abnormalities in primary degenerative dementias such as AD and FTD predominantly affect cortical structures, and those in VaD mostly reflect white matter (WM) damage. Although qualitatively similar metabolite abnormalities have been reported in FTD and AD, it has been suggested that their anatomical distribution (predominantly anterior in FTD, more posterior in AD) may be diagnostically discriminant. Metabolite abnormalities (notably elevated mI) are also detectable in patients with mild cognitive impairment (MCI) who are at greatly increased risk of progressing to AD. The quantitative relationships which have been observed between metabolite

ratios and indices of cognitive impairment also support the view that the former reflect pathology which causes cognitive impairment, and progress as part of the disease process.

DWI To date, although diffusion-weighted imaging (DWI) has shown minor regional increases in apparent diffusion coefficient (ADC) and decreased anisotropy indices (cf. Chapter 33) in subjects with AD, these are not sufficiently robust to be used diagnostically. It has been suggested that slower diffusion processes detected at higher b-values (b  2000–3000) than those typically used in routine duffusion-weighted imaging (DWI) (b  1000), may be more sensitive to the pathological processes of AD in WM (Yoshiura et al., 2003). Fiber tractography has yet to be evaluated in a clinical context.

Perfusion imaging Measurement of regional blood flow using dynamic susceptibility MR perfusion has not been widely applied to the study of dementia, but yields information similar to that from hexamethylpropyleneamine oxime single photon emission computed tomography (HMPAO-SPECT), which is quite commonly used. Preliminary comparisons suggest that the MR technique is as specific and at least as sensitive (Harris et al., 1998). It is also less costly, carries no ionizing radiation burden and can be performed at the same time as high-resolution structural imaging and MRS.

Other techniques, multi-modal approaches and economic issues A number of other MRI methods have been applied to the study of AD. These include MRI volumetry, which measure atrophy as a surrogate marker of neuronal loss; based on either focal measures of hippocampal volume, or global brain volume measurements from registered serial imaging (Fox and Freebrough, 1997). Others may be considered “pathology sensitive” and reflect changes in tissue

525

526

Adam D. Waldman

ultrastructure, such as magnetization transfer (MT) techniques which are sensitive to macromoleculebound water (e.g. van de Flier et al., 2002). Both types of approach have shown diagnostic potential in dementia. A recent study which examined the sensitivity and specificity of MRS, ADC measurements and MRIbased hippocampal volumetry in patients with AD, MCI and normal controls, concluded that combined measures gave the highest diagnostic accuracy (Kantarci et al., 2002). The most discriminant parameters, however, varied with disease severity, suggesting that imaging protocols should be tailored to clinical details of the individual under investigation. The broader question as to how MR and other neuroimaging methods (notably SPECT, and glucose studies with positron emission tomography, PET) contribute to management has also been addressed in a recent review (Kantarci and Jack, 2003). The evidence is that MRS, hippocampal volumetry, SPECT and PET each have similar accuracy to clinical evaluation in identifying those with established AD from normal elderly subjects; but these methods are of added value in distinguishing AD from other dementias, may help to identify individuals at risk of progressing to dementia and provide surrogate markers of disease and therapeutic response. Cost-efficiency of physiological and functional neuroimaging in dementia management is an important issue in resource-limited or managed health care systems, particularly as the population at risk grows due to demographical changes. A study based on decision-modeling, and comparing standard clinical work-up with additional costs and from perfusion MRI and PET against quality adjusted life years (QUALY) benefit revealed highly variable, but spectacularly high, costs per QUALY with the added imaging techniques (ranging between $24,000 and $8 million). The conclusion was the addition of functional neuroimaging to usual diagnostic regimens at AD clinics, was clearly not cost-effective, given the efficacy of current treatments (McMahon et al., 2000). The advent of effective, specific disease-modifying therapies for AD is, however, likely to change the cost–benefit balance dramatically. In the meantime, study and refinement of these techniques may be considered as “proof-of-concept”; they are likely to play important roles in the evaluation of, and

patient selection for, such treatments as they become available.

Physiological imaging of prion disease Neuroimaging plays an important role in the diagnosis of prion diseases, which are fatal, progressive neurodegenerative disorders caused by abnormal isomers of normally occurring (PrP) brain proteins. Prion diseases include sporadic Creutzfeldt–Jakob disease (sCJD), familial forms with gene mutations such as Gerstmann–Straussler–Scheinker (GSS) disease, variant Creutzfeldt–Jakob disease (vCJD) and iatrogenic disease from contaminated allograft tissue and surgical instruments. The typical imaging abnormality is of hyperintensity of gray matter (GM) structures on T2-weighted sequences. Imaging patterns in sCJD are variable, and range from isolated abnormal signal in the caudate nuclei to diffuse basal ganglia, thalamic and cortical involvement (Collie et al., 2001). A characteristic pattern of bilateral pulvinar hyperintensity “the pulvinar sign” has been described in a large number of cases with a diagnosis of possible vCJD (Zeidler et al., 2000). The sensitivity of these signs is uncertain in early disease. Study of sCJD cases suggests that DWI offers increased sensitivity for detecting GM abnormalities in these diseases; striking signal change has been detected in the absence of abnormalities on conventional proton density and T2-weighted images (Bahn and Parchi, 1999; Demaerel et al., 1999). The limited published data on DWI in vCJD suggests that it also increases sensitivity to pulvinar abnormalities, but that diffusion characteristics may change over the course of the disease; ADC may be lower or higher than normal brain (Oppenheim et al., 2000; Matoba et al., 2001; Waldman et al., 2003). The sensitivity of DWI in familial disease is not known. A number of studies of MRS in prion disease have been published, mostly isolated case reports in sCJD and familial disease using long-TE techniques. The most consistent finding was of quite marked, but non-specific reduction in NAA in various regions of the brain (e.g. Graham et al., 1993). More recently, short-TE measurements from the thalami of vCJD patients (all abnormal on T2-weighted

Psychiatric and neurodegenerative disease: overview

imaging) have shown striking elevation of mI and very-low NAA (Galanaud et al., 2002; Cordery et al., 2003). The magnitude of the abnormalities suggest that they may be detectable in early disease, before structural imaging abnormalities are present, and would therefore be useful diagnostically. They have been attributed to the severe neuronal depletion and intense gliosis, which are pathological features of this brain region in vCJD. These are similar but more extreme than those seen, for example, in AD. Metabolite changes which are associated with conformational diseases may therefore be thought of as surrogates which reflect final common pathways of neuronal loss and gliosis. We are not aware of any published work on perfusion MR in prion disease; there are some reports of radionuclide perfusion studies (e.g. Miller et al., 1998), but no characteristic pattern of perfusion abnormality has emerged.

a research context, and may provide biomarkers of disease progression and therapeutic response (reviewed recently by Brooks et al., 2003).

Physiological MRI in psychiatry A clinical role for physiological MRI in psychiatric disease has yet to be established; to date, neither MRS, diffusion or perfusion MRI have proven diagnostic or prognostic value in any common psychiatric condition. The findings of 1H and 31H MRS and diffusion imaging in diseases such as schizophrenia, however, give important insight into pathophysiological and anatomical correlates of what were once considered to be “functional psychoses”. They may be used to test conceptual models for the etiology of psychiatric disease, reveal potential therapeutic targets and provide objective surrogate markers of response to novel and established drug treatment.

Physiological MRI in neurodegeneration (movement disorders)

Conclusion

No role has emerged for MRS in the clinical investigation of degenerative movement disorders. To date, the findings in the most common of these, Idiopathic Parkinson’s Disease (IPD), are inconsistent and those in other parkinsonian syndromes, notably multisystem atrophy (MSA) and progressive supranuclear palsy (PSP), differences are not sufficiently specific or discriminant to be useful diagnostically. The limited number of diffusion studies in these conditions show increased ADC in the basal ganglia of MSA and PSP subjects, with high sensitivity and specificity in distinguishing them from IPD or control subjects (e.g. Seppi et al., 2003). Radionuclide imaging with SPECT and PET allow more functional evaluation of local dopaminergic and cholinergic fuction using specific ligands. Dopamine uptake transporter (DAT) imaging using SPECT is the most commonly used in a clinical environment; low uptake in the striatum appears to be a useful diagnostic marker for IPD (which may be distinguishable from PSP, but not MSA; e.g. cf. Antonini et al., 2003). A number of other SPECT and PET ligands and metabolic markers which reflect different aspects of neuronal function are used in

Physiological imaging techniques provide complementary information relevant to the investigation of neurodegenerative and psychiatric diseases. MRS provides biochemical markers of local neuronal health and integrity, and of glial cell populations in structures which are known to subserve important functions; for example, memory and executive function or control of motor processes. They offer potentially sensitive indices which will reflect cellular damage before it is apparent as even subtle bulk volume loss. DWI and diffusion tensor imaging (DTI) yield information about ultrastructure, from which connectivity and neural circuits between different relevant brain regions can be inferred through fiber tractography. The integrity of these circuits and connections are necessary for normal cognition, memory and executive function, as well as motor control. These may be disrupted directly, for example by ischaemic or other damage to relevant WM regions, or may undergo change secondary to loss of function in the regions which they connect. In the case of some psychiatric conditions they may develop abnormally.

527

528

Adam D. Waldman

Perfusion imaging is perhaps of less primary interest in this group of diseases, although the increasing overlap between vascular risk factors and AD, and known abnormalities in regional blood flow in this condition suggests that it may have a further role in dementia imaging.

REFERENCES Antonini A, Benti R, De Notaris R, Tesei S, Zecchinelli A, Sacilotto G, Meucci N, Canesi M, Mariani C, Pezzoli G, Gerundini P. 2003. 123I-Ioflupane/SPECT binding to striatal dopamine transporter (DAT) uptake in patients with Parkinson’s disease, multiple system atrophy, and progressive supranuclear palsy. Neurol Sci 24: 149–150. Bahn MM, Parchi P. 1999. Abnormal diffusion-weighted magnetic resonance images in Creutzfeldt–Jakob Disease. Arch Neurol 56: 577–583. Brooks DJ, Frey KA, Marek KL, Oakes D, Paty D, Prentice R, Shults CW, Stoessl AJ. 2003. Assessment of neuroimaging techniques as biomarkers of the progression of Parkinson’s disease. Exp Neurol 184(suppl 1): S68–S79. Collie DA, Sellar RJ, Zeidler M, Colchester ACF, Knight R, Will RG. 2001. MRI of Creutzfeldt–Jakob disease: imaging features and recommended MRI protocol Clin Rad 56: 726–739. Cordery R, MacManus D, Collinge J, Rossor M, Waldman A. 2003. Short TE proton spectroscopy in variant and familial Creutzfeld–Jakob disease. Proc Int Soc Mag Res Med (Abstract). 11: 438. Demaerel P, Heiner L, Robberecht W, Sciot R, Wilms G. 1999. Diffusion-weighted MRI in sporadic Creutzfeldt–Jakob disease. Neurology 52: 205–208. Fox NC, Freeborough PA. 1997. Brain atrophy progression measured from registered serial MRI: validation and application to Alzheimer’s disease. J Magn Reson Imaging 7(6): 1069–1075. Galanaud D, Dormont D, Grabli D, Charles P, Hauw JJ, Lubetzki C, Brandel JP, Marsault C, Cozzone PJ. 2002. MR spectroscopic pulvinar sign in a case of variant Creutzfeldt–Jakob disease. J Neuroradiol 29(4): 285–287. Graham GD, Petroff OAC, Blamire AM, Rajkowska G, Goldman-Rakic P, Prichard JW. 1993. Proton magnetic resonance spectroscopy in Creutzfeldt–Jakob disease. Neurology 43: 2065–2068. Harris GJ, Lewis RF, Satlin A, English CD, Scott TM, Yurgelun-Todd DA, Renshaw PF. 1998. Dynamic susceptibility contrast MR imaging of regional cerebral blood volume in Alzheimer disease: a promising alternative to nuclear medicine. Am J Neuroradiol 19(9): 1727–1732.

Kantarci K, Jack Jr CR. 2003. Neuroimaging in Alzheimer disease: an evidence-based review. Neuroimaging Clin N Am 13(2): 197–209. Kantarci K, Xu Y, Shiung MM, O’Brien PC, Cha RH, Smith GE, Ivnik RJ, Boeve BF, Edland SD, Kokmen E, Tangalos EG, Petersen RC, Jack Jr CR. 2002. Comparative diagnostic utility of different MR modalities in mild cognitive impairment and Alzheimer’s disease. Dement Geriatr Cogn Disord 14(4): 198–207. Matoba M, Tonami H, Miyaji H, Yokota H, Yamamoto I. 2001. Creutzfeldt–Jakob disease: serial changes on diffusionweighted MRI. J Comput Assist Tomogr 25(2): 274–277. McKhann G, Drachman D, Folstein M, Katzman R, Price D, Stadlan EM. 1984. Clinical diagnosis of Alzheimer’s disease: report of the NINCDS-ADRDA Work Group under the auspices of Department of Health and Human Services Task Force on Alzheimer’s Disease. Neurology 34: 939–944. McMahon PM, Araki SS, Neumann PJ, Harris GJ, Gazelle GS. 2000. Cost-effectiveness of functional imaging tests in the diagnosis of Alzheimer disease. Radiology 217(1): 58–68. Miller DA, Vitti RA, Maslack MM. 1998. The role of 99m-Tc HMPAO SPECT in the diagnosis of Creutzfeldt–Jakob disease. Am J Neuroradiol 19: 454–455. Oppenheim C, Brandel J-P, Hauw J-J, Deslys JP, Fontaine B. 2000. MRI and the second French case of vCJD. Lancet 15: 356(9225): 253–254. Scahill RI, Schott JM, Stevens JM, Rossor MN, Fox NC. 2002. Mapping the evolution of regional atrophy in Alzheimer’s disease: unbiased analysis of fluid-registered serial MRI. Proc Natl Acad Sci USA 99(7): 4703–4707. Seppi K, Schocke MF, Esterhammer R, Kremser C, Brenneis C, Mueller J, Boesch S, Jaschke W, Poewe W, Wenning GK. 2003. Diffusion-weighted imaging discriminates progressive supranuclear palsy from PD, but not from the Parkinson variant of multiple system atrophy. Neurology 60(6): 922–927. van der Flier WM, van den Heuvel DM, WeverlingRijnsburger AW, Bollen EL, Westendorp RG, van Buchem MA, Middelkoop HA. 2002. Magnetization transfer imaging in normal aging, mild cognitive impairment, and Alzheimer’s disease. Ann Neurol 52(1): 62–67. Waldman AD, Jarman P, Merry RTG. 2003. Rapid echoplanar imaging in variant Creutzfeldt–Jakob disease: where speed is of the essence. Neuroradiology 45(8): 528–531. Yoshiura T, Mihara F, Tanaka A, Ogomori K, Ohyagi Y, Taniwaki T, Yamada T, Yamasaki T, Ichimiya A, Kinukawa N, Kuwabara Y, Honda H. 2003. High b value diffusion-weighted imaging is more sensitive to white matter degeneration in Alzheimer’s disease. Neuroimage 20(1): 413–419. Zeidler M, Sellar RJ, Collie DA, et al. 2000. The pulvinar sign on magnetic resonance imaging in variant Creutzfeldt–Jakob disease. Lancet 355: 1412–1418.

32

MR spectroscopy in psychiatry John D. Port Department of Radiology, Mayo Clinic, Rochester, Minnesota, USA

Key points • Major depression may be associated with decreased cellular metabolism and increased membrane turnover. • Abnormal membrane phospholipids metabolism may play a part in bipolar disorders. • Schizophrenia may be associated with abnormal frontal lobe phospholipids metabolism.

Introduction MR spectroscopy (MRS) holds great promise for the diagnosis and management of psychiatric disease. The technique has been available on human MR machines since 1973, but was limited to major medical centers where the experimental spectroscopy sequences and dedicated support personnel were available. As such, MRS was primarily a tool for researchers, not clinicians. That all changed in 1995 when the first commercial clinical spectroscopy software (PROton Brain Exam (PROBE) sequence) was approved by the food and drug administration (FDA). Soon everyone could do MRS, and the number of psychiatric MRS studies proliferated. The increasing interest in psychiatric MRS results from a simple fact: despite decades of research, most anatomical studies of psychiatric disease have yielded low sensitivity and specificity for detecting disease (e.g. Krystal et al., 2001). To date, there are

few anatomical markers of psychiatric disease that are considered “reliable”. This finding correlates well with the clinical impression that psychiatric diseases are primarily functional, caused by chemical imbalances or microscopic structural differences that are not detectable with current technology. While clinical MR examinations of psychiatric patients are occasionally performed, the clinical indication is usually to rule out any organic causes for the patient’s behavioral anomalies, rather than to make a diagnosis of a particular psychiatric disease. Clearly what is needed is a new imaging technique that can reliably and reproducibly determine imaging markers of psychiatric disease. MRS may be that “new” technique, but unfortunately there are many hurdles, which still need to be overcome before new MRS protocols can even be tested. The psychiatric MRS literature to date contains many studies of marginal quality with significant design flaws and other technical shortcomings. The gaps in the literature are so great that statistical tools such as metaanalysis can be performed only in rare conditions (e.g. Yildiz et al., 2001b). As such, the overall value of the existing MRS psychiatric literature base is somewhat limited. A number of major issues need to be considered when reviewing a given psychiatric MRS article to insure its quality and therefore the validity of its findings. These will be discussed in detail below. Then, with these issues in mind, a large subset of psychiatric MRS literature will be reviewed with emphasis on the best studies performed to date. 529

530

John D. Port

Psychiatric MRS issues Subject selection Purity of diagnosis Spatially selective MRS has been available as a research tool since the mid 1980s. At that time, psychiatric disorders were classified using the Diagnostic and Statistical Manual of Mental Disorders (DSM) – third edn (DSM-III, American Psychiatric Association. Task Force on Nomenclature and Statistics, and American Psychiatric Association. Committee on Nomenclature and Statistics, 1980). The DSM-III represented a major change in thinking within the psychiatric community; it was in this version that a multi-axis analysis of psychiatric disease was developed. Since 1980, three additional versions have been released (DSM-III-R, 1987; DSM-IV, 1994; DSM-IV-TR, 2000, American Psychiatric Association. Task Force on Nomenclature and Statistics), each with slightly different criteria for diagnosing and classifying psychiatric disease. These changes complicate comparisons of past and current literature, since one cannot be sure that experimental cohorts truly represent the same psychiatric disease. Several studies have combined patients with different psychiatric diagnoses into a single experimental cohort. Some of these combinations make sense in order to study particular symptoms, such as combining major depression patients and bipolar patients into a cohort in order to study changes in the depressed vs. euthymic states (Frey et al., 1998; Auer et al., 2000). However, some combinations have no apparent logical, for example, combining major depression patients with obsessive-compulsive disorder (OCD) patients (Renshaw et al., 1992). Ideally, study cohorts should consist of patients with a single, well-defined psychiatric disease in order to detect MRS changes specific to that particular diagnosis. Sample size Many of the psychiatric MRS studies performed to date have evaluated only limited numbers of psychiatric patients; as such, the results are effectively pilot studies with little statistical power. Since the magnitude of MRS metabolite differences between psychiatric patients and controls is currently unknown,

a reasonable method to arrive at a sample size estimate for a study is to use the effect size methodology of Cohen (1977). Specifically, in order to detect a “large” effect size (0.8) for a comparison of two independent sample groups (assuming a Gaussian distribution for the measurements), for 80% power at an -level of 0.05, a study must have at least 26 subjects in each group (i.e. a total of 52 subjects). To detect a “medium” effect size (0.5), the study must have at least 64 subjects in each group (128 subjects total). If a case–control study is designed with a matched control group (e.g. controls are age and sex matched to their patient counterparts), the numbers are somewhat better: a study must have at least 15 subjects in each group (30 subjects total) to detect a “large” effect size and 34 in each group (68 subjects total) to detect a “medium” effect size. So assuming that MRS is a powerful technique and will demonstrate large differences in metabolite concentrations or ratios between patients and controls (i.e. a large effect size), a study comparing two independent patient groups should have at least N  26 psychiatric patients and N  26 controls; for the same study conducted in a matched setting, at least N  15 patients and N  15 controls would be needed to have at least 80% power. Age and sex matching It is well established that brain metabolite concentrations change considerably in the developing brain, reaching roughly normal adult patterns by about age 21⁄2 years (van der Knaap et al., 1990; Kreis et al., 1993). It is also fairly well accepted that these metabolite concentrations slowly vary with increasing old age (Charles et al., 1994a; Fukuzako et al., 1997; Ross and Bluml, 2001). A recent study found differences in the rate of decrease of choline (Cho) between elderly men and women, indicating that there also may be sex differences in metabolite concentrations (Sijens et al., 2003). While normal metabolite concentrations for a given age and sex have yet to be determined, by choosing a matched case–control design, such concentration differences are automatically taken into account. If a study does not use such a design, and the age ranges of the cohorts are large, differences in metabolite concentration between the cohorts may be obscured by age-related variations.

MR spectroscopy in psychiatry

Medications Another large subject selection issue is the problem of patient medications. For many of the studies reviewed, cohorts of psychiatric subjects were taking a large variety of medications at many different doses. While the short-term effects of some of these medications have been determined (e.g. the selective serotonin reuptake inhibitors (SSRIs) used for depression block reuptake of serotonin, thereby increasing serotonin concentrations in the synapse), the long-term mechanisms of action on psychiatric symptoms remain unclear. While many of these drugs have sideeffect profiles that occur almost immediately (once blood levels reach therapeutic levels), relief of psychiatric symptoms often occurs 1–2 weeks later. The effects of these medications on MR spectra are unknown. As such, if subjects within a study cohort are taking different doses of different medications, there is the potential to “dilute” any real MRS effects within the cohort, thereby reducing the sensitivity for detecting differences between patients and controls. In an ideal world, patients would be free of medications. There several are clinical situations where this occurs, including new diagnosis of psychiatric illness in a drug-naive patient, or a patient stops taking medications on their own. Unfortunately, such drugnaive patients are rare, even in major clinical psychiatry centers, and even more difficult to recruit due to the presence of psychiatric symptoms. Furthermore, it is ethically difficult to stop a patient’s medications for a study, especially if the patient is gaining some benefit from the drug: there is always the potential for the patient to cause harm to self or others. Several investigators have successfully incorporated “drug washout periods” into their studies, supervising patients during the washout period to insure that patients will be safe. Unfortunately, these washout periods are often only long enough to eliminate the drug from the body (e.g. three to four serum half-lives); the theoretical changes in neuronal function may not have completely reversed by the time MRS is performed. Optimal washout periods have yet to be established. Drug abuse Unfortunately, psychiatric patients often selfmedicate with a variety of illegal substances including

opiates, hallucinogens, and alcohol. While many psychiatric MRS studies attempt to control for such substance abuses, many do not. Since the acute and chronic effects of many of these substances on MR spectra are unknown, the concern is again that such drugs would dilute any real MRS effects, thereby reducing the sensitivity for detecting patientcontrol differences. Study design Voxel placement Since the exact location of the postulated biochemical abnormalities are unknown in psychiatric disease, the “optimal” location for placement of the MRS voxel becomes an issue. Often voxels are placed on regions of the brain thought to be involved in a particular psychiatric disease, as determined by other imaging modalities such as positron emission tomography (PET) or single photon emission computed tomography (SPECT). However, the real possibility exists that the brain abnormality is located in a brain region not sampled by the MRS voxel. As such, the abnormality could be missed altogether. Many studies attempt to sample several (2–4) locations in a given subject’s brain, but in reality it is only a small percentage of the entire brain volume. Newer studies are using the technique of MRS imaging (MRSI) to sample dozens of voxels at a time, thus reducing this potential for sampling errors (Figure 32.1). Another issue with voxel placement relates to semantics. As an example, Ebert et al. (1997) placed voxels on the “right and left anterior cingulate gyrus […], the right striatum, and the right occipital part of the parietal lobes”. Unfortunately this description is ambiguous; from the description it appears that four voxels were placed, when in reality only a single voxel was placed over both cingulate gyri. Furthermore, the exact location of the parietal lobe voxel is unclear. The lack of more specific anatomical description makes comparisons with other studies impossible, since one cannot be sure that the same regions were sampled. A single image indicating the locations of the voxels is indispensable in order to prevent such confusion. On the topic of semantics, often voxel descriptions do not correspond exactly with the regions imaged.

531

532

John D. Port

(a)

Reference

[ml]

[Cr]

[NAA  NAAG]

Fig. 32.1 Example MRSI analysis, with data acquired from a 3 T MRI scanner. (a) The reference image (upper left) shows a T1-weighted reference slice overlaid with the location of the PRESS voxel (white box) and the phase encode grid (gray lines). Quantitative creatine (Cr), myo-inositol (MI) and NAA levels (upper right, lower left, and lower right images, respectively) as determined by linear combination (LCmodel ) are also shown. In these metabolite maps, values are linearly scaled such that yellow voxels have the maximum concentration of a given metabolite, with darker shades representing lesser concentrations. Green voxels have Cramer–Rao bounds greater than 19% and are considered “unreliable” by LC model criteria. Two voxels have been selected: the blue arrow points to a voxel containing predominantly left thalamus, while the red arrow points to a voxel containing predominantly the right thalamus. LC model spectra from each of these voxels are presented in (b) and (c), respectively. Note the quantitative values for each metabolite, shown in the table (upper right). Such quantitative data can be used for statistical comparisons, preferably following tissue volume correction to compensate for any cerebrospinal fluid (CSF) in the voxel.

MR spectroscopy in psychiatry

TR/TE  1500/30; TG  115; R1  11; R2  29; Volume of interest (VOI)  2.03 ml; Calibration curve  1.00 Data of: Mayo Clinic, Rochester, Minnesota

LC model (Version 5.2-2) Copyright: S.W. Provencher.

Wed May 14 01:17:32 2003

Ref.: Magn Reson Med 30: 672–679 (1993). Concentration

0.000 0.075 0.697 0.787 0.341 0.323 0.000 0.186 0.077 0.557 0.818 0.290 0.000 0.000 1.108 0.663

347967

0

55343

(b) Left thalamus (blue arrow)

SD (%) 999 228 11 33 57 59 999 10 84 11 13 37 999 999 6 34

1 info 1 info

/Cr

Metabolite

0.000 Ala 0.107 Asp 1.000 Cr 1.129 GABA 0.489 Gln 0.463 Glu 0.000 GPC 0.267 PC 0.110 Lac 0.799 ml 1.174 NAA 0.417 NAAG 0.000 Scyllo 0.000 Tau 1.591 NAANAAG 0.952 GluGln Diagnostics MYBASI 8 FINOUT 9

Miscellaneous output S/N  7 FWHM  0.098 ppm Data shift  0.019 ppm 5°

Ph:

7°/ppm

0

Input changes LTABLE7 FILTAB’/export/homel/sagedat a/MAYO_CLINIC_BIPOLAR_STUDY /Subject-16X/1/lower-MRSI/1 3_12_0P23552.7.tab‘ FILPS=‘/export/homel/sagedata /MAYO_CLINIC_BIPOLAR_STUDY/ Subject-16X/1/lower-MRSI/13 _12_0P23552.7.PS‘ FILRAW‘/export/homel/sagedat

(c) Right thalamus (red arrow)

TR/TE  1500/30; TG  115; R1  11; R2  29; VOI  2.03 ml; Calibration curve  1.00

526411

0

56751

Data of: Mayo Clinic, Rochester, Minnesota LC model (Version 5.2-2) Copyright: S.W. Provencher.

Wed May 14 01:15:57 2003

Ref.: Magn Reson Med 30: 672– 679 (1993). Concentration

0.000 0.146 0.632 0.319 0.000 0.208 0.198 0.000 0.334 0.396 1.018 0.406 0.000 0.000 1.425 0.208

SD (%) /Cr 999 127 10 71 999 90 9 999 24 17 11 42 999 999 7 90

0.000 0.231 1.000 0.505 0.000 0.329 0.313 0.000 0.528 0.628 1.612 0.644 0.000 0.000 2.256 0.329

Metabolite Ala Asp Cr GABA Gln Glu GPC PC Lac ml NAA NAAG Scyllo Tau NAANAAG GluGln

Diagnostics 1 info 3 info‘s 1 info

MYBASI 8 RFALSI 11 TWOREG 8

Miscellaneous output FWHM  0.081 ppm S/N  9 Data shift  0.000 ppm 11° 4°/ppm Ph:

0

Input changes

Fig. 32.1 (cont.)

LTABLE7 FILTAB’/export/homel/sagedat a/MAYO_CLINIC_BIPOLAR_STUDY /Subject-16X/1/lower-MRSI/1 1_12_0P23552.7.tab‘ FILPS=‘/export/homel/sagedata /MAYO_CLINIC_BIPOLAR_STUDY/ Subject-16X/1/lower-MRSI/11 _12_0P23552.7.PS‘

533

534

John D. Port

For example, prefrontal cortex is a large but specific region of the human brain defined as Brodmann’s areas 9–12 (BA, Brodmann, 1909) located just rostral to premotor cortex (BA 6 and 8). Unfortunately, the popular brain area known as dorsolateral prefrontal cortex (DLPFC) is only loosely defined as the lateral portion of BA 9 coupled with BA 46. Several MRS studies of this area have been performed (e.g. Frey et al., 1998, Winsberg et al., 2000), in which voxels were supposedly placed on DLPFC. However, this is extremely difficult from a technical perspective. For example, like most cortex, DLPFC is located peripherally near the skull and scalp. Voxels cannot be placed too close to the skull or scalp; if they are, susceptibility artifacts and lipids will contaminate the acquired spectra, making it useless. Thus any voxel involving DLPFC is usually placed a little deeper in the brain substance. Furthermore, most MRS voxels are usually cubic or rectangular, while DLPFC is thin and linear. As such, DLPFC voxels realistically only contain a small fraction of DLPFC, with a larger fraction of adjacent subcortical white matter (WM). Thus one is not really sampling much of DLPFC at all. Furthermore, DLPFC is a large region of cortex; different studies might sample different portions of DLPFC, making comparisons difficult at best. Many psychiatric MRS studies do not show examples of their spectra. For deeper brain areas, this is less of an issue, since susceptibility artifacts and lipid contamination are less of a problem in these brain regions. However, for brain regions near the sinuses (such as the anterior temporal, frontopolar, and orbitofrontal cortices), mastoid air cells or brainstem, the very location of the sampled region raises suspicions of susceptibility/lipid contamination. It is difficult to have faith in the MRS findings in these regions, if the spectra cannot be explicitly viewed by the reviewer (e.g. a study of orbitofrontal cortex by Kinney et al., 2000). MRS pulse sequence Specific details of MRS pulse sequences are detailed in Chapter 1. Many of the older psychiatric MRS studies used the early localization techniques such as Depth REsolved Surface coil Spectroscopy (DRESS) or Image Selective In vivo Spectroscopy (ISIS). While effective, these techniques suffer from resonance offset artifacts which cause asymmetric

contamination from tissues outside of the volume of interest (Bottomley, 1991). Since most recent MRS studies use techniques such as point resolved spectroscopy (PRESS) and stimulated echo acquisition mode (STEAM) (or their variants), these artifacts are no longer an issue. However, care should be taken when comparing data from these older studies with new studies; there may be differences due to the technical aspects of the MRS acquisition. Tissue volume correction Most MRS techniques use cubic or rectangular voxels, which do not usually correspond with the curved shapes of the sampled brain regions. As such, a given voxel often samples a combination of cerebrospinal fluid (CSF), gray matter (GM) and WM. Since CSF has no measurable proton MRS metabolites, the presence of a large fraction of CSF within a voxel will artifactually lower the metabolite concentrations in that voxel. Furthermore, metabolite concentrations are different in GM and WM (Pfefferbaum et al., 1999; McLean et al., 2000). While this is less of a problem when ratios are used for metabolite comparisons, if absolute metabolite concentrations are desired, this problem could potentially mask important differences between patients and controls. New post-processing techniques have been developed using anatomical images to take these tissue components in to account (e.g. McLean et al., 2000). It is also possible to incorporate voxel tissue composition data into the statistical analysis (e.g. Auer et al., 2000) or correct metabolite concentrations (e.g. Deicken et al., 2001, Schuff et al., 2001, Villarreal et al., 2002). Furthermore, for MRSI studies, it is possible to shift the acquisition grid to better fill a voxel with a desired brain region. MR spectra quantification The details of MRS analysis have already been presented in Chapter 2 and will not be reviewed here. However, one should be familiar with the different analysis techniques and their limitations as psychiatric MRS studies are reviewed. Specifically, much of the earlier psychiatric MRS research used simple peak heights for analysis of their MRS data. Today, at least for proton spectroscopy, curve-fitting analyses should be performed, and studies that do not use these techniques should be viewed with skepticism.

MR spectroscopy in psychiatry

The sensitivity and specificity of analyses which depend entirely on metabolite ratios are vulnerable to changes in the metabolite used as reference, commonly creatine (Cr). Differences in quantification may explain in part the discrepancies between different studies of the same brain region.

Review of psychiatric MRS studies With these issues in mind, a critical review of the literature shows that over the past two decades there have been several excellent MRS studies performed on psychiatric cohorts. Such studies have been performed for many of the diagnoses listed in the current DSM-IV-TR (Table 32.1) and past versions of the DSM. It is impossible to cover all of literature associated with these diagnoses within the bounds of this brief chapter. Therefore, the remainder of the chapter will focus primarily on four major diagnostic areas: affective disorders including major depressive disorder and bipolar disorder, anxiety disorders including OCD, panic disorder, and post-traumatic stress disorder (PTSD), substance-related disorders including alcohol abuse, and schizophrenia. Only articles published in peer-reviewed journals are presented; published abstracts, while valuable, do not provide enough information about the methods and results to be evaluated critically. Affective disorders Major depression Major depression is among the most common of the psychiatric diseases. More than 8 million people in the US suffer through a major depressive episode each year, with an overall disease prevalence of 16% of the general population (Kessler et al., 1994, 1997). Overall, patients experience sleep, mood, appetite, and cognitive disturbances, with specific symptoms including feelings of guilt, hopelessness, helplessness, and despair. The disease has considerable mortality risk; an unfortunate fraction of people with major depression die each year from suicide. Over the past decade a number of MRS studies have examined several brain regions thought to be involved in depression, namely, the basal ganglia and various cortical areas (cf. Table 32.2). In several

Table 32.1. Selected summary of DSM-IV-TR classification system Childhood disorders Attention-deficit and disruptive behavior disorders (attention-deficit/hyperactivity, conduct, oppositional defiant) Learning disorders (mathematics, reading, written expression) Oppositional defiant disorder Pervasive developmental disorders (autism, Asperger’s, Rett’s, childhood disintegrative) Cognitive disorders Amnestic disorders Delirium disorders Dementia (Alzheimer’s, Creutzfeldt–Jakob, Pick’s, Huntington’s, Parkinson’s, vascular) Substance-related disorders Alcohol, amphetamine, caffeine, cannabis, cocaine, hallucinogen, inhalant, nicotine, opioid, and sedative related disorders Psychotic disorders Brief psychotic disorder Delusional disorder Schizoaffective disorder Schizophreniform disorder Schizophrenia (catatonic, disorganized, paranoid, residual, undifferentiated) Shared psychotic disorder Mood disorders Bipolar disorders (bipolar I, bipolar II, cyclothymia) Depressive disorders (dysthymic disorder, major depressive disorder) Anxiety disorders Acute stress disorder Agoraphobia without panic disorder Generalized anxiety disorder OCD Panic disorder (with or without agoraphobia) PTSD Social phobia Eating disorders Anorexia nervosa Bulimia nervosa

of the brain regions studied, Cho levels are abnormal, either elevated or depressed, depending on the study. Furthermore, these Cho levels tend to return to normal following treatment. Preliminary 31P studies also find that -nucleoside triphosphate (NTP)

535

Design Case–control Case–control

Pre–post-rx

Case–control

Case–control Case–control

Pre–post-rx

Case–control

Pre–post-rx

Case–control

Cross-section

Reference

Sharma et al., 1992

Charles et al., 1994b

Renshaw et al., 1997

Frey et al., 1998

Hamakawa et al., 1998

Sanacora et al., 1999a

Sonawalla et al., 1999

Auer et al., 2000

Ende et al., 2000

Steingard et al., 2000

Murata et al., 2001

Table 32.2. Major depression studies

47S

17S 28N

17S 30N

19S 18N

15S

14S 18N

22S 20N

22S 22N

41S 22N

7S 10N

1S 9N

Subjects

Left frontal WM

Orbitofrontal cortex

Both hippocampi

Anterior cingulate parietal WM

Left striatum

Midline occipital cortex

Left basal ganglia

Both DLPFC

Left striatum

Basal ganglia and thalamus

Left head caudate

Region(s) studied

H PRESS

H STEAM

H PRESS-MRSI

H PRESS

H STEAM

H ISIS-DANTE

H STEAM

H STEAM

H STEAM

H STEAM

H STEAM

Nucleus

8 cc

3.4 cc

2.4 cc

6.8–12 cc

8 cc

13.5 cc

27 cc

12 cc

8 cc

27 cc

16 cc

Voxel size

Quant Loc Spectra VTS None

Loc Spectra AgeM Quant VTS  Loc Spectra

Loc Spectra  Spectra NoMeds Quant NoMeds

Spectra

None: N too small, descriptive

Loc Spectra Spectra NoMeds AgeM Loc Spectra NoMeds Loc

NAA/Cr late-onset MD  early-onset MD

Cho/Cr and Cho/NAA MD  normals

[Cho] pre-rx MD  normals; [Cho] after five or more ECT rx MD  pre-rx MD

Cho/Cr true responder MD  pre-rx; Cho/Cr placebo and non-responder MD  pre-rx [Glx] depressed (MD  BP)  normals

[GABA] MD  normals

R frontal mI/Cr treated depressed (MD  BP)  normals; L frontal mI/Cr treated (MD  BP)  untreated (MD  BP) Cho/NAA MD  normals

Cho/Cr MD  normals; Cho/Cr MD post-rx  pre-rx; NAA/Cho MD post-rx  pre-rx Cho/Cr MD  normals

Significant findings

Quality info

Case–control Case–control Case–control

Pre–post-rx

Cross-section

Cross-section

Kato et al., 1992 Moore et al., 1997 Volz et al., 1998

Renshaw et al., 2001

Riedl et al., 1997

Renshaw et al., 1992

3S

12S

12S 17N

12S 10N 35S 18N 14S 8N

11S 11N

Whole brain

Whole brain

Left striatum Both basal ganglia

Frontal lobes Both basal ganglia Frontal lobes

Both amygdala

F FID

Li FID

H STEAM; P ISIS

P DRESS P ISIS P ISIS

H-MRSI

–

–

8 ccH; 45 ccP

– 45 cc 39 cc

Spectra None

Loc

Spectra AgeM Spectra NoMeds

Spectra AgeM SexM Spectra NoMeds Loc

[fluoxetine or norfluoxetine] (MD  OCD) in brain 2.6 times  plasma

-NTP MD  normals -NTP/metab female responder MD  non-responder MD Good correlation between daily Li dose, plasma Li and brain Li in long-term patients

pH variability MD  BP -NTP MD  normals PME MD  normals; -NTP MD  normals

L amygdala Cho/Cr MD  normals

Subject column: S, number of subjects studied with the specified psychiatric illness; N, number of normal people studied. Nucleus column: ISIS, Image Selective In vivo Spectroscopy; DANTE, Delays Alternating with Nutations for Tailored Excitation; DRESS, Depth-REsolved Surface coil Spectroscopy; MPCSI, MultiPlanar Chemical Shift Imaging; RAA, Recently Abstinent Alcoholics. Quality info column (the more qualifiers, the better the study): Loc, the location of the voxel is displayed on an image; Spectra, a representative spectrum is displayed; NoMeds, subjects are medication naive or have gone through a medication washout period; AgeM, patients are age matched with controls; SexM, patients are sex matched with controls; Quant, a quantitative curve-fitting analysis has been performed; VTS, voxel tissue segmentation performed; TVC, tissue volume correction performed (corrects metabolite values based on the voxel’s GM and WM composition). Significant findings column: R, right; L, left; H, proton; P, phosphorus; [x], concentration of a given metabolite; FID, free induction decay; MD, major depression; NAA, N-acetyl aspartate; Cho, choline; PME, phosphomonoesterases; PDE, phosphodiesterases; PCr, phosphocreatine.

Case–control

Kusumakar et al., 2001

538

John D. Port

levels are decreased in depressed patients relative to controls. These data suggest that cellular energetics are decreased and membrane turnover increased in patients with depression. Clearly more work is needed on a well-defined cohort of major depression patients in order to determine more consistent imaging markers of depression. Bipolar disorder The lifetime prevalence of bipolar disease is estimated at 1–1.5% of the US population, or approximately 3 million people. Bipolar patients suffer considerable morbidity and mortality, and overall the disease imposes a significant public health problem. Bipolar patients are at greatly increased risk of death from suicide (Michels and Marzuk, 1993a, 1993b). The diagnosis of bipolar disease can be clinically difficult to establish; at least two episodes are necessary to confirm the diagnosis, and individual episodes can often be clinically ambiguous. Many bipolar MRS studies have been performed over the last decade (Table 32.3). These have found metabolic abnormalities located in the frontal lobes, temporal lobes, basal ganglia, thalami, and anterior cingulate cortex. Specifically, Kato, Deicken and their teams have found abnormal phosphomonoesterase (PME) levels in the frontal and temporal lobes of bipolar patients relative to normal controls, and various investigators have found abnormal Cho/Cr levels in the basal ganglia and anterior cingulate cortex of bipolar patients relative to normals. This suggests that bipolar patients have abnormal membrane phospholipid metabolism. Other studies demonstrated decreased 31P phosphocreatine (PCr) and 1H Cr levels in the frontal lobes supporting the hypothesis of decreased frontal lobe energetics.Unfortunately the majority of these studies suffer from the confounding issues described above; study results are complicated by variable diagnoses (bipolar I, II and nitric oxide synthase (NOS)) and patient states (manic, euthymic, depressed), medications and relatively small sample sizes. Despite these limitations, Yildiz et al. (2001a) were able to perform a metaanalysis of the first eight phosphorus MRS studies in Table 32.3, concluding that PME levels in euthymic bipolars are significantly less than in normals; PME

levels in depressed bipolars are significantly greater than euthymic bipolars, lending further support to the idea of abnormal membrane phospholipid metabolism in bipolar disease. Lithium is an effective medication for bipolar disease. Unfortunately, lithium treatment has a narrow therapeutic window, and therapeutic monitoring is needed in order to prevent toxic side-effects during treatment. Furthermore, some patients do not experience symptom relief despite therapeutic serum levels. In the past decade, a number of studies have utilized 7Li-MR to compare brain lithium concentration in bipolar patients with other measures such as serum, red blood cell, white blood cell, CSF, and even muscle lithium concentrations; these studies are nicely described in a recent review by Soares et al. (2001), coupled with an additional recent study by Moore et al. (2002). Only two studies to date have correlated brain lithium concentrations with symptoms and side-effects (Table 32.3). These relatively small studies show small but significant relationships between brain lithium concentration, sideeffects and response to treatment. Clearly more work is needed to assess the value of lithium MRS, with a well-planned pre- and post-treatment study design on a large number of lithium-treated bipolar patients. Lithium has many biological effects, but is thought to act on the brain primarily by blocking the enzyme inositol monophosphatase. This enzyme is responsible for the synthesis of myo-inositol (mI) an important precursor in the phosphatidylinositol intracellular second messenger system. As such, patients taking lithium should have decreased mI levels; this theory is known as the inositol-depletion hypothesis, and is nicely described elsewhere (Atack et al., 1995; Belmaker et al., 1996; Soares and Mallinger, 1997). Two recent studies have found preliminary evidence supporting this hypothesis; both Moore et al. (1999) and Davanzo et al. (2001) have found decreased mI/Cr or [mI] in bipolar patients following commencement of lithium therapy. General affective disorder Occasionally studies are performed on patients with “affective disorder”, combining cohorts of unipolar and bipolar patients in order to increase the study sample size. While potentially reducing the power of

Design Case–control

Case–control

Case–control

Case–control

Case–control Case–control

Pre–post-rx

Case–control

Pre–post-rx

Pre–post-rx

Reference

Sharma et al., 1992

Stoll et al., 1992

Kato et al., 1996b

Hamakawa et al., 1998

Ohara et al., 1998

Hamakawa et al., 1999

Moore et al., 1999

Castillo et al., 2000

Moore et al., 2000b

Moore et al., 2000a

Table 32.3. Bipolar studies

9S 14N

12S 9N

10S 10N

12S

23S 20N

10S 10N

18S 20N

19S 19N

7S 6N

4S 9N

Subjects

H STEAM

H STEAM

H PRESS

H STEAM

H STEAM

H STEAM

H STEAM

Nucleus

Anterior cingulate cortex

R frontal, L temporal, C occipital, L parietal

H STEAM-MRSI

H STEAM

Both frontal, temporal H PRESS

R frontal, L temporal, C occipital, L parietal

Both medial frontal cortex

Both lentiform nuclei

Left basal ganglia

Left basal ganglia

Parietal lobes

Left head caudate

Region(s) studied

2 cc

8 cc

8 cc or 27 cc

8 cc

16 cc

8 cc

27 cc

27 cc

27 cc

16 cc

Voxel size

VTS

Spectra Quant VTS Loc Spectra

Spectra NoMeds Loc

Spectra Quant VTS Loc

SexM Loc Spectra AgeM Loc Spectra Quant Loc Spectra Loc Spectra Quant Loc

(cont.)

R Cho/Cr BP  normals; R Cho/Cr BP without antidepress  BP on antidepress and normals

Summed region [NAA] post-rx  pre-rx

Frontal and temporal Glx/Cr BP  normals; lipids in frontal lobes of BP

R frontal [mI] early post-rx and late post-rx  pre-rx

L [Cr] depressed BP  euthymic BP; R [Cr] male BP  female BP

R NAA/Cr BP  L BP

[Cho] BP  normals; Cho/Cr and Cho/NAA BP  normals

Cho/Cr BP  normals

None: N too small; TREND: NAA/Cr BP on lithium  normals; TREND: Cho/Cr and mI/Cr BP  normals None; no significant difference in Cho between BP and normals

Loc Spectra

Spectra

Significant findings

Quality info

Design Case–control

Pre–post-rx

Case–control

Case–control Case–control

Pre–post-rx

Case–control

Case–control

Reference

Winsberg et al., 2000

Davanzo et al., 2001

Deicken et al., 2001

Kato et al., 1991

Kato et al., 1992

Kato et al., 1993

Kato et al., 1994b

Kato et al., 1994c

Table 32.3. (cont.)

29S 59N

40S 60N

17S 17N

10S 10N

9S 9N

15S 15N

11S 11N

20S 20N

Subjects

Frontal lobes

Frontal lobes

Frontal lobes

Frontal lobes

Frontal lobes

Both thalami

Anterior cingulate cortex

Both DLPFC

Region(s) studied

P DRESS

P DRESS

P DRESS

P DRESS

P DRESS

H MM1-MRSI

H PRESS

H PRESS

Nucleus

–

–

–

–

–

1.5 cc

8 cc

8 cc

Voxel size

Spectra Loc

SexM Loc

Loc Spectra AgeM

Spectra

SexM Quant VTS TVC Spectra

R and L NAA/Cr BP  normals

Loc Spectra NoMeds AgeM SexM VTS Loc Spectra AgeM SexM Quant Loc Spectra

PCr BP-II all states  normals;

pH BP  normals; PME female BP-I  normals

PME manic BP  euthymic BP and normals PME and pH depressed BP  euthymic BP; PME euthymic BP  normals and euthymic MD; PCr severe depressed BP  mild depressed BP PME manic BP  euthymic BP; PME euthymic BP  normals; pH euthymic BP  normals and manic BP

R and L [NAA] and [Cr] BP  normals; L [NAA] BP and normals  R [NAA] BP and normals

mI/Cr BP post-Li  pre-Li

Significant findings

Quality info

Case–control Case–control

Case–control Pre–post-rx

Cross-section

Deicken et al., 1995b

Deicken et al., 1995a

Kato et al., 1998

Kato et al., 1994a

Kato et al., 1996a

BP, bipolar. For abbreviations see footnote of Table 32.2.

Case–control

Kato et al., 1995

17S

14S

7S 60N

12S 16N

12S 14N

25S 21N

Frontal lobes

Frontal lobes

Frontal lobes

Both medial temporal lobes Both anterior frontal lobes

Both frontal lobes

Li DRESS

Li FID

P DRESS

P SE-MRSI

P SE-MRSI

–

–

–

25 cc

25 cc

P phase encoded –

None

Loc NoMeds Loc NoMeds SexM Spectra NoMeds None

Loc Spectra

Spectra

significant correlation between improvement in mania scores and brain lithium concentration Sun et al., (2003) with hand tremor  Sun et al., (2003) without hand tremor

R and L PME BP  normals; R and L PDE BP  normals; R PCr BP  L PCr BP pH BP  normals

PME hypomanic, depressed BP-II  normals; PME euthymic BP-I  depressed BP-I and normals; pH euthymic BP-I  normals L PCr depressed BP  normals; R PCr manic and euthymic BP  normals; L PME depressed BP  normals; R PCr euthymic BP  normals R and L PME BP  normals;

542

John D. Port

the study to detect differences between normals and controls, this technique can yield interesting results regarding trait vs. state issues. Specifically, one can look for an imaging marker of the depressed “state” by combining cohorts of patients with different “traits” such as bipolar disorder or major depression, both of which have depression as a symptom. An excellent study by Auer et al. (2000) found decreased Glx levels (a combination of glutamate (Glo)  glutamine (Gln) in depressed patients with unipolar or bipolar diagnoses, suggesting that Glx may serve as a marker of the depressed state. This study is notable for meticulous technique including metabolite quantification and voxel tissue segmentation, two techniques, which possibly increased the power of the study enough to detect differences between patients and controls. Anxiety disorders Anxiety disorders are also quite common, with 15% of the population of the US affected by an anxiety disorder during their lifetime (Kessler et al., 1994). The DSM-IV-TR shows many subtypes of anxiety disorder (Table 32.1), however, many have specific diagnostic criteria that allow for proper classification. Several anxiety disorders have been relatively well studied; these are expounded below. Panic disorder Panic disorder affects one out of 75 people at some point in their lifetime. It usually appears in adolescence or early adulthood, and is characterized by panic attacks, sudden surges of overwhelming fear that occur without warning or obvious cause. The fear is out of proportion to the actual situation (often it is completely unrelated), and passes within a few minutes. Patient’s experience typical “fight or flight” symptoms, including racing heartbeat, difficulty breathing, dizziness, trembling, and nausea. MRS studies of panic disorder are shown in Table 32.4. It has been known for many years that panic attacks can be provoked by intravenous infusion of lactate (Lac) (Pitts and McClure, 1967). Panic disorder patients appear to produce excess blood Lac, suggesting some underlying metabolic or physiological abnormality, which may be the cause of this

condition. Dager and his group have taken advantage of this fact to induce panic attacks in panic disorder patients and controls using hyperventilation and Lac injections (Dager et al., 1994, 1995, 1997, 1999). They found that N-acetylaspartate (NAA)/Lac levels increased more for panic disorder patients than controls, confirming the above hypothesis. However, treated patients experienced the same changes in NAA/Lac levels without experiencing any symptoms. As such, the NAA/Lac ratio represents a reliable trait marker rather than a state marker, and can be helpful for the diagnosis of panic disorder. While interesting, this work has not been particularly fruitful for determining the underlying mechanisms of panic disorder. Goddard et al. (2001) have recently performed an excellent study demonstrating decreased -amino-butyric acid (GABA) levels in the occipital cortex of panic disorder patients relative to controls. This finding is potentially exciting since panic disorder treatments often involve benzodiazepines, which interact with the GABA system. Much interesting work remains to be performed localizing brain GABA distribution and monitoring response to therapy. Obsessional compulsive disorder Obsessional compulsive disorder (OCD) is characterized by intrusive unwanted thoughts, ideas, or images that are distressing (obsessions) and urges to perform ritualistic behaviors or mental acts (compulsions) to reduce this distress. The lifetime prevalence of this anxiety disorder is estimated to be 2–3% around the world (Weissman et al., 1994). OCD is associated with impairment in occupational, academic, and social functioning (Koran et al., 1996) and can sometimes involve self-injury, such as skin damage from excessive hand washing. Although symptoms tend to wax and wane through the course of the disorder, OCD symptoms rarely remit spontaneously. Several MRS studies have been performed on OCD cohorts over the past decade (Table 32.5). Unfortunately, there are not enough good quality studies to draw any conclusions about spectroscopic findings in OCD. However, an exciting, potentially important finding was initially observed by Moore et al. (1998) and further examined by

Case–control

Case–control

Case–control

Case–control

Case–control

Pre–post-rx

Case–control

Case–control

Dager et al., 1994

Dager et al., 1995

Dager et al., 1997

Dager et al., 1999

Goddard et al., 2001

Layton et al., 2001

Massana et al., 2002

Shioiri et al., 1996

18S 18N

11S 11N

6S

14S 14N

15S 10N

13S 10N

7S 7N

8S 8N

Subjects

PD, panic disorder. For abbreviations see footnote of Table 32.2.

Design

Reference

Table 32.4. Panic disorder studies

H PEPSI MRSI

H PRESS

H PRESS

H PRESS

Nucleus

Frontal lobes Hemispheres

Medial PFC right amygdala/hippo

Slice at level of lateral ventricles

Spectra NoMeds AgeM SexM Loc Spectra AgeM 12 cc PFC

SexM

Loc

NoMeds Spectra NoMeds AgeM SexM NoMeds

Spectra

Spectra NoMeds Loc

Spectra NoMeds Loc

PCr L PD  R panic

Lac/NAA PD rises equally during Lac infusion on/off gabapentin; symptoms were less with rx R amygdala and hippocampal [Cr] PD  normals

[GABA] PD  normals

Global Lac/NAA PD rises  normals; No lateralization or focal regions of increased Lac/NAA

Lac/NAA PD during Lac infusion rises  normals

Lac/NAA PD during hyperventilation rises  normals

Lac/NAA PD during Lac infusion rises  normals and medicated Lac non-responders

Loc

Spectra Loc

Significant findings

Quality info

8 cc amyg

1 cc

13.5 cc

1 cc

27 cc

27 cc

27 cc

Voxel size

P DRESS, 1D-CSI –

H PRESS

H PEPSI MRSI

Both occipital cortices H ISIS-DANTE

Slice at level of lateral ventricles

Left insular cortex

Left insular cortex

Left insular cortex

Region(s) studied

Design Case–control Case–control

Pre–post-rx

Case–control

Case–control

Pre–post-rx

Pre–post-rx

Reference

Ebert et al., 1997

Bartha et al., 1998

Moore et al., 1998

Ohara et al., 1999

Fitzgerald et al., 2000

Rosenberg et al., 2000

Bolton et al., 2001

1S

11S 11N

11S 11N

12S 12N

1S

13S 13N

12S 6N

Subjects

Table 32.5. Obsessional compulsive disorder studies

Left head caudate

Left head caudate

Both thalami

Both lentiform

Left head caudate

Anterior cingulate right striatum Left head striatum

Region(s) studied

H PRESS

H PRESS

H MPCSI/MRSI

H PRESS

H PRESS

H STEAM

H PRESS

Nucleus

0.7 cc

0.7 cc

0.8 cc

8 cc

0.7 cc

4.5 cc

2 cc

Voxel size

Spectra NoMeds Quant

Spectra NoMeds AgeM SexM Quant Loc Spectra NoMeds AgeM SexM Quant Loc

Spectra AgeM SexM Loc

[NAA] OCD  normals

Loc Spectra NoMeds AgeM SexM Quant Loc Spectra NoMeds Quant Loc

[Glx] pre-rx OCD  [Glx] during-rx  [Glx] post-rx

[Glx] pre-rx OCD  normals; [Glx] decreased to normal post-rx

NAA/Cr L lateral thalamus OCD  normals; other significant non-standard ratios

None: no differences found between patients and controls

[Glx] pre-rx OCD  [Glx] during-rx OCD

NAA/Cr R striatum OCD  normals

Significant findings

None

Quality info

Cross-section

Pre–post-rx

Renshaw et al., 1992

Strauss et al., 1997

For abbreviations see footnote of Table 32.2.

Case–control

Rosenberg et al., 2001

7S

3S

11S 11N

Whole brain

Whole brain

Both thalami

F FID

F FID

H MPCSI/MRSI

–

–

0.8 cc

NoMeds

Spectra

Spectra NoMeds AgeM SexM Quant None

Loc

[fluoxetine or norfluoxetine] (MD  OCD) in brain 2.6 times  plasma Reached brain fluvoxamine steady state in 30 days Brain 24 times  plasma

Reanalysis of fitzgerald 2000: [Cho] R and L medial thalami OCD  normals

546

John D. Port

Bolton et al. (2001) in two excellent studies of single OCD patients. These studies performed serial MRS on the left caudate head before and after treatment with paroxetine, and found that Glx concentrations fell dramatically during treatment. Bolton further demonstrated that Glx remained low once medication was discontinued. The implication is that there is a reversible glutamatergically mediated dysfunction of the caudate nucleus in OCD, which potentially can be used to diagnose OCD and monitor treatments. Further research using a longitudinal case–control design on a large group of patients will be most helpful to verify this interesting finding. A word must be said about the methodology in one OCD study by Fitzgerald et al. (2000). This study was methodologically sound from a patient selection, MRS acquisition and quantification perspective. However, results were expressed as non-standard ratios (e.g. NAA/(Cho  Cr) and NAA/Cho); some of these ratios turned out to be statistically significant, and a long discussion of their significance was presented. The interpretation of these ratios is difficult at best; one gets the sense that the authors tried several different combinations of ratios in an effort to find a statistically significant result to put into their study. The only interesting statistically significant item is barely mentioned in the discussion, i.e. the increased NAA/Cr in the left lateral thalamus in OCD patients relative to controls. In the author’s defense, they subsequently performed a quantitative analysis of the same data (Rosenberg et al., 2001) using a standardized established technique, finding that Cho levels were elevated in the medial thalami of OCD patients relative to controls. This example demonstrates how different analysis techniques generate different “significant” findings. Post-traumatic stress disorder Post-traumatic stress disorder (PTSD) develops following a frightening or traumatic, often life-threatening event. These events can be quite diverse, including combat experiences (shell-shock), child abuse, rape, natural disasters, and accidents. PTSD patients typically experience strong feelings of horror, helplessness or fear following the event, repeatedly reliving the traumatic event in the form of flashbacks or nightmares. Often flashbacks are triggered by particular

events, and PTSD patients will often avoid these situations, occasionally severely restricting behavior. Other symptoms include sleep problems, depression, emotional numbing (anhedonia), and substance abuse (self-medication). Table 32.6 shows the five MRS studies performed to date in PTSD patients. The general trend is that NAA levels are decreased in various brain regions in PTSD patients relative to normals. However, as expounded by several authors, it is unclear whether this is apparent neuronal loss is due to coexistent substance abuse, or to PTSD itself. Further work needs to be performed in this field. Substance-related disorders Substance-related disorders are the most prevalent of psychiatric diseases, with almost 27% of the population of the US estimated to have substance abuse with or without substance dependence during their lifetime. However, such disorders are often difficult to study in their pure form; people who have substance abuse disorders also have other comorbid psychiatric disease (Kessler et al., 1994). Of the substance-related disorders, alcoholism has been the disease most studied using MRS; the current literature is discussed below. Alcohol abuse/dependence The lifetime prevalence of alcohol dependence is 14.1%, second only to major depression in terms of pervasiveness in the general population (Kessler et al., 1994). Because of the large number of people suffering from alcohol dependence, it has been possible to find large cohorts of alcohol-dependent subjects who do not have other comorbid psychiatric illness. Therefore, the samples of patients are relatively pure in terms of diagnosis, as the definition of alcohol dependence has remained relatively stable across the DSM versions. Early MRS studies of alcohol dependence were focused primarily on measuring brain ethanol concentrations. Ethanol has a single dominant peak at 1.2 ppm, as detected using proton MRS. Therefore, using phantom replacement methods for “calibration,” brain ethanol levels can, in theory, be measured and compared (Hanstock et al., 1990;

Case–control

Case–control

Pre–post-rx Case–control

Case–control

Freeman et al., 1998

De Bellis et al., 2000

De Bellis et al., 2001

Schuff et al., 2001

Villarreal et al., 2002

For abbreviations see footnote of Table 32.2.

Design

Reference

8S 5N

18S 19N

1S

11S 11N

21S 8N

Subjects

Table 32.6. Post-traumatic stress disorder studies

Both hippocampi occipital WM

Both hippocampi

Anterior cingulate

Anterior cingulate

Both hippocampi

Region(s) studied

1.1 cc

3 cc

3 cc

12 cc

Voxel size

H PRESS, STEAM 9.3 cc

H PRESS MRSI

H STEAM

H STEAM

H STEAM

Nucleus

AgeM SexM Quant VTS TVC

Spectra

SexM Quant VTS TVC Loc

Spectra

None: limited due to small sample size. Trend: L hippocampal [NAA] PTSD  normals

[NAA] PTSD  normals; R hippocampal [Cr] PTSD  normals; hippocampal volumes in PTSD same as normals

NAA/Cr post-rx PTSD  pre-rx

NAA/Cr PTSD  normals

R NAA/Cr PTSD  L PTSD; R NAA/Cr and L Cho/Cr PTSD  normals

Loc Spectra SexM AgeM SexM Quant NoMeds Quant Loc

Significant findings

Quality info

548

John D. Port

Mendelson et al., 1990; Spielman et al., 1993). Chiu et al. (1994) used these methods to find that brain-to-blood ethanol ratios were significantly higher in heavy drinkers (alcohol tolerant) than in occasional drinkers, suggesting that long-term ethanol use alters cell membranes in brain tissues. While interesting, these studies will not be considered further here. More recent MRS studies of alcohol dependence have focused on metabolic changes that occur in the brains of recently abstinent alcoholics (RAA) (Table 32.7). Anatomical MR imaging (MRI) has shown brain volume reductions in alcohol-dependent individuals, most pronounced in the cerebellar vermis but also in the total amount of GM and WM (for a summary, cf. Schweinsburg et al., 2001). Therefore, these have been the areas most studied using MRS techniques. In general, these studies have found decreased NAA in the cerebellums of recently abstinent alcoholics relative to normal subjects, and these low NAA levels have increased to near normal as the period of abstinence increases. There is considerable disagreement about NAA levels in the frontal lobes, with some studies finding decreases in recently abstinent alcoholics (Jagannathan et al., 1996; Bendszus et al., 2001; Schweinsburg et al., 2001, 2003) and others finding no differences (O’Neill et al., 2001; Parks et al., 2002). Two studies from Schweinsburg have found increased mI levels in recently abstinent alcoholics subjects (Schweinsburg et al., 2000, 2001); this data should be replicated by others. Phosphorus MRS studies have suggested membrane abnormalities in alcoholics (Meyerhoff et al., 1995; Estilaei et al., 2001a, 2001b). One exciting preliminary study measured GABA levels in the brains of alcoholics (Behar et al., 1999); more work measuring neurotransmitters such as GABA and Glu will help confirm this finding. Schizophrenia Schizophrenia is perhaps one of the most disturbing psychiatric diseases, affecting approximately 1% of the general population (Kasai et al., 2002). It is composed of a heterogeneous group of illnesses, diagnosed according to the presence or absence of positive symptoms (hallucinations, delusions, thought disorder) and negative symptoms (flattened

affect, social withdrawal, apathy). Schizophrenia starts insidiously in adolescence or early adulthood, and usually progresses into a horribly disabling chronic mental illness. Previously successful patients often lose their jobs, friends, spouses and homes, and many end up living on the streets or in shelters. Since 1990, there have been over 100 MRS studies of schizophrenic patients (Lyoo and Renshaw, 2002). It would be impossible to cover all of them with the above level of detail given the limitations of this brief chapter. However, others have performed such extensive reviews; selected review articles of schizophrenia are listed in Table 32.8, chosen to emphasize different aspects of this complex disease. Several consistent findings have emerged from this body of literature. 31P MRS has repeatedly found decreased PME and increased phosphodiesterase (PDE) in the frontal lobes of schizophrenics relative to normals (Keshavan et al., 2000). It appears that the increased PDE levels are only found in the early phases of schizophrenia, not in chronic cases (Stanley et al., 1994, 1995). This suggests that schizophrenics have abnormal frontal lobe membrane phospholipid metabolism, with decreased synthesis and increased breakdown of neuronal membrane phospholipids. 1H MRS has also consistently found decreased NAA levels in the frontal and temporal lobes of schizophrenics (Keshavan et al., 2000). Both of these findings support the synaptic pruning hypothesis of schizophrenia which states that there is “overpruning” of frontal lobe synapses in late childhood and adolescence in schizophrenics (Feinberg, 1982, 1990; Pettegrew et al., 1993). This synaptic pruning eventually damages the brain sufficiently for schizophrenic symptoms to occur. Unfortunately, much of the schizophrenia MRS literature suffers from the flaws described earlier in this chapter. Study results are complicated by heterogeneity of diagnosis; until recently, many of the schizophrenic subtypes (e.g. catatonic, disorganized, paranoid, residual, undifferentiated) were not defined, ignored or simply combined into a single patient cohort. Diverse medications have been less of a problem, with many of the studies evaluating first-episode, medication-naive (FEMN) schizophrenics (Stanley, 2002). Sample sizes have been better than in studies of other psychiatric illness.

Case–control

Case–control

Case–control

Case–control

Schweinsburg et al., 2001

O’Neill et al., 2001

Bendszus et al., 2001

S10 N10 11S 10N

Case–control Case–control

Schweinsburg et al., 2000

10S 27N

Case–control

Jagannathan et al., 1996 Behar et al., 1999 Seitz et al., 1999

R parietal WM

R frontal WM

Anterior cingulate Anterior centrum

R thalamus

Cerebellum frontal lobe thalamus Occipital lobe Cerebellar vermis

Cerebellar vermis

Region(s) studied

17S 12N L cerebellum

Frontal GM

8S 121“N” Various GM, WM

37S 15N

9S 5N

10S 9N

Pre–post-detox

Martin et al., 1995

Subjects

Design

Reference

Table 32.7. Alcohol abuse studies

H PRESS

H PRESS MRSI

H PRESS

H PRESS

H ISIS H STEAM

H STEAM

H PRESS

Nucleus

8 cc

0.96 cc

8 cc

8 cc 8 cc

3.4 cc

13.5 cc 8 cc

8 cc

8 cc

Voxel size

Quant

Spectra

VTS TVC Loc

TVC Quant

Spectra Quant

Spectra SexM Quant TVC VTS Loc

Loc

Spectra

NAA/Cr frontal GM and cerebellum early RAA  normals NAA/Cr frontal GM and cerebellum late RAA  early RAA Cho/Cr cerebellum early RAA  normals Cho/Cr cerebellum late RAA  early RAA (cont.)

None: no significant region-specific differences in metabolite values

[mI] frontalparietal WM RAA  normals [NAA] frontal WM RAA  normals [NAA] frontalparietal WM RAA  normals

NAA/Cr and NAA/Cho all areas in alcoholics  normals GABA/Cr alcoholics  normals NAA/Cr and Cho/Cr vermis RAA  normals [mI] thalamus RAA  long-term abstinent alcoholics and normals [mI] ant cingulate RAA  normals

Cho/NAA vermis with abstinence  before abstinence

Loc Spectra Spectra

Significant findings

Quality info

Pre–post-detox

Case–control

Case–control

Case–control

Case–control

Parks et al., 2002

Schweinsburg et al., 2003

Meyerhoff et al., 1995

Estilaei et al., 2001b

Estilaei et al., 2001a

For abbreviations see footnote of Table 32.2.

Design

Reference

Table 32.7. (cont.)

36S 14N

13S 17N

92S 8N

25S 25N

31S 12N

Subjects

Centrum semiovale

Centrum semiovale

“GM”

“WM”

R frontal WM

Anterior cingulate

Cerebellar vermis

R frontal WM

Region(s) studied

P ISIS

P ISIS

P ISIS

H PRESS

H PRESS

Nucleus

168 cc

168 cc

90 cc

100 cc

8 cc

8 cc

Voxel size

Spectra Quant

Spectra Quant Loc

Spectra Quant Loc

TVC Loc

Quant

Spectra

Quant VTS TVC Loc

No significant difference in the broad component integral between light/non-drinkers and abstinent alcoholics

Broad component integral in heavy drinkers  light/non-drinkers

[PDE] and [PCr] WM HIV-negative alcoholics  normals

[NAA] frontal WM M  F alcoholics  M  F normals [NAA] frontal WM F alcoholics  M alcoholics [NAA] ant cingulate cortex F alcoholics  F normals

[NAA] and [Cho] cerebellum early RAA  normals [NAA] cerebellum late RAA  early RAA

Loc Spectra

Significant findings

Quality info

MR spectroscopy in psychiatry

Table 32.8. Schizophrenia – selected recent review articles Reference

Emphasis

Soares and Innis, 1999

Neurotransmitter theories and evidence Comparison of schizophrenia and BP disease Methodological issues and H, P MRS findings Neurotransmitter theories and evidence Methodological issues and H, P MRS findings General H, P, F MRS review Genetics, neuropathology, neurotransmitters Excellent general overview of schizophrenia Links H, P MRS to pathogenesis theories First-episode, medication-naïve schizophrenic H, P MRS

Curtis et al., 2000 Keshavan et al., 2000 Laruelle, 2000 Stanley et al., 2000 Vance et al., 2000 Krystal et al., 2001 Kasai et al., 2002 Lyoo and Renshaw, 2002 Stanley, 2002

Such “first generation” MRS studies of schizophrenia (Keshavan et al., 2000) have been helpful, but more work must be done in order to create reliable MRS markers of schizophrenia.

Future directions No chapter on spectroscopy would be complete without a brief comment on the recommended future direction for MRS work in psychiatric illness. First and foremost, it is important that future psychiatric MRS studies address all of the technical issues outlined above. The most important of these issues is the selection of a pure cohort of study subjects; inclusion and exclusion criteria must be carefully chosen to insure that only patients with a certain precise diagnosis will be studied in order to prevent “dilution” of findings. Ideally, subjects should be medication naive (e.g. new diagnosis) or go through a medication washout period before being studied. An age- and sex-matched case–control design is useful since fewer subjects are needed to reach statistical significance. However, a study should have as many

subjects as possible (given budgetary and time limitations) to increase the likelihood of discovering statistically significant findings. Finally, from a technical aspect, quantitative analyses should be performed using any number of automated or semi-automated curve-fitting programs (outlined in Chapter 2). Examples of good studies that account for many of these issues include Auer et al., 2000; Rosenberg et al., 2000; Deicken et al., 2001; Villarreal et al., 2002. Future psychiatric MRS studies should take advantage of new experimental MRS methods that are relevant to psychiatric illness and treatment. A few pioneering studies have tentatively examined GABA, the major inhibitory neurotransmitter in the brain (Behar et al., 1999; Sanacora et al., 1999b; Goddard et al., 2001). For an excellent review of the state of the art in GABA MRS, cf. Sanacora (Sanacora et al., 2000). Furthermore, Glu is the principle excitatory neurotransmitter in the brain. Using short echo times (TE) techniques, it is possible to measure Glx levels, however, to date only a few psychiatric MRS studies have examined this peak (Moore et al., 1998; Auer et al., 2000; Rosenberg et al., 2000; Bolton et al., 2001). Since a considerable amount is known about these neurotransmitter systems, detecting GABA or Glu MRS abnormalities in psychiatric patients would be quite helpful to understanding the underlying biochemical defects as well as to determine optimal treatments. Perhaps someday we may be able to obtain MRS data for other important neurotransmitters such as dopamine, serotonin and acetylcholine; for now their extremely low brain concentrations limit such studies to the realm of PET studies. New psychiatric MRS studies should take advantage of new hardware technologies such as veryhigh-field MRI machines (3 T and above). While susceptibility artifacts are considerably increased at high fields, high-order shimming can help smooth out magnetic fields enough to obtain reliable spectroscopic data. Since the signal-to-noise ratio is increased at higher magnetic field strengths, imaging time can be decreased, voxels can be made smaller, or more voxels can be obtained using MRSI techniques (Chapter 1). The improvement in available signal afforded at higher fields also allow possibilities for more sophisticated spectral editing techniques

551

552

John D. Port

which will allow measurement of some clinically relevant neurotransmitters described above. The MRSI technique is also valuable since several different brain regions can be studied simultaneously in a single well-positioned slice. This is quite useful since we do not yet know where the biochemical abnormalities are located in the various psychiatric diseases. While quantitative MRSI analysis methods need to be further refined, pioneers such as Deicken and his team (Deicken et al., 2001) currently set the standard for excellence in MRSI study design. Very-high-field magnets may also lead to the development of new psychiatric imaging techniques. For example, animal research suggests a heterogeneous distribution of lithium within the brain parenchyma (Soares et al., 2000). Ideally, since lithium has a single peak, it should be possible to perform lithium imaging, much as has been done for sodium. However, such techniques have been hampered by low signal to noise ratios due to the small concentrations of lithium within the brain. Higher-field magnets may be able to increase the signal-to-noise such that lithium images can be obtained. Soares and his team have already performed lithium MRS at 3T (Soares et al., 2001); lithium imaging is a logical and very exciting next step. Modern spectroscopy techniques have only been around for the past 10–15 years, and available clinically for the past 8 years. Psychiatric MRS is still in its infancy; there is much to do.

REFERENCES American Psychiatric Association. Task Force on Nomenclature and Statistics, and American Psychiatric Association. Committee on Nomenclature and Statistics. 1980. Diagnostic and Statistical Manual of Mental Disorders, American Psychiatric Association, Washington DC. American Psychiatric Association. 2000. Diagnostic and Statistical Manual of Mental Disorders: DSM-IV-TR, American Psychiatric Association, Washington DC. Atack JR, Broughton HB, Pollack SJ. 1995. Inositol monophosphatase – a putative target for Li in the treatment of bipolar disorder. Trend Neurosci 18: 343–349. Auer DP, Putz B, Kraft E, Lipinski B, Schill J, Holsboer F. 2000. Reduced glutamate in the anterior cingulate cortex in

depression: an in vivo proton magnetic resonance spectroscopy study. Biol Psychiatr 47: 305–313. Bartha R, Stein MB, Williamson PC, Drost DJ, Neufeld RW, Carr TJ, Canaran G, Densmore M, Anderson G, Siddiqui AR. 1998. A short echo 1H spectroscopy and volumetric MRI study of the corpus striatum in patients with obsessivecompulsive disorder and comparison subjects. Am J Psychiatr 155: 1584–1591. Behar KL, Rothman DL, Petersen KF, Hooten M, Delaney R, Petroff OA, Shulman GI, Navarro V, Petrakis IL, Charney DS, Krystal JH. 1999. Preliminary evidence of low cortical GABA levels in localized 1H-MR spectra of alcohol-dependent and hepatic encephalopathy patients. Am J Psychiatr 156: 952–954. Belmaker RH, Bersudsky Y, Agam G, Levine J, Kofman O. 1996. How does lithium work on manic depression? Clinical and psychological correlates of the inositol theory. Annu Rev Med 47: 47–56. Bendszus M, Weijers HG, Wiesbeck G, Warmuth-Metz M, Bartsch AJ, Engels S, Boning J, Solymosi L. 2001. Sequential MR imaging and proton MR spectroscopy in patients who underwent recent detoxification for chronic alcoholism: correlation with clinical and neuropsychological data. Am J Neuroradiol 22: 1926–1932. Bolton J, Moore GJ, MacMillan S, Stewart CM, Rosenberg DR. 2001. Case study: caudate glutamatergic changes with paroxetine persist after medication discontinuation in pediatric OCD. J Am Acad Child Adolesc Psychiatr 40: 903–906. Bottomley PA. 1991. The trouble with spectroscopy papers. Radiology 181: 344–350. Brodmann K. 1909. Vergleichende Lokalisationslehre der Grosshirnrinde in ihren Prinzipien dargestellt auf Grund des Zellenbaues, JA Barth, Leipzig. Castillo M, Kwock L, Courvoisie H, Hooper SR. 2000. Proton MR spectroscopy in children with bipolar affective disorder: preliminary observations. Am J Neuroradiol 21: 832–838. Charles HC, Lazeyras F, Krishnan KR, Boyko OB, Patterson LJ, Doraiswamy PM, McDonald WM. 1994a. Proton spectroscopy of human brain: effects of age and sex. Prog Neuropsychopharmacol Biol Psychiatr 18: 995–1004. Charles HC, Lazeyras F, Krishnan KR, Boyko OB, Payne M, Moore D. 1994b. Brain choline in depression: in vivo detection of potential pharmacodynamic effects of antidepressant therapy using hydrogen localized spectroscopy. Prog Neuropsychopharmacol Biol Psychiatr 18: 1121–1127. Chiu TM, Mendelson JH, Woods BT, Teoh SK, Levisohn L, Mello NK. 1994. In vivo proton magnetic resonance spectroscopy detection of human alcohol tolerance. Magn Reson Med 32: 511–516.

MR spectroscopy in psychiatry

Cohen J. 1977. Statistical Power Analysis for the Behavioral Sciences, Academic Press, New York. Curtis VA, van Os J, Murray RM. 2000. The Kraepelinian dichotomy: evidence from developmental and neuroimaging studies [comment]. J Neuropsychiatr Clin Neurosci 12: 398–405. Dager SR, Friedman SD, Heide A, Layton ME, Richards T, Artru A, Strauss W, Hayes C, Posse S. 1999. Two-dimensional proton echo-planar spectroscopic imaging of brain metabolic changes during lactate-induced panic. Arch Gen Psychiatr 56: 70–77. Dager SR, Marro KI, Richards TL, Metzger GD. 1994. Preliminary application of magnetic resonance spectroscopy to investigate lactate-induced panic. Am J Psychiatr 151: 57–63. Dager SR, Strauss WL, Marro KI, Richards TL, Metzger GD, Artru AA. 1995. Proton magnetic resonance spectroscopy investigation of hyperventilation in subjects with panic disorder and comparison subjects. Am J Psychiatr 152: 666–672. Dager SR, Richards T, Strauss W, Artru A. 1997. Single-voxel 1H-MRS investigation of brain metabolic changes during lactate-induced panic. Psychiatr Res 76: 89–99. Davanzo P, Thomas MA, Yue K, Oshiro T, Belin T, Strober M, McCracken J. 2001. Decreased anterior cingulate myoinositol/creatine spectroscopy resonance with lithium treatment in children with bipolar disorder. Neuropsychopharmacology 24: 359–369. De Bellis MD, Keshavan MS, Harenski KA. 2001. Anterior cingulate N-acetylaspartate/creatine ratios during clonidine treatment in a maltreated child with posttraumatic stress disorder. J Child Adolesc Psychopharmacol 11: 311–316. De Bellis MD, Keshavan MS, Spencer S, Hall J. 2000. N-acetylaspartate concentration in the anterior cingulate of maltreated children and adolescents with PTSD. Am J Psychiatr 157: 1175–1177. Deicken RF, Eliaz Y, Feiwell R, Schuff N. 2001. Increased thalamic N-acetylaspartate in male patients with familial bipolar I disorder. Psychiatr Res 106: 35–45. Deicken RF, Fein G, Weiner MW. 1995a. Abnormal frontal lobe phosphorous metabolism in bipolar disorder. Am J Psychiatr 152: 915–918. Deicken RF, Weiner MW, Fein G. 1995b. Decreased temporal lobe phosphomonoesters in bipolar disorder. J Affect Disord 33: 195–199. Ebert D, Speck O, Konig A, Berger M, Hennig J, Hohagen F. 1997. 1H-magnetic resonance spectroscopy in obsessivecompulsive disorder: evidence for neuronal loss in the cingulate gyrus and the right striatum. Psychiatr Res 74: 173–176. Ende G, Braus DF, Walter S, Weber-Fahr W, Henn FA. 2000. The

therapy: a proton magnetic resonance spectroscopic imaging study. Arch Gen Psychiatr 57: 937–943. Estilaei MR, Matson GB, Payne GS, Leach MO, Fein G, Meyerhoff DJ. 2001a. Effects of abstinence from alcohol on the broad phospholipid signal in human brain: an in vivo 31P magnetic resonance spectroscopy study. Alcohol Clin Exp Res 25: 1213–1220. Estilaei MR, Matson GB, Payne GS, Leach MO, Fein G, Meyerhoff DJ. 2001b. Effects of chronic alcohol consumption on the broad phospholipid signal in human brain: an in vivo 31P MRS study. Alcohol Clin Exp Res 25: 89–97. Feinberg I. 1982. Schizophrenia: caused by a fault in programmed synaptic elimination during adolescence? J Psychiatr Res 17: 319–334. Feinberg I. 1990. Cortical pruning and the development of schizophrenia. Schizophr Bull 16: 567–570. Fitzgerald KD, Moore GJ, Paulson LA, Stewart CM, Rosenberg DR. 2000. Proton spectroscopic imaging of the thalamus in treatment-naive pediatric obsessive-compulsive disorder. Biol Psychiatr 47: 174–182. Freeman TW, Cardwell D, Karson CN, Komoroski RA. 1998. In vivo proton magnetic resonance spectroscopy of the medial temporal lobes of subjects with combat-related posttraumatic stress disorder. Magn Reson Med 40: 66–71. Frey R, Metzler D, Fischer P, Heiden A, Scharfetter J, Moser E, Kasper S. 1998. Myo-inositol in depressive and healthy subjects determined by frontal 1H-magnetic resonance spectroscopy at 1.5 tesla. J Psychiatr Res 32: 411–420. Fukuzako H, Hashiguchi T, Sakamoto Y, Okamura H, Doi W, Takenouchi K, Takigawa M. 1997. Metabolite changes with age measured by proton magnetic resonance spectroscopy in normal subjects. Psychiatr Clin Neurosci 51: 261–263. Goddard AW, Mason GF, Almai A, Rothman DL, Behar KL, Petroff OA, Charney DS, Krystal JH. 2001. Reductions in occipital cortex GABA levels in panic disorder detected with 1H-magnetic resonance spectroscopy. Arch Gen Psychiatr 58: 556–561. Hamakawa H, Kato T, Murashita J, Kato N. 1998. Quantitative proton magnetic resonance spectroscopy of the basal ganglia in patients with affective disorders. Eur Arch Psychiatr Clin Neurosci 248: 53–58. Hamakawa H, Kato T, Shioiri T, Inubushi T, Kato N. 1999. Quantitative proton magnetic resonance spectroscopy of the bilateral frontal lobes in patients with bipolar disorder. Psychol Med 29: 639–644. Hanstock CC, Rothman DL, Shulman RG, Novotny Jr EJ, Petroff OA, Prichard JW. 1990. Measurement of ethanol in the human brain using NMR spectroscopy. J Stud Alcohol 51: 104–107. Jagannathan NR, Desai NG, Raghunathan P. 1996. Brain

hippocampus in patients treated with electroconvulsive

metabolite changes in alcoholism: an in vivo proton

553

554

John D. Port

magnetic resonance spectroscopy (MRS) study. Magn Reson Imaging 14: 553–557. Kasai K, Iwanami A, Yamasue H, Kuroki N, Nakagome K, Fukuda M. 2002. Neuroanatomy and neurophysiology in schizophrenia. Neurosci Res 43: 93–110. Kato T, Fujii K, Shioiri T, Inubushi T, Takahashi S. 1996a. Lithium side effects in relation to brain lithium concentration measured by lithium-7 magnetic resonance spectroscopy. Prog Neuropsychopharmacol Biol Psychiatr 20: 87–97. Kato T, Hamakawa H, Shioiri T, Murashita J, Takahashi Y, Takahashi S, Inubushi T. 1996b. Choline-containing compounds detected by proton magnetic resonance spectroscopy in the basal ganglia in bipolar disorder. J Psychiatr Neurosci 21: 248–254. Kato T, Inubushi T, Takahashi S. 1994a. Relationship of lithium concentrations in the brain measured by lithium-7 magnetic resonance spectroscopy to treatment response in mania. J Clin Psychopharmacol 14: 330–335. Kato T, Murashita J, Kamiya A, Shioiri T, Kato N, Inubushi T. 1998. Decreased brain intracellular pH measured by 31PMRS in bipolar disorder: a confirmation in drug-free patients and correlation with white matter hyperintensity. Eur Arch Psychiatr Clin Neurosci 248: 301–306. Kato T, Shioiri T, Murashita J, Hamakawa H, Inubushi T, Takahashi S. 1994b. Phosphorus-31 magnetic resonance spectroscopy and ventricular enlargement in bipolar disorder. Psychiatr Res 55: 41–50. Kato T, Shioiri T, Murashita J, Hamakawa H, Takahashi Y, Inubushi T, Takahashi S. 1995. Lateralized abnormality of high energy phosphate metabolism in the frontal lobes of patients with bipolar disorder detected by phase-encoded 31P-MRS. Psychol Med 25: 557–566. Kato T, Shioiri T, Takahashi S, Inubushi T. 1991 Measurement of brain phosphoinositide metabolism in bipolar patients using in vivo 31P-MRS. J Affect Disord 22: 185–190. Kato T, Takahashi S, Shioiri T, Inubushi T. 1992. Brain phosphorous metabolism in depressive disorders detected by phosphorus-31 magnetic resonance spectroscopy. J Affect Disord 26: 223–230. Kato T, Takahashi S, Shioiri T, Inubushi T. 1993. Alterations in brain phosphorous metabolism in bipolar disorder detected by in vivo 31P and 7Li magnetic resonance spectroscopy. J Affect Disord 27: 53–59. Kato T, Takahashi S, Shioiri T, Murashita J, Hamakawa H, Inubushi T. 1994c. Reduction of brain phosphocreatine in bipolar II disorder detected by phosphorus-31 magnetic resonance spectroscopy. J Affect Disord 31: 125–133. Keshavan MS, Stanley JA, Pettegrew JW. 2000. Magnetic resonance spectroscopy in schizophrenia: methodological issues and findings – Part II. Biol Psychiatr 48: 369–380.

Kessler RC, McGonagle KA, Zhao S, Nelson CB, Hughes M, Eshleman S, Wittchen HU, Kendler KS. 1994. Lifetime and 12-month prevalence of DSM-III-R psychiatric disorders in the United States. Results from the National comorbidity survey. Arch Gen Psychiatr 51: 8–19. Kessler RC, Rubinow DR, Holmes C, Abelson JM, Zhao S. 1997. The epidemiology of DSM-III-R bipolar I disorder in a general population survey. Psychol Med 27: 1079–1089. Kinney DK, Steingard RJ, Renshaw PF, Yurgelun-Todd DA. 2000. Perinatal complications and abnormal proton metabolite concentrations in frontal cortex of adolescents seen on magnetic resonance spectroscopy. Neuropsychiatr Neuropsychol Behav Neurol 13: 8–12. Koran LM, Thienemann ML, Davenport R. 1996. Quality of life for patients with obsessive-compulsive disorder. Am J Psychiatr 153: 783–788. Kreis R, Ernst T, Ross BD. 1993. Development of the human brain: in vivo quantification of metabolite and water content with proton magnetic resonance spectroscopy. Magn Reson Med 30: 424–437. Krystal JH, D’Souza DC, Sanacora G, Goddard AW, Charney DS. 2001. Current perspectives on the pathophysiology of schizophrenia, depression, and anxiety disorders. Med Clin N Am 85: 559–577. Kusumakar V, MacMaster FP, Gates L, Sparkes SJ, Khan SC. 2001. Left medial temporal cytosolic choline in early onset depression. Can J Psychiatr 46: 959–964. Laruelle M. 2000. The role of endogenous sensitization in the pathophysiology of schizophrenia: implications from recent brain imaging studies. Brain Res Rev 31: 371–384. Layton ME, Friedman SD, Dager SR. 2001. Brain metabolic changes during lactate-induced panic: effects of gabapentin treatment. Depress Anxiety 14: 251–254. Lyoo IK, Renshaw PF. 2002. Magnetic resonance spectroscopy: current and future applications in psychiatric research. Biol Psychiatr 51: 195–207. Martin PR, Gibbs SJ, Nimmerrichter AA, Riddle WR, Welch LW, Willcott MR. 1995. Brain proton magnetic resonance spectroscopy studies in recently abstinent alcoholics. Alcohol Clin Exp Res 19: 1078–1082. Massana G, Gasto C, Junque C, Mercader JM, Gomez B, Massana J, Torres X, Salamero M. 2002. Reduced levels of creatine in the right medial temporal lobe region of panic disorder patients detected with (1)H magnetic resonance spectroscopy. Neuroimage 16: 836–842. McLean MA, Woermann FG, Barker GJ, Duncan JS. 2000. Quantitative analysis of short echo time (1)H-MRSI of cerebral gray and white matter. Magn Reson Med 44: 401–411. Mendelson JH, Woods BT, Chiu TM, Mello NK, Lukas SE, Teoh SK, Sintavanarong P, Cochin J, Hopkins MA, Dobrosielski M.

MR spectroscopy in psychiatry

1990. In vivo proton magnetic resonance spectroscopy of alcohol in human brain. Alcohol 7: 443–447. Meyerhoff DJ, MacKay S, Sappey-Marinier D, Deicken R, Calabrese G, Dillon WP, Weiner MW, Fein G. 1995. Effects of chronic alcohol abuse and HIV infection on brain phosphorus metabolites. Alcohol Clin Exp Res 19: 685–692. Michels R, Marzuk PM. 1993a. Progress in psychiatry (1). New Engl J Med 329: 552–560. Michels R, Marzuk PM. 1993b. Progress in psychiatry (2). New Engl J Med 329: 628–638. Moore CM, Breeze JL, Gruber SA, Babb SM, Frederick BB, Villafuerte RA, Stoll AL, Hennen J, Yurgelun-Todd DA, Cohen BM, Renshaw PF. 2000a. Choline, myo-inositol and mood in bipolar disorder: a proton magnetic resonance spectroscopic imaging study of the anterior cingulate cortex. Bipolar Disord 2: 207–216. Moore CM, Christensen JD, Lafer B, Fava M, Renshaw PF. 1997. Lower levels of nucleoside triphosphate in the basal ganglia of depressed subjects: a phosphorous-31 magnetic resonance spectroscopy study. Am J Psychiatr 154: 116–118. Moore CM, Demopulos CM, Henry ME, Steingard RJ, Zamvil L, Katic A, Breeze JL, Moore JC, Cohen BM, Renshaw PF. 2002. Brain-to-serum lithium ratio and age: an in vivo magnetic resonance spectroscopy study. Am J Psychiatr 159: 1240–1242. Moore GJ, Bebchuk JM, Hasanat K, Chen G, Seraji-Bozorgzad N, Wilds IB, Faulk MW, Koch S, Glitz DA, Jolkovsky L, Manji HK. 2000b. Lithium increases N-acetyl-aspartate in the human brain: in vivo evidence in support of bcl-2’s neurotrophic effects? Biol Psychiatr 48: 1–8. Moore GJ, Bebchuk JM, Parrish JK, Faulk MW, Arfken CL, Strahl-Bevacqua J, Manji HK. 1999. Temporal dissociation between lithium-induced changes in frontal lobe myoinositol and clinical response in manic-depressive illness. Am J Psychiatr 156: 1902–1908. Moore GJ, MacMaster FP, Stewart C, Rosenberg DR. 1998. Case study: caudate glutamatergic changes with paroxetine therapy for pediatric obsessive-compulsive disorder. J Am Acad Child Adolesc Psychiatr 37: 663–667. Murata T, Kimura H, Omori M, Kado H, Kosaka H, Iidaka T, Itoh H, Wada Y. 2001. MRI white matter hyperintensities, (1)H-MR spectroscopy and cognitive function in geriatric depression: a comparison of early- and late-onset cases. Int J Geriatr Psychiatr 16: 1129–1135. Ohara K, Isoda H, Suzuki Y, Takehara Y, Ochiai M, Takeda H, Igarashi Y. 1998. Proton magnetic resonance spectroscopy of the lenticular nuclei in bipolar I affective disorder. Psychiatr Res 84: 55–60. Ohara K, Isoda H, Suzuki Y, Takehara Y, Ochiai M, Takeda H, Igarashi Y. 1999. Proton magnetic resonance spectroscopy

of lenticular nuclei in obsessive-compulsive disorder. Psychiatr Res 92: 83–91. O’Neill J, Cardenas VA, Meyerhoff DJ. 2001. Effects of abstinence on the brain: quantitative magnetic resonance imaging and magnetic resonance spectroscopic imaging in chronic alcohol abuse. Alcohol Clin Exp Res 25: 1673–1682. Parks MH, Dawant BM, Riddle WR, Hartmann SL, Dietrich MS, Nickel MK, Price RR, Martin PR. 2002. Longitudinal brain metabolic characterization of chronic alcoholics with proton magnetic resonance spectroscopy. Alcohol Clin Exp Res 26: 1368–1380. Pettegrew JW, Keshavan MS, Minshew NJ. 1993. 31P nuclear magnetic resonance spectroscopy: neurodevelopment and schizophrenia. Schizophr Bull 19: 35–53. Pfefferbaum A, Adalsteinsson E, Spielman D, Sullivan EV, Lim KO. 1999. In vivo brain concentrations of N-acetyl compounds, creatine, and choline in Alzheimer disease. Arch Gen Psychiatr 56: 185–192. Pitts Jr FN, McClure Jr JN, 1967. Lactate metabolism in anxiety neurosis. New Engl J Med 277: 1329–1336. Renshaw PF, Guimaraes AR, Fava M, Rosenbaum JF, Pearlman JD, Flood JG, Puopolo PR, Clancy K, Gonzalez RG. 1992. Accumulation of fluoxetine and norfluoxetine in human brain during therapeutic administration. Am J Psychiatr 149: 1592–1594. Renshaw PF, Lafer B, Babb SM, Fava M, Stoll AL, Christensen JD, Moore CM, Yurgelun-Todd DA, Bonello CM, Pillay SS, Rothschild AJ, Nierenberg AA, Rosenbaum JF, Cohen BM. 1997. Basal ganglia choline levels in depression and response to fluoxetine treatment: an in vivo proton magnetic resonance spectroscopy study. Biol Psychiatr 41: 837–843. Renshaw PF, Parow AM, Hirashima F, Ke Y, Moore CM, Frederick Bde B, Fava M, Hennen J, Cohen BM. 2001. Multinuclear magnetic resonance spectroscopy studies of brain purines in major depression. Am J Psychiatr 158: 2048–2055. Riedl U, Barocka A, Kolem H, Demling J, Kaschka WP, Schelp R, Stemmler M, Ebert D. 1997. Duration of lithium treatment and brain lithium concentration in patients with unipolar and schizoaffective disorder – a study with magnetic resonance spectroscopy. Biol Psychiatr 41: 844–850. Rosenberg DR, Amponsah A, Sullivan A, MacMillan S, Moore GJ. 2001. Increased medial thalamic choline in pediatric obsessive-compulsive disorder as detected by quantitative in vivo spectroscopic imaging. J Child Neurol 16: 636–641. Rosenberg DR, MacMaster FP, Keshavan MS, Fitzgerald KD, Stewart CM, Moore GJ. 2000. Decrease in caudate glutamatergic concentrations in pediatric obsessive-compulsive disorder patients taking paroxetine. J Am Acad Child Adolesc Psychiatr 39: 1096–1103.

555

556

John D. Port

Ross B, Bluml S. 2001. Magnetic resonance spectroscopy of the human brain. Anat Rec 265: 54–84. Sanacora G, Mason GF, Krystal JH. 2000. Impairment of GABAergic transmission in depression: new insights from neuroimaging studies. Crit Rev Neurobiol 14: 23–45. Sanacora G, Mason GF, Rothman DL, Behar KL, Hyder F, Petroff OA, Berman RM, Charney DS, Krystal JH. 1999a. Reduced cortical gamma-aminobutyric acid levels in depressed patients determined by proton magnetic resonance spectroscopy. Arch Gen Psychiatr 56: 1043–1047. Sanacora G, Rothman DL, Krystal JH. 1999b. Applications of magnetic resonance spectroscopy to psychiatry. Neuroscientist 5: 192–196. Schuff N, Neylan TC, Lenoci MA, Du AT, Weiss DS, Marmar CR, Weiner MW. 2001. Decreased hippocampal N-acetylaspartate in the absence of atrophy in posttraumatic stress disorder. Biol Psychiatr 50: 952–959. Schweinsburg BC, Alhassoon OM, Taylor MJ, Gonzalez R, Videen JS, Brown GG, Patterson TL, Grant I. 2003. Effects of alcoholism and gender on brain metabolism. Am J Psychiatr 160: 1180–1183. Schweinsburg BC, Taylor MJ, Alhassoon OM, Videen JS, Brown GG, Patterson TL, Berger F, Grant I. 2001. Chemical pathology in brain white matter of recently detoxified alcoholics: a 1H magnetic resonance spectroscopy investigation of alcohol-associated frontal lobe injury. Alcohol Clin Exp Res 25: 924–934. Schweinsburg BC, Taylor MJ, Videen JS, Alhassoon OM, Patterson TL, Grant I. 2000. Elevated myo-inositol in gray matter of recently detoxified but not long-term abstinent alcoholics: a preliminary MR spectroscopy study. Alcohol Clin Exp Res 24: 699–705. Seitz D, Widmann U, Seeger U, Nagele T, Klose U, Mann K, Grodd W. 1999. Localized proton magnetic resonance spectroscopy of the cerebellum in detoxifying alcoholics. Alcohol Clin Exp Res 23: 158–163. Sharma R, Venkatasubramanian PN, Barany M, Davis JM. 1992. Proton magnetic resonance spectroscopy of the brain in schizophrenic and affective patients. Schizophr Res 8: 43–49. Shioiri T, Kato T, Murashita J, Hamakawa H, Inubushi T, Takahashi S. 1996. High-energy phosphate metabolism in the frontal lobes of patients with panic disorder detected by phase-encoded 31P-MRS. Biol Psychiatr 40: 785–793. Sijens PE, Heijer Td T, Origgi D, Vermeer SE, Breteler MM, Hofman A, Oudkerk M. 2003. Brain changes with aging: MR spectroscopy at supraventricular plane shows differences between women and men. Radiology 226: 889–896. Soares JC, Boada F, Keshavan MS. 2000. Brain lithium measurements with (7)Li magnetic resonance spectroscopy (MRS): a literature review. Eur Neuropsychopharmacol 10: 151–158.

Soares JC, Boada F, Spencer S, Mallinger AG, Dippold CS, Wells KF, Frank E, Keshavan MS, Gershon S, Kupfer DJ. 2001. Brain lithium concentrations in bipolar disorder patients: preliminary (7)Li magnetic resonance studies at 3 T. Biol Psychiatr 49: 437–443. Soares JC, Innis RB. 1999. Neurochemical brain imaging investigations of schizophrenia. Biol Psychiatr 46: 600–615. Soares JC, Mallinger AG. 1997. Intracellular phosphatidylinositol pathway abnormalities in bipolar disorder patients. Psychopharmacol Bull 33: 685–691. Sonawalla SB, Renshaw PF, Moore CM, Alpert JE, Nierenberg AA, Rosenbaum JF, Fava M. 1999. Compounds containing cytosolic choline in the basal ganglia: a potential biological marker of true drug response to fluoxetine. Am J Psychiatr 156: 1638–1640. Spielman DM, Glover GH, Macovski A, Pfefferbaum A. 1993. Magnetic resonance spectroscopic imaging of ethanol in the human brain: a feasibility study. Alcohol Clin Exp Res 17: 1072–1077. Stanley JA. 2002. In vivo magnetic resonance spectroscopy and its application to neuropsychiatric disorders [comment]. Can J Psychiatr 47: 315–326. Stanley JA, Pettegrew JW, Keshavan MS. 2000. Magnetic resonance spectroscopy in schizophrenia: methodological issues and findings – Part I. Biol Psychiatr 48: 357–368. Stanley JA, Williamson PC, Drost DJ, Carr TJ, Rylett RJ, Malla A, Thompson RT. 1995. An in vivo study of the prefrontal cortex of schizophrenic patients at different stages of illness via phosphorus magnetic resonance spectroscopy. Arch Gen Psychiatr 52: 399–406. Stanley JA, Williamson PC, Drost DJ, Carr TJ, Rylett RJ, Morrison-Stewart S, Thompson RT. 1994. Membrane phospholipid metabolism and schizophrenia: an in vivo 31P-MR spectroscopy study. Schizophr Res 13: 209–215. Steingard RJ, Yurgelun-Todd DA, Hennen J, Moore JC, Moore CM, Vakili K, Young AD, Katic A, Beardslee WR, Renshaw PF. 2000. Increased orbitofrontal cortex levels of choline in depressed adolescents as detected by in vivo proton magnetic resonance spectroscopy. Biol Psychiatr 48: 1053–1061. Stoll AL, Renshaw PF, Sachs GS, Guimaraes AR, Miller C, Cohen BM, Lafer B, Gonzalez RG. 1992. The human brain resonance of choline-containing compounds is similar in patients receiving lithium treatment and controls: an in vivo proton magnetic resonance spectroscopy study. Biol Psychiatr 32: 944–949. Strauss WL, Layton ME, Hayes CE, Dager SR. 1997. 19F magnetic resonance spectroscopy investigation in vivo of acute and steady-state brain fluvoxamine levels in obsessive-compulsive disorder. Am J Psychiatr 154: 516–522.

MR spectroscopy in psychiatry

Sun Z, Wang F, Cui L, Breeze J, Du X, Wang X, Cong Z, Zhang H, Li B, Hong N, Zhang D. 2003. Abnormal anterior cingulum in patients with schizophrenia: a diffusion tensor imaging study. Neuroreport 14: 1833–1836. van der Knaap MS, van der Grond J, van Rijen PC, Faber JA, Valk J, Willemse K. 1990. Age-dependent changes in localized proton and phosphorus MR spectroscopy of the brain. Radiology 176: 509–515. Vance AL, Velakoulis D, Maruff P, Wood SJ, Desmond P, Pantelis C. 2000. Magnetic resonance spectroscopy and schizophrenia: what have we learnt? Aust NZ J Psychiatr 34: 14–25. Villarreal G, Petropoulos H, Hamilton DA, Rowland LM, Horan WP, Griego JA, Moreshead M, Hart BL, Brooks WM. 2002. Proton magnetic resonance spectroscopy of the hippocampus and occipital white matter in PTSD: preliminary results. Can J Psychiatr 47: 666–670. Volz HP, Rzanny R, Riehemann S, May S, Hegewald H, Preussler B, Hubner G, Kaiser WA, Sauer H. 1998. 31P

magnetic resonance spectroscopy in the frontal lobe of major depressed patients. Eur Arch Psychiatr Clin Neurosci 248: 289–295. Weissman MM, Bland RC, Canino GJ, Greenwald S, Hwu HG, Lee CK, Newman SC, Oakley-Browne MA, Rubio-Stipec M, Wickramaratne PJ, et al. 1994. The cross national epidemiology of obsessive compulsive disorder. The Cross National Collaborative Group. J Clin Psychiatr 55(suppl.): 5–10. Winsberg ME, Sachs N, Tate DL, Adalsteinsson E, Spielman D, Ketter TA. 2000. Decreased dorsolateral prefrontal N-acetylaspartate in bipolar disorder. Biol Psychiatr 47: 475–481. Yildiz A, Demopulos CM, Moore CM, Renshaw PF, Sachs GS. 2001a. Effect of lithium on phosphoinositide metabolism in human brain: a proton decoupled (31)P magnetic resonance spectroscopy study. Biol Psychiatr 50: 3–7. Yildiz A, Sachs GS, Dorer DJ, Renshaw PF. 2001b. 31P Nuclear magnetic resonance spectroscopy findings in bipolar illness: a meta-analysis. Psychiatr Res 106: 181–191

557

33

Diffusion MR imaging in neuropsychiatry and aging Adolf Pfefferbaum and Edith V. Sullivan Neuroscience Program, SRI International, CA, USA; Department of Psychiatry and Behavioral Sciences, Stanford University School of Medicine, Stanford, CA, USA

Key points • Diffusion tensor imaging (DTI) detects subtle degradation of white matter microstructure with normal aging. • DTI may facilitate early detection of Alzheimer’s disease.

Introduction Over the past decade, the quest for identification of brain mechanisms underlying complex cognitive, motor, and other behavioral functioning has shifted from single structures or loci to systems and circuits. Among the forces guiding this change have been the many functional MR imaging (fMRI) studies that regularly confirm that multiple brain regions are invoked to execute even ostensibly simple tasks. While undeniable that a single, focal lesion can produce impairment in a complex function, such as word naming, the systems conceptualization of brain functioning has logical appeal for understanding the neural bases of the highly variable and vastly complex behaviors characteristic of neuropsychiatric conditions and may serve to explain patterns of functional degradation typifying normal aging. With this shift is the recognition of the relevance of connecting elements of brain circuitry and the possibility that disruption of the connections may be as effective as lesions in gray matter (GM) nodes in producing functional impairment. The neural systems Zeitgeist has provided the adequate impetus for the rapid development of MR diffusion 558

imaging as a non-invasive, in vivo method for characterizing the integrity of microstructure of white matter (WM) fibers in the brain. This chapter provides a review of diffusion imaging findings in normal aging and neuropsychiatric diseases and adds to a growing list of such overviews (Kubicki et al., 2002c; Lim and Helpern, 2002; Moseley, 2002; Moseley et al., 2002; Sullivan and Pfefferbaum, 2003). To provide a context, we first review the physical structure of WM and then briefly summarize the principles of diffusion imaging, with the aim of illustrating how this imaging modality is suitable for visualizing and quantifying disruptions to WM microstructure with aging and disease. More detailed descriptions of diffusion methodology and analysis are found in Chapters 4, 5, and 6. White matter structure The principal component of the brain’s WM is the axon, the long extension from neuronal soma that reaches towards another cell body within a neural circuit (for a thorough review, cf. (Waxman et al., 1995)). Axons can be quite long, centimeters in the brain, are of widely varying gauge (0.2–20 m) but can taper significantly and can be myelinated. The myelin sheath, which is mainly oligodendrocyte, is common on central nervous system (CNS) axons of over 0.2 m and adds girth to the axon diameter, except at myelin’s breaches, the nodes of Ranvier. Within the axon are axoplasm and organelles, including neurofilaments, mitochondria, and microtubules making up the axon’s cytoskeleton. With physical

Diffusion MR imaging in neuropsychiatry and aging

trauma, the cytoskeleton, including the linear orientation of neurofilaments, can be perturbed (Arfanakis et al., 2002). The gross morphology of axons resembles cable wiring laid in bundles, coursing between targets. Basic axonal configurations include fasciculi, commissures, and fibers. The extracellular spaces between fibers harbor fluid, which provides one of the main vehicles of molecular movement in volumes of WM. In general, the physical macro- and microstructure of WM is linear, and deviations from linearity can indicate abnormality. DTI measurement Conventional brain structural MRI maps protons, primarily from water and contributions from fat, with tissue contrast possible because of the fundamental differences in water content in the primary tissues of the brain (WM consists of about 70% water, GM 80%, and cerebrospinal fluid (CSF) 99%), differences in the degree to which water is bound, and differences in the local environment. Water molecules in the brain are in constant Brownian motion, and although the movement of these protons affects conventional structural imaging, diffusion weighted imaging (DWI) and diffusion tensor imaging (DTI) allow quantification of this microscopic movement within each voxel (cf. Chapters 4 and 5). In regions with few or no constraints imposed by physical boundaries, such as CSF in the ventricles, water movement is random in every direction and is isotropic. In contrast to CSF, the path of a water molecule in a WM fiber is constrained by the physical boundaries, such as the axon sheath, causing the movement to be greater along the long axis of the fiber than across it. This movement is called anisotropic; the linear diffusion along the long axis of a fiber is greater than radial diffusion across the fiber (Song et al., 2002) (cf. Chapters 4 and 5). Among the fruitful applications of diffusion imaging has been the examination of WM integrity (Moseley et al., 1990; Basser, 1995). Disruption of WM microstructure detectable with diffusion imaging can reflect breakdown of myelin, cytoskeletal structure, and axon density (Basser, 1995; Basser and Pierpaoli, 1996; Spielman et al., 1996).

The application of additional magnetic gradients during image acquisition allows detection of the microscopic water motion of diffusion. Freely diffusing particles will move more during the time of image acquisition than those with physical restrictions. In order to fully characterize the orientation of the diffusion motion in three dimensional space, observations are made by applying the diffusion gradients in at least six non-collinear orientations. For each voxel, the amount of diffusion is quantified by calculating the ratio of the signals without and with the diffusion gradients for each of the six or more gradient orientations, resulting in at least six different DWI, each comprising signal decrease due to the movement of protons in the orientation of that particular gradient application. The full technique is called DTI because the tensor, a mathematical description of a threedimensional ellipsoid which describes the magnitude and orientation of diffusion, is computed. The tensor provides three eigenvalues, each with three orientational vectors (eigenvectors), describing the diffusion ellipsoid. The eigenvalue average, or trace, reflects the magnitude of diffusion. The extent to which one eigenvalue dominates the other two determines the degree of anisotropy within a voxel. A variety of metrics describing the ratio of the eigenvalues has been proposed, with fractional anisotropy (FA) (Pierpaoli and Basser, 1996) the most common. FA varies in magnitude in different brain structures and tissue types. For example, FA of the ventricular system, which contains mostly CSF, is near 0, whereas FA of the corpus callosum, where fibers are arranged in a regular and parallel fashion, approaches 100%; that is, an infinitely long and thin cylinder. FA, however, is quite sensitive to tissue inhomogeneity from crossing fibers within a voxel (Virta et al., 1999; Pierpaoli et al., 2001) and partial voluming (Pfefferbaum and Sullivan, 2003). Thus, lower than expected FA in a region of fully volumed WM can be an index of the loss of WM integrity. Alternatively, if the fully volumed WM voxels are in a region where multiple WM tracts cross in different directions, such as adjacent to the corpus callosum, FA will be reduced, not necessarily because of reduced fiber integrity but because no single orientation predominates within a voxel (Virta et al., 1999; Pierpaoli et al., 2001).

559

560

Adolf Pfefferbaum and Edith V. Sullivan

The diffusion tensor contains information about spatial orientation of fiber tracts, and the eigenvectors from the tensor can be used to define the orientation of such tracts. Although the connectivity and coherence between different brain regions on vector maps is readily apparent on visual inspection, these maps are difficult to quantify because no simple method exists for averaging vectors. Various “connectivity maps” for quantitative analysis of structural connectivity of WM have been proposed including the lattice index (LI) (Pierpaoli and Basser, 1996), “dot product” maps (Tang et al., 1997), fibertract trajectories (Basser, 1998; Mori et al., 2002; Masutani et al., 2003), and maps of the degree of “alignment” among neighboring vectors, that is on a voxel-to-voxel basis, resulting in a measure of intervoxel coherence (C) (Jones et al., 1999; Pfefferbaum et al., 2000b). Several approaches have been taken to identify regions of interest (ROI) in DTI analysis, including identification of regions on FA maps themselves or voxel-by-voxel comparison with statistical parametric mapping (SPM) calculated on brains normalized to a common size or atlas. A prudent, alternative approach is to identify brain ROI on native structural images, segmentation maps, or basis images without diffusion weighting (b  0 images) and then to overlay the identified anatomy onto anisotropy or diffusivity images for anisotropy or diffusion quantification (Pfefferbaum et al., 2000b). This method avoids the problem of using the dependent variable, for example FA or trace, to define the independent variable, that is the anatomical ROI. The following sections review published findings on diffusion imaging in normal adult aging and neuropsychiatric disorders, including Alzheimer’s disease (AD), schizophrenia, alcoholism, acquired immuno deficiency syndrome (AIDS), and depression.

Normal adult aging Following the developmental years, the normal adult brain continues to undergo considerable morphological change as it ages. Cross-sectional and longitudinal studies using conventional structural MRI provide consistent evidence for systematic

age-related volume increases in CSF-filled spaces, including sulci, fissures and ventricles, that occur at the expense of cortical GM and with little volume change in WM (e.g. (Pfefferbaum et al., 1994; Blatter et al., 1995; Raz et al., 1997; Sullivan et al., 2001b, 2004)); however, a minority of studies have reported the opposite pattern, with greater agerelated volume decline in WM than GM (Guttmann et al., 1998; Jernigan et al., 2001). When regional WM area or volume does show age-related loss, it is typically small, estimated at 2% per decade in a neuropathology study (Miller et al., 1980) and 1% per year in the corpus callosum of elderly men examined longitudinally with MRI (Sullivan et al., 2002). Nonetheless, such volume shrinkage may accelerate in very old age (Salat et al., 1999). Post-mortem investigations reveal degradation of WM microstructure (Kemper, 1994), and include evidence for decline in numbers of myelinated fibers of the precentral gyrus and corpus callosum. Small connecting fibers of the anterior corpus callosum are especially vulnerable in aging, and their disruption may contribute to age-related declines in cognitive processes dependent on functioning of the prefrontal cortical circuitry (Craik et al., 1990; Raz 1999). Degradation of myelin and microtubules and even axon deletion also accompanies normal aging (Meier-Ruge et al., 1992; Aboitiz et al., 1996). Although these defects in WM microstructure are beyond detection with conventional MRI, they fall within the scope of DTI quantification. DTI studies of normal aging commonly report on measures of diffusivity, such as apparent diffusion coefficient (ADC), or the trace of the diffusion tensor. Such studies reveal little variability in whole brain measures, where they have been estimated to increase by 3% per decade after 40 years of age (Chen et al., 2001) but by only 5% across the entire adult age range (20–89 years) (Naganawa et al., 2003). An analysis of diffusion images, which were collected in a clinical setting but read as normal, indicated a quadratic relationship between ADC and age that accelerated from age 60 years onward (Engelter et al., 2000). Another study, in which tissue types were segmented and measured separately, reported absence of ADC increase with age (Helenius et al., 2002).

Diffusion MR imaging in neuropsychiatry and aging

Fig. 33.1 An axial image segmented into CSF (black), GM, (dark gray) and WM (light gray), FA image at the same level with segmentation boundaries superimposed in red, and the same FA image after spatial warping.

