Osteoarthritic Joint Pain (Novartis Foundation Symposia)

Osteoarthritic Joint Pain: Novartis Foundation Symposium 260. Volume 260 Edited by Derek J. Chadwick and Jamie Goode Copyright ¶ Novartis Foundation 2004. ISBN: 0-470-86763-9

OSTEOARTHRITIC JOINT PAIN

The Novartis Foundation is an international scienti¢c and educational charity (UK Registered Charity No. 313574). Known until September 1997 as the Ciba Foundation, it was established in 1947 by the CIBA company of Basle, which merged with Sandoz in 1996, to form Novartis. The Foundation operates independently in London under English trust law. It was formally opened on 22 June 1949. The Foundation promotes the study and general knowledge of science and in particular encourages international co-operation in scienti¢c research. To this end, it organizes internationally acclaimed meetings (typically eight symposia and allied open meetings and 15^20 discussion meetings each year) and publishes eight books per year featuring the presented papers and discussions from the symposia. Although primarily an operational rather than a grant-making foundation, it awards bursaries to young scientists to attend the symposia and afterwards work with one of the other participants. The Foundation’s headquarters at 41 Portland Place, London W1B 1BN, provide library facilities, open to graduates in science and allied disciplines. Media relations are fostered by regular press conferences and by articles prepared by the Foundation’s Science Writer in Residence. The Foundation o¡ers accommodation and meeting facilities to visiting scientists and their societies.

Information on all Foundation activities can be found at http://www.novartisfound.org.uk

Novartis Foundation Symposium 260

OSTEOARTHRITIC JOINT PAIN

2004

Copyright & Novartis Foundation 2004 Published in 2004 by John Wiley & Sons Ltd, The Atrium, Southern Gate, Chichester PO19 8SQ, UK National 01243 779777 International (+44) 1243 779777 e-mail (for orders and customer service enquiries): [email protected] Visit our Home Page on http://www.wileyeurope.com or http://www.wiley.com All Rights Reserved. No part of this book may be reproduced, stored in a retrieval system or transmitted in any form or by any means, electronic, mechanical, photocopying, recording, scanning or otherwise, except under the terms of the Copyright, Designs and Patents Act 1988 or under the terms of a licence issued by the Copyright Licensing Agency Ltd, 90 Tottenham Court Road, London W1T 4LP, UK, without the permission in writing of the Publisher. Requests to the Publisher should be addressed to the Permissions Department, John Wiley & Sons Ltd,The Atrium, Southern Gate, Chichester,West Sussex PO19 8SQ, England, or emailed to [email protected], or faxed to (+44) 1243 770620. This publication is designed to provide accurate and authoritative information in regard to the subject matter covered. It is sold on the understanding that the Publisher is not engaged in rendering professional services. If professional advice or other expert assistance is required, the services of a competent professional should be sought. Other Wiley Editorial O⁄ces John Wiley & Sons Inc., 111 River Street, Hoboken, NJ 07030, USA Jossey-Bass, 989 Market Street, San Francisco, CA 94103-1741, USA Wiley-VCH Verlag GmbH, Boschstr. 12, D-69469 Weinheim, Germany John Wiley & Sons Australia Ltd, 33 Park Road, Milton, Queensland 4064, Australia John Wiley & Sons (Asia) Pte Ltd, 2 Clementi Loop #02-01, Jin Xing Distripark, Singapore 129809 John Wiley & Sons Canada Ltd, 22 Worcester Road, Etobicoke, Ontario, Canada M9W 1L1 Wiley also publishes its books in a variety of electronic formats. Some content that appears in print may not be available in electronic books. Novartis Foundation Symposium 260 ix+292 pages, 40 ¢gures, 13 tables

British Library Cataloguing in Publication Data A catalogue record for this book is available from the British Library ISBN 0 470 86761 2 Typeset in 101
2 on 121
2 pt Garamond by DobbieTypesetting Limited, Tavistock, Devon. Printed and bound in Great Britain by T. J. International Ltd, Padstow, Cornwall.

This book is printed on acid-free paper responsibly manufactured from sustainable forestry, in which at least two trees are planted for each one used for paper production.

Contents Symposium on Osteoarthritic joint pain, held atthe Novartis Foundation, London,1^3 July 2003 Editors: Derek J. Chadwick (Organizer) and Jamie Goode This symposium is based on a proposal by Stuart Bevan and John Rediske David Felson

Chair’s introduction 1

Hans-Georg Schaible Discussion 22

Spinal mechanisms contributing to joint pain

4

Blair D. Grubb Activation of sensory neurons in the arthritic joint 28 Discussion 36 Neuromuscular aspects of osteoarthritis: a perspective 49

Kenneth D. Brandt Discussion 58

Paul Creamer Current perspectives on the clinical presentation of joint pain in human OA 64 Discussion 74 Walter Herzog, Andrea Clark and David Longino Joint mechanics in osteoarthritis 79 Discussion 95 General discussion I Gunnar Ordeberg Discussion 115

Developing animal models of RA 100

Characterization of joint pain in human OA 105

Bruce L. Kidd, Andrew Photiou and Julia J. Inglis The role of in£ammatory mediators on nociception and pain in arthritis 122 Discussion 133 James L. Henry Molecular events of chronic pain: from neuron to whole animal in an animal model of osteoarthritis 139 Discussion 145 v

vi

CONTENTS

C. S. McCabe, R. C. Haigh, N. G. Shenker, J. Lewis and D. R. Blake Phantoms in rheumatology 154 Discussion 174 Peter A. Simkin Bone pain and pressure in osteoarthritic joints 179 Discussion 186 Philip G. Conaghan and David T. Felson Structural associations of osteoarthritis pain: lessons from magnetic resonance imaging 191 Discussion 201 Martin Koltzenburg The role of TRP channels in sensory neurons 206 Discussion 213 Patrick W. Mantyh and Stephen P. Hunt bone cancer pain 221 Discussion 238

Mechanisms that generate and maintain

N. G. Shenker, D. R. Blake, C. S. McCabe, R. Haigh and P. I. Mapp T cells and neurogenic arthritis 241 Discussion 252

Symmetry,

Laurence A. Bradley, Brian C. Kersh, Jennifer J. DeBerry, Georg Deutsch, Graciela A. Alarco¤n and David A. McLain Lessons from ¢bromyalgia: abnormal pain sensitivity in knee osteoarthritis 258 Discussion 270 David Felson

Chair’s summing up

Index of contributors Subject index

282

280

277

Participants David R. Blake The Royal National Hospital for Rheumatic Diseases, Upper BoroughWalls, in conjunction withThe Department of Medical Sciences and The Department of Pharmacy and Pharmacology, University of Bath, Bath BA1 1RL, UK Laurence A. Bradley Division of Clinical Immunology and Rheumatology, University of Alabama at Birmingham, 805 Faculty O⁄ceTower, 510 20th street south, Birmingham, AL 35294, USA Kenneth D. Brandt Indiana University School of Medicine, Indiana University Multipurpose Arthritis and Musculoskeletal Diseases Center, 1110 West Michigan Street, Room 545, Indianapolis, IN 46202-5100, USA Philip G. Conaghan Academic Unit of Musculoskeletal Disease, University of Leeds & Department of Rheumatology, Leeds General In¢rmary, Great George Street, Leeds LS1 3EX, UK Paul Creamer

Southmead Hospital,Westbury onTrym, Bristol BS10 5NB, UK

Paul Dieppe Department of Social Medicine, University of Bristol, Canynge Hall,Whiteladies Road, Bristol BS8 2PR, UK Christopher H. Evans Center for Molecular Orthopaedics, Harvard Medical School, 221 Longwood Avenue, BL-152, Boston, MA 02115, USA David T. Felson (Chair) Boston University School of Medicine, 715 Albany Street, A203, Boston, MA 02118, USA Janet K. Fernihough Novartis Institute for Medical Sciences, 5 Gower Place, London WC1E 6BN, UK Alyson Fox Novartis Institute for Medical Sciences, 5 Gower Place, London WC1E 6BN, UK vii

viii

PARTICIPANTS

Blair D. Grubb Department of Cell Physiology and Pharmacology, University of Leicester, PO Box 138, Leicester LE1 9HN, UK James L. Henry Department of Physiology and Pharmacology, University of Western Ontario, Medical Sciences Building, M221, London, Ontario, N6A 5C1, Canada Walter Herzog Human Performance Laboratory, Faculty of Kinesiology, Department of Mechanical Engineering, Faculty of Engineering, The University of Calgary, 2500 University Drive NW B205, Calgary, AB, CanadaT2N 1N4 Stephen Hunt Department of Anatomy and Developmental Biology, Anatomy Building, Gower Street, London WC1E 6BT, UK David J. Hunter Clinical Epidemiology Research and Training Unit, Boston Medical Center, 715 Albany Street, Room A-203, Boston, MA 02118-2526, USA Bruce L. Kidd Bone andJoint Unit, Bart’s andThe London School of Medicine, Charterhouse Square, London EC1M 6BQ, UK Martin Koltzenburg Neural Plasticity Unit, Neural Plasticity, Institute of Child Health, 30 Guilford Street, London WC1N 1EH, UK Klaus E. Kuettner Department of Biochemistry, Rush Medical College, RushPresbyterian-St Luke’s Medical Center, 1653 W. Congress Parkway, Chicago, IL 60612, USA Stefan Lohmander Department of Orthopaedics, University Hospital Lund, 22185 Lund, Sweden Andrew Mackenzie Switzerland

Novartis Pharma AG,WSJ-386.10.35, CH-4002, Basel,

Marzia Malcangio Chronic Pain Programme, Novartis Institute for Medical Science, University College London, Gower Street, London WC1E 6BT, UK Anthony M. Manning 94304-1397, USA

Roche Bioscience, 3401HillviewAvenue, Palo Alto, CA

PARTICIPANTS

ix

Gunnar Ordeberg Division of Orthopaedic Surgery, Karolinska Institutet, Danderyd Hospital, Stockholm, SE-18288, Sweden David S. Pisetsky Duke University Medical School,151G DurhamVA Hospital, 509 Fulton Street, Durham, North Carolina, 27705, USA John Rediske Research, Arthritis and Bone MetabolismTherapeutic Area, Novartis Pharmaceuticals Corporation, 556 Morris Avenue, Summit, NJ 07901, USA Hans-Georg Schaible Institut fˇr Physiologie 1/Neurophysiologie, Teichgraben 8, 07740 Jena, Germany Hua Shen (Novartis Foundation Bursar) Center of Experimental Rheumatology, and WHO Collaborating Center for Molecular Biology, and Novel Therapeutic Strategies for Rheumatic Diseases, University Hospital Zu«rich, Gloriastrasse 23, CH-8091 Zu«rich, Switzerland Peter A. Simkin Department of Medicine, University of Washington, Box 356428, Seattle,WA 98195, USA Wim van den Berg Center of Rheumatology Research and Advanced Therapeutics, Nijmegen Center of Molecular Life Sciences, Geert Grooteplein 26-28, 6500 HB Nijmegen,The Netherlands Thasia G.Woodworth Switzerland

Arthritis/Bone, Novartis Pharma AG, CH-4002, Basel,

KiranYashpal Department of Physiology and Pharmacology, University of Western Ontario, Medical Sciences Building, M221, London, Ontario, N6A 5C1, Canada

Osteoarthritic Joint Pain: Novartis Foundation Symposium 260. Volume 260 Edited by Derek J. Chadwick and Jamie Goode Copyright ¶ Novartis Foundation 2004. ISBN: 0-470-86763-9

Chair’s introduction David Felson Clinical Epidemiology Research and Training Unit, Boston University School of Medicine, 715 Albany Street, A 203, Boston, MA 02118-2526, USA

To begin with, I’d like to mention some results from Lois Verbrugge, who gave a group of people in Detroit a diary to take home to record their daily health symptoms (Verbrugge 1979). In people aged 65 and over she tallied the most common daily health symptom. In women over 65, it was knee trouble; in men it was backache followed by knee trouble. The most common category of symptoms in older people was musculoskeletal problems. Is this just older people? No. In women aged 45^64 the second most common daily symptom was knee trouble. Thus, pain in knees, often due to osteoarthritis (OA), is a remarkably prevalent problem. Osteoarthritis is the most common form of arthritis. In elderly subjects the prevalence of each of hand and hip OA is around 2^3% of the population aged 65 and over. Prevalence in this case is de¢ned as symptoms on most days and evidence of structural OA as a cause of these symptoms. Knee OA (see below) is even more common. In the elderly, OA is a more common cause of lower extremity disability than any other disease. It causes 1 in 8 days of restricted activity in US elders and is the reason why most people have knee and hip replacements. The burden of this disease on society is truly remarkable. I’d also like to mention data on knee OA from a variety of recent studies in the UK (Peat et al 2001). Of 10 000 people aged 55 and over, about 25% have radiographic OA, but not necessarily symptoms. Irrespective of radiographs, about 25% of people in this over-55 age group have had an episode of knee pain lasting at least 4 weeks. About half of these have radiographic OA. This 12.5% of the population would be the group that we would categorize as having symptomatic OA. The other thing to note is that arthritis is not going to decrease in prevalence in western societies. Because of the demographics, it is only going to increase in prevalence. We are getting older, but we are also getting more overweight, which is a major risk factor for OA in weight-bearing joints. Add to this the increasing number of sports injuries. The estimate is that the current level of 20 million a¡ected in the USA will rise to 40 million by 2020. In the past the focus of OA research was on hyaline articular cartilage loss. I am not going to suggest that cartilage loss is not pivotal; rather that the whole joint is 1

2

FELSON

a¡ected in OA, with bony changes (sclerosis and osteophytes), cartilage loss, joint capsule stretching and thickening, modest synovial in£ammation all often present and weak periarticular muscles. If we were to discover this disorder today, we probably would not call it OA, but in medical textbooks next to chronic hepatic failure and chronic renal failure, there would be a chapter on ‘chronic joint failure’. What are the symptoms and signs of OA? In OA, pain is worse on use of the joint, mild in morning, severe after immobility. There is loss of movement, pain on movement and restricted range of motion. The pain symptoms of OA will be a focus of this symposium. One of the central rationales for this symposium
and one of the reasons I am excited about it
is that pain and disease structure do not necessarily go hand in hand. We have one group that has symptomatic knee pain, and only about half of them have radiographic OA. Then there is the population with radiographic OA, and a large percentage of them have no symptoms at all. Often people with severe structural disease have no symptoms. What accounts for this discordance between symptoms and structure? There is another problem: that of drug development in this challenging situation. There is an interleukin (IL)1 inhibitor called diacerein that, in theory, would prevent cartilage loss. In a randomized trial of hip OA (Dougados et al 2001), compared with placebo, the diacerein group experienced less joint space loss which signi¢es less cartilage loss. This suggests that diacerein did indeed prevent cartilage loss. But when the readout is pain improvement, the diacerein group and placebo group had exactly the same scores. That is, IL1 inhibition seemed to prevent cartilage loss but had no measureable e¡ect on pain. We have a situation where not only is there an observed discordance between pain and structural change, but the results of this study suggest that drug development may produce therapies which prevent cartilage loss but have no e¡ects on the symptoms we are most concerned about. In OA treatments are limited and those that are widely used are dangerous, expensive and suboptimal. Most treatments being developed focus on preventing cartilage loss, and this may or may not have e¡ects on pain and disability. What do we want to focus on at this meeting? Some of the larger perspective questions are as follows: . . . .

What is the pathophysiology of OA joint pain? Why do some but not all people with OA get joint pain? Why do some people get severe pain and others mild, intermittent pain? Are there treatment opportunities that go along with a better understanding of OA joint pain?

These are ambitious questions. Hopefully, we will get some handle on where we might go to answer those questions as we proceed.

CHAIR’S INTRODUCTION

3

References Dougados M, Nguyen M, Berdah L et al 2001 Evaluation of the structure-modifying e¡ects of diacerein in hip osteoarthritis: ECHODIAH, a three-year, placebo-controlled trial. Evaluation of the Chondromodulating E¡ect of Diacerein in OA of the Hip. Arthritis Rheum 44:2539^2547 Peat G, McCarney R, Croft P 2001 Knee pain and osteoarthritis in older adults: a review of community burden and current use of primary health care. Ann Rheum Dis 60:91^97 Verbrugge LM 1979 Female illness rates and illness behavior: testing hypotheses about sex di¡erences in health. Women Health 4:61^79

Osteoarthritic Joint Pain: Novartis Foundation Symposium 260. Volume 260 Edited by Derek J. Chadwick and Jamie Goode Copyright ¶ Novartis Foundation 2004. ISBN: 0-470-86763-9

Spinal mechanisms contributing to joint pain Hans-Georg Schaible Department of Physiology, Friedrich-Schiller-University of Jena, Teichgraben 8, D-07740 Jena, Germany

Abstract. Nociceptive input from the joint is processed in di¡erent types of spinal cord neurons. A proportion of these neurons are only activated by mechanical stimulation of the joint and other deep tissue, e.g. adjacent muscles. Other neurons are activated by mechanical stimulation of joint, muscles and skin. The majority of the neurons are wide dynamic-range neurons (small responses to innocuous pressure to deep tissue and stronger and graded responses to noxious mechanical stimulation). Importantly, neurons with joint input show pronounced hyperexcitability during development of joint in£ammation (enhanced responses to mechanical stimulation of the in£amed joint as well as to healthy adjacent deep structures, reduction of mechanical threshold in high threshold neurons and expansion of the receptive ¢eld). Thus in£ammation induces neuroplastic changes in the spinal cord which alter nociceptive processing. This state of hyperexcitability is maintained during persistent in£ammation. The neurons are under strong control of descending inhibition which increases at least during the acute phase of in£ammation. Several transmitters and mediators contribute to the generation and maintenance of in£ammation-induced spinal hyperexcitability including glutamate, substance P, neurokinin A, CGRP, prostaglandins and probably others. The latter compounds show enhanced release and an altered release pattern during in£ammation in the joint. 2004 Osteoarthritic joint pain. Wiley, Chichester (Novartis Foundation Symposium 260) p 4^27

Pain sensation in the joint The major sensation in deep tissue such as joint and muscle is pain. Sensory information from muscle and joint in£uences the motoric system and is involved in the sense of movement and position but usually this does not reach consciousness. In humans pain in the normal joint can be elicited particularly when noxious mechanical, thermal and chemical stimuli are applied to the ¢brous structures such as ligaments and ¢brous cartilage (Lewis 1942, Kellgren & Samuel 1950). In a normal joint pain is most commonly elicited by twisting or hitting the 4

SPINAL MECHANISMS OF JOINT PAIN

5

joint. Stimulation of ¢brous structures with innocuous mechanical stimulation can evoke pressure sensations. No pain is elicited by stimulation of cartilage, and stimulation of normal synovial tissue rarely evokes pain (Kellgren & Samuel 1950). Joint in£ammation is characterized by hyperalgesia and persistent pain at rest which is usually dull and badly localized (Lewis 1938, 1942, Kellgren 1939, Kellgren & Samuel 1950). The application of noxious stimuli causes stronger pain than normal. Pain is even evoked by mechanical stimuli whose intensity is normally not su⁄cient to elicit pain, i.e. movements in the working range and gentle pressure, e.g. during palpation. This heightened pain sensitivity results from peripheral sensitization (increase of sensitivity of nociceptive primary a¡erent neurons) and central sensitization (hyperexcitability of nociceptive neurons in the central nervous system). Pain during degenerative osteoarthritis shows similarities and di¡erences to arthritic pain. Similarly, as with arthritis, pain may increase when the joint is being loaded. However, pain may also be reduced during walking, and pain may be particularly severe during rest at night when the joint is immobile. It is likely that mechanisms of pain during arthritis and osteoarthritis are di¡erent in some aspects. Spinal cord neurons that respond to mechanical stimulation of the joint The articular nerves supplying the knee or elbow joint of rat, cat and monkey enter the spinal cord via several dorsal roots thus projecting to several spinal segments. Due to the widely distributed projection area joint a¡erents in£uence sensory neurons and re£ex pathways in several spinal segments. Within the grey matter knee joint a¡erents project to the super¢cial lamina I and to the deep laminae V^VII (c.f. Schaible & Grubb 1993). Figure 1A shows the spinal termination ¢elds of horseradish peroxidase-labelled knee joint a¡erents in the segment L7 in the cat spinal cord. Correspondingly, spinal cord neurons that are synaptically activated by joint a¡erents can be identi¢ed in the super¢cial and deep dorsal horn and also in the ventral horn (Schaible et al 1986). Receptive ¢elds and activation thresholds of neurons with joint input In both cat and rat, mechanonociceptive inputs from the joint are processed in dorsal horn neurons that respond solely to mechanical stimulation of deep tissue, or in neurons that respond to mechanical stimulation of both deep tissue and the skin. Receptive ¢elds of single sensory neurons (regions from which neurons can be activated) are usually not restricted to the joint but more extended. Figure 1C shows the receptive ¢eld of a spinal cord neuron with convergent inputs from skin,

FIG. 1. Spinal projection of primary a¡erent ¢bres of the knee joint and response properties of spinal cord neurons with input from the knee joint. (A) Spinal termination ¢eld of horseradish peroxidase-labelled primary a¡erent ¢bres of the posterior articular nerve of the knee joint in the grey matter of the segment L7 in the cat. (B) Responses of a wide dynamic range (WDR) neuron with input from the knee joint to innocuous and noxious movements of the knee joint. The histograms show the number of action potentials/s that were elicited by the movements (bin width 1 s). Flex, £exion of the knee joint; Ext, extension of the knee joint; f. Ext, forced extension of the knee joint; OR, outward rotation of the knee joint (supination); n. OR, noxious outward rotation of the knee joint; IR, inward rotation of the knee joint (pronation); n. IR, noxious inward rotation of the knee joint. (C) Receptive ¢eld of spinal cord neuron with input from the knee joint. This neuron was excited by pressure applied to the skin of the paw, the deep tissue of thigh (quadriceps muscle) and lower leg (gastrocnemius-soleus muscle) and the structures of the knee joint. (D) Receptive ¢eld of spinal cord neuron that was only excited by pressure applied to deep tissue (muscles) and the knee joint. (A) reprinted with permission from Craig et al 1988; (B^D) from Schaible et al 1987a.

6 SCHAIBLE

SPINAL MECHANISMS OF JOINT PAIN

7

deep tissue and the knee joint. The neuron was activated by pressure applied to the knee joint (capsule, ligaments) and also by compression of the quadriceps muscle in the thigh and the gastrocnemius-soleus muscle in the lower leg, and in addition it had a cutaneous receptive ¢eld at the paw. However, many neurons have receptive ¢elds that are restricted to the deep tissue. Figure 1D shows the receptive ¢eld of a spinal cord neuron with a receptive ¢eld in the deep tissue of the leg and in the knee joint. Some neurons have bilateral receptive ¢elds (c.f. Schaible & Grubb 1993). Concerning mechanical thresholds, neurons are either nociceptive-speci¢c (NS) or wide-dynamic-range (WDR) neurons. NS neurons respond only to intense pressure and/or to painful movements such as forceful supination and pronation. These stimuli elicit pain. WDR neurons respond to both innocuous pressure and noxious pressure, encoding stimulus intensity by the frequency of action potentials. They may also be weakly activated by movements in the working range, and they show much stronger responses to painful movements. Figure 1B displays the response pattern of a wide dynamic range neuron with joint input. The neuron exhibited small responses to £exion, extension, and outward rotation (OR) of the knee in its physiological range, but pronounced responses were elicited by forced extension (f. ext) and by noxious outward rotation (n. OR) exceeding the working range of the joint. By and large, NS neurons have smaller receptive ¢elds restricted to deep tissue in joint and muscle, and they do not have a receptive ¢eld in the skin (c.f. Schaible & Grubb 1993). Projections of spinal neurons with joint input Neurons with joint input project to di¡erent supraspinal sites (cerebellum, spinocervical nucleus, thalamus, reticular formation) or to intraspinal (segmental) interneurons and motoneurons (c.f. Schaible & Grubb 1993). Ascending projections to the thalamus (in the spinothalamic tract) are important to activate the thalamocortical systems that generate the conscious pain sensation. Segmental projections are important for the generation of motor and sympathetic re£exes. Furthermore, in the cat neurons have been identi¢ed that have cell bodies in the ventral horn, belong to the spinoreticular tract and are predominantly or exclusively excited by noxious stimulation of deep tissue (Fields et al 1977, Meyers & Snow 1982). Inhibition by heterotopic and descending inhibitory systems Neurons with joint input are inhibited by heterotopic stimuli, in line with the concept of di¡use noxious inhibitory controls (DNIC). The latter means that painful stimulation at one site of the body may reduce the pain at another site of the body (LeBars & Villanueva 1988). In addition, most spinal cord neurons with

8

SCHAIBLE

joint input are tonically inhibited by descending inhibitory systems that keep the spinal cord under continuous control (Cervero et al 1991, Schaible et al 1991). The interruption of descending inhibition can lower the excitation threshold of spinal cord neurons for mechanical input from the knee, substantially increase the receptive ¢elds of neurons and cause (increased) ongoing discharges. Thus the response properties of neurons with joint input are controlled by the primary a¡erent input, by intrinsic properties of the spinal cord neurons, by local circuits and by descending pathways. Hyperexcitability in spinal cord neurons during in£ammation in the joint As described in the introduction, pain and hyperalgesia are usually elicited during in£ammation of the joint. Hence experimental models have been used to study neuronal mechanisms underlying these pain symptoms. In£ammation in the joint can be induced by the intra-articular injections of crystals such as urate and kaolin, or by carrageenan. The injection of kaolin and carrageenan (K/C) into the joint produces an oedema and granulocytic in¢ltration within 1^3 hours with a plateau after 4^6 hours. Awake animals show limping of the injected leg and enhanced sensitivity to pressure onto the joint. By contrast, the injection of Freund’s complete adjuvant (FCA) into one joint produces a monoarthritis that is present for 2 to 4 weeks. Usually the lesion is restricted to the injected joint, although bilateral e¡ects are observed sometimes. Hyperalgesia (limping or guarding of the leg, enhanced sensitivity to pressure onto the joint) develops within a day, reaches a peak within 3 days and is maintained to some degree up to several weeks. When FCA is injected at a high dose into the tail base or lymph node, a polyarthritis develops (Schaible & Grubb 1993). Generation of hyperexcitability During the development of a K/C-induced in£ammation in the joint, both NS and WDR neurons with joint input show enhanced responses within 1^3 hours to noxious stimuli applied to the in£amed joint (central sensitization). NS neurons exhibit a reduction in their mechanical threshold such that the application of innocuous stimuli to the in£amed joint is su⁄cient to excite the neurons. Figure 2A shows the generation of hyperexcitability in a spinal cord neuron with joint input. Initially, while the joint was normal, the neuron responded only to noxious pressure applied to the knee (and adjacent muscles in thigh and lower leg, Fig. 2B, left side). No responses were elicited by pressure onto the ankle and the paw. After injection of kaolin and carrageenan into the knee joint (K/C) the responses to noxious compression of the knee

SPINAL MECHANISMS OF JOINT PAIN

9

increased markedly, and at a latency of about half an hour the neuron started also to respond to pressure applied to the ankle and the paw. Thus the receptive ¢eld expanded from the knee towards the paw (Fig. 2B, right side), and the previously high threshold neuron was then even activated by gentle innocuous pressure. The increased responses to stimuli applied to the in£amed joint result most likely from the enhanced synaptic input from a¡erent units which are sensitized during stimulation. However, the appearance of responses to stimulation of ankle and paw must result from a mechanism in the spinal cord because these regions were not in£amed. Thus nociceptive spinal cord neurons obviously develop a state of hyperexcitability in which the responsiveness to both inputs from in£amed and non-in£amed areas is increased (Dougherty et al 1992, Neugebauer & Schaible 1990, Neugebauer et al 1993, Schaible et al 1987b). The increased responses to stimulation of the in£amed area are thought to be the neuronal mechanism of primary hyperalgesia (hyperalgesia at the site of in£ammation) whereas the increased responses to stimuli applied to healthy tissue are thought to be the neuronal mechanism underlying secondary hyperalgesia (hyperalgesia in healthy tissue adjacent to and remote from in£amed tissue). Figure 2C and 2D show the working hypothesis how these changes are produced. When the tissue is normal the neuron is only excited by stimuli applied to the restricted receptive ¢eld (circle in Fig. 2C) but not by stimuli applied to adjacent areas. When an in£ammation develops in the receptive ¢eld (shaded area, Fig. 2D), primary a¡erents in this region are sensitized and they induce a process of spinal sensitization. When the spinal neuron is hyperexcitable it shows stronger responses to stimuli applied to the original receptive ¢eld (stimulation sites 2 and 3), and in addition the neuron responds to inputs that are normally too weak to excite the neuron above threshold (stimulation sites 1 and 4). Hence the receptive ¢eld expands (Fig. 2D). This central sensitization can persist during chronic in£ammation. In rats with unilateral arthritis (Grubb et al 1993) as well as in rats su¡ering from chronic polyarthritis (Menetrey & Besson 1982) spinal cord neurons appear on average more sensitive and have expanded receptive ¢elds. Interestingly, the stimulation of primary a¡erents from deep tissue (muscle and joint) evokes more prolonged facilitation of a nociceptive £exor re£ex than stimulation of cutaneous a¡erents (Woolf & Wall 1986), and capsaicin injection into deep tissue elicits more prolonged hyperalgesia than injection of capsaicin into the skin (Sluka 2002) suggesting that deep input is particularly able to induce long term changes in the nociceptive system. However, spinal sensitization is counteracted to some extent by inhibitory in£uences. Descending inhibition (Schaible et al 1991) as well as heterotopic inhibitory in£uences (see above) are increased during in£ammation (Calvino et al 1987).

10

SCHAIBLE

SPINAL MECHANISMS OF JOINT PAIN

11

FIG. 2. Development of in£ammation-evoked hyperexcitability in a spinal cord neuron with input from the knee joint. (A) Histogram showing the responses (action potentials/response) of the neuron to noxious pressure applied to the knee joint, the ankle and the paw before and after injection of kaolin and carrageenan (K/C) into the ipsilateral knee joint. (B) Receptive ¢eld (shaded area) of the neuron before (control) and during knee joint in£ammation (3 h post K/C). C and D. Model showing the responses and the receptive ¢eld of a spinal cord neuron before in£ammation and (C) and after development of hyperexcitability (D). Before in£ammation the neuron was only excited by pressure to the initial receptive ¢eld (stimulation sites 2 and 3). After in£ammation the neuron was activated from a larger area (stimulation sites 1^4). A,B reprinted with permission from Neugebauer et al (1993).

12

SCHAIBLE

Transmitters, mediators and receptors involved in synaptic activation of spinal cord neurons with joint input The generation and maintenance of central sensitization is produced by the action of transmitter/receptor systems in the spinal cord. After sensitization primary a¡erent neurons release more transmitter from their spinal terminations upon peripheral stimulation (presynaptic component). Furthermore, spinal cord neurons are rendered more excitable by changes in receptor sensitivity (postsynaptic component).

Excitatory amino acids Glutamate is the major transmitter in the synaptic activation of spinal cord neurons with joint input. On the postsynaptic site, glutamate activates N-methyl-Daspartate (NMDA) receptors and non-NMDA receptors. The activation of nonNMDA receptors leads to basic excitation of neurons. By contrast, the activation of NMDA receptors leads to a calcium in£ux into neurons and causes processes of neuronal plasticity in many neuronal circuits, such as long-term changes of responses. In our hands the ionophoretic application of antagonists at AMPA/ kainate (non-NMDA) receptors close to neurons with joint input reduced the responses to innocuous and noxious pressure whereas the application of NMDA receptor antagonists reduced only the responses to noxious mechanical stimulation. Thus, in our hands, NMDA receptors are only activated by noxious stimulation (Neugebauer et al 1993). The ionophoretic application of NMDA antagonists at AMPA/kainate and NMDA receptors to spinal cord neurons as well as systemic application of NMDA antagonists prevents the development of in£ammation-evoked spinal hyperexcitability (Neugebauer et al 1993). Figure 3 shows the e¡ect of ketamine, an antagonist at NMDA receptors. In six control neurons without ketamine the induction of in£ammation in the knee joint by injection of K/C caused increases of the responses to noxious pressure applied to the injected knee and the non-injected ankle. When ketamine was administered before and during induction of in£ammation, the induction of in£ammation in the knee joint did not change responses in the four neurons tested as long as the antagonist was applied (Fig. 3C,D). Importantly, antagonists at both receptor types can reduce responses of the neurons to mechanical stimulation of the joint also after in£ammation is established, and this is even seen in a chronic model of in£ammation (Neugebauer et al 1993, 1994a). Thus glutamate receptors play a key role in the generation and maintenance of in£ammation-evoked spinal hyperexcitability even in the longterm range.

SPINAL MECHANISMS OF JOINT PAIN

13

Neuropeptides Numerous joint a¡erents contain the neuropeptides substance P, neurokinin A and CGRP that are coexpressed with glutamate. Noxious compression, but not innocuous compression of the normal joint enhances the intraspinal release of these peptides above baseline. This pattern of release changes when the joint is in£amed. During acute in£ammation release occurs when the joint is stimulated at innocuous intensity. Thus, under in£ammatory conditions a ‘cocktail’ of transmitters and/or modulators is released in the spinal cord that changes synaptic processing under in£ammatory conditions (Hope et al 1990, Schaible et al 1990, 1994). Excitatory neuropeptides facilitate the responses of spinal cord neurons. The e¡ect of substance P is shown in Figs 4A,B. The WDR neuron in Fig. 4A showed graded responses to innocuous and noxious pressure applied to the knee joint. A short ionophoretic application of substance P to the spinal cord neuron caused reversible increases of ongoing discharges and responses to mechanical stimulation. In the NS neuron in Fig. 4B substance P caused an increase of responses to noxious pressure and a small response to innocuous pressure and to ankle stimulation. Ionophoretic application of antagonists at neurokinin 1, neurokinin 2 and CGRP receptors attenuate the development of in£ammation-evoked hyperexcitability. Figure 4C shows the e¡ect of CP96,345, an antagonist at the neurokinin 1 receptor, on the development of in£ammation-evoked hyperexcitability. Compared to control neurons (top graph, induction of in£ammation in the absence of the antagonist) the neurons treated with spinal administration of the neurokinin 1 receptor antagonist showed a smaller increase in their responses after induction of in£ammation (middle graph). The inactive enantiomer, CP96,344, did not attenuate the magnitude of in£ammation-evoked hyperexcitability (bottom graph). The antagonists also reduce hyperexcitability when it is established (Neugebauer et al 1995, 1996a,b). Probably, the activation of these peptide receptors enhances the sensitivity of glutamatergic synaptic transmission (Ebersberger et al 2000). However, it is important to point out that the antagonists at neuropeptide receptors are less antinociceptive than antagonists at glutamate receptors.

Prostaglandins Spinal prostaglandins (PGs) are synthesized in DRG neurons and in the spinal cord by both cyclooxygenases 1 and 2. PG receptors are located on primary a¡erent neurons and on spinal cord neurons indicating that PGs can act presynaptically

14

SCHAIBLE

SPINAL MECHANISMS OF JOINT PAIN

15

FIG. 3. Blockade of the development of hyperexcitability in spinal cord neurons by i.v. administration of the NMDA receptor antagonist ketamine. (A, B) Changes of the responses of spinal cord neurons to noxious pressure applied to the knee joint and the ankle during development of in£ammation in the knee joint after the injection of K/C into the ipsilateral knee joint. (C, D) Same experimental approach as in A and B, but in these experiments ketamine was given i.v. during induction and in the initial period of in£ammation in the knee joint. Reprinted with permission from Neugebauer et al (1993).

16

SCHAIBLE

SPINAL MECHANISMS OF JOINT PAIN

17

FIG. 4. E¡ect of substance P on the responses of spinal cord neurons to mechanical stimulation of the knee joint and e¡ect of a neurokinin 1 receptor antagonist on the development of in£ammation-evoked hyperexcitability of spinal cord neurons. (A, B) The histograms (action potentials/s) shows that ionophoretic application of substance P at 70 nA or 100 nA enhances responses to mechanical stimulation. In the neuron in A substance P also caused enhanced ongoing discharges. (C) Development of in£ammation-evoked hyperexcitability in spinal cord neurons (responses to innocuous pressure) in the absence of the antagonist (top), in the presence of the neurokinin 1 receptor antagonist CP96,345 at the spinal cord neurons (middle) and in the presence of the inactive enantiomer CP96,344 of the neurokinin 1 receptor antagonist (bottom). (A, B) reprinted with permission from Neugebauer et al 1994b; (C) from Neugebauer et al 1995.

18

SCHAIBLE

FIG. 5. Up-regulation of cyclooxygenase 2 (COX-2) in the spinal cord during in£ammation in the knee joint and the e¡ect of spinal administration of indomethacin on the development of in£ammation-evoked hyperexcitability of spinal cord neurons. (A, B) During in£ammation in the joint mainly spinal COX-2 shows an increase. (C) The spinal application of indomethacin, a blocker of COX-1 and COX-2 attenuates spinal hyperexcitability. Open squares show the in£ammation-evoked changes of responses after kaolin/carrageenan injection in control neurons, ¢lled squares show the changes of the responses during development of in£ammation after topical administration of indomethacin to the spinal cord. Top graphs show responses to noxious pressure, bottom graphs responses to innocuous pressure. (A, B) reprinted with permission from Ebersberger et al (1999); (C) from Vasquez et al (2001).

SPINAL MECHANISMS OF JOINT PAIN 19

20

SCHAIBLE

(in£uencing the release of synaptic mediators) and postsynaptically (in£uencing excitability) (Vanegas & Schaible 2001). During in£ammation in the joint, there is a tonic release of PGE2 within the dorsal and ventral horn (Ebersberger et al 1999). This is likely to result from an upregulation of spinal COX-2 that is already increased at 3 hours after induction of knee joint in£ammation (Figs 5A,B). The application of PGE2 to the spinal cord surface facilitates the responses of spinal cord neurons to mechanical stimulation of the joint, and the pattern of e¡ects is similar to that observed during peripheral in£ammation. When the cyclooxygenase inhibitor indomethacin is applied to the spinal cord before in£ammation, the development of hyperexcitability is signi¢cantly attenuated compared to control rats in which only vehicle is applied to the spinal cord (Fig. 5C). Thus spinal PGs are involved in the generation of in£ammationevoked spinal hyperexcitability (Vasquez et al 2001). The e¡ect of spinal PGE2 is mimicked by the application of agonists at EP1, EP2 and EP4 receptors to the spinal cord.

Concluding remarks This review has its focus on the synaptic activation and development of hyperexcitability of spinal cord neurons in the course of joint in£ammation. The data clearly show that the spinal cord undergoes neuroplastic changes during in£ammation, and this will in£uence the expression of pain in patients. The transmission of spinal information to the thalamocortical cortex will generate the subjective experience of pain with its di¡erent dimensions. However, hyperexcitability might also in£uence the in£ammatory process in the joint. The group of Willis has recently reported that the spinal cord may in£uence the peripheral in£ammatory process through dorsal root re£exes (Willis 1999). This will act in concert with other central nervous mechanisms that interact with the in£ammatory process (Straub & Cutolo 2001). Thus the nervous system does not only act as a sensor for damaging stimuli, it also has an active part in the expression of the in£ammatory lesion. It remains to be clari¢ed how important the mechanisms described are in the generation and maintenance of pain during degenerative osteoarthritis.

References Calvino B, Villanueva L, LeBars D 1987 Dorsal horn (convergent) neurons in the intact anaesthetized arthritic rat. II. Heterotopic inhibitory in£uences. Pain 31:359^379 Cervero F, Schaible H-G, Schmidt RF 1991 Tonic descending inhibition of spinal cord neurons driven by joint a¡erents in normal cats and in cats with an in£amed knee joint. Exp Brain Res 83:675^678

SPINAL MECHANISMS OF JOINT PAIN

21

Craig AD, Heppelmann B, Schaible H-G 1988 The projection of the medial and posterior articular nerves of the cat’s knee to the spinal cord. J Comp Neurol 276:279^288 Dougherty PM, Sluka KA, Sorkin LS, Westlund KN, Willis WD 1992 Neural changes in acute arthritis in monkeys. I. Parallel enhancement of responses of spinothalamic tract neurons to mechanical stimulation and excitatory amino acids. Brain Res Brain Res Rev 17:1^13 Ebersberger A, Grubb BD, Willingale HL, Gardiner NJ, Nebe J, Schaible H-G 1999 The intraspinal release of prostaglandin E2 in a model of acute arthritis is accompanied by an upregulation of cyclooxygenase-2 in the rat spinal cord. Neuroscience 93:775^781 Ebersberger A, Charbel Issa P, Vanegas H, Schaible H-G 2000 Di¡erential e¡ects of calcitonin gene-related peptide and calcitonin gene-related peptide 8-37 upon responses to N-methyl-Daspartate or (R, S)-alpha-amino-3-hydroxy-5-methylisoxazole-4-propionate in spinal nociceptive neurons with knee input in the rat. Neuroscience 99:171^178 Fields HL, Clanton CH, Anderson SD 1977 Somatosensory properties of spinoreticular neurons in the cat. Brain Res 120:49^66 Grubb BD, Stiller RU, Schaible H-G 1993 Dynamic changes in the receptive ¢eld properties of spinal cord neurons with ankle input in rats with unilateral adjuvant-induced in£ammation in the ankle region. Exp Brain Res 92:441^452 Hope PJ, Jarrott B, Schaible H-G, Clarke RW, Duggan AW 1990 Release and spread of immunoreactive neurokinin A in the cat spinal cord in a model of acute arthritis. Brain Res 533:292^299 Kellgren JH 1939 Some painful joint conditions and their relation to osteoarthritis. Clin Sci 4:193^205 Kellgren JH, Samuel EP 1950 The sensitivity and innervation of the articular capsule. J Bone Joint Surg Am 32-B:84^91 LeBars D, Villanueva L 1988 Electrophysiological evidence for the activation of descending inhibitory controls by nociceptive pathways. In: Fields HL, Besson J-M (eds) Prog Brain Res, Elsevier, Amsterdam, vol 77:275^299 Lewis T 1938 Suggestions relating to the study of somatic pain. Br Med J 1:321^325 Lewis T 1942 Pain. Macmillan, London Menetrey D, Besson J-M 1982 Electrophysiological characteristics of dorsal horn cells in rats with cutaneous in£ammation resulting from chronic arthritis. Pain 13:343^364 Meyers DER, Snow PJ 1982 The responses to somatic stimuli of deep spinothalamic tract cells in the lumbar spinal cord of the cat. J Physiol 329:355^371 Neugebauer V, Schaible H-G 1990 Evidence for a central component in the sensitization of spinal neurons with joint input during development of acute arthritis in cat’s knee. J Neurophysiol 64:299^311 Neugebauer V, Lˇcke T, Schaible H-G 1993 N-methyl-D-aspartate (NMDA) and non-NMDA receptor antagonists block the hyperexcitability of dorsal horn neurons during development of acute arthritis in rat’s knee joint. J Neurophysiol 70:1365^1377 Neugebauer V, Lˇcke T, Grubb BD, Schaible H-G 1994a The involvement of N-methyl-Daspartate (NMDA) and non-NMDA receptors in the responsiveness of rat spinal neurons with input from the chronically in£amed ankle. Neurosci Lett 170:237^240 Neugebauer V, Schaible H-G, Weiretter F, Freudenberger U 1994b The involvement of substance P and neurokinin-1 receptors in the responses of rat dorsal horn neurons to noxious but not to innocuous mechanical stimuli applied to the knee joint. Brain Res 666:207^215 Neugebauer V, Weiretter F, Schaible H-G 1995 The involvement of substance P and neurokinin-1 receptors in the hyperexcitability of dorsal horn neurons during development of acute arthritis in rat’s knee joint. J Neurophysiol 73:1574^1583 Neugebauer V, Rˇmenapp P, Schaible H-G 1996a The role of spinal neurokinin-2 receptors in the processing of nociceptive information from the joint and in the generation and

22

DISCUSSION

maintenance of in£ammation-evoked hyperexcitability of dorsal horn neurons in the rat. Eur J Neurosci 8:249^260 Neugebauer V, Rˇmenapp P, Schaible H-G 1996b Calcitonin gene-related peptide is involved in spinal processing of mechanosensory input from the rat’s knee joint and in the generation and maintenance of hyperexcitability of dorsal horn-neurons during development of acute in£ammation. Neuroscience 71:1095^1109 Schaible H-G, Grubb BD 1993 A¡erent and spinal mechanisms of joint pain. Pain 55:5^54 Schaible H-G, Schmidt RF, Willis WD 1986 Responses of spinal cord neurons to stimulation of articular a¡erent ¢bres in the cat. J Physiol 372:575^593 Schaible H-G, Schmidt RF, Willis WD 1987a Convergent inputs from articular, cutaneous and muscle receptors onto ascending tract cells in the cat spinal cord. Exp Brain Res 66:479^488 Schaible H-G, Schmidt RF, Willis WD 1987b Enhancement of the responses of ascending tract cells in the cat spinal cord by acute in£ammation of the knee joint. Exp Brain Res 66:489^499 Schaible H-G, Jarrott B, Hope PJ, Duggan AW 1990 Release of immunoreactive substance P in the spinal cord during development of acute arthritis in the knee joint of the cat: a study with antibody microprobes. Brain Res 529:214^223 Schaible H-G, Neugebauer V, Cervero F, Schmidt RF 1991 Changes in tonic descending inhibition of spinal neurons with articular input during the development of acute arthritis in the cat. J Neurophysiol 66:1021^1032 Schaible H-G, Freudenberger U, Neugebauer V, Stiller RU 1994 Intraspinal release of immunoreactive calcitonin gene-related peptide during development of in£ammation in the joint in vivo
a study with antibody microprobes in cat and rat. Neuroscience 62:1293^1305 Sluka KA 2002 Stimulation of deep somatic tissue with capsaicin produces long-lasting mechanical allodynia and heat hypoalgesia that depends on early activation of the cAMP pathway. J Neurosci 22:5687^5693 Straub RH, Cutolo M 2001 Involvement of the hypothalamicpituitary adrenal/gonadal axis and the peripheral nervous system in rheumatoid arthritis: viewpoint based on a systemic pathogenetic role. Arthritis Rheum 44:493^507 Vanegas H, Schaible H-G 2001 Prostaglandins and cyclooxygenases in the spinal cord. Prog Neurobiol 64:327-363 [Erratum in Prog Neurobiol 65:609] Vasquez E, Br K-J, Ebersberger A, Klein B, Vanegas H, Schaible H-G 2001 Spinal prostaglandins are involved in the development but not the maintenance of in£ammationinduced spinal hyperexcitability. J Neurosci 21:9001^9008 Willis WD Jr 1999 Dorsal root potentials and dorsal root re£exes: a double-edged sword. Exp Brain Res 124:395^421 Woolf CJ, Wall PD 1986 Relative e¡ectiveness of C primary a¡erent ¢bers of di¡erent origins in evoking a prolonged facilitation of the £exor re£ex in the rat. J Neurosci 6:1433^1442

DISCUSSION Hunter: As someone who thinks a lot about the pathophysiology of OA, I am interested in your thoughts as to what e¡ect the sensitization may have, particularly in the role of potential impairments to proprioception and trophic changes to the muscle. You mentioned that the sensitization might have a role in changes in motor re£exes. Could you expand on that? Schaible: There aren’t many studies that have addressed this. Usually, a noxious stimulus would result in a nociceptive re£ex: you withdraw your limb from the noxious stimulus. Some years ago we conducted a study in which we examined

SPINAL MECHANISMS OF JOINT PAIN

23

these re£exes (He et al 1988). We induced joint in£ammation to see whether the re£exes were altered. We found that the pattern is changing so that not all noxious stimuli evoke the nociceptive re£ex. That is, there is some inhibition of the re£exes. This is important, because usually if you have an in£amed joint you don’t keep it in an extreme position, because this will activate the joint a¡erents the most. When this re£ex is increasingly inhibited, you may be able to keep the joint in a position where the joint a¡erents aren’t activated very much. This is an extremely important ¢eld which has been neglected in recent years by neurophysiologists: we need to know how this pathology in£uences the re£ex pattern. Hunter: Do the spinocortical tracts interact at all with the tracts which control proprioception and any of the trophic response that it may have on the muscle? That is, does the sensitization of the upward ascension of pain ¢bres have any interaction with proprioceptive and trophic spinal ¢bres? Schaible: Intuitively, I would say yes. There are few data addressing this, though. In some re£ex studies this has been seen, but it hasn’t been studied under these conditions. Grubb: One of the things we use as a measure of central sensitization is the expansion of the receptive ¢elds to nociceptive stimuli, and also to nonnociceptive stimuli, following stimulation of skin. This is most marked in the rat, but is also seen in larger animals such as the cat. When you examine patients with chronic OA, do you see changes in the responsivness well beyond the con¢nes of the joint itself, into the lower thigh or calf, for example, where we see these expanded receptive ¢elds? How well can the features we see be mapped into human? Creamer: We don’t know. I don’t think the equivalent studies have been done. We know that pain often radiates from a¡ected joints, so that pain from the hip is often felt in the thigh or pain in the knee is often felt lower down. But this is usually thought to be because of radiation rather than increased sensitivity. We know that around the knee people seem to be tender at points which don’t tend to correspond to particular pathological structures. There may be some sort of increased general sensitivity to pain round the knee. Pisetsky: The data are clear-cut that, after a single stimulus, there are prolonged, rapid changes in receptive ¢elds of neurons from in£ammation in the joint. How much in£ammation in the joint do you need to get this kind of change? This model involves very intense, acute in£ammation where there are potential systemic e¡ects (e.g. cytokine or glucocorticoid mediated). Would you see the same thing in a chronic OA model, if you could do it in dogs? Or do you need a lot of in£ammation over a short time scale to get remodelling? Schaible: It is di⁄cult to specify how much in£ammation is needed. What we see is a robust phenomenon. The in£ammation does not need to be very thick to produce these changes, and we see some variation. We also have the test

24

DISCUSSION

stimulation (repeated mechanical stimulation of the joint), which could help to promote the sensitization processes. From what I know from all my experience, I would assume that the process of central sensitization is in principle the same when there is a mild form of in£ammation. However, if we just use mechanical stimulation and stimulate the knee in a noxious manner every ¢ve minutes for three hours, we don’t get this central sensitization. The additional stimulus of some kind of sensitization by in£ammation is needed. The models of in£ammatory pain that have been used in pain research are not very mild. We use in£ammatory processes which can be measured. But I would assume that even a mild in£ammation would do the same job, with a slower time course and the phenomenon might be as large. Pisetsky: Could you block this with anti-tumour necrosis factor (TNF) or an interleukin (IL)1 inhibitor? How much peripheral component is there to the central sensitization, which could be cytokine mediated? Schaible: We have some work on this, but the picture still isn’t clear. We can partly reduce the sensitization by using COX inhibitors. We haven’t tried using anti-TNF, but that is something we intend to do. We have tried to anaesthetize the joint after in£ammation to see whether the central sensitization is stable. In pain research we have two ways of thinking. One group says that there is a pain memory: you have one stimulus and the brain never forgets. The other group thinks that there is sensitization as long as there is a process going on in the periphery. From all of our recordings, I assume that the second is the case: if you reduce the peripheral input, you can get rid of the central sensitization. When we anaesthetized the joint we were able to get rid of some of the activity. The problem is that the local anaesthetics don’t work very well in in£amed tissue, so we don’t know the extent of the pain block. Dieppe: I want to come back to Blair Grubb’s point. One of the striking clinical features of knee OA is the symmetry of the pain pattern. So to what extent do you see bilateral spread across the spinal cord which might help explain this, as well as ipsilateral spreading? Schaible: We have done spinal cord recordings. There are some spinal cord neurons which show an altered response form. If they have a bilateral receptive ¢eld, the responses of the neurons will change whether the ipsilateral or contralateral side is stimulated. But there is also a bilateral e¡ect at the level of the primary a¡erents. We have investigated receptor expression in dorsal root ganglion (DRG) primary a¡erents. If there is in£ammation on one side there is an upregulation of receptors on both sides, but there is only pain on one side (Segond von Banchet et al 2000). The reason why you don’t have pain on the other side might be because the ligands are not there if there is no in£ammatory process. Grubb: In reply to the earlier question about what type of stimuli the spinal cord neurons are experiencing, we are possibly back to front in the order of papers in this

SPINAL MECHANISMS OF JOINT PAIN

25

meeting. The primary a¡erents not only become more sensitive to noxious stimuli, but also many of them become spontaneously active. During this acute in£ammatory model which lasts hours, the spinal cord neurons experience a constant a¡erent barrage around 1 Hz from C ¢bres and the A-d pain a¡erents. The combination of this enhanced barrage from the increased mechanical sensitivity and the increased spontaneous activity contributes to the central sensitization that Professor Schaible describes. Henry: You described the distribution of projections in the spinal cord. It would be interesting to know what happens through the development of your RA model. We are seeing some contralateral e¡ects, which are a little surprising to us. Can anyone comment on the central distribution of primary a¡erent ¢bres in an RA model? Grubb: If you are talking about contralateral e¡ects, there was a panel at the bottom of one of Professor Schaible’s slides showing that if you record the ipsilateral side, in animals in which the in£ammation has been for 20 days, about 20% have developed receptive ¢elds from the contralateral side. So it is a signi¢cant proportion of the WDR neurons in the deep dorsal horn that develop these contralateral receptive ¢elds in the rat. It may vary for other species and in di¡erent models. Kidd: I am very interested in the mirroring of in£ammation within the CNS. We know of in£ammatory cells interacting with peripheral neurons. Might the same processes be occurring at a central level which might then produce these contralateral changes? There is quite good evidence that the glial cells interact with the neuron^neuron communication at a spinal level. What is the importance of these glial^neuron communications at a spinal level? Schaible: I can’t comment directly on this because we haven’t done any work on it. I know, however, that much of the COX is probably not located in the neurons but in the glial cells. Intuitively, I think that the glial cells are quite important for the process of central sensitization. Simkin: I gather that all of these stimuli are starting in the synovium. Has anyone done comparable work with stimuli in subchondral bone? Grubb: If you look at the peptidergic a¡erents that come from joint tissues, they have been identi¢ed in almost every single joint tissue capsule, subchondral bone, cartilage, tendons and ligaments. Pisetsky: We are taught that cartilage has no nociceptive function. Grubb: There is one paper suggesting innervation on the edge of the cartilage. It is a low density of innervation, but it has been reported (Schwab & Funk 1998). Pisetsky: Is any of this joint speci¢c? Or can you get this kind of sensitization with in£ammation in some other structure? Schaible: It is possible. It has been demonstrated also for skin in£ammation. However, the joint sensitization is very sensitive to mechanical stimulation. The

26

DISCUSSION

skin in£ammation has mainly been studied with temperature stimulation. It has always been di⁄cult to do this with mechanical stimulation in skin. Lohmander: You pointed out that most of the relevant COX enzymes might be located in the glial cells with receptors in the spinal cells. Would a systemically distributed COX inhibitor penetrate and be able to a¡ect the COX-related mechanisms in the spinal cord? Schaible: Yes, as long as the inhibitors can penetrate. This is the main problem: most of the compounds we have are not useful for this. Lohmander: The point I am trying to make is that the medications currently used would not be very likely to a¡ect the central spinal mechanisms we are discussing. Schaible: In order to be sure we need to test each one. You cannot guarantee that this is the case. Manning: There is certainly a good distribution of the COX inhibitors that are on the market into the CNS. The challenge is to show that they are centrally acting. Rediske: In addition to playing a role in the activation of sensory neurons, a number of in£ammatory mediators also have neurotrophic activities. Could some of the changes in the receptive ¢eld sensitivity be due to changes in innervation patterns and increases in C ¢bre density? Schaible: There are changes like this in some models, although I must add that in chronic in£ammation the number of C ¢bres may be reduced. There are con£icting data in the periphery. In the spinal cord we don’t know whether there is any sprouting during in£ammation. This has not been tested to my knowledge. This whole ¢eld of neurotrophic factors hasn’t been addressed that much. A peripheral application of nerve growth factor (NGF) will cause changes in primary a¡erent neurons. I am not sure how relevant this is for the central sensitization. Henry: I’d like to return to the mirror phenomenon. This bears on the question of whether OA is an in£ammatory response or not. If it is, then mirror doesn’t necessarily mean neural sprouting, but if it isn’t then it probably would be a neural-mediated e¡ect. We are seeing some evidence of bone sclerosis on the contralateral side in our animal model of OA, which would suggest a systemic in£ammatory response. Schaible: How is that induced? Henry: Cutting the anterior cruciate ligament (ACL) and medial meniscus. Felson: There is evidence in human studies that limping and pain in one leg changes the gait pattern so as to overload the contralateral side. The e¡ects you are seeing might be related to biomechanical changes in gait in the animal that has been injured. Manning: I’d like to address some of your prostaglandin pharmacology data. Cyclooxygenases produce a range of prostaglandins, not just PGE2. The use of neutralizing antibodies to PGE2 has certainly substantiated that it is a key pain modulator, but PGI2 is also produced and that interacts with discrete inositol

SPINAL MECHANISMS OF JOINT PAIN

27

phosphate (IP) receptors. A number of companies have developed IP receptor antagonists that look very good in pain models. Have you looked at PGI2 or IP receptors in the spinal cord? Schaible: Not so far. We have measured prostaglandins in the spinal cord, and in terms of quantity the most important are PGE2 and PGD2. I2 is not that high, but the receptors are there. This needs to be measured. Grubb: In the periphery there is no doubt that both are extremely important for the sensitization of the a¡erents in terms of primary hyperalgesia and allodynia. Mackenzie: You said that COX-2 inhibitors were more e¡ective than nonselective COX inhibitors. Does this imply that at the spinal level COX-2 is constitutively expressed? Schaible: This is de¢nitely known. It might not be true that indomethacin is nonselective in the CNS. I recently read a report that considered indomethacin as more speci¢c for COX-1 in the hippocampus. It really depends on the tissue as to how e¡ective these inhibitors are. Pisetsky: Does paracetamol work in this? Schaible: We haven’t studied this. Grubb: We did one set of experiments which we published in 1990 in which we compared aspirin and paracetamol, and we got a modest reduction in primary a¡erent ¢ring, even though its IC50 for COX inhibition is very poor. There was also a story about COX-3, which may not exist in humans. There is a single base missing in the DNA sequence and it appears that the whole sequence is not in frame (Dinchuk et al 2003). Quite why paracetamol works in humans I can’t tell you since the mechanism of action is unknown. Mackenzie: Experiments at our laboratories in Novartis have con¢rmed that there is a single base missing in the human sequence. Grubb: One other possibility is that they got the sequence wrong in the original paper in dogs, although this seems unlikely. References Dinchuk JE, Liu RQ, Trzaskos JM 2003 COX-3: In the wrong frame in mind. Immunol Lett 86:121 He X, Proske U, Schaible HG, Schmidt RF 1988 Acute in£ammation of the knee joint in the cat alters responses of £exor motoneurons to leg movements. J Neurophysiol 59:326^340 Schwab W, Funk RHW 1998 Innervation pattern of di¡erent cartilaginous tissues in the rat. Acta Anat (Basel) 163:184^190 Segond von Banchet GG, Petrow PK, Brauer R, Schaible HG 2000 Monoarticular antigeninduced arthritis leads to pronounced bilateral upregulation of the expression of neurokinin 1 and bradykinin 2 receptors in dorsal root ganglion neurons of rats. Arthritis Res 2:424^427

Osteoarthritic Joint Pain: Novartis Foundation Symposium 260. Volume 260 Edited by Derek J. Chadwick and Jamie Goode Copyright ¶ Novartis Foundation 2004. ISBN: 0-470-86763-9

Activation of sensory neurons in the arthritic joint Blair D. Grubb Department of Cell Physiology and Pharmacology, University of Leicester, PO Box 138, Leicester LE1 9HN

Abstract. Joints are richly innervated with a range of sensory nerve ¢bres that convey information to the central nervous system about forces exerted on articular tissues by both low and high threshold mechanical stimuli. High threshold nociceptive a¡erents terminate primarily in the synovium and periosteum, and normally respond only to movement of the joint beyond the working limits. Following joint damage, two factors combine to alter the mechanical sensitivity of articular nociceptors. Firstly, physical changes (joint e¡usion and tissue oedema) alter the resting and movement-induced forces exerted on the joint tissues and secondly, in£ammatory mediators released within the damaged tissue sensitize articular nociceptive a¡erents by binding to receptors on the nerve endings. These factors result in a reduction of the mechanical threshold for activation of articular nociceptors such that manipulation of the joint within the normal range is easily su⁄cient to activate them. Acute and chronic animal models of joint in£ammation have been used to study the mechanisms of articular nociceptor sensitization and a number of in£ammatory mediators and their receptors have been implicated. The focus of this paper will be to introduce some of the important issues involved in the sensitization of nociceptive articular a¡erents. 2004 Osteoarthritic joint pain. Wiley, Chichester (Novartis Foundation Symposium 260) p 28^48

Joint innervation Joints are innervated by distinct articular nerves although additional innvervation through accessory nerves is also known to be important. The network of primary a¡erent nerve ¢bres can detect both non-noxious and noxious mechanical stimuli applied to the joint. In animal studies, most recordings from primary a¡erent articular nociceptors have been made in either rat or cat where the anatomy of the articular nerves is well understood and the physiological characteristics of the primary a¡erent nerve ¢bres have been investigated. The precise location of articular nociceptors is an area of considerable interest and several studies have identi¢ed a¡erent nerve ¢bres in joint capsule, synovium, periosteum, proximal 28

JOINT SENSORY NEURONS

29

ligament and tendon suggesting that damage to any part of the joint structure can excite nociceptive a¡erents thus eliciting pain (Schaible & Grubb 1993). The two main systems of classi¢cation for primary a¡erent ¢bres based on conduction velocity or ¢bre diameter, combined with the physiological properties, have been long established. Articular a¡erents can easily be classi¢ed using the same methods and data from several studies have established the range of a¡erent nerve ¢bres innervating joint structures. Detailed morphological analysis of articular nerves has shown that approximately 20% of articular a¡erent ¢bres are myelinated with the majority of this group falling into the group III (Ad conduction velocity¼2.5^20 m s
1) category of ¢nely myelinated « nociceptors. In addition, a relatively small number of large diameter myelinated low threshold group II (Ab conduction velocity¼20^65 m s
1) ¢bres also « innervate each joint. The remaining 80% of articular ¢bres are unmyelinated and studies suggest that approximately half of these are sensory group IV (C-¢bres, conduction velocity¼52.5 m s
1) whilst the remainder are e¡erent sympathetic neurons innervating the joint. Low threshold a¡erents typically have largely corpuscular endings with Ru⁄ni, Golgi and Paccinian endings all found in joint tissues (Johansson et al 1991). The high threshold mechanociceptive articular a¡erents show very little terminal specialization and can only be identi¢ed on the basis of beaded varicosities containing accumulations of mitochondria. These nerve ¢bres are often found in association with blood or lymphatic vessels where they often form a dense network of nerve ¢bres running for considerable distances along each vessel (Heppelmann et al 1990). Mechanical responses of articular a¡erents Articular a¡erents have also been characterized according to their response properties with a range of mechanical sensitivities, from those activated by innocuous manipulation or rotation of the joint, to those only activated by noxious manipulation of the joint or rotation/£exion of the joint beyond the normal working range. More precisely, the majority of articular a¡erents characterized as belonging to group II are low threshold mechanoreceptor and can be either slowly or rapidly adapting (Dorn et al 1991). Interestingly, a few of these ¢bres are only activated by noxious stimuli applied to the joint indicating that they may have a role in nociception. Articular a¡erents belonging to groups III and IV typically have much higher threshold than group II ¢bres, and are most often activated by noxious movements or manipulations of the joint (Schaible & Grubb 1993). In addition, a further class of group III and IV nociceptors has been reported, the ‘silent nociceptors’, which are mechano-insensitive in the normal animal but which develop mechanical sensitivity following the development of joint in£ammation (Grigg et al 1986, Schaible & Schmidt 1988a).

30

GRUBB

Phenotype classi¢cation of primary a¡erent nerve ¢bres Primary a¡erent nerve ¢bres have also been classi¢ed according to their neuropeptide phenotype which provides a guide to the subclasses of nociceptors that exist. In very general terms, primary a¡erent neurons have been classi¢ed into three main groups: . large diameter non-nociceptive neurons and nerve ¢bres that bind the RT97+ve antibody (Bergman et al 1999) which recognizes phosphorylated epitopes on identi¢ed neuro¢lament proteins (Johnstone et al 1997) . isolectin B4 (IB4) positive neurons, which are non-peptidergic nociceptive neurons (Silverman & Kruger 1990), and . calcitonin gene-related peptide (CGRP)-expressing neurons (McCarthy & Lawson 1990), classed as peptidergic nociceptive neurons which can be recognized using selective CGRP antibodies. It should be noted that whilst CGRP is used to identify a distinct sub-population of primary a¡erent nociceptors, many other neuropeptides are also expressed in primary a¡erent neurons innervating joint structures. These include substance P, which is present in about half of the CGRP-positive neurons, and many others including neurokinin A, galanin, opioid peptides and neuropeptide Y. Approximately 25% of all dorsal root ganglion neurons contain CGRP in the normal animal and neuroanatomical studies have clearly demonstrated CGRPpositive nerve terminals in a number of joint structures including synovium, intervertebral disks, ligaments, joint capsule and soft tissues. When Freund’s adjuvant monoarthritis is induced in the rat ankle joint, the proportion of CGRP positive cell bodies in lumbar dorsal root ganglion (DRG) neurons increases markedly presumably as a consequence of the increase in neuronal activation (Hanesch et al 1993). The precise functional consequence of this up-regulation of CGRP expression in sensory nerves innervating the joint is unclear. It is known, however, that these neuropeptides are released from both the central and peripheral terminals of nociceptors where they are involved in central transmission and peripheral neurogenic in£ammation, respectively. Models of in£ammation used to study the properties of articular nociceptors Osteoarthritis (OA) is a signi¢cant clinical problem for which suitable analgesic therapies are required. In recent years a number of animal models have been developed which share similarities with the human disease process. These include naturally occurring OA models in guinea pig and instability-induced OA in mice with a natural predisposition to spontaneous OA. In addition, there are also a

JOINT SENSORY NEURONS

31

number of surgically induced models of OA in dog, rat and guinea pig (Bendele 2002). Unfortunately, none of these models has been used to study the e¡ects of in£ammation on the properties of articular a¡erents. The original studies, which were performed in the 1980s and early 1990s, were designed to look at the process of a¡erent sensitization without any speci¢c disease model in mind. Most studies have used either: . the carrageenan/kaolin model of acute arthritis which produces noticeable changes in a¡erent activity in 1^2 hours and which is normally continued for up to 24 hours, or . the complete Freund’s adjuvant (CFA)-induced mono- or poly-arthritis where injections of heat-killed Mycobacterium tuberculosis or Mycobacterium butyricum induce an arthritic process lasting several days to months. The joint in£ammation normally peaks early at 2^5 days and then declines and either resolves (monoarthritic model) or enters a second phase that can last several weeks (polyarthritic model).

Articular a¡erents and in£ammation Following trauma to the joint we experience an increased sensitivity to load bearing and to movement of the joint within the normal range (allodynia). In addition, there is also a markedly increased sensitivity to any further noxious mechanical stimulation (hyperalgesia). These conscious manifestations of joint damage are brought about by changes in the sensitivity of primary a¡erent nerve ¢bres, by spinal processing of joint input and by processing within higher centres. Following the development of joint in£ammation, low threshold group II articular a¡erents only show acute and relatively transient changes in their responses to joint manipulation which resolve within a few hours. By contrast, articular a¡erents belonging to groups III and IV often start to show ongoing spontaneous activity in the absence of joint movement, and show enhanced responses to manipulation of the joint (rotation/extension/£exion of the joint or capsular indentation using a blunt probe; Coggeshall et al 1983, Guilbaud et al 1985, Schaible & Schmidt 1988a). Furthermore, many units which were previously mechanoinsensitive develop receptive ¢elds and may also show ongoing spontaneous activity. These alterations in the ¢ring characteristics of group III and IV a¡erents are the result of a marked reduction in the mechanical threshold for activation of articular mechanoreceptors and this contributes, in part, to the psychophysical measures of allodynia and hyperalgesia experienced in humans.

32

GRUBB

Mechanisms underlying articular mechanonociceptor sensitization In£ammation is a complex process in which a number of in£ammatory mediators are released in response to tissue damage. In acute in£ammation, these in£ammatory mediators are part of the normal process that is required for tissue repair and regeneration. A number of these in£ammatory mediators are responsible for the sensitization of primary a¡erent articular mechanonociceptors and this serves to limit mobility and thus protect the joint so that it is not damaged further during the repair process. In chronic disease, the in£ammatory response may be protracted due to abnormal pathology in the joint tissues. In the case of OA this is largely due to the destruction of cartilage and bone remodelling (osteophytes or bony spurs) that is a response of bone to local damage. The loss of the normal articulating surfaces and the abnormal bone pathology results in chronic in£ammation that can last years. To understand the mechanisms underlying the sensitization of articular mechanonociceptors in chronic in£ammation it is vital to understand the in£ammatory mediators that are present during each stage of the disease process. One problem is that the in£ammation is not constant, but rather episodic and dependent on a number of factors including activity. This can mean that the in£ammatory mediators present in the joint will vary, thus making the targeting of treatment potentially di⁄cult. Since speci¢c animal models of OA have not been used to study mechanisms of sensitization of articular mechanonociceptors, the information available regarding their sensitization has been generated using the animal models of joint in£ammation mentioned above. An important group of in£ammatory mediators that regulate the mechanical sensitivity of articular mechanonociceptors are the prostaglandins. Prostaglandins are formed when membrane phospholipids are broken down by the enzyme PLA2 to form arachidonic acid. This is then converted by the enzyme cyclooxygenase (COX) to the biologically active prostaglandins, PGE2, PGI2, PGD2 and PGF2a. There is good evidence from the literature that non-selective COX inhibitors (non-steroidal anti-in£ammatory drugs) such as indomethacin and the salicylic acid derivatives reverse the sensitization of articular mechanonociceptors seen in animal models of arthritis (Guilbaud & Iggo 1985, Heppelmann et al 1986). This evidence combined with ample evidence that COX inhibitors are clinically usefully treatments for arthritis in man shows that the joint tissues are a major target for this class of drugs. Further studies have shown that PGE2 (and the related PGE1) and PGI2 seem to be particularly important for the sensitization of nerve ¢bres since these compounds are able to either directly excite articular mechanonociceptors or sensitize them to mechanical stimuli (Heppelmann et al 1985, Schaible & Schmidt 1988b, Grubb et al 1991, Birrell et al 1991, Schepelmann et al 1992).

JOINT SENSORY NEURONS

33

The COX enzyme exists in at least two forms in human (possibly three in dog, Chandrasekharan et al 2002). COX-1 (Funk et al 1991) is a constitutive enzyme that is present in a large number of cell types and is seen as the housekeeping isoform. COX-2 (Hla & Neilson 1992) is only constitutively expressed in a small number of tissues, e.g. brain and spinal cord, but is readily up-regulated in many tissues in response to damage (Seibert et al 1994). In the periphery, an in£ammatory insult will readily up-regulate COX-2 expression resulting in an increase in prostaglandin synthesis. Whilst no a¡erent ¢bre studies have been published showing the roles of COX-1 and COX-2 in producing articular mechanonociceptor sensitization, the clinical success of coxibs, the new selective COX-2 inhibitors, suggests that prostaglandins synthesized through this pathway are responsible in large part for nociceptor sensitization in chronic joint in£ammation. In addition to the prostaglandins, a number of other in£ammatory mediators have been shown to excite and or sensitize articular mechanonociceptors. Bradykinin, for example, has been shown to directly excite the majority of group III and IV articular a¡erents in cat knee and rat ankle joint (Kanaka et al 1985, Birrell et al 1993) and to sensitize them to movements (Neugebauer et al 1989). In addition, it is well established that prostaglandins (PGE2/PGI2) may augment the excitatory or sensitizing actions of bradykinin (Schaible & Schmidt 1988b, Birrell et al 1993, Schepelmann et al 1992). Likewise, another important in£ammatory mediator, serotonin, has been shown to excite approximately two-thirds of articular mechanonociceptors in the rat ankle and cat knee joint. This list is by no means extensive and there exists a large literature on the e¡ects of leukotrienes, histamine, neuropeptides and other in£ammatory mediators on the sensitization of articular mechanonociceptors (Schaible & Grubb 1993). In£uence of joint mechanics An important aspect of joint damage is the e¡usion that accumulates in the joint space. This is particularly important when considering the responses of joint mechanonociceptors to joint £exion or extension. Joint pressure is normally subatmospheric and lowest when the joint is in resting or in the midrange position. However, up to 70 ml of e¡usate can be removed from the in£amed human knee joint (Jayson & Dixon 1970) and this markedly increases the baseline pressure within the joint. More importantly, however, upon £exion or extension, joint pressure increases dramatically, thus increasing stretching of the capsule and surrounding tissues. Since these are densely innervated by articular mechanonociceptors, there can be little doubt that these physical changes in the condition of the joint will also contribute to the enhanced ¢ring of these articular a¡erents.

34

GRUBB

Summary A review of the recent literature shows that relatively little has been achieved in recent years to push forward these early studies of articular mechanonociceptor sensitization in mammalian species. It is certainly true that the majority of in£ammatory mediators have been tested in one of the animal models and there is little doubt that they each have signi¢cant e¡ects on the baseline ¢ring and response properties of articular mechanonociceptors. From these we have a good appreciation of the respective roles of each in£ammatory mediator and of some of the interactions that occur between them. Unfortunately, there are limitations to the information that can be gleaned from these studies since direct recordings cannot be made from the very ¢ne nerve terminals in the joint. The way forward with this type of research is not entirely clear. It is obvious to all who work on joints that articular a¡erents are distinct in that they are present in a number of di¡erent locations, each designed to detect potential damage to the joint through di¡erent perturbations of the joint tissues. On the negative side, there is clearly a gap in the present knowledge regarding the location and responsiveness of joint mechanonociceptors in osteoarthritic joints. If the mechanisms of joint mechanonociceptor sensitization are also di¡erent in OA (or equivalent animal models), i.e. di¡erent to the data obtained in more general models of joint in£ammation, then we will need to re-evaluate the role of these a¡erents in OA. In addition, we still do not know which proteins/ion channels present in joint a¡erent terminals are responsible for mechanosensation although a number have been suggested for di¡erent tissues. An important question, however, is whether articular a¡erents really are unique in terms of the basic mechanisms by which they detect and respond to mechanical stimuli and to in£ammatory mediators. It is clear that there are relatively few data that would allow one to distinguish a joint a¡erent from a skin or muscle a¡erent in terms of phenotype. An interesting paradox is that many joint, skin and muscle a¡erents express the heat-sensitive TRP channel (TRPV1) since they can respond to capsaicin. This means that joint a¡erents have the capacity to respond to damaging or potentially damaging heat stimuli which they are unlikely to experience. A recent hypothesis (Liang et al 2001) suggests that the sensitization of skin a¡erents by bradykinin produces a reduction in the thermal threshold for opening of the vanilloid receptor/channel complex that normally opens only in response to noxious heat. The threshold is reduced so far, indeed, that normal body temperature is su⁄cient to open these channels and excite the nociceptor. Such mechanisms could also be an explanation for the presence of these heat responsive channels in deep tissues and could indicate that there is homogeneity in the mechanisms by which noxious stimuli are detected in di¡erent tissues.

JOINT SENSORY NEURONS

35

In order to progress, we need to establish whether basic common principles apply to all nociceptors irrespective of the tissue that they innervate. We need a better understanding of the basic transduction mechanisms for thermal and mechanical stimuli, the mechanisms of sensitization including an appreciation of the G protein-coupled receptors, signalling pathways and the ion channels that regulate nerve terminal excitability and ¢ring. Only by combining electrophysiological measurements from single cells, molecular biology, protein biochemistry, and modern neuroanatomical and imaging techniques can we hope to start to understand the molecular mechanisms of articular mechanonociception in OA. References Bendele AM 2002 Animal models of osteoarthritis in an era of molecular biology. J Musculoskel Neur Interact 2:501^503 Bergman E, Carlsson K, Liljeborg A, Manders E, Hokfelt T, Ulfhake B 1999 Neuropeptides, nitric oxide synthase and GAP-43 in B4-binding and RT97 immunoreactive primary sensory neurons: normal distribution pattern and changes after peripheral nerve transection and aging. Brain Res 832:63^83 Birrell GJ, McQueen DS, Iggo A, Coleman RA, Grubb BD 1991 PGI2-induced activation and sensitization of articular mechanoreceptors. Neurosci Lett 124:5^8 Birrell GJ, McQueen DS, Iggo A, Grubb BD 1993 Prostanoid-induced potentiation of the excitatory and sensitizing e¡ects of bradykinin on articular mechanonociceptors in the rat ankle joint. Neuroscience 54:537^544 Chandrasekharan NV, Dai H, Roos KLT et al 2002 COX-3, a cyclooxygenase-1 variant inhibited by acetaminophen and other analgesic/antipyretic drugs: Cloning, structure, and expression. Proc Natl Acad Sci USA 99:13926^13931 Coggeshall RE, Hong KAP, Langford LA, Schaible H-G, Schmidt RF 1983 Discharge characteristics of ¢ne medial articular a¡erents at rest and during passive movements of in£amed knee joints. Brain Res 272:185^188 Dorn T, Schaible H-G, Schmidt RF 1991 Response properties of thick myelinated group II a¡erents in the medial articular nerve of normal and in£amed knee joints of the cat. Somatosens Mot Res 8:127^136 Funk CD, Funk LB, Kennedy ME, Pong AS, Fitzgerald GA 1991 Human platelet/ erythroleukemia cell prostaglandin G/H synthase: cDNA cloning, expression, and gene chromosomal assignment. FASEB J 5:2304^2312 Grigg P, Schaible H-G, Schmidt RF 1986 Mechanical sensitivity of group III and IV a¡erents from posterior articular nerve in normal and in£amed cat knee. J Neurophysiol 55:635^643 Grubb BD, Birrell GJ, McQueen DS, Iggo A 1991 The role of PGE2 in the sensitization of joint mechanoreceptors in normal and in£amed ankle joints of the rat. Exp Brain Res 84:383^392 Guilbaud G, Iggo A 1985 The e¡ect of acetylsalicylate on joint mechanoreceptors in rats with polyarthritis. Exp Brain Res 61:164^168 Guilbaud G, Iggo A, Tegner R 1985 Sensory receptors in ankle joint capsules of normal and arthritic rats. Exp Brain Res 58:29^40 Hanesch U, Pfrommer U, Grubb BD, Schaible H-G 1993 Acute and chronic phases of unilateral in£ammation in rat’s ankle are associated with an increase in the proportion of calcitonin generelated peptide-immunoreactive dorsal root ganglion cells. Eur J Neurosci 5:154^161 Hla T, Neilson K 1992 Human cyclooxygenase-2 cDNA. Proc Natl Acad Sci USA 89:7384^7388

36

DISCUSSION

Heppelmann B, Schaible H-G, Schmidt RF 1985 E¡ects of prostaglandins E1 and E2 on the mechanosensitvity of group III a¡erents from normal and in£amed cat knee joints. Adv Pain Res Ther 9:91^101 Heppelmann B, Pfe¡er A, Schaible H-G, Schmidt RF 1986 E¡ects of acetylsalicylic acid and indomethacin on single groups III and IV sensory units from acutely in£amed joints. Pain 26:337^351 Heppelmann B, Messlinger K, Neiss WF, Schmidt RF 1990 Ultrastructural three-dimensional reconstruction of group III and group IV sensory nerve endings (‘free nerve endings’) in the knee joint capsule of the cat: evidence for multiple receptive sites. J Comp Neurol 292:103^ 116 Jayson MIV, Dixon ASJ 1970 Intraarticular pressure in rheumatoid arthritis of the knee. I. Pressure changes during passive joint distension. Ann Rheum Dis 261^265 Johansson H, Sjolander P, Sojka P 1991 Receptors in the knee joint ligaments and their role in the biomechanics of the joint. Crit Revs Biomed Eng 18:341^368 Johnstone M, Goold RG, Fischer I, Gordon-Weeks PR 1997 The neuro¢lament antibody RT97 recognises a developmentally regulated phosphorylation epitope on microtubule-associated protein 1B. J Anat 191:229^244 Kanaka R, Schaible H-G, Schmidt RF 1985 Activation of ¢ne articular a¡erent units by bradykinin. Brain Res 327:81^90 Liang Y-F, Haake B, Reeh PW 2001 Sustained sensitization and recruitment of rat cutaneous nociceptors by bradykinin and a novel theory of its excitatory action. J Physiol 532:229^239 McCarthy PW, Lawson SN 1990 Cell type and conduction velocity of rat primary sensory neurons with calcitonin gene-related peptide-like immunoreactivity. Neuroscience 34:623^ 632 Neugebauer V, Schaible H-G, Schmidt RF 1989 Sensitization of articular a¡erents to mechanical stimuli by bradykinin. P£ugers Arch 415:330^335 Schaible H-G, Schmidt RF 1988a Time course of mechanosensitivity changes in articular a¡erents during a developing experimental arthritis. J Neurophysiol 60:2180^2195 Schaible H-G, Schmidt RF 1988b Excitation and sensitization of ¢ne articular a¡erents from cat’s knee joint by prostaglandin E2. J Physiol 403:91^104 Schaible H-G, Grubb BD 1993 A¡erent and spinal mechanisms of joint pain. Pain 55:5^54 Schepelmann K, Messlinger K, Schaible H-G, Schmidt RF 1992 In£ammatory mediators and nociception in the joint: excitation and sensitization of slowly conducting a¡erent ¢bers of cat’s knee by prostaglandin I2. Neuroscience 50:237^247 Seibert K, Zhang Y, Leahy K et al 1994 Pharmacological and biochemical demonstration of the role of cyclooxygenase 2 in in£ammation and pain. Proc Natl Acad Sci USA 91:12013^12017 Silverman JD, Kruger L 1990 Selective neuronal glycoconjugate expression in sensory and autonomic ganglia: relation of lectin reactivity to peptide and enzyme markers. J Neurocytol 19:789^801