Cross-sectional studies examining age effects on anisotropy, usually expressed as FA, have revealed age-related FA declines in WM in normal healthy men and women ((Chun et al., 2000; Pfefferbaum et al., 2000b; Nusbaum et al., 2001; O’Sullivan et al., 2001; Stebbins et al., 2001; Sullivan et al., 2001a; Pfefferbaum and Sullivan, 2003), but cf. (Chepuri et al., 2002)) in whom volume declines were not detectable. Pfefferbaum and colleagues developed and applied post-acquisition imaging methods to reduce the spatial distortion arising from diffusion images acquired with echo-planar imaging (Figure 33.1); in addition, regions for DTI quantification were identified on segmented high-resolution images (Pfefferbaum et al., 2000b). With this approach, DTI measures showed substantial regional variability, on average 43% for FA and 16% for trace, and aging effects in individuals spanning the normal age range (23–85 years). Significant negative correlations were observed between advancing age and FA in the genu and splenium of the corpus callosum (Pfefferbaum and Sullivan, 2003) and bilateral frontal and parietal pericallosal WM (Pfefferbaum et al., 2000a), and positive correlations were present between older age and diffusivity (trace) in the genu,

splenium, and centrum WM (Pfefferbaum and Sullivan, 2003) (Figures 33.2 and 33.3). Regional variation and age correlations were equivalent in men and women (Sullivan et al., 2001a). One common approach to standardize the placement and size of ROI and to minimize partial voluming from surrounding tissue or CSF is to overlay circles or ellipses of a particular size onto each brain region. One such study examined 30 men and 20 women, age: 21–69 years, using seven WM and GM ROI to measure regional effects of age on ADC and FA (Abe et al., 2002). The only significant correlations were between advancing age and greater ADC in frontal WM and the lentiform nucleus, and between age and lower FA in the genu of the corpus callosum. The genu was the only region to show a complementary relationship between ADC and FA, an effect to which partial voluming from adjacent ventricular CSF may have contributed. This possibility was tested directly by systematically manipulating the size of anatomically-defined ROI by morphometric erosion and dilation of the perimeter of a region (Pfefferbaum and Sullivan, 2003). The purposeful inclusion of non-WM voxels into regional DTI measures, of ADC (or trace) resulted in a significant increase

561

Adolf Pfefferbaum and Edith V. Sullivan

FA 0.70

Genu r  0.37, P  0.003

0.80

0.65

0.75

0.60

0.70

0.55

0.65

0.50

0.60

0.45

0.55

0.40 20 30 40 50 60 70 80 90 Age

0.50

Splenium r  0.29, P  0.02

0.50

Centrum r  0.79, P  0.0001

FA

0.45

0.40

0.35

20 30 40 50 60 70 80 90 Age

0.30 20 30 40 50 60 70 80 90 Age

Trace 460 Trace (10 6 mm2/s)

562

Genu r  0.50, P  0.0001

440 420

380

Splenium r  0.24, P  0.07

370

360

Centrum r  0.58, P  0.0001

350

360

340

350 400

330 340

380

320

330

360

320

310

340 20 30 40 50 60 70 80 90 Age

310 20 30 40 50 60 70 80 90 Age

300 20 30 40 50 60 70 80 90 Age

Fig. 33.2 Cross-sectional measures of FA and trace across the adult age span in three WM ROI. FA decreases and trace increases with advancing age.

in trace together with an even greater decrease in FA as the target region was dilated. Further affected were the correlations between each DTI measure and age, such that morphological dilation resulted in progressively lower correlations between trace or FA and age in WM and higher correlations in GM. This experiment highlights the importance of accurate segmentation and how partial voluming can result in spuriously high or low correlations with age. Anisotropy of corpus callosum is under genetic regulation even in old age (Pfefferbaum et al., 2001b). In a study comparing heritability of regional callosal FA in monozygotic (MZ) twin pairs relative to dizygotic (DZ) twin pairs, the relative proportion of genetic to environmental contributions to FA variance differed regionally, being 5 : 1 for total midsagittal callosal area, 3 : 1 for the splenium, and 1 : 1

for the genu. This heritability pattern suggested greater environmental influences on anterior than posterior reaches of the corpus callosum. This finding complements an observation made in a crosssectional developmental study, showing that ADC declines rapidly over the first 6 months of life in frontal but not occipital WM and implicates earlier myelination of occipital than frontal fibers (Nomura et al., 1994). Several brain structure–function relationships in normal individuals have been observed with DTI measures. Using the alternating finger tapping task, a test of interhemispheric information transfer, Sullivan et al. (2001a) observed a relationship between splenium and parietal pericallosal WM FA and finger tapping output that was selective to the alternating condition but not to the unimanual

Diffusion MR imaging in neuropsychiatry and aging

31 yo

71 yo

Fig. 33.3 Axial images (SPGR on left and FA on right) at the level of the lateral ventricles in young and elderly healthy adult volunteers.

conditions (Figure 33.4). These healthy controls also showed more generalized relationships between FA in a number of brain regions and quantitative tests of balance. In addition, O’Sullivan et al. (2001) reported that lower attentional set shifting scores (Trails B and A) correlated with greater diffusivity in an anterior brain region and that lower verbal fluency scores correlated with lower FA in central sample of WM. Stebbins and colleagues (2001) reported a

correlation between age-related decline in frontal WM FA in healthy adults and performance on a test of reasoning (Raven’s progressive matrices). Studies of normal, healthy men and women across the adult age range document normative agerelated, DTI-detected changes in brain WM that is requisite for interpreting DTI data from individuals with neuropsychiatric disorders and other conditions affecting the brain. Norms also require the

563

Adolf Pfefferbaum and Edith V. Sullivan

0.80

0.70

0.65 r  0.56 P  0.02

0.60 100

150

200

250

300

Alternated finger tapping output

Parietal pericallosal FA

0.60

0.75 Splenium FA

564

0.55 0.50 0.45 0.40 0.35 100

r  0.61 P  0.008 150

200

250

300

Alternated finger tapping output

Fig. 33.4 Relationship between posterior callosal and pericallosal FA and an interhemispheric transfer task of alternated finger tapping task. Higher FA is associated with better performance.

establishment of measurement reliability, which is also essential for longitudinal and multi-site studies. A study of five normal adults imaged five times each resulted in 5% across-subject variation and a mean variation of 2.3  1.2 SD% in ADC (Naganawa et al., 2003). In another study, the reliability of global and regional FA and trace measurements was estimated in 10 healthy, young adults, who were imaged three times on two different 1.5-T scanners made by the same manufacturer (Pfefferbaum et al., 2003). When reliability was measured on a voxel-byvoxel basis or on a slice-by-slice basis, FA and trace values were equivalently and significantly higher within than across scanners. For the regional analysis of the corpus callosum, the coefficients of variation of anisotropy and diffusivity were lower within scanners (3%) than across scanners (4.5% for FA and 7.5% for trace). A common assumption is that age and disease result in decrease in anisotropy and increase in diffusivity; however, selective deletion of WM fibers with a uniform orientation from a tissue sample of crossing fibers can result in increased anisotropy and increased or decreased diffusivity (Pierpaoli et al., 2001). Therefore, knowledge of the underlying regional architecture can guide interpretation (Shimony et al., 1999; Virta et al., 1999; Pierpaoli et al., 2001). FA varies widely in different WM regions and structures; for example, when partial voluming is controlled, FA is approximately 0.70 in the

splenium to 0.60 in the genu to 0.40 in centrum semiovale (Pfefferbaum and Sullivan, 2003). Other likely influences on DTI measures include increased interstitial CSF in WM per se (e.g. leukoaraiosis) and partial volume effects (PVE) from inclusion of GM and/or CSF in the WM sample. Consequently, it is useful to evaluate regional anisotropy values within the context of published values from studies that have rigorously controlled for regional anatomy and PVE.

Alzheimer’s disease AD is a progressive neurodegenerative disorder, occurring in 10% of individuals over 65 and nearly 50% of those over 85 years of age. AD initially affects medial temporal lobe structures, most notably the hippocampus and entorhinal cortex (in vivo: (Laakso et al., 1995; Jack et al., 1998, 2002; Killiany et al., 2000)), with later involvement of temporal and parietal neocortex (post-mortem: (Braak and Braak, 1994; Brun, 1994; Kemper, 1994)). The classical pattern of neuropsychological deficits follow this pattern, with earliest and most prominent deficits in ability to form new memories and additional deficits likely to include visuospatial and language abilities (Cummings, 2000). It is now recognized that volume loss of WM in corpus callosum (Teipel et al., 2002) and central WM (Salat et al.,

Diffusion MR imaging in neuropsychiatry and aging

Fig. 33.5 Coronal images (SPGR on left, FA in middle, and trace on right) at the level of the lateral ventricles and hippocampus in a 74-year-old, healthy woman and a 74-year-old woman with probable AD.

1999) can accompany the prominent GM loss and is suggestive of Wallerian degeneration and regression of axonal processes with neuronal death. Such WM degradation may well be detected earlier in the course of the disease with DTI than with conventional imaging. Diffusion imaging is potentially sensitive to severity of AD (Figure 33.5). One study revealed a stepwise decrease in regional anisotropy related to classification of AD patients as “possible” and “probable”. Anisotropy in the temporal stem, but not the hippocampus, of the “possible” AD group was significantly lower (mean was approximately 0.4) than that in the controls (mean  0.6), and anisotropy of the “probable” AD group was lower still (mean  0.2) (Hanyu et al., 1998). Decreased anisotropy in the presence of normal diffusivity was interpreted as reflecting gliosis or accumulation of cell inclusion

material, such as tangles and plaques in the AD group. Another study noted lower anisotropy in posterior WM and higher diffusivity in the hippocampus of AD patients compared with controls (Sandson et al., 1999). The AD effect endured after covarying for global measures of atrophy and WMH, although the measures did not correlate with dementia severity. The fiber structure of the corpus callosum is relatively homogeneous in orientation as it courses between the hemispheres. Orientationally-specific diffusivity parallel (but not perpendicular) to callosal fibers was observed to be higher in AD patients than age-matched controls. These group differences were present in the genu and splenium but not the body of the corpus callosum, and Mini-Mental State Examination (MMSE) scores correlated significantly with splenium DTI measures in these AD patients

565

566

Adolf Pfefferbaum and Edith V. Sullivan

(Hanyu et al., 1999a). Another study, however, failed to detect any significant abnormalities in diffusivity in AD patients in brain regions that exhibited significant perfusion deficits in temporoparietal and sensorimotor regions and hippocampus; group differences in atrophy were controlled by specifying each ROI to be 50 pixels in size (Bozzao et al., 2001). Examination of diagnostic specificity using diffusion imaging has been done by contrasting diffusion tensor measures in patients with different agerelated dementing disorders. Patients with vascular dementia of the Binswanger type and patients with AD plus periventricular hyperintensities were shown to have equivalently abnormally high ADC globally in WM; however, the groups had regional differences, with the vascular group having higher ADC in anterior WM, and the AD group having higher ADC in posterior WM (Hanyu et al., 1999b). Another study noted the superior sensitivity of DWI over conventional T2-weighted imaging in detecting abnormal brain WM in patients with vascular dementia relative to healthy controls and observed abnormally high diffusion in WM regions extending well beyond those identified with conventional imaging (Assaf et al., 2002). Regional differences in diffusivity and anisotropy were measured in patients with mild cognitive impairment (MCI), which is considered to be a precursor of clinically-detectable AD, compared with AD and control groups (Kantarci et al., 2001). Both the MCI and the AD groups had higher ADC globally than controls, whereas only the AD patients showed significant diffusion and anisotropy abnormalities in temporal and parietal regions. These authors have interpreted these imaging results as reflecting increased extracellular space attributed to myelin and axonal loss (cf. (Schwartz, 2001)). Another approach used for DTI quantification is to calculate the voxel-based distribution of diffusivity of large regions of brain tissue presented as an histogram, before and after segmentation into GM and WM. A study based on histogram analysis observed significant AD-related abnormalities in the form of lower peaks and greater diffusion in GM of the temporal but not parietal lobe, and peak values correlated with MMSE scores. Diffusivity in WM in both regions examined was greater in AD patients than controls (Bozzali et al., 2001). In a later study

(Bozzali et al., 2002), this research group used three DTI indices – mean diffusivity, FA, and intervoxel coherence (C) – to examine regional WM in AD patients and controls. Whereas FA is a measure of the magnitude and orientation of diffusion on an intravoxel basis, C provides an orientational measure on an intervoxel basis, i.e. the degree to which the diffusion orientation of a voxel is similar to its neighbors (Pfefferbaum et al., 2000b). Relative to the controls, the AD group showed high diffusivity and low FA values in the corpus callosum and WM of the frontal, temporal, and parietal lobes, whereas the groups showed no statistically significant differences in these DTI measures in the WM of the occipital lobe, internal capsule, or pericallosal regions. In contrast to these measures, C did not distinguish the groups in any brain region examined, suggesting local disruption rather than disconnection per se as characterizing the WM abnormality. The LI, a measure of intervoxel orientational coherence similar to C, has also been used to evaluate WM integrity, and, like anisotropy, typically decreases in conditions disrupting WM microstructure. Relative to controls, the LI of AD patients was significantly lower in the splenium of the corpus callosum, superior longitudinal fasciculus, and left cingulum (Rose et al., 2000). By contrast, the AD and control groups did not differ on diffusion measures in the pyramidal tract, thus providing evidence for selectivity of the identified regional abnormalities. Further, LI of the splenium was functionally relevant in that it correlated significantly with MMSE scores. Perhaps the first DTI study of AD performed on a 3 T system examined FA in nine WM brain regions (Takahashi et al., 2002). Controls and AD patients showed similar variability in FA across regions, with the highest FA in the internal capsule and lowest in lobar WM. Relative to age-matched controls, the AD group had significantly lower FA temporal WM, anterior and posterior cingulate bundles, and posterior corpus callosum. This pattern of abnormality comports with patterns of regional GM degeneration typically observed in AD and may be indicative of axonal deletion secondary to neuronal death. Taken together, these studies provide support for the utility of MR diffusion imaging to detect disruption of regional WM microstructure in vivo,

Diffusion MR imaging in neuropsychiatry and aging

perhaps to permit early detection of AD, and to indicate likely pathological mechanisms underlying the disruption. Simultaneous application of different measures of diffusivity and anisotropy have the potential of distinguishing between types of microstructural damage in AD. For example, with disruption of the microtubule system and myelination, anisotropy should decrease while diffusivity increases, but with gliotic reaction to death of cells, including oligodendrocytes, diffusivity along with anisotropy can decrease.

Schizophrenia Schizophrenia is a debilitating disorder, occurring in approximately 1% of the population, with higher frequency in men than women. Although an initial schizophrenia break can occur in late life, the more common form presents between late adolescence and the fourth decade of life. Symptoms manifest as hallucinations, social withdrawal, and cognitive disability, often in victims who had previously performed adequately or even well in school and life in general. Brain abnormality was long suspected to underlie schizophrenia, but it was not until the introduction of in vivo brain imaging that subtle yet significant brain dysmorphology was identified systematically. In addition to enlarged third and lateral ventricles observable on computed tomography (CT) ((e.g. Illowsky et al., 1988; Pfefferbaum et al., 1988; Vita et al., 1988); for reviews (Marsh et al., 1996; Pearlson and Marsh, 1999; Shenton et al., 2001)), quantitative MRI revealed widespread cortical GM volume deficits (Zipursky et al., 1992), most prominent in prefrontal and anterior temporal cortices (Schlaepfer et al., 1994; Sullivan et al., 1998; Kubicki et al., 2002a; Bartzokis et al., 2003). The selectivity of frontal GM volume deficits, relative to posterior GM or frontal WM volumes, was recently confirmed in neuropathological study (Selemon et al., 2002). Although WM volume deficits have seldom been reported with in vivo methods (Breier et al., 1992; Buchanan et al., 1998; Wolkin et al., 1998), postmortem study has provided evidence for delayed myelination in frontal WM (Benes et al., 1994; Benes, 1997) and displacement of interstitial WM neurons

in prefrontal and temporal regions (Akbarian et al., 1996). In vivo MR spectroscopic imaging (MRSI) identified low levels of N-acetyl aspartate (NAA) in the WM of patients with schizophrenia (Lim et al., 1998), suggesting compromised tissue integrity despite normal volume and disordered neuronal connectivity (e.g. McGuire and Frith, 1996), setting the stage for later investigations with DTI. Buchsbaum and colleagues (1998) published the first application of DTI to the study of schizophrenia. This analysis was based on a SPM analysis (Friston et al., 1995) and indicated abnormally low relative anisotropy (RA) in prefrontal WM tracts, especially of the right hemisphere. A growing number of DTI studies has reported either widespread or regional deficits in FA in groups of schizophrenic patients compared with controls. Lim and colleagues observed a 1.5 SD effect size, where the schizophrenic group had abnormally low FA throughout the brain’s WM relative to controls despite absence of group differences in WM volume (Lim et al., 1999). A follow-up analysis showed that the FA abnormality was not related to T2 prolongation in WM; such a correlation would have been indicative of greater presence of interstitial water in the schizophrenics (Pfefferbaum et al., 1999). Widespread, abnormally low WM FA was again observed in a mixed group of schizophrenic and schizo-affective patients compared with age-matched controls (Nierenberg et al., 2003). Other studies investigated regional microstructural integrity of the corpus callosum and reported greater diffusivity or lower FA in the splenium but not the genu (Foong et al., 2000; Agartz et al., 2001). The functional relevance of diffusion measures to psychiatric symptomatology, measured quantitatively, was supported by two separate analyses of schizophrenic men. In the first study, lower FA was a significant predictor of high motor impulsivity, and high trace was a predictor of greater aggression (Hoptman et al., 2002). In the second study, low FA was related to more severe negative symptoms (Wolkin et al., 2003). That these clinical measures were related to inferior but not superior frontal WM samples demonstrates selectivity of these relationships and suggests that disruption or abnormality of pathways involving inferior reaches of frontal lobes

567

568

Adolf Pfefferbaum and Edith V. Sullivan

contribute to these symptoms. Similarly, low FA in inferior WM regions was present in schizophrenic patients with abnormal visual evoked potentials, suggestive of deficits in primary visual processing (Butler et al., 2003). Not all studies of schizophrenia have observed abnormalities in WM with DTI (Steel et al., 2001; Foong et al., 2002). Although Kubicki and colleagues (2002b) also did not identify group differences in DTI measures, tractography revealed that schizophrenics did not exhibit the asymmetry detected in the normal control group, in which left hemisphere FA was greater than right hemisphere FA in the uncinate fasciculus, a WM tract that connects the anterior temporal and frontal cortices. In another study, examination of first episode schizophrenic patients failed to reveal group differences in DTI measures; however, within the patient group, regionally lower FA was related to poorer performance on cognitive tasks (Trail Making, Wisconsin Card Sorting, and verbal fluency) known to rely on frontal lobe functioning (Carbon et al., 2003). Disruption of these frontal and temporal connections may contribute to cognitive disturbances in organized thinking and memory characteristic of schizophrenia (e.g. Ford et al., 2002).

Alcoholism According to the World Health Organization, chronic alcohol abuse is a universal problem with enormous social and economic costs that is growing among the young and has exceptionally high prevalence among indigenous peoples in developed countries. The cognitive, motor, and social behavioral sequelae are mild to devastating, and with prolonged sobriety certain domains of dysfunction may improve or resolve (Brandt et al., 1983; Rourke and Grant 1999; Sullivan et al., 2000). Neuropathological studies of patients with chronic alcoholism indicate that brain WM is especially affected (Harper et al., 1985; De la Monte, 1988; Badsberg-Jensen and Pakkenberg, 1993; Harper, 1998) regardless of sex (Harper et al., 1990). Abnormalities identified in WM on post-mortem examination include volume reduction, demyelination, loss of myelinated fibers, and axonal deletion possibly arising from regional neuronal loss (Alling

and Bostrom, 1980; Harper et al., 1987; Harper and Kril, 1989; Kril et al., 1997). The corpus callosum is also affected (Harper and Kril, 1988; TarnowskaDziduszko et al., 1995), especially in alcoholics who have experienced nutritional deficiencies. In their most extreme forms, these WM conditions can be lifethreatening and are associated with MarchiafavaBignami disease and central pontine myelinolysis (Victor et al., 1989; Charness, 1993). MRI of brain macrostructure in alcoholic men are consistent with the post-mortem studies with respect to WM volume shrinkage in the cerebrum (Pfefferbaum et al., 1992; Pfefferbaum et al., 1997; Hommer et al., 2001), including corpus callosum (Pfefferbaum et al., 1996; Estruch et al., 1997). Alcohol’s effect on the brains of women is controversial. One set of quantitative studies (Hommer et al., 2001) provides evidence for significant whole brain WM volume deficits of greater proportion in alcoholic women than alcoholic men with similar sex-dependent differences observed in cross-sectional area of the corpus callosum (Hommer et al., 1996). Another study observed WM volume deficits, estimated from a large sample of the centrum semiovale, in alcoholic men but not women, even when subgroups of alcoholic men and women were matched for age and lifetime alcohol consumption (Pfefferbaum et al., 2001a). These differences notwithstanding, chronic alcoholism results in compromised cognitive and motor functions in both men and women and is indicative of brain dysfunction. Surprisingly few studies in chronic alcoholics have successfully detected selective brain structure– function relationships based on in vivo MRI acquired with conventional methods despite the consistency with which abnormalities in brain macrostructure and function are observed. Considering the toll alcohol toxicity takes on brain WM, DTI measures may be more sensitive than conventional imaging in this endeavor. To date, only two group studies have been published based on DTI in chronic alcoholism. In the first study (Pfefferbaum et al., 2000a), alcoholic men were observed to have lower FA than age- and sex-matched controls in the corpus callosum and centrum semiovale (Figure 33.6). A significant correlation between FA in the genu and duration of sobriety suggests chronic

Diffusion MR imaging in neuropsychiatry and aging

Control

Alcoholic

Fig. 33.6. Coronal images (SPGR on left, FA on right) at the level of the lateral ventricles and hippocampus in a 61-year-old, healthy man and a 61-year-old man with chronic alcoholism.

alcohol consumption as the mechanism of the FA deficit. Although a more macrostructural measure based on intervoxel-to-intervoxel orientational diffusion coherence (C) (Pfefferbaum et al., 2000b) was upwards of 1 SD lower in the alcoholics than controls, the overall group difference was not statistically significant. These indices of regional WM microstructure showed evidence of functional meaningfulness because working memory performance correlated selectively with intravoxel anisotropy (FA) in the splenium, whereas scores on attention correlated selectively with intervoxel coherence (C) in the genu. The second study examined alcoholic women (Pfefferbaum and Sullivan, 2002) and questioned whether detection of WM abnormalities, which was

not forthcoming with conventional MRI, was possible with DTI. Relative to age- and sex-matched controls, alcoholic women had lower FA and C in the genu of the corpus callosum and centrum semiovale despite absence of group differences in midsagittal regional callosal areas. Morphometric erosion of the ROI to minimize the potential effects of partial voluming yielded the same alcoholism-related abnormality. Lower genu FA was related to greater lifetime consumption of alcohol even after controlling for age, again suggesting alcohol as the mechanism of WM microstructural changes. A further analysis compared the DTI data of alcoholic women with those of the alcoholic men from the previous study. This comparison indicated a similar FA deficit in the genu and centrum semiovale in alcoholic men and

569

570

Adolf Pfefferbaum and Edith V. Sullivan

women. A deficit in splenium FA was detected only in the alcoholic men, whereas a deficit in centrum C was detected only in the alcoholic women. Additional reports present case studies of patients with alcoholic Wernicke’s encephalopathy, which results from thiamine deficiency, is marked by neurological signs of ophthalmoplegia, ataxia of gait, and altered mental state, is characterized by mammillary body lesions, and if left untreated can be the precursor of potentially permanent global amnesia (Korsakoff’s Syndrome) (Victor et al., 1989; Caine et al., 1997; Sheedy et al., 1999). Bergui et al. (2001) used conventional and DWI in a 31-year-old alcoholic man with Wernicke’s encephalopathy. T2-weighted images and ADC revealed high signal intensity specific to tissue in the mammillary bodies on admission to hospital that resolved 2 weeks later and following thiamine treatment. The authors concluded that the high diffusion was indicative of extracellular edema rather than cellular damage, which would not have resolved in follow-up. Doherty et al. (2002) reported two cases of Wernicke’s encephalopathy, one resulting from alcohol abuse and the other from anorexia. ADC was low in the alcoholic, thus paralleling the pattern observed in the chronic phase of ischemic stroke. The anorexic showed no evidence for abnormalities in diffusion imaging parameters but only observed a “T2-shinethrough effect” (i.e. increased T2 in the ROI), suggestive of remote injury. Concurring with this result is a study of a 71-year-old man with a 2 week history of behavioral disorders comporting with Wernicke’s encephalopathy and hyperintense DWI signal in midbrain and medial thalamic regions less visible with T2-weighted or fluid attenuated inversion recovery (FLAIR) images (Kashihara and Irisawa, 2002). A third case study indicated that, although DWI was useful in visually identifying thalamic signal abnormalities in an alcoholic with Wernicke’s encephalopathy, ADC values were normal (Ducreux et al., 2002). Finally, a case of non-alcohol substance dependence presented with symptoms of Wernicke’s encephalopathy. Serial imaging revealed signal hyperintensity with DWI and FLAIR at initial scanning with normalization of signal intensity, especially on DWI, following thiamine treatment (Niclot et al., 2002). Discrepancies among these case studies

may reflect differences in stage of recovery from Wernicke’s encephalopathy during which DWI was acquired or etiological differences. A case study of an alcoholic woman with Marchiafava–Bignami disease reported low diffusion in regions of the corpus callosum with high signal intensity seen with T1- and T2-weighted images (Inagaki and Saito, 2000). The authors noted that this pattern is different from other demyelinating conditions, such as multiple sclerosis (MS), but may be a precursor of demyelination. In light of the consistency of neuropathological reports of selective WM degradation in alcoholism (e.g. Harper and Kril, 1990), DTI offers an especially relevant and safe imaging method for longitudinal studies. Repeat examination would serve to track the condition of WM microstructure over the course of alcoholism and possibly detect neuropathological markers of serious but potentially reversible conditions, such as Wernicke’s encephalopathy or central pontine myelinolysis, at a presymptomatic stage.

Human immunodeficiency virus infection (HIV+) Over 42 million children and adults were estimated to have been living with human immunodeficiency virus (HIV) at the end of 2002 (UNAIDS, 2002), with 5 million new infections occurring during that year. HIV infection has a strong predilection for the brain (Masliah et al., 2000), with upwards of 90% of AIDS patients manifesting CNS lesions (Trillo-Pazos and Everall, 1997). The initial stages of HIV infection are asymptomatic but an AIDS-defining illness will eventually emerge to mark the clinical diagnosis of AIDS and the onset of concomitant events affecting the brain. Among these events are lymphomas and opportunistic infections that generally present with well-defined space occupying lesions on neuroimaging examination (e.g. (Post et al., 1993, 1999; Ruiz et al., 1997)). More difficult to detect and quantify, however, are the sequelae of direct HIV infection of the brain that precede full-blown giant cell encephalopathy. Although traditional clinical MR proton density and T2-weighted imaging do not

Diffusion MR imaging in neuropsychiatry and aging

consistently report abnormalities in asymptomatic HIV-infected individuals, such abnormalities can be detected (Manji et al., 1994) and quantified with segmentation algorithms (Jernigan et al., 1993). Particularly affected are frontal WM and basal ganglia where tissue volume reduction occurs (Aylward et al., 1993; Jernigan et al., 1993). Evidence from contrast-enhancement studies suggests a cerebrovascular genesis, related to increased permeability of the blood brain barrier, for neuropathology in the basal ganglia (Sacktor et al., 1996; Berger et al., 2000). In symptomatic cases, abnormalities such as ventricular enlargement, global atrophy, and frontal WM and caudate volume loss (Aylward et al., 1995; Di Sclafani et al., 1997; Stout et al., 1998; Symonds et al., 1999) have been reported. Cerebellar degeneration ((Tagliati et al., 1998); but cf. Sclar et al., (2000)) and central pontine myelinolysis have also been reported (Miller et al., 1998). Chang and Ernst (1997) published the initial application of DTI to the study of HIV infection and reported that DTI was capable of detecting toxoplasmosis and progressive multifocal leukoencephalitis. Following this study, Ulug et al. (2000) reported abnormal ADC and FA in normal appearing periventricular WM and corpus callosum in HIV-infected subjects, with diffusion parameters correlating with signs of disease severity, namely CD4 count and viral load. DTI studies (Filippi et al., 2001; Pomara et al., 2001; Schaefer et al., 2001) have identified microstructural abnormalities in tissue that appears normal with conventional MRI. In one study (Filippi et al., 2001), HIV-infected patients with viral loads ranging from undetectable to 400,000 copies/mm3 showed reduction in WM anisotropy and increase in average water diffusivity in the splenium, genu, and frontal and parietal subcortical WM samples; these DTI measures were associated with viral load. Another study (Pomara et al., 2001) identified abnormally low FA in the WM of the frontal lobes of six HIV-infected patients compared with nine controls. In addition, abnormally high FA was present in the internal capsule, possibly reflecting disruption or deletion of selective crossing fibers. By contrast, group differences in these regions were not detectable with mean diffusivity, computed proton density, T2 times, or T2-weighted MRI. These studies provide

evidence to support the hypothesis that inflammation affecting WM fibers and their cytoskeletal constituents is at least one mechanism by which HIV infection affects the brain. Furthermore, DTI studies may provide an early indication of the potential for an individual HIV patient to ultimately develop dementia as a result of their illness (Berger and Avison, 2001). Examination with DTI may also serve to identify interactions with common comorbidities, most notably alcohol abuse (Pfefferbaum et al., 2002), and to detect and track repair with positive pharmacological treatment.

Depression DTI has been applied in two studies of depression, and both studies focused on late-life depression. The first study (Taylor et al., 2001) examined potential differences between controls and depressed geriatric patients in the microstructure of MR-detected WM signal hyperintensities, which are known to occur with greater than normal frequency in latelife depression (e.g. Kramer-Ginsberg et al., 1999; O’Brien et al., 2000). DTI revealed higher ADC and lower FA in WM marked by hyperintensities than in normal WM, but the control and depressed groups did not differ from each other in any DTI measure. The second study used regional FA measures as predictors of pharmacological response in older patients with depression (Alexopoulos et al., 2002). Depressed patients with high FA in WM superior to the anterior commissure–posterior commissure (AC–PC) plane had a shorter time to remission of depressive symptoms than those with low FA in this WM region. Rates of remission were not related to FA in WM inferior to the AC–PC plane.

Conclusion This growing body of research based on diffusion imaging substantiates the utility of this MR imaging method in detecting patterns of sparing and involvement of regional WM microstructural integrity in normal aging and neuropsychiatric conditions, many

571

572

Adolf Pfefferbaum and Edith V. Sullivan

of which are characterized by changes in brain structural integrity, subtle on the macrostructural level but more reliably detectable on the microstructural level. The target of DTI is microstructure, which is a departure from the traditional macrostructural methods of volume assessment, and may permit insight into neural mechanisms, regional patterns, and course of WM and cell degeneration of normal aging and degenerative diseases. Because of initial successes in distinguishing clinically similar disorders and in detecting abnormalities in brain microstructure overlooked by MRI techniques targeting macrostructure, additional neuropsychiatric conditions, such as obsessive-compulsive disorder, and developmental syndromes, such as autism, will be investigated with diffusion imaging. Improvement in tractography should yield a method to monitor WM fiber regrowth following trauma or disease and to test models of WM recovery, such as the possibility that WM growth is facilitated in parallel but inhibited in non-parallel orientation (Pettigrew and Crutcher, 2001). Finally, DTI tractography combined with MR functional imaging should guide in differentiating brain areas activated while performing a task that form a circuit from those activated coincidentally during task engagement (cf. (Molko et al., 2002)).

ACKNOWLEDGEMENTS

This project was supported by the United States National Institute on Alcohol Abuse and Alcoholism (AA05965, AA10723, AA12888, AA12999) and the National Institute on Aging (AG17919). The authors wish to thank Margaret J. Rosenbloom, M.A. for invaluable assistance in manuscript preparation.

REFERENCES Abe O, Aoki S, Hayashi N, Yamada H, Kunimatsu A, Mori H, Yoshikawa T, Okubo T, Ohtomo K. 2002. Normal aging in the central nervous system: quantitative MR diffusion-tensor analysis. Neurobiol Aging 23: 433–441. Aboitiz F, Rodriguez E, Olivares R, Zaidel E. 1996. Age-related changes in fibre composition of the human corpus callosum: sex differences. Neuroreport 7: 1761–1764.

Agartz I, Andersson JL, Skare S. 2001. Abnormal brain white matter in schizophrenia: a diffusion tensor imaging study. Neuroreport 12: 2251–2254. Akbarian S, Kim JJ, Potkin SG, Hetrick WP, Bunney WE, Jones EG. 1996. Maldistribution of interstitial neurons in prefrontal white matter of the brains of schizophrenic patients. Archiv Gen Psychiatry 53: 425–436. Alexopoulos GS, Kiosses DN, Choi SJ, Murphy CF, Lim KO. 2002. Frontal white matter microstructure and treatment response of late-life depression: a preliminary study. Am J Psychiatry 159: 1929–1932. Alling C, Bostrom K. 1980. Demyelination of the mamillary bodies in alcoholism. A combined morphological and biochemical study. Acta Neuropathol (Berl) 50: 77–80. Arfanakis K, Cordes D, Haughton VM, Carew JD, Meyerand ME. 2002. Independent component analysis applied to diffusion tensor MRI. Magn Reson Med 47: 354–363. Assaf Y, Mayzel-Oreg O, Gigi A, Ben-Bashat D, Mordohovitch M, Verchovsky R, Reider II G, Hendler T, Graif M, Cohen Y, Korczyn AD. 2002. High b value q-space-analyzed diffusion MRI in vascular dementia: a preliminary study. J Neurol Sci 15: 203–204, 235–239. Aylward EH, Brettschneider PD, McArthur JC, Harris GJ, Schlaepfer TE, Henderer JD, Barta PE, Tien AY, Pearlson GD. 1995. Magnetic resonance imaging measurement of gray matter volume reductions in HIV dementia. Am J Psychiatry 152: 987–994. Aylward EH, Henderer JD, McArthur JC, Brettschneider PD, Harris GJ, Barta PE, Pearlson GD. 1993. Reduced basal ganglia volume in HIV-1-associated dementia: results from quantitative neuroimaging. Neurology 43: 2099–2104. Badsberg-Jensen G, Pakkenberg B. 1993. Do alcoholics drink their neurons away? Lancet 342: 1201–1204. Bartzokis G, Nuechterlein KH, Lu PH, Gitlin M, Rogers S, Mintz J. 2003. Dysregulated brain development in adult men with schizophrenia: a magnetic resonance imaging study. Biol Psychiatry 53: 412–421. Basser PJ. 1995. Inferring microstructural features and the physiological state of tissues from diffusion-weighted images. NMR Biomed 8: 333–344. Basser PJ. 1998. Fiber-tractography via diffusion tensor MRI (DT-MRI). Proceedings of the International Society of Magnetic Resonance in Medicine, 6th Meeting, 1226. Basser J, Pierpaoli C. 1996. Microstructural and physiological features of tissues elucidated by quantitative diffusion tensor MRI. J Magn Reson 111: 209–219. Benes FM. 1997. What an archaeological dig can tell us about macro- and microcircuitry in brains of schizophrenia subjects. Schizophrenia Bull 23: 503–507. Benes FM, Turtle M, Khan Y, Farol P. 1994. Myelination of a key relay zone in the hippocampal formation occurs in the

Diffusion MR imaging in neuropsychiatry and aging

human brain during childhood, adolescence, and adulthood. Archiv Gen Psychiatry 51: 477–484. Berger JR, Avison MJ. 2001. Diffusion tensor imaging in HIV infection: what is it telling us? Am J Neuroradiol 22: 237–238. Berger JR, Nath A, Greenberg RN, Andersen AH, Greene RA, Bognar A, Avison MJ. 2000. Cerebrovascular changes in the basal ganglia with HIV dementia. Neurology 54: 921–926. Bergui M, Bradac GB, Zhong JJ, Barbero PA, Durelli L. 2001. Diffusion-weighted MR in reversible Wernicke Encephalopathy. Neuroradiology 43: 969–972. Blatter DD, Bigler ED, Gale SD, Johnson SC, Anderson C, Burnett BM, Parker N, Kurth S, Horn S. 1995. Quantitative volumetric analysis of brain MRI: normative database spanning five decades of life. Am J Neuroradiol 16: 241–245. Bozzali M, Falini A, Franceschi M, Cercignani M, Zuffi M, Scotti G, Comi G, Filippi M. 2002. White matter damage in Alzheimer’s disease assessed in vivo using diffusion tensor magnetic resonance imaging. J Neurol Neurosurg Psychiatry 72: 742–746. Bozzali M, Franceschi M, Falini A, Pontesilli S, Cercignani M, Magnani G, Scotti G, Comi G, Filippi M. 2001. Quantification of tissue damage in AD using diffusion tensor and magnetization transfer MRI. Neurology 57: 1135–1137. Bozzao A, Floris R, Baviera ME, Apruzzese A, Simonetti G. 2001. Diffusion and perfusion MR imaging in cases of Alzheimer’s disease: correlations with cortical atrophy and lesion load. Am J Neuroradiol 22: 1030–1036. Braak H, Braak E. 1994. Morphological criteria for the recognition of Alzheimer’s disease and the distribution pattern of cortical changes related to this disorder. Neurobiol Aging 15: 355–356. Brandt J, Butters N, Ryan C, Bayog R. 1983. Cognitive loss and recovery in long-term alcohol abusers. Archiv Gen Psychiatry 40: 435–442. Breier A, Buchanan RW, Elkashef A, Munson RC, Kirkpatrick B, Gellad F. 1992. Brain morphology and schizophrenia – a magnetic resonance imaging study of limbic, prefrontal cortex, and caudate structures. Archiv Gen Psychiatry 49: 921–926. Brun A. 1994. Regional rather than global pathology decides symptoms in senile dementia of Alzheimer’s type. Neurobiol Aging 15: 367–368. Buchanan RW, Vladar K, Barta PE, Pearlson GD. 1998. Structural evaluation of the prefrontal cortex in schizophrenia. Am J Psychiatry 155: 1049–1055. Buchsbaum MS, Tang CY, Peled S, Gudbjartsson H, Lu D, Hazlett EA, Downhill J, Haznedar M, Fallon JH, Atlas SW. 1998. MRI white matter diffusion anisotropy and PET metabolic rate in schizophrenia. Neuroreport 9: 425–430. Butler PD, Lim KO, Nierenberg J, Hoptman MJ, Choi SJ,

visual dysfunction in schizophrenia: relationship to white matter integrity inferred from diffusion tensor imaging (abstract). Schizophrenia Res 60: 190S. Caine D, Halliday GM, Kril JJ, Harper CG. 1997. Operational criteria for the classification of chronic alcoholics: identification of Wernicke’s encephalopathy. J Neurol Neurosurg Psychiatry 62: 51–60. Carbon M, Bates J, Bilder RM, Lim KO. 2003. Microstructural white matter changes as correlates of impaired performance in first episode schizophrenia (abstract). Schizophrenia Res 60: 191S. Chang L, Ernst T. 1997. MR spectroscopy and diffusionweighted MR imaging in focal brain lesions in AIDS. Neuroimaging Clin North America (Neuroimaging of AIDS II) 7: 409–426. Charness ME. 1993. Brain lesions in alcoholics. Alcoholism: Clinical and Experimental Research 17: 2–11. Chen ZG, Li TQ, Hindmarsh T. 2001. Diffusion tensor trace mapping in normal adult brain using single-shot EPI technique. A methodological study of the aging brain. Acta Radiologica 42: 447–458. Chepuri NB, Yen YF, Burdette JH, Li H, Moody DM, Maldjian JA. 2002. Diffusion anisotropy in the corpus callosum. Am J Neuroradiol 23: 803–808. Chun T, Filippi CG, Zimmerman RD, Ulug AM. 2000. Diffusion changes in the aging human brain. Am J Neuroradiol 21: 1078–1083. Craik FIM, Morris LW, Morris RG, Loewen ER. 1990. Relations between source amnesia and frontal lobe functioning in older adults. Psychol Aging 5: 148–151. Cummings JL. 2000. Cognitive and behavioral heterogeneity in Alzheimer’s disease: seeking the neurobiological basis. Neurobiol Aging 21: 845–861. De la Monte SM. 1988. Disproportionate atrophy of cerebral white matter in chronic alcoholics. Archiv Neurol 45: 990–992. Di Sclafani V, Mackay RD, Meyerhoff DJ, Norman D, Weiner MW, Fein G. 1997. Brain atrophy in HIV infection is more strongly associated with CDC clinical stage than with cognitive impairment. J Int Neuropsychol Soc 3: 276–287. Doherty MJ, Watson NF, Uchino K, Hallam DK, Cramer SC. 2002. Diffusion abnormalities in patients with Wernicke encephalopathy. Neurology 58: 655–657. Ducreux D, Petit-Lacour MC, Benoudiba F, Castelain V, Marsot-Dupuch K. 2002. Diffusion-weighted imaging in a case of Wernicke encephalopathy. J Neuroradiol 29: 39–42. Engelter ST, Provenzale JM, Petrella JR, DeLong DM, MacFall JR. 2000. The effect of aging on the apparent diffusion coefficient of normal-appearing white matter. Am J Roentgenol 175: 425–430. Estruch R, Nicolas JM, Salamero M, Aragon C, Sacanella E,

Schechter I, Zemon V, Saperstein A, Javitt DC. 2003. Primary

Fernandez-Sola J, Urbano-Marquez A. 1997. Atrophy of the

573

574

Adolf Pfefferbaum and Edith V. Sullivan

corpus callosum in chronic alcoholism. J Neurol Sci 146: 145–151. Filippi CG, Ulug AM, Ryan E, Ferrando SJ, van Gorp W. 2001. Diffusion tensor imaging of patients with HIV and normalappearing white matter on MR images of the brain. Am J Neuroradiol 22: 277–283. Foong J, Maier M, Clark C, Barker G, Miller D, Ron M. 2000. Neuropathological abnormalities of the corpus callosum in schizophrenia: a diffusion tensor imaging study. J Neurol Neurosurg Psychiatry 68: 242–244. Foong J, Symms MR, Barker GJ, Maier M, Miller DH, Ron MA. 2002. Investigating regional white matter in schizophrenia using diffusion tensor imaging. Neuroreport 13: 333–336. Ford JM, Mathalon DH, Whitfield S, Faustman WO, Roth WT. 2002. Reduced communication between frontal and temporal lobes during talking in schizophrenia. Biol Psychiatry 21: 485–492. Friston KJ, Holmes AP, Worsley J-P, Poline CD, Frith CD, Frackowiak RSJ. 1995. Statistical parametric maps in functional imaging: a general linear approach. Hum Brain Mapp 2: 189–210. Guttmann CRG, Jolesz FA, Kikinis R, Killiany RJ, Moss MB, Sandor T, Albert MS. 1998. White matter changes with normal aging. Neurology 50: 972–978. Hanyu H, Asano T, Sakurai H, Imon Y, Iwamoto T, Takasaki M, Shindo H, Abe K. 1999a. Diffusion-weighted and magnetization transfer imaging of the corpus callosum in Alzheimer’s disease. J Neurol Sci 167: 37–44. Hanyu H, Asano T, Sakurai H, Iwamoto T, Takasaki M, Shindo H, Abe K. 1999b. Magnetization transfer ratio in cerebral white matter lesions of Binswanger’s disease. J Neurol Sci 166: 85–90. Hanyu H, Sakurai H, Iwamoto T, Takasaki M, Shindo H, Abe K. 1998. Diffusion-weighted MR imaging of the hippocampus and temporal white matter in Alzheimer’s disease. J Neurol Sci 156: 195–200. Harper C. 1998. The neuropathology of alcohol-specific brain damage, or does alcohol damage the brain? Neuropathol Exp Neurol 57: 101–110. Harper CG, Kril JJ. 1988. Corpus callosal thickness in alcoholics. Br J Addict 83: 577–580. Harper C, Kril JJ. 1989. Patterns of neuronal loss in the cerebral cortex in chronic alcoholic patients. J Neurol Sci 92: 81–89. Harper CG, Kril JJ. 1990. Neuropathology of alcoholism. Alcohol Alcoholism 25: 207–216. Harper CG, Kril JJ, Holloway RL. 1985. Brain shrinkage in chronic alcoholics: a pathological study. Br Med J 290: 501–504. Harper CG, Kril JJ, Daly JM. 1987. Are we drinking our neurones away? Br Med J 294: 534–536. Harper CG, Smith NA, Kril JJ. 1990. The effects of alcohol on the female brain – a neuropathological study. Alcohol Alcoholism 25: 445–448.

Helenius J, Soinne L, Perkio J, Salonen O, Kangasmaki A, Kaste M, Carano RA, Aronen HJ, Tatlisumak T. 2002. Diffusionweighted MR imaging in normal human brains in various age groups. Am J Neuroradiol 23: 194–199. Hommer D, Momenan R, Rawlings R, Ragan P, Williams W, Rio D, Eckardt M. 1996. Decreased corpus callosum size among alcoholic women. Archiv Neurol 53: 359–363. Hommer DW, Momenan R, Kaiser E, Rawlings RR. 2001. Evidence for a gender-related effect of alcoholism on brain volumes. Am J Psychiatry 158: 198–204. Hoptman MJ, Volavka J, Johnson G, Weiss E, Bilder RM, Lim KO. 2002. Frontal white matter microstructure, aggression, and impulsivity in men with schizophrenia: a preliminary study. Biol Psychiatry 52: 9–14. Illowsky B, Juliano DM, Bigelow LB, Weinberger DR. 1988. Stability of CT scan findings in schizophrenia: results of an 8 year follow-up study. J Neurol Neurosurg Psychiatry 51: 209–213. Inagaki T, Saito K. 2000. A case of Marchiafava-Bignami disease demonstrated by MR diffusion-weighted image. No To Shinkei 52: 633–637. Jack C, Petersen R, Xu Y, O’Brien P, Smith G, Ivnik R, Tangalos E, Kokmen E. 1998. Rate of medial temporal-lobe atrophy in typical aging and Alzheimer’s disease. Neurology 51: 993–999. Jack Jr CR, Dickson DW, Parisi JE, Xu YC, Cha RH, O’Brien PC, Edland SD, Smith GE, Boeve BF, Tangalos EG, Kokmen E, Petersen RC. 2002. Antemortem MRI findings correlate with hippocampal neuropathology in typical aging and dementia. Neurology 58: 750–757. Jernigan TL, Archibald S, Hesselink JR, Atkinson JH, Velin RA, McCutchan JA, Chandler J, Grant I. 1993. Magnetic resonance imaging morphometric analysis of cerebral volume loss in human immunodeficiency virus infection. Archiv Neurol 50: 250–255. Jernigan TL, Archibald SL, Fennema-Notestine C, Gamst AC, Stout JC, Bonner J, Hesselink JR. 2001. Effects of age on tissues and regions of the cerebrum and cerebellum. Neurobiol Aging 22: 581–594. Jones D, Simmons A, Williams S, Horsfield M. 1999. Noninvasive assessment of axonal fiber connectivity in the human brain via diffusion tensor MRI. Magn Reson Med 42: 37–41. Kantarci K, Jack Jr CR, Xu YC, Campeau NG, O’Brien PC, Smith GE, Ivnik RJ, Boeve BF, Kokmen E, Tangalos EG, Petersen RC. 2001. Mild cognitive impairment and Alzheimer disease: regional diffusivity of water. Radiology 219: 101–107. Kashihara K, Irisawa M. 2002. Diffusion weighted magnetic resonance imaging in a case of acute Wernicke’s encephalopathy. J Neurol Neurosurg Psychiatry 73: 181. Kemper TL. 1994. Neuroanatomical and neuropathological changes during aging and dementia. In Clinical Neurology

Diffusion MR imaging in neuropsychiatry and aging

of Aging (Eds., Albert ML, Knoefel JE), Oxford University Press, New York, pp. 3–67. Killiany RJ, Gomez-Isla T, Moss M, Kikinis R, Sandor T, Jolesz F, Tanzi R, Jones K, Hyman BT, Albert MS. 2000. Use of structural magnetic resonance imaging to predict who will get Alzheimer’s disease. Ann Neurol 47: 430–439. Kramer-Ginsberg E, Greenwald BS, Krishnan KR, Christiansen B, Hu J, Ashtari M, Patel M, Pollack S. 1999. Neuropsychological functioning and MRI signal hyperintensities in geriatric depression. Am J Psychiatry 156: 438–444. Kril JJ, Halliday GM, Svoboda MD, Cartwright H. 1997. The cerebral cortex is damaged in chronic alcoholics. Neuroscience 79: 983–998. Kubicki M, Shenton ME, Salisbury DF, Hirayasu Y, Kasai K, Kikinis R, Jolesz FA, McCarley RW. 2002a. Voxel-based morphometric analysis of gray matter in first episode schizophrenia. Neuroimage 17: 1711–1719. Kubicki M, Westin CF, Maier SE, Frumin M, Nestor PG, Salisbury DF, Kikinis R, Jolesz FA, McCarley RW, Shenton ME. 2002b. Uncinate fasciculus findings in schizophrenia: a magnetic resonance diffusion tensor imaging study. Am J Psychiatry 159: 813–820. Kubicki M, Westin CF, Maier SE, Mamata H, Frumin M, ErsnerHershfield H, Kikinis R, Jolesz FA, McCarley R, Shenton ME. 2002c. Diffusion tensor imaging and its application to neuropsychiatric disorders. Harvard Rev Psychiatry 10: 324–336. Laakso MP, Soininen H, Partanen K, Helkala E-L, Hartikainen P, Vainio P, Hallikainen M, Hanninen T, Riekkinen PJ. 1995. Volumes of hippocampus, amygdala and frontal lobes in the MRI-based diagnosis of early Alzheimer’s disease: correlation with memory functions. J Neural Transm 9: 73–86. Lim KO, Adalsteinson A, Spielman D, Rosenbloom MJ, Sullivan EV, Pfefferbaum A. 1998. Proton magnetic resonance spectroscopic imaging of cortical gray and white matter in schizophrenia. Archiv Gen Psychiatry 55: 346–352. Lim KO, Hedehus M, Moseley M, De Crespigny A, Sullivan EV, Pfefferbaum A. 1999. Compromised white matter tract integrity in schizophrenia inferred from diffusion tensor imaging. Archiv Gen Psychiatry 56: 367–374. Lim KO, Helpern JA. 2002. Neuropsychiatric applications of DTI – a review. NMR in Biomed 15: 587–593. Manji H, Connolly S, McAllister R, Valentine AR, Kendall BE, Fell M, Durrance P, Thompson AJ, Newman S, Weller IVD, Harrison MJG. 1994. Serial MRI of the brain in asymptomatic patients infected with HIV: results from the UCMSM/ Medical Research Council neurology cohort. J Neurol Neurosurg Psychiatry 57: 144–149. Marsh L, Lauriello J, Sullivan EV, Pfefferbaum A. 1996. Neuroimaging in neuropsychiatric disorders. In Neuroimaging II: Clinical Applications (Ed., Bigler E), Plenum Press, New York, pp. 73–125.

Masliah E, DeTeresa RM, Mallory ME, Hansen LA. 2000. Changes in pathological findings at autopsy in AIDS cases for the last 15 years. AIDS 14: 69–74. Masutani Y, Aoki S, Abe O, Hayashi N, Otomo K. 2003. MR diffusion tensor imaging: recent advance and new techniques for diffusion tensor visualization. Eur J Radiol 46: 53–66. McGuire PK, Frith CD. 1996. Disordered functional connectivity in schizophrenia. Psychol Med 26: 663–667. Meier-Ruge W, Ulrich J, Bruhlmann M, Meier E. 1992. Agerelated white matter atrophy in the human brain. Ann NY Acad Sci 673: 260–269. Miller AKH, Alston RL, Corsellis JAN. 1980. Variations with age in the volumes of grey and white matter in the cerebral hemispheres of man: measurements with an image analyzer. Neuropathol Appl Neurobiol 6: 119–132. Miller RF, Harrison MJ, Hall-Craggs MA, Scaravilli F. 1998. Central pontine myelinolysis in AIDS. Acta Neuropathol (Berlin) 96: 537–540. Molko N, Cohen L, Mangin JF, Chochon F, Lehericy S, Le Bihan D, Dehaene S. 2002. Visualizing the neural bases of a disconnection syndrome with diffusion tensor imaging. J Cogn Neurosci 14: 629–36. Mori S, Kaufmann WE, Davatzikos C, Stieltjes B, Amodei L, Fredericksen K, Pearlson GD, Melhem ER, Solaiyappan M, Raymond GV, Moser HW, van Zijl PC. 2002. Imaging cortical association tracts in the human brain using diffusiontensor-based axonal tracking. Magn Reson Med 47: 215–223. Moseley M. 2002. Diffusion tensor imaging and aging – a review. NMR Biomed 15: 553–560. Moseley M, Bammer R, Illes J. 2002. Diffusion-tensor imaging of cognitive performance. Brain Cognition 50: 396–413. Moseley ME, Mintorovitch J, Cohen Y, Asgari HS, Derugin N, Norman D, Kucharczyk J. 1990. Early detection of ischemic injury: comparison of spectroscopy, diffusion-, T2-, and magnetic susceptibility-weighted MRI in cats. Acta Neurochir Supplementa 51: 207–209. Naganawa S, Sato K, Katagiri T, Mimura T, Ishigaki T. 2003. Regional ADC values of the normal brain: differences due to age, gender, and laterality. Eur Radiol 13: 6–11. Niclot P, Guichard JP, Djomby R, Sellier P, Bousser MG, Chabriat H. 2002. Transient decrease of water diffusion in Wernicke’s encephalopathy. Neuroradiology 44: 305–307. Nierenberg J, Hoptman MJ, Choi SJ, Butler PD, Ardekani BA, Revheim N, Javitt DC, Lim KO. 2003. Abnormal white matter integrity in schizophrenia and schizoaffective disorder revealed by diffusion tensor imaging (abstract). Schizophrenia Res 60: 203S. Nomura Y, Sakuma H, Takeda K, Tagami T, Okuda Y, Nakagawa T. 1994. Diffusional anisotropy of the human brain assessed with diffusion-weighted MR: relation with normal brain development and aging. Am J Neuroradiol 15: 231–238.

575

576

Adolf Pfefferbaum and Edith V. Sullivan

Nusbaum AO, Tang CY, Buchsbaum MS, Wei TC, Atlas SW. 2001. Regional and global changes in cerebral diffusion with normal aging. Am J Neuroradiol 22: 136–142. O’Brien J, Perry R, Barber R, Gholkar A, Thomas A. 2000. The association between white matter lesions on magnetic resonance imaging and noncognitive symptoms. Ann NY Acad Sci 903: 482–489. O’Sullivan M, Jones D, Summers P, Morris R, Williams S, Markus H. 2001. Evidence for cortical “disconnection” as a mechanism of age-related cognitive decline. Neurology 57: 632–638. Pearlson G, Marsh L. 1999. Structural brain imaging in schizophrenia: a selective review. Biol Psychiatry 46: 627–649. Pettigrew DB, Crutcher KA. 2001. Myelin contributes to the parallel orientation of axonal growth on white matter in vitro. BioMed Cent Neurosci 2: 9–20. Pfefferbaum A, Sullivan EV. 2002. Microstructural but not macrostructural disruption of white matter in women with chronic alcoholism. Neuroimage 15: 708–718. Pfefferbaum A, Sullivan EV. 2003. Increased brain white matter diffusivity in normal adult aging: relationship to anisotropy. Magn Reson Med 49: 953–961. Pfefferbaum A, Adalsteinsson E, Sullivan EV. 2003. Replicability of diffusion tensor imaging measurements of FA and trace in brain. J Magn Reson Imaging 18: 427–433. Pfefferbaum A, Lim KO, Desmond J, Sullivan EV. 1996. Thinning of the corpus callosum in older alcoholic men: a magnetic resonance imaging study. Alcohol Clin Exp Res 20: 752–757. Pfefferbaum A, Lim KO, Zipursky RB, Mathalon DH, Lane B, Ha CN, Rosenbloom MJ, Sullivan EV. 1992. Brain gray and white matter volume loss accelerates with aging in chronic alcoholics: a quantitative MRI study. Alcohol Clin Exp Res 16: 1078–1089. Pfefferbaum A, Mathalon DH, Sullivan EV, Rawles JM, Zipursky RB, Lim KO. 1994. A quantitative magnetic resonance imaging study of changes in brain morphology from infancy to late adulthood. Archiv Neurol 51: 874–887. Pfefferbaum A, Rosenbloom MJ, Deshmukh A, Sullivan EV. 2001a. Sex differences in the effects of alcohol on brain structure. Am J Psychiatry 158: 188–197. Pfefferbaum A, Rosenbloom M, Sullivan E. 2002. Alcoholism and AIDS: MR imaging approaches for detecting interaction neuropathology. Alcohol Clin Exp Res 26: 1031–1046. Pfefferbaum A, Sullivan EV, Carmelli D. 2001b. Genetic regulation of regional microstructure of the corpus callosum in late life. Neuroreport 12: 1677–1681. Pfefferbaum A, Sullivan EV, Hedehus M, Adalsteinsson E, Lim KO, Moseley M. 2000a. In vivo detection and functional correlates of white matter microstructural disruption in chronic alcoholism. Alcohol Clin Exp Res 24: 1214–1221.

Pfefferbaum A, Sullivan EV, Hedehus M, Lim KO, Adalsteinsson E, Moseley M. 2000b. Age-related decline in brain white matter anisotropy measured with spatially corrected echo-planar diffusion tensor imaging. Magn Reson Med 44: 259–268. Pfefferbaum A, Sullivan EV, Hedehus M, Moseley M, Lim KO. 1999. Brain gray and white matter transverse relaxation time in schizophrenia. Schizophrenia Res Neuroimaging Sec 91: 93–100. Pfefferbaum A, Sullivan EV, Mathalon DH, Lim KO. 1997. Frontal lobe volume loss observed with magnetic resonance imaging in older chronic alcoholics. Alcohol Clin Exp Res 21: 521–529. Pfefferbaum A, Zipursky RB, Lim KO, Zatz LM, Stahl SM, Jernigan TL. 1988. Computed tomographic evidence for generalized sulcal and ventricular enlargement in schizophrenia. Archiv Gen Psychiatry 45: 633–640. Pierpaoli C, Barnett A, Pajevic S, Chen R, Penix L, Virta A, Basser P. 2001. Water diffusion changes in Wallerian degeneration and their dependence on white matter architecture. Neuroimage 13: 1174–1185. Pierpaoli C, Basser PJ. 1996. Towards a quantitative assessment of diffusion anisotropy. Magn Reson Med 36: 893–906. Pomara N, Crandall DT, Choi SJ, Johnson G, Lim KO. 2001. White matter abnormalities in HIV-1 infection: a diffusion tensor imaging study. Psychiatry Res 106: 15–24. Post MJ, Berger JR, Duncan R, Quencer RM, Pall L, Winfield D. 1993. Asymptomatic and neurologically symptomatic HIVseropositive subjects: results of long-term MR imaging and clinical follow-up. Radiology 188: 727–733. Post MJ, Yiannoutsos C, Simpson D, Booss J, Clifford DB, Cohen B, McArthur JC, Hall CD. 1999. Progressive multifocal leukoencephalopathy in AIDS: are there any MR findings useful to patient management and predictive of patient survival? AIDS Clinical Trials Group, 243 Team. Am J Neuroradiol 20: 1896–906. Raz N. 1999. Aging of the brain and its impact on cognitive performance: integration of structural and functional findings. In Handbook of Aging and Cognition II (Eds, Craik FIM, Salthouse TA) Mahwah, Erlbaum, NJ, pp. 1–90. Raz N, Gunning FM, Head D, Dupuis JH, McQuain J, Briggs SD, Loken WJ, Thornton AE, Acker JD. 1997. Selective aging of the human cerebral cortex observed in vivo: differential vulnerability of the prefrontal gray matter. Cereb Cortex 7: 268–282. Rose SE, Chen F, Chalk JB, Zelaya FO, Strugnell WE, Benson M, Semple J, Doddrell DM. 2000. Loss of connectivity in Alzheimer’s disease: an evaluation of white matter tract integrity with colour coded MR diffusion tensor imaging (In Process Citation). J Neurol Neurosurg Psychiatry 69: 528–530.

Diffusion MR imaging in neuropsychiatry and aging

Rourke SB, Grant I. 1999. The interactive effects of age and length of abstinence on the recovery of neuropsychological functioning in chronic male alcoholics: a 2-year follow-up study. J Int Neuropsychol Society 5: 234–246. Ruiz A, Post JD, Ganz WI, Georgiou M. 1997. Nuclear medicine applications to the neuroimaging of AIDS. A neuroradiologist’s perspective. Neuroimaging Clin North America 7: 499–511. Sacktor NC, Bacellar H, Hoover DR, Nance-Sproson TE, Selnes OA, Miller EN, Dal Pan GJ, Kleeberger C, Brown A, Saah A, McArthur JC. 1996. Psychomotor slowing in HIV infection: a predictor of dementia, AIDS and death. J Neurovirol 2. Salat DH, Kaye JA, Janowsky JS. 1999. Prefrontal gray and hite matter volumes in healthy aging and Alzheimer disease. Archiv Neurol 56: 338–344. Sandson TA, Felician O, Edelman RR, Warach S. 1999. Diffusion-weighted magnetic resonance imaging in Alzheimer’s disease. Dement Geriatr Cogn Disord 10: 166–171. Schaefer PW, Gonzalez RG, Hunter G, Wang B, Koroshetz WJ, Schwamm LH. 2001. Diagnostic value of apparent diffusion coefficient hyperintensity in selected patients with acute neurologic deficits. J Neuroimaging 11: 369–380. Schlaepfer TE, Harris GJ, Tien AY, Peng LW, Lee S, Federman EB, Chase GA, Barta PE, Pearlson GD. 1994. Decreased regional cortical gray matter volume in schizophrenia. Am J Psychiatry 151: 842–848. Schwartz RB. 2001. Apparent diffusion coefficient mapping in patients with Alzheimer disease or mild cognitive impairment and in normally aging control subjects: present and future. Radiology 219: 8–9. Sclar G, Kennedy CA, Hill JM, McCormack MK. 2000. Cerebellar degeneration associated with HIV infection [letter]. Neurology 54: 1012–1013. Selemon LD, Kleinman JE, Herman MM, Goldman-Rakic PS. 2002. Smaller frontal gray matter volume in postmortem schizophrenic brains. Am J Psychiatry 159: 1983–1991. Sheedy D, Lara A, Garrick T, Harper C. 1999. Size of mamillary bodies in health and disease: useful measurements in neuroradiological diagnosis of Wernicke’s encephalopathy. Alcohol Clin Exp Res 23: 1624–1628. Shenton M, Dickey C, Frumin M, McCarley R. 2001. A review of MRI findings in schizophrenia. Schizophrenia Res 49: 1–52. Shimony JS, McKinstry RC, Akbudak E, Aronovitz JA, Snyder AZ, Lori NF, Cull TS, Conturo TE. 1999. Quantitative diffusiontensor anisotropy brain MR imaging: normative human data and anatomic analysis. Radiology 212: 770–784. Song SK, Sun SW, Ramsbottom MJ, Chang C, Russell J, Cross AH. 2002. Dysmyelination revealed through MRI as increased radial (but unchanged axial) diffusion of water. Neuroimage 17: 1429–1436.

Spielman D, Butts K, de Crespigny A, Moseley M. 1996. Diffusion-weighted imaging of clinical stroke. Int J Neuroradiol 1: 44–55. Stebbins G, Carrillo MD, Medina D, de Toledo-Morrell L, Klingberg T, Poldrack RA, Moseley M, Karni O, Wilson RS, Bennett DA, Gabrieli JDE. 2001. Frontal white matter integrity in aging and its relation to reasoning performance: a diffusion tensor imaging study (abstract 456.3). Soc Neurosci Abst 27: 1204. Steel R, Bastin M, McConnell S, Marshall I, CunninghamOwens D, Lawrie S, Johnstone E, Best J. 2001. Diffusion tensor imaging (DTI) and proton magnetic resonance spectroscopy (1H MRS) in schizophrenic subjects and normal controls. Psychiatry Res 106: 161–170. Stout J, Ellis R, Jernigan T, Archibald S, Abramson I, Wolfson T, McCutchan J, Wallace M, Atkinson J, Grant I. 1998. Progressive cerebral volume loss in human immunodeficiency virus infection: a longitudinal volumetric magnetic resonance imaging study. Archiv Neurol 55: 161–168. Sullivan EV, Pfefferbaum A. 2003. Diffusion tensor imaging in normal aging and neuropsychiatric disorders. Eur J Radiol 45: 244–255. Sullivan EV, Adalsteinsson E, Hedehus M, Ju C, Moseley M, Lim KO, Pfefferbaum A. 2001a. Equivalent disruption of regional white matter microstructure in aging healthy men and women. Neuroreport 12: 99–104. Sullivan EV, Lim KO, Mathalon DH, Marsh L, Beal DM, Harris D, Hoff A, Faustman WO, Pfefferbaum A. 1998. A profile of cortical gray matter volume deficits characteristic of schizophrenia. Cereb Cortex 8: 117–124. Sullivan EV, Pfefferbaum A, Adalsteinsson E, Swan GE, Carmelli D. 2002. Differential rates of regional change in callosal and ventricular size: a 4-year longitudinal MRI study of elderly men. Cereb Cortex 12: 438–445. Sullivan EV, Rosenbloom MJ, Desmond JE, Pfefferbaum A. 2001b. Sex differences in corpus callosum size: relationship to age and intracranial size. Neurobiol Aging 22: 603–611. Sullivan EV, Rosenbloom MJ, Lim KO, Pfefferbaum A. 2000. Longitudinal changes in cognition, gait, and balance in abstinent and relapsed alcoholic men: relationships to changes in brain structure. Neuropsychology 14: 178–188. Sullivan EV, Rosenbloom MJ, Serventi KL, Pfefferbaum A. 2004. Effects of age and sex on volumes of the thalamus, pons, and cortex. Neurobiol Aging 25: 185–192. Symonds LL, Archibald SL, Grant I, Zisook S, Jernigan TL. 1999. Does an increase in sulcal or ventricular fluid predict where brain tissue is lost? J Neuroimaging 9: 201–209. Tagliati M, Simpson D, Morgello S, Clifford D, Schwartz RL, Berger JR. 1998. Cerebellar degeneration associated with human immunodeficiency virus infection. Neurology 50: 244–251.

577

578

Adolf Pfefferbaum and Edith V. Sullivan

Takahashi S, Yonezawa H, Takahashi J, Kudo M, Inoue T, Tohgi H. 2002. Selective reduction of diffusion anisotropy in white matter of Alzheimer disease brains measured by 3.0 T magnetic resonance imaging. Neurosci Lett 332: 45–48. Tang CY, Lu D, Wei TC, Spiegel J, Atlas SW, Buchsbaum MS. 1997. Image processing techniques for the eigenvectors of the diffusion tensor (abstract). Int Soc Magn Reson in Med, 5th Meeting 2054. Tarnowska-Dziduszko E, Bertrand E, Szpak G. 1995. Morphological changes in the corpus callosum in chronic alcoholism. Folia Neuropathol 33: 25–29. Taylor WD, Payne ME, Krishnan KR, Wagner HR, Provenzale JM, Steffens DC, MacFall JR. 2001. Evidence of white matter tract disruption in MRI hyperintensities. Biol Psychiatry 50: 179–183. Teipel SJ, Bayer W, Alexander GE, Zebuhr Y, Teichberg D, Kulic L, Schapiro MB, Moller HJ, Rapoport SI, Hampel H. 2002. Progression of corpus callosum atrophy in Alzheimer disease. Archiv Neurol 59: 243–248. Trillo-Pazos G, Everall IP. 1997. From human immunodeficiency virus (HIV) infection of the brain to dementia. Genitourin Med 73: 343–347. Ulug AM, Filippi CG, Ruyan E, Ferrando SJ, Van Gorp W. 2000. Utility of DWI, tensor imaging, and MR spectroscopy in HIV patients with normal brain MR scans (abstract). Proc Int Soc Magn Reson in Med 8: 1200. UNAIDS. 2002. AIDS Epidemic Update. UNAIDS joint United Nations Programme on HIV/AIDS.

Victor M, Adams RD, Collins GH. 1989. The WernickeKorsakoff Syndrome and Related Neurologic Disorders Due to Alcoholism and Malnutrition (2nd edn.). F.A. Davis Co., Philadelphia. Virta A, Barnett A, Pierpaoli C. 1999. Visualizing and characterizing white matter fiber structure and architecture in the human pyramidal tract using diffusion tensor MR. Magn Reson Imaging 17: 1121–1133. Vita A, Sacchetti E, Valvassori G, Cazzullo CL. 1988. Brain morphology in schizophrenia: a 2- to 5-year CT scan follow-up study. Acta Psychiat Scand 78: 618–621. Waxman SG, Kocsis JD, Stys PK. 1995. The Axon: Structure, Function and Pathophysiology. Oxford University Press, New York. Wolkin A, Choi SJ, Szilagyi S, Sanfilipo M, Rotrosen JP, Lim KO. 2003. Inferior frontal white matter anisotropy and negative symptoms of schizophrenia: a diffusion tensor imaging study. Am J Psychiatry 160: 572–574. Wolkin A, Rusinek H, Vaid G, Arena L, Lafargue T, Sanfilipo M, Loneragan C, Lautin A, Rotrosen J. 1998. Structural magnetic resonance image averaging in schizophrenia. Am J Psychiatry 155: 1064–1073. Zipursky RB, Lim KO, Sullivan EV, Brown BW, Pfefferbaum A. 1992. Widespread cerebral gray matter volume deficits in schizophrenia. Archiv Gen Psychiatry 49: 195–205.

34

MR spectroscopy in aging and dementia Kejal Kantarci and Clifford R. Jack, Jr. Department of Radiology, Mayo Clinic, Rochester, Minnesota, USA

Key points • Choline (Cho) and Creatine (Cr) tend to increase with aging, and cortical gray matter N-acetylaspartate (NAA) levels are stable throughout aging. • NAA is consistently found to be lower and the metabolite myo-inositol (mI) higher in the 1 H MR spectra of patients with Alzheimer’s disease (AD) than cognitively normal elderly. The changes in Cho are conflicting. • NAA/Cr and mI/Cr ratios correlate with neuropsychological measures of cognitive function in patients with AD, and this correlation is more significant with NAA/mI ratios. • White matter (WM) NAA/Cr is lower in patients with vascular dementia (VaD) than in patients with AD, reflecting the WM ischemic damage in VaD. • Patients with Lewy bodies have normal NAA/Cr levels, whereas patients with AD, VaD and frontotemporal dementia have lower NAA/ Cr levels than normal in the posterior cingulate gyrus. • mI/Cr is increased in patients with amnesic mild cognitive impairment very mild AD, and pre-dementia phase of Down’s syndrome, while NAA/Cr levels are normal compared to age-matched controls.

Neuroimaging techniques may have an important role in the clinical evaluation of dementia for early diagnosis, differential diagnosis, and monitoring of

disease activity. The goal of this chapter is to review over a decade of MR spectroscopy (MRS) literature in aging and dementia in order to demonstrate the potential clinical applications of the technique and its limitations in this field. This chapter will primarily focus on 1H MRS because a majority of the MRS literature in aging and dementia utilizes this technique, and it can readily be performed as part of a routine structural MRI study. The literature on 31P MRS and other nuclei will be discussed in the “future perspectives” section owing to their limited applications at this time.

1

H MRS in aging

Age related changes in 1H MRS measurements of the metabolites N-acetyl aspartate (NAA), choline (Cho), creatine (Cr) and myo-inositol (mI) were investigated by several groups, and the results have been conflicting (Table 34.1). There are reports showing that metabolite measurements are stable throughout aging (Saunders et al., 1999), one study showed a decrease in NAA, Cho and Cr in the gray matter (GM) (Charles et al., 1994), and others an increase in Cho and Cr in the GM and white matter (WM) with aging (Chang et al., 1996; Pfefferbaum et al., 1999a; Leary et al., 2000; Schuff et al., 2001). As a whole, most studies agree that Cho and Cr increase with aging, and almost all studies agree that cortical GM NAA levels are stable throughout aging. NAA in the brain is primarily located in neuron bodies, axons and dendrites, thus NAA is a sensitive marker for neuronal density. Based on the correlation 579

580

Kejal Kantarci and Clifford R. Jack, Jr.

Table 34.1. 1H MRS findings in normal aging Study

Charles et al. 1994 (2) Chang et al. 1996 (3) Pfefferbaum et al. 1999a (4) Saunders et al. 1999 (1) Leary et al. 2000 (6) Schuff et al. 2001 (5)

Age range

21–75 19–78 young vs. old 24–89 22–62 56–89

GM

WM

NAA

Cr

Cho

mI

NAA

Cr

Cho

mI

↓* Stable Stable Stable NS ↑

↓* ↑ ↑ Stable NS ↑

↓* ↑ ↑ Stable NS Stable

NS ↑ NS Stable NS NS

Stable Stable Stable Stable Stable Stable

Stable Stable Stable Stable ↑ Stable

Stable Stable ↑ Stable ↑ Stable

NS Stable NS Stable Stable NS

*Corpus striatum; NS: not studied; ↑: increase with aging; ↓: decrease with aging.

between mitochondrial ATP production and NAA levels, production of NAA in the neuron is thought to take place in the mitochondria (Bates et al., 1995). Depressed NAA levels may, however, also normalize after recovery from head trauma (Brooks et al., 2000) or cessation of seizures after epilepsy surgery (Hugg et al., 1996). Normalization of NAA levels after therapy in neurological disorders implies that NAA is also a marker for neuronal integrity, and possibly for neuronal mitochondrial function. Neuron counts using stereological methods indicate that the number of neurons remain stable through out the human life span (Long et al., 1999). Therefore a change in NAA levels would not be expected in a normally aging elderly person unless there is a neuronal functional disruption. The potential of NAA as a marker for neuronal function in aging was investigated through associations between neuropsychological measures of cognitive function and regional NAA levels in the brains of the elderly. A recent study showed that, while frontal WM NAA level correlated with the measures of executive-attentional ability, it did not correlate with the measures of memory or verbal fluency in cognitively normal elderly (Valenzuela et al., 2000). Furthermore, this specific functional correlation between executive-attentional ability and NAA levels was observed in the frontal WM but not in the occipitoparietal region, in accordance with the functional anatomy. Understanding the regional variations in NAA levels in association with performance in specific cognitive domains may help strategic selection of regions in the brain to study with MRS in people with cognitive dysfunction or dementia.

1

H MRS in dementia

Dementia is a clinical diagnosis defined as an acquired syndrome of decline in memory and other cognitive functions that affect the daily life in an alert patient (American Psychiatric Association, 1987). Dementia caused by neurological disorders that can accurately be diagnosed with clinical evaluation, laboratory tests and anatomic imaging such as tumors, subdural hematoma, or vitamin B12 deficiency will not be mentioned in this section. This section will rather focus on 1H MRS in common dementia syndromes, differential diagnosis of which is challenging on clinical grounds. The most prevalent pathological causes of dementia in autopsy series are Alzheimer’s disease (AD), dementia with Lewy bodies (DLB) and vascular dementia (VaD) (Holmes et al., 1999a). Different pathological substrates of dementia may also co-exist creating a heterogenous group of clinical dementia syndromes referred to as mixed dementia, the most common of which, is the combination of cerebrovascular pathology and AD (Massoud et al., 1999). Operational criteria have been generated for these pathological subtypes of dementia in order to define the associated clinical syndromes. Autopsy series however indicate that the accuracy of antemortem differential diagnosis of these dementing disorders by means of the current clinical criteria is not optimal at this time. This generates the incentive for identifying specific neuroimaging markers for various dementing pathologies in order to aid in clinical diagnosis.

MR spectroscopy in aging and dementia

1

H MRS in AD

Control

NAA

1

Most of the H MRS literature in dementia is focused on differentiating patients with AD from cognitively normal elderly. The metabolite NAA is consistently found to be lower and the metabolite mI higher in the 1 H MR spectra of patients with AD than cognitively normal elderly (Klunk et al., 1992; Miller et al., 1993; Tedeschi et al., 1996; Mohanakrishnan et al., 1997; Schuff et al., 1998; Rose et al., 1999; Jessen et al., 2000; Kantarci et al., 2000; Huang et al., 2001). There are conflicting reports on Cho levels. Some studies identified higher Cho levels (Meyerhoff et al., 1994; Pfefferbaum et al., 1999b; Kantarci et al., 2000) in people with AD than normal, and some did not find any difference from normal (Moats et al., 1994; Parnetti et al., 1997; Schuff et al., 1997; Jessen et al., 2000). As absolute quantification of these metabolite peaks requires long imaging times, they are frequently studied with respect to the metabolite peak Cr. While one study have found elevated Cr levels in AD (Huang et al., 2001), many have shown this metabolite peak to be fairly stable in AD compared to age matched controls (Moats et al., 1994; Christiansen et al., 1995; Shonk et al., 1995a; Ernst et al., 1997; Mohanakrishnan et al., 1997; Parnetti et al., 1997; Schuff et al., 1997, 1998, 2002) making Cr a suitable reference peak. NAA is reduced both in the cortical GM and WM in patients with AD. The regional decrease in cortical NAA level is in agreement with the regional neuropathological involvement in AD. The pathology of AD and associated neuronal damage involves the limbic cortical regions earlier and more profoundly than the primary sensory-motor and visual cortex (VC), which is involved with the pathology only at the end stages of the disease (Braak and Braak, 1991). In keeping with that, NAA/Cr ratios of people with mild to moderate AD are lower than normal in a paralimbic cortical region, the posterior cingulate gyri. NAA/Cr measurements from the medial occipital lobe covering the VC however are not different from normal in these patients who are at the mild to moderate stages of the neurodegenerative disease (Kantarci et al., 2000) (Figure 34.1). As NAA is located in neuron bodies, axons and dendrides, reduction of NAA levels in patients with AD would either be the

Cr

AD NAA Cr

Control

NAA

Cr

NAA

AD

Cr

Fig. 34.1 Examples of proton spectra obtained from the posterior cingulate voxel (a) and medial occipital voxel (b) with an echo time (TE) of 135 ms in a control subject (top), and a patient with probable AD (bottom). NAA/Cr from the posterior cingulate voxel is lower in the patient with AD compared to the control subject. NAA/Cr ratio from the medial occipital voxel in the patient with probable AD is similar to the control subject. (Kantarci et al., 2000).

outcome of a loss of the neuronal components, neuronal functional disruption or both. Most of the metabolite mI in the brain is present in glial cells (Glanville et al., 1989) and increased mI levels correlate with glial proliferation in inflammatory central nervous system (CNS) demyelination (Bitsch et al., 1999). It is possible that elevation of mI levels is

581

Kejal Kantarci and Clifford R. Jack, Jr.

associated with glial proliferation in AD (Ross et al., 1998). The metabolite mI has a shorter transverse relaxation time than the other prominent metabolites in the human brain 1H MR spectra such as NAA, Cr and Cho. For this reason, quantification of the mI peak necessitates use of short echo times (TE) approximately within the 30–35 ms range. There are technical challenges to short TE 1H MR spectroscopic imaging (1H MRSI) such as water suppression, signal contribution from subcutaneous fat, and gradient eddy current distortions, all of which decrease the test-retest reproducibility of 1H MRS (Soher et al., 2000). Therefore single voxel 1H MRS is the preferred method for quantification of the metabolite mI at this time. Another challenge with short TE 1H MRS in AD is obtaining reproducible measurements from the anteromedial temporal lobe, a region which is known to be affected with the pathology of AD earlier and more severely than any other region in the brain. The anteromedial temporal lobe is in proximity to the tissue–air interface near the petrous bone. The air–tissue susceptibility difference, make establishing a homogenous magnetic field and water suppression within the 1H MRS voxel difficult. Short TE single voxel 1H MRS acquisitions from this critical region would benefit from improvements in shimming algorithms. One drawback of single voxel 1H MRS is that the voxel size generally used (8 cm3) to obtain the sufficient signal to noise ratio (SNR) is larger than the volume of the anteromedial temporal lobe limbic structures, causing partial voluming of the surrounding tissue and decreasing the anatomic specificity of the measurements. 1H MRS at higher field strengths such as 3 T may allow for comparable SNR with smaller voxels, which will be covered later in this chapter. 1 H MRS measurements of NAA/Cr and mI/Cr ratios correlate with neuropsychological measures of cognitive function in patients with AD, and this correlation is more significant with NAA/mI ratios (Kantarci et al., 2002a) (Figure 34.2). There is a region-specific association between neuropsychological performance and 1H MRS metabolite changes depending on the cognitive domain being studied. For example, metabolic changes in patients with AD correlate with loss of verbal memory in the left medial temporal lobe, and correlate with language impairment and

152.00 142.00 132.00 DRSTOT

582

122.00 112.00 102.00 92.00 82.00 1

1.5

2

2.5 NAA/ml

3

3.5

Fig. 34.2 Plot of dementia rating scale total scores (DRSTOT) and NAA/mI ratio in 67 cognitively normal elderly represented by (●) and solid regression line, 18 patients with mild cognitive impairment (MCI) represented by (♦), and 33 patients with AD represented by (●). Dotted regression line represents patients with cognitive impairment (both MCI and AD). There is a significant association between NAA/mI and DRSTOT in patients with cognitive impairment (P  0.01). No association is present in cognitively normal elderly (P  0.05). (Kantarci et al., 2002).

visuoconstructional abilities in the left parietotemporal cortex (Chantal et al., 2002). The association between 1H MRS measurements and neuropsychological performance suggests that 1H MRS may be a sensitive marker for disease progression in AD. As mI levels are higher and NAA levels are lower in patients with AD than normal elderly, the ratio of NAA/mI has been the most diagnostically accurate 1 H MRS measurement to differentiate patients with AD from normal elderly. NAA/mI ratios distinguished clinically diagnosed patients with AD from cognitively normal elderly with a sensitivity of 83% and specificity of 98% in one cohort (Shonk et al., 1995a), and sensitivity of 82% and specificity of 80% in another (Kantarci et al., 2000). As definitive diagnosis of AD requires histopathological confirmation, the diagnostic value of NAA/mI would ideally be assessed with pathological diagnosis. 1H MRS has not been evaluated for the diagnosis of AD in pathologically confirmed cohorts yet. According to the most recent guidelines by the American Academy of Neurology, quantitative MR techniques are not recommended

4

MR spectroscopy in aging and dementia

Table 34.2. 1H MRS metabolite measurements (mean  SD) in common dementias

N NAA/Cr Cho/Cr mI/Cr

Normal elderly

AD

VaD

DLB

FTD

130 1.54  0.12 0.66  0.08 0.65  0.08

95 1.41  0.12* 0.69  0.08* 0.73  0.12*

9 1.45  0.14* 0.69  0.10 0.64  0.10

17 1.52  0.12 0.73  0.09* 0.69  0.11

32 1.42  0.16* 0.71  0.08* 0.78  0.09*

*Metabolite ratios are statistically significantly different from normal (t-tests P  0.05).

for routine use of dementia evaluation at this time because superiority to clinical criteria has not been demonstrated (Knopman et al., 2001).

(CADASIL), examples of imaging and MRS findings of which are shown in Chapter 25. 1

H MRS in Dementia with Lewy Bodies (DLB)

1

H MRS in Vascular Dementia (VaD)

Cerebrovascular pathology is another common pathology observed in patients with dementia in autopsy series. In most cases however, vascular pathology co-exists with the pathology of AD, and pure vascular pathology is relatively uncommon, accounting for about 10% of the dementia cases in autopsy series (Holmes et al., 1999b). Vascular lesions are more common in patients with dementia than normal elderly. In a patient with the clinical diagnosis of AD and cerebrovascular disease, the challenge is to identify how much if any of the two pathologies are contributing to dementia, so that appropriate therapies can be planned. 1H MRS studies indicate that NAA and NAA/Cr levels are reduced in patients with VaD. WM NAA/Cr is lower in patients with VaD than in patients with AD, reflecting the WM ischemic damage in VaD with respect to the cortical degenerative pathology in AD (Kattapong et al., 1996; MacKay et al., 1996). Cortical mI/Cr levels on the other hand are normal in patients with VaD (Waldman et al., 2002; Kantarci et al., 2003a) (Table 34.2). As mI/Cr is elevated in patients with AD, mI/Cr may help to identify concommitant AD in a demented patient with cerebrovascular disease. Studies that include histopathological confirmation are necessary in order to clarify the role of mI/Cr in differential diagnosis of VaD, mixed dementia (VaD and AD), and AD. Dementia is also rarely associated with vascular disease in the context of specific familial conditions such as cerebral autosomal dominant arteriopathy with subcortical infarcts and leukoencephalopathy

The presence of Lewy bodies in substantia nigra is the pathological feature of Parkinson’s disease. Although cortical Lewy bodies can occasionally be detected in Parkinson’s disease, cortical Lewy bodies presenting with dementia is recognized as a distinct neurodegenerative disease with established clinical criteria (McKeith et al., 1996). Lewy bodies are commonly encountered in people with dementia, although usually accompanied by AD pathology as well. Lewy body pathology by itself is less common than the mixed (AD and Lewy body) type (Gomez-Isla et al., 1999). In our 1H MRS series (Kantarci et al., 2003a), patients clinically identified as having DLB have normal NAA/Cr levels, whereas patients with AD, VaD and frontotemporal dementia (FTD) have lower NAA/Cr levels than normal, in the posterior cingulate gyri (Table 34.2). Patients with DLB have preserved neuronal numbers at autopsy (Gomez-Isla et al., 1999), and normal limbic cortical volumes on voxelbased morphometry, which distinguishes them from patients with AD (Burton et al., 2002). Likewise, normal NAA/Cr levels suggest integrity of neurons in the posterior cingulate gyri, and may be useful in distinguishing patients with DLB from other dementia syndromes. It is however possible that NAA/Cr is decreased in other brain regions of people with DLB, that have not yet been studied with 1H MRS. 1

H MRS in Fronto Temporal Degeneration (FTD)

FTD involves selective and progressive degeneration of the frontal and temporal cortex. The associated

583

584

Kejal Kantarci and Clifford R. Jack, Jr.

clinical syndrome matches the anatomic distribution of pathological degeneration with dysfunction in the frontal lobe cognitive domains such as executive function (Neary et al., 1998). 1H MRS metabolite changes in FTD are similar to the changes encountered in AD; lower NAA/Cr and higher mI/Cr than normal (Table 34.2) (Kantarci et al., 2003a). One study identified lower NAA/Cr and higher mI/Cr levels in the frontal cortex of patients with FTD than patients with early AD, indicating that regional 1H MRS measurements may help differentiate neurodegenerative disorders that display regionally specific involvement (Ernst et al., 1997). It should be noted however, that while regional differences may be prominent during early stages of the pathological process in neurodegenerative diseases, these differences would be lost as neurodegenerative pathology involves the majority of the cerebral cortex in later stages.

1

H MRS as a biochemical marker in AD

Recent trials of disease modifying treatments, and advances in understanding the molecular biology of AD offer the promise of useful therapeutic interventions in the near future (Hsiao et al., 1996; Schenk et al., 1999). Disease modifying therapies in AD will be most useful before irreversible damage characterized by neuron loss takes place prior to the onset of dementia (Cummings et al., 1998; Kordower et al., 2001). For this reason, improved methods for early diagnosis have become imperative. An ideal neuroimaging marker should be able to accurately; (1) detect early neurodegenerative pathology, (2) reflect pathological stage across the entire severity spectrum, (3) predict when an individual with early pathology will become demented, and (4) monitor the effect of a therapeutic intervention on the neurodegenerative pathology. Different neuroimaging markers are being assessed to identify those that fulfill these criteria. For example, MR-based volumetry is being investigated as a structural imaging marker, fluoro2-deoxyglucose (FDG) positron emission tomography (PET) as a metabolic imaging marker, amyloid imaging as a plaque or tangle density marker, and 1H MRS as a biochemical imaging marker. This section will focus on the strengths and limitations of 1H MRS

as a biochemical imaging marker for the pathological progression of AD. First the clinical and pathological progression of AD and clinical-pathological correlation will be discussed, so that the investigation of imaging markers in AD can be interpreted. Clinical and pathological progression of AD There is agreement on clinical grounds that the earliest symptom of AD is memory dysfunction (Petersen et al., 1994), yet a significant number of elderly individuals with memory difficulties do not meet the clinical criteria for dementia. In the cognitive continuum, these individuals reside between normal cognition and dementia. Several clinical definitions have been proposed for these individuals with memory impairment such as: amnestic mild cognitive impairment (MCI) (Flicker et al., 1991), age associated memory impairment (Crook et al., 1987–1988), clinical dementia rating score of 0.5 (Morris, 1993), or minimally impaired. The clinical outcome of people with amnestic MCI has been investigated by several groups, and all (Flicker et al., 1991; Bowen et al., 1997; Petersen et al., 1999; Ritchie et al., 2001; Bennett et al., 2002) have concluded that patients with amnestic MCI have a higher risk of developing AD than cognitively normal elderly. Histopathological hallmarks of AD are neurofibrillary tangles and senile plaques. According to Braak and Braak (1991), the pathological progression of AD follows a hierarchical topographical course in the brain. Neurofibrillary pathology has a predilection for the limbic cortical structures, first involving the entorhinal cortex, and hippocampus, and later the neocortex. It involves the visual and primary sensory motor cortex at the very late stages of the disease. AD pathology encountered in cognitively normal elderly individuals also follows the same pattern in an age dependent manner but at a rate slower than in AD (Arriagata et al., 1992; Morris et al., 1996; Morris and Price, 2001; Delacourte et al., 1999; Gerober et al., 1999; Price and Morris, 1999; Schmitt et al., 2000). Thus, the number and extent of neurofibrillary tangle and senile plaque formation in the brain of the elderly who are destined to develop AD exists in a continuum, and the clinical diagnosis of AD is made only after a certain pathological threshold is reached (Gerober et al., 1999).

1

H MRS in pre-clinical, prodromal and clinical AD

Data from independent sites reveal that mI/Cr is increased in patients with amnestic MCI (Kantarci et al., 2000; Catani et al., 2001), very mild AD (Huang et al., 2001), and pre-dementia phase of Down’s syndrome (Shonk and Ross, 1995b; Huang et al., 1999), while NAA/Cr levels are normal compared to elderly controls. Regional elevation of mI/Cr levels in prodromal AD without a decrease in NAA/Cr suggests that 1H MRS may be sensitive to the biochemical changes in the pathological progression of AD before there is a significant loss of neuronal integrity in that

AD

Prodromal AD

Pathological continuum

Pre-clinical AD

The clinical correlates of Braak and Braak pathological staging of AD demonstrate an association between cognitive dysfunction and neurofibrillary pathology (Arriagata et al., 1992; Delacourte et al., 1999; Gerober et al., 1999). Cognitive function declines as neurofibrillary pathology extends from the entorhinal cortex to the neocortex. People who are destined to develop AD pass through three phases: (1) pre-clinical AD: early AD pathology with no clinical symptoms, (2) prodromal AD: pathological involvement extends into the neocortex with cognitive impairment that does not fulfill the diagnostic criteria for dementia and (3) clinical AD: with further pathological involvement in the neocortex, and cognitive impairment fulfills the clinical criteria for dementia and AD. Neuron loss in the entorhinal cortex differentiates people with MCI from cognitively normal elderly individuals with neuropathological or preclinical AD, which suggests that neuropathological AD becomes symptomatic with loss of entorhinal cortex neurons and impairment in memory function (Morris and Price, 2001). This finding is further strengthened with the knowledge that entorhinal cortex and hippocampus are involved in declarative memory function, and memory impairment in people with amnestic MCI is probably a manifestation of selective loss of neurons in these structures. Therefore people with amnestic MCI, who eventually develop AD can reasonably be regarded as having prodromal AD, a transitional phase that the elderly who develop AD pass through. Figure 34.3 illustrates the possible clinical-pathological correlations in preclinical, prodromal and clinical phases of AD.

AD

MR spectroscopy in aging and dementia

Amnestic MCI with AD Pathology Normal Elderly with AD Pathology Normal

Memory impairment

Dementia

Clinical continuum Fig. 34.3 Clinical-pathological correlation during the progression of AD. Elderly people who are at the earliest stages in the pathological progression of AD usually have normal cognition. Memory is the first cognitive domain to get impaired with the progression of the neurodegenerative pathology. People with amnestic MCI, who develop AD in the future are at this intermediate stage with pathological involvement, which matches with the prodromal symptoms of memory impairment. The diagnosis of AD is established when a certain pathological threshold is reached and the individual meets the clinical criteria for dementia. Both the clinical and pathological progression of AD exists in a continuum, and pathological findings in cognitively normal elderly, people with MCI and AD often overlap.

region. It also implies that different metabolites may be useful for predicting and monitoring different pathological stages in the course of AD. For example, we identified higher mI/Cr at baseline in cognitively normal elderly people who subsequently converted to amnestic MCI than normals who remained stable. NAA/Cr was lower at baseline in people with amnestic MCI who later converted to AD than MCI who remained stable at longitudinal follow-up (Table 34.3) (Kantarci et al., 2002b). Therefore mI/Cr may be useful for predicting pre-clinical AD in cognitively normal elderly and monitoring individuals with pre-clinical AD, and NAA/Cr may be useful for predicting prodromal AD in people with MCI, and monitoring patients with prodromal AD (Figure 34.4). Figure 34.5 summarizes the proposed sequence of

585

586

Kejal Kantarci and Clifford R. Jack, Jr.

Table 34.3. Baseline 1H MRS metabolite ratios (median and range) in the stable and converter clinical groups

N NAA/Cr MI/Cr

Cognitively stable normal elderly control

Control converter to amnestic MCI or AD

Amnestic MCI stable

Amnestic MCI converter to AD

62 1.54 (1.24–1.75) 0.63 (0.49–0.83)*

7 1.57 (1.41–1.59) 0.71 (0.49–0.78)*

17 1.51 (1.39–1.68)* 0.70 (0.54–0.82)

7 1.38 (1.22–1.5)* 0.69 (0.59–0.84)

*Difference between stable and converter clinical groups is statistically significant (rank-sum test P  0.05).

NAA

Normal aging

Cr

ml

Pre-clinical AD

Prodromal AD

Mild AD

ml/Cr ↑

ml/Cr ↑↑

ml/Cr ↑↑

NAA/Cr↓

NAA/Cr↓↓

Control Fig. 34.5 Proposed sequence of 1H MRS changes in the posterior cingulate voxel during the progression of AD. mI/Cr ratio increases in pre-clinical AD. NAA/Cr ratio starts decreasing in the prodromal phase. NAA/Cr ratio continues to decrease and mI/Cr ratio remains elevated in patients with the clinical diagnosis of AD.

NAA

Cr 1

ml

MCI

NAA ml

Cr

AD

4.0

3.0

2.0

1.0

ppm

Fig. 34.4 Examples of proton spectra obtained from the posterior cingulate volume of interest (VOI) with an TE of 30 ms in a control subject (top), in a patient with MCI (middle), and in a patient with AD (bottom). NAA/Cr ratio is lower in the patient with AD than both the patient with MCI and the control subject. mI/Cr ratios are higher in patients with MCI and AD than the control subject. The mI/Cr ratio is also higher in the patient with AD than the patient with MCI. (Kantarci et al., 2000).

H MRS changes in the posterior cingulate voxel during the pathological and clinical progression of AD. NAA/Cr measurements may also predict cognitive decline and monitor disease activity in patients with clinically established AD (Adalsteinsson et al., 2000; Jessen et al., 2001). As neurofibrillary pathology and associated neuron loss progresses stereotypically in the cortex of people with AD, regional decreases in the neuronal viability marker NAA/Cr may determine the extent of pathological involvement in patients with AD. Furthermore, regional NAA/Cr measures could be a non-invasive method of staging pathological progression of AD. For example NAA/Cr levels from a limbic cortical region may decline faster in mild AD but the NAA/Cr in a neocortical region may decline faster in moderate AD. Longitudinal change in metabolite levels in people with pre-clinical and prodromal AD remain to be investigated. It should be noted however, that the pathological and the clinical progression exist in a continuum, and the same should be expected of 1 H MRS metabolite levels. Overlap between the metabolite levels in different clinical groups is

MR spectroscopy in aging and dementia

expected both because of the biological overlap and also because of the possible measurement errors introduced by the technique itself. The value of 1H MRS in early AD is likely to be in applications where group effects are sought such as monitoring effectiveness of therapies in drug trials.

NAA 3T Cr

ml

Cho GluGln

Future perspectives on MRS in aging and dementia

NAA 1.5 T

This section will provide an overview on the less explored areas of MRS in aging and dementia, in order to give a perspective on present research questions.

Cr ml

Cho

GluGln

MRS at magnetic fields higher than 1.5 T MRS performed at higher magnetic field strength has the advantages of higher SNR and improved spectral resolution. These gains are partially lost however, with the decrease in transverse relaxation times, and increase in magnetic susceptibility effects at higher magnetic field strengths (Hetherington et al., 1997; Barker et al., 2001). With integration of higher magnetic field machines into clinical practice, there is growing interest in the diagnostic performance of MRS at higher fields with respect to the established magnetic field strength of 1.5 T (Gonen et al., 2001; Kantarci et al., 2003b). Higher SNR at higher field strength would on theoretical grounds be expected to increase the diagnostic accuracy and test-retest reproducibility of MRS. Improved spectral resolution would allow for accurate quantification of metabolites such as glutamine (Gln) and glutamate (Glu) that cannot be consistently resolved in the 1H MR spectra at 1.5 T (Figure 34.6). Quantification of Gln and Glu may be diagnostically useful in dementia, because a decrease in Gln and Glu levels have been reported in patients with AD (Antuono et al., 2001; Hattori et al., 2002). In our experience, the advantages of the higher field strength of 3 T are partially offset by the sensitivity of 3 T to field inhomogeneity (Kantarci et al., 2003b). Improving field homogeneity within the voxel may help to achieve optimum diagnostic performance from 1H MRS at 3 T in AD. One way of improving field homogeneity is regionally selective

4.0

3.0

2.0

1.0

ppm

Fig. 34.6 Examples of 1H MR spectra with an TE of 30 ms obtained from the posterior cingulate voxel at 3 T (top) and 1.5 T (bottom). Glu and Gln peak resonances are better resolved at 3 T than at 1.5 T. The resonance at 3.4 ppm. (immediately left of Cho), that is more prominent in the 1.5 T than in the 3 T spectra is scyllo-inositol and/or taurine. (Kantarci et al., 2003b).

higher-order shimming over the 1H MRS voxel (Kim et al., 2002). It is reasonable to expect the diagnostic performance of 1H MRS at higher magnetic field strengths to improve as the flexibility of the shimming software increases, and other engineering advances geared toward clinical imaging at fields higher than 1.5 T become commercially available.

31

P MRS in AD

The only other nucleus studied with MRS in dementia is 31P. Some of the earlier studies did not find any changes in people with AD on 31P MRS (Bottomley et al., 1992; Murphy et al., 1993). Others identified a consistent increase in phosphomonoesters (PME), phosphodiesters (PDE) and PME/PDE in people with AD (Gonzales et al., 1996; Pettegrew et al.,

587

588

Kejal Kantarci and Clifford R. Jack, Jr.

1997). The change in PME and PDE levels in AD is thought to originate from defects in membrane metabolism in AD. In a postmortem study performed at high field, the decrease in PME and PDE was further exagerated in carriers of the apolipoprotein–E4 allele, compared to carriers of the apolipoprotein–E3/3 allele, indicating that membrane phospholipid metabolite alterations are more severe in carriers of apolipoprotein–E4 allele (Klunk et al., 1998). Longitudinal follow-up of a patient with pre-clinical AD revealed higher than normal PME levels in the pre-clinical phase of AD and decreasing PME levels as the patient progressed to AD (Pettegrew et al., 1995). It is possible that 31P MRS findings may change during the progression of AD. Longitudinal studies in larger populations are needed to understand the time course of 31P MRS changes in AD. 31P MRS has a much lower sensitivity than 1H MRS. Due to lower SNR values, 31P MRS requires larger voxels, which limits the specificity of findings from a certain region in the brain. This is a drawback when regional differences are of interest. 31 P MRS may benefit from the integration of higher field machines into clinical practice, because the increased SNR at higher magnetic field strengths would allow for smaller voxels. Three-dimensional MRSI Volumetric MRSI techniques were developed in order to obtain multi-voxel MRS data from the whole brain in one acquisition. Regional MRS data acquired from a three-dimensional (3D) MRSI acquisition may be useful in differential diagnosis of dementia syndromes that affect certain regions in the brain more severely than others. For example, NAA/Cr levels are lower, mI/Cr levels are higher than normal both in patients with FTD and AD. However, patients with FTD have more severe changes in the frontal lobes than patients with AD, which may help in distinguishing the two disorders (Ernst et al., 1997). One difficulty with 3D MRSI of the brain is that a volume covering the whole brain normally contains different tissue types with variable magnetic susceptibility creating local inhomogeneities in the static magnetic field. MR spectra that are close to the

frontal sinuses or temporal bone air cells are affected prominently by the air–tissue susceptibility differences. Unfortunately, the most severely pathologically affected region in AD is the anteromedial temporal lobe, which is close to the petrous bone– air–tissue interface. Rigorous shimming algorithms such as higher-order shimming may improve spectral quality from this critical region. Due to the abundance of data generated by the 3D MRSI sequences, spectra that are of sub-optimal quality need to be eliminated manually. Automated techniques have been proposed for elimination of poor quality spectra. For example in one study, 31% of brain 3D MRSI spectra with TE of 135 ms had to be discarded through automated analysis (Ebel et al., 2001). The percentage of poor quality spectra increased to 48% when TE of 25 ms was used, owing to lipid contamination from the skull at short TEs. Advances in lipid suppression techniques may allow for better quality spectra with short TE 3D MRSI.

Conclusions Present data support the concept that 1H MRS may in the future become an adjunct to clinical evaluation for differential diagnosis of dementing illnesses. The value of 1H MRS in monitoring the disease activity in AD is expected to be in areas where group effects are sought such as monitoring effectiveness of therapies in drug trials. Elevation of mI/Cr levels in people with amnestic MCI and very early AD without a decrease in the neuronal viability marker NAA/Cr suggests 1H MRS may also be valuable in monitoring early disease progression for preventive therapies. Machines with magnetic field strengths of greater than 1.5 T are now being integrated to clinical practice. Higher SNR at higher field strength would be expected to increase the diagnostic accuracy and test-retest reproducibility of MRS. Improved spectral resolution would allow for more accurate quantification of metabolites such as Gln and Glu, which may provide additional diagnostic information in dementia. MRS with nuclei other than 1H such as 31P, may also benefit from the SNR gain at higher magnetic field strengths, improving the diagnostic accuracy of these techniques.

MR spectroscopy in aging and dementia

Regional MRS data acquired with 3D MRSI acquisitions may especially be useful in differential diagnosis of dementia syndromes that affect certain regions in the brain more severely than others. Overall, MRS is a promising investigational technique in aging and dementia at this time. The potential clinical application of MRS in aging and dementia however, is growing with technical advances in the field.

REFERENCES Adalsteinsson E, Sullivan EV, Kleinhans N, Spielman DM, Pfefferbaum A. 2000. Longitudinal decline of the neuronal marker N-acetyl aspartate in Alzheimer’s disease. Lancet 355: 1696–1697. American Psychiatric Association. DSM-III-R. 1987. Diagnostic and Statistical Manual of Mental Disorders, 3rd edn. revised. Washington, DC. Antuono PG, Jones JL, Wang Y, Li S.J. 2001. Decreased glutamate  glutamine in Alzheimer’s disease detected in vivo with 1H-MRS at 0.5 T. Neurology 56(6): 737–742. Arriagata PV, Growdon JH, Hedley-Whyte ET, Hyman BT. 1992. Neurofibrillary tangles but not senile plaques parallel duration and severity of Alzheimer’s disease. Neurology 42: 631–639. Barker PB, Hearshen DO, Boska MD. 2001. Single-voxel proton MRS of the human brain at 1.5 T and 3 T. Magn Reson Med 45: 765–769. Bates T, Strandward M, Keelan J, Davey G, Munro P, Clarke J. 1995. Inhibition of N-acetylaspartate production: Implications for 1H MRS studies in vivo. Neuroreport 7: 1397–1400. Bennett DA, Wilson RS, Schneider JA, et al. 2002. Natural history of mild cognitive impairment in older persons. Neurology 59: 198–205. Bitsch A, Bruhn H, Vougioukas V, et al. 1999. Inflammatory CNS demyelination: histopathologic correlation with in vivo quantitative proton MR spectroscopy. Am J Neuroradiol 20: 1619–1627. Bottomley P, Cousins J, Pendrey D, et al. 1992. Alzheimer dementia: quantification of energy metabolism and mobile phosphoesters with p-31 NMR spectroscopy. Radiology 183: 695–699. Bowen J, Teri L, Kukull W, et al. 1997. Progression to dementia in patients with isolated memory loss. Lancet 349: 763–765. Braak H, Braak E. 1991. Neuropathological staging of Alzheimer’s disease. Acta Neuropathol (Berl) 82: 239–259.

Brooks WM, Stidley CA, Petropoulos H, et al. 2000. Metabolic and cognitive response to human traumatic brain injury. J Neurotrauma 17: 629–640. Burton EJ, Karas G, Paling SM, et al. 2002. Patterns of cerebral atrophy in dementia with Lewy bodies using voxel-based morphometry. Neuroimage 17: 618–630. Catani M, Cherubini A, Howard R. 2001. 1H MR spectroscopy differentiates mild cognitive impairment from normal brain aging. Neuroreport 12(11): 2315–2317. Chang L, Ernst T, Poland R, Jenden DJ. 1996. In vivo proton magnetic resonance spectroscopy of the normal aging human brain. Life Sci 58(22): 2049–2056. Chantal S, Labelle M, Bouchard R, Braun CM, Boulanger Y. 2002. Correlation of regional proton magnetic resonance spectroscopy changes with cognitive deficits in mild Alzheimer disease. Arch Neurology 59: 955–962. Charles HC, Lazeyas F, Krishnan KRR, et al. 1994. Proton spectroscopy of human brain: effects of age and sex. Prog Neuro-Psychopharmachol and Biol Psychiatr 18: 995–1004. Christiansen P, Schlosser A, Henriksen O. 1995. Reduced N-acetylaspartate content in the frontal part of the brain in patients with probable Alzheimer’s disease. Magn Reson Imaging 13(3): 457–462. Crook T, Bahar H, Sudilovski A. 1987–88. Age associated memory impairment: diagnostic criteria and treatment strategies. Int J Neurol 21–22: 73–82. Cummings JL, Vinters HV, Cole GM, Khachaturian ZS. 1998. Alzheimer’s disease etiologies, pathophysiology, cognitive reserve, and treatment opportunities. Neurology 51(suppl 1): S2–S17. Delacourte A, David JP, Sergeant N, et al. 1999. The biochemical pathway of neurofibrillary degeneration in aging and Alzheimer’s disease. Neurology 52: 1158–1165. Ebel A, Soher BJ, Maudsley AA. 2001. Assessment of 3D proton MR echo-planar spectroscopic imaging using automated spectral analysis. Magn Reson Med 46: 1072–1078. Ernst T, Chang L, Melchor R, Mehringer M. 1997. Frontotemporal dementia and early Alzheimer disease: Differentiation with frontal lobe H-1 MR spectroscopy. Radiology 203: 829–836. Flicker C, Ferris SH, Reisberg B. 1991. Mild cognitive impairment in the elderly: predictors of dementia. Neurology 41: 1006–1009. Gerober E, Dickson D, Sliwinski MJ, et al. 1999. Memory and mental status correlates of Braak staging. Neurobiol Aging 20(6): 573–579. Glanville NT, Byers DM, Cook HW, Spence MW, Palmer FB. 1989. Differences in the metabolism of inositol and phosphoinositides by cultured cells of neuronal and glial origin. Biochimica et Biophysica Acta 1004(2): 169–179.

589

590

Kejal Kantarci and Clifford R. Jack, Jr.

Gomez-Isla T, Gowdon WB, McNamara M, et al. 1999. Clinicopathologic correlates in temporal cortex in dementia with Lewy bodies. Neurology 53(9): 2003–2009. Gonen O, Gruber S, Li BSY, Mlynarik V, Moser E. 2001. Multivoxel 3D proton spectroscopy in the brain at 1.5 versus 3.0 T: Signal-to-noise ratio and resolution comparison. Am J Neuroradiol 22: 1727–1731. Gonzales RG, Guimaraes AR, Moore GJ, Crawley A, Cupples LA, Growdon JH. 1996. Quantitative in vivo31P magnetic resonance spectroscopy of Alzheimer’s disease. Alzh Dis Assoc Dis 10(1): 42–52. Hattori N, Abe K, Sakoda S, Sawada T. 2002. Proton MR spectroscopic study at 3 Tesla on glutamate/glutamine in Alzheimer’s disease. Neurochemistry 13(1): 183–186. Hetherington HP, Pan JW, Chu W-J, Mason GF, Newcomer BR. 1997. Biological and clinical MRS at ultra-high field. NMR Biomed 10: 360–371. Holmes C, Cairns N, Lantos P, Mann A. 1999a. Validity of current clinical criteria for Alzheimer’s disease, vascular dementia and dementia with Lewy bodies. Br J Psychiatr 174: 45–50. Holmes C, Cairns N, Lantos P, Mann A. 1999b. Validity of current clinical criteria for Alzheimer’s disease, vascular dementia and dementia with Lewy bodies. Br J Psychiatr 174: 45–50. Hsiao K, Chapman P, Nilsen S, et al. 1996. Correlative memory deficits, Abeta elevation, and amyloid plaques in transgenic mice. Science 274(5284): 99–102. Huang W, Alexander GE, Chang L, et al. 2001. Brain metabolite concentration and dementia severity in Alzheimer’s disease. A 1H MRS study. Neurology 57: 626–632. Huang W, Alexander GE, Daly EM, et al. 1999. High brain myo-inositol levels in the predementia phase of Alzheimer’s disease in adults with Down’s syndrome: a 1H MRS study. Am J Psychiatr 156: 1879–1886. Hugg JA, Kuzniecky RI, Gilliam FG, Morawetz RB, Faught RE, Hetherington HP. 1996. Normalization of contralateral metabolic function following temporal lobectomy demonstrated by 1H magnetic resonance spectroscopic imaging. Ann Neurol 40: 236–239. Jessen F, Block W, Träber F, et al. 2000. Proton MR spectroscopy detects a relative decrease of N-acetyl aspartate in the medial temporal lobe of patients with AD. Neurology 55: 684–688. Jessen F, Block W, Träber F, et al. 2001. Decrease of N-acetylaspartate correlates with cognitive decline of AD patients. Neurology 57(5): 930–932. Kantarci K, Jack CR, Xu YC, et al. 2000. Regional metabolic patterns in mild cognitive impairment and Alzheimer’s disease, a 1H MRS study. Neurology 55(2): 210–217. Kantarci K, Petersen RC, Boeve BF, et al. 2002b. 1H MRS predicts development of Alzheimer’s disease in patients with mild

cognitive impairment and disease progression in patients with Alzheimer’s disease. Neurobiol Aging 23(18): S350. Kantarci K, Petersen RC, Boeve BF, et al. 2003a. 1H MRS in differential diagnosis of common dementia syndromes. Neurology 60(5) (suppl 1): A161. Kantarci K, Reynolds GH, Petersen RC, et al. 2003b. 1H MRS in mild cognitive impairment and Alzheimer’s disease; Comparison of 1.5 T and 3 T. Am J Neuroradiol 24: 843–849. Kantarci K, Smith GE, Ivnik RJ, et al. 2002a. 1H MRS, cognitive function, and apolipoprotein E genotype in normal aging, mild cognitive impairment and Alzheimer’s disease. J Int Neuropsychol Soc 8: 934–942. Kattapong VJ, Brooks WM, Wesley MH, Kodituwakku PW, Rosenberg GA. 1996. Proton magnetic resonance spectroscopy of vascular- and Alzheimer-type dementia. Arch Neurol 53: 678–680. Kim D, Adalsteinsson E, Glover GH, Spielman DM. 2002. Regularized higher-order in vivo shimming. Magn Reson Med 48: 715–722. Klunk WE, Panchalingam K, McClure RJ, Stanley JA, Pettegrew JW. 1998. Metabolic alterations in postmortem Alzheimer’s disease brain are exaggerated by Apo-E4. Neurobiol Aging 19(6): 511–515. Klunk WE, Panchalingam K, Moosy J, Mc Clure RJ, Pettegrew JW. 1992. N-acetyl-L-aspartate and other amino acid metabolites in Alzheimer’s disease brain: a preliminary proton nuclear magnetic resonance study. Neurology 42: 1578–1585. Knopman DS, DeKosky ST, Cummings JL, et al. 2001. Practice parameter: Diagnosis of dementia (an evidence based review). Report of the quality standards subcommittee of the American Academy of Neurology. Neurology 56:1143–1153. Kordower JH, Chu Y, Stebbins GT, et al. 2001. Loss and atrophy of layer II entorhinal cortex neurons in elderly people with mild cognitive impairment. Ann Neurol 49: 202–213. Leary SM, Brex PA, MacManus DG, et al. 2000. A 1H magnetic resonance spectroscopy study of aging in parietal white matter: implications with trials in multiple sclerosis. Magn Reson Imaging 18: 455–459. Long J, Mouton P, Jucker M, Ingram D. 1999. What counts in brain aging? Design-based stereological analysis of cell number. J Gerontol Ser A Biol Sci Med Sci 54: B407–B417. MacKay S, Ezekiel F, Di Sclafani V, et al. 1996. Alzheimer’s disease and subcortical ischemic vascular dementia: evaluation by combining MR imaging segmentation and H-1 MR spectroscopic imaging. Radiology 198: 537–545. Massoud F, Devi G, Stern Y, et al. 1999. A clinicopathological comparison of community-based and clinic-based cohorts of patients with dementia. Arch Neurol 56(11): 1368–1373. McKeith IG, Galasko D, Kosaka K, et al. 1996. Consensus guidelines for the clinical and pathologic diagnosis of dementia with Lewy bodies. Neurology 47: 1113–1124.

MR spectroscopy in aging and dementia

Meyerhoff DJ, MacKay S, Norman D, Van Dyke C, Fein G, Weiner MW. 1994. Axonal injury and membrane alterations in Alzheimer’s disease suggested by in vivo proton magnetic resonance spectroscopic imaging. Ann Neurol 36: 40–47. Miller BL, Moats RA, Shonk T, Earnst T, Wooley S, Ross BD. 1993. Alzheimer disease: Depicition of increased cerebral myo-inositol with proton MR spectroscopy. Radiology 187: 433–437. Moats RA, Ernst T, Shonk TK, Ross BD. 1994. Abnormal cerebral metabolite concentrations in patients with probable Alzheimer disease. Magn Reson Med 32: 110–115. Mohanakrishnan P, Fowler AH, Vonsattel JP, et al. 1997. Regional metabolic alterations in Alzheimer’s disease: An in vitro 1H NMR study of the hippocampus and cerebellum. J Gerontol Biol Sci 52A(2): B111-B117. Morris JC. 1993. The Clinical Dementia Rating (CDR): current version and scoring rules. Neurology 43: 2412–2414. Morris JC, Price JL. 2001. Pathologic correlates of nondemented aging, mild cognitive impairment, and early-stage Alzheimer’s disease. J Mol Neurosci 17: 101–118. Morris JC, Storandt M, McKeel DW, Jr. et al. 1996. Cerebral amyloid deposition and diffuse plaques in normal aging: Evidence for presymptomatic and very mild Alzheimer’s disease. Neurology 46: 707–719. Murphy D, Bottomley P, Salerno J, et al. 1993. An in vivo study of phosphorous and glucose metabolism in Alzheimer’s disease using magnetic resonance spectroscopy and PET. Arch Gen Psychiatr 50: 341–349. Neary D, Snowden JS, Gustafson, et al. 1998. Frontotemporal lobar degeneration: A consensus on clinical diagnostic criteria. Neurology 51(6): 1546–1554. Parnetti L, Tarducci R, Presciutti O, et al. 1997. Proton magnetic resonance spectroscopy can differentiate Alzheimer’s disease from normal aging. Mech Ageing Devel 97: 9–14. Petersen RC, Smith GE, Ivnik RJ, Kokmen E, Tangalos EG. 1994. Memory function in very early Alzheimer’s disease. Neurology 44: 867–872. Petersen RC, Smith GE, Waring SC, Ivnik RJ, Tangalos EG, Kokmen E. 1999. Mild cognitive impairment clinical characterization and outcome. Arch Neurol 56: 303–308. Pettegrew JW, Klunk K, Panchalingam K, McClure RJ, Stanley JA. 1997. Magnetic resonance spectroscopic changes in Alzheimer’s disease. Ann NY Acad Sci 826: 282–306. Pettegrew JW, Klunk WE, Kanal E, Panchalingam K, McClure RJ. 1995. Changes in brain membrane phospholipid and highenergy phosphate metabolism precede dementia. Neurobiol Aging 16(6): 973–975. Pfefferbaum A, Adalsteinsson E, Spielman D, Sullivan EV, Lim KO. 1999a. In vivo spectroscopic quantification of the N-acetyl moiety, creatine, and choline from large volumes

of brain gray and white matter: effects of normal aging. Magn Reson Med 41: 276–284. Pfefferbaum A, Adalsteinsteinsson E, Spielman D, Sullivan E, Lim K. 1999b. In vivo brain concentration of N-acetyl compounds, creatine, and choline in Alzheimer’s disease. Arch General Psychiatr 56: 185–192. Price JL, Morris JC. 1999. Tangles and plaques in nondemented aging and preclinical Alzheimer’s disease. Ann Neurol 45: 358–368. Ritchie K, Artero S, Touchon J. 2001. Classification criteria for mild cognitive impairment. A population based validation study. Neurology 56: 37–42. Rose SE, de Zubicaray GI, Wang D, et al. 1999. A 1H MRS study of probable Alzheimer’s disease and normal aging: Implications for longitudional monitoring of dementia progression. Magn Reson Imaging 17(2): 291–299. Ross BD, Bluml S, Cowan R. 1998. In vivo MR spectroscopy of human dementia. Neuroimag Clin N Am 8(4): 809–822. Saunders DE, Howe FA, Boogaart A, Griffiths JR, Brown MM. 1999. Aging of the adult human brain: In vivo quantitation of metabolic content with proton magnetic resonance spectroscopy. J Magn Reson Imaging 9: 711–716. Schenk D, Barbour R, Dunn W, et al. 1999. Immunization with amyloid-beta attenuates Alzheimer-disease-like pathology in the PDAPP mouse. Nature 400(6740): 173–177. Schmitt FA, Davis DG, Wekstein DR, Smith CD, et al. 2000. Preclinical AD revisited. Neuropathology of cognitively normal older adults. Neurology 55: 370–376. Schuff N, Amend DL, Ezekiel F, et al. 1997. Changes of hippocampal N-acetyl aspartate and volume in Alzheimer’s disease A proton MR spectroscopic imaging and MRI study. Neurology 49: 1513–1521. Schuff N, Amend DL, Meyerhoff DJ, et al. 1998. Alzheimer disease: Quantitative H-1 MR Spectroscopic imaging of frontoparietal brain. Radiology 207: 91–102. Schuff N, Capizzano AA, Du AT, et al. 2002. Selective reduction of N-acetylaspartate in medial temporal and parietal lobes in AD. Neurology 58: 928–935. Schuff N, Ezekiel F, Gamst AC, et al. 2001. Region and tissue differences of metabolites in normally aged brain using multislice 1H magnetic resonance spectroscopic imaging. Magn Reson Med 45: 899–907. Shonk TK, Moats RA, Gifford PG, et al. 1995a. Probable Alzheimer’s disease: Diagnosis with proton MR spectroscopy. Radiology; 195: 65–72. Shonk T, Ross BD. 1995b. Role of increased cerebral myoinositol in the dementia of Down syndrome. Magn Reson Med 33: 858–861. Soher BJ, Vermathen P, Schuff N, et al. 2000. Short TE in vivo 1 H MR spectroscopic imaging at 1.5 T: acquisition and automated spectral analysis. Magn Reson Imaging 18: 1159–1165.