DISCUSSION Brandt: Earlier on, Peter Simkin asked a question about bone versus synovial space. I was reminded of the elegant work by Arnoldi and Lemberg in the 1970s (Arnoldi et al 1972), who measured blood£ow through subchondral bone in OA joints and showed that it is decreased. In OA, the CO2 concentration in blood from the medullary space is elevated, lactate concentration is increased and O2 tension is decreased. Furthermore, Arnoldi and Lemberg demonstrated that the pain of

JOINT SENSORY NEURONS

37

osteoarthritis could be relieved immediately and completely by an osteotomy, a surgical procedure that changes the pressure within the medullary space but which has nothing to do with treating the in£ammatory change directly. The synovium is easier to study than the bone, but I wouldn’t limit considerations of OA pain to the articular space. Grubb: In the models that we have used to study nociceptors, there is little damage to the bone. The studies of a¡erent ¢bre sensitization have been done in the wrong sorts of models. They were done in animal models where the primary aim was to look at a¡erent ¢bre sensitization, and they were not studies aimed particularly at OA. The contribution of bone pain to OA and the contribution of a¡erent ¢bres is simply a huge de¢cit in our knowledge. Kuettner: In clinical terms, we are studying OA in older individuals. In your animal models, have you taken age as a component? Most animal work is done on young animals. Grubb: It is almost all done on young animals. Kuettner: The question is, can we interpret data from young animals and apply them to the adult population either in humans or animals? With the exception of the guinea pig, we have di⁄culties reproducing some of the animal models of OA in old animals. Grubb: That is a valid question and we don’t know the answer. Henry: This raises another issue: when does OA start? Does it start with the clinical signs, or before? Dieppe: Good question. I don’t have an answer, but it is probably long before we see clinical signs. Moving to a completely di¡erent issue, one of the fascinating things about musculoskeletal pain is that it is sensitive to barometric pressure: patients can predict the weather. Is it possible from what you are saying that the increased sensitization of mechanoreceptors is such that they can respond to these miniscule changes in barometric pressure? The patients can certainly detect them. If it is not this, then what on earth is going on? Grubb: I don’t know the answer to that, but I can speculate. I understand that people su¡er more in damp, cold weather when the pressure is falling. It is true that a number of joint mechano-nociceptors are sensitive to gentle pressure on the capsule in an in£amed joint. It may be that modest capsular stretching as a result of a di¡erence between the articular and exterior pressure could be big enough to produce some additional activation of the nociceptive nerve endings. It is interesting that this only happens when pressure is falling, because this would be when any change in the capsule stretching would be occurring. I have no evidence for this, though, and I stress that this is pure speculation. Koltzenburg: It may be that the source of pain in OA is much more the bone than the joint itself. Reduction in bone pressure by bone marrow aspiration is also very

38

DISCUSSION

painful. I don’t know how this relates to a fall in barometric pressure: this would be interesting to discuss. I have a question about the TRPV1 (VR1) receptor. How do you think this is implicated in mechanosensation? I can see that this might be playing an important role in heat sensitization, but when the vanilloid receptors are activated they don’t change the mechanosensitivity in the same sensory neurons and there was also no change in mechanically-evoked cutaneous pain in mice lacking TRPV1. Grubb: All I can say is that many neurons respond to both mechanical and thermal stimuli. All I am o¡ering in putting forward Peter Reid’s model is the idea that bradykinin present in the damaged tissue may be activating a proportion of the neurons through activation of the TRPV1 receptor, thus perhaps changing the background spontaneous activity of the a¡erent ¢bres. Whether this occurs in addition to the changes in mechanosensitivity is not clear. In our original experiments, we never thought to apply any thermal stimuli to deep tissues because you would think that they wouldn’t normally experience this. But if they respond to capsaicin they must contain the TRPV1 heat sensitive ion channel. If it is true that sensitization drops their heat threshold such that they are activated by body temperature, then bradykinin could produce its sensitizing e¡ects in this way. This is an interesting idea. Schaible: There is some work by Besson’s group on the polyarthritic rat. They found that these animals are mechanically hypersensitive, but with respect to heat they are hypoalgesic (for reference see Schaible & Grubb 1993). I am not sure whether this Peter Reid model applies to the joint. Grubb: The problem with the CFA polyarthritis model is that it is a profound arthritic model with arthritis in all joints, skin lesions and skin in£ammation. I am unsure what you can say about heat sensitivity of joint a¡erents in this model. Pisetsky: How are gravity, weight bearing and compression perceived? Grubb: There have been studies done on capsular a¡erents from way back, looking at loading of the capsules in di¡erent planes. Loading the capsule axially produces the most appropriate stimulus for activating these a¡erents. Pisetsky: What about just weight bearing? Grubb: I am not sure. Evans: I was interested in the suggested pathogenesis of OA that you showed. You had one chain of arrows going through the in£ammatory route, and then you had the particles o¡ to the side shunted directly into pain. Were you implying that independently of their in£ammatory properties they have the ability to induce pain? If so, how? Grubb: These articular nociceptors have become extremely mechanically sensitive. If the subchondral bone has become damaged and the nociceptors are very close to the surface, and you have particular matter in there, this could be contributing. Of course, the articular surfaces rubbing against each other could

JOINT SENSORY NEURONS

39

be a more powerful stimulus. I just had them in there to suggest that they contribute to the nociceptor activation. Henry: That would overlook the idea that there isn’t a 100% correlation between the pathology and the pain. Brandt: Can you give us a quantitative sense of the duration of the latent period that precedes ¢ring of a nociceptor in the joint versus that of a nerve ending that deals with proprioception? Grubb: The conduction velocities of C ¢bres are usually 2^2.5 m/s, but large ¢bres will conduct at 30^100 m/s. The nociceptors are slow conducting ¢bres and the proprioceptors are much faster. Pisetsky: You de¢ned noxious movement in terms of the extent of movement. What about the rate at which a joint is moved? Grubb: I presented a lot of work from Hans-Georg Schaible, and he has done most of the work on joint movement, so I think he should answer that question. Schaible: Whenever we move our joints in the normal range there is not much torque. But when we twist the joint against the resistance of the tissue this is what we call noxious movement. The rate of movement doesn’t matter. Moving the joint quickly will only cause the low threshold mechanoreceptor ¢bres to ¢re. This will not activate a C ¢bre. A nociceptor is usually not even activated during the ramp of a stimulation, but when the stimulus is there, it is activated one or two seconds after the joint has reached an extreme position. In the normal movement range C ¢bres will not be activated. Pisetsky: Does duration of joint use matter? Schaible: It isn’t easy to answer this question. The tests were usually 15 seconds long, and this short duration doesn’t change the response pattern. Hunter: In an abnormal joint which is mechanically unstable, would you expect that this would produce a nociceptive stimulus in a lower range of movement than is needed for a normal joint? Schaible: The question is whether mechanical changes alone would be su⁄cient to change the response properties. From our experience so far we believe that if you apply just mechanical stimuli, the response properties will not change a lot. You need something additional. We can’t really answer this sort of speci¢c question because our studies have not addressed it. Felson: Can you give us a sense of the mechanical stimuli you have applied? The local stresses within osteoarthritic joints are often very high. Is there some way that you can characterize the magnitude of the twisting beyond the normal range of motion that you apply? Schaible: It has been measured but I can’t recall the numbers. Fernihough: What is seen as a noxious twisting movement by ligaments may not be seen at all by the synovium, which is a fully deformable tissue. Do you think that some of the areas that contribute to a pain response are also the response of the

40

DISCUSSION

nerves within those tissues to di¡ering mechanical stresses? Following on from this, has anyone seen things like TRPV1 expression in di¡erent tissues within a joint? Grubb: The responses are to the twisting of the joint and to £exion of the joint beyond the normal range. As shown in work by Ferrell et al (1986), massive changes in the ¢ring of the articular nerves are seen if there is an e¡usion. It really makes a huge di¡erence. This could contribute to the ¢ring of synovial a¡erents as well. It depends on the volume of e¡usion, which varies in OA. Herzog: One of the interesting questions is raised when you have unstable joints, such as in some forms of OA where the anterior cruciate ligament (ACL) is absent and twisting movements like medial rotation become much easier. The ACL is one of the primary structures to prevent excessive anterior translation and medial rotation of the tibia relative to the femur. Therefore, if the ACL is absent (torn), anterior tibial translation and medial rotation encounter much less resistance than in the intact knee (e.g. Maitland et al 1998). When you apply this noxious torque, and then take one of the structures out that primarily resist the torque (e.g. the ACL) and do the same displacement but this time in the absence of torque, do you still see the same nociceptive response? Does it come from a speci¢c tissue or is it a general joint response? Schaible: The anterior cruciate ligament seems to be quite important. I had a collaboration with an orthopaedic surgeon, and they said if they treated the joint and the ACL wasn’t properly repaired then there were a lot of problems afterwards. We have studied the response properties of mechanoreceptors supplying the ACL (Krauspe et al 1992). On each movement they exactly indicate the torque and the stress which is applied to the joint. They ¢re all the time when you move the joint and increase the stress level. This proprioception is quite necessary. There is a disorder called Charcot’s joint, where proprioception is absent or reduced and the joints are damaged. We need proprioceptive control all the time, and the ACL with its mechanoreceptors seems to be particularly important for this reason. Koltzenburg: Charcot joints are often a sequel of severe neuropathy. It is not necessarily something speci¢c for a large ¢bre neuropathy and it seems to be more frequent in patients that have demonstrable small ¢bre neuropathy. Brandt: Coming back to the issue of joint e¡usion, it is worth keeping in mind that stresses and torques on joints are induced by muscle contraction. A phenomenon called arthrogenous re£ex inhibition comes into play here. As we develop a knee e¡usion, contraction of our quadriceps muscle is inhibited. This protects us from rupturing the joint. Herzog: That’s a good point. In the presence of symptoms (joint pain, swelling of the joint, loss of joint stability caused by ligament damage etc.), the muscles that produce joint movements have been shown to behave distinctly di¡erent (e.g. the force they produce may be reduced, and the coordination among muscles of a

JOINT SENSORY NEURONS

41

synergistic group may be disrupted) from normal. For example, in our cat model of OA (ACL transaction: Herzog et al 1993), we demonstrated that joint instability, caused by ligament transection, was associated with decreased knee extensor forces, and a completely di¡erent activation pattern of the knee £exors relative to steps cycles during walking and running (Hasler et al 1998). Qualitatively, similar results have been obtained for humans, where swelling was associated with a decrease in knee extensor activation compared to normal (e.g. de Andradne et al 1965, Fahrer et al 1988, Stratford 1981), although the muscle forces in these situations were never measured, but were inferred through the electrical activity patterns (measured using surface electromyograms). Regarding ‘re£ex inhibition’ of muscles in injured or painful joints, there is a vast body of literature showing that OA, joint swelling, pain, joint instability and other joint pathologies are associated with a reduced ability for muscle force production (e.g. Brandt 1997, Hurley et al 1994, Suter et al 1998, Suter & Herzog 2000). Re£ex inhibition is measured by asking a subject to perform a maximal voluntary contraction of the muscle group of interest, measuring the joint moment exerted by the muscle contraction, and then superimposing an arti¢cial electrical stimulation on the maximal e¡ort contraction (Merton 1954, Belanger & McComas 1981). The superimposed stimulation to the muscle or the corresponding nerve will result in additional force production if the maximal e¡ort did not recruit all motor units maximally. Such muscle inhibition is normal (i.e. in the knee extensors it varies from about 5^10%, depending on the knee angle, for normal subjects), but reaches high, pathological levels (50^80%) in some patients with knee (joint) pathologies. Simkin: I am going to mention some results from Arnoldi’s work in my paper which show that when a horse metacarpophalangeal joint £exes to 90 degrees the intraosseus pressure soars dramatically (Arnoldi et al 1980). A number of studies in various species have been done, usually with a small e¡usion, that show a very similar response of intraosseus pressure. This feeds into Ken Brandt’s earlier point about intraosseus pressure and pain, and perhaps into the sensitization. Grubb: One important issue that I haven’t covered is that if we really believe that these a¡erents in the subchondral bone are particularly important for sensing pain in OA, then we need to develop a model for recording from them. What may also be interesting is that if the nociceptors are truly in the bone, do you get remodelling of the terminals within the bone as a result of the ongoing damage in OA? They may be sprouting or changing, constantly having to re-grow and sprout. Brandt: They may also be present in the walls of the blood vessels. Grubb: In the joint there is very good evidence that a lot of the a¡erents are running along the blood vessel. This could be very true. As you get changes

42

DISCUSSION

in the vascularization of the bone, you may get changes in the sensitivity of the nociceptors. Who knows? There may be a neuropathic component of OA. Brandt: With regard to the issue of innervation or not of articular cartilage, with advanced disease when there is capillary invasion these capillaries do contain nerve endings. Fernihough: I think the innervation of cartilage is right at the edge where osteophyte formation is seen (Schwab & Funk 1998). Creamer: I can see how local in£ammation causes peripheral sensitization, so that mechanical stimuli are perceived as pain. If you take away the in£ammation, does this sensitization remain? Grubb: I have never studied a model of in£ammation where the in£ammation has been allowed to resolve. But when we use enzyme inhibitors to take away some of the in£ammatory mediators, the sensitization goes away fairly rapidly, within the time period you would expect to see the breakdown of the prostaglandins. Those tend to be fairly quick e¡ects in terms of the prostaglandins. But I am not sure whether this is true for all cytokines and in£ammatory mediators. Creamer: Do you therefore think that a constant level of in£ammation is needed in an OA joint for the maintenance of the sensitization? Grubb: That is an interesting question. I don’t know whether the damaged articulating surfaces and the enhanced mechanical stimuli applied are su⁄cient to produce activation of the nociceptors in the absence of in£ammation in OA. As we have heard, there is a poor correlation between in£ammation and pain. I would always imagine that if you have that amount of damage, there must be some in£ammation in the background all the time. Pisetsky: Have people done these kinds of experiments in either transgenic or knockout mice with relevant cytokines? For example, if an interleukin (IL)6 transgenic mouse is exposed to in£ammation for a long time, is it more or less likely to be sensitized? There are enough mouse models around for this to be approached. Lohmander: What is the degree of biological variability in the sensitization response in these standardized animal models? Do you have any understanding at this time of the biological basis for this variability, if it exists? And is there anything in these models that could teach us about the basis of the variability that appears to be present in the response of patients to what seems to be a standardized structural change in the joint? Grubb: I presented a model where there were two classes of pain a¡erents: peptidergic and non-peptidergic. This is hugely simplistic. In skin, a¡erents have been characterized or broken down into a number of di¡erent types
nociceptors in particular. There are many subgroups depending on their sensitivity to thermal or mechanical stimuli and so on. We know to some extent which ones exist within

JOINT SENSORY NEURONS

43

the joint, but we don’t know how each of these di¡erent types of nociceptor respond nor how each one is sensitized by in£ammatory mediators. Koltzenburg: At this point in time it is unclear whether, and if so how, the phenotype of primary a¡erent neurons innervating the skin and joint di¡ers. Part of the di¡erence may simply arise because of the fact that the terminal endings are embedded in a di¡erent tissue so that for example a given mechanical stimulus acts di¡erentially on the receptors. Lohmander: Is there a genotype for pain response? In other words, what do we know about the in£uence of genetic variation on the variation in pain response? Koltzenburg: The central termination patterns of cutaneous and deep somatic a¡erents are clearly di¡erent. This would imply that the molecules that specify this di¡erent speci¢city presumably are a result of di¡erential gene expression pattern during development. Lohmander: Part of the reason I am asking is because of the current searches for the ‘OA gene’, which are often linked to OA pain as the phenotype. Are we in these studies looking for the pain genotype instead of the OA genotype? Manning: There certainly are di¡erences between strains and within strains in their response to various noxious stimuli. People such as Je¡ Mogul have been phenotyping across strains and then looking for the genetic basis for this phenotype. Koltzenburg: There is one caveat. We often assume that di¡erence in behaviour in inbred mouse strains is attributed to the genetic di¡erences. However, the e¡ect of upbringing in di¡erent inbred strains has not been evaluated for nociception. The onus is on him to do cross-fostering experiments to show whether this change he attributes to genetics will hold. Manning: The observation is clear that there are variations within a population and across strains. The basis for these is unclear at present. No one has really looked at this in the kinds of non-surgical OA models that we believe are the most relevant to human OA. Grubb: It is important to remember that the phenotypes of the di¡erent nociceptors are not constant. This changes during in£ammation. Work done years ago shows that CGRP and substance P expression changes markedly in in£ammation. We see other changes in receptors and ion channels in these models. Nociceptors are of many types and they are also plastic: they change their phenotype in response to di¡erent insults, which makes studying them complicated. Koltzenburg: What do people think is the crucial tissue that generates the pain in crystal-induced arthritis, rheumatoid arthritis and OA? Have people done local anaesthetic blocks or in¢ltration anaesthesia? The fact that joint replacement on OA eliminates pain suggests that the synovium isn’t so much involved.

44

DISCUSSION

Felson: Most of us would be unsure which tissue is the source of pain in everyone. Creamer: We looked at what happens when we inject local anaesthetic into OA limbs. E¡ectively, pain is abolished in about 80% of people. On a visual analogue score a lot of people will score zero, but for some people it appeared to have no e¡ect at all. We concluded that most people had a locally driven cause for their pain, but others had a very central cause. Identifying the tissue that is the cause of pain in OA is a bit of a holy grail. We interpreted those experiments as saying that whatever it is, it must be in contact with the intra-articular environment, or somewhere that the anaesthetic could di¡use into. Koltzenburg: So not bone. Creamer: We weren’t really sure whether it would get through. In people with established OA and lots of cartilage loss, the bone might be accessible to intraarticular £uid. Simkin: If I recall correctly your endpoint was 1 h. What was it like after 10 min? All intracapsular tissues should have been anaesthetized by then? Creamer: We didn’t measure this. Some e¡ects persisted for 24 h and even seven days. The other thing that came out of this study is that we looked at what happened on the uninjected knee. Interestingly, there was some symmetry of response. These were all people who had bilateral knee pain, and their worse knee was injected with either saline or anaesthetic. In the group with anaesthetic not only did their index knee pain reduce, but it also decreased in the non-injected knee. Felson: As someone who reads that paper often and cites it frequently, the n of response is 6 out of 10 for me, not 8 out of 10. That is, 4 out of 10 did not have a response that was dramatic. I would interpret this as being related to bone origin pain. Pisetsky: There is also an animal model side of this question. You create an animal model of OA and look at cartilage. But, from what I am hearing, I should also be looking at ligaments, tendons, capsule, innervation and so on. Everyone just scores cartilage, but perhaps we should be expanding how we evaluate the animal models. Hunter: The structural changes that occur in OA and the measurement of these will probably go through a phase leap over the next two or three years, with the use and involvement of magnetic resonance imaging (MRI). This will give us a better structural characterization of which structures are important. Also, using knowledge that we already have from MRI, we can show that bone is an important source of pain in so far as bone marrow oedema is a signi¢cant source of pain in patients. Pisetsky: Even in human disease, much of the emphasis is on managing cartilage.

JOINT SENSORY NEURONS

45

Grubb: Joint replacement alleviates the pain. Immediately, this makes us think that the pain must come from the articulating surfaces. If those grinding articular surfaces generate an in£ammation which produces in£ammatory mediators, you may have pain not only from the articulating surfaces but also from nociceptors that are close to the contents of the joint capsule itself. By getting rid of the source of the in£ammation, this may eliminate the sensitization, which need not be limited to the subchondral bone. Dieppe: As many people in this room know, I think OA is a bone disease. Carlo Arnoldi and his group did critical experiments on bone, carrying out decompression and showing that pain, particularly night pain, improved. I think we are forgetting about the heterogeneity of the pain experience in OA. I don’t think we have done enough work trying to dissect out the possibility that di¡erent experiences of pain might relate to di¡erent tissues generating it
night pain, for example, being related to intraosseous pressure. There is also some work suggesting that some types of pain might relate more to the presence of e¡usion. To suggest that there is a single answer and tissue to explain OA pain is na|« ve, given the huge heterogeneity of the pain experience. Schaible: With regard to the reproducibility of these e¡ects, I think the phenomenon of sensitization in the models that we use is very similar across models. However, this doesn’t mean that the molecular events that are leading to this sensitization are always the same. We have experience from a chronic in£ammatory model lasting over 50 days, where we have a thick joint all the time but we can see an ongoing disease process and an up- and down-regulation of receptors (Segond von Banchet et al 2000). It is possible that OA is di¡erent at di¡erent time points in the same joint, and the pain process could also di¡er. This is another complication. Brandt: I agree with Paul Dieppe about the heterogeneity within tissues. Heterogeneity between sites also exists. In the hip, a small e¡usion has a vastly greater e¡ect on the intra-articular pressure than it does in a knee joint. No one has yet mentioned periarticular pain, which is common in arthritis. In our experience, 1 of every 4 patients with knee OA has this, at least intermittently, as the basis for their ‘knee pain’. It is associated with local sharp tenderness over the anserine bursa, which can be relieved by a local injection of anaesthetic or steroid. ‘OA’ pain varies not only from patient to patient, but also within the same patient from visit to visit. Kidd: Coming back to age-related changes, there is evidence from the psychophysical studies that as we get older, our response to capsaicin is dramatically reduced, and there is a nice linear decrease in capsaicin responsiveness in humans over a period of time. Paradoxically, though, the pain response is actually enhanced. So here you have a nice example of nociceptor response not being mirrored by the pain response. I have a question relating to

46

DISCUSSION

the silent nociceptors which Dr Schaible described over a decade ago. This was a pivotal observation that within the joint many of the a¡erents are silent. This fed back into observations made in the 1930s, where a group of heroic surgeons here in London went into joints unanaesthetized, poked and prodded them and got no pain response whatsoever. This raises the issues of whether we always need an in£ammatory mediator to get synovium-driven pain. Do you feel that there are an equal number of silent nociceptors supplying bone tissues? Pat Mantyh recently showed a much denser innervation of bone than we had hitherto suspected. They seem to be unmyelinated a¡erents. Do you feel that we are likely to have an equal proportion of silent nociceptors supplying the bone as are seen in synovium? This would imply that an in£ammatory response is needed to sensitize these ¢bres and hence give pain? Schaible: It is always di⁄cult to say how many silent nociceptors there are. They need an additional stimulus such as sensitization to become responsive. They are found in the skin at about 10%, which is not that much. My guess in the deep tissue of the joint is that about 30% are silent. It could well be that the density of silent nociceptors is important with regard to how strong pain is. Pisetsky: Is there any relatively simple operational way to ¢nd out whether a patient’s joint has been sensitized? Schaible: We cannot di¡erentiate between a peripheral and a central sensitization. You can just ¢nd out whether the patient is hypersensitive or not. Pisetsky: Are people with OA sensitized in that joint area? Schaible: We can’t tell. Pisetsky: People get better rapidly
sometimes the day after
following joint replacement surgery. What does this imply about sensitization? Or is the pain from a diseased joint OA so much greater than the pain from sensitization, so that when the joint is replaced, the pain essentially goes away even if there is some sensitization. Schaible: We can counteract sensitization by increasing inhibitory systems, for example. This is another complication. Koltzenburg: It is much better understood for cutaneous pain and some forms of neuropathic pain, where central sensitization in the absence of peripheral a¡erent drive is very rare. In most experimental models you can show that a large component of central sensitization is in fact a physiological re£ection of the peripheral nociceptive a¡erent drive. There are some phenomena that go beyond this. Hyperalgesia to pinprick stimuli that is commonly found in the capsaicin model is initiated by nociceptor a¡erent input but does not require primary a¡erent input for its maintenance. However, in most other cases of hyperalgesia, central sensitization will regress without primary a¡erent input. Pisetsky: Over how long a period of time?

JOINT SENSORY NEURONS

47

Koltzenburg: With capsaicin it is in the range of minutes. In some conditions it may take much longer. As far as it has been tested we are talking hours at the most. Grubb: What proportion of OA patients respond to non-steroidal antiin£ammatory drugs (NSAIDs)? This is an indication of whether they are sensitized or not: if they respond to NSAIDs, there is at least prostaglandininduced sensitization of the a¡erents. Brandt: About 30% have marked improvement, a few get worse and the majority exhibit modest improvement. If you add paracetamol to the NSAID some additional analgesia is seen, but a very high proportion of OA patients treated with both agents still complain of residual pain. Grubb: Is the rest of the pain due to in£ammatory mediators other than prostaglandin, or is it not due to sensitization? This is an interesting question. Hunter: If you block the sensitization with substance P or a NMDA inhibitor, what impairment do you see on other neural re£exes and responses? In particular, what e¡ect do you think this may have on proprioception and motor control? Koltzenburg: First, substance P antagonists have not worked clinically. If you give NMDA antagonists in an experimental setting, most subjects will complain about psychotropic side-e¡ects. The majority of the e¡ects are not necessarily on proprioception but more on cognitive e¡ects. This has so far been the limiting factor in the use of NMDA antagonists.

References Arnoldi CC, Linderholm H, Mussbichler H 1972 Venous engorgement and intraosseous hypertension in osteoarthritis of the hip. J Bone Joint Surg Br 54:409^421 Arnoldi CC, Reimann I, Mortensen S et al 1980 The e¡ect of joint position on juxta-articular bone marrow pressure. Relation to intra-articular pressure and joint e¡usion
an experimental study on horses. Acta Orthop Scand 51:893^897 Belanger AY, McComas AJ 1981 Extent of motor unit activation during e¡ort. J Appl Physiol 51:1131^1135 Brandt KD 1997 Putting muscle into osteoarthritis. Ann Intern Med 127:154^155 de Andradne JR, Grant C, Dixon ASJ 1965 Joint distension and re£ex muscle inhibition in the knee. J Bone Joint Surg Am 47:313^322 Fahrer H, Rentsch HU, Gerber NJ, Beyeler C, Hess CW, Gruenig B 1988 Knee e¡usion and re£ex inhibition of the quadriceps. J Bone Joint Surg Br 70:635^638 Ferrell WR, Nade S, Newbold PJ 1986 The interrelation of neural discharge, intra-articular pressure, and joint angle in the knee of the dog. J Physiol 373:353^365 Hasler EM, Herzog W, Leonard TR, Stano A, Nguyen H 1998 In-vivo knee joint loading and kinematics before and after ACL transection in an animal model. J Biomech 31:253^262 Herzog W, Adams ME, Matyas JR, Brooks JG 1993 A preliminary study of hindlimb loading, morphology and biochemistry of articular cartilage in the ACL-de¢cient cat knee. Osteoarthritis Cartilage 1:243^251 Hurley MV, Jones DV, Newham DJ 1994 Arthrogenic quadriceps inhibition and rehabilitation of patients with extensive traumatic knee injuries. Clin Sci (Lond) 86:305^310

48

DISCUSSION

Krauspe R, Schmidt M, Schaible HG 1992 Sensory innervation of the anterior cruciate ligament. An electrophysiological study of the response properties of single identi¢ed mechanoreceptors in the cat. J Bone Joint Surg Am 74:390^397 Maitland ME, Leonard TR, Frank CB, Shrive NG, Herzog W 1998 Longitudinal measurement of tibial motion relative to the femur during passive displacements and femoral nerve stimulation in the ACL-de¢cient cat model of osteoarthritis. J Orthop Res 16:448^454 Merton PA 1954 Voluntary strength and fatigue. J Physiol (Lond) 123:553^564 Segond von Banchet GG, Petrow PK, Brauer R, Schaible HG 2000 Monoarticular antigeninduced arthritis leads to pronounced bilateral upregulation of the expression of neurokinin 1 and bradykinin 2 receptors in dorsal root ganglion neurons of rats. Arthritis Res 2:424^427 (available on http://arthritis-research.com/content/2/5/424) Schaible HG, Grubb BD 1993 A¡erent and spinal mechanisms of joint pain. Pain 55:5^54 Schwab W, Funk RH 1998 Innervation pattern of di¡erent cartilaginous tissues in the rat. Acta Anat (Basel) 163:184^190 Stratford P 1981 Electromyography of the quadriceps femoris muscles in subjects with normal knees and acutely e¡used knees. Phys Ther 62:279^283 Suter EW, Herzog W 2000 Does muscle inhibition after knee injury increase the risk of osteoarthritis? Exerc Sport Sci Rev 28:15^18 Suter EW, Herzog W, DeSouza K, Bray RC 1998 Inhibition of the quadriceps muscles in patients with anterior knee pain. J Appl Biomechan 14:360^373

Osteoarthritic Joint Pain: Novartis Foundation Symposium 260. Volume 260 Edited by Derek J. Chadwick and Jamie Goode Copyright ¶ Novartis Foundation 2004. ISBN: 0-470-86763-9

Neuromuscular aspects of osteoarthritis: a perspective Kenneth D. Brandt Multipurpose Arthritis and Musculoskeletal Diseases Center, and Performing Arts Medicine Program, Indiana University School of Medicine, 1110 West Michigan Street, Indianapolis, IN 46202-5100, USA

Abstract. Osteoarthritis (OA) represents failure of the diarthrodial joint and may be due to a primary abnormality in any of the tissues of the joint, e.g. articular cartilage, subchondral bone, synovium, periarticular muscle, or sensory nerves whose termini lie within the joint. Neuropathic arthropathy, due to severe sensory neuropathy, causes severe and rapid breakdown of joints. We have shown that interruption of sensory input from the ipsilateral hind limb strikingly accelerates progression of OA after anterior cruciate ligament transection in the dog; a clinical correlate exists in humans with diabetic neuropathy who sustain even minor joint trauma. Knee OA in humans is accompanied by defects in proprioception, although it is not clear whether the neurological abnormality is primary or a consequence of intra-articular pathology. The magnitude of the load on a joint and, especially, the rate of impulsive loading, in£uence development of OA. It is relevant, therefore, that quadriceps weakness may precede development of knee OA in some people, insofar as this may diminish the e¡ectiveness of protective muscular re£exes and thereby increase deleterious joint loading. Individuals vary with respect to how they load their joints, perhaps because of genetic di¡erences in central program generators. 2004 Osteoarthritic joint pain. Wiley, Chichester (Novartis Foundation Symposium 260) p 49^63

Although the most striking pathologic changes in osteoarthritis (OA) are usually found in the articular cartilage, OA is not a disease of any single tissue (e.g. cartilage) but a disease of an organ, the synovial joint. It is analogous, therefore, to heart disease, in which a primary problem in the endocardium, epicardium or myocardium may produce a syndrome of congestive heart failure. In some cases the joint may fail because of a primary problem in the articular cartilage; in other cases the problem may reside in the subchondral bone, synovium, periarticular muscles, supporting ligaments or the nerves whose sensory termini lie within the joint or surrounding tissues. Although it is popular to categorize OA as being either primary (aetiopathic) or secondary, in a general sense all OA is secondary. A genetic abnormality, such as a 49

50

BRANDT

FIG. 1. Disability in patients with OA is due to the inactivity associated with the disease and to the e¡ects of aging, as well as to the direct e¡ects of the intra-pathology. Reproduced with permission from Stenstrom & Minor 2003.

point mutation in the cDNA coding for Type II collagen, may lead to spondyloepiphyseal dysplasia, resulting in wide-spread and early-onset OA and providing an example of OA arising as the result of a primary cartilage disorder. Genetically based systemic metabolic abnormalities, such as haemochromotosis, Wilson’s disease and calcium pyrophosphate crystal deposition (CPPD) disease are also genetic conditions leading to chondropathies and to secondary OA. However, the OA phenotypes associated with these genetic disorders account for only a very small proportion of the universe of clinical OA. Considering OA as a clinical entity, it is the major cause of chronic disability among the elderly, in whom it is the most common cause of di⁄culty in climbing stairs and in walking. OA is chie£y a disorder of the elderly. In considering disability in OA, one must take into account not only the direct consequences of the structural changes in the joint due to the arthritis, but also the consequences of ageing and of inactivity (Stenstrom & Minor 2003; Fig. 1).

Adverse e¡ects of immobilization of a joint Much of the discussion centring around the relationship between biomechanics and OA relates to overloading of the joint, i.e. the concentration of peak dynamic loads. It is less widely recognized that a reduction in loading is also detrimental to joint tissues. For example, immobilization of the hind limb of a normal dog by application of an orthopaedic cast, restricting motion of the knee and unloading of the joint during ambulation, rapidly results in striking changes in the articular cartilage (Palmoski et al 1979). The cartilage thins, the net rate of

NEUROMUSCULAR ASPECTS

51

FIG. 2. Articular cartilage, stained with Safranin-O, from the femoral condyle of the left knee (left hand panel) of a dog whose right hind limb had been immobilized in an orthopaedic cast for 8 weeks (magni¢cation
35). Immobilization resulted in a marked decrease in the thickness of the articular cartilage of the immobilized knee, with depletion of proteoglycans and a decrease in the net rate of proteoglycan synthesis. Note also the marked disuse osteoporosis that occurred during the period of immobilization (right hand panel).

proteoglycan synthesis is reduced by as much as 50%, and reductions of similar magnitude occur in the proteogylcan concentration and proteogylcan content of the cartilage (Fig. 2). These changes are associated with striking disuse osteoporosis of the subchondral bone and atrophy of the periarticular muscles. If the period of immobilization is prolonged, ¢brofatty ankylosis of the joint occurs. If, however, the period of immobilization is limited to a few months, the above changes are completely reversible following removal of the cast and remobilization of the extremity. Immediately upon restoration of loading, the chondrocytes resume synthesis of matrix proteoglycans, which are produced at a greatly accelerated rate
comparable to, or greater than, that seen in OA. The osteoporosis in the subchondral bone is also reversed. Within only a few weeks, biochemical, biomechanical, metabolic and ultrastructural analyses of the cartilage provide no hint of the profound changes that had been present only a few weeks earlier
i.e. the changes are completely reversible (Fig. 2).

52

BRANDT

If, however, after articular cartilage atrophy has been induced by immobilization, an attempt is made to accelerate recovery by loading of the limb during remobilization, e.g. by a few weeks of low-intensity treadmill running, reversal of the atrophic changes in the cartilage is prevented (Palmoski & Brandt 1981). Although proteoglycan synthesis resumes at a greatly augmented level, as described above, the newly synthesized proteoglycans are not e¡ectively laid down in the cartilage matrix but are lost into the joint space, and the depletion of matrix proteoglycans appears to persist inde¢nitely. Notably, this ‘post-rehabilitation arthropathy’ occurs with levels of treadmill activity so low (e.g. 4 miles/h, 30 min/d; 5/d per week) that they produce no discernable changes in the normal canine joint. Thus, after a period of disuse, such as may occur in some subjects with painful hip or knee OA, atrophic changes develop in the articular cartilage, bone and periarticular muscles. The chondrocytes, lying in their lacunae within the cartilage matrix, no longer have a proteoglycan cushion to protect them from the stresses of normal load bearing. We have shown that the cartilage atrophy that follows immobilization of the knee is due chie£y to the lack of normal contraction of the periarticular muscles (hamstrings, quadriceps) in the stance phase of gait that results in loading of the knee joint, rather than to a lack of £exion/extension of the joint (Palmoski et al 1980). When knees of rabbits were immobilized in an orthopaedic cast, stimulation of the quadriceps by an electrode inserted into the muscle through the cast prevented the atrophy of articular cartilage and subchondral bone that developed in the absence of muscle stimulation (Burr et al 1984). How do the experimental data presented above relate to humans with OA? Many patients with knee OA exhibit atrophy of the muscles surrounding the involved joint and, in particular, the quadriceps muscle. It has generally been considered that this represents the consequences of disuse of the arthritic extremity as the patient avoids bearing load on the painful joint to minimize discomfort. OA pain is classically worse with activity and relieved by rest. Several studies (see Slemenda et al 1997) have shown that the strength of a maximum voluntary contraction is some 30^40% lower in patients with knee OA than that in age- and sex-matched controls. It is important to recognize, therefore, that the quadriceps weakness in patients with OA is reversible through strength-training
even among the elderly. The statement: ‘Well, the 75 year old patient with OA isn’t going to be compliant with an exercise program, even if it should work,’ is nonsense. Adherence to therapeutic exercise programs by patients with OA is no poorer than their adherence to

NEUROMUSCULAR ASPECTS

53

prescribed dosing of non-steroidal anti-in£ammatory drugs (NSAIDs), and may be better than the latter. More to the point are the questions: ‘Why should the patient with OA exercise?’ ‘What good will it do?’ The answers are important: quadriceps strengthening exercises may result in marked improvement in knee pain and function among patients with knee OA. In some cases, that improvement may be as great as or greater than that achieved with NSAIDs. Furthermore, with respect to knee OA, in particular, quadriceps weakness may not only be the result of joint pain, it plays an important role in pathogenesis of structural damage in the disease, i.e. it may be a risk factor for OA. Johansson et al (1991) and others have demonstrated the importance of the quadriceps muscle as a stabilizer of the knee joint. Quadriceps contraction reduces anteroposterior displacement of the tibia on the femur. In further support of the aetiopathogenetic importance of quadriceps weakness in knee OA. In a three year study of an elderly community-based cohort from central Indiana (Slemenda et al 1997), found that incident radiographic knee OA was associated with a decrease in the baseline level of peak knee extensor torque. Notably, the quadriceps weakness in these subjects was not the result of disuse atrophy. Indeed, because of obesity, lower extremity muscle mass in these subjects was increased, relative to the controls, and weakness was seen even among individuals who denied having knee pain and considered themselves to be as active as, or more active than, their peers. In the absence of OA or knee pain, obesity is associated with an increase in lower extremity muscle mass. The obese subject needs more muscle to support the adiposity. The data suggest that, despite the increase in muscle bulk, for reasons that are unclear, the quadriceps muscle may be functionally impaired in subjects with knee OA. Radin et al (1986) have shown that about 30% of individuals exhibit a gait pattern that results in a generation of heel strike transient at the knee (Je¡erson et al 1990). Because the quadriceps muscle is the main anti-gravity muscle in the lower extremity and serves as a brake to decelerate the leg during gait, thereby minimizing the impact of heel strike, quadriceps weakness may lead to excessive impulsive loading of the knee with gait and may predispose to joint pain and to OA (Radin et al 1991). The fact that a heel strike transient can be produced by the injection of local anaesthetic around the femoral nerve of subjects who do not normally generate a heel strike transient (Radin et al 1986) provides biomechanical support for that argument. Furthermore, subjects who generate a heel strike transient can be trained through biofeedback to alter their gait, so as to eliminate impulsive loading of the knee (E. L. Radin, personal communication). Whether that will protect against development of knee pain or OA remains to be shown. The neurophysiological mechanism underlying generation of a heel strike transient during gait in some individuals may relate to di¡erences among us with

54

BRANDT

FIG. 3. The ‘vicious circle of events’ that may lead to OA. Reproduced with permission from Stokes & Young (1984).

respect to central pattern generators (CPGs) in the spinal cord, as discussed by Vilensky (2003). In any event, it is clear that periarticular muscle is a major shock-absorber for joints. Radin et al (1984) showed that repetitive impulsive loading of the hind limb of rabbits led to rapid breakdown of articular cartilage and bone in the ipsilateral knee when the load was delivered rapidly (50 ms from onset to peak) even if the magnitude of load was not excessive. In contrast, loads of comparable or even greater magnitude did not result in joint damage if the rate of loading was ramped up more gradually (onset to peak¼500 ms), permitting time for the periarticular muscles to absorb the load by eccentric contraction. Notably, Mikesky et al (2000) found that the rate of loading of the knee was much slower among strength-trained normal females (e.g. weight-lifters, rowers) than that among sedentary or aerobically trained women. Longitudinal follow-up of these subjects may ascertain whether the former group is protected from development of knee OA. Figure 3 illustrates the relationships between immobilization, muscle weakness and joint damage, as discussed above. In addition, it focuses on another aspect of the pathophysiology of knee OA
re£ex inhibition of the quadriceps muscle (arthogenous muscle inhibition)
which may provide an additional mechanism for weakness of the muscle surrounding the arthritic joint (Stokes & Young 1984). E¡ects of therapeutic exercise in patients with knee OA Several mechanisms may explain the salutory e¡ects of exercise on pain and function in patients with knee OA. Exercise may result in an increase in

NEUROMUSCULAR ASPECTS

55

endurance, improvement in proprioceptive acuity, and a decrease in the severity of arthogenous muscle inhibition. In addition, it may result in improvement in comorbidity (e.g. obesity, cardiovascular disease) and psychological status may facilitate weight loss in the obese subject. It is of interest, therefore, that Sharma et al (2003) have suggested recently that quadriceps strength may be detrimental to subjects with knee OA who have signi¢cant varus^valgus malalignment or ligamentous laxity and that in the presence of these local abnormalities the exercise program for the patient with OA should be individualized to reduce the risk of accelerating joint damage. More work is needed in this area. The bene¢ts of exercise in patients with OA, furthermore, may extend beyond the obvious improvements in joint pain and function. Analyses of the data from the FAST trial illustrate this point (Ettinger et al 1997). In this 18 month randomized placebo-controlled trial that compared the e¡ects of aerobic exercise and strengthening exercise in subjects with knee OA, only modest improvements in knee strength and aerobic ¢tness were achieved. Only minimal di¡erences in outcomes were seen upon comparison of the two exercise treatment groups, probably due to the relatively modest levels of exercise employed by the investigators in an attempt to minimize the drop-out rate. However, the results of the FAST trial were interesting with respect to the e¡ects of exercise on depression (Penninx et al 2002) and to loss of activities of daily living (ADL) (Penninx et al 2001). Among subjects who were not depressed at the onset of the study and exhibited a high level of compliance with the exercise protocol, none of those in the aerobic exercise group but 26% of those in the strength-training group and 31% of the controls developed evidence of depression during the trial. Furthermore, the risk of developing loss of ADL was only about 40% as great among subjects who were highly compliant with the exercise regimen as among those with poorer compliance. Neuropathic arthropathy It has been recognized for many years that impairment of sensory input for an extremity due to an underlying neurological disorder (e.g. diabetes mellitus, tabes dorsalis, syringomyelia) may result in rapid, severe breakdown of the joints in that extremity. The mechanism most widely implicated in this phenomenon is the accumulation of repeated microtrauma associated with usage of the joint, due to impairment of proprioception or nociception, or both. However, a wide variety of ablative neurosurgical procedures have failed to consistently produce neuropathic arthropathy experimentally. Some insight into this apparent inconsistency is provided by our studies in the canine cruciate-de¢ciency model of OA. After transection of the anterior cruciate

56

BRANDT

ligament (ACL) in the neurologically intact dog, end-stage OA develops gradually, so that full-thickness ulceration of the articular cartilage is not seen sooner than four years after the creation of joint instability (Brandt et al 1991). On the other hand, if sensory input from the ipsilateral extremity is markedly reduced (e.g. by L4-S1 dorsal root ganglionectomy or transection of the articular nerves supplying the knee), advanced changes of OA, with full-thickness ulceration of the cartilage, may occur within only a few weeks (O’Connor et al 1992). In the normal dog with a stable knee, the neurosurgical procedure alone does not lead to arthropathy even after follow-up periods as long as 2 years. Gait studies have shown that the combination of dorsal root ganglionectomy and ACLT results in a marked alteration in knee kinematics, with hyperextension at touchdown, accounting for the rapid acceleration of joint destruction (Vilensky et al 1997). A human correlate emphasizes the clinical relevance of the above ¢ndings: among patients with severe diabetic neuropathy, neuropathic joint disease, usually in the foot, develops in about 10^12% of cases. Why not more frequently? We have seen a series of patients with diabetic neuropathy who had no evidence of joint disease until they sustained relatively minor trauma (e.g. a simple ankle sprain), following which neuropathic arthopathy developed acutely and the involved joint broke down within a few weeks (Slowman et al 1990). The analogy to the results of the canine experiments described above is striking: in the experimental model, dea¡erentation of the hind limb was achieved through an extensive neurosurgical procedure, whereas our patients presented with diabetic neuropathy. In the canine model, joint instability was established by ACL transection, whereas the diabetic patients developed an unstable joint after relatively minor trauma. In both cases, however, the result was the same
in the presence of an underlying neurosensory defect, joint trauma was followed by rapid breakdown of the joint. Notably, in the dog, if ACL is performed prior to interruption of sensory input from the limb, the development of knee OA is not accelerated. This suggests that the neurologically intact dog is able to modify its CPGs to minimize the changes in loading of the knee induced by ACL transection. In contrast, when the neurosurgical procedure precedes ACL transection, the central nervous system (CNS) is unable to recognize the acute changes in loading and joint mechanics resulting from ACL de¢ciency and is, therefore, unable to accommodate to these alterations by modifying the pattern of gait (Vilensky 2003). Proprioception Awareness by the CNS of the position of the joint within space generates muscle activity that results in the control of motion, stability of the joint and protection of

NEUROMUSCULAR ASPECTS

57

the joints from injury. Muscle is not only an e¡ector organ, essential for movement of a joint, it is an important sensory organ, containing specialized sensory nerve endings, muscle spindles (whose a¡erent ¢bres transmit proprioceptive impulses through the dorsal root ganglion into the spinal cord where re£exes are established in which the e¡erent arm is a motor nerve to periarticular muscle. Several points may be made with respect to proprioception in OA: receptors in muscle, tendons, joint capsule, ligaments, horns of the menisci and skin provide proprioceptive input that contributes to protective muscular re£exes and to voluntary muscle activity. Proprioceptive activity diminishes with age. Furthermore, in subjects with unilateral knee OA, proprioception is less accurate
even in the apparently normal contralateral knee
than in nonarthritic control subjects, raising the possibility that a subclinical neurological defect is of aetiological importance in some subjects with idiopathic knee OA
a view supported by the above discussion (Vilensky et al 1997). Proprioceptive impairment, furthermore, may be associated with impaired physical function. It is notable, therefore, that proprioception may be improved with elastic bandages, knee sleeves, orthoses or exercise. However, the impact of the improvement in proprioception on joint pain, function, or progression of structural damage in OA is unclear. References Brandt KD, Braunstein EM, Visco DM, O’Connor B, Heck D, Albrecht M 1991 Anterior (cranial) cruciate ligament transection in the dog: a bona ¢de model of osteoarthritis, not merely of cartilage injury and repair. J Rheumatol 18:436^446 Burr DB, Frederickson RG, Pavlinch C, Sickles M, Burkart S 1984 Intracast muscle stimulation prevents bone and cartilage deterioration in cast-immobilized rabbits. Clin Orthop Rel Res 189:264^278 Ettinger WH, Burns R, Messier SP et al 1997 A randomized trial comparing aerobic exercise and resistance exercise with a health education program in older adults with knee osteoarthritis. The Fitness Arthritis and Seniors Trial (FAST). J Am Med Assoc 277:25^31 Je¡erson RJ, Collins JJ, Whittle MW, Radin EL, O’Connor JJ 1990 The role of the quadriceps in controlling impulsive forces around heelstrike. Proc Inst Mech Eng 204:21^28 Johansson H, Sjolander P, Sojka P 1991 A sensory role for the cruciate ligaments. Clin Orthop 268:161^178 Mikesky AE, Meyer A, Thompson KL 2000 Relationship between quadriceps strength and rate of loading during gait in women. J Orthop Res 18:171^175 O’Connor BL, Visco DM, Brandt KD, Myers SL, Kalasinski L 1992 Neurogenic acceleration of osteoarthritis: the e¡ects of prior articular nerve neurectomy on the development of osteoarthritis after anterior cruciate ligament transection in dogs. J Bone Joint Surg Am 74:367^376 Palmoski MJ, Brandt KD 1981 Running inhibits the reversal of atrophic changes in canine knee cartilage after removal of a leg cast. Arthritis Rheum 24:1329^1337 Palmoski MJ, Perricone E, Brandt KD 1979 Development and reversal of a proteoglycan aggregation defect in normal canine knee cartilage after immobilization. Arthritis Rheum 22:508^517

58

DISCUSSION

Palmoski MJ, Colyer RA, Brandt KD 1980 Joint motion in the absence of normal loading does not maintain normal articular cartilage. Arthritis Rheum 23:325^334 Penninx BW, Messier SP, Rejeski WJ et al 2001 Physical exercise and the prevention of disability in activities of daily living in older persons with osteoarthritis. Arch Intern Med 161:2309^ 2316 Penninx BW, Rejeski WJ, Pandya J et al 2002 Exercise and depressive symptoms: a comparison of aerobic and resistance exercise e¡ects on emotional and physical function in older persons with high and low depressive symptomatology. J Gerontol B Psychol Sci Soc Sci 57:P124^ P132 Radin EL, Martin RB, Burr DB, Caterson B, Boyd RD, Goodwin C 1984 E¡ects of mechanical loading on the tissues of the rabbit knee. J Orthop Res 2:221^234 Radin EL, Whittle MW, Yang KH et al 1986 The heelstrike transient, its relationship with the angular velocity of the shank, and the e¡ects of quadriceps paralysis. In: Lantz SA, King AI (eds) Advances in bioengineering. American Society of Mechanical Engineering, New York, p 121^123 Sharma L, Dunlop DD, Cahue S, Song J, Hayes KW 2003 Quadriceps strength and osteoarthritis progression in malaligned and lax knees. Ann Intern Med 138:613^619 Slemenda C, Brandt KD, Heilman DK et al 1997 Quadriceps weakness and osteoarthritis of the knee. Ann Intern Med127:97^104 Slowman-Kovacs S, Braunstein EM, Brandt KD 1990 Rapidly progressive Charcot arthropathy following minor joint trauma in patients with diabetic neuropathy. Arthritis Rheum 33:412^ 417 Stenstrom CH, Minor MA 2003 Evidence for the bene¢t of aerobic and strengthening exercise in rheumatoid arthritis. Arthritis Rheum 49:428^434 Stokes M, Young A 1984 The contribution of re£ex inhibition to arthrogenous muscle weakness. Clin Sci (Lond) 67:7^14 Vilensky J 2003 Neuromuscular System. Innervation of the joint and its role in osteoarthritis. In: Brandt KD, Doherty M, Lohmander LS (eds) Textbook on osteoarthritis, 2nd edn. Oxford University Press, p 161^167 Vilensky JA, O’Connor BL, Brandt KD, Dunn EA, Rogers PI 1997 Serial kinematic analysis of the canine hind limb joints after dea¡erentation and anterior cruciate ligament transection. Osteoarthritis Cartilage 5:173^182

DISCUSSION Pisetsky: In your work on muscle strength, there was a comparison of males and females. I think muscle strength in females showed a relationship with development of OA, but the males did not. Can you comment on that? Brandt: No, it was an observation. The numbers were too small to say more about this. Fernihough: I believe Eric Radin was asked that question at a conference a long time ago and he said it was because women wear silly shoes! Woodworth: Was body mass index (BMI) included, to determine whether this recognized risk factor for OA progression might be an explanation for the e¡ect in women but not men?

NEUROMUSCULAR ASPECTS

59

Brandt: Yes, we adjusted for obesity and also for lean tissue mass (muscle) in the thigh, as determined by DXA. Herzog: I was interested in your two di¡erent categories of people according to how they walk
the diggers and gliders. I noticed that even in the diggers where the impact occurred relatively fast, the total load was one times bodyweight. I know from people who run and contact the ground with their heel ¢rst that the initial impact is much higher. Would this suggest that if you are a runner who strikes the ground with your heel ¢rst, that you are more likely to develop osteoarthritis? Brandt: The data would suggest that. It isn’t so much the magnitude of the load as the rate of loading. The rabbit experiments support this. Herzog: We have just ¢nished a small study looking at loading of joints by maximal muscular stimulation, and also by blunt impact. When you match the loading inside the joint to the pressure that goes across the joint, we ¢nd that the maximum muscular loading at the maximum rate will not produce cell death whatsoever. However, if you reach the same load in 3 ms by impact loading, there is an enormous amount of cell death. These preliminary data would agree with your qualitative argument that the rate of joint loading has a greater in£uence on the adaptive/degenerative response of articular cartilage than the absolute load. However, we should keep in mind that studies in this area are still preliminary, and the precise link between in vivo joint loading, stress^strain ¢elds inside the joint, and the corresponding biological response of joint structures (such as the articular cartilage) is not known at present. Brandt: To come back to bone, rapid impulsive loading may also reactivate the secondary centre of ossi¢cation, and initiate changes at the tide mark. It a¡ects more than only muscle and cartilage. Hunter: The heel strike is very interesting. You made a comment that you could negate the e¡ect of the heel strike by anaesthetizing a femoral nerve. Was that purely by way of inactivating a quadriceps muscle? The quadriceps muscle is presumably responsible for that heel strike impulse. Brandt: Yes, and it attenuates it. Hunter: Are there other ways that you think the heel strike could be attenuated? Brandt: Shoes may make a di¡erence. Radin has designed a study using biofeedback to change diggers into gliders, by placing an accelerometer anchored around the knee that beeps if the knee is loaded too rapidly in gait. Hunter: I understood that their load through mid-stance was greater, not just on the heel strike. Brandt: Yes, but the most striking change is the heel strike transient. John O’Connor estimated this to be some 15 times greater in the knees of diggers than of gliders. This would ¢t the changes seen in the animal models.

60

DISCUSSION

Hunter: For a subject with symptomatic OA, do you think this has analogies from that perspective also? Do these repeat stimuli coming from the foot with heel impact have any role to play in triggering symptoms? Brandt: The study I have just mentioned should answer that question. One thing we can say is that exercise generally has a positive e¡ect on OA symptoms, with improvement in pain and function. Rediske: Could that re£ect more the systemic response to exercise? For example, could exercise trigger endorphin release? Brandt: It is not intensity related. Felson: I wanted to ask about the dark side of strength. You commented on Sharma’s recent work which suggests that in malaligned joints, strong muscle contraction will cause loading in a very localized area. We published data prior to this from the Framingham study showing that hand-grip strength was associated with a marked increase in the development of OA in proximal hand joints. Perhaps strength isn’t so good. You have commented on exercise in general being salutary: I am wondering what the trade-o¡s are here. When is exercise good, and when is strengthening and exercise bad? Brandt: Those are the questions that Leena’s data raise (Sharma et al 2003). We need to segregate this out. Exercise in OA needs to pay attention to ¢tness (cardiovascular condition) as well as to strengthening and also range of motion. Pisetsky: Going back to the diggers and gliders, to what extent is this male versus female? Is it correlated with something else? Brandt: It is not correlated with gender. It is one-thirds diggers, two-thirds gliders. Pisetsky: Is that common among populations? Brandt: As far as I know it hasn’t been studied in other populations. Dieppe: There has been a lot of epidemiological work on the issue of whether doing a lot of exercise is a cause of OA, but there has been precious little on whether exercise is a good or bad thing once you have OA. The trials have been largely short-term and have concentrated on pain relief. From what you are saying today, Ken Brandt, one of the take home messages is that the evidence is beginning to accumulate suggesting that exercise is a bad thing for damaged joints. What is your perspective on this? All my patients ask me about it, and I have been telling them to keep active. Brandt: It is something we need to understand better. It is clear that exercise to improve strength and cardiovascular ¢tness can be performed without exacerbating the symptoms. Whether we are doing something detrimental structurally, depending on the local mechanics and the speci¢cs of the exercise, is unclear. Felson: I am certain that it isn’t so simple. Ken Brandt brought up a nice way of thinking about an una¡ected normal joint, and a vulnerable joint. Those two are

NEUROMUSCULAR ASPECTS

61

likely to be di¡erent entities with respect to the e¡ect of strengthening and exercise. I was hearing echoes of this in our earlier discussion about central sensitization induced by in£ammation which produces a very di¡erent joint in its response to pain than one that would otherwise be normal. The same is probably true here of the e¡ects of strength of muscle contraction, and of exercise. The surprise to me in Dr Sharma’s data was that the neutrally aligned joints weren’t also damaged. In long enough follow-up I think they will be. It is likely therefore that strength provokes damage, ultimately. It is in part because all of the joints she is studying are more or less vulnerable joints. Dieppe: I would go along with that. Leena Sharma’s work is very nice, but it is a very short-term study. Brandt: On the other hand, quadriceps weakness has some analogy to cruciate ligament de¢ciency. The quadriceps is a stabilizer of the knee. It very much comes down to the speci¢cs: what is being loaded and how? Felson: The non-vulnerable, normal joint is the one you study in those women and men whom you follow longitudinally. Then, the e¡ect of strength may be in the deceleration of the limb before heel strike and the distribution of a stabilizing force in dynamic loading, as you said. These are the bene¢cial elements of quadriceps contraction. Then when the joint becomes a vulnerable one, the dark side starts to overwhelm the favourable side of strength for a lot of biomechanical reasons. Rediske: Could the kind of exercise that is contributing to increased quadriceps strength be a contributing factor? Each form of exercise could have a di¡erent impact on the a¡ected joint. Felson: They also have a di¡erent impact at di¡erent angulations of limbs, where limbs have more or less joint space area to load across. Brandt: It also depends on the milieu of the joint. Studies looking at marathon runners to see whether running leads to OA were looking at a select population of people who had proved their ability to run 26 miles. By de¢nition, the people able to do this have a favourable mechanical environment in their joints. People who drop out with soft tissue injuries are protected by their aches and pains. Rediske: In these studies being discussed, what kinds of exercise contributed to increased quad strength? Felson: One of the arguments against strength would be that aerobic exercise makes people feel better and it doesn’t have any measureable e¡ect on strength. This would argue against strength playing a primary role in the salutary bene¢ts of exercise in knee OA. Resistance training also has a salutary e¡ect on symptoms, but the long-term e¡ects on structure aren’t clear. Pisetsky: In the dog models, have you looked at capsules, ligaments and tendons around the joints to see what changes result from immobility?

62

DISCUSSION

Brandt: With the duration of immobility that we employed, few gross changes were seen in these structures. If the immobilization is sustained for a considerably longer period, ¢bro-fatty ankylosis of the joint will occur. It looks a bit like the aftermath of a badly treated Staphylococcus infection. In our experiments considerable muscle atrophy occurs, but it is reversible. Tendons and ligaments will tighten and shorten. Simkin: It would be worth mentioning occupational information. As I understand it, the heavy lifters are at signi¢cantly increased risk. Felson: The data show that the stereotyped occupational activities which tend to cause OA are those which require bending and lifting. Squatting increases loading across the tibiofemoral joint in particular, and lifting has a multiplying e¡ect on this. If you do this for 8 h a day for many years, it more likely than not causes OA. There are lots of occupational studies showing that people who use joints in a stereotyped pattern for many years have a high risk of getting OA in those joints. Herzog: What about people in some Asian countries who squat all the time? Felson: We have recently ¢nished a population prevalence study in Beijing. One of the questions we asked subjects was the duration of squatting, which is frankly shocking there. Instead of standing and waiting for a bus they squat and wait, and they often do their household chores in a squatting position. Especially among women, this is associated with a higher than expected rate of tibiofemoral disease. Interestingly, the Chinese women had more symptomatic knee OA than women in the USA, even though they are substantially thinner than women in the USA. One of the factors accounting for this is the high rate of squatting in China. Fernihough: In the ACL model, do you see any behaviour in the animal that you would report as pain? If there are any weight bearing changes, would you describe this as a pain behaviour? Brandt: Di¡erences may be seen by forceplate analysis, such as those in vertical ground reaction force, a measure widely accepted as a surrogate for joint pain. I am not sure that this is legitimate, because the animal may be reacting to a perception of instability, rather than the pain. The animal uses the limb with the unstable knee quite happily, however. We exercise them to promote the model, and they do this without any obvious discomfort. Lohmander: In humans, pain is not a dominant symptom of the ACL de¢ciency. Instability and insecurity are typically the dominant problems. Grubb: Is that true even in cases where there is profound damage to the knee? Lohmander: Then it is no longer an ACL-de¢cient knee, but an OA knee. Grubb: At what stage does the pain develop? Is there a relationship between the pain and the damage in your model? There isn’t a good correlation in OA. Lohmander: I don’t think we can tell.

NEUROMUSCULAR ASPECTS

63

Brandt: In the dog model, we have shown a fairly good relationship between the severity of chondropathy and the magnitude of vertical ground reaction forces. The more loading, the more cartilage damage. Reference Sharma L, Dunlop DD, Cahue S, Song J, Hayes KW 2003 Quadriceps strength and osteoarthritis progression in malaligned and lax knees. Ann Intern Med 138:613^619

Osteoarthritic Joint Pain: Novartis Foundation Symposium 260. Volume 260 Edited by Derek J. Chadwick and Jamie Goode Copyright ¶ Novartis Foundation 2004. ISBN: 0-470-86763-9

Current perspectives on the clinical presentation of joint pain in human OA Paul Creamer Southmead Hospital, Bristol, BS10 5NB, UK

Abstract. Pain is the commonest symptom of osteoarthritis (OA), the principal reason why individuals seek medical care and a major determinant of other outcomes such as disability and joint replacement. Most studies have examined knee OA: little is known about other sites. Community studies indicate only a modest relationship between structural change on X-ray and reporting of pain. Many community subjects, for example, fail to complain of pain despite extensive X-ray change, while others report pain with normal X-rays. Pain severity of patients attending hospital is even less related to X-ray change, being more dependent on body mass index (BMI), coping strategies and psychosocial variables. Many patients can identify more than one type of pain. It is increasingly clear that OA pain is heterogeneous, being classi¢able on the basis of location, precipitating factors, response to anti-in£ammatory and steroid medication and the e¡ects of local anaesthetic. This potential to classify OA pain represents a useful tool with which to test hypotheses regarding structural origin of pain. 2004 Osteoarthritic joint pain. Wiley, Chichester (Novartis Foundation Symposium 260) p 64^78

Osteoarthritis (OA) is a disorder of synovial joints characterized by destruction of articular cartilage and overgrowth of marginal and subchondral bone. Pain is the principal symptom of OA and the major reason why subjects seek medical attention, which may include costly interventions such as joint replacement. Pain is also the most signi¢cant determinant of disability. Given the current lack of disease modifying drugs in OA, the treatment of OA is essentially the treatment of OA pain. Although OA may a¡ect many peripheral joints (knees, hands, hips, feet) most of our knowledge of pain in OA derives from the knee. It is important to note that mechanisms may vary from joint to joint and data from the knee may not necessarily be transferable to other joints. 64

CLINICAL PRESENTATION OF JOINT PAIN IN HUMAN OA

65

Knee pain in the community Knee pain is usually assessed as a dichotomous variable (present or absent) using, for example, the NHANES-1 screening question: ‘Have you ever had pain in or around the knee on most days for at least one month?’ Subtle changes in the phrasing of the question can result in large di¡erences in the apparent prevalence of pain but in general about 24^28% of community dwellers aged 40^70 respond positively to such a question (O’Reilly et al 1996). Prevalence of knee pain increases with radiographic severity of OA (Felson et al 1987, Hochberg et al 1989, Carman 1989, Spector et al 1993, Lethbridge-Cejku et al 1995). In the NHANES-I study, for example, among subjects aged 65^74 knee pain was reported by 8.8% of subjects with normal X-rays, 20.4% with Kellgren and Lawrence (K+L) grade 1 OA, 36.9% with grade 2 and 60.4% with grades 3^4 (Davis et al 1992). Similar ¢ndings of a progressive increase in the risk of pain reporting with worsening radiographic change have been reported at other joint sites (Table 1). It is, however, clear that there are many subjects in whom X-ray changes and reported pain are discordant. Pain may be reported in the absence of X-ray changes
the prevalence of self-reported knee pain with normal X-rays is about 10.0%. There are several potential reasons for this. First, most studies utilize only supine or weight-bearing views of the tibio-femoral joint; failure to assess the patellofemoral joint could result in a subject being classi¢ed as ‘X-ray negative’ when in fact changes were present but not seen. Indeed, up to 24% of females reporting knee pain have isolated patellofemoral disease and if lateral views are included the predictive value of pain for radiographic change increases (McAlinden et al 1993). Second, a positive response to the NHANES-I knee question does not di¡erentiate between isolated knee pain and widespread pain of which the knee is but a part. The prevalence of ‘widespread chronic pain’ is about

TABLE 1 Prevalence (%) of reported pain by radiographic severity at 1st carpometacarpal (ICMC), distal (DIP) and proximal (PIP) interphalangeal, and hip joints Joint site Radiographic severity (KL grade)

ICMC a

DIP/PIP a

Hipb

0/1 2 3 4

10.6 34.2 65.1

15.2 48.7 80.9

8.0 (M) 12.0 (F) 10.0 (M) 14.0 (F) 44.0 (M) 86.0 (F)

a

Hart et al 1994. Lawrence 1997.

b

66

CREAMER

11% (Croft et al 1993). Such patients may answer a⁄rmatively about knee pain but this would not necessarily imply local pathology. Third, X-rays are relatively insensitive: they may be normal when other diagnostic studies such as arthroscopy show clear evidence of OA (Fife et al 1991). X-rays do not allow visualization of non-bony sources of pain, such as capsule, synovium or ligaments. Finally, not all knee pain is due to OA: causes such as anserine bursitis, internal derangements and referred pain from hip or spine would not be identi¢ed on X-rays of the knees. The second group (X-ray positive, pain negative) is larger. Pain reporting in grade 3^4 OA ranges from 40^79%: thus, up to half the patients in the community with, by any standard, established radiographic OA deny pain. The relationship improves if osteophytes rather than global change are used (Spector et al 1993, Lethbridge-Cejku et al 1995, Cicuttini et al 1996). The precise question that is asked may a¡ect the response in terms of pain reporting. The NHANES-I question may underestimate prevalence: patients may have had pain but not on ‘most days of a month’ or they may simply fail to recall previous episodes of pain. Further, OA may be a phasic condition with episodes of pain separated by remissions: the question may fail to capture the painful episode. Another approach to examining the relationship between structural change and pain is to consider the prevalence of X-ray change in those presenting with joint pain. A community survey of 4057 subjects aged 40^70 found a prevalence of knee pain of 28.3%. Of these, 74% had at least grade 1 osteophyte and 40.9% had at least grade 2 (O’Reilly et al 1996). In 195 subjects aged over 40 presenting to their GP with a ¢rst episode of hip pain, Birrel et al (2000) found 44% had a KL grade 52 and 34% had KL 53. A minimum joint space of 42.5 mm was seen in 30%. By the time subjects present to primary care with hip pain, therefore, a signi¢cant number will already have established OA change on X-ray. The risk factors for radiographic knee OA (age, sex, race, obesity) are di¡erent from those for knee pain reporting in the community. In addition to X-ray change, psychological well-being and health status (Davis et al 1992), anxiety (in women only) (Creamer et al 1999a), feeling ‘low’ or ‘very low’ in spirits (Hochberg et al 1989), hypochondriasis (Lichtenberg et al 1986) and ‘negative a¡ect’ (Dekker 1993) have all been associated with higher levels of knee pain reporting. Lower educational level is an independent risk factor for pain reporting (Hannan et al 1992). OA pain in the clinic Some individuals with knee or hip pain elect to present to medical care. The reasons for this choice are unclear but co-morbidity (especially psychosocial), coping beliefs, social support, availability of services and degree of empowerment are all

CLINICAL PRESENTATION OF JOINT PAIN IN HUMAN OA

67

FIG. 1. Circadian rhythm for pain in patients with OA of the hand. Self measurements/ratings were made by 20 or 21 patients every 24 hours during waking for 10 days. Individual values had trends removed and were converted to a percentage of the mean before combining for group analysis by population mean cosinor. For rhythm characteristics P value is from the zero amplitude test; amplitude ¼ half peak trough di¡erence of cosine; bathyphase ¼ lowest point of cosine (referenced from 0000). P50.001 for each variable from ANOVA for time e¡ect. Reproduced with permission from Bellamy et al (2002).

likely to be more important than pain severity, radiographic change, age or functional limitation. A community study of subjects with hip or knee pain found that depression scores were signi¢cantly higher in those that had elected to seek medical care (Dexter & Brandt 1994). A similar role for psychological factors in the promotion of healthcare seeking behaviour has been suggested in other conditions such as ¢bromyalgia. It is safe to assume that almost all individuals with OA presenting to healthcare will have pain. Although pain is clearly important to patients and is discussed at 98% consultations, potential causes are discussed minimally or not at all in up to 46% cases (Bellamy & Bradley 1996). Furthermore, physicians and patients may disagree about the severity of their pain and e¡ect on life (Hogkins et al 1985). Suarez-Almazor et al (2001) in a study of 105 patients with musculoskeletal disease found that intraclass correlation coe⁄cients (ICCs) were only 0.42 for pain. Physicians tended to rate their patients’ health status higher than the patients themselves and were less willing to gamble on the risk of death versus perfect health. The importance of pain to patients with OA and the relationship

68

CREAMER

between pain severity and its importance has been little studied but clearly has great relevance. For patients presenting to healthcare, pain becomes a continuous variable
pain severity. A major advance in OA pain research has been the adoption of standardized, validated questionnaires such as the WOMAC, Lequesne, McGill Pain Questionnaire (MPQ) or a simple VAS. The WOMAC has recently been shown to be more sensitive than the SF-36 (Davies et al 1999) and the Lequesne Index (Theiler et al 1999) and appears not to be in£uenced by anxiety and depression as much as the MPQ (Creamer et al 1999b). It also allows pain occurring in di¡erent situations to be separately assessed. The risk factors for pain severity reporting are di¡erent from those for pain as a dichotomous variable. In a group of hospital outpatients with knee OA (Creamer et al 1999b) risk factors for pain severity reporting di¡ered slightly according to the scale used though obesity, helplessness and education remained associated with pain severity after adjustment for confounding variables. Age, disease duration and quality of life were not related to severity of pain. Others have reported links between pain severity and psychological factors: Summers et al (1988) reporting on 65 patients with OA of hip or knee found that depression (as measured by the Beck Depression Inventory) and anxiety correlated with some measures of the MPQ. Another study of 61 patients with knee OA found signi¢cant correlations between MPQ and Zung Anxiety and Depression Inventory scores (Sala⁄ et al 1991). In the community chronicity and severity of knee pain were associated with higher psychosocial disability (as measured by subscales of the Sickness Impact Pro¢le) compared to age- and sex-matched controls from the same community (Hopman-Rock et al 1996). A number of studies have shown that, in hospital patients, radiographic change is not related to pain severity (Creamer et al 1999b, Bruyere et al 2002). It may be that a threshold needs to be reached for joints to become painful but beyond that, other factors (coping strategies, depression, co-morbidity, BMI) determine the perceived severity for an individual. The nature of OA pain Generally quoted descriptions of OA pain are largely anecdotal, supported by surprisingly little patient-based evidence. ‘Typical’ OA pain is said to be insidious, variable and intermittent (‘good days and bad days’); mainly occurring on use, movement or weight bearing and later in the day. Nearly all symptomatic patients have use-related pain but many also have rest or night pain. Knee pain is generally anterior or medial; hip pain classically is felt in the groin but may radiate to the knee. Thumb base OA is more likely to cause pain than interphalangeal OA and may be felt di¡usely ‘around the wrist’. A diurnal variation has been described

CLINICAL PRESENTATION OF JOINT PAIN IN HUMAN OA

69

at both the knee (Bellamy et al 1990; see Fig. 1) and the hand (Bellamy et al 2002) with pain worse in the evenings and easier in mornings. The reason for good and bad days is unclear: in£uences of weather or barometric pressure are often cited by patients and may have some validity. Strusberg et al (2002) found that in OA, pain correlated with low temperature (r ¼
0.23, P50.001) and high humidity (r ¼0.24, P50.001). Seasonal variation (worse in winter) is often reported but this may be more due to perception than reality since reported symptoms do not necessarily agree with measured clinical scores (Hawley et al 2001). Pain may also be reported more strongly at weekends (Bellamy et al 1990). Although the cause remains uncertain it is increasingly clear that pain in OA is heterogeneous, varying between individuals and with di¡erent phases of the disease. Recently e¡orts have been made to identify di¡erent patterns of pain, in the hope that they may indicate di¡erent pathological or anatomical processes. The location of pain at the knee, for example, is not random, but falls into two well de¢ned groups: generalized anterior pain and localized inferomedial pain. These di¡erences are not explicable by radiographic change and may represent local bony or soft tissue sources (Creamer et al 1998a). Another example is the response to local anaesthetic (Creamer et al 1996, Hassan et al 2002): in many patients this will abolish pain temporarily whilst in others no e¡ect is seen. In simple terms, some patients may have local sources of pain whilst in others the pain is centrally driven. The complexity of pain mechanisms is further emphasised by the fact that intra-articular anaesthetic can also abolish pain in contralateral, untreated joints, implying central or spinal mechanisms (Creamer et al 1996). Finally, the e¡ect of intra-articular steroids overall is short lived, but individual patients derive sustained bene¢t
do they have a more in£ammatory cause for their pain? Night pain (often used by orthopaedic surgeons as an indicator of the need for joint surgery) is said to be an unusual feature, limited to advanced disease. We found (Creamer et al 1998b) that 43% subjects with knee OA reported pain of 530 mm on a VAS for night pain and 14.7% actually felt the night to be the most painful time. 27.9% felt that resting in bed made their pain worse. Using the latter de¢nition, we were unable to con¢rm a relationship between night pain and disease severity as assessed by pain severity, disability, examination ¢ndings or radiographic change. A modest relationship with disease duration was seen, but most signi¢cantly, night pain was associated with high levels of helplessness and worse perceived quality of life, perhaps due to underlying fatigue. Such clinical observations allow testable hypotheses to be generated. Night pain, for example is often thought to be due to raised intraosseous pressure: the ability of MRI to detect focal changes in subchondral bone linked with pain (Felson et al 2001) allows this to be investigated further. If inferomedial knee

70

CREAMER

pain is due to collateral ligament pathology, again this may be detected by magnetic resonance imaging (MRI). If failure of intra-articular anaesthetic to abolish pain indicates a central source this may be associated with higher depression or helplessness. Such studies have the potential to allow a more tailored, individual approach to pain treatment. Longitudinal studies show that most patients feel that their pain gets worse with time though there is considerable variability. In the Bristol OA 500 study (Dieppe et al 2000), for example, the proportion of subjects with knee OA reporting their pain to be ‘severe’ was 25% at baseline, 17% at 3 years and 27% at 8 years. However, 80% of patients felt they had worsened overall. E¡ect of pain in disease We have considered the risk factors for pain reporting but what about the e¡ect pain may have on the underlying disease? Reduction in pain, for example by intraarticular local anaesthetic, results in increased maximum voluntary contraction (MVC) of quadriceps (Hassan et al 2002). The in£uence of pain on other potential risk factors such as proprioception and balance is unclear: Hassan et al (2002) reported that pain reduction did not result in improvements in proprioception or static postural stability. Jadelis et al (2001), examined dynamic balance in a cross sectional study of older patients with knee OA. Balance was most strongly related to quadriceps strength, but in those subjects with weak quadriceps pain severity became an independent predictor of poor balance. We do not know if long term pain reduction can reduce progression of disease but there is some evidence that pain predicts incident knee OA and that subjects with pain progress faster than those with similar radiographic change without pain. Hart (Hart et al 1999) found odds ratios of 1.91 (95% con¢dence interval 1.18^3.09) for knee pain predicting development of osteophyte at follow up. Cooper et al (2000) in a follow up study of 354 community subjects found that baseline knee pain predicted incident knee OA at 5 years (odds ratio 2.9 [1.2^6.7] for KL 51; odds ratio 1.3 [0.6^2.7] for KL 52). Knee pain also predicted progression over 5 years. Causes of pain in OA The anatomic cause of pain in OA remains unknown. Any theory has to consider that the principal structure involved (cartilage) possesses few pain-sensitive ¢bres. Bone pain may be a factor in many subjects: perhaps via osteophyte growth with stretching of periosteum, raised intraosseous pressure or microfractures. Felson et al (2001) examined the relationship between ‘bone marrow lesions’ (thought to represent oedema) on MRI and knee pain. Lesions were found in 77.5% persons

CLINICAL PRESENTATION OF JOINT PAIN IN HUMAN OA

71

with painful knees compared with 30% with no knee pain (P50.001). ‘Large’ lesions were present almost exclusively in persons with knee pain (35.9% vs. 2%; P50.001). Although lesions were associated with more severe radiographic change in general, the relation with pain persisted even after adjustment for severity of radiographic disease, e¡usion, age and sex. No relation was seen with pain severity. Other sources of pain include ligament damage, capsular tension, meniscal injury and synovitis. In£ammation may be present in OA and may cause pain either by direct stimulation of primary a¡erent peripheral a¡erent nociceptive ¢bres (PANs) or by sensitizing PANs to mechanical or other stimuli. Systemic markers of in£ammation such as C reactive protein (CRP) are raised in many patients with OA and may predict future progression of disease (Spector et al 1997). In addition there is a central component to pain and in£uences such as anxiety, depression and comorbidity are likely to operate in some patients as described above. Conclusions Many questions remain about OA pain. What makes a person with OA pain seek medical attention? Is it worsening of the disease (little evidence for this)? Loss of coping skills? Socioeconomic or ¢nancial factors? What would be the e¡ect of early aggressive pain control in reducing intensity or duration of chronic pain? In other words, does control of pain a¡ect the natural history of the disease? To what extent is pain protective and to what extent does it reduce function and result in physical deconditioning? How can we improve our understanding of our patients’ health perceptions and risk-bene¢t preferences so that we may suggest more appropriate interventions? Many individuals with radiographic OA do not report pain and perhaps we should ask not ‘why is OA painful?’ but ‘why is it so often pain free?’ Much e¡ort is being expended on ¢nding drugs capable of modifying the disease process, notably on cartilage loss. We would expect an e¡ective disease-modifying drug to also have an e¡ect on pain but, given the poor correlation currently seen between structural change (at least on X-ray) and symptoms, a word of caution might reasonably be sounded. There are grounds to at least consider the wisdom of investing large resources in expensive technologies designed to reduce structural change when this may not, in fact, a¡ect the problems that are important for the patient. References Bellamy N, Bradley L 1996 Workshop on chronic pain, pain control and patient outcomes in rheumatoid arthritis and osteoarthritis. Arthritis Rheum 39:357^362

72

CREAMER

Bellamy N, Sothern R, Campbell J 1990 Rhythmic variations in pain perception in osteoarthritis of the knee. J Rheumatol 17:364^372 Bellamy N, Sothern R, Campbell J, Buchanan WW 2002 Rhythmic variations in pain, sti¡ness and manual dexterity in hand osteoarthritis. Ann Rheum Dis 61:1075^1080 Birrell F, Croft P, Cooper C, Hosie G, Macfarlane GJ, Silman A 2000 Radiographic change is common in new presenters in primary care with hip pain. PCR Hip Study Group. Rheumatology (Oxford) 39:772^775 Bruyere O, Honore A, Rovati LC et al 2002 Radiologic features poorly predict clinical outcomes in knee osteoarthritis. Scand J Rheumatol 31:13^16 Carman WJ 1989 Factors associated with pain and osteoarthritis in the Tecumseh Community Health Study. Semin Arthritis Rheum 18:S10^S13 Cicuttini FM, Baker J, Hart DJ, Spector TD 1996 Association of pain with radiological changes in di¡erent compartments and views of the knee joint. Osteoarthritis Cartilage 4:143^147 Cooper C, Snow S, McAlindon TE et al 2000 Risk factors for the incidence and progression of radiographic knee osteoarthritis. Arthritis Rheum 43:995^1000 Creamer P, Hunt M, Dieppe P 1996 Pain mechanisms in osteoarthritis of the knee: e¡ect of intraarticular anaesthetic. J Rheumatol 23:1031^1036 Creamer P, Lethbridge-Cejku M, Hochberg MC 1998a Where does it hurt? Pain localization in osteoarthritis of the knee. Osteoarthritis Cartilage 6:318^323 Creamer P, Lethbridge-Cejku M, Hochberg M 1998b What is the signi¢cance of night pain in knee osteoarthritis? Br J Rheumatol 37:S153 Creamer P, Lethbridge-Cejku M, Costa P, Tobin J, Herbst JH, Hochberg MC 1999a The relationship of anxiety and depression with self reported knee pain in the community: data from the Baltimore Longitudinal Study of Aging. Arthritis Care Res 12:3^7 Creamer P, Lethbridge-Cejku M, Hochberg MC 1999b Determinants of pain severity in knee osteoarthritis: e¡ect of demographic and psychosocial variables using 3 pain measures. J Rheumatol 26:1785^1792 Croft P, Rigby AS, Boswell R, Schollum J, Silman A 1993 The prevalence of chronic widespread pain in the general population. J Rheumatol 20:710^173 Davies GM, Watson DJ, Bellamy N 1999 Comparison of the responsiveness and relative e¡ect size of the Western Ontario and McMaster Universities Osteoarthritis Index and the shortform Medical Outcomes Study Survey in a randomised, clinical trial of osteoarthritis patients. Arthritis Care Res 12:172^179 Davis M, Ettinger W, Neuhaus JM, Barclay JD, Degal MR 1992 Correlates of knee pain among US adults with and without radiographic knee osteoarthritis. J Rheumatol 19:1943^1949 Dekker J, Tola P, Aufdemkampe G, Winckers M 1993 Negative a¡ect, pain and disability in osteoarthritis patients: the mediating role of muscle weakness. Behav Res Ther 31:203^206 Dexter P, Brandt K 1994 Distribution and predictors of depressive symptoms in osteoarthritis. J Rheumatol 21:279^286 Dieppe P, Cushnaghan J, Tucker M, Browning S, Shepstone L 2000 The Bristol ‘OA 500’ study: progression and impact of the disease after 8 years. Osteoarthritis Cartilage 8:63^68 Felson DT, Naimark A, Anderson J, Kazis L, Castelli W, Meenan RF 1987 The prevalence of knee osteoarthritis in the elderly: the Framingham Osteoarthritis Study. Arthritis Rheum 30:914^918 Felson DT, Chaisson CE, Hill CL et al 2001 The association of bone marrow lesions with pain in knee osteoarthritis. Ann Internal Med 134:541^549 Fife RS, Brandt KD, Braunstein EM et al 1991 Relationship between arthroscopic evidence of cartilage damage and radiographic evidence of joint space narrowing in early osteoarthritis of the knee. Arthritis Rheum 34:377^382