591

592

Kejal Kantarci and Clifford R. Jack, Jr.

Tedeschi G, Bertolino A, Lundbom N, et al. 1996. Cortical and subcortical chemical pathology in Alzheimer’s disease as assessed by multislice proton magnetic resonance spectroscopic imaging. Neurology 47: 696–704. Valenzuela MJ, Sachdev PS, Wen W, Shnier R, Brodaty H, Gillies D. 2000. Dual voxel proton magnetic resonance spectroscopy in the healthy elderly: Subcortical-frontal axonal N-acetylaspartate levels are correlated with fluid cognitive

abilities independent of structural brain changes. Neuroimage 12: 747–756. Waldman ADB, Rai GS, McConnell JR, Chaudry M, Grant D. 2002. Clinical brain proton magnetic resonance spectroscopy for management of Alzheimer’s and sub-cortical ischemic vascular dementia in older people. Arch Gerontol Geriat 35: 137–142.

MR spectroscopy in aging and dementia

Case Study 34.1 MRS for investigation of Alzheimer’s disease Adam D. Waldman, M.D., Ph.D., Hammersmith Hospitals and Institute of Neurology, London, UK.

Background Elevated mI and decreased NAA are characteristic findings in Alzheimer’s disease (AD).

Technique Short TE Single voxel-MRS posterior cingulate gyrus (PRESS TE  30 ms).

Real

Comments •

•

• •

•

Posterior cingulate gyrus voxel allows reproducible and quantifiable spectra predominantly from cortex in a region affected throughout the course of AD. mI/NAA ratios or NAA vs. mI plots optimize separation of the groups for diagnostic discrimination; overlap varies between series sensitivity 70–90% and specificity 75–95%. Local reference standards are essential for diagnostic use. Ratios from automated MRS processing and quantitative off-line processing give similar discrimination. Spectroscopic abnormalities in AD are predominantly GM, whereas in vascular dementia are mostly in WM.

NAA

ml

0.014

AD

0.012 0.010 0.008 0.006 0.004 0.002 0.000 4

3 2 1 Frequency (ppm)

1.2

0

Controls AD

1

Key points MRS distinguishes AD from normal subjects and vascular dementia, and may be used diagnostically. mI and NAA are the discriminant metabolites.

mI/Cr

0.8 0.6 0.4 0.2 0 0.8

1

1.2

1.4

1.6

1.8

2

NAA/Cr References Kantarci K, Jack Jr CR, Xu YC, Campeau NG, et al. 2000. Regional metabolic patterns in mild cognitive impairment and Alzheimer’s disease, a 1H MRS study. Neurology 55: 210–217. Shonk TK, Moats RA, Gifford P, Michaelis T, Mandigo JC, Izumi J, Ross BD. 1995. Probable Alzheimer disease: diagnosis with proton MR spectroscopy. Radiology 195: 65–72.

593

35

MR spectroscopy in neurodegeneration C.A. Davie Department of Clinical Neurosciences, Royal Free and University College Medical School, London, UK

Key points • Neurodegenerative diseases are characterized by neuronal dysfunction followed by cell death. • Metabolite changes in idiopathic Parkinson’s disease (IPD) are inconsistent. • Multiple system atrophy patients demonstrate significant reduction in absolute N-acetylaspartate (NAA) and NAA/creatine ratios when compared with IPD. • Lactate appears to be increased in Huntington’s disease.

Introduction Neurodegenerative disease encompasses a very wide group of disorders affecting the central nervous system (CNS). Many of these disorders arise from the combined effects of genetic predisposition as well as largely unidentified environmental factors. Although the neurodegenerative disorders show wide diversity in etiology and clinical phenotype they are all characterized by neuronal dysfunction followed by cell death. The basic mechanisms that underlie neurodegenerative diseases are unknown. Loss of function of specific regions of the brain is due to incapacitation of cells that constitute those regions. Cells can simply stop functioning normally (neurons may cease to transmit signals), or they may die. There is now evidence that the pathology of several neurodegenerative diseases is due to inappropriate apoptosis or programmed cell death (PCD). Recent developments have shown that inappropriate 594

activation of apoptotic pathways is a contributing event in many neurodegenerative diseases. A non-invasive tool that has the potential to monitor the progress of neuronal degeneration from a very early stage in the disease could have a profound impact on our understanding of the mechanisms that lead to neuronal dysfunction and ultimately death. Ideally, such a tool would have the ability to monitor treatment of neurodegenerative diseases from an early stage, even in some cases before the development of neurological symptoms. Similarly, techniques that give an insight into microglial activation would also be useful in the development of more specific treatment moieties in neurodegenerative disease. Molecular pathways of PCD are activated in various neurodegenerative disorders including Parkinson’s disease (idiopathic Parkinson’s disease, IPD) and Huntington’s disease (HD).

MR spectroscopy techniques MR spectroscopy (MRS) allows the quantification and serial measurement of major brain metabolites (Vion-Dury et al., 1994). The most widely used applications are to quantify chemicals containing phosphorus (31P MRS) or hydrogen (1H MRS). MRS allows the quantification of different chemicals in a single acquisition, which can be repeated serially. 31 P MRS allows the direct, non-invasive measurement of key metabolites involved in energy metabolism including phosphocreatine (PCr), inorganic phosphate (Pi) and adenosine triphosphate (ATP). 1 H MRS allows quantification of a number of metabolites including N-acetylaspartate (NAA) (a putative

MR spectroscopy in neurodegeneration

neuronal marker), Lactate (Lac) (a marker of anaerobic metabolism), -amino-butyric acid (GABA), glutamate (Glu) and glutamine (Gln), total creatine (Cr) (including PCr), choline (Cho) containing compounds and myo-inositol (mI) (cf. Chapter 1). In the study of neurodegenerative brain diseases 1H MRS is normally carried out on a localized volume of brain tissue down to 1 ml. The spectroscopic acquisition can be repeated allowing the study of several brain regions in one examination. Some studies however, have employed techniques which allow either acquisition of a strip or area of localized volumes chemical shift imaging (CSI), or 1H MR spectroscopic imaging (1H MRSI). CSI/MRSI allows simultaneous acquisition of signal from one or several brain slices. Each slice consists of many individual volumes (usually a 32 32 matrix) with each volume being 1 ml or less in size. Low-resolution images can be produced for each of the individual metabolites, which has the great advantage of providing metabolite information from several regions simultaneously. However, the downside to this is poorer signal-to-noise ratio (SNR) and often less spatially accurate information. Single slice and multi-slice 1H MRSI permits the mapping of brain metabolite distribution which makes them particularly useful in studying diffuse diseases and heterogeneous lesions of the CNS; but until recently were only available with long echo time (TE) using commercial, automated sequences. MRS carried out in spectrometers with higher field strengths can be used to examine much smaller regional brain volumes in animals. Furthermore, in vitro spectroscopy can be applied to brain tissue ex-vivo and to cell extracts or cultures. NAA (Birken and Oldendorf, 1989) is an amino acid of unknown function, which has been shown in experimental studies on primary cell cultures from neonatal rat brains and optic nerves to be contained almost exclusively within neurons (Urenjak et al., 1993). NAA is also present in oligodendrocyte progenitor (O2A) cells though these are only present in very small numbers in healthy adult human brain (Scolding et al., 1998). Although mature oligodendrocytes have been shown to express NAA in vitro under specific conditions (Bhakhoo and Pearce, 2000), a loss of neurons in vivo would predispose to a persistent reduction in the concentration of NAA. Such a reduction has

been demonstrated reproducibly in several diseases characterized by neuronal loss (Rudkin and Arnold, 1999) including HD (Jenkins et al., 1993, 1998) and autosomal-dominant cerebellar ataxia. Reversible changes in NAA have been demonstrated in human disease including mitochondrial cytopathy (De Stefano et al., 1995). NAA is reduced in isolated rat brain mitochondria after exposure to specific inhibitors of the mitochondrial respiratory chain (Bates et al., 1996). A decrease in NAA levels seen in disease states by 1H nuclear MR (1H NMR) spectroscopy in vivo may reflect impaired but potentially reversible mitochondrial energy production rather than neuronal cell loss. A reversible decrease in NAA has been demonstrated in patients with motor neuron disease after treatment with the Glu antagonist riluzole (Kalra et al., 1998). The ability of MRS to detect neuronal dysfunction in addition to cell death provides a powerful tool for the assessment of putative therapies early in the course of neurodegeneration. The topic of neurodegenerative illnesses presenting most commonly with dementia covered elsewhere (cf. Chapter 34). This chapter will review the application of MRS in the neurodegenerative illnesses which present primarily with hyperkinetic or akinetic rigid syndromes.

MRS in the parkinsonian syndromes Parkinsonian syndromes are characterized by slowness of movement (bradykinesia), and thought (bradyphrenia), tremor at rest and extrapyramidal rigidity. Differentiating between the various clinical syndromes can be difficult, particularly early in the clinical picture. It is estimated in various series that approximately one quarter of patients diagnosed as having IPD during life have another neuropathological diagnosis at postmortem (Hughes et al., 1992). Although the accuracy of diagnosis increases in a specialist movement disorder service (Hughes et al., 2002) the “correct” clinical diagnosis of parkinsonian syndromes other than IPD is reached after a mean of 5.4 years after the symptom onset (Hughes et al., 2002). A number of MR studies have attempted to differentiate between the various parkinsonian syndromes and to improve diagnostic specificity.

595

596

C.A. Davie

Parkinson’s disease IPD is a neurodegenerative disorder of unknown cause. Age is the most consistent risk factor. Although the mechanisms underlying neurodegeneration in Parkinson’s disease are not fully understood, considerable evidence suggests that genetic factors can influence susceptibility to the disease. The discovery of the -synuclein (PARK1) and other parkin (PARK2) genes has demonstrated that genetic mutations can lead to the development of phenotypes of Parkinson’s disease. Conventional MR imaging (MRI) is normal in IPD though this investigation is usually performed to exclude a structural cause for the development of parkinsonism. The results of MRS studies on patients with IPD have been mixed. The first and largest study of localized 1H MRS in IPD by Holshouser et al. (1995) studied 151 patients. In this study no significant reduction was seen in NAA/Cr ratio from a volume localized to the basal ganglia. The authors noted a decrease in the NAA/Cho ratio in the older IPD patients and concluded that their findings may indicate a slight decrease in NAA or increase in Cho and Cr in this subgroup. A much smaller study by Clarke et al. (1997) carried out absolute quantitation of metabolites from the basal ganglia in five IPD patients using tissue water as an internal concentration reference. No abnormality was detected in Cho, Cr, NAA, Gln or Glu evaluated separately or as a combination of both (Glx). The same authors carried out a later study (Clarke and Lowry, 2000) in six patients with IPD again using absolute quantitation and observed an increase in Cho concentration (of approximately 60%) from the lentiform nucleus which was not observed in a group of six patients with clinically probable multiple system atrophy (MSA). A study by Hoang et al. (1998) in five patients with PD performed 1H MRS from single volumes localized to putamen, occipital gray matter (GM) and posterior parietal white matter (WM) using absolute metabolite concentrations. Quantitative phosphorus and protondecoupled phosphorus MRS of superior biparietal WM and GM was also performed. This study failed

to show any statistically significant difference between the PD patients and controls. Energy metabolism was normal in all brain regions measured by MRS. The group from King’s College London (Ellis et al., 1996) initially reported a reduction in NAA/Cr and NAA/Cho in IPD patients. Subsequent reports from this group showed normal NAA/Cr in untreated IPD patients and those without levodopa-induced complications. However, NAA/Cho was reduced in de novo drug-naive IPD patients and normal in the levodopa-treated group (Ellis et al., 1997 ). The same group (Hu et al., 1999) have performed 1H MRS from volumes localized to cortex in IPD patients without dementia. They observed significant temporoparietal cortex reductions in NAA/Cr ratios compared to controls. There was a significant correlation between reduction in NAA/Cr ratios and measures of global cognitive decline, occurring independently of motor impairment. Hu and co-authors (Hu et al., 2000) also carried out regional (31P MRS) from right and left temporoparietal cortex, occipital cortex and a central voxel incorporating basal ganglia and brainstem in 10 non-demented IPD patients and nine age-matched control subjects. This showed significant bilateral increases in the P(i)/beta-ATP ratio of approximately 50% in temporoparietal cortex in the IPD patients compared with controls. In the right temporoparietal cortex, there was also a significant increase in the mean relative percentage suggesting that both glycolytic and oxidative pathways are impaired. The authors concluded that this dysfunction might reflect either the presence of primary cortical pathology or deafferentation of striato-cortical projections. Federico et al. (1999) reported a reduction in NAA/Cr from basal ganglia in 19 patients with IPD. A number of studies (Davie et al., 1995a; Tedeschi et al., 1997a; Turjanski et al., 1997) using metabolite ratios have shown no difference in NAA/Cr and NAA/Cho ratios from basal ganglia in IPD. A study by Abe (2000) looked at localized regions of 1 ml from the frontal cortex and putamen. This study employed metabolite ratios rather than absolute quantification. There was no alteration in metabolites from the frontal cortex in 23 IPD patients compared to

MR spectroscopy in neurodegeneration

controls and various groups of other parkinsonian syndromes. The NAA/Cr ratio from putamen was reduced in IPD patients by approximately 30%. A more recent quantitative study using MRSI has been carried out by O’Neill and colleagues (2002). This study assessed localized 1H MRS from substantia nigra, basal ganglia and cerebral cortex in 10 patients with IPD. The authors reported a significant reduction in Cr concentration of 24% from the substantia nigra in the IPD group. No other metabolite changes were observed in this and other brain regions. The significance of this finding in the substantia nigra has to be guarded since the 1H MRS spectra from the region of the substantia nigra contained contributions from non-nigral tissue. The only consistent finding in all these studies is a lack of consistency. This is in part due to a large number of very small studies which until recently have relied on metabolite ratio data. A further large multinational study using absolute quantification with longitudinal data and pathological correlation of diagnosis is needed. Multi-System Atrophy (MSA) MSA is a sporadically occurring neurodegenerative disease that presents with varying combinations of parkinsonism, cerebellar ataxia, autonomic failure and pyramidal signs of varying severity during the course of illness. MSA includes within its spectrum striatonigral degeneration, olivopontocerebellar atrophy and isolated autonomic failure. The majority of MSA patients present with parkinsonism and like IPD, this manifests as asymmetrical limb rigidity and bradykinesia with complaints of stiffness, muscular aching and hand clumsiness. Rest tremor is seen in MSA but is far less frequent than in IPD. Neuropathological findings consist of neuronal loss, gliosis and demyelination with widespread regional involvement particularly including the striatonigral, olivopontocerebellar and autonomic nervous systems. Argyrophilic neuronal and glial cytoplasmic inclusions that stain positive for synuclein are characteristic but are also occasionally seen in progressive supranuclear palsy (PSP) and corticobasal degeneration (CBD). Lewy bodies are not a feature.

Conventional MRI may show a rim of increased signal in the lateral putamen on T2-weighted imaging as a consequence of gliosis. A reduced signal may also be observed at this region probably as a consequence of iron deposition. Cerebellar and pontine atrophy may be observed and in some patients a so-called ‘hot cross bun’ is seen in the pons due to increased visibility of the lateral and longitudinal pontine fibers occurring in the context of pontocerebellar degeneration. The majority of MRS studies on patients with MSA have attempted to differentiate these patients from other parkinsonian syndromes and most frequently from IPD. Davie et al. (1995b) carried out 1H MRS localized to the lentiform nucleus in patients with asymmetrical, levodopa responsive IPD and patients with clinically probable MSA who were subdivided into striatonigral and olivopontocerebellar atrophy variants. The MSA patients showed a significant reduction in absolute [NAA] and NAA/Cr ratios from the lentiform nucleus compared with controls and IPD patients. This was most striking in the striatonigral patients who showed an approximate reduction of the NAA/Cr ratio of 30% compared to controls. Federico et al. (1999) also showed a reduction in the lentiform nucleus of NAA/Cr and NAA/Cho in MSA patients compared to controls and a reduction of NAA/Cr in MSA patients compared to those with IPD. A study by Ellis et al. (1996) showed a reduction in NAA/Cr and NAA/Cho in MSA compared to controls though a similar reduction was observed in patients with IPD. A study by Clarke and Lowry (Clarke et al., 1997) using absolute quantitation showed no changes in NAA, Cr or Cho from the lentiform nucleus in MSA. The study by Abe et al. (2000) of various parkinsonian syndromes studied 18 patients with clinically probable MSA and reported a reduction of NAA/Cr from 1 ml voxels in frontal cortex and putamen. Progressive Supranuclear Palsy (PSP) PSP, also known as Steele–Richardson–Olszewski (SRO) syndrome, presents as a symmetrical rather than asymmetrical akinetic–rigid syndrome – in contrast to IPD and MSA. It tends to present with

597

598

C.A. Davie

axial involvement rather than the limbs, causing early postural and gait instability with falls and a dystonic posture where the trunk is flexed but the neck extended. Any elderly patient with truncal or neck rigidity either in the absence of or with only mild limb involvement and impaired postural reflexes should be suspected of having PSP. Later, limb rigidity and bradykinesia develop in a symmetrical fashion but rest tremor is uncommon. Voluntary gaze problems are common but may be a late feature or even absent. This targets the pallidum (rather than the striatum), the substantia nigra compacta and reticulata, peri-aqueductal GM, oculomotor, vestibular and cerebellar dentate nuclei and the superior colliculi. Cortical involvement, particularly superior frontal areas, is present but to a lesser extent. Taupositive neurofibrillary tangles are found as neuronal inclusions. Conventional MRI in PSP patients may show mid brain atrophy and associated dilatation of the third ventricle. Davie et al. (1997) carried out single-voxel MRS localized to the lentiform nucleus in nine patients with a clinical diagnosis of PSP and in eight healthy age-matched controls. One of these patients subsequently had pathologically proven PSP at postmortem. The PSP group showed a significant reduction in the absolute concentration of NAA by the order of 50% compared with the control group. The NAA concentration was significantly reduced in seven of the nine patients studied. The authors felt that the reduction of the NAA/Cr ratio from the lentiform nucleus in the PSP group may reflect neuronal loss, occurring predominantly in the globus pallidus. Federico et al. (1999) also showed a reduction in NAA/Cr from lentiform nucleus. Abe et al. (2000) studied 12 patients with a clinical diagnosis of PSP. Compared to normal controls, patients with PSP, but not IPD, had significant reduction of the NAA/Cr ratio in the frontal cortex, whereas patients with PSP and IPD had significant reduction of the NAA/Cr ratio in the putamen. Patients with PSP showed a significant reduction of the NAA/Cr ratio in the putamen as compared with patients with PD and MSA. Tedeschi and colleagues (1997b) carried out MRSI in groups of patients with IPD, PSP and CBD. In the

12 PSP patients studied, they were able to show specific metabolite changes with a reduction of NAA/Cr observed in this group from brainstem, centrum semi-ovale, frontal lobe and precentral cortex. There was a reduction of NAA/Cho in the PSP patients from the lentiform nucleus. These findings are particularly interesting since the regions of metabolic abnormality correlate well with the brain regions where one would expect the most prominent neuropathological abnormality (Figure 35.1). Brain and muscle energy metabolism was also assessed in vivo in five patients with PSP using 31 P MRS (Martinelli et al., 2000). This disclosed a reduced PCr and an increased calculated free adenosine diphosphate (ADP) in the occipital lobes of all patients. Furthermore, Pi was significantly increased and Mg2+ was reduced. In the gastrocnemius muscle, Pi at rest was increased in four patients, and the three patients who were able to perform an incremental exercise showed a rate of PCr post-exercise recovery slower than control subjects. These findings imply widespread multisystem deficit of energy metabolism in PSP, and suggest that such a deficit may play a role in the pathogenesis of the disorder. Cortico Basal Degeneration (CBD) CBD is an uncommon neurodegenerative disorder which is strikingly asymmetrical in it’s clinical presentation. Typically the patient becomes aware of a clumsy stiff limb. Other features of CBD include limb myoclonus that may be stimulus sensitive, asymmetric postural limb tremor and intense limb muscular aching which is far more severe than that seen in IPD. Dementia is usually a late feature of CBD, but an early frontal dementia mimicking Pick’s disease can occasionally be seen. This consists of asymmetrical degeneration of posterior frontal, inferior parietal and superior temporal cortices, the thalamus, substantia nigra and cerebellar dentate nuclei. Swollen, achromatic, tau-positive neurons (Pick cells) are characteristic histological findings in the absence of argyrophilic Pick bodies. Conventional MRI may show asymmetric atrophy in the cerebral henispheres. The MRSI study by Tedeschi et al. (1997) assessed a group of nine CBD patients. They observed significant reduction in

MR spectroscopy in neurodegeneration

BS Cho

TC

NAA

TC BS

Cr

TH

LN

OC

PC

FC CD

LN

TH

CSO PC OC

PCC

PCC

Fig. 35.1 Location of the regions of interest and corresponding representative spectra from one control subject. Regions of interest are the brainstem (BS), caudate (CD), lentiform nucleus (LN), thalamus (TH), centrum semi-ovale (CSO) and the frontal (FC), parietal (PC), precentral (PCC), temporal (TC) and occipital cortices (OC).

NAA/Cho from the centrum semi-ovale, lentiform nucleus and parietal lobe. The most striking reduction was in the lentiform nucleus where the reduction was in the region of 30%. This again correlates well with the areas most affected neuropathologically in this condition.

The study by Abe et al. using a single voxel technique from frontal cortex and putamen studied 19 patients with a clinical diagnosis of CBD. Patients with CBD had significant reduction of the NAA/Cr ratio in the frontal cortex and putamen compared to those with PD, MSA and vascular parkinsonism (Figure 35.2).

599

600

C.A. Davie

MRI

NA

Cho

Cr

(a)

(b)

Fig. 35.2 MRI and 1H MRSI of a control subject (a) and a PSP patient (b) at the level of the basal ganglia. The MRI slice is 3 mm thick, and corrresponds to the center of the 15 mm thick 1H MRSI slice. 1H MRSI data are displayed at their nominal voxel in plane resolution (7.5 7.5 mm), using a color scale which depicts the strongest signal integral with red and the weakest with dark blue. The 1H MRSI images are scaled to the highest value of metabolite signal intensities for each 1H MRSI slice, so that the regional distribution of metabolite signal intensities can be expressed within the same slice, although color images from the same anatomical location cannot be compared in different subjects or in different slices. Figures 35.1 and 35.2 taken from: Tedeschi G, Litvan I, Bonavita S, Bertolino, A, Lundbom N, Patronas NJ, Hallett M. 1997a. Proton MRSI in PSP, Parkinson’s disease and CBD. Reproduced with kind permission from Oxford University Press and Professor M. Hallet.

Huntingtons disease (HD) HD is an autosomal-dominant neurodegenerative disorder caused by a CAG polyglutamine repeat expansion in exon 1 of the HD gene. The protein huntingtin expressed by this gene shows a widespread distribution. The gene defect in HD results in an impairment of energy metabolism within the striatum. However, the mechanisms that underlie selective neuronal cell death and dysfunction in HD and other neurodegenerative illnesses remain poorly understood. A number of experimental animal models for HD exist. The molecular pathways that underlie selective neuronal dysfunction and cell death in HD remain poorly understood. There is evidence to indicate

that the HD gene defect may result in an impairment of energy metabolism within the striatum. Mitochondrial dysfunction has been demonstrated in the brains of HD patients. Other mechanisms, which may also be relevant, include the level of mutant Huntington expression, chronic excitotoxicity of the glutamatergic cortical projection neurons or interaction of Huntington with specific intracellular proteins either in the cytoplasm or nucleus. Whether some or all of these mechanisms are causative or occur as a consequence of the disease remains unanswered. Conventional MRI may reveal atrophy of the head of caudate at a relatively early stage in the disease though it is an insensitive diagnostic finding. As the

MR spectroscopy in neurodegeneration

condition progresses, cerebral atrophy may develop and some patients develop striatal hypointensity on T2-weighted MR images probably as a result of iron deposition. MRS has been used as an in-vivo tool to assess disease mechanisms and the effect of treatment in patients with HD. (Jenkins et al., 1993, 1998; Davie et al., 1994; Koroshetz et al., 1997). In the first MRS study in HD patients, Jenkins and colleagues (Jenkins et al., 1993) carried out localized proton NMR spectroscopy in 18 patients at high risk for, or suffering from, HD as compared with normal controls. Spectra were collected from the basal ganglia and occipital cortex using long-TE MRS and quantitated using ratios. Lac concentrations were increased in the occipital cortex of symptomatic HD patients by approximately 300% when compared with normal controls, and the Lac level correlated with duration of illness. In addition, several patients showed highly elevated Lac levels in the basal ganglia. Basal ganglia levels of NAA were lowered and Cho dramatically elevated relative to Cr by approximately 50%. In a subsequent study by the same group (Koroshetz et al., 1997), oral treatment with coenzyme Q10, an essential cofactor of the electron transport chain, resulted in significant decreases in cortical Lac concentrations in 18 patients with HD, which reversed following withdrawal of treatment. The same group (Jenkins et al., 1998) showed an elevation of Lac signal from striatum in a few presymptomatic carriers of the HD gene. This was not observed from occipital cortex. In patients with symptomatic HD there was a decrease in NAA and increased Lac in the striatum which correlated with duration of symptoms. 31P MRS demonstrated a significant decrease in the PCr to Pi ratio in resting muscle of eight patients as compared with eight control subjects. The cerebrospinal fluid (CSF) Lac–pyruvate ratio was significantly increased in 15 patients as compared with 13 control subjects. The presence of Lac from a localized volume of frontal cortex was reported by (Harms et al., 1997) in just under half a group of symptomatic HD patients. Lac was also observed in four asymptomatic gene carriers. A reduction of NAA/Cho was observed from the symptomatic patients but not the asymptomatic group.

Other groups have failed to observe elevated Lac with 1H MRS in HD patients (Davie et al., 1994; Taylor-Robinson et al., 1996). The study by TaylorRobinson et al. (1996) showed an elevation from the striatum of the combined Gln and Glu signal relative to Cr. An asymptomatic gene carrier showed no metabolic abnormalities. The authors felt that these findings supported disordered striatal Glu metabolism in HD supporting the theory of Glu excitotoxicity in HD. However, MRS studies in transgenic mice with the HD gene (Jenkins et al., 2000) suggest that it may have been an elevation of Gln rather than Gln that the authors observed. A study by Hoang et al. (1998) in 15 patients with HD performed 1H MRS from single volumes localized to putamen, occipital GM, and posterior parietal WM using absolute metabolite concentrations. Quantitative phosphorus and proton-decoupled phosphorus MRS of superior biparietal WM and GM was also performed. This study failed to show any statistically significant difference between the HD patients and controls. Energy metabolism was normal in all brain regions measured by MRS. No increase in cerebral Lac or decrease in PCr and ATP was detected. Small, systematic abnormalities in NAA (decreased), Cr (decreased), Cho (increased) and mI (increased) were demonstrable in all patient’s individually and in summed spectra, but were insufficient to make diagnosis possible in the individual patient. In the very advanced stages, HD patients often enter into an akinetic rigid state rather than the marked choreoform movements which are seen earlier in the disease. A study by Sanchez-Pernaute and colleagues (1999) looked at six longstanding akinetic patients, four presymptomatic patients with the HD gene and five age-matched controls. Localized, single-voxel MRS was performed in the basal ganglia. This study failed to show elevation of Lac signal from either of the two HD patient groups. NAA and Cr were reduced markedly in both groups of patients, particularly in the advanced akinetic group. The decrease was also significant in presymptomatic patients in whom motor and cognitive performances were within the normal range. The authors felt that the Cr signal may be a useful marker for progression in HD and could be useful in assessing

601

602

C.A. Davie

therapeutic outcome, particularly during the initial stages when most clinical indices are still within the normal range. A 31P MRS study (Lodi et al., 2000) has been undertaken from skeletal muscle in patients with HD and has shown a deficit of in vivo mitochondrial oxidative metabolism, supporting a role for mitochondrial dysfunction as a factor involved in the pathogenesis of the disease. Significant abnormality was also detected in presymptomatic carriers.

Conclusions The ability to perform in vivo longitudinal studies from several brain regions and to quantitate metabolite changes provides exciting opportunities for research and as a surrogate tool in assessing putative treatments in many neurodegenerative disorders. It is however disappointing that there has been little consistency in the above studies performed to date. There are a number of reasons why this is likely to have occurred. Perhaps most relevant is the very few patient numbers employed in most studies. In a number of rare disorders (e.g. SRO and CBD) this is perhaps justifiable. However, with multi-center collaboration across countries this can be overcome. Clarke and Lowry (2001) has calculated a need for a minimum number of 29 patients per group in the study of parkinsonian disorders. A corollary of collaborative studies is standardization of spectroscopic procedures between centers and countries. The use of varying pulse sequences, TE and repetition times (TR), a failure to obtain absolute metabolic quantitation and a lack of standardization of brain regions assessed for any given disease make comparison between groups impossible. To date of all the movement disorders only one large multicenter study has been conducted, in patients with IPD (Holshouser et al., 1995). There is also a need to obtain clinico-pathological correlation in future studies particularly in conditions where diagnostic sensitivity and specificity are poor. Most of the studies so far, have concentrated on metabolites such as NAA, Cr, Cho and Lac which are easier to measure using standard sequences at standard clinical field strengths. With the introduction

of higher field strengths up to 4.7 T in humans and with specific editing techniques, there should be a greater emphasis on detecting changes in the absolute concentrations of other NMR metabolites including GABA, Gln and Glu. Furthermore, there is interest in the use of mI as a possible surrogate marker of gliosis. This has yet to be tested in the conditions discussed above. At the present time, 1H MRS and 1H MRSI have several limitations. To obtain a good SNR, the exam duration is still long and may be compromised by patient movement. Furthermore, some brain regions such as the cerebellum and temporal lobes which are of pathological interest, are difficult to assess due to magnetic field inhomogeneities which can influence the quality of spectra. Methodological improvements in localized shimming will allow more reproducible studies from these brain regions in the future. 1 H MRS and 1H MRSI data are still frequently measured as metabolite ratios, which decreases the sensitivity of the technique when two or more metabolites change simultaneously in the same direction. For this reason, absolute quantification techniques should be routinely employed though ratio data can supplement this where for example there is a question of partial volume effects (PVE) from surrounding CSF. However, despite the limitations of many of the current studies, there is still great potential for this technique to improve our understanding of mechanisms of neurodegeneration and to allow in vivo monitoring of the effects of treatment on disease progression. In conclusion, although MRS has yet to find a clinical role in neurodegenerative disorders, it continues to provide a non-invasive tool for the in vivo study of brain metabolism in this group of diseases.

REFERENCES Abe K, Terakawa H, Takanashi M, Watanabe Y, Tanaka H, Fujita N, Hirabuki N, Yanagihara T. 2000. Proton magnetic resonance spectroscopy of patients with parkinsonism. Brain Res Bull 52(6): 589–595. Bates TE, Strangward M, Keelan J, Davey GP, Munro PM, Clark JB. 1996. Inhibition of N-acetylaspartate production: implications for 1H MRS studies in vivo. Neuroreport 31;7(8): 1397–1400.

MR spectroscopy in neurodegeneration

Bhakoo KK, Pearce D. 2000. In vitro expression of N-acetyl aspartate by oligodendrocytes: implications for proton magnetic resonance spectroscopy signal in vivo. J Neurochem 74(1): 254–262. Birken DL, Oldendorf WH. 1989. N-acetyl-L-aspartic acid: a literature review of a compound prominent in 1H-NMR spectroscopic studies of brain. Neurosci Biobehav Rev 13(1): 23–31. Review. Clarke CE, Lowry M. 2000. Basal ganglia metabolite concentrations in idiopathic Parkinson’s disease and multiple system atrophy measured by proton magnetic resonance spectroscopy. Eur J Neurol 7(6): 661–665. Clarke CE, Lowry M. 2001. Systematic review of proton magnetic resonance spectroscopy of the striatum in parkinsonian syndromes. Eur J Neurol 8(6): 573–577. Clarke CE, Lowry M, Horsman A. 1997. Unchanged basal ganglia N-acetylaspartate and glutamate in idiopathic Parkinson’s disease measured by proton magnetic resonance spectroscopy. Movement Disord 12: 297–301. Davie CA, Barker GJ, Quinn N, Tofts PS, Miller DH. 1994. Proton MRS in Huntington’s disease. Lancet 18; 343(8912): 1580. Davie CA, Barker GJ, Machado C, Miller DH, Lees AJ. 1997. Proton magnetic resonance spectroscopy in Steele– Richardson–Olszewski syndrome. Movement Disord 12(5): 767–771. Davie CA, Wenning GK, Barker GJ, Tofts PS, Kendall BE, Quinn N, McDonald WI, Marsden CD, Miller DH. 1995a. Differentiation of multiple system atrophy from idiopathic Parkinson’s disease using proton magnetic resonance spectroscopy) Ann Neurol 37(2): 204–210. Davie CA, Wenning GK, Barker GJ, et al. 1995b. Differentiation of multiple system atrophy from idiopathic Parkinson’s disease using proton magnetic resonance spectroscopy. Ann Neurol 37: 204–210. De Stefano N, Matthews PM, Arnold DL. 1995. Reversible decreases in N-acetylaspartate after acute brain injury. Magn Reson Med 34(5): 721–727. Ellis C, Lemmens G, Williams SCR, Simmons A, Leigh PN, Chaudhuri KR. 1996. Striatal changes in striatonigral degeneration and Parkinson’s disease: a proton magnetic resonance spectroscopy study. Movement Disord 11: 104–104. Ellis CM, Lemmens G, Williams SC, et al. 1997. Changes in putamen N-acetylaspartate and choline ratios in untreated and levodopa-treated Parkinson’s disease: a proton magnetic resonance spectroscopy study. Neurology 49: 438–444. Federico F, Simone IL, Lucivero V, Mezzapesa DM, Mari Md Lamberti P, Petruzzellis M. 1999. Usefulness of proton magnetic resonance spectroscopy in differentiating parkinsonian syndromes. Ital J Neurol Sci 20: 223–229. Harms L, Meierkord H, Timm G, Pfeiffer L, Ludolph AC. 1997. Decreased N-acetyl-aspartate/choline ratio and increased

lactate in the frontal lobe of patients with Huntington’s disease: a proton magnetic resonance spectroscopy study. J Neurol Neurosurg Psychiatr 62(1): 27–30 Hoang TQ, Bluml S, Dubowitz DJ, Moats R, Kopyov O, Jacques D, Ross BD. 1998. Quantitative proton-decoupled 31P MRS and 1H MRS in the evaluation of Huntington’s and Parkinson’s diseases. Neurology 50(4): 1033–1040. Holshouser BA, Komu M, Moller HE, et al. 1995. Localized proton NMR spectroscopy in the striatum of patients with idiopathic Parkinson’s disease: a multicenter pilot study. Magn Reson Med 33: 589–594. Hu MT, Taylor-Robinson SD, Chaudhuri KR, Bell JD, Morris RG, Clough C, Brooks DJ, Turjanski N. 1999. Evidence for cortical dysfunction in clinically non-demented patients with Parkinson’s disease: a proton MR spectroscopy study. J Neurol Neurosurg Psychiatr 67(1): 20–26. Hu MT, Taylor-Robinson SD, Chaudhuri KR, Bell JD, Labbe C, Cunningham VJ, Koepp MJ, Hammers A, Morris RG, Turjanski N, Brooks DJ. 2000. Cortical dysfunction in nondemented Parkinson’s disease patients: a combined (31)P-MRS and (18)FDG-PET study. Brain 123 (Pt 2): 340–352. Hughes AJ, Daniel SE, Kilford L, Lees AJ. 1992. Accuracy of clinical diagnosis of idiopathic Parkinson’s disease: a clinicopathological study of 100 cases. J Neurol Neurosurg Psychiatr 55(3): 181–184. Hughes AJ, Daniel SE, Ben-Shlomo Y, Lees AJ. 2002. The accuracy of diagnosis of parkinsonian syndromes in a specialist movement disorder service. Brain 125(Pt 4): 861–870. Jenkins BG, Koroshetz WJ, Beal MF, Rosen BR. 1993. Evidence for impairment of energy metabolism in vivo in Huntington’s disease using localized 1H NMR spectroscopy. Neurology 43: 2689–2695. Jenkins BG, Klivenyi P, Kustermann E, Andreassen OA, Ferrante RJ, Rosen BR, Beal MF. 2000. Nonlinear decrease over time in N-acetyl aspartate levels in the absence of neuronal loss and increases in glutamine and glucose in transgenic Huntington’s disease mice. J Neurochem 74(5): 2108–2119. Jenkins BG., Rosas HD, Chen YC, Makabe T, Myers R, Macdonald M, et al. 1998. 1H NMR spectroscopy studies of Huntington’s disease: correlations with CAG repeat numbers. Neurology 50: 1357–1365. Kalra S, Cashman NR, Genge A, Arnold DL. 1998. Recovery of N-acetylaspartate in corticomotor neurons of patients with ALS after riluzole therapy. Neuroreport 9(8): 1757–1761. Koroshetz WJ, Jenkins BG, Rosen BR, Beal MF. 1997. Energy metabolism defects in Huntington’s disease and effects of coenzyme Q10. Ann Neurol 41(2): 160–165. Lodi R, Schapira AH, Manners D, Styles P, Wood NW, Taylor DJ, Warner TT. 2000. Abnormal in vivo skeletal muscle energy metabolism in Huntington’s disease and dentatorubropallidoluysian atrophy. Ann Neurol 48(1): 72–76.

603

604

C.A. Davie

Martinelli P, Scaglione C, Lodi R, Iotti S, Barbiroli B. 2000. Deficit of brain and skeletal muscle bioenergetics in progressive supranuclear palsy shown in vivo by phosphorus magnetic resonance spectroscopy. Movement Disord 15(5): 889–893. O’Neill J, Schuff N, Marks Jr WJ, Feiwell R, Aminoff MJ, Weiner MW. 2002. Quantitative 1H magnetic resonance spectroscopy and MRI of Parkinson’s disease. Movement Disord 17(5): 917–927. Rudkin TM, Arnold DM. 1999. Proton magnetic resonance spectroscopy for the diagnosis and management of cerebral disorders. Arch Neurol 56(8): 919–926. Review. Sanchez-Pernaute R, Garcia-Segura JM, del Barrio Alba A, Viano J, de Yebenes JG. 1999. Clinical correlation of striatal 1H MRS changes in Huntington’s disease. Neurology 11;53(4): 806–812. Scolding N, Franklin R, Stevens S, Heldin CH, Compston A, Newcombe J. 1998. Oligodendrocyte progenitors are present in the normal adult human CNS and in the lesions of multiple sclerosis. Brain 121 (Pt 12): 2221–2228. Taylor-Robinson SD, Weeks RA, Bryant DJ, Sargentoni J, Marcus CD, Harding AE, Brooks DJ. 1996. Proton magnetic resonance spectroscopy in Huntington’s disease: evidence in favour of the glutamate excitotoxic theory. Movement Disord 11(2): 167–173.

Tedeschi G, Litvan I, Bonavita S, Bertolino A, Lundbom N, Patronas NJ, Hallett M. 1997a. Proton magnetic resonance spectroscopic imaging in progressive supranuclear palsy, Parkinson’s disease and corticobasal degeneration. Brain 120 (Pt 9): 1541–1552. Tedeschi G, Litvan I, Bonavita S, et al. 1997b. Proton magnetic resonance spectroscopic imaging in progressive supranuclear palsy, Parkinson’s disease and corticobasal degeneration. Brain 120: 1541–1552. Turjanski N, Bhattacharya S, Seery JP, et al. 1997. Subclinical cortical dysfunction in Parkinson’s disease: a proton magnetic resonance spectroscopy study. Movement Disord 12: 6060. Urenjak J, Williams SR, Gadian DG, Noble M. 1993. Proton nuclear magnetic resonance spectroscopy unambiguously identifies different neural cell types. J Neurosci 13(3): 981–989. Vion-Dury J, Meyerhoff DJ, Cozzone PJ, Weiner MW. 1994. What might be the impact on neurology of the analysis of brain metabolism by in vivo magnetic resonance spectroscopy? J Neurol 241(6): 354–371.

MR spectroscopy in neurodegeneration

Case Study 35.1 DTI in primary lateral sclerosis Aziz M.Ulug, Ph.D., M. Flint Beal MD., Robert D. Zimmerman MD. Weill Medical College of Cornell University, New York, USA History 76-year-old female presenting with mild weakness in the left leg and arm, hyperreflexia and bilateral Babinski signs.

T2w

Dav

Technique Conventional MRI and MR diffusion tensor imaging (DTI).

Imaging findings

Anisotropy map

Conventional MR images are normal. DWI and average diffusion constant map are unremarkable. Diffusion anisotropy map shows marked hypointensity at the right posterior limb of the internal capsule (PLIC) which contains the corticospinal tract. When compared to normals, the anisotropy measured from PLIC is decreased bilaterally. The decrease is more prominent at the right PLIC compared to the left PLIC in agreement with the symptoms of motor deficit of left extremities.

Anisotropy in false color Discussion Conventional imaging in this case was completely normal and did not show any evidence of upper motor neuron disease. Quantitative diffusion imaging stained the pathology at the right PLIC in agreement with clinical symptoms.

Key points Conventional MRI is not remarkable in Primary lateral Sclerosis (PLS). DTI shows upper motor neuron damage. Reference Ulug AM, Grunewald T, Lin MT, Kamal AK, Filippi CG, Zimmerman RD, Beal MF. 2004. Diffusion tensor imaging in the diagnosis of primary lateral sclerosis. J Magn Reson Imaging 19: 34–39.

605

Section 7 Trauma

36

Potential role of MR spectroscopy, diffusion-weighted/ diffusion-tensor imaging and perfusion-weighted imaging in traumatic brain injury: overview John D. Pickard Academic Department of Neurosurgery, Addenbrooke’s Hospital, Cambridge, UK

Head injury is one of the major causes of death and disability at all ages (Jennett and Teasdale, 1981). Considerable success has been achieved with its prevention and management. Legislation for seat belts, air bags, crash helmets, drink driving, gun control and safer playgrounds have reduced significantly its incidence. Rigorous application of management guidelines for the resuscitation and early management of the head-injured patient, starting at the roadside, has resulted in a significant reduction in mortality and morbidity (Becker et al., 1977; www.nice.org.uk). The advanced trauma and life support (ATLS) and recently published British National Institute of Clinical Excellence (NICE) guidelines seek to prevent and treat such avoidable secondary insults as hypoxia, hypotension, fits and raised intracranial pressure (ICP) due to intracranial hematomas, contusions and brain swelling. Following resuscitation, early computed tomography (CT) will identify intracranial masses, unilateral hydrocephalus due to brain shift and diffuse swelling but is very insensitive at detecting diffuse axonal injury (Gean, 1994). Following appropriate surgery to remove any hematoma, the patient should be admitted to a Neuroscience or Intensive Care Unit. Prognosis for head-injured patients admitted to a Specialist Neuroscience Intensive Care Unit is significantly better than if admitted to a General Intensive Care Unit. Both systemic (blood pressure, arterial blood gases, etc.) and intracranial parameters (ICP and brain oxygen tension) are continuously

monitored. Where ICP rises, the CT is repeated to exclude any new or progressive lesions. Algorithms have now been developed to guide intensive care management (Figure 36.1). If conventional care is so straightforward, what is the potential role of MR imaging (MRI) after traumatic brain injury (TBI), particularly given that patients on ventilators pose considerable logistical problems for the MR team?

Initial severity and prognosis Traditionally, the Glasgow Coma Scale (GCS) (Teasdale and Jennett, 1974) on admission combined with the early CT appearance (Marshall Score) has been used to indicate the severity of injury. The duration of post-traumatic amnesia (Russell, 1971) has been used retrospectively to assess severity. However, with early intubation and ventilation, the GCS is no longer a reliable indicator of severity (Balestrerim et al., 2004). CT is poor at defining diffuse axonal injury as well as the sort of diffuse damage that we suppose to be the basis for late cognitive and behavioral disability (Gean, 1994). As discussed in the succeeding chapters, diffusionweighted imaging (DWI) and proton spectroscopy may have considerable potential to detect diffuse axonal injury early but larger validation studies are required.

609

610

John D. Pickard

Addenbrooke's NCCU: ICP/CPD management algorithm All patients with or at risk of intracranial hypertension must have invasive arterial monitoring, central venous pressure (CVP) line, ICP monitor and Rt SjvO2 catheter at admission to Neurosciences critical care unit (NCCU). • • • •

This algorithm should be used in conjunction with the full protocols for patient management. Aim to establish multimodality monitoring within the first 6h of NCCU stay. Interventions in stage III to be targeted to clinical picture and multimodality monitoring. Cerebral perfusion pressure (CPP) 70 mmHg set as a target, although CPP  60 mmHg is often acceptable. • Check recruitment to research studies. Evacuate significant space occupying lesions (SOLs) and drain CSF before escalating medical Rx. Rx in italics and Grades IV and V only after approval by NCCU consultant.

(I)

• • • • •

10–15° head up, no venous obstruction CPP  70 mmHg (CVP 6–10 mmHg;  pulmonary arterial catheter (PAC) SpO2  97%; PaO2  11kPa, PaCO2 4.5–5.0kPa Temperature  37°C; SjO2  55%; blood sugar 4–7 mmol/l Propofol 2–5 mg/kg/h; fentanyl 1–2g/kg/h; atracurium 0.5mg/kg/h (consider indications for midazolam and remifentanil) • Ranitidine 50 mg 8º intra venous (i.v.) (or sucralfate 1 g 6º Nasogastric (NG) if enteral access) • Phenytoin 15 mg/kg if indicated (fits, depressed, etc.) (II)

ICP  20 CPP  60

No

Drain CSF via external ventricular drain (EVD) if possible and evacuate significant SOLs

Yes (III)

5% NaCl 2 ml/kg (repeat if Na  155mmol/l, Posm  320) 20% mannitol 2 ml/kg 3 or till plasma 320mosm/l PAC, volume, vasoactives: trial of ↑↑ CPP (70mmHg) Temperature ⯝ 35°C, daily lipid screen if still on propofol Electroencephalograms (EEG): ? fits → institute or escalate antiepileptic therapy • Reduce PaCO2 to ~4.0 kPa providing SjO2 stays 55% • Consider 0.3 M THAM 1–2 ml/kg if chronically ↓ PaCO2

• • • • •

(IV)

Yes

CPP  60; ICP  25 (check probe, ? RE-CT)

• Recent CT? • Low risk of new SOL? No No

CT SOL?

Yes – evacuate

Temp 33 °C (discontinue propofol) (V)

CPP  60; ICP  25 (check probe, ? RE-CT)

Try i.v. anesthetic (e.g. propofol 1mg/kg), maintain CPP (fluids and vasoactives). If ICP and CPP improve start thio (250mg boluses up to 3–5 g, then 4–8 mg/kg/h to maintain burst suppression). Monitor EEG if available.

Fig. 36.1 Addenbrooke’s NCCU: ICP/CPP management algorithm.

Consider decompressive craniectomy as an alternative to medical therapy for uncontrolled intracranial hypertension

Potential role of MRS, DEI/DTI and PWI in traumatic brain injury: overview

Evolving pathophysiology Brain swelling may be the result of cerebral ischemia (intracellular swelling), cytotoxic edema (e.g. astroglial swelling due to toxic accumulation of glutamate (Glu) and potassium and exacerbated by ischemia), vasogenic edema (associated with delayed changes in blood–brain barrier function) and vascular engorgement (associated with vascular dysautoregulation) (Reilly and Bullock, 1997). The combination of these processes may lead to microvascular injury and collapse with a resulting increase in diffusion distances for oxygen from the capillaries to the neurons (Menon et al., 2004). Monitors may give either global (Pickard et al., 2003) (ICP, jugular venous oxygen) or focal information (brain tissue oxygen, microdialysis) (Hutchinson et al., 2002; Kett-White et al., 2002). The pathophysiological processes following trauma may display enormous regional heterogeneity so that a global monitor may not always be sufficient and may miss key focal insults. Focal monitors must be very accurately placed with regard to the pathology and may not always provide useful information about global insults (Steiner et al., 2003). Imaging will reveal regional heterogeneity but provide only a snapshot that must be placed in context by continuous multimodality bedside monitoring. For example, brain swelling can be monitored by an ICP transducer but there is growing evidence that the placement of such a transducer may give different values depending on the development of pressure gradients. Combined arterial blood pressure and ICP waveform analysis may be used to indicate the state of global autoregulation of cerebral blood flow (CBF) (Czosnyka et al., 1996). However, ICP cannot be used to distinguish between various pathophysiological processes such as cytotoxic and vasogenic edema as they evolve at different stages after the head injury. Clearly treatment for one type of swelling may differ from that for another – for example, mannitol may prove very dangerous for vascular engorgement. Xenon CT and single photon emission completed tomography (SPECT) can be used to provide regional maps of cerebral blood perfusion but recent positron emission tomography (PET) studies have confirmed that it is essential to monitor both CBF

and metabolism if any changes in CBF are to be interpretable. A mismatch between CBF and metabolism may be reflected in brain lactate (Lac) and diffusionweighted/perfusion-weighted MRI. More work needs to be done by cross validation with PET and microdialysis in a much larger series of patients followed over time. Reliable signatures for each type of swelling must be developed that are reproducible, quantifiable and define regional heterogeneity. Randomized control trials of Glu antagonists have failed largely because it is not yet possible to define the homogeneous subgroups of patients who might be supposed to benefit (Narayan et al., 2002). Many head injury patients do not have elevated Glu levels and hence any potential benefit may be masked by the noise of the non-responders and by any side effects. It is too early yet to say in what way MR spectroscopy and functional MR will help resolve this conundrum.

Late sequelae Head injury, even the relatively minor, may be followed by devastating late behavioral and cognitive disability (Russell, 1971; Brooks, 1984). It is supposed that this is the result of stereotyped injuries to the inferior frontal lobes and anterior temporal lobes combined with diffuse axonal injury elsewhere in the brain. Voxel-based MR morphometry, N-acetyl aspartate (NAA) spectroscopy and diffusion-tensor imaging (DTI) are beginning to provide quantitative data in favor of this concept. Early studies found a poor correlation between late atrophy on MR and late neuropsychological defects but recent improvements in MR imaging and image analysis are more promising. For example, children who require prolonged ventilation for raised ICP are more likely to have reduced brain volumes subsequently.

Summary The thoughtful use of MR may prove to be very helpful in the future management of patients following TBI but larger scale, longitudinal studies are required. One good example is the use of DWI/DTI in the diagnosis of non-accidental injury in children, a subject which naturally provokes considerable

611

612

John D. Pickard

emotion and in which objective data is invaluable (Uzcinski, 2002). The Court of Appeal in the United Kingdom has recently made very clear that expert opinion must be supported by objective investigations wherever possible.

REFERENCES Balestrerim M, Czosnyka M, Chatfield DA, et al. 2004. Predictive value of Glasgow Coma Scale after brain trauma: change in trend over the past ten years. J Neurol Neurosurg Psychiatr 75: 161–162. Becker DP, Miller JD, Ward JD, et al. 1977. The outcome from severe head injury with early diagnosis and intensive management. J Neurosurg 47: 491–502. Brooks N 1984. Closed Head Injury – Psychological, Social and Family Consequences. OUP, Oxford. Czosnyka M, Smielewski P, Kirkpatrick P, et al. 1996. Monitoring of cerebral autoregulation in head injured patients. Stroke 27: 829–834. Gean AD 1994. Imaging of Head Trauma. Raven Press, NY. Hutchinson PJ, Gupta AK, Fryer TF, et al. 2002. Correlation between cerebral blood flow, sub stroke delivery and metabolism in head injury: a combined microdialysis and triple oxygen positron emission tomography study. J Cereb Blood Flow Metab 2: 735–745.

Jennett B, Teasdale G 1981. Management of Head Injuries. FA Davies, Philadelphia. Kett-White R, Hutchinson PJ, Czosnyka M, Boniface S, Pickard JD, Kirkpatrick PJ. 2002. Multi-modal monitoring of acute brain injury. Adv Tech Stand Neurosurg 27: 87–134. Menon DK, Coles JP, Gupta AK, et al. 2004. Diffusion limited oxygen delivery following head injury. Crit Care Med 32: 1384–1390. Narayan RK, Michel ME, Ansell V, et al. 2002. Clinical trials in head injury. J Neurotrauma 19: 503–557. Pickard JD, Czosnyka M, Steiner LA. 2003. Raised Intracranial Pressure in Neurological Emergencies. BMJ Books, London, pp. 188–246. Reilly P, Bullock R. 1997. Head Injury – Pathophysiology and Management of Severe Closed Injury. Chapman and Hall Medical, London. Russell WR. 1971. The Traumatic Amnesias. OUP. Steiner LA, Coles JP, Johnston AJ, et al. 2003. Responses to post-traumatic pericontusional cerebral blood flow and blood volume to an increase in cerebral perfusion pressure. J Cereb Blood Flow Metab 23: 1371–1377. Teasdale G, Jennett B. 1974. Coma and impaired consciousness. A practical scale. Lancet 2: 81–84. Uzcinski R. 2002. Shaken baby syndrome: fundamental questions. Br J Neurosurg 16(3): 217–219. www.nice.org.uk

37

MR spectroscopy in traumatic brain injury William M. Brooks Hoglund Brain Imaging Center, Departments of Neurology and Molecular and Integrative Physiology, University of Kansas Medical Center, Kansas City, USA

Key points • Conventional MR and computed tomography findings, although providing essential information acutely, are only weakly related to cognitative outcome. • High lactate levels acutely, particularly in areas with normal conventional imaging findings, are associated with poor outcome. • Longitudinal falls in N-acetyl aspartate may continue for months after the initial insult.

Traumatic brain injury (TBI) affects approximately 2 million people in the US each year, and many of these injuries result in long-term disability (Kraus et al., 1994). Indeed, TBI is the leading cause of death in patients under the age of 45 in the US, with an annual mortality rate of at least 20 per 100,000 (Rice and MacKenzie, 1989; Sosin et al., 1989). The longterm impact of TBI is extreme and even patients with mild or moderate brain injury suffer lingering effects including persistent symptoms, impaired function, and increased medical and social costs (Gualtieri, 1995). In addition, there is the enormous social impact of TBI on communities and families who bear the cost of long-term care and rehabilitation of patients whose outcomes range from good function and social integration to a hopeless and persistent vegetative state (McIntosh et al., 1996). Reliable methods to assess severity of injury and to predict outcome of patients soon after injury are necessary for overall clinical management and evaluation of pharmaceutical interventions. Although

existing clinical assessment tools based on the evolving clinical presentation provide some prediction of general outcome, they are inadequate for predicting cognitive functioning of individual patients. An assessment tool based on an understanding of the fundamental pathological processes affecting the brain may provide new methods for predicting outcome, detecting responders to treatment, and assessing therapeutic intervention. Histological studies post-mortem reveal extensive diffuse axonal damage, i.e. injury to individual neuronal cells, associated with TBI (Gennarelli et al., 1982; Grady et al., 1993; Gentleman et al., 1995; Graham et al., 1995; Povlishock and Christman, 1995). After TBI, reactive cytoskeletal disorganization causes axonal swelling that can lead to disconnection and ultimately cell death (Povlishock and Christman, 1995). Staining studies suggest that axonal swelling may be a near-universal consequence of fatal brain trauma (Gentleman et al., 1995). Moreover, although focal white matter (WM) abnormalities related to shearing injury are commonly seen in the subcortical WM, corpus callosum, and mid-brain on MR images, selective antibody staining for the 68-kDa cytoskeleton subunit shows widespread axonal insults throughout the WM, within the gray–white interface, and within subcortical gray matter (GM) structures (Grady et al., 1993). Specific areas of vulnerability include those where axons change anatomic course, travel around blood vessels, or enter tissue of differing densities (i.e. gray–white interface) (Grady et al., 1993; Povlishock and Christman, 1995). Reactive axonal changes are seen in the majority of tissue samples and in areas 613

614

William M. Brooks

distant from focal findings (i.e. occipital GM) (Grady et al., 1993). Diffuse axonal injury after TBI, as seen in histological findings, is influenced by injury severity and may be the major injury component relating to behavioral impairment. Primate studies show that the number and distribution of damaged axons predict morbidity in primates (Gennarelli et al., 1982). Similarly, human studies report relationships between morbidity and number of injured axons, head velocity on impact, and severity of injury as determined from the Glasgow Coma Scale (GCS) score (Adams et al., 1989; Blumbergs et al., 1994). Accordingly, an assessment of brain injury severity based upon the extent of axonal damage might well provide a useful approach to predicting behavioral outcome and cognitive function in patients who survive TBI. Although the neuroimaging modalities of computed tomography (CT) and conventional MR provide essential anatomic information, such as shearing, hemorrhage, and edema, for acute clinical management, quantitative measurements of lesion severity from these imaging techniques are only weakly related to gross outcome or detailed neuropsychological dysfunction (Waxman et al., 1991; Levin et al., 1992; Bigler 1996). Atrophy can be measured with high-resolution MR imaging (MRI) and CT and is correlated with long-term cognitive status assessed by neuropsychological testing (Bigler 1996; Blatter et al., 1997). However, since atrophy is a relatively slow process it is not a useful clinical measure for predicting outcome. Thus, although conventional anatomic neuroimaging modalities are useful in acute clinical care and to quantify overall brain injury, they are not helpful during the early stages of recovery for predicting cognitive outcome. This limitation likely stems from the fact that the spatial sensitivity and contrast mechanisms of existing structural imaging techniques probe the anatomic disruption of small tissue structures rather than cellular mechanisms of injury. MR spectroscopy (MRS) offers an alternative approach to imaging the brain. The examination of the metabolic status of brain cells makes MRS especially attractive for TBI. In particular, after TBI, a metabolic cascade is unleashed in the brain

resulting in widespread metabolic depression that has been reported in animal studies as well as in positron emission studies in patients (Dail et al., 1981; O’Neill et al., 2000). As a result, and depending on injury severity, some cells will recover while others will ultimately die. The capability of MRS to quantify metabolic concentrations non-invasively, to probe the metabolic status of the neurons and glia during inflammatory responses, and the fact that MRS acquisition capability is commonly available on MRI machines in many hospitals makes it useful for repeated studies in patients as they recover from injury. Despite these obvious strengths, MRS has only been used to a limited extent to study TBI. Moreover, interpretation of existing studies is complicated by the range of clinical and technical conditions under which MRS data have been acquired. For example, design considerations pertinent to patients include the age of patients, the severity of injury, the time between injury and MRS examination, and the outcome measures used to document recovery, e.g. clinical rating scales (Glasgow Outcome Scale, Disability Rating Scale (DRS), Functional Independence Measure (FIM)) vs. fine-grained outcome (intelligence quotient (IQ) or neuropsychological testing). Technique-specific issues include location of the tissue actually sampled in each patient, e.g. frontal vs. occipitoparietal, GM vs. WM, or normal appearing vs. lesional, the acquisition technique employed, e.g. stimulated echo vs. spin echo (SE), short vs. long echo time (TE), single-voxel vs. multi-voxel/spectroscopic imaging (SI), the analysis approach used, e.g. time domain vs. frequency domain and absolute quantification of metabolite vs. ratios of spectroscopic peaks, and finally which specific metabolites are quantified. For a full discussion of the technical factors affecting acquisition and analysis of MRS data, please see Chapters 1–3. The interpretation of individual studies and generalization to clinical practice and rehabilitation is further complicated by the small numbers of patients that have been studied on each occasion. Indeed, to date, few published studies have examined more than 25 subjects, the first criterion for inclusion in the recent discussion of “Early indicators of prognosis in severe TBI” (Chesnut et al., 2000).

MR spectroscopy in traumatic brain injury

Moreover, only two studies describe patients within 1 day of injury (Haseler et al., 1997; Condon et al., 1998), although more subjects have been studied several days post-injury once medical conditions have stabilized (Rango et al., 1990; Sutton et al., 1995; Ashwal et al., 1997; Holshouser et al., 1997, 2000; Cecil et al., 1998; Ross et al., 1998a; Wild et al., 1999; Garnett et al., 2000b; MacMillan et al., 2002), and during the subacute and long-term recovery interval (Brooks et al., 2000; Cadoux-Hudson et al., 1990; Cecil et al., 1998; Choe et al., 1995; Friedman et al., 1998; Ricci et al., 1997; Ross et al., 1998a; Wild et al., 1999). Most studies have focused on examining tissue that appears normal on MRI, i.e. avoiding the focal effects of contusion, hematoma, and obvious shearing injury. In this way, the current spectroscopic data provide information about the cellular injury that is often seen at histology but that is rarely observed by conventional radiological assessment. As detailed below, observations from proton MRS fall into three categories: (1) acute post-injury phase observations of elevated lactate (Lac) suggesting hypoxic injury, (2) evidence of decreased N-acetyl aspartate (NAA) and elevated choline (Cho) suggesting neuronal loss or dysfunction and inflammation which are related to severity of injury, and (3) prediction of behavioral outcome.

Lac Proton MRS offers a unique, non-invasive approach for detecting Lac, the presence of which is generally thought to indicate hypoxia/ischemia. Microdialysis studies show that elevated Lac levels (⬃2 mM), up to 2 days post-injury, are associated with poor outcome (Goodman et al., 1999). Coincidentally, this is just above the sensitivity threshold for detecting Lac by MRS in vivo (⬃0.5–1.0 mM). Accordingly, the presence of Lac in MRS from radiologically normal-appearing tissue should generally be viewed with some concern. This is supported by acute-phase MRS studies that identify Lac in tissue which appears normal by MRI in some patients, many of whom suffer poor outcomes. In studies of radiologically normal-appearing tissue, Holshouser et al. found elevated Lac in eight of 24 pediatric TBI subjects, of whom three died and four

sustained severe disability (Holshouser et al., 1997). In other studies, MRS-visible Lac was also associated with death (Condon et al., 1998) or persistent disability (Haseler et al., 1997; Ross et al., 1998a). Ashwal et al. (2000) found in 26 infants and 27 children that 91% of infants and 80% of children with poor functional and cognitive outcomes had MRS-visible Lac. In a subsequent study, Brenner et al. (2003) found Lac in the majority of those patients who were below average on IQ and neuropsychological performance measured approximately 4 years after injury. Although the majority of patients with detectable Lac in normalappearing tissues are children, two adults demonstrating elevated Lac by MRS died rapidly following injury (Condon et al., 1998). Elevations in large lipid/macromolecule peaks, suggesting neuronal injury, have also been observed in some children with positive Lac findings (Haseler et al., 1997; Holshouser et al., 1997, 2000; Ross et al., 1998a). Several studies reveal marked Lac in focal injury following TBI (Felber et al., 1993; Sutton et al., 1995; Condon et al., 1998). However, in contrast to Lac in normal-appearing tissue, this focal Lac has not been clearly linked to poor outcomes, except when also accompanied by diffusely visible Lac as noted above. The most striking example is one patient detailed in Condon et al. who showed dramatically elevated Lac at initial exam (8 h post-injury) and at 6 days after injury in the contusion locus (Condon et al., 1998). In the contralateral hemisphere however, no Lac was visible at either exam, and a good recovery was reported.

NAA Other MRS abnormalities are seen in patients having elevated Lac in normal-appearing tissues. Reduced NAA (or NAA/creatine (Cr), NAA/Cho) is apparent soon after injury (Haseler et al., 1997; Holshouser et al., 1997, 2000; Ross et al., 1998a) and possibly as early as 8 h after TBI (Condon et al., 1998), indicating neuronal loss or metabolic depression. Subacute (⬃2 weeks to 2 months post-TBI) changes following trauma are also characterized

615

616

William M. Brooks

NAA

Cre Cho

4

3

2 ppm

1

Fig. 37.1 Images and spectra illustrating results from two patients following TBI. As illustrated in the images, MR spectroscopic data were acquired from posterior GM that was normal on MRI. The upper series is from a patient with good outcome following TBI. In this case NAA is normal. The lower series is from a patient with moderate long-term disability and shows reduced NAA. In each case, the spectroscopic data were acquired in about 6 min at 1.5 T.

by marked loss of NAA or NAA/Cr within WM. Specifically, marked decreases have been demonstrated in the corpus callosum, a common site of MRI-visible “shearing” injury (Cecil et al., 1998; Sinson et al., 2001), frontal lobe (Choe et al., 1995; Ricci et al., 1997; Cecil et al., 1998; Garnett et al., 2000a, 2000b), parietal lobe (Holshouser et al., 1997; Friedman et al., 1998, 1999; Ross et al., 1998a; Brooks et al., 2000; Holshouser et al., 2000; Yoon et al., 2000), mid-brain (Zampolini et al., 1997), and throughout a section of axial brain (Wild et al., 1999). Several studies of posterior GM voxels reveal reduced NAA within 6 days (Felber et al., 1993) and at later time points after injury (Holshouser

et al., 1997, 2000; Friedman et al., 1998, 1999; Ross et al., 1998a; Brooks et al., 2000). Figure 37.1 shows spectra recorded from two patients following TBI. In spectra acquired from normal-appearing GM, reduced NAA corresponded to patients with persistent disability. Although much of the interest in TBI has focused on brain changes manifested soon after injury, MRS studies confirm that brain metabolism changes likely evolve over weeks to months from trauma, possibly extending out to several years post-injury. Longitudinal studies carried out at 6 weeks, 3 months, and 6 months post-injury, indicate that NAA in WM is still falling between 6 weeks and

MR spectroscopy in traumatic brain injury

3 months although it possibly recovers subsequently as shown by higher NAA concentrations at 6 months post-injury (Friedman et al., 1999; Brooks et al., 2000). NAA measurements taken in predominantly GM were significantly lower than age-matched controls at both early time points although an apparent subsequent recovery almost reached statistical significance between 3 and 6 months (P  0.06). This result was obtained by using a paired comparison of those patients who were studied at both time points thus reducing the influence of between subject variability (Brooks et al., 2000). Recovering NAA levels are visible in Figure 37.2 which shows spectra recorded from frontal WM at 3 and 6 months after severe injury. Long-term cross-sectional studies of patients up to 6 years after injury indicate that NAA levels may continue to recover for a considerable period and also supports the possibility of extended neuronal recovery (Brief et al., 2000; Sinson et al., 2001). However, there are conflicting reports. Garnett et al. (2000b) studied frontal WM in 21 patients at 12 days and 15 at a mean of just over 6 months after TBI. Over this period, they found that NAA/Cr actually became a little lower. Indeed, in the 10 patients who received both studies facilitating a within-subjects design, they also found a persistent fall of NAA/Cr over time. This conflicting result between Brooks et al. and Garnett et al. may stem from several factors. The Brooks result largely arose from examination of GM, although a similar, but weaker, observation was made in WM. Also Brooks et al. examined posterior rather than frontal tissue where injury severity and recovery profiles might differ, pointing to the possibility that anatomic location might be an important factor in studies of trauma. Finally, the early time point in the Garnett study was made at a mean of 12 days after injury and included some patients as early as 3 days after injury. One study reporting NAA concentrations plotted against time since injury suggests that NAA concentrations might still be falling at about 1 month after TBI (MacMillan et al., 2002). Accordingly, NAA concentrations might still have been decreasing at the time of initial examination by Garnett et al. (2000b). The possibility of recovery of NAA following injury is further supported by studies in experimental injury in rodents. One study using cortical slices

(a) Cho

Cre

NAA

mIns

(b)

4

2

3

1

ppm Fig. 37.2 Spectra, obtained from a SI study carried out at 3 T, illustrating recovery of metabolites following TBI. This 19-year-old male patient sustained a motor cycle accident and was not wearing a protective helmet. At admission to the emergency room and at 24 h his GCS was 3. His coma lasted for 6 days. Spectrum (a) obtained 3 months after injury from the right frontal lobe, shows decreased NAA and substantially elevated Cho and mI from normal appearing white matter (NAWM). Spectrum (b) is from the same location at 6 months after injury – NAA has made a modest recovery and Cho has fallen sharply.

has shown that NAA falls significantly by 2 days and then recovers at 6 days following weight drop injury in rats (Gasparovic et al., 2000). A similar time course has been obtained recently using spatially selective MRS in longitudinal studies of rat (Schuhmann et al., 2003). Thus, timing of examinations might be an important issue in the study and use of MRS in clinical care of TBI. The anatomic variability found in metabolite levels after TBI is illustrated in Figure 37.3. In this SI study,

617

618

William M. Brooks

(a)

Cho B

A

Cre

NAA

mIns

(b)

(c)

4

3

2

1

4

ppm

3

2

1

ppm

Fig. 37.3 Images and spectra following TBI illustrating anatomic inhomogeneity of metabolic response. Data are from the patient described in Figure 37.2. Spectrum (a) obtained 3 months after injury from the right frontal lobe, shows decreased NAA and substantially elevated Cho and mI from NAWM. Spectrum (b) is from the contralateral hemisphere with approximately normal levels of NAA, Cho, and mI. Spectrum (c) is from an uninjured subject of similar age (similar location to spectrum (a)). Arrows in the image indicate the location of spectra A and B. The white rectangle represents the stimulated echo acquisition mode (STEAM) excitation region of interest (ROI) chosen to avoid artifacts from skull lipid signals.

one region of the right hemisphere WM revealed substantially reduced NAA compared with the contralateral locations that were similar to normal levels. SI offers the capability of evaluating metabolites from spectra from different locations throughout the brain. The majority of studies find that NAA falls soon after TBI and there is evidence that some return towards normal is possible. However, the interpretation of these observations is less clear. This is largely due to the fact that the role (or roles) of NAA in the brain has not been unambiguously established. In TBI, several mechanisms that might give rise to altered NAA concentrations are possible. Reduced

NAA can reflect neuronal death as in stroke (Barker et al., 1994) and would be consistent with the neuronal loss that is present in TBI. However, a mechanism of neuronal death cannot account for reversible decreases in NAA. NAA plays a role in myelin synthesis in developing brain (Burri et al., 1991; D’Adamo and Yatsu, 1966) and has been implicated in lipid repair following TBI (Rubin et al., 1997; Cecil et al., 1998; Smith et al., 1998). Studies showing decreased NAA in mitochondria after blocking of oxidative phosphorylation (OXPHOS) suggest that reversible alterations in NAA may reflect mitochondrial impairment followed by functional recovery

MR spectroscopy in traumatic brain injury

(de Stefano et al., 1995; Bates et al., 1996). Indeed, several studies have found a correlation between NAA concentrations and metabolic rate determined from positron emission tomography (PET) imaging suggesting that NAA indicates neuronal number and metabolic activity (Lu et al., 1997; O’Neill et al., 2000). This mechanism is further supported by PET studies of patients following TBI that show reduced glucose metabolism a month after TBI suggesting metabolic abnormality (Bergsneider et al., 2000). Longer-term studies confirm gradual increases in glucose metabolism over 6 months post-injury (Bergsneider et al., 2001). Thus, alterations in NAA may reflect contributions from neuronal death, reversible neuronal metabolic derangement, or membrane repair.

Cho and myo-inositol Cho results are more variable. A number of studies report Cho or Cho/Cr elevations (Holshouser et al., 1997, 2000; Ricci et al., 1997; Zampolini et al., 1997; Ross et al., 1998a; Friedman et al., 1999; Brief et al., 2000; Brooks et al., 2000; Garnett et al., 2000a, 2000b), although several have not found significant differences (Choe et al., 1995; Cecil et al., 1998; Wild et al., 1999). Myo-inositol (mI) was elevated in some studies within mid-brain (Zampolini et al., 1997; Garnett et al., 2000b) and posteriorly (Brooks et al., 2000; Yoon et al., 2000), but not in others (Choe et al., 1995). Some studies found elevated mI in individual patients (e.g. Ross et al., 1998a). Figure 37.3 shows spectra from one patient who sustained a severe head injury that resulted in substantially elevated Cho and mI. Findings on Cho vary with the post-injury time points examined. Initially, Cho has been found to be elevated and then reduced toward normal levels by 6 months, suggestive of resolving injury in the majority of patients. However, individual patients with poorer outcomes demonstrated persistent elevated Cho at 3 and 6 months, often greater than at 1.5 months after injury, suggesting continuing inflammatory processes (Brooks et al., 2000). In contrast, a similar study that compared spectroscopic findings at a mean of 12 days post-injury with those obtained

from the same patients at about 6 months, found that Cho/Cr was still elevated. Another study of patients measured serially by Ross et al. (1998a), also found persistently elevated Cho 9 months after TBI. This study also found sustained elevation of three out of four major metabolites over time, a category termed “hyperosmolar” and suggested to affect 30–40% of pediatric head trauma patients (Lee et al., 1994). However, a study of four patients initially examined during coma but who eventually were re-examined after regaining awareness (Ricci et al., 1997) found that initially high Cho/Cr fell in three patients. Figure 37.2 illustrates the resolving Cho levels observed in one patient studied at 3 months and then again at 6 months after injury. Interpretation of altered Cho and mI is also challenging in TBI. The Cho peak is elevated following tissue breakdown or inflammation (SappeyMarinier et al., 1992; Davie et al., 1994). Elevated mI, a glial osmolyte, suggests glial proliferation (BadarGoffer et al., 1992; Brand et al., 1993), However, in chronic hypernatremia, mI and glycerophosphocholine (GPC) and betaines (two osmolyte components of the Cho peak), are elevated (Lien et al., 1990). Thus, if TBI were associated with necrosis or inflammatory processes, Cho and mI elevation might be prominent. Since diffusion-weighted MRI studies indicate cellular edema after TBI (Barzo et al., 1997) elevated mI and Cho might, in part, result from post-traumatic edema. Finally, the membrane component phosphatidyl inositol, which co-resonates with inositol in the 1.5 T 1H-MR spectrum, may be released during trauma-induced membrane damage, elevating the mI peak by this mechanism (Ross et al., 1998b). Thus, since several pathological processes might be present in TBI, alterations in Cho and mI should be interpreted cautiously.

Severity of injury, outcome and cognitive function Although most studies report altered metabolic status following TBI indicative of cellular injury or dysfunction, many also report correlation between spectroscopic measurements and severity of injury, outcome, and cognitive recovery. One of the clinical

619

620

William M. Brooks

challenges in the management of patients following TBI is to assess the actual severity of the initial injury. This has largely been achieved using the GCS score a five-point scale ranging from normal (GCS  5) to dead (GCS  1) (Teasdale and Jennett, 1974). An alternative clinical measure for severity is the duration of post-traumatic amnesia (PTA; Russell, 1932). Although each of GCS and PTA provide information about the condition of the patient at the time of assessment and they are weakly correlated with long-term outcome and recovery, they are not reliable predictors of outcome for individual patients. Outcome is commonly assessed using the Glasgow Outcome Score (GOS; Jennett and Bond, 1975) or the DRS (Clifton et al., 1992). Cognitive recovery, which is assessed by more finely grained assessment tools such as neuropsychological or intelligence testing, has also been used to quantify outcome from TBI. MRS may provide a sensitive tool to characterize injury and predict behavioral outcome. In a study of 19 head-injured patients imaged between 3 and 38 days of TBI, Garnett et al. (2000a) reported that NAA/Cr was generally lower and Cho/Cr higher in patients following TBI. Further analysis showed that NAA/Cr generally decreased and Cho/Cr increased with injury severity assessed by the GCS or the duration of PTA (Garnett et al., 2000a). Additionally, in a study by Cecil et al. (1998) patients with more severe injuries demonstrated NAA/Cr reductions in both the splenium and lobar WM suggesting widespread neuronal injury, while patients with mild TBI demonstrated primarily callosal findings. Other studies have shown that concentrations of NAA measured 2 weeks to 2 months after injury were correlated with general cognitive function measured at the time of scanning (Friedman et al., 1998). Together, these studies suggest that MRS measures of cellular injury reflect individual severity of injury and may be used to augment common clinical indices for characterizing injury severity. Neurometabolite concentrations obtained soon after injury may also be useful for predicting individual outcome. Although it remains unclear how early it is possible to observe reliable neurometabolite changes by MRS following TBI and, by extension,

the optimal duration from injury for spectroscopic assessment, several studies have demonstrated relationships between normal-appearing tissue NAA and outcome with reduced NAA, NAA/Cr, or NAA/Cho being related to poor recovery. Ricci et al. (1997) showed that NAA/Cho, measured in the left and right frontal cortex, discriminated patients who recovered from coma from those who died or remained in a persistent vegetative state. A similar study using single-voxel MRS to examine the thalamus showed that NAA/Cr discriminated patients who remained in a persistent vegetative state after TBI from those who had been in a vegetative state but who regained awareness. Both groups were significantly different from control (Uzan et al., 2003). Some studies have predicted outcome from spectroscopic analysis of the posterior GM (Holshouser et al., 1997, 2000; Friedman et al., 1999) while others have demonstrated clear group discrimination from WM voxels localized within the frontal lobes (Choe et al., 1995; Ricci et al., 1997). Recent work using the FIM to quantify outcome has also shown relationships between WM NAA and patient function at time of examination, evidence that life-skill measures may be similarly predicted (Yoon et al., 2000). Several studies have focused on predicting fine-grained behavioral functioning assessed by neuropsychological testing following TBI. NAA measured within 2 months of injury and subsequently (Friedman et al., 1998, 1999; Brooks et al., 2000) was strongly correlated with overall neuropsychological function measured 6 months after injury. These results suggest a metabolic or cellular injury basis, detectable by 1H-MRS, to the cognitive impairment seen after TBI. Although these preliminary studies suggest that 1 H-MRS is informative on the likely outcome of individual patients, the optimal anatomic location and time after injury for study is yet to be determined. Most studies to date have typically acquired one or more single voxels from large regions of normalappearing tissue. However, MR spectroscopic imaging (MRSI) techniques which simultaneously sample a large array of voxels across the brain within TBI patients are now available (Wild et al., 1999) (cf. Chapters 1 and 2). Figure 37.4 illustrates the power

MR spectroscopy in traumatic brain injury

2.5 mM Cho

0

10 mM NAA

0 TBI

Uninjured control

Fig. 37.4 Anatomic and SI from a 6-year-old boy who suffered a severe head injury after being knocked from his bicycle by a passing car. The T1-weighted image on the left shows the location of the SI excitation volume (black rectangle). The spectroscopic images to the right show illustrate the altered metabolic status. NAA is generally lower throughout the slice with focal abnormality adjacent to the obvious injury seen on MRI. Cho is elevated throughout the slice. The rightmost column of spectroscopic images are data from an age-matched, uninjured volunteer.

of SI where metabolite maps corresponding to individual metabolites reveal patterns of injury previously undetected by conventional MRI. This multi-voxel approach is widely available on most 1.5 and 3 T scanners being installed, using vendorsupplied acquisition and analysis software.

related to the hemisphere of greater associated spasticity were found (Cadoux-Hudson et al., 1990). Although duration from injury considerations make a comparison of these results challenging, future work will aid in associating human 31P metabolite changes with injury time courses derived from animal investigations.

Phosphorus MR spectroscopy Even fewer clinical phosphorus MR spectroscopy 31 (P-MRS) studies have been performed (CadouxHudson et al., 1990; Rango et al., 1990; Garnett et al., 2000c), reflecting the challenge of studying TBI patients using the less sensitive phosphorus nucleus. Each study examined only small numbers of subjects and conflicting results are reported. Two studies of patients between 1 and 24 days after injury found alkalotic WM, which resolved to normal values by 3 weeks post-TBI (Rango et al., 1990; Garnett et al., 2000c). GM tissue pH was not statistically different from normal. Six to 18 months later, acidotic brain pH and metabolite abnormalities

Future potential: clinical application of MR spectroscopy in TBI MRS is still almost exclusively a research technique and has yet to become clinically accepted in TBI. However, it appears likely that it may in the future be useful for the evaluation of TBI in several ways. As noted above, the presence of MRS-visible Lac in radiologically normal-appearing tissue soon after TBI is a possible indicator of poor outcome, although the significance of Lac in abnormal tissue appears to be less serious. Although there have been no MRS studies of therapy monitoring to date, this would also appear to be a promising area for investigation.

621

622

William M. Brooks

For instance, in stroke study patients, MRS has been used to monitor reductions in Lac concentrations following administration of drug therapy (Graham et al., 2000). In studies of polysubstance abuse, 31 P-MRS has been used to detect altered cerebral highenergy phosphates and phospholipid metabolism changes in subjects undergoing methadone maintenance therapy (Christensen et al., 1996), indicating long-term improvement in neurochemistry (Kaufman et al., 1999). These examples suggest that MRS may be a useful non-invasive tool to monitor cellular response to therapeutic interventions. Most applicable to treatment may be the ability to map the pattern of neuronal integrity with MRS while patients are still in the early stages of recovery. Combining MRI information with a MRSI-based cellular neurochemistry “map” may augment the monitoring of recovery during treatment and, ultimately, suggest interventional strategies. An important observation from work to date is that brain changes persist long beyond the acute injury response. Treatments targeted at this extended window of opportunity might prove helpful in some patients with persistent injurious processes (i.e. as indicated by persistently high Cho). MR spectroscopic quantification of NAA, Cho, and mI offers a non-invasive approach to assessing brain cellular injury and response. The association of these markers with recovery and outcome suggests that they are coupled, either directly or indirectly, with the cellular mechanisms underlying brain injury and offer great potential for clinical management. However, considerable developmental work is still required before MRS can be used as a routine clinical modality. Firstly, the optimal time after injury for studies to provide the most useful information needs to be determined. Indeed multiple studies of individual patients might be required to obtain a patient-specific spatiotemporal profile. Secondly, the value of single-voxel methods compared with MRSI approaches for prognostic purposes is yet to be determined. Thirdly, it is still unclear whether the extra effort required to use shorter echo times to obtain quantifiable data for such metabolites mI and glutamate (Glu) provides a substantial benefit over the technically less challenging and more reliable longer TE acquisitions. These questions can

only be addressed by longitudinal studies that enroll large numbers of patients and that compare the different technical aspects as well as taking account of such clinical factors as GCS, patient age and apolipoprotein status directly. Since MRS can be relatively easily implemented on standard MR scanners, MRS may serve an increasing role in acute patient management and rehabilitation.

REFERENCES Adams JH, Doyle D, Ford I, Graham DI, McLellan D. 1989. Diffuse axonal injury in head injury: definition, diagnosis and grading. Histopathology 15(1): 49–59. Ashwal S, Holshouser BA, Tomasi LG, et al. 1997. 1H-magnetic resonance spectroscopy-determined cerebral lactate and poor neurological outcomes in children with central nervous system disease. Ann Neurol 41(4): 470–481. Ashwal S, Holshouser BA, Shu SK, Simmons PL, Perkin RM, Tomasi LG, Knierim DS, Sheridan C, Craig K, Andrews GH, Hinshaw Jr DB. 2000. Predictive value of proton magnetic resonance spectroscopy in pediatric closed head injury. Pediatr Neurol 23(2): 114–25. Badar-Goffer RS, Ben-Yoseph O, Bachelard HS, Morris PG. 1992. Neuronal-glial metabolism under depolarizing conditions. A 13C-n.m.r. study. Biochem J 282(Pt 1): 225–230. Barker PB, Gillard JH, van Zijl PCM, et al. 1994. Acute stroke: evaluation with serial proton magnetic resonance spectroscopic imaging. Radiology 192(3): 723–732. Barzo P, Marmarou A, Fatouros P, Hayasaki K, Corwin F. 1997. Contribution of vasogenic and cellular edema to traumatic brain swelling measured by diffusion-weighted imaging. J Neurosurg 87(6): 900–907. Bates TE, Strangward M, Keelan J, Davey GP, Munro PMG, Clark JB. 1996. Inhibition of N-acetylaspartate production: implications for 1H-MRS studies in vivo. Neuroreport 7(8): 1397–1400. Bergsneider M, Hovda DA, Lee SM, et al. 2000. Dissociation of cerebral glucose metabolism and level of consciousness during the period of metabolic depression following human traumatic brain injury. J Neurotrauma 17(5): 389–401. Bergsneider M, Hovda DA, McArthur DL, Techepare M, Huang SC, Sehati N, Satz P, Phelps ME, Becker DP. 2001. Metabolic recovery following human traumatic brain injury based on FDG-PET: time course and relationship to neurological disability. J Head Trauma Rehabil 16(2): 135–148. Bigler ED. 1996. Neuroimaging and traumatic brain injury. Neuroimaging II: Clinical applications. Plenum Press, New York, pp. 261–278.

MR spectroscopy in traumatic brain injury

Blatter DD, Bigler ED, Gale SD, Johnson SC, Anderson CV, Burnett BM, Ryser D, Macnamara SE, Bailey BJ. 1997. MR-based brain and cerebrospinal fluid measurement after traumatic brain injury: correlation with neuropsychological outcome. Am J Neuroradiol 18(1): 1–10. Blumbergs PC, Scott G, Manavis J, Wainwright H, Simpson DA, McClean AJ. 1994. Staining of amyloid precursor protein to study axonal damage in mild head injury. Lancet 344(8929): 1055–1056. Brand A, Richter-Landsberg C, Leibfritz D. 1993. Multinuclear NMR studies on the energy metabolism of glial and neuronal cells. Dev Neurosci 15(3–5): 289–298. Brenner T, Freier MC, Holshouser BA, Burley T, Ashwal S. 2003. Predicting neuropsychological outcome after traumatic brain injury in children. Pediatr Neurol 28(2): 104–114. Brief E, Vernon-Wilkinson R, MacKay AL, Scamvougeras A, Feldman H, Forster B. 2000. 1H MRS, MRI and neuropsychological testing of patients with traumatic brain injury at long elapsed time since injury. Proc Intl Reson Med (Abstract) 1138. Brooks WM, Stidley CA, Petropoulos H, Jung RE, Weers DC, Friedman SD, Barlow MA, Sibbitt Jr WL, Yeo RA. 2000. Metabolic and cognitive response to human traumatic brain injury: a quantitative proton magnetic resonance study. J Neurotrauma 17(8): 629–640. Burri R, Steffen C, Herschkowitz N. 1991. N-acetyl-L-aspartate is a major source of acetyl groups for lipid synthesis during rat brain development. Dev Neurosci 13(6): 403–411. Cadoux-Hudson TAD, Wade D, Taylor DJ, Rajogopalan B, Ledingham JGG, Briggs M, Radda GK. 1990. Persistant metabolic sequelae of severe head injury in humans in vivo. Acta Neurochir (Wien) 104(1–2): 1–7. Cecil KM, Hills EC, Sandel E, et al. 1998. Proton magneticresonance spectroscopy for detection of axonal injury in the splenium of the corpus-callosum of brain-injured patients. J Neurosurg 88(5): 795–801. Chesnut RM, Ghajar J, Maas AIR. 2000. Part 2: Early indicators of prognosis in severe traumatic brain injury. J Neurotrauma 17: 557–627. Choe BY, Suh TS, Choi KY, Shinn KS, Park CK, Kang JK. 1995. Neuronal dysfunction in patients with closed head injury evaluated in vivo 1H magnetic resonance spectroscopy. Invest Radiol 30(8): 502–506. Christensen JD, Kaufman MJ, Levin JM, Mendelson JH, Holman BL, Cohen BM, Renshaw PF. 1996. Abnormal cerebral metabolism in polydrug abusers during early withdrawal: a 31P spectroscopy study. Magn Reson Med 35(5): 658–663. Clifton GL, Hayes RL, Levin HS, Michel ME, Choi SC. 1992. Outcome measures for clinical-trials involving traumatically brain-injured patients: report of a conference. Neurosurgery 31(5): 975–978.

Condon B, Oluoch-Olunya D, Hadley D, Teasdale F, Wagstaff A. 1998. Early 1H magnetic resonance spectroscopy of acute head injury: four cases. J Neurotrauma 15(8): 563–571. D’Adamo Jr AF, Yatsu FM. 1966. Acetate metabolism in the nervous system. N-acetyl-L-aspartic acid and the biosynthesis of brain lipids. J Neurochem 13(10): 961–965. Dail WG, Feeney DM, Murray HM, Linn RT, Boyeson MG. 1981. Responses to cortical injury 2: widespread depression of the activity of an enzyme in cortex remote from a focal injury. Brain Res 211(1): 79–89. Davie CA, Hawkins CP, Barker GJ, Brennan A, Tofts PS, Miller DH, McDonald WI. 1994. Serial proton magnetic resonance spectroscopy in acute multiple sclerosis lesions. Brain 117:(Pt 1): 49–58. de Stefano N, Matthews PM, Ford B, Genge A, Karpati G, Arnold DL. 1995. Short-term dichloroacetate treatment improves indices of cerebral metabolism in patients with mitochondrial disorders. Neurology 45(6): 1193–1198. Felber SR, Ettl AR, Birbarner GG, Luz G, Aichner FT. 1993. MR imaging and proton spectroscopy of the brain in posttraumatic cortical blindness. J Magn Reson Imag 3(6): 921–924. Friedman SD, Brooks WM, Jung RE, Hart BL, Yeo RA. 1998. Proton MR spectroscopic findings correspond to neuropsychological function in traumatic brain injury. Am J Neuroradiol 19(10): 1879–1885. Friedman SD, Brooks WM, Jung RE, Chiulli SJ, Sloan JH, Montoya BT, Hart BL, Yeo RA. 1999. Quantitative 1H-MRS predicts outcome following traumatic brain injury. Neurology 52(7): 1384–1391. Garnett MR, Blamire AM, Rajagopalan B, Styles P, CadouxHudson TAD. 2000a. Evidence for cellular damage in normalappearing white matter correlates with injury severity in patients following traumatic brain injury: a magnetic resonance study. Brain 123: 1403–1409. Garnett MR, Blamire AM, Corkill RG, Cadoux-Hudson TAD, Rajagopalan B, Styles P. 2000b. Early proton magnetic resonance spectroscopy in normal-appearing brain correlates with outcome in patients following traumatic brain injury. Brain 123: 2046–2054. Garnett MR, Corkill RG, Cadoux-Hudson TAD, Blamire AM, Rajogopalan B, Young JD, Styles P. 2000c. Altered metabolite profile in normal appearing brain following acute brain injury: a phosphorous MRS study. Proc Intl Reson Med (Abstract) 1139. Gasparovic C, Arfai N, Smid N, Feeney DM. 2000. Decrease and recovery of N-acetylaspartate/creatine in rat brain remote from focal Injury. J Neurotrauma 18(3): 241–246. Gennarelli TA, Spielman GM, Langfitt TW, Gildenberg PL, Harrington T, Jane JA, Marshall LF, Miller JD, Pitts LH. 1982. Influence of the type of intracranial lesion on outcome from severe head injury. J Neurosurg 56(1): 26–36.

623

624

William M. Brooks

Gentleman SM, Roberts GW, Gennarelli TA, Maxwell WL, Adams JH, Kerr S, Graham DI. 1995. Axonal injury: a universal consequence of fatal closed head injury? Acta Neuropathol 89(6): 537–543. Goodman JC, Valadka AB, Gopinath SP, Uzura M, Robertson CS. 1999. Extracellular lactate and glucose alterations in the brain after head injury measured by microdialysis. Crit Care Med 27(9): 1965–1973. Grady MS, McLaughlin MR, Christman CW, Valadka AB, Fligner CL, Povlishock JT. 1993. The use of antibodies targeted against the neurofilament subunits for the detection of diffuse axonal injury in humans. J Neuropathol Exp Neurol 52(2): 143–152. Graham DI, Gentleman SM, Lynch A, Roberts GW. 1995. Distribution of b-amyloid protein in the brain following severe head injury. Neuropathol Appl Neurobiol 2(1): 27–34. Graham GD, Barker PB, Brooks WM, Morris DC, Ahmed W, Bryniarski E, Hearshen DO, Sanders JA, Holshouser B.A, Turkel CC. 2000. MR spectroscopy study of dichloroacetate treatment after ischemic stroke. Neurology 55(9): 1376–1378. Gualtieri CT. 1995. The problem of mild brain injury. Neuropsychiatry Neuropsychology Behav Neurol 8: 127–136. Haseler LJ, Arcinue E, Danielsen ER, Bluml S, Ross BD. 1997. Evidence from proton magnetic-resonance spectroscopy for a metabolic cascade of neuronal damage in shaken baby syndrome. Pediatrics 99(1): 4–14. Holshouser BA, Ashwal S, Luh GY, Shu S, Kahlon S, Auld KL, Tomasi LG, Perkin RM, Hinshaw Jr DB. 1997. Proton MR spectroscopy after acute nervous system injury: outcome prediction in neonates, infants, and children. Radiology 202(2): 487–496. Holshouser BA, Ashwal S, Shu S, Hinshaw DB. 2000. Proton MR spectroscopy in children with acute brain injury: comparisons of short and long echo time acquisitions. J Magn Reson Imag 11(1): 9–19. Jennett B, Bond M. 1975. Assessment of outcome after severe brain damage: a practical scale. Lancet 1(7905): 480–484. Kaufman MJ, Pollack MH, Villafuerta RA, Kukes TJ, Rose SL, Mendelson JH, Cohen BM, Renshaw PF. 1999. Cerebral phosphorus metabolite abnormalities in opiate-dependent polydrug abusers in methadone maintenance. Psychiatry Res 90(3): 143–152. Kraus JF, McArthur DL, Silberman TA. 1994. Epidemiology of mild brain injury. Semin Neurol 14(1): 1–7. Lee JH, Arcinue E, Ross BD. 1994. Brief report: organic osmolytes in the brain of an infant with hypernatremia. N Engl J Med 331(7): 439–442. Levin HS, Williams DH, Eisenberg HM, High Jr WM, Guinto Jr FC. 1992. Serial MRI and neurobehavioural findings after

mild to moderate closed head injury. J Neurol Neurosurg Psychiatry 55(4): 255–262. Lien YH, Shapiro JI, Chan L. 1990. Effects of hypernatremia on organic brain osmolites. J Clin Inv 85(5): 1427–1435. Lu D, Margouleff C, Rubin E, Labar D, Schaul N, Ishikawa T, Kazumata K, Antonini A, Dhawan V, Hyman RA, Eidelberg D. 1997. Temporal lobe epilepsy: correlation of proton magnetic resonance spectroscopy and 18F-fluorodeoxyglucose positron emission tomography. Magn Reson Med 37(1): 18–23. MacMillan CSA, Wild JM, Wardlaw JM, Andrews PJD, Marshall I, Easton VJ. 2002. Traumatic brain injury and subarachnoid hemorrhage: in vivo occult pathology demonstrated by magnetic resonance spectroscopy may not be “ischaemic’’. A primary study and review of the literature. Acta Neurochir 144: 853–862. McIntosh TK, Smith DH, Meaney DF, Kotapka MJ, Gennarelli TA, Graham DI. 1996. Neuropathological sequelae of traumatic brain injury: relationship to neurochemical and biomechanical mechanisms. Lab Invest 74(2): 315–342. O’Neill J, Eberling JL, Schuff N, Jagust W, Reed B, Soto G, Ezekiel F, Klein G, Weiner MW. 2000. Method to correlate 1H MRSI and 18FDG-PET. Magn Reson Med 43(2): 244–250. Povlishock JT, Christman CW. 1995. The pathobiology of traumatically induced axonal injury in animals and humans: a review of current thoughts. J Neurotrauma 12(4): 555–564. Rango M, Lenkinski RE, Alves WM, Gennarelli TA. 1990. Brain pH in head injury: an image-guided 31P magnetic resonance study. Ann Neurol 28(5): 661–667. Ricci R, Barbarella G, Musi P, Boldrini P, Trevisan C, Basaglia N. 1997. Localised proton MR spectroscopy of brain metabolism in vegetative patients. Neuroradiology 39(5): 313–319. Rice D, MacKenzie E. 1989. Cost of Injury in the United States. San Francisco: Institute for Health and Aging”, University of California. Ross BD, Bluml S, Cowan R, Danielsen E, Farrow N, Tan J. 1998b. In vivo MR spectroscopy of human dementia. Neuroimag Clin N Am 8(4): 809–822. Ross BD, Ernst T, Kreis R, Haseler LJ, Bayer S, Danielsen E, Bluml S, Shonk BS, Mandigo J, Caton W, Clark, Jensen SW, Lehman NL, Arcinue E, Pudenz R, Shelden J. 1998a. 1H MRS in acute traumatic brain injury. J Magn Reson Imag 8(4): 829–840. Rubin Y, Cecil K, Wehrli S, McIntosh TK, Lenkinski RE, Smith DH. 1997. High-resolution 1H NMR spectroscopy following experimental brain trauma. J Neurotrauma 14(7): 441–449. Russell WR. 1932. Cerebral involvement in head injury. Brain 55: 549–603. Sappey-Marinier D, Calabrese G, Hetherington HP, Fisher SN, Deicken R, Van Dyke C, Fein G, Weiner MW. 1992. Proton MRS of human brain: applications to normal white matter,

MR spectroscopy in traumatic brain injury

chronic infarction, and MRI white matter signal hyperintensities. Magn Reson Med 26(2): 313–327. Schuhmann MU, Stiller D, Skardelly M, Bernarding J, Klinge PM, Samii A, Samii M, Brinker T. 2003. Metabolic changes in the vicinity of brain contusion: a proton magnetic resonance spectroscopy and histology study. J Neurotrauma 20(8): 725–743. Sinson G, Bagley LJ, Cecil KM, Torchia M, McGowan JC, Lenkinski RE, McIntosh TK, Grossman RI. 2001. Magnetization transfer imaging and proton MR spectroscopy in the evaluation of axonal injury: correlation with clinical outcome after traumatic brain injury. Am J Neuroradiol 22(1): 143–151. Smith DH, Cecil KM, Meaney DF, Chen XH, McIntosh TK, Gennarelli TA, Lenkinski RE. 1998. Magnetic resonance spectroscopy of diffuse brain trauma in the pig. J Neurotrauma 15(9): 665–674. Sosin DM, Sacks JJ, Smith SM. 1989. Head-injury associated deaths in the United States from 1979 to 1986. J Am Med Assoc 262(16): 2251–2255. Sutton LN, Wang Z, Duhaime AC, Costarino D, Sauter R, Zimmerman R. 1995. Tissue lactate in pediatric head

trauma: a clinical study using 1H NMR spectroscopy. Pediatr Neurosurg 22(2): 81–87. Teasdale G, Jennett B. 1974. Assessment of coma and impaired consciousness. Lancet 2(7872): 81–84. Uzan M, Albayram S, Dashti SGR, Aydin S, Hanci M, Kuday C. 2003. Thalamic proton magnetic resonance spectroscopy in vegetative state induced by traumatic brain injury. J Neurol Neurosurg Psychiatry 74(1): 33–38. Waxman K, Sundine MJ, Young RF. 1991. Is early prediction of outcome in severe head injury possible? Arch Surg 126(10): 1237–1242. Wild JM, MacMillan CSA, Wardlaw JM, Marshall I, Cannon J, Easton VJ, Andrews PJD. 1999. 1H spectroscopic imaging of acute head injury – evidence of diffuse axonal injury. Magma 8(2): 109–115. Yoon SJ, Lee JH, Kim ST, Chun MH. 2000. Evaluation of traumatic brain injured patients in correlation with functional status in rehabilitation medicine by localized 1H MR spectroscopy. Proc Intl Reson Med (Abstract) 1140. Zampolini M, Tarducci R, Gobbi G, Franceschini M, Todeschini E, Presciutti O. 1997. Localized in vivo 1H-MRS of traumatic brain injury. Euro J Neurol 4: 246–254.

625

38

Diffusion- and perfusion-weighted MR imaging in head injury Peter G. Bradley and David K. Menon Division of Anaesthesia, University of Cambridge, Addenbrooke’s Hospital, Cambridge, UK

Basic pathophysiology

Acceleration magnitude (g)

A detailed discussion of the pathological features of acute traumatic brain injury (TBI) is beyond the scope of this chapter and has been discussed elsewhere (Menon, 1999). However, it is important to appreciate that the severity and type of impact will substantially influence the structural lesions that ensue (Figure 38.1). The acceleration–deceleration forces that follow from impact during falls and motor vehicle accidents can produce axonal dysfunction and injury, brain contusions, and axial and extraaxial hematomas. The generation of such macroscopic injury is associated with microscopic and ultramicroscopic changes, including ischemic cytotoxic oedema, astrocyte swelling with microvascular effacement and dysfunction, microglial activation SDH

DAI

300

200

Concussion

100 No injury 0 0

10 20 30 40 Acceleration duration (ms)

Fig. 38.1 Biomechanics of head injury: diagrammatic representation of the duration and intensity of acceleration–deceleration insult on the type of injury produced. (Redrawn from Bullock, 1997). SDH, Sub-dural hemorrhage; DAI, Diffuse axonal injury.

626

50

and recruitment, blood brain barrier (BBB) disruption with vasogenic edema, and phasic inflammatory cell recruitment. The pathophysiological processes underlying these pathological changes have been extensively discussed in previous articles (Graham et al., 2002) and will not be addressed in detail here. These varied pathophysiological consequences of a single structural pathology are well reflected by sequential changes in cerebrovascular physiology that are observed following head injury. Classically, cerebral blood flow (CBF) is thought to show triphasic behavior (Martin et al., 1997; Bullock, 1997) (Figure 38.2). Early after head injury (within 12 h), global CBF is reduced, sometimes to ischemic levels. Between 12 and 24 h post-injury, gradually CBF increases towards baseline levels, and the brain may exhibit supranormal CBF 2–5 days post-injury. While many reports refer to this phenomenon as hyperemia, the absence of consistent reductions in cerebral oxygen extraction suggest that metabolism and blood flow often remain coupled, and a more appropriate label would be hyperperfusion. CBF values begin to fall several days following head injury, and in some patients these reductions in CBF may be associated with marked increases in large vessel flow velocity on transcranial Doppler ultrasound that suggest vasospasm. These time-varying hemodynamic responses also define the vascular contribution to intracranial pressure elevation in time (Figure 38.2). Immediately after head injury there is no vascular engorgement, and though a transient BBB leak has been reported in the first hour after impact in animal models, there are no data regarding BBB

Diffusion- and perfusion-weighted MR imaging in head injury

CBF (ml/100 g/min)

↑ ICP mechanisms

Cytotoxic edema Vasogenic edema Vascular engorgement

50 40 30 20 10

Survival Ischemic threshold Death

0 Whole brain ADC (% of base line)

120 110 100 90

2% 1% Normal (⬃79%)

Brain water content

80

0 1 2 3 4 5 6 7 8 9 10 Trauma ↑

Days post-injury

30

Fig. 38.2 Pathophysiology of brain water and CBF: diagrammatic representation of changes in brain water, average apparent diffusion coefficient (ADC), and CBF, and the associated physiological processes responsible for intracranial hypertension at various points following head injury. The temporal patterns of the various processes depicted are based on a consensus for both clinical and experimental literature cited in this chapter.

disruption at this stage in humans. Apart from mass lesions (e.g. intracranial hematomas), intracranial pressure (ICP) elevation during this phase is assumed to be the consequence of cytotoxic edema, usually secondary to cerebral ischemia. Increase in CBF and cerebral blood volume (CBV) from the second

day post-injury onwards make vascular engorgement an important contributor to intracranial hypertension. Following a transient initial dysfunction, the BBB appears to become leaky between the second and fifth days post-trauma, and vasogenic edema then

627

628

Peter G. Bradley and David K. Menon

(a)

(c)

(b)

(d)

Fig. 38.3 Macrovascular and microvascular ischemia: conventionally, reductions in cerebral perfusion are due to reductions in cerebral perfusion pressure (CPP) or obstruction of proximal conductance vessels. This results in relatively uniform ischemia in the territory of the vessel involved, and greater oxygen extraction (b) when compared to the normal setting (a). However, microvascular injury associated with head trauma can cause astrocyte swelling, perivascular edema, endothelial swelling, and microvascular collapse (d). This results in tissue hypoxia with heterogeneous patterns of ischemia and oxygen extraction (c).

contributes more to brain swelling. It is important to recognize that brain water is increased within the first hour following injury, and remains high for up to 2 weeks (Figure 38.2). Over this period the excess brain water probably redistributes to varying extents between the intra- and extracellular compartments. In the setting of ischemic stroke, early intracellular edema is usually referred to as “cytotoxic”, and often attributed to (neuronal) ischemia. The situation in head injury is much more complex. Cell swelling may involve astrocytes and be the consequence of glutamate (Glu) reuptake rather than the direct consequence of ischemia. Alternatively, the extracellular edema that accumulates following brain trauma may be protein rich and restrict the diffusion of extracellular water (Bullock R, 1997), endowing it with attributes that result in MR appearances that are conventionally associated with intracellular edema. Finally, the cell swelling and extracellular edema may result in microvascular injury and collapse (Bullock et al., 1991; Vaz et al.,

1997), and be the cause, rather than the consequence of ischemia (Figure 38.3). While ischemia is common at postmortem in fatal head injury, CBF reductions are generally modest in the first few days following injury. Two different factors may provide explanations for this discordance. First, ischemia may be an ultra-early event (12 h), and may be missed by many clinical studies, which are commonly undertaken at later time points. Second, there may be substantial pathophysiological heterogeneity in the injured brain (Figure 38.4), and the impact of regional critical ischemia at later time points may be diluted by surrounding normal tissue, with no significant net effect on global measures of CBF. Since structural changes, as detected by X-ray computed tomography (CT) or conventional MR imaging (MRI) are relatively late and often irreversible, these considerations have lead to the conclusion that there is a need to image physiology and metabolism in such patients.

Diffusion- and perfusion-weighted MR imaging in head injury

FLAIR

0

CBF (ml/100 g/min)

60

Fig. 38.4 Heterogeneity in cerebral pathophysiology following head injury: FLAIR (left) and 15O positron emission tomography (15O-PET) derived CBF (right). There is a reduction in (regional cerebral blood flow) rCBF around the right frontal contusion, however the pericontusional tissue in the left basal ganglia shows regions of both increased and decreased rCBF.

Increased CBF

CBF MET

CBF MET

CBF MET

OEF Hyperemia Normal

Appropriate perfusion

Ischemia

CBF MET

Decreased CBF

CBF MET

CBF MET

CBF MET

Fig. 38.5 Conceptual paradigms in cerebral ischemia: conventionally, hypoperfusion is thought to represent ischemia, and hyperperfusion, hyperemia. However, these definitions are no longer valid when there are independent changes in cerebral metabolic rate (MET), as have been shown in head injury. In these settings, ischemia can only be defined by documenting an increase in oxygen extraction fraction (OEF), or by showing that CBF reductions produce cytotoxic edema.

The best-established technique for physiological imaging is the use of stable xenon computed tomography (Xe-CT) studies for measurement of regional cerebral blood flow (rCBF). While perfusion imaging confirms the presence of regional hypoperfusion in head injury, interpretation of these findings is

confounded by the fact that metabolism is independently depressed in head injury, and the CBF decreases may represent appropriately coupled hypoperfusion rather than ischemia (Figure 38.5). Concordant imaging of both flow and metabolism can address this issue, but this approach requires 15O positron

629

630

Peter G. Bradley and David K. Menon

emission tomography (15O-PET) (Figure 38.4), which is not widely available in this setting. There has hence been substantial enthusiasm to apply diffusion-weighted MRI in this setting, in the hope that it may be used to image the consequences of ischemia, if not ischemia itself. In addition, the ability to employ perfusion-weighted MRI could allow, at least qualitative, imaging of cerebral perfusion within the same modality.

Insights from MRI in experimental models Models A variety of animal models have been used to mimic human head injury. While the details of these models are covered in other publications, it is important to recognize that none of these covers the entire spectrum of pathology observed in humans. Broadly speaking, the controlled cortical impact model primarily produces a focal cortical contusion with some deeper injury, while the impact-acceleration model is similar but produces a greater degree of acceleration–deceleration insult, and may produce more axonal shearing. The lateral fluid percussion injuries produces a more widespread and diffuse injury over the injured hemisphere, and also results in transmitted pressure wave through the brain. Other models mimic subdural and epidural hematomas, but diffusion-weighted imaging (DWI) studies of these models are limited.

DWI changes Despite the important role ascribed to cerebral ischemia in the context of head injury, early MR studies with DWI suggested that the pathophysiology in animal models of head trauma was significantly different from that seen in simple ischemia. Hanstock et al. (1994) showed that within the first 4 h post-injury, rats subjected to a moderate fluid percussion injury showed increases, rather than decreases in apparent diffusion coefficient (ADC), contrary to the findings in acute ischemic stroke. These data suggested early vasogenic edema, and

were concordant with previous studies that had demonstrated early opening of the BBB using a variety of techniques, including Evans blue extravasation (Cortez et al., 1989) and Gd-DTPA enhancement on MRI (Barzo et al., 1996). More recent data from an impact-acceleration model suggest that this early increase in ADC is a real and consistent finding in a context that mimics contusion in human head injury (Barzo et al., 1996, 1997). However, it has become clear that this initial phase of high ADC (which implies vasogenic edema) is rapidly succeeded by a prolonged phase of low ADC, which lasts up to 2 weeks following injury (Barzo et al., 1997). These initial increases in ADC are consistently associated with increases in T2, demonstrated on T2 weighted spin echo (SE) or fluid attenuation inversion recovery (FLAIR) MRI sequences, appearances which characterise vasogenic edema. At least at early stages, the ADC increases are also temporally concordant with increased BBB leak as demonstrated by local enhancement following administration of Gd-DTPA (Barzo et al., 1996). Interestingly, both the early increases in ADC, and the subsequent reductions in ADC occur against a background of increased brain water content, which may only be small (a rise from 79% to 81%), but is clearly associated with disruption of normal physiology and characteristic imaging findings (Barzo et al., 1997). The severity of ADC reduction in such models may be substantially enhanced by hypotension and hypoxia, which add physiological insults to the picture produced by traumatic injury (Ito et al., 1996). The detection and temporal patterns of the high and low ADC phases is also crucially dependent on the precise model used, the severity of insult, and the time points at which imaging is undertaken. Consequently, departures from this pattern have been reported by other authors (Assaf et al., 1997, 1999; Albensi, 2000), who have used somewhat different methodology. ADC reductions, which are usually interpreted as signifying ischemic cytotoxic edema, are not confined to contusion models. Other studies, in models of experimental subdural hematoma, show reductions in ADC values in underlying cortex by 1 h postinsult, which become more extensive over the following 2 h (Tsuchida et al., 1999).

Diffusion- and perfusion-weighted MR imaging in head injury

PWI changes Perfusion weighted imaging (PWI) in experimental models of head injury has employed both susceptibility contrast agents (Assaf et al., 1999; Schneider et al., 2002) and arterial spin labeling (ASL) techniques (Hendrich et al., 1999). In summary, these studies show significant reductions in relative cerebral blood flow and volume (rCBF and rCBV). Significantly, early PWI (performed one to 3 h post-injury) showed far more severe and widespread perfusion deficits when compared to imaging studies that were undertaken 24 h post-head injury (Hendrich et al., 1999). These studies also confirm the perfusion heterogeneity in the injured brain, both at early and late time points, and demonstrate pericontusional hypoperfusion (Assaf et al., 1999; Hendrich et al., 1999), the severity of which correlates with early reductions in regional ADC. However, the extent of subsequent histological injury is better reflected by the extent of ADC change, rather than the perfusion deficit, which is typically much larger (Assaf et al., 1999). One important methodological finding is the observed heterogeneity in T1 values across the injured brain (Hendrich et al., 1999), which results in marked inaccuracies in calculated CBF with ASL techniques, if normal T1 values are used.

Interpretation of experimental MR data While these data provide important pointers to understanding the imaging findings on diffusionweighted MRI in humans, they possess two major shortcomings. First, no data are available from animal studies that clearly delineate DWI responses in diffuse axonal injury, since experimental models of this pathology are less well developed and characterised. This is a major failing since axonal shearing is an important pathological finding and prognostically relevant lesion in human head injury. Second, human head injury often combines several lesions including contusions, extra-axial hematomas, and axonal shearing injury, and although some MR studies have attempted to address this issue (Bendszus et al., 2002), the models used may not be representative, and the effect of interaction of these varying pathologies on imaging appearances remains unclear.

DWI in acute head injury Detection of cytotoxic edema Early MR studies in head injury using conventional MR sequences demonstrated the sensitivity of the technique for detecting axonal injury, with demonstration of patterns of pathology that had major prognostic significance (Kampfl et al., 1998). Subsequently several authors showed that DWI revealed lesions that were in keeping with axonal shearing injury, days to months following head injury (Weishmann et al., 1999; Liu et al., 1999; Takayama et al., 2000; Rugg-Gunn et al., 2001; Chan et al., 2003). In some of these instances, conventional MRI was entirely normal (Rugg-Gunn et al., 2001). Several of these studies showed major reductions in regional anisotropy on diffusion tensor images (DTI), particularly in subcortical white matter (WM) in the frontal and temporal regions, and in the splenium of the corpus callosum, all of which are classical sites of diffuse axonal injury. The sensitivity of DWI and DTI in demonstrating these lesions was significant, and they were thought to represent loss of water anisotropy in sheared WM tracts. However, increases in signal intensity on DWI images were seen as early as 1 day post-injury (Liu et al., 1999; Arfanakis et al., 2002) and were seen to be transient in other studies (Takayama et al., 2000; Arfanakis et al., 2002), suggesting that the observed changes might represent acute dynamic pathophysiology. Clear interpretation of some early studies remains difficult as ADC values were not reported, and the coexistence of T2 changes made it difficult to exclude the presence of T2 “shine through” in DWI images with significant T2 weighting.

Detection of vasogenic edema A better framework for classifying changes on DWI images is provided by Hergan et al. (2002), who classify lesions based on DWI and ADC appearances: • Type 1 lesions are DWI and ADC hyperintense and are interpreted as showing vasogenic edema, with the DWI hyperintensity presumably arising from T2 shine through effects.

631

632

Peter G. Bradley and David K. Menon

Fig. 38.6 Vasogenic edema: FLAIR (left) and ADC map (right) obtained at 3 T from a patient 22 h post-head injury. The T2 hyperintensity in the left frontal contusion is associated with an increased ADC, typical of vasogenic edema.

• Type 2 lesions are DWI hyperintense and ADC hypointense, and are thought to represent cytotoxic edema. • Type 3 lesions consist of a central hemorrhagic lesion surrounded by an area of increased ADC representing pericontusional vasogenic edema. In addition, all images were classified into three groups based on the size and extent of lesions, with group A consisting of focal injury, group B consisting of regional or confluent injury and group C representing diffuse or extensive injury. Complex changes on DWI images While this approach is useful, it may represent an oversimplification. For example, we have shown that regions of low ADC may be found in areas that are clearly defined as vasogenic edema on FLAIR sequences (Figure 38.6), suggesting that cytotoxic edema (or some other microstructural cause of restricted diffusion) can coexist with regions where there is continuing BBB leak (Bradley et al., 2003a). In other patients we have been able to show narrow rims of low ADC just outside regions of pericontusional vasogenic edema (Figure 38.7) characterised by a high ADC and high signal intensity on FLAIR

images (Bradley et al., 2003a). These findings are reminiscent of low ADC cuffs seen around intracerebral hemorrhage in some patients (Kidwell et al., 2001), and may represent the biological consequences of local thrombin release in hemorrhage or hemorrhagic contusions. It is not uncommon for low ADC and high ADC lesions to coexist (suggesting a greater temporal heterogeneity in lesion evolution than in experimental models), and for low ADC regions to be found in areas where T2 and FLAIR imaging show high signal, which would imply vasogenic edema (Figure 38.8). The biological basis of these findings and the identity of cells involved in their production remain uncertain (but cf. below). In other studies, we have shown that ADC changes seen in these settings can be highly dynamic, and change with cerebrovascular physiology. Thus, in pilot studies, we have shown that hyperventilation produces clear changes on regional ADC, with increases in ADC in some regions and reductions in others (Bradley et al., 2003b). These findings require confirmation, since there are no good data on the variability of DWI imaging in this setting. However, if proven, they may provide the first conclusive evidence to support the suggestion that hyperventilation-induced reductions in

Diffusion- and perfusion-weighted MR imaging in head injury

Fig. 38.7 Cytotoxic edema: FLAIR (left) and ADC map (right) obtained at 3 T from a patient 34 h following head injury. The T2 hyperintensity in the left frontoparietal contusion is associated with a decreased ADC, typical of cytotoxic edema.

Fig. 38.8 Mixed ADC lesions: FLAIR (left) and ADC map (right) obtained at 3 T from a patient 59 h following head injury. The T2 hyperintensity in the large right sided contusion is associated with areas of both increased and decreased ADC. This heterogeneity of ADC, particularly within large contusions, is not uncommon.

CBF may produce regional ischemia (Coles et al., 2002). Detection of axonal injury What remains clear is that DWI and DTI possess (cf. Case Study 38.1) unparalleled sensitivity in

detecting axonal injury. The burden of axonal injury is probably best appreciated on fractional anisotropy (FA) maps, which provide a quantitative measure of the anisotropy within an imaging voxel (Figure 38.9). The presence of intact myelinated WM tracts results in marked anisotropy, and FA maps clearly show the WM pathways in the brain.

633

634

Peter G. Bradley and David K. Menon

While T 2* weighted gradient echo (GE) images may be more sensitive at detecting the microhemorrhages associated with axonal injury (Lin et al., 2001; Tong et al., 2003), there seems to be little doubt that DWI is better at detecting axonal shearing. Huisman et al. (2003) reported on 427 lesions in 25 patients who underwent MRI within 48 h of head injury (Table 38.1). They found that DWI identified 310 shearing injuries, 70 of which were not seen on conventional MR sequences. Sixty-five percent of lesions

identified using DWI showed increases in ADC, but they provided no information on the relationship between ADC values and time after injury.

PWI While some experimental studies have reported on the use of PWI, there are no published clinical studies using the technique in head injury (although

On arrival

1 year after injury

DWI

DTI

Fig. 38.9 DTI imaging in acute head injury: T*2 weighted GE (left), diffusion weighted images (middle), and DTI FA map (R), obtained within 24 h of head injury to demonstrate the value of DTI imaging in demonstrating axonal injury. The loss of WM anisotropy in the right frontal region is evident, and coincides with a hyperintense lesion on the DWI frame.

Diffusion- and perfusion-weighted MR imaging in head injury

Table 38.1. Number of traumatic lesions detected by MRI Number of (%) lesions

DWI, T2/FLAIR, GRE  DWI, T2/FLAIR, GRE

DWI, T2/FLAIR , GRE

DWI, T2/FLAIR , GRE  DWI , T2/FLAIR, GRE  DWI , T2/FLAIR, GRE

DWI, T2/FLAIR, GRE 

84 (20) 147 (34) 70 (16) 9 (2) 9 (2) 8 (2) 100 (23)

Total

427 (100)

DSC MR-derived CBF

Sequence combination showing abnormality

blood oxygen level dependent functional MR imaging (BOLD) fMRI techniques have been used for functional imaging in the follow up of patients after head injury (McAllister et al., 2001). We have preliminary evidence of using susceptibility contrast PWI in this setting, and while comparison with gold standard techniques, such as PET, shows a similar overall pattern the poor correlation means that caution should be used when interpreting the results (Figure 38.10). This poor correlation may be because the methodology used for calculating CBF from conventional PWI images requires

PET derived CBF

Fig. 38.10 Correlation between 15O-PET CBF and PWI: images from a patient 38 h post head injury, acquired at 3 T, showing a spoiled gradient recalled acquisition (SPGR) image (bottom left) and 15O-PET derived CBF map (bottom right) and dynamic susceptibility contrast (DSC) MR derived CBF map (top left). The lower panel shows the voxel-by-voxel correlation between the two coregistered images (top right). While there is general concordance, substantial scatter is evident.

635

636

Peter G. Bradley and David K. Menon

further work. Alternatively, the susceptibility changes produced by intracerebral hemorrhage and local alterations in oxygen extraction (and hence deoxyhemoglobin levels) may make the assumptions commonly used in deriving CBF with this technique inaccurate.

Specific settings DWI has received special attention in two settings where central nervous system (CNS) trauma is important; namely non-accidental injury in children and boxing related brain injury.

demonstrated include cerebral atrophy, focal or generalized ventricular enlargement, cavum septum pellucidum (often fenestrated), and superficial siderosis. However, it remains unclear whether the diagnostic yield is consistently increased when conventional MRI is compared to CT (Moseley, 2000). Recently, it has been suggested (Zhang et al., 2003) that there may be diffuse changes in the brains of professional boxers (Figure 38.12), with increases in directionally averaged diffusion on a voxel-based analysis across the brain (Figure 38.13). Clearly, further studies are required, and the sensitivity of this approach compared to CT and conventional MRI. However, the technique holds some promise as an imaging technique for detecting and quantifying boxing related CNS injury.

Non accidental trauma in children There is an emerging belief that DWI may be a useful technique in this setting. Two publications (Suh et al., 2002; Biousse et al., 2002) compared DWI with conventional MRI in a total of 46 patients with presumptive diagnosis of non-accidental head injury (Figure 38.11). In both publications, DWI defined lesions were more extensive than those demonstrated on conventional MRI. While these findings suggest that DWI may be more sensitive in detection of such lesions, further studies are required before its role can be clearly defined in this setting. Indeed, a recently published scheme for imaging of non-accidental pediatric injury (Jaspan et al., 2003) proposes only an optional role for DWI. However, regardless of potential diagnostic benefits, the presence of DWI abnormalities in this setting provides important support for ischemia and axonal shearing as important pathophysiological mechanisms in the condition; such information may be important in identifying therapeutic targets and improving clinical management.

Boxing When compared to X-ray CT, conventional MRI appears to provide better imaging of CNS (cf. Case Study 38.2) abnormalities in boxers (reviewed by Moseley,2000). Abnormalities that are commonly

Interpretation of findings The data presented in this chapter attest to the prominent and growing role of diffusion and perfusion MR in defining, diagnosing and understanding TBI. However, while several experimental and clinical studies have revealed intriguing results, their interpretation is not straightforward. Our understanding of the pathophysiology underlying acute CNS insults is based on a conceptual framework generated in the setting of stroke. While several questions remain to be answered in the setting of stroke, the large clinical database and a substantial experimental literature have allowed us to at least define the boundaries of our understanding and provide an integrated (if incomplete) picture of how clinical DWI and PWI changes may be interpreted. Unfortunately, it is unlikely that these principles can be directly translated to TBI, where pathophysiology is more complex, and the sequence of processes that affect brain water content and distribution may be different. For example, the widespread perfusion abnormalities in early head injury (Coles et al., 2003) may make it inappropriate to use contralateral brain as a “normal” control, even when this seems structurally normal. This makes the interpretation of relative CBF images much more complex. Similarly, the coexistence of low CBV and CBF in many patients

Fig. 38.11 Non-accidental pediatric head injury: non-contrast CT scan (top left and top middle), axial MRI-T2-weighted image (top right and middle left), diffusion-weighted image (middle middle and middle right), and the calculated ADC map (bottom left and bottom middle) in an 11-month-old child with confirmed shaken baby syndrome. The CT study was obtained upon admission to the hospital and the DWI was performed two days later. There is an interhemispheric subdural hematoma (arrow) well seen on the CT scan (top middle). The T2-weighted images (top right and middle left) are remarkable only for possible cortical swelling involving the left occipital lobe (arrow). The DWI (middle and middle right) demonstrate symmetric hyperintensity involving the occipital lobes bilaterally, the right parietal lobe, and the posterior right frontal lobe. These hyperintensities on DWI correspond to decreased signal on the ADC map (bottom left and bottom middle), suggesting restricted diffusion. These abnormalities are consistent with bilateral watershed infarcts predominating in the occipito-parietal junctions. MR angiography (MRA) showed normal intracranial arteries. The venous sinuses were normal on the conventional MRI (From Biousse et al., 2002).

(a)

(b)

(c)

(d)

(e)

(f)

Fig. 38.12 MRI in boxing-related CNS injury; T1 weighted images of MR findings in boxers: Cavum septum pellucidum (a and b), FLAIR images showing nonspecific WM disease (c and d), and mild subcortical WM demyelination (e and f). (From Zhang et al., 2003).

Diffusion- and perfusion-weighted MR imaging in head injury

1.6 104

BDav

Normal Boxer

1.4 104 Number of pixels

techniques. In particular, correlational studies with other physiological imaging techniques such as PET, and longitudinal studies within patients or experimental animals could help to define the physiological characteristics and biological fate of the lesions that are demonstrated using these methodologies.

1.2 104 1 104 8000 2

6000 4000

REFERENCES

2000 0

0

0.5

1.0

1.5

BDav

2.0

2.5

3.0

3.5

(10 5 cm2/s)

Fig. 38.13 Statistical analysis of DWI images in boxers: normalized brain diffusion distribution histograms in a control subject and a boxer. The areas under the two curves are the same. The Dav data (dots and circles) are fitted with a triple Gaussian function to represent the two-compartment nature and the mixing between the two compartments (lines). The narrow peak represents the distribution of the brain tissue about its mean. The second and the third compartments have a broader distribution. The mean of the brain tissue pixel distribution is recognized as a mean diffusion constant for the entire brain (BDav). The distribution width of the brain tissue compartment is also recorded. The fitted curve of the boxer (circles) shifts to the right as compared with the curve of the control subject (dots). The second compartment level of the boxer’s curve is higher than that of the control subject. (From Zhang et al., 2003).

suggests abnormal microvascular physiology, while the presence of vascular engorgement in the subacute phase can substantially increase CBV and thus confound the application of algorithms that derive absolute or relative CBF maps. Finally, the experimental data suggest that ADC reductions in head injury are not always accompanied by reductions in CBF to levels that are typical of ischemic cytotoxic edema, raising the possibility that other mechanisms may be responsible for these imaging findings. It seems clear that DWI and PWI provide important pathophysiological insights and diagnostic information in head injury, but further studies are required to make the best possible use of these

Albensi BC, Knoblach SM, Chew BGM, O’Reilly MP, Faden AI, Pekar JJ. 2000. Diffusion and high resolution MRI of traumatic brain injury in rats: time course and corellation with histology. Exp Neurol 162: 61–72. Arfanakis K, Haughton VM, Carew JD, Rogers BP, Dempsey RJ, Meyerand ME. 2002. Diffusion tensor MR imaging in diffuse axonal injury. Am J Neuroradiol 23: 794–802. Assaf Y, Beit-Yannai E, Shohami E, Berman E, Cohen Y. 1997. Diffusion- and T2-weighted MRI of closed-head injury in rats: a time course study and correlation with histology. Magn Reson Imaging 15 (1): 77–85. Assaf Y, Holokovsky A, Berman E, Shapira Y, Shohami E, Cohen Y. 1999. Diffusion and perfusion magnetic resonance imaging following closed head injury in rats. J Neurotraum 16 (12): 1165–1176. Barzo P, Marmarou A, Fatouros P, Corwin F, Dunbar J. 1996. Magnetic resonance imaging-monitored acute blood–brain barrier changes in experimental traumatic brain injury. J Neurosurg 85: 1113–1121. Barzo P, Marmarou A, Fatouros P, Hayasaki K, Corwin F. 1997. Contribution of vasogenic and cellular edema to traumatic brain swelling measured by diffusion-weighted imaging. J Neurosurg 87: 900–907. Bendszuz M, Burger R, Vince GH, Solymosi L. 2002. A reproducible model of an epidural mass lesion in rodents; Part II: characterization by in vivo magnetic resonance imaging. J Neurosurg 97 (6): 1419–1423. Biousse V, Suh DY, Newman NJ, Davis PC, Mapstone T, Lambert SR. 2002. Diffusion-weighted magnetic resonance imaging in shaken baby syndrome. Am J Opthalmol 133 (2): 249–255. Bradley PG, Harding SG, Pena A, et al. 2003a. Diffusion weighted magnetic resonance imaging in early severe head injury. Crit Care 7 (suppl 2): P89. Bradley PG, Harding SG, Pena A, et al. 2003b. Hyperventilation induced changes in diffusion weighted imaging in early severe head inury. J Cereb Blood Flow Metab (supp Brain 03): P406.

639

640

Peter G. Bradley and David K. Menon

Bullock R. 1997; In Head Injury, Chapman and Hall Medical, London. Chapter 7. http://www.edc.gsph.pitt.edu/neurotrauma/thebook/Chap07.pdf Bullock R, Maxwell WL, Graham DI, Teasdale GM, Adams JH. 1991. Glial swelling following human cerebral contusion: an ultrastructural study. J Neurol Neurosurg Psychiatr 54: 427–434. Coles JP, Minhas PS, Fryer TD, et al. 2002. Effect of hyperventilation on cerebral blood flow in traumatic head injury: clinical relevance and monitoring correlates. Crit Care Med 30: 1950–1959. Coles JP, Fryer TD, Smielewski P, et al. 2004. Defining ischemic burden after traumatic brain injury using 15O PET imaging of cerebral physiology. J Cereb Blood Flow Metab 24: 191–201. Chan JHM, Tsui EYK, Peh WCG, et al. 2003. Diffuse axonal injury: detection of changes in anisotropy of water diffusion by diffusion-weighted imaging. Neuroradiology 45: 34–38. Cortez SC, McIntosh TK, Noble LJ. 1989. Experimental fluid percussion brain injury: vascular disruption and neuronal and glial alterations. Brain Res 482: 271–282. Graham DI, Gennarelli TA, McIntosh TK. 2002. In Greenfield’s Neuropathology, Vol. 1 Arnold, London, pp. 828–830. Hanstock CC, Faden AI, Bendall R, Vink R. 1994. Diffusionweighted imaging differentiates ischaemic tissue from Traumatized tissue. Stroke 25 (4): 843–848. Hendrich KS, Kochanek PM, Williams DS, Schiding JK, Marion DW, Ho C. 1999. Early perfusion after controlled cortical impact in rats: quantification by arterial spinlabeled MRI and the influence of spin-lattice relaxation time heterogeneity. Magn Reson Med 42: 673–681. Hergan K, Schaefer PW, Sorensen AG, Gonzalez RG, Huisman TAGM. 2002. Diffusion-weighted MRI in diffuse axonal injury of the brain. Eur Radiol 12: 2536–2541. Huisman TAGM, Sorensen AG, Hergan K, Gonzalez RG, Schaefer PW. 2003. Diffusion-weighted imaging for the evaluation of diffuse axonal injury in closed head Injury. J Comput Assist Tomogr 27 (1): 5–11. Ito J, Marmarou A, Barzo P, Fatouros P, Corwin F. 1996. Characterization of edema by diffusion-weighted imaging in experimental traumatic brain injury. J Neurosurg 84: 97–103. Jaspan T, Griffiths PD, McConachie NS, Punt JAG. 2003. Neuroimaging for non-accidental head injury in childhood: a proposed protocol. Clin Radiol 58: 44–53. Kampfl A, Schmutzhard E, Franz G, et al. 1998. Prediction of recovery from post-traumatic vegetative state with cerebral magnetic-resonance imaging. Lancet 351: 1763–1767.

Kidwell CS, Saver JL, Mattiello J, et al. 2001. Diffusionperfusion MR evaluation of perihematomal injury in hyperacute intracerebral hemorrhage. Neurology 57:1611–1617. Lin DDM, Filippi CG, Steever AB, Zimmerman RD. 2001. Detection of intracranial hemorrhage: comparison between gradient-echo images and b0 images obtained from diffusion-weighted echo-planar sequences. Am J Neuroradiol 22: 1275–1281. Liu AY, Maldjian JA, Bagley LJ, Sinson GP, Grossman RI. 1999. Traumatic brain injury: diffusion-weighted MR imaging findings. Am J Neuroradiol 20: 1636–1641. Martin NA, Patwardhan RV, Alexander MJ, et al. 1997. Characterization of cerebral hemodynamic phases following severe head trauma: hypoperfusion, hyperemia, and vasospasm. J Neurosurg 87: 9–19. McAllister TW, Sparling MB, Flashman LA, Guerin SJ, Mamourian AC, Saykin AJ. 2001. Differential working memory load effects after mild traumatic brain injury. NeuroImage 14: 1004–1012. Menon DK. 1999. Cerebral protection in severe brain injury: physiological determinants of outcome and their optimisation. Br Med Bull 55: 226–258. Moseley IF. 2000. The neuroimaging evidence for chronic brain damage due to boxing. Neuroradiology 42: 1–8. Rugg-Gunn FJ, Symms MR, Barker GJ, 2001. Greenwood R, Duncan JS. Diffusion imaging shows abnormalities after blunt head trauma when conventional magnetic resonance imaging is normal. J Neurol Neurosurg Psychiatr 70: 530–533. Schneider G, Fries P, Wagner-Jochem D, et al. 2002. Pathophysiological changes after traumatic brain injury: comparison of two experimental animal models by means of MRI. MAGMA 14 (3): 233–241. Suh DY, Davis PC, Hopkins KL, Fajman NN, Mapstone TB. 2002. Nonaccidental pediatric head injury: diffusionweighted imaging findings. Neurosurgery 49(2): 309–318; discussion 318–320. Takayama H, Kobayashi M, Sugishita M, Mihara B. 2000. Diffusion-weighted imaging demonstrates transient cytotoxic edema involving the corpus callosum in a patient with diffuse brain injury. Clin Neurol Neurosurg 102: 135–139. Tong KA, Ashwal S, Holshouser BA, et al. 2003. Hemorrhagic shearing lesions in children and adolescents with posttraumatic diffuse axonal injury: improved detection and initial results. Radiology 227: 332–339. Tsuchida E, Alessandri B, Corwin F, Fatouros P, Bullock R. 1999. Detection of ultra-early brain damage after acute subdural hematoma in the rat by magnetic resonance imaging. J Neurotraum 16(7): 595–602.

Diffusion- and perfusion-weighted MR imaging in head injury

Vaz R, Sarmento A, Borges N, Cruz C, Azevedo I. 1997. Ultrastructural study of brain microvessels in patients with traumatic cerebral contusions. Acta Neurochir (Wien) 139: 215–220. Wieshmann UC, Symms MR, Clark CA, et al. 1999. Blunthead trauma associated with widespread water-diffusion changes. The Lancet 353: 1242–1243.

Zhang L, Ravdin LD, Relkin N, et al. 2003. Increased diffusion in the brain of professional boxers: a preclinical sign of traumatic brain injury? Am J Neuroradiol 24: 52–57.

641

642

Peter G. Bradley and David K. Menon

Case Study 38.1 Diffuse axonal injury Karen Tong, M.D., Stephen Ashwal, M.D., E. Mark Haacke, Ph.D., Loma Linda University School of Medicine, Loma Linda, California History 9-year-old boy struck by car, with initial GCS of three, later declared brain dead, and taken off life support.

CT

FLAIR

DWI

ADC

GRE

SWI

Technique CT, conventional MRI, diffusion-weighted and susceptibility weighted imaging (SWI).

Imaging findings CT shows diffuse sulcal effacement. Scattered hyperintense injuries are seen on FLAIR and DWI. Numerous tiny hemorrhages are better seen on SWI, compared to the corresponding conventional GE image.

Discussion Diffuse axonal injury (DAI) is a significant but poorly detected form of TBI. Small hemorrhages at the GM–WM junction or deep brain are surrogate markers of axonal injury. SWI is a highly modified gradient echo MR technique (Reichenbach et al., 1997) that identifies six times more lesions and twice the volume of hemorrhages than conventional GE (Tong et al., 2003).

Key points DAI is underestimated by CT or conventional MRI. SWI is highly sensitive for small hemorrhages indicative of DAI.

References Reichenbach JR, Venkatesan R, Schillinger DJ, Kido DK, Haacke EM. 1997. Small vessels in the human brain: MR venography with deoxyhemoglobin as an intrinsic contrast agent. Radiology 204: 272–277. Tong K, Ashwal S, Holshouser B, Shutter L, Herigault G, Haacke EM, Kido D. 2003. Improved detection of hemorrhagic shearing lesions in children with post-traumatic diffuse axonal injury–initial results. Radiology 227: 332–339.

Diffusion- and perfusion-weighted MR imaging in head injury

Case Study 38.2 Occult brain damage in a professional boxer Lijuan Zhang, M.D., Robert D. Zimmerman, M.D., Aziz M. Ulug Ph.D., Weill Medical College of Cornell University, New York, USA History 36-year-old professional boxer with no history of neurological dysfunction or cognitive defects.

T1w

T2w

FLAIR

DWI

Technique Conventional and diffusion MRI. Imaging findings Normal routine T1 weighted, T2 weighted FLAIR and DWI examination. Quantitative diffusion findings Diffusion parameters measured from the entire brain, are elevated in the subject compared to a normal.

Discussion

Key points Conventional MR is unremarkable. Quantitative diffusion imaging may suggest microstructural damage in the subject’s brain that is invisible to routine MRI.

1.5 104

The data points from the entire brain of a normal control and a boxer with normal MRI are normalized and fitted to a triple Gaussian function (Zhang, 2003). The mean diffusion constant of the brain tissue (BDav) is represented by the peak. The fitted curve of the boxer shifts to the right as compared with the curve of the normal subject.

normal subject boxer with normal MRI

4

1.2 10 Number of pixels

Dementia Pugilistica is commonly reported in retired boxers (Jordan, 2000). Routine MRI including DWI in this case are normal; however, increased diffusion in the whole brain indicates microstructural brain damage invisible to routine MRI (Zhang, 2003). Increased diffusion was also reported in demented subjects (Ulug, 2001) and therefore increased brain diffusion constant may represent preclinical signs of cognitive decline.

9000

6000

3000

0

0

1

2

3

BDav (10 5 cm2/s)

References Jordan BD. 2000. Chronic traumatic injury associated with boxing. Semin Neurol 20: 179–185. Ulug AM, Relkin N, Zimmerman RD. 2001. Diagnosis of normal pressure hydrocephalus using diffusion imaging. In Proceedings of the 30th Annual Meeting of the American Aging Association. Madison, Wisc: American Aging Association, 72. Zhang LJ, Ravdin LD, Relkin N, Zimmerman RD, Jordan B, Lathan WE, Ulug, AM. 2003. Increased diffusion in the brain of professional boxers: A preclinical sign of traumatic brain injury. Am J Neuroradial 24: 52–57.

643

Section 8 Pediatrics

39

Physiological MR of the pediatric brain: overview Elias R. Melhem1 and Xavier Golay2 1

Department of Radiology, Hospital of the University of Pennsylvania, Philadelphia, USA Department of Radiology, The Johns Hopkins University, Baltimore, USA

2

Introduction MR imaging (MRI) has made important contributions toward the study of the developing pediatric brain. In addition to morphological information, advanced MRI methodologies are being relied on to interrogate non-invasively brain chemistry, physiology, and microstructure. Altogether, the application of such advanced MR methodologies, including spectroscopy (MRS), perfusion imaging, and diffusiontensor imaging (DTI) in the pediatric population has the potential for providing more in-depth information in the daily pediatric radiology practice. In an ideal world, one should be able to apply all these techniques together to more appropriately differentiate between several pathologies. However, despite the obvious advantages of the combination of such techniques, most of these procedures are actually applied separately. The main reason for this partitioning comes from the prolonged acquisition times associated with each of these techniques. Furthermore, most of these methods are by their very nature sensitive to motion, and can be challenging to apply to difficult patient populations, such as unsedated children with disabilities or developmental delay. Recently, however, the incorporation of fast spatial-encoding methods, such as those provided by parallel imaging (Sodickson and Manning, 1997; Pruessmann et al., 1999), has made standard use of advanced MRI for the evaluation of the pediatric brain more feasible and has allowed for the routine implementation of isotropic, high spatial resolution three-dimensional (3D) morphological imaging.

Furthermore, the greater availability of high field (3 T) MR scanners and phased-array receiver coils designed for brain imaging has permitted the tradeoff of high image signal-to-noise for faster acquisition time. These improvements should allow in the future comprehensive physiological MR studies to be performed in children with clinically acceptable scan times.

MR spectroscopy Background MRS provides the ability to generate non-invasively biochemical profile characteristics of the pediatric brain in vivo. As described in Chapter 1, through careful analysis of spectra one can obtain information about neuronal viability, cell membrane integrity or cellular bioenergetics (Cecil and Jones, 2001). Several brain metabolites can be separated based on differences in resonance frequency using MRS (Figure 39.1) (Miller, 1991). Metabolites that are identified on proton MRS and commonly used in clinical assessment of pediatric brain include (cf. also Chapter 1): (a) N-acetyl aspartate (NAA), which comprises a majority of the resonances at 2.02 ppm, is synthesized almost exclusively in the mitochondria of the neurons and hence is a marker of neuronal density and function. (b) Creatine and creatine phosphate (Cr), with resonances at 3.02 ppm, are involved in the regulation of cellular energetics. In general, Cr is used 647

648

Elias R. Melhem and Xavier Golay

NAA

Cr Cho ml Glx

Glx Lac

4.0

3.0

2.0

1.0

Fig. 39.1 Normal short echo time (TE) proton MR spectrum (TR/TE: 2000/35) obtained from a single voxel placed in left frontal lobe of a 9-year-old boy.

(c)

(d)

(e)

(f)

as an internal reference against which other metabolites are compared. Exceptions to that rule are conditions affecting cellular energetics such as ischemia and necrosis, and rare conditions like creatine-deficiency syndromes (Cecil et al., 2003). Phosphocholine (PC) and glycerophosphocholine (GPC), with resonances at 3.21 ppm, are markers of membrane biochemistry. Cho is a sensitive but not very specific marker, which can be elevated in a multitude of conditions that cause myelin breakdown or cellular proliferation. Myo-inositol (mI), with resonance at 3.5 ppm, is involved in the regulation of intracellular sodium content and glial activation. Again not a very specific metabolite that increases in regions of gliosis and astrocytic activation. Glutamate (Glu) and glutamine (Gln) (Glu  Gln  Glx), with resonances at 2.2–2.6 and 3.6–3.8 ppm, are markers that have been reported to increase in hypoxic–ischemic and hepatic encephalopathies. Lactate (Lac), with a doublet at 1.3 ppm, is a marker of anaerobic metabolism, which can be elevated in conditions that cause brain hypoxia and necrosis. Lac has been studied in children with stroke, brain tumors, and mitochondrial encephalopathies. It is important to note at this point that a proton spectrum’s appearance will depend on the chosen echo time (TE) and that

only four metabolites (Cho, Cr, NAA, and Lac) can be clearly seen and quantified at long (136 or 272 ms) TEs. In vivo proton MRS can be performed and analyzed using a multitude of methods (Salibi and Brown, 1998), the description of which is beyond the scope of this chapter (cf. Chapters 1 and 2). Suffice it to say, there are several robust localizing techniques (single voxel, multi-voxel, or chemical shift imaging), with a large range of acquisition parameters (short, intermediate or long TE) and semi-quantitative analysis packages made increasingly available on clinical MR scanners. From a practical standpoint, we encourage radiologists to adopt one proton MRS method in their clinical practice in order to enhance familiarity and optimize utility. If single voxel MRS is chosen, then localization becomes problematic when investigating neurological diseases that have no apparent abnormalities on conventional MRI. One approach to this problem in the pediatric population consists of routine interrogation of basal ganglia and white matter (WM) of the frontal lobe using two separate voxels. Another approach involves interrogation of brain regions that are often affected based on the pathophysiology of a particular disease. Obviously, this becomes less of an issue when multi-voxel MRS and chemical shift imaging techniques are implemented. Furthermore, radiologists interpreting proton MRS in the pediatric age group should be mindful of the effects of medications, such as propylene glycol, commonly used to sedate children during MRI and of normal developmental changes affecting brain metabolites. Considerable evolution in brain metabolites normally occurs during the first 3 years of life, which have been attributed to dendritic arborization, axonal pruning, and myelination. Quantitatively, there is a gradual decrease in mI and Cho concentrations, and a commensurate increase in NAA and Cr concentrations during that period of time (Kreis et al., 1993). Qualitatively, there is a critical change in the appearance of the metabolic spectrum in the first month of life, during which time NAA replaces mI and Cho as the dominant peak (cf. Chapters 40 and 44). Finally, the practicing radiologists should be cognizant of variations in brain metabolite levels that exist between

Physiological MR of the pediatric brain: overview

different regions of the brain particularly in gray matter (GM) vs. WM (Pouwels and Frahm, 1998).

Applications There is a plethora of published reports on the use of MRS in the evaluation of neurological and neurosurgical diseases affecting the pediatric brain. In this chapter we will limit our review to those major categories of disease in which MRS shows greatest promise. Specifically, we will discuss the role of MRS in the assessment of brain neoplasms, hypoxic– ischemic injury of the brain, and inborn errors of metabolism. These topics and others are discussed in detail in the following sections. Brain neoplasms MRS in pediatric brain neoplasms has been used to indicate degree of malignancy, to differentiate neoplasms from mimics, and to assess injurious effects of oncological therapies on the developing brain (Waldrop et al., 1998; Tzika et al., 2002) (cf. Chapter 43). Typical MR spectrum from a primary brain neoplasm will demonstrate elevated Cho and reduce NAA peaks. Initial hopes that the presence of a Lac peak, which reflects anaerobic metabolism, signifies malignant behavior (Figure 39.2) did not materialize in the pediatric population. Several reports of juvenile pilocytic astrocytomas (JPA) (WHO I), common low-grade pediatric neoplasms, showing elevated Lac peak (Figure 39.3) and a report by Sutton et al. (1994), showing elevated Lac in all pediatric gliomas studied, regardless of grade, have lessened the enthusiasm regarding MRS as a tool for assigning tumor grade. As far as we know, there are no single or combination of proton MRS-based metabolic markers that can reliably differentiate malignant from benign primary pediatric brain neoplsms. Hence, we encourage practicing radiologists to use MRS conservatively as an adjunct tool only in situations where grade cannot be predicted by the characteristics of the neoplasm on conventional MRI. On conventional MRI, it is often difficult to distinguish between non-enhancing neoplasms and dysplastic/hamartomatous brain lesions. Furthermore, in both types of abnormalities, there may be

elevated Cho and reduced NAA making the distinction difficult on MRS (Li et al., 1998). Recently, investigators have suggested that mI, which was found to be far more elevated in dysplastic lesion than in lowgrade gliomas, may aid in differentiating the two entities (Aasly et al., 1999). Unfortunately, in our hands mI has not been useful in this differentiation, especially since we have experienced a few welldocumented glial neoplasms with markedly elevated mI levels (Figure 39.4). MRS can also help exclude or confirm the possibility of neoplasia in brain abnormalities that are non-specific on conventional MRI (Figures 39.5 and 39.6). It is particularly useful in the differentiation between recurrent neoplasia and radiation necrosis (Rabinov et al., 2002). MRS in radiation necrosis typically demonstrates global reduction in all brain metabolites except for Lac and lipid (Figure 39.7). Hypoxia–ischemia In the neonate, acute hypoxic–ischemic brain injury may be extremely difficult to identify on MRI, including diffusion-weighted imaging (DWI), in the first 24 h after insult (McKinstry et al., 2002; Melhem, 2002). Recent reports demonstrate that elevated Lac and Glx on proton MRS may be early markers for acute injury especially in the region of the basal ganglia and thalami (BGT) (Pu et al., 2000). This observation, if substantiated by further longitudinal studies incorporating follow up imaging and neurological assessment, may place MRS at the helm of diagnostic methods destined to guide and justify acute interventions aimed at limiting the extent of neonatal brain injury (cf. Chapter 41). In children with adult-like brain infarction, MRS typically shows elevated Lac due anaerobic metabolism, and normal levels of NAA, Cho and Cr during the acute phase (first 72 h). In the subacute phase, NAA declines as a result of neuronal death, Cho elevates, and Lac remains elevated for up 3 weeks after the insult (Figure 39.8). In the chronic stage of brain infarction, all metabolites decline including Lac (Graham et al., 1993). Inborn errors of metabolism Extensive experience with MRS in metabolic diseases of the brain over the last 15 years has led to a

649

650

Elias R. Melhem and Xavier Golay

(a)

(b)

(c)

(d)

Fig. 39.2 A 5-year-old boy with choroids plexus carcinoma in the right lateral ventricle. Axial fast fluid attenuated inversion recovery (FLAIR) (TR/TE/TI: 8800/140/2200) (a) demonstrates mildly hyperintense mass in the trigone of the right lateral ventricle with hydrocephalus and periventricular edema. Corresponding metabolite maps of Cho (b), NAA (c), and Lac (d) obtained from multi-slice proton MRS imaging (TR/TE: 2300/280) demonstrate increased Cho and Lac intensity and absence of NAA in the mass as well as elevated Lac in CSF.

large number of publications documenting spectral abnormalities in these diseases (Hunter and Wang, 2001). These are reviewed in detail in Chapter 46. Again, in its current application, proton MRS tends to be quite sensitive but not specific to a particular metabolic disease. In other words, in most instances

MRS cannot provide a distinct metabolic profile characteristic of a specific metabolic disease. This has been attributed to the complexity and overlap in the metabolic pathways of the brain. Representative examples of this complexity are the large number of metabolic diseases that result in anaerobic glycolysis

Physiological MR of the pediatric brain: overview

(a)

(b)

Cho

Cr NAA

ml

Lac Glx

4.0

3.0

2.0

1.0

ppm

Fig. 39.3 A 7-year-old girl with midline cerebellar pilocytic astrocytoma. Axial contrastenhanced T1-weighted (TR/TE: 550/10) (a) demonstrates heterogeneously enhancing mass arising from the cerebellar vermis and compressing the fourth ventricle. Short TE proton MR spectrum (TR/TE: 2000/35) (b) obtained from a single voxel placed in the solid component of the mass demonstrates elevated Cho and Lac, and decreases NAA.

(a)

TR: 2000 TE: 35

(b)

ml

Cho Cr NAA

Fig. 39.4 A 9-year-old boy with brainstem fibrillary astrocytoma. Axial fast FLAIR (TR/TE/TI: 8800/140/2200) (a) demonstrates hyperintense mass enlarging the pons and bulging into the fourth ventricle and prepontine cistern. Short TE proton MR spectrum (TR/TE: 2000/35) (b) obtained from a single voxel placed in the mass demonstrates mild elevation of Cho and marked decrease of NAA. Interestingly, however, mI is the most markedly elevated metabolite in the mass and is the dominant peak in the spectrum.

and lactic acidosis. Inborn errors that disrupt the function of the Kreb’s cycle (Trinh et al., 2001), oxidative phosphorolation, or electron transport chain may all lead to elevated Lac in the brain and cerebrospinal fluid (CSF) (Figure 39.9). Still, MRS has the potential for determining the extent and

severity of metabolic derangement, predicting brain regions at risk for irreversible injury, and monitoring the effects of therapy in these metabolic diseases. Exceptions to the lack of distinct MRS profile in metabolic disease include: (a) Canavan disease, which has characteristic elevation in NAA on MRS

651

652

Elias R. Melhem and Xavier Golay

(a)

(b)

Fig. 39.5 A 12-year-old boy with seizures and an indeterminate lesion in the medial aspect of the left frontal lobe on MRI. Axial fast FLAIR (TR/TE/TI: 8800/140/2200) (a) demonstrates an illdefined lesion with central hypointensity, similar to CSF, with surrounding hyperintensity and no appreciable mass effect or enhancement (not shown). Corresponding metabolite map of Cho (b) obtained from multi-slice proton MRS imaging (TR/TE: 2300/280) demonstrates decreased Cho in the lesion, making neoplasia extremely unlikely. There was no change in the appearance of the lesion on 2-year follow up (not shown).

(a)

(b)

(c)

Fig. 39.6 A 19-year-old boy with a history of posterior fossa medulloblastoma cured with gross total resection and craniospinal radiation 12 years before these images. Currently the boy presents with behavioral change and progressive right arm and leg weakness. Axial fast FLAIR (TR/TE/TI: 8800/140/2200) (a) demonstrates hyperintensity in the WM of the left front-parietal lobe with mild associated mass effect. In addition, there is encephalomalacia in the right occipital lobe from resection of an overlying meningioma 2 years prior to this presentation. Axial contrast-enhanced T1-weighted (TR/TE: 550/10) (b) demonstrates no corresponding enhancement. Corresponding metabolite map of Cho (c) obtained from multi-slice proton MRS imaging (TR/TE: 2300/280) demonstrates marked elevation of Cho in the lesion. On open biopsy a glioblastoma was diagnosed.

Physiological MR of the pediatric brain: overview

(a)

(b)

3.0

2.0

1.0

ppm

Fig. 39.7 A 24-year-old female with petrous apex mass treated with radiation therapy (RT) 2 years prior to this presentation. Currently she presents with intractable seizures. Coronal contrast-enhanced T1-weighted (TR/TE: 550/10) (a) demonstrates ring-enhancing mass in the left medial temporal lobe worrisome for intracranial neoplastic extension vs. radiation induced brain necrosis. Short TE proton MR spectrum (TR/TE: 2000/35) (b) obtained from a single voxel that includes the enhancing portion of the mass, is strongly suggestive of radiation necrosis with marked reduction of all brain metabolites, and elevation Lac and lipids.

(a)

Cho Cr

(b) NAA

Lac

3.0

2.0

1.0

ppm

Fig. 39.8 A 25-year-old male who presented with acute onset of headache and disorientation. MRI and MRS were performed 4 days after presentation. Axial fast T2-weighted (TR/TE: 4500/80) (a) demonstrates swelling and diffuse T2-hyperintensity limited to the corpus callosum. Intermediate TE proton MR spectrum (TR/TE: 2000/140) (b) obtained from a single voxel placed in the corpus callosum demonstrates mild elevation of Cho, mild reduction of NAA and elevation of Lac, consistent with subacute infarction confirmed on follow up MRI.

653

654

Elias R. Melhem and Xavier Golay

(a)

(b)

(c)

Fig. 39.9 A 16-month-old girl with known methylmalonic academia presented with acute onset lethargy, vomiting, and upper arm tremors. Axial fast FLAIR (TR/TE/TI: 8800/140/2200) (a) performed 6 days after presentation demonstrates bilateral hyperintensity and swelling limited to the globi pallidi. Corresponding metabolite map of NAA (b) and Lac (c) obtained from multi-slice proton MRS imaging (TR/TE: 2300/280) demonstrate symmetric decreased NAA and increased Lac typical of metabolic diseases disrupting aerobic metabolism of the brain.

resulting from an error in the metabolism of NAA due to deficiency in aspartoacylase (ASPA) enzyme (Grodd et al., 1990), and (b) phenylketonuria (PKU), a disorder of phenylalanine (Phe) metabolism, which has characteristic elevation of Phe on MRS (Novotny et al., 1995). The role of MRS in metabolic disease is discussed further in Chapters 45 and 46.

MR perfusion imaging Background Brain perfusion refers to the delivery of oxygen and nutrients to a particular tissue, and removal of metabolic waste. Measurement of cerebral perfusion relies on the measurement of the concentration of a tracer in the brain at a particular point in time (Meier and Zierler, 1954; Zierler, 1965; Grubb et al., 1978). So far, several modality-specific tracers have been used to measure brain perfusion including iodinated contrast and xenon gas for computed tomography (CT) (Axel, 1980) and radionuclides for single photon emission computed tomography (SPECT) (Osborne et al., 1981) or PET (Go et al., 1981; Rhodes et al., 1981; Raichle et al., 1983). In children, MR-based perfusion imaging offers several

advantages over other modalities such as greater safety (no ionizing radiation), better temporal and spatial resolution, and lower cost. Two general categories of tracers have been used for MR perfusion imaging: (a) exogenous non-diffusible agents, such as paramagnetic contrast material (Villringer et al., 1988) and (b) endogenous diffusible agents, such as magnetically labeled hydrogen atoms in blood (Williams et al., 1992; Zhang et al., 1993). These are covered in detail in Chapters 7–9. Exogenous non-diffusible tracer methods are based on the assumption that the tracer remains in the intravascular compartment, and that a linear relationship exists between the intravascular concentration of the susceptibility agent and change in the MR relaxation rate. Most commonly, imaging is performed dynamically during an intravenous bolus injection of a paramagnetic agent using an ultrafast T 2*-weighted echo planar imaging (EPI) sequence. Less commonly, perfusion imaging is done in steady-state (imaging after a constant infusion has reached an equilibrium concentration of the tracer in the blood) or using T1-weighted sequences (Wong et al., 2000). Based on the central volume theorem and principle of tracer kinetics, perfusion parameters such as cerebral blood volume (CBV), cerebral blood flow (CBF) and mean transit time (MTT) of the tracer

Physiological MR of the pediatric brain: overview

through the brain can be calculated. However, these parameters are influenced by the amount of tracer injected and its rate of injection, the paramagnetic properties of the tracer, and the hemodynamic properties of the subject’s vascular system, including total vascular volume and cardiac output (Wong et al., 2000). In order to facilitate comparisons of these parameters over time and across different subjects, relative values are generated using internal standard of reference from normal appearing brain. Off course, this semi-quantitative approach can be applicable to diseases that result in local alterations in cerebral perfusion. In diseases where alterations may be more diffuse (such as sickle cell disease) the use of an internal standard of reference may lead to erroneous estimation of perfusion parameters, hence the need for quantitative approaches. One such attempt at absolute quantification of perfusion parameters relies on the accurate measurement of the arterial input function (AIF), which can only be determined using dynamic methods (Østergaard et al., 1996; cf. also Chapter 7). Unfortunately, the accuracy of this approach remains difficult to assess. Endogenous diffusible tracer methods are based on the assumption that the tracer can move freely from the intravascular compartment into the extravascular compartment. The endogenous tracer consists of magnetically labeled (“tagged”) arterial blood water and does not require the intravenous injection of exogenous agents. Contrast in this type of perfusion imaging, known as arterial spin labeling (ASL), is based on the signal obtained by the flow of the labeled arterial water protons into an imaging plane or volume and its exchange with unlabeled protons in the brain parenchyma (Williams et al., 1992). The inflowing arterial water protons can be labeled either by continuous application of a radio frequency (RF) pulse in a plane at the level of the cervical carotid (Detre et al., 1994; Alsop and Detre, 1996), or by spatial labeling of a large slab (Edelman et al., 1994; Wong et al., 1997; Golay et al., 1999), both proximal to the imaging plane or volume of interest (Figure 39.10). Alternatively, measurement of CBF can be obtained from the difference between selective and non-selective inversionrecovery sequences (Kim, 1995; Kwong et al., 1995). Typically, this is followed by imaging the brain

downstream from the level of spin labeling. A control image without spin labeling is also obtained, and the calculated signal difference between the labeled and control image allows for the measurement of CBF (for a review of all available methods, cf. Barbier et al., 2001 and Chapter 8). ASL methods suffer from several potential artifacts resulting in overestimation of brain perfusion. Among those, magnetization transfer (MT) effects will affect the magnetization in the imaging plane or volume of interest by the off-resonance labeling RF pulse, which will selectively saturate the broad resonance frequency peak of macromolecular-bound protons in the brain. Several strategies have been developed to compensate or eliminate the MT problem in both pulsed (Golay et al., 1999; Jahng et al., 2003) and continuous arterial spin labeling (CASL) methods (Alsop and Detre, 1998). Another source of inaccuracies is related to the variability in loss of spin labeling due to T1 relaxation of arterial water protons that occurs between the time of labeling and the time of imaging. Recently, the successful merger of ASL methods with single shot EPI, hence making them more efficient, has generated a lot of interest in the application of these completely non-invasive methods for the imaging of brain perfusion in the pediatric population. Applications There are few reports on the applications of MR perfusion imaging using exogenous or endogenous tracers in the assessment of children with stroke, neoplasms, and neurodegenerative disorders (Ball and Holland, 2001). Stroke Most applications of MR perfusion imaging are in adult stroke, and have been used for the evaluation of perfusion–diffusion mismatch in brain tissue at risk for ischemia. Recently, however, MR perfusion imaging has been used to generate perfusion parameters in children with sickle cell anemia who are at risk for stroke (Ball and Holland, 2001). In a study of 48 patients with sickle cell disease, perfusion abnormality, measured using dynamic perfusion MRI after an intravenous bolus injection of a paramagnetic agent, was larger than the area

655

656

Elias R. Melhem and Xavier Golay

Fig. 39.10 General scheme used in CASL. Two consecutive images are acquired, the first one as a control, and the second one as labeling, and the subtraction of both images provides a contrast sensitive to brain perfusion. In that particular implementation (Alsop and Detre, 1998), care has been taken to avoid MT artifacts in the final perfusion-weighted image.

of infarction in nine patients and was seen in an arterial distribution with no infarction in a further nine (Kirkham et al., 2001). In another study of 14 neurologically asymptomatic children with sickle cell disease (HbSS) and seven age-matched controls, CBF, measured with CASL-based MR perfusion imaging, was higher in patients than in controls in all cerebral artery territories (Figure 39.11). Four patients had significantly decreased CBF in right hemisphere compared to left (Figure 39.12). Finally, an important observation in this report was that three patients with brain perfusion abnormalities (reduction in baseline CBF) had no demonstrable abnormalities on conventional

MRI (Oguz et al., 2003). The latter observation underscores the value of advanced MR-based imaging techniques as preclinical markers for brain ischemia (see Chapters 42 and 44). Brain neoplasms Another important application of perfusion MRI is in the assessment of brain neoplasms. Specifically, measurements of CBV and vascular permeability are being used to assess the degree of neovascularity in tumors, which correlate with tumor grade and malignant histology (Cha et al., 2002). Furthermore, these measures may help to monitor the effect of therapeutic agents that target tumor angiogenesis. Perfusion

Physiological MR of the pediatric brain: overview

200 ml/min/100 g (a)

0 ml/min/100 g 250 ml/min/100 g (b)

0 ml/min/100 g Fig. 39.11 Whole brain CBF maps generated using CASL-based MR perfusion imaging (TR/TE  5000/36) obtained from 8-year-old female volunteer (a) and an 8-year-old girl with sickle cell disease (b). Note significantly brighter GM (higher CBF) in the patient with sickle cell disease.

657

658

Elias R. Melhem and Xavier Golay

255 ml/min/100 g

0 ml/min/100 g Fig. 39.12 A 9-year-old girl with sickle cell disease, but no history of stroke and no abnormalities on conventional MRI, MR angiography (MRA) and transcranial doppler sonography (TCD). Whole brain CBF maps generated using CASL-based MR perfusion imaging (TR/TE  5000/36) demonstrates more than 30% decrease in CBF in the entire right cerebral hemisphere.

MRI can also aid in better defining indeterminate brain lesions on conventional imaging (Figure 39.13) (Cha et al., 2002), and differentiating recurrent neoplasm from radiation necrosis (Wong et al., 2000). Typically, in regions of radiation necrosis CBV is lower than GM, whereas in regions of recurrent high-grade neoplasm CBV is equal or greater than GM. Neurodegenerative disorders A recent application of MR perfusion imaging has been in girls with Rett syndrome, a neurodegenerative disorder characterized by profound cognitive impairment, communication dysfunction, stereotypic movements, and pervasive growth failure. In addition to global reductions in both GM and WM demonstrated by volumetric analyses of anatomical brain MRI, preferential reduction in CBF to the frontal lobes is evidenced by ASL MR perfusion imaging (Naidu et al., 2001). This selective vulnerability of the frontal lobes, which was corroborated by MR spectroscopic imaging (MRSI) and 18FDG-PET, is felt

to provide considerable insight into mechanisms underlying the clinical features of this disease. There is no doubt that in the future MR perfusion imaging, in combination with other advanced imaging tools, will play an important role in better understanding of the pathophysiology and preclinical diagnosis of many neurodegenerative disorders affecting children.

Diffusion imaging Background The diffusion phenomenon can be defined as the random translational motion (Brownian motion) of molecules in matter. Diffusion depends on the type of molecule and its environment, and is typically described by a coefficient known as the diffusion coefficient (Le Bihan et al., 1986). DWI allows in vivo measurement of free water diffusion coefficient in human brain and, as importantly, permits generation

Physiological MR of the pediatric brain: overview

(a)

(b)

Fig. 39.13 Same patient as in Figure 39.6. Axial fast FLAIR (TR/TE/TI: 8800/140/2200 (a) demonstrates hyperintensity in the WM of the left front-parietal lobe with mild associated mass effect. There is corresponding increase CBV in the lesion on the axial regional cerebral blood volume (rCBV) map (b) suggestive of malignant neoplasms. On open biopsy a glioblastoma was diagnosed.

of diffusion coefficient brain maps that help localize derangements in specific brain regions. In the brain, the diffusion of free water is influenced by several factors including cellular structure, myelin, and circulation of blood in capillary networks. Hence, what can be measured in the image voxel, usually on the order of millimeters in dimension, is an apparent diffusion coefficient (ADC) that accounts for the effect of these factors on the diffusion of free water in vivo. The principles of diffusion imaging are described in detail in Chapters 4–6. Generation of diffusionweighted images of the brain requires the application of strong diffusion sensitizing magnetic gradients. The degree of diffusion weighting is described by the so-called b-value, a parameter that is determined by the type of gradient scheme implemented. For the Stejskal–Tanner spin-echo scheme (Figure 39.14), a pulsed pair of approximately rectangular gradients around a 180° RF pulse that is most commonly implemented on clinical MR scanners, the b-value is determined by the duration () and strength (G) of the sensitizing pulsed-gradients, and the time

90°

180°

RF

Gx

Gx

Gy

Gy

Gz

Gz

Gx

Gy

Gz d

D

Fig. 39.14 Diagram shows Stejskal–Tanner pulsed bipolar gradient scheme. This scheme is commonly implemented on clinical MR scanners for diffusion sensitization. Degree of diffusion sensitization (b-value) is determined by duration () and strength (height) of sensitizing pulsed-gradients (G), and the time interval between the two pulsed-gradients ().

659

660

Elias R. Melhem and Xavier Golay

(a)

(b) cc slf

cst

ilf Fig. 39.15 Color-coded WM fiber maps (a) and 3D fiber tracts (b) are generated based on the fractional anisotropy (FA) and vector maps. Corpus callosum (cc), superior longitudinal fasciculus (slf), inferior longitudinal fasciculus (ilf ), and cortico-spinal tract (cst).

interval between the two pulsed-gradients () according to: b-value  2G22( /3) (Melhem et al., 2002). Acquiring diffusion-weighted images with at least two different b-values (commonly 20 and 1000 s/mm2) while keeping the TE fixed allows the determination of the ADC value for each image voxel. Assigning a gray scale to the range of ADC values in the different voxels constitutes an ADC map. The ADC map provides contrast based on differences in diffusivity of water in biological tissue uncontaminated by differences in T2-relaxation times (T2-shine through) (Melhem et al., 2002). Interestingly, ADC values in WM vary depending on the direction of the diffusion sensitizing gradients (Chenevert et al., 1990; Doran et al., 1990; Moseley et al., 1990). This observation can be explained by the fact that diffusion of water molecules in WM is not the same in all directions of a 3D space (i.e. the diffusion is anisotropic). Diffusion anisotropy is predominantly caused by the orientation of fiber tracts in WM and is influenced by its micro- and macrostructural features. Of the microstructural features, intra-axonal organization appears to be of greatest influence on diffusion anisotropy, others include density of fiber and neuroglial cell packing, degree of myelination and individual fiber diameter. On a macroscopic scale, the variability in the orientation of all WM tracts within an imaging

voxel influences the degree of anisotropy assigned to that voxel (Basser et al., 1994a, 1994b). In a first approximation, one can assign a diffusion tensor to describe diffusion anisotropy in WM by characterizing the surface of an ellipsoid, which represents the root mean square diffusive displacement of free water in space. To characterize the diffusion ellipsoid, diffusion-sensitizing gradients are applied in at least six non-colinear directions (e.g. xx, yy, zz, xy, xz, and yz) (Basser et al., 1994b). This is followed by any kind of fast imaging readouts to obtain what is known as DTI. In addition to providing several important measures of diffusion anisotropy, such as fractional anisotropy (FA), relative anisotropy (RA), and volume ratio, DTI is the only method that allows in vivo mapping of WM tracts responsible for brain connectivity (Figure 39.15) (Conturo et al., 1999; Stieltjes et al., 2001; Mori et al., 2002). Applications There are several reported applications of DWI and DTI in the assessment of normal brain development, hypoxic–ischemic injury, neoplasms, infections, cerebral palsy, congenital malformations, and inborn errors of metabolism that affect the central nervous system (CNS). These are also discussed further in the following chapters.

Physiological MR of the pediatric brain: overview

(a)

(b)

(c)

Fig. 39.16 Term neonate with hypoxic–ischemic encaphalopathy. Axial fast T2-weighted (TR/TE: 4500/120) (a) demonstrates subtle loss of GM ribbon in both frontal and parietal lobes. Corresponding axial DWI (b  1000 s/mm2) (b) and ADC map (c) there is prominent decrease diffusion in both frontal and parietal lobes involving GM and WM compatible with infarctions in a watershed distribution in term neonates.

Brain development Following birth, ADC and anisotropy measures in WM change in fashion commensurate with the expected process of myelination (Mukherjee et al., 2001, 2002). Specifically, ADC values decrease while the anisotropy measures increase during the first 6 months in a pattern similar to but preceding the appearance of myelin in WM fibers. DWI may prove to be a more accurate tool for studying brain development than conventional MRI. Hypoxia–ischemia Recent studies examining the role of DWI in term neonatal hypoxic–ischemic encephalopathy show that DWI better demonstrate and define the extent of brain injury than conventional MRI (Figure 39.16). The implications of this finding are that DWI may be the technique of choice for defining diagnostic entry criteria for future clinical studies of neonatal encephalopathy, and for determining long-term neurological outcome (McKinstry et al., 2002). Unfortunately, however, the ability of DWI to reliably demonstrate the effects of hypoxic– ischemic encephalopathy on the brain within 24 h of the insult remains questionable (Melhem, 2002). This is especially true when the brain injury is limited to deep GM nuclei where MRSI may be more

sensitive. On the other hand, similar to the adult experience, DWI is extremely sensitive for the early demonstration of hyperacute thromboembolic brain infarctions in neonates and young children (Figure 39.17). In preterm neonates, where acute periventricular leukomalacia (PVL) is the principal form of brain injury, again DWI is reported to better demonstrate and define the extent of brain injury compared to cranial ultrasonography and conventional MRI (Figure 39.18) (Inder et al., 1999). Lastly, in our limited experience, DWI has proven helpful in differentiating irreversible brain injury from reversible edema secondary to venous infarction in the neonate (Figure 39.19). This observation, however, warrants further investigation (Bernstein and Albers, 2001). Brain neoplasms DWI can be useful in separating solid from cystic or necrotic portions of brain neoplasms (Figure 39.20). Typically, in the cystic or necrotic portions the ADC values are extremely elevated and approach that of CSF (Le Bihan et al., 1993). Another important application of DWI is in differentiating intracranial epidermoids from arachnoid cysts, both of which can be isointense to CSF on conventional MRI

661

662

Elias R. Melhem and Xavier Golay

(a)

(b)

(c)

Fig. 39.17 Term neonate imaged 8 hrs after a left middle cerebral artery (MCA) territory infarction. Axial fast T2-weighted (TR/TE: 4500/120) (a) demonstrates hyperintensity in the left basal ganglia and cerebral hemisphere with loss of GM ribbon. Corresponding axial DWI (b  1000 s/mm2) (b) and ADC map (c) there is prominent decrease diffusion in the left basal ganglia and cerebral hemisphere confirming infarction.

(Tsuruda et al., 1990) (see also Case study 20.1). Typically, ADC values from epidermoids are slightly reduced compared to brain parenchyma (Figure 39.21), whereas ADC values from arachnoid cysts approach that of CSF. Finally, based on the recent observation of a strong negative correlation between tumor nuclear-to-cytoplasmic ratio and ADC, DWI can add histological specificity to conventional MRI (Guo et al., 2002). This has been best demonstrated by the low ADC values in primative neuroectodermal tumor (PNET) and primary CNS lymphoma compared to other brain neorplasms (Figures 39.22 and 39.23).

Fig. 39.18 Preterm neonate with perinatal asphyxia. Axial ADC map demonstrates decrease diffusion in the periventricular WM (arrows) compatible with acute infarction in a watershed distribution for preterm neonates.

Brain infections and abscesses Children are at risk for developing brain abscesses secondary to infectious meningitis, direct extension from infections involving the paranasal sinuses and mastoid/middle ear apparatus, or hematogenous spread of infection particularly in children with pulmonary arteriovenous malformations (AVMs) or congenital heart malformations. DWI is very specific in the diagnosis of pyogenic brain abscesses and differentiating them from other enhancing masses with necrotic centers. As opposed to neoplasms, the necrotic portion of pyogenic abscesses has very low

Physiological MR of the pediatric brain: overview

(a)

(b)

(c)

Fig. 39.19 A 6-month-old infant with acute central venous sinus thrombosis. Contrast-enhanced sagittal T1-weighted (TR/TE: 550/10) (a) demonstrates acute thrombus filling and enlarging the vein of Galen and straight sinus. There is also well-defined hypointense lesion in the thalamus. On DWI (b  1000 s/mm2) (b) and ADC map (c) there is prominent decrease diffusion in both thalami surrounded by increase diffusion extending into the internal capsules and basal ganglia consistent with venous “infarction”. On follow up MRI (not shown) brain injury is limited only to regions of decrease diffusion.

(a)

(b)

(c)

Fig. 39.20 A 9-year-old girl with cerebellar pilocytic astrocytoma. Contrast-enhanced axial T1-weighted (TR/TE: 550/10) (a) demonstrates heterogeneously enhancing mass arising from the cerebellar vermis compressing the fourth ventricle. It is difficult to decide based on this image whether the non-enhancing component of the mass is solid or cystic. DWI (b  1000 s/mm2) (b) and ADC map (c) clearly separate the solid (mild increase in diffusion) from cystic component (marked increase in diffusion) in the right-posterior portion of the mass.

663

664

Elias R. Melhem and Xavier Golay

(a)

(b)

(c)

Fig. 39.21 A 17-year-old male with recurrent epidermoid in the right suprasellar cistern. Contrast-enhanced coronal T1-weighted (TR/TE: 550/10) (a) demonstrates non-enhancing hypointense mass in right suprasellar cistern deviating the pituitary stalk. The question in this case was whether the mass was due to post-operative arachnoid cyst or recurrent epidermoid. DWI (b  1000 s/mm2) (b and c) demonstrates a hyperintense mass strongly suggestive of recurrent epidermoid.

(a)

(b)

(c)

Fig. 39.22 A 5-year-old boy with medulloblastoma. Axial T1-weighted (TR/TE: 550/10) (a) and axial fast T2-weighted (TR/TE: 4500/80) (b) demonstrates hemorrhagic mass with heterogeneous signal intensity arising from the cerebellar vermis compressing the fourth ventricle. ADC map (c) demonstrates mild decrease in diffusion compared to normal brain parenchyma in the non-hemorrhagic portions of the mass attributed to high cellularity and nuclear-to-cytoplasmic ratio.

ADC values, which has been attributed to the restrictive effect of pus on the diffusivity of free water molecules (Figure 39.24) (Desprechins et al., 1999). Cerebral palsy Cerebral palsy has been divided based on the nature of the motor deficits into pyramidal, extrapyramidal, and mixed types. The pathological basis for cerebral

palsy and commonly associated cognitive impairment may be subtle and limited to disruptions in the integrity of WM. Recent studies (Hoon et al., 2002) using DTI and WM tractography confirm the importance of connections between the thalamus and somatosensory cortex (posterior thalamic radiations) in the regulation of walking in children with cerebral palsy (Figure 39.25). Further work is

Physiological MR of the pediatric brain: overview

(a)

(b)

(c)

Fig. 39.23 A 24-year-old female with primary CNS lymphoma. Axial fast FLAIR (TR/TE/TI: 8800/140/2200) (a) demonstrates mildly hyperintensity mass in the right frontal lobe and corpus callosum with surrounding vasogenic edema. Contrast-enhanced axial T1-weighted (TR/TE: 550/10) (b) demonstrates marked homogeneous enhancement of the mass. ADC map (c) demonstrates mild decrease in diffusion compared to normal brain parenchyma in the mass attributed to high cellularity and nuclear-to-cytoplasmic ratio.

(a)

(b)

(c)

Fig. 39.24 An 8-year-old girl with chronic ear infections presents with headaches. Contrast-enhanced sagittal T1-weighted (TR/TE: 550/10) (a) demonstrates ring-enhancing mass in the right temporal lobe as well as thick dural enhancement underneath the mass. Axial DWI (b  1000 s/mm2) (b) and ADC map (c) show decreased diffusion in the necrotic center of the mass strongly suggestive of pus in a brain abscess.

currently underway to study the effect and degree of injury to specific WM tracts on motor and cognitive function in children with cerebral palsy. Congenital brain malformations DTI has been reported in holoprosencephaly (HPE), a term that defines a spectrum of complex congenital

brain and facial malformations resulting from defective dorso-ventral patterning and cleavage of the developing forebrain during the first 5 weeks of embryonic development. One study shows that DTI may better identify abnormalities of specific WM tracts in the brainstem than conventional MRI (Albayram et al., 2002). Specifically, DTI-based

665

666

Elias R. Melhem and Xavier Golay

PLIC Lateral (b)

(a)

lv lv

PTR Lateral (c)

(d)

Fig. 39.25 Three-dimensional WM fiber tracts showing the anterior fibers and posterior fibers of the posterior limb of the internal capsule (PLIC) in a control subject and a 6-year old boy with spastic quadriplegia secondary to PVL. In the lateral view (lv), the anterior fibers of the PLIC, which normally contain cortico-spinal fibers, can clearly be seen in the control (a) and patient (b), but posterior fibers in the PLIC (arrow) are not visible in the patient. The posterior thalamic radiation (PTR), which projects through the retrolenticular division in the ternal capsule, is visible in the control (c) but appears severely attenuated in the patient (d).

brainstem WM tract maps demonstrated absence of the cortico-spinal tracts in two children with alobar HPE (Figure 39.26), corroborating what has been previously documented on autopsy series by neuropathologists. The role of DTI in deciphering abnormal WM connections in other congenital brain malformations remains to be explored. Inborn errors of metabolism DTI has been used to evaluate children with specific inborn errors of metabolism that target

the brain and its WM. In X-linked adrenoleukodystrophy (ALD), a peroxisomal disorder characterized by rapid demyelination of cerebral WM, DTI may better define the extent of WM involvement, predict disease activity and progression, and categorize affected WM based on well-defined histopathological zones (Figure 39.27). Preliminary results demonstrate that affected WM can be divided into three distinct zones based on differences in diffusion values which may reflect varying degrees of axonal and myelin loss (Table 39.1) (Ito et al., 2001; Eichler et al., 2002).

Physiological MR of the pediatric brain: overview

(a)

(b)

Fig. 39.26 An 8-month-old girl with alobar HPE. Axial fast T2-weighted (5000/92: TR/TE) (a) and corresponding axial WM color map (b) generated using DTI at the level of the pons (arrowheads). On the color map the cortico-ponto-spinal tract cannot be identified in the basis pontis (anterior to the solid white line) despite the apparently normal configuration of the pons on the T2-weighted image.

(a)

(b)

(c)

Fig. 39.27 A 13-year-old boy with X-linked adrenoleukodystrophy (ALD). Axial FA map (a), fast T2-weighted (b), and isotropic ADC maps (c) at the level of the lateral ventricles show reduction of FA, T2-hyperintensity, and increase of ADC in the deep WM of both parieto-occipital lobes and corpus callosum. Note the differential in the reduction of FA and the increase of ADC between the central and peripheral zones of the WM abnormality.

Furthermore, initial observations suggest that the presence of reduced ADC values at the leading edge of the WM lesion predicts rapid progression of the disease process over a short period of time. In Krabbe disease (KD), a lysosomal disorder affecting the central and peripheral nervous system,

DTI has been reported to provide a quantitative measure of abnormal WM, which can be used as a marker to monitor response to treatment (Guo et al., 2001). DTI has also been shown to better demonstrate WM abnormality than conventional T2-weighted MRI.

667

668

Elias R. Melhem and Xavier Golay

Table 39.1. Averages and standard deviations (SD) of ADCi (10 3 mm2/s) and FA values in normal appearing white matter (NAWM) and three well-defined histological zones in 11 boys with cerebral X-linked ALD (Ito et al., 2001) Histological zones

ADCi (SD)

FA (SD)

NAWM Area A Area B Area C

0.66 (0.12) 0.84 (0.16) 1.36 (0.19) 1.73 (0.15)

0.52 (0.16) 0.31 (0.12) 0.22 (0.07) 0.13 (0.05)

Fast spatial encoding for imaging the pediatric brain It is clear that MR imaging has become an exciting window through which one peeks into the anatomy, physiology, and chemistry of the living brain. Clinical physiological MR allows radiologists to be more sensitive in detecting and defining the extent of disease, and more specific regarding the type of disease affecting the pediatric brain. It is tempting to implement as many of these techniques as possible in the hope of providing the most accurate answer to the clinical problem at hand. However, the penalty with this comprehensive approach is longer scanning times, which are particularly problematic in the pediatric population, where remaining still for long periods is unrealistic and prolonged sedation carries risk. Methods which reduce the acquisition time of each of these advanced MR techniques increase their applicability in routine clinical practice. Since the inception of clinical MRI, steady improvements in magnetic field gradient performance on clinical scanners have allowed progressively faster acquisitions, which have facilitated routine applications in children. However, the limits for improvement in imaging based on gradient spatial encoding are quickly approaching; in part due to physiological limitations such as neuromuscular stimulation from rapid switching of encoding gradients and levels of energy deposition from dense RF pulse trains. The advent of parallel imaging techniques (Sodickson et al., 1997; Pruessmann et al., 2001) potentially provides very significant further reductions in acquisition time.

Parallel imaging exploits information related to the distinct spatial sensitivities of coil array elements; a large number of surface coils each independently and simultaneously imaging a given volume potentially negate the need for time-consuming gradient encoding steps (Hutchinson and Raff, 1988). In practice, hybrid spatial encoding, based on the combined use of the sensitivity of multiple receivers together with gradient encoding, reduces scan time by requiring fewer phase encoding steps whilst maintaining their maximum values. At the time of writing, applications of parallel MR imaging allow up to a six-fold reduction (limited by the number of surface coils operating in parallel) in scan acquisition time (van den Brink et al., 2003). This approach leads to reduction in scan time, signal-to-noise ratio (SNR), and field-of-view (FOV) while preserving spatial resolution (Pruessmann et al., 2001). Reduced FOV normally leads to an image with strong aliasing artifacts, however, using the spatial sensitivity maps of each surface coil, this aliasing effect is undone and a full FOV image is reconstructed (Figure 39.28). Successful implementation of sensitivity-encoding to comprehensive MR imaging of the pediatric brain requires: (a) specially designed multiple surface coils operating in a phased array configuration to maximize signal and to minimize constructive interaction of noise coming from the coil elements; (b) parallel imaging reconstruction software which operates either in frequency domain (SiMultaneous Acquisition of Spatial Harmonics, SMASH) preceding discrete Fourier transformation (Sodickson et al., 1997) or in image domain (SENSitivity Encoding, SENSE) following DFT (Pruessmann et al., 2001); and (c) incorporating sensitivity encoding schemes

Physiological MR of the pediatric brain: overview

Body coil reference Array coil references (a)

SENSE acquisition (d)

(b)

Sensitivity maps

Sense reconstruction (e)

(c)

Fig. 39.28 Schematic representation of sensitivity-encoding reconstruction in image domain (SENSE). Reference image acquired with the body coil (a). Full-FOV reference images acquired with each array element (b). Sensitivity maps, obtained from (a) and (b) (c). Conventional sum-of-squares representation of multiple coil SENSE data (d). The same data as in (d) after SENSE reconstruction using the sensitivity maps (c).

(a)

(b)

(c)

Fig. 39.29 WM color maps from a volunteer generated in the axial (a), sagittal (b), and coronal (c) planes using a high-resolution (256 256 matrix) DTI sequence showing superior–inferior projecting tracts in red, anterior–posterior tracts in blue, and left–right tracts in green. Implementing a SENSE reduction factor of 2, this whole brain DTI sequence required 12 min instead of the typical 30-min acquisition.

into morphological and functional MR imaging sequences. Several MR device-manufacturing companies now produce dedicated brain phased array coils for parallel imaging, and the major scanner vendors are increasingly incorporating either SENSE or SMASH into their latest imaging platforms.

Furthermore, research groups, including ours, have been successful in significantly enhancing the speed of acquisition of proton MRSI (Dydak et al., 2001), PI (van den Brink et al., 2003) and DTI (Bammer et al., 2001, 2002) with sensitivity-encoding schemes (Figure 39.29).

669

670

Elias R. Melhem and Xavier Golay

Table 39.2. Summary of comprehensive brain MRI protocol Sequence

1. Localizer 2. SENSE-Ref 3. T1-wa 4. Dual-PD-T2-wa 5. FLAIRa 6. T1-w localizer 7. Field mapb 8. MRSIb 9. Field mapf 10. ASLc 11. T1-mape 12. DTI 13. High Res 3D

Scan plane

Type

TR/TE (ms)

FOV (cm)

Matrixg

Thickness/ gap (mm)

Slices

SENSE Factor

Scan time

Three planes Coronal Axial Axial Axial Axial Axial Axial Axial Axial Axial Coronal Coronal

GRE

40/5.2

25 25

256 128

10/10

3/1/1

1

15 sec

3D-GRE SE FSE FSE SE Dual GRE MRSI Dual GRE EPI 3D-EPI SE-EPI 3D-SPGR

3.6/0.7 582/15 2000/(7.1/100) 11,000/140 300/14 250/(10/20) 5200/280 250/(10/20) 17/8.6 17/8.6 10,000/90 35/6.0

30 30 23 18 23 18 23 18 23 18 23 18 23 18 23 18 29 20 29 20 23 17 23 18

48 48 256 165 256 165 256 165 128 82 128 82 28 20 128 82 64 50 64 50 128 95 256 82

7.5/0.0 5.0/1.0 5.0/1.0 5.0/1.0 15/2.5 15/2.5 15/2.5 15/2.5 4.5/0.0 4.5/0.0 3.0/0 1.5/0.0

40 22 22 22 8d 8d 8d 8d 26 26 62 124

1 1 1 1 1 1 2 2 1 2 2 2 2 1.5

26 min 1 min 36 sec 1 min 16 sec 2 min 12 sec 26 sec 22 sec 10 min 22 sec 5 min 15 sec 1 min 2 sec 15 sec 3 min 30 sec

Total

41 min 42 sec

None of the diagnostic sequences (3–5) will be performed using a SENSE factor 1, since it would degrade the image quality too much. However, these sequences can be performed with one average only, because of the high signal-to-noise gain from the phased-array coil. b All preparation scans, as well as the field map scan will be done using the body coil. There is an additional 5 minutes preparation time for the MRSI sequence. c For the ASL sequence the time between two acquisitions is 5 s, and therefore 10 s/pair of labeled-control scans. Thirty averages will be collected. d It is assumed that eight slices are enough to cover the whole brain. e For the T1-maps, two averages will be acquired for increased precision. Identical readouts will be used in both ASL and T1-map, as the results of one will be applied to the other. f The second field map is required for remaining susceptibility corrections. g All matrix sizes are given after SENSE reconstruction. a

SENSE-based MR imaging protocol

REFERENCES

We would like to conclude by introducing a comprehensive 45-minutes-long advanced MR imaging protocol that allows the delineation of microstructure, and mapping of perfusion parameters and metabolites of the pediatric brain. It combines routine clinical imaging, high-resolution anatomical and physiological imaging, and is made feasible by using parallel methods (in this example SENSE). The protocol includes: T1-weighted, PD-weighted, T2-weighted, FLAIR, high resolution 3D imaging, whole brain DTI, whole brain ASL perfusion imaging, and whole brain proton MRS imaging (Table 39.2).

Aasly J, Silfvenius H, Aas TC, Sonnewald U, Olivecrona M, Juul R, White LR. 1999. Proton magnetic resonance spectroscopy of brain biopsies from patients with intractable epilepsy. Epilepsy Res 35: 211–217. Albayram S, Melhem ER, Mori S, Zinreich SJ, Barkovich AJ, Kinsman SL. 2002. Holoprosencephaly in children: diffusion tensor MR imaging of white matter tracts of the brainstem-initial experience. Radiology 223: 645–651. Alsop DC, Detre JA. 1996. Reduced transit-time sensitivity in noninvasive magnetic resonance imaging of human cerebral blood flow. J Cereb Blood Flow Metab 16: 1236–1249.

Physiological MR of the pediatric brain: overview

Alsop DC, Detre JA. 1998. Multisection cerebral blood flow MR imaging with continuous arterial spin labeling. Radiology 208: 410–416. Axel L. 1980. Cerebral blood flow determined by rapidsequence computed tomography. Radiology 137: 679–686. Ball Jr WS, Holland SK. 2001. Perfusion imaging in the pediatric patient. Magn Reson Imaging Clin N Am 9: 207–230. Bammer R, Keeling SL, Augustin M, et al. 2001. Improved diffusion-weighted single-shot echo-planar imaging (EPI) in stroke using sensitivity encoding (SENSE). Magn Reson Med 46: 548–554. Bammer R, Auer M, Keeling SL, et al. 2002. Diffusion tensor imaging using single-shot SENSE-EPI. Magn Reson Med 48: 128–136. Barbier EL, Lamalle L, Decorps M. 2001. Methodology of brain perfusion imaging. J Magn Reson Imaging 13: 496–520. Basser PJ, Mattiello J, Le Bihan D. 1994a. MR diffusion tensor spectroscopy and imaging. Biophys J 66: 259–267. Basser PJ, Mattiello J, LeBihan D. 1994b. Estimation of the effective self-diffusion tensor from the NMR spin echo. J Magn Reson B 103: 247–254. Bernstein R, Albers GW. 2001. Potential utility of diffusionweighted imaging in venous infarction. Arch Neurol 58: 1538–1539. Cecil KM, Jones BV. 2001. Magnetic resonance spectroscopy of the pediatric brain. Top Magn Reson Imaging 12: 435–452. Cecil KM, DeGrauw TJ, Salomons GS, Jakobs C, Egelhoff JC, Clark JF. 2003. Magnetic resonance spectroscopy in a 9-dayold heterozygous female child with creatine transporter deficiency. J Comput Assist Tomogr 27: 44–47. Cha S, Knopp EA, Johnson G, Wetzel SG, Litt AW, Zagzag D. 2002. Intracranial mass lesions: dynamic contrastenhanced susceptibility-weighted echo-planar perfusion MR imaging. Radiology 223: 11–29. Chenevert TL, Brunberg JA, Pipe JG. 1990. Anisotropic diffusion in human white matter: demonstration with MR technique in vivo. Radiology 177: 401–405. Conturo TE, Lori NF, Cull TS, et al. 1999. Tracking neuronal fiber pathways in the living human brain. Proc Natl Acad Sci USA 96: 10422–10427. Desprechins B, Stadnik T, Koerts G, Shabana W, Breucq C, Osteaux M. 1999. Use of diffusion-weighted MR imaging in differential diagnosis between intracerebral necrotic tumors and cerebral abscesses. Am J Neuroradiol 20: 1252–1257. Detre JA, Zhang W, Roberts DA, et al. 1994. Tissue specific perfusion imaging using arterial spin labeling. NMR Biomed 7: 75–82. Doran M, Hajnal JV, van Bruggen N, King MD, Young IR, Bydder GM. 1990. Normal and abnormal white matter tracts shown by MR imaging using directional

diffusion weighted sequences. J Comput Assist Tomogr 14: 865–873. Dydak U, Weiger M, Pruessmann KP, Meier D, Boesiger P. 2001. Sensitivity-encoded spectroscopic imaging. Magn Reson Med 46: 713–722. Edelman RR, Siewert B, Darby DG, et al. 1994. Qualitative mapping of cerebral blood flow and functional localization with echo-planar MR imaging and signal targeting with alternating radio frequency. Radiology 192: 513–520. Eichler FS, Itoh R, Barker PB, Mori S, Garrett ES, van Zijl PC, Moser HW, Raymond GV, Melhem ER. 2002. Proton MR spectroscopic and diffusion tensor brain MR imaging in Xlinked adrenoleukodystrophy: initial experience. Radiology 225: 245–252. Go KG, Lammertsma AA, Paans AMJ, Vaalburg W, Woldring MG. 1981. Extraction of water labeled with oxygen 15 during single-capillary transit: influence of blood pressure, osmolarity, and blood–brain barrier damage. Arch Neurol 38: 581–584. Golay X, Stuber M, Pruessmann KP, Meier D, Boesiger P. 1999. Transfer insensitive labeling technique (TILT): application to multislice functional perfusion imaging. J Magn Reson Imaging 9: 454–461. Graham GD, Blamire AM, Rothman DL, Brass LM, Fayad PB, Petroff OA, Prichard JW. 1993. Early temporal variation of cerebral metabolites after human stroke. A proton magnetic resonance spectroscopy study. Stroke 24: 1891–1896. Grodd W, Krageloh-Mann I, Petersen D, Trefz FK, Harzer K. 1990. In vivo assessment of N-acetylaspartate in brain in spongy degeneration (Canavan’s disease) by proton spectroscopy. Lancet 336: 437–438. Grubb RL, Raichle ME, Higgins CS, Eichling JO. 1978. Measurement of regional cerebral blood volume by emission tomography. Ann Neurol 4: 322–328. Guo AC, Petrella JR, Kurtzberg J, Provenzale JM. 2001. Evaluation of white matter anisotropy in Krabbe disease with diffusion tensor MR imaging: initial experience. Radiology 218: 809–815. Guo AC, Cummings TJ, Dash RC, Provenzale JM. 2002. Lymphomas and high-grade astrocytomas: comparison of water diffusibility and histologic characteristics. Radiology 224: 177–183. Hoon Jr AH, Lawrie Jr WT, Melhem ER, Reinhardt EM, Van Zijl PC, Solaiyappan M, Jiang H, Johnston MV, Mori S. 2002. Diffusion tensor imaging of periventricular leukomalacia shows affected sensory cortex white matter pathways. Neurology 59: 752–756. Hunter JV, Wang ZJ. 2001. MR spectroscopy in pediatric neuroradiology. Magn Reson Imaging Clin N Am 9: 165–189.

671

672

Elias R. Melhem and Xavier Golay

Hutchinson M, Raff U. 1988. Fast MRI data acquisition using multiple detectors. Magn Reson Med 6: 87–91. Inder T, Huppi PS, Zientara GP, Maier SE, Jolesz FA, di Salvo D, Robertson R, Barnes PD, Volpe JJ. 1999. Early detection of periventricular leukomalacia by diffusion-weighted magnetic resonance imaging techniques. J Pediatr 134: 631–634. Ito R, Melhem ER, Mori S, Eichler FS, Raymond GV, Moser HW. 2001. Diffusion tensor brain MR imaging in X-linked cerebral adrenoleukodystrophy. Neurology 56: 544–547. Jahng GH, Zhu XP, Matson GB, Weiner MW, Schuff N. 2003. Improved perfusion-weighted MRI by a novel double inversion with proximal labeling of both tagged and control acquisitions. Magn Reson Med 49: 307–314. Kim SG. 1995. Quantification of relative cerebral blood flow change by flow-sensitive alternating inversion recovery (FAIR) technique: application to functional mapping. Magn Reson Med 34: 293–301. Kirkham FJ, Calamante F, Bynevelt M, et al. 2001. Perfusion magnetic resonance abnormalities in patients with sickle cell disease. Ann Neurol 49: 477–485. Kreis R, Ernst T, Ross BD. 1993. Development of the human brain: in vivo quantification of metabolite and water content with proton magnetic resonance spectroscopy. Magn Reson Med 30: 424–437. Kwong KK, Chesler DA, Weisskoff RM, et al. 1995. MR perfusion studies with T1-weighted echo planar imaging. Magn Reson Med 34: 878–887. Le Bihan D, Breton E, Lallemand D, Grenier P, Cabanis E, Laval-Jeantet M. 1986. MR imaging of intravoxel incoherent motions: application to diffusion and perfusion in neurologic disorders. Radiology 161: 401–407. Le Bihan D, Douek P, Argyropoulou M, Turner R, Patronas N, Fulham M. 1993. Diffusion and perfusion magnetic resonance imaging in brain tumors. Top Magn Reson Imaging 5: 25–31. Li LM, Cendes F, Bastos AC, Andermann F, Dubeau F, Arnold DL. 1998. Neuronal metabolic dysfunction in patients with cortical developmental malformations: a proton magnetic resonance spectroscopic imaging study. Neurology 50: 755–759. McKinstry RC, Miller JH, Snyder AZ, Mathur A, Schefft GL, Almli CR, Shimony JS, Shiran SI, Neil JJ. 2002. A prospective, longitudinal diffusion tensor imaging study of brain injury in newborns. Neurology 59: 824–833. Meier P, Zierler KL. 1954. On the theory of the indicatordilution method for measurement of blood flow and volume. J Appl Physiol 6: 731–744. Melhem ER. 2002. Time-course of apparent diffusion coefficient in neonatal brain injury: the first piece of the puzzle. Neurology 59: 798–799.

Melhem ER, Mori S, Mukundan G, Kraut MA, Pomper MG, van Zijl PC. 2002. Diffusion tensor MR imaging of the brain and white matter tractography. Am J Roentgenol 178: 3–16. Miller BL. 1991. A review of chemical issues in 1H NMR spectroscopy: N-acetyl-L-aspartate, creatine and choline. NMR Biomed 4: 47–52. Mori S, Frederiksen K, van Zijl PC, et al. 2002. Brain white matter anatomy of tumor patients evaluated with diffusion tensor imaging. Ann Neurol 51: 377–380. Moseley ME, Cohen Y, Kucharczyk J, et al. 1990. Diffusionweighted MR imaging of anisotropic water diffusion in cat central nervous system. Radiology 176: 439–445. Mukherjee P, Miller JH, Shimony JS, Conturo TE, Lee BC, Almli CR, McKinstry RC. 2001. Normal brain maturation during childhood: developmental trends characterized with diffusion-tensor MR imaging. Radiology 221: 349–358. Mukherjee P, Miller JH, Shimony JS, Philip JV, Nehra D, Snyder AZ, Conturo TE, Neil JJ, McKinstry RC. 2002. Diffusion-tensor MR imaging of gray and white matter development during normal human brain maturation. Am J Neuroradiol 23: 1445–1456. Naidu S, Kaufmann WE, Abrams MT, et al. 2001. Neuroimaging studies in Rett syndrome. Brain Dev 23(suppl. 1): S62–71. Novotny Jr EJ, Avison MJ, Herschkowitz N, Petroff OA, Prichard JW, Seashore MR, Rothman DL. 1995. In vivo measurement of phenylalanine in human brain by proton nuclear magnetic resonance spectroscopy. Pediatr Res 37: 244–249. Oguz KK, Golay X, Pizzini FB, Freer CA, Winrow N, Ichord R, Casella JF, Van Zijl PC, Melhem ER. 2003. Sickle cell disease: continuous arterial spin-labeling perfusion MR imaging in children. Radiology 227: 567–574. Osborne D, Jaszczak R, Coleman RE, Drayer B. 1981. Single photon emission computed tomography in the canine lung. J Comput Assist Tomogr 5: 684–689. Østergaard L, Sorensen AG, Kwong KK, Weisskoff RM, Gyldensted C, Rosen BR. 1996. High resolution measurement of cerebral blood flow using intravascular tracer bolus passages. Part II: Experimental comparison and preliminary results. Magn Reson Med 36: 726–736. Pouwels PJ, Frahm J. 1998. Regional metabolite concentrations in human brain as determined by quantitative localized proton MRS. Magn Reson Med 39: 53–60. Pruessmann KP, Weiger M, Scheidegger MB, Boesiger P. 1999. SENSE: sensitivity encoding for fast MRI. Magn Reson Med 42: 952–962. Pu Y, Li QF, Zeng CM, Gao J, Qi J, Luo DX, Mahankali S, Fox PT, Gao JH. 2000. Increased detectability of alpha brain glutamate/glutamine in neonatal hypoxic–ischemic encephalopathy. Am J Neuroradiol 21: 203–212. Rabinov JD, Lee PL, Barker FG, Louis DN, Harsh GR, Cosgrove GR, Chiocca EA, Thornton AF, Loeffler JS, Henson JW,

Physiological MR of the pediatric brain: overview

Gonzalez RG. 2002. In vivo 3-T MR spectroscopy in the distinction of recurrent glioma versus radiation effects: initial experience. Radiology 225: 871–879. Raichle ME, Martin WRW, Herscovitch P, Mintun MA, Markham J. 1983. Brain blood flow measured with intravenous H215O. II. Implementation and validation. J Nucl Med 24: 790–798. Rhodes CG, Lenzi GL, Frackowiak RSJ, Jones T, Pozzilli C. 1981. Measurement of CBF and CMRO2 using the continuous inhalation of C15O2 and 15O2. J Neurol Sci 50: 381–389. Salibi NM, Brown MA. 1998. Clinical MR Spectroscopy: First principles. New York, Wiley-Liss. Sodickson DK, Manning WJ. 1997. Simultaneous acquisition of spatial harmonics (SMASH): fast imaging with radiofrequency coil arrays. Magn Reson Med 38: 591–603. Stieltjes B, Kaufmann WE, van Zijl PC, et al. 2001. Diffusion tensor imaging and axonal tracking in the human brainstem. Neuroimage 14: 723–735. Sutton LN, Wehrli SL, Gennarelli L, Wang Z, Zimmerman R, Bonner K, Rorke LB. 1994. High-resolution 1H-magnetic resonance spectroscopy of pediatric posterior fossa tumors in vitro. J Neurosurg 81: 443–448. Trinh BC, Melhem ER, Barker PB. 2001. Multi-slice proton MR spectroscopy and diffusion-weighted imaging in methylmalonic acidemia: report of two cases and review of the literature. Am J Neuroradiol 22: 831–833. Tsuruda JS, Chew WM, Moseley ME, Norman D. 1990. Diffusion-weighted MR imaging of the brain: value of differentiating between extraaxial cysts and epidermoid tumors. Am J Neuroradiol 11: 925–931.

Tzika AA, Zarifi MK, Goumnerova L, Astrakas LG, Zurakowski D, Young-Poussaint T, Anthony DC, Scott RM, Black PM. 2002. Neuroimaging in pediatric brain tumors: Gd-DTPAenhanced, hemodynamic, and diffusion MR imaging compared with MR spectroscopic imaging. Am J Neuroradiol 23: 322–333. van den Brink JS, Watanabe Y, Kuhl CK, et al. 2003. Implications of SENSE MR in routine clinical practice. Eur J Radiol 46: 3–27. Villringer A, Rosen BR, Belliveau JW, et al. 1988. Dynamic imaging with lanthanide chelates in normal brain: contrast due to magnetic susceptibility effects. Magn Reson Med 6: 164–174. Waldrop SM, Davis PC, Padgett CA, Shapiro MB, Morris R. 1998. Treatment of brain tumors in children is associated with abnormal MR spectroscopic ratios in brain tissue remote from the tumor site. Am J Neuroradiol 19: 963–970. Williams DS, Detre JA, Leigh JS, Koretsky AP. 1992. Magnetic resonance imaging of perfusion using spin inversion of arterial water. Proc Natl Acad Sci USA 89: 212–216. Wong EC, Buxton RB, Frank LR. 1997. Implementation of quantitative perfusion imaging techniques for functional brain mapping using pulsed arterial spin labeling. NMR Biomed 10: 237–249. Wong JC, Provenzale JM, Petrella JR. 2000. Perfusion MR imaging of brain neoplasms.Am J Roentgenol 174: 1147–1157. Zhang W, Williams DS, Koretsky AP. 1993. Measurement of rat brain perfusion by NMR using spin labeling of arterial water: in vivo determination of the degree of spin labeling. Magn Reson Med 29: 416–421. Zierler KL. 1965. Equations for measuring blood flow by external monitoring of radioisotopes. Circ Res 16: 309–321.

673

40

Physiological MRI of normal development and developmental delay A. James Barkovich, Pratik Mukherjee and Daniel B. Vigneron Neuroradiology Section, University of California, San Francisco, USA

674

Key points

Summary table

• Metabolite profiles in the fetus and neonates are very different from those of adults; they vary regionally, and change rapidly and nonlinearly with gestational age. • Decrease in apparent diffusion coefficient (ADC) of developing white matter (WM) precedes myelination. • WM ADC is greater than that in gray matter, this difference disappears later in life. • The developing cerebral cortex shows anisotropic diffusion, which diminishes with gestational age and disappears before term. • Brain ADC and diffusion anisotropy (unlike T1- and T2-dependent imaging) continues to evolve well beyond 3 years of age. • MR spectroscopy finding may reveal specific features in some metabolic conditions, but is frequently abnormal but non-specific in developmental delay. • Current data is limited, but diffusion tensor imaging shows promise for correlating WM damage with developmental delay.

• Technique of neonatal and infantile proton MR spectroscopy (MRS) and MRS imaging (MRSI). • Technique of neonatal, infantile, and childhood diffusion tensor imaging (DTI). • Normal brain development, as assessed by proton MRSI. • Normal brain development, as assessed by DTI. • Uses of proton MRS and DTI in neonatal injury. • Uses of proton MRS and DTI in developmental delay. • Early investigations of perfusion imaging in infants and children.

Introduction Imaging is a crucial tool in the evaluation of neonates and infants with neurological abnormalities: the encephalopathic neonate, the epileptic child, and the developmentally delayed infant (Kuzniecky and Jackson, 1995; Barkovich, 1997; Volpe, 2000; Ment et al., 2002; Shevell et al., 2003). Transfontanelle sonography was the non-invasive imaging tool to be used to evaluate the neonate and

Physiological MRI of normal development and developmental delay

was extremely successful, providing a new window for the evaluation of the pathological processes in the preterm and term neonate. Ultrasound was valuable in the assessment of intracranial hemorrhage in premature neonates (Papile et al., 1978), but less successful in the assessment of nonhemorrhagic parenchymal injury in both term and preterm infants (Grant et al., 1983; Schouman-Clays et al., 1993; Rutherford et al., 1994; Volpe, 2000), and of little use after closure of the fontanelles. Therefore, computed tomography (CT) and anatomic MR imaging (MRI) have been investigated as imaging tools for neurological disease in infants. MRI has shown some success (Schouman-Clays et al., 1993; Rutherford et al., 1995, 1996; Barkovich et al., 1998; Inder et al., 1999) in this regard and has been shown to be clearly superior to sonography and CT in the assessment of developmental delay due to structural and metabolic disorders (Raymond et al., 1995; van der Knaap and Valk, 1995). Many children with neurodevelopmental disorders, however, have normal anatomic imaging studies. The ability to assess physiological and biochemical parameters through the use of MR, in particular diffusion tensor imaging (DTI) and proton spectroscopy, has considerably broadened the potential use of MR in the assessment of neurologically impaired neonates and infants. In this chapter we discuss the normal developmental processes of the brain as assessed by these new techniques and then discuss the early application of these techniques in the evaluation of neurologically impaired and developmentally delayed neonates.

Technical aspects of neonatal/infant MRI Transportation/sedation/technical aspects Obviously, no technique is useful if the patient cannot be transported to the MR suite and imaged safely, and high quality images obtained. Transportation and sedation of neonates for MR is critical for obtaining high quality images safely, but is beyond the scope of this chapter. It is discussed in all major pediatric imaging textbooks and readers should become completely familiar with techniques for

sedation and monitoring during routine neonatal and infant MRI before using the techniques discussed in this chapter. Although MR can sometimes be performed in premature infants without sedation, it is nearly impossible to obtain high quality imaging data in term neonates and infants without sedation. Another difficulty in the imaging of neonates, particularly premature neonates, is obtaining adequate signal-to-noise ratio (SNR). Neonates have very small heads; therefore, it is critical to obtain thin imaging sections and small spectroscopy voxels. The unmyelinated brains of neonates, however, have a very high water concentration, resulting in rather poor contrast resolution and relatively little signal. Most commercially available head coils are built for adult heads and, because of the large difference in size and water concentration, are not optimal for the imaging of neonatal heads. Therefore, it is recommended that dedicated pediatric head coils, which are commercially available from many manufacturers, be used if possible. Dedicated coils can increase SNR by 200–300%, resulting in a marked improvement in voxel size (smaller), signal-to-noise (higher), imaging time (shorter), or all of the above. MR spectroscopy MR spectroscopy (MRS) provides the non-invasive detection of a number of important cellular metabolites and has become a valuable research and clinical tool to monitor normal and abnormal metabolism in the pediatric brain (Wang and Zimmerman, 1998; Scarabino et al., 1999; Shevell et al., 1999; Cecil and Jones, 2001; Hunter and Wang, 2001). Both phosphorus and proton MRS have been applied in pediatric studies, with the majority of the studies being proton MRS due to its greater inherent sensitivity. Phosphorus MRS studies have demonstrated the ability to monitor key metabolites such as adenosine triphosphate (ATP), phosphocreatine (PCr), and inorganic phosphate (Pi) in the pediatric brain and to detect significant bioenergetic changes following hypoxic–ischemic brain injury (Azzopardi et al., 1989; Laptook et al., 1989; Peden et al., 1990). While of great biological interest, phosphorus MRS is severely limited in its clinical applicability due to low

675

676

A. James Barkovich, Pratik Mukherjee and Daniel B. Vigneron

inherent sensitivity and its limited availability on clinical MR scanners. Proton MRS can be performed routinely on most clinical MR scanners and has been applied in a wide range of pediatric applications (Wang and Zimmerman, 1998; Scarabino et al., 1999; Shevell et al., 1999; Cecil and Jones, 2001; Hunter and Wang, 2001; Huppi and Inder, 2001) and the techniques are discussed in Chapter 1.

Diffusion tensor MRI DTI is an elaboration of diffusion-weighted imaging (DWI) that allows for the quantitative characterization of the three-dimensional (3D) spatial distribution of water diffusion in each MRI voxel, and is discussed in Chapters 4 and 5.

Physiological imaging of normal brain maturation Proton MRS of normal human brain maturation The neonatal MR spectra are strikingly different from those of adult brain. The N-acetyl aspartate (NAA) resonance is much smaller than the choline (Cho) resonance in the newborn brain, whereas it is typically twice Cho in the adult brain. The metabolite concentrations and ratios in babies change non-linearly with age, and the most rapid changes occur between premature and term newborns (Huppi et al., 1991; Kreis et al., 1993, 2002; Huppi, 2001). Single voxel MRS studies of the developing brain were acquired from specific regions (e.g. occipital cortex, parietooccipital white matter (WM), and thalamus) with voxel sizes of typically 8 cm3 (Huppi et al., 1991; Kreis et al., 1993, 2002; Huppi, 2001). By acquiring water unsuppressed spectra and measuring the relaxation times for each resonance, the absolute quantification of each metabolite relative to water could be calculated providing estimates of concentration (Kreis et al., 1993, 2002). These studies showed that Cho (Cho containing compounds) was significantly higher in concentration by a factor of two as compared to adult values. These studies also showed the NAA was

significantly lower in concentration in newborns than older infants and increased rapidly between 30 and 40 weeks gestational age. A quantitative MRS study demonstrated that although some compounds (such as NAA) were significantly lower and some (such as Cho) higher in premature (32 weeks gestational age) when compared with term infants, the total brain metabolite content was not significantly different (Kreis et al., 2002). See also Chapter 44. The single voxel studies provided important quantitative data but were limited in the assessment of the anatomic variation in metabolite levels provided by MRS. This was due to both the limited spatial coverage of 2–3 voxels and coarse spatial resolution of 5–8 cm3. To address this, 3D MR spectroscopic imaging (MRSI) was applied to detect the spatial distribution of MRS-detectable compounds in premature and term infants (Vigneron et al., 2001). The goals were to test the feasibility of obtaining 3D MRSI in newborns, assess the spatial variations of metabolite levels, and to determine age-dependent differences in MRSI data. The number of spectral voxels (each about 1 cm3) obtained in each examination ranged from 68 to 130 in premature infants and 73 to 204 in the term babies. This study demonstrated the feasibility of detecting the 3D distributions of Cho, creatine (Cr), and NAA resonances in the neonatal brain and significant (P  0.05) spectral differences were detected among anatomic locations and between the premature and term groups. In premature (30–34 weeks postconceptional age), regions that mature earliest, such as the thalamus, demonstrated the highest levels of NAA while later maturing frontal WM showed the lowest (Figure 40.1). The basal ganglia spectra showed the largest increase in NAA between term and premature infants consistent with rapid maturation over this time period (Figure 40.2). This study showed that MRSI can detect “metabolic maturation” in cellular metabolite levels and thus may be an important tool in the assessing of both normal and abnormal cerebral development in the pediatric brain. The significant differences in the metabolite distribution and peak area ratios between the term and preterm infants (Figure 40.2) show that metabolites vary with both topology and with brain maturation (Vigneron et al., 2001). This study also indicates the need for

Physiological MRI of normal development and developmental delay

Cho Cr NAA

Fig. 40.1 Images and spectral array from the brain of a premature, 30 weeks post-conceptional age neonate with normal outcome. Note the marked variation in metabolite levels with location in the brain. The 3D MRSI data were acquired in 17 min with a spatial resolution of 1 cm3. Reprinted from Vigneron et al. (2001).

Basal ganglia

Frontal WM

Cho

Cr

NAA

Term (40 weeks)

Premature (30 weeks)

Thalamus

Fig. 40.2 Representative spectra from 1 cm3 voxels from three regions in the brain of a premature (30 week post-conceptional age) and a term (40 week post-conceptional age) neonate. Note higher relative NAA levels in the thalamus and especially the basal ganglia of the term infant. Note also a similar metabolite pattern in the frontal WM between the premature and term neonate, which is consistent with later maturation of this region. Reprinted from Vigneron et al. (2001). Short TE spectra of neonatal brain also show much higher levels of myo-inositol (ml) than in adults (see Chapter 44, Fig. 44.2).

677

678

A. James Barkovich, Pratik Mukherjee and Daniel B. Vigneron

determining topological and age-matched normative values before attempting to assess metabolic abnormalities in neonates by MRS. Diffusion tensor MRI of normal human brain maturation MRI has become an important clinical tool for the assessment of brain maturation in children. The signal intensity changes on T1- and T2-weighted images during brain development are thought to result from decreases in brain water content and increases in the concentration of macromolecules such as myelin (Barkovich et al., 1988). These maturational processes also cause alterations in brain water diffusion that can be analyzed quantitatively with diffusion MRI. An early study of human brain maturation with diffusion-weighted MRI, measured in two orthogonal directions, reported decreasing brain apparent diffusion coefficient (ADC) and increasing WM anisotropy during the first 6 months of life (Nomura et al., 1994). The development of anisotropy in WM tracts precedes the onset of myelination as detected by T1- and T2-weighted MRI and by histology (Wimberger et al., 1995). This early phase of rising WM anisotropy is known as “premyelination”, and may be due to non-structural factors such as ion channel activity in the developing axolemma (Prayer et al., 2001). Morriss et al. (1999), who examined 30 children ages 1 day to 17 years using diffusionweighted MR images acquired in three orthogonal directions, detected decreasing brain ADC and increasing WM anisotropy over the first 3 years of life. These prior DWI studies were limited by relatively small numbers of subjects and by the small number of diffusion-encoding directions. The anisotropy measurements in these studies underestimate the true diffusion anisotropy, as the majority of anisotropy information lies in the off-diagonal elements of the diffusion tensor, to which orthogonal diffusion encoding in three or fewer directions is not sensitive (Shimony et al., 1999). Moreover, computation of the full diffusion tensor, requiring at least six diffusionencoding directions, is needed to generate rotationally invariant measures of anisotropy. More recently, DTI studies of premature newborns (Huppi et al., 1998; Neil et al., 1998), children during

the first decade of life (Mukherjee et al., 2001, 2002; McGraw et al., 2002), as well as older children and adolescents (Klingberg et al., 1999; Schmithorst et al., 2002) have confirmed that ADC decreases with age in both gray matter (GM) and WM (Figure 40.3), and that rotationally invariant anisotropy increases with age, especially in WM (Figures 40.4 and 40.5). In premature and term newborns, ADC is greater in WM than in GM (Huppi et al., 1998; Neil et al., 1998). This GM–WM difference in ADC gradually disappears during the first year of life, and ADC remains relatively uniform throughout the brain parenchyma into adulthood (Mukherjee et al., 2001). It should be noted that all of these DWI and DTI studies of brain development were performed at b-factors of 1000 s/mm2 or less, and that the measured diffusion parameters may change with b-factors greatly exceeding 1000 s/mm2 (Mulkern et al., 2001). A DTI study of 153 children, ages 1 day to 12 years, demonstrated that changes in ADC and anisotropy extend well beyond the first 3 years of life, even in rapidly maturing regions such as the deep GM nuclei and the central WM tracts of the internal capsule and corpus callosum (Mukherjee et al., 2001). The agedependent changes of ADC in these central GM and WM regions show a bi-exponential time course consisting of a large, rapid early component during the first 2 years life and a smaller, slower late component predominating thereafter. This maturational time course of ADC closely parallels that of changes in brain water content with age, although the 46% drop in ADC between term birth and adulthood (Mukherjee et al., 2001) is much larger than the 12% decrease in brain water content over that interval (Dobbing and Sands, 1973). Hence, the age-dependent reduction of ADC reflects more than just tissue water loss. Other developmental factors that may influence ADC include increased binding of water to macromolecules such as myelin, reducing free water content, as well as the formation of new structural barriers to water diffusion, such as sprouting neurites forming synapses in GM and progressive myelination in WM. Mukherjee et al. (2001) demonstrated that the anisotropy increase in central WM tracts also followed the same bi-exponential time evolution as the ADC decrease (Figure 40.5). This time course

Physiological MRI of normal development and developmental delay

17 days old

4 months old

1 year old

2 years old

10 years old

(a)

(b)

(c)

(d)

(e)

(f)

(g)

(h)

(i)

(j)

Fig. 40.3 The decrease in the ADC during the first decade of life, illustrated in five children with transverse sections at the level of the corona radiata above the bodies of the lateral ventricles (a–e) and at the level of the basal ganglia (f–j). All images are displayed at the same scale, and with identical window and level settings, to allow direct comparison of size and image intensity across subjects. ADC was computed from a single-shot echo-planar diffusion tensor sequence. Increased ADC denotes greater spatially invariant magnitude of water diffusion. The larger ADC values of WM relative to the cortical GM (arrows in (f )) are striking in the 17-day-old newborn, but this GM–WM differentiation is much less conspicuous in the 4-month old (open arrows in (g)) and is not evident in the subjects ages 1 year and older. These images were reproduced with permission from Mukherjee et al. (2001).

parallels that of brain cholesterol content (Dobbing and Sands, 1973), a crude marker of myelination. However, correlation does not necessarily equal causation, and the biophysical processes responsible for the developmental increase in WM anisotropy are still not clearly established. Comparison of older children, of mean age 10 years, with adults showed that frontal lobe WM anisotropy continues to increase after the first decade of life (Klingberg et al., 1999). A DTI study of 66 children up to 6 years of age showed evidence for slower maturation of peripheral WM compared to central WM tracts (McGraw et al., 2002). No quantitative analysis of the time course of the anisotropy changes was performed in this study to assess how well it conforms to a bi-exponential model. Assuming that the

maturational anisotropy increase in peripheral WM is also bi-exponential, one might hypothesize that the slow component is relatively stronger, and the rapid component is relatively weaker, than in central WM tracts. Alternatively, the time constants of one or both exponential components may be longer in peripheral WM than in central WM. Experimental testing of one of these different quantitative models would have implications for the underlying biological processes that cause developmental increases in WM anisotropy. The changes in the three diffusion tensor eigenvalues during brain development have also been characterized in a DTI study of 167 subjects ranging from 31 weeks estimated gestational age to 12 years of postnatal age (Mukherjee et al., 2002). The maximum

679

680

A. James Barkovich, Pratik Mukherjee and Daniel B. Vigneron

17 days old

4 months old

1 year old

2 years old

10 years old

(a)

(b)

(c)

(d)

(e)

(f)

(g)

(h)

(i)

(j)

Fig. 40.4 The growth of diffusion anisotropy during the first decade of life, as quantified by the rotationally invariant measure A and illustrated in the same five children as Figure 40.3, with transverse sections at the level of the corona radiata above the bodies of the lateral ventricles (a–e) and at the level of the basal ganglia (f–j). All images are displayed at the same scale, and with identical window and level settings, to allow direct comparison of size and image intensity across subjects. These anisotropy images were calculated from the same diffusion tensor sequence as the ADC images of Figure 40.3, and are co-registered with the ADC images at identical anatomic levels. The only visible anisotropy in the 17-day-old infant is in the internal capsule, especially its posterior limb (arrows in (f )), and in the corpus callosum, especially its splenium (arrowhead in (f)). In the 4-month-old, more peripheral WM tracts can be identified, such as the optic radiations (arrows in g). Increasing anisotropy in the optic radiations with age (arrows in (g–j)) reflects continued WM maturation. These images were reproduced with permission from Mukherjee et al. (2001).

eigenvalue max, corresponding to the magnitude in the direction of greatest diffusion, is larger in white matter than in GM throughout brain development. In contradistinction, the intermediate and minimum eigenvalues, int and min respectively, are larger in WM than in GM during the newborn period, but become smaller in WM than in GM with progressive myelination. This reflects the preferential reduction of diffusion in directions orthogonal to the fiber orientation in developing WM tracts. In early maturing WM such as the posterior limb of the internal capsule (PLIC), which is partially myelinated at term, int and min are already less than in the surrounding deep GM nuclei.

DTI has been used to delineate non-invasively the WM connectivity between arbitrary regions of the adult brain (Conturo et al., 1999; Mori et al., 1999), a technique known as diffusion tensor fiber tracking or diffusion tensor tractography (DTT). This method relies on iteratively following the primary eigenvector of the diffusion tensor at each voxel, which is the direction of maximal diffusivity and is presumed to be tangential to the orientation of the fiber tract at that location. WM tracts such as the corpus callosum and internal capsule contain many functionally distinct axonal pathways, which can be dissociated with DTT. Hence, DTT has great potential for examining the maturation of functionally specific WM pathways

Physiological MRI of normal development and developmental delay

(a) 1.6

(b) 1.0

PLIC ALIC

0.8

1.2 0.6 RA

Dav (10 3 mm2/s)

1.4

1.0

0.4 0.8 0.2

0.6

0

0.4 0

2

4 6 Age (years)

8

10

0

2

4 6 Age (years)

8

10

Fig. 40.5 Plots of ADC and relative anisotropy (RA) vs. age for the anterior limbs of the internal capsule (ALIC) and PLIC in 166 subjects ranging from 26 weeks estimated gestational age to 12 years postnatal age. The zero point of the abscissa is at 40 weeks estimated gestational age. Data are included from Neil et al. (1998) and Mukherjee et al. (2001). (a) The values for ADC (denoted as Dav) are somewhat higher in the ALIC than the posterior limb early in life, but by a few months of age the values become identical. (b) The values of RA, on the other hand, tend to be higher in the PLIC than in the anterior limb throughout life, although this anisotropy difference becomes smaller with increasing age. These ADC and RA differences early in life indicate earlier maturation of the PLIC compared with the anterior limb. Figure reproduced with permission from Neil et al. (2002).

in the developing brain. However, there are technical challenges to the application of DTT to the pediatric brain, including the smaller size of the WM pathways and their lower anisotropy compared with the adult brain (Neil et al., 2002). In addition to elucidating WM maturation, diffusion anisotropy can also be used to assess development of the human cerebral cortex in premature newborns. In a study of preterm infants of estimated gestational age of 26–41 weeks, McKinstry et al. (2002a) showed that the cerebral cortex has detectable anisotropy (Figure 40.6) that gradually diminishes with maturation, reaching noise levels at approximately 36 weeks estimated gestational age. The orientation of the maximum eigenvector, corresponding to the direction of greatest diffusion, was consistently radial to the pial surface of the cortex. This radial diffusivity is presumed to be caused by the coherent parallel organization of radial glial fibers and apical dendrites of the migrating neurons during this early phase of cortical development. The outgrowth of basal dendrites, as well as innervation of the cortical plate by thalamocortical and corticocortical axons, may be responsible for disrupting this coherent

30 -weeks premature newborn FA image

Fig. 40.6 Diffusion anisotropy of the developing cerebral cortex illustrated in a premature newborn imaged at 30 weeks estimated gestational age. Axial fractional anisotropy (FA) image above the level of the lateral ventricles reveals high anisotropy throughout the cortical plate, including the parietal lobe (arrow) and the interhemispheric region (arrowhead). Note also the virtually complete absence of anisotropy in the immature WM at this early developmental stage.

681

682

A. James Barkovich, Pratik Mukherjee and Daniel B. Vigneron

parallel architecture and eventually eliminating diffusion anisotropy. This hypothesis is supported by the data of Mukherjee et al. (2003a), which show that the loss of preterm cortical anisotropy is due primarily to diminution of the maximum eigenvalue max, with relatively little maturational change in the two minor eigenvalues, int and min. This pattern of eigenvalue change is the opposite of that observed in WM during post-term development, where the anisotropy increase is due to reductions in int and min, with relatively little change in max (Mukherjee et al., 2002). DTI can also reveal the region-specific development of the cerebral cortex, with earlier decreases in anisotropy in primary sensorimotor cortex than in association cortex of the frontal lobe or the parietal lobe (Mukherjee et al., 2003b). This heterochronous cortical development is consistent with an electron microscopy study showing earlier synaptogenesis in a primary sensory region, such as auditory cortex, than in an association region such as prefrontal cortex (Huttenlocher and Dabholkar, 1997).

Assessment of neonatal brain injury Proton spectroscopy has shown significant potential for the early detection of brain injury in encephalopathic neonates (Leth et al., 1996; Penrice et al., 1996; Hanrahan et al., 1998, 1999; Barkovich et al., 1999, 2001). In the normal term infant, lactate (Lac) is not seen in the brain parenchyma (although it may be present in the cerebrospinal fluid of normal neonates) (McGuinness et al., 1983; Fernandez et al., 1986). Lac is seen in the brain within hours after injury, probably due to mitochondrial impairment and subsequent anaerobic glycolysis in brain parenchyma. The concentration of NAA, which increases as neurons mature and decreases with neuronal injury, is reduced within a few days of any injury (Groenendaal et al., 1994); the degree of reduction of NAA seems to correlate well with neurodevelopmental outcome (Groenendaal et al., 1994; Barkovich et al., 1999). Some investigators report an increase in glutamine (Gln)/glutamate (Glu) as well (Pu et al., 2000). In bithalamic voxels, elevated Lac peaks (Lac/NAA ratios above 0.5, Lac/Cr ratios

above 1, and elevated Lac/Cho) are associated with impaired neurological outcome at age 12 months (Penrice et al., 1996; Amess et al., 1999; Barkovich et al., 1999; Hanrahan et al., 1999). Elevated Lac and reduced NAA can also be seen in neonatal infection, such as viral encephalitides and complicated meningitides, although the frequency of these findings are not established. No large studies have been performed looking at the association of proton MRS with neurodevelopmental outcome in preterm neonates, probably because of the difficulty of determining normal metabolite concentrations or metabolite ratios in all the different regions of the brain at all of the different ages during development. Our preliminary studies indicate that NAA values are lower and Lac values higher in preterm neonates with other evidence of brain injury than in preterm neonates who have otherwise normal imaging and development. The primary clinical role of DWI has traditionally been for the evaluation of acute ischemia. Whereas anatomic MRI is usually normal in the first day or two after injury (Barkovich et al., 1995), DWI has proven its utility in the assessment of acute perinatal brain injury, which is characterized by reduced ADC (Inder et al., 1999; Robertson et al., 1999; Barkovich et al., 2001; Soul et al., 2001; Wolf et al., 2001). A recent DTI study indicates that the maximum diagnostic yield for detecting reduced ADC in perinatal brain injury occurs at 2–3 days following the insult (McKinstry et al., 2002b). However, DTI may also have potential for the evaluation of more chronic forms of brain injury or other abnormalities resulting in developmental delay. Measurement of rotationally invariant scalar DTI parameters such as ADC, anisotropy, and the three eigenvalues may provide clinically useful quantitative normative milestones of brain maturation for the assessment of developmental delay (Barkovich, 2000; Mukherjee et al., 2001). Advantages of ADC for this purpose are insensitivity to noise and relative uniformity throughout the brain parenchyma after the first year of life. Anisotropy and eigenvalue measurements are more strongly biased by noise, and therefore require longer acquisitions with higher SNR. Also, since anisotropy and eigenvalues are heterogeneous throughout the brain,

Physiological MRI of normal development and developmental delay

(a)

(b)

Fig. 40.7 Non-ketotic hyperglycinemia: T2-weighted image (a) shows subtle gyral anomaly in the left parietal lobe (arrows). Proton MRS (b) shows large glycine peak (arrow) at 3.55 ppm.

reproducible region-of-interest (ROI) placement is very important for accurate quantitation across subjects. However, anisotropy and the three eigenvalues may be more sensitive than ADC to the process of WM development, especially for evaluation of the more subtle maturational changes occurring in older children and adolescents (Klingberg et al., 1999). Assessment in the developmentally delayed infant Proton spectroscopy In addition to its utility in the assessment of encephalopathic neonates (Barkovich et al., 1999; Hanrahan et al., 1999), proton MRS is useful in the MR assessment of children with developmental delay, both resulting from inborn errors of metabolism (Grodd et al., 1991) and brain injury of other causes (Filippi et al., 2002). Intuitively, one would expect that MRS would be most useful in the diagnosis of inborn errors of metabolism and, indeed, abnormalities are seen on the MRS in many inborn errors of metabolism, often before the MRI becomes abnormal. Abnormal spectra are seen in X-linked adrenoleukodystrophy (ALD)

(Bruhn et al., 1992; Tzika et al., 1993), phenylketonuria (PKU) (Pietz et al., 1996), maple syrup urine disease (MSUD) (Felber et al., 1993), non-ketotic hyperglycinemia (Viola et al., 2002) (Figure 40.7), and Sjögren–Larsson syndrome (Mano et al., 1999), among many others. Inborn errors of metabolism and WM diseases are also discussed in Chapters 44 and 45. Perhaps the disorders in which MRS has been the most useful, however, have been the disorders of Cr deficiency. Cr, which stores energy as Cr phosphate, is essential for normal brain function. Cr is synthesized from arginine by the action of enzymes L-arginine : glycine amidinotransferase (AGAT) and guanidinoacetate methyltransferase (GAMT), then actively transported into the brain. It is catabolized to Cr and excreted in the urine. For maintenance of the body pool of Cr, the daily urinary loss of Cr must be balanced by endogenous synthesis and dietary intake of Cr (van der Knaap et al., 2000). Patients with Cr deficiency present with developmental delay, followed by regression, muscle hypotonia, extrapyramidal movement abnormalities, and intractable epilepsy (Stöckler et al., 1996; Ganesan et al., 1997; Schulze et al., 1997; van der Knaap et al., 2000). These symptoms can be partially

683

684

A. James Barkovich, Pratik Mukherjee and Daniel B. Vigneron

(a)

(b)

Fig. 40.8 GAMT deficiency: coronal FLAIR image (a) shows increased signal intensity in the bilateral globi pallidi. Proton MRS (b) shows absence of the normal Cr peak at 3.0 ppm.

or completely reversed by dietary Cr supplement, so it is crucial to make the diagnosis in affected patients. MRS has helped to identify three disorders that result in a reduction of Cr in the brain: GAMT deficiency, AGAT deficiency, and deficiency of a Cr transporter protein (Bianchi et al., 2000; Item et al., 2001; Salomons et al., 2001; deGrauw et al., 2002). All of these disorders can be diagnosed by the absence or marked reduction of the normal Cr peak on proton MRS (Figure 40.8). Patients with AGAT and GAMT deficiencies respond well to dietary supplementation with Cr monohydrate (Kahler and Fahey, 2003). Proton MRS can also be revealing in children with less specific causes of developmental delay. Filippi et al. (2002) performed single voxel MRS in a series of children with developmental delay with normal anatomic MRI. They found reduced NAA/Cr and increased Cho/Cr ratios when compared to normal age-matched controls (Filippi et al., 2002). This finding implies an underlying neuronal or axonal deficiency, injury, or immaturity in developmentally delayed children. Diffusion imaging Although there have been no definitive DTI studies of developmental delay, there are promising initial results in premature newborns with WM injury.

Hüppi et al. (2001) found reduced anisotropy in injured central WM tracts of 10 premature infants that were studied at term-equivalent age, compared with 10 age-matched preterm infants without injury. There was no significant difference in ADC between the two groups, indicating that anisotropy was more sensitive than ADC for detection of WM injury. Using serial DTI studies in 23 premature newborns studied shortly after birth and again at term-equivalent age, Miller et al. (2002) found abnormal development in more widespread regions of central and peripheral WM in patients with conventional MR evidence of WM injury. WM anisotropy in these injured subjects did not increase across the two serial scans as much as in those subjects without WM injury, and even decreased in frontal WM. In the subjects with a greater degree of WM injury, ADC did not decrease across the two serial scans as much as in the uninjured patients and, in some cases, showed increases. However, ADC results did not differ between the minimally injured group and the uninjured group. In agreement with the results of Hüppi et al. (2001), this showed that anisotropy is the more sensitive measure of WM injury. More suggestive evidence for the utility of DTI in assessing developmental delay was provided by Klingberg et al. (2000), who showed that WM

Physiological MRI of normal development and developmental delay

anisotropy in the temporoparietal region of the left hemisphere was significantly correlated with reading ability in a group of poor readers and a group of controls with normal reading ability. MR tractography has also been shown to be useful for the evaluation of motor delay. Berman et al. (2002) used MR tractography to delineate the corticospinal tracts in subjects with congenital hemiplegia, and demonstrated reduced anisotropy in the affected corticospinal tract compared to the contralateral side. An MR tractography study by Hoon et al. (2002) of two 6-year-old boys with spastic quadriparesis due to premature birth revealed decreased connectivity of the somatosensory cortex via pathways through the internal capsule and the corpus callosum. This suggests that WM projections to and from the somatosensory cortex may play an important role in the pathophysiology of periventricular leukomalacia (PVL). DTI with tractography has the potential to provide a new window into all forms of developmental delay, including motor delay, visual delay, speech delay, reading disorders, and other forms of cognitive delay. Perfusion imaging Perfusion imaging is mainly useful in the setting of neurological impairment or global developmental delay in the setting of an underlying vasculopathy such as those associated with sickle cell disease, moyamoya disease, or neurofibromatosis (NF) type I. In these settings, perfusion studies can help to assess the need for revascularization procedures (Calamante et al., 2001; Kirkham et al., 2001). As the perfusion data acquired by arterial spin labeling (ASL) is potentially quantifiable, and because it can be performed without the use of intravenous contrast, it may have some advantages over bolus infusion techniques.