CLINICAL PRESENTATION OF JOINT PAIN IN HUMAN OA

73

Hannon MT, Anderson JJ, Pincus T, Felson DT 1992 Educational attainment and osteoarthritis: di¡erential associations with radiographic changes and symptom reporting. J Clin Epidemiol 45:139^147 Hassan BS, Doherty SA, Mockett S, Doherty M 2002 E¡ect of pain reduction on postural sway, proprioception, and quadriceps strength in subjects with knee osteoarthritis. Ann Rheum Dis 61:422^428 Hart DJ, Spector TD, Egger P, Coggon D, Cooper C 1994 De¢ning osteoarthritis of the hand for epidemiological studies: the Chingford Study. Ann Rheum Dis 53: 220^223 Hart DJ, Doyle DV, Spector TD 1999 Incidence and risk factors for radiographic knee osteoarthritis in middle-aged women: the Chingford Study. Arthritis Rheum 42:17^24 Hawley DJ, Wolfe F, Lue FA, Moldofsky H 2001 Seasonal symptom severity in patients with rheumatic diseases: a study of 1,424 patients. J Rheumatol 28:1900^1909 Hochberg MC 1996 Prognosis of osteoarthritis. Ann Rheum Dis 55:685^688 Hochberg MC, Lawrence RC, Everett DF, Cornoni-Huntley J 1989 Epidemiological associations of pain in osteoarthritis of the knee: data from the National Health and Nutrition Examination Survey and the National Health and Nutrition Examination-I Epidemiologic Follow-up Survey. Semin Arthritis Rheum 18:S4^S9 Hodgkins M, Albert D, Daltroy L 1985 Comparing patients’ and their physicians’ assessments of pain. Pain 23:273^277 Hopman-Rock M, Odding E, Hofman A, Kraaimaat FW, Bijlsma JW 1996 Physical and psychosocial disability in elderly subjects in relation to pain in the hip and/or knee. J Rheumatol 23:1037^1044 Jadelis K, Miller ME, Ettinger WH Jr, Messier SP 2001 Strength, balance, and the modifying e¡ects of obesity and knee pain: results from the Observational Arthritis Study in Seniors (OASIS). J Am Geriatr Soc 49:884^891 Lawrence JS 1977 Rheumatism in populations. William Heinemann, London Lethbridge-Cejku M, Scott WW Jr, Reichle R et al 1995 Association of radiographic features of osteoarthritis of the knee with knee pain: data from the Baltimore Longitudinal Study of Aging. Arthritis Care Res 8:182^188 Lichtenberg PA, Swensen CH, Skehan MW 1986 Further investigation of the role of personality, lifestyle and arthritic severity in predicting pain. J Psychosom Res 30:327^337 McAlindon TE, Cooper C, Kirwan JR, Dieppe PA 1993 Determinants of disability in osteoarthritis of the knee. Ann Rheum Dis 52:258^262 O’Reilly SC, Muir KR, Doherty M 1996 Screening for pain in knee osteoarthritis: which question? Ann Rheum Dis 55:931^933 Sala⁄ F, Cavalieri F, Nolli M, Ferraccioli G 1991 Analysis of disability in knee osteoarthritis. Relationship with age and psychological variables but not with radiographic score. J Rheumatol 18:1581^1586 Spector TD, Hart DJ, Byrne J, Harris PA, Dacre JE, Doyle DV 1993 De¢nition of osteoarthritis for epidemiological studies. Ann Rheum Dis 52:790^794 Spector TD, Hart DJ, Nandra D et al 1997 Low-level increases in serum C-reactive protein are present in early osteoarthritis of the knee and predict progressive disease. Arthritis Rheum 40:723^727 Strusberg I, Mendelberg RC, Serra HA, Strusberg AM 2002 In£uence of weather conditions on rheumatic pain. J Rheumatol 29:335^338 Suarez-Almazor ME, Conner-Spady B, Kendall CJ, Russell AS, Skeith K 2001 Lack of congruence in the ratings of patients’ health status by patients and their physicians. Med Decis Making 21:113^121 Summers MN, Haley WE, Reveille JD, Alarcon GS 1988 Radiographic assessment and psychological variables as predictors of pain and functional impairment in osteoarthritis of the knee or hip. Arthritis Rheum 31:204^209

74

DISCUSSION

Theiler R, Sangha O, Schaeren S et al 1999 Superior responsiveness of the pain and function sections of the Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) as compared to the Lequesne-Algofunctional Index in patients with osteoarthritis of the lower extremities. Osteoarthritis Cartilage 7:515^519

DISCUSSION Bradley: I want to comment on your data on anxiety. The role of anxiety in pain reporting is greatly underestimated. With regard to the McGill Pain Questionnaire, one reason why you ¢nd low correlations between the WOMAC and the visual analogue scale from the McGill is that, unlike the WOMAC pain scale, the McGill is multidimensional and contains a large subgroup of words dealing with a¡ect and emotion. When you look at ethnic group di¡erences on the McGill, do you ¢nd variation as a function of the types of words chosen to describe pain? My reason for asking this is that when we apply quanti¢ed stimuli in the laboratory to patients, we ¢nd that African^Americans tend to use higher intensity a¡ective words compared with Caucasians, even if their sensory intensity responses are the same. Do you ¢nd similar phenomena in your larger population studies? Creamer: There are di¡erences between English English and American English. In the more detailed study we didn’t have enough African Americans to really address this. They did report higher McGill pain scores, but most of this e¡ect disappeared when we adjusted for BMI. There was a sense that the words chosen might have been di¡erent. And with the McGill we have a sense that it isn’t really measuring what I want it to measure. On the McGill I also looked at whether there were words people would choose given the opportunity that aren’t on the McGill. There aren’t many, because the McGill has 76 words, but there were a few. There are some words that are never chosen by people. McGill was developed for all sorts of pain, including cancer and dental pain. It may not be the best tool to look at some of these issues in OA. Schaible: What is the minimum set of symptoms that you need to diagnose OA? In e¡ect, one could ask if someone reports pain but you don’t ¢nd anything in the joint, why do we call it OA pain? And if someone has joint changes visible by X-ray but the pain doesn’t correlate, is this OA? Creamer: It depends on how you de¢ne OA. OA can be de¢ned pathologically, radiographically or clinically. There is some overlap, but there are also some di¡erences. In the community, pain is de¢ned as ‘yes’ or ‘no’. Patients are asked a question about whether they have pain, and we try to de¢ne this on the basis of it being experienced on most days for at least a month. When we are looking at pain severity then it is much more arbitrary, but sometimes people try to develop

CLINICAL PRESENTATION OF JOINT PAIN IN HUMAN OA

75

cut-o¡s for studies to ensure that they have people in the study with adequate pain, so a bene¢t can be shown for whatever intervention is used. Schaible: For a clinician, if someone reports joint pain, is this su⁄cient to make the diagnosis of OA in the absence of any visible in£ammation? Creamer: Not all knee pain is due to OA. One of the explanations for the kneepain-positive people who are X-ray negative is that they have another pathology in the knee joint. The other thing is that X-rays are relatively insensitive. Arthroscopy may well reveal cartilage changes long before the X-ray has changed. Some people propose that we should talk less about knee OA and more about knee pain. Dieppe: I think the sub-setting of pain is a very important concept. I think we are not getting this sorted because we don’t know what hypotheses to test and we don’t know what questions to ask. We have got into this habit of thinking about night pain, rest pain and walking pain as being entities. I think this is based on nothing. Similarly, the other methods of subtyping that are being attempted are potentially unhelpful. I have been rather impressed by what the social scientists can o¡er ¢elds like this. We should be doing qualitative research before we do quantitative research. We should be doing in-depth, unstructured interviews with people with OA, however de¢ned, trying to take out themes from this sort of qualitative research as to what the issues are, and then derive the hypotheses and do the sorts of studies you have done having ¢rst got some hypotheses about pain subgroups. Creamer: I agree with you. Qualitative research is ¢endishly di⁄cult, so it is much easier to go for the tools that have already been developed. But at least this sort of work shows that potentially there are di¡erences. Dieppe: Yes, your sort of work stimulates me to think we really should go for this. Kuettner: We are mixing the di¡erent forms of OA. Aren’t knee and hip OA totally di¡erent in their aetiology, and don’t they require di¡erent clinical approaches? Felson: There is no clear cut answer to this. Pisetsky: There is another way you could subset: those patients who have surgery and those who don’t. If you look at the people who have operations, how are they describing their pain as opposed to those who don’t? Do you get any insight by dividing the patients up in this way? Dieppe: We have been looking at those issues. We have been studying the barriers and facilitators to people seeking medical help in the ¢rst place, and we have been trying to take this through to referrals and surgery. A lot of this is being done with qualitative research, so the numbers of people we have information on is small. Where we are so far suggests to me that healthcare utilization for OA has little to do with pain or disease severity, but that other sociocultural factors are determining it. Some of my social scientist colleagues

76

DISCUSSION

go as far as saying that there is no disease here, and that it is all purely a sociocultural phenomenon! Brandt: With respect to the issue of OA pain and progression, there was some nice work from Hurwitz et al (2000) measuring gait and pain in patients with arthritic medial compartment knee OA. They showed that when the patients were taking pain medication they increased the loading of the medial compartment. When the pain medication was washed out and joint pain became more severe, the subjects changed their gait so as to protect the damaged cartilage. However, long-term data are not available to show whether this results in analgesic arthropathy. But this also relates to Leena’s study (Sharma et al 2003), because she didn’t measure joint pain. One of the possibilities that needs to be considered is whether those people who were stronger had less pain and therefore loaded their knee more than others. Felson: Pain is a protective mechanism. I’d like to ask an almost rhetorical question. In RA, it is my understanding that anxiety and depressive symptoms contribute to pain severity also. Yet therapies for RA seem to have terri¢c e¡ects on pain. Does this mean that we can address pain anyway without grappling with this concern? In RA, a third of the patients don’t have morning sti¡ness, for example, so there is the same variability in pain description and reporting that you have described in OA. Yet we don’t seem to have too much trouble in developing therapies for RA while we ignore the qualitative aspects of pain. Will this be true for OA also? Creamer: Do you think that in RA we have a more de¢ned pathology and site of origin of pain? There is synovitis and in£ammation. Felson: Is that where the pain comes from in RA? Creamer: Treatments such as a steroid injection into an in£amed knee are highly e¡ective ways of reducing pain in RA. I have a better feel for the pathology of RA than OA, and in£ammation seems to be what is driving most of the pain in RA. But your point is well made. Grubb: A number of us are interested in the development of animal models for the study of OA, and what worries me is that we have this clear lack of correlation between the radiological scores of the disease and pain. How can we develop a model if we don’t have a clear idea of what typical OA is? What features should we be looking for in an animal model that would well represent human OA? We can’t develop a good animal model of human OA without that correlation. Brandt: It also depends on what you want to use the model for. If you want to use it to study a drug that might inhibit cartilage loss, then you want a model that demonstrates a certain rate of cartilage loss. If you want a model to evaluate pain, this imposes an entirely di¡erent set of requirements. This is challenging. Grubb: That is what many of us here are interested in
the pain aspect. It is not clear to me what we should be doing here.

CLINICAL PRESENTATION OF JOINT PAIN IN HUMAN OA

77

Brandt: There is an obvious di⁄culty in evaluating OA pain. It is, however, no easier to assess structural damage. While there are similarities in pathology, no animal models have been clearly shown to predict the e¡ects of ‘chondroprotective’ drugs in humans. Schaible: This comes down to the question of nociception and pain. A model is urgently required to ¢nd out whether there is any change in nociception, or whether there is nociception at all in degenerative processes in a joint. Then there is a discussion about what this means for pain. I wouldn’t be too negative about this. Grubb: I am not being negative. Complete Freund’s adjuvant (CFA) polyarthritis is a very good animal model with a lot of joint pathology in which changes in nociception are seen. It is not, however, a model of OA. Schaible: I am biased. We should say that there is a de¢ned process in the joint, and we should answer the question about whether this evokes nociception. This is something we could answer and should answer. Henry: The idea of subgroupings raises a lot of issues. The people looking for biomarkers must feel lost as well. The real answer lies in making a stab at developing animal models. When we develop an animal model, what can we learn about the process? From this we might stumble across one model that will be particularly useful in terms of understanding nociception. Even humans don’t have a good model of OA. Some have pain without clinical signs, and some have clinical signs without pain. What are the basic scientists trying to model? It is not as simple as it was a few years ago when we had OA and models. Pisetsky: I have a question about the value of pathology. There are many operative specimens in OA. Are we getting the most information out of them? Given the heterogeneity of the disease, should we be doing more pathology? In the RA world where there is not much surgery any more, when people did pathological studies, di¡erent forms of RA were histologically distinguishable. Not all people are alike, and subsets that were informative could be identi¢ed. Kuettner: If you explant the cartilage from di¡erent animals, you get distinctive responses to di¡erent mediators. It becomes very di⁄cult to say, for example, that the rabbit is a good model for the human disease. Even within the human, the di¡erent cartilages from di¡erent joints respond quite di¡erently. Pisetsky: I am asking, for example, should I be looking at nerve ¢bres in capsules? Lohmander: The problem with surgical specimens in OA is that they represent end-stage disease in most cases, and also that the patients receiving surgery have been ¢ltered through the ¢lters we have heard about here, so they might not be representative. Pisetsky: But, even bearing these limitations in mind, we have the opportunity to get pathological tissue.

78

DISCUSSION

Kuettner: You can also get normal tissue from tissue donor banks. Hunter: I’d like to address the qualitative research issues. You presented some nice data from Nick Bellamy showing huge diurnal changes within subjects. There is also that huge dichotomy between people who have structural OA and those people who are symptomatic. Surely there is room for research into those particular subjects who are asymptomatic with structural changes and those people who have big diurnal changes, to try to explore what is going on there. Do you have any ideas what this may be? Dieppe: I agree with you. Unfortunately qualitative research is di⁄cult, expensive and time-consuming. The only data we have are more to do with accessing healthcare utilization, so I don’t have any useful qualitative data on pain. Mackenzie: You tried to control for pain threshold di¡erences in a structured way. How clear are you that this really re£ects the ability of di¡erent patients to tolerate pain in the real world in very di¡erent ways? Creamer: It can only ever be a surrogate. It is coming from the ¢bromyalgia literature where there has been some work on the pain threshold in general. This was just an attempt to get a bit of a handle on this. Mackenzie: You can perhaps get a handle on this by trying to understand the di¡erences in the way in which people deal with pain. Creamer: Some of the brain imaging studies might be relevant here. Bradley: One thing that comes out from the imaging literature is that we haven’t paid as much attention as we should to the emotional/a¡ective dimension of pain. This a¡ective dimension is very important in the way patients present in the clinic. However, even in the laboratory, psychological factors have a much greater association with pain tolerance tasks as compared to pain threshold tasks. In the imaging world, where people are only just beginning to study pain through neuroimaging, most of the e¡ort is focused on mapping the neural correlates of intensity, and much less attention is paid to the neural correlates of a¡ect. So, neuroimaging of pain a¡ect responses is a very important question. References Hurwitz DE, Ryals AR, Block JA, Sharma L, Schnitzer TJ, Andriacchi TP 2000 Knee pain and joint loading in subjects with osteoarthritis of the knee. J Orthop Res 18:572^579 Sharma L, Dunlop DD, Cahue S, Song J, Hayes KW 2003 Quadriceps strength and osteoarthritis progression in malaligned and lax knees. Ann Intern Med 138:613^619

Osteoarthritic Joint Pain: Novartis Foundation Symposium 260. Volume 260 Edited by Derek J. Chadwick and Jamie Goode Copyright ¶ Novartis Foundation 2004. ISBN: 0-470-86763-9

Joint mechanics in osteoarthritis Walter Herzog, Andrea Clark and David Longino Faculties of Kinesiology, Engineering, and Medicine, University of Calgary, 2500 University Drive NW, Calgary, AB T2N 1N4, Canada

Abstract. The primary goal of our research has been to quantify the in vivo loading of normal and osteoarthritic (OA) joints, and to determine the corresponding biological responses. Much of the research in this area has been performed using articular cartilage explants. We feel that, although critically important to our understanding of cartilage mechanics and biology, these experiments may not be directly transferable to interpreting the in vivo joint mechanics and elucidating the detailed mechanisms of onset and progression of OA. Therefore, we have attempted to measure the loading of the knee in freely moving feline and lapine models of OA. We have found that, upon anterior cruciate ligament transection in the cat, knee joints are more £exed, muscle forces are decreased and muscle control patterns are destroyed. Articular cartilage initially becomes thicker, softer and more permeable, resulting in generally increased joint contact areas and decreased peak pressures in the initial stages of joint degeneration compared to control values. Based on our results, we speculate that unloading of the joint (rather than overloading), combined with poor muscular control and weakness, might constitute risks for the onset of joint degeneration. 2004 Osteoarthritic joint pain. Wiley, Chichester (Novartis Foundation Symposium 260) p 79^99

Biomechanics is the science that deals with the external and internal forces acting on a de¢ned biological system, and the e¡ects that are produced by these forces. When we consider joint mechanics in osteoarthritis (OA), it is a biomechanical problem. The biological system is the joint. The external forces may be represented by gravitational, inertial and contact forces; and the internal forces are the forces acting on and within joint structures, such as the articular cartilage, ligaments, muscles, tendons, menisci, etc. The speci¢c e¡ect we are interested in, at least in the context of this chapter, is OA. OA is a joint disease of primarily unknown origin. Its end-stage is de¢ned by complete cartilage erosion from all, or parts, of the articular surfaces. As such, it is often considered a disease of the articular cartilage. However, we would like to de¢ne it in much broader terms. OA is a joint disease associated with damage to the articular surfaces, accompanied by osteophyte formation and changes in the subchondral bone structure, the ligaments, menisci, the synovial £uid, etc. 79

80

HERZOG ET AL

It has typically (one might even say always) been assumed that OA is associated, in one way or another, with joint loading. However, to our knowledge this has not been proven with absolute certainty. Nevertheless, we will adopt the notion that joint loading, which can be quanti¢ed through the in vivo joint mechanics, triggers the onset of OA, and once triggered, the disease continues and, at present, cannot be stopped or reversed. This progression of the disease, typically over years, may also be associated with joint loading. However, this notion is much more tentative than the one assuming that it merely triggers the onset of OA. If we adopt the premise that a certain type of joint loading produces joint degeneration leading to OA, the following basic scienti¢c questions emerge: . What is the type of loading that is harmful to the joint in terms of magnitude, rate of application, frequency, and other mechanical descriptors? . What is the biological response to this load that produces the OA response? . And probably most importantly, what are the pathways (or mechanisms) linking the load to the degenerative response? In order to answer the ¢rst two questions, it seems imperative that the normal in vivo joint loading is quanti¢ed and compared to situations that are known to produce end-stage OA, and that the biological responses to such in vivo loading are measured. Theoretically, such experiments should be straightforward, but very little progress has been made towards these goals, probably because the work involved is extremely time consuming and therefore expensive. The third question is the one that really needs answering. However, with little systematic knowledge concerning the ¢rst two questions, solutions to the third question are likely far away. Impressive research on the third question has been made in studies using articular cartilage explants that are exposed to well-de¢ned loading conditions in a laboratory setting (e.g. Burton-Wurster et al 1993, Quinn et al 1998, Sah et al 1989, Torzilli et al 1997). However, the boundary conditions of explants under these arti¢cial loading conditions are far removed from real in vivo joint loading, therefore, the results from these studies, although interesting from a materials property point of view, may have little relevance for the scenario occurring in human or animal OA. Of course, there is the possibility that articular cartilage cylinders with perfectly £at-shaved surfaces subjected to con¢ned, or uncon¢ned, in vitro loading conditions between two metal plates may behave similarly to the naturally rounded cartilage surfaces attached to their native bone that are exposed to the gliding and compression loading from their adjacent cartilage surfaces. However, if so, this must be demonstrated. Here, we will not discuss the vast number of excellent studies on articular cartilage explants, partly because of a lack of relevance in the current context, and partly because

JOINT MECHANICS

81

those studies have received much attention and are summarized in a series of excellent reviews (e.g. Guilak et al 1997, Hasler et al 1999, Mow et al 1994, Sah et al 1992). Instead, we will focus our attention on studies aimed at quantifying the internal and external forces occurring in diarthrodial joints of an experimental model of bona ¢de end-stage OA. This will be done by presenting research on the ground reaction forces and kinematic patterns of the anterior cruciate ligament (ACL) transected cat, by discussing work on the corresponding muscle forces and activation patterns, and by showing results on the associated degenerative responses. Following this discussion, we would like to present preliminary work on the biological response of articular cartilage to controlled in vivo loading of joints, and on an experimental model of muscle weakness. The ¢rst of these studies is a natural extension of the ¢rst part of our work, aimed at approaching a solution to the second question above. The second of these studies is motivated by the idea that muscle forces provide the dominant loads on joints (e.g. Crowninshield & Brand 1981), and that muscle weakness and the associated adaptations in neuromuscular control of movement, might constitute powerful risk factors for joint degeneration and OA.

Methods, techniques and approaches The cat model of OA There are only a few experimental animal models of OA that have been followed longitudinally over years. Two such models are the ACL transected dog (Brandt et al 1991) and cat (Herzog et al 2003, Suter et al 1998). Both these models have been found to present a history of onset and progression of OA similar to the human disease, and have led to complete erosion of articular cartilage from speci¢c areas of the articular surfaces. We chose ACL transection of the cat as our model of choice for in vivo investigation for two primary reasons: . the cat hind limb is the best known mammalian system in terms of muscular anatomy and in vivo force production, and neurophysiological movement control and . the cat knee does not su¡er from naturally occurring OA, therefore degenerative changes in joint structure, control and movement can be associated directly with targeted intervention, even when animals are followed over years where agee¡ects might play a role in other experimental models, such as the dog. The detailed approach to arthroscopic and open joint ACL transection surgery has been described previously (Herzog et al 1993, 1998).

82

HERZOG ET AL

External forces and kinematics The external ground reaction forces for walking animals, before and after experimental intervention, were measured using a pair of animal-sized (10
15 cm) force-platforms (AMTI, Amherst, USA) embedded in a speci¢cally designed walkway (Suter et al 1998). These platforms measure the threedimensional forces and moments acting from the ground on the animal’s paws, and simultaneously provide the location of the centre of pressure of the resultant ground reaction force. Hind limb kinematics were measured using a high speed video system (Motion analysis, Sta. Rosa, CA, 200 Hz), with a correction algorithm to account for skin marker movement (Goslow et al 1973). Muscle forces and activation patterns Forces in selected ankle extensor muscles were measured using E-shaped, external tendon force transducers based on a strain-gauged design (Walmsley et al 1978). Forces in the knee extensors (patellar tendon) were quanti¢ed using an omegashaped, implantable force transducer based on the design of Xu et al (1992) and tested and adapted by Herzog et al (1996) and Hasler et al (1998). Activation patterns of selected knee and ankle extensor and £exor muscles were measured using indwelling, Te£on coated, bipolar, ¢ne wire electrodes embedded into the mid-belly of the target muscles with an approximate inter-electrode distance of 5 mm, and aligned in the approximate direction of the muscle ¢bres (Guimaraes et al 1995). Joint pressure distribution In situ joint pressure distributions in normal and ACL transected knees were measured using Fuji low and medium grade, pressure sensitive ¢lms (Herzog et al 1998). Films were packaged in 12
120 mm strips for multiple measurements, and were sealed in polyethylene adhesive layers for moisture proo¢ng (Liggins et al 1995). Film strips were inserted into the joint space through a lateral opening (approximately 20 mm). Once inserted, controlled knee extensor activation was produced by stimulating the femoral nerve via a bipolar cu¡ electrode and a Grass (S88) stimulator. Knee extensor force was measured using a speci¢cally designed tibial restraining bar, and the force was adjusted by changing the voltage and frequency of stimulation. Fuji ¢lm calibration and analysis were performed as outlined in detail by Clark et al (2002). Results and discussion Osteoarthritis Transection of the cat ACL causes increases in thickness, softness, and permeability of the articular cartilage; and produces osteophyte formation, increases in thickness

JOINT MECHANICS

83

and mass of the medial collateral ligament and medial joint capsule, and atrophy of the hind limb musculature within weeks (Herzog et al 1993, 1998). Over time, sitespeci¢c articular surfaces are eroded away, cartilage becomes soft and loses internal structure, joint space becomes narrower, and osteophytes increase in number and size (Fig. 1). Mechanics Mechanically, ACL transection is associated with an immediate instability of the knee that results in an excessive anterior translation and internal rotation of the tibia relative to the femur for a given amount of force or moment, respectively. Muscle forces, external ground reaction forces and knee angle decrease immediately following ACL transection, and activation patterns of the muscles show two distinct features: . a burst-like (rather than the normal, continuous) ¢ring pattern of the knee extensors and . an additional phase of activation of the knee £exors (semitendinosus) to, as it appears, compensate for the loss of mechanical function associated with ACL transection (Fig. 2). Within approximately four months of ACL transection, joint stability in the anterior^posterior and internal rotation direction are fully re-established (Maitland et al 1998). This result is in contrast to the notion that continuing joint instability, particularly the pronounced anterior shift of the tibia relative to the femur in many animal species (Tashman et al 1995) and humans, is responsible, at least in part, for the progression of joint degeneration. In fact, despite accurate measurements of relative movements between tibia and femur using a sonomicrometry (Sonometrics Corporation, London, ON, Canada) system with a 16 mm spatial resolution, we have been unable to detect the anterior tibial shift typically observed in ACL de¢cient humans, dogs and sheep at the instant of foot contact during cat locomotion (e.g., Korvick et al 1994, Tashman et al 1995). We speculate that the activation of the knee £exors just prior to paw contact in the cat (Fig. 2) might prevent the anterior shift of the tibia relative to the femur. If so, one could argue that proper neuromuscular adaptation o¡sets the mechanical loss of function associated with ACL de¢ciency. However, the crucial clinical question of whether such an adaptation is meaningful in terms of preventing or slowing down joint degeneration has never been addressed to our knowledge (although it is an intuitively appealing idea). Similarly, the reason for the quick re-establishment of knee stability following ACL transection in the cat is not known. We speculate that the initial increase in

FIG. 1. Tibial plateau and retropatellar articular cartilage samples from an anterior cruciate ligament de¢cient cat, ¢ve years post intervention. Left panel, experimental side; right panel, contralateral side. Note, the complete erosion of articular cartilage from the medial tibial plateau of the experimental hindlimb (arrow), and the loss of structure and ¢ssuring in the surface zone (arrow) of the experimental articular cartilage. The contralateral tibial plateau and articular cartilage (right panel) look relatively normal.

84 HERZOG ET AL

JOINT MECHANICS

85

FIG. 2. Gastrocnemius (Gastroc) and knee extensor (Quad) forces, as well as semitendinosus (ST) and vastus lateralis (VL) EMG before (Intact knee) and ten days after anterior cruciate ligament transection (ACLT knee). The values shown are for slow walking (0.4 m/s) on a motor driven treadmill. Observe the decrease in muscle force, and the change in EMG patterns and continuity from before to after ACL transection. Vertical line TD shows the touch down of the hindlimb during the step cycle; the corresponding vertical line PO indicates the instant of paw o¡.

articular cartilage thickness, the strengthening of the medial capsule, the increase in mass of the medial collateral ligament, and the appearance of osteophytes at the joint margins might contribute to the observed increase in joint stability (Maitland 1996). However, despite the apparent lack of anterior translation of the tibia during locomotion, and the reduction of anterior translation and medial rotation to normal values within four months of ACL transection, the cat knee continues to degenerate. Therefore, joint instability might be a factor contributing to the onset and progression of osteoarthritis. However, the results of our studies would indicate that joint instability is not a necessary, but possibly a su⁄cient factor for joint degeneration and OA. We hinted above at the idea that ground reaction forces decrease immediately following ACL transection. However, as with the stability of the knee discussed

86

HERZOG ET AL

FIG. 3. Peak vertical ground reaction forces (means  1 SD, n ¼ 7) as a percentage of body weight for the ACL transected (open symbols) and the corresponding contralateral, intact hindlimb (¢lled symbols) during cat walking and running before (pre) and at various time periods after unilateral ACL transection (1 week^12 months). Note that the forces di¡er statistically at 1 week, 3 weeks and 3 months post-intervention from those observed preintervention, but not for the other time points. A minimum of 10 step cycles for both hindlimbs was used at each time point and each animal to calculate the mean values shown.

above, the ground reaction-force patterns recover to normal values within about 16 weeks for static (quiet standing) tasks, and within about 6 months for dynamic (straight walking and running) tasks (Fig. 3). These results are of interest insofar as in the ACL transected dog, neither the ground reaction forces nor knee stability appear to recover following intervention, not even after 54 months. Therefore, we have two models of experimentally induced OA whose pathogenesis is very similar (e.g. Adams 1989, Herzog et al 1993) but the long-term external mechanics are completely di¡erent. In the dog, the ACL transected joint remains unstable over years, while in the cat, such instability is never observed during walking, and in anterior or medial rotation displacement tests, the instability (of the passive joint) is abolished within four months of ACL transection. Similarly, the ACL transected dog shows reduced vertical ground reaction forces over years, while such force reductions are all but gone in the cat after 4^6 months, depending on the task. Although the interpretation of the pathogenesis of these two animal models of OA, in conjunction with the directly measured mechanics, is wide open, the following one appears the most obvious, and although by no means proven, might stimulate new thinking about the idea of how altered joint mechanics may produce OA. First, one has to acknowledge that the mechanical changes to the joint are similar in the cat and dog in the early stages following intervention, and

JOINT MECHANICS

87

the pathogenesis is similar too. Then, one has to consider that, after approximately 4^6 months, the mechanics start to di¡er substantially, but the progression of the disease remains similar. This scenario suggests that ACL transection in the cat and dog disrupts the mechanics of the knee, and that this event is responsible for triggering events that lead to degeneration and OA of the joint. Obviously, we are not sure what these events are, but in general, we see less muscular forces, less contact pressures, and less external ground reaction forces, thus leading to the hypothesis that there is an initial ‘unloading’ of the joint. Therefore, unloading the joint, rather than the more commonly accepted ‘overloading’ might actually be the key to initiating the degenerative events. Furthermore, once these degenerative processes have been initiated, the stabilization of the joint, and the normalization of its mechanics, do not appear to in£uence the progress of the disease, suggesting that once degenerative processes are initiated, the progression of the disease is not very sensitive to changes in joint loading. If so, this result would imply that rehabilitation strategies aimed at establishing normal movement patterns and joint loading would be most successful early in the rehabilitation process. Early, here, refers to the idea of as early as possible following a traumatic event that might pose the joint at risk for OA. Unfortunately, this result, if correct, would also imply that rehabilitation strategies based on physiotherapy principles would likely not work late in the degenerative process. This is tragic insofar as most degenerative joint diseases are typically diagnosed at an advanced state only, when, according to our interpretation of the results, any mechanical or neuromuscular intervention is too late to be very e¡ective. Or, in other words, diagnosing the possibility for joint degeneration needs to be done immediately upon an insult or trauma. Therefore, the early detection of joint degeneration and OA is likely of prime importance for the successful ¢ght against this disease. Muscle forces and EMG As indicated above (Fig. 2), ankle and knee extensor forces are dramatically reduced in the ¢rst few weeks following ACL transection in the cat. No muscle force measurements in any experimental model of OA have ever been made months or years after intervention. However, it is trivial to show that a given direction and magnitude of the ground reaction force is associated with given muscle force patterns, assuming that there is no substantial amount of cocontraction. Therefore, based on the fact that external ground reaction forces and moments are, on average, the same at about six months following intervention in the cat, we may assume that the individual muscle forces are also similar prior to and about six months following ACL transection. Electromyographical patterns of eight muscles in the cat hind limb are indistinguishable in terms of timing

88

HERZOG ET AL

relative to the step cycle, the magnitude and the frequency content, six months following ACL transection compared to normal. We feel that one result concerning muscle forces and electromyograms (EMGs) deserves attention. Figure 2 shows that gastrocnemius and quadriceps forces are reduced following ACL transection. But not only are the muscle forces reduced, they are also ‘jerky’ rather than ‘smooth’. When comparing the EMG patterns in the vastus lateralis (VL, one of the quadriceps muscles) before and after intervention, it is apparent that the continuous EMG during the stance phase prior to intervention changes to a 3^5 burst pattern during stance following intervention. This result suggests that the ‘jerkiness’ of the knee extensor force patterns following intervention is caused by this burst-like (rather than the continuous) activation pattern. We assume that these individual bursts during stance represent a ¢ght between an excitatory extensor mechanism that sends the commands to extend the knee so that the animal does not collapse, and an inhibitory extensor mechanism of unknown origin, possibly associated with the instability of the knee in the early phase following intervention. We further speculate that this perturbed activation pattern, which results in poorly coordinated muscle activation patterns, and thus presumably poorly controlled ¢ne mechanics of joint movement, might be the mechanism for triggering some of the degenerative responses that are seen quickly following joint perturbation. Pressure patterns In the most perfect world, joint loading could be accurately described by the instantaneous and time-evolving stress and strain states of all the individual tissues that make up a joint. However, that is not possible. An acceptable compromise to this perfect scenario would be the instantaneous and time evolving pressure distribution patterns in diarthrodial contact surfaces during normal movement. However, even such measurements could not be made to date, although pressure-time histories in arti¢cial joints have been made on isolated patients with instrumented joint prostheses (e.g. Bergmann et al 1993, Krebs et al 1991). We have measured the joint surface pressure distributions in the in situ cat knee using Fuji pressure sensitive ¢lm for a variety of knee angles and knee extensor forces representing those observed during normal locomotion (Herzog et al 1998, 2000), in normal and ACL-de¢cient knees. Among the many results, arguably the most important was that pressure distribution in a given knee (for a given knee extensor force and knee angle) changed dramatically from pre- to post-intervention. In 38 measurements of joint contact patterns in ¢ve cats, we found that the contact area in the patellofemoral joint increased by 22% (37%), and the corresponding peak pressure (n ¼34) decreased by 55% (21%) at 16 weeks post ACL transection compared

JOINT MECHANICS

89

FIG. 4. Pressure distribution in the patellofemoral joint of a cat 16 weeks post ACL transection. The contact area in the intact joint (Intact, right) is smaller, and the peak pressure (grey scale) is greater compared to the ACL transected joint (ACLT, left) for the same force across the joint. This result was statistically signi¢cant across measurements in ¢ve di¡erent cats (38 measurements total).

to the normal, contralateral joint (Fig. 4). This result demonstrated for the ¢rst time that joint contact loading changes very quickly following ACL transection, and that a given joint loading (produced by muscular forces via nerve stimulation) for a given joint kinematics (knee angle and angular velocity) might produce completely di¡erent local loading conditions on the articular surfaces, depending on the progression of joint disease. The increase in contact area and the decrease in peak pressure observed in this study with the progression of disease could readily be explained by the altered mechanical properties of the articular cartilage associated with joint degeneration (Herzog et al 1998). This result should illustrate at least two points: ¢rst, measurement of ground reaction forces, and even the muscular forces, says little about the local loading (stress strain states) of the joint. However, it is likely that the detailed local loading is responsible for the di¡erentiated local degeneration of the joint. Therefore, understanding the local, in vivo joint loading might be an essential piece of information that scientists should try to obtain in the near

90

HERZOG ET AL

future, and although we, like many others, have attempted to do this using theoretical approaches, it is our ¢rm belief that meaningful results must come from experimental measurement. Second, the local loading changes continuously in a degenerating joint, thus the biological responses for a given load are likely to change in accordance. Of course, we have not even considered the possibility that a precisely identical load given to a normal and a diseased articular cartilage might produce completely di¡erent biological responses, because of adaptations in the mechanisms that relate loading to the biological response as the joint degenerates. If this should happen (which seems reasonable), the relationship between how, and how fast, a joint will adapt to mechanical stimuli becomes even more complex than for the situation in which the relationship between mechanical loading and biological response remains constant during an adaptive/ degenerative process. Final comments Physiological joint loading and biological response In order to elucidate the detailed mechanisms underlying joint degeneration leading to OA, it seems essential that the in vivo local loading of the target joint is known. Aside from the standard measurements in humans and animal models of OA involving gait analysis, measurement of the external ground reaction forces, and possibly EMG, quanti¢cation of the muscle forces and pressure distribution patterns are possible and provide much needed insight into the joint mechanics. For lack of space, we restricted our focus primarily on this aspect, and we used the cat ACL transection as the model for illustration for the simple reason because, as far as we know, it is the only bona ¢de model of end-stage OA for which systematic measurements of individual muscle forces, EMGs and joint pressure distributions have been made in combination with all the other standard measurements of joint kinematics and external ground reaction forces. Because of this focus, we have neglected two issues that we intended to discuss provided that space was available. The ¢rst of these is related to the idea of obtaining the biological response to controlled in vivo joint loading. Using arti¢cial electrical nerve or direct muscle stimulation, a muscle, or group of muscles, can be activated to produce a precisely de¢ned force, through varying the frequency and current of stimulation, for a precisely de¢ned joint kinematics (joint angle and angular displacement). We have started such work in the rabbit knee. An example of a one hour loading protocol (2 s of knee extensor stimulation every 30 s at about 50% of the total maximal isometric force) was produced and, following the loading protocol, the mRNA expression of some key proteins and proteinases was measured (Fig. 5). In principle, this approach

JOINT MECHANICS

91

FIG. 5. Normalized mRNA levels of metalloproteinase 3 (MMP3) from articular cartilage of central and peripheral regions of patella and femoral groove. The experimental samples were loaded in vivo for one hour as shown in the top panel. The contralateral knee, and control knees from normal animals, were used as unloaded controls.

92

HERZOG ET AL

FIG. 6. Normal articular cartilage from the retropatellar surface of a control animal (A), and articular cartilage from the retropatellar surface of an experimental animal with quadriceps weakness (about 70% loss of force) four weeks past intervention (B). The articular cartilage from the experimental animal has surface ¢ssures, and a great loss of structure. It received a Mankin score of 11 (out of 14) from a blinded observer, while the cartilage from the control animal was scored perfectly normal.

JOINT MECHANICS

93

can be re¢ned to mimic precise movement conditions and muscle forces (and thus joint loading) that occur during unrestrained movements, such as locomotion. We feel that this presents a powerful approach, as physiologically relevant joint loading conditions can be simulated experimentally, the corresponding articular surface pressure distributions can be quanti¢ed, and the expected biological responses can be measured. It seems imperative that these physiologically relevant loading conditions and biological responses are compared to experiments performed on isolated cartilage explants that are subjected to arti¢cial in vitro loading conditions, and to test whether or not the in vitro results are meaningful for the physiological system. The second issue is related to muscle force and weakness. We have described here the important role that muscles play in joint loading, and have hinted that neuromuscular malfunctioning might produce fertile conditions for the onset of OA. However, there is another aspect to the story of muscles: the one of muscle weakness. Muscle weakness has been implicated as a risk factor for joint degeneration (Slemenda et al 1997, 1998). However, there is no direct evidence linking muscle weakness to joint degeneration. We have developed a rabbit model of knee extensor weakness, and have found that within four weeks of muscle weakness, there are gross morphological and histological signs (Fig. 6) of knee and articular cartilage changes that are consistent with a degenerative response (Longino 2003). Realizing that muscle weakness was associated with changes in loading patterns similar to those in the ACL de¢cient cat, the idea that ‘unloading’ or ‘under loading’ may predispose a joint to degeneration, becomes a recurrent theme. Here, we could only summarize the salient features of selected work on the in vivo loading of joints in experimental animal models of joint degeneration and OA. Nevertheless, we hope that we might inspire additional work in this exciting and wide-open ¢eld of scienti¢c investigation. It is our belief that ultimate understanding of human OA will not come from theoretical models of joint contact mechanics and loading, nor from the experimental analysis of articular cartilage explants subjected to loading and boundary conditions far removed from those of the native joint, but must come from the precise understanding of the in vivo loads in diarthrodial joints and the corresponding biological response.