Conclusion The use of physiological MR techniques in the evaluation of newborns and infants is still in its early stages of use, but seems to have great promise. Uses will likely include assessment of brain injury in preterm infants, assessment of brain injury or

malformation in encephalopathic term infants, assessment of developmentally delayed infants, and assessment of infants with vasculopathy. A major current challenge is the establishment of ranges of normal values for the different regions of the brain at different ages, so that we can identify those children with mild-to-moderate abnormalities in addition to severely affected children. Imaging at higher field strengths and other technical advances to improve SNR will allow detection of metabolites that have low natural concentrations in the brain. Improved SNR will also allow tracking of smaller axonal fascicles and, ultimately, may allow the identification of missing or aberrant WM pathways in children with specific developmental problems.

REFERENCES Amess PN, Penrice J, Wylezinska M, Lorek A, Twonsend J, Wyatt JS, et al. 1999. Early brain proton magnetic resonance spectroscopy and neonatal neurology related to neurodevelopmental outcome at 1 year in term infants after presumed hypoxic–ischaemic brain injury. Dev Med Child Neurol 41: 436–445. Azzopardi D, Wyatt JS, Cady EB, Delpy DT, Baudin J, Stewart AL, et al. 1989. Prognosis of newborn infants with hypoxic– ischemic brain injury assessed by phosphorus magnetic resonance spectroscopy. Pediatr Res 25: 445–451. Barkovich AJ. 1997. The encephalopathic neonate: choosing the proper imaging technique. Am J Neuroradiol 18: 1816–1820. Barkovich AJ. 2000. Concepts of myelin and myelination in neuroradiology. Am J Neuroradiol 21: 1099–1109. Barkovich AJ, Baranski K, Vigneron D, Partridge JC, Hallam DK, Latal Hajnal B, et al. 1999. Proton MR spectroscopy in the evaluation of asphyxiated term neonates. Am J Neuroradiol 20: 1399–1405. Barkovich AJ, Hajnal BL, Vigneron D, Sola A, Partridge JC, Allen F, et al. 1998. Prediction of neuromotor outcome in perinatal asphyxia: evaluation of MR scoring systems. Am J Neuroradiol 19: 143–150. Barkovich AJ, Kjos BO, Jackson Jr DE, Norman D. 1988. Normal maturation of the neonatal and infant brain: MR imaging at 1.5 T. Radiology 166: 173–180. Barkovich AJ, Westmark KD, Bedi HS, Partridge JC, Ferriero DM, Vigneron DB. 2001. Proton spectroscopy and diffusion imaging on the first day of life after perinatal asphyxia: preliminary report. Am J Neuroradiol 22: 1786–1794.

685

686

A. James Barkovich, Pratik Mukherjee and Daniel B. Vigneron

Barkovich AJ, Westmark KD, Ferriero D, Sola A, Partridge C. 1995. Perinatal asphyxia: MR findings in the first 10 days. Am J Neuroradiol 16: 427–438. Berman JI, Glenn OA, Vigneron DB, Barkovich AJ, Henry RG. 2002. Towards quantitative DTI tractography evaluating congenital hemiplegia in corticospinal tracts. Proc Intl Soc Mag Reson Med, Honolulu, 1124. Bianchi MC, Tosetti M, Fornai F, Alessandri MG, Cipriani P, De Vito G, et al. 2000. Reversible brain creatine deficiency in two sisters with normal blood creatine level. Ann Neurol 47: 511–513. Bruhn H, Kruse B, Korenke GC, et al. 1992. Proton NMR spectroscopy of cerebral metabolic alterations in infantile peroxisomal disorders. J Comput Assist Tomogr 16: 335–344. Calamante F, Ganesan V, Kirkham FJ, Jan W, Chong WK, Gadian DG, et al. 2001. MR perfusion imaging in Moyamoya syndrome: potential implications for clinical evaluation of occlusive cerebrovascular disease. Stroke 32: 2810–2816. Cecil KM, Jones BV. 2001. Magnetic resonance spectroscopy of the pediatric brain. Top Magn Reson Imaging 12: 435–452. Conturo TE, Lori NF, Cull TS, Akbudak E, Snyder AZ, Shimony JS, et al. 1999. Tracking neuronal fiber pathways in the living human brain. Proc Natl Acad Sci USA 96: 10422–10427. deGrauw TJ, Salomons GS, Cecil KM, Chuck G, Newmeyer A, Schapiro MB, et al. 2002. Congenital creatine transporter deficiency. Neuropediatrics 33: 232–238. Dobbing J, Sands J. 1973. Quantitative growth and development of human brain. Arch Dis Child 48: 757–767. Felber SR, Sperl W, Chemelli A, Murr C, Wendel U. 1993. Maple syrup urine disease: metabolic decompensation monitored by proton magnetic resonance imaging and spectroscopy. Ann Neurol 33: 396–401. Fernandez F, Verdu A, Quero J, Ferreiros MC, Daimiel E, Roche MC, et al. 1986. Cerebrospinal fluid lactate levels in term infants with perinatal hypoxia. Pediatr Neurol 2: 39–42. Filippi CG, Ulug AM, Deck MDF, Zimmerman RD, Heier LA. 2002. Developmental delay in children: assessment with proton MR spectroscopy. Am J Neuroradiol 23: 882–888. Ganesan V, Johnson A, Connelly A, Eckhardt S, Surtees RA. 1997. Guanidinoacetate methyltransferase deficiency: new clinical features. Pediatr Neurol 17: 155–157. Grant EG, Schellinger D, Richardson JD, Coffey ML, Smirniotopoulous JG. 1983. Echogenic periventricular halo: normal sonographic finding or neonatal cerebral hemorrhage? Am J Neuroradiol 4: 43–46. Grodd W, Krageloh-Mann I, Klose U, Sauter R. 1991. Metabolic and destructive brain disorders in children: findings with localized proton MR spectroscopy. Radiology 181: 173–181. Groenendaal F, Veenhoven EH, van der Grond J, Jansen GH, Witkamp TD, de Vries L. 1994. Cerebral lactate and N-acetylaspartate/choline ratios in asphyxiated full-term neonates

demonstrated in-vivo using proton magnetic resonance spectroscopy. Pediatr Res 35: 148–151. Hanrahan JD, Azzopardi D, Cowan FM, Rutherford MA, Cox IJ, Edwards AD. 1998. Persistent increases in cerebral lactate concentration after birth asphyxia. Pediatr Res 44: 304–311. Hanrahan JD, Cox IJ, Azzopardi D, Cowan FM, Sargentoni J, Bell JD, et al. 1999. Relation between proton magnetic resonance spectroscopy within 18 hours of birth asphyxia and neurodevelopment at 1 year of age. Dev Med Child Neurol 41: 76–82. Hoon Jr AH, Lawrie WT, Melham ER, Reinhardt EM, van Zijl PCM, Solaiyappan M, et al. 2002. Diffusion tensor imaging of periventricular leukomalacia shows affected sensory cortex white matter pathways. Neurology 59: 752–756. Hunter JV, Wang ZJ. 2001. MR spectroscopy in pediatric neuroradiology. Magn Reson Imaging Clin N Am 9: 165–89, ix. Huppi PS. 2001. MR imaging and spectroscopy of brain development. Magn Reson Imaging Clin N Am 9: 1–17, vii. Huppi PS, Inder TE. 2001. Magnetic resonance techniques in the evaluation of the perinatal brain: recent advances and future directions. Semin Neonatol 6: 195–210. Huppi PS, Posse S, Lazeyras F, Burri R, Bossi E, Herschkowitz N. 1991. Magnetic resonance in preterm and term newborns: 1H-spectroscopy in developing human brain. Pediatr Res 30: 574–578. Huppi P, Maier S, Peled S, Zientara GP, Barnes PD, Jolesz FA, et al. 1998. Microstructural development of human newborn cerebral white matter assessed in vivo by diffusion tensor magnetic resonance imaging. Pediatr Res 44: 584–590. Hüppi PS, Murphy B, Maier SE, Zientara GP, Inder TE, Barnes PD, et al. 2001. Microstructural brain development after perinatal cerebral white matter injury assessed by diffusion tensor magnetic resonance imaging. Pediatrics 107: 455–460. Huttenlocher PR, Dabholkar AS. 1997. Regional differences in synaptogenesis in human cerebral cortex. J Comp Neurol 387: 167–178. Inder T, Huppi P, Zientara G, Maier S, Jolesz F, di Salvo D, et al. 1999. Early detection of periventricular leukomalacia by diffusion-weighted magnetic resonance imaging techniques. J Pediatr 134: 631–634. Item CB, Stockler-Ipsiroglu S, Stromberger C, Muhl A, Alessandri MG, Bianchi MC, et al. 2001. Arginine:glycine amidinotransferase deficiency: the third inborn error of creatine metabolism in humans. Am J Hum Genet 69: 1127–1133. Kahler SG, Fahey MC. 2003. Metabolic disorders and mental retardation. Am J Med Genet 117C: 31–41. Kirkham FJ, Calamante F, Bynevelt M, Gadian DG, Evans JP, Cox TC, et al. 2001. Perfusion magnetic resonance abnormalities in patients with sickle cell disease. Ann Neurol 49: 477–485.

Physiological MRI of normal development and developmental delay

Klingberg TC, Hedehus M, Temple E, Salz T, Gabrieli JDE, Moseley ME, et al. 2000. Microstructure of temporo-parietal white matter as a basis for reading ability: evidence from diffusion tensor magnetic resonance imaging. Neuron 25: 493–500. Klingberg TC, Vaidya J, Gabrieli JDE, Moseley ME, Hedehus M. 1999. Myelination and organization of the frontal white matter in children: a diffusion tensor MRI study. Neuroreport 10: 2817–2821. Kreis R, Ernst T, Ross BD. 1993. Development of the human brain: in vivo quantification of metabolite and water content with proton magnetic resonance spectroscopy. Magn Reson Med 30: 424–37. Kreis R, Hofmann L, Kuhlmann B, Boesch C, Bossi E, Huppi PS. 2002. Brain metabolite composition during early human brain development as measured by quantitative in vivo 1H magnetic resonance spectroscopy. Magn Reson Med 48: 949–958. Kuzniecky R, Jackson G. 1995. Magnetic Resonance in Epilepsy. Raven Press, New York. Laptook AR, Corbett RJ, Uauy R, Mize C, Mendelsohn D, Nunnally RL. 1989. Use of 31P magnetic resonance spectroscopy to characterize evolving brain damage after perinatal asphyxia. Neurology 39: 709–712. Leth H, Toft PB, Peitersen B, Lou HC, Henriksen O. 1996. Use of brain lactate levels to predict outcome after perinatal asphyxia. Acta Paediatr 85: 859–864. Mano T, Ono J, Kaminga T, Imai K, Sakurai K, Harada K, et al. 1999. Proton MR spectroscopy of Sjögren–Larsson’s syndrome. Am J Neuroradiol 20: 1671–1673. McGraw P, Liang L, Provanzale JM. 2002. Evaluation of normal age-related changes in anisotropy during infancy and childhood as shown by diffusion tensor imaging. Am J Roentgenol 179: 1515–1522. McGuinness GA, Weisz SC, Bell WE. 1983. CSF lactate levels in neonates. Effects of asphyxia, gestational age, and postnatal age. Am J Dis Child 137: 48–50. McKinstry RC, Mathur A, Miller JH, Ozcan A, Snyder AZ, Schefft GL, et al. 2002a. Radial organization of developing preterm human cerebral cortex revealed by noninvasive water diffusion anisotropy MRI. Cerebral Cortex 12: 1237–1243. McKinstry RC, Miller JH, Snyder AZ, Mathur A, Schefft GL, Almli CR, et al. 2002b. A prospective, longitudinal diffusion tensor imaging study of brain injury in newborns. Neurology 59: 824–833. Ment LR, Bada HS, Barnes P, Grant PE, Hirtz D, Papile LA, et al. 2002. Practice parameter: neuroimaging of the neonate: report of the Quality Standards Subcommittee of the American Academy of Neurology and the Practice Committee of the Child Neurology Society. Neurology 58: 1726–1738.

Miller SP, Vigneron DB, Henry RG, Bohland MA, Ceppi-Cozzio C, Hoffman C, et al. 2002. Serial quantitative diffusion tensor MRI of the premature brain: development in newborns with and without injury. J Magn Reson Imaging 16: 621–632. Mori S, Crain BJ, Chacko VP, van Zijl PCM. 1999. Three dimensional tracking of axonal projections in the brain by magnetic resonance imaging. Ann Neurol 45: 265–269. Morriss M, Zimmerman R, Bilaniuk L, Hunter J, Haselgrove J. 1999. Changes in brain water during childhood. Neuroradiology 41: 929–934. Mukherjee P, Miller JH, Shimony JS, Conturo TE, Lee BC, Almli CR, et al. 2001. Normal brain maturation during childhood: developmental trends characterized with diffusion-tensor MR imaging. Radiology 221: 349–358. Mukherjee P, Miller JH, Shimony JS, Philip JV, Nehra D, Snyder AZ, et al. 2002. Diffusion-tensor MR imaging of gray and white matter development during normal human brain maturation. Am J Neuroradiol 23: 1445–1456. Mukherjee P, Gill KRS, Veeraraghavan S, Henry RG, Miller SP, Vigneron DB, et al. 2003a. Anisotropy loss during development of cerebral cortex in premature newborns is due to decreasing water diffusion perpendicular (but not parallel) to the cortical surface. Proc Intl Soc Mag Reson Med, Toronto. Mukherjee P, Gill KRS, Veeraraghavan S, Henry RG, Miller SP, Vigneron DB, et al. 2003b. Region-specific maturation of cerebral cortex in premature newborns demonstrated with high-resolution diffusion tensor imaging. Proc Intl Soc Mag Reson Med, Toronto. Mulkern RV, Vajapeyam S, Robertson RL, Caruso PA, Rivkin M, Maier SE. 2001. Biexponential apparent diffusion coefficient parametrization in adult vs newborn brain. Magn Reson Imaging 19: 659–668. Neil JJ, Miller JH, Mukherjee P, Huppi P. 2002. Diffusion tensor imaging of normal and injured developing human brain – a technical review. NMR Biomed 15: 543–552. Neil JJ, Shiran SI, McKinstry RC, Schefft GL, Snynder AZ, Almli CR, et al. 1998. Normal brain in human newborns: apparent diffusion coefficient and diffusion anisotropy measured by using diffusion tensor MR imaging. Radiology 209: 57–66. Nomura Y, Sakuma H, Takeda K, Tagami T, Okuda Y, Nakagawa T. 1994. Diffusional anisotropy of the human brain assessed with diffusion-weighted MR: relation with normal brain development and aging. Am J Neuroradiol 15: 231–238. Papile LA, Brustein J, Burstein R, Koffer H. 1978. Incidence and evolution of subependymal and intraventricular hemorrhage: a study of infants with birth weights less than 1500 gm. J Pediatr 92: 529–534.

687

688

A. James Barkovich, Pratik Mukherjee and Daniel B. Vigneron

Peden CJ, Cowan FM, Bryant DJ, Sargentoni J, Cox IJ, Menon DK, et al. 1990. Proton MR spectroscopy of the brain in infants. J Comput Assist Tomogr 14: 886–894. Penrice J, Cady EB, Lorek A, Wylezinska M, Amess PN, Aldridge RF, et al. 1996. Proton magnetic resonance spectroscopy of the brain in normal preterm and term infants and early changes after perinatal hypoxia–ischemia. Pediatr Res 40: 6–14. Pietz J, Kreis R, Schmidt H, Meyding-Lamadé U, Rupp A, Boesch C. 1996. Phenylkentonuria: findings at MR imaging and localized in vivo H-1 MR spectroscopy of the brain in patients with early treatment. Radiology 201: 413–420. Prayer D, Barkovich AJ, Kirschner DA, Prayer LM, Roberts TPL, Kucharczyk J, et al. 2001. Visualization of nonstructural changes in early white matter development on diffusion weighted MR images: evidence supporting premyelination anisotropy. Am J Neuroradiol 22: 1572–1576. Pu Y, Li Q-F, Zeng C-M, Gao J, Qi J, Luo D-X, et al. 2000. Increased detectability of alpha brain glutamate/glutamine in neonatal hypoxic–ischemic encephalpathy. Am J Neuroradiol 21: 203–212. Raymond AA, Fish DR, Sisodiya SM, Alsanjari N, Stevens JM, Shorvon SD. 1995. Abnormalities of gyration, heterotopias, tuberous sclerosis, focal cortical dysplasia, microdysgenesis, dysembryoplastic neuroepithelial tumor and dysgenesis of the archicortex in epilepsy: clinical, EEG and neuroimaging features in 100 adult patients. Brain 118: 629–660. Robertson R, Ben-Sira L, Barnes P, Mulkern R, Robson C, Maier S, et al. 1999. MR line scan diffusion weighted imaging of term neonates with perinatal brain ischemia. Am J Neuroradiol 20: 1658–1670. Rutherford M, Pennock J, Schwieso JE, Cowan FM, Dubowitz LM. 1995. Hypoxic–ischemic encephalopathy: early magnetic resonance imaging findings and their evolution. Neuropediatrics 26: 183–191. Rutherford M, Pennock J, Schwieso J, et al. 1996. Hypoxic ischaemic encephalopathy: early and late magnetic resonance findings in relation to outcome. Arch Dis Child 75: 145–151. Rutherford MA, Pennock JM, Dubowitz LMS. 1994. Cranial ultrasound and magnetic resonance imaging in hypoxic– ischemic encephalopathy: a comparison with outcome. Dev Med Child Neurol 36: 813–825. Salomons GS, van Dooren SJ, Verhoeven NM, Cecil KM, Ball WS, Degrauw TJ, et al. 2001. X-linked creatine-transporter gene (SLC6A8) defect: a new creatine-deficiency syndrome. Am J Hum Genet 68: 1497–1500. Scarabino T, Popolizio T, Bertolino A, Salvolini U. 1999. Proton magnetic resonance spectroscopy of the brain in pediatric patients. Eur J Radiol 30: 142–153.

Schmithorst VJ, Wilke M, Dardzinski BJ, Holland SK. 2002. Correlation of white matter diffusivity and anisotropy with age during childhood and adolescence: a crosssectional diffusion tensor MR imaging study. Radiology 222: 212–218. Schouman-Clays E, Henry-Feugeas M-C, Roset F, et al. 1993. Periventricular leukomalacia: correlation between MR imaging and autopsy findings during the first 2 months of life. Radiology 189: 59–64. Schulze A, Hess T, Wevers R, Mayatepek E, Bachert P, Marescau B, et al. 1997. Creatine deficiency syndrome caused by guanidinoacetate methyltransferase deficiency: diagnostic tools for a new inborn error of metabolism. J Pediatr 131: 616–631. Shevell M, Ashwal S, Donley D, Flint J, Gingold M, Hirtz D, et al. 2003. Practice parameter: evaluation of the child with global developmental delay: report of the Quality Standards Subcommittee of the American Academy of Neurology and The Practice Committee of the Child Neurology Society. Neurology 60: 367–380. Shevell MI, Ashwal S, Novotny E. 1999. Proton magnetic resonance spectroscopy: clinical applications in children with nervous system diseases. Semin Pediatr Neurol 6: 68–77. Shimony JS, McKinstry RC, Akbudak E, Aronovitz JA, Snyder AZ, Lori NF, et al. 1999. Quantitative diffusion-tensor anisotropy brain MR imaging: normative human data and anatomic analysis. Radiology 212: 770–784. Soul JS, Robertson RL, Tzika AA, du Plessis AJ, Volpe JJ. 2001. Time course of changes in diffusion-weighted magnetic resonance imaging in a case of neonatal encephalopathy with defined onset and duration of hypoxic–ischemic insult. Pediatrics 108: 1211–1214. Stöckler S, Hanefeld F, Frahm J. 1996. Creatine replacement in guanidinoacetate methyltransferase deficiency, a novel inborn error of metabolism. Lancet 348. Tzika A, Ball Jr W, Vigneron D, Dunn RS, Nelson SJ, Kirks D. 1993. Childhood adrenoleukodystrophy: assessment with proton MR spectroscopy. Radiology 189: 467–480. van der Knaap MS, Valk J. 1995. Magnetic Resonance of Myelin, Myelination, and Myelin Disorders, 2nd edn. Springer, Berlin. van der Knaap MS, Verhoeven NM, MaaswinkeMooij P, Pouwels PJW, Onkenhout W, Peeters EAJ, et al. 2000. Mental retardation and behavioral problems as presenting signs of a creatine synthesis defect. Ann Neurol 47: 540–543. Vigneron DB, Barkovich AJ, Noworolski SM, von dem Bussche M, Henry RG, Lu Y, et al. 2001. Three-dimensional proton MR spectroscopic imaging of premature and term neonates. Am J Neuroradiol 22: 1424–1433.

Physiological MRI of normal development and developmental delay

Viola A, Chabrol B, Nicoli F, Confort-Gouny S, Viout P, Cozzone PJ. 2002. Magnetic resonance spectroscopy study of glycine pathways in nonketotic hyperglycinemia. Pediatr Res 52: 292–300. Volpe JJ. 2000. Neurology of the Newborn, 4th edn. Saunders, Philadelphia. Wang ZJ, Zimmerman RA. 1998. Proton MR spectroscopy of pediatric brain metabolic disorders. Neuroimaging Clin N Am 8: 781–807.

Wimberger DM, Roberts TP, Barkovich AJ, Prayer LM, Moseley ME, Kucharczyk J. 1995. Identification of ‘premyelination’ by diffusion-weighted MRI. J Comput Assist Tomogr 19: 28–33. Wolf RL, Zimmerman RA, Clancy R, Haselgrove JH. 2001. Quantitative apparent diffusion coefficient measurements in term neonates for early detection of hypoxic–ischemic brain injury: initial experience. Radiology 218: 825–833.

689

41

MR spectroscopy of hypoxic brain injury Brian Ross1, Cathleen Enriquez2 and Alexander Lin2 1

Huntington Medical Research Institutes, Pasadena, USA Rudi Schulte Research Institute, Santa Barbara, USA

2

Key points • Proton MR Spectroscopy (MRS) is sensitive to hypoxic injury; loss of N-acetyl aspartate (NAA), increase in lactate (Lac) and glutamine, and ultimately decrease in creatine (Cr) may occur. • High Lac and low NAA and Cr are bad prognostic indicators in hypoxic brain injury. • Reperfusion may “washout” Lac, even though permanent neuronal injury may be present. • MRS may assist in the diagnosis of brain death. • A knowledge of normal, age-related regional metabolic variations is essential in interpreting spectra from children or neonates. • Some neonates may receive anti-seizure medication containing propan-1,2-diol which accumulates in the brain; this is a doublet signal at 1.1 ppm which should not be confused with Lac.

Introduction Hypoxic or hypoxic–ischemic encephalopathy is the result of prolonged oxygen deprivation of the central nervous system (CNS). The pathophysiology is reasonably well understood, from extensive studies in experimental animals (Heiss, 1983; Hossman, 1988; Siesjo, 1988). At a critical reduced level of oxygen delivery (or blood flow), the electroencephalogram (EEG) slows, potassium increases, adenosine triphosphate (ATP) and phosphocreatine (PCr) are depleted. These effects are largely reversible; however, if oxygen deprivation is prolonged, increased intracellular calcium and acidosis induce histological signs of necrosis which become 690

apparent at a much later time (24–48 h). Free-fatty acids appear as phospholipases are activated; cells swell as part of cytotoxic (hypoxic) edema. Excitatory neurotransmitters, glutamate (Glu) and aspartate released from ischemic cells and lactate (Lac) produced by glycolysis when oxidative metabolism is inhibited by hypoxia, are all believed to contribute to cytotoxicity. Clinically, hypoxic encephalopathy is encountered in two quite distinct situations. First, neonatal or perinatal asphyxia which is mild, moderate or severe, and associated with long-term neurological sequelae including spastic diplegia and mental retardation. Second, in children and adults who have suffered hypoxia as a result of drowning, asphyxia, cardiac arrest, carbon-monoxide poisoning, or other disastrous cerebral accident. Neonatal hypoxia Many other factors than hypoxia are involved in the outcome of pregnancy so that interpretation of MR spectroscopy (MRS) examinations can be complex. Nevertheless, it is clear from what is known of the pathogenesis of hypoxic encephalopathy, that MRS may have the potential to clarify etiology in any given infant. Hypoxic–ischemic encephalopathy Many medical conditions can precipitate this emergency, including cardiac arrest, suffocation (e.g. from near-drowning or aspiration of food), carbonmonoxide poisoning, trauma or other neurological conditions interrupting respiration and general anesthesia with gas that is oxygen deficient. Among the most important issues are diagnosis, particularly of

MR spectroscopy of hypoxic brain injury

mI

(a)

Cr

Cho

(b)

Cho NAA Lac

(c)

Cr NAA Lac mI Lipid Control

Parietal Day 5

Occipital Day 5

mI Cho Cr

Lac Patient 1

NAA Normal

Parietal Day 12 Lipid Lac

Hypoxia Day 5

Parietal Day 42 Cho mI Cr

Patient 2

4

3

2 ppm

1

0

4

3

FrontoParietal Day 42

NAA

2 ppm

1

0

NAA

4

3

2 ppm

Hypoxia 12 Weeks

1

0

Fig. 41.1 (a) Good prognosis after hypoxia (Patient 1) compared to severe hypoxic encephalopathy with very poor prognosis (Patient 2). (b) Anatomical heterogeneity of hypoxic damage and progression of biochemical damage over time. (c) Hypoxic encephalopathy in prematurity. 1H MRS spectra illustrating the paradoxical recovery of cerebral metabolites after severe hypoxic encephalopathy in an infant at 31 weeks gestational age (normal  34–35 weeks).

the syndrome of brain death, and prognosis. While several “logistical” models are in current use, it is clear that MRS could contribute by adding objective measures of a hypoxic brain profile and, most significantly by the direct determination of neuronal injury and survival on a regional basis.

MRS of the hypoxia–ischemia cascade For the diagnosis of neonatal hypoxia, ultra-sound, computed tomography (CT), MR imaging (MRI), positron emission tomography (PET) or single photon emission computed tomography (SPECT) can all

be considered. Although there have been relatively few comparitive studies, localized MRS of the brain promises to provide additional diagnostic information. Loss of N-acetyl aspartate (NAA), appearance of Lac, increase of Glutamine (Gln) and Glu and ultimately loss of Creatine (Cr) are logically anticipated by the hypoxia–ischemia cascade illustrated in Figure 41.3. Empirically, excess lipid is a frequent accompaniment in MRS. The distinction between severe hypoxic encephalopathy with poor prognosis and a prognosis of “full-recovery” is usually made on the basis of: (1) excess Lac, (2) decreased NAA, (3) loss of Cr (Figure 41.1). Sensitivity of MRS for hypoxic

691

692

Brian Ross, Cathleen Enriquez and Alexander Lin

Hypoxia–ischemia

Redox failure (cyto) NADH Lac (mito) NADH Glu

Energy failure

PCr

Cr “Membrane” Phospholipids

Ca

Phospholipase Free fatty acid

4

3

2 ppm

1

0

Fig. 41.2 The spectrum obtained from white matter (WM) 14 days after cardiac arrest in a 17 year old asthmatic. There is notable Lac, and almost complete loss of NAA consistent with hypoxic encephalopathy. The elevated Cho/Cr and mI/Cr probably represents gliosis, which develops only after a week or more.

injury appears superior to either ultra-sound or CT, but specificity is unproven. Complicating factors in spectral interpretation include the effects of gestational age, drugs, parental feeding solutions and xenobiotics which appear in the 1H MR spectrum, including propanediol (a doublet centered at 1.1 ppm which can easily be mis-assigned as Lac), mannitol, ethanol and glucose. In children and adults with hypoxic–ischemic encephalopathy, as in the neonatal brain, increased lipid, Lac, decreased NAA, increased alpha, beta and gamma Glx, decreased Cr are likely findings (Figure 41.2). In this patient group MRS appears highly sensitive to severe global hypoxic insult, probably providing the best prognostic information after near-drowning, for instance (Kreis et al., 1996). Specificity is also high, but with some overlap with the posttrauma MR spectrum, where the etiology of brain injury is more complex and less well understood.

Pathobiology of human hypoxia–ischemia In the 1960s, Lowry and others (Duffy et al., 1972) used the then novel “rapid-freezing’’ technique to

Free radicals Fig. 41.3 Metabolic cascade in cerebral hypoxic injury: based on earlier experimental studies and in vivo MRS experience of human hypoxic encephalopathy, a series of likely events is portrayed. 1H MRS is directly capable of reporting on several aspects of this cascade, including Lac, Glu (see Figure 41.4), lipid membrane destruction, energy status and delayed neuronal injury (as NAA).

re-evaluate Pasteur’s and Warburg’s predictions about hypoxic metabolism. In whole-brain, reduced energy metabolism (low ATP, increased adenosine diphosphate (ADP), reduced PCr, increased Cr) and accelerated glycolysis (increased Lac, decreased pyruvate, increased hydroxybutyrate, decreased acetoacetate, increased Glu, decreased oxoglutarate) were found to fit nicely with the balance of “redox-states” introduced by Bucher and Krebs and meticulously documented by Veech (Miller et al., 1973). In vivo, 31P NMR appeared to favor this interpretation, with reductions in ATP and PCr and reduced pH (believed to reflect increased Lac). With the advent of the more powerful techniques of 1H MRS (more powerful because of at least an order of magntiude greater sensitivity with a broader range of metabolites assayed, higher spatial resolution, and much greater availability than 31P MRS), the prediction of increased Lac was confirmed. 1H MRS data recorded loss of NAA that confirmed the selective injury of neurons, demonstrated by histological studies in animals and implied by a century of clinical neurology. At this point human MRS diverges

MR spectroscopy of hypoxic brain injury

Hypoxia mI

Cho mI Gln

Cr NAA

Glu

Lac

Glx

NAA

Lip Control Gln

mI

Glu

4

3

2

1

ppm

80

70

60 ppm

Cr  Cho NAA

50

40

Fig. 41.4 Evidence for Gln formation in human hypoxic encephalopathy. Representative 1H (left) and natural abundance 13C MR spectra (right) are from a 1 month-old female infant (gestational age 41weeks) who had suffered moderate hypoxic brain injury necessitating extra-corporeal membrane oxygenator (ECMO) therapy. The left spectrum is 1H MRS acquired with STEAM echo time (TE) 30 ms and illustrates (a) small excess lipid; (b) small excess Lac; (c) 20% reduced NAA/Cr by age and (d) excess of Glx  and . (e) Note that Cho/Cr and mI/Cr although apparently elevated, are normal for gestation and postnatal age respectively. Normative data available for STEAM TE 30 ms is roughly applicable to point resolved spectroscopy (PRESS) TE 35 ms. 13C spectra are from the patient (top right) and an age-matched normal control (bottom right). Glx of which the principal metabolic constituents are poorly resolved in 1 H MR spectra is now resolved into two resonances, (Glu and Gln). The concentration of Gln is clearly elevated, while Glu is not.

from animal studies and classical post-mortem human evidence, providing an entirely novel view of hypoxic brain injury.

New insights into human hypoxic brain injury Universal truths of biochemistry were not always reflected by the new findings from 31P and 1H MRS. The new complexity comes from the realization that, in contrast to Lowry’s mice, human hypoxia is rarely so simple: in the first place strenuous efforts have been made to restore blood flow and oxygen to the brain in most patients. In our clinical experience, in only a single patient it was observed (as predicted by animal experiments of irreversible hypoxic injury) that ATP and PCr were completely depleted; intracellular pH, as assayed from the chemical shift of the enormously increased Pi, was strongly acidic, and Lac greatly increased. This represents the full MRS picture

of brain death, a pattern characteristic of the many MRS studies of local brain death reported after stroke (irreversible arterial occlusion). A much more familiar pattern of human hypoxic brain injury is perhaps more accurately described as “secondary hypoxic injury’’ (Cady et al., 1994; Kreis et al., 1996). ATP, PCr Pi and pHi are often normal so that, contrary to our expectations based on Pasteur, Warburg, Lowry, Bucher and Veech, energy metabolism appears to be “normal’’ (this is indeed only apparent (cf. Bluml et al., 2001)). In these cases, the phosphorus spectrum is not particularly helpful for diagnostic purposes, but there are nevertheless significant changes that can typically be observed in the proton spectrum (Figure 41.4). This pattern includes excess lipid (not predicted by most classical biochemistry, but readily appreciated as a release of triglyceride from NMR-invisible macro molecule lipid pools of membrane and myelin), excess Lac, excess Gln (rather than the anticipated

693

694

Brian Ross, Cathleen Enriquez and Alexander Lin

GS



GLN

Gln C2/Glu C2 Glu C2/Glu C4 TCA

GLU

PAG

TC Glu C2

Fig. 41.5 Mechanism of Gln formation during hypoxic injury: as illustrated by the cases in Figure 41.4, Glu is rapidly converted to Gln in the in vivo human brain. This diagram shows the likely enzymatic conversion to Gln, of neurotransmitter Glu, released in excessive amounts during hypoxia-induced hyper-excitation. Glutamine synthetase (GS) is located exclusively in the astrocytes which also have active transporters to remove Glu from the synaptic space.

accumulation of Glu, Figure 41.5). NAA, generally believed to be a neuronal and axonal metabolite (absent from mature glia), is reduced, consistent with the presence of neuronal injury or cell death. This neurochemical profile of hypoxic–ischemic brain injury has been extensively studied, and is the best documented of all MRS findings with the possible exception of brain tumors (cf. Chapter 19). The following cases are representative and are offered as illustrations of the major questions frequently asked in applying 1H MRS in practice. 1. The diagnostic and predictive value of 1H MRS in hypoxic encephalopathy depends upon careful standardization of the MRS technique. Two spectra are presented (Figure 41.6). In each case MRS was performed at the neonatologist’s request after a neurological opinion suggested severe hypoxic

Patient 1: Good neurological outcome

GM voxel position

GM PRESS

Patient 2: Poor neurological outcome

GM voxel position

GM PRESS

Fig. 41.6 Predictive value of 1H MRS in hypoxic encephalopathy: two pairs of studies are presented, in each case MRS was performed at the neonatologist’s request after a neurological opinion suggested severe hypoxic brain injury with very guarded prognosis: Patient 1, in whom the neurological outcome was excellent, was a comatose infant examined on Day 1 and again on Day 4 (not shown). Patient 2, in whom neurological outcome was poor, was examined in coma on Day 5 and again on Day 35 (not shown).

MR spectroscopy of hypoxic brain injury

brain injury with very guarded prognosis. The same brain location (mesial posterior gray matter (GM), cingulate gyrus) and the same 1H MRS method (point resolved spectroscopy (PRESS) TE 35 ms; voxel size 8 cm3) was selected for each infant: Patient 1 was comatose when examined on Day 1, but the neurological outcome was excellent, and the child was alert when examined with 1H MRS again on Day 4 (not shown). Patient 2 was examined in coma on Day 5 and again on Day 35 (not shown). The clinical neurological outcome was poor. The presence of lipid and Lac as well as lower than age-matched normal NAA/Cr differentiated these two spectra sufficiently to provide prognostic information at the first MRS examination. 2. When acquired under the same MRS conditions (PRESS TE 35 ms), multi-voxel MR spectroscopic imaging (MRSI) and single-voxel MRS yield identical information and are equally diagnostic and predictive of neurological outcome (Figure 41.7). Three infants are compared. Each was given the same (poor) prognosis after a clinical diagnosis of severe neurological injury following perinatal hypoxia. Patient 4 showed normal MR spectra throughout the MRSI, and subsequently made a full neurological recovery. Patients 3 and 5 demonstrated pathological MR spectra, with excess lipid and Lac and much reduced NAA/Cr over most of the volume examined by MRSI. These volumes include those selected for single-voxel MRS in patients 1 and 2 described in Figure 41.6, and can therefore reliably be interpreted to indicate poor neurological outcome. Long-term neurological follow-up over 7 months, 18 months and 1 year, respectively, confirmed both the clinical prediction and the MRSI interpretation of these three infants. 3. Despite clear emphasis in the pediatric literature (Volpe) on heightened susceptibility of basal ganglia structures to oxygen deprivation that has not been our experience or that of most observers. In both the MRS literature and in our own experience, anatomical location is less critical than adherence to a standardized MRS method in achieving reliable prognostic information. A series of singlevoxel MRS studies are illustrated in Figure 41.8 to demonstrate the impact of changing from stimulated echo acquisition mode (STEAM) to PRESS (at equivalent short TE) and of moving from voxels

containing predominantly GM in a watershed region, white matter (WM) in a posterior parietal location to basal ganglia. As discussed in Chapter 1, for the most part STEAM or PRESS localization techniques generally yield equivalent results, and in the three cases illustrated, spectra from all of the three locations, and using either PRESS or STEAM were equally reliable in predicting poor neurological outcome. Since interpretation of these spectra was mainly based on the peaks of NAA, Lac and Cr (which are present in both short and long TE spectra), it would appear that short TE is equally effective as long TE in diagnosis of hypoxic encephalopathy, although Lac is more readily resolved from lipids at long TE. 4. Age appears not to be an important differentiator when it comes to the application of 1H MRS to diagnosis of hypoxic encephalopathy. A systematic examination of the effect of near-drowning included children aged 18 months to 12 years (Kreis et al., 1996; Dubowitz et al., 1998). Patterns of adult hypoxic brain injury are illustrated in Figure 41.9. Three different adult patients are illustrated to demonstrate: (a) global hypoxic encephalopathy with normal MRI following airway obstruction (poor neurological outcome, patient died); (b) focal venous congestion (sagittal sinus thrombosis) with limited neurological injury (good neurological outcome after treatment); and (c) global hypoxic encephalopathy with abnormal MRI and focal MRS changes coinciding within the basal ganglia and striatum bilaterally (uncertain neurological outcome). Note the similarity between the MRS criteria of poor neurological outcome in the first adult patient and the preceding neonates. Venous thrombosis, despite the dramatic appearance of abnormalities on MRI, largely spares the neurons (NAA is well preserved since there is only minimal hypoxia), as shown in the second patient. Instead, there is massive accumulation of Lac, reflecting the major acid–base disturbance caused by accumulation of CO2 in the regions of venous congestion. Because of this preservation of neurons, neurological recovery was almost complete. The third patient suffered hypoxic brain injury during a prolonged coma induced by drug overdose. While the standard watershed GM and parietal WM

695

Patient 3: Poor outcome

Patient 4: Good outcome

Patient 5: Poor outcome

MR Units 1

3

2

4

0 Fig. 41.7 Value of multi-voxel MRS (chemical shift imaging, CSI) in hypoxic encephalopathy: multi-voxel MRS when acquired with the same MRS method (PRESS TE 35 ms) yields identical information. Three infants are compared, each with a clinical diagnosis of severed neurological injury after perinatal hypoxia. Patient 3 demonstrates pathological MR spectra over most of the volume examined. These volumes include those selected for single-voxel MRS (Figure 41.6) and indicated poor neurological outcome. Long-term follow-up confirmed both the clinical prediction and the MRS interpretation of severe neurological injury. However, Patient 4 showed normal MRS, and subsequently made a full neurological recovery. Patient 5 demonstrates an alternative method of displaying CSI data that focuses on select areas of the image such that spectra are enlarged and more interpretable.

Patient 6 : PRESS vs. STEAM

WM voxel position

WM PRESS

WM STEAM

Patient 7 : PRESS vs. STEAM

WM voxel position

WM STEAM

WM PRESS

Patient 8 : GM vs. WM vs. BG

4

3

2

1

0

4

3

2

4

3

2

1

0

ppm WM PRESS

ppm GM PRESS

GM and WM voxel position

1

0

ppm

BG voxel position

BG PRESS

Fig. 41.8 Anatomical location less critical than MRS method: a series of single-voxel MRS studies are illustrated to demonstrate the impact of changing from STEAM to PRESS (both at equivalent short TE) and of moving from voxels containing predominantly GM in watershed region, WM in posterior parietal location to basal ganglia.

698

Brian Ross, Cathleen Enriquez and Alexander Lin

(a)

WM: PRESS 35 ms 4

3

2 ppm

1

0

1

0

Cr

NAA Cho

NAA

Cr Cho mI

Glx

Age-matched Healthy Control

Glx

mI

Lac

0 GM: PRESS 35 ms

GM: PRESS 35 ms 4

3

2 ppm

1

0

4

3

2 ppm

Fig. 41.9 (a) Global hypoxia following airway obstruction (poor outcome, fatal). MRI was normal. T2-W localiser, Single voxel PRESS spectra from GM and WM showing elevated lactate and decreased NAA (normal GM spectrum for comparison). (b) Venous ischemia due to sagittal sinus thrombosis (good neurological outcome with treatment). (i) FLAIR localiser with 4 single voxels showing elevated lactate but preserved NAA in the posterior lesion, reduced NAA and elevated lactate in the anterior lesion and normal spectrum in normal appearing white matter. (ii) T2-W localiser and CSI showing elevated lactate within the focal lesion (first series), but not within adjacent medial normal-appearing tissue (second series). (c) Global hypoxic injury (uncertain neurological outcome). DWI and T2-W images show abnormality in watershed territories bilaterally. CSI shows heterogenous WM abnormality (elevated lactate / decreased NAA), generally more severe in the right hemisphere. Single voxel spectra show decreased NAA without detectable lactate in the parietal WM voxel, and relatively preserved metabolites in the posterior cingulate voxel (predominantly GM).

suffered only modest neuronal loss (NAA/Cr reduced by 15%), the striking MRI abnormalities were accompanied by up to 50% loss of NAA and significant lipid and Lac accumulations in 18/18 spectra selected from the MRSI dataset. This very unusual distribution of neuronal loss in a complex clinical setting

made prediction of neurological outcome difficult, if not impossible. To summarize the contributions of MRS studies of hypoxic brain injury: 1. “Pure’’ acute anoxic spectra confirm classic biochemical theory, but are often of little relevance

(b) (i)

probe–p TR:1500 TE:35

probe–p TR:1500 TE:35

probe–p TR:1500 TE:35

Anterior lesion

probe–p TR:1500 TE:35

Anterior lesion (small voxel)

NAWM

(ii)

1

CSI in lesion grid

Fig. 41.9(b) (cont.)

Posterior lesion

1

2

3

4

MRI 1

2

3

4

5

6

7

8

9

700

Brian Ross, Cathleen Enriquez and Alexander Lin

(c)

1

1

1

1

1

2

3

1

2

3

1

2

3

1

2

3

4

5

6

4

5

6

4

5

6

4

5

6

7

8

9

7

8

9

7

8

9

7

8

9

Single-voxel results

Posterior cingulate gyrus

Right parietal WM

Fig. 41.9(c) (cont.)

to diagnosis of human hypoxic brain injury; this is because most patients have been partially or completely resuscitated. 2. Neuronal-glial substrate cycling is active in hypoxic human brain: The use of neuronal cells in culture to study anoxia is out moded by the realization, now largely accepted that the neuron cannot be considered as a biochemical unit. Its most relevant processes of metabolism and

neurotransmission include the surrounding glia. Hence, Glx not Glu is the amine that accumulates to excess. In intact brain Glu toxicity is largely neutralized by Glx synthetase. Excess Gln is a useful diagnostic finding but, unlike Lac, appears to have no predictive value in children recovering from hypoxic brain damage (Kreis et al., 1996). 3. Lac is a reliable marker of prior hypoxic brain injury if present, but the converse may not be

MR spectroscopy of hypoxic brain injury

true, as Lac can be cleared from the brain by secondary events. Similarly, it is believed (Rothman et al., 1991) that Lac can reappear in “recovering’’ brain, perhaps as a result of leukocyte migration. 4. Short-chain fatty acid and/or triglyceride and other macromolecules seen in MR spectra provide confirming evidence, as yet incomplete, that a cascade of membrane related events including Ca++ and free-radical release activates phospholipases, and contributes to the irreversible neurochemical injury (Haseler et al., 1997). 5. Neuronal injury is indeed the most important longterm consequence of hypoxic–ischemic injury in the brain at all ages, from newborns, through young children and adolescents to the elderly and holds high predictive value of neurological outcome at 1 year. In contrast to anoxia, venous stasis with CO2 accumulation, a rare but important neurological emergency, may only lead to tissue acidosis and does not immediately reduce NAA. 6. Basal ganglia and hippocampus appear to be no more susceptible to hypoxia than other brain regions, so that MRS can safely be performed in “easier’’ locations. Exceptions may “prove-the-rule’’ however (cf. Figure 41.9c). Referring clinicians may still be more confident if basal ganglia are included in MRSI studies until MRS becomes more familiar to them.

Part 2. What is the diagnostic value of MRS in hypoxic brain injury? While 31P, 13C, 23Na and 1H MRS have all played an important role in elucidating mechanisms of human hypoxic brain injury, in diagnostic terms, only 1 H MRS has yet shown potential for routine clinical application. Since the early reports of long-TE (Grodd et al., 1991), short-TE (Kreis et al., 1991) or MRSI (Peden et al., 1990), a recent “PubMed” (National Library of Medicine, Bethesda, Maryland) literature search identified 137 published original papers, reviews and full length abstracts on 1H MRS in hypoxic–ischemic brain injury. Of these, 20 studies covering 459 patients found correlation between MRS performed early (1–20 days after birth) and neurological or neurodevelopmental outcome at up to

1 year (Table 41.1). In all subjects for whom inclusion criteria can be evaluated as meeting International Clasification of Disease (ICD)-9 codes for hypoxic– ischemic injury (ICD-9  348.1; 348.3; 348.5 plus E 950, 952, 910, 911–912, 954; ICD-9 767; 768; 768.5-. 6-.9) drowning, cardiac arrest, etc. the group mean data show reduced NAA (greater than 20%) and excess Lac (greater than 100%). These data are highly promising for the ability of 1H MRS to detect hypoxic–ischemic brain injury and predict prognosis.

Part 3. Efficacy of MRS in evaluation of hypoxic–ischemic brain injury Despite the literature cited above, none records in either abstract or text the term “efficacy”. Hence, in evidence-based analysis, MRS does not yet feature among the many approved diagnostic management tools for hypoxic brain injury. There are several reasons for this. First, we must consider competing technologies. Depending upon the age of the patient, clinical neurological examination, bed-side ultrasound for detection of intracerebral hemorrhage, infrared spectroscopy for assay of redox-state, CT for detection of hemorrhage, cortical atrophy and ventricular dilatation are the standard techniques now available to clinicians. In addition, advanced MRI techniques in particular diffusion-weighted MRI, are becoming an important tool for evaluation of hypoxic–ischemic injury in neonates (cf. Chapter 42). Therefore, for MRS to become widely accepted, further studies are needed to establish the additional diagnostic or prognostic value compared to these other techniques. Although 1H MRS studies need only to report efficacy equal to one other of the procedures to pass the test of efficacy-based medicine (EBM), the widespread adoption of MRS in evaluating hypoxic–ischemic injury will probably require demonstration of information unavailable from other techniques, in order to justify its additional cost and complexity. Only two published reports have been found which meets these stringent requirements. (Kreis et al., 1996; Ashwal et al., 2000). A multi-site trial involving a large number of cases would undoubtedly go a long way towards a more widespread acceptance of this methodology.

701

702

Brian Ross, Cathleen Enriquez and Alexander Lin

Table 41.1. Sensitivity and predictive value of MRS in hypoxic–ischemic encephalopathy. Source: PubMed, June 2003 Study

N

1 2 3 4 5 6 7 8 9 10

Cappellini et al., 2002 Robertson et al., 2002 Malik et al., 2002 Maneru et al., 2001 Cady, 2001 Groenendaal et al., 2001 Roelants-Van Rijn et al., 2001 Pu et al., 2000 Cady et al., 1997 Amess et al., 1999

20 59 16 18 ? 26 21 40 31 28

11 12 13 14 15 16 17 18 19 20

Chateil et al., 1999 Dubowitz et al., 1998 Hanrahan et al., 1998 Falini et al., 1998 Groenendaal et al., 1997 Suzuki et al., 1996 Penrice et al., 1996 Groenendaal et al., 1996 Felber et al., 1993 Peden et al., 1993

30 22 24 1 19 8 19 32 79 11

Major finding

Outcome measure

Interval

NAA/Cr  Lac/Cr Lac/Cr  pH Glx MRI; Lac NAA/Cho NAA; Lac; ATP NAA NAA; LAC Glx Lac/Cho; NAA Lac/NAA (93% specific; 92% PPV) NAA/Cho; Lac MRI vs. NAA/Cr Lac NAA NAA/Cho NAA/Cr; Cho/Cr Lac/NAA NAA/Cr; Lac NAA/Lac NAA/Cho

Neurology Neurology Neurology Neurology SARNAT Score SARNAT Score Clinical SARNAT Score Clinical Neurological

6 months 12 months 2 months 12 months 12 months 3 months ? ? 12 months 12 months

Neurological Neurological Clinical Neurology Neurology Neurology Neurology SARNAT Score (adults) Neurology

? 3 years 12 months ? 12 months ? 1 year ? ? 1 year

459

EBM also asks another question: does MRS improve patient outcome? For diagnostic imaging tests, this question is often difficult to answer, since arriving at a diagnosis does not always imply improved outcome, and also it raises difficult clinical trial design issues. MRS studies, which often measure outcome in the context of neurological sequence of the original insult, badly miss the point of outcome measure in EBM. Currently, there is only one report in which 1H MRS actually influenced patient outcome (Kreis et al., 1996) with approval of the local Hospital Ethics Committee. In that study, ventilator support was terminated if 1H MRS findings, clinical neurology and EEG all concurred in predicting vegetative outcome as the result of continued life-support following near-drowning. Finally, the complexity of researching efficacy in a disease when there is no treatment is well recognized. In the near term, it seems likely that the role of MRS may be to act as a selection criteria or

2–36 months

outcome measure (surrogate marker) for experimental therapy trials (e.g. neuroprotectants) in hypoxic–ischemic brain injury.

Conclusions Multinuclear MRS has illuminated a new syndrome of “secondary’’ hypoxic brain injury more allied to “no-reflow’’ phenomenon and suggesting different approaches to treatment. Despite promising preliminary results and a substantial amount of literature, the clinical efficacy of 1 H MRS remains to be proven.

ACKNOWLEDGEMENTS

The work described was funded by Rudi Schulte Research Institute, Santa Barbara, CA and the Jameson Foundation.

MR spectroscopy of hypoxic brain injury

REFERENCES Amess PN, Penrice J, Wylezinska M, Lorek A, Townsend J, Wyatt JS, Amiel-Tison C, Cady EB, Stewart A. 1999. Early brain proton magnetic resonance spectroscopy and neonatal neurology related to neurodevelopmental outcome at 1 year in term infants after presumed hypoxic–ischaemic brain injury. Dev Med Child Neurol 41(7): 436–445. Ashwal S, Holshouser BA, Shu SK, Simmons PL, Perkin RM, Tomasi LG, Knierim DS, Sheridan C, Craig K, Andrews GH, Hinshaw DB. 2000. Predictive value of proton magnetic resonance spectroscopy in pediatric closed head injury. Pediatr Neurol 23(2): 114–125. Bluml S, Moreno A, Hwang JH, Ross BD. 2001. 1-(13)C glucose magnetic resonance spectroscopy of pediatric and adult brain disorders. NMR Biomed 14(1): 19–32. Cady EB. 2001. Magnetic resonance spectroscopy in neonatal hypoxic-ischaemic insults. Childs Nerv Syst 17(3): 145–149. Cady EB, Amess P, Penrice J, Wylezinska M, Sams V, Wyatt JS. 1997. Early cerebral-metabolite quantification in perinatal hypoxic–ischaemic encephalopathy by proton and phosphorus magnetic resonance spectroscopy. Magn Reson Imaging 15(5): 605–611. Cady EB, Lorek A, Penrice J, Wylezinska M, Cooper CE, Brown GC, Owen-Reece H, Kirkbride V, Wyatt JS, Osmund E et al. 1994. Brain-metabolite transverse relaxation times in magnetic resonance spectroscopy increase as adenosine triphosphate depletes during secondary energy failure following acute hypoxia-ischaemia in the newborn piglet. Neurosci Lett 182(2): 201–204. Cappellini M, Rapisardi G, Cioni ML, Fonda C. 2002. Acute hypoxic encephalopathy in the full-term newborn: correlation between magnetic resonance spectroscopy and neurological evaluation at short and long term. Radiol Med (Torino) 104(4): 332–340. Chateil JF, Quesson B, Brun M, Thiaudiere E, Sarlangue J, Delalande C, Billeaud C, Canioni P, Diard F. 1999. Localised proton magnetic resonance spectroscopy of the brain after perinatal hypoxia: a preliminary report. Pediatr Radiol 29(3): 199–205. Danielsen ER, Ross BD. 1999. Magnetic Resonance Spectroscopy Diagnosis of Neurological Diseases. Marcel Dekker, New York. Dubowitz DJ, Bluml S, Arcinue E, Dietrich RB. 1998. MR of hypoxic encephalopathy in children after near drowning: correlation with quantitative proton MR spectroscopy and clinical outcome. Am J Neuroradiol 19(9): 1617–1627. Duffy TE, Nelson SR, Lowry OH. 1972. Cerebral carbohydrate metabolism during acute hypoxia and recovery. J Neurochem 19(4): 959–977.

Falini A, Barkovich AJ, Calabrese G, Origgi D, Triulzi F, Scotti G. 1998. Progressive brain failure after diffuse hypoxic–ischemic brain injury: a serial MR and proton MR spectroscopic study. Am J Neuroradiol 19(4): 648–652. Felber SR. 1993. 1H magnetic resonance spectroscopy in intracranial tumors and cerebral ischemia. Radiology 33(11): 626–632. Grodd W, Krageloh-Mann I, Klose U, Sauter R. 1991. Metabolic and destructive brain disorders in children: findings with localized proton MR spectroscopy. Radiology 181(1): 173–181. Groenendaal F, Roelants-Van Rijn AM, van Der Grond J, Toet MC, de Vries LS. 2001. Glutamate in cerebral tissue of asphyxiated neonates during the first week of life demonstrated in vivo using proton magnetic resonance spectroscopy. Biol Neonate 79(3–4): 254–257. Groenendaal F, van der Grond J, van Haastert IC, Eken P, Mali WP, de Vries LS. 1996. Findings in cerebral proton spin resonance spectroscopy in newborn infants with asphyxia, and psychomotor development. Ned Tijdschr Geneeskd 140(5): 255–259. Groenendaal F, van der Grond J, Eken P, van Haastert IC, Rademaker KJ, Toet MC, de Vries LS. 1997. Early cerebral proton MRS and neurodevelopmental outcome in infants with cystic leukomalacia. Dev Med Child Neurol 39(6): 373–379. Hanrahan JD, Cox IJ, Edwards AD, Cowan FM, Sargentoni J, Bell JD, Bryant DJ, Rutherford MA, Azzopardi D. 1998. Persistent increases in cerebral lactate concentration after birth asphyxia. Pediatr Res 44(3): 304–311. Haseler LJ, Arcinue E, Danielsen ER, Bluml S, Ross BD. 1997. Evidence from proton magnetic resonance spectroscopy for a metabolic cascade of neuronal damage in shaken baby syndrome. Pediatrics 99(1): 4–14. Heiss WD. 1983. Flow thresholds of functional and morphological damage of brain tissue. Stroke 14(3): 329–331. Hossman K-A. 1988. Pathophysiology of cerebral infarction. Handbook of Clinical Neurology vol 53; Vascular Disease Pt I. Elsevier, Amsterdam, pp. 27–46. Kreis R, Ernst T, Arcinue E, Lieberman R, Ross BD. 1991. Myoinositol in short TE 1H-MRS: a new indicator of neonatal brain development and pathology. In Proc Soc Magn Reson Med (2): 1007. Kreis R, Arcinue E, Ernst T, Shonk TK, Flores R, Ross BD. 1996. Hypoxic encephalopathy after near-drowning studied by quantitative 1H-magnetic resonance spectroscopy. J Clin Invest 97(5): 1142–1154. Malik GK, Pandey M, Kumar R, Chawla S, Rathi B, Gupta RK. 2002. MR imaging and in vivo proton spectroscopy of the brain in neonates with hypoxic–ischemic encephalopathy. Eur J Radiol 43(1): 6–13.

703

704

Brian Ross, Cathleen Enriquez and Alexander Lin

Maneru C, Junque C, Bargallo N, Olondo M, Botet F, Tallada M, Guardia J, Mercader JM. 2001. (1)H-MR spectroscopy is sensitive to subtle effects of perinatal asphyxia. Neurology 57(6): 1115–1118. Miller AL, Hawkins RA, Veech RL. 1973. The mitochondrial redox state of rat brain. J Neurochem 20(5): 1393–1400. Peden CJ, Cowan FM, Bryant DJ, Sargentoni J, Cox IJ, Menon DK, Gadian DG, Bell JD, and Dubowitz LM. 1990. Proton spectroscopy of the brain in infants. J Comput Assisted Tomogr 14: 886–894. Peden CJ, Rutherford MA, Sargentoni J, Cox IJ, Bryant DJ, Dubowitz LM. 1993. Proton spectroscopy of the neonatal brain following hypoxic–ischaemic injury. Dev Med Child Neurol 35(6): 502–510. Penrice J, Cady EB, Lorek A, Wylezinska M, Amess PN, Aldridge RF, Stewart A, Wyatt JS, Reynolds EO. 1996. Proton magnetic resonance spectroscopy of the brain in normal preterm and term infants, and early changes after perinatal hypoxia-ischemia. Pediatr Res 40(1): 6–14. Pu Y, Li QF, Zeng CM, Gao J, Qi J, Luo DX, Mahankali S, Fox PT, Gao JH. 2000. Increased detectability of alpha brain glutamate/ glutamine in neonatal hypoxic–ischemic encephalopathy. Am J Neuroradiol 21(1): 203–212.

Robertson NJ, Cowan FM, Cox IJ, Edwards AD. 2002. Brain alkaline intracellular pH after neonatal encephalopathy. Ann Neurol 52(6): 732–742. Roelants-Van Rijn AM, van der Grond J, de Vries LS, Groenendaal F. 2001. Value of (1)H-MRS using different echo times in neonates with cerebral hypoxia-ischemia. Pediatr Res 49(3): 356–362. Rothman DL, Howseman AM, Graham GD, Petroff OA, Lantos G, Fayad PB, Brass LM, Shulman GI, Shulman RG, Prichard JW. 1991. Localized proton NMR observation of [3-13C]lactate in stroke after [1–13C]glucose infusion. Magn Reson Med 21(2): 302–307. Siesjo BK. 1988. Historical overview. Calcium, ischemia, and death of brain cells. Ann N Y Acad Sci 522: 638–661. Suzuki S, Ichijo M, Fujii H, Ikeda K, Hitosugi M. 1996. Cerebral metabolic disturbance in hypoxic encephalopathy: evaluation with H-1 MR spectroscopy. Rinsho Shinkeigaku 36(7): 844–849. Volpe JJ. Hypoxic–ischemic encephalopathy: clinical aspects. In: neurology of the newborn. 3rd ed. Philadelphia. WB. Saunders; pp. 314–369.

MR spectroscopy of hypoxic brain injury

Case Study 41.1 Reye’s syndrome: MRSI Peter Barker, D.Phil., David Hearshen, Ph.D. and Suresh Patel, M.D., Henry Ford Hospital, Detroit History Comatose 3 year old female, enlarged liver, 1 week after administration of aspirin for a viral infection.

Lac

Cho

Technique Conventional MRI and multi-slice MRSI (TE 280 ms).

Imaging findings MRI (not shown) exhibited brain swelling, herniation and lack of flow voids in the major intracranial vessels. Long TE MRSI shows a global absence of NAA and increase in Lactate, consistent with severe ischaemic neuronal injury. Choline and Cr are unremarkable. The patient died a few days later.

NAA

Lac

Cho Cr

Discussion Reye’s syndrome (RS) is an acute encephalopathy and fatty degeneration of the liver in children. Initially, profuse vomiting occurs, accompanied by a varying degree of neurological impairment, ranging from irritability, to coma and death. If recognized early, it may be successfully treated. Brain spectra in the acute phase are similar to those of adult patients with hepatic encephalopathy (Kreis, 1990), with elevated Gln (due to increased blood ammonia levels), and decreased Cho (Kreis, 1995). The case presented here shows the end stages of the disease, 7 days after symptom onset, with brain swelling, a large, global increase in Lac and virtually absent NAA. These changes are consistent with braindeath.

ppm

3.0

2.0

Key points RS is reversible if diagnosed early. Early RS shows elevated Gln and decreased Cho on short TE MRS. The late RS case shown here is consistent with brain death, with an absence of NAA and elevated Lac throughout the brain.

References Ernst T, Ross BD, Flores R. 1992. Cerebral MRS in an infant with suspected Reye’s syndrome. Lancet 340(8817): 486. Kreis R, Farrow N, Ross BD. 1990. Diagnosis of hepatic encephalopathy by proton magnetic resonance spectroscopy. Lancet 336(8715): 635–636. Kreis R, Pfenninger J, Herschkowitz N, Boesch C. 1995. In vivo proton magnetic resonance spectroscopy in a case of Reye’s syndrome. Intens Care Med 21(3): 266–269.

1.0

705

42

The role of diffusion and perfusion weighted brain imaging in neonatology Mary A. Rutherford and Serena J. Counsell Robert Steiner MR Unit, Imaging Sciences Department, Clinical Sciences Center, Imperial College, Hammersmith Hospital, London, UK

Key points • Diffusion weighted imaging (DWI) is easy to perform on the neonatal brain and should be part of the routine MR examination. • Apparent diffusion coefficient (ADC) values decrease and regional anisotropy values increase with increasing age reflecting brain maturation. • DWI is clinically useful for identifying ischemic tissue in the neonatal brain, the pattern of which can predict outcome. • ADC values should always be measured in addition to visual analysis of the DW images. • ADC values 1 10 3/mm2/s are associated with hemispheric white matter infarction and values 0.8 10 3/mm2 with thalamic damage. • Techniques such as, diffusion tensor imaging, 3 T imaging and high b value imaging will have important applications for clinical neonatology. • Perfusion weighted imaging is challenging but already allows insights into the response of the immature brain to hypoxic–ischemic injury.

establishing patterns of perinatal injury and predicting outcome (Barkovich, 1992; Kuenzle et al., 1994; Mercuri et al., 1995, 2000, 2002; Rutherford et al., 1996, 1998). MRI provides detailed information about the pattern of lesions following perinatal brain injury (Barkovich, 1992; Kuenzle et al., 1994; Rutherford et al., 1996; Rutherford, 2002) and is an excellent predictor of outcome in infants with hypoxic–ischemic encephalopathy (HIE) (Rutherford et al., 1998; Mercuri et al., 1999, 2000). Conventional MRI has also been used to study perinatal stroke, later hemiplegia develops if there is involvement of three sites namely, hemispheric white matter (WM), basal ganglia and thalami (BGT) and posterior limb of the internal capsule (PLIC) (Rutherford, 2002). In preterm infants with unilateral focal lesions the development of a hemiplegia is related to the MR signal intensity within the ipsilateral PLIC at term equivalent age (De Vries et al., 1999). Diffusion weighted imaging (DWI) may be used to assess the developing brain and its response to injury but is not yet routinely used to predict outcome. Whilst perfusion weighted imaging (PWI) may have many applications in studies of the neonatal brain, there are very few published studies using PWI in the very immature brain (Tanner et al., 2000).

Introduction Practical issues MR imaging (MRI) of the neonatal brain is a relatively new field but there are now many publications that illustrate its role in defining malformations, 706

Successful imaging of the neonatal brain requires careful preparation of the infant and knowledge of

The role of diffusion and perfusion weighted brain imaging in neonatology

(a)

(b)

Table 42.1. ADC values (median (range)) in different brain regions in all infants Region

Fig. 42.1 Normal appearance to the neonatal brain at term T1-weighted (spin echo (SE), SE 500/15) sequence (a) and T2-weighted (fast spin echo (FSE), FSE 4200/210) sequence (b).

the normally developing brain (Rutherford, 2002). These issues are not insurmountable but require close cooperation between radiologist, radiographer and neonatologist. Neonates are uncooperative. A successful image relies on a still baby. To this end neonates may be successfully imaged during natural sleep, following a feed or under light sedation with, for instance, chloral hydrate. All neonates, sedated or not, should be monitored during scanning with MR compatible pulse oximetry and electrocardiogram (ECG). A qualified pediatrician should be in attendance throughout the scan. Excessive noise, particularly with fast sequences, may wake a sleeping infant or even harm the developing auditory system and ear protection should be used. We use moldable dental putty as individualized earplugs and neonatal earmuffs. Infants may move even when asleep: molded air bags or foam placed snugly around the infant’s head will keep this to a minimum. Swaddling the infants will keep them warm and also reduce movements. All the usual metal checks need to be carried out with particular attention, in this population, to the presence of intravenous scalp lines, long lines, electroencephalograms (EEG) electrodes, intraventricular shunts and metal fasteners on baby clothes. Neonates are small: the average term born neonate, weighing approximately 3.5 kg. Improved signal to noise will be obtained by using as small a coil as possible. In the absence of a dedicated pediatric coil an adult knee coil is well suited, however these may not

Controls ADC

All patients ADC

( 10 3 mm2/s)

( 10 3 mm2/s)

Thalami VLN Lentiform PLIC WM CSO WM ant WM post Cerebellar Hemispheres Vermis

1 (1–1.15) 0.88 (0.76–0.95) 1.1 (1–1.3) 1 (0.83–1.2) 1.5 (1.3–1.7) 1.6 (1.5–1.7) 1.55 (1.35–1.85)

1 (0.5–1.4) 0.85 (0.39–1.2) 1.05 (0.5–1.65) 0.9 (0.48–1.5) 1.43 (0.5–2.0) 1.5 (0.6–1.95) 1.5 (0.5–1.9)

1.1 (1–1.25) 0.97 (0.8–1.2)

1 (0.8–1.3) 0.98 (0.7–1.2)

Brainstem

0.98 (0.86–1.1)

0.92 (0.5–1.25)

VLN: ventrolateral nuclei of thalami; CSO: centrum semi-ovale; ant: anterior; post: posterior.

be large enough to accommodate an endotracheal tube in infants who are ventilated. MR sequences will need to be adjusted for neonatal brain imaging. The neonatal brain has high water content and tissue signal intensity contrasts are very different from the mature adult brain (Figure 42.1). The neonatal brain is largely unmyelinated (Figure 42.1) but reaches near adult levels of myelination by the end of the second year. The pathologies of the immature brain also differ. Abnormalities are often symmetrical and may be mistaken for normal appearances by the inexperienced observer. In addition the evolution of perinatal pathology takes place on the background of a brain that is developing rapidly both in terms of actual size and in its tissue properties.

DWI The normal neonatal brain DWMRI techniques have been used to study the normal neonatal brain (Tanner et al., 2000; Forbes et al., 2002; Miller et al., 2003). Apparent diffusion coefficient (ADC) values are higher than in more mature brain. There is regional variation: ADC values are lower in gray matter (GM) than in myelinated WM and highest in unmyelinated WM. ADC values for normal term brain are shown in Table 42.1. These control

707

708

Mary A. Rutherford and Serena J. Counsell

values compared favorably with those from other studies (Tanner et al., 2000; Forbes et al., 2002). All three studies obtained DWI at 1.5 T with two b values (0 and 1000 mm2/s). In Forbes’ study of 40 children from birth to 1 year, 14 infants were term born but all were scanned for clinical reasons and were therefore not true controls. Their ADC values for subcortical frontal WM were slightly higher than the values in Table 42.1 (median 1.88 compared to 1.6 10 3/mm2/s respectively) but for the PLIC, ADC values were comparable (median 1.09 vs. 1.0 10 3/mm2/s). Forbes et al. also noted an increase in ADC values within anterior WM compared with posterior WM. The median ADC in controls was 1.6 for anterior WM (Table 42.1), which was not significantly different from the 1.55 for posterior WM. Variation in regions of interest which were measured within the WM may explain some of the differences between these studies. Tanner’s study measured ADC values in 10 term born neurologically normal neonates, less than 43 days old, and their values for anterior WM were comparable to those in Table 42.1 (1.62 vs. 1.6 10 3/mm2/s). ADC values reach mature adult levels by about 2 years although small decreases may still be found until early adulthood. There are few diffusion tensor imaging (DTI) studies of the normal term brain. However, anisotropy of WM tracts has been identified prior to myelination (Huppi et al., 1998; Mukherjee et al., 2002; Miller et al., 2003). Measurements of regional anisotropy (RA) are lower than in the adult brain in WM and in central GM (DeLano and Cao, 2002). The main clinical role for DWI in neonatal medicine is in establishing the presence of hypoxic– ischemic injury. Hypoxic–ischemic lesions in the term baby Following a perinatal hypoxic–ischemic insult, abnormalities detected with conventional MRI may take several days to evolve, a period during which maximum benefit from neuroprotective strategies, such as hypothermia, to modify brain injury may be obtained and when important clinical decisions have to be made. Very early confirmation of the site and severity of tissue injury enables appropriate targeting of therapeutic interventions in a far more specific

way than is currently available. In addition even established abnormalities may not be obvious to the inexperienced radiologist. An additional and more objective method of assessing tissue integrity early after injury, such as DWI, is therefore very useful. Neonates with perinatal hypoxic–ischemic injury often present with seizures. The pattern of lesions sustained depends on the nature of the injury and this may be reflected in the clinical history and findings. Those infants who present with seizures but have near normal Apgar scores usually show WM lesions such as perinatal stroke or parasagittal infarction. The WM lesions may occasionally be hemorrhagic (Mercuri et al., 1995). Infants with depressed Apgar scores and a necessity for resuscitation at delivery present with seizures as part of a more generalized encephalopathy. Infants with such HIE usually sustain bilateral BGT lesions with or without WM infarction. Perinatal stroke DWI has been used to assess tissue injury in neonates with perinatal stroke (Cowan et al., 1994; Krishnamoorthy et al., 2000). DWI abnormalities are most obvious 1–4 days after delivery at a time when conventional imaging may not be that abnormal (Figure 42.2). ADC values may decrease to approximately 30% of normal. Abnormal signal intensity gradually reduces by the end of the first week as the conventional imaging appearances become more abnormal. The evolution of the diffusion abnormality in perinatal stroke is consistent with and appears to be similar to that seen in adults (Fiesbach et al., 2002). It has been suggested that ADC values may pseudonormalize more quickly than in the adult patient (Mader et al., 2002). However, as the etiology of perinatal stroke is poorly understood and seizures may not become clinically obvious until 48 h, it is difficult to time the onset of infarction. DWI may appear to overestimate the size of a perinatal infarct (Figure 42.3) but this is difficult to prove in the face of a growing and developing brain. Registration of serially acquired images have shown excessive growth in and around areas of infarction following perinatal stroke (Rutherford et al., 1997a; 1997b) but the infarcted hemisphere is always smaller and shows less myelin on follow up images (cf. Case Study 42.1).

The role of diffusion and perfusion weighted brain imaging in neonatology

(a)

(b)

(c)

(d)

Fig. 42.2 5-day old infant with left MCA infarct (arrows). The abnormalities are quite subtle on the T1-weighted (SE 15/500) images (a) more obvious on the T2-weighted (FSE 4200/210) images (b) but most obvious on the DWI (c) and ADC trace (d) map.

(a)

(b)

(d)

(e)

(c)

Fig. 42.3 DWI in an infant with a right middle cerebral artery (MCA) infarct aged 2 days (a) and 4 days (b) and 6 weeks (c) T2-weighted imaging at 6 weeks (d) shows a small posterior infarct which appears to be much smaller than the original diffusion abnormality and at 6 months (e) when all that remains is a smaller hemisphere with some loss of WM and asymmetry of the cortical folding.

709

710

Mary A. Rutherford and Serena J. Counsell

(a)

(b)

(c)

(d)

DWI may also show changes in the brainstem in acute stroke consistent with Wallerian degeneration (Figure 42.4). This may not influence the outcome which is related to the actual sites of infarction. Perinatal stroke only results in hemiplegia if there is involvement of three sites hemisphere, BGT and PLIC (Mercuri et al., 1999). This study used conventional MRI only and it is not known whether this rule applies to abnormalities that may only be visible on DWI (Figure 42.4). The combination of DTI and functional imaging shows promise for understanding the surprisingly good outcomes after perinatal stroke (Seghier et al., 2004).

HIE In infants with a global hypoxic–ischemic insult the most frequent site of injury is to the central GM. Bilateral BGT lesions are strongly associated with the development of motor impairment and the extent of the lesions is closely related to the severity of that impairment (Rutherford et al., 1996). Severe WM infarction leads to cognitive impairment but a less severe motor impairment (Mercuri et al., 2000). There are relatively few studies using DWI in infants with HIE and these have had conflicting results (Forbes et al., 2000; Barkovich et al., 2001; Wolf et al., 2001; Mckinstry et al., 2002; Takeoka et al., 2002; Zarifi et al., 2002). DWI may show obvious abnormalities following bilateral WM infarction (Figure 42.5) in HIE but not if the BGT are also affected (Figure 42.6).

Fig. 42.4 Wallerian degeneration (acute phase). MCA infarct in an infant aged 6 days. T1-weighted (SE 500/15) imaging (a). T2-weighted (FSE 4200/210) imaging (b) DWI showing abnormal high signal intensity within the left internal capsule (c) There is an additional abnormal high signal intensity in the left crus (arrow) (d).

(a)

(b)

(c)

(d)

Fig. 42.5 Three-day old infant with neonatal encephalopathy. Severe WM infarction. The changes are subtle on T1-weighted (SE 500.15) images (a) but more obvious on T2-weighted (FSE 4200/210) images with some loss of GM/WM differentiation (b). On DWI there is bilateral marked abnormal high signal intensity throughout the hemispheres (c). This was consistent with infarction which is evident on T1-weighted (SE 500/15) images at 3 weeks (d).

The role of diffusion and perfusion weighted brain imaging in neonatology

(a)

(d)

(c)

ADC 0.9 ADC 0.65

ADC 0.53 (b)

ADC 0.85

Fig. 42.6 Serial imaging in a neonate with global perinatal hypoxic–ischemic injury. (a) T1-weighted (SE 500/15) imaging. (b) T2-weighted (FSE 4200/210) imaging (c) DWI does not show any focal high signal intensity at 2 days of age despite reduced ADC values (range 0.53–0.9) and subsequent tissue breakdown at 15 days of age (d).

In these infants a clue may be found by observing the appearance of the normal cerebellum (Vermeulen et al., 2003) (Figure 42.7) and measuring the ADC values will correctly detect the presence of ischemic tissue. DWI appearances in the presence of isolated but clinically significant BGT lesions may not be abnormal (Figure 42.8). In a recent study we studied the relationship between contemporaneous DWMRI and conventional MRI in 63 term born neonates infants with HIE and compared the results to control term born infants during the neonatal period. DWI was acquired using single shot echo planar imaging (EPI) at multiple levels. Fifteen slices of 5 mm thickness were

(a)

(b)

Fig. 42.7 DWI in an infant with widespread ischemia showing white cerebrum with widespread markedly reduced ADC values 1 10 3/mm2/s and a normal cerebellum (arrow).

711

Mary A. Rutherford and Serena J. Counsell

1.5

(b) ADC thalami (10 3 mm2/s)

(a)

Fig. 42.8 Mild BGT lesions with foci of increased signal intensity on early T1-weighted (SE 500/15) images (arrows). There are no corresponding areas of high signal on the DWI Trace image. ADC values are often within the normal range in mild and moderate BGT lesions, even when imaged early. These BGT lesions are usually associated with some motor impairment.

2.0

ADC (10–3 mm2/s)

712

1.5

1.0

0.5

0.0 0

5 10 Age at time of scan in days

Controls Normal WM

15

Moderate WM abnormalities Severe WM abnormalities

Fig. 42.9 ADC values for posterior WM in infants with HIE and controls.

obtained (repetition time (TR) 6000 ms, echo time (TE) 110 ms, field of view (FOV) 24 cm, b values of 0 and 1000 s/mm2) in three orthogonal directions. In infants with HIE, ADC values were significantly reduced in the first week following severe injury to either WM (P  0.0001) or BGT (P  0.0001) but values normalized at the end of the first week and then increased during week two (Figures 42.9, 42.10).

Normal BGT Moderate BGT lesions Severe BGT lesions

1.3 1.1 0.9 0.7 0.5 0

10 20 30 Age at time of scan in days

40

Fig. 42.10 Thalamic ADC vs. age at scan in infants with HIE and controls.

ADC values were either normal or increased in moderate BGT and WM lesions when compared to controls. ADC values 1.1 10 3 mm2/s were always associated with WM infarction and values 0.8 10 3 mm2/s with thalamic infarction. We also found that DWI was not very helpful in confirming the presence of either moderate WM or moderate BGT lesions. Moderate WM abnormalities are associated with relatively good outcome, normal motor development but higher risk of cognitive impairment. However, moderate BGT lesions are usually associated with significant motor impairment in the form of quadriplegic cerebral palsy and it is therefore very important to be able to correctly diagnose their presence. It is essential therefore to consider the timing of the injury, the age of the child, the conventional imaging appearances and to look at the DWI in addition to measuring the ADC. This combined approach will improve the predictive abilities of diffusion techniques to correctly identify ischemic tissue. Preterm brain injury There are few MR studies in preterm brain pathology because of difficulties in safely transporting and scanning small and often sick preterm neonates (Inder et al., 2000; Roelants-van Rijn et al., 2001; Miller et al., 2002). DWI has been used to identify areas of restricted diffusion in infants with periventricular leukomalacia (PVL) prior to cyst formation

The role of diffusion and perfusion weighted brain imaging in neonatology

(a)

(b)

(c)

(d)

Fig. 42.11 Preterm infant with cystic PVL. Imaged at 33 weeks with T1-weighted (SE 500/15) (a) T2-weighted (FSE 4200/210) (b) and diffusion weighted (c) sequences. There are obvious cysts (arrows) on the conventional images (a) (b). On DWI there are additional areas of abnormal high signal intensity (arrows) adjacent to the ventricles consistent with pre-cystic ischemic tissue. At term T2-weighted imaging shows further ventricular dilatation and loss of WM (arrows) in the region of the previous DWI abnormalities.

(Inder et al., 2000; Roelants-van Rijn et al., 2001). DWI may show areas of restricted diffusion adjacent to established cystic lesion in PVL (Figure 42.11). DWI has also been used to identify abnormal WM in areas distant from cystic lesions the ADC values in these areas being higher than normal tissue (Counsell et al., 2003). Whilst DWI provides most clinical benefit by identifying areas of ischemia it may also be used to characterize brain development in the preterm infant without lesions. The clinical significance of preterm lesions such as PVL and hemorrhagic venous infarction is fully appreciated: the former giving rise to a spastic diplegia and the latter to hemiplegia in those where the infarct involves the corticospinal tracts.

Whilst the imaging correlates for these motor impairments are recognized there is as yet no imaging correlate for the neurocognitive and neurobehavioral disorders that frequently follow preterm delivery. MRI techniques may allow us to detect abnormalities responsible for these non-motor impairments. The preterm infant when imaged at term may show a diffuse long T1, long T2 within the WM on conventional imaging, so called diffuse excessive high signal intensity (DEHSI) in the absence of any focal lesions (Maalouf et al., 1999). It has been reported that the preterm infant without focal lesions, imaged at term equivalent age, has higher ADC values in WM than term born controls (Huppi et al., 1998). We have recently shown that preterms at term with DEHSI have high ADC values compared to term born controls, but preterms without DEHSI had values that were comparable to controls (Huppi et al., 1998; Counsell et al., 2003). This suggests that there is a subset of preterm infants who show abnormal WM development even when there is no focal pathology such as cystic PVL. Further long-term studies are needed to ascertain the clinical significance of this abnormal development. Congenital malformations In addition to identifying areas of acute ischemia in the immature brain DWI has a few additional clinical roles. In infants with callosal agenesis DWI may be used to identify abnormal tracts, which run medial to the lateral ventricles giving them their characteristic shape, so called Probst bundles (Figure 42.12). Metabolic disorders Conventional MRI is able to identify brain malformations that may accompany some metabolic disorders, such as callosal agenesis in non-ketotic hyperglycinaemia (NKH) or cortical migration defects in perioxisomal disorders such as Zellweger’s syndrome. WM in these infants may have a long T1 and long T2 and this may be reflected in increased ADC values. Occasionally the results of DWI are more enlightening and show more distinct abnormalities. These may be reversible with treatment and could potentially be used to monitor therapy.

713

714

Mary A. Rutherford and Serena J. Counsell

(b) (a)

Fig. 42.12 Infant with agenesis of the corpus callosum (a) Fetal T2-weighted EXPRESS (half-fourier single-shot turbo spin echo) sequence. Transverse plane. There is absence of callosal fibers crossing the midline and an abnormal orientation to the ventricles (b) Diffusion-weighted image at term showing WM tracts running parallel and medial to the lateral ventricles (arrow). These are known as Probst bundles.

(a)

(b)

peduncles, internal capsule, may mimic those found with more straightforward HIE and therefore the diagnosis of NKH may be missed. This further emphasizes the need to fully investigate any infant presenting with neonatal seizures, not only with a careful history and examination, but with metabolic investigations as well as neuroimaging. The latter should where possible be MR imaging and include DWI.

Future developments and applications Fig. 42.13 Maple syrup urine disease (MSUD). Term born neonate presenting with seizures on day 5 following an uneventful delivery. (a) T2-weighted (FSE 4200/210) sequence. There is abnormal high signal intensity throughout the PLIC (arrow). There is no low signal intensity from myelin. (b) There is obvious high signal intensity on the diffusion-weighted image. This would be consistent with cytotoxic edema, although these changes have been shown to be reversible. Similar changes throughout all of the early myelinating areas in the brain were observed.

In infants with maple syrup urine disease (MSUD)DWI is pathognomonic resulting in a “map” of actively myelinating areas (Cavalleri et al., 2002) (Figure 42.13). There are also reports of abnormal DWI in NKH (Khong et al., 2003). This is an important finding as the areas of abnormality e.g. cerebral

Fetal DWI There have now been several studies that have used DWI to assess the brain of the preterm infant at term equivalent age. This has highlighted differences in ex utero development of the immature brain. In these studies term born infants have acted as controls. This is not ideal for following the developmental processes between 23 and 40 weeks. The ideal control would be to study the brain developing in utero with fetal MRI. Fetal MRI is mainly used to confirm or detect abnormalities suspected on antenatal ultrasound. There has been one report to date using DWI of the normal fetal brain (Baldoli et al., 2002) and one of an abnormal fetus (Righini et al.,

The role of diffusion and perfusion weighted brain imaging in neonatology

Fig. 42.14 DTI at 3 T. WM tractography showing the fiber orientation in the PLIC in a preterm infant at 2 years of age.

2004). Fetal DWI is challenging because of the effects of fetal motion, in some centers successful DWI of the fetus is only obtained in the presence of maternal sedation. It does have promise as a tool for assessing early ischemic change in the fetal brain either in association with a possible injury e.g. maternal trauma or in high-risk pregnancies where ischemia may occur as a complication e.g. monochorionic twins (McKinstry et al., 2002). DTI and WM tracking DTI studies of the very immature brain have shown increased RA in the developing cortex consistent with the simple radiating organization of fibers (McKinstry et al., 2002). As for the adult brain, DTI may be used in order to map the WM tracts in the immature brain and may have many useful applications for understanding normal development or the response of the brain to injury (Huppi et al., 2001; Bammer et al., 2003; Zhai et al., 2003) (Figure 42.14). Imaging at 3 T Imaging at 3 T opens up possibilities for improving signal to noise in neonatal imaging. This may be

used to shorten examination times. For diffusion sequences imaging at 3 T, allows sufficient signal to noise to increase b values to over 1000 mm2/s. Initial experience shows that the immature brain shows marked changes in normal tissue signal intensity with increasing b values (Figure 42.15). In addition, as has been reported in the adult literature, lesion conspicuity may be improved with higher b values (Burdette and Elster, 2002; DeLano and Cao, 2002) (Figure 42.16). Further improvements may also be made by using sensitivity encoding (SENSE) imaging (Jaermann et al., 2004).

PWI PWI is used in adult stroke usually in combination with DWI to assess tissue viability. The majority of studies have used a contrast-enhanced technique. At present non-contrast methods such as arterial spin labeling (ASL) remain problematic because of poor signal to noise. This is likely to be even more of a problem in neonatal brain studies, although imaging on a 3 T scanner may improve results.

715

716

Mary A. Rutherford and Serena J. Counsell

350

700

1500

3000

Fig. 42.15 Diffusion tensor imaging of a normal brain of a term born infant imaged at 3 T. Single shot EPI (TR 2500/TE 100) Slice thickness 4 mm. Signal averages 2–8. Variation in tissue signal intensity with increasing b value (range 350–3000 mm2/s).

200

350

500

700

1000

1250

2000

3000

Fig. 42.16 3 T diffusion tensor imaging. Single shot EPI (TR 2500/TE 100). Slice thickness 4 mm. Signal averages 2–8. MCA infarct at different b values (range 300–3000 mm2/s) in a term infant aged 5 days. Lesion conspicuity increases with increasing b value. Hyperintensity is also seen in the splenium of the corpus callosum, which may be due to acute Wallerian degeneration, diaschisis or edema. On follow-up MRI the corpus callosum was normal.

The role of diffusion and perfusion weighted brain imaging in neonatology

Intensity vs. Time 519

Start

Stop

Last

2  WM

1  Basal ganglia

331 0.00

11.5

23.0

34.6

46.1

57.6

69.1

80.6 s

Fig. 42.17 “Normal” perfusion at 10 days shows regional variation in the signal intensity vs. time following contrast administration.

As yet there is only one published study using PW MRI to study perfusion in the neonatal brain. (Tanner et al., 2003) The main difficulties with PWI studies of the neonatal brain are motion artifact, difficulties in obtaining data of the normal brain, and problems with quantification. We perform PWI using a gradient-echo (T*2 )weighted EPI (5 mm thickness/gap 2.5 mm, TR 1000 ms, TE 37.3 ms, FOV 24 cm). Gadolinium dimeglumine gadopentetate (Gd-DTPA), (0.2-mls/kg), is injected by hand via an available intravenous line in the lower arm or foot 15 s after the start of the sequence, allowing a consistent baseline to be achieved. We do not use a power injector for our studies and have not as yet given a small predose of contrast to avoid T1 effects caused by a leaky blood–brain barrier (BBB) (Figure 42.18). We collect 10 slices through the brain, 75–80 frames were acquired per slice. Perfusion of the neonatal brain shows regional variation (Figure 42.17). Neonates with extensive WM

infarction may show both under and over perfusion (Figure 42.18). In subacute scans performed at around 5 days this does not appear to relate to the timing of the injury but those with severely underperfused WM tend to have severe brain swelling, this may result in a secondary under-perfusion from impedance to blood flow. We have also demonstrated marked increased blood flow in areas of focal infarction that show markedly reduced ADC values (Figure 42.19). However, it is likely that if perfusion studies were performed on day 1 after injury hypoperfusion would be detected in early infarction. Just as in adults, semi-quantitative measurements of mean transit time (MTT), cerebral blood flow (CBF) and cerebral blood volume (CBV) using an arterial input factor (AIF), placed in the middle cerebral artery (MCA) are possible (cf. Chapter 4). In neonates tit may also be helpful to use software to check and/or correct for head motion during the perfusion scan. One advantage of neonatal studies is that pathology is not usually due to vascular disease, so that a more reliable

717

718

Mary A. Rutherford and Serena J. Counsell

512

(b)

(a)

Basal ganglia WM 302 0.00 11.5 23.0 34.6 46.1 57.6 69.1 80.6 s Fig. 42.18 (a) WM hyperperfusion in an infant with severe BGT and WM lesions aged 5 days; WM ADC values were 1 10 3 mm 2 s 2. (There is overshoot of the post peak baseline consistent with a T1 effect from BBB breakdown) (b) 10 days later there has been extensive WM infarction.

(a)

(b) Intensity vs. Time 816

Last

1 2

383 0.00 11.5 23.0 34.6 46.1 57.6 69.1 80.6 s Fig. 42.19 PWI in a 5-day old infant with a left MCA infarction seen on DWI (a) The ADC value in the center of the infarct was reduced at 0.56 10–3/mm2/s, less than 50% of normal (a). Of interest is that there is greater perfusion through the center of the infarct (1) than through the apparently normal WM on the right (2) (b).

AIF may perhaps be obtained. Preliminary results have shown regional variation in MTT. In those infants with normal imaging (N  4) the mean MTT was almost doubled in the WM (13.5 s) compared to the BGT (5.7 s). The pathologies and the response of the immature brain are very different from those in the adult and PWI is a suitable technique for further understanding these processes. Further insights will be made by comparing the results of PWI to those, from both conventional and DWI.

ACKNOWLEDGEMENTS

We would like to thank Professor Graeme Bydder and all the staff of the Robert Steiner MR Unit, Hammersmith Hospital. In particular, Joanna Allsop, David Larkman, Yuji Shen, Jo Hajnal, David Edwards and Frances Cowan. I am also grateful to The Medical Research Council, The Academy of Medical Sciences, The Health Foundation and Philips Medical Systems for their support.

The role of diffusion and perfusion weighted brain imaging in neonatology

REFERENCES Baldoli C, Righini A, Parazzini C, Scotti G, Triulzi F. 2002. Demonstration of acute ischemic lesions in the fetal brain by diffusion magnetic resonance imaging. Ann Neurol 52: 243–246. Bammer R, Acar B, Moseley ME. 2003. In vivo MR tractography using diffusion imaging. Eur J Radiol 45: 223–234. Barkovich AJ. 1992. MR and CT evaluation of profound neonatal and infantile asphyxia Am J Neuroradiol 13: 959–972. Barkovich AJ, Westmark KD, Bedi HS, Partridge JC, Ferriero DM, Vigneron DB. 2001. Proton spectroscopy and diffusion imaging on the first day of life after perinatal asphyxia: preliminary report. Am J Neuroradiol 22: 1786–1794. Burdette JH, Elster AD. 2002. Diffusion-weighted imaging of cerebral infarctions: are higher B values better? J Comput Assist Tomogr 26: 622–627. Cavalleri F, Berardi A, Burlina AB, Ferrari F, Mavilla L. 2002. Diffusion-weighted MRI of maple syrup urine disease encephalopathy. Neuroradiology 44: 499–502. E pub 2002 Apr 24. Counsell SJ, Allsop JM, Harrison MC, Larkman DJ, Kennea NL, Kapellou O, Cowan FM, Hajnal JV, Edwards AD, Rutherford MA. 2003. Diffusion-weighted imaging of the brain in preterm infants with focal and diffuse white matter abnormality. Pediatrics 112: 1–7. Cowan FM, Pennock JM, Hanrahan JD, Manji KP, Edwards AD. 1994. Early detection of cerebral infarction and hypoxic–ischemic encephalopathy in neonates using diffusion-weighted magnetic resonance imaging. Neuropediatrics 25: 172–175. De Vries LS, Groenendaal F, van Haastert IC, Eken P, Rademaker KJ, Meiners LC. 1999. Asymmetrical myelination of the posterior limb of the internal capsule in infants with periventricular haemorrhagic infarction: an early predictor of hemiplegia. Neuropediatrics 30: 314–319. DeLano MC, Cao Y. High b-value diffusion imaging. 2002. Neuroimag Clin N Am 12: 21–34. Fiesbach JB, Jansen O, Schellinger PD, Heiland S, Hacke W, Sartor K. 2002. Serial analysis of the apparent diffusion coefficient time course in human stroke. Neuroradiology 44: 294–298. Forbes KP, Pipe JG, Bird CR. 2002. Changes in brain water diffusion during the 1st year of life. Radiology 222: 405–409. Forbes KP, Pipe JG, Bird R. 2000. Neonatal hypoxic–ischemic encephalopathy: detection with diffusion-weighted MR imaging. Am J Neuroradiol 21: 1490–1496. Huppi PS, Maier SE, Peled S, Zientara GP, Barnes PD, Jolesz FA, Volpe JJ. 1998. Microstructural development of human newborn cerebral white matter assessed in vivo by diffusion tensor magnetic resonance imaging. Pediatr Res 44: 584–590.

Huppi PS, Murphy B, Maier SE, Zientara GP, Inder TE, Barnes PD, Kikinis R, Jolesz FA, Volpe JJ. 2001. Microstructural brain development after perinatal cerebral white matter injury assessed by diffusion tensor magnetic resonance imaging. Pediatrics 107: 455–460. Inder T, Huppi PS, Zientara GP, Maier SE, Jolesz FA, di Salvo D, Robertson R, Barnes PD, Volpe JJ. 2000. Early detection of periventricular leukomalacia by diffusion-weighted magnetic resonance imaging techniques. J Pediatr 136: 421. Jaermann T, Crelier G, Pruessmann KP, Golay X, Netsch T, Van Muiswinkel AM, MoriS, Van Zijl PC, Valavanis A, Kollias S, Boesiger P. 2004. SENSE-DTI at 3 T. Magn Reson Med 51(2): 230–236. Khong PL, Lam BC, Chung BH, Wong KY, Ooi GC. 2003. Diffusion-weighted MR imaging in neonatal nonketotic hyperglycinemia. Am J Neuroradiol 24: 1181–1183. Krishnamoorthy KS, Soman TB, Takeoka M, Schaefer PW. 2000. Diffusion-weighted imaging in neonatal cerebral infarction: clinical utility and follow-up. J Child Neurol 15: 592–602. Kuenzle C, Baenziger O, Martin E, Thun-Hohenstein L, Steinlin M, Good M, Fanconi S, Boltshauser E, Largo RH. 1994. Prognostic value of early MR imaging in term infants with severe perinatal asphyxia. Neuropediatrics 25: 191–200. Maalouf EF, Duggan PJ, Rutherford MA, Counsell SJ, Fletcher AM, Battin M, Cowan F, Edwards AD. 1999. Magnetic resonance imaging of the brain in a cohort of extremely preterm infants. J Pediatr 135: 351–357. Mader I, Schoning M, Klose U, Kuker W. 2002. Neonatal cerebral infarction diagnosed by diffusion-weighted MRI: pseudonormalization occurs early. Stroke 33: 1142–1145. McKinstry RC, Mathur A, Miller JH, Ozcan A, Snyder AZ, Schefft GL, Almli CR, Shiran SI, Conturo TE, Neil JJ. 2002. Radial organization of developing preterm human cerebral cortex revealed by non-invasive water diffusion anisotropy MRI. Cereb Cortex 12: 1237–1243. McKinstry RC, Miller JH, Snyder AZ, Mathur A, Schefft GL, Almli CR, Shimony JS, Shiran SI, Neil JJ. 2002. A prospective, longitudinal diffusion tensor imaging study of brain injury in newborns. Neurology 59: 824–833. Mercuri E, Cowan F, Rutherford M, Acolet D, Pennock J, Dubowitz L. 1995. Ischaemic and haemorrhagic brain lesions in newborns with seizures and normal Apgar scores. Arch Dis Child Fetal Neonatal Ed 73: 67–74. Mercuri E, Ricci D, Cowan F, Lessing D, Frisone M, Haatja L, Counsell S, Dubowitz L, Rutherford MA. 2000. Head growth in infants with hypoxic–ischaemic encephalopathy: Correlation with neonatal magnetic resonance imaging. Pediatrics 106: 235–243. Mercuri E, Rutherford M, Cowan F, Azzopardi D, Bydder G, Pennock J, Counsell S, Papadimitriou M, Dubowitz L. 1999. Early prognostic indicators of outcome in infants with

719

720

Mary A. Rutherford and Serena J. Counsell

neonatal cerebral infarction: A clinical EEG and MRI study. Paediatrics 103: 39–46. Mercuri E, Rutherford M, Barnett A, Foglia C, Haataja L, Counsell S, Cowan F, Dubowitz L. 2002. MRI lesions and infants with neonatal encephalopathy. Is the Apgar score predictive? Neuropediatrics 33: 150–156. Miller JH, McKinstry RC, Philip JV, Mukherjee P, Neil JJ. 2003. Diffusion-tensor MR imaging of normal brain maturation: a guide to structural development and myelination. Am J Roentgenol 180: 851–859. Miller SP, Vigneron DB, Henry RG, Bohland MA, Ceppi-Cozzio C, Hoffman C, Newton N, Partridge JC, Ferriero DM, Barkovich AJ. 2002. Serial quantitative diffusion tensor MRI of the premature brain: development in newborns with and without injury. J Magn Reson Imaging 16: 621–632. Mukherjee P, Miller JH, Shimony JS, Philip JV, Nehra D, Snyder AZ, Conturo TE, Neil JJ, McKinstry RC. 2002. Diffusion-tensor MR imaging of gray and white matter development during normal human brain maturation. Am J Neuroradiol 23: 1445–1456. Righini A, Bianchini E, Parazzini C, Gementi P, Ramenghi L, Baldoli C, Nicolini U, Mosca F, Triulzi F. 2003. Apparent diffusion coefficient determination in normal fetal brain: a prenatal MR imaging study. Am J Neuroradiol 24: 799–804. Righini A, Salmona S, Bianchini E, Zirpoli S, Moschetta M, Kustermann A, Nicolini U, Triulzi F. 2004. Prenatal magnetic resonance imaging evaluation of ischemic brain lesions in the survivors of monochorionic twin pregnancies: report of 3 cases. J Comput Assist Tomogr 28: 87–92. Roelants-van Rijn AM, Nikkels PG, Groenendaal F, van Der Grond J, Barth PG, Snoeck I, Beek FJ, de Vries LS. 2001. Neonatal diffusion-weighted MR imaging: relation with histopathology or follow-up MR examination. Neuropediatrics 32: 286–294. Rutherford MA. 2002. Ed: MRI of the Neonatal Brain. WB Saunders. Rutherford MA, Pennock JM, Schwieso JE, Cowan FM and Dubowitz LMS. 1996. Hypoxic–ischaemic encephalopathy: Early and late MRI findings and clinical outcome. Arch Dis Child 75: 145–151. Rutherford MA, Pennock JM, Cowan FM, Dubowitz LM, Hajnal JV, Bydder GM. 1997a. Does the brain regenerate after perinatal infarction? Eur J Paediatr Neurol 1: 13–17.

Rutherford MA, Pennock JM, Cowan FM, Saeed N, Hajnal JV, Bydder GM. 1997b. Detection of subtle changes in the brains of infants and children via subvoxel registration and subtraction of serial MR images. Am J Neuroradiol 18: 829–835. Rutherford MA, Pennock JM, Counsell SJ, Mercuri E, Cowan FM, Dubowitz LM, Edwards AD. 1998. Abnormal magnetic resonance signal in the internal capsule predicts poor neurodevelopmental outcome in infants with hypoxic–ischemic encephalopathy. Pediatrics 102: 323–328. Seghier ML, Lazeyras F, Zimine S, Maier SE, Hanquinet S, Delavelle J, Volpe JJ, Huppi PS. 2004. Combination of event-related fMRI and diffusion tensor imaging in an infant with perinatal stroke. Neuroimage 21: 463–472. Takeoka M, Soman TB, Yoshii A, Caviness VS Jr, Gonzalez RG, Grant PE, Krishnamoorthy KS. 2002. Diffusion-weighted images in neonatal cerebral hypoxic–ischemic injury. Pediatr Neurol 26: 274–281. Tanner SF, Cornette L, Ramenghi LA, Miall LS, Ridgeway JP, Smith MA, Levene MI. 2003. Cerebral perfusion in infants and neonates: preliminary results obtained using dynamic susceptibility contrast enhanced magnetic resonance imaging. Arch Dis Child Fetal Neonatal 88: 525–530. Tanner SF, Ramenghi LA, Ridgway JP, Berry E, Saysell MA, Martinez D, Arthur RJ, Smith MA, Levene MI. 2000. Quantitative comparison of intrabrain diffusion in adults and preterm and term neonates and infants. Am J Roentgenol 174: 1643–1649. Vermeulen RJ, Fetter WP, Hendrikx L, Van Schie PE, van der Knaap MS, Barkhof F. 2003. Diffusion-weighted MRI in severe neonatal hypoxic–ischaemia: the white cerebrum. Neuropediatrics 34: 72–76. Wolf RL, Zimmerman RA, Clancy R, Haselgrove JH. 2001. Quantitative apparent diffusion coefficient measurements in term neonates for early detection of hypoxic–ischemic brain injury: initial experience. Radiology 218: 825–833. Zarifi MK, Astrakas LG, Poussaint TY, Plessis Ad A, Zurakowski D, Tzika AA. 2002. Prediction of adverse outcome with cerebral lactate level and apparent diffusion coefficient in infants with perinatal asphyxia. Radiology 225: 859–870. Zhai G, Lin W, Wilber KP, Gerig G, Gilmore JH. 2003. Comparisons of regional white matter diffusion in healthy neonates and adults performed with a 3.0-T head-only MR imaging unit. Radiology 229: 673–681.

The role of diffusion and perfusion weighted brain imaging in neonatology

Case Study 42.1 Perinatal asphyxic injury Ronald L. Wolf, M.D., Ph.D. and Robert A. Zimmerman, M.D., University of Pennsylvania Medical Center and Children’s Hospital of Philadelphia, Philadelphia, PA History 1-day-old term male infant, seizure activity, Apgar scores 0/0/1.

T2W

T1W

DWI

ADC

T2W (normal)

ADC (normal)

Technique Conventional and diffusion-weighted MRI, with ADC map.

Imaging findings T2-weighted TSE image appears almost normal, with the exception of subtle increased intensity in ventrolateral thalamus (see normal T2-weighted image in bottom row for comparison). T1-weighted SE image is normal. DWI (b  1000 s/mm2) and ADC map clearly show abnormal (decreased) diffusion in ventrolateral thalamus and PLIC.

Discussion Perinatal asphyxic injury (or hypoxic–ischemic injury) can be difficult to detect on conventional images in the acute/subacute setting due to high water content of neonatal brain and incomplete myelination. In this case, the injury is more clearly depicted on DWI/ADC map.

Key points Conventional imaging limited in early detection of perinatal asphyxic injury. DWI and ADC maps can improve conspicuity of acute/subacute hypoxic–ischemic injury.

References Cowan FM, Pennock JM, Hanrahan JD, Manji KP, Edwards AD. 1994. Early detection of cerebral infarction and hypoxic–ischemic encephalopathy in neonates using diffusion-weighted magnetic resonance imaging. Neuropediatrics 25: 172–175. Wolf RL, Zimmerman RA, Clancy R, Haselgrove JH. 2001. Quantitative ADC measurements in term neonates for early detection of hypoxic–ischemic brain injury: initial experience. Radiology 218: 825–833.

721

43

Physiological MR imaging of pediatric brain tumors Jill V. Hunter Texas Children’s Hospital, Dallas, TX, USA

Key points • The incidence of pediatric brain tumors is greatest during the first decade of life. • Diffusion-weighted imaging and apparent diffusion coefficient maps may be helpful in the diagnosis of posterior fossa tumors, and also distinguishing radiation necrosis from tumor regrowth. • Aggressive pediatric brain tumors have high cerebral blood volume, low N-acetyl aspartate/ choline (Cho) and high Cho/creatine ratios. These imaging characteristics are generally associated with a poor prognosis.

Historically, assessment of pediatric brain tumors has focused on their morphology. With the introduction of higher field strengths, faster gradients, parallel imaging and new sequence design, together with new contrast agents, the ability to combine parameters of function with anatomy can give us insights into tumor physiology. Dynamic T1- and T*2 -weighted dynamic gadolinium (Gd)-enhanced MR imaging (MRI) can be used to assess perfusion, vascularity, permeability and microcirculation of brain neoplasms. Diffusion-weighted imaging (DWI) can be used to better delineate and even differentiate tumors. MR spectroscopic imaging (MRSI), allows for metabolic mapping both within and around tumors, and helps to differentiate tumor recurrence from radiation necrosis. The combined use of these new techniques yields information about lesion pathophysiology to be obtained in vivo, which it is 722

hoped will improve our prognostic abilities and aid in the planning of therapy. The incidence of brain tumors peaks in the first decade of life then declines, with a second peak in older adulthood (Strother et al., 2002). The first peak is characterized by a predominance of male patients; this is attributable to a disproportionate incidence of both medulloblastoma (MB) and ependymoma in males, and for other tumor types the sexes are equally affected. In the first 2–3 years of life more white than non-white children are affected, although the incidence in later childhood is equal in the two groups. During the first 2 years of life cerebral lesions predominate; cerebellar lesions are more common during the rest of the first decade. The approximate incidence of central nervous system (CNS) tumors in childhood is: high-grade supratentorial astrocytomas 10–15%, low-grade supratentorial astrocytomas 15–25%, MB 10–20%, cerebellar astrocytoma 10–20%, ependymoma 5–10%, brainstem glioma 10–20%, craniopharyngioma 6–9%, pineal tumors 0.5–2% and others 12–14%. Primitive neuroectodermal brain tumors (PNETs), including MB (PNET/MB), and supratentorial PNETs, constitute 20% of all pediatric brain tumors according to (Huber et al., 2001). Cerebral metastases from neoplasms outside the CNS are rare in children, as are primary extra-axial tumors such as meningiomas and sarcomas . The range of tumors seen in children is somewhat different from that seen in adults, in part because of the occurrence of embryonal tumors of neuroepithelial origin and genetic predispositions, notably neoplasms related to neurocutaneous syndromes. Inherited disorders associated with brain tumors

Physiological MR imaging of pediatric brain tumors

include: neurofibromatosis type 1 (NF-1, chromosome 17), associated with neurofibromata, optic nerve glioma (typically present by age 3 years of age), astrocytoma; neurofibromatosis type 2 (NF-2, chromosome 22), associated with bilateral vestibular schwannomas, peripheral schwannoma, meningioma and spinal ependymoma; tuberous sclerosis (TSC1, TSC2), associated with subependymal giant cell astrocytoma at the foramen of Monro; von Hippel-Lindau syndrome (VHL gene), associated with hemangioblastoma; Li-Fraumeni syndrome associated with multiple brain tumor types, most commonly astrocytoma and MB; Turcot syndrome associated with an incidence of MB and astrocytoma and ataxia telangiectasia associated with an increased incidence of tumors. Cowden syndrome has a recognized association with Lhermitte-Duclos or dysplastic gangliocytoma of the cerebellum.

Conventional MRI The investigation of a brain tumor with MRI traditionally uses conventional T1- and T2-weighted spin-echo (SE) imaging, usually with post-Gd chelate enhancement. Maximal information on the pathological characteristics of the tumor prior to surgery is helpful to the treatment team; for planning relevant further investigations, radiotherapy, chemotherapy and surgical resection. Although conventional structural imaging may allow confident diagnosis in many cases, its discriminant power is limited. For example, distinguishing reliably posterior fossa MB from juvenile pilocytic astrocytoma (JPA), and ependymoma presents a great challenge to the radiologist. Although JPAs classically present as a small nubbin of solid tumor associated with a large cystic component and MBs are typically solid masses, there is great variability. Also, because the posterior fossa is a confined “box”, once tumor mass has filled or effaced the fourth ventricle it becomes very hard to identify the epicenter of the tumor. The anatomical extension of the tumor may be helpful diagnostically, for example ependymomas have a far greater propensity to extrude laterally out through the foramina of Luschka or centrally through the foramen of Magendie in the midline than either JPA or MB.

Some ultrastructural information is available from the T1- and T2-weighted images (WI) themselves, in that densely packed small round cell tumors such as posterior fossa MBs are hyperintense on T1-weighted sequences with relative loss of signal on T2-weighted sequences, reflecting the high nuclear-to-cytoplasm ratio. Despite the use of differentially weighted SE sequences, when used in isolation, differentiation of a cystic MB from a solid-appearing JPA, can at times be impossible. On T1-WI ependymomas are typically isointense or hypointense to normal gray matter (GM); on T2-WI typically isointense to slightly hyperintense to normal GM. Whilst approximately 50% of ependymomas contain calcification on computed tomography (CT) scanning, this is not always easily appreciated on MRI. Gradient-echo (GE) imaging is performed to look for evidence of hemorrhage, the presence of which can preclude the use of certain chemotherapeutic agents. GE imaging can also be helpful in the detection of calcification if it is present in sufficient quantity. A distinction can sometimes be drawn between altered blood and calcium based on the quality of “blooming” which is less marked with calcium and darker and more intense when blood products are present.

Dynamic susceptibility-contrast imaging Dynamic susceptibility-contrast imaging (DSCI) is a form of “super” GE sequence. High-resolution 3D GE imaging with long-echo times (TEs) is used to create high contrast phase images (Haacke et al., 2002a). A phase mask is used to enhance the tissue contrast in the magnitude images making it possible to differentiate tissues with small differences in iron content, to separate fat and water, and to better visualize clot and microhemorrhage. Phase images contain direct information about the background magnetic field and chemical shift of tissues. Haacke et al. (2002b) report that DSCI can be used: (1) to separate veins from arteries, (2) to image vessel wall, (3) to image microhemorrhage and brain iron, and (4) to enhance T1 contrast in GM/white matter (WM). Tong et al. (2003), have shown that DSCI is superior to conventional GE imaging in the detection of hemorrhage in traumatic brain injury (TBI). All of the above

723

724

Jill V. Hunter

Fig. 43.1 Axial T1-WI pre-contrast in an 8-year-old female presenting with headache and demonstrating a large left frontal extra-axial tumor displacing both frontal horns posteriorly. Note the presence of signal flow voids along the left-hand margin of the lesion which demonstrates convex bowing to the right of midline.

Fig. 43.2 Axial T1-WI of the same patient post-Gd injection demonstrating heterogeneous enhancement of the right side of the mass with more uniform marked enhancement of the left-hand margin of the lesion consistent with a highly vascular component. The tumor proved to be a spindle cell sarcoma at biopsy.

properties lend themselves to a better understanding of the physiology of tumors. From the same data set acquired for DSCI, the phase images can be potentially used for quantification of magnetic susceptibility, as reported by Wang et al. (1999) and Li and Wang (2003).

Fluid attenuated inversion recovery Fluid attenuated inversion recovery (FLAIR) and proton-density imaging is used to better assess the extent of cerebral edema surrounding a tumor. It is well recognized that local tumor can extend into the vasogenic edema immediately surrounding the main lesion, although efforts to predict local metastatic spread by imaging alone have so far not proved very successful. Goo and Choi (2003) have shown that post-Gd FLAIR imaging may increase the conspicuity of extra-axial lesions compared to

Fig. 43.3 Coronal collapsed 3D time of flight (TOF) MRA maximum intensity projection image without contrast in the same patient as in Figures 43.1 and 43.2, demonstrating mass effect causing displacement of the anterior cerebral arteries (ACAs) to the right of midline in association with abnormal vasculature.

Physiological MR imaging of pediatric brain tumors

Fig. 43.4 AP view of a left-sided internal carotid artery (ICA) injection performed as part of a catheter digital subtraction angiogram. Note the highly abnormal circulation with small aneurysm formation. This area of angiogenesis looks almost like an arterio-venous malformation. Following external beam radiation therapy (RT) the tumor was sufficiently devascularized that it became amenable to gross total resection.

post-contrast T1-WI in children, but is worse than T1-WI at demonstrating intra-axial lesions.

Contrast enhancement Assessment of a new tumor will always include the administration of contrast to look for evidence of breakdown of the blood–brain barrier (BBB) in addition to identifying large feeding vessels or draining veins to indicate the presence of any unusual vascular component to the lesion and demonstrating the normal background vasculature. MR angiography (MRA) with or without the administration of contrast can be used to better define an abnormal arterial supply (Figures 43.1–43.4). This may be helpful in both lesion characterization and surgical planning. T1-weighted SE imaging of tumors may be performed with magnetization transfer (MT) saturation pulses, both before and after Gd, in order to improve the conspicuity of contrast enhancement in the tumor. The use of MT with post-Gd imaging has been shown in children to produce contrast-to-noise equivalent to the use of a double dose of Gd agent

with conventional MRI (Zimmerman et al., 1994). Haba et al. (2001), have reported on equivalent conspicuity in tumors and state that “half-dose (Gd) T1-weighted SE with MT can replace standard-dose T1-weighted SE without MT with no loss of contrast in the investigation of meningiomas”. Yuh et al. (1997), reviewed the literature on varying dosages of Gd in conjunction with T1-weighted SE imaging and while he concluded that there was a trend for increased visualization with increased contrast he did not assess the impact of MT sequences, and concluded that dosage needed to be tailored to the pathology and location of the tumor. Attempts at utilizing patterns of enhancement to differentiate tumor types in the posterior fossa of children have met with limited success. Arle et al. (1997), using a neural network presented with MRI characteristics (which did not include diffusion imaging), age, sex and tumor size had an accuracy of 72% in correctly predicting posterior fossa tumor type, consistent with the neuroradiologist who was using the same information. However, it is generally true that JPAs, despite their low grade, often demonstrate marked enhancement, due to breakdown of the BBB, when compared with PNET or MB which

725

726

Jill V. Hunter

has a more variable but generally lower level of enhancement. This contrasts with hemangioblastomas that due to their very vascular nature enhance vigorously, but also demonstrate signal flow voids and feeding vessels with draining veins following the administration of Gd. Ependymomas tend to show a more heterogeneous pattern of enhancement with areas of intense focal enhancement mixed with a more diffuse background of lower level contrast uptake. The extent of enhancement is important as it allows for a more accurate description of tumor extent, which can improve the accuracy of biopsy. Also, improved visualization of the edges of the tumor can result in better-targeted radiotherapy. The extent of leptomeningeal spread pre-operatively and the number of metastases can have significant impact on patients’ prognosis, management and quality of life.

Tumor volume Assessment of tumor size is usually based on a combination of the pre- and post-Gd T1-WI and is usually expressed as the three maximum dimensions of the lesion; 1.5 cm3 or greater maximum volume of residual tumor following post-operative debulking has been shown by Albright et al. (1996), to be associated with an increased rate of MB treatment failure, and hence, is used in many oncology protocols as a risk stratification criterion. In addition an arbitrary interval growth of 25% is often used as evidence of treatment failure, and will exclude the child from continuing on a set protocol.

Radiation therapy Gd-enhanced MRI is now incorporated in many radiation therapy (RT) planning protocols often combined with fusion to the patient’s CT scans. After RT treatment, there can be a number of physiological changes. Radiation necrosis in the brain is often difficult to distinguish from recurrent tumor; like recurrent tumor, it may have mass effect and enhancement and typically occurs 6 months after treatment, peaking at around a year to 18 months following completion of radiotherapy. MR spectroscopy

(MRS) can be helpful in distinguishing radiation necrosis from recurrent tumor (cf. below). We have also seen a particular variant of what is presumed to be radiation necrosis following conformal RT to the posterior fossa in children. It presents as a very superficial enhancement, outlining the folia over the surface of the cerebellar hemispheres (Figures 43.5–43.8); it appears within 3 months of completion of conformal radiotherapy and disappears by 6 months following completion of treatment.

DWI DWI has proved helpful for the differentiation of MB from JPA or ependymoma, especially in the posterior fossa. Apparent diffusion coefficient (ADC) maps are routinely generated from the DWI data in order to avoid the problems of interpreting T2 “shine-through”, as described by Provenzale et al. (1999). The solid component of a MB or PNET/atypical terato-rhabdoid tumor (ATRT) shows restricted diffusion or (decreased in ADC), as reported by Erdem et al. (2001). This is felt to be on the basis of restricted water motion secondary to the presence of a high nuclear-to-cytoplasmic ratio in these tightly packed small round blue cell tumors, and contrasts with the solid component of a JPA which almost always displays increased ADC. Other highly cellular tumors such as lymphoma or aggressive malignancies with high mitotic activity, for instance glioblastoma multiforme (GBM) and anaplastic ependymoma, may also show areas of restricted diffusion (decrease in signal) on ADC maps. JPA and classic ependymoma demonstrate increased signal on ADC maps, especially in any cystic components. In the immuno-compromised child (e.g. a patient with acquired immuno-deficiency syndrome (AIDS), or post-transplantation), where a mass lesion could be the result of infection, neoplasm, fungal infection or toxoplasmosis, the use of DWI alone is probably not sufficient to make a diagnosis, and other techniques such as MRS may be required. However, in many cases, a biopsy is still required to make the diagnosis. DWI may have a part to play in the diagnosis of acute radiation necrosis following some of the

Physiological MR imaging of pediatric brain tumors

Fig. 43.5 Axial T1-WI through the posterior fossa post-Gd in a child with recent posterior fossa craniectomy for removal of PNET. Note the absence of any enhancement.

Fig. 43.6 Axial T1-WI through the posterior fossa post-Gd in the same child 3 months post-conformal RT. Note interval development of superficial enhancement right greater than left following the pattern of the folia. The differential diagnosis lies between leptomeningeal tumor spread and radiation necrosis.

Fig. 43.7 Axial T1-WI post-contrast in same child at same level through posterior fossa, 1 month following prior examination. Note the improvement in enhancement of the folia, without additional therapy.

Fig. 43.8 Axial T1-weighted post-Gd imaging in same child 6 months following completion of conformal RT to posterior fossa without any interim chemotherapy. There has been complete resolution of radiation necrosis changes.

727

728

Jill V. Hunter

newer delivery methods of RT such as gamma knife surgery. Hein et al. (2004) reported the use of DWI to distinguish tumor recurrence from radiation injury, with tumor recurrence showing lower ADC values than necrosis. Two examples of restriction on ADC maps following focal high-dose RT for cerebral metastases in adult patients have also been seen at our institution (Hunter, unpublished data). As described in Chapter 4, anisotropy parameters such as fractional anisotropy (FA) can be calculated from diffusion MRI data, and may in the future provide additional information on pediatric brain tumors. Diffusion-tensor imaging (DTI) performed utilizing anywhere from 6 to 30 (or more) different diffusion-encoding directions, is used to measure ADC and FA, and can also be used at high resolution to demonstrate WM tracts (tractography). Mori et al. (2002), have demonstrated that interruption of, or distortions to, the normal WM tracts can be displayed in patients with (adult) brain tumors. Zhou et al. (2003), have indicated that this technology may be of value in surgical planning and prediction of outcomes prior to resection of tumors, especially those in close proximity to anatomically eloquent areas. To date, however, there is little experience of this technique in pediatric brain tumors.

Perfusion-weighted imaging Measurements of tumor blood flow are important for understanding tumor physiology and are of potential value in both selecting and evaluating therapies. Brain tumors typically demonstrate reduced blood flow, and slower transit time, when compared to normal brain tissue. Selective sensitization of tumors to hyperthermia is generally ascribed to the poor tumor blood flow relative to surrounding normal tissue, according to Hetzel (1989). RT preferentially kills those cells that contain substantial amounts of dissolved oxygen (i.e. high blood flow), while anoxic and hypoxic tumor cells are relatively radio-resistant. Microvessel density has been correlated with relapse and poor survival in children with primary brain tumors. Neovascularity is a hallmark of adult GBM. Vascular endothelial growth factor (VEGF) is

overexpressed in malignant gliomas compared to normal brain. Platelet derived growth factor (PDGF) can induce VEGF secretion in human glioma cell line. Stimulation of angiogenesis can rapidly increase blood vessel density by as much as 10-fold (Dellian et al., 1996). Quantitative measurements of tumor blood flow might therefore be useful for the evaluation of the efficacy of anti-angiogenic therapies, such as thalidomide, and to tailor these therapies to individual tumors. MR perfusion imaging is performed either by utilizing a first-pass Gd bolus technique or by using arterial spin tag labeling (ASL) (cf. Chapters 4 and 5). The technique takes advantage of the T *2 or T2 decrease in signal caused by the first pass of a bolus of Gd contrast using a GE or SE sequence, respectively. This enables a time vs. signal intensity curve to be drawn, from which a variety of parameters, such as mean transit time (MTT), time-to-peak, area under the curve (a measure of relative cerebral blood volume (rCBV) and relative cerebral blood flow (rCBF)), can be calculated. This technique has been applied to a range of tumors over the last 10 years, by workers such as Provenzale et al. (2002), and Huisman and Sorensen (2004). Interpretation is made more difficult because tumors have a complex vascular supply well as areas of necrosis. BBB breakdown is also a complicating factor for the analysis of tumor perfusion, as has been discussed in detail in Chapter 21. Despite these problems in analysis, in general more aggressive malignancies do appear to have greater rCBF and rCBV, as reported in adults by Aronen et al. (1994). While techniques for quantifying absolute CBF with the Gd-bolus technique have been developed (cf. Chapter 7), most studies of brain tumors have used relative measurements, often expressed relative to WM in the contralateral hemisphere. Arterial spin labeling (ASL) perfusion MRI utilizes arterial blood water as an endogenous diffusible tracer (cf. Chapter 8). In ASL, arterial blood water is magnetically labeled proximal to the tissue of interest and the effects of pre-labeling are determined by pair-wise comparison with images acquired using control labeling. Pediatric perfusion imaging based on ASL may have unique advantages over first-pass Gd techniques in that it is totally non-invasive, i.e.

Physiological MR imaging of pediatric brain tumors

does not require an intravenous injection. It also offers better signal-to-noise ratio (SNR) than in adults, due to the higher level of CBF in children (Wang et al., 2003). The higher water content in the pediatric brain, and its smaller size, should also enhance SNR in children. Despite these potential advantages, there are currently no reports in the literature on the use of ASL in pediatric brain tumors.

MRS Proton MRS techniques and the information they yield have been reviewed in Chapters 1 and 2. As discussed in Chapter 40, children are “not little adults” in that there is a changing pattern of metabolites with age, especially during the first 3 years of life. In a normal child, at birth the choline (Cho) peak is higher than N-acetyl aspartate (NAA) but while Cho falls during infancy, NAA rises so that by 18 months of age the NAA peak is as high as if not higher than Cho (depending on acquisition conditions and brain region). Myo-inositol (mI) is also very high in the neonate. Creatine (Cr) remains fairly stable, rising only slowly during infancy, and for this reason, historically, has been used to normalize NAA and Cho values – expressed as a ratio relative to Cr. Due to the changing metabolite ratios in childhood, when interpreting MRS data from a brain tumor it is vital to compare the pathological tissue with mirror-imaged normal tissue from the opposite side where possible, or, if the lesion is in a midline location, with a normal age- and sex-matched control. It should be noted that, in both adults and children, normal levels and ratios of metabolites vary depending on which part of the brain is being sampled, so that it is important to match pathological data with appropriate brain regions. MRS has been used to assist in the separation of posterior fossa MB from JPA and ependymoma (Figures 43.9–43.12). Wang et al. (1995) demonstrated the possibility of using discriminant analysis applied to single voxel long-TE MRS data, by plotting Cr/Cho against NAA/Cho ratios (Figure 43.13). A follow-up study (Hunter et al., 2001) reported on the outcomes in this same group of patients. Of the 26 children (18 male, 8 female, age 1–15 years and mean 7.3 years), with follow-up ranging from 11 days to

PNET Cho

Cr/ PCr NAA

mI

4

3.5

3 2.5 2 1.5 Chemical shift (ppm)

1

0.5

SE, TE  135 ms, TR  1.6 s. L.S. Fig. 43.9 Single voxel MRS performed at TE  135 ms, demonstrating near complete absence of NAA, normal Cr and grossly elevated Cho in this child with histologically proven cerebellar MB/PNET.

Cerebellar control Cr/ Cho PCr

4

3.5

NAA

3 2.5 2 1.5 Chemical shift (ppm)

1

0.5

SE, TE  135 ms, TR  1.6 s. S.M. Fig. 43.10 Single voxel MRS cerebellum in age-matched normal control performed at TE  135 ms, for comparison and interpretation of patient with posterior fossa PNET.

Astrocytoma Cho NAA

4

3.5

3 2.5 2 1.5 Chemical shift (ppm)

1

0.5

SE, TE  135 ms, TR  1.6 s. M.Z. Fig. 43.11 Single voxel MRS acquired at TE  135 ms in a child with posterior fossa JPA. Note the relative preservation of NAA which is only slightly reduced, moderate increase in Cho and decrease in Cr.

729

Jill V. Hunter

GBM

Ependymoma Cho Cr/ PCr

Cho

Lipids/ Lac

NAA

mI

4

3.5

3 2.5 2 1.5 Chemical shift (ppm)

1

0.5

SE, TE  270 ms, TR  1.5 s. S.S. Fig. 43.12 Single voxel MRS acquired at TE  135 ms in a child with a proven posterior fossa ependymoma. Note the moderate decrease in NAA, with moderate increase in Cho and relative preservation of Cr. Note also the presence of a lipid/Lac peak consistent with the presence of an area of necrosis within the tumor. Lac: Lactate.

2.0 Astrocytoma Ependymoma PNET Normal

1.5

NAA / Cho

730

Normal

1.0 Astrocytoma

0.5 Ependymoma

PNET

0.0 0.0

0.5

1.0

1.5

Cr / Cho Fig. 43.13 Scattergram comparison of NAA/Cho vs. Cr/Cho for astrocytoma, ependymoma, PNET and normal cerebellar tissue. The straight lines are boundaries between the three tumor types found by discriminant analysis. Solid triangle: astrocytomas; square: ependymoma; circle: PNET; open diamond: normal control. (From Wang Z et al., 1995; with permission.)

4

3.5

Cr/ PCr NAA

Lipids/ Lac

3 2.5 2 1.5 Chemical shift (ppm)

1

0.5

SE, TE135 ms, TR  1.5 s. M.A. Fig. 43.14 Single voxel MRS of a supratentorial GBM, in a child, performed at TE  135 ms, demonstrating marked decrease in NAA with grossly elevated Cho and the presence of a lipid/Lac peak. This is a rare tumor in childhood but aggressive like its adult counterpart. Note the presence of a small mI peak. Compare with PNET.

11 years (mean 7.3 years), 1/11 with astrocytoma progressed, 4/11 with MB progressed and 3 died, while 3/4 with ependymoma progressed – 2 of whom died. The Cr/Cho ratio was significantly higher in the ependymoma group compared to the PNET and JPA patients. A lower NAA/Cho and higher Cr/Cho were predictive for progression in the whole cohort (P  0.037). NAA/Cho was decreased in the group that died (P  0.017). Even taking account of the bias from the high death rate in the group with ependymoma, there was still an effect from an elevated Cr/Cho. There was a statistically significant effect of low NAA/Cho as a predictor of death. MRS findings may also contain prognostic information for supratentorial tumors (Figures 43.14–43.18). Girard et al. (1998), assessed peak areas of NAA, Cho and Cr in 14 young people (average 10 years), with newly diagnosed cerebral hemispheric tumors. There were three glioblastomas and one each of PNET, ganglioglioblastoma, ependymoma, anaplastic ependymoma, rhabdoid teratoid tumor, pilocytic astrocytoma a gliomatosis cerebri and four gangliogliomas. Four patients died and 10 survived. The NAA/Cho and Cr/Cho were low in patients who died and high in survivors.

Physiological MR imaging of pediatric brain tumors

Superatentorial control Cr/ Cho PCr

Cr/ PCr

4

Gliomatosis cerebri

NAA

Cho Cr/ PCr

Glu/Gln

3.5

NAA

3 2.5 2 1.5 Chemical shift (ppm)

1

0.5

SE, TE  135 ms, TR  1.6 s. R.D.

4.5

Fig. 43.15 Normal supratentorial control at TE  135 ms. Glu: Glutamate; Gln: Glutamate.

3.5 3 2.5 2 1.5 1 Chemical shift (ppm) SE, TE  135 ms, TR  1.6 s. R.D.

0.5

Fig. 43.17 Single voxel MRS performed at TE  135 ms in a child with biopsy proven gliomatosis cerebri. Note the slight decrease in NAA in the presence of preserved Cr/PCr and moderate elevation of Cho. A second Cr peak is noted at approximately 3.8 ppm.

Ganglio-GBM Cho

Cr/ PCr

4

NAA Lac Rhabdoid tumor Cho

4

3.5

3 2.5 2 1.5 1 Chemical shift (ppm) SE, TE  135 ms, TR  1.6 s. N.R.

0.5

Fig. 43.16 Single voxel MRS of ganglio-GBM acquired at TE  135 ms. This is a less aggressive tumor than GBM and demonstrates a slight decrease in NAA with a moderate increase in Cho while Cr/PCr is relatively well preserved. Note the elevation in Lac consistent with some necrosis.

Lazareff et al. (1999), studied the role of MRS in monitoring response of histologically proven pediatric glioma to adjuvant chemotherapy or radiotherapy. Thirty-eight scans were performed, with the follow-up period ranging from 6 to 40 months. The Cho signal (normalized relative to that in normal brain – Chon) correlated with tumor volume and clinical response. In four patients whose tumor progressed after treatment, Chon increased and, in six patients who had a stable or decreased tumor volume, the ratio decreased. A study of 27 children with recurrent brain tumors (Warren et al., 2000) also demonstrated the prognostic significance of the Cho/NAA ratio in pediatric brain tumors. Diagnoses included high-grade

Cr/ PCr

4.5

4

3.5

NAA

Lac

3 2.5 2 1.5 1 Chemical shift (ppm) STEAM, TE  270 ms, TR  1.5 s. A.J.

0.5

Fig. 43.18 Single voxel MRS acquired at TE  135 ms in a child with a supratentorial rhabdoid tumor. This is an aggressive lesion with poor survival by 18 months post-diagnosis. Note the near complete absence of NAA in the presence of marked elevation of Cho with relative preservation of Cr/PCr. Compare with other aggressive tumors, such as PNET and GBM. Note also small mI peak and Lac peak, STEAM: stimulated echo acquisition mode.

glioma (10), brainstem glioma (7), MB/peripheral neuroectodermal tumor (6), ependymoma (3) and pineal germinoma (1). The concentrations of Cho and NAA in the tumor and normal brain were quantified and maximum Cho/NAA ratio was determined for each patient’s tumor. The maximum Cho/NAA ratio ranged from 1.1 to 13.2 (median 4.5).

731

732

Jill V. Hunter

The ratio in apparently normal brain tissue was 1. Children with maximum Cho/NAA ratio of 4.5 had a projected survival of 50% at 63 weeks and those with a ratio of 4.5 (n  13) had a median survival of 22 weeks, with none surviving at 63 weeks. Another study examined the combined value of Gd-perfusion imaging and MRS (Tzika, 2003). It was found that “Cho (suggestive of tumor cellularity and proliferative activity) correlated with rCBV, while the relationship between Cho and ADC (suggestive of cellular density) was inverse”. It was noted that this combination of findings may both improve our understanding of tumor physiology, and also might be important for distinguishing between high- from low-grade tumors. This information could be particularly relevant in inoperable tumors. MRS has also demonstrated benefit in the presurgical diagnosis of large suprasellar tumors. Sutton et al. (1997), were able to separate craniopharyngioma from pituitary adenoma and hypothalamic region astrocytoma using single voxel acquisitions. Craniopharyngiomas showed a dominant peak at 1–2 ppm consistent with lipids/lactate (lipids/Lac). Chiasmatic gliomas showed a profile similar to JPAs, with a moderate reduction in NAA in association with mild to moderate elevation of Cho. Pituitary adenomas showed a Cho peak or no metabolites at all. Another application of MRSI is in distinguishing recurrent tumor from radiation necrosis, which may also exhibit enhancement and mass effect on conventional MRI. In adult gliomas (Rock et al., 2002) it was shown that MRSI could distinguish pure tumor from pure necrosis: radiation necrosis had Chon  1.0, while tumors had Chon 1.0. Also, Lac and/or lipid tended to be higher in necrosis than in tumor regrowth. However, there have been relatively few studies to date on the use of spectroscopy for distinguishing radiation necrosis from tumor regrowth in children.

Combined sequences and technologies Generally, the confidence with which diagnosis in pediatric brain tumors can be made is improved by the use of multi-modal MRI. For instance, DWI in combination with conventional imaging

(including contrast-enhanced scanned with MT), MRS and knowledge of the patient’s age, can often lead to a high degree of certainty in distinguishing between the three most common posterior fossa tumors. Physiological parameters can also improve evaluation of degree of dissemination of a tumor. The combination of perfusion, diffusion and spectroscopy imaging also offers great potential in surveillance of patients with known or treated tumors; in particular, looking for recurrent or residual disease, and looking for early signs of progression or treatment response. It is not yet clear how these will be combined to inform clinical management of pediatric neoplasia, and other imaging methods, such as positron emission tomography (PET) may also contribute. Serial monitoring with structural MRI is currently the mainstay of imaging follow-up. Serial application of physiological imaging techniques is likely to augment this, and may be combined with registration methods for the structural images to allow more objective and sensitive evaluation of disease progression. In view of the relative rarity of many of these tumors, the development and implementation of consistent evidence-based protocols across multiple centers is central to progress in the clinical management of neurological neoplasms in the pediatric population.

REFERENCES Albright AL, Wisoff JH, Zeltzer PM, Boyett JM, Rorke LB, Stanley P. 1996. Effects of medulloblastoma resections on outcome in children: a report from the Children’s Cancer Group. Neurosurgery 38(2): 265–271. Arle JE, Morriss C, Wang ZJ, Zimmerman RA, Phillips PG, Sutton LN. 1997. Prediction of posterior fossa tumor type in children by means of magnetic resonance image properties, spectroscopy, and neural networks. J Neurosurg 86(5): 755–761. Aronen HJ, Gazit IE, Louis DN, Buchbinder BR, Pardo FS, Weiskoff RM, Harsh GR, Cosgrove GR, Halpern EF, Hochberg FH, et al. 1994. Cerebral blood volume maps of gliomas: comparison with tumor grading and histologic findings. Radiology 191(1): 41–51.

Physiological MR imaging of pediatric brain tumors

Dellian M, Witwer BP, Salehi HA, Yuan F, Jain RK. 1996. Quantitation and physiological characterization of angiogenic vessels in mice: effect of basic fibroblast growth factor, vascular endothelial growth factor/vascular permeability growth factor, and host microenvironment. Am J Pathol 149: 59–71. Erdem E, Zimmerman RA, Haselgrove JC, Bilaniuk LT, Hunter JV. 2001. Diffusion weighted imaging (DWI) and fluid attenuated inversion recovery (FLAIR) imaging in the evaluation of primitive neuroectodermal tumors. Neuroradiology 43(11): 927–933. Girard N, Wang ZJ, Erbetta A, Sutton LN, Phillips PC, Rorke LB, Zimmerman RA. 1998. Prognostic value of proton MR spectroscopy of cerebral hemisphere tumors in children. Neuroradiology 40(2): 121–125. Gonen O, Wang ZJ, Viswanathan AK, Molloy PT, Zimmerman RA. 1999. Three-dimensional multivoxel proton MR spectroscopy of the brain in children with neurofibromatosis type I. Am J Neuroradiol 20: 1333–1341. Goo HW, Choi C-G. 2003. Post-contrast FLAIR MR imaging of the brain in children: normal and abnormal intracranial enhancement. Pediatr Radiol 33: 843–849. Haacke EM, Herigault G, Yu Y, Tong K, Obenaus A, Reichenbach JR. 2002b. Observing tumor vascularity noninvasively using magnetic resonance imaging. Image Anal Stereol 21: 107–113. Haacke EM, Ogg R, Reichenbach JR, Gurleyik K, Xu Y, Herigault G. 2002a. Susceptibility weighted imaging (SWI): a new means to enhance image contrast. Proc Intl Soc Mag Reson Med 10. Haba D, Papon AP, Tanguy JY, Aube C, Caron-Poitreau C. 2001. Use of half-dose gadolinium-enhanced MRI and magnetization transfer saturation in brain tumors. Eur Radiol 11: 117–122. Hein PA, Clifford JE, Dunn JF, Hug EB. 2004. Diffusionweighted imaging in the follow-up of treated high-grade gliomas: tumor recurrence versus radiation injury. Am J Neuroradiol 25: 201–209. Hetzel FW. 1989. Biological rationale for hyperthermia. Radiol Clin North Am 27: 499–508. Huber H, Eggert A, Janss AJ, Wiewrodt R, Zhao H, Sutton LN, Rorke LB, Phillips PC, Grotzer MA. 2001. Angiogenic profile of childhood primitive neuroectodermal brain tumors/ medulloblastomas. Eur J Cancer 37: 2064–2072. Huisman TA, Sorensen AG. 2004. Perfusion-weighted magnetic resonance imaging of the brain: techniques and applications in children. Eur Radiol 14: 59–72. Hunter JV, Tavani F, Wang Z, Zhao H, Janss A, Zimmerman RA. 2001. Prognostic value of proton magnetic resonance spectroscopy in posterior fossa tumors in children. (Abstract) RSNA Proc. 87th Annual Meeting. p. 849.

Lazareff JA, Gupta RK, Alger J. 1999. Variation of post-treatment H-MRSI choline intensity in pediatric gliomas. J Neurooncol 41(3): 291–298. Li L, Wang Z. 2003. Magnetic susceptibility quantitation with MRI by solving boundary value problems. Med Phys 30: 449–453. McNight TR, Noworolski SM, Vigneron DB, Nelson SJ. 2001. An automated technique for the quantitative assessment of 3D-MRSI data from patients with glioma. J Magn Reson Imaging 13: 167–177. Mori S, Frederiksen K, van Zijl PCM, Stieltjes B, Kraut MA, Solaiyappan M, Pomper MG. 2002. Brain white matter anatomy of tumor patients evaluated with diffusion tensor imaging. Ann Neurol 51(3): 377–380. Provencher SW. 1993. Estimation of metabolite concentration from localized in vivo proton NMR Spectra. Magn Reson Med 30: 672–679. Provenzale JM, Engelter ST, Petrella JR, et al. 1999. Use of MR exponential diffusion-weighted images to eradicate T2 “shine-through” effect. Am J Roentgenol 172: 537–539. Provenzale JM, Wang GR, Brenner T, Petrella JR, Sorensen AG. 2002. Comparison of permeability in high-grade and lowgrade brain tumors using dynamic susceptibility contrast MR imaging. Am J Roentgenol 178(3): 711–716. Rock JP, Hearshen D, Scarpace L, Croteau D, Guitierrez J, Fisher JL, Rosenblum ML, Mikkelsen T. 2002. Correlations between magnetic resonance spectroscopy and imageguided histopathology, with special attention to radiation necrosis. Neurosurgery 51(4): 912–923. Silva AC, Kim S-C, Garwood M. 2002. Imaging blood flow in tumors using arterial spin labeling. Magn Reson Med 44(2): 169–173. Strother DR, Pollack IF, Fisher PG, Hunter JV, Woo SY, Pomeroy SL, Rorke LB. 2002. In Principles and Practice of Pediatric Oncology, 4th edn. (Eds., Pizzo, Poplack), Chapter 27, Lippincott, Williams and Wilkins, Philadelphia. Sutton LN, Wang ZJ, Wehrli SL, Marwaha S, Molloy P, Phillips PC, Zimmerman RA. 1997. Proton spectroscopy of suprasellar tumors in pediatric patients. Neurosurgery 41(2): 388–397. Tong KA, Ashwal S, Holshauser BA, Shutter LA, Herigault G, Haacke EM, Kido DK. 2003. Hemorrhagic shearing lesions in children and adolescents with posttraumatic diffuse axonal injury. Radiology 227(2): 332–339. Tzika AA, astrakas LG, Zarifi MK, Petridou N, YoungPoussaint T, Goumnervoa L, Zurakowski D, Anthony DC, Black PMcL. 2003. Neuroradiology 45: 1–10. Wang JJ, Licht DJ, Jahng G-H, Liu C-S, Rubin JT, Haselgrove J, Zimmerman RA, Detre JA. 2003. Pediatric perfusion imaging using pulsed arterial spin labeling. J Magn Reson Imaging 18(4): 404–413.

733

734

Jill V. Hunter

Wang Z, Sutton LN, Cnaan A, Haselgrove JC, Rorke LB, Zhao H, Bilaniuk LT, Zimmerman RA. 1995. Proton MR spectroscopy of pediatric cerebellar tumors. Am J Neuroradiol 16: 1821–1833. Wang ZJ, Li S, Haselgrove JC. 1999. Magnetic resonance imaging measurement of volume magnetic susceptibility using a boundary condition. J Magn Reson 140: 477–481. Wang ZJ, Bergqvist C, Hunter JV, Jin D, Wang DJ, Wehrli S, Zimmerman RA. 2003. In vivo measurement of brain metabolites using two-dimensional double-quantum MR spectroscopy – exploration of GABA levels in a ketogenic diet. Magn Reson Med 49: 615–619. Warren KE, Frank JA, Black JL, Hills RS, Duyn JH, Aikin AA, Lewis B, Adamson PC, Balis FM. 2000. Proton magnetic resonance spectroscopic imaging in children with recurrent primary brain tumors. J Clin Oncol 18(5): 1020–1026.

Yuh WTC, Parker JR, Carvlin MJ. 1997. Indication-related dosing for magnetic resonance contrast media. Eur Radiol 7: s269–s275. Zhou XJ, Leeds NE, Poonawalla AH, Weinberg J. 2003. Assessment of tumor cell infiltration along white-matter fiber tracts using diffusion tensor imaging. Eleventh Scientific Meeting and Exhibition ISMRM, Abstract no. 2238, p. 440 Toronto. Zimmerman RA, Haselgrove JC, Bilaniuk LT, Hunter JV. 1994. Magnetization transfer suppression in gadolinium enhancement of the pediatric brain. Proceedings of the XV Symposium Neuro-radiologicum, (Eds., Takahashi M, Koregi Y, Moseley I) Kumamoto, Japan, 25 September–1 October, Springer, pp. 267–268.

Physiological MR imaging of pediatric brain tumors

Case Study 43.1 Pediatric astrocytoma Peter Barker, D.Phil., Johns Hopkins University School of Medicine, Baltimore History A 3-year-old female with a left cerebellar-pontine astrocytoma.

T1 Gd

FLAIR

Technique Conventional MRI and multi-slice MRSI (TE

a

280 ms).

b

Imaging findings The solid, enhancing part of the tumour exhibits only a Cho signal (b). However, Cho is no higher than in the contralateral hemisphere. The right side of the pons has low NAA (a), even though it has normal MRI appearance. The cystic portion of the lesion in the cerebellum has low levels of all metabolites.

Cho

NAA

a

b

Discussion MRSI has been shown to be helpful in predicting survival in children with primary brain tumours (Warren, 2000), high Cho being a bad prognostic indicator. As in adults, pediatric brain tumours can be very heterogeneous. Key points MRSI may be helpful in distinguishing malignant and benign brain lesions in children.

Cho

Aggressive lesions tend to have high Cho and low NAA.

Cr

Cho

NAA

It is important to recognize normal age-related regional spectral variations in children. Some posterior fossa lesions may be difficult to evaluate by MRS because of field inhomogeneity.

ppm

3.0

2.0

1.0

ppm

3.0

2.0

1.0

Reference Warren KE, Frank JA, Black JL, Hill RS, Duyn JH, Aikin AA, Lewis BK, Adamson PC, Balis FM. 2000. Proton magnetic resonance spectroscopic imaging in children with recurrent primary brain tumors. J Clin Oncol 18(5): 1020–1026.

735

44

Physiological MRI techniques and pediatric stroke Dawn Saunders1, W. Kling Chong1 and Vijeya Ganesan1,2 1

Great Ormond Street Hospital, London, UK Institute of Child Health, London, UK

2

Key points • Stroke in childhood is an under-recognized disorder. • Normal age-related variations in diffusion, perfusion and metabolite levels should be kept in mind. • Acute stroke in children shows similar patterns to that in adults, with reduced apparent diffusion coefficient (ADC) (diffusionweighted imaging hyperintensity), reduced flow, increased lactate and reduced N-acetyl aspartate. • For the most part, as in adults, a low ADC predicts subsequent infarction, but occasionally ADC changes may be reversible. • ADC is elevated in reversible posterior leukoencephalopathy syndrome (i.e. vasogenic edema). • Perfuson deficits (increased mean transit time) are common in sickle-cell disease. • Perfusion MRI may be helpful in the evaluation of lobar hypoperfusion (delayed time to peak, cf. to the posterior fossa) in moyamoya disease, and monitoring the effects of revascularization surgery.

Introduction Stroke is an important and under-recognized disorder in children and is one of the top 10 causes of childhood death (Fullerton et al., 2002). Arterial 736

ischemic stroke affects around 8 of 100 000 children annually. Up to a quarter of these children will have a recurrence and two-thirds have long-term disability directly attributable to stroke (Ganesan et al., 2000). Many advances in the understanding of childhood stroke have arisen due to the insights available from modern imaging techniques, in particular from MR imaging (MRI). The aims of conventional MRI are not only to detect the infarct but also to provide information to establish the cause of the stroke and to exclude other causes (such as tumor). Clinical applications of physiological MRI techniques (MR diffusion imaging, MR perfusion imaging, and MR spectroscopy, (MRS) in this group of patients are still largely in the research domain. This chapter will consider arterial ischemic stroke (henceforth abbreviated as stroke) in children over 1 month of age. There are some important differences in stroke etiology between adults and children. Around half of the children affected by stroke will have another recognized medical condition, most commonly sickle-cell disease (SCD) or congenital heart disease. Thus, many children may have dual pathologies on imaging, as well as factors, which may influence the interpretation of physiological MRI (e.g. chronic hypoxia or polycythemia). Rather than having a single identified etiology, the majority of children with stroke will have a combination of multiple risk factors. As well as those already mentioned, other important risk factors for stroke in children are anemia (which is found in up to 40% of cases), prothrombotic disorders and infections such as varicella zoster (Ganesan et al., 2003). Cerebral arteriopathy is found in up to 80% of children with stroke. This most commonly affects focal

Physiological MRI techniques and pediatric stroke

areas of large intracranial arteries, although more diffuse pathologies affecting smaller vessels (e.g. cerebral vasculitis) are also recognized. The intracranial distribution of arterial disease is in contrast to the preponderance of cervical atheroma in adults with stroke. Although in some cases it is possible to identify specific diagnostic entities such as arterial dissection or moyamoya syndrome, the commonest abnormality identified is occlusion or stenosis of unknown etiology affecting the terminal internal carotid artery (ICA) or proximal middle cerebral artery (MCA). Some cases of stenotic arterial disease are associated with preceding varicella infection. Embolic sources are only identified in the minority of cases. Arterial abnormalities resolve over time in a proportion of cases, suggesting that they may be related to a transient inflammatory process. As will be discussed further below, the nature and location of arterial pathology is important in relation to interpretation of data obtained from MR perfusion imaging. In contrast to adults, in whom the diagnosis of stroke can often be made with confidence on clinical grounds, the differential diagnosis of acute hemiparesis in children is wide, encompassing neoplastic, metabolic and inflammatory disorders. In this context, physiological MRI techniques can provide critical diagnostic information, as will be discussed further later. The wide differential diagnosis and perceived rarity of pediatric stroke means there are significant delays in recognition and referral for imaging investigations.

Practical issues of MRI in children Commercial MR machines are designed for adult practice and few manufacturers make provisions for the issues encountered in pediatric practice. Currently, most children are imaged in units that have a wider adult practice with only a handful of institutions in the developed world that have dedicated pediatric MR facilities. Technologists need to be creative in the use of available receiver coil technology in order to obtain the highest quality images for the particular pediatric body part to be examined. The higher spatial resolution demanded to study the smaller anatomic

parts will often require modifications of the default adult sequences that are supplied by the manufacturers for adult practice. In many situations, the pulse sequences need to run for longer acquisition times. As discussed in the next section, the rapidly changing and developing physiology of the brain in the first months and years of life necessitate the creative modification of pulse sequence parameters in order to optimize the soft tissue contrast between anatomic and pathological structures of interest in those age groups. Once these hardware and software options have been optimized, it is then necessary to consider whether or not the subject is able to keep sufficiently still for the duration of the examination. Most subjects approximately under the age of 3 months and over the age of 5 or 6 years will undergo MR studies without the need for sedation or general anesthesia. It is worth noting that where sedation is required, it is usually in the form of heavy sedation, often heavier than is required for computed tomography (CT) studies, nuclear medicine studies or in adult practice because of the longer examination times and higher noise levels encountered in MR scanners. At our institution, general anesthesia is used only if there is a specific contraindication to sedation. The sedation is applied by specially trained nurses who are trained to decide when and how much sedation to be given, equipped to administer and monitor its effects and skilled to recover and clinically discharge the subject at the end of the procedure (Sury et al., 1999). Many other institutions prefer to only offer the use of general anesthesia for subjects who are unable to keep still. During the course of investigations, children may also need to undergo painful or uncomfortable procedures, such as lumbar punctures or transoesophageal echocardiography. If sedation or general anesthesia is to be used to facilitate an MR scan, it is often helpful to be able to perform these other investigations at the same time.

Normal values of physiological imaging in childhood Changes that occur within the developing brain must be considered when investigating children to

737

738

Dawn Saunders, W. Kling Chong and Vijeya Ganesan

(a)

(b)

(c)

Fig. 44.1 (a) Coronal T1-weighted image and (b) axial short tau inversion recovery (STIR) image of a 3 week normal child showing areas of increased water content in the frontal and parietal WM (white arrows). (c) Corresponding areas of high signal on the DW image may be mistaken for WM damage.

allow the correct interpretation of MRI. Changes in normal values of measured parameters in very young children are expected as the brain undergoes very rapid development during the first 2–3 years of life. The first descriptions of brain maturation during childhood with MRI demonstrated T1 and T2 shortening which was thought to reflect the progression of myelination in white matter (WM) tracts (Barkovich, 1988). Perfusion and metabolic changes that occur during normal development have been studied using single photon emission computed tomography (SPECT) (Chiron et al., 1992) and fluorodeoxyglucose (FDG) positron emission tomography (PET) (Chugani et al., 1987), respectively. MR studies have also shown dramatic changes over this period in the transverse relaxation rate (Scott et al., 2001), apparent diffusion coefficient (ADC) values (Forbes et al., 2002), diffusivity (Mukherjee et al., 2001) and metabolite concentrations (Kries et al., 1993; Pouwels et al., 1999). Changes in diffusion MRI with age Diffusion MRI is a technique that is sensitive to the microscopic diffusion of water within tissues (Moseley et al., 1990; Pierpaoli et al., 1996) (cf. Chapter 4). The more water movement detected within the brain, the greater the ADC. The greater the order of the environment, for example within

WM tracts, the greater the directionality of movement and greater the anisotropy. Diffusion measurements in the brains of newborns and infants show that there is more movement of water (high ADC) with less directionality (low anisotropy), within the immature and unmyelinated brain than in the mature and myelinated adult brain (Morriss et al., 1999; Mukherjee et al., 2001). The most dramatic decrease in ADC values occur within the first few months and the greatest changes is seen in the frontal and parietal WM (Forbes et al., 2002). This is visible on the ADC image and confirmed quantitatively with signal intensity measurements (Englebrecht et al., 2002). Failure to appreciate normal changes in the ADC with age may result in the erroneous reporting of diffusionweighted imaging (DWI). For example, the high ADC seen in the neonatal period in the WM could be interpreted as an area of free diffusion resulting from a mature infarct (Figure 44.1). Reduced diffusivity with age is seen within WM tracts such as the corpus callosum and posterior limb of the internal capsule (PLIC) and is thought to be due to myelination, whilst gray matter (GM) structures such as basal ganglia demonstrated very small changes (Neil et al., 1998; Mukherjee et al., 2001). The decrease in diffusion in the WM during brain maturation has been explained initially in terms of the development of myelin, which acts as a

Physiological MRI techniques and pediatric stroke

barrier to diffusion. However, the initial decrease in ADC and increase in anisotropy predates the T1- and T2-weighted changes on conventional MR images and is thought to represent changes of “premyelination” (Wimberger et al., 1995). Recent experiments using non-myelinated nerves demonstrate that high spatial anisotropy exists prior to maturation of myelination (Beaulieu et al., 1994; Huppi et al., 1998). It has been proposed that this pre-myelination anisotropy could be related to an increase in fiber diameter, axonal membrane changes or early wrapping of axons by oligodendroglial processes. A detailed study of diffusion anisotropy by Mori et al., in the developing mouse brain detected early changes in the organization of the cortex and underlying WM which predated myelination and correlated well with cellular changes seen on histology (Mori et al., 2001). Changes in MR perfusion imaging with age MR perfusion is increasingly used in the measurement of cerebral perfusion in children. The technique is particularly suited to children because it does not use ionizing radiation (unlike CT perfusion imaging, SPECT, or PET) but there are only two small published studies demonstrating MR perfusion imaging changes with age. The first study was carried out in normal children by Perthen et al. (2002), in a group of children between 5 and 30 months of age, with a skin nevus or febrile convulsions and subsequently found to be normal. Using bolus-tracking perfusion MRI, they showed an increase in cerebral blood flow (CBF) with age in both cortical and deep GM, WM and the cerebellum which correlated well with SPECT studies (Perthen et al., 2002). A second study using pulsed arterial spin labelling (ASL) in seven children aged 1 month to 7 years reported a 70% increase in ASL signal and a 30% increase in absolute CBF in children compared with adults. A significant linear decrease in ASL signal was seen with age (Wang et al., 2003). The changes with age of CBF, determined by perfusion imaging, correlate well with those reported from studies using SPECT. Chiron et al. (1992) used [133Xe]-SPECT to measure CBF in 42 normal children aged 2 days to 19 years (29 were 30months old), and 32 normal adults. Mean CBF was

found to rise from birth to a maximum (30% increase) that was maintained between approximately 4 and 8 years of age, before decreasing towards adult values. These results are supported by transcranial Doppler measurements of blood flow in the major arteries supplying the brain, which have been shown to increase from birth to a maximum at around 3 or 4 years, which is sustained until a decline from approximately 7 years towards adult values (Newell and Aaslid, 1992). The increase in perfusion values is thought to be related to the increase in cerebral metabolic demand by the maturing brain, as demonstrated by FDG PET (Chugani et al., 1987). However the maximum metabolic rates seen around 3 or 4 years of age were maintained until 9 years of age, when they began to decline towards adult values. The timecourse of metabolic changes better corresponds to the process of initial overproduction and subsequent elimination of excessive neurons and synapses known to occur in the developing brain (Huttenlocher and Dabholkar, 1997). Changes in metabolite concentrations with age During the development of the brain, the most important spectral changes are due to changes in the metabolic composition of the brain. Less marked changes result from the process of myelination, the establishment of neuronal connections, the accumulation of iron-containing substances, and decreasing water content, which results in changes in the relaxation times of water and metabolites. The interpretation of MRS in unilateral or focal disease allows the use of the contralateral hemisphere as the control. However, in the study of children with bilateral or global disease, the use of either water as an internal standard or metabolite ratios requires a detailed understanding of the changes with age of intracerebral water and metabolite concentrations. At birth the myo-inositol (mI) peak dominates the spectra, choline (Cho) is responsible for the strongest peak in older infants and total N-acetyl aspartate (tNAA) and creatine (Cr) dominate the spectra of older children and adults (Kreis et al., 1993). Lactate (Lac) is not observed in the normal brain of either children or adults (Figure 44.2). Due to the vast differences in the metabolite concentrations it is

739

740

Dawn Saunders, W. Kling Chong and Vijeya Ganesan

necessary to define normal age-related peak ratios before attempting to interpret a spectrum of a child with a suspected pathology. The total NAA peak is composed of both NAA and N-acetyl aspartate glutamate (NAAG), which increases continually during the first few months of life. Throughout childhood the total NAA gradually increases until 10 years of age with some regional differences. mI and Cho show a very strong decrease during development, whilst total Cr, glutamate (Glu) and glutamine (Gln) increases throughout the first year of life and then remains constant after 1 year of age (Pouwels et al., 1999).

4 days

5 months

Clinical applications of diffusion imaging in pediatric stroke

4 years

Adult

NAA Cr Cho Cr mI NAA Glx Glx

4

3

2 ppm

Lac CH2 CH3

1

0

Fig. 44.2 Typical 1H-MRS proton spectra (TR 1500 ms, TE 30 ms) acquired from the parietal lobes of normal children of different ages. This is compared to an adult spectrum where the NAA peak at 2 ppm dominates the spectrum (lowest trace) whereas the Cho and mI are much higher then the NAA peak in the newborn (top spectra). The middle two rows contain spectra from a 5-month old and a 4-year old, demonstrating the transition between the two extremes. The spectra are scaled individually to the largest peak in each trace. (Mag Res Med 1993; 30: 424–437, with permission from John Wiley publishers). Glx: Glutamate (Glu)  Glutamine (Gln); Lac: lactate; CH2, CH3: methyl groups; Cr: Creatine.

As in adults, diffusion imaging has the potential to detect hyperacute cerebral infarction within minutes of stroke onset, prior to changes visible on CT and T2-weighted MR which may not be visible for several hours (Moseley et al., 1990). In practice, CT is still the first line investigation in children with stroke and, children are rarely imaged using MR before T2weighted changes occur. However, children who are already inpatients (e.g. cardiac patients, recent onset stroke patients) may be imaged early and in these children DWI can be used to detect infarcts of different ages. In the absence of atrophy, the appearance of both acute and chronic infarction on T2-weighted imaging is demonstrated by increased signal on T2-weighted sequence and a low signal on a T1weighted sequence. Contrast enhancement is usually apparent 5–10 days following infarction. By comparing the diffusion characteristics of infarcted tissue to normal tissue, diffusion imaging can easily distinguish between acute and chronic cerebral infarction, as they show low signal (restricted diffusion) or high signal (free diffusion) abnormalities, respectively, on the ADC map (Figure 44.3). DWI studies may show abnormalities that are not solely the result of diffusion characteristics and can be contaminated by “T2 shine through effects” (Provenzale et al., 1999), and will not always differentiate acute from chronic infarcts, as both will return high signal.

Physiological MRI techniques and pediatric stroke

(a)

(b)

(c)

(d)

Fig. 44.3 Images of a 13-year-old child who presented with a left hemiparesis found to be secondary to a dissection in the neck. (a) Axial T2-weighted image at the level of the head of the caudate demonstrating typical increased T2 signal due to infarction, in the distribution of the right MCA territory. (b) A corresponding DW image. The region of infarction is high signal; a combination of both restricted diffusion and T2-signal. (c) The corresponding ADC map demonstrates a reduced signal in the MCA territory in keeping with an acute infarct. (d) Hematoma is seen within the right internal carotid at the level of the C1 vertebral body on the axial fat-saturated T1-weighted image of the neck.

741

742

Dawn Saunders, W. Kling Chong and Vijeya Ganesan

T2-weighted EPI

DWI

ADCAV

Fig. 44.4 MR images from a 12-year-old boy who developed new symptoms 2 days following onset of hemiplegia. The T2-weighted images show two areas of abnormality, the deep watershed territory of the left frontal WM and the left fronto-parietal cortex. The DWI and ADC maps demonstrate restricted diffusion in the deep watershed infarct (high signal on DWI and low on ADC map) and areas of free diffusion in the cortical lesion (low on DWI and high on ADC map). The area of restricted diffusion on the ADC map (low signal) is the new infarct.

The temporal changes seen in regions of infarction allows the differentiation of acute from chronic stroke in children with multiple ischemic lesions seen on conventional MRI at the time of their first symptomatic event (Figure 44.4). This is of particular value in children with SCD as 25% of children have silent infarcts and 17% go on to develop further infarcts (Saunders et al., 2002). Embolism arising from a cardiac source or dissection of the extracranial vessels can also cause multiple infarcts. In these patients, it can be difficult to determine whether new neurological symptoms represent a new ischemic event or the unmasking of a prior deficit due to trauma or an intercurrent illness. Animal models have convincingly shown that DWI changes in ischemia are reversible (Minematsu et al., 1992; Lo et al., 1994). There are, however, few reports in humans of the reversibility of DWI changes, and the vast majority of children presenting with stroke with changes on DWI go on to develop cerebral infarction. Reversible DWI abnormalities have been seen in two children in a group of 28 with stroke that did not progress to infarction. One patient who presented with a right hemiplegia showed extensive high signal on DWI, with evidence of tissue

swelling but without T2-weighted hyperintensity change. DWI signal abnormality persisted for over 2 weeks, with no imaging indication of infarction (Figure 44.5). A second child with a MCA territory infarct had a concomitant area of T2-weighted and DWI, and also had an area of signal hyperintensity more posteriorly with no corresponding signal change on T2-weighted imaging. Follow-up imaging revealed maturation of the area with hyperintensity on T2-weighted imaging but not imaging change in the more posterior area (Connelly et al., 1997). There have been a few isolated cases of ADC reversal in the adult population in patients receiving aggressive treatment such as thrombolysis. In a recent review of 116 MRI studies with low ADC on the initial scan, five cases of ADC reversal were identified which occurred in the clinical setting of venous sinus thrombosis and seizure (n  1), hemiplegic migraine (n  1) and hyperacute infarction following thrombolysis (n  1). The locations of ADC reversal were WM, deep gray nuclei and cortical GM (Grant et al., 2001). The identification of these rare but potentially reversible changes may have implications to some children in the future, with respect to treatments such as thrombolysis.

Physiological MRI techniques and pediatric stroke

(a)

(b)

(c)

(d)

Fig. 44.5 Axial T2-weighted images (a and c) and DWI (b and d) of a 2-year-old child acquired 20 h after presentation with a right hemiparesis. The T2-weighted images show an area of swollen cortex without hyperintensity in the left parieto-occipital region (the apparent focus of WM high signal in the left hemisphere) (b) is partial volume artifact from the upper part of the lateral ventricle. The corresponding diffusion-weighted images show hyperintensity in the equivalent anatomical locations to the areas of cortical swelling. A cerebral angiogram was normal and the patient made a full recovery. (Connelly et al., 1997, with permission from the Br Med J Publishing Group).

Use of DWI in the differential diagnosis of ischemia Diffusion imaging can differentiate ischemia from cerebral edema seen in the reversible posterior leukoencephalopathy syndrome (RPLS) (Coley et al., 1999). The syndrome is characterized by headaches, confusion, seizures and visual disturbances

associated with transient, predominantly posterior cortical and juxta-cortical lesions seen on imaging (Hinchey et al., 1996). The syndrome has been seen in association with cyclosporin toxicity (Coley et al., 1999), hyertensive encephalopathy (Hinchey et al., 1996) and the administration of chemotherapeutic

743

744

Dawn Saunders, W. Kling Chong and Vijeya Ganesan

agents (Edwards et al., 2001; Shin et al., 2001). A longitudinal study of two cases of RPLS from our institution, revealed high signal areas on ADC maps within the cortex and juxta-cortical WM of the parietal and occipital lobes, indicative of vasogenic edema, which corresponded to areas of high signal on the T2-weighted images. At no time was the restricted diffusion (low signal on ADC map) of ischemia demonstrated on the ADC map (Coley et al., 1999). This supports the theory that RPLS is secondary to the passive extravasation of fluid from blood vessels (vasogenic edema) and it has been proposed that it is due to disruption of the blood– brain barrier either as a result of persistent high blood pressure or direct damage to the blood vessels by drugs (Figure 44.6). Diffusion tensor imaging in children with basal ganglia infarcts and language difficulties Diffusion tensor imaging (DTI) is rarely carried out in children due to the difficulty of keeping children still for prolonged periods. The relationship between language function and MRI abnormalities has been investigated in 17 older children with acquired unilateral basal ganglia infarctions. Analysis of 3D structural MR images and DTI revealed that poor language performance was associated with decreased GM density in Broca’s and Wernicke’s areas, decreased diffusion anisotropy and WM density in deep left frontal WM, and decreased diffusion anisotropy in the internal capsule. Perfusion abnormalities within the fronto-parietal regions and Wernicke’s area were seen in the three patients with the poorest language function (Tournier et al., 2003).

Clinical applications of MR perfusion imaging in pediatric stroke MR perfusion imaging has the potential to provide information about the hemodynamic status of cerebral tissue in children presenting with stroke. The concept, developed in evaluation of adult stroke patients, that the combination of structural, diffusion and perfusion MRI can distinguish between tissue

which is irreversibly damaged, unaffected tissue and compromised but potentially salvageable tissue is also relevant to the evaluation of children with stroke. Although, for the reasons mentioned in the introduction, children are rarely seen in the hyperacute phase of stroke, children with cerebrovascular pathology may have more chronic tissue compromise. Two specific groups of such patients are children with SCD and those with moyamoya syndrome. MR perfusion imaging in SCD Stroke is 250 times more common in children with SCD than in other children; 25% of people with homozygous SCD will have had a stroke by the age of 45. In addition, up to 25% of children with SCD have evidence of clinically silent cerebral infarction, which adversely affects cognitive function. The majority of children with SCD who go on to develop stroke have an arteriopathy affecting the terminal ICA or proximal MCA, anterior or posterior cerebral arteries (ACA or PCA) which can be detected using trans-cranial Doppler ultrasound or MR angiography (MRA). However, such arteriopathy is not found in up to 20% of cases. Other factors such as small vessel disease, anemia or hypoxia may also contribute to pathogenesis. In a study undertaken by Kirkham et al. (2001), 48 children with SCD (including 10 who had never had any neurological symptoms or signs) were studied with structural, diffusion and bolus-tracking MR perfusion imaging and MRA. T2-weighted MRI revealed areas of cerebral infarction in 22 cases with MR diffusion abnormalities (increased ADC) in a corresponding distribution but not elsewhere. MR perfusion imaging revealed abnormalities (reduced CBF and increased mean transit time (MTT)) in 25 cases (52%), extending beyond areas of infarction in nine cases and distinct from infarcts in another nine. All the patients had neurological or cognitive dysfunction. Perfusion abnormalities were significantly more likely, but not exclusively confined to, patients with abnormal MRA. The conclusions from this study were that the combination of structural and physiological MRI techniques enabled identification of tissue with hemodynamic compromise in structurally intact tissue that was a potential target

Physiological MRI techniques and pediatric stroke

(a)

(b)

(c)

(d)

(e)

(f)

Fig. 44.6 Axial MR images of a 9-year-old patient with SCD and a viral-induced nephritic syndrome. The patient developed a generalized seizure 4 days after commencing CSA therapy and experienced prolonged impairment of higher neurological function. Regions of increased diffusion matched all the areas of T2 signal change. Many of the areas resolved without any residual T2 or diffusion abnormality. These findings suggest that the neurotoxic effects of CSA were associated with a partially reversible extravasation of fluid into the brain. (a) A T2-weighted image (day 2 after seizure) shows cortical and juxta-cortical signal alteration within both frontal and parietal lobes (arrows). (b) The ADC map (day 2) reveals increased diffusion in all areas of T2 abnormality (arrows). (c) A T2-weighted image (day 6) shows new, more superior abnormalities (arrows). (d) The ADC map (day 6) again reveals that all the T2-weighted abnormalities correspond to all areas of increased diffusion (arrows). (e) A T2-weighted image (day 49), obtained at the same level as the images in (a) and (b), show complete resolution of most of the lesions that were present on day 2 (open arrows). A new lesion is present in the left posterior frontal lobe (solid arrow). A pre-existing abnormality in the left parietal lobe (curved arrow) is essentially unchanged. (f) The ADC map (day 49) reveals increased diffusion in the new area of T2 prolongation (solid arrows). No diffusion abnormality is present where the lesions have resolved (open arrows). (Reprinted with permission from Am J Neuroradiol 1999; 20: 1507–1510).

745

746

Dawn Saunders, W. Kling Chong and Vijeya Ganesan

(a)

(b)

T2-weighted TSE

DW-EPI

ADCAV

MTT

CBV

CBF

Fig. 44.7 A 5-year-old boy with SCD and bilateral moyamoya syndrome. Although the T2-weighted and diffusion images show a chronic left occipital infarct (a), there is also an extensive CBF abnormality in the right temporal lobe (b) with increased MTT in equivalent regions (Stroke 2002; 33: 1146–1151, with permission from Lippincott, Williams & Wilkins). EPI: echo planar imaging; TSE: turbo spin echo.

for intervention. It was suggested that such imaging could be used as an adjunct to current evaluation strategies to refine the need for and effects of treatments such as regular blood transfusion. An example of a large perfusion abnormality is shown in (Figure 44.7). More recently, Oguz et al. (2003) studied cerebral hemodynamics in 14 asymptomatic children with

SCD with MRI using continuous arterial spin labelling (CASL). None of the patients had disease of major intracranial arteries; one had silent frontal infarcts. CBF was significantly increased in all vascular territories in the patient group compared to the controls. Four children (including one with silent infarcts) had lateralised reductions of CBF but the implications of this for future stroke or silent

Physiological MRI techniques and pediatric stroke

infarction are unknown at present. CASL is likely to be particularly attractive in patients with SCD in whom, despite any demonstrable adverse effects, there is often reluctance to administer gadolinium. MR perfusion imaging in moyamoya syndrome Moyamoya syndrome is a cerebral arteriopathy with stenosis or occlusion of the terminal ICA and profuse basal collaterals. The commonest clinical presentation is with stroke or transient ischemic attack (TIA) in childhood and affected patients have a high rate of stroke recurrence and cognitive decline. Studies using a variety of techniques (such as SPECT, xenon-CT and PET) have established that patients commonly have evidence of chronic cerebral tissue hypoperfusion. However, MRI has the advantage of providing both structural and physiological information in a single examination and, of particular relevance to this condition, information about the structural integrity of hemodynamically compromised tissue. Tsuchiya et al. (1998) studied 19 patients of varying age with moyamoya syndrome using bolustracking MR perfusion imaging and generated semi-quantitative flow maps based on the signal intensity–time curves and the bolus arrival time. Asymmetry or focal hypoperfusion was found in 17 cases and there was good correlation with SPECT studies. In a study of 13 children with moyamoya syndrome using structural, diffusion and perfusion MRI, hemodynamic abnormalities affecting structurally intact tissue were observed in all cases (Calamante et al., 2001). This included apparently low CBF as well as abnormalities of the perfusion summary parameters MTT and maximum peak concentration (MPC). Quantitative analysis of CBF using deconvolution was shown to be unreliable due to delay and dispersion effects, which are issues relevant to all MR perfusion studies of this condition. In a quantitative regional analysis of summary parameters, using the cerebellum as a reference, we were able to demonstrate a relationship between abnormal perfusion summary parameters and the patients’ clinical status. The most clinically unstable patients (with frequent TIA or recurrent stroke) were shown to have the most severe abnormalities. Although

a cut-off of 5 seconds for the peak width distinguished between these and the other patients, this threshold is unlikely to be reliably generalizable to other patients because of the small numbers included in this study. It is important to note that, in the context of moyamoya syndrome, the relationship between abnormalities on MR perfusion imaging and clinical status is not a linear one and that, at present, MR perfusion imaging cannot be used to predict reliably the clinical outcome in individual cases, as illustrated in (Figure 44.8) and (cf. Case Studies 44.1 and 44.2). Study of cerebrovascular reactivity with MR perfusion imaging (using acetazolamide or carbon dioxide provocation), which has been undertaken in adults with carotid atheroma, has potential to provide additional dynamic information about cerebral hemodynamic status in patients with moyamoya syndrome but has not been reported to date. A further application of perfusion imaging in this patient group is in the evaluation of treatment efficacy in children undergoing surgical revascularization. Lee et al. (2003) serially studied 13 children with moyamoya treated with encephalo-arteriosynangiosis. Compared to controls, the patients tended to have increased rCBV and delayed time to peak (TTP) pre-operatively. Post-operatively, TTP was significantly reduced in the revascularized MCA territory and, in patients with increased rCBV preoperatively, rCBV was reduced. In a single case TTP reduction was shown to be mild after 1 month but greater 6 months after surgery, suggesting that this improvement was due to development of collaterals in the revascularized territory. The relationship between changes in perfusion status and clinical outcome were not discussed in detail. Improvement in both MR perfusion imaging and clinical status after encephalomyo-synangiosis was reported in a single adult patient by Wityk et al. (2002). In our experience, persistent hemodynamic abnormalities were apparent in patients who had undergone revascularization (usually direct arterial anastomosis in our centre), including those who had had benefited symptomatically from surgery (Calamante et al., 2001). We have seen patients in whom MR perfusion abnormalities have clearly

747

748

Dawn Saunders, W. Kling Chong and Vijeya Ganesan

(a)

(b)

(d)

(c)

(e)

Fig. 44.8 (a) Normal axial T2-weighted images and (b) ADC map of a 13-year-old boy with bilateral moyamoya syndrome. (c) the corresponding TTP map shows bilateral areas of reduced TTP (slower flow) in the fronto-parietal lobes despite normal imaging. (d and e) right and left internal carotid angiograms show bilateral terminal carotid occlusion and typical moymoya collaterals best seen in (d) collaterals from the ophthalmic artery (OphtA) and posterior leptomeningeal collaterals are also visible. The PCA arises from the right ICA (d) which is a normal variant.

improved following surgery as well as those in whom these remain unchanged (Calamante et al., 2001; Ganesan, personal communication) and are currently undertaking a prospective study exploring the relationship between perfusion status and clinical outcomes.

Clinical applications of MRS in pediatric stroke MRS in arterial ischemic stroke Although, in vivo MRS has been used to study cerebral infarction in adults, only a single published

Physiological MRI techniques and pediatric stroke

MRS in metabolic stroke Spectroscopy has been carried out in four children with clinically and genetically defined mitochondrial encephalopathy with lactic acidosis and stroke (MELAS). The [1H] MRS findings in lesions within the occipital lobe were characterized by elevated Lac and glucose and severely reduced concentrations of total NAA, Glu and total Cr (Wilichowski et al., 1999). In the acute period, these findings are explained by a period of hypoxia in lesions caused by stroke-like episodes. The increase in the non-oxidative glycosolyation results in the accumulation of Lac, which may damage or cause loss of neuronal tissue as evidenced by the decrease of the neuronal markers of NAA and NAAG (Birken and Oldendorf, 1989), (Figure 44.9). The reduced concentrations of total Cr and Cho in MELAS are further indication of the cellular degeneration of neurons and axons, even though some lesions in MELAS are reversible. In the chronic phase, the visualization of Lac in these lesions may be due to infiltrating macrophages which rely on anaerobic glycolysis, as the neuronal tissue is no longer metabolically active (Graham et al., 1993).

Lac 100

80

TE (ms)

study has been carried out in children. Wang et al. in a study of children with SCD, demonstrated a reduced NAA and raised Lac in regions of chronic infarction determined by MRI (Wang et al., 1992). Reduced NAA and Cr and raised Lac were detected in a child with an anterior cerebral infarct secondary to a vasculitis caused by neurocystercycosis (Kohli et al., 1997). These metabolite changes do not differ from [1H] MRS changes seen in regions of infarction in adults (Saunders et al., 1995). To date, MRS has not been carried out in areas of normal appearing brain and reduced CBF determined by MR perfusion studies. Lac, the end product of glycolysis, is a particularly useful measure of metabolism, as it does not occur in the normal brain. The concentration of Lac rises when the glycolytic rate exceeds the tissues’ capacity to catabolize it or remove it from the blood stream. The rise in brain Lac results from mismatch between glycolysis and oxygen supply and is a hallmark for the detection of ischemia.

60 Cr 40

Cho mI

Glu NAA Gln

20

4.0

3.5

3.0

2.5 2.0 ppm

1.5

1.0

0.5

Fig. 44.9 Single voxel stimulated echo acquisition mode (STEAM), TE  10 ms, spectra acquired from a child with an occipital infarct and demonstrates grossly elevated Lac doublet with a reduction in NAA. This is a patient with lactic acidosis and MELAS-type syndrome. (Reprinted from MRI Clinics North America 2001; 9(1): 165–189, with permission from Jill Hunters and Elsevier).

We have observed raised Lac and reduced NAA within the basal ganglia of a child with methylmalonic acidaemia (Figure 44.10). These findings are comparable to 1H MRS findings seen in both adult and pediatric stroke (Wang et al., 1992; Saunders et al., 1995) and illustrate the fact that metabolite changes of infarction determined by [1H] MRS are non-specific and reflect the underlying pathology and not the etiology. As a result, 1H MRS does not currently have a clinical role in the investigation of children with stroke. Of particular interest in children with metabolic stroke is the demonstration of the partial recovery of the NAA metabolite concentration, which has been reported in a single patient with MELAS (DeStafano et al., 1995). Recovery of NAA has also been demonstrated in multiple sclerosis (MS) plaques (Arnold et al., 1992; Davie et al., 1994) and in the

749

750

Dawn Saunders, W. Kling Chong and Vijeya Ganesan

(a)

(b)

Fig. 44.10 Proton (1H MRS) spectroscopy in a 6-year-old girl of Asian origin with methylmalonic acidemia, who presented with hypotonia and the “locked in” syndrome. (a) Bilateral basal ganglia infarction is seen on the axial T2-weighted imaging and confirmed by (b) 1H MRS. The spectra was acquired at 135 ms and reveals an inverted doublet (1.35 ppm) and a mildly reduced NAA signal (2.02 ppm).

brain of patients with acquired immunodeficiency syndrome (AIDS) dementia complex following zidoviduine therapy (Vion-Dury et al., 1995). It is thought that the reduction in NAA concentration results from impaired mitochondrial function during the acute insult (e.g. an intercurrent infection in MELAS) and the recovery of NAA represents the recovery of the metabolic function of the cell. These observations have led to the view that NAA is a marker of neuronal integrity and not a marker of neuronal numbers.

Summary Clinical applications of physiological MRI techniques in children with arterial ischemic stroke are still largely in the research domain. There are limited numbers of multi-modal MRI studies carried out in children with stroke due to the length of time required to image the patients and the ethical issues raised concerning the use of sedation and general anesthesia for the purposes of research. As a result, there are few studies that include more than two physiological MR sequences, for example diffusion and perfusion,

and those children who are scanned are often older and more co-operative. Increased use of MRI in children may result from decreased scan times as a result of improvements in imaging hardware (gradient strengths, field strength and parallel imaging). As the etiologies of childhood stroke are diverse, the combination of structural and physiological MRI techniques will continue to provide vital diagnostic information. However, in the future they are also likely to play an important role in identifying patients who may be suitable for specific treatments. In addition, these techniques show great promise in pre-symptomatic investigation of high risk groups and thus are likely to be an important component of future trials of primary and secondary interventions for pediatric stroke.

ACKNOWLEDGEMENTS

Many thanks to our colleagues in the Department of Radiology and Physics (Fernando Calamante, Alan Connelly, Joanna Perthen, Donald Tournier) at the Institute of Child Health for their thoughts and use of images.

Physiological MRI techniques and pediatric stroke

REFERENCES Arnold DL, Mathews VP, Francis GS, O’Connor J, Antel JP. 1992. Proton magnetic resonance spectroscopic imaging for metabolic characterization of demyelinating plaques. Ann Neurol 31: 235–241. Barkovich AJ, Kjos BO, Jackson Jr DE, Norman D. 1998. Normal brain maturation of the neonatal and infant brain: MR imaging at 1.5T. Radiology 166: 173–180. Beaulieu C, Allen PS. 1994. Determinants of anisotrophic water diffusion in nerves . Magn Reson Med 31: 394–400. Birken DL, Oldendorf WH. 1989. N-Acetyl-L-aspartic acid: a literature review of a compound prominent in 1H-MRS spectroscopic studies of brain. Neurosci Biobehav Rev 13: 23–31. Calamante F, Ganesan V, Kirkham FJ, Jan W, Chong WK, Gadian DG, Connelly A. 2001. MR perfusion imaging in moyamoya syndrome: potential implications for clinical evaluation of occlusive cerebrovascular disease. Stroke 32: 2810–2816. Chiron C, Raynaud C, Mazière B, Zilbovicius M, Laflamme L, Masure M-C, Dulac O, Bourguignon M, Syrota A. 1992. Changes in regional cerebral blood flow during brain maturation in children and adolescents. J Nucl Med 33: 696–703. Chugani HT, Phelps ME, Mazziotta JC. 1987. Positron emission tomography of human brain functional development. Ann Neurol 22: 487–497. Coley SC, Porter DA, Calamante F, Chong WK, Connelly A. 1999. Quantitative MR diffusion mapping and cyclosporineinduced neurotoxicity. Am J Neuroradiol 20: 1507–1510. Connelly A, Chong WK, Johnson CL, Ganesan V, Gadian DG, Kirkham FJ. 1997. Diffusion-weighted magnetic resonance imaging of compromised tissue in stroke. Arch Dis Child 77: 38–41. Davie CA, Hawkins CP, Barker GJ, Tofts PS, Miller DH, McDonald WI. 1994. Serial proton magnetic resonance spectroscopy in acute multiple sclerosis lesions. Brain 117: 49–58. DeStafano N, Matthews PM, Arnold DL. 1995. Reversible decreases in N-acetylaspartate after acute brain injury. Mag Reson Med 34: 721–727. Edwards MJ, Walker R, Vinnicombe S, Barlow C, MacCallum P, Foran JM. 2001. Reversible posterior leukoencephalopathy syndrome following CHOP chemotherapy for diffuse large B-cell lymphoma. Ann Oncol 12: 1327–1329. Englebrecht V, Scherer A, Rassek M, Wittsack HJ, Modder U. 2002. Diffusion-weighted MR imaging of the paediatric brain: Findings in normal brain and white matter disease. Radiology 222: 410–418. Forbes KPN, Pipe JG, Bird CR 2002. Changes in brain water during the first year of life. Radiology 222: 405–409. Fullerton HJ. Chetkovich DM, Wu YW, et al. 2002. Deaths from stroke in US children, 1979 to 1998. Neurology 59: 34–39.

Ganesan V, Hogan A, Jones A, Shack N, Kirkham FJ. 2000. Parent-reported outcome in ischaemic stroke. Dev Med Child Neurol 42: 455–461. Ganesan V, Prengler M, McShane MA, Wade A, Kirkham FJ. 2003. Investigation of risk factors in children with arterial ischaemic stroke. Ann Neurol 53: 167–173. Graham GD, Blamire AM, Rothman DL, Brass LM, Fayad PB, Petroff OAC, Prichard JW. 1993. Early temporal variation of cerebral metabolites after human stroke. A proton magnetic resonance study. Stroke 24: 1891–1896. Grant PE, He J, Halpern EF, et al. 2001. Frequency and clinical context of decreased apparent diffusion coefficient reversal in the human brain. Radiology 221: 43–50. Hinchey J, Chaves C, Appigani B, et al. 1996. A reversible posterior leukoencephalopathy syndrome. New Engl J Med 334: 494–500. Huppi P, Maier S, Pelad S, Zientara G, Barnes P, Jolesz F, Volpe J. 1998. Microstructural development of human newborn cerebral white matter assessed in vivo diffusion tensor magnetic resonance imaging. Pediatr Res 44: 584–590. Huttenlocher PR, Dabholkar AS. 1997. Regional differences in synaptogenesis in human cerebral cortex. J Comp Neurol 387: 167–178. Kirkham FJ, Calamante F, Bynevelt M, Gadian DG, Evans JPM, Cox TC, Connelly A. 2001. Perfusion magnetic resonance abnormalities in patients with sickle cell disease. Ann Neurol 49: 477–485. Kohli A, Gupta R, Kishore J. 1997. Anterior cerebral artery territory infarction in neurocysticercosis: evaluation by MR angiography and in vivo proton MR spectroscopy. Pediatr Neurosurg 26: 93–96. Kreis R, Ernst T, Ross BD. 1993. Development of human brain: in vivo quantification of metabolite and water content with proton magnetic resonance spectroscopy. Mag Reson Med 30: 424–437. Lee S-K, Kim DI, Jeong E-K, Kim S-Y, Kim SH, In YK, Kim D-S, Choi J-U. 2003. Postoperative evaluation of moyamoya disease with perfusion-weighted MR imaging: initial experience. Am J Neuroradiol 24: 741–747. Lo EH, Matsumoto K, Pierce AR, Garrido L, Luttinger D. 1994. Pharmacological reversal of acute changes in diffusionweighted magnetic resonance imaging in focal cerebral ischaemia. J Cereb Blood Flow Metab 14: 597–603. Minematsu K, Li L, Sotak CH, Davis MA, Fischer M. 1992. Reversible focal ischaemic injury demonstrated by diffusion-weighted magnetic resonance imaging in rats. Stroke 23: 1304–1310. Mori S, Itoh R, Zhang J, et al. 2001. Diffusion tensor imaging of the developing mouse brain. Mag Reson Med 46: 18–23. Morriss MC, Zimmerman RA, Bilaniuk LT, Hunter JV, Haselgrove JC. 1999. Changes in brain water diffusion during childhood. Neuroradiology 41: 929–934.

751

752

Dawn Saunders, W. Kling Chong and Vijeya Ganesan

Moseley ME, Cohen Y, Mintorovitch J, et al. 1990. Early detection of regional cerebral ischaemia in cats: comparison of diffusion- and T2-weighted MRI and spectroscopy. Magn Reson Med 14: 330–346. Mukherjee P, Miller JH, Shimony JS, Conturo TE, Lee BCP, Almli CR, McKinstry RC. 2001. Normal brain maturation during childhood: developmental trends characterized with diffusion-tensor MR imaging. Radiology 221: 349–358. Newell DW, Aaskid R. 1992. Transcranial Doppler: clinical and experimental uses. Cerebrovasc Brain Metab Rev 4: 122–143. (Review) Oguz KK, Golay X, Pizzini FB, Freer CA, Winrow N, Ichord R, Casella JF, van Zijl PCM, Melhem ER. 2003. Sickle cell disease: continuous arterial spin-labelling perfusion MR imaging in children. Radiology 227: 567–574. Perthen JE, Calamante F, Gadian DG, Connelly A. 2003. A novel pulsed arterial spin labelling sequence to alloe the investigation of transit times. Proceedings of the International Society of Magnetic Resonance in Medicine: 2211 (abstract). Pouwels PJW, Brockmann K, Kruse B, Wilken B, Wick M, Hanefeld F, Frahm J. 1999. Regional dependence of human brain metabolites from infancy to adulthood as detected by quantitative localised proton MRS. Paediatr Res 46: 474–485. Provenzale JR, Engleter ST, Petrells JR, Smith JS, MacFall JR. 1999. Use of exponential diffusion-weighted images to eradicate T2 shine-through effect. Am J Neuroradiol 172: 537–539. Saunders DE, Bynevelt M, Hewes DKM, Cox TC, Chong K, Evans JP, Kirkham FJ. 2002. MRI in children with sickle cell disease without overt stroke (abstract). Dev Med Child Neurol 42(suppl 90): 27. Saunders DE, Howe FA, van den Boogaart A, McLean MA, Griffiths JR, Brown MM. 1995. Continuing ischemic damage following acute middle cerebral artery infarction in man demonstrated by short echo proton spectroscopy. Stroke 26: 1007–1013. Scott RC, Gadian DG, Cross JH, Wood SJ, Neville BGR, Connelly A. 2001. Quantitative magnetic resonance characterization of mesial temporal sclerosis in childhood. Neurology 56: 1659–1665.

Shin RK, Stern JW, Janss AJ, Hunter JV, Liu GT. 2001. Reversible posterior leukoencephalopathy during treatment for acute lymphoblastic leukaemia. Neurology 56: 388–391. Sury MRJ, Hatch DJ, Deeley T, Dick-Mireaux C, Chong WK. 1999. Development of a nurse-led sedation service for paediatric magnetic resonance imaging. Lancet 353: 1667–1671. Tournier JD, Rowan A, Calamante F, et al. 2003. Changes in gray and white matter structures associated with language in patients with acquired unilateral basal ganglia infarction revealed by structural and diffusion tensor MRI (abstract). Proceedings of the International Society of Magnetic Resonance in Medicine, p. 401. Tsuchiya K, Inaoka S, Mizutani Y, Hachiya J. 1998. Echo-planar perfusion MR of moyamoya disease. Am J Neuroradiol 19: 211–216. Vion-Dury J, Nicoli F, Salvan AM, Confort-Gouny S, Dhiver C, Cozzone PJ. 1995. Reversal of brain metabolic alterations with zidovudine detected by proton localised magnetic resonance spectroscopy. Lancet 345: 60–61. Wang J, Licht DJ, Liu C, Jahng G, Haselgrove J, Detre J. 2003. Pediatric perfusion imaging using arterial spin labelling (abstract). Proceedings of the International Society of Magnetic Resonance in Medicine, p. 121. Wang Z, Bogdan AR, Zimmerman RA, Gusnard DA, Leigh JS, Ohene-Frempong K. 1992. Investigation of stroke in sickle cell disease by 1H nuclear magnetic resonance spectroscopy. Neuroradiology 35: 57–65. Wilichowski E, Pouwels PJW, Frahm J, Hanefeld F. 1999. Quantitative proton magnetic resonance spectroscopy of cerebral metabolic disturbances in patients with MELAS. Neuropaediatrics 30: 256–263. Wimberger DM, Roberts TP, Barkovich AJ, Prayer LM, Moseley ME, Kucharczyk J. 1995. Identification of “premyelination” by diffusion-weighted MRI. J Comput Assist Tomogr 19: 28–33. Wityk RJ, Hillis A, Beauchamp N, Barker PB, Rigamonti D. 2002. Perfusion-weighted magnetic resonance imaging in adult moyamoya syndrome: characteristic patterns and change after surgical intervention: case report. Neurosurgery 51: 1499–1506.

Physiological MRI techniques and pediatric stroke

Case Study 44.1 Moyamoya disease: MR perfusion Peter Barker, D.Phil. and Doris Lin, M.D. Ph.D., Johns Hopkins University School of Medicine, Baltimore History 8-year-old right-handed female with headaches and episodic left arm and leg numbness that worsen with activity.

FLAIR

DWI

MRA

MTT

Technique Conventional MRI, MRA, diffusion- and DSC perfusion-weighted MRI. Imaging findings Fluid alternated inversion recovery (FLAIR) and DWI were unremarkable. MRA shows irregularity and reduced flow bilaterally in the MCA, ACA and PCA with collateral vasculature in the suprasellar cistern extending to the deep GM. Perfusion weighted imaging (PWI) demonstrates multi-focal areas of perfusion deficits involving the bilateral frontal and left parieto-occipital regions. Discussion

Key points PWI often shows a characteristic hypoperfusion pattern in moyamoya. PWI may also be useful for monitoring reperfusion therapies.

6 ΔR2 (sec)

Bilateral moyamoya syndrome typically shows a pattern of diffuse hypoperfusion in the anterior circulation compared with the posterior circulation and basal ganglia, with the most severe changes in the WM (Wityk, 2002). MR perfusion imaging may be helpful for identifying tissue at risk of infarction in moyamoya, and can also be used to monitor the effectiveness of revascularization therapies.

4

*

2 0 20

40

60

Time (s) Reference Wityk RJ, Hillis A, Beauchamp N, Barker PB, Rigamonti D. 2002. Perfusion-weighted magnetic resonance imaging in adult moyamoya syndrome: characteristic patterns and change after surgical intervention: case report. Neurosurgery 51(6): 1499–1505.

753

754

Dawn Saunders, W. Kling Chong and Vijeya Ganesan

Case Study 44.2 Sturge–Weber syndrome: MR perfusion imaging Peter Barker, D.Phil. and Doris Lin, M.D. Ph.D., Johns Hopkins University School of Medicine, Baltimore History 9-month-old male with congenital left glaucoma, facial port wine-stain and seizures.

T2

T1Gd

CBV

MTT

Technique Conventional and perfusion MRI. Imaging findings T2 MRI shows mild left occipital and parietal volume loss and hypointensity in the sub-cortical WM. Post-gadolinium T1 MRI shows leptomeningeal enhancement. PWI reveals increased MTT in the left parietal cortex, with signal intensity–time curve showing protracted venous-phase washout. Discussion

*

ΔR2

Sturge–Weber Syndrome (SWS) is a congenital neurocutaneous disease characterized by ipsilateral cutaneous capillary angioma and intracranial leptomeningeal angiomatosis. Venous stasis occurs leading to parenchymal hypoperfusion and bioenergetic insufficiency, particularly in the presence of seizures (Maria, 1999). Slow gadolinium dimeglumine gadopentate (Gd-DTPA) washout suggests impaired venous drainage (Lin, 2003). Perfusion deficits may occur before the development of structural abnormalities (Reid, 1997). Key points PWI shows hypoperfusion and delayed contrast clearance in SWS. Perfusion imaging may be sensitive for detecting intracranial involvement in SWS.

10

20

30

40

50

Time (s)

References Lin DD, Barker PB, Kraut MA, Comi A. 2003. Early characteristics of Sturge–Weber syndrome shown by perfusion MR imaging and proton MR spectroscopic imaging. Am J Neuroradiol 24(9): 1912–1915. Maria BL, Hoang KBN, Robertson RL, Barnes PD, Drane WE, Chugani HT. 1999. Imaging brain structure and function in Sturge–Weber syndrome. In Sturge–Weber syndrome (Eds, BodensteinerJB, Roach ES), pp. 43–69. Reid DE, Maria BL, Drane WE, Quisling RG, Hoang KB. 1997. Central nervous system perfusion and metabolism abnormalities in Sturge–Weber syndrome. J Child Neurol 12: 218–222.

45

MR spectroscopy in pediatric white matter disease Folker Hanefeld1, Knut Brockmann1 and Peter Dechent2 1

Department of Pediatrics and Neuropediatrics, Georg-August-University, Göttingen, Germany Biomedizinische NMR Forschungs GmbH am Max-Planck-Institut für Biophysikalische Chemie, Göttingen, Germany

2

Key points • Multiple, diverse genetic disorders result in white matter (WM) disease. • MR spectroscopy (MRS) has specific patterns which assist diagnosis in Canavan’s disease (high N-acetyl aspartate (NAA) and N-acetylaspartylglutamate (tNAA)), succinate dehydrogenase deficiency (SDH, complex II deficiency, elevated succinate), myelinopathia centralis diffusa/vanishing white matter (even reduction, later total loss of all metabolites). Elevated lactate indicates an anaerobic metabolic state of tissue, that can be found in a variety of demyelinating disorders, especially in mitochondrial leukoencephalopathies. • In other leukodystrophies, MRS may provide information on axonal integrity (tNAA), demyelination (high choline-containing compounds (Cho) and/or myo-inositol (mI)), hypomyelination (low Cho, normal NAA) or gliosis, but does not assist making a specific diagnosis. • Leukodystrophies often exhibit a wide range of phenotypic variability (clinically and spectroscopically).

MR spectroscopy (MRS) provides in vivo information into the metabolic alterations occurring during the different stages of white matter (WM) diseases. Inborn metabolic/genetic disorders are separate from acquired, mainly inflammatory or hypoxic– ischemic conditions (which are treated in Chapters 23, 25, and 41).

In this chapter we describe metabolic characteristics as shown by MRS of WM caused primarily by hereditary conditions (leukodystrophies – Table 45.1). Our classification of leukoencephalopathies is based on the concept of hypo-, and demyelination (Table 45.2). Some of these disorders are also discussed in Chapters 25 and 46. There are only few comprehensive reviews on this topic available (Grodd et al., 1991; van der Knaap et al., 1992; Tzika et al., 1993a; van der Knaap and Valk, 1995; Frahm and Hanefeld, 1997). Most publications deal with individual disease entities and will be referred to in the respective sections.

Lysosomal disorders Metachromatic leukodystrophy Metachromatic leukodystrophy (MLD, MIM #250100) is a lysosomal storage disorder transmitted as an autosomal-recessive trait (von Figura et al., 2001). A deficiency of arylsulfatase A (ASA) leads to accumulation of cerebroside sulfates in cerebral WM and peripheral nerves. Mutations in the ARSA gene on chromosome 22q13.31-qter constitute the molecular basis (Polten et al., 1991). As in Krabbe disease (KD) (globoid cell leukodystrophy, GLD), three variants have been delineated according to age of onset. The rapidly progressing late infantile form is characterized by spasticity and ataxia together with peripheral neuropathy and optic atrophy. Juvenile MLD shows similar symptoms with a more protracted course. Adult MLD may start with psychopathological symptoms (Betts et al., 1968). 755

756

Folker Hanefeld, Knut Brockmann and Peter Dechent

Table 45.1. Classification of hereditary WM disorders (adapted from van der Knaap and Valk) Lysosomal disorders MLD Globoid cell leukodystrophy (Morbus Krabbe, GLD) Peroxisomal disorders Zellweger cerebrohepatorenal syndrome X-linked adrenoleukodystrophy (X-ALD) Neonatal ALD Refsum’s disease (classic/infantile) Mitochondrial encephalomyopathies Succinate dehydrogenase (SDH) deficiency (complex II) Others Disorders of organic acid metabolism Disorders of amino acid metabolism Canavan’s disease (aspartoacylase, ASPA deficiency) Phenylketonuria (PKU) Maple syrup disease Others Mucopolysaccharidoses Hurler disease Hunter disease Disorders of DNA repair Cockayne’s disease Disorders of neurofilaments Alexander disease Leukoencephalopathy in giant axonal neuropathy (GAN) Leukoencephalopathies with hypomyelination Pelizaeus–Merzbacher disease Salla disease (SDE) 18q-syndrome Disorders of extracellular matrix Congenital muscular dystrophy and leukoencephalopathy due to laminin 2-deficiency Miscellaneous Aicardi–Goutieres syndrome Myelinopathia centralis diffusa (MCD)/vanishing white matter disease (VWM) Megalencephalic leukoencephalopathy with subcortical cysts (MLC) Trichothiodystrophy Cerebrotendinous xanthomatosis

Table 45.2. Proton MRS pattern in hypomyelination vs. demyelination

Hypomyelination Demyelination

tNAA

tCr

Cho

mI

Lac

n ↓↓

↑ (↑)

↓ ↑

↑ ↑

– (↑)

n: normal, up arrow increased, down arrow decreased, brackets indicate changes not always seen. tNAA: total N-acetyl aspartyl compounds; tCr: total creatine; Cho: cholinecontaining compounds; mI: myo-inositol; Lac: lactate.

MR imaging (MRI) reveals WM changes which are initially subtle and progress to diffuse widespread symmetric alterations later in the course. The U-fibers are characteristically spared (van der Knaap and Valk, 1995). Reports on use of MRS in MLD are confined to few publications. Two patients aged 17 and 21 years with MLD were included in a large series of degenerative cerebral disorders investigated by 1H and 31P MRS of the brain (van der Knaap et al., 1992). Based on their MRI abnormalities of WM, they were assigned to a heterogeneous group of patients (n  10) with severe demyelination affecting almost all or all of the WM. In these patients, localized proton MRS of WM revealed significantly decreased N-acetyl aspartate (NAA)/creatine (Cr) ratios but choline (Cho)/Cr ratios within the normal range. Seven patients with MLD (four with late infantile, three with juvenile MLD) were investigated by localized proton MRS of cerebral WM and gray matter (GM) (Kruse et al., 1993). In six of seven patients duration of the disease was longer than 6 months resulting in severe leukodystrophic MRI changes. In these patients MRS revealed markedly decreased NAA/Cr ratios and strongly elevated myo-inositol (mI)/Cr ratios. These alterations were more pronounced in WM than in GM but not related to late infantile vs. juvenile subtype. A trend to increase of Cho/Cr was observed, which did not reach significance. When this observation was reported 10 years ago, the authors stated that the striking increase of mI was not detected in other leukodystrophies and might indicate a specific role of this metabolite in the pathophysiology of MLD. As experience with MRS in WM disorders grew, strong elevation of mI was found in various leukoencephalopathies including other lysosomal disorders (KD, cf. below),

MR spectroscopy in pediatric white matter disease

(a)

(b)

mI

Cho Cr NAA

Cr

4.0

(c)

3.5

2.5 2.0 3.0 Chemical shift (ppm)

1.5

1.0

1.5

1.0

(d) mI Cho Cr

4.0

3.5

3.0 2.5 2.0 Chemical shift (ppm)

Fig. 45.1 MRI and MRS of a 7-year-old boy with MLD. (a and c) T2-weighted MRI; (b and d) MRS (stimulated echo acquisition mode, STEAM, TR/TE/TM  6000/20/10 ms) of (b) right frontal WM and (d) left parieto-occipital WM. MRI shows diffuse symmetrical WM changes, U-fibers are spared. MRS reveals loss of tNAA and elevation of mI. Cr is increased in frontal and reduced in parietooccipital WM. Cho is normal in frontal and elevated in parieto-occipital WM.

Alexander disease (cf. below), and cerebral involvement of giant axonal neuropathy. There is evidence that this finding indicates the proliferation of astrocytes (Brand et al., 1993). The MRS findings in MLD therefore present non-specific features of metabolic changes secondary to demyelination, neuro-axonal loss, and astrocytic gliosis. These histopathological alterations constitute the hallmarks of MLD as well as KD and other demyelinating WM disorders. Figure 45.1 depicts MRI and MRS of a 7-year-old boy with MLD investigated in our institution using localized proton MRS. Globoid cell leukodystrophy Globoid cell leukodystrophy (GLD) or Krabbe’s disease (KD) is an autosomal-recessive disorder (MIM #245200) affecting the central nervous system (CNS)

and peripheral nervous system (Wenger et al., 2001). The biochemical basis is a deficiency of galactocerebroside -galactosidase (GALC) activity caused by various mutations in the GALC gene on chromosome 14q24.3–32.1. Galactocerebrosides are selectively highly concentrated in myelin. Deficiency of GALC results in accumulation of cerebrosides and of psychosine (galactosylsphingosine). Initially normal development of myelin is followed by a declining rate of myelination together with demyelination. Oligodendrocytes vanish and are replaced by proliferated astrocytes. Large numbers of globoid cells accumulated around blood vessels are the pathognomonic feature of KD. Their formation is probably induced by phagocytosis of accumulated cerebrosides (Friede, 1989). Three subtypes of the disorder have been delineated: the most common, infantile form presents

757

758

Folker Hanefeld, Knut Brockmann and Peter Dechent

with hypotonia followed by extreme irritability, spasticity, and rapid motor and mental deterioration leading to death within few years. Juvenile and adult subtypes have been identified, characterized by insidious visual impairment, cognitive deterioration, gait disturbance, and a more slowly protracted course (Hagberg et al., 1969; Crome et al., 1973; Thomas et al., 1984). In infantile KD, computed tomography (CT) shows hypodensity of cerebral WM along with hyperdensity of the thalami and centrum semiovale in the beginning of the disease (Choi and Enzmann, 1993). MRI detects T1-hypointensity and T2-hyperintensity in periventricular and cerebellar WM, but may be deceptively normal (Sasaki et al., 1991; Finelli et al., 1994). Altogether, neuroradiological investigations reveal a progressive and diffuse cerebral atrophy. WM alterations in juvenile and adult subtypes detectable by MRI are characteristically confined to periventricular parieto-occipital regions in the beginning of the disease. MRS has been applied to KD in different subtypes and using different methods. Twins with late infantile KD have been studied using MRS of cerebral WM only (Zarifi et al., 2001). Single-voxel stimulated echo acquisition mode (STEAM) spectroscopy was performed; loss of NAA as well as elevation of Cr, Cho, and mI were observed based on qualitative spectral evaluation, without quantification of metabolite concentrations. Proton MRS was performed in a pair of siblings with adult KD, presenting with spastic paraparesis and demyelinating neuropathy (Farina et al., 2000). With use of point resolved spectroscopy (PRESS) technique increased Cho/Cr and mI/Cr ratios in affected WM in the centrum semiovale in a 33-year-old woman and her 35-year-old brother were found. Quantitative localized proton MRS of standardized brain regions was used to assess abnormalities of cerebral metabolite concentrations in seven patients with biochemically proven KD, four with infantile, two with juvenile, and one with adult subtype (Brockmann et al., 2003c). Figure 45.2(a)–(c) shows representative MR images and spectra of cerebral WM and cortical GM of three patients with KD. MRS revealed different patterns of metabolic abnormalities in infantile (Fig. 45.2(a)), juvenile

(Fig. 45.2(b)), and adult (Fig. 45.2(c)) KD, which correlate with respective clinical courses and neuropathological features in these subtypes. In infantile KD, pronounced elevation of both mI and cholinecontaining compounds in affected WM reflected demyelination and glial proliferation. The accompanying decrease of NAA pointed to neuroaxonal loss. GM showed similar, albeit much milder alterations. In juvenile KD, MRS indicated astrocytosis with minor neuroaxonal damage in WM. In a patient with adult KD, MRS of affected WM was close to normal. These MRS data are in agreement with histopathological features of KD.

Adrenoleukodystrophy Cerebral adrenoleukodystrophy (ALD) is part of the variable phenotype of X-linked adrenoleukodystrophy (X-ALD, MIM #300100) caused by mutations in the ALD gene on chromosome Xq28 (Dubois-Dalcq et al., 1999; Moser et al., 2000). The gene encodes a peroxisomal membrane protein of the ATP-binding-cassette (ABC) transporter family (Mosser et al., 1993). Mutations result in impairment of peroxisomal degradation of very-long-chain-fatty-acids (VLCFA) with subsequent elevation of plasma levels of VLCFA. X-ALD is the most common leukodystrophy in children and adults and affects approximately 1 in 20,000 males. Organ manifestations include adrenal insufficiency and demyelination in cerebral WM. The phenotype of X-ALD is highly variable and not predictable even in patients carrying the same ALD mutation, as illustrated by diverse clinical courses in monozygotic (MZ) twins (Korenke et al., 1996). The cerebral form affects approximately 35% of hemizygotic boys with onset of clinical symptoms mainly between 4 and 12 years of age. Most affected boys die within 3–5 years after onset of symptoms (cf. Case Study 45.1). Pathological investigation reveals usually symmetrical demyelination of parieto-occipital WM. Three zones with different histological characteristics have been delineated: (i) destruction of myelin and proliferation of sudanophilic macrophages, (ii) associated inflammatory response in myelinated and demyelinated axons, and (iii) dense gliosis with loss of oligodendrocytes, myelin, and axons (Schaumburg et al., 1975).

(a) (A)

(B)

Cho mI

Cr

4.0

(C)

3.5

3.0 2.5 2.0 1.5 Chemical shift (ppm)

Cr mI Cho

(D)

WM 0.5

NAA

Cr

1.0

NAA

Cr Lac

4.0

(b) (A)

3.5

3.0 2.5 2.0 1.5 Chemical shift (ppm)

(B)

GM

1.0

NAA mI Cho

Cr

Cr WM

4.0

(C)

3.5

3.0 2.5 2.0 1.5 Chemical shift (ppm)

1.0

NAA

(D) Cr Cr

mI

Cho GM

4.0

3.5

3.0

2.5

2.0

1.5

1.0

Chemical shift (ppm) Fig. 45.2 (a) MRI and MRS of a 9-month-old boy with infantile KD: (A and B) T2-weighted MRI and MRS (STEAM, TR/TE/TM  6000/20/10 ms) of right parieto-occipital WM scaled down by a factor of two; (C and D) T1-weighted MRI and MRS of paramedian parietal GM. Lac: Lactate. See text. (b) MRI and MRS of a 6-year-old girl with juvenile KD: (A and B) T2-weighted MRI and MRS (STEAM, TR/TE/TM  6000/20/10 ms) of left parieto-occipital WM; (C and D) T1-weighted MRI and MRS of paramedian parietal GM. Lac: Lactate. See text. (c) MRI and MRS of a 47-year-old woman with adult KD: (A and B) T2-weighted MRI and MRS (STEAM, TR/TE/TM  6000/20/10 ms) of right parieto-occipital WM; (C and D) T1-weighted MRI and MRS of paramedian parietal GM. See text. (From Brockmann et al. (2003c).)

760

Folker Hanefeld, Knut Brockmann and Peter Dechent

(c) (A)

(B) NAA Cho Cr mI Cr WM

4.0

(C)

3.5

3.0 2.5 2.0 1.5 Chemical shift (ppm)

1.0

(D) NAA Cr Cr

mI Cho GM

4.0

3.5

3.0 2.5 2.0 1.5 Chemical shift (ppm)

1.0

Fig. 45.2 (cont.)

MRI abnormalities often precede the clinical symptoms and reflect the pathological alterations showing symmetrical WM lesions in parietal and occipital regions in most cases. Less frequently, frontal predominance or asymmetric patterns may occur. Inflammatory changes are recognized by contrast enhancement. MRIs are evaluated using a demerit score ranging from 0 to 34 developed by Loes (1994). The only effective therapy known to date is hematopoietic stem cell transplantation (HSCT) leading to stabilization or reversal of clinical and MRI abnormalities if performed at an early stage of the disease (Aubourg et al., 1990; Shapiro et al., 2000; Baumann et al., 2003). If carried out in an advanced stage of demyelination HSCT is not beneficial, and may even accelerate the neurological deterioration.

Proton MRS has been shown to add important data regarding the onset of demyelination and the extent of WM damage, thus providing valuable information for decision making with respect to HSCT. Various different MRS methods have been applied which have shown similar results. Single-voxel proton MRS in 11 patients with X-ALD, seven with neurological symptoms or WM lesions on MRI and four asymptomatic patients, revealed reduced NAA/Cr ratios and increased ratios of Cho, mI, glutamate (Glu) and glutamine (Gln) to Cr in WM lesions. In asymptomatic patients an increase of Cho/Cr ratio was found in normal appearing white matter (NAWM) (Tzika et al., 1993b). In 25 patients with different phenotypes of ALD multi-slice proton MR spectroscopic imaging (MRSI) revealed a reduction of NAA and an elevation of Cho

MR spectroscopy in pediatric white matter disease

in abnormal and NAWM of patients from all disease subgroups (Kruse et al., 1994). An increase of Cho was associated with active demyelination, and an elevation of lactate (Lac) was detected in severely affected occipital WM of one patient with cerebral ALD. MRSI was thus judged to be more sensitive in detection of metabolic abnormalities and incipient demyelination than MRI. This finding has been confirmed by other groups. A study of 39 male patients with X-ALD using quantitative localized proton MRS identified an increase of both Cho and mI as indicative of the onset of demyelination (Pouwels et al., 1998). Monitoring of disease progression by MRS revealed continuous reduction of NAA together with elevation of Cho and mI, accompanied by elevated Lac, reflecting neuroaxonal damage or loss, active demyelination, astrocytic proliferation, and infiltration of macrophages (Figure 45.3(a)–(c)). Single-voxel proton MRS without quantification of metabolite concentrations showed decreased ratios of NAA/Cr and NAA/Cho while Cho/Cr was increased in frontal and occipital WM of patients without treatment or after HSCT (Rajanayagam et al., 1996). These alterations were detected in abnormal WM and, to a lesser extent, as well in normal appearing regions as assessed by MRI (Rajanayagam et al., 1997). Even without any clinical and MRI evidence for cerebral involvement WM spectra are abnormal (Salvan et al., 1999). Multi-slice proton MRSI was performed in 25 patients with X-ALD and revealed reduced NAA/Cho ratios in NAWM of those patients, who showed progressive MRI alterations at follow-up. MRSI thus identified impending or beginning degeneration of WM and predicted lesion progression on MRI (Eichler et al., 2002a). In 11 patients with X-ALD and 11 healthy control subjects conventional MRI, MRSI, and diffusion tensor imaging (DTI) were compared. In patients with X-ALD, MRSI detected abnormalities in WM that had normal appearance in both MRI and DTI. MRSI was therefore considered the most sensitive technique to identify early metabolic changes in WM due to demyelination or axonal loss (Eichler et al., 2002b). In 12 patients with X-ALD localized quantitative proton MRS of frontal and occipital WM was performed before and up to 5 years after HSCT. Clinical

outcome after HSCT represented either a stable condition, a further deterioration, or death. While neurological deterioration due to active demyelination was characterized by a further increase of elevated Cho concentrations, a positive outcome after HSCT was correlated with high NAA levels in affected WM before HSCT (Wilken et al., 2003).

Mitochondrial encephalomyopathies Disorders of oxidative phosphorylation (OXPHOS) are related to diverse genetic and biochemical defects and heterogeneous clinical symptoms ranging from isolated organ dysfunction to multi-system disorder (Smeitink et al., 2001). Leukoencephalopathies constitute a small part of the spectrum of neurological diseases caused by disturbances of cellular energy generation due to defects of the mitochondrial as well as the nuclear genome (Bourgeron et al., 1995; Brown and Squier, 1996; Rahman et al., 2001). The main findings of proton MRS studies in disorders of OXPHOS are the elevation of Lac with regional variation and reduction of NAA (Detre et al., 1991). Basal ganglia and cortical GM are predominantly involved, whereas cerebral or cerebellar WM are more or less spared (Barkovich et al., 1993; Cross et al., 1993, 1994; Kraegeloh-Mann et al., 1993). Retrospective analysis of a series of 110 children with mitochondrial encephalopathies revealed eight patients with MRI features of a leukoencephalopathy (Moroni et al., 2002). Proton MRSI was performed in six of these eight patients. A decrease of NAA, Cho, and Cr together with Lac accumulation within the affected WM was observed in three patients, and a mild elevation of Lac as the sole abnormality was found in another two children. Specific MRS features in mitochondrial leukoencephalopathies are confined to complex II (succinate dehydrogenase (SDH)) deficiency to date. In three such patients localized proton MRS revealed accumulated succinate in dystrophic WM (Brockmann et al., 2002). Two sisters, children of consanguineous healthy parents, presented with a rapidly progressive loss of all motor skills at 10 months of age. A 4-year-old boy from another family showed increasing spasticity and clumsiness from the age of 20 months onwards.

761

(a)

(b)

mI Cr

4.0

Cho Cr

NAA 5Y 10M

3.5

3.0 2.5 2.0 1.5 Chemical shift (ppm)

1.0

5Y 10M

7Y 7M Lac

4.0

3.5

3.0 2.5 2.0 1.5 Chemical shift (ppm)

1.0

7Y 7M

8Y 10M

4.0

3.5

3.0 2.5 2.0 1.5 Chemical shift (ppm)

1.0

8Y 10M

9Y 5M

4.0

3.5

3.0 2.5 2.0 1.5 Chemical shift (ppm)

1.0

9Y 5M Fig. 45.3 MRI (a) and MRS (b) in right parieto-occipital WM of a boy with cerebral adrenoleukodystrophy (cALD) at 5 years, 10 months (TR/TE/TM  3000/20/30 ms, 128 accumulations, 12 ml), 7 years, 7 months (TR/TE/TM  6000/20/30 ms, 128 accumulations, 8 ml), 8 years, 10 months (TR/TE/TM  6000/20/30 ms, 64 accumulations, 8 ml), and 9 years, 5 months (TR/TE/TM  6000/20/30 ms, 64 accumulations, 4.1 ml). (c) Mean proton MR spectra (STEAM, TR/ TE/TM  6000/20/30–10 ms) representing summed data from NAWM (n  47) and affected WM (n  21) in asymptomatic ALD patients as well as affected WM in patients with cALD (n  9). Note the loss of NAA, an increase of Cho-containing compounds and finally the appearance of Lac in cALD patients compared to the asymptomatic group. (From Pouwels et al. (1998).)

MR spectroscopy in pediatric white matter disease

examination revealed the neuropathological characteristics of Leigh syndrome (LS). One additional patient with similar findings on MRS was described in a series of 25 children with confirmed diagnosis of mitochondrial encephalomyopathy by Bizzi et al. (2002).

(c) Asymptomatic NAWM Cho Cr

Cr

mI

4

Glu/ NAA Gln

3 2 Chemical shift (ppm)

1

Asymptomatic affected

4

3 2 Chemical shift (ppm)

1

cALD affected

Lac

4

3 2 Chemical shift (ppm)

1

Fig. 45.3 (cont.)

MRI revealed extensive T2-hyperintensities in central parts of supratentorial WM with predominance in frontal and occipital lobes in all patients. Localized proton MRS revealed a prominent singlet at 2.40 ppm in cerebral and cerebellar WM not present in GM or basal ganglia (Figure 45.4(a) and (b)). The signal was also found to be elevated in cerebrospinal fluid (CSF) and could be identified as originating from the two equivalent methylene groups of succinate. Subsequently, isolated deficiency of complex II was demonstrated in all patients in muscle or fibroblasts. In addition, MRS from cerebral WM revealed markedly decreased concentrations of NAA and Cr, elevated Lac, and a mild increase of mI. In contrast to WM, no succinate or Lac were detectable in cortical GM and basal ganglia. One of the sisters died at the age of 18 months. Post mortem

Canavan’s disease Canavan’s disease (spongy degeneration of the brain, Canavan–van Bogaert–Bertrand disease, MIM #271900) is a rare autosomal-recessive leukodystrophy with macrocephaly. Canavan’s disease is caused by mutations in the gene for aspartoacylase (ASPA) located on chromosome 17pter-p13 (Kaul et al., 1993). Three clinical variants have been delineated. The most frequent infantile form manifests within the first 6 months of life. Symptoms include hypotonia, irritability, macrocephaly, and feeding difficulties, followed by spasticity, blindness, choreoathetosis and relentless motor and mental deterioration leading to death within few years. A congenital form is characterized by severe hypotonia, lethargy, and rapid degeneration, while the juvenile form shows later onset, normal head circumference, and a more protracted course. Pathological findings include an abnormally enlarged brain and a spongy appearance of WM due to profuse intramyelinic vacuolization (Friede, 1989). Oligodendroglia and axonal fibers are intact, peripheral nerves are not involved. Matalon et al. (1988) discovered increased concentrations of NAA in urine and plasma of patients with spongy degeneration and demonstrated a deficiency of ASPA as the biochemical basis of the disease. CT shows diffuse symmetrical hypodensity of WM. MRI reveals a centripetal fashion of demyelination, starting in the U-fibers and progressing to periventricular regions. NAA is one of the major metabolites detectable by proton MRS. Subsequently to the first description of increased NAA signals in Canavan’s disease by Grodd et al. (1990) several reports on use of MRS in Canavan’s disease have appeared (Austin et al., 1991; Marks et al., 1991; Barker et al., 1992; Toft et al., 1993; Aydinli et al., 1998). Limitations in these studies were mainly associated with the range of identifiable metabolites (e.g. due to long-echo time (TE) rather

763

764

Folker Hanefeld, Knut Brockmann and Peter Dechent

(a) (A)

(B)

Suc

(C)

(D) 0.5 WM L

Cho Cr mI Cr 4.0

NAA

mI

Cr Cho

Suc NAA

Cereb R Lac

Lac

3.5 3.0 2.5 2.0 1.5 Chemical shift (ppm)

1.0

4.0

3.5 3.0 2.5 2.0 1.5 Chemical shift (ppm)

1.0

Fig 45.4 (a) MRI and MRS of a 12-month-old girl with complex II (SDH) deficiency: (A and B) T2- and T1-weighted MRI and (C and D) localized proton MR spectra (STEAM, TR/TE/TM  6000/20/10 ms, 64 accumulations) from (C) left-hemispheric parieto-occipital WM, scaled down by a factor of 2 and (D) right-hemispheric cerebellum (Cereb). (b) MRI and MRS of a 50-month-old boy with complex II SDH deficiency: (A and B) T2- and T1-weighted MRI and (C and D) localized proton MR spectra from (C) left-hemispheric parieto-occipital WM and (D) paramedian parietal GM, scaled down by a factor of 2. (From Brockmann et al. (2002).)

than short-TE conditions) and neurochemical data evaluation (e.g. use of short TR values and peak ratios rather than absolute metabolite concentrations). In the study by Barker et al. (1992) proton MR spectra from predominantly frontal lobe WM were recorded, which showed a near normal ratio of NAA/water, but markedly decreased levels of all other compounds (Cho, Cr, etc.). These results are best explained by the decreased cellularity and increased water content in Canavan’s, with an elevation of the intracellular NAA levels.

Wittsack et al. (1996) studied two children and determined absolute concentrations of metabolites. They were able to demonstrate that the increase of NAA/Cho ratio was not only due to a reduction of Cho-containing compounds but also due to an increase of NAA concentrations. In our institution, five patients with CD were investigated. Localized proton MRS and quantification of metabolite concentrations confirmed an absolute elevation of NAA in WM and even more pronounced in GM and basal ganglia. This finding is

MR spectroscopy in pediatric white matter disease

(b) (A)

(B)

Cho

(C) mI

(D)

Cr Suc

Cr

4.0

NAA

NAA

0.5 GM

Cr WM L Lac

3.5 3.0 2.5 2.0 1.5 Chemical shift (ppm)

1.0

Cr mI Cho

4.0

3.5 3.0 2.5 2.0 1.5 Chemical shift (ppm)

1.0

Fig 45.4 (cont.)

usually accompanied by a decrease of Cho, an increase of mI, and unchanged levels of Cr (Frahm and Hanefeld, 1997). Figure 45.5 depicts these changes in WM of a 4-year-old girl with CD. In conclusion, the detection of raised NAA concentrations in GM and WM of a macrocephalic child strongly supports the diagnosis of Canavan’s disease.

NAA

mI Cr Cr Cho

Pelizaeus–Merzbacher disease This X-linked myelin disease (MIM #312080) is caused by a mutation of the proteolipid protein (PLP) gene, which has a structural role in myelin formation and also in glial cell development. According to onset a more severe connatal type can be separated from the classical form which progresses

4.0

3.5

3.0 2.5 2.0 Chemical shift (ppm)

1.5

1.0

Fig. 45.5 Localized proton MRS (STEAM, TR/TE/TM  3000/20/30 ms, 128 accumulations) of parieto-occipital WM of a 4-year-old girl with Canavan’s disease. NAA is markedly increased.

765

766

Folker Hanefeld, Knut Brockmann and Peter Dechent

at a variable pace. Affected children present with muscular hypotonia and rotatory nystagmus during the first year of life, later dystonic movements, ataxia, optic atrophy, and pyramidal symptoms develop. Neuropathology describes a paucity or even absence of myelin with preservation of islets of almost normal myelin around blood vessels (tigroid pattern). Axons are often spared and no active demyelination (destruction of myelin) is described. The severe connatal cases show almost no myelin in the brain and spinal cord and an absence of oligodendroglial cells (Koeppen and Robitaille, 2002). The GM is also affected with lack of myelin and loss of neurons in later stages of the disease. MRI shows diffuse hyperintense WM signal in T2-weighted images. In T1-weighted images the contrast between GM and WM is missing. MRS studies have been performed in small series or single cases, mainly of the classical form. Most investigators relied on peak ratios rather than absolute concentrations of metabolites. This is not without problems in a disorder of hypo- as well as demyelination and a slowly progressive demyelination. Grodd et al. (1991) comparing relative peak intensities report a reduction of NAA and an increase of Cho in one case of early onset, while a late onset patient showed normal NAA and decreased Cho concentrations. In an abstract published in 1997 Pouwels et al. report their findings in six Pelizaeus–Merzbacher disease (PMD) patients aged 1.5–21 years including one follow-up case. Fully relaxed, short TE spectra were recorded with STEAM. Absolute concentrations of metabolites were determined with the linear combination (LCM) model. Metabolite concentrations were normal in GM and basal ganglia. In WM a low concentration of Cho, but high levels of Cr, NAA, mI, and Glu was recorded (Figure 45.6). The content of Cr, NAA, and Cho in WM of PMD patients was much more similar to GM than to normal WM. Altogether the metabolite profile seen on MRS in PMD resembles hypomyelination and is completely different from demyelinating disease (Table 45.2).

Nezu et al. (1998) studied four boys aged 10–17 years with classical PMD. Follow-up after 5 years was performed in three patients. Proton MRS revealed no abnormal peaks. Peak area ratios of NAA/Cr and Cho/Cr were within normal limits. Spalice et al. (2000) performed proton MRS in two connatal cases of PMD. A decrease of Cho was noticed. Bonavita et al. (2001) reported a significant decrease of brain NAA in nine patients with PLP mutations. In a most recent publication from Japan by Takanashi et al. (2002) five boys with PMD and PLP1 duplications were studied. They report an increase of absolute concentrations of NAA (16%), Cr (43%) and mI (31%) in PMD patients as compared to agematched controls. Cho concentrations were within the same limit in both groups. An axonal involvement secondary to demyelination or increased cell population of oligodendrocyte progenitors are discussed. The elevated Cr and mI concentrations are thought to reflect astrocytic gliosis. Summarizing the reports on MRS findings of PMD it becomes evident, that they depend greatly on the methods used and the type as well as the stage of the disease. However, PMD is probably a key disorder for characterizing hypo- and demyelination in vivo by its metabolic profile as demonstrated by MRS.

Alexander disease Alexander disease, originally described as “Progressive fibrinoid degeneration of fibrillary astrocytes associated with mental retardation in a hydrocephalic infant” by Alexander (1949), is a rare, mostly sporadic disorder of the CNS) (MIM #203450) (Borrett and Becker, 1985). The most common infantile form presents as a progressive leukodystrophy with developmental delay, macrocephaly, seizures, spasticity, ataxia, and rapid deterioration within few years. Juvenile and autosomal dominant adult subtypes have been identified (Russo et al., 1976; Pridmore et al., 1993; Schwankhaus et al., 1995). The characteristic pattern of neuroradiological features (van der Knaap et al., 2001) comprises extensive

MR spectroscopy in pediatric white matter disease

(a)

NAA Pat_933 (1Y 7M): PMD WM

GM Cr

Cho mI

mI Cho Cr Glu/Gln NAA

4.0

3.5

3.0 2.5 2.0 Chemical shift (ppm)

1.5

1.0

4.0

3.5

3.0 2.5 2.0 Chemical shift (ppm)

1.5

1.0

1.5

1.0

(b) Pat_933 (1Y 7M): PMD Basal ganglia

4.0

3.5

3.0 2.5 2.0 Chemical shift (ppm)

Cerebellum central

1.5

1.0

4.0

3.5

3.0 2.5 2.0 Chemical shift (ppm)

Fig. 45.6 (a and b) Localized proton MRS (STEAM, TR/TE/TM  3000/20/30 ms, 128 accumulations) of GM and WM, basal ganglia and central cerebellum of a 19-month-old boy with PMD. See text.

signal changes of cerebral WM with frontal predominance, a periventricular rim with high signal on T1-weighted images and low signal on T2-weighted images, abnormalities of basal ganglia, thalami, and brain stem, as well as contrast agent enhancement of particular GM and WM structures.

Neuropathological examination reveals astrocytosis, intracellular deposits in astrocytes, and demyelination as the main features. The morphological hallmark is the presence of large numbers of Rosenthal fibers predominantly in the subpial, subependymal, and perivascular regions of the CNS (Friede 1964; Cole et al., 1979).

767

768

Folker Hanefeld, Knut Brockmann and Peter Dechent

These intracytoplasmic inclusions contain the intermediate filament protein glial acidic fibrillary protein (GFAP) and small heat-shock proteins (Iwaki et al., 1989). GFAP is highly specific for cells of astroglial lineage and coded by GFAP on 17q21 (Reeves et al., 1989). Heterozygous de novo mutations in GFAP have been identified in patients with histopathologically proven Alexander disease (Brenner et al., 2001; Rodriguez et al., 2001). The connections between GFAP mutations and severe neurodegeneration in Alexander disease are not completely understood. Widespread destruction of WM with rostro-caudal gradient, even resulting in cavitation, has accounted for grouping Alexander disease among the leukodystrophies. Three studies of MRS in Alexander disease have been published so far. Two cases of infantile Alexander disease in a review article by Grodd et al. (1991) demonstrated reduced NAA and normal Cho in frontal WM and close to normal NAA but decreased Cho in an occipital region. Elevated Lac was found in the frontal lobe of a child with early onset of the disease. Takanashi et al. (1998) reported a 17-year-old male with a mild course of infantile Alexander disease which led to a low NAA/Cr ratio and the presence of Lac in parietal WM. Three boys and one girl with heterozygous de novo mutations in GFAP (Meins et al., 2002) and clinical and MRI features of infantile Alexander disease were investigated using localized quantitative proton MRS (Brockmann et al., 2003c). Their ages ranged from 1.6 to 6.7 years at MRS investigation. MRI and proton MR spectra of a 6-year-old girl with Alexander disease are depicted in Figure 45.7 for frontal WM, parietal GM, basal ganglia, and cerebellum. They demonstrate representative metabolic abnormalities observed in all patients. The main findings are the detection of strongly elevated mI in WM, basal ganglia, and cortical GM and a reduction of NAA, most pronounced in WM and accompanied by the accumulation of Lac. Affected cerebral WM as identified by MRI revealed a strong elevation of mI and a distinct decline of NAA. These alterations were accompanied by the accumulation of Lac in all cases. The findings in cerebral WM were supported by similar observations in cerebellar WM obtained in one patient at the age of 6.7 years (Figure 45.7).

In general, parietal GM showed similar though milder metabolic disturbances than WM. Basal ganglia exhibited an NAA reduction and a striking elevation of both mI (similar to WM) and Cho (similar to GM). Elevated Lac in basal ganglia was detected in one patient only (3.1 mM). This pattern of altered metabolite levels may be assigned to three major pathophysiological processes, i.e. (i) a generalized glial proliferation in all regions investigated as well as (ii) active demyelination and (iii) neuroaxonal degeneration of predominantly WM. Marked elevation of mI in cerebral WM and basal ganglia as well as, though less pronounced, in cortical GM reflects the generalized astrocytosis in Alexander disease. The observation of increased Cho in basal ganglia (all patients), GM (three patients), and WM (two patients) must be ascribed to ongoing demyelinating processes which take place in parallel to astrocytic proliferation. The generalized reduction of NAA points to a damage or even loss of vital neuroaxonal tissue predominantly in cerebral and cerebellar WM but also in cortical and subcortical GM.

Myelinopathia centralis diffusa/vanishing white matter disease In 1993, Hanefeld et al. reported a “diffuse WM disease in three children with unique features on MRI and MRS” and later proposed the term myelinopathia centralis diffusa (MCD) for it (Hanefeld et al.,1996). Subsequent publications by Schiffmann et al. (1994) and van der Knaap et al. (1997) confirmed this identification of a separate, genetically determined leukoencephalopathy, giving it the terms vanishing white matter disease (VWM) (MIM #603896) and childhood ataxia with central hypomyelination (CACH). Recently it was shown that mutation of the five subunits of the eukaryotic translation initiation factor eIF2B can cause leukoencephalopathy with VWM/ MCD (Leegwater et al., 2001; van der Knaap et al., 2002). MCD/VWM begins usually in late infancy or early childhood but acute fatal infantile forms have also been described (Francalanci et al., 2001).

(A)

(E) Ins WM

Cho

Lac

Cr

4.0

(B)

3.5

Cr

(F)

1.0

2.5 1.5 2.0 3.0 Chemical shift (ppm) NAA

Ins Cho

Cr

4.0 (C)

(G)

3.5

Ins

GM

1.0

2.5 1.5 2.0 3.0 Chemical shift (ppm) Cr Cho NAA

BG

Cr

Lac

4.0 (D)

3.5

2.5 1.5 2.0 3.0 Chemical shift (ppm)

1.0

(H)

Cr

4.0

Ins

3.5

Cho Cr

Cereb

2.5 1.5 2.0 3.0 Chemical shift (ppm)

1.0

Fig. 45.7 Cerebral MRI and proton MRS of the brain of a 6-year-old girl with severe form of Alexander disease. (A and E) T2-weighted MRI and proton MRS (STEAM, TR/TE/TM  6000/20/10 ms) of right frontal WM, (B and F) T1-weighted MRI and proton MRS of paramedian parietal GM, (C and G) T2-weighted MRI and proton MRS of basal ganglia, and (D and H) T1-weighted MRI and proton MRS of left-hemispheric cerebellar WM. (From Brockmann et al. (2003a).)

770

Folker Hanefeld, Knut Brockmann and Peter Dechent

Acute ataxia, frequently associated with mild trauma or infection is followed by spasticity and optic atrophy while mental functions remain relatively preserved over a prolonged period. Neuropathology shows severe destruction of WM in cerebral hemispheres with cobweb like structures, spongy degeneration and cavitations (van der Knaap et al., 1997; Rodriguez et al., 1999). There are conflicting reports about increased or reduced density of oligodendrocytes. In a personal early onset fatal infantile case typical signs of apoptosis of oligodendrocytes could be demonstrated within areas of an active demyelination (Brück et al., 2001). Cerebellar WM is also affected as well as the central tegmental tract. The MRI of MCD/VWM reflects the abnormalities described by neuropathologists. Both hemispheres show signal intensities close to CSF on T1-weighted, T2-weighted, and FLAIR images. A fine meshwork of strands and cavitations can be seen in the advanced stages. The involvement of the central tegmental tract is clearly visible. The unique MRS characteristics in the original description presented as almost total loss of all metabolites in affected WM (Figure 45.8). These findings have been confirmed in the studies by van der Knaap et al. (1997, 1998), Schiffmann et al. (1994) and Tedeschi et al. (1995). A prospective study of 14 personal cases of MCD/VWM showed a progressing decrease of NAA in affected WM. Cr was decreased at a stable level. Concentrations of choline-containing compounds were within normal limits in more than 50% of all patients. The mI showed no constant pattern. It was increased during the initial stage in one patient and decreased in the majority as the disease progressed. Lac was also increased in affected WM (Figure 45.9). MR spectra of GM revealed a mild but definite decrease of NAA, which reached significant levels in four patients when investigated 14–22 years after onset of the disease. Cr showed also a mild decrease during disease progress, but remained within normal limits. Choline-containing compounds and mI remained normal throughout. The interpretation of MRS findings in MCD/VWM is difficult. Clearly the metabolic profile and its evolution indicate an early involvement of axons

Lac

Glc

Case 1

4.0

3.5

3.0 2.5 2.0 1.5 Chemical shift (ppm)

1.0

0.5

Lac

Glc

Case 2

4.0

3.5

3.0 2.5 2.0 1.5 Chemical shift (ppm)

1.0

0.5

NAA

Cho Cr

Cr mI

Control

4.0

3.5

3.0 2.5 2.0 1.5 Chemical shift (ppm)

1.0

0.5

Fig. 45.8 Localized proton MRS (STEAM, TR/TE/TM  3000/20/30 ms, 128 accumulations, 8 ml) of parietal WM of a 5-year-old girl (case 1) and a 10-year-old girl (case 2) with MCD/leukoencephalopathy with VWM (MCD/VWM) and a 9-year-old control. A loss of all major metabolites is obvious. Only Lac and glucose (Glc) are detectable. MRI of corresponding regions showed a signal intensity almost identical to that of CSF.

and myelin. It is different from the pattern seen in hypomyelination. It is also not typical for acute demyelination, which is characterized by a significant increase of Cho and mI concentrations. One can only speculate whether a so far unknown anomaly caused by the mutated eIF2B factor on chromosome 3q27 and 14q24 may damage both myelin and axons. Active demyelination will then be triggered by insults like trauma, infections or fever.