References Adams ME 1989 Cartilage hypertrophy following canine anterior cruciate ligament transection di¡ers among di¡erent areas of the joint. J Rheumatol 16:818^824 Bergmann G, Graichen F, Rohlmann A 1993 Hip joint loading during walking and running, measured in two patients. J Biomech 26:969^990

94

HERZOG ET AL

Brandt KD, Braunstein EM, Visco DM, O’Connor B, Heck D, Albrecht M 1991 Anterior (cranial) cruciate ligament transection in the dog: a bona ¢de model of osteoarthritis, not merely of cartilage injury and repair. J Rheumatol 18:436^446 Burton-Wurster N, Vernier-Singer M, Farquhar T, Lust G 1993 E¡ect of compressive loading and unloading on the synthesis of total protein, proteoglycan, and ¢bronectin by canine cartilage explants. J Orthop Res 11:717^729 Clark AL, Herzog W, Leonard TR 2002 Contact area distribution in the feline patellofemoral joint under physiologically meaningful loading conditions. J Biomech 35:53^60 Crowninshield RD, Brand RA 1981 The prediction of forces in joint structures: distribution of intersegmental resultants. Exercise and Sport Sciences Reviews. The Franklin Institute Press, Philadelphia, p 159^181 Goslow GE, Reinking RM, Stuart DG 1973 The cat step cycle: hind limb joint angles and muscle lengths during unrestrained locomotion. J Morphol 141:1^41 Guilak F, Sah RL, Setton LA 1997 Physical regulation of cartilage metabolism. In: Mow VC, Hayes WC (eds) Basic orthopaedics biomechanics. Lippincott-Raven Publishers, Philadelphia, p 179^207 Guimaraes ACS, Herzog W, Allinger TL, Zhang YT 1995 The EMG-force relationship of the cat soleus muscle and its association with contractile conditions during locomotion. J Exp Biol 198:975^987 Hasler EM, Herzog W, Leonard TR, Stano A, Nguyen H 1998 In-vivo knee joint loading and kinematics before and after ACL transection in an animal model. J Biomech 31:253^262 Hasler EM, Herzog W, Wu JZ, Muller W, Wyss U 1999 Articular cartilage biomechanics: theoretical models, material properties, and biosynthetic response. Crit Rev Biomed Eng 27:415^488 Herzog W, Adams ME, Matyas JR, Brooks JG 1993 A preliminary study of hindlimb loading, morphology and biochemistry of articular cartilage in the ACL-de¢cient cat knee. Osteoarthritis Cartilage 1:243^251 Herzog W, Hasler EM, Leonard TR 1996 In-situ calibration of the implantable force transducer. J Biomech 29:1649^1652 Herzog W, Diet S, Suter E et al 1998 Material and functional properties of articular cartilage and patellofemoral contact mechanics in an experimental model of osteoarthritis. J Biomech 31:1137^1145 Herzog W, Hasler EM, Leonard TR 2000 Experimental determination of in vivo pressure distribution in biologic joints. J Musculoskel Res 4:1^7 Herzog W, Longino C, Clark A 2003 The role of muscles in joint adaptation and degeneration. Langenbecks Arch Surg 388:305^315 Korvick DL, Pijanowski GJ, Schae¡er DJ 1994 Three-dimensional kinematics of the intact and cranial cruciate ligament-de¢cient sti£e of dogs. J Biomech 27:77^87 [Erratum in: J Biomech 27:1295] Krebs DE, Elbaum L, Riley PO, Hodge WA, Mann RW 1991 Exercise and gait e¡ects on in vivo hip contact pressures. Phys Ther 71:301^309 Liggins AB, Hardie WR, Finlay JB 1995 Spatial and pressure resolution of Fuji pressuresensitive ¢lm. Exp Mech 35:166^173 Longino D 2003 Botulinum toxin and a new animal model of muscle weakness. MSc thesis, University of Calgary, Calgary, AB, Canada Maitland ME 1996 Longitudinal measurement of tibial motion relative to the femur during passive displacements and femoral nerve stimulation in the ACL-de¢cient cat model of osteoarthritis. PhD thesis, University of Calgary, Calgary, AB, Canada Maitland ME, Leonard TR, Frank CB, Shrive NG, Herzog W 1998 Longitudinal measurement of tibial motion relative to the femur during passive displacements in the cat before and after anterior cruciate ligament transection. J Orthop Res 16:448^454

JOINT MECHANICS

95

Mow VC, Bachrach NM, Setton LA, Guilak F 1994 Stress, strain, pressure and £ow ¢elds in articular cartilage and chondrocytes. In: Mow VC, Guilak F, Tran-Son-Tray R, Hochmuth RM (eds) Cell mechanics and cellular engineering. Springer Verlag, New York, p 345^379 Quinn TM, Grodzinsky AJ, Buschmann MD, Kim YJ, Hunziker EB 1998 Mechanical compression alters proteoglycan deposition and matrix deformation around individual cells in cartilage explants. J Cell Sci 111:573^583 Sah RL, Kim YL, Doong J-YH, Grodzinsky AJ, Plaas AHK, Sandy JD 1989 Biosynthetic response of cartilage explants to dynamic compression. J Orthop Res 7:619^636 Sah RL, Grodzinsky AJ, Plaas AHK, Sandy JD 1992 E¡ects of static and dynamic compression on matrix metabolism in cartilage explants. In: Kuettner K, Peyron JG, Schleyerbach R, Hascall VC (eds) Articular cartilage and osteoarthritis. Raven Press, New York, p 373^392 Slemenda C, Brandt KD, Heilman DK et al 1997 Quadriceps weakness and osteoarthritis of the knee. Ann Intern Med 127:97^104 Slemenda C, Heilman DK, Brandt KD et al 1998 Reduced quadriceps strength relative to body weight. A risk factor for knee osteoarthritis in women? Arthritis Rheum 41:1951^1959 Suter E, Herzog W, Leonard TR, Nguyen H 1998 One-year changes in hindlimb kinematics, ground reaction forces and knee stability in an experimental model of osteoarthritis. J Biomech 31:511^517 Tashman S, DuPre¤ K, Goitz H, Lock T, Kolowich P, Flynn M 1995 A digital radiographic system for determining 3D joint kinematics during movement. American Society of Biomechanics, p 249^250 Torzilli PA, Grigiene R, Huang C et al 1997 Characterization of cartilage metabolic response to static and dynamic stress using a mechanical explant test system. J Biomech 30:1^9 Walmsley B, Hodgson JA, Burke RE 1978 Forces produced by medial gastrocnemius and soleus muscles during locomotion in freely moving cats. J Neurophysiol 41:1203^1216 Xu WS, Butler DL, Stou¡er DC, Grood ES, Glos DL 1992 Theoretical analysis of an implantable force transducer for tendon and ligament structures. J Biomech Eng 114:170^177

DISCUSSION Brandt: With your botox model, where you reduce muscle strength around the joint, do you see osteophytes? And what does the subchondral bone look like? Herzog: We haven’t looked at that yet. By gross morphology, the osteophytes don’t appear to be there after four weeks. It is really an adaptation model at the moment, and not anything that I would like to say much more about until we have followed it longer-term. What we do see after four weeks of inducing severe muscle weakness through botox is a reddening of parts of the joint surfaces (medial tibial plateau), and degenerative signs (based on the Mankin score) on the articular cartilage, assessed in histological sections. Felson: Ken Brandt, do you want to say something about the Bole animal model for OA? Brandt: Giles Bole, who was the Head of Rheumatology at the University of Michigan, tried, some 20 years ago, to establish OA in the guinea pig with a procedure that did not invade the joint. He performed gluteal tenotomy and myotomy, producing OA in both hips and both knees. This resulted in gait abnormalities that he was able to demonstrate elegantly by dipping the paws of

96

DISCUSSION

the animal into India ink and having it then walk across a piece of paper. It is somewhat similar to what Walter Herzog is doing with botox
destablizing muscle forces. Herzog: This is a very interesting animal model of OA that I was not aware of. In the past two years, we looked at many di¡erent possibilities for inducing muscle weakness, including denervation of the muscle, immobilizing the joint, unloading the hindlimb (for example as done in rats through hindlimb suspension where rats are hung up by their tails, and the hindlimbs are thus lifted o¡ the ground and no loading through ground contact can occur). Although, it is possible in the hindlimb suspension model that muscle loading occurs through the activation of muscles. We decided on the botox model because we wanted to control the amount of muscle weakness, and in this ¢rst study, wanted enough muscle strength left so that the animal could function normally if it wanted to. We played around with the number of botox units. At the end, we injected 3.5 units per kilogram and that seemed to be the magic ¢gure: it reduced muscle force by about 70^80%, and thus left about 20^30% of muscle force, enough to walk and run around, if the animals wanted to. Brandt: Were there any changes in the contralateral joint after four weeks? Herzog: No. Henry: There are some of us in the room who are curious about how long it will take for your cats to develop pain. Herzog: That’s a good question. After ¢ve years the joints look terrible. But having worked for 18 years with cats, I am 99% sure that even after ¢ve years these joints are causing no pain, because the cats walk normally and climb around. When cats have pain, they let you know. I think I have a fantastic model of very bad joint disease with apparently no associated pain. I would be very curious to hear from pain experts about this. Felson: In the break we were talking about structural equivalents of pain. This raises a question about the cat model at 5 years. When you say there are all the changes of OA, would you include synovitis and bone marrow oedema? Did you see this in the cat model after 5 years? Herzog: No, we have not looked at that. Conaghan: Have you scored the synovium for severity? Herzog: No. Pisetsky: If I understood correctly, you are saying that the muscle is doing something in the ACL model to compensate. Herzog: Yes, we think the knee £exor muscles are doing something. In the intact cat knee (as in human knees), the ACL prevents anterior translation of the tibia relative to the femur. In animal models (except the cat), and in humans, it has been reported that after the loss of the ACL, the tibia slips forward relative to the femur at ¢rst foot contact in the stance phase of the step cycle. In the cat, this has

JOINT MECHANICS

97

never been observed, neither by us nor other people. However, what we have observed is that following ACL transection in the cat, the knee £exor muscles are activated at the instant of ¢rst foot contact (they are not activated at that time in the intact hindlimb). Knee £exor activity and force mechanically prevent the tibia from slipping forward on the femur, because the knee £exors are pulling the tibia backward. Therefore, it appears that following the loss of the ACL, the knee £exors take over the mechanical stabilization of the knee that is normally provided by the ACL. The interesting question here is: why does this happen in the cat and not in other animals, and not in humans? Pisetsky: If it is not pain, what is the driving force for the muscle to change? And is this compensation somehow a¡ecting the time until the onset of the arthritis? It is a very long model. Herzog: It is a long model. Except for Ken Brandt, I don’t think I know of anyone here who has run an animal model for several years. I don’t know what the muscle senses to make it contract. My hunch would be that the initial instability would cause some stretch of the muscles that was not there in the intact knee, and that this stretch might activate the muscles. If this is the case, it would all happen very quickly, as we measure this adapted activation pattern of the knee £exors within three days of intervention. Pisetsky: Is that pain? Herzog: In the ¢rst few days the animals have pain. It could be pain, although it is more likely to be a spinal re£ex mediated response. We must also consider that these animals are on pain killers for the ¢rst few days following intervention, thus pain as the source for these adaptations seems rather unlikely. Pisetsky: Can you add a drug to slow or speed up this model? Herzog: We haven’t tried this. Grubb: In your ACL transection model, where you saw the decreased loading of the quadriceps muscle, was there any evidence of guarding? Was the loading larger on the contralateral side? Herzog: Yes. The loading on the contralateral side is initially greater than normal, and greater than on the experimental hindlimb. For static loading (standing still), these di¡erences disappear after approximately 3^4 months (Herzog et al 1993). For dynamic loading (walking), di¡erences in loading between the experimental and the contralateral hindlimbs seem to disappear after about 6 months (Suter et al 1998). Grubb: So were there also gait abnormalities? Herzog: Yes. The results I showed were initial results from 2^4 weeks after intervention. There was a distinct limp and an overloading of the muscles and the ground reaction forces on the contralateral limb. This seemed to recover after about six months. Incidentally, the anterior drawer (i.e. the anterior displacement of the tibia relative to the femur for a precisely controlled anterior force), even in

98

DISCUSSION

the passive sense, goes away in the cat. It seems that the passive structures in that knee take up the slack within about 4 months (Maitland et al 1998). Evans: Why did the cartilage in the femoral groove remain intact? Herzog: There are a couple of possible answers from a mechanical point of view. At any given instant in time, the pressure exerted from one cartilage on the other has to be equal and opposite. If you measure the pressure distribution in the patella at di¡erent knee angles, you realise that the mid-part of the patella is always loaded, whereas there is no area of the femoral condyle that is always loaded. In the patella there is one area that is always loaded, but there is not in the femur. A second explanation could be that the articular cartilage properties di¡er from one side of the joint to the other. For example, we know that the retropatellar surface articular cartilage is always about twice as thick as the one on the femoral groove side. A third explanation might be that it is advantageous having a convex versus a concave surface. Dieppe: Coming back to your painless cats, I would like to suggest that you have a very good model of human elbow OA. There is evidence from pathological studies that elbow OA is extremely common, but it is hardly ever symptomatic. Here we have one human joint which seems to get the disease but which doesn’t get pain, which is what you seem to be modelling in the cat. This also reinforces the earlier point about the heterogeneity between joints, which we are lumping together under the single banner of OA. Herzog: I didn’t know that. I should add that the reason I am holding on to these long-term animals is that I would love to see them develop clinical signs (i.e. pain), to see why after such a long time, when these joints ¢nally might become painful, these joints might be di¡erent from the joints that look terrible at ¢ve years, and what might have been responsible for the sudden appearance of clinical signs. Kuettner: We have been looking at the ankle and knee joints, and we recently extended this study to the elbow and the shoulder. We ¢nd that the elbow behaves similarly to the knee joint, whereas the shoulder is more like the ankle joint, with regard to the capacity ex vivo to synthesize matrix components. Brandt: Did all of your animals acquire a compensatory hamstring gait? Herzog: Yes. Brandt: In humans who have an ACL rupture, physical therapists try to teach that, but some people simply can’t learn it. Herzog: I would suggest that in humans it is not quite as e¡ective because we (humans) essentially walk with a straight knee. In contrast, the cat knee angle goes from about 80^120 degrees during walking, and the semitendinosus (a knee £exor whose activity patterns we measured in our animal model of OA) is attaching about a third down on the tibia. In mid-stance it will perfectly pull the tibia back relative to the femur, whereas in human gait this is never really the case because we walk with an extended joint.

JOINT MECHANICS

99

Hunter: You showed quite a large di¡erence in the muscle size and strength after the botox treatment. Does botox have any e¡ect on muscle quality in addition to muscle size, and if so does this have any relationship to the joint changes you see subsequently? Herzog: I can’t answer that. But we do know that there is a disparity between the muscle mass and the loss of force. The initial loss of force is about twice as big as the loss of mass. This presumably has to do with the fact that botox reacts at the neural junction to the muscle, inhibiting acetylcholine release. Therefore what we presumably see is that we have an initial very high inhibition of acetylcholine release. The muscle then adapts by atrophying and losing mass. Over the next few weeks it will probably lose even more muscle mass before it recovers again. Then of course the e¡ect of botox is reversible, as the beauty industry knows: people who use it to reduce wrinkles have to inject it every three or four months, because after this time there is enough sprouting of new terminal axons that the muscle will be virtually fully re-innervated. Similarly, in the animals, we know that the e¡ect of botox wears o¡ and the muscles recover virtually their full strength. In fact, one of the beautiful aspects of the botox model is that you can control quite precisely the time frame of muscle weakness. Repeat injections can keep the muscle force low for inde¢nite periods, whereas a single injection loses most of its e¡ects within 3^4 months. Grubb: Botulinum toxin injected into muscle classically causes neuromuscular preterminal sprouting but loss of junctional contact. It has always been assumed that the muscle simply atrophies for the period where it is inactive, and then when the botox disappears the new neuromuscular junctions form very well from resprouting of the e¡erents into the neuromuscular or junctional ¢eld. Then you get recovery. I don’t think there is any muscular involvement. References Herzog W, Adams ME, Matyas JR, Brooks JG 1993 A preliminary study of hindlimb loading, morphology and biochemistry of articular cartilage in the ACL-de¢cient cat knee. Osteoarth Cartilage 1:243^251 Maitland ME, Leonard TR, Frank CB, Shrive NG, Herzog W 1998 Longitudinal measurement of tibial motion relative to the femur during passive displacements in the cat before and after anterior cruciate ligament transection. J Orthop Res 16:448^454 Suter E, Herzog W, Leonard TR, Nguyen H 1998 One-year changes in the hindlimb kinematics, ground reaction forces and knee stability in an experimental model of osteoarthritis. J Biomech 31:511^517

Osteoarthritic Joint Pain: Novartis Foundation Symposium 260. Volume 260 Edited by Derek J. Chadwick and Jamie Goode Copyright ¶ Novartis Foundation 2004. ISBN: 0-470-86763-9

General discussion I

Developing animal models of RA Felson: I wanted to reintroduce a question that was posed earlier by Blair Grubb. What is osteoarthritis (OA) and how are we ever going to design an animal model to deal with this particular set of problems? How would it be reasonable to develop or study animal models for drug development or other therapeutic purposes that might have painful OA, and not just OA? Grubb: One of the things that made me ask the question is that there seems to be a very poor correlation between pain and the radiological evidence. In the discussion after my paper, Paul Dieppe made a strong point that the synovium probably wasn’t involved in the pain of OA; that it was probably subchondral bone. What is the evidence concerning the source of pain in OA, and why do clinicians think it is possibly from subchondral bone? And why don’t you think that it is from other structures? Felson: Let me defer you for a minute. There will be lots of discussion later on about bone as a potential source of pain in OA. Grubb: From the point of view of an animal model of OA, if you want one that produces the same pain as OA, you want to know where the pain is coming from. Kuettner: Should we call it an animal model of OA, or should we say it is an animal model of degenerative joint disease? Calling it OA already has clinical implications. If we call it an animal model for degenerative joint disease then we address certain questions. Dieppe: I agree. Also, given the facts that we have alluded to: that hip OA is di¡erent from knee OA or elbow OA, to have a single model doesn’t seem to be the right way to go. I think we are muddling up three things and calling them all OA. There is a group of people, relatively young, whose joints fall to pieces and who need joint replacement. This is a terrible disease that is relatively uncommon. Then there is something that happens to joints that involves adaptation to mechanical change. Then there is a huge burden of pain in the community. I think they are potentially quite di¡erent things, yet we are muddling them up and calling them all OA. Unless we start to unpack this muddle, I don’t see how we can go forward. To say we will have a model of OA seems to be going in the wrong direction, because it is confounding that muddle rather than unpacking it. 100

GENERAL DISCUSSION I

101

Felson: I disagree with you. I’m nervous about doing this because you are someone I look to for ideas that I then slavishly follow! I think OA is a spectrum of a single end-stage disorder. I would take rheumatoid arthritis (RA) and other forms of rheumatic disease such as ankylosing spondylitis as examples where some people have mild structural disease and no reason for severe pain but who are plagued by pain, and other people who have much more structural disease but little pain, and we seem to be able to treat all these people with new agents. This is what OA is going to be to. I would argue that the elbow that isn’t symptomatic would be if we walked on our hands. The cat may not be symptomatic because it is somehow adapted in ways we don’t understand and walks on the other three paws. So much of the symptom relation is mechanical: the incipient symptoms are reproducibly mechanical. I am trying to be a lumper more than a splitter. I’m inclined to say let’s make it easier for people trying to develop therapies and not put impossible barriers in front of them by splitting OA into 20 di¡erent diseases, each of which needs to be treated di¡erently. I think it is ultimately one mechanically driven disease for which we need to understand the aetiology of symptoms better. Pisetsky: In the animal models of RA, I don’t think pain has ever been a requirement for validation. Nor can I remember anyone saying that this should be part of an assessment. Investigators have looked for pathology such as in£ammation. The other thing investigators have studied is whether drugs that work in human disease work in animal disease. If you said that non-steroidal anti-in£ammatory drugs (NSAIDs), for example, had an e¡ect in human OA, it would be of interest to test it in your projected animal models to see whether there is a bene¢t. In lupus, another question has been whether there is a di¡erence in males versus females in disease. One of the attractions of the NZB/W mice is that disease is much worse in females. Are there any animal models of OA where the disorder is worse in female animals? Brandt: We are looking at somewhat di¡erent things. The drugs developed recently for RA, that are proving to be so e¡ective in this disease, were developed with some understanding of the pathogenesis of the disease. They are not general anti-in£ammatories. The drugs we use today for symptomatic therapy of OA are general analgesics. Hunter: That is an important point. You may help structure but you may not necessarily help symptoms. This is a lesson that has been learned in humans. Any animal model development that occurs needs to marry any structural abnormality with any symptomatic consequence that may be revealed. It is important to try to clarify the disease as opposed to trying to divide the disease as much as possible. Animal models are a means of trying to eliminate the pyschosocial elements and just concentrate on the biology. I believe that in the next two to three years a lot of answers will come through from some big longitudinal cohort studies that are

102

GENERAL DISCUSSION I

about to begin, investigating various abnormalities with structurally more sensitive tools than we currently have available to us. It would be reasonable to have a prioritization list of structures within the knee that have a¡erent abnormalities, which people can then walk away with and then develop an animal model to help them explore each structural abnormality further. Rediske: There were a couple of abstracts at the recent national In£ammation Research Association meeting (October 2002) by groups at P¢zer that described the joint pain component of the mouse iodoacetate model of OA. They used a pressure plate device, and showed a di¡erence between the OA knee versus the una¡ected knee in terms of weight-bearing sensitivity. It was this sensitivity that could be reduced by NSAIDs which eliminates any neuropathic component. There is some potential in this model. At Roche, Tony Manning has also started to look at this question of joint pain in OA animal models. Could he comment on his ongoing studies? Manning: Two points. Are there pain measurements that can be made in animal models? Yes there are, and groups are beginning to take these and apply them to caustic and surgical models. Some of the most interesting work has been done in the Harltey guinea pig, a spontaneous model. The quandary is that we are in a catch 22 situation. If the clinicians don’t know how to diagnose the clinical endpoints needed for the registration of new disease-modifying products, how can the researchers then re¢ne their models to ask speci¢c questions about therapies? Any advances in standardizing magnetic resonance imaging (MRI) or biomarkers or classi¢cation of human disease would be helpful. At this point it is unrealistic to ask which is a better animal model of disease that we haven’t yet de¢ned. Felson: Paul Dieppe was pessimistic about OA as a monolithic disease. What if I were to tell you that ¢ve years from now we won’t have any MRI or biochemical markers that help us understand human painful OA? Manning: Until then, you cannot build a better animal model and you will not have better therapeutic opportunities. It will be a long time until we have treatments that protect the structure and the symptoms. Felson: Why is that? You said we can evaluate painful paws and look at OA. Why can’t you test therapies? Manning: The challenge is as you just said: which patients do you do this in? If you pick patients with radiological OA, how many of these truly have pain? If part of the registration requirements are to impact symptoms as well as joint structure, this places an additional burden. Fernihough: What we have with the animal models is a very complicated testtube, so we can only look at the mechanistic side of things. Since the animal model is induced, and is controlled to be consistent, it cannot re£ect the full range of features that characterize painful human OA. Therefore, we can only answer questions regarding the biochemistry and physiology of nociception in a

GENERAL DISCUSSION I

103

whole animal setting. We can look at nerves at the level of the dorsal root ganglion (DRG) and there are no huge changes between normal and OA rats. We can look in the knee, but you see nerves everywhere in a rat knee joint. We are required to demonstrate that we are looking at OA from a histological point of view, so we want a model of OA. But where it falls down is that clinicians can talk to patients in a psychosocial environment and ask them questions about what pain is important to them, but this doesn’t give the animal model people a handle on the mechanistic nociception. There may be subsets such as activity related pain that we can mirror, but the communication is not good at the moment as to what the clinicians would de¢ne as more important from a mechanistic point of view, and separating this from the social environment. Conaghan: I liked David Felson’s use of the phrase ‘joint failure’. What if we said that an individual animal model is just a model of joint failure with predominant subchondral bone damage, or synovitis, or cartilage failure? This is what we have the models of at present. I would agree with Paul Dieppe that there is an important di¡erence between the people who seek help and the people who don’t, but as someone who has run a very busy knee service for a long time, I don’t see this discrepancy between community OA, the young person with premature failure, and a 60 year old with age-related joint failure. I agree with David Felson that this is a common end-stage process. I think the animal models are still valid for particular parts of the joint failure model. Pisetsky: It would be very hard to develop a drug for OA without an animal model. Registration organizations would not be happy, and nor would people testing the drugs. There are so many mouse knockouts available that we should be able test di¡erent combinations of strains and knockouts to produce a suitable animal model. Kuettner: But should we concentrate on stopping the degeneration, or try to ¢nd a drug that will restore and repair the tissue? The surgeon is going in when say 60% of the cartilage is gone. If we want to use a drug, we need to use it way before this and stop the degeneration earlier, possibly in tandem with a repair mechanism. Pisetsky: There are recent studies in RA suggesting that early anti-tumour necrosis factor (TNF) can actually lead to repair. One of the speculations is that, at early stages of disease, there is a repair capability such that, if the in£ammation is blocked, repair ensues. If you go too long before treatment starts, then you lose the repair process. Manning: The challenge is to identify those patients very early so there is an opportunity to intervene. In RA we are in a slightly better situation of being able to detect those patients early. We need better diagnostics in OA. Creamer: Klaus Kuettner said that by the time people get to surgery they have lost 60% of their cartilage and it is then too late. But there are plenty of people who have lost this much cartilage who don’t seem to have any problem.

104

GENERAL DISCUSSION I

Fernihough: From the animal model perspective we are only worried about the histology to demonstrate that it is an OA or degenerative joint process. Working on an animal pain model of OA I don’t care whether the histology gets better or worse
we are focused on the pain. But we don’t have good readouts for that in a clinical environment to keep our model relevant. Felson: I’m confused as to why you are struggling with the clinical readout issues. There are well-validated scales and all kinds of criteria for approval of therapies that relate to signs and symptoms. Fernihough: One of the comments early on was that the psychosocial environment and qualitative analysis of patients is much more important. But I can’t have an in-depth unstructured interview with a rat! Manning: There are good, well-validated readouts. I don’t think there is so much of a problem here. My comments were more concerned with the pursuit of the hypothesis that protecting joint structure will impact signs and symptoms. This is as yet unproven, even in the animal models. Felson: There are drugs that improve symptoms and signs of OA that have been well validated. WOMAC and Lequesne Index have been used and they work OK. Brandt: The problem is that you can say that an animal model mirrors the chondroprotective e¡ects of a new drug for OA only after you demonstrate the e⁄cacy of that drug in humans. Only then can you say that the model is a predictor of the e¡ect of the drug in humans. No animal model of OA has yet been shown to have predictive value. Pisetsky: Is there anything di¡erent about the biomechanics of the animals that has been a hindrance here? I am impressed that the cats go a long time without developing problems, but the dogs get problems in a month or two. Should we be thinking of a particular animal that is more likely to have some correspondence to humans? Brandt: The dogs do not develop joint pain, as far as we know. They change their gait, as do the cats, but they don’t look as though they are in discomfort.

Osteoarthritic Joint Pain: Novartis Foundation Symposium 260. Volume 260 Edited by Derek J. Chadwick and Jamie Goode Copyright ¶ Novartis Foundation 2004. ISBN: 0-470-86763-9

Characterization of joint pain in human OA Gunnar Ordeberg Division of Orthopaedic Surgery, Karolinska Institutet, Danderyd Hospital, Stockholm, SE-18288, Sweden

Abstract. Characterization and di¡erentiation of joint pain is di⁄cult. Though degenerative changes in joints are frequent causes of pain in hip and knee, these changes are not always painful, and other possible causes of pain must also be considered. In degenerative changes in the spine, the problem is even more complex, as peripheral neuropathic pain, caused by mechanical compression and/or leakage of cytokines irritating nerve roots may be di⁄cult to di¡erentiate from nociceptive pain from intervertebral joints, discs or muscles. We know now that nociceptive pain has often referred to areas of pain with numbness and parestesthethic sensations, previously regarded as characteristic for neurogenic pain. Furthermore, in patients with painful coxarthrosis quantitative sensory testing (QST) has shown disturbed sensory thresholds not only in regions adjacent to the a¡ected hip but also contralaterally. These sensory disturbances, previously noted in neuropathic pain, normalized after successful surgery with relief of pain, thus con¢rming the relation to the hip joint. Patients with painful coxarthrosis also have moderately increased substance P activity in cerebrospinal £uid. Thus the ¢ndings show some similarities with ¢bromyalgic patients with highly increased substance P in cerebrospinal £uid and sensory disturbances. In conclusion, joint pain has a profound impact on the sensory system and need a multimodal approach. 2004 Osteoarthritic joint pain. Wiley, Chichester (Novartis Foundation Symposium 260) p 105^121

Pain is usually characterized as nociceptive, neuropathic, idiopathic or psychogenic. Di¡erent receptors and pain transmitters are involved, and responses to analgesic agents di¡er in these categories as does the pattern of pain distribution. Pain is also characterized regarding its quality (stabbing, aching, shooting or paresthetic), whether it is permanent or occasional, or whether it is related to the time of the day, exercise, strain and physical or mental stress. Pain in osteoarthritis (OA) is most frequent in hip and knee, i.e. the big joints under mechanical load. Degenerative changes concomitant with pain are also extremely common in the spine; however there is often controversy as to whether the pain is produced from OA in the intervertebral joints, degeneration of discs or in other structures such as muscles and ligaments (Schwarzer et al 1994). 105

106

ORDEBERG

Furthermore osteophytes, synovitis and capsular thickening in OA of the intervertebral joints
as well as herniation from the degenerated disc
with mechanical and chemical irritation of the nerve structures may cause pain of peripheral neurogenic origin that is sometimes hard to di¡er from degenerative nociceptive pain (Brisby et al 2002). Pain in OA may start either from subchondral bone, as when OA develops as a cause of an avascular necrosis in the femoral head (Arlet et al 1978), from primary lesions in the cartilage (Broom et al 2001) or from joint swelling and in£ammatory reactions with distension of the capsule. As the hip is a ball and socket joint, hip OA will in most cases follow a fairly similar course, as there is only one joint compartment. Though the localization of pain in hip OA may vary, as I will come to later, the surgical treatment in hip OA is also fairly uniform; total hip replacement with endoprosthesis is by far the most common surgical treatment. The questions are mainly of design of prosthesis and whether bone cement or other ¢xation is to be preferred. Osteotomy may still be used in a few young patients with focal degeneration, but it is becoming increasingly rare (Millis & Kim 2002). In the knee there are three compartments from a functional view (McAlinder et al 1992): the medial and the lateral femurotibial joints and the femuropatellar joint. In patients with advanced knee OA usually all three compartments are engaged. However, in patients with moderate knee OA symptoms di¡er according to which compartment is primarily engaged. Femuropatellar joint ¢brillation or degeneration of patellar cartilage is common even in young individuals, especially in athletes, and pain is provoked when the knee is under load in £exion, as in climbing stairs, squatting or in sports. In most cases pain from this joint is moderate; in patients with very severe pain and with malalignment in the femuropatellar joint this can be addressed surgically either by a lateral release of the capsule or transfer of the tibial tubercle (Aderinto & Cobb 2002, Wang 2001). Arthroscopic lavage or smoothening of the cartilage has been used widely in these cases as in cases with slight or moderate OA in other compartments. However, in a recent double-blind investigation (Bradley et al 2002) this treatment turned out to be no better than a sham operation with only skin incision. Results from surgery with patellar prosthesis in patients with isolated OA in the femuropatellar compartment have also been questioned, and have not gained widespread use. In contrast a patellar prosthesis is often used in total knee replacement with replacement of all three compartments (Waters & Bentley 2003). In most cases clinical symptoms in knee OA start in the medial femurotibial compartment (McAlinder et al 1992). Lateral femurotibial onset is seen mainly in patients with deformities or previous fractures, especially compression fractures of the lateral tibial condyle, which is a common fracture in elderly women. Pain usually starts in the part of the knee ¢rst a¡ected; there is also localized tenderness

MECHANISMS IN JOINT PAIN

107

of the a¡ected joint space with provocation of pain in passive extension and rotation of the knee. In isolated medial femurotibial OA in younger cases, proximal tibial osteotomy with shift of the load axis to the lateral compartment is sometimes performed. This operation used to be more common before the evolution of knee endoprosthesis. Interestingly this operation had the reputation of relieving pain at rest. It may be speculated that part of the e¡ect was caused by relief of increased intramedullary pressure in the tibia by the fenestration performed with the surgery (Arnoldi et al 1980). Today surgical implantation of arti¢cial joint components is by far the most common surgery in knee OA. The most common procedure is a total knee replacement with prosthetic replacement of both medial and lateral femurotibial compartments, with or without replacement of the patellofemoral joint (Waters & Bentley 2003), even though isolated replacement of the medial or lateral femurotibial compartment is sometimes performed. Symptoms of hip OA usually start with localized pain either in the groin or trochanteric region. Pain in the ventral part of the thigh and knee is also common and occasionally a patient with hip OA may present with pain predominantly in the knee. The true source of pain is revealed by the clinical examination with restriction and pain with hip rotation and normal physical examination of the knee. In the beginning the patient usually has pain only when the joint is under load. Later, pain at rest may follow. Furthermore, in the beginning the pain can be eliminated or at least signi¢cantly reduced by di¡erent kinds of analgesics but later the response may become insu⁄cient or absent. The area with pain will often spread to the thigh and knee, but pain localization in the leg or even widespread pain all over the body is not uncommon. In fact, having started to use these pain drawings in patients with clinical signs of hip OA, I have been surprised how many patients have wide pain distribution, and also despite this most patients report almost total pain relief following surgery with total hip replacement. Of course, enlargement of the area of pain could be due to concomitant OA in other joints or in the spine. However, we think that sensitization is the most common cause, as this mechanism has been reported in animal experiments, and increased pain area has also been reported in patients with pain from burns (Pedersen 2000). The wide variation of pain intensity in hip OA, with poor correlation with the degree of radiological severity as well as of the histological evidence of synovitis, turned our interest to the plasticity of the nervous system and especially to factors such as central sensitization and activation of endogenous pain inhibitory mechanisms for the modulation of pain sensation. We studied an endogenous pain inhibitory mechanism, namely di¡use noxious inhibitory control (DNIC). In normal individuals, DNIC-like heterotopic noxious condition stimulation can induce reduction of pain sensitivity in other parts of the body. However, in a

108

ORDEBERG

FIG. 1. Mean pressure pain thresholds (PPTs) (
SEM) in OA patients and healthy controls. Before surgery no statistically signi¢cant change in PPT was seen during tourniquets in patients. In controls, PPT increased during tourniquet (P50.002), decreased following tourniquet (P50.001), and returned to baseline values. Following surgery PPT increased during tourniquet in patients (P50.02), decreased following tourniquet in patients (P50.03) and returned to baseline values. In controls PPT increased during tourniquet (P50.001) and decreased following tourniquet (P50.002), but remained slightly elevated compared to baseline values (P50.01). With permission from Lindh et al (1997).

previous study, Eva Kosek showed that ¢bromyalgia patients exhibited dysfunction of this mechanism (Kosek & Hansson 1997) as well as hyperalgesia/ allodynia not restricted to painful areas (Kosek et al 1996). In the study of DNIC in OA, 15 patients with painful hip OA underwent examination and, as 13 of them were later operated on, these were also examined ¢ve months after surgery, when they were almost pain-free. Comparison was also made with sex- and age-matched controls. We used a submaximal e¡ort tourniquet test to induce ischaemic pain. A blood pressure cu¡ was applied on the upper arm ipsilateral to the painful hip and in£ated to 170 mmHg. The subjects were instructed to perform a standardized exercise which gave rise to ischaemic pain in the arm. The pain remained as long as the cu¡ was in£ated. Quantitative sensory testing (QST) was performed contralaterally to the painful area in the leg in the patients and pressure pain thresholds were determined with pressure algometry (Jensen et al 1986), while perception thresholds to light touch were assessed by

MECHANISMS IN JOINT PAIN

109

FIG. 2. Mean light-touch perception thresholds (LTTs) (
SEM) in OA patients and healthy controls before and following surgery. Before surgery, LTTs increased during tourniquet in both groups (P50.001) and decreased following tourniquet in controls (P50.001), but not in patients. After tourniquet LTTs returned to baseline in controls, but there was a strong tendency to elevated LTTs in patients (P50.05). Following surgery, LTTs increased signi¢cantly during tourniquet in both groups (P50.001), decreased following tourniquet in both groups (P50.001) and returned to baseline values. With permission from Lindh et al (1997).

von Frey ¢laments (Aestesiometer; Weinstein 1962). The thermal sensitivity was analysed by a Thermotest with cold and warm stimuli (Hansson et al 1988). Registrations were performed before, during and 45 minutes after the tourniquet test. As shown in Fig. 1, patients had a tendency to lower pressure pain thresholds (PPTs) compared to controls before tourniquet (P ¼0.05), while with applied tourniquet the PPTs were signi¢cantly lower (P50.001) in patients than in controls, and after tourniquet the PPTs were also signi¢cantly lower in patients (P50.002). However, at the second registration, when the patients had been operated on, PPTs increased during tourniquet in patients and controls alike, and there was no statistically signi¢cant di¡erence in PPTs between patients and controls either before, during or after tourniquet. Thus the main ¢nding was that no modulation of pressure pain sensitivity was found in patients before surgery, in contrast to controls, thus suggesting a dysfunction of DNIC mechanisms. However following surgery this dysfunction

110

ORDEBERG

FIG. 3. Mean thresholds to innocuous cold (CTs) (
SEM) in OA patients and healthy controls before and following surgery. Before, as well as following surgery, CTs increased during tourniquet in both groups (P50.001), decreased following surgery in both groups (P50.001) and returned to baseline. With permission from Kosek & Ordeberg (2000b).

had subsided. As shown in Fig. 2 light-touch perception thresholds (LTTs) were more similar in patients and controls; before surgery the thresholds increased in both groups and the only di¡erence was a tendency towards elevated LTTs in patients after tourniquet. Following surgery no di¡erences in LTTs were seen between patients and controls. In the registration of temperature thresholds there was no e¡ect of tourniquet on thresholds to either innocuous warmth or heat pain, before and after surgery. Perception thresholds to innocuous cold increased during tourniquet in patients as well as in controls, decreased in both groups after tourniquet before and after surgery (Fig. 3), and no signi¢cant di¡erences were found between the groups. In the study of hyperalgesia in 14 of the patients with painful hip OA, quantitative sensory testing was performed in the most painful area, in most patients in the trochanteric region, and in their respective controls (see Table 1) before and (in the 12 patients who had subsequent operation) after surgery. Pressure pain sensitivity was assessed with a pressure algometer and thermal sensitivity with a Thermotest. Compared to controls, patients had increased sensitivity to pressure pain (P50.002), innocuous warmth (P50.03), cold pain

MECHANISMS IN JOINT PAIN

111

TABLE 1 Quantitative sensory testing in patients with OA of the hip and healthy controls (mean
SEM) Before surgery (n ¼ 14) Pain side PPT (kPa) Patients 198.8
22.7**{ Controls 338.7
26.3 LTT log 10 (mg) Patients 3.25
0.4 Controls 3.06
0.2 CT (D8C) Patients 1.4
0.2 Controls 1.6
0.3 WT (D 8C) Patients 2.5
0.4* Controls 4.2
0.7 CT+WT (D 8C) Patients 3.8
0.5 Controls 5.8
0.9 HPT (8C) Patients 41.2
0.8 Controls 42.9
0.6

Following surgery (n ¼ 12)

Contralateral side Pain side

252.8
30.1 333.2
22.0

290.8
44.7{ 333.8
32.3

Contralateral side

298.7
52.7 315.4
29.8

3.11
0.2 3.00
0.2

3.01
0.3{{ 3.03
0.3

3.03
0.2 2.97
0.2

1.6
0.2 1.6
0.2

1.6
0.3 1.5
0.2

1.8
0.4 1.6
0.1

3.0
0.5* 3.8
0.5

3.8
0.5{ 3.2
0.3

3.8
0.8{ 3.2
0.3

4.6
0.6 5.4
0.6

5.5
0.7 4.6
0.4

5.6
1.1 4.8
0.3

41.2
0.7 42.8
0.8

43.1
1.0 42.8
0.7

43.3
0.9 43.0
0.6

CT, perception threshold to innocuous cold; HPT, heat pain threshold; LTT, light-touch perception threshold; PPT, pressure pain threshold; WT, perception threshold to innocuous warmth; (CT + WT), sum of innocuous thermal thresholds. Statistically signi¢cant di¡erences are indicated (a) between groups *P50.05; **P50.01; (b) between treatments {P50.05; {{P50.01; (c) between sides {P50.05. With permission from Kosek & Ordeberg (2000a).

(P50.05) and a tendency to increased sensitivity to heat pain (P ¼0.054) before surgery. However, at the second examination after surgery, no signi¢cant di¡erences in sensitivity of any kind were seen between patients and controls. Furthermore patients also had reduced PPTs on the contralateral side before surgery. This was also normalized following operation. Interestingly, the same type of sensory aberrations seen before surgery in these patients with hip OA have previously also been registered by my collaborator Eva Kosek in patients with ¢bromyalgia (Kosek et al 1996). However, these abnormalities have not been found in similar studies of patients with rheumatoid arthritis and with trapezius myalgia (Le¥er et al 2002a,b).

112

ORDEBERG

FIG. 4. sP levels in cerebrospinal £uid from patients with hip OA (coxarthrosis) and ischialgia (radiating pain from herniated lumbar disc) (mean
SEM). ***Patient groups compared to control group (P50.001). Postoperative activity compared to preoperative values (P50.01). With permission from Kosek & Ordeberg (2000b).

It is known from several studies that patients with ¢bromyalgia have increased substance P-like activity (SPLI) in cerebrospinal £uid (CSF) (Vaeroy et al 1988 ). In earlier investigations we have also analysed CSF samples from 11 patients with painful hip or knee OA for SPLI and compared them to a group of nine pain-free controls and nine patients with rhizopathic pain from a herniated lumbar disc (Lindh et al 1997). The SPLI in CSF from the OA patients was increased in comparison to the controls (Fig. 4) but also in comparison to the patients with rhizopathic pain. As shown in Fig. 5 there was a correlation between SPLI and pain score as recorded pre-operatively, but the increased SPLI was less than that seen in patients with ¢bromyalgia (Russell et al 1994). The OA patients in the study were all candidates for surgical joint replacement and had another CSF sample 5 months after the operation. SPLI had decreased, but was still higher than in the controls. Thus these ¢ndings suggest a gradual transition in OA from uncomplicated nociceptive pain to secondary sensory disturbances having similarities with ¢ndings in patients with ¢bromyalgia. An interesting question is of course when these disturbances are reversible; it is obvious that they were reversible in the studies above; however we all know that widespread pain for example in patients

MECHANISMS IN JOINT PAIN

113

FIG. 5. Correlation between CSF levels of SPLI and pain scores as recorded preoperatively. y ¼ 1.78 + 13.9 r 2 ¼ 0.51 (P50.01). With permission from Kosek & Ordeberg (2000b).

with diagnosis of ¢bromyalgia is very di⁄cult to counteract. In the orthopaedic literature, especially regarding patients with back pain, wide pain distribution is often regarded as a sign of behavioural symptoms (Waddell & Richardson 1992). Accordingly there is a reluctance to treat patients with widespread pain surgically, even in the presence of a morphological, potentially pain-inducing lesion. However, we have recently taken part in a study of fusion surgery in lower back pain of segmental origin, where preoperative pain distribution was evaluated from pain drawings. In this study the clinical result two years postoperatively was the same in the patients with wide pain distribution on their pain drawings as in those with more localized pain (Hgg et al 2003). In conclusion, though our knowledge of pain mechanisms in OA and other pain conditions is still very fragmentary, improved classi¢cation and understanding of underlying mechanisms have already increased treatment options in recent years. We believe that further research will give additional improvements in the treatment of pain. References Aderinto J, Cobb AG 2002 Lateral release for patellofemoral arthritis. Arthroscopy 18:399^403 Arlet J, Ficat P, Mazieres B 1978 Coxarthroses due to ischemia. Ischemic coxarthropathy. (French) Rev Rhum Mal Osteoartic 45:549^560

114

ORDEBERG

Arnoldi CC, Djurhuus JC, Heerfordt J, Karle A 1980 Intraosseous phlebography, intraosseous pressure measurements and 99mTC-polyphosphate scintigraphy in patients with various painful conditions in the hip and knee. Acta Orthop Scand 51:19^28 Bradley JD, Heilman DK, Katz BP, Gsell P, Wallick JE, Brandt KD 2002 Tidal irrigation as treatment for knee osteoarthritis: a sham-controlled, randomized, double-blinded evaluation. Arthritis Rheum 46:100^108 Brisby H, Olmarker K, Larsson K, Nutu M, Rydevik B 2002 Proin£ammatory cytokines in cerebrospinal £uid and serum in patients with disc herniation and sciatica. Eur Spine J 11:62^66 Broom N, Chen MH, Hardy A 2001 A degeneration-based hypothesis for interpreting ¢brillar changes in the osteoarthritic cartilage matrix. J Anat 199:683^698 Hgg O, Fritzell P, Hedlund R, Moller H, Ekselius L, Nordwall A 2003 Swedish Lumbar Spine Study. Pain-drawing does not predict the outcome of fusion surgery for chronic low-back pain: a report from the Swedish Lumbar Spine Study. Eur Spine J 12:2^11 Hansson P, Ekblom A, Lindblom U, Marchettini P 1988 Does acute intraoral pain alter cutaneous sensibility? J Neurol Neurosurg Psychiatry 51:1032^1036 Jensen K, Andersen HO, Olesen J, Lindblom U 1986 Pressure-pain threshold in human temporal region. Evaluation of a new pressure algometer. Pain 25:313^323 Kosek E, Ordeberg G 2000a Abnormalities of somatosensory perception in patients with painful osteoarthritis normalizes following successful treatment. Eur J Pain 4:229^238 Kosek E, Ordeberg G 2000b Lack of pressure pain modulation by heterotopic noxious conditioning stimulation in patients with painful osteoarthritis before, but not following, surgical pain relief. Pain 88:69^78 Le¥er AS, Kosek E, Lerndal T, Nordmark B, Hansson P 2002a Somatosensory perception and function of di¡use noxious inhibitory controls (DNIC) in patients su¡ering from rheumatoid arthritis. Eur J Pain 6:161^176 Le¥er AS, Hansson P, Kosek E 2002b Somatosensory perception in a remote pain-free area and function of di¡use noxious inhibitory controls (DNIC) in patients su¡ering from long-term trapezius myalgia. Eur J Pain 6:149^159 Lindh C, Liu Z, Lyrens S, Ordeberg G, Nyberg F 1997 Elevated cerebrospinal substance P-like immunoreactivity in patients with painful osteoarthritis, but not in patients with rhizopatic pain from a herniated lumbar disc. Scand J Rheumatol 26:468^472 Lindh C, Liu Z, Welin M, Ordeberg G, Nyberg F 1999 Low calcitonin gene-related, peptide-like immunoreactivity in cerebrospinal £uid from chronic pain patients. Neuropeptides 6:517^521 McAlindon TE, Snow S, Cooper C, Dieppe PA 1992 Radiographic patterns of osteoarthritis of the knee joint in the community: the importance of the patellofemoral joint. Ann Rheum Dis 51:844^849 Millis MB, Kim YJ 2002 Rationale of osteotomy and related procedures for hip preservation: a review. Clin Orthop 405:108^121 Pedersen JL 2000 In£ammatory pain in experimental burns in man. Dan Med Bull 47:168^195 Russell IJ, Orr MD, Littman B et al 1994 Elevated cerebrospinal £uid levels of substance P in patients with the ¢bromyalgia syndrome. Arthritis Rheum 37:1593^1601 Schwarzer AC, Aprill CN, Derby R, Fortin J, Kine G, Bogduk N 1994 The relative contributions of the disc and zygapophyseal joint in chronic low back pain. Spine 19:801^806 Vaeroy H, Helle R, Forre O, Kass E, Terenius L 1988 Elevated CSF levels of substance P and high incidence of Raynaud phenomenon in patients with ¢bromyalgia: new features for diagnosis. Pain 32:21^26 Waddell G, Richardson J 1992 Observation of overt pain behaviour by physicians during routine clinical examination of patients with low back pain. J Psychosom Res 36:77^87 Wang CJ 2001 Management of patellofemoral arthrosis in middle-aged patients. Chang Gung Med J 24:672^680

MECHANISMS IN JOINT PAIN

115

Waters TS, Bentley G 2003 Patellar resurfacing in total knee arthroplasty. A prospective, randomized study. J Bone Joint Surg Am 85A:212^217 Weinstein S 1962 Tactile sensitivity of the phalanges. Percept Mot Skills 14:351^154

DISCUSSION Lohmander: You have some very impressive data. I have a question about the normalization of the pain threshold after OA surgery. The data you showed were group averages, yet we know that not all patients respond equally well to what we think is standardized and perfect surgery. Have you looked behind these group average data and tried to determine whether those patients who do not respond as well to surgery as you might expect also do not normalize with regards to their pain thresholds? Is there any sign of a relationship between the di¡erent ways of monitoring the outcome? Ordeberg: We tried to do this. In this group there were 15 patients, two of whom were excluded from operation because of medical reasons. So the post-operative ¢gures are from the 13 patients who were operated on. Fortunately (or unfortunately) they were all well. It would be interesting to address the point you raised. Kidd: You showed that the DNICs were reduced in the painful OA preoperatively. Interestingly, this was reversed following joint replacement surgery. What clinical relevance does this reduction in DNIC have for the patient? Ordeberg: I can only speculate. It may be that the DNIC is one of the reasons why we feel less pain when we are highly occupied with something else. It may be that patients with sensory disturbances don’t have that relief from other activities. Grubb: Yesterday we discussed whether the sensitization in OA is central or peripheral. The data you presented on the changes in skin thresholds are very interesting. Are the skin threshold measures simple and easy to apply? Ordeberg: The instruments are described in our papers and are commercially available. However calibration and instructions to patients are crucial. We think all registrations in a study should be done by the same investigator, as was done in our study. Fox: Are these sensory measurements relieved by analgesics? Is the reduced threshold altered? Ordeberg: I can’t answer that. From what I remember there have been a few reports with con£icting results. This may be di¡erent in controls and patients with sensory disturbances. Bradley: We did a small placebo-controlled trial using sertraline (ZoloftTM) and we found that this did produce a signi¢cant increase in pain thresholds at the tender points in our ¢bromyalgia patients compared to placebo (Alberts et al 1998). This is an unusual ¢nding. When you look at most of the studies of serotonin inhibitors

116

DISCUSSION

you don’t see much e¡ect on pain sensitivity. One reason for this may be that in most of the previous studies, the investigators were simply content to look at changes in the number of tender points. By virtue of the fact that one of the classi¢cation criteria for ¢bromyalgia is extreme tenderness at 11 or more of 18 anatomic sites, it is very unlikely that any pharmacological treatment will substantially reduce the number of sites that are found to be extremely tender. It probably is more likely that an e¡ective treatment will increase pain threshold levels at multiple sites, even if most of those sites are still tender. At least our data suggest that it is really worthwhile to examine the value of using quanti¢able measurements of pain sensitivity, such as pain threshold or tolerance levels, rather than simply counting the number of tender points, to assess change produced by a pharmacological intervention. Since patients with ¢bromyalgia are sensitive to a wide array of stimuli at multiple body sites, several other measures of pain sensitivity may also be used to assess the e¡ects of treatment interventions, such as magnitude estimates of the intensity or unpleasantness of pressure, thermal, or ischaemic stimuli or temporal summation tasks involving pressure or thermal stimuli (e.g. Edwards & Fillingim 1999, Staud et al 2003a). Felson: You commented on your study and contrasted it with others. Would it be fair to say that in ¢bromyalgia trials, by and large, the trials that show e⁄cacious therapy with respect to self-reported pain don’t show any e¡ect on tender points? Bradley: That is absolutely right. It is the number of tender points that doesn’t really change. It would be very unusual for a single pharmacological intervention to reduce signi¢cantly the number of tender points. But certainly it could potentially reduce sensitivity at those points. Fox: It is potentially important from an industry perspective. We are trying to design more sensitive clinical trials which will pick up e⁄cacy. We are relying on the VAS score all the time. It may be that if these sorts of changes are consistent, and they are quanti¢able in the way that VAS is, then they may be more sensitive. This could encourage companies to do another trial with their compound rather than discard it. Hunter: Do they have solid standardization sessions where they actually go in and anatomically localize where they are going to take these measurements from? Bradley: We do that by training our research assistants to reliably identify the tender points and to show a high level of consistency with one another in their pain threshold measurements on the same individuals (Aaron et al 1996). I don’t know to what extent other investigators do that. One other variable to consider in these trials is that they will involve multiple study sites. How can we ensure reliability of our measurements across these sites? In addition, even within a single site, the reliability of pressure stimuli delivered manually by one research assistant may vary across multiple trials. There are at least two investigators I know of who have tried to solve these problems by building automated

MECHANISMS IN JOINT PAIN

117

piston-driven devices that apply consistent levels of pressure stimulation across trials. One group is at the University of Florida with Don Price and Roland Staud (Staud et al 2003b) and the other is at the University of Michigan (Gracely et al 2002). These devices are used to apply pressure to ¢ngertips. There also are devices that deliver thermal stimuli in a reliable manner that are commercially available. However, even with these devices, you get into the question of the extent to which they will be used consistently across study centres. So, I think there is still a need to focus on the inter-site and inter-judge reliability in planning these treatment trials. Creamer: Yesterday I mentioned that we have looked at pain threshold in OA of the knee, and it didn’t correlate with reported pain severity. Pisetsky: In your studies, were patients allowed to be on medication in the preoperative state? Ordeberg: From what I remember, not in the days immediately preceding the operation. Pisetsky: Generally, is it necessary to wash people o¡ their medications before this kind of assessment? Ordeberg: We had no speci¢c washout. Most patients did not use analgesics as they reported that the e¡ect had subsided. Those who did use analgesics were instructed not to do so on the day of investigation. Pisetsky: Would existing medicines, such as a non-steroidal anti-in£ammatory drug (NSAID) or a tricyclic, a¡ect these measurements? Ordeberg: It is likely that they would. On the other hand at the follow-up, if they were on medication and had some concentration left at the time of the operation, it would a¡ect the results in the opposite way. After surgery they were not on drugs. Dieppe: I was intrigued by your emphasis on avascular necrosis. This raises an issue that we didn’t discuss yesterday when we were talking about pain features of OA, and this is that in many patients it is episodic. Many people talk about £ares of OA pain. The tendency in the literature is to assume that £ares and episodes of pain are related to synovitis, although the evidence for this is almost non-existent. The sort of thing you have shown raises the possibility that £ares and episodes of pain are actually vascular in origin, and minor events of avascular necrosis. The vascular hypothesis of OA and OA pain is something we have only talked about with respect to raised intraosseus pressure. In the 1950s, when they were developing some of these ideas, Trueta and others talked about the possibility that minor episodes of tiny events of avascular necrosis could be very important in pathogenesis and pain. This vascular hypothesis is worthy of further investigation. Ordeberg: The problem is how we should register these small ¢ndings. Can they be registered by MRI? Dieppe: I don’t know. There is a bit of pathological work in the literature which is supportive of this idea, but I have no idea how you could do this in vivo.

118

DISCUSSION

Felson: There is avascular necrosis noted in most pathological specimens in OA, and there is work in the spine by Leena Kauppila and others that shows association of vascular insu⁄ciency with back pain in spinal disease. Hunter: There have been some interesting pathological studies done on joint replacements, suggesting that the pathological changes that occur in the bone marrow of people with end-stage OA are not dissimilar to what you would expect with a person who has ischaemic necrosis otherwise. Potentially there are some reported vascular changes which you can measure reasonably well using standard MRI. Ordeberg: I would like to add that when hip prostheses are ¢xed with cement, they show immediate pain relief, whereas with many of the constructs with other ¢xation to the bone marrow it takes some time for pain relief after surgery. Implants without bone cement may have some initial motion in the interface between prosthesis and bone marrow, which may cause pain from receptors in the bone. Hunter: Are you suggesting with your substance P ¢ndings within the CSF that this is a process similar to what Professor Schaible was talking about yesterday in terms of a central sensitization? With the blood pressure cu¡, is this really just a distraction that allows it to increase the sensitivity? Ordeberg: I think it is very much a question of sensitization. Fox: It seems that your attention is going towards the spinal cord. Do your studies imply that substance P would be a good target? Ordeberg: It is a marker. To call it a target would be taking a step further. Fox: There has been at least one published trial with NK1 antagonists. Grubb: With regard to the issue of central or peripheral sensitization, I like your mapping diagrams. Earlier, we were discussing whether this was truly nociceptive pain or a sensitization process going on in the joint. Can you say from your data whether all patients showed what looked like extended receptive ¢elds? Do all patients show such marked changes, or is it often quite restricted to the joint region itself? Ordeberg: I think it varies quite widely. Most patients have some degree of pain distribution, but on the other hand those patients selected for these studies all had pain at rest. Schaible: I would be very pleased to see whether substance P is in fact a marker. This would be some way of getting at the issue of whether there is a central sensitization in pain states, even if substance P and substance P receptors are not a very good target for therapy. Ordeberg: On the other hand, we don’t know whether substance P has an action in CSF or whether its presence represents an out£ow from the dorsal root ganglion. Kidd: It would seem logical that substance P would be a good marker. When released in the periphery, it is quickly degraded by degradative enzymes so it

MECHANISMS IN JOINT PAIN

119

can’t di¡use. But from your work and the other work that Blair Grubb has done it would seem that the purpose of substance P in the spinal cord is to di¡use. It is logical that it is going to spill over into the CSF. I would think that it is quite a good surrogate marker. It is as good as anything else. Fox: If it is a marker, isn’t it a surprise that the apparent central sensitization, which is perhaps leading to this referred pain and expanded receptive ¢elds, is relieved so quickly by surgery? Ordeberg: It was six months afterwards. There was still some elevation of substance P compared with controls. Fox: But the pain goes very quickly. Grubb: Hans-Georg Schaible, in your carrageenan/kaolin model of OA where you see expansion of receptive ¢elds, did the receptive ¢elds recede at the same time as you saw the changes in neuronal ¢ring when you applied the NK1 antagonist in the spinal cord? Schaible: Yes, they became a bit smaller. If we then take away the compound there is again an expansion. Gunnar Ordeberg, how stable is the increase of the substance P concentration that you see? Is it a big increase compared with normals? Ordeberg: It is not as big as has been previously reported in ¢bromyalgia patients, for example. In these studies I was limited in that I had only 11 patients. Hunter: It looked very much like the cold sensitivity was being a¡ected in the same way as the pain thresholds were. Is there an analogous way of measuring this? Ordeberg: These systems are quite di¡erent. The e¡ects on the di¡erent thresholds might be varying with the di¡erent pain source. Hunter: I understand that pain ¢bres may also act as the temperature ¢bres. Grubb: There are C cold ¢bres and there are di¡erent ¢bres which respond to warmer temperatures (432 8C). There are also C ¢bres that are mechano^heat sensitive, showing that there is considerable variety in the responsiveness of a¡erent nerve ¢bres. There have been many attempts to try to divide these up into subgroups in skin. I was in a lab recently where this has been done and they have a list of 14 di¡erent types of skin C ¢bres, grouped according to their di¡erent mechanical and heat thresholds. When considering human receptive ¢elds for deep tissues such as the joint, we are also thinking about convergent inputs onto spinal cord neurons. These neurons are receiving inputs from joint mechanoreceptors and from skin mechanoreceptors which may also be thermally sensitive. This means that the situation is complex. Bradley: Most investigators who look at pain sensitivity in patients with OA have primarily focused on mechanical pressure. Our group has looked at di¡erences in heat thresholds and have not found a substantial di¡erence between patients and controls. It would be interesting to perform a temporal summation paradigm using both mechanical pressure as well as thermal stimulation.

120

DISCUSSION

Looking for changes or di¡erences in wind-up would allow us to take a better look at the question of identifying sensitivity markers for central sensitization. Regardless of what we ¢nd, it will be interesting. Will patients respond to both pressure and heat/cold, or will they respond primarily to pressure? Kidd: There is an Australian study looking at pain threshold for mechanical pressure over OA joints and reference sites over the forearms, showing changes in the threshold (Farrell et al 2000). The correlation with pain came from the reference sites. This argues in favour of a central component. Your temporal summation idea would be another way of testing this. Felson: I’d like to ask about capsaicin. Its e⁄cacy in OA might relate to some of what you commented on. Presumably capsaicin has as its mechanism of action this distraction of substance P, similar maybe to the tourniquet that you were administering. Is this a reasonable parallel? Henry: I don’t think so. What we saw earlier was a heterosegmental inhibitory mechanism, whether it loops through the brainstem or lies within the spinal cord. But capsaicin activates the vanilloid receptors, which after a while desensitize. Thus after a while you don’t get activation of the C ¢bres that are releasing substance P and CGRP. This is a totally di¡erent mechanism. It is a desensitization, because at the beginning the e¡ects of capsaicin are not anti-nociceptive but pro-nociceptive. If you eat a meal with a lot of chilli pepper in it, there is a fairly rapid desensitization so 10 minutes into the meal it won’t have the same burning sensation as that experienced initially. Malcangio: You said that capsaicin disrupts C ¢bres. It doesn’t do this unless it is given to neonates. Fox: It is very well established that capsaicin produces a functional desensitization of C ¢bres, which underlies its analgesic activity. Rediske: Along this theme of central versus peripheral sensitization, is there much known about the e⁄cacy of centrally acting analgesics in OA? Does this give us some insights into what is happening mechanistically? Schaible: Has anyone tried dipyrone in OA? This is a compound that hasn’t been used much because it causes some problems. When we tested it for other reasons it reduced spinal activity a lot, but it doesn’t have much e¡ect on a¡erent ¢bres. Has anyone used this? It might be worth trying. Bradley: I have a comment about the paper you published with Dr Kosek. This is one of the papers that drove my interest in pain in people with OA. When we look at DNIC, one factor we need to take into account is whether there are sex di¡erences in the response. The reason I mention this is that in the ¢bromyalgia literature there was a recent paper by Roland Staud and Don Price where they found, just as you did, that there was an abnormal di¡use noxious inhibitory control (DNIC) response in ¢bromyalgia patients (Staud et al 2003b). Their sample was primarily composed of women. When they performed the same

MECHANISMS IN JOINT PAIN

121

experiment with healthy female controls, there was no di¡erence in DNIC between the patients and controls. In contrast, healthy men showed a normal DNIC response. So the question that is now raised is to what extent is abnormal DNIC response related to female sex. I don’t remember the ratio of men to women in your sample. Ordeberg: There were 9 men and 6 women. Bradley: If you looked at the men and women separately you might ¢nd a greater abnormal DNIC response in the men, and perhaps a greater response to the surgical intervention. References Aaron LA, Bradley LA, Alarco¤n GS et al 1996 Psychiatric diagnoses are related to health careseeking behavior rather than to illness. Arthritis Rheum 39:436^445 Alberts KR, Bradley LA, Alarco¤n GS et al 1998 Sertraline hydrochloride (Zoloft) alters pain threshold, sensory discrimination ability, and functional brain activity in patients with ¢bromyalgia (FM): a randomized, controlled trial (RCT). Arthritis Rheum 41(Suppl):S259 Edwards RR, Fillingim RB 1999 Ethnic di¡erences in thermal pain responses. Psychosom Med 61:346^354 Farrell M, Gibson S, McMeeken J, Helme R 2000 Pain and hyperalgesia in osteoarthritis of the hands. J Rheumatol 27:441^447 Gracely RH, Petzke F, Wolf JM, Clauw DJ 2002 Functional magnetic resonance imaging evidence of augmented pain processing in ¢bromyalgia. Arthritis Rheum 46:1333^1343 Staud R, Cannon RC, Mauderli AP, Robinson ME, Price DD, Vierck CJ Jr 2003a Temporal summation of pain from mechanical stimulation of muscle tissue in normal controls and subjects with ¢bromyalgia syndrome. Pain 102:87^95 Staud R, Robinson ME, Vierck CJ Jr, Price DD 2003b Di¡use noxious inhibitory controls (DNIC) attenuate temporal summation of second pain in normal males but not in normal females or ¢bromyalgia patients. Pain 101:167^174

Osteoarthritic Joint Pain: Novartis Foundation Symposium 260. Volume 260 Edited by Derek J. Chadwick and Jamie Goode Copyright ¶ Novartis Foundation 2004. ISBN: 0-470-86763-9

The role of in£ammatory mediators on nociception and pain in arthritis Bruce L. Kidd, Andrew Photiou and Julia J. Inglis Bone & Joint Unit, Bart’s & The London, Queen Mary School of Medicine & Dentistry, Charterhouse Square, London EC1M 6BQ, UK

Abstract. Pain is the most common complaint of individuals with osteoarthritis but the cause of symptoms in this disorder remains unclear. Quantitative sensory testing reveals that in patients with chronic joint disease there is di¡use and persistent alteration of nociceptive (pain) pathways, irrespective of the level of activity of the underlying disease. In£ammatory mediators contribute to this plasticity either by directly activating high threshold receptors or more commonly by sensitizing nociceptive neurons to subsequent everyday stimuli. This involves early post-translational modi¢cation of receptors/ion channels and later, longer-lasting transcription-dependent mechanisms involving changes to the chemical phenotype of the neuron. Included amongst these changes are the increased production and release of various pro- and anti-in£ammatory neuropeptides which have diverse actions on both circulating and resident cell populations. These neurally derived mediators act synergistically with cytokines and growth factors to contribute to ongoing tissue injury. It is becoming apparent that the interaction between a damaged joint and the sensory nervous system is far from straightforward and that activity arising from such interactions may produce not only pain but may also in£uence the subsequent course of the underlying disease. 2004 Osteoarthritic joint pain. Wiley, Chichester (Novartis Foundation Symposium 260) p 122^138

Pain is the most common complaint of individuals with osteoarthritis (OA) but the cause of symptoms in this disorder remains unclear. The broad aim of this review is to examine the complex relationship between tissue injury and the sensory nervous system. More speci¢cally, the emerging concept of neural plasticity, its importance to symptoms and the in£uence of in£ammatory mediators will be described. This is followed by a discussion concerning the likely contribution of sensory nerves to the outcome of chronic joint disease. Neural plasticity and pain At the outset, it is useful to identify several processes as being associated with pain including nociception, pain perception (interoception) and a number of secondary events including communication of distress and disability. For the purposes of this review, nociception is de¢ned as the detection of noxious stimuli by nerves and the 122

PAIN AND INFLAMMATION

123

subsequent transmission of encoded information to the brain. Interoception, on the other hand, is a perceptual and ultimately private experience that arises in response to nociceptive activity but is separate from it. In contrast to interoception, verbal and non-verbal communications can be observed and therefore measured, as can disability, although these all appear to be as strongly in£uenced by a range of cultural, social, demographic and environmental factors as by internal factors. It follows that both the context in which symptoms occur as well as the underlying nociceptive process need to be taken into account when assessing an individual’s report of pain. A characteristic feature of all disorders involving persistent or chronic pain is that there is an exaggerated response to everyday stimuli. In the most obvious example, simple mechanical stimuli such as walking produce varying levels of discomfort in patients with any persistent arthropathy. Whereas in past centuries it was believed that an immutable nervous system delivered a faithful representation of the damaged tissue to a single ‘pain’ centre in the brain, it is now appreciated that the entire nociceptive system is capable of considerable functional change, or plasticity. The extent of this change varies according to di¡erent conditions and appears to be a key determinant of symptoms arising from tissue disease. Minor injuries produce short-lived excitation of specialized high threshold nociceptors with brief, spatially localized pain. More severe tissue damage results in excitation of relevant nociceptors as well as longer lasting changes in the response to subsequent stimuli (peripheral sensitization) (Raja et al 1999). Heightened skin sensitivity following sunburn provides a convenient example. In turn, sustained or repetitive activity within peripheral ¢bres leads to substantial changes to the function and activity of central nociceptive pathways. At a spinal level, this involves exaggerated responses to normal stimuli together with expansion of receptive ¢eld size producing tenderness and referred pain in areas away from the site of injury (spinal sensitization) (Coderre et al 1993). Functional imaging studies of the brain following noxious stimuli using functional magnetic resonance imaging (fMRI) and positron emission tomography (PET) show a complex pattern with discrete areas of activity throughout both cerebral hemispheres (Gracely et al 2002) and it is likely that sensitization also occurs at a supraspinal level (cortical senstization). The consequences of cortical sensitization remain unclear but may produce states of hypervigilance and other more general phenomena observed in patients with chronic pain. Assessing nociception Given the importance of plasticity in the genesis of persistent pain it is perhaps surprising that relatively few studies have directly assessed nociceptive

124

KIDD ET AL

mechanisms in chronic human diseases. It is well recognized that studies of reported pain may not necessarily be reliable guides of underlying nociceptive mechanisms. Quantitative sensory testing (QST) and related procedures attempt to overcome these shortcomings by using psychophysical methods and multiple stimulus modalities to provide a systematic analysis of nociceptive function. The techniques have proved useful for making individual clinical decisions as well as for basic investigation of underlying mechanisms (Gracely et al 2003). The clinical observation that pain may be referred away from OA joints and reports of increased tenderness over apparently normal tissues have led to speculation that changes in central modulation of nociceptive input might contribute to symptoms in OA. Recent quantitative psychophysical studies evaluating pain mechanisms in OA lend support to this idea. Cutaneous and deep hyperalgesia to thermal and mechanical stimuli in subjects with OA have been tested over symptomatic carpometacarpal joints and control sites in the forearm (Farrell et al 2000). Signi¢cantly, variance in movement-related pain ratings was predicted by forearm pain thresholds. Similar results were obtained in a study in which muscle hyperalgesia was assessed by intramuscular infusion of 6% hypertonic saline (Bajaj et al 2001). OA subjects had increased pain intensity with signi¢cantly larger referred and radiating pain areas than matched controls. It is highly unlikely that local changes to nociceptive activity account for either set of results and points to the presence of enhanced central mechanisms. In patients with other musculoskeletal disorders, capsaicin-based techniques have been used to quantify cutaneous axon-re£ex skin £ares as an indirect marker of peripheral sensory activity. Capsaicin is the active ingredient in hot chilli peppers and has proved a useful tool for nociceptive research as it selectively excites unmyelinated sensory ¢bres concerned with nociceptive transmission. In degenerative arthritis of the spine, capsaicin-induced skin £ares were found to be reduced (LeVasseur et al 1990), whereas selective increases over in£amed joints have been reported in rheumatoid arthritis (RA) (Jolli¡e et al 1995). At other reference sites, away from in£amed joints, skin £ares were similar to those observed in normal controls, arguing against a generalized up-regulation of peripheral sensory ¢bre activity in this condition. Capsaicin-based assessments have also been used to explore abnormal central nociceptive activity in patients with chronic joint disease. In addition to producing skin £ares, intradermal injections of capsaicin also produce measurable areas of hyperalgesia and allodynia around the injection site as a result of enhanced sensitization of spinal neurons (Torebjork et al 1992). Pinprick hyperalgesia induced by capsaicin has been shown to be substantially greater over the forearms of RA patients compared to normal controls (Morris et al 1997; Fig. 1) Peripheral sensory activity as measured by von Frey threshold over the forearms of rheumatoid patients was normal. In the rheumatoid patients the maximal area of

PAIN AND INFLAMMATION

125

A

B

FIG. 1. (A) Comparison of the area of capsaicin-induced pinprick hyperalgesia in normal and rheumatoid subjects (solid line indicates mean value) demonstrating enhanced central nociceptive (pain) processing in arthritic disease. (B) Comparison of sensory thresholds for pinprick stimuli using von Frey hairs of increasing force in normal (circles) and rheumatoid (squares) subjects (n ¼ 35 each group) illustrating similar peripheral nociceptive processing at control sites over the forearm. (From Morris et al 1997 with permission.)

126

KIDD ET AL

hyperalgesia was found to correlate with a composite score of upper limb tenderness, but not with overall pain score or a systemic marker of in£ammation. Taken together, these studies show that in patients with chronic joint disease there is a di¡use and persistent alteration of central nociceptive processing, irrespective of the level of activity of the underlying in£ammatory disease. Paradoxically, the studies also reveal a selective increase of peripheral activity over diseased joints but not at control sites elsewhere. These results raise the intriguing possibility that prior stimuli can produce long term and di¡use changes within central nociceptive pathways. They also lead to the question as to what the consequences of peripheral neurogenic (axon re£ex) activity on the longterm outcome of joint disease are likely to be.

E¡ects of in£ammatory mediators on nociception During in£ammation a range of pro- and anti-in£ammatory mediators are released with the balance between these mediators playing a critical role in the subsequent activity of nociceptive ¢bres. Characteristically, the terminals of nociceptive ¢bres express multiple receptors for in£ammatory mediators that become active across relatively narrow ranges of stimulus intensity. The cellular mechanisms by which these changes occur involve early post-translational changes to receptors/ion channels and later, longer-lasting transcription-dependent mechanisms involving changes to the chemical phenotype of the cell (Kidd & Urban 2001; Fig. 2).

Acute in£ammation Whilst some mediators such as bradykinin contribute to pain by directly activating nociceptors others are generally considered to be sensitizing agents. This may arise as a result of changes to the sensitivity of receptor molecules or via modulation of voltage-gated ion channels. Prostaglandins, for example, have been shown to upregulate receptors on primary nociceptive ¢bres using an adenylyl cyclase^protein kinase A pathway (Segond von Banchet et al 2003). A second example is provided in the vallinoid VR1 (TRPV1) receptor which is involved in the transduction of noxious heat stimuli and is also the receptor for capsaicin, discussed earlier. Mutation analysis of the TRPV1 receptor has revealed an inhibitory binding site for phosphatidylinositol 4,5-bisphosphate (PIP2) which can be hydrolyzed following activation of phospholipase C (PLC) by mediators including prostaglandins, bradykinin and NGF. This releases TRPV1 from basal inhibition and lowers the threshold for activation (Prescott & Julius 2003).

PAIN AND INFLAMMATION

127

FIG. 2. In£uences on primary nociceptive neurons leading to ‘peripheral sensitization’. Under normal circumstances, high intensity stimuli are encoded by specialized membrane-bound receptors. Conduction of message to central terminals is mediated by ion channels and excitatory amino acids respectively. During acute in£ammation (A) prostaglandins (PG) and other mediators change the sensitivity of receptors and reduce activation threshold for ionchannels. Longer-term changes include transcriptional events (B) mediated by cytokines and growth factors resulting in enhanced production of receptors, ion channels and transmitters/ modulators. (From Kidd & Urban 2001.)

Nerve growth factor Neurotrophin growth factors, including NGF, make signi¢cant and long lasting contributions to the changes of neuron sensitivity observed during in£ammation. NGF mRNA and/or protein has been identi¢ed in various cell types and a large number of in£ammatory mediators act to increase production, particularly interleukin (IL)1b and tumour necrosis factor (TNF)a (Woolf et al 1997). During the acute stages of an in£ammatory response, nerve growth factor (NGF), acting via neuronal TrkA receptors, produces tyrosine phosphorylation of intracellular targets including ion channels. Over the longer term, NGF exerts a more global in£uence by regulating gene expression leading to prolonged changes in neuron function (Levine & Reichling 1999). In studies using primary cultures of adult dorsal root ganglia neurons, treatment with NGF has been shown to increase both TRPV1 mRNA expression and capsaicin-invoked release of CGRP in a dose-dependent fashion (Winston et al 2001). Other studies have shown NGF regulation of mRNA for the

128

KIDD ET AL

neuropeptides substance P and CGRP as well as the tetrodotoxin resistant sodium channel SNS/PN3. More recently, NGF has also been shown to exert long term in£uences on nociceptor activity in a transcription-independent fashion via the p38 mitogen-activated protein kinase (MAPK) (Ji et al 2002). Cytokines TNFa, IL1b and IL6 have been shown to induce both heat and mechanical hypersensitivity when injected peripherally (Opree & Kress 2000, Sa¢ehGarabedian et al 1995). Antisera to these cytokines reduce hyperalgesia in in£ammatory models (Woolf et al 1997) and the use of novel anti-TNF therapies in rheumatoid arthritis is accompanied by a signi¢cant reduction in pain scores (Andreakos et al 2002). Exact mechanisms of underlying cytokine-induced hyperalgesia remain controversial. As cytokines exert e¡ects on numerous cell types, their hyperalgesic actions have largely been attributed to indirect mechanisms involving production of prostaglandins and other mediators. Consistent with this, IL1b-, TNFa- and IL6-induced hyperalgesia has been shown to be signi¢cantly attenuated by pre-treatment with indomethacin (Cunha et al 1992, Samad et al 2001). Furthermore, in£ammation of peripheral tissues induces COX-2 expression in the CNS and endothelial cells of the blood^brain barrier through an IL1-dependent pathway (Samad et al 2001, Ek et al 2001). Evidence is also accumulating for more direct cytokine e¡ects on nociceptive neurons. Electrophysiological studies of the hyperalgesic actions of subcutaneous TNFa have shown ectopic activity in nociceptive neurons 2 min after injection, whilst signi¢cant mechanical hyperalgesia is present within 15 min (Junger & Sorkin 2000). Likewise, heat-evoked CGRP release from nociceptive neurons was increased 15 min following incubation with IL1b, TNFa and IL6 incubation (Opree & Kress 2000). TNFa increases the sensitivity of cultured sensory neurons to capsaicin, whilst IL1b increases substance P synthesis in cultured dorsal root ganglion (DRG) (Nicol et al 1997). In our own (unpublished) studies we have shown low-level expression of TNF receptors on primary nociceptive neurons and a subsequent bilateral increased expression following tissue in£ammation. This increased expression of receptors for TNF during in£ammation may result in hypersensitivity to the cytokine and may explain the rapid analgesic properties of anti-TNF therapy. Neurogenic in£ammation The nervous system in£uences the immune responses systemically via humoral substances such as cortisol following activation of the hypothalamic^pituitary^

PAIN AND INFLAMMATION

129

adrenal (HPA) axis. A second and potentially more selective mechanism by which neuro^immune interactions may occur is via local sensory and autonomic nerves. Such interactions, which are well documented in disorders a¡ecting the respiratory and gastrointestinal systems (Joos & Pauwels 2000, Chavolla-Calderon et al 2003), may also in£uence the clinical features of many rheumatic disorders. For example, a neurogenic role in the production of symmetrical joint disease has been suggested following the observation that in£ammation in one joint can result in a neurogenically mediated response in the contralateral joint (Kidd et al 1995). Neurogenic in£ammation is mediated, at least in part, by the neuropeptides, substance P (SP) and neurokinin A (NKA) (Maggi 1997). These are members of the tachykinin family of peptides which are encoded by the pre-pro-tachykinin A (PPT-A) gene and are produced by small diameter sensory nerves as well as many types of immune cell (Ho et al 1997, Joos & Pauwels 2000). They act via the neurokinin family of receptors that includes neurokinin-1 (NK-1) and neurokinin-2 (NK-2) with SP having the greatest a⁄nity for the NK-1 receptor whereas NKA acts primarily via the NK-2 receptor. Both receptors are present on leukocytes and many other cell populations (Maggi 1997, Joos & Pauwels 2000). Given the presence of PPT mRNA and SP-like immunoreactivity in many immune cells which also express neurokinin receptors the possibility exists that in£ammatory cells use tachykinins as a paracrine or autocrine signalling mechanism to propagate in£ammation (Chavolla-Calderon et al 2003). In vitro studies indicate that SP stimulates the release of IL1, IL6 and TNF from human monocytes (Lotz et al 1988). SP also mediates mast cell degranulation and production of PGE2 and collagenase from synoviocytes (Maggi 1997). It has also been shown to be chemotactic for a number of cell types including neutrophils, eosinophils, monocytes and more recently, lymphocytes (Hood et al 2000). The expression of vascular endothelial cell adhesion molecules is also up-regulated (Quinlan et al 1999). Importantly, disruption of the NK-1 receptor protects against the injury induced by antigen-antibody complex formation in the airways, revealing that neurokinins serve an important role in this type of injury (Chavolla-Calderon et al 2003). The true arthrogenic potential of substance P and other neuropeptides in naturally occurring human arthropathy is unknown, but we and others have assessed their contribution to the pathogenesis of experimental arthritis in a number of models (Cruwys et al 1995, Kidd et al 2003). For example, mice with a disruption of the NK-1 receptor had signi¢cantly less footpad swelling and mechanical hyperalgesia than wild-type animals in a complete Freund’s adjuvant (CFA) arthritis model (Kidd et al 2003; Fig. 3). Histological and radiological scores were markedly reduced in the knock-out group. The di¡erential e¡ect was particularly marked with respect to scores for synovial hyperplasia and in£ammatory cell in¢ltrate. We are currently exploring the in£uence of SP on

130

KIDD ET AL

FIG. 3. Footpad diameter (a) and mechanical hyperalgesia (b) in wild-type (circles) and NK-1 knockout (squares) mice following adjuvant-induced in£ammation (10 mg/ml) demonstrating pro-in£ammatory e¡ects of substance P during persistent disease. *Indicates signi¢cant di¡erences (P50.05) (from Kidd et al 2003 with permission).