MR spectroscopy in pediatric white matter disease

(a)

tNAA Cr Ins

4.0

Cho

3.5

3.0

2.5

2.0

1.5

1.0

Chemical shift (ppm) Cho Cr

(b)

4.0

3.5

3.0

tNAA

2.5

2.0

Lac

1.5

1.0

Chemical shift (ppm) (c)

Glc Lac

4.0

3.5

3.0

2.5

2.0

1.5

1.0

Chemical shift (ppm) Fig. 45.9 Localized proton MR spectra of WM of a boy with MCD/VWM at (a) 3 years, 3 months (11 months after onset of symptoms), (b) 6 years, and (c) 10 years, 10 months of age (8 years after onset). A progressive diffuse signal change of WM on MRI (not shown) is paralleled by a decrease of concentrations of all metabolites in WM (whereas GM spectra (not shown) remain normal. (From Hanefeld et al. (1993).)

Megalencephalic leukoencephalopathy with subcortical cysts Megalencephalic leukoencephalopathy with subcortical cysts (MLC, MIM #604004) (Leegwater et al., 2001) has been delineated during the last decade as a novel entity among the hitherto unclassified leukoencephalopathies in childhood. It had been originally called “leukoencephalopathy with swelling and a discrepantly mild clinical course” (van der Knaap et al., 1995a). Mutations in a gene renamed MLC1 on 22qter have been identified as the cause of MLC (Leegwater et al., 2001).

Macrocephaly developing within the first year of life, gradually increasing ataxia and spasticity, and remarkably well preserved cognitive functions constitute the clinical pattern characteristic for this disorder (van der Knaap et al., 1995a; Goutieres et al., 1996; Singhal et al., 1996; Topcu et al., 1998; Ben-Zeev et al., 2001). MRI shows extensive signal changes of hemispherical cerebral WM early in the course and subcortical cysts predominantly in the temporal and parietal region (van der Knaap et al., 1995b). Histopathology revealed a cavitating spongiform leukoencephalopathy with vacuoles in the outermost lamellae of myelin sheaths (van der Knaap et al., 1996). Subcortical WM was altered by intense fibrillary astrogliosis, myelinated axons were well preserved. Cortical neuronal structure was normal. Since 1995, more than 70 patients with this disorder have been reported, predominantly infants and children. MRS investigations are confined to few studies. The original report delineating this disorder (van der Knaap et al., 1995a) included proton MRS findings in abnormal WM of four patients aged 3.5–20 years. In all patients the PRESS sequence and in addition in three patients the STEAM sequence was used. With increasing age a decreasing NAA/Cr ratio was found. Cho/Cr was elevated, mI/Cr was normal in all patients. Mejaski-Bosnjak et al. (1997) reported MRS features of an 8-year-old girl with MLC. The child had a severe clinical course leading to wheelchairdependency at 5 years of age. MRI revealed multiple large cystic lesions in subcortical WM frontally, temporally, and parietally. Localized MRS of affected WM (free of cysts) showed a loss of all metabolites, pointing to a complete disintegration of neuroaxonal and glial tissue. Figure 45.10(a) depicts MRI and MRS of a 9-yearold boy with a much milder clinical course who was investigated in our institution. De Stefano et al. (2001) used proton MRSI and found decreased ratios of NAA/Cr and normal Cho/Cr ratios in WM of a 28-year-old male. MRS was used in a mentally normal 37-year-old woman (Figure 45.10(b)) with MLC who was investigated for increasing gait disturbance caused by longstanding spasticity and ataxia (Brockmann et al., 2003b). MRI showed marked enlargement of inner

771

772

Folker Hanefeld, Knut Brockmann and Peter Dechent

(a) (A)

(B) NAA Cho Cr

mI

4.0

3.5

3.0

2.5

2.0

1.5

1.0

Chemical shift (ppm)

(b) (A)

(B) WM RPO

mI Cho Cr

4

(C)

NAA

3 2 Chemical shift (ppm)

(D) GM PMP

1

NAA Cr Cho

mI

4

3 2 Chemical shift (ppm)

1

Fig. 45.10 (a) MRI and MRS of a 9-year-old boy with MLC. MRI shows diffuse WM abnormalities. MRS reveals a mild decrease of NAA and Cr and an increase of mI. Choline-containing compounds are within normal limits. (b) MRI and MRS of a 37-year-old woman with MLC. Axial T2-weighted (A) and T1-weighted (C) MRI shows diffuse signal changes of WM. Localized proton spectra (STEAM, TR/TE  6000/20 ms) reveal marked reduction of NAA, Cr, and Cho in WM (B) and solely mild decrease of NAA in parietal GM (D). (From Brockmann et al. (2003b).)

MR spectroscopy in pediatric white matter disease

and outer CSF spaces and widespread bilateral signal changes with diffuse T2-hyperintensity of cerebral WM including the U-fibers. Numerous subcortical cysts were visible in anterior-temporal and parietal regions. Quantitative localized proton MRS of WM revealed marked reduction of NAA, Cr, and Cho with normal values for mI, consistent with axonal loss and astrocytic proliferation (Figure 45.10). A spectrum of parietal GM showed only mild reduction of NAA, and a basal ganglia spectrum was normal. MRS findings obtained in MLC underline the MRI evidence of a WM disorder without cortical involvement. Initial MRS investigations performed early in the course may be almost normal. In patients with more longstanding or severe clinical symptoms a decrease of NAA consistent with axonal damage or loss and elevation of mI in line with astrocytic gliosis are prominent findings. Cho levels are normal or reduced in most investigations. Accordingly, histopathological studies showed no signs of demyelination.

Other metabolic diseases involving white matter MRS findings in other rare metabolic diseases which involve WM have also been reported. These include cerebrotendinous xanthomatosis, Sjogren–Larsson syndrome and leukodystrophy associated with ovarian dysgenesis and are described in Chapter 25.

Table 45.3. Proton MRS findings in pediatric WM diseases NAA

Cr

Cho mI

Lac

MLD GLD ALD SDH-Def. Canavan PMD Alexander MCD/VWM

↓↓ ↓↓ ↓↓ ↓↓ ↑↑ (↑) ↓↓ ↓↓

↑ ↑ (↓) ↓ n ↑ ↓ ↓↓

↑ ↑ ↑ n ↓ ↓ ↑ ↓↓

↑↑ ↑↑ ↑↑ n ↑ ↑ ↑↑ ↓↓

– (↑) (↑) ↑↑ – – ↑ –

MLC

↓↓

↓↓

↓↓

n

–

Others

Suc (2.4 ppm)

n: normal, SDH-Def.: SDH deficiency, Canavan: Canavan’s disease, Alexander: Alexander disease, MCD/VWM: Myelinopathia centralis diffusa/Leukoencephalopathy with vanishing white matter.

Addendum: Recently, a deficiency of ribose-5-phosphate isomerase (RPI) has been identified as the cause of a slowly progressive leukoencephalopathy in a boy. Proton MRS of white matter showed a specific pattern characterized by highly elevated peaks in the sugar and polyol region, and increased concentrations of the pentitols arabitol and ribitol were also found in body fluids. Subsequently, deficient activity of RPI and mutations in the gene encoding for RPI were detected (van der Knaap et al., 1999; Huck et al., 2004).

REFERENCES

Conclusion MRS is a powerful tool for the classification and follow-up of WM disorders in childhood. From data available so far the metabolic profiles observed can be classified as (i) specific in Canavan’s disease and SDH deficiency, (ii) characteristic in mitochondrial leukoencephalopathies, MCD/VWM, and PMD, and (iii) unspecific in the majority of demyelinating disorders (MLD, KD, ALD, others) (Table 45.3). MRS thus provides important information for the characterization of unclassified WM disorders, follow-up studies, and the monitoring of therapeutic interventions.

Alexander WS. 1949. Progressive fibrinoid degeneration of fibrillary astrocytes associated with mental retardation in a hydrocephalic infant. Brain 72: 373–381. Aubourg P, Blanche S, Jambaque I, Rocchiccioli F, Kalifa G, Naud-Saudreau C, Rolland MO, Debre M, Chaussain JL, Griscelli C, Fischer A, Bougneres PF. 1990. Reversal of early neurologic and neuroradiologic manifestations of X-linked adrenoleukodystrophy by bone marrow transplantation. New Engl J Med 322: 1860–1866. Austin SJ, Connelly A, Gadian DG, Benton JS, Brett EM. 1991. Localized 1H NMR spectroscopy in Canavan’s disease: a report of two cases. Magn Reson Med 19: 439–445. Aydinli N, Caliskan M, Calay M, Ozmen M. 1998. Use of localized proton nuclear magnetic resonance spectroscopy in Canavan’s disease. Turk J Pediatr 40: 549–557.

773

774

Folker Hanefeld, Knut Brockmann and Peter Dechent

Barker PB, Bryan RN, Kumar AJ, Naidu S. 1992. Proton NMR spectroscopy of Canavan’s disease. Neuropediatrics 23: 263–267. Barkovich AJ, Good WV, Koch TK, Berg BO. 1993. Mitochondrial disorders: analysis of their clinical and imaging characteristics. Am J Neuroradiol 14: 1119–1137. Baumann M, Korenke GC, Weddige-Diedrichs A, Wilichowski E, Hunneman DH, Wilken B, Brockmann K, Klingebiel T, Niethammer D, Kuhl J, Ebell W, Hanefeld F. 2003. Haematopoietic stem cell transplantation in 12 patients with cerebral X-linked adrenoleukodystrophy. Eur J Pediatr 162: 6–14. Ben-Zeev B, Gross V, Kushnir T, Shalev R, Hoffman C, Shinar Y, Pras E, Brand N. 2001. Vacuolating megalencephalic leukoencephalopathy in 12 Israeli patients. J Child Neurol 16: 93–99. Betts TA, Smith WT, Kelly RE. 1968. Adult metachromatic leukodystrophy (sulphatide lipidosis) simulating acute schizophrenia: report of a case. Neurology 18: 1140–1142. Bizzi A, Danesi U, Moroni I, Erbetta A, Bugiani M, Uziel G, Savoiardo M. 2002. Incidence of cerebral lactic acidosis with mitochondrial encephalomyopathy (Abstract). Proc Intl Soc Magn Reson Med 10: 984. Bonavita S, Schiffmann R, Moore DF, Frei K, Choi B, Patronas MDN, Virta A, Boespflug-Tanguy O, Tedeschi G. 2001. Evidence for neuroaxonal injury in patients with proteolipid protein gene mutations. Neurology 56: 785–788. Borrett D, Becker LE. 1985. Alexander’s disease. A disease of astrocytes. Brain 108: 367–385. Bourgeron T, Rustin P, Chretien D, Birch-Machin M, Bourgeois M, Viegas-Pequignot E, Munnich A, Rotig A. 1995. Mutation of a nuclear succinate dehydrogenase gene results in mitochondrial respiratory chain deficiency. Nat Genet 11: 144–149. Brand A, Richter-Landsberg C, Leibfritz D. 1993. Multinuclear NMR studies on the energy metabolism of glial and neuronal cells. Dev Neurosci 15: 289–298. Brenner M, Johnson AB, Boespflug-Tanguy O, Rodriguez D, Goldman JE, Messing A. 2001. Mutations in GFAP, encoding glial fibrillary acidic protein, are associated with Alexander disease. Nat Genet 27: 117–120. Brockmann K, Bjornstad A, Dechent P, Korenke CG, Smeitink J, Trijbels JM, Athanassopoulos S, Villagran R, Skjeldal OH, Wilichowski E, Frahm J, Hanefeld F. 2002. Succinate in dystrophic white matter: a proton magnetic resonance spectroscopy finding characteristic for complex II deficiency. Ann Neurol 52: 38–46. Brockmann K, Dechent P, Meins M, Haupt M, Sperner J, Stephani U, Frahm J, Hanefeld F. 2003a. Cerebral proton magnetic resonance spectroscopy in infantile Alexander disease. J Neurol 250: 300–306.

Brockmann K, Finsterbusch J, Terwey B, Frahm J, Hanefeld F. 2003b. Megalencephalic leukoencephalopathy with subcortical cysts in an adult: quantitative proton MR spectroscopy and diffusion tensor MRI. Neuroradiology 45: 137–142. Brockmann K, Dechent P, Wilken B, Rusch O, Frahm J, Hanefeld F. 2003c. Proton MRS profile of cerebral metabolic abnormalities in Krabbe disease. Neurology 60: 819–825. Brown GK, Squier MV. 1996. Neuropathology and pathogenesis of mitochondrial diseases. J Inherit Metab Dis 19: 553–572. Brück W, Herms J, Brockmann K, Schulz-Schaeffer W, Hanefeld F. 2001. Myelinopathia centralis diffusa (vanishing white matter disease): evidence of apoptotic oligodendrocyte degeneration in early lesion development. Ann Neurol 50: 532–536. Choi S, Enzmann DR. 1993. Infantile Krabbe disease: complementary CT and MR findings. Am J Neuroradiol 14: 1164–1166. Cole G, De Villiers F, Proctor NS, Freiman I, Bill P. 1979. Alexander’s disease: case report including histopathological and electron microscopic features. J Neurol Neurosurg Psychiatr 42: 619–624. Crome L, Hanefeld F, Patrick D, Wilson J. 1973. Late onset globoid cell leucodystrophy. Brain 96: 841–848. Cross JH, Connelly A, Gadian DG, Kendall BE, Brown GK, Brown RM, Leonard JV. 1994. Clinical diversity of pyruvate dehydrogenase deficiency. Pediatr Neurol 10: 276–283. Cross JH, Gadian DG, Connelly A, Leonard JV. 1993. Proton magnetic resonance spectroscopy studies in lactic acidosis and mitochondrial disorders. J Inherit Metab Dis 16: 800–811. De Stefano N, Balestri P, Dotti MT, Grosso S, Mortilla M, Morgese G, Federico A. 2001. Severe metabolic abnormalities in the white matter of patients with vacuolating megalencephalic leukoencephalopathy with subcortical cysts. A proton MR spectroscopic imaging study. J Neurol 248: 403–409. Detre JA, Wang ZY, Bogdan AR, Gusnard DA, Bay CA, Bingham PM, Zimmerman RA. 1991. Regional variation in brain lactate in Leigh syndrome by localized 1H magnetic resonance spectroscopy. Ann Neurol 29: 218–221. Dubois-Dalcq M, Feigenbaum V, Aubourg P. 1999. The neurobiology of X-linked adrenoleukodystrophy, a demyelinating peroxisomal disorder. Trends Neurosci 22: 4–12. Eichler FS, Barker PB, Cox C, Edwin D, Ulug AM, Moser HW, Raymond GV. 2002a. Proton MR spectroscopic imaging predicts lesion progression on MRI in X-linked adrenoleukodystrophy. Neurology 58: 901–907. Eichler FS, Itoh R, Barker PB, Mori S, Garrett ES, van Zijl PC, Moser HW, Raymond GV, Melhem ER. 2002b. Proton MR spectroscopic and diffusion tensor brain MR imaging in X-linked adrenoleukodystrophy: initial experience. Radiology 225: 245–252.

MR spectroscopy in pediatric white matter disease

Farina L, Bizzi A, Finocchiaro G, Pareyson D, Sghirlanzoni A, Bertagnolio B, Savoiardo M, Naidu S, Singhal BS, Wenger DA. 2000. MR imaging and proton MR spectroscopy in adult Krabbe disease. Am J Neuroradiol 21: 1478–1482. Finelli DA, Tarr RW, Sawyer RN, Horwitz SJ. 1994. Deceptively normal MR in early infantile Krabbe disease. Am J Neuroradiol 15: 167–171. Frahm J, Hanefeld F. 1997. Localized proton magnetic resonance spectroscopy of brain disorders in childhood. In Magnetic Resonance Spectroscopy and Imaging in Neurochemistry (Ed. Bachelard HS). Plenum, New York, pp. 329–402. Francalanci P, Eymard-Pierre E, Dionisi-Vici C, Boldrini R, Piemonte F, Virgili R, Fariello G, Bosman C, Santorelli FM, Boespflug-Tanguy O, Bertini E. 2001. Fatal infantile leukodystrophy: a severe variant of CACH/VWM syndrome, allelic to chromosome 3q27. Neurology 57: 265–270. Friede RL. 1964. Alexander’s disease. Arch Neurol 11: 414–422. Friede RL. 1989. Developmental Neuropathology, 2nd edn. Springer, Berlin/New York. Goutieres F, Boulloche J, Bourgeois M, Aicardi J. 1996. Leukoencephalopathy, megalencephaly, and mild clinical course. A recently individualized familial leukodystrophy. Report on five new cases. J Child Neurol 11: 439–444. Grodd W, Krageloh-Mann I, Petersen D, Trefz FK, Harzer K. 1990. In vivo assessment of N-acetylaspartate in brain in spongy degeneration (Canavan’s disease) by proton spectroscopy. Lancet 336: 437–438. Grodd W, Krageloh-Mann I, Klose U, Sauter R. 1991. Metabolic and destructive brain disorders in children: findings with localized proton MR spectroscopy. Radiology 181: 173–181. Hagberg B, Kollberg H, Sourander P, Akesson HO. 1969. Infantile globoid cell leucodystrophy (Krabbe’s disease). A clinical and genetic study of 32 Swedish cases 1953–1967. Neuropadiatrie 1: 74–88. Hanefeld F, Brockmann K, Christen HJ, Pouwels P, Frahm J. 1996. Myelinopathia centralis diffusa – eine neue autosomal rezessiv vererbte Leukenzephalopathie. Monatsschr Kinderheilkunde 144: S26. Hanefeld F, Holzbach U, Kruse B, Wilichowski E, Christen HJ, Frahm J. 1993. Diffuse white matter disease in three children: an encephalopathy with unique features on magnetic resonance imaging and proton magnetic resonance spectroscopy. Neuropediatrics 24: 244–248. Huck JH, Verhoeven NM, Struys EA, Salomons GS, Jakobs C, van der Knaap MS. 2004. Ribose-5-phosphate isomerase deficiency: new inborn error in the pentose phosphate pathway associated with a slowly progressive leukoencephalopathy. Am J Hum Genet 74: 745–751. Iwaki T, Kume-Iwaki A, Liem RK, Goldman JE. 1989. Alpha Bcrystalline is expressed in non-lenticular tissues and accumulates in Alexander’s disease brain. Cell 57: 71–78.

Kaul R, Gao GP, Balamurugan K, Matalon R. 1993. Cloning of the human aspartoacylase cDNA and a common missense mutation in Canavan disease. Nat Genet 5: 118–123. Koeppen AH, Robitaille Y. 2002. Pelizaeus–Merzbacher disease. J Neuropathol Exp Neurol 61: 747–759. Korenke GC, Fuchs S, Krasemann E, Doerr HG, Wilichowski E, Hunneman DH, Hanefeld F. 1996. Cerebral adrenoleukodystrophy (ALD) in only one of monozygotic twins with an identical ALD genotype. Ann Neurol 40: 254–257. Krageloh-Mann I, Grodd W, Schoning M, Marquard K, Nagele T, Ruitenbeek W. 1993. Proton spectroscopy in five patients with Leigh’s disease and mitochondrial enzyme deficiency. Dev Med Child Neurol 35: 769–776. Kruse B, Barker PB, van Zijl PC, Duyn JH, Moonen CT, Moser HW. 1994. Multislice proton magnetic resonance spectroscopic imaging in X-linked adrenoleukodystrophy. Ann Neurol 36: 595–608. Kruse B, Hanefeld F, Christen HJ, Bruhn H, Michaelis T, Hanicke W, Frahm J. 1993. Alterations of brain metabolites in metachromatic leukodystrophy as detected by localized proton magnetic resonance spectroscopy in vivo. J Neurol 241: 68–74. Leegwater PA, Vermeulen G, Konst AA, Naidu S, Mulders J, Visser A, Kersbergen P, Mobach D, Fonds D, van Berkel CG, Lemmers RJ, Frants RR, Oudejans CB, Schutgens RB, Pronk JC, van der Knaap MS. 2001a. Subunits of the translation initiation factor eIF2B are mutant in leukoencephalopathy with vanishing white matter. Nat Genet 29: 383–388. Leegwater PA, Yuan BO, van der Steen J, Mulders J, Konst AA, Boor PK, Mejaski-Bosnjak V, van der Maarel SM, Frants RR, Oudejans CB, Schutgens RB, Pronk JC, van der Knaap MS. 2001b. Mutations of MLC1 (KIAA0027), encoding a putative membrane protein, cause megalencephalic leukoencephalopathy with subcortical cysts. Am J Hum Genet 68: 831–838. Loes DJ, Hite S, Moser H, Stillman AE, Shapiro E, Lockman L, Latchaw RE, Krivit W. 1994. Adrenoleukodystrophy: a scoring method for brain MR observations. Am J Neuroradiol 15: 1761–1766. Marks HG, Caro PA, Wang ZY, Detre JA, Bogdan AR, Gusnard DA, Zimmerman RA. 1991. Use of computed tomography, magnetic resonance imaging, and localized 1H magnetic resonance spectroscopy in Canavan’s disease: a case report. Ann Neurol 30: 106–110. Matalon R, Michals K, Sebesta D, Deanching M, Gashkoff P, Casanova J. 1988. Aspartoacylase deficiency and N-acetylaspartic aciduria in patients with Canavan disease. Am J Med Genet 29: 463–471. Meins M, Brockmann K, Yadav S, Haupt M, Sperner J, Stephani U, Hanefeld F. 2002. Infantile Alexander disease: a GFAP mutation in monozygotic twins and novel mutations in two other patients. Neuropediatrics 33: 194–198.

775

776

Folker Hanefeld, Knut Brockmann and Peter Dechent

Mejaski-Bosnjak V, Besenski N, Brockmann K, Pouwels PJ, Frahm J, Hanefeld FA. 1997. Cystic leukoencephalopathy in a megalencephalic child: clinical and magnetic resonance imaging/magnetic resonance spectroscopy findings. Pediatr Neurol 16: 347–350. Moroni I, Bugiani M, Bizzi A, Castelli G, Lamantea E, Uziel G. 2002. Cerebral white matter involvement in children with mitochondrial encephalopathies. Neuropediatrics 33: 79–85. Moser HW, Smith KD, Watkins PA, Powers J, Moser AB. 2000. X-linked adrenoleukodystrophy. In The Metabolic and Molecular Bases of Inherited Disease (Eds., Scriver CR, Beaudet AL, Sly WS, Valle D). McGraw-Hill, New York, pp. 3257–3301. Mosser J, Douar AM, Sarde CO, Kioschis P, Feil R, Moser H, Poustka AM, Mandel JL, Aubourg P. 1993. Putative X-linked adrenoleukodystrophy gene shares unexpected homology with ABC transporters. Nature 361: 726–730. Nezu A, Kimura S, Takeshita S, Osaka H, Kimura K, Inoue K. 1998. An MRI and MRS study of Pelizaeus–Merzbacher disease. Pediatr Neurol 18: 334–337. Polten A, Fluharty AL, Fluharty CB, Kappler J, von Figura K, Gieselmann V. 1991. Molecular basis of different forms of metachromatic leukodystrophy. New Eng J Med 324: 18–22. Pouwels P, Hanefeld F, Frahm J. 1997. Proton MRS in Pelizaeus–Merzbacher disease. Neuropediatrics 28: 355–356. Pouwels PJ, Kruse B, Korenke GC, Mao X, Hanefeld FA, Frahm J. 1998. Quantitative proton magnetic resonance spectroscopy of childhood adrenoleukodystrophy. Neuropediatrics 29: 254–264. Pridmore CL, Baraitser M, Harding B, Boyd SG, Kendall B, Brett EM. 1993. Alexander’s disease: clues to diagnosis. J Child Neurol 8: 134–144. Rahman S, Brown RM, Chong WK, Wilson CJ, Brown GK. 2001. A SURF1 gene mutation presenting as isolated leukodystrophy. Ann Neurol 49: 797–800. Rajanayagam V, Grad J, Krivit W, Loes DJ, Lockman L, Shapiro E, Balthazor M, Aeppli D, Stillman AE. 1996. Proton MR spectroscopy of childhood adrenoleukodystrophy. Am J Neuroradiol 17: 1013–1024. Rajanayagam V, Balthazor M, Shapiro EG, Krivit W, Lockman L, Stillman AE. 1997. Proton MR spectroscopy and neuropsychological testing in adrenoleukodystrophy. Am J Neuroradiol 18: 1909–1914. Reeves SA, Helman LJ, Allison A, Israel MA. 1989. Molecular cloning and primary structure of human glial fibrillary acidic protein. Proc Natl Acad Sci USA 86: 5178–5182. Rodriguez D, Gelot A, della Gaspera B, Robain O, Ponsot G, Sarlieve LL, Ghandour S, Pompidou A, Dautigny A, Aubourg P, Pham-Dinh D. 1999. Increased density of oligodendrocytes in childhood ataxia with diffuse central

hypomyelination (CACH) syndrome: neuropathological and biochemical study of two cases. Acta Neuropathol (Berl) 97: 469–480. Rodriguez D, Gauthier F, Bertini E, Bugiani M, Brenner M, N’Guyen S, Goizet C, Gelot A, Surtees R, Pedespan JM, Hernandorena X, Troncoso M, Uziel G, Messing A, Ponsot G, Pham-Dinh D, Dautigny A, Boespflug-Tanguy O. 2001. Infantile Alexander disease: spectrum of GFAP mutations and genotype–phenotype correlation. Am J Hum Genet 69: 1134–1140. Russo Jr LS, Aron A, Anderson PJ. 1976. Alexander’s disease: a report and reappraisal. Neurology 26: 607–614. Salvan AM, Confort-Gouny S, Chabrol B, Cozzone PJ, VionDury J. 1999. Brain metabolic impairment in non-cerebral and cerebral forms of X-linked adrenoleukodystrophy by proton MRS: identification of metabolic patterns by discriminant analysis. Magn Reson Med 41: 1119–1126. Sasaki M, Sakuragawa N, Takashima S, Hanaoka S, Arima M. 1991. MRI and CT findings in Krabbe disease. Pediatr Neurol 7: 283–288. Schaumburg HH, Powers JM, Raine CS, Suzuki K, Richardson Jr EP. 1975. Adrenoleukodystrophy. A clinical and pathological study of 17 cases. Arch Neurol 32: 577–591. Schiffmann R, Moller JR, Trapp BD, Shih HH, Farrer RG, Katz DA, Alger JR, Parker CC, Hauer PE, Kaneski CR, Heiss JD, Kaye EM, Quarles RH, Brady RO, Barton NW. 1994. Childhood ataxia with diffuse central nervous system hypomyelination. Ann Neurol 35: 331–340. Schwankhaus JD, Parisi JE, Gulledge WR, Chin L, Currier RD. 1995. Hereditary adult-onset Alexander’s disease with palatal myoclonus, spastic paraparesis, and cerebellar ataxia. Neurology 45: 2266–2271. Shapiro E, Krivit W, Lockman L, Jambaque I, Peters C, Cowan M, Harris R, Blanche S, Bordigoni P, Loes D, Ziegler R, Crittenden M, Ris D, Berg B, Cox C, Moser H, Fischer A, Aubourg P. 2000. Long-term effect of bone-marrow transplantation for childhood-onset cerebral X-linked adrenoleukodystrophy. Lancet 356: 713–718. Singhal BS, Gursahani RD, Udani VP, Biniwale AA. 1996. Megalencephalic leukodystrophy in an Asian Indian ethnic group. Pediatr Neurol 14: 291–296. Smeitink J, van den Heuvel L, DiMauro S. 2001. The genetics and pathology of oxidative phosphorylation. Nat Rev Genet 2: 342–352. Spalice A, Popolizio T, Parisi P, Scarabino T, Iannetti P. 2000. Proton MR spectroscopy in connatal Pelizaeus–Merzbacher disease. Pediatr Radiol 30: 171–175. Takanashi J, Inoue K, Tomita M, Kurihara A, Morita F, Ikehira H, Tanada S, Yoshitome E, Kohno Y. 2002. Brain N-acetylaspartate is elevated in Pelizaeus–Merzbacher disease with PLP1 duplication. Neurology 58: 237–241.

MR spectroscopy in pediatric white matter disease

Takanashi J, Sugita K, Tanabe Y, Niimi H. 1998. Adolescent case of Alexander disease: MR imaging and MR spectroscopy. Pediatr Neurol 18: 67–70. Tedeschi G, Schiffmann R, Barton NW, Shih HH, Gospe Jr SM, Brady RO, Alger JR, Di Chiro G. 1995. Proton magnetic resonance spectroscopic imaging in childhood ataxia with diffuse central nervous system hypomyelination. Neurology 45: 1526–1532. Thomas PK, Halpern JP, King RH, Patrick D. 1984. Galactosylceramide lipidosis: novel presentation as a slowly progressive spinocerebellar degeneration. Ann Neurol 16: 618–620. Toft PB, Geiss-Holtorff R, Rolland MO, Pryds O, Muller-Forell W, Christensen E, Lehnert W, Lou HC, Ott D, Hennig J, et al.. 1993. Magnetic resonance imaging in juvenile Canavan disease. Eur J Pediatr 152: 750–753. Topcu M, Saatci I, Topcuoglu MA, Kose G, Kunak B. 1998. Megalencephaly and leukodystrophy with mild clinical course: a report on 12 new cases. Brain Dev 20: 142–153. Tzika AA, Vigneron DB, Ball Jr WS, Dunn RS, Kirks DR. 1993a. Localized proton MR spectroscopy of the brain in children. J Magn Reson Imaging 3: 719–729. Tzika AA, Ball Jr WS, Vigneron DB, Dunn RS, Nelson SJ, Kirks DR. 1993b. Childhood adrenoleukodystrophy: assessment with proton MR spectroscopy. Radiology 189: 467–480. van der Knaap MS, Valk J. 1995. Magnetic Resonance of Myelin, Myelination and Myelin Disorders, 2nd edn. Springer, Berlin. van der Knaap MS, Barth PG, Gabreels FJM, Franzoni E, Begeer JH, Stroink H, Rotteveel JJ, Valk J. 1997. A new leukoencephalopathy with vanishing white matter. Neurology 48: 845–855. van der Knaap MS, Barth PG, Stroink H, van Nieuwenhuizen O, Arts WF, Hoogenraad F, Valk J. 1995a. Leukoencephalopathy with swelling and a discrepantly mild clinical course in eight children. Ann Neurol 37: 324–334. van der Knaap MS, Barth PG, Vrensen GF, Valk J. 1996. Histopathology of an infantile-onset spongiform leukoencephalopathy with a discrepantly mild clinical course. Acta Neuropathol (Berl) 92: 206–212. van der Knaap MS, Kamphorst W, Barth PG, Kraaijeveld CL, Gut E, Valk J. 1998. Phenotypic variation in leukoencephalopathy with vanishing white matter. Neurology 51: 540–547.

van der Knaap MS, Leegwater PA, Konst AA, Visser A, Naidu S, Oudejans CB, Schutgens RB, Pronk JC. 2002. Mutations in each of the five subunits of translation initiation factor eIF2B can cause leukoencephalopathy with vanishing white matter. Ann Neurol 51: 264–270. van der Knaap MS, Naidu S, Breiter SN, Blaser S, Stroink H, Springer S, Begeer JC, van Coster R, Barth PG, Thomas NH, Valk J, Powers JM. 2001. Alexander disease: diagnosis with MR imaging. Am J Neuroradiol 22: 541–552. van der Knaap MS, Valk J, Barth PG, Smit LM, van Engelen BG, Tortori Donati P. 1995b. Leukoencephalopathy with swelling in children and adolescents: MRI patterns and differential diagnosis. Neuroradiology 37: 679–686. van der Knaap MS, van der Grond J, Luyten PR, den Hollander JA, Nauta JJ, Valk J. 1992. 1H and 31P magnetic resonance spectroscopy of the brain in degenerative cerebral disorders. Ann Neurol 31: 202–211. von Figura K, Gieselmann V, Jaeken J. 2001. Metachromatic leukodystrophy. In The Metabolic and Molecular Bases of Inherited Disease (Eds., Scriver CR, Beaudet AL, Sly WS, Valle D), 8th edn. McGraw-Hill, New York, pp. 3695–3724. Wenger DA, Suzuki K, Suzuki Y, Suzuki K. 2001. Galactosylceramide lipidosis: Globoid cell leukodystrophy (Krabbe disease). In The Metabolic and Molecular Bases of Inherited Disease (Eds. Scriver CR, Beaudet AL, Sly WS, Valle D), 8th edn. McGraw-Hill, New York, pp. 3669–3693. Wilken B, Dechent P, Brockmann K, Finsterbusch J, Baumann M, Ebell W, Korenke GC, Pouwels PJW, Hanefeld FA, Frahm J. 2003. Quantitative proton magnetic resonance spectroscopy of children with adrenoleukodystrophy before and after hematopoietic stem cell transplantation. Neuropediatrics 34: 237–246. Wittsack HJ, Kugel H, Roth B, Heindel W. 1996. Quantitative measurements with localized 1H MR spectroscopy in children with Canavan’s disease. J Magn Reson Imaging 6: 889–893. Zarifi MK, Tzika AA, Astrakas LG, Poussaint TY, Anthony DC, Darras BT. 2001. Magnetic resonance spectroscopy and magnetic resonance imaging findings in Krabbe’s disease. J Child Neurol 16: 522–526.

777

778

Folker Hanefeld, Knut Brockmann and Peter Dechent

Case Study 45.1 Adenoleukodystrophy (ALD): MRSI Peter Barker, D.Phil., Johns Hopkins University School of Medicine, Baltimore, MD, USA History 7-year-old male with elevated very long chain fatty acids and cortical blindness.

T2

Cho a b

Technique Conventional MRI and multi-slice MRSI (TE 280 ms). c Imaging findings T2 MRI shows confluent, bilateral posterior hyperintensity consistent with demyelination. Regions of T2 hyperintensity show a large elevation of Cho, increased Lac and decreased NAA, consistent with demyelination, inflammation and neuronal loss/dysfunction, respectively (c). WM (b) anterior to the lesion with normal MRI appearance also has higher Cho than normal.

Lac

NAA

Discussion MRSI is useful for detecting demyelination and axonal loss in X-ALD, and may be able to predict disease progression (Eichler, 2002). Chronic, “burnt-out” ALD lesions typically have low levels of all metabolites.

NAA Cho Cr

a

Key points MRSI can detect demyelination and axonal loss in ALD. b

NAWM on MRI may have abnormal metabolism, predicting lesion progression. Short TE spectra of ALD lesions show increased mI.

Lac c ppm

3.0

2.0

1.0

Reference Eichler FS, Barker PB, Cox C, Edwin D, Ulug AM, Moser HW, Raymond GV. 2002. Proton MR spectroscopic imaging predicts lesion progression on MRI in X-linked adrenoleukodystrophy. Neurology 58(6): 901–907.

46

MR spectroscopy of inborn errors of metabolism Alberto Bizzi1, Marianna Bugiani2 and Ugo Danesi1 1

Department of Neuroradiology, Istituto Nazionale Neurologico “Carlo Besta”, Milan, Italy Department of Child Neurology, Istituto Nazionale Neurologico “Carlo Besta”, Milan, Italy

2

Key points • MR spectroscopy (MRS) supplements conventional imaging and biochemical studies in diagnosis of several inborn errors of metabolism. • The disorders involving brain creatine deficiency, and brain N-acetyl aspartate (NAA) deficiency, can be diagnosed by MRS. • Canavan and Salla disease show an elevated “NAA” signal at 2.0 ppm. • Branched-chain amino acids at 0.9 ppm can be observed due to acute metabolic decompensation in maple syrup urine disease. • Untreated phenylketonuria patients may show a small phenylalanine signal at 7.36 ppm (i.e. downfield of water). • Patients with non-ketotic hyperglycinemia show a glycine resonance at 3.55 ppm (use a long echo time to distinguish this from myoinositol, mI). • Some patients with succinate dehydrogenase deficiency show a unique spectrum with elevation of brain succinate (2.42 ppm) and lactate. • Elevated arabitol and ribitol (3.5 and 3.8 ppm) have been observed in a leukodystrophy with abnormal polyol metabolism. • Many leukodystrophies, amino- and organicacidopathies have spectra which show changes related to axonal integrity (NAA loss) or demyelination (increased choline, mI), or gliosis, but do not have specific diagnostic patterns.

Genetics of metabolic diseases Hereditary inborn errors of metabolism are the results of an enzyme defect involving one or more metabolic pathways. An enzymatic block may act by inducing deficiency of metabolites normally produced beyond the block; by interfering with other metabolic pathways as a result of deviation from normal to accessory or normally unused pathways; by producing accumulation of substances that may interfere with the cell’s function and/or survival; or by interfering in various ways with other essential metabolic processes. Classic genetic disorders are due to an abnormality in a single gene, in a chromosome or may be multifactorial. In addition there are other more recently described categories, such as mitochondrial inheritance, fragile site, and genomic imprinting. Most metabolic diseases are inherited according to single-gene Mendelian mode of inheritance. The inheritance of the phenotypic set follows the Mendelian rules of inheritance: autosomal-dominant, autosomal-recessive, X-linked. Canavan, Krabbe, Gaucher (1q), galactosemia (9p), Hallervorden–Spatz and Wilson disease (13q) are examples of autosomal-recessive single-gene disorders. Adrenoleukodystrophy (ALD), Aicardi syndrome, Pelizaeus–Merzbacher disease (PMD) are examples of X-linked single-gene disorders. The incidence of single-gene disorders is between 2% and 3% by the age of 1 year, closer to 5% by the age of 25 years (Baird et al., 1988). Newborn screening programs are available for single-gene disorders that respond well to dietary therapy, such as phenylalanine (Phe) hydroxylase deficiency, galactosemia, and maple syrup urine disease (MSUD). The completion 779

780

Alberto Bizzi, Marianna Bugiani and Ugo Danesi

of the Human Genome Mapping Project will make it possible to screen the population for many other single-gene disorders. This novel possibility raises many ethical and legal questions that have yet to be addressed (Gilliam et al., 1990). Chromosomes are the structures in which genes are packaged. Chromosome disorders are the result of either deficiency or excess of chromosomal material. It is estimated that about 5 in 1000 live newborns will have a chromosome abnormality. Deficiency or excess of chromosomal material can be the result of a change in chromosome number (polyploidy, aneuploidy) or structure. Multifactorial inheritance is characteristic of congenital malformations, such as neural tube defects. These malformations tend to be familial, but careful analyses of large number of pedigrees exclude single-gene inheritance. The magnitude of the risk varies from malformation to malformation; when many family members are affected, the recurrence risk for the disorder is greater. Mitochondrial inheritance is an example of non-Mendelian genetic mechanism. Human mitochondrial DNA (mtDNA) is a small double-stranded molecule, including 13 structural genes for proteins involved in oxidative phosphorylation (OXPHOS). The mtDNA is transmitted via the egg but not the sperm. Variability in phenotype is related to the proportion of mutant mtDNA as compared with wild-type mtDNA present in the tissue. Maternal transmission has been demonstrated for several mitochondrial encephalomyopathies (ME): Leber’s hereditary optic atrophy (LHON), ME with lactic acidosis and strokes (MELAS), myoclonus epilepsy with ragged red fibers (MERRFs) and Kearns–Sayre syndrome (KSS). Other MEs follow the classic Mendelian transmission (single-gene disorders) because the deficient protein (involved in OXPHOS) is coded by a nuclear gene. Most cases of KSS (external ophthalmoplegia, retinal degeneration, heart block, and high cerebrospinal fluid (CSF) protein content) appear to be sporadic, therefore they may represent new mutations.

Criteria of clinical diagnosis The majority of hereditary metabolic disorders become manifest during childhood; they pose a

diagnostic challenge to pediatricians because they often affect the brain more often than any other organ system, and the same symptoms and signs occur in different diseases. Time of presentation and selective vulnerability are key concepts for understanding metabolic diseases. Most metabolic disorders are progressive. The time of presentation and the mode of progression may be very different even for diseases with the same biochemical disorder. Most diseases set in after an early phase of normal development, and the clinical course is often characterized by slowing of psychomotor development, and then by a plateau phase before entering a phase of neurological deterioration. In the most severely affected patients the deterioration phase has already begun in utero, and significant clinical and neuroradiological features are present at birth. In most patients, a normal early psychomotor development is observed followed by a slowing of variable duration which in some cases can be very long. After developmental slowing, most patients reach a plateau stage, again of variable duration. Deterioration, the most easily recognizable clinical phase of illness progression, may be slow or rapid. The concept of selective vulnerability was introduced in the first half of the 20th century when it was suggested that vascular, metabolic, and degenerative diseases of the central nervous system (CNS) followed a specific pattern of damage that could be explained by unique properties of specific populations of glial and neuronal cells. Elements of the CNS have varying degree of sensitivity to different biochemical injuries. Levels of local metabolic activity are an important factor influencing selective vulnerability. Energy depletion by metabolic deficiency will have the greatest effect on the most active cell populations. In infants, active myelination zones have a high level of activity and are, therefore, more vulnerable to damage. In adults the gray matter (GM) has a higher level of activity and it will be damaged most severely. It is well known that deficiency of an enzyme or the abnormal accumulation of toxic metabolites have different effects on different brain cell populations. Differences in selective vulnerability may be explained by differences in residual activity of enzymes in each population of cells; differences in importance of the enzyme function missing from different cells;

MR spectroscopy of inborn errors of metabolism

differences in the effects of the accumulation of abnormal substances, differences in sensitivity to lack of substances that are not formed, and presence of other factors within the cell with synergistic or antagonist effects. In inborn errors of metabolism, the selection of the primary target and the pattern of spreading of lesions may be influenced by these factors. The recognition of patterns of selective vulnerability is of practical value and it contributes to the diagnostic value of MR imaging (MRI). The pivotal role of conventional MRI In 1983 Dyken and Krawiecki (1983) proposed to group progressive neurodegenerative disorders of infancy and childhood into five major categories according to their anatomo-pathological selective vulnerability. The five categories are: polioencephalopathies, leukoencephalopathies, corencephalopathies, spinocerebellopathies, and diffuse encephalopathies. This classification based on the pattern of brain involvement is particularly useful to the radiologist who is trying to identify the primary targets of the disease and the pattern of spread of the lesions. The recognition of these morphological features will help the radiologist to narrow the differential diagnosis from an imaging perspective. This approach is most useful in the early stages of the disease. On the contrary, in the end stage of the illness a broader involvement of areas of the brain, that are not the primary target of the disease, will cause overlap of imaging findings among different diseases. Classification of diseases is changing as we gain more knowledge. The advent of MRI in the 1980s has provoked a dramatic acceleration in the recognition and diagnosis of metabolic diseases. MRI has become an important test in the initial diagnostic work-up of children with a suspected metabolic disorder. It is likely that inborn errors of metabolism will ultimately be diagnosed and treated according to the underlying genetic mutation. In the meantime MRI, will continue to have a pivotal role in the initial work-up of these patients, and the morphological approach will continue to allow the radiologist to contribute to the diagnosis in many cases. The sensitivity of MRI to detect subtle brain abnormalities and its multiplanar capability are far superior

to computed tomography (CT) and ultrasound. Fluid attenuated inversion recovery (FLAIR) images are useful for detecting signal changes in children beyond the age of two, when maturation of myelin in the white matter (WM) is normally completed. Spin-echo (SE) T2-weighted images are also useful in the early years of life. In most disorders, T1-weighted images are generally less sensitive to pathological change in tissue than T2-weighted images, nonetheless they may help to discriminate between areas that have been involved early in the course of disease and areas that have been affected late. An important initial assessment of the MR image should be whether the disease involves primarily cortical GM (polioencephalopathy), the deep gray nuclei (corencephalopathy), the WM (leukoencephalopathy) or is a diffuse process (diffuse encephalopathy). If MR images show that the cortex is the primarily involved structure, with cortical thinning and enlarged cortical sulci, consideration should be given to such disorders as neuronal ceroid lipofuscinoses or glycogen storage diseases. If signal abnormalities in the deep nuclei are the most prominent features, the location of the affected structures becomes very important. Abnormalities of the striatum are seen in mitochondrial disorders (Leigh syndrome (LS) and MELAS syndrome), organic acidemias, Wilson’s disease, juvenile Huntington disease, and hypoglycemia. Abnormalities in the globus pallidus are characteristic of Hallervorden–Spatz (T1 and T2 shortening), methylmalonic acidemia, and L-2-hydroxyglutaric (L-2-OHG) aciduria (T2 prolongation). In many metabolic diseases, abnormalities in the deep gray nuclei may be associated with WM involvement. When the WM is the primarily affected, it is important to determine whether the subcortical, lobar, or deep WM are involved. Early involvement of the subcortical arcuate (U) fibers is seen in galactosemia, in some organic acidopathies and in KSS. If involvement of the arcuate fibers is found with macrocephaly, the diagnosis of L-2-OHG aciduria should be suggested; if early involvement of U-fibers and macrocephaly are associated with T2 prolongation in the globi pallidi and thalami the diagnosis of Canavan disease (CD) becomes the most likely. Alexander disease is the most likely diagnosis when extensive cerebral WM changes with

781

782

Alberto Bizzi, Marianna Bugiani and Ugo Danesi

frontal predominance and relative sparing of the occipital areas are found in a macrocephalic patient. In Alexander disease, additional MRI criteria may be present: a periventricular rim with high signal on T1-weighted images and low signal on T2-weighted images, abnormalities of basal ganglia and thalami (BGT), brain stem abnormalities, and contrast enhancement of particular GM and WM structures. The presence of four of these five criteria should suggest the diagnosis of Alexander disease (van der Knaap et al., 2001). The definitive diagnosis is now possible with molecular analysis showing a mutation in the glial acidic fibrillary protein (GFAP) gene (Brenner et al., 2001). Early symmetric involvement restricted to deep WM with sparing of the subcortical U-fibers suggests metachromatic leukodystrophy (MLD), Krabbe disease (KD), and GM1 or GM2 gangliosidosis. Extensive deep WM signal abnormalities in the posterior cerebral regions and in the splenium of the corpus callosum with relative sparing of the arcuate fibers is a characteristic pattern and it is seen in about 80% of symptomatic patients with childhood X-linked adrenoleukodystropy (X-ALD). The MRI pattern of PMD is also usually highly suggestive: imaging shows a delay or an arrest in myelination in a stage that is in itself normal but it is not normal for the age of the patient. Follow-up MRI studies confirm the lack of progress of myelination. Laboratory work-up for the diagnosis of hereditary metabolic diseases Pediatric neurologists frequently request MRI as an early investigation. The suggestion of an accurate and narrow differential diagnosis in the radiology report will have important consequences on the selection of relevant subsequent laboratory tests. Important biochemistry and molecular biology tests are listed in Table 46.1. Many of these procedures are run on urine, blood, and CSF samples. Some tests should be performed routinely, whenever a hereditary metabolic disease is suspected, other highly specialized and expensive investigations should always be guided by clinical, radiological and other biochemical findings. DNA studies and molecular biology techniques are essential to confirm the diagnosis and counseling of hereditary disease. The combination of

conventional MRI and MR spectroscopy (MRS) has the potential to diagnose or exclude selected rare metabolic diseases non-invasively. MRS in hereditary metabolic diseases In vivo proton MRS is capable of detecting mobile metabolites with concentrations in the millimolar range. Total scan time must be limited to no longer than 50–60 min during a clinical study. A relatively large number of hereditary metabolic diseases have been investigated by MRS in the last 15 years. Clinical spectroscopic findings may be considered in three groups: (a) Metabolic diseases with specific MRS findings which in themselves allow the confident diagnosis of a specific disease; (b) Metabolic diseases with characteristic MRS findings that are shared by few disorders; (c) Metabolic diseases with non-specific MRS findings. Metabolic diseases with specific MRS findings There are only a few metabolic diseases that can be diagnosed immediately by MRS. Whilst an unusual occurrence, it is important for the clinical spectroscopist to identify the pathognomonic spectra in these cases. Creatine deficiency syndrome Stockler discovered the first defect in creatine (Cr) metabolism with MRS in 1994 (Stockler et al., 1994); in vivo MRS unexpectedly revealed the absence of Cr in the brain of a boy with development delay since the age of 5 months and severe extrapyramidal movement signs. Oral Cr supplementation led to significant improvement of the patient’s symptoms. This child was found to have a deficiency of guanidinoacetate methyl-transferase (GAMT). Two other Cr deficiency syndromes have been indentified; the arginine: glycine amidinotransferase (AGAT) deficiency and the Cr transporter (CrT1) defect. GAMT and AGAT deficiency have autosomal-recessive traits, whereas the CrT1 defect is an X-linked disorder. Cr deficiency syndromes result in a progressive encephalopathy with mental retardation, severe expressive language impairment, extrapyramidal movement disorder and drug-resistant epilepsy. The clinical presentation of the disease is heterogeneous, irrespective of the

MR spectroscopy of inborn errors of metabolism

Table 46.1. Diagnostic work-up: main laboratory investigations Routine laboratory blood investigations Macrocytic anemia Vacuolized leukocytes ⇓ Cholesterol ⇑ Triglycerides ⇑ CPK ⇑ Uric acid ⇓ Uric acid ⇑ Iron, transferrin ⇑ Copper ⇓ Copper, coeruloplasmin Biochemical investigations Blood gas analysis ⇑ Plasma Lac ⇑ NH3 Aminoacids (plasma, urine, CSF) Urinary organic acids

VLCFA, phytanic acid Carnitine status (free carnitine, acylcarnitine, total carnitine) Urinary purine, pyrimidines Orotic acid Glycosaminoglycane Oligosaccharides, free neuraminic acid Transferrin electrophoresis Urinary pterins

Cobalamin or folic acid metabolism disorders Lysosomal storage disorders Steroid synthesis defects, lipoprotein disorders Glycogen storage disorders, lipoprotein disorders Mitochondrial disorders, fatty acid oxidation defects, glycogen storage disorders, glycolysis defects Disorders of purine metabolism, fatty acid oxidation defects, mitochondrial disorders, glycogen strorage disorders Disorders of purine metabolism, molybdenum cofactor deficiency Peroxisomal disorders Peroxisomal disorders Wilson disease, Menkes disease

Metabolic acidosis Respiratory chain defects, PDH defect, some organic acidurias Some organic acidurias, urea cycle defects Aminoacid metabolism disorders, disorders of energy metabolism, urea cycle disorders Organic acidurias, aminoacid metabolism disorders, fatty acid oxidation defects, disorders of energy metabolism, urea cycle defects, disorders of vitamin B12 metabolism Peroxisomal disorders Disorders of intermediate metabolism, fatty acids oxidation defects Disorders of purine and pyrimidine metabolism OTC deficiency, urea cycle defects, disorders of pyrimidine metabolism Lysosomal storage disorders/mucopolysaccharidoses Lysosomal storage disorders/oligosaccharidoses Disorders of protein glycosylation (CDG syndromes) Hyperphenylalaninemia

Determination of selected enzymatic activities Leukocyte Biopsy (skin for cultured fibroblasts, muscle, liver) Molecular genetic investigations (nDNA, mtDNA) Post-mortem investigations

underlying gene mutations or residual enzyme activity. Failure of expressive language is, however, the striking common feature in these patients. Cr and its phosphorylated form (PCr) are essential for energy storage and transfer. The depletion of the Cr/PCr pool within the CNS makes this syndrome easily detectable by MRS. In brain, total Cr concentration

is lower than in skeletal muscle, heart, retina, and spermatozoa (Wyss and Kaddurah-Daouk, 2000), but there is ample evidence for close correlations between Cr metabolism and proper brain function. Cr biosynthesis (Figure 46.1) occurs mainly in liver, pancreas and kidney, and involves two enzymes: L-AGAT and GAMT. Cr is transported through the

783

784

Alberto Bizzi, Marianna Bugiani and Ugo Danesi

1. Cr synthesis (Liver, Pancreas) Glycine

3. Phosphate system (Brain, Muscle)

Arginine

AGAT Ornithine

ATP

GAA

Cr

S-adenosylmethionine

Cr

PCr

CK

GAMT S-adenosylhomethionine

ADP

Nonenzymatic conversion

Cr

Nonenzymatic conversion

CrT1 4. Urinary excretion

2. Cr uptake (Brain, Muscle) Fig. 46.1 The Cr pathway in the human body is indicated. There are two enzymatic defects that may occur during the synthesis in the liver and pancreas. A third defect may involve the transporter and it is organ specific (brain, heart, muscle, kidney). GAA: Guanidinoacetate.

blood to the tissues void of Cr synthesis, where it is taken up through the CrT1. Most of cases reported so far are due to a lack of the second enzyme. GAMT deficiency is an autosomal-recessive disease characterized by accumulation of GAA and reduced Cr in brain, body fluids and urine. Patients with GAMT deficiency benefit from oral Cr monohydrate supplementation (Figure 46.2), which helps to control epilepsy and partially restores neurological development over time (Stockler et al., 1996). Improvement of movement disorders is more controversial, since it appears to be due to the toxic effect of GAA rather than to Cr depletion. Treatment with combined arginine restriction and ornithine substitution in GAMT deficiency is capable to decrease guanidinoacetic acid permanently and improves the clinical outcome. A defect of the first enzyme AGAT was disclosed in three related patients with reversible Cr deficiency (Bianchi et al., 2000; Battini et al., 2002). In these patients, GAA is decreased and there are no extrapyramidal movement disorders. Clinical MRS is important to document restoring of the brain Cr pool during therapy. The persistent

absence of the Cr peak in the spectrum should rule out both enzymatic defects of Cr biosynthesis, suggesting a CrT1 defect and prompting a search for molecular defects in the CrT1 gene SLC6A8. Mutations in the CrT1 gene on the X chromosome have been identified in boys with developmental delay, no language acquisition, partial epilepsy, and irreversible cerebral Cr absence (Cecil et al., 2001; Bizzi et al., 2002). Patients with CrT1 may have normal Cr levels in plasma and elevated Cr levels in urine; GAA is not increased. Additional patients with the CrT1 from four unrelated families living in a single metropolitan area have been diagnosed (deGrauw et al., 2002). This implies that the syndrome may be more common than originally thought and it emphasizes the importance of screening patients with spectroscopy. Female carriers heterozygous for the CrT1 mutation may demonstrate a low IQ, learning disabilities, and lower Cr in the brain (Cecil et al., 2001; Bizzi et al., 2002; deGrauw et al., 2002). A significant reduction of Cr on MRS has been reported in a 9-day-old heterozygous female suggesting that consideration must be given to oral Cr supplementation

MR spectroscopy of inborn errors of metabolism

1,2 1 0,8 0,6 0,4 0,2 0 Diagnosis

4 months Cr/Cho

10 months

16 months

Cr/NAA

Fig. 46.2 Cr (GAMT) deficiency. A spectrum (multivoxel point resolved spectroscopy (PRESS): repetition time (TR)/echo time (TE): 1500/136; 16 16 matrix) from the posterior third of the right centrum semiovale at time of diagnosis, 4, 10, and 16 months after starting diet with Cr monohydrate (700 mg/day) supplementation in a 2-year-old male diagnosed with GAMT deficiency. Semi-quantitative analysis of the choline (Cho)/Cr and Cr/N-acetyl aspartate (NAA) ratios show progressive Cr recovery in the brain. The slow recovery in this particular case was enhanced by a period of non-compliance in diet supplementation between 8 and 10 months of therapy. H- MR spectroscopic imaging (MRSI) confirmed the suspicion of the pediatrician that the parents had stopped supplementing Cr to the diet of the infant. Resuming the diet allowed further recovery of Cr levels.

in carriers because it could lead to normalized level of brain Cr and improved metabolic or cognitive function (Cecil et al., 2003). The Cr deficiency syndromes have also been reported in adults with history of developmental motor delay, seizures, and severe speech language impairment. Two adult patients with CrT1 were able to communicate using signs and gestures or understand spoken language well (deGrauw et al., 2002). After 3 weeks of Cr supplementation and arginine restrictive diet, epileptic seizures disappeared permanently in a 26-year-old patient with GAMT deficiency (Schulze et al., 2003). This patient did not show any active speech development, but he learned to draw simple pictures and to play games after 1 year of therapy. Evidence of GAMT deficiency in striated muscle has been demonstrated with P-MRS only in this adult patient. Muscle strength was normal, therefore, alternative mechanisms in high-energy phosphate storage and transport may be present in striated muscle.

Conventional MRI findings are non-specific in Cr deficiency syndrome. So far, bilateral T2-weighted hyperintensity in the pallidum have been reported in five of seven patients with GAMT. Maturation of myelin follows normal milestones in the majority of patients. We observed association with mesial temporal sclerosis (Bizzi et al., 2002) in one patient with CrT1 and in another with GAMT; a hypoplastic left cerebral hemisphere has also been reported (deGrauw et al., 2002). In two adult patients with CrT1, follow-up MRI scans have shown progressive cerebral atrophy in the second and third decade (deGrauw et al., 2002). Normal MRI at the age of 26 was reported in a patient with GAMT deficiency, whose symptoms were already present in infancy and showed no further progression during adolescence (Schulze et al., 2003). In conclusion, MRS using either short or long echo times (TEs) is a sensitive method for detecting Cr deficiency syndromes. Metabolic and genetic analysis

785

786

Alberto Bizzi, Marianna Bugiani and Ugo Danesi

will identify those patients that will benefit from oral Cr supplementation distinguishing reversible (GAMT and AGAT) from irreversible (CrT1) defects. MRS may be used to monitor restoration of the Cr pool in brain. Oral supplementation (400 mg/kg/day) will slowly restore Cr pools in the brain. After 6 weeks, brain Cr should have reached almost 50% of its normal concentration, and it should be nearly normal after 24 months of treatment. We observed transient depletion of Cr in a patient with diagnosed GAMT deficiency who had already restored his Cr pool. The parents later admitted that they had suspended oral supplementation for 2 months. In established cases, MRS may also be used to monitor therapy. Mental retardation with global NAA deficiency NAA in the CNS was discovered in 1956 (Tallan et al., 1956), although it was not until the advent of widely available in vivo MRS that interest in this amino acid became the subject of extensive study. NAA is predominantly located in neurons, dendrites and axons; however, it has been shown to be also present in oligodendrocytes (progenitors, immature, and even mature cells) (Bhakoo and Pearce, 2000) and mast cells (Burlina et al., 1997). The well-defined methyl group of NAA dominates the proton MR spectrum beyond the age of 12 months. Reduction of NAA determined by MR spectroscopic imaging (MRSI) in piriform cortex, amygdala, and hippocampus correlated well with neuronal injury and decreased axonal density determined by histology (Ebisu et al., 1994). In most contexts NAA is considered a surrogate marker of neural tissue integrity. Decreased NAA is associated with neuronal loss in a wide range of diseases. Reports of reversible NAA signal loss may be in multiple sclerosis (MS) (De Stefano et al., 1995), acute disseminated encephalomyelitis (Bizzi et al., 2001) and MEs (Clark, 1998) suggest that low levels may also reflect reversible neuronal dysfunction. The idea of NAA as a completely non-specific marker of neuroaxonal integrity is challenged by a report of a 3-year-old child with severe developmental delay and microcrania with normal MRI and electroencephalograms (EEG), in whom single-voxel MRS showed complete absence of NAA in four different brain regions (Figure 46.3). (Martin et al., 2001). The concentrations of myo-inositol (mI), choline (Cho), and Cr were

within age-appropriate normal limits. The authors hypothesized a block in the biosynthesis of NAA at the level of acetyl-CoA-L-aspartate-N-acetyltransferase (ANAT), which converts L-aspartate (L-asp) to NAA (Figure 46.4). Both precursors (L-asp and acetylCoA) are involved in many metabolic pathways and therefore are not expected to accumulate and show up in the spectrum. This case suggests a specific biosynthetic defect of NAA, and supports a role of MRS in detecting new neurometabolic disorders. Canavan’s disease Canavan’s disease is caused by a deficiency of aspartoacylase (ASPA), an enzyme that deacetylates NAA to generate free acetate in the brain (Figure 46.4) (cf. Chapter 45). Canavan’s disease is caused by singlegene mutations in the ASPA gene. It is most common among children of Ashkenazi Jewish descent, but it has been diagnosed in many diverse ethnic groups. There are three forms: the infantile is the most frequent form; the congenital form is characterized by lethargy that becomes manifest a few days after birth; few cases of the juvenile form have been reported. A brief history of the discovery of this disease is a good example of how progresses of knowledge in medicine are made. In 1931 Canavan described a case of what he thought was “Schilder’s disease” with spongy degeneration of WM (Canavan, 1931). In 1949 van Bogaert and Bertrand recognized that “spongy degeneration of the brain” was a genetic disease and in the following years several reports showed that Canavan’s disease had an autosomalrecessive mode of inheritance with high prevalence among Ashkenazi Jews (Gambetti et al., 1969). It was only in 1988 that elevation of N-acetylaspartic acid in urine and the deficiency of ASPA in cultured skin fibroblasts were described (Matalon et al., 1988). A couple of years later, single-voxel MRS in vivo studies showed that NAA is elevated in the brain of these patients (Grodd et al., 1990; Barker et al., 1992). In recent years, more cases of Canavan’s disease have been reported, suggesting that this disease is more common than previously thought. In the majority of cases of the infantile form time of presentation is at 3 months of age with signs of delayed development. The triad of hypotonia, head lag, and macrocephaly should make one consider

MR spectroscopy of inborn errors of metabolism

787

IR T1-weighted (a)

(2)

(3) (b) NAA Cr Cho Glucose taurine mI

(4)

Gln, Glu

NAAG Gln, Glu, GABA

Mobile lipids Lac

(1)

(5) SE T2-weighted 4.5

4.0

3.5

3.0 2.5 ppm

2.0

1.5

1.0

4.5

4.0

3.5

3.0 2.5 ppm

2.0

1.5

Fig. 46.3 NAA deficiency. Axial inversion recovery turbo SE (IRTSE) (a) and T2-weighted MR images (b) show a normal brain except for minimal patchy signal abnormality in the posterior periventricular WM: these mild signal changes are considered normal in a 3-year-old child. Single-voxel spectra (PRESS: TR/TE  6000/30 ms) acquired from the occipital GM (1 and 2), parieto-occipital WM (3), and basal ganglia (4) showing complete absence of NAA. The absence of NAA is even more clearly shown in the long TE (TE  270 ms) spectrum (5) acquired in the parietal-occipital WM. (Gln: Glutamine; Glu: Glutamate; Lac: Lactate)

the diagnosis. MRI shows a mild swollen aspect in the subcortical WM with symmetrical and confluent T2 prolongation, bilateral involvement of the globi pallidi and thalami, cerebellum and brainstem; the neostriatum is spared. In such cases the differential diagnosis may include maple syrup urine disease. MRS can readily confirm a presumed diagnosis of Canavan’s disease by showing elevated NAA. Multivoxel MRSI may show a heterogeneous distribution of metabolic changes: NAA is most elevated in the subcortical WM and in the posterior third of the centra semiovalia and thalami which are the

areas affected earlier by the disease (Figure 46.5). In the same areas Cr is mildly elevated, while Cho is mildly decreased. In areas with normal signal intensity on T2-weighted MR images (anterior deep WM) NAA signal may be only slightly above normal levels. These results are in accordance with neuropathology studies that have shown spongiform changes developing first at the WM and GM junction, in the globus pallidus and thalamus. Vacuolating myelinopathy with intralamellar vacuoli is the histopathological features of Canavan’s disease. As the disease progresses, the deep WM also becomes involved following a

1.0

788

Alberto Bizzi, Marianna Bugiani and Ugo Danesi

Pyr

Glu

Glu

20 G

Pyr OAA

Asp  Acetyl CoA

NAA

Asp NAA ASPA Acetate Fig. 46.4 NAA pathway inside the mitochondria is illustrated in this cartoon. In the case with complete absence of NAA, Martin et al. (2001) have hypothesized a block of the transferase involved in the biosynthesis of NAA from aspartate and acetyl-CoA. Glu; Pyr: pyruvate; Asp: aspartate; OAA: oxaloacetic acid; 2OG: 2-oxoglutarate.

(a)

T2-weighted

Cho

NAA

(b) 2

Cho Cr

1

Cr

NAA

4.0

3.0

2.0

1.0

ppm Fig. 46.5 Canavan disease. Multivoxel H-MRSI (PRESS: TR/TE  1500/136 ms; 16 16 matrix; field of view (FOV)  160 160 mm3; slice thickness  20 mm) in a 2-year-old male diagnosed with Canavan disease. (a) Cho, Cr and NAA maps and two selected spectra (as indicated in the axial T2-weighted MRI image) at the level of the centrum semiovalis are illustrated. (b) NAA signal intensity is elevated in the posterior subcortical WM (voxel 1) that is hyperintense on T2-weighted MRI image; NAA is near normal levels in the deep WM (voxel 2) that is relatively spared by signal abnormality.

MR spectroscopy of inborn errors of metabolism

centripetal spread. The definitive diagnosis is made with determination of NAA in the urine and ASPA activity in cultured skin fibroblasts. NAA may be an essential component in a series of reactions required for the conversion of lignoceric acid to cerebronic acid, a component of myelin, and the formation of glutamic acid (Shigematsu et al., 1983). NAA is synthesized in GM; ASPA is primarily restricted to oligodendroglia and the expression of the ASPA gene in the postnatal brain closely parallels myelination in the CNS. Experimental data in the rat brain support a role of NAA in myelin synthesis, restricted to the CNS. The inability to hydrolyze NAA leads to lack of acetate and severe spongy degeneration in the WM. These findings provide strong additional support for insufficient myelin synthesis as the pathogenic basis of Canavan’s disease and make a compelling case for acetate supplementation as a simple and non-invasive therapy for what has to date been a fatal disease (Kirmani et al., 2003). Salla disease Salla disease is an autosomal-recessive disorder characterized by accumulation of free sialic acid (cf. Chapter 25) (Varho et al., 1999). Free sialic acid (N-acetylneuraminic acid, NANA) accumulates in lysosomes due to defective transport across the lysosomal membrane. Almost 100 patients with Salla disease have been diagnosed in Finland, and sporadic cases have been identified in other populations as well. The main feature of the disease is psychomotor retardation with ataxia, hypotonia, and transient nystagmus, later followed by progressive spasticity and athetosis. The majority of patients are able to walk and to speak words or short sentences; the most severely affected patients never become ambulant. Life expectancy is only slightly decreased. MRI has demonstrated diffuse signal abnormalities in the cerebral WM and an extremely thin corpus callosum (Varho et al., 1999). The BGT are normal in all patients. Histopathology has shown loss of axons and myelin sheaths with pronounced astrocytic reaction in WM. Recently, an abnormally elevated signal at 2.02 ppm, co-resonant with that from NAA, was reported in the affected parietal WM of patients with Salla disease. The signal increase may be due to accumulation

of free NANA (which shares the N-acetyl moiety corresponding to the singlet resonance) that offsets the probable loss of NAA inside the axons. NAA is therefore apparently increased. Cr signal was increased in the WM and basal ganglia, while Cho was decreased. The elevated Cr signal may be related to the presence of an astrocytic reaction or to increased glucose utilization, as shown by fluoro-2-deoxyglucose-positron emission tomography (FDG-PET). The low Cho signal may be due to a smaller pool of compounds involved in myelin membrane turnover. In the basal ganglia the signal intensities of the 2.02 ppm and Cho resonances were not different from control subjects. Maple syrup urine disease Maple syrup urine disease is a genetically heterogeneous, autosomal-recessive aminoacidopathy, resulting in severe impairment or death if untreated. It is caused by a deficiency of branched-chain alfa-keto acid (BCKA) dehydrogenase, a mitochondrial multienzyme complex catalyzing the oxidative decarboxylation of BCKA, which are produced after the transamination of essential branched-chain amino acids (BCAA) isoleucine, leucine and valine. The enzyme complex has at least six components, and a mutation in any of the six genetic loci which code for the subunits can result in dysfunction of the BCKA dehydrogenase complex. This enzyme deficiency leads to the accumulation of toxic levels of BCAA and BCKA in the body resulting in severe metabolic acidosis. Three clinical phenotypes have been described: classic, intermediate, and intermittent. The time of presentation seems to correlate with the degree of residual enzymatic activity. The classic form presents in the first week of life with poor feeding and vomiting, leading to lethargy, alternating periods of hypertonia and hypotonia, irregular respiration and apnea. Neurological findings may be mistaken for sepsis or meningitis; without a restricted diet, death occurs in the first year of life. The severity of longterm neurological deficits is strongly correlated with the duration of the acute toxic phase in the neonatal period. A characteristic maple syrup odor in urine may be the identifying sign in less severe disease. The intermediate form is less severe and presents later in the first year of life with lethargy, behavioral

789

790

Alberto Bizzi, Marianna Bugiani and Ugo Danesi

(a)

(b) 40

20

30

15

10 20 5 10 90

0

4.0 3.5 3.0 2.5 2.0 1.5 1.0 0.5 0.0 ppm

4.0 3.5 3.0 2.5 2.0 1.5 1.0 0.5 0.0 ppm

Fig. 46.6 Maple syrup disease. Neonate diagnosed with maple syrup disease at 9 days of life. Single-voxel H-MRS (PRESS: TR/TE  1600/270 ms) acquired from a voxel (15 15 15 mm3) placed in the lateral ventricles and bilateral anterior thalami during acute metabolic decompensation and after treatment. (a) At day 2 there is elevation of Lac (1.3 ppm) and of the methyl group of the BCAA/BCKA (0.9 ppm). (b) After therapy at day 12 the two abnormal peaks have disappeared; abnormal signal on T2-weighted persisted in the thalami (not shown).

disorders, ataxia, generalized hypotonia, and seizures. The intermittent form usually affects older children, and may be triggered by infection, other stress conditions or sudden increase in dietary protein. During acute metabolic decompensation, MRS shows the reversible presence of a abnormal peak at 0.9 ppm due to the methyl groups of BCAA and BCKA, associated with accumulation of lactate (Lac) and loss of NAA (Felber et al., 1993; Jan et al., 2003). A diffusion MR study performed in six patients during acute decompensation demonstrated marked restriction of diffusion (decreased apparent diffusion coefficient (ADC)) compatible with cytotoxic or intramyelinic sheath edema in pons, midbrain, pallidi, thalami, cerebellar, and periventricular WM (Jan et al., 2003). MRS changes (Figure 46.6) and diffusion-weighted images (DWIs) abnormalities were reversed with treatment in all six patients. The presence of Lac and decreased ADC are usually considered bad prognostic signs for tissue recovery in for example ischemic disease; in these cases, the reversible changes of Lac and NAA may be related to the compromise of mitochondrial function during metabolic decompensation. The severe intramyelinic edema indicates

that the acute toxic effect of BCAA and BCKA may be associated with a reversible disturbance of the fluid retention mechanisms of the myelin sheath (Harper et al., 1990). In untreated cases of classic maple syrup urine disease, CT and MRI shows a characteristic pattern with signal abnormalities and swelling of the WM structures which are undergoing myelination, and signal abnormalities in the globus pallidus and lateral thalamus. MRI of patients with the intermediate form may mimic Canavan’s disease with diffuse symmetrical hyperintensity in the subcortical WM, midbrain, globi pallidi, and most of the thalami (Uziel et al., 1988). Clinical history, laboratory findings and MRS will differentiate the two entities. A definite diagnosis is made by a positive 2,4-dinitrophenylhydrazine reaction in the urine. Prenatal diagnosis and heterozygote testing are possible in families with a known mutation. Dietary restriction of the BCAA and caloric supplementation may allow survival beyond the neonatal period, but often with residual neurological damage. Lifelong treatment by diet is feasible. Isoleucine, leucine, and valine are essential amino acids and

MR spectroscopy of inborn errors of metabolism

they must be maintained within a narrow range of values. MRS is also useful to monitor response during diet restriction. The use of a long TE is important to avoid contamination with the signal of lipids and macromolecules. At TE  136 ms the signal of BCAA and BCKA is inverted (Heindel et al., 1995). Phenylketonuria This aminoacidopathy is an autosomal-recessive disorder caused by a mutation in a gene mapped to chromosome 12q22–24.1. In most patients, phenylketonuria (PKU) is caused by deficiency of Phe hydroxylase, and in a minority of patients by deficiency of tetrahydrobiopterin. The deficiency of the Phe hydroxylating system results in the production of compounds that are toxic to the developing brain. Classic PKU presents in the first year of life with psychomotor retardation, eczema, irritability, vomiting, and insufficient growth. A mousy, musty urine and body odor is often a revealing sign. Early detection of PKU patients is now possible thanks to worldwide neonatal screening programs. Early treatment with low Phe diet prevents most of the neurological abnormalities. Treatment needs to be monitored regularly with assessment of serum Phe levels. MRI shows signs of delayed and defective myelination in the periatrial and periventricular WM. The frontal areas and the subcortical fibers are initially spared. Symmetrical WM abnormalities have also been demonstrated in patients who do not maintain restricted diets. Follow-up MRI after following a strict diet shows resolution of the abnormalities within a few months in some but not all patients. The early reversible MRI WM changes may represent intramyelinic vacuole formation, and the late changes may represent permanent myelin damage and loss. The WM changes are clinically correlated with lower limb spasticity, difficulty in processing visual information and changes in mood and personality. CT scan shows calcifications bilaterally in the globi pallidi and frontal subcortical regions. In few cases these deposits may show up as hyperintensities on T1-weighted MR images. In untreated PKU patients, MRS detects an elevated Phe signal at 7.36 ppm in the affected WM (Kreis et al., 1995). No significant changes were found for NAA, Cho, and Cr in GM or WM. Plasma Phe may not be a reliable indicator of brain Phe level in subjects

Phe

9.0

8.5 8.0 7.5 Frequency (ppm)

7.0

6.5

Fig. 46.7 Classic PKU. Two consecutive evaluations of brain Phe performed in a 19-year-old female showing an increase of the Phe peak (7.36 ppm) after diet discontinuation (solid line). Single-voxel H-MRS (stimulated echo acquisition mode (STEAM): TR/TE/mixing time (TM)  2010/30/13.7 ms; NEX  256) were acquired from the WM of the centrum semiovale and part of adjacent cortical GM, just above the lateral ventricle of one hemisphere. The Phe peak was very low before diet discontinuation (dotted line).

with PKU. MRS can be a useful tool in evaluating the individual vulnerability of PKU patients to different values of plasma Phe. A recent study performed in 10 off-diet PKU patients demonstrated elevated Phe in the brain of all patients (Leuzzi et al., 2000). In patients diagnosed late, brain Phe concentration correlated with clinical phenotype better than plasma Phe. The discrepancy between brain and plasma Phe was relevant from a clinical point of view in two late diagnosed cases: in one patient with normal mental development, a high level of plasma Phe was associated with a relatively low concentration of brain Phe; in the other patient with severe neurological impairment, a very high level of brain Phe was associated with plasma Phe compatible with a diagnosis of mild PKU. MRI showed WM abnormalities in all patients; and no correlation was found between WM alterations and concurrent brain Phe concentrations. MRS can therefore provide a marker of clinically relevant brain Phe for diagnosis and therapeutic monitoring (Figure 46.7).

791

792

Alberto Bizzi, Marianna Bugiani and Ugo Danesi

(a)

Cho Gly Cr

(c) NAA

Cho

NAA

Cr Gly

(b)

(d)

Cho NAA

Cho Cr

Cr

NAA

Gly

Gly

3.75

3.0

2.25 1.50 0.75 ppm

0

3.75

3.0

2.25 1.50 ppm

0.75

0

Fig. 46.8 NKH : non-ketotic hyperglycinemia. Neonate diagnosed with NKH at 10 days. Single-voxel H-MRS (PRESS: TR/TE  2000/136 ms; NEX  256) were acquired from the thalami and basal ganglia at 10(a), 31(b), 45(c) and 105(d) days of life. The location of the voxel (30 30 40 mm3) is indicated in the axial T1-weighted SE MR image showing brain development appropriate for 10 days of age. An abnormal elevation of glycine (Gly) at 3.55 ppm with a Gly/Cr ⬃ 1 is seen at time of diagnosis. Note the progressive decrease of Gly signal during treatment with a protein restriction diet. The clinical course was characterized by severe encephalopathy and respiratory failure.

Non-ketotic hyperglycinemia Non-ketotic hyperglycinemia (NKH) is a heterogeneous genetic autosomal-recessive disorder related to a defect in one of the enzymes of the glycine cleavage system. Elevated glycine levels are found in the urine, serum, CSF, and brain without significant ketoacidosis. Two clinical phenotypes have been described: the neonatal and the late-onset forms. Patients with the neonatal form are normal at birth, but within a few days show progressive neurological signs, including lethargy, hypotonia, convulsions, and apneic spells. Most patients die within a few weeks. In the late-onset form, patients are normal throughout the neonatal period, with subsequent developmental slowing. Widespread spongiosis of myelinated WM is the histopathological feature of NKH, with tracts undergoing active myelination during the neonatal period (optic nerves and tracts, cerebellar peduncles, and corticospinal tracts) being the most severely affected.

Corona radiata, posterior limb of internal capsulae (PLIC), and posterior columns are less affected. Tracts that have completed myelination before birth, such as the anterior and posterior spinal roots, are usually spared. Highly elevated glycine in the CSF and absence of ketoacidosis are the clues to the diagnosis. A large glycine peak at 3.55 ppm in the MR spectrum of two infants with NKH (Figure 46.8) has been reported (Heindel et al., 1993). The same author found that the glycine level in brain tissue better correlated with the clinical picture than glycine concentration in plasma and CSF. In this context, a long TE acquisition is necessary, to distinguish the glycine resonance from that of mI (at 3.56 ppm) which is normally high in neonates. Definite diagnosis is made testing the glycine cleavage system in transformed lymphoblasts. Prenatal diagnosis on chorionic villi is feasible. Molecular

MR spectroscopy of inborn errors of metabolism

genetic diagnosis is also possible when the mutation in the family is known. Succinate dehydrogenase deficiency Reduction of succinate dehydrogenase and complex II activities of the respiratory chain in muscle and cultured skin fibroblasts has been reported in at least nine children with leukoencephalopathy and LS, a ME (Burgeois et al., 1992; Brockmann et al., 2002; Bizzi et al., 2002; Moroni et al., 2002) (cf. Chapter 45). If so far the diagnosis of succinate dehydrogenase has been elusive, things may change with the increasing use of MRS in children with leukoencephalopathy. In 1997 Hanefeld described the first case of a child with an additional resonance at 2.42 ppm in the abnormal WM (Frahm and Hanefeld, 1997). This peak has been assigned to the two methylene groups of succinate. More recently, three additional children with succinate and Lac accumulation in the WM were diagnosed (Brockmann et al., 2002; Bizzi et al., 2002). In our case, multivoxel spectroscopy confirmed that accumulation of succinate and Lac were restricted to the WM and were associated with marked bilateral NAA signal loss, and mild Cho and Cr depletion. In the adjacent cortical GM the resonances of the major metabolites appeared unaltered. In spite of normal levels of Lac in plasma and CSF in these cases, MRSI strongly suggested the correct diagnosis of mitochondrial leukoencephalopathy due to succinate dehydrogenase deficiency. Succinate dehydrogenase consists of only four subunits, all nuclearly encoded, and is part of both the respiratory chain and the Krebs cycle. succinate dehydrogenase catalyzes the formation of fumarate from succinate in tricyclic acid (TCA) cycle; complex II transfers hydrogen ions from flavin adenosine dinucleotide (FADH) to the Coenzyme Q in the respiratory chain. succinate dehydrogenase and complex II share the same two subunits: the iron–sulphur protein subunit (Ip) and the flavoprotein subunit (Fp). Decreased succinate dehydrogenase activity may be related to a defect of either the Ip or the Fp subunit of complex II. Mutations in the Fp subunit have been reported in three patients with leukoencephalopathy (Bourgeron et al., 1995; Parfait et al., 2000). It is important to emphasize that not all cases with succinate dehydrogenase deficiency and leukoencephalopathy show elevated succinate on MRS. In

fact succinate was detected in four of the seven cases with succinate dehydogenase/complex II deficiency studied by MRS (the two cases reported by Burgeois did not have MRS). The reason why some cases do not accumulate succinate is not yet understood; it may be due to a difference in the enzyme subunits involved. Lac accumulation was more common, and was present in six of these seven cases. The accumulation of succinate is, of course, more specific than Lac because it signifies a block at a specific step in Krebs cycle. ME with succinate dehydrogenase deficiency may present acutely in the first few years of life in a child with normal psychomotor development. A progressive motor deterioration with loss of all acquired motor skills may occur in a few weeks. Spasticity, hyperreflexia and a positive bilateral Babinski’ sign dominate the clinical picture. In the early phase Lac and pyruvate may be normal in plasma and in the CSF as well. In the few reported patients with succinate dehydrogenase deficiency, MRI demonstrated diffuse symmetric signal changes in the cerebral WM, often with characteristic cavitations in the periventricular zones. Involvement of the corpus callosum, PLIC, cerebellar WM and other myelinated long tracts has been observed in most cases. The U-fibers, the cortical GM and the deep gray nuclei are spared. It is important to remark that ME cases with leukoencephalopathy do not have the characteristic MRI features of LS, namely T2-weighted signal abnormalities in the basal ganglia and brainstem. Symmetrical abnormalities in the subthalamic nuclei, with abnormalities at different levels of the brainstem (medulla, pontine tegmentum, periaqueductal area) and lack of basal ganglia involvement, may also be observed in LS; this is a rather frequent feature of LS patients with SURF-1 mutation (Farina et al., 2002). Histopathology has shown extensive spongiform degeneration, astrocytosis, microglial reaction, intense vascular proliferation, and cystic cavitations in the cerebral WM of one patient with succinate dehydrogenase deficiency (Brockmann et al., 2002). Defect in polyhydric alcohols metabolism Polyhydric alcohols (polyols) are particularly abundant in the human brain and are formed by reduction of sugars. Abnormal levels of polyols have

793

794

Alberto Bizzi, Marianna Bugiani and Ugo Danesi

(a)

(b)

Fig. 46.9 Leukoencephalopathy associated with disturbance in the metabolism of polyols. (a) Axial T2-weighted MR images showing extensive signal abnormalities of the cerebral hemispheric WM with partial sparing of the frontal periventricular areas in a 11-yearold boy. Note the slightly swollen aspect of WM with broadening of gyri (open arrows). (b) Single-voxel H-MRS (STEAM: TR/TE/TM  6000/20/10 ms; NEX  128) acquired from parietal WM (1 and 2) and parietal cortex (3 and 4) of the patient (1 and 3) are compared with those of a normal control (2 and 4). In the patient there are several additional abnormal peaks between 3.5 and 3.8 ppm. STEAM spectra of pure solutions of 50 mM arabitol (5) and 50 mM ribitol (6), acquired with the same parameters of the in vivo studies, showed the same resonance pattern of chemical shifts and J-coupling effects between 3.5 and 3.8 ppm. Ara: arabitol; Rib: ribitol.

been described in hepatic encephalopathy, hypergalactosemia, Alzheimer, and Down’s syndrome. More recently, the case of a 14-year-old boy with leukoencephalopathy and neuropathy of unknown origin with highly elevated levels of arabitol and ribitol in the brain was demonstrated by MRS (van der Knaap et al., 1999). Elevations of arabitol and ribitol were higher in CSF than in plasma, and thousands of times higher than in the CSF of control subjects. In contrast, CSF mI was decreased. These findings suggested a presumed inborn error in polyol metabolism. In the WM and GM of the parietal lobe single-voxel MRS showed elevated abnormal resonances between 3.5 and 3.8 ppm that were detectable at short (Figure 46.9) and long TEs. The resonances matched those from spectra of pure solutions of 50 mM arabitol and ribitol. Severe reduction

in signal intensities of Cho, Cr, and NAA in the WM were also seen. At 11 and 14 years of age, MRI revealed stable extensive abnormalities of the cerebral WM with a slightly swollen appearance, prominent involvement of the U-fibers, relative sparing of periventricular WM, and complete sparing of corpus callosum and internal capsule. The leukoencephalopathy of this patient is likely due to polyol toxicity. Little is known about the metabolism of polyols in humans. The concentration of most polyols is higher in CSF than in blood, suggesting that polyol pathways are particularly active in CNS, where they may have their most important function. Elevated levels of polyols have been found in diabetes mellitus and galactose intoxication, which are associated with a peripheral neuropathy characterized by myelin vacuolization. The slight swelling

MR spectroscopy of inborn errors of metabolism

Ara/Rib

NAA

(1)

NAA

Cho Cr

Cr

mI

(2)

Cr Cho

Lac

4.0

3.5

3.0

2.5

2.0

1.5

1.0

4.0

3.5

3.0

2.5

2.0

1.5

1.0 (4)

(3)

scyllo Ins

4.0

3.5

3.0

2.5

2.0

1.5

1.0

4.0

3.5

3.0

2.5

2.0

1.5

(5)

4.0

3.5

3.0

2.5

2.0

1.5

1.0

1.0

(6)

4.0

3.5

3.0

ppm

2.5

2.0

1.5

1.0

ppm

Fig. 46.9 (cont.)

of the WM seen on MRI may also be due to myelin vacuolization. A presumed defect in polyol metabolism should be considered when MRS reveals an abnormal elevated resonance between 3.5 and 3.8 ppm in a patient with leukoencephalopathy and peripheral neuropathy of unknown origin.

Diseases with suggestive MRS findings In this category we have grouped those metabolic diseases showing spectroscopic changes that may help clinicians to narrow their differential diagnosis, despite the fact that they are non-specific and they are common to different diseases sharing a similar pathological correlate. Only the general spectroscopic feature

795

796

Alberto Bizzi, Marianna Bugiani and Ugo Danesi

of each group will be briefly described to avoid overlapping. Please refer to Chapter 45 for a further description of the hereditary leukoencephalopathies. Mitochondrial encephalopathy Mitochondria are membranous organelles that are responsible for providing and storing most of the energy required by the cell in the form of high-energy bond of ATP. This process is called OXPHOS and is the coordinated result of multiple components. Mitochondrial function is dependent upon the coordinated expression of two separate genomes: nuclear DNA (nDNA) present in two copies in each cell and mtDNA, present in 2–10 copies per mitochondrion. Inborn errors of mitochondrial function constitute an heterogeneous group of disorders caused by a defect in either nuclear or mitochondrial genome. mtDNA is particularly sensitive to alterations. A rapid rate of spontaneous mutations and the presence of a poor repair system are probably important to explain the extraordinary heterogeneity of the clinical features of mitochondrial encephalopathies (MEs). MRI has proven useful in the diagnostic work-up of ME and specific neuroradiological features have been described by several authors (Barkovich et al., 1993; Savoiardo et al., 1995; Valanne et al., 1998; Farina et al., 2002) in the following syndromes: LS, MELAS, KSS, LHON, MERRF syndrome. ME should be considered in any infant or child who has abnormalities of the deep GM, in particular if WM is also involved. Disorders of organic acid metabolism lead to abnormal mitochondrial function, therefore, it is important to remark that organic acidemias and ME are not necessarily separate and distinct entities. Since its infancy, MRS has been indicated as an important and practical tool in the diagnostic evaluation of ME. The detection of Lac has been used as a signature for mitochondrial disease. Several studies have reported accumulation of Lac in patients with diagnosed ME (Barkovich et al., 1993; Wilichowski et al., 1999; Moroni et al., 2002). Cross studied six children with pyruvate dehydrogenase (PDH) deficiency and Lac was demonstrated in the brain of all patients (Cross et al., 1994). Regional variations in the Lac signal were observed in those patients in whom two regions were examined. In another study, Cross performed MRS on 24 patients

with ME and showed good concordance between observations of brain Lac with MRS and measurements of Lac in CSF (Cross et al., 1993). KragelohMann studied five children with LS with bilateral lesions of the putamina and caudate heads. Serum Lac was abnormal in four children, and CSF Lac slightly elevated in three of the five children. Singlevoxel MRS of the basal ganglia revealed elevated Lac, giving further evidence for a defect of energy metabolism in the brain (Krageloh-Mann et al., 1993). The authors concluded that a MRS study is useful to support a possible diagnosis of ME in the absence of peripheral Lac elevation. The frequency of detection of Lac in children referred for suspected ME has been recently addressed by Lin (Lin et al., 2003). In this retrospective study of 29 patients, Lac was detected in five of eight patients with a diagnosis of ME confirmed by genetic, biochemical, or pathological features; Lac was elevated in four of 16 patients with a final diagnosis of probable ME. Lac was not detected in any of the five patients in which the diagnosis of ME was excluded (cf. Case Study 46.1). Clinically, serum Lac level is only a fair marker of mitochondrial disease. In some cases lactic acidosis might be even artifactual due to difficulties in drawing blood from a poorly cooperating child. In their analysis of 36 children with highly suspected ME, Nissenkorn et al. found that 21 patients had significantly elevated serum lactic acid (Nissenkorn et al., 2000); serum Lac was normal in 15 patients, including three children with LS. In a study on 25 children with confirmed diagnosis of ME Lac elevation was detected in 17 patients in at least one of three compartments (serum, CSF, or brain tissue) (Bizzi et al., 2002). The incidence of lactic acidosis in the brain was less than 50% and it was about the same in each of the three compartments. In four patients Lac elevation was found only in the brain with MRS (Figure 46.10). These studies emphasize the value of adding MRS to the imaging work-up of patients with suspected ME. It is also important to keep in mind that increased Lac occurs in other neurological disorders such as hypoxia, ischemia, genetic leukodystrophies, abscess, inflammation, and neoplasms. In most cases, however, the clinical presentation, MRI and other MRS findings will narrow significantly the

MR spectroscopy of inborn errors of metabolism

(2)

2

3

4

3.0

2.0

1.0

ppm 1 (3)

3.0 (1)

2.0

1.0

ppm (4)

3.0

2.0 ppm

1.0

3.0

2.0 ppm

1.0

Fig. 46.10 Mitochondrial leukoencephalomyopathy. Multivoxel H-MRSI (PRESS: TR/TE  1500/ 136 ms; 16 16) performed at the level of the centrum semiovalis. The position of selected spectra (nominal size 10 10 20 mm3) is illustrated on the axial T2-weighted MR image showing diffuse symmetrical hyperintensity and volume loss in the WM. Spectra (2, 3 and 4) show mild Lac accumulation in the WM, associated with moderate NAA and mild Cho and Cr signal losses in this 13-year-old female patient with normal Lac levels in plasma. Spectrum in the adjacent cortical GM (1) shows only mild NAA signal loss. Biochemical analyses confirmed the diagnosis of ME due to a defect of cytochrome-c-oxidase (complex IV).

differential diagnosis. The association of mild Lac with elevated Cho will suggest the diagnosis of a demyelinating leukoencephalopathy (ALD, MLD, Krabbe, Alexander’s disease (AD)) or of neoplasm. MRI will often help to distinguish these. Hereditary leukoencephalopathy with depletion of the main metabolites secondary to cavitations Three diseases share the following spectroscopic findings: depletion of all main metabolite signals in the WM and near normal spectra in the GM. They are leukoencephalopathy with vanishing white matter disease (LVWD), Megalencephalic leukoencephalopathy with subcortical cysts (MLC) and ME with predominant WM signal abnormalities on MRI. All

three diseases may show Lac in the WM. The highest Lac accumulation is usually a suggestive sign of ME. These entities have been already described in detail in a previous chapter. Hereditary leukoencephalopathies with Cho elevation as a sign of active demyelination The detection of an increased Cho signal suggests instability of cellular membranes and increased free Cho and intermediate Cho-compounds (phosphocholine (PC) and glycerophosphocholine (GPC)). MRSI is very sensitive to detect phospholipid alterations in patients with demyelinating diseases. Whether the increase of Cho is due to destruction of myelin or build up of new myelin is still unknown.

797

798

Alberto Bizzi, Marianna Bugiani and Ugo Danesi

Moderate NAA and Cr signal losses are signs of severe axonal loss in the WM; on the contrary, near normal spectra are found in the adjacent cortical GM indicating relative sparing of the cortex at least in the early phase of the disease. A mildly elevated Lac signal is often associated to these findings. These WM spectroscopic abnormalities are frequently symmetric and may vary in severity along the anterior–posterior axis. Disorders sharing the spectroscopic findings described above are MLD, KD, X-ALD, and Alexander disease. Metabolic diseases with non-specific MRS findings Hereditary leukoencephalopathies with near normal appearing proton spectra despite diffuse signal abnormalities on conventional MRI. Pelizaeus-Merzbacher disease Pelizaeus-Merzbacher disease (PMD) is caused by alteration in the gene encoding the proteolipid protein (PLP) and it is characterized by a failure to form and maintain myelin. Very mild and inconsistent metabolic changes have been reported in the few patients studied with MRS. Some authors have shown normal NAA/Cr values with decreased Cho/Cr ratios (Takanashi et al., 1997; Spalice et al., 2000), while Bonavita et al. (2001) has shown decreased NAA/Cr with normal Cho/Cr ratios. The only study that used absolute quantification has shown that NAA, Cr, and mI are elevated, while there are no statistical changes in Cho (Takanashi et al., 2002). In our series of patients with a genetically confirmed diagnosis of PMD we found spectra with a normal profile throughout the WM and GM in the slice of interest at the level of the centra semiovalia. There were no significant spectroscopic changes between the WM and the GM, despite diffuse extensive signal hyperintensity in the WM on T2-weighted MR images (Figure 46.11). Near normal spectral profile may be found in cases with leukoencephalopathy without documented mutations in the PLP gene. The normal spectroscopic findings, in association with the MRI findings, suggest that these undefined cases are likely to be hypomyelinating diseases of unknown etiology.

Diseases with focal NAA signal loss Metabolic diseases with non-specific MR spectroscopic findings include diseases with focal NAA signal loss.

Other amino and organic acidopathies L-2-OHG aciduria L-2-hydroxyglutaric

(L-2-OHG) aciduria is an inherited autosomal-recessive disease with megalencephaly. Laboratory investigations reveal elevated urinary excretion of L-2-OHG acid; the basic defect is not known. Conventional MRI shows asymmetric bilateral focal T2-weighted signal changes beginning in the cerebral subcortical WM, basal ganglia and dentate nuclei. Cerebellar atrophy is often present. Although similar appearances may be seen in other metabolic disorders, the distribution of signal abnormalities in L-2-OHG aciduria is highly characteristic and may suggest the correct diagnosis (D’Incerti et al., 1998). MRS reports on L-2-OHG aciduria are rare. Moderate decrease of NAA and Cho signals with moderate increase of mI, with no elevation of Lac has been reported (Hanefeld et al., 1994). We have seen three patients with only scattered mild NAA signal loss and mild Cr signal increase in the subcortical affected WM; Cho signal was nearly normal throughout the WM (Figure 46.12).

3-Hydroxy-3-methylglutaryl-coenzyme A lyase deficiency 3-Hydroxy-3-methylglutaryl-coenzyme A lyase (HMGCoA lyase) deficiency is an inborn error of leucine catabolism characterized by recurrent severe metabolic attacks of hypoglycemia, metabolic acidosis, hepatomegaly, lethargy or coma and apnea in the neonatal period. The enzyme HMG-CoA lyase catalyzes the final step of leucine degradation in the mitochondrial matrix, converting HMG-CoA to acetyl-CoA, and acetoacetic acid. Diagnosis is suggested by a particular pattern of organic acids in urine with large amounts of 3-hydroxy-3-methylglutaric, 3-methyl-glutaconic, 3-methylglutaric, and 3-hydroxyisovaleric acids. Lactic acid may be present in sizable amounts at times of acute illness. The diagnosis is confirmed by demonstration of deficient

MR spectroscopy of inborn errors of metabolism

3.0

2.0 ppm

1.0

3.0

2.0 ppm

1.0

3.0

2.0 ppm

1.0

3.0

2.0 ppm

1.0

3.0

2.0

799

1.0

ppm

3.0

2.0 ppm

1.0

3.0

2.0 ppm

1.0

3.0

2.0 ppm

1.0

3.0

2.0 ppm

1.0

Fig. 46.11 Multivoxel H-MRSI (PRESS: TR/TE  1500/136 ms; 16 16) performed at the level of the centrum semiovalis. The position of selected spectra (nominal size 10 10 20 mm3) is illustrated on the axial T2-weighted MR image showing diffuse symmetrical hyperintensity in the WM. All spectra have a normal spectral profile; there is no difference between spectra in the WM compared with those in the adjacent GM. In this 2-year-old male the diagnosis of PMD was confirmed with genetic studies.

activity of the enzyme in leukocytes and cultured fibroblasts. On MR images there is a diffuse signal abnormality of the cerebral WM with superimposed, more marked patchy areas of T2-dependent signal hyperintensity. Most patients are clinically normal or have intelligence just below average. There is a striking, consistent discrepancy between the extensive WM abnormalities on MRI and the lack of clinical findings. MRS findings in three patients with HMG-CoA lyase deficiency have shown mild to moderate diffuse NAA and Cr signal loss in the WM

with relative preservation of Cho (van der Knaap et al., 1998). In patients with signs of cognitive impairment, bilateral signal hyperintensity on T2-weighted images may be seen in the occipital WM and GM, the globi pallidi, and the dorsolateral part of the thalami. These are signs of hypoglycemic injury in regions that are the most vulnerable during the neonatal period. History may reveal that these patients have suffered episodes of life-threatening metabolic decompensation with severe hypoglycemia in the neonatal period.

800

Alberto Bizzi, Marianna Bugiani and Ugo Danesi

3.0

2.0 ppm

1.0

3.0

2.0 ppm

1.0

3.0

2.0 ppm

1.0

2.0 ppm

1.0

3.0

3.0

2.0 ppm

1.0

3.0

2.0 ppm

1.0

3.0

2.0 ppm

1.0

3.0

2.0 ppm

1.0

3.0

2.0 ppm

1.0

Fig. 46.12 L-2-OHG aciduria. Multivoxel H-MRSI (PRESS: TR/TE  1500/136 ms; 16 16; NEX  2) performed at the level of the centrum semiovalis. The position of selected spectra (nominal size 7.5 7.5 20 mm3) is illustrated on the axial T2-weighted MR image showing symmetrical focal hyperintensity in the subcortical WM of the anterior frontal and parietal lobes. The combination of these signal abnormalities in the WM, associated with those in dentate nuclei, globi pallidi, and thalami should suggest the diagnosis of a metabolic disease. Unfortunately, the spectroscopic findings are not specific in L-2-OHG aciduria. In this 25-year-old female patient spectra from the subcortical WM show mild NAA signal loss, associated with mild Cr signal increase; Cho signal is within normal throughout the WM. The diagnosis of L-2-OHG aciduria was confirmed biochemically demonstrating increased concentration of L-2-OHG in urine, blood and CSF.

REFERENCES Baird P, Anderson T, et al. 1988. Genetic disorders in children and young adults: a population study. Am J Hum Genet 42: 677. Barker PB, Bryan RN, et al. 1992. Proton NMR spectroscopy of Canavan’s disease. Neuropediatrics 23(5): 263–267. Barkovich AJ, Good WV, et al. 1993. Mitochondrial disorders: analysis of their clinical and imaging characteristics. Am J Neuroradiol 14(5): 1119–1137.

Battini R, Leuzzi V, et al. 2002. Creatine depletion in a new case with AGAT deficiency: clinical and genetic study in a large pedigree. Mol Genet Metab 77(4): 326–331. Bhakoo KK, Pearce D. 2000. In vitro expression of N-acetylaspartate by oligodendrocytes: implications for proton magnetic resonance spectroscopy signal in vivo. J Neurochem 74(1): 254–262. Bianchi MC, Tosetti M, et al. 2000. Reversible brain creatine deficiency in two sisters with normal blood creatine level. Ann Neurol 47(4): 511–513.

MR spectroscopy of inborn errors of metabolism

Bizzi A, Bugiani M, et al. 2002. X-linked creatine deficiency syndrome: a novel mutation in creatine transporter gene SLC6A8. Ann Neurol 52(2): 227–231. Bizzi A, Danesi U, et al. 2002. Incidence of cerebral lactic acidosis in children with mitochondrial encephalomyopathy. Int Soc Magn Res Med (ISMRM), Honolulu, Hawai’I, USA. Bizzi A, Ulug AM, et al. 2001. Quantitative proton MR spectroscopic imaging in acute disseminated encephalomyelitis. Am J Neuroradiol 22(6): 1125–1130. Bonavita S, Schiffmann R, et al. 2001. Evidence for neuroaxonal injury in patients with proteolipid protein gene mutations. Neurology 56(6): 785–788. Bourgeron T, Rustin P, et al. 1995. Mutation of a nuclear succinate dehydrogenase gene results in mitochondrial respiratory chain deficiency. Nat Genet 11(2): 144–149. Brenner M, Johnson A, et al. 2001. Mutations in GFAP, encoding glial fibrillary acidic protein, are associated with Alexander disease. Nat genet 27: 117–120. Brockmann K, Bjornstad A, et al. 2002. Succinate in dystrophic white matter: a proton magnetic resonance spectroscopy finding characteristic for complex II deficiency. Ann Neurol 52(1): 38–46. Burgeois M, Goutieres F, et al. 1992. Deficiency in complex II of the respiratory chain, presenting as a leukodystrophy in two sisters with Leigh syndrome. Brain Dev 14(6): 404–408. Burlina A, Ferrari V, et al. 1997. Mast cells contain large quantities of secretagogue-sensitive N-acetylaspartate. J Neurochem 69(3): 1314–1317. Canavan MM. 1931. Schilder’s encephalitis periaxialis diffusa. Arch Neurol Psychiatr 25: 299. Cecil KM, DeGrauw TJ, et al. 2003. Magnetic resonance spectroscopy in a 9-day-old heterozygous female child with creatine transporter deficiency. J Comput Assist Tomogr 27(1): 44–47. Cecil KM, Salomons GS, et al. 2001. Irreversible brain creatine deficiency with elevated serum and urine creatine: a creatine transporter defect? Ann Neurol 49(3): 401–404. Clark JB. 1998. N-acetylaspartate: a marker for neuronal loss or mitochondrial dysfunction. Dev Neurosci 20(4–5): 271–276. Cross JH, Connelly A, et al. 1994. Clinical diversity of pyruvate dehydrogenase deficiency. Pediatr Neurol 10(4): 276–283. Cross JH, Gadian DG, et al. 1993. Proton magnetic resonance spectroscopy studies in lactic acidosis and mitochondrial disorders. J Inherit Metab Dis 16(4): 800–811. D’Incerti L, Farina L, et al. 1998. L-2-Hydroxyglutaric aciduria: MRI in seven cases. Neuroradiology 40(11): 727–733. De Stefano N, Matthews PM, et al. 1995. Reversible decreases in N-acetylaspartate after acute brain injury. Magn Reson Med 34(5): 721–727.

deGrauw TJ, Salomons GS, et al. 2002. Congenital creatine transporter deficiency. Neuropediatrics 33(5): 232–238. Dyken P, Krawiecki N. 1983. Neurodegenerative diseases of infancy and childhood. Ann Neurol 13(4): 351–364. Ebisu T, Rooney WD, et al. 1994. N-acetylaspartate as an in vivo marker of neuronal viability in kainate-induced status epilepticus: 1H magnetic resonance spectroscopic imaging. J Cereb Blood Flow Metab 14(3): 373–382. Farina LC, Uziel L, Bugiani G, Zeviani M, Savoiardo M. 2002. MR findings in Leigh syndrome with COX deficiency and SURF-1 mutations. Am J Neuroradiol 23(7): 1095–1100. Felber SR, Sperl W, et al. 1993. Maple syrup urine disease: metabolic decompensation monitored by proton magnetic resonance imaging and spectroscopy. Ann Neurol 33(4): 396–401. Frahm J, Hanefeld F. 1997. Localized proton magnetic resonance spectroscopy of brain disorders in childhood. Magnetic Resonance Spectroscopy and Imaging in Neurochemistry, Vol. 8, Bachelard H. (Ed.). Plenum Press, New York, pp. 329–402. Gambetti P, Mellman WJ, et al. 1969. Familial spongy degeneration of the central nervous system (Van Bogaert–Bertrand disease). An ultrastructural study. Acta Neuropathol (Berl) 12(2): 103–115. Gilliam TC, Brzustowicz LM, et al. 1990. Genetic homogeneity between acute and chronic forms of spinal muscular atrophy. Nature 345(6278): 823–825. Grodd W, Krageloh-Mann I, et al. 1990. In vivo assessment of N-acetylaspartate in brain in spongy degeneration (Canavan disease) by proton spectroscopy. Lancet 336: 437. Hanefeld F, Kruse B, et al. 1994. In vivo proton magnetic resonance spectroscopy of the brain in a patient with L-2-hydroxyglutaric acidemia. Pediatr Res 35(5): 614–616. Harper PA, Healy PJ, et al. 1990. Maple syrup urine disease (branched chain ketoaciduria). Am J Pathol 136(6): 1445–1447. Heindel W, Kugel H, et al. 1993. Noninvasive detection of increased glycine content by proton MR spectroscopy in the brains of two infants with nonketotic hyperglycinemia. Am J Neuroradiol 14(3): 629–635. Heindel W, Kugel H, et al. 1995. Proton magnetic resonance spectroscopy reflects metabolic decompensation in maple syrup urine disease. Pediatr Radiol 25(4): 296–299. Jan W, Zimmerman RA, et al. 2003. MR diffusion imaging and MR spectroscopy of maple syrup urine disease during acute metabolic decompensation. Neuroradiology 45(6): 393–399. Kirmani BF, Jacobowitz DM, et al. 2003. Developmental increase of aspartoacylase in oligodendrocytes parallels CNS myelination. Brain Res Dev Brain Res 140(1): 105–115. Krageloh-Mann I, Grodd W, et al. 1993. Proton spectroscopy in five patients with Leigh’s disease and mitochondrial enzyme deficiency. Dev Med Child Neurol 35(9): 769–776.

801

802

Alberto Bizzi, Marianna Bugiani and Ugo Danesi

Kreis R, Pietz J, et al. 1995. Identification and quantitation of phenylalanine in the brain of patients with phenylketonuria by means of localized in vivo 1H magnetic-resonance spectroscopy. J Magn Reson B 107(3): 242–251. Leuzzi V, Bianchi MC, et al. 2000. Clinical significance of brain phenylalanine concentration assessed by in vivo proton magnetic resonance spectroscopy in phenylketonuria. J Inherit Metab Dis 23(6): 563–570. Lin DD, Crawford TO, et al. 2003. Proton MR spectroscopy in the diagnostic evaluation of suspected mitochondrial disease. Am J Neuroradiol 24(1): 33–41. Martin E, Capone A, et al. 2001. Absence of N-acetylaspartate in the human brain: impact on neurospectroscopy? Ann Neurol 49(4): 518–521. Matalon R, Michals K, et al. 1988. Aspartoacylase deficiency and N-acetylaspartic aciduria in patients with Canavan disease. Am J Med Genet 29(2): 463–471. Moroni I, Bugiani M, et al. 2002. Cerebral white matter involvement in children with mitochondrial encephalopathies. Neuropediatrics 33(2): 79–85. Nissenkorn A, Zeharia A, et al. 2000. Neurologic presentations of mitochondrial disorders. J Child Neurol 15(1): 44–48. Parfait B, Chretien D, et al. 2000. Compound heterozygous mutations in the flavoprotein gene of the respiratory chain complex II in a patient with Leigh syndrome. Hum Genet 106(2): 236–243. Savoiardo M, Ciceri E, et al. 1995. Symmetric lesions of the subthalamic nuclei in mitochondrial encephalopathies: an almost distinctive Mark of Leigh disease with COX deficiency. Am J Neuroradiol 16(8): 1746–1747. Schulze A, Bachert P, et al. 2003. Lack of creatine in muscle and brain in an adult with GAMT deficiency. Ann Neurol 53(2): 248–251. Shigematsu H, Okamura N, et al. 1983. Purification and characterization of the heat-stable factors essential for the conversion of lignoceric acid to cerebronic acid and glutamic acid: identification of N-acetyl-L-aspartic acid. J Neurochem 40(3): 814–820. Spalice A, Popolizio T, et al. 2000. Proton MR spectroscopy in connatal Pelizaeus–Merzbacher disease. Pediatr Radiol 30(3): 171–175.

Stockler S, Hanefeld F, et al. 1996. Creatine replacement therapy in guanidinoacetate methyltransferase deficiency, a novel inborn error of metabolism. Lancet 348: 789–790. Stockler S, Holzbach U, et al. 1994. Creatine deficiency in the brain: a new, treatable inborn error of metabolism. Pediatr Res 36(3): 409–413. Takanashi J, Inoue K, et al. 2002. Brain N-acetylaspartate is elevated in Pelizaeus–Merzbacher disease with PLP1 duplication. Neurology 58(2): 237–241. Takanashi J, Sugita K, et al. 1997. Proton MR spectroscopy in Pelizaeus–Merzbacher disease. Am J Neuroradiol 18: 533–535. Tallan H, Moore S, et al. 1956. N-acetyl-L-aspartic acid in brain. J Biol Chem 219: 257–264. Uziel G, Savoiardo M, et al. 1988. CT and MRI in maple syrup urine disease. Neurology 38(3): 486–488. Valanne L, Ketonen L, et al. 1998. Neuroradiologic findings in children with mitochondrial disorders. Am J Neuroradiol 19(2): 369–377. van der Knaap MS, Bakker HD, et al. 1998. MR imaging and proton spectroscopy in 3-hydroxy-3-methylglutaryl coenzyme A lyase deficiency. Am J Neuroradiol 19(2): 378–382. van der Knaap M, Naidu S, et al. 2001. Alexander disease: diagnosis with MR imaging. Am J Neuroradiol 22: 541–552. van der Knaap MS, Wevers RA, et al. 1999. Leukoencephalopathy associated with a disturbance in the metabolism of polyols. Ann Neurol 46(6): 925–928. Varho T, Komu M, et al. 1999. A new metabolite contributing to N-acetyl signal in 1H MRS of the brain in Salla disease. Neurology 52(8): 1668–1672. Wilichowski E, Pouwels PJ, et al. 1999. Quantitative proton magnetic resonance spectroscopy of cerebral metabolic disturbances in patients with MELAS. Neuropediatrics 30(5): 256–263. Wyss M, Kaddurah-Daouk R. 2000. Creatine and creatinine metabolism. Physiol Rev 80(3): 1107–1213.

MR spectroscopy of inborn errors of metabolism

Case Study 46.1 Mitochondrial encephalopathy, lactic acidosis and stroke like episodes (MELAS) Peter Barker, D.Phil., Johns Hopkins University School of Medicine, Baltimore, USA History

T2

31-year-old male with MELAS, documented by a point mutation in the mitochondrial tRNA.

Cho

Technique b

Conventional MRI and multi-slice MRSI (TE  280ms).

a Imaging findings T2 MRI shows an stroke-like lesion in the left parietal lobe. The lesion (a) has very high Lac and almost absent NAA. Other regions, particularly GM (b) and CSF, also show an elevation of Lac above normal.

NAA

Lac

Discussion MELAS is usually characterized by elevated brain and/or CSF Lac levels, as are other mitochondrial diseases (Lin et al., 2003). Expression of Lac occurs as a function of phenotype and stage of disease. Stroke-like lesions in MELAS have low NAA, which can sometimes recover. (a)

(b) Lac

Key points MELAS shows elevated Lac in lesions, GM and CSF.

NAA

Cho Detection of brain and/or CSF Lac may help make a diagnosis of mitochondrial disease (and brain involvement). Lac is not always specific for mitochondrial disease.

Cho Cr

Cr NAA

3.0 2.0 ppm

1.0

3.0 2.0 ppm

Lac

1.0

Reference Lin DD, Crawford TO, Barker PB. 2003. Proton MR spectroscopy in the diagnostic evaluation of suspected mitochondrial disease. Am J Neuroradiol 24(1): 33–41.

803

Index

Note: page numbers in italics refer to figures and tables abscess, brain 354–6 capsule 354–5, 414 central cavity hyperintensity 409, 411–12 development 409, 410 diagnosis 413 diffusion-weighted imaging 409, 410, 411–14 pediatric patients 662, 664, 665 empyema 357 fungal 394–6 MRS 381–2, 383–4, 385 organisms 382, 383, 384 pediatric patients 662, 664, 665 subdural empyema spread 357 tuberculous 382, 388, 390, 471 acetate, hydatid cysts 398, 401 acetazolamide 215, 216, 254–5 N-acetyl aspartate (NAA) 4, 594–5 absence 10, 11 adrenoleukodystrophy 760–1 aging 579–80 AIDS dementia complex 462 alcohol abuse/dependence 548 Alexander disease 768, 769 Alzheimer’s disease 581, 582, 583, 585–6, 588, 593 border-zone infarcts 242 brain developing 20, 22 normal maturation 676, 677 brain tumor MRS 283, 290, 292, 299 glioma 306 pediatric 729–30, 731–2 radiation necrosis 301 recurrent astrocytoma 310 therapy response assessment 299–300 brain tumor proton MRS 289 Canavan’s disease 651, 654, 763–5, 786, 787, 788, 789 cerebral ischemia 172 cerebrotendinous xanthomatosis 438 cerebrovascular reactivity 241 cortico-basal degeneration 599 Creutzfeldt–Jakob disease 527 cryptococcoma 471 decrease 11–12 deficiency 786, 787, 788 dementia 525, 583

dementia with Lewy bodies 583 demyelinating diseases 373, 374 developmental delay 684 diseases with focal signal loss 798 encephalitis 367 epilepsy 484, 491, 493, 497 extra-temporal lobe 497, 498 focal 495 temporal lobe 494 Epstein–Barr virus 392 fronto-temporal degeneration 583, 584 gray matter signals 20, 21, 22 hereditary leukoencephalopathies 798 herpes simplex encephalitis 391 HIV dementia 461, 462 abnormality reversal 462–3 pediatric patients 463–4 HSV1 368, 369 3-hydroxy-3-methylglutaryl-coenzyme A lyase deficiency 799 hypoxic brain injury 693, 694, 695, 699 hypoxic–ischemic cascade 691 hypoxic–ischemic encephalopathy 692 internal carotid artery occlusion/stenosis 236–8 Japanese encephalitis 393 Krabbe’s disease 758 lymphoma 467, 469 malaria 403 maple syrup urine disease 790 megalencephalic leukoencephalopathy with subcortical cysts 771, 772, 773 meningitis 381 metabolic activity indication 619 metabolic rate 619 mitochondrial levels 618 MRS in stroke 165 measurement 168–9, 170, 171 mucormycosis 396 multiple sclerosis 431, 432, 433–4, 435, 451 multi-system atrophy 597 myelinopathia centralis diffusa/vanishing white matter disease 770, 771 neural tissue integrity surrogate marker 786 neurocysticercosis 398, 399 neuronal loss relationship 10–12, 786 neuronal number indication 619

L-2-OHG aciduria

798, 800 oligodendrocytes 595 panic disorder 542 para-infectious encephalopathy 393, 394 Parkinson’s disease 596 pediatric imaging 647, 648, 739–40 brain infarction 649 metabolic stroke 749–50 neoplasms 649, 650, 652–3 Pelizaeus–Merzbacher disease 766, 767, 798 post-traumatic stress disorder 546 primary neoplastic lesion differentiation 297 progressive multifocal leukoencephalopathy 469, 478 progressive supranuclear palsy 598 proton MRS 9, 10–12 neoplastic/non-neoplastic differentiation 296 signal 294 tumefactive multiple sclerosis 308 proton spectra peaks 30 Reye’s syndrome 705 Salla disease 439, 789 schizophrenia 548, 567 severe obstructive carotid artery disease 239–40, 242–3 spectroscopy 611 stroke 173, 174–5, 181 outcome measures 176 patient selection for thrombolysis 175 subacute sclerosing panencephalitis 392 transient ischaemic attacks 166 traumatic brain injury 615–19, 620, 621, 622 vascular dementia 583 white matter normal appearing 431, 432, 433–4 signals 20, 21, 22 N-acetylneuraminic acid (NAN) 439 acoustic neuroma 343 extravascular extracellular space 343–4 acute disseminated encephalomyelitis 373, 375, 393–4 diffusion-weighted imaging 446–7, 450 MRS 437, 442 normal appearing white matter 450

805

806

Index

adenosine triphosphate 594, 675 hypoxic brain injury 690 adrenoleukodystrophy 762 diffusion tensor imaging 761 MRS 758, 760–1, 778 X-linked 666–7, 668, 683, 758, 761 advanced trauma and life support (ATLS) 609 affective disorders 535, 536–7, 538, 539–41, 542 aging apparent diffusion coefficient 560–1 corpus callosum 561, 562, 563 depression 571 diffusion tensor imaging 560–4 frontal lobe 561 genu 561 higher magnetic field strength MRS 587 lentiform nucleus 561 normal 558, 560–4 proton MRS 579–80 splenium 561, 562–3 white matter 561–4, 562, 563 AIDS dementia complex 372, 420, 462 AIDS-defining illness 570 see also HIV infection alanine brain tumor proton MRS 289 hydatid cysts 398, 401 malaria 403 neurocysticercosis 398, 399 alcohol abuse/dependence 546, 548, 549–50 long-term use 548 alcoholism centrum semiovale 569 cognitive function 568 corpus callosum 569 diffusion tensor imaging 568–70 fluid attenuated inversion recovery 570 fractional anisotropy 568–70 motor function 568 white matter 568, 569–70 Alexander disease 766–8, 769, 781–2 Alzheimer’s disease 524–6 anisotropy 565, 566, 567 arterial spin labeling 133, 134 clinical progression 584–5 cognitive decline 586 combined imaging measures 526 corpus callosum 565–6 diagnosis 582–3 accuracy 526 early 584 diffusion tensor imaging 564–7 diffusivity 567 disease activity 586 disease progression 582 disease-modifying therapies 584 fractional anisotropy 566 genu 565–6 gray matter 581 loss 565 histopathology 584 intervoxel orientational coherence 566 Lewy bodies 583 MRS 525, 593 three-dimensional imaging 588

myo-inositol 13, 581–2 neurofibrillary tangles 584 neuroimaging marker 584–7 neuronal damage 581 neuropsychological deficit pattern 564–5 pathologic progression 584–5 phosphorus-31 MRS 587–8 proton MRS 580, 581–3 biochemical marker 584–7 senile plaques 584 splenium 565–6 white matter 581 loss 564–5 amaurosis fugax 246 amino acidopathies 798–9, 800 amino acids Aspergillus abscess 394, 395 branched-chain 789, 790 epilepsy 491 hydatid cysts 398, 401 MRS of brain abscess 382 neurocysticercosis 400 -aminobutyric acid (GABA) 595 alcohol abuse/dependence 548 brain concentration 485 epilepsy 491, 492 estimates 31 MRS epilepsy 502 psychiatric disease 551 panic disorder 542 -aminobutyric acid (GABA)-A receptor mutation 488–9 amnesia global 570 post-traumatic 620 anatomy experimental 1–2 static 2 anemia, pediatric stroke 736 aneurysms 258 angiogenesis 279–80, 329–30 brain tumors 414 inhibitors 281, 282 microvascular structure indicators 330 pathogenic tissues 330 pediatric patients 656 angiogenic cascade 280 angiography see magnetic resonance angiography (MRA) angiomas 258 angiostatin 280 anisotropy 315 Alzheimer’s disease 565, 566, 567 artifactual 105 cerebral cortex 681–2 characterization 89–90 corpus callosum 562 dementia 525 diffusion 100–1 changes with age 739 image artifacts 106 diffusion tensor MRI 204 fractional 68–9, 70–1, 81, 90, 445 aging 561, 562, 563 alcoholism 568–70 Alzheimer’s disease 566

axonal injury detection 633 chronic carotid disease 249 diffuse ischemic diseases 228 diffusion direction 513 diffusion tensor MRI 204 gray matter 92 gray matter/white matter differences in stroke 227 HIV dementia 465 index 100 maps 318 multiple sclerosis 446, 447, 451, 452 pediatric brain tumors 728 schizophrenia 567–8 stroke 223, 224–5, 226, 227–8 stroke clinical outcome correlation 227–8 stroke ischemic penumbra 165 surgical treatment planning 318 time course of changes in stroke 225, 226, 227 tissue viability prediction in stroke 228 white matter 559, 561–3, 564 gray matter 223 HIV dementia 465 HIV infection 421 indices 223 intracellular compartment 224 lesion border masking 100 measurement 88–9 mild cognitive impairment 566 myelin sheath 224 partial volume effects 100–1 pediatric imaging 660 prematurity 681 quantification 560 relative 68, 69, 90 diffusion tensor MRI 204 two-dimensional visualization 89–91 water diffusion in normal tissues 224 white matter 99–101, 223 brain maturation 678, 679, 680, 681 fibers 559 prematurity 684 anorexia, Wernicke’s encephalopathy 570 antiangiogenic therapies 329–30 antiepileptic drugs 481, 485–6 diffusion-weighted imaging 514 pharmacodynamics 491 antihypertensive therapy 215 antiretroviral treatment 462–3 anxiety disorders 542, 543–5, 546 apodization 28, 29 apoptosis neurodegenerative disease 594 oligodendrocytes 770 apparent diffusion coefficient (ADC) 61, 62, 445 acute disseminated encephalomyelitis 446–7 aging 560–1 brain abscess 355, 382, 409, 411–14 abscess in pediatric patients 662, 664 damage 189 development 661, 738 maturation 678, 679, 680, 681

Index

brain tumors 284, 316–17, 413–14 radiosurgery 322, 323 chronic carotid artery disease 248, 249 dementia 525 diffusion characterizing in vivo 63 data 223 DWI in stroke 193 threshold 197 estimate 73, 79 gradient orientation 87 head injury 627, 630, 631–2, 634 non-accidental in child 637 hippocampal sclerosis 515 HIV dementia 464–5 HIV infection 420 hyperventilation 632 hypoxic–ischemic encephalopathy 711, 712 images 69 infections 354 ischemic stroke 188 ischemic tissues 512 labeled water 127 maple syrup urine disease 790 maps biopsy guidance 317 brain abscess 409, 411–14 Creutzfeldt–Jakob disease 422–3 cytotoxic edema 633 head injury 633 moyamoya syndrome 748 pediatric tumors 726 reversible posterior leukoencephalopathy syndrome 744 status epilepticus 515 therapy response 322, 323 toxoplasmosis 365 West Nile encephalitis 427 measurement 100, 445 multiple sclerosis 445, 450 neonates brain injury 682–3 normal 707–8 stroke 708 neuronal cell death after seizures 516 optic nerve 453–4 pediatric patients 659, 660 brain tumors 728 changes with age 739 reversal 742 stroke 708, 740, 741, 742 primitive neuro-ectodermal tumor 314, 316 ratio of orthogonal 68 regional of tissue water 312–14 seizures 513–14 stroke acute 227 DWI 193, 197 ischemic penumbra 165 pediatric 708, 740, 741, 742 subdural hematoma 630 therapeutic changes 319, 320 trace 67 white matter injury in prematurity 684 X-linked adrenoleukodystrophy 667

arabitol 773, 794 arachnoid cyst 317 diffusion-weighted imaging 326 intracranial epidermoid differential diagnosis 661–2 arbovirus 370 arcuate fibers 781 area under peak (AUP) 147, 149 arginine:glycine amidinotransferase (AGAT) 683 deficiency 683, 782, 784 creatine supplementation 786 arterial input function (AIF) 112, 114 absolute measurement 144 curve amplitude 114 DCE-MRI 339 DRCE-MRI 335, 336, 337 dynamic susceptibility contrast imaging 143–4, 145, 146 neonates 717, 718 partial volume effects 144, 146 peak saturation 144 pediatric patients 655 relaxation rate 142 single 115 underestimation 144 arterial magnetization 122, 123 labeled 122 arterial spin labeling 141 accuracy 127, 208–9 Alzheimer’s disease 133, 134 anaplastic oligodendroglioma 349 applications to brain 126–33 artifacts 150–2, 153, 154–7 blood spin–lattice relaxation time 156 blood volume imaging 129–30 blood–brain partition coefficient 156 brain ischemic region definition 132–3 trauma assessment 132–3 tumors in pediatric patients 728–9 cerebral perfusion assessment 215 cerebrovascular reserve testing 215–16 cerebrovascular transit artifact 210–11 chronic carotid artery disease 248 chronic cerebrovascular disease 214–15 clinical application improvement 218 continuous 123–5, 126 background suppressed 133, 134 cerebral blood flow 209, 210 cerebrovascular disease 210, 212, 213 cerebrovascular reserve 252 cerebrovascular reserve testing 215, 216 comparison to other techniques 127 DSCI comparison 209–10 hypoperfusion 214 interictal 510–11 intravascular signal 152, 153 labeling efficiency 152, 154 magnetization transfer 154 modifications for transit times 150–2 multi-slice 154 off resonance radio frequency 128 pediatric patients 655, 656, 657 pediatric stroke 746–7 post-tagging delay 150–1 sickle cell disease 746–7

subtraction errors 155 whole brain coverage 128, 129 delay introduction 124 development 126–7 developmental delay 685 dynamic 129 dynamic susceptibility contrast imaging comparison 209–10, 211 flow-driven 123 functional MRI of brain 130–2 glioma 282 global ischemia/hypoxia 132–3 head injury 631 histology guidance 132–3 hypoperfusion 210 image acquisition schemes 218 labeling efficiency 152, 154 magnetization transfer 123, 154, 208 method 207–8 middle cerebral artery occlusion 214 model 122–3 MRI detection of regional blood flow 119–35 multi-slice imaging 124–5, 154 subtraction errors 155 neonates 715 neural activity changes 130 neurological disorder assessment 132–3, 134 partial volume effects 155–6 pediatric patients 655, 656, 657 brain tumors 728–9 changes with age 739 pitfalls 150–2, 153, 154–7 precision 127 blood volume measurement 129 BOLD fMRI 131 perfusion fMRI 131 pulsed 125–6 background suppressed 130 cerebrovascular disease 210 comparison to other techniques 127 inflow time 152 labeling efficiency 152, 154 magnetization transfer 154 modifications for transit times 150–2 quantification model 155 saturation pulses 151 subtraction errors 155 quantification model 123–4 regional blood flow quantification 127–8 reliability 127 sensorimotor activation detection 209 signal-to-noise ratio 208, 218 single brain compartment 123–4 single volume coil 123, 124 single-slice pulsed 209 strategy 122–3 stroke 207–18 acute 211–12, 213, 214 assessment 132–3 subject movement 156–7 subtraction errors 155, 156 technique development 121 temporal lobe epilepsy 133 theoretical basis 122–6 tissue magnetization 125–6

807

808

Index

arterial spin labeling (contd) transit times 127, 150–2 transit-related errors 208–9 two-coil setup 124, 125 two-compartment exchange model 155 validity 208–9 vessel territory mapping 128–9 whole brain coverage 128 arteriovenous malformations 258 pulmonary pediatric 662 arylsulfatase A deficiency 755 aspartate, hypoxic brain injury 690 aspartoacylase (ASPA) 789 deficiency 654, 786, 789 Aspergillus (aspergillosis) 359, 361, 364 abscess 394, 395 granuloma 396 intracranial vascular 362 aspiration monitoring during MRS 185 astrocytes 628 gliosis 773 astrocytoma 297 anaplastic 279, 297 white matter fiber tract displacement 318–19 brainstem fibrillary 651 contrast enhancement 298 giant cell 723 multiple sclerosis differential diagnosis 431 pediatric 722, 723 MRS 730, 735 pilocytic 313 presurgical diagnosis 732 recurrent 310 transverse relaxation time 298 see also juvenile pilocytic astrocytoma ataxia telangiectasia 723 atherosclerosis 249 aura, migraine 266–9 axons 558–9 aging deletion 560 membrane changes 739 gross morphology 559 injury detection in traumatic brain injury 633–4 diffusion-weighted imaging 633–4 shearing 634 traumatic brain injury 613, 614, 631, 642 loss 778 aging 560 metachromatic leukodystrophy 757 water diffusivity 447 membrane changes with age 739 Pelizaeus–Merzbacher disease 766 regression in Alzheimer’s disease 565 b-value 57–8, 659–60 3 T diffusion tensor imaging 715, 716 high 99 increasing 58 optimal 73 optimal number of measurements 73 signal attenuation 101 Babinski sign, positive bilateral 793

bacteria aerobic 382, 383 facultative anaerobes 382, 384 see also meningitis, bacterial band-selective inversion with gradient diphasing (BASING) 45 basal ganglia acute disseminated encephalomyelitis 450 Alexander disease 768, 769 depression 535, 536–7 HIV dementia 465 HIV infection 571 hypoxic–ischemic brain injury 649, 700, 701 infarcts 744 mitochondrial encephalopathy 796 multiple sclerosis 449–50 pediatric imaging 648 3D MRSI 676, 677 infarcts 744 betaines 619 bipolar disorder 538, 539–41 general affective disorder 542 Bloch equations 122, 126, 155 blood spin–lattice relaxation time 156 T1 209 transit times of labeled 150–2 blood oxygen level-dependent (BOLD) contrast 119 migraine 264 response 247 blood pressure, arterial 611 blood vessels large vessel dispersion of contrast agent 114 partial volume effects 114, 115 see also angiogenesis blood–brain barrier breakdown 146–7, 725 pediatric brain tumors 728 integrity assessment 127 leak in head injury 626–8, 632 permeability assessment 248 to contrast agents 116 to gadolinium contrast agents 115 blood–brain partition coefficient, arterial spin labeling 156 bolus arrival time (BAT) 147, 149, 251 bolus tracking 111, 112 brain changes with age 739 flow derivation 115–16 leaky blood–brain barrier 115, 116 moyamoya syndrome 747 pediatric stroke 744 sickle cell disease 744 transit time derivation 115–16 border-zone 208 abnormal perfusion timing 251 border-zone infarcts 236 severe obstructive carotid artery disease 241–2 boxing (sport) diffusion-weighted imaging 636, 639 head injury 636, 638, 639 occult brain damage 643

brain aging 22–3 anatomy 1 arterial spin echo-based functional MRI 130–2 atrophy 614 congenital malformations 665–6, 667 death 693 development diffusion-weighted imaging 661, 713 MRS 20, 22 prematurity 713 epileptogenic lesions 489 function 1 functional impairment 558 hemodynamic status 247 iron homeostasis 271 maturation 738 apparent diffusion coefficient 678 diffusion tensor imaging 678–82 diffusion-weighted imaging 678 normal 676, 677, 678–82 proton MRS 676, 677, 678 normal appearing tissue 451, 452 parenchyma hypoperfusion 251 water diffusivity 59 peritumoral infiltration 279–80 spatial heterogeneity 2 structure–function relationships 562–3 swelling 611, 628 T1 of tissue 209 temperature measurement 13 stereotactic biopsy site selection 283 water anisotropic diffusion 223 neonates 675, 707 pathophysiology 627, 628 pediatric 729, 739 signal 34, 35 see also abscess, brain; parenchymal disease brain injury hypoxic adult 695, 698–9 diagnosis 700 MRS 690–703, 705 neuronal–glial substrate cycling 700 point resolved spectroscopy 694, 695, 696–8 ischemic late secondary 188 MRSI in acute 181 neonates 682–5 diffusion tensor imaging 682–3 diffusion-weighted imaging 682 proton MRS 682, 683–4 pathophysiology 611 preterm 712–13 swelling 611 see also head injury; hypoxic–ischemic brain injury brain tumors N-acetyl aspartate signal 294 angiogenesis 414 imaging 339, 340, 341–4 angiogenic neovasculature 330

Index

apparent diffusion coefficient 312–14, 316–17 biopsy site choice 283–4 brain abscess differentiation 354–5 capillary permeability 281 choline signal 291–3, 295 contrast diffusion rates 334 creatine signal 293–4 diagnosis 283 diffusion tensor MRI 284, 323 diffusion-weighted imaging 284, 312–24, 326–8, 412–13 pediatric patients 726, 728 early neoplastic changes 115 edema 317, 323 peritumoral 341 external beam radiotherapy 300 focal radiation planning 283 grading 283 heterogeneity 322–3 imaging endpoints in trials 280 incidence 279 interoperative biopsy correlation 283 intratumoral hemorrhage 413 lactate 171 signal 294 metabolism 294–5 microcirculation 281 microvascular features of pathologic tissues 331 microvasculature quantification 330–1 mobile lipid signal 294 MRS 288–301, 306–11 necrosis 294–5 treatment-induced 316 outcome 279 pediatric 649–50, 651–3, 653 combined sequences/technologies 732 contrast enhancement 725–6 diffusion tensor imaging 728 diffusion-weighted imaging 661–2, 664, 665, 726, 728 microvessel density 728 MRS 729–32 perfusion-weighted imaging 656, 658, 659, 728–9 physiological MRI 722–32, 735 prognosis 731–2 radiation therapy 726, 727 serial monitoring 732 volume 726 perfusion-weighted imaging 656, 658, 659, 728–9 peritumoral brain infiltration 279–80 proton MRS 282–4 clinical significance of diagnosis 298 diagnostic uses 295–8 echo time 289, 290 general features 288–9, 290, 291 grading 297 guidance of therapeutic procedures 298–9 malignant transformation 300 metastatic tumors 296 neoplastic recurrence 300–1 pattern classification techniques 297

preoperative guidance 299 primary neoplastic lesion differentiation 296–8 radiation necrosis 300–1 recurrence 301 spectra variability 291 therapy response assessment 299–300 tumor margin definition 298 voxel position 297–8 radiation therapy 320–1, 322 radiosurgery 322, 323 recurrence 350 seizures 492, 494 superimposed infection 413 surgical cytoreduction 298 surgical planning 317–19 survival rate 312 therapeutic regimen effectiveness 312 tissue compartment volumes 281 tumefactive demyelinating lesions 296 tumor margin determination 323 vascular proliferation 414 water mobility magnitude of change 322 brainstem fibrillary astrocytoma 651 glioma 722 infarcts 164 migraine headache 269–70 perinatal stroke 710 white matter tract maps 665–6, 667 branched-chain alfa-keto acid (BCKA) 790 branched-chain alfa-keto acid (BCKA) dehydrogenase deficiency 789 branched-chain amino acids (BCAA) 789, 790 Broca PP 1 Brownian motion 54 CADASIL 228, 437 callosal agenesis 713, 714 Canavan’s disease 651, 654 maple syrup urine disease differential diagnosis 790 MRS 763–5, 786–7, 788, 789 NAA 10, 651, 654 Candida, brain abscess 394 capillary permeability 127 cardiac cycle 74 cardiac output cerebral perfusion 256 see also heart/heart disease carotid artery, internal angiography 725 giant aneurysms 258, 259 infarction 362 ischemic damage symptoms 249 magnetic resonance angiography 249, 250 obstructive disease 235 occlusion cerebrovascular reactivity 241 MR spectra 236–8 MRS 171 multi-slice MRSI 173, 174–5, 181 pediatric stroke 737 stenosis bolus delay/dispersion 144, 145 carotid endarterectomy 237–8, 240

cerebrovascular reactivity 241 interventional techniques 253 mean transit time 251–2 medical interventions 253 MR spectra 236–8 pediatric stroke 737 summary parameters 149 surgical interventions 253 transient ischaemic attacks 166 carotid artery disease, chronic 246–59 cerebral blood volume 252 cerebral perfusion 253–4 cerebrovascular reserve 252 collateral macrovascular flow 249–51 hemodynamic alterations 253–5 imaging techniques 248–9 interventional techniques 253 moyamoya syndrome 256, 258 MR findings 249–56, 257, 258–9 perfusion 251–2 Takayasu arteritis 258 types 249–56, 257, 258–9 vascular reserve 252 carotid artery disease, severe obstructive 234–43 anaerobic glycolysis 235 border-zone infarcts 241–2 cerebral flow 239–40 cerebral perfusion pressure 234–5 cerebrovascular reactivity 241 collateral flow 234 patterns 240, 241 hemodynamic changes 234–5 hemodynamic failure 241–2 hemodynamic measure association 238–42 MR spectra 236–8 MRS 238–42 clinical relevance 242–3 neuronal damage 235 perfusion abnormalities 239–40 carotid bifurcation, transient ischaemic attacks 250 carotid endarterectomy 176, 215–16 cerebrovascular reserve 254–5 hyperperfusion syndrome following 229 internal carotid artery stenosis 237–8, 240 MR perfusion assessment 253 carotid stenting 253 cavum septum pellucidum 636, 638 CD4 cell count 466, 571 cell density 314, 315–17 central nervous system (CNS), granulomatous infections 359 central pontine myelinolysis 447, 570, 571 centrum semiovale, alcoholism 569 cerebellar ataxia, autosomal dominant 595 cerebellar atrophy 597 L-2-OHG aciduria 798 cerebellar degeneration, HIV infection 571 cerebral arterioles, dilatation 241 cerebral arteriopathy, pediatric stroke 736–7 cerebral artery, anterior, border-zone lesions 242 cerebral artery, middle border-zone lesions 236, 242 collateral flow 250

809

810

Index

cerebral artery, middle (contd) infarction 362 neonatal 709, 718 pediatric 741, 742 occlusion 163, 164 N-acetyl aspartate MRS measurement 169, 170, 171 arterial spin labeling 214 multi-slice MRSI 173, 174–5, 181 pediatric stroke 737 stenosis continuous arterial spin labeling 212, 213 pediatric stroke 737 transit time 151 stroke apparent diffusion coefficient 189 perinatal 709, 718 subarachnoid hemorrhage 259 cerebral atrophy boxing 636 creatine deficiency 785 cerebral autosomal dominant arteriopathy with subcortical infarcts and leukoencephalopathy see CADASIL cerebral blood flow 4, 5 absolute quantification 144 absolute values 115 arterial input function inaccuracy 143–4, 145 autoregulation 215 bolus delay/dispersion 144, 145 cerebral homeostasis changes 172 changes with age 739 chronic carotid disease 247 continuous arterial spin labeling 209, 210 deconvolution 113–14 derivation from kinetic principles 116 dynamic susceptibility contrast imaging 115, 142–3 epilepsy 510–11 global autoregulation 611 head injury 626, 627, 628, 629, 631, 636, 639 hemodynamic failure 241–2 measurement 115 model-dependent/-independent approaches 113 moyamoya syndrome 747 neonates 717 overestimation 144 pediatric patients 654, 655 stroke 656, 657, 658 perfusion-weighted imaging 183 pharmacological augmentation 215, 216 reduced 208 regional gray matter 120 measurement techniques 127 MRI detection 119–35 quantification 126, 127–8 stroke 133 visual cortex 128, 131 wash-in/wash-out of tracers 119, 120 relative columnar neuronal element specificity 131

glioma 281–2 HIV dementia 464, 465 migraine 267, 268, 269 pediatric brain tumors 728 residue tracer analysis 111–13 severe obstructive carotid artery disease 239–40 stroke 192, 194–5, 208 underestimation 143 cerebral blood volume 4, 5 bolus tracking 111, 112 chronic carotid disease 252 derivation from kinetic principles 116 dynamic susceptibility contrast imaging 142 epilepsy 510 head injury 627, 631, 636 measurements 111 moyamoya syndrome 747 neonates 717 pediatric patients 654 perfusion-weighted imaging 183 relative brain abscess 414 brain tumor 414 DSCI-MRI 333 glioma 281–2, 341 HIV dementia 464 mapping 282 pediatric brain tumors 728 peripheral tumors 344, 345 tumors 342–3 seizures 510 cerebral cortex anisotropy 681–2 prematurity 681 cerebral dysfunction, interictal/postictal 482 cerebral embolization, hypoperfusion 214 cerebral hemodynamic compromise 211 cerebral hemodynamic impairment grading 211, 213 cerebral hemorrhage, diffusion-weighted imaging 190, 191 cerebral injury, lactate correlation 238 cerebral ischemia N-acetyl aspartate 172 carotid intervention 255–6 head injury 628, 629, 633 hypoperfusion 215 lactate 172 metabolic changes 172–3 reperfusion 172–3 water diffusivity imaging 247–8 cerebral metabolic rate of oxygen metabolism (CMRO2) 235, 247 internal carotid artery occlusive disease 237 cerebral metastases 341 pediatric 722, 726 cerebral palsy, diffusion-weighted imaging 664–5, 666 cerebral perfusion 119 assessment 215 cardiac output 256 chronic carotid disease 253–4 deficit 247 regional 132

status 247 vascular abnormalities 258 see also cerebral blood flow cerebral perfusion imaging contrast mechanisms 109 delay 115 dispersion 115 echo time 110 exogenous contrast agents 109–16 gradient echo sequence 109, 110 mean transit time 114 perfusion non-linearity 111 pitfalls 114–16 quantification 114–15 residue function 113–14 residue tracer analysis 111–13 spin echo sequence 110 susceptibility contrast 109–11 tissue concentration time curves 115–16 see also bolus tracking cerebral perfusion pressure heart disease 256 hemodynamic compensation 234–5 severe obstructive carotid artery disease 234–5 cerebritis 381 cerebrospinal fluid (CSF) partial volume 34 pulsatile motion 102 water diffusivity 59 cerebrotendinous xanthomatosis (CTX) 437–8 cerebrovascular blood supply deficit, chronic 246–7 cerebrovascular disease 163–6 arterial spin labeling 210, 214–15 pediatric 216–18 chronic 214–15 diffuse ischemic 228 hypoperfusion 214 management 230 pediatric 216–18 stroke mimics 229–30 cerebrovascular reserve 247 carotid endarterectomy 254–5 chronic carotid artery disease 252 intravascular stenting 254–5 testing arterial spin labeling 215–16 carotid artery disease 239 chemical shift imaging 33 bandwidth 40–1 epilepsy 497 hypoxic encephalopathy 696, 698–9 neurodegenerative disease 595 chemical shift-selective water suppression (CHESS) 45 pulses 18 chemotherapy 743–4 cholesterol 679 choline (Cho) 4, 595 adrenoleukodystrophy 760, 761 aging 579, 580 brain 22–3 Alexander disease 768, 769 Alzheimer’s disease 581, 582, 583 border-zone infarcts 242

Index

brain developing 20, 22 normal maturation 676, 677 brain tumor MRS 283, 292, 295, 299 glioma 306 malignant transformation 300 meningioma 309 pediatric 729–32 radiation necrosis 301 signal elevation 291–3 therapy response assessment 299–300 brain tumor proton MRS 289, 291 Canavan’s disease 764, 765, 787, 788 cerebrovascular reactivity 241 cortico-basal degeneration 599 cryptococcoma 471 dementia 583 developmental delay 684 epilepsy 491, 492, 493, 497 focal 495 fronto-temporal degeneration 583 gray matter signals 20, 21, 22 hereditary leukoencephalopathies 797–8 HIV dementia 461, 462 abnormality reversal 462–3 pediatric patients 464 HSV1 368, 369 Huntington’s disease 601 3-hydroxy-3-methylglutaryl-coenzyme A lyase deficiency 799 hypoxic encephalopathy 693 internal carotid artery occlusion/stenosis 236–8 Japanese encephalitis 393 Krabbe’s disease 758 lymphoma 467, 469 malaria 403 megalencephalic leukoencephalopathy with subcortical cysts 771, 772, 773 meningitis 381 metabolism 292 mitochondrial encephalopathy 797 mucormycosis 396 multiple sclerosis 431, 432, 434, 435 multi-system atrophy 597 myelinopathia centralis diffusa/vanishing white matter disease 770, 771 neonatal brain injury 682 neurocysticercosis 398, 399 obsessional compulsive disorder 546 L-2-OHG aciduria 798, 800 para-infectious encephalopathy 394 Parkinson’s disease 596 pediatric imaging 739–40 brain infarction 649 metabolic stroke 749 neoplasms 649, 650, 651, 652–3 Pelizaeus–Merzbacher disease 766, 767, 798 post-contrast spectroscopy 49, 50 primary neoplastic lesion differentiation 297 progressive multifocal leukoencephalopathy 469, 478 progressive supranuclear palsy 598 proton MRS 9, 12

neoplastic/non-neoplastic differentiation 295–6 signal 291–3 tumefactive multiple sclerosis 308 proton spectra peaks 30 Salla disease 439, 789 severe obstructive carotid artery disease 239–40, 242–3 stroke 173, 174–5 subacute sclerosing panencephalitis 392 transient ischaemic attacks 166 traumatic brain injury 615, 617, 618, 619, 620, 621, 622 tuberculoma 472 tuberculous abscess 389 white matter signals 20, 21, 22 chordoma 314, 316 choroid plexus carcinoma 650 chromosomes 780 circle of Willis collateral flow 234, 240, 241, 250 role 250–1 clinical lesion/deficit research 1–2 coenzyme Q10 601 cognitive function, alcoholism 568 cognitive impairment, mild 525 amnestic 584, 586 anisotropy 566 combined imaging measures 526 diagnostic accuracy 526 diffusivity 566 coherence control 41–2 complex II deficiency 761, 763, 764–5, 793 computed tomography (CT) 2 head injury 609 non-accidental in child 637 pediatric stroke 740 perfusion imaging in stroke 165 stroke 163, 164, 184, 198 xenon 164, 165, 611 head injury 629 pediatric patients 654 concentration gradients, diffusion 54–5 congenital heart disease, pediatric stroke 736 congenital malformations cortical development in epilepsy 483 diffusion-weighted imaging 713, 714 heart 662 connectivity maps 560 contrast agents bolus delay 143–4, 145, 149 bolus dispersion 143–4, 145 enhancement in pediatric tumors 725–6 large vessel dispersion 114 mean transit time 114, 116 post-contrast spectroscopy 49, 50 re-circulation 146 residue tracer analysis 111–13 tissue concentration 110 time curves 115–16 transit times 150–2 wash-in/wash-out 119, 120 see also gadolinium contrast agents contrast mechanisms 109 corencephalopathy 781 corpus callosum agenesis 713, 714

aging 561, 562, 563 alcoholism 569 Alzheimer’s disease 565–6 anisotropy 562 cortical development malformations, epilepsy 483 cortical dysplasia focal 498 subtle focal 490 cortical hypointensity, gyriform 514 cortical spreading depression 264, 267, 269 cortico-basal degeneration dementia 598 MRS 598–9 corticocortical axons 681–2 coupled spin systems 15, 44 Cowden syndrome 723 craniopharyngioma, pediatric 722, 732 creatine (Cr) 595, 683–4 adrenoleukodystrophy 760, 761 aging 579, 580 brain 22–3 Alexander disease 768, 769 Alzheimer’s disease 581, 582, 583, 585–6, 588 biosynthesis 783–4 brain injury hypoxic 693 traumatic 615, 616, 617, 619, 620 brain normal maturation 676, 677 brain tumor MRS 283, 292 pediatric 729, 730 radiation necrosis 301 brain tumor proton MRS 289 Canavan’s disease 765, 787, 788 cortico-basal degeneration 599 cryptococcoma 471 deficiency 683–4 dementia 583 dementia with Lewy bodies 583 developmental delay 684 epilepsy 491, 492, 493, 497 Epstein–Barr virus 392 fronto-temporal degeneration 583, 584 hereditary leukoencephalopathies 798 HIV dementia 461, 462 abnormality reversal 462–3 pediatric patients 464 Huntington’s disease 601 3-hydroxy-3-methylglutaryl-coenzyme A lyase deficiency 799 hypoxia–ischemia cascade 691 hypoxic brain injury 695, 699 hypoxic–ischemic encephalopathy 692 internal carotid artery occlusion/stenosis 236–8 Japanese encephalitis 393 Krabbe’s disease 758 lymphoma 467, 469 megalencephalic leukoencephalopathy with subcortical cysts 771, 772, 773 meningitis 381 mucormycosis 396 multiple sclerosis 431, 432, 434, 435 multi-system atrophy 597 myelinopathia centralis diffusa/vanishing white matter disease 770, 771

811

812

Index

creatine (Cr) (contd) neonatal brain injury 682 neurocysticercosis 398, 399 obsessional compulsive disorder 546 L-2-OHG aciduria 798, 800 para-infectious encephalopathy 393, 394 Parkinson’s disease 596, 597 pediatric imaging 647–8, 739–40 brain infarction 649 metabolic stroke 749 Pelizaeus–Merzbacher disease 766, 767, 798 primary neoplastic lesion differentiation 297 progressive multifocal leukoencephalopathy 469, 478 progressive supranuclear palsy 598 proton MRS 9, 12 signal 293–4 proton spectra peaks 30 Salla disease 439, 789 severe obstructive carotid artery disease 239–40, 242–3 stroke 169 supplementation 784–5, 786 transient ischaemic attacks 166 transporter protein deficiency 684 vascular dementia 583 creatine deficiency syndrome 782–6 creatine transporter (CrT1) carriers 784–5 defect 782, 784 adult 785 Creutzfeldt–Jakob disease 372, 373, 526–7 diffusion-weighted imaging 421–3, 428, 526 hyperintense signal abnormalities 422 MRS 526–7 neurodegeneration markers 421 striatal involvement 422 variant 372, 526–7 diffusion-weighted imaging 526 MRS 526–7 crusher gradients 14 strong 42, 43 variable 42 cryptococcal meningitis 361, 362, 466 cryptococcoma 396, 397 AIDS patients 466, 470–1 cryptococcosis 471 cyclosporin toxicity 229, 743, 745 cysticercosis 363–5 MRS 396–8, 399, 400 cytotoxic agents 319 Dawson’s fingers 374, 375 deconvolution 113–14 dementia cortico-basal degeneration 598 cost-efficiency of neuroimaging 526 Creutzfeldt–Jakob disease 372, 421 diffusion-weighted imaging 525, 566 with Lewy bodies 580 proton MRS 583 metabolite measurements 583 MRS 525 higher magnetic field strength 587

perfusion imaging 525 physiological MRI 524–6 proton MRS 580–4 pugilistica 643 vascular 525 Binswanger type 566 MRS 525 proton MRS 580, 583 see also AIDS dementia complex; Alzheimer’s disease; HIV dementia demyelinating diseases 353, 372–6, 429, 437–9 diffusion-weighted imaging 444–5, 446–7 lesion load 373–4 metachromatic leukodystrophy 757 myo-inositol 13 treatment response 373–4 see also acute disseminated encephalomyelitis; multiple sclerosis demyelination 778 depression aging 571 diffusion tensor imaging 571 major 535, 536–7, 538 general affective disorder 542 SSRIs 531 Descartes, Renée 1 developmental delay diffusion-weighted imaging 684–5 infants 683–5 proton MRS 683–4 perfusion-weighted imaging 685 diagonalization 89 diffuse excessive high signal intensity (DEHSI) 713 diffusion anisotropic 61–3, 87–8, 100–1 characterization 89–90 image artifacts 106 source 61–3 surgical treatment planning 318 b-factor 57–8 Brownian motion 54 coefficient 316 concentration gradients 54–5 displacement probability profiles 55, 56 encoding 57 flow 102–4 isotropic 87–8 magnetic resonance signal 55–8 measurement 445 motion 102–4 quantification 560 random walk 55, 56, 58 signal-to-noise ratio 57 white matter 65–6 diffusion ellipsoid 63–4, 65, 66, 66, 67 characterization 88, 89 orientation 88 visualization 91 pediatric imaging 660 shape definition 88 diffusion spectrum imaging 80 diffusion tensor 63–9, 70–2, 73, 512 anisotropy indices 68–9, 70–1 diagonalization 89 diffusion ellipsoid relation 63–4

eigenvalues 63, 67, 68, 512, 559 brain development 679–80, 682 eigenvector 63, 101, 318, 512 prematurity 681 elements 63, 64, 89 estimation 64–6 normalization of data sets 106–7 orientation 63, 69, 72, 73 registration of data sets 106–7 rotationally invariant measures 66–7 trace 223 diffusion tensor fiber tracking 680–1 diffusion tensor imaging 409, 452–3 3 T 715, 716 adrenoleukodystrophy 761 aging 560–4 alcoholism 568–70 Alzheimer’s disease 564–7 applications 513 brain maturation 678–82 structure–function relationships 562–3 tumors 284, 323 congenital brain malformations 665–6, 667 cytotoxic edema detection 631 demyelinating diseases 373, 374 depression 571 developmental delay 684–5 diffusion anisotropy 100–1 diffusion ellipsoid orientation 91 experiments 89 fiber orientation mapping 95, 228–30 glioma differentiation from metastases 328 infiltration 327 head injury 611, 633, 634 HIV dementia 466 HIV infection 571 interictal 515 Krabbe disease 667 limitations 76–7, 78, 79–81 MR tractography 86–96, 685 neonates brain injury 682–3 normal 708 optimal echo time 73–4 optimal repetition time 74 pediatric brain tumors 728 pediatric patients 660, 675, 676 basal ganglia infarcts 744 perinatal stroke 710 prematurity 678 progressive multifocal leukoencephalopathy 571 quantification 566 quantitative measurements 66–9, 70–2, 73 sampling vectors 73 schizophrenia 567–8 sequence optimization 73–4 speed of acquisition enhancement 669 stroke 223–31 mimics 229–30 surgical treatment planning 317–19 termination criteria 92–3 tissue anisotropy 204 toxoplasmosis 571

Index

trace 67–8 Wallerian degeneration 228–30 white matter aging 561–4 measurement 559 tracking 715 diffusion tensor tractography (DTT) 680–1 diffusion-weighted imaging 4–5, 54, 58–9, 58–81, 60, 61 acute disseminated encephalomyelitis 446–7, 450 angular resolution 101 anisotropic diffusion 61 arachnoid cyst 326 axonal injury detection 633–4 biopsy guidance 317 boxing 636, 639 brain maturation 678 tissue damage 450–1 brain abscess 355–6, 382, 409, 410, 411–14 pediatric patients 662, 664, 665 brain development 661 preterm infant 713 brain injury hypoxic–ischemic 661, 662, 663, 700, 702 hypoxic–ischemic in neonates 708, 710–12, 721 ischemic 187–8, 252–3 neonates 682, 706 preterm 712–13 stroke 252–3 traumatic 619 brain tumors 284, 312–24, 326–8, 412–13 biopsy 323 diagnosis 317 distinction from other lesions 316 edema 317 malignant lesion differentiation 323 pediatric patients 661–2, 664, 665, 722, 726, 728 water mobility magnitude of change 322 cellular density 314, 315–17 central pontine myelinolysis 447 cerebral palsy 664–5, 666 chronic carotid artery disease 248, 249 clinical findings correlation 451–2 congenital malformations 713, 714 Creutzfeldt–Jakob disease 372, 373, 421–3, 428, 526 dementia 525, 566 demyelinating diseases 446–7 dephasing 315 developmental delay 684–5 diffusion tensor 63–9, 70–2, 73 diffusion time dependence 104 eddy currents 104–6 edema detection cytotoxic 631 vasogenic 631–2 empyema 417–18 encephalitis 367, 416–18, 419 epidermoids 326 epidural effusions 356 epilepsy 512–16 epileptic focus 520 fetus 714–15

fiber direction estimation 101 field gradient pulses 102 gradient field non-linearity 106 head injury 630 acute 631–4 complex changes 632–3 non-accidental in children 636, 637 hemorrhage identification 190, 191 high b-value images 453 hippocampal sclerosis 515 HIV dementia 464–6 HIV infection 418, 420–1 HSV1 368, 369 hyperintensity 185, 188, 248, 317 hypoxic–ischemic brain injury 661, 662, 663, 700, 702 neonates 708, 710–12, 721 ictal 512–16 human studies 514–15 imaging gradients 104 infections 354, 408–23 pediatric patients 662, 664, 665 interictal 512–16 human studies 515–16 intracranial vascular disease 363 intravoxel dephasing 102 ischemic injury 187–8 stroke 252–3 ischemic lesions 186 maple syrup urine disease 790 measurement system-related sources of error 104–7 meningitis 356, 359, 414–15, 416 metastases 412–13 motion-induced phase errors 102–3 multiple sclerosis 187–8, 189, 376, 445–6 clinical findings correlation 451–2 multi-slice imaging 104 neonates 706–15 brain injury 682, 706 hypoxic–ischemic brain injury 708, 710–12, 721 normal 707–8 non-echo planar imaging 182–3 normal appearing white matter 447–50 object motion 102–4 object-related sources of error 99–104 optic nerve 453–4 para-infectious encephalopathy 393, 394 parenchymal disease 361, 363 partial volume effects 100, 101 pediatric patients 658–67 applications 660–2, 663, 664–7 brain congenital malformations 665–6, 667 brain tumors 661–2, 664, 665, 722, 726, 728 cerebral palsy 664–5, 666 changes with age 738–9 inborn errors of metabolism 666–7 infections 662, 664, 665 ischemia differential diagnosis 743–4 stroke 740, 741, 742–4 X-linked adrenoleukodystrophy 666–7, 668 perfusion influence 101 phase gradient 106

preterm brain injury 712–13 primary lateral sclerosis 605 progressive multifocal leukoencephalopathy 370, 447 pulse sequences 314–15 radiation necrosis 413 rephasing 315 rigid body motion 102, 103 ring-enhancing lesion 412, 420, 421 multiple sclerosis 446 seizures 513–14 sequence optimization 73–4 signal attenuation 59 spinal cord lesions 454 stroke 182–3 apparent diffusion coefficient 193, 197 CT comparison 198 increased signal duration 190, 191, 192 ischemic injury 252–3 ischemic penumbra 164–5 lesion development 190, 192–4, 195, 196 lesion measurement 196–7, 198 lesion reversal 205 lesion volume measurement 197 outcome 197 patient selection for thrombolysis 175 perinatal 708, 709, 710 severity 197, 198 TIA differentiation 186–7 uses 185–6 subarachnoid hemorrhage 206 subdural empyema 357, 415–16, 417–18 susceptibility gradients 104 systematic error due to noise 106 T2 effect 99 T2-shine through 59, 61, 99 technology 182–3 therapeutic monitoring 319–23 toxoplasmosis 365, 366, 467, 468 transient ischaemic attacks 165–6, 186–7 traumatic brain injury 619 uses 119 visible lesions 188–9, 190 white matter anisotropy as artifact 99–101 tract mapping 453 X-linked adrenoleukodystrophy 666–7, 668 see also diffusion tensor imaging; echo planar imaging; perfusion-weighted imaging/diffusion-weighted imaging (PWI/DWI) mismatch region diffusivity Alzheimer’s disease 567 imaging 247–8 mean 67–8, 223 stroke 223–4 mild cognitive impairment 566 quantitative estimates 61 voxel-based distribution 566 see also apparent diffusion coefficient (ADC) digital subtraction angiography (DSA) 163–4 disease conditions metabolite ratios 32 proton MRS 13 dopamine uptake transporter (DAT) imaging 527

813

814

Index

Doppler ultrasound, duplex 163 dorsolateral prefrontal cortex 534 dot product maps 560 DSM-IV-TR 535 dual-band spectral–spatial method 45 dynamic contrast enhanced imaging techniques using MR (DCE-MRI) 331 antiangiogenic therapy response assessment 344 arterial input function 335, 336, 337, 339 contrast leakage 333–5 data analysis 336–8, 339 collection 331–2 dual echo methods 335 extracellular distribution volume 343 flip angles 335, 336 glioma 339 imaging sequence 335, 336 low flip angle methods 335 maximal enhancement value 337 maximal intensity change per time interval ratio (MITR) 337 pharmacokinetic analysis 335–6, 337–8, 339 pre-enhancement techniques 333–5, 335 pseudopermeability effect 338, 339 relative recirculation 333 value 342 relaxivity technique 331, 335–8, 339, 343–4, 345 residual contrast effects 334 signal intensity 333–4 signal-to-noise ratio 332 surrogate markers 331 susceptibility technique 331, 332–5 gliomas 339, 340, 341–2 T1 sensitivity reduction 333 T1-based acquisition 332, 333 therapeutic monitoring 344 tissue enhancement 333–5 transfer coefficient 337–8, 339, 340, 341, 343 peripheral tumors 344, 345 tumefactive demyelinating lesions 343 tumors capillary permeability 343 grade 341 histologic heterogeneity 341 dynamic susceptibility contrast imaging (DSCI) 109, 141 arterial input function 143–4, 145, 146 arterial spin labeling comparison 209–10, 211 artifacts 142–4, 145, 146–50 blood–brain barrier breakdown 146–7 cerebral blood flow 115 contrast agents 146 conversion factors 143 correction factors 143 dual echo sequence 146 HIV dementia 464 limitations 142–4, 145, 146–50 lymphoma 469 pediatric tumors 723–4 severe obstructive carotid artery disease 239–40

stroke 207 subject movement 150 summary parameters 147–9 voxel shift 147 echinococcosis 398–9, 401, 403 Echinococcus granulosus 398 echo planar imaging 74–6 artifacts 75–6, 147 blurring 103 bolus tracking 111, 112 chronic carotid artery disease 249 eddy currents 105–6 epilepsy 515–16 ghosting artifacts 103 gradient echo 110–11 image post-processing 104 intravascular signal 152, 153 motion-induced phase errors 102–4 navigator echoes 103 neoplasia imaging 331 parallel imaging 78 pediatric patients 654 perfusion non-linearity 111 phase gradient 103 single-shot 77, 78, 103 spin echo 110–11 spiral navigated interleaved 75, 76, 77 stroke 182 ischemic penumbra 164 vessel misregistration 147, 148 echo planar imaging-signal tagging with alternating RF (EPISTAR) 125, 209 inflow time 152 echo time averaging 49, 50 optimal 73–4 pediatric imaging 648 eddy currents 104–6 edema, brain brain tumors 317, 323, 341 cytotoxic 229, 230, 247–8 head injury 611, 627, 628, 631, 639 West Nile encephalitis 427 extracellular 628 intracellular 628 peritumoral 341 reactive 354 vasogenic 229, 230, 247–8 glioma 341 head injury 611, 627–8, 631–2 pediatric tumors 724 West Nile encephalitis 427 eIF2B factor 768, 770 Einstein equation 55 elderly patients 524 electroencephalogram (EEG) 483 hypoxic brain injury 690 electron transport, inborn errors of metabolism 651 empyema 356, 357, 358 diffusion-weighted imaging 415–16, 417–18 epidural 357 encephalitis 366–8, 369, 370–2 AIDS 371–2 arbovirus 370

Creutzfeldt–Jakob disease 372, 373 diffusion-weighted imaging 416–18, 419 enterovirus 368, 370 herpes simplex 388–9, 391 HSV1 369 Japanese 392–3 progressive multifocal leukoencephalopathy 370–1 toxoplasmosis 365 see also herpes simplex encephalitis encephaloduroarteriosynangiosis (EDAS) 258 encephalomyelitis Epstein–Barr virus 418, 419 see also acute disseminated encephalomyelitis encephalopathy acute necrotizing 393–4 creative deficiency syndrome 782 diffuse 781 giant cell 570–1 hypertensive 743 hypoxic 690, 691, 695 proton MRS 694–5, 696–9, 699–700 mitochondrial 749, 750, 761, 763, 764–5, 793, 796–7 MELAS 749, 750, 803 succinate dehydrogenase deficiency 793 Reye’s syndrome 705 see also hypoxic–ischemic brain injury; Wernicke’s encephalopathy endostatin 280 enterovirus 368, 370 enzyme-linked immunoelectrotransfer blot (EITB) 397 enzyme-linked immunosorbent assay (ELISA) 397 ependymoma MRS 729–30 pediatric 722, 723, 726 diffusion-weighted imaging 726 epidermoids, diffusion-weighted imaging 326 epidural empyema 357 epilepsy 481 ablative lesions 483 alien tissue lesions 483 autosomal dominant nocturnal frontal lobe 488 brain imaging pathophysiological investigations 484–5 presurgical evaluation 483–4 brain metabolism 488 cerebral blood flow 510–11 cerebral blood volume 510 chemical shift imaging 497 classification 482 cortical development malformations 483 diagnosis 481 brain imaging 482–3 diffusion-weighted imaging 512–16, 520 drug-resistant 481, 483 echo planar imaging 515–16 electroencephalogram 483 gene mutations 488 gliosis 516 glucose metabolic dysfunction 484 hyperperfusion 510, 512

Index

hypoperfusion 510–11, 512 interictal 484 MR investigation 489–91 MRS 491–502, 507–8 data analysis 497–9 high-field 502 nuclei for radiolabeling 500–1 nocturnal frontal lobe 515 partial 483 acquired lesions 489 postictal switch 484 presurgical evaluation 483–4 primary 482 refractory focal 489, 490 secondary 482 temporal lobe 482 arterial spin labeling 133 bilateral MRS abnormalities 495 conventional MR 489–91 diffusion-weighted imaging 514 extra 497, 498 interictal hypoperfusion 511 limbic 483, 484–5 MRSI 508 phosphorus-31 MRS 500 proton MRS 494–5, 496, 497 rodent model 513, 514 severe 495 widespread MRS abnormalities 495, 496, 497 therapeutics research 485–6 see also seizures/seizure disorders Epstein–Barr virus (EBV) 368 diffusion-weighted imaging 418, 419 lymphoma 468, 469 MRS 391–2 external reference method 34–5 extravascular extracellular space 331, 335, 337, 338 tumors 343–4 fast automatic shimming technique by mapping along projections (FASTMAP) 16 3D fast spoiled gradient recalled echo acquisition at steady state (3D-SPGR) 489 fast-spin echo 76 fatty acids, short chain 700 fetus, diffusion-weighted imaging 714–15 fiber assignment by continuous tracking (FACT) 92 fiber orientation mapping 223 stroke 228–30 fiber reconstruction techniques 91 fiber tracking 4, 69, 73 algorithms 318 fiber tract trajectories 560 fibroblast growth factor, basic 279–80 Fick’s first law 55 flip angle 40 flow encoding arterial spin tagging (FEAST) 211 flow-sensitive alternating inversion recovery (FAIR) 125 inflow time 152 subtraction errors 156

flow-sensitive alternating inversion recovery with an extra radiofrequency pulse (FAIRER) 125 fluid attenuated inversion recovery (FLAIR) 79 alcoholism 570 anaplastic oligodendroglioma 349 aspergillosis 364 brain abscess 382 chronic carotid artery disease 248, 254 Creutzfeldt–Jakob disease 372, 373, 428 cysticercosis 365 empyema 357, 358, 416, 417 encephalitis 367 Epstein–Barr virus 392 head injury 630, 632, 633 boxing 638 HSV1 368, 369 hypoxic encephalopathy 698 inborn errors of metabolism 781 leptomeningeal enhancement 254 meningitis 356, 414 pediatric tumors 650, 724–5 progressive multifocal leukoencephalopathy 370 radiation necrosis vs. recurrence 350 reversible posterior leukoencephalopathy 230 subdural empyema 416, 417 toxoplasmosis 365, 421 tuberculous meningitis 360 West Nile encephalitis 427 flumazenil, carbon-11 labeled 484, 485 Fourier transform 3, 7, 28, 29 convolution theorem 113 free-induction decay 27–8, 29 free-radical scavengers 272 frequency domain 27–8, 29 frontal lobe aging 561 Huntington’s disease 601 phosphomonoesterase 538 schizophrenia phospholipid membrane metabolism abnormality 548 fronto-temporal degeneration (FTD) 525 MRS 525 proton MRS 583–4 functional MRI arterial spin echo-based of brain 130–2 blood oxygen level dependent 130–2 perfusion 131 functional MRI-BOLD 264 head injury 635 migraine 268, 269 red nucleus 270 substantia nigra 270 fungal disease 359 granuloma 396, 397 intracranial vascular 362 MRS 394–6 gadolinium contrast agents 49, 109 blood–brain barrier permeability 115 brain abscess distinction from tumor 413 FLAIR imaging 254 pediatric brain tumors 722, 727 MRS combination 732

pediatric tumors 725 perfusion-weight imaging 183 radiation therapy protocol 726 gadolinium dimeglumine gadapentetate (GdDTPA) 155, 156, 207, 280 DSCI-MRI 334 head injury 630 seizure disorders 510 tumor diagnosis 317 galactocerebroside b-galactosidase deficiency 757 Gall FJ 1 gamma-variate fitting 144, 146 DSCI-MRI 332, 333, 334 gangliocytoma, dysplastic 723 ganglioglioblastoma 730 ganglioglioma 494, 730 general affective disorder 538, 542 general anesthesia, pediatric patients 737 genu aging 561 alcoholism 569 Alzheimer’s disease 565–6 Gerstmann–Straussler–Scheinker disease 526–7 giant cell encephalopathy 570–1 Glasgow Coma Scale (GCS) 609, 614 traumatic brain injury severity 620 Glasgow Outcome Score (GOS) 620 glial acidic fibrillary protein (GFAP) 768, 782 glial proliferation 619 Alexander disease 768 glioblastoma multiforme 279, 297 angiogenesis 279 brachytherapy 301 brain abscess differential diagnosis 382 MRS 730 multiple sclerosis differential diagnosis 431 pediatric 726 relative cerebral blood volume 343 transfer coefficient 343 glioma brain abscess differentiation 354–5 brainstem 722 chiasmatic 732 differentiation from metastases 328 diffusion tensor imaging 327 DRCE-MRI 339, 343 DSCI-MRI 339, 340, 341–2 extravascular extracellular space 343–4 infiltration 327 optic nerve 723 peri-tumoral region 341–2 relative cerebral blood volume 341 relative recirculation value 342 transfer coefficient 343 VEGF production 342 glioma, malignant 279 angiogenic therapy response 280–2 chemotherapy response 300, 731 contrast enhancement 299 imaging endpoints 281 metabolic heterogeneity 306 metastases 340 microscopic anatomy 283 necrotic 315, 316

815

816

Index

glioma, malignant (contd) pediatric 731 radiotherapy response 731 recurrence 279 relative cerebral blood flow/volume 281–2 surrogate markers of drug effectiveness 281 tumor burden 283 gliomatosis cerebri 730 gliosis 516, 567 global energy minimization 91 global ischemia/hypoxia, arterial spin labeling 132–3 globoid cell leukodystrophy see Krabbe’s disease globus pallidus abnormalities 781 glucose metabolism and seizures 516 glutamate (Glu) 595 adrenoleukodystrophy 760 aging 587 dementia 587 epilepsy 491, 492 general affective disorder 542 Huntington’s disease 601 hypoxic brain injury 690, 694, 700 hypoxic encephalopathy 693 hypoxic–ischemic brain injury 649 hypoxic–ischemic cascade 691 hypoxic–ischemic encephalopathy 692 neonatal brain injury 682 obsessional compulsive disorder 546 pediatric imaging 648, 740 Pelizaeus–Merzbacher disease 766, 767 post-contrast spectroscopy 49, 50 proton MRS 9, 13 proton spectra peaks 30 psychiatric disease 551 reuptake 628 traumatic brain injury 622 glutamate (Glu) antagonist 595, 611 glutamine (Gln) 595 adrenoleukodystrophy 760 aging 587 dementia 587 epilepsy 491, 492 general affective disorder 542 Huntington’s disease 601 hypoxic brain injury 693, 694, 700 hypoxic encephalopathy 693 hypoxic–ischemic brain injury 649 hypoxic–ischemic cascade 691 hypoxic–ischemic encephalopathy 692 neonatal brain injury 682 obsessional compulsive disorder 546 pediatric imaging 648, 740 proton MRS 9, 13 proton spectra peaks 30 psychiatric disease 551 glycerophosphocholine 291 chronic hypernatremia 619 malaria 403 pediatric imaging 648 glycine hydatid cysts 398, 401 non-ketotic hyperglycinemia 792 glycolipids, phenolic 388 glycolytic pathway 382, 385

gradient echo scans HSV1 368 neurocysticercosis 398 pediatric tumors 723 gradient fields, non-linearity 106 granulomatous disease 358–9, 360, 361–3 fungal 396, 397 gray matter aging 560 changes 579 Alexander disease 767, 768, 769 Alzheimer’s disease 565, 581 anisotropy 223 biochemical injuries 780 brain maturation 678, 679 Cho signals 20, 21, 22 diffusion 224 diffusion tensor eigenvalues 680 electron density 2 fractional anisotropy 92 stroke 227 frontal volume in schizophrenia 567 lesions 558 loss in Alzheimer’s disease 565 metabolite concentrations 36, 534 MRS data analysis 499 metachromatic leukodystrophy 756 NAA signals 20, 21, 22 neonatal diffusion-weighted imaging 707 normal appearing 444, 448–9 multiple sclerosis 448–50, 451 Parkinson’s disease 596 partial volume effect 100 pediatric patients 649 Pelizaeus–Merzbacher disease 766 periaqueductal 270–1, 272 polyhydric alcohols metabolism defect 794 regional blood flow 120 segmentation 90 stroke diffusion 224 fractional anisotropy 227 toxoplasmosis lesions 466 traumatic brain injury 613–14, 616, 617 water diffusion characteristics 448 mean diffusivity 79 water content 33 relaxation time 4 guanidinoacetate methyltransferase (GAMT) 683 deficiency 684, 782, 784 creatine supplementation 785, 786 Haemophilus influenzae 380 hamartoma, hypothalamic 499 head coils, pediatric 675, 707, 737 head injury 609–12, 613–22, 626–39, 642–3 acquisition technique 614 analysis 614 apparent diffusion coefficient 630, 631–2, 633, 634 non-accidental in child 637 arterial spin labeling 631 MRI assessment 132–3 axonal injury 613, 614, 631 detection 633–4

blood–brain barrier leak 626–8, 632 boxing 636, 638, 639, 643 brain atrophy 614 brain metabolism changes 616–17 cerebral blood flow 626–9, 631, 636, 639 cerebral blood volume 627, 631, 636 clinical rating scales 614 cognitive function 619–21 diffusion tensor imaging 633, 634 diffusion-weighted imaging 619, 630, 631–4 complex changes 632–3 disability 613 edema cytotoxic 611, 627, 631, 639 vasogenic 631–2 experimental data interpretation 631 fluid attenuated inversion recovery 632, 633 focal 615 focal monitor 611 global monitor 611 histology 613–14 hyperperfusion 626 incidence 613 interpretation of findings 636, 639 ischemia 628, 629, 633 late sequelae 611 management algorithm 609, 610 metabolic cascade 614 metabolites for quantification 614 outcome prediction 620 microvascular injury 611 mortality 613 MRI in experimental models 630–1 MRS 614–22 clinical application 621–2 neuronal injury 615 neuronal integrity pattern mapping 622 outcome 619–21 oxygen extraction fraction 629 pathophysiology 626–30, 636 perfusion abnormalities 636 perfusion-weighted imaging 631, 634–6 phosphorus-31 MRS 621 post-injury time metabolite analysis 619 prognosis 609 proton MRS 615–22 severity 609, 614, 619–20 assessment 613 tissue location 614 white matter boxing 638 diffusion tensor imaging 631 see also brain injury headache hemochromatosis 272 mechanisms 269–72 see also migraine/migraine pathogenesis heart/heart disease cerebral perfusion pressure 256 congenital in pediatric stroke 736 congenital malformations 662 embolic stroke 256, 257 diagnosis 186, 188 embolism 742 hemangioblastoma 723, 726

Index

hematopoietic stem cell transplantation (HSCT) 760, 761 hemiparesis, pediatric 737, 741 hemiplegia congenital 685 pediatric stroke 742, 742, 743 perinatal stroke 710 hemochromatosis 272 hemodynamic failure 247 severe obstructive carotid artery disease 241–2 hepatic encephalopathy, myo-inositol 13 hereditary leukoencephalopathy choline elevation as sign of active demyelination 797–8 depletion of main metabolites secondary to cavitations 797 herpes simplex encephalitis 388–9, 391 diffusion-weighted imaging 416–18, 419 Japanese encephalitis distinction 392 herpes simplex virus 1 (HSV-1) 367–8, 369, 388–9, 416–18, 419 herpes simplex virus 2 (HSV-2) 388–9 herpes zoster 368 highly active antiretroviral therapy (HAART) 462–3 progressive multifocal leukoencephalopathy 470 hippocampal sclerosis 481, 483 conventional MR 490 diffusion-weighted imaging 515 limbic temporal lobe epilepsy 484–5 histological techniques 2 HIV dementia 460–6, 571 brain tissue microstructural integrity 465 diffusion-weighted imaging 464–6 metabolites 461, 462 abnormality reversal with antiretroviral treatment 462–3 regional variations in abnormalities 462 perfusion magnetic resonance imaging 464, 465 proton MRS 461, 462 HIV encephalitis 371–2 HIV encephalopathy 229 pediatric patients 463–4 HIV infection 570–1 anisotropy 421 brain lesion differential diagnosis 466 brain sequelae 570–1 CD4 cell count 466 cerebellar degeneration 571 cryptococcal meningitis 361 diffusion tensor imaging 571 diffusion-weighted imaging 418, 420–1 encephalitis 371–2 focal brain lesions of combined etiologies 472 giant cell encephalopathy 570–1 granulomatous disease 359 minor cognitive motor disorder 461 MRS of brain disorders 460–73, 478 multi-modality imaging 472 neuropsychiatric impairment 420 opportunistic focal brain lesions 466–72 physiological MRI of brain disorders 460–73, 478

therapy 420–1 tuberculosis 385, 471–2 viral load 571 holoprosencephaly 665–6, 667 Human Genome Mapping Project 780 Huntington’s disease 595, 600–2 akinetic rigid state 601 conventional MRI 600–1 MRS 601 presymptomatic gene carriers 601, 602 hydatic disease 398–9, 401, 403 hydrocephalus meningitis 359, 360 obstructive 359 3-hydroxy-3-methylglutaryl-coenzyme A lyase deficiency 798–9 L-2-hydroxyglutaric aciduria 798, 800 hyperglycemia, non-ketotic 683 hypernatremia, chronic 619 hyperperfusion in head injury 626 hypertension management 215 hypertensive encephalopathy 229 hyperventilation, apparent diffusion coefficient 632 hypervolemic hypertensive therapy 176 hypoglycemia, neonates 799 hypoperfusion cerebral embolization 214 cerebral ischemia 215 cerebrovascular disease 214–15 stroke 208 hypoxia monitoring during MRI 185 prognosis 691 see also brain injury, hypoxic hypoxia–ischemia cascade 691–2 hypoxic–ischemic brain injury 690–1, 692 apparent diffusion coefficient 711, 712 diffusion-weighted imaging 661, 662, 663, 700, 702 neonate 708, 710–12 MRS 700, 702 neonates 649, 661, 662, 663, 706 diffusion-weighted imaging 708, 710–12 neuronal injury 700 pathobiology 692–3 patient outcome 702 proton MRS 649, 694–5, 696–9, 699–700, 702 imaging dysfunctional 1 endpoints 281 gradients 104 inborn errors of metabolism 649–51, 654 amino acidopathies 798–9, 800 clinical diagnostic criteria 780–7, 788, 789–99, 800 conventional MRI 781–2 diffusion-weighted imaging 666–7 genetics 779–80 hereditary 780–1 laboratory work-up 782, 783 MRS 782–7, 788, 789–98 MRS 650–1, 654, 683, 779–99, 800, 803 non-specific findings 798, 799 suggestive findings 795–6

neonates 713–14 organic acidopathies 798–9, 800 polyhydric alcohols metabolism defect 793–8 infants developmental delay 683–5 sedation 675 technical aspects of MRI 675–6 see also neonates infections 296, 353 CNS granulomatous 359 diffusion-weighted imaging 408–23 hematogenous spread 354 host response 354 MRS 380–403, 407 parasitic 363–6 MRS 396–9, 400, 401, 402, 403 pediatric patients 662, 664, 665 stroke 736 pyogenic 354–7, 358 MRS 380–2, 383–4, 385 systemic signs 354 tuberculous 385–6, 387, 388 viral 366–7 MRS 388–9, 390, 391–4 inflammation 353 meninges 359, 360, 361 inflow time 152 informative techniques 2 inheritance, mitochondrial/multifactorial 780 inositol depletion hypothesis 538 subacute sclerosing panencephalitis 392 see also myo-inositol (mI) interferon 280 International Classification of Disease 9 (ICD-9), hypoxic brain injury 700 intervoxel coherence 566 intervoxel orientational coherence 566 intracranial epidermoids 661–2, 664 intracranial pressure raised in head injury 626–7 waveform analysis 611 intracranial vascular abnormalities 258–9 intracranial vascular disease 361–3 intravoxel dephasing 102 intravoxel incoherent motion (IVIM) 101 iron homeostasis in brain 271 migraine pathogenesis 270–1, 272 ischemia see cerebral ischemia; hypoxic–ischemic brain injury Japanese encephalitis 392–3 JC virus 447, 469 juvenile pilocytic astrocytoma 649, 651, 723 contrast enhancement 725–6 diffusion-weighted imaging 726 MRS 729–30 ketogenic diet 485 knee coil, adult 707 Korsakoff’s syndrome 570 Krabbe’s disease 667, 757–8, 759 Kreb’s cycle 385, 793 inborn errors of metabolism 651

817

818

Index

k-space acquisition 76, 79, 105–6 segmentation 103 trajectories 74–5 traversal 74–5 kuru 372 lactate (Lac) 4, 595 accumulation in brain 49 N-acetyl aspartate absence 11 adrenoleukodystrophy 761 Alexander disease 768, 769 Aspergillus abscess 394, 395 brain tumor MRS 289, 292 meningioma 309 recurrent astrocytoma 310 therapy response assessment 299–300 brain tumors 171 carotid endarterectomy 238 cerebral injury correlation 238 cerebral ischemia 172 cerebrovascular reactivity 241 epilepsy 491, 492 herpes simplex encephalitis 391 HIV dementia in pediatric patients 464 Huntington’s disease 601 hydatid cysts 398, 401 hypoxia–ischemia cascade 691 hypoxic brain injury 690, 693, 700 hypoxic–ischemic brain injury 649 hypoxic–ischemic encephalopathy 692 Japanese encephalitis 393 lymphoma 467, 469 malaria 403 maple syrup urine disease 790 MELAS 803 meningitis 381 mitochondrial disease 171 mitochondrial encephalopathy 796–7 MRS in stroke 165 multiple sclerosis 431, 432 myelinopathia centralis diffusa/vanishing white matter disease 770, 771 neonatal brain injury 682 neurocysticercosis 398, 399 neurological impairment correlation 238 panic disorder 542 para-infectious encephalopathy 393, 394 pediatric imaging 648, 739, 740 brain infarction 649 HIV dementia 464 neoplasms 649, 650 post-contrast spectroscopy 49, 50 primary neoplastic lesion differentiation 297 progressive multifocal leukoencephalopathy 469, 478 proton MRS 9, 12 signal 294 tumefactive multiple sclerosis 308 Reye’s syndrome 705 severe obstructive carotid artery disease 242–3 stroke 169, 171, 173, 174–5, 181 MRS 165 patient selection for thrombolysis 175 subacute sclerosing panencephalitis 392

succinate dehydrogenase deficiency 793 traumatic brain injury 615, 622 tuberculous abscess 388 venous thrombosis 695 visual cortex 171 lactic acid 516 lactic acidosis 796 lacunar infarcts 164 multiple 228 language difficulties 744 Larmor equation 57 lattice index 223, 560 Leigh syndrome 763 lentiform nucleus aging 561 multi-system atrophy 597 leptomeningeal enhancement in meningitis 414 leptomeningeal vessels collateral flow 234, 240 unilateral enhancement 254 leukariosis 228 leukodystrophy Alexander disease 766 metachromatic 755–7 with ovarian dysgenesis 439–40 see also adrenoleukodystrophy; Krabbe’s disease leukoencephalopathy 781 pediatric 10 polyhydric alcohols metabolism defect 794 posterior 229 reversible 229–30, 230, 443 see also Canavan’s disease; progressive multifocal leukoencephalopathy Lewy bodies 583 Lhermitte–Duclos syndrome 723 Li–Fraumeni syndrome 723 linear combination (LC) model 30 line-propagation alogrithms 91–2 termination criteria 92–3 white matter branching 93 lipids brain injury hypoxic 693 traumatic 615 cryptococcoma 471 hydatid cysts 398 hypoxic–ischemic encephalopathy 692 lymphoma 467, 469 progressive multifocal leukoencephalopathy 469, 478 proton MRS signal 294 Sjogren–Larsson syndrome 438 stroke 171 suppression 18, 20, 51 3D MRSI in Alzheimer’s disease 588 very selective saturation pulses 49 tuberculous abscess 388, 389 Listeria 380 lithium 538 psychiatric disease 552 low-flow infarcts 235 MRS technique 238–9 low-pass filtering 28, 29

lymphoma, CNS 468–9 AIDS patients 420, 466, 467, 468–9 contrast enhancement 299 diagnosis 366 multicentric 469 pediatric patients 662, 665 proton spectra 27, 28 ring-enhanced lesions 467, 469 lysosomal disorders 755–8, 759–60 lysosomes, sialic acid accumulation 789 macrocephaly 771, 781, 786 macromolecules hypoxic brain injury 700 stroke 171 traumatic brain injury 615 magnesium migraine pathogenesis 266, 267 progressive supranuclear palsy 598 magnetic resonance angiography (MRA) 163 internal carotid artery 249, 250 intracranial vascular disease 363 meningitis 356 pediatric tumors 725 spindle cell sarcoma 724 magnetic resonance imaging (MRI) angiography 120 comprehensive protocol 670 conventional boxing 636, 638 head injury experimental models 630–1 inborn errors of metabolism 781–2 neurodegenerative diseases 596, 598–9, 600–1 pediatric tumors 723 Reye’s syndrome 705 development 3–4 multi-modal 732 signal for metabolite spectra 35 volumetry 525 see also functional MRI; functional MRI-BOLD magnetic resonance morphometry, voxel-based 611 magnetic resonance signal diffusion 55–8 magnetic resonance spectroscopic imaging (MRSI) 14, 38, 39 acquisition mode 42–5 epilepsy 497 temporal lobe 508 multiple-voxel techniques 15–16, 19, 51, 695, 696 data set quantification 35–6 gray matter concentrations 36 pediatric brain 648 Reye’s syndrome 705 single-voxel technique comparison 16–18, 19 stroke 173, 174–5, 181 traumatic brain injury 620–1 white matter concentrations 36 out-of-band SAT 49 pediatric brain tumors 722 psychiatric disease 551–2 Rasmussens’s encephalitis 507 stroke 173, 174–5, 181 outcome measures 176 three-dimensional 588, 676

Index

magnetic resonance spectroscopy (MRS) 4, 7–23, 31–6 acute disseminated encephalomyelitis 437, 442 adrenoleukodystrophy 758, 760–1, 778 Alexander disease 766–8, 769 algorithm fitting 30 Alzheimer’s disease 593 amino acidopathies 798–9, 800 automatic integration 30 baseline correction 29 brain abscess 356, 381–2, 383–4, 385 brain spectra anatomical variations 20, 21, 22–3 brain tumors 288–301, 306–11 CADASIL 437 Canavan’s disease 763–5, 786–7, 788, 789 cerebrotendinous xanthomatosis 437–8 coherence control 41–2 control region spectrum 33, 35 cortico-basal degeneration 598–9 coupled spins 44 creatine deficiency syndrome 782–6 Creutzfeldt–Jakob disease 372, 526–7 cryptococcoma 471 data acquisition 40–7, 48, 49, 50, 51 data analysis/quantification 20 dementias 525 demyelinating diseases 373, 374 disease-related abnormalities 429–30 echinococcosis 398–9, 401, 403 echo time averaging 49, 50 encephalitis 367 epidural empyema 357 Epstein–Barr virus 391–2 external reference method 34–5 fungal disease 394–6 hereditary leukoencephalopathy 797–8 hereditary metabolic disorders 782–7, 788, 789–98 higher magnetic field strength 587, 595 high-field 502 HIV encephalitis 372 HIV-associated brain disorders 460–73, 478 focal brain lesions of combined etiologies 472 pediatric 463–4 HSV1 368, 369 Huntington’s disease 601 3-hydroxy-3-methylglutaryl-coenzyme A lyase deficiency 798–9 hypoxic brain injury 690–703, 705 hypoxic–ischemic brain injury 700, 702 improved coverage methods 51 inborn errors of metabolism 650–1, 654, 683, 779–99, 800, 803 infections 354, 380–403, 407 pyogenic 380–2, 383–4, 385 tuberculous 385–6, 387, 388 viral 388–9, 390, 391–4 inherent spatial resolution 430 internal carotid artery 169, 171 intracranial vascular disease 363 Japanese encephalitis 392–3 leukodystrophy with ovarian dysgenesis 439–40

lysosomal disorders 755–8, 759–60 magnetization transfer effects 45 malaria 402, 403 manual integration 30 maple syrup urine disease 789–91 megalencephalic leukoencephalopathy with subcortical cysts 771, 772, 773 meningitis 359 pyogenic 380–1 mental retardation with global NAA deficiency 786, 787 metabolic disorders, hereditary 782–7, 788, 789–98 metabolites concentration quantification 32–5 peak areas 29–31 signal strength 27 middle cerebral artery 169, 171 mitochondrial encephalopathy 761, 763, 764–5, 796–7 with lactic acid and stroke 803 model spectra fitting 30 mucolipidosis type IV 437 multiple sclerosis 376, 430–1, 432, 433–4, 435, 436–7 myelinopathia centralis diffusa/vanishing white matter disease 768, 770, 771 neurocysticercosis 396–8, 399, 400 neurodegeneration 594–602, 605 non-ketotic hyperglycinemia 792–3 nuclei for radiolabeling 500–1 L-2-OHG aciduria 798, 800 organic acidopathies 798–9, 800 outer volume suppression 47, 48, 49 para-infectious encephalopathy 393–4 parasitic infections 396–9, 400, 401, 402, 403 parenchymal disease 361 parkinsonian syndromes 595–9, 600 patient movement 176 peak fitting 30 pediatric brain 647–51, 652–3, 654 applications 649–51, 652–3, 654 stroke 748–50 technical aspects 675–6 tumors 729–32 white matter disease 755–73, 778 Pelizaeus–Merzbacher disease 765–6, 767, 798, 799 phase correction 29 phenylketonuria 791 pitfalls 38 pre-scan 38–40 polyhydric alcohols metabolism defect 793–8 prion diseases 526–7 progressive multifocal leukoencephalopathy 469 progressive supranuclear palsy 598, 600 proton MR spectra 8, 9, 10–13 psychiatry 529–52 studies 535–51 pulse sequence 534 quantification of spectral peaks 27, 28, 31–6, 51 radiation necrosis 726 relaxivity 49

ribose-5-phosphate isomerase deficiency 773 Salla disease 439, 789 schizophrenia 567 seizure disorders 491–502, 507–8 severe obstructive carotid artery disease 234–43 shimming 39 single-voxel techniques 13–15, 648 echo time 16, 17, 18 epilepsy 497 hypoxic encephalopathy 694, 695, 697 SI comparison 16–18, 19 Sjogren–Larsson syndrome 438 slice order 42–5 software development 498–9 spatial localization 13–18, 19, 20 spectra quantification 534–5 spectral analysis 27–31 spectroscopic imaging acquisition mode 42–5 status epilepticus 491 stroke 168–77 clinical applications 175–6 ischemic penumbra 165 prognostic indicator 176 subacute sclerosing panencephalitis 392 succinate dehydrogenase deficiency 793 techniques 594–5 time-domain fitting 30 tissue volume correction 534–5 toxoplasmosis 466–7 transient ischaemic attacks 165–6 transmitter gain setting 39 traumatic brain injury 614–22 clinical application 621–2 in vivo 7 water reduction 40 reference 46 reference signal 33–4 sidebands 46–7, 51 suppression 44–5 see also phosphorus-31 MRS; proton MRS magnetic resonance tractography 79, 86–96, 452–3, 685 limitations 94–5 validation 95 magnetic resonance venography (MRV) 356 magnetization prepared rapid acquisition gradient echo (MPRAGE) 489 magnetization transfer arterial spin labeling 208 dementia 526 maps 470 neurocysticercosis 398 pediatric patients 655, 725 progressive multifocal leukoencephalopathy 469–70 toxoplasmosis 467, 468 tuberculoma 386, 388, 389 tuberculous meningitis 385 magnetization transfer (MT) 123, 154 effects 45, 123 techniques 120 magnetization transfer contrast (MTC) 127, 128

819

820

Index

magnetization transfer ratio neurocysticercosis 398 progressive multifocal leukoencephalopathy 470 tuberculoma 386 tuberculous meningitis 385 magneto-encephalography 264–5 cortical spreading depression 268 headache mechanism 269 malaria 402, 403 maple syrup urine disease 683, 714, 789–91 Canavan’s disease differential diagnosis 787, 790 clinical phenotypes 789–90 diagnosis 790 Marchiafava–Bignami disease 570 mastoiditis, subdural empyema 357, 416 maximum peak concentration (MPC) 147, 148, 149 Maxwell gradients 106 mean transit time (MTT) 251 bolus delay/dispersion 144, 145 cerebral perfusion imaging 114, 116 dynamic susceptibility contrast imaging 142 neonates 717, 718 overestimation 143 pediatric patients 654 brain tumors 728 perfusion-weighted imaging 183 seizures 510 medulloblastoma 652, 664 pediatric 722, 723 diffusion-weighted imaging 726 MRS 729–30 treatment failure 726 megalencephalic leukoencephalopathy with subcortical cysts 771, 772, 773, 797 MELAS 749, 750, 803 membrane phospholipid metabolism 538 memory impairment, age-associated 584 meninges granulomatous disease 359 inflammation 359, 360, 361 meningioma 343 extravascular extracellular space 343–4 MRS 309 pediatric 723 relative cerebral blood volume 342–3 transfer coefficient 343 meningitis 359, 360, 361 bacterial 356 subdural empyema spread 356, 357 carcinomatous 359 cryptococcal 361, 362, 466 diffusion-weighted imaging 414–15, 416 pyogenic 359, 380–1 tuberculous 360, 385, 386 ventriculitis 415, 416 viral 359 mental retardation 690 creative deficiency syndrome 782 with global N-acetyl aspartate deficiency 786, 787, 788 Mescher–Garwood pulse (MEGA) 45 metabolic disorders see inborn errors of metabolism

metabolites chemical shift registration 44 concentration quantification 32–5 peak area determination 29–31 pediatric 648 changes with age 739–40 ratios 32–3, 35 sequence timing 44 signal strength 27, 29, 31 signal-to-noise ratio of spectrum 39 spectroscopic imaging 38 see also named metabolites metachromatic leukodystrophy 755–7 metastases brain abscess differentiation 354–5 brain tumors 296 cerebral 341 pediatric 722, 726 diffusion-weighted imaging 412–13 gliomas 340 differentiation 328 transfer coefficient 343 methylmalonic acidemia 749, 750 microtubules, degradation with aging 560 microvasculature pharmacokinetic analyses 337 tumors 330–1 migraine/migraine pathogenesis 263–72, 275 with aura 266–9 BOLD signal 264 brainstem 269–70 concepts 263–4 familial hemiplegic 266, 271 free-radical damage 272 functional MRI-BOLD 264, 268, 269 headache mechanisms 269–71 hemochromatosis 272 interictal status 264–6 iron levels 270–1, 272 magnesium changes 266, 267 neuron cell membrane excitability 265 periaqueductal gray matter 270–1, 272 phosphocreatine 265, 266 phosphodiesters 265 phosphomonoesters 265 phosphorus spectroscopy 266 regional cerebral blood flow 267, 268, 269 without aura 269–71 misery perfusion 211 mitochondria NAA levels 618 NAA pathway 788 mitochondrial cytopathy 595 mitochondrial disorders 171, 781 mitochondrial DNA (mtDNA) 780, 796 mitochondrial dysfunction, Huntington’s disease 600 mitochondrial encephalomyopathy 761, 763, 764–5 maternal transmission 780 mitochondrial encephalopathy 749, 750, 761, 763, 764–5, 796–7 with lactic acid and stroke (MELAS) 749, 750, 803 predominant white matter signal abnormalities 797 succinate dehydrogenase deficiency 793

mixing time 74 modulating factors 31 Morgagni GB 1 motion-induced phase differences 75 motor function, alcoholism 568 movement, subject 150 arterial spin labeling 156–7 dynamic susceptibility contrast imaging 150 MRSI in acute stroke 176 neonate sedation 675, 707 pediatric sedation 737 movement disorders, physiological MRI 527 moyamoya syndrome 144, 145, 256, 258 cerebral blood flow/volume 747 pediatric stroke 737 perfusion-weighted imaging 685, 747–8, 753 sickle cell disease 746 surgical revascularization efficacy 747–8 mucolipidosis type IV 437 mucormycosis 359, 361 intracranial vascular 362 MRS 395–6 multi-modality imaging 472 multiple sclerosis 372–6 acute lesions 446 acute phase 431, 432 axons damage 434, 436–7 loss 447 Wallerian degeneration 448 basal ganglia 449–50 brain volume reduction 446 clinical form 430, 434 demyelination foci 376 diffusion-weighted imaging 187–8, 189, 445–6 clinical findings correlation 451–2 disability 433–4, 436, 452 disease evolution 434 disease-modifying therapies 436–7 gray matter 433 normal appearing 448–50, 451 hypointense lesions 444, 445 macroscopic lesions 451 MR tractography 452–3 MRS 376, 430–1, 432, 433–4, 435, 436–7 normal appearing brain tissue 451, 452 pathology 430 plaque 375–6, 430, 431 primary–progressive type 430, 448, 449, 452 progression 434 relapsing–remitting type 430, 452 secondary–progressive type 430, 448, 449, 452 therapy effectiveness monitoring 434, 437 tissue damage 445–6, 446 tumefactive 308, 459 white matter lesions 376, 430, 431 normal appearing 431, 432, 433–4, 447–50 structure segmentation 453 multiple-region of interest technique 93–4

Index

multiple-voxel (SI) techniques 15–16, 19, 51, 695, 696 2-D 16 3-D 16 data set quantification 35–6 echo time 16 gray matter concentrations 36 multi-slice 16, 19 pediatric brain 648 Reye’s syndrome 705 single-voxel technique comparison 16–18, 19 stroke 173, 174–5, 181 traumatic brain injury 620–1 white matter concentrations 36 multi-system atrophy (MSA) 527, 597 Mycobacterium tuberculosis 385, 388 myelin 61–2 degradation with aging 560 hydrophobic 372 maturation 781 synthesis 618, 789 water diffusivity 447 myelin sheath anisotropy 224 fluid retention mechanisms 790 myelination 680 cholesterol as marker 679 infant brain 675 white matter 738 zones 780 see also demyelinating diseases myelinopathia centralis diffusa/vanishing white matter disease 768, 770, 771 myelinopathy, vacuolating 787 myo-inositol (mI) 595 adrenoleukodystrophy 760, 761 aging 579, 580 Alexander disease 768, 769 Alzheimer’s disease 581–2, 583, 585, 586, 593 Canavan’s disease 765 Creutzfeldt–Jakob disease 527 cryptococcoma 471 dementia 525 epilepsy 491, 493 fronto-temporal degeneration 583, 584 HIV dementia 461 Krabbe’s disease 758 lithium therapy 538 lymphoma 469 megalencephalic leukoencephalopathy with subcortical cysts 773 pediatric imaging 648, 649, 651, 739 Pelizaeus–Merzbacher disease 766, 767 progressive multifocal leukoencephalopathy 470 proton MRS 9, 12–13 traumatic brain injury 617, 618, 619, 622 vascular dementia 583 NAA see N-acetyl aspartate (NAA) National Institute for Clinical Excellence (NICE, UK) guidelines 609 National Institute of Health Stroke Scale (NIHSS) 212 navigator echoes 103 necrotizing encephalopathy, acute 393–4

Neisseria meningitidis 380 neonates 706–18, 721 apparent diffusion coefficient 707–8 arterial input function 717, 718 arterial spin labeling 715 brain injury 682–5 diffusion-weighted imaging 682, 706 perfusion-weighted imaging 706 proton MRS 682, 683–4 brain water content 675, 707 cerebral blood flow/volume 717 diffusion tensor imaging 708 diffusion-weighted imaging 707–15 hypoxic–ischemic brain injury 708, 710–12, 721 normal 707–8 genetic screening 779–80 head size 675 hypoglycemia 799 hypoxic–ischemic brain injury 649, 661, 662, 663 diffusion-weighted imaging 708, 710–12, 721 mean transit time 717, 718 metabolic disorders 713–14 MR sequence adjustment 707 perfusion-weighted imaging 715, 717–18 periventricular leukomalacia 661, 662, 685 preparation for imaging 706–7 sedation 675, 707 seizures 708 size 707 swaddling 707 technical aspects of MRI 675–6 three-dimensional MRSI 676 transportation 675 venous infarction 661, 663 white matter infarction 710, 712, 717, 718 tractography 715 see also prematurity; stroke, perinatal neurinoma, relative cerebral blood volume 342–3 neurocutaneous syndromes, neoplasms 722–3 neurocysticercosis 396–8, 399, 400 neurodegenerative disease 523–8 apoptosis 594 categories 781 diagnostic validation 524 MRS 594–602, 605 patient factors 524 perfusion-weighted imaging in pediatric patients 658 physiological MRI 527 proton MRS 595 technical considerations 524 neurodevelopmental disorders 675 neurofibrillary tangles 584 neurofibroma 723 neurofibromatosis type 1 258, 723 perfusion-weighted imaging 685 neurofibromatosis type 2 723 neurological disorders arterial spin labeling MRI assessment 132–3, 134 internal carotid artery occlusion 237

neurological impairment, lactate 238 neuronal cell death Alzheimer’s disease 565 seizures 516 neuronal–glial substrate cycling 700 neurons Alzheimer’s disease 565, 581 cell membrane excitability 265 elimination of excessive 739 injury hypoxic brain injury 700 traumatic brain injury 615 loss metachromatic leukodystrophy 757 NAA decrease 10–12, 786 migrating in cortical development 681 suppression 267–8 neuropsychological assessment 489 nociceptive centers, cortical–subcortical connection recruitment 269 nodes of Ranvier 558 non-accidental trauma in children 636, 637 non-ketotic hyperglycinemia (NKH) 713, 714, 792–3 diagnosis 792 nuclear magnetic resonance (NMR) 3, 7 obsessional compulsive disorder 542, 544–5, 546 occipital cortex, hyper-excitable 268–9 L-2-OHG aciduria 798, 800 oligoastrocytoma, anaplastic 321, 322 oligodendrocytes N-acetyl aspartate 595 apoptosis 770 oligodendroglioma anaplastic 349 transverse relaxation time 298 ophthalmic artery, collateral flow 234, 240 optic nerve diffusion-weighted imaging 453–4 glioma 723 optic neuritis 454 organic acidopathies 798 osteomyelitis 357 outer volume suppression 47, 48, 49 ovarian dysgenesis 439–40 oxidative phosphorylation disorders 761 inborn errors of metabolism 651 mitochondrial encephalopathy 796 NAA levels in mitochondria 618 proteins 780 oxygen extraction fraction 247 head injury 629 oxygen metabolism seizures 516 see also cerebral metabolic rate of oxygen metabolism (CMRO2) panic disorder 542, 543 para-infectious encephalopathy 393–4 parallel imaging, pediatric 647, 668 paranasal sinuses, fungal disease 359 parasitic infections 363–6 MRS 396–9, 400, 401, 402, 403

821

822

Index

parasympathetic nervous system, headache 269 parenchymal disease 361, 363 PARK1 genes 596 PARK2 genes 596 parkinsonian syndromes diagnosis 595 MRS 595–9, 600 Parkinson’s disease 527, 596–7 glycolytic pathway impairment 596 gray matter 596 Lewy bodies 583 oxidative pathway impairment 596 phosphorus-31 MRS 596 proton MRS 596–7 white matter 596 paroxysmal depolarization shifts 482 partial volume effects (PVE) 144 pediatric patients angiogenesis 656 apparent diffusion coefficient 659, 660 arterial input function 655 arterial spin labeling 655, 656, 657 brain normal maturation 676, 677, 678–82 water content 675, 707, 729, 739 brain tumors 649–50, 651–2, 652–3, 653 combined sequences/technologies 732 diffusion-weighted imaging 661–2, 664, 665 MRS 729–32 perfusion-weighted imaging 656, 658, 659, 728–9 prognosis 731–2 serial monitoring 732 cerebral blood flow 654, 655 stroke 656, 657, 658 cerebral blood volume 654 diffusion tensor imaging 675, 676 basal ganglia infarcts 744 diffusion-weighted imaging 658–67 applications 660–2, 663, 664–7 brain congenital malformations 665–6, 667 cerebral palsy 664–5, 666 changes with age 738–9 inborn errors of metabolism 666–7 ischemia differential diagnosis 743–4 non-accidental trauma 636, 637 stroke 740, 741, 742–4 X-linked adrenoleukodystrophy 666–7, 668 echo time 648 fast spatial encoding 668–9 HIV infection MRS abnormalities 463–4 infections 662, 664, 665 magnetization transfer 655 mean transit time 654 metabolites 648 changes with age 739–40 MRS 647–51, 652–3, 654 applications 649–51, 652–3, 654 stroke 748–50 technical aspects 675–6 white matter disease 755–73, 778 neoplasms 649, 650, 651, 652–3 neurodegenerative disease 658

neurodevelopmental disorders 675 non-accidental trauma 636, 637 normal values of physiological imaging 737–8 parallel techniques 647, 668 perfusion-weighted imaging 654–6, 657, 658 applications 655–6, 657, 658 changes with age 739 stroke 655–6, 657, 744, 745, 746–8 phosphorus-31 MRS 675–6 physiological MRI 647–70, 674–85 normal values 737–8 practical issues 737 proton MRS 648, 675, 676 pulse sequence parameter modification 737 radiation necrosis 649, 653, 658 reconstruction software 668–9 sensitivity-encoding 668–9 sickle cell disease 655–6, 657, 658 stroke arterial ischemic 748–9 diffusion-weighted imaging 740, 741, 742–4 MELAS 749, 750 metabolic 749–50 moyamoya syndrome 747–8 MRS 748–50 perfusion-weighted imaging 744, 745, 746–8 sickle cell disease 744, 745, 746–7 technical aspects 675–6 three-dimensional MRSI 676 transportation 675 white matter disease MRS 755–73, 778 see also inborn errors of metabolism; infants; neonates Pelizaeus–Merzbacher disease 765–6, 767, 798, 799 percutaneous transluminal angioplasty 253, 254, 255–6 perfusion-weighted imaging 4, 5, 131 brain abscess distinction from tumor 413 brain injury 706 brain tumors in pediatric patients 656, 658, 659, 728–9 cerebral blood flow/volume 183 chronic carotid artery disease 248–9 dementia 525 developmental delay 685 endogenous diffusible tracer methods 655 epilepsy 484 gadolinium bolus 183 head injury 631, 634–6 hemodynamic status 247 HIV dementia 464, 465 ictal 509–12 interictal 509–12 mean transit time 183 moyamoya syndrome 747–8, 753 neonates 715, 717–18 brain injury 706 neoplasias 329–44, 345, 349–50 neurodegenerative disease in pediatric patients 658

pediatric patients 654–6, 657, 658 applications 655–6, 657, 658 brain tumors 656, 658, 659, 728–9 changes with age 739 stroke 655–6, 657, 744, 745, 746–8 seizures 509–12 sickle cell disease 744, 745, 746–7 stroke 182, 183 ischemic penumbra 164 lesion development 190, 192–4, 195, 196 lesion reversal 205 patient selection for thrombolysis 175 pediatric patients 655–6, 657, 744, 745, 746–8 recoverable vs. permanent damage 193–4 Sturge–Weber syndrome 754 subarachnoid hemorrhage 206 time to reach peak concentration 183 toxoplasmosis 420, 467, 468 tumefactive multiple sclerosis 459 see also arterial spin labeling; dynamic susceptibility contrast imaging (DSCI) perfusion-weighted imaging/diffusionweighted imaging (PWI/DWI) mismatch region 175, 182 stroke 194, 195, 196 acute 212, 213 pediatric 217–18 perfusion-weighted signal 152, 153, 154 Periodically Rotated Overlapping ParallEL Lines with Enhanced Reconstruction (PROPELLER) 76, 79, 80 periventricular leukomalacia, neonates 661, 662, 685 diffusion-weighted imaging 712–13 peroxisomal disorders 713 phantom replacement 35 phase differences, motion-induced 75 phase encoding techniques 120 phase gradient 103, 106 phase-encoding direction 78 phenylalanine 791 phenylketonuria 654, 683, 791 phosphate, inorganic 594, 675 polysubstance abuse 622 progressive supranuclear palsy 598 phosphatidylcholine 292 phosphatidylethanolamine 292 phosphocholine 291 pediatric imaging 648 phosphocreatine 12, 293–4, 594, 675, 783 anaerobic metabolism 516 brain tumor proton MRS 289 hypoxic brain injury 690 migraine 265, 266 phosphodiesterase 548 phosphodiesters 265 Alzheimer’s disease 587–8 phosphoesters 499–500 phospholipid metabolism 293 polysubstance abuse 622 phosphomonoesterase 538, 548 phosphomonoesters Alzheimer’s disease 587–8 migraine 265

Index

phosphorus-31 MRS 7, 499–500, 501, 594 alcohol abuse/dependence 548 Alzheimer’s disease 587–8 brain injury hypoxic 693 traumatic 621 Huntington’s disease 602 hypoxic–ischemic encephalopathy 692 migraine 266 Parkinson’s disease 596 pediatric patients 675–6 progressive supranuclear palsy 598 schizophrenia 548 substance-related disorders 622 physiological MRI 353–4, 523 dementias 524–6 HIV-associated brain disorders 460–73, 478 pediatric patients 674–85 brain tumors 722–32, 735 stroke 736–50, 753–4 Pick cells 598 pineal tumors, pediatric 722 pituitary adenoma 732 pixels of interest (POIs) 93–4, 94 Plasmodium falciparum 403 platelet-derived growth factor (PDGF) 728 point resolved spectroscopy (PRESS) 14–15, 17, 18, 38 coherence control 41–2 coupled spin systems 44 hypoxic brain injury 694, 695, 696–8 hypoxic encephalopathy 693 J-refocused 49 Krabbe’s disease 758 TE-averaged 46–7 VSS pulse combination 47, 48, 49 polioencephalopathy 781 polyhydric alcohols, metabolism defect 793–8 polymorphonuclear neutrophils 354 pontine atrophy 597 positron emission tomography (PET) 2, 489 brain development 738 cerebral metabolic demand in brain development 739 epilepsy 509 FDG 483, 484 antiepileptic drug pharmacodynamic imaging 485 head injury 611, 629, 630 HIV dementia 464 neurodegenerative disease 527 pediatric patients 654 perfusion-weighted imaging correlation 635–6 radio-labeled water detection 119 stroke cerebral blood flow correlation with tissue damage 192–3 ischemic penumbra 164 patient selection for thrombolysis 175 post-contrast spectroscopy 49, 50 pediatric patients 725 postpartum vasculopathy 215 post-traumatic stress disorder 546, 547 potassium, hypoxic brain injury 690

P/Q voltage-gated calcium channel 266, 271 prefrontal cortex, dorsolateral 534 prematurity 706 anisotropy 681 brain development 713 injury 712–13 cerebral cortex 681 diffusion tensor imaging 678 diffusion-weighted imaging 712–13 hypoxic encephalopathy 691 proton MRS 682 spastic quadriparesis 685 white matter injury 684 premyelination phase 678, 739 primary lateral sclerosis 605 primitive neuro-ectodermal tumor 316, 662 pediatric 722, 726, 727 MRS 730 prion diseases 372, 373 physiological imaging 526–7 spread 422 see also Creutzfeldt–Jakob disease Probst bundles 713, 714 progressive multifocal leukoencephalopathy 370–1 diffusion tensor imaging 571 diffusion-weighted imaging 447 HIV infection 466, 469–70, 478 MRS 469 progressive supranuclear palsy (PSP) 527, 597–8, 600 propanediol 692 proteolipid protein (PLP) gene mutation 765 proton decoupling 499–500 proton magnetic resonance spectroscopic imaging (MRSI) 582 proton MRS 7–8, 594 acceleration 51 N-acetyl aspartate 9, 10–12, 294 adrenoleukodystrophy 760–1 aging 579–80 Alzheimer’s disease 580, 581–3 biochemical marker 584–7 clinical 586  587 preclinical 585 prodromal 585–6 brain normal maturation 676, 677, 678 temperature measurement 13 brain tumors 282–4, 288–301, 306–11 clinical significance of diagnosis 298 diagnostic uses 295–8 echo time 289, 290 general features 288–9, 290, 291 grading 297 guidance of therapeutic procedures 298–9 malignant transformation 300 metastatic 296 necrosis 295 neoplastic recurrence 300–1 pattern classification techniques 297 pediatric 729–32 preoperative guidance 299 primary neoplastic lesion differentiation 296–8

radiation necrosis 300–1 recurrence 301 spectra variability 291 therapy response assessment 299–300 tumor margin definition 298 voxel position 297–8 Canavan’s disease 764–5, 788 choline 9, 12 signal 291–3 compound concentration 13 creatine 9, 12 signal 293–4 dementia 580–4 developmental delay 683–4 disease conditions 13 echo times 8 epilepsy 484, 491–2, 493, 494–5, 496, 497 extra-temporal lobe 497, 498 temporal lobe 494–5, 496, 497 fronto-temporal degeneration 583–4 glutamate 9, 13 glutamine 9, 13 HIV dementia 461, 462 human brain 168 hypoxic brain injury 693, 694–5, 696–9, 699–700 hypoxic–ischemic brain injury 649, 694–5, 696–9, 699–700, 702 hypoxic–ischemic encephalopathy 692 inborn errors of metabolism 650 intracranial infections 296 Krabbe’s disease 758 lactate 9, 12 signal 294 lipid signal 294 megalencephalic leukoencephalopathy with subcortical cysts 771 metabolic ratios 282–3 metachromatic leukodystrophy 756 mitochondrial encephalopathies 763, 764–5 multi-slice 761 multi-system atrophy 597 myo-inositol 9, 12–13 neonatal brain injury 682 neoplastic/non-neoplastic differentiation 295–6 neurodegenerative disease 595 Parkinson’s disease 596–7 pediatric imaging 648, 676 pediatric patients 675 post-contrast spectroscopy 49 prematurity 682 prescription 38–9 primary neoplastic lesion differentiation 296–8 progressive supranuclear palsy 598, 600 ribose-5-phosphate isomerase deficiency 773 single voxel 582 spectra information content of brain 8, 9, 10–13 speed of acquisition enhancement 669 stroke 173, 174–5 traumatic brain injury 615–22, 620–1 tumefactive demyelinating lesions 296 vascular dementia 583

823

824

Index

proton MRSI 595 proton nuclear MR (1H NMR) 595 pro-urokinase 163 pseudo-diffusion 101 pseudopermeability effect 338, 339 pseudostroke 163 psychiatric disease 523–8 affective disorders 535, 536–7, 538, 539–41, 542 age matching 530 anxiety disorders 542, 543–5, 546 diagnosis 535 purity 530 diagnostic validation 524 drug abuse 531 medications 531 MRS 529–52 future studies 551–2 medication effects 531 pulse sequence 534 spectra quantification 534–5 studies 535–51 study design 531, 532–3, 534 subject selection 530–1 tissue volume correction 534 patient factors 524 physiological MRI 527 sample size 530 schizophrenia 548, 551 sex matching 530 subject selection 530–1 substance-related disorders 546, 548, 549–50 technical considerations 524 voxel placement 531, 532–3, 534 pulse sequences 14 pulsed gradient spin-echo (PGSE) experiment 57 pulvinar sign 526 pyruvate 382, 385 q-space imaging 80 three-dimensional 101 quality adjusted life years 526 quantitative imaging of perfusion using a single subtraction (QUIPSS) 125, 151, 152 radiation adverse effects of exposure 300 necrosis 300–1, 311 diffusion-weighted imaging 413, 726, 728 pediatric 649, 653, 658, 726, 727, 728, 732 recurrence distinction 350 treatment of brain tumors 320–1, 322 pediatric 726, 727 radio frequency off resonance 128 receiver chain 31 radio frequency coil matching 31 single for continuous arterial spin labeling 123, 124 tuning 31 two-coil 124, 125

radio frequency pulses 7 180° 57, 106 design 40–1 spectral–spatial 41 radio tracers, metabolic 2 radiotherapy, external beam 300 Rantion Scale scores 212 Rasmussens’s encephalitis 491, 492, 507 red nucleus, functional MRI-BOLD 270 region of interest (ROI), reference 94 relaxation time 31 blood spin–lattice 156 rate change 142 relaxivity 49 repetition time coherence control 42 optimal 74 resuscitation in head injury 609 Rett syndrome 658 reversible posterior leukoencephalopathy syndrome (RPLS) 743–4 Reye’s syndrome 705 rhabdoid teratoid tumor 730 ribitol 773, 794 ribose-5-phosphate isomerase (RPI) deficiency 773 riluzole 595 Rolando L 1 Rosenthal fibers 767 Salla disease 439, 789 sarcoidosis 359, 361 scaling factors 31–2 schizophrenia 548, 551 diffusion tensor imaging 567–8 fractional anisotropy 567–8 frontal lobe phospholipid membrane metabolism abnormality 548 medications 548 MRS 567 synaptic pruning 548 schwannoma extravascular extracellular space 344 pediatric 723 relative cerebral blood volume 342–3 sedation neonates 675, 707 pediatric patients 737 seizures/seizure disorders 481–6 absence 482 brain metabolism 488–9 brain tumors 492, 494 cerebral blood volume 510 complex partial 481–2 diffusion-weighted imaging 512–16 experimentally-induced 513–14 epileptic 481 epileptogenic 489 generalized onset 482 generalized tonic-clonic 482, 497 generation investigation 502 glucose levels 516 hyperperfusion 510 hypoperfusion 510–11 hypoxic–ischemic brain injury in neonates 708 ictal period 510

interictal discharges 502 interictal period 510–11 MRS 491–502, 507–8 neuronal cell death 516 oxygen metabolism 516 partial-onset 481, 482 pathophysiology 482 perfusion MRI 509–12 phenomenology 481–2 sickle cell disease 745 simple partial 481 see also epilepsy selective serotonin reuptake inhibitors (SSRIs) 531 selective vulnerability 780–1 senile plaques 584 SENSitivity Encoding (SENSE) 51, 76, 78, 668, 669, 670 sensorimotor activation detection by arterial spin labeling 209 sequence time 31 shaken baby syndrome 637 shimming 3D MRSI in Alzheimer’s disease 588 high-order 16 localized 17–18 pitfalls 39 slice-by-slice 16 Shinnar–LeRoux (SLR) linear phase pulse 40 short saturation (SAT) pulses 47, 48, 49 sialic acid 789 sickle cell disease bolus delay/dispersion 144, 145 moyamoya syndrome 746 pediatric patients 655–6, 657, 658 stroke 736, 742 perfusion-weighted imaging 685, 744, 745, 746–7 seizures 745 siderosis, superficial 636 signaling pathways 280 SiMultaneous Acquisition of Spatial Harmonics (SMASH) 668, 669 single photon emission computed tomography (SPECT) 489 brain development 738 cerebral blood flow 239 epilepsy 509 head injury 611 HIV dementia 464 HMPAO 483 ictal 484, 511, 512 lymphoma 366, 469 neurodegenerative disease 527 pediatric patients 654 stroke ischemic penumbra 164 single-gene disorders 779–80 single-voxel techniques 13–15, 648 echo time 16, 17, 18 epilepsy 497 hypoxic encephalopathy 694, 695, 698 SI technique comparison 16–18, 19 singular value decomposition (SVD) 114 sinusitis, subdural empyema 357, 358, 416 Sjogren–Larsson syndrome 438, 683 slice order 42–5 slice selection techniques 120

Index

sodium ion/myoinositol cotransporter (SMIT) 494 spastic diplegia 690 spastic quadriparesis 685 spectral analysis 27–31 time domain pre-processing 28, 29 spectroscopy phantom as quality control 51 post-contrast 49, 50 spin echo head injury 630 neoplasia imaging 331–2 spinal cord lesions 454 spindle cell sarcoma 724 spinocerebellopathies 781 splenium aging 561, 562–3 Alzheimer’s disease 565–6 statistical parametric mapping 560 status epilepticus 482, 484, 485 apparent diffusion coefficient maps 515 diffusion-weighted imaging 514 MRS 491 Steele–Richardson–Olszewski syndrome see progressive supranuclear palsy (PSP) Stejskal–Tanner diffusion weighting 106, 659 Stejskal–Tanner equation 73 Stejskal–Tanner sequence 58 stenting, intravascular cerebrovascular reserve 254–5 filter system 256, 257 ischemia 255, 257 stimulated echo acquisition mode (STEAM) 14–15, 18, 38 coherence control 41–2 coupled spin systems 44 hypoxic encephalopathy 693, 695, 697–9 T2 relaxation minimizing 74 tuberculoma 386 Streptococcus pneumoniae 380 striatum abnormalities 781 stroke 163–5 N-acetyl aspartate 173, 174–5, 181 MRS measurement 168–9, 170, 171 arterial spin labeling 207–18 MRI assessment 132–3 bolus delay/dispersion 143 cardioembolic 186, 188 cerebral blood flow 192, 194–5 reduced 208 cerebral ischemia metabolic changes 172–3 choline 173, 174–5 computed tomography 184, 198 creatine MRS measurement 169 diffusion tensor MRI 223–31 diffusion-weighted imaging 182–3 apparent diffusion coefficient 193, 197 CT comparison 198 increased signal duration 190, 191, 192 lesion development 190, 192–4, 195, 196 lesion measurement 196–7, 198 lesion reversal 205 lesion volume measurement 197 outcome 197 pediatric patients 740, 741, 742–4 severity 197, 198

dynamic susceptibility contrast imaging (DSCI) 207 echo planar imaging 78 embolic 256, 257 feasibility of MRI 183–5 fiber orientation mapping 228–30 fractional anisotropy 224–5, 226, 227–8 hypoperfusion 208 ischemic penumbra 164–5, 192 ischemic/ischemic injury 163, 252–3 fractional anisotropy changes 225, 226, 227 late secondary 188 patterns 253 lactate 173, 174–5, 181 MRS measurement 169, 171 lacunar 186 apparent diffusion coefficient 189 late-presenting 185–6 lesion growth 190, 192–4, 195, 196 lipids 171 macromolecules 171 magnetic resonance spectroscopy 168–77 management 208 MCA territory apparent diffusion coefficient 189 mean diffusivity 223–4 mild 185–6, 187, 197 apparent diffusion coefficient 189 chronic cerebrovascular blood supply deficit 246 mimics 229–30 mitochondrial encephalopathy with lactic acid 749–50 MRS outcome measures 176 patient selection for thrombolysis 175–6 prognostic indicator 176 thrombolysis 163 multi-slice MRSI 173, 174–5, 181 patient movement 176 pediatric patients 216–18, 736–50, 753–4 arterial ischemic 748–9 diffusion-weighted imaging 740, 741, 742–4 metabolic 749–50 moyamoya syndrome 747–8 sickle cell disease 744, 745, 746–7 perfusion deficits 164 perfusion-weighted imaging 182, 183 lesion development 190, 192–4, 195, 196 lesion reversal 205 pediatric patients 655–6, 657 recoverable vs. permanent damage 193–4 perinatal 706 diffusion tensor imaging 710, 716 diffusion-weighted imaging 708, 709, 710 PET cerebral blood flow correlation with tissue damage 192–3 proton MRS 173, 174–5 PWI/DWI mismatch region 194, 195, 196 recurrent infarct 186, 187 regional blood flow 133 screening method limitations 163–4 staging of progression 175–6

thrombolysis 163 patient selection 175–6 TIA differentiation 186–7 tissue viability prediction 228 Wallerian degeneration 228–30, 231 Sturge–Weber syndrome 754 subacute sclerosing panencephalitis 392 subarachnoid hemorrhage diffusion-weighted imaging 206 middle cerebral artery 259 perfusion-weighted imaging 206 subdural effusions, post-meningitic 356 subdural empyema 356, 357, 358, 415–16, 417–18 subdural hematoma apparent diffusion coefficient 630 shaken baby syndrome 637 substance-related disorders 546, 548, 549–50 phosphorus-31 MRS 622 Wernicke’s encephalopathy 570 substantia nigra, functional MRI-BOLD 270 succinate hydatid cysts 398, 401 neurocysticercosis 398, 399 succinate dehydrogenase deficiency 761, 763, 764–5, 793 sugar signals, Aspergillus abscess 394 surrogate endpoints 281 surrogate markers 319, 523 synaptic pruning in schizophrenia 548 T1 enhancement 115 T1-based acquisition 332 T1-shine through 335 elimination 334 T1-weighted images Aspergillus abscess 394, 395 boxing-related injury 638 brain abscess 354, 355, 382, 383, 409, 410 brain maturation 678 cysticercosis 365 demyelinating diseases 444 encephalitis 367 epilepsy 489 herpes simplex encephalitis 391 Japanese encephalitis 392–3 meningitis 359, 360, 361, 362, 386, 414, 415, 416 multiple sclerosis 446 neurocysticercosis 397–8 parenchymal disease 361, 363 pediatric tumors 723, 724, 725 sinusitis 357, 358 toxoplasmosis 365, 366 tuberculoma 386, 389, 471 T2 enhancement 115 T2 measurement 34 T2 relaxation 74 rate 110 T2-based acquisition 332 T2-relaxometry 490, 491 T2-shine through 59, 61, 99 Epstein–Barr virus encephalitis 418, 419 multiple sclerosis 376 pediatric stroke 740 pediatric tumors 726

825

826

Index

T2-weighted images Aspergillus abscess 394, 395 axonal injury 634 brain abscess 354, 355, 382, 383, 409, 410 brain maturation 678 Canavan’s disease 787, 788 Creutzfeldt–Jakob disease 372 cysticercosis 365 demyelinating diseases 444 encephalitis 367 Epstein–Barr virus encephalitis 418, 419 gradient echo in axonal injury 634 head injury 630 non-accidental in child 637 herpes simplex encephalitis 391 3-hydroxy-3-methylglutaryl-coenzyme A lyase deficiency 799, 800 Japanese encephalitis 392–3 meningitis 359, 362, 386 multiple sclerosis 446 multi-system atrophy 597 neurocysticercosis 397–8, 399, 400 para-infectious encephalopathy 393, 394 parenchymal disease 361, 363 pediatric patients 654 stroke 740, 741, 742, 743, 745 pediatric tumors 723 Pelizaeus–Merzbacher disease 798, 799 phenylketonuria 791 sinusitis 357, 358 stroke in pediatric patients 740, 741, 742, 743, 745 toxoplasmosis 365, 366 tuberculoma 386, 387, 389, 471 Taenia solium 396–7 Takayasu arteritis 258 temporal bone, fungal disease 359 temporal lobe, phosphomonoesterase 538 temporal lobectomy 490 tensor calculation 87, 88 process 89 thalamocortical axons 681–2 thalamus hypoxic–ischemic brain injury 649 infarction 712 pediatric 3D MRSI 676, 677 thiamine deficiency 570 thrombolysis in stroke 163 patient selection 175–6 pediatric patients 742 thrombospondins 1 and 2 280 time to peak (TTP) 147, 148, 149, 251 perfusion-weighted imaging 183 tissue liquefaction 354 brain abscess 382 tissue necrosis 354 brain abscess 382 hypoxic brain injury 690 toxoplasmosis 365–6, 421 AIDS patients 420, 466–8 diffusion tensor imaging 571 ring-enhancing lesions 466 treatment 467, 468 tracking–editing technique 93–4 tractography see fiber tracking; magnetic resonance tractography

transcranial magnetic stimulation (TMS), occipital cortex 264 transient ischaemic attacks (TIAs) carotid bifurcation 250 chronic cerebrovascular blood supply deficit 246 diffusion imaging 165–6 magnetic resonance spectroscopy 165–6 moyamoya syndrome 747 stroke differentiation 186–7 transit times 150–2 arterial spin labeling 127 measurement 151–2 transmitter gain setting 39 transportation of neonates/pediatric patients 675 trauma/traumatic brain injury see head injury tricarboxylic acid (TCA) cycle 385, 793 inborn errors of metabolism 651 trigeminal activation 269 trigeminal nucleus caudalis 269 triglycerides 700 tuberculoma 385–6, 387, 388, 389 HIV infection 471–2 tuberculosis 359 brain abscess 388, 390 CNS 385–6, 387, 388 HIV infection 471–2 meningitis 360, 385, 386 miliary 471 tuberous sclerosis 723 tumefactive demyelinating lesions 296, 373, 375 multiple sclerosis 308 relative cerebral blood volume 343 turbo-stimulated echo acquisition mode (STEAM) sequences 249 Turcot syndrome 723 ultra small particulates of iron oxides (USPIO) 282 un-inverted flow-sensitive alternating inversion recovery (UNFAIR) 125 valproate 485–6 vanishing white matter disease 768, 770, 771, 797 varicella zoster pediatric stroke 736 stenotic arterial disease 737 vascular endothelial growth factors (VEGF) 279, 329, 338 glioma 342 pediatric tumors 728 vascular permeability 4 venous infarction, neonates 661, 663 venous magnetization 122 venous thrombosis 229, 695 ventricular enlargement 636 ventriculitis 354 meningitis 415, 416 very high-field MRI, psychiatric disease 551, 552 very selective saturation (VSS) pulses 47, 48, 49 lipid suppression 49

vessel misregistration 147, 148 video-electroencephalogram telemetry 489 vigabatrin 502 Virchow-Robin spaces 471 visual cortex hyper-excitability in migraine 264 lactate levels 171 regional blood flow 128, 131 vitamin E 272 volume of interest (VOI) 31 shift 147 water as reference signal 33–4 volume ratio 90 von Hippel–Lindau syndrome 723 voxel see multiple-voxel (SI) techniques; single-voxel techniques; volume of interest (VOI) Wallerian degeneration 228–30, 231 Alzheimer’s disease 565 axons in multiple sclerosis 448 perinatal stroke 710, 716 water apparent diffusion coefficient 127 diffusion anisotropic in brain 223 anisotropy in normal tissues 224 surrogate marker 319 white matter 318 diffusion-weighted imaging signal source 314–15 diffusivity 59 imaging 247–8 multiple sclerosis 447 head injury 627, 628 labeled 119, 127 motion translation 408 reduction 40, 44–5 reference signal 33–4, 35 unsuppressed 46 sidebands 46–7, 51 T2 measurement 34 see also edema, brain water suppression 15, 18, 20, 44–5 enhanced through T1 effects (WET) 45 residual water 40 water-to-metabolite magnetization transfer (MT) 45 Wernicke’s encephalopathy 570 West Nile encephalitis 367, 370, 427 white matter adrenoleukodystrophy 758, 760–1, 762 aging 560, 561–4 changes 579, 580 alcoholism 568, 569–70 Alexander disease 766–8, 769 Alzheimer’s disease 581 anisotropy 99–101, 223 brain maturation 678, 679, 680, 681 prematurity 684 border-zone lesions 242 brain maturation 678, 679 brainstem maps 665–6, 667 Canavan’s disease 763–5 cerebral palsy 664–5, 666 Cho signals 20, 21, 22

Index

connectivity between brain regions 680 diffusion 65–6 anisotropy 61 stroke 224 diffusion tensor eigenvalues 680 measurement 559 diffusion tensor imaging 561–4 electron density 2 fiber tracts displacement by tumor 318–19 location in brain tumor area 317–18, 323 spatial orientation 560 water mobility 315 water molecule path 559 fractional anisotropy 90, 559, 561–3, 564 stroke 227 frontal and pediatric 3D MRSI 676, 677 head injury boxing 638 diffusion tensor imaging 631 hereditary leukoencephalopathies 797–8 HIV infection 571 3-hydroxy-3-methylglutaryl-coenzyme A lyase deficiency 799 hyperintensities 186 hypoxic–ischemic encephalopathy 692 infarction 208 neonates 710, 712, 717, 718 injury 684 Krabbe disease 667 loss in Alzheimer’s disease 564–5 macroscopic structure characterization 95

maple syrup urine disease 790 maturation 781 megalencephalic leukoencephalopathy with subcortical cysts 771, 772, 773 metabolic diseases 781–2 metabolite concentration 36, 534 MRS data analysis 499 metachromatic leukodystrophy 756 microstructure integrity 558 mitochondrial encephalomyopathies 761, 763, 764–5 myelination 738 myelinopathia centralis diffusa/vanishing white matter disease 768, 770, 771 NAA signals 20, 21, 22 neonatal diffusion-weighted imaging 707–8 normal appearing 431, 432, 433–4, 444 acute disseminated encephalomyelitis 450 diffusion-weighted imaging 447–50 multiple sclerosis 431, 432, 433–4, 447–50, 451 parietal pericallosal 561, 562–3, 564 Parkinson’s disease 596 partial volume effect 100 pediatric patients 648, 649, 660 changes with age 738–9 Pelizaeus–Merzbacher disease 765–6, 767 phenylketonuria 791 polyhydric alcohols metabolism defect 794, 795 posterior callosal 564

ribose-5-phosphate isomerase deficiency 773 schizophrenia 567 segmentation 90 somatosensory cortex projections 685 stroke diffusion 224 fractional anisotropy 227 structure 558–9 succinate dehydrogenase deficiency 793 toxoplasmosis lesions 466 tracts branching 93 editing using multiple-region of interest 93–4 reconstruction 91 traumatic brain injury 613, 616–17 water content 33 relaxation time 4 water diffusion 318 characteristics 448 white matter tractography 453, 664, 666 neonates 715 pediatric tumors 728 Willis, Thomas 1 xenon inhaled 119 see also computed tomography (CT), xenon X-ray computerized tomography 119 Zellweger’s syndrome 713 zero-quantum interference 15

827