PAIN AND INFLAMMATION

131

leukocyte recruitment by endothelium using intra-vital microscopy. In ongoing (unpublished) studies we have shown that in NK-1 knockout mice, neutrophil migration across the endothelial cell barrier is substantially reduced. Further studies have suggested that SP acts to promote leukocyte migration by a mechanism involving the chemokine, MCP-1. Summary A characteristic feature of symptomatic OA is an exaggerated response to everyday stimuli occurring as a result of functional changes within the nociceptive system. Quantitative sensory testing reveals that in patients with chronic joint diseases there is a di¡use and persistent alteration of central nociceptive processing, irrespective of the level of activity of the underlying disease. In£ammatory mediators contribute to pain either by directly activating high threshold receptors or more commonly by sensitizing these receptors to mechanical and other stimuli. Sensory sensitization involves early post-translational changes to receptors/ion channels and later, longer-lasting transcription-dependent mechanisms involving changes to the chemical phenotype of the sensory neuron. This includes increased production and release of various pro- and antiin£ammatory neuropeptides which have diverse actions on immune and other cells in the periphery. It is therefore apparent that the interaction between a damaged joint and the sensory nervous system is far from straightforward and that activity arising from such interactions may produce not only pain but may also in£uence the subsequent course of the underlying disease. References Andreakos ET, Foxwell BM, Brennan FM, Maini RN, Feldmann M 2002 Cytokines and anticytokine biologicals in autoimmunity: present and future. Cytokine Growth Factor Rev 45:299^313 Bajaj P, Bajaj P, Graven-Nielsen T, Arendt-Nielsen L 2001 Osteoarthritis and its association with muscle hyperalgesia: an experimental controlled study. Pain 93:107^114 Chavolla-Calderon M, Bayer MK, Fontan JJP 2003 Bone marrow transplantation reveals an essential synergy between neuronal and hemopoietic neurokinin production in pulmonary in£ammation. J Clin Invest 111:973^980 Coderre T, Katz J, Vaccarino A, Melzack R 1993 Contribution of central neuroplasticity to pathological pain: review of clinical and experimental evidence. Pain 52:259^285 Cruwys SC, Garrett NE, Kidd BL 1995 Sensory denervation with capsaicin attenuates in£ammation and nociception in arthritic rats. Neurosci Lett 193:205^207 Cunha FQ, Poole S, Lorenzetti BB, Ferreira SH 1992 The pivotal role of tumour necrosis factor alpha in the development of in£ammatory hyperalgesia. Br J Pharmacol 107:660^664 Ek M, Engblom D, Saha S, Blomqvist A, Jakobsson PJ, Ericsson-Dahlstrand A 2001 In£ammatory response: pathway across the blood-brain barrier. Nature 410:430^431 Farrell M, Gibson S, McMeeken J, Helme R 2000 Pain and hyperalgesia in osteoarthritis of the hands. J Rheumatol 27:441^447

132

KIDD ET AL

Gracely RH, Petzke F, Wolf JM, Clauw DJ 2002 Functional magnetic imaging evidence of augmented pain processing in ¢bromyalgia. Arthritis Rheum 46:1333^1343 Gracely RH, Grant MA, Giesecke T 2003 Evoked pain measures in ¢bromyalgia. Best Pract Res Clin Rheumatol 17:593^609 Hood VC, Cruwys SC, Urban L, Kidd BL 2000 Di¡erential role of neurokinin receptors in human lymphocyte and monocyte chemotaxis. Regul Pept 96:17^21 Ji R-R, Samad TA, Jin S-X, Schmoll R, Woolf CJ 2002 p38 MAPK activation by NGF in primary sensory neurons after in£ammation increases TRPV1 levels and maintains heat hyperalgesia. Neuron 36:57^68 Jolli¡e VA, Anand P, Kidd BL 1995 Assessment of cutaneous sensory and autonomic axon re£exes in rheumatoid arthritis. Ann Rheum Dis 54:251^255 Joos GF, Pauwels RA 2000 Pro-in£ammatory e¡ects of substance P: new perspectives for the treatment of airways disease? Trends Pharmacol Sci 21:131^133 Junger H, Sorkin LS 2000 Nociceptive and in£ammatory e¡ects of subcutaneous TNFalpha. Pain 85:145^151 Kidd BL, Urban LA 2001 Mechanisms of in£ammatory pain. Br J Anaesth 87:3^11 Kidd BL, Cruwys SC, Garrett NE et al 1995 Neurogenic in£uences on contralateral responses during rat monoarthritis. Brain Res 688:72^76 Kidd BL, Inglis JJ, Vetsika E et al 2003 Inhibition of in£ammation and hyperalgesia in NK-1 receptor knock-out mice. Neuroreport 14:2189^2192 LeVasseur SA, Gibson SJ, Helme RD 1990 The measurement of capsaicin-sensitive sensory nerve ¢bre function in elderly patients with pain. Pain 41:19^25 Levine JD, Reichling DB 1999 Peripheral mechanisms of in£ammatory pain. In: Wall PD, Melzac R (eds) Textbook of pain, 4th edn. Churchill Livingstone, Edinburgh, p 59^84 Lotz M, Vaughan JH, Crason DA 1988 E¡ect of neuropeptides on production of in£ammatory cytokines by human monocytes. Science 241:1218^1221 Maggi CA 1997 The e¡ects of tachykinins on in£ammatory and immune cells. Regul Pept 70: 75^90 Ho WZ, Lai JP, Zhu XH, Uvaydova M, Douglas SD 1997 Human monocytes and macrophages express substance P and neurokinin^1 receptor. J Immunol 159:5654^5660 Morris VH, Cruwys SC, Kidd BL 1997 Characterisation of capsaicin-induced mechanical hyperalgesia as a marker for altered nociceptive processing in patients with rheumatoid arthritis. Pain 71:179^186 Nicol GD, Lopshire JC, Pa¡ord CM 1997 Tumor necrosis factor enhances the capsaicin sensitivity of rat sensory neurons. J Neurosci 17:975^982 Opree A, Kress M 2000 Involvement of the proin£ammatory cytokines tumor necrosis factoralpha, IL-1b and IL-6 but not IL-8 in the development of heat hyperalgesia: e¡ects on heatevoked calcitonin gene-related peptide release from rat skin. J Neurosci 20:6289^6293 Prescott ED, Julius DA 2003 Modular PIP2 binding site as a determinant of capsaicin receptor sensitivity. Science 300:1284^1288 Quinlan KL, Song IS, Naik SM et al 1999 VCAM-1 expression on human dermal microvascular endothelial cells is directly and speci¢cally up-regulated by substance P. Am J Immunol 162:1656^1661 Raja S, Meyer R, Kingkamp MJNC 1999 Peripheral neural mechanisms of nociception. In: Wall PD, Melzac R (eds) Textbook of pain, 4th edn. Churchill Livingstone, Edinburgh, p 11^58 Sa¢eh-Garabedian B, Poole S, Allchorne A, Winter J, Woolf CJ 1995 Contribution of interleukin-1 beta to the in£ammation-induced increase in nerve growth factor levels and in£ammatory hyperalgesia. Br J Pharmacol 115:1265^1275 Samad TA, Moore KA, Sapirstein A et al 2001 Interleukin-1beta-mediated induction of COX-2 in the CNS contributes to in£ammatory pain hypersensitivity. Nature 410:471^475

PAIN AND INFLAMMATION

133

Segond von Banchet G, Scholze A, Schaible H-G 2003 Prostaglandin E2 increases the expression of the neurokinin1 receptor in adult sensory neurones in culture: a novel role of prostaglandins. Br J Pharmacol 139:672^680 Torebjork H, Lundberg LER, LaMotte RH 1992 Central changes in processing of mechanoreceptive input in capsaicin-induced secondary hyperalgesia in humans. J Physiol 448:765^780 Winston J, Toma H, Shenoy M, Pasricha P 2001 Nerve growth factor regulates VR-1 mRNA levels in cultures of adult dorsal root ganglia neurons. Pain 89:181^186 Woolf CJ, Allchorne A, Sa¢eh-Garabedian B, Poole S 1997 Cytokines, nerve growth factor and in£ammatory hyperalgesia: the contribution of tumor necrosis factor alpha. Br J Pharmacol 121:417^424

DISCUSSION Blake: What is the de¢nition of interoception? Kidd: What I have gone to some lengths to di¡erentiate is the internal unpleasant experience that we all feel, and the external expression of that distress. Interoception is a term used to de¢ne the internal and essentially private experience we have. It is a useful term to di¡erentiate what we experience as a result of nociceptive activity from what we can measure. The word ‘pain’, like ‘in£ammation’, covers a multitude of sins. I think we need to be more speci¢c with our terminology if we are ever going to progress. Fernihough: TNF expression at day 7 after CFA treatment is quite striking. Do you think that this kind of cytokine up-regulation in this phenotypic switch in nerves is then leading to an up-regulation of all the usual candidates, which mechanistically makes in£ammatory pain no di¡erent from neuropathic pain or any other pain? Or do you think that the phenotypic switch is directly linked to ion channel changes? Kidd: When we get down to a molecular level the de¢nitions we use for neuropathic pain begin to break down. Nerve injury is followed by the ectopic expression of receptors. I have shown you that in£ammation is also followed by the ectopic expression of receptors. So you could perhaps argue that one is the same as the other. How you resolve that dilemma is that in neuropathic pain you get ectopic expression of di¡erent sorts of receptors. Maybe it is the cluster of receptors that are expressed that matters, and we now need to de¢ne what these clusters are. Pisetsky: Your results on TNF explain a lot of the clinical bene¢t of anti-TNF. These agents result in an extraordinarily rapid improvement in symptoms, and can work far faster than you would expect clinically. Do you have any information on other cytokines, such as IL1? Kidd: We haven’t looked at IL1 receptor expression speci¢cally, but we would presume that it would increase. It seemed to me that there was a big problem with TNF. There just wasn’t any TNF receptor expression on sensory nerves in normal

134

DISCUSSION

situations. There is some IL1 receptor expression, and it too may increase with in£ammation. Pisetsky: Just as there are pro-in£ammatory cytokines, there are antiin£ammatory ones. Do you think cytokines like IL10 will have another e¡ect on neurons, just as they do on other cells such as lymphocytes? Kidd: As a rheumatologist, the more I look at the nervous system the more it seems to resemble the immune system. It is completely under the control of both positive and negative in£uences. There may well be receptors for antiin£ammatory cytokines. Nancy Watkins has shown IL10 receptor expression and some impressive analgesic e¡ects using IL10. Mackenzie: By using anti-TNF antibodies have you been able to show an antinociceptive e¡ect in your CFA model? Kidd: This is work we are doing at the moment. We have shown that anti-TNF can knock this e¡ect down. We are in the process of quantifying this. Schaible: We also have some data showing TNF receptors on root ganglia. I have a question about the TRPV1 receptor up-regulation: is this the reason why you see an increased capsaicin response in the skin? Kidd: The argument goes that intradermal capsaicin acting through TRPV1 gives an intense a¡erent barrage into the nervous system which then induces secondary sensitization and hence the areas that we have shown. If there is preexisting up-regulation of TRPV1 then those areas will be increased, and so the alternative explanation of these data is that they are not showing central sensitization at all, but are just re£ecting some change in the periphery. The answer to your question is my original observation that i.v. capsaicin did not increase £are areas over the control sites. I can say with con¢dence therefore that TRPV1 was not up-regulated at the control site. This is pretty strong evidence to support a central e¡ect rather than a peripheral one. Henry: You would need a change in TRPV1 receptors to see this. The TRPV1 receptor is simply a Ca2+ channel, and an increase in intracellular Ca2+ will cause the release of whatever is releasable from the terminals, both central and peripheral. If there is an up-regulation of substance P or CGRP that can produce those changes, then simply activating the TRPV1 receptor is going to cause the release of more substance P or CGRP both centrally and peripherally. Fox: I think I’m missing something. The capsaicin response is greater, implying that you think this seems to be an ongoing sensory process. If this is the case, why weren’t they hyperalgesic to start with? Kidd: The £are area was greater in the in£amed joints. We didn’t test over the in£amed joints because we reasoned that the data would not give us meaningful results, because the interpretation would be di⁄cult for the reasons we have just been discussing. We had to go to an area where we believed that peripheral nociceptive function was not modi¢ed. Our hypothesis was that there was

PAIN AND INFLAMMATION

135

pre-existing central sensitiztion in RA. To test this, we had to have an unmodi¢ed site in the periphery. This is why we looked at forearms, rather than over the joints themselves. Fox: TRPV1 expression has certainly been shown to be increased in human skin. Shen: I am interested in substance P. Where does the substance P come from at the local level? Kidd: I was slightly disingenuous to the extent that I just talked about substance P being produced within DRG cells. It is quite clear that substance P is also produced both in resident macrophages and circulating leukocytes. There is a body of evidence now coming out suggesting that this peripheral release of substance P is important. The peripheral cells that have substance P, or have been shown to have the precursor to substance P, also have NK-1 receptors. There is now evidence that these peripheral cells act in an autocrine or paracrine fashion to amplify the in£ammatory cascade. Most of this work has been done in the lung. The nerves release substance P, and acting through an NK-1 mechanism, further enhance substance P release from resident macrophages, which then ampli¢es the response. This is further evidence to indicate that the nervous system is an important player in persistence of an in£ammatory response. Grubb: I’m very interested in the central sensitization argument and the use of the capsaicin model either in a na|« ve ¢eld, away from the RA joint, or over the joint. Let us think about the model that Hans-Georg Schaible proposed yesterday, with these skin inputs which are normally very small and insu⁄cient to excite the neuron. When the neuron becomes sensitized by a constant input, minor inputs from the skin may be su⁄cient to depolarize and activate these neurons. If you believe that this is required and if you choose a site well away from the RA joint, is this really central sensitization? Why didn’t you do it over the joint? Kidd: We had already shown that there was up-regulation of peripheral nociceptive neurons over diseased joints. We know that the capsaicin responses are critically dependent on the state of the peripheral activity. If we were working in an area where there was already pre-existing activity and then showed enhanced hyperalgesia we couldn’t make a claim that this was telling us about central sensitization. Grubb: But if these neurons in the na|« ve areas don’t have inputs to the same neurons that receive inputs in the RA areas, there must be something very di¡use. Kidd: Yes, we feel that there is likely to be a di¡use e¡ect. We come back to substance P and these other neuromodulators in the CNS which di¡use widely. One has to think that there is a di¡use up-regulation of spinal activity: this is the only way to explain the data. Dieppe: As I see it there are some people with OA where the peripheral or noxious stimulus and the peripheral sensitization are probably critical, and there are others where the interoception and the context are critical. My worry is that we

136

DISCUSSION

don’t know how to di¡erentiate them. Can you give us some clues: if this hypothesis is in any way correct, how can we di¡erentiate what are potentially two quite di¡erent pain scenarios? Kidd: Disease in a joint does result in nociceptive activity. I’ve argued that we can now measure that using quantitative sensory testing. We now need to go back to a group of people who we have de¢ned as having OA pain and assess the nociceptive mechanisms. I would argue that we would identify a group who had nociceptive activity and symptoms, and a group in whom there is less nociceptive activity. We would speculate that this group would have more of a psychosocial contextual input. This would be the way to sort it out. We have to look at nociception and we have to look at context. Trying to do one on its own won’t give the picture. The answer is to go back and do more quantitative sensory tests. Blake: I think Paul Dieppe is fundamentally wrong. This comes back to the comment made before that the visual analogue scale is an unreliable measure of pain. Intrinsically, it has to be because we are using a linear system to measure a non-linear event. A study that was recently sponsored by Glaxo involved someone with a burn injury. Using a MRI machine they showed a correlation between the visual analogue score and accelerated uptake around the limbic system, which was 0.9. The only thing that lights up is the hypothalamus, cingulate gyrus and limbic system. The sensory cortex doesn’t come into play. The fact that the correlation is 0.9 tells us that the visual analogue scale (VAS) is your emotional response to whatever, and the lighting up of the cingulate system is an index of your emotional response. This indicates that the distinction between these things is zilch. Kidd: I’d say that you are wrong! The fMRI studies are showing nociception not pain. You are failing to address the di¡erence between what the patient is telling us and what the patient is experiencing. The fMRI studies are no more valid than a capsaicin skin test in telling us about nociceptive activity. How the patient then chooses to communicate the resulting distress is an entirely di¡erent matter and is critically dependent on context. The VAS score does not just assess the emotional response; it tells us about the sensory discriminative aspect also. Bradley: I’d like to a⁄rm what Bruce Kidd has just said. What you get in a fMRI study looking at a correlation between brain activity and VAS depends in part on how you instruct the subject to use the VAS. It is possible with about an hour of training to teach people to di¡erentiate between the intensity of whatever it is they are experiencing versus the unpleasantness of what they are experiencing. I will show some data in my paper on ethnic group di¡erences in emotional pain responses, and how this correlates with brain imaging. Another simple way of looking at this is as follows. We are working on a study now in which we are exposing patients with ¢bromyalgia and controls to stressors in the laboratory. Beforehand, we train people to use the VAS in one way to measure the intensity

PAIN AND INFLAMMATION

137

of the sensation produced by a thermal stimulus and in another way to measure the unpleasantness. What we are ¢nding is that stressors in the laboratory don’t di¡erentially a¡ect perceptions of pain intensity between patients and controls, but patients with ¢bromyalgia show much greater increases in perceptions of pain unpleasantness after exposure to the stressors in the laboratory. We also ¢nd some reductions in blood levels of cortisol as a function of the stressors. With quanti¢ed sensory testing we can train people to di¡erentiate between di¡erent dimensions of the pain experience, and you see di¡erential e¡ects when you have properly trained subjects. Malcangio: Do antidepressant drugs relieve OA pain? From what you say there is some emotional component. Yashpal: When you look at intracellular mechanisms of action, it is becoming clearer that there are some proteins that might be common to pain and epilepsy. An anti-epileptic such as gabapentin is very e¡ective for some chronic pain patients. On the other hand, we have published a paper showing that NK-1 antagonists are very e¡ective antidepressants. The emerging theory is that some of the intracellular mechanisms might be common. Fox: What you say is correct but I think it only applies to certain pain states. We are used to recognizing that anticonvulsants and the newer anti-epileptic drugs work in neuropathic pain, as do SSRIs and tricyclics. It would be new to us if we knew that these agents were working in nociceptive or in£ammatory pain mechanisms. This would be a real surprise. The other thing about the NK-1 story is that one of the studies showed NK-1 antagonists are e¡ective in depression, but the same drug was ine¡ective in chronic pain. Pisetsky: Clinically, antidepressants are commonly used in OA. Felson: Although frankly there is no evidence that they work. Dieppe: The clinical answer is that there isn’t evidence, but occasional patients respond extremely well. The problem is that we don’t know how to pick who is going to respond to an anxiolytic or antidepressant and a non-steroidal antiin£ammatory drug (NSAID). This is because we don’t know how to sort out the heterogeneity of the pain in OA. Yashpal: We don’t know the intracellular mechanism by which gabapentin works, but it is commonly prescribed. It works pretty well in quite a few chronic pain patients. Creamer: You said that one of the ¢rst things we should do is to look at quantifying pain responses in OA. Could you expand on how we might begin to do this? Kidd: The algometer applies mechanical pressure and measures thresholds for mechanical stimuli. You can also measure thermal thresholds. There is a big literature on this. These studies need to be done carefully so they can’t be used in a big population. The tools are there.

138

DISCUSSION

Rediske: I thought your last point was provocative, that mediators of the neurogenic in£ammatory processes could be playing a role in OA. Are you aware of studies looking for the potential disease modifying activity of analgesics that target some of the potential neurotransmitters? Kidd: I think it is the other way round. The data are that NSAIDs are catastrophic for the joints. There have been no long-term studies. How would you do it? How would you reduce peripheral e¡erent activity? It would be very di⁄cult. Rediske: With substance P antagonists, which have been tried in a number of clinical situations. Kidd: My understanding of those studies is that they have just looked at pain responses.

Osteoarthritic Joint Pain: Novartis Foundation Symposium 260. Volume 260 Edited by Derek J. Chadwick and Jamie Goode Copyright ¶ Novartis Foundation 2004. ISBN: 0-470-86763-9

Molecular events of chronic pain: from neuron to whole animal in an animal model of osteoarthritis James L. Henry Department of Physiology and Pharmacology, Medical Sciences Building, University of Western Ontario, London, Ontario, N6A 5C1, Canada

Abstract. A novel animal model of osteoarthritis has been established for studies on the disease process as well as on mechanisms underlying the symptoms of osteoarthritis, principally pain and fatigue. The model is established by cutting the anterior cruciate ligament in the rat to introduce instability of the joint, removing the medial meniscus to induce further derangement of the joint, and exercising the rat on a modi¢ed rota-rod to generate weight-bearing mobility of the joint; this exercise can be regulated in terms of duration to govern severity of the model. The model exhibits many of the characteristics of osteoarthritis in humans, including bone and cartilage remodelling, in¢ltration of the joint tissues by immune cells and alterations in sensory mechanisms. This model will ¢nd application in development of novel interventions for the treatment of osteoarthritis and its symptoms as well as development of diagnostic tools for early detection. 2004 Osteoarthritic joint pain. Wiley, Chichester (Novartis Foundation Symposium 260) p 139^153

An astounding 18% of the population in the western world is expected to su¡er from some type of arthritis by the year 2020 (Lawrence et al 1998). Osteoarthritis (OA) remains the most common form of arthritis, a¡ecting 12% of US adults. OA is the second most common diagnosis, after chronic heart disease, leading to Social Security disability payments due to long-term absence from work. Although it is more prevalent in older patients, it has been shown to be di¡erent than the ageing process (Aurich et al 2002). The features of OA constitute a group of conditions that are diagnosed upon common pathological and radiological characteristics. It is prevalent in the knees, hands, hips and apophyseal joints of the spine. The cause of OA is likely multifactorial, although there are two general concepts. One theory is that there is material failure of the cartilage network leading to tissue breakdown (Poole 1999). The second theory is directed toward injury of chondrocytes with increased degradative responses (Aigner & Kim 2002). Underlying factors contributing to OA include obesity, developmental and anatomic abnormalities, 139

140

HENRY

aging and immune responses. Recent studies have suggested genetic contributions to OA (Loughlin 2001). Despite a myriad of causative factors, clinical and radiological features of the disease are similar. Clinical signs, such as radiological features, are useful for the diagnosis and prognosis of OA. However, patients complain of pain and fatigue. Pain that is a medical problem is actually a family of disorders that can be broken down based on origin: nociceptive pain, due to activation of pain sensory receptors; neuropathic pain, due to activation of the axons of pain ¢bres; and idiopathic pain, for which there is no discernable aetiology. Recent advances have led to our understanding of pain as a process rather than a phenomenon (Stucky et al 2001). The earliest pains are brought about by a set of processes that trigger additional mechanisms, including plastic changes in the nervous system. Thus, those processes that underlie acute pain can eventually trigger mechanisms that lead to chronic pain, where pain becomes unnecessary and debilitating (Woolf & Salter 2000). In the past, it was not understood that to treat pain we need to treat the mechanisms that produce the pain, rather than just the disease itself. Previous thought was that if the disease were treated successfully, the pain would disappear. This is not consistent with current thought among pain specialists, who would cite phantom limb pain as a striking example; the pain persists even after amputation of the a¡ected site (Flor 2003, Haigh et al 2003). Chronic pain is seen as a disorder of the nervous system. There is an obvious relationship between OA and the symptoms of OA, but we need to identify the mechanisms triggering these symptoms, in particular, pain and fatigue. These concepts underlie the hypotheses that follow. To recap, diseases can trigger changes in the nervous system that then generate ongoing persistent pain. Because of the nature of signal transduction mechanisms, untreated or even undertreated acute pain, if allowed to persist, can develop into chronic pain that arises by di¡erent mechanisms (Woolf & Salter 2000, Ji & Woolf 2001, Stucky et al 2001). These longer-term mechanisms are usually more insurmountable, and thus present greater medical challenges. Early intervention is thus important. Irrespective of what triggers the original pain, a virtual soup of chemicals participates in the activation of pain neurons in peripheral tissues. There is a similar soup of chemicals participating in conduction of the pain signal at the ¢rst sensory synapse, where nerve cells projecting from the periphery pass information to the next neuron in the pathway. At both peripheral and central nerve terminals, certain chemicals not only act over the short term to create and transfer the pain signal but also alter the properties of the pain neurons themselves (Herbert & Schmidt 2001, Zimmerman 2001). Thus, these neurons can be seen as being plastic or malleable. It is important to block the pain signal before it can act long enough to induce these plastic changes. Once established, the chronic pain that results is di⁄cult to overcome (Goucke 2003, Pohl et al 2003).

ANIMAL MODEL OF OSTEOARTHRITIS

141

Early intervention requires early detection. In fact, detection of the disease process before it can initiate pain is important in preventing plastic changes. To this end, there is a search for early markers of pain. There are typically three components to a comprehensive pain assessment: pain diagnosis based on inferred pathophysiology, identi¢cation of contributing factors and identi¢cation of barriers. We always try to identify ¢rst the primary cause of pain. However, by applying a single-dimensional pain assessment to a case of chronic pain, one risks slipping into a quagmire of failed therapies, increasing symptoms, angry patients and frustrated physicians. Reliable surrogate markers of pain may ¢nd inestimable usefulness in objectifying and quantifying assessment measurements. Surrogates can be used in a comparative way to characterize the disease (diagnostic value) and to measure disease progression (prognostic value) (Wang et al 2002). Surrogate markers also o¡er a means to determine e⁄cacy of pain management interventions. The contribution of the nervous system to OA has been the subject of only limited qualitative assessment, surprising in view of the obvious involvement of the nervous system in OA pain. One more concept that needs to be introduced is how sensory nerves in£uence the health of peripheral tissues. Sensory nerves can initiate and maintain in£ammation by the release of mediators from a¡erent terminals. Simple examples are the £are and wheal of the triple response that is seen after a scratch to the skin. The £are and the wheal are eliminated by prior anaesthesia of the sensory nerves, and are absent in animal experiments in which sensory nerves have been cut and allowed to degenerate (Herbert & Schmidt 2001). Neurogenic in£ammation has been implicated in a number of conditions, including in£ammatory arthritis (Sharif et al 2004). Sharif et al (2003) provide evidence that di¡erent nerve ¢bre types regulate joint physiology in di¡erent directions, with some acting in an anti-in£ammatory capacity, while others are pro-in£ammatory. Hypotheses Primary sensory neurons pick up the sensory signal in peripheral tissues through either specialized or undi¡erentiated peripheral nerve terminals, called sensory receptors. These receptors can transduce only three types of adequate stimulus
mechanical, thermal and chemical. Our primary hypothesis is that the pain of OA arises as a result of chemical activation of sensory receptors that speci¢cally carry the pain signal. This hypothesis demands identifying the chemicals that are altered in and around the joint at di¡erent stages of OA, as one or more of these may generate the pain. A second hypothesis is that joint chemicals in OA enter the circulation and lead to fatigue. A third hypothesis is that neurotransmitters released from the peripheral terminals of sensory ¢bres regulate disease progression and may also participate in the initiation of OA.

142

HENRY

Objectives Our objectives are to (i) understand aetiology of OA and its symptoms of pain and fatigue, (ii) identify biomarkers and other tools for early diagnosis before onset of clinical signs, (iii) discover targets for development of innovative intervention strategies, and (iv) provide reliable markers of prognosis in longitudinal studies. The proposed programme will provide a rich transdisciplinary environment for the training of basic and clinical scientists. As well, this programme will provide novel research models, reagents and techniques for academia and industry. We have established an animal model of OA to enable longitudinal studies that can focus on the earliest stages of OA, including a period before the onset of clinical signs. The long-term objective using this model is to investigate the earliest signs of OA, and to identify biomarkers and gene markers of the disease and its symptoms
pain and fatigue
to develop methods of early detection; to gather fundamental knowledge regarding factors that will modify disease progression; to identify mechanisms underlying the pain of OA; to develop novel interventions based on mechanisms, and to measure outcomes of these and other interventions under development. Animal models of OA allow longitudinal studies to begin before manifestation of any clinical signs. Early detection and intervention are of critical importance to circumvent the debilitation of OA, if we assume that clinical signs become detectable only when the condition is already at an advanced stage. There are several animal models of OA available, including horse, rabbit, dog, guinea pig and rat (Smith & Ghosh 2001, Brandt 2002). In selecting the model, we were aware that any animal model has its shortcomings when compared to the human condition. These allowances aside, the rat model was chosen for several reasons: . High-content A¡ymetrix microarray analyses can be done only in rat or mouse. Electrophysiological studies of the type required cannot be done in mouse. Therefore, the rat is the only species where function can be tied directly with gene chip analysis. . Physiological data from the rat for pain and fatigue are abundantly available in the published literature. . Electrophysiological data on rat spinal and peripheral sensory mechanisms are also extensive. . The rat is a common species for studies on transcription factors, neurotransmitters, cytokines, etc. . In our model, the severity of the disease can be controlled (see below). . Small animals such as rats are preferred by industry as models for preclinical drug development.

ANIMAL MODEL OF OSTEOARTHRITIS

143

The model We have developed a novel OA model in the rat through modi¢cation of a previously published procedure (Williams et al 1982, Stoop et al 2001). The knee joint is destabilized by cutting the anterior cruciate ligament (ACL) and removing the medial meniscus (Karahan et al 2001). Next, we developed three methods to regulate severity. One is through forced weight-bearing articulation of the OA joint; to achieve this, a rota-rod device was modi¢ed in-house. The rotating rod is elevated to one metre as a disincentive for the rat to fall. The rotating rod is also covered with carpeting to increase grip by the rat. A DC motor was installed to slow the turning rate to a level that the rat can remain comfortably on the rod, but is required to walk continuously. The diameter of the walking surface is 6.7 cm so that the hind legs are continuously walking uphill. Each rat is exposed to forced weight-bearing articulation for one hour. Pilot studies indicate that forced movement exacerbates cartilage degradation in the ACL/medial meniscus model. A second method is forced articulation without loading, achieved by daily 5 min periods of swimming in a circular tank ¢lled with water to a depth of 27 cm. The third method is voluntary weight-bearing articulation in an open ¢eld. In this case, rats are exposed for daily 10 min periods in an open ¢eld with rat toys. In all cases, severity can be controlled by changing the duration of the activity. The model we have developed exhibits a number of signs that allow quantitative measurement of disease progression and its associated symptoms of pain and fatigue. The OA joint exhibits sclerosis of bone, osteophytes, altered 3D conformation and decreased gap. Other data indicate decreased cartilage thickness in the OA joint, cartilage lesions, increased sensitivity of nociceptive neurons and decreased threshold for activation of nociceptive neurons arising from the joint. The overarching strategy for developing a new model in this way is that the severity can be regulated by the duration of the activity paradigm. A common complaint about preclinical animal models is that they are so severe that it is di⁄cult to assess responses to interventions. On the other hand, it is important to be able to quantify physiological parameters of the model (particularly pain and fatigue) and to identify biomarkers. The ability to regulate severity was considered very important for this study, and thus this model stands distinct from other animal models of OA. Results Signs of osteoarthritis Radiological examination using micro-computed tomography reveals joint space narrowing, osteoarthritis and osteophyte formation in the arthritic joint.

144

HENRY

Histological examination indicates in¢ltration of immune cells. Injection of Evans Blue dye intravenously and analysis of joint tissue demonstrated extravasation of plasma proteins, indicating loss of integrity of the endothelial barrier. Behavioural tests indicate a withdrawal response to von Frey hair stimulation at intensities below those that provoke withdrawal in normal animals, indicating the development of an animal equivalent to allodynia. Exploration in open ¢eld testing was also reduced. Electrophysiological experiments Pilot electrophysiological experiments have been run on this model. Recordings from single dorsal horn neurons have demonstrated increased receptive ¢eld size, reorganization of receptive ¢elds to include both deep and cutaneous territories, spontaneous discharge developing in silent neurons following passive articulation of the arthritic joint, and reversal of the e¡ects of g-amino butyric acid (GABA) applied by iontophoresis from inhibition to excitation. Summary and conclusions This model therefore demonstrates several signi¢cant features, generally remodelling of bone and cartilage, in¢ltration of immune cells, disruption of homeostatic mechanisms maintaining the integrity of soft tissues in the joint, alterations in sensory withdrawal thresholds and remodelling of spinal pain mechanisms. Our results thus indicate that this novel animal model of OA may prove useful as one step in understanding mechanisms underlying initiation and development of OA, and has potential applicability to identifying biomarkers and gene markers of disease onset and disease progression. This mechanistic approach notwithstanding, it is important to appreciate that this is only one approach to getting control of OA and its symptoms, and many important issues remain to be answered. Among these are the development of innovative biosocial approaches to management of the disease, understanding comorbidity of OA and the causal relationships with and impact of other signs, the basis for gender di¡erences in the propensity to develop OA, optimization of social support, and factors in£uencing access to care. References Aigner T, Kim HA 2002 Apoptosis and cellular vitality: issues in osteoarthritic cartilage degeneration. Arthritis Rheum 46:1986^1996 Aurich M, Poole AR, Reiner A et al 2002 Matrix homeostasis in aging normal human ankle cartilage. Arthritis Rheum 46:2903^2910 Brandt KD 2002 Animal models of osteoarthritis. Biorheology 39:221^235 Flor HJ 2003 Cortical reorganisation and chronic pain: implications for rehabilitation. J Rehabil Med (41 suppl):S66^S72

ANIMAL MODEL OF OSTEOARTHRITIS

145

Goucke CR 2003 The management of persistent pain. Med J Aust 178:444^447 Haigh RC, McCabe CS, Halligan PW, Blake DR 2003 Joint sti¡ness in a phantom limb: evidence of central nervous system involvement in rheumatoid arthritis. Rheumatology (Oxford) 42:888^892 Herbert MK, Schmidt RF 2001 Sensitization of group III articular a¡erents to mechanical stimuli by substance P. In£amm Res 50:275^282 Ji RR, Woolf CJ 2001 Neuronal plasticity and signal transduction in nociceptive neurons: implications for the initiation and maintenance of pathological pain. Neurobiol Dis 8:1^10 Karahan S, Kincaid SA, Kammermann JR, Wright JC 2001 Evaluation of the rat sti£e joint after transection of the cranial cruciate ligament and partial medial meniscectomy. Comp Med 51:504^512 Lawrence RC, Helmick CG, Arnett FC et al 1998 Estimates of the prevalence of arthritis and selected musculoskeletal disorders in the United States. Arthritis Rheum 41:778^799 Loughlin J 2001 Genetic epidemiology of primary osteoarthritis. Curr Opin Rheumatol 13:111^ 116 Pohl M, Meunier A, Hamon M, Braz J 2003 Gene therapy of chronic pain. Curr Gene Ther 3:223^238 Poole AR 1999 An introduction to the pathophysiology of osteoarthritis. Front Biosci 4:D662^ D670 Sharif RN, Cahill CM, Ribeiro da Silva A, Menard HA, Henry JL 2004 Remodelling of somatosensory synapses. A mechanism for allodynia in arthritis. Submitted Smith MM, Ghosh P 2001 Experimental models of osteoarthritis. In: Moskowitz RW, Buckwalter JA, Altman RD, Howell DS, Goldberg VM (eds) Osteoarthritis: diagnosis and medical/surgical management, 3rd edn. Saunders, Philadelphia Stoop R, Buma P, van der Kraan PM et al 2001 Type II collagen degradation in articular cartilage ¢brillation after anterior cruciate ligament transection in rats. Osteoarthritis Cartilage 9:308^ 315 Stucky CL, Gold MS, Zhang X 2001 Mechanisms of pain. Proc Natl Acad Sci USA 98:11845^ 11846 Wang H, Sun H, Della Penna K et al 2002 Chronic neuropathic pain is accompanied by global changes in gene expression and shares pathobiology with neurodegenerative diseases. Neuroscience 114:529^546 Williams JM, Felten DL, Peterson RG, O’Connor BL 1982 E¡ects of surgically induced instability on rat knee articular cartilage. J Anat 134:103^139 Woolf CJ, Salter MW 2000 Neuronal plasticity: increasing the gain in pain. Science 288:1765^ 1769 Zimmermann M 2001 Pathobiology of neuropathic pain. Eur J Pharmacol 429:23^37

DISCUSSION Schaible: Your results with the nerve cuts seem to be at variance with the earlier results of Levine. Is that true? He claimed that cutting of nerves will make arthritis better. Henry: Yes. I can’t account for his results. Blake: You don’t have to. In a clinical setting, all of us who have tried to reproduce this have got it to swing in both directions. The most common observation is that it reduces it. Likewise, in rheumatoid arthritis (RA) in a clinical setting, if someone has an ulnar nerve lesion they do not get the RA in

146

DISCUSSION

that area, by and large. Everything depends on the time of the injury. Lesions can suppress it prior, aggravate it after, and mild switches which we don’t understand can £ick it one way or the other. Henry: I agree: the timing is critically important. Patients who have stroke with RA are a good example. We have to make sure that we don’t see the peripheral changes as being due to sensory neurons only. In stroke the sensory neurons aren’t altered, it is the motor neurons and autonomic neurons. We should avoid taking a simplistic approach. Kuettner: I was surprised to see your pictures of the receptor for substance P on chondrocytes, calci¢ed cartilage and newly formed bone. Do you have any idea of the function of substance P speci¢cally during cartilage remodelling? Do you see that the collagen breakdown products are derived from type 2 collagen or are they gelatin? Henry: They are type 2. Those slides were from RA animals, so this suggests that there is probably an up-regulation of the receptors in those tissues. Keep in mind also that despite the fact that there is an abundance of degrading enzymes £oating around, in this area there are fewer enzymes to break down substance P, which means that it can then di¡use to other places. The issue of sprouting of neurons was raised yesterday, and whether this could account for the pain of OA. I don’t know, but if substance P can di¡use from one site to another, then you don’t have to have the hard wiring because of the di¡usion. I’ll be able to tell you more next year. Herzog: Coming back to cutting the nerve, you were talking primarily about trophic e¡ects and the release of chemicals from the nerve. How do you distinguish this from all the events that happen mechanically when the nerve is cut, such as muscle function loss? Henry: We haven’t yet. But by giving the antagonists we can eliminate the sensory neurons without altering gait. Grubb: This experiment depends on when and how you cut the nerve. The temporal aspects are important. There are two interpretations. If you cut the nerve you could say that you are getting rid of the sensory nerves and any things they might be protecting. But also if you cut the nerve there will be a massive release of substance P for a short period from that dying nerve terminal. Instead of getting rid of the nerve this may actually exacerbate nerve activity transiently. Henry: We made our measurements well after the cutting of the nerve. Mackenzie: With regard to the discrepancy between the Levine experiments and yours, I think the models are actually di¡erent. If I remember correctly, the Levine experiments were in adjuvant polyarthritis where they were looking at the development of the secondary lesions which ¢rst occur about 10 or 12 days after injection of adjuvant. Your experiments involve the injection of adjuvant directly into one joint.

ANIMAL MODEL OF OSTEOARTHRITIS

147

Henry: Ours is a model of mono-arthritis a¡ecting one joint. Rediske: Have you looked to see whether chondrocytes are also a source of substance P? Henry: No. Evans: A recent paper showed that they are (Millward-Sadler et al 2003). They have the capacity to make it under the right conditions. Pisetsky: I have a question about this model, and trying to distinguish e¡ects of local versus systemic in£ammation. Even if you are putting complete Freund’s adjuvant (CFA) in a joint, it is likely to be doing a lot of things peripherally, given what it contains. How do you distinguish e¡ects that are local from those on general in£ammation? Henry: We have looked contralaterally and we don’t see any evidence of joint destruction. Pisetsky: There are local problems plus systemic problems. You wouldn’t see contralateral joint damage, but these animals probably have high levels of TNF and other cytokines. How do you distinguish the systemic e¡ect which may be acting on the brain from the local e¡ects? Henry: We don’t. Until we have these data I am assuming that there are going to be systemic e¡ects beyond the local joints. The way I see the biological system, this has to be occurring. But this is also why I was alluding earlier to common aetiology. I would expect there to be other e¡ects. Pisetsky: I am just trying to ¢gure out how you extrapolate this to an OA model, where the amount of in£ammation is much less. Felson: Can you comment on the OA model? Do you have some preliminary data from that? Henry: Yesterday I also raised the question of whether OA is a systemic disease. It is very early for us to comment on our results, but we have seen an osteophyte contralaterally in our OA model. This stunned us. I can’t explain this. Fernihough: The model has comparisons with Geza Pap’s overuse model (Pap et al 1998). This is an established model of OA and has histological features of the disease. In conjunction with pain, the importance of time has been discussed. How does the pain sensation change with time? Do all of your animals share a common range of pain sensation? Henry: I can’t comment on the OA model. In the RA model, however, there was a very consistent reduction in pain that started immediately. By day 2 and 3 it was down to baseline and was stable after that. So there is a very rapid change in sensitivity of the nociceptive mechanisms. There wasn’t a lot of variability from one animal to another. Fox: There will be some di⁄culty in interpreting the behavioural data from animals. This is a perennial problem. We have been doing similar models of RA and Janet Fernihough has produced some beautiful histology showing

148

DISCUSSION

changes in the knee. There are changes in nociceptive markers in DRGs and such like. This is ongoing. But behaviourally it is very di⁄cult to distinguish nociception changes compared with changes in movement, just because the joint is now unstable. This will be a real challenge. I will be interested to see how you intend measuring this. Henry: We don’t intend to do it in isolation, for these very reasons. Van den Berg: I have some di⁄culty with the discussion, linking your ¢ndings to either an RA or OA model. What you have done in these acute models is simply making in£ammation. Whatever stimulus you use, you will probably get the same kind of data out of it. Whether you use Freund’s or something else, it doesn’t really matter: you are making an in£amed site somewhere in the body. The only thing you can learn from this is what in£ammation is doing in terms of feedback on your pain system. In the discussion about cutting the ligament, it might be critical whether you cut the ligament before or after the nerve. Do you recall what you did in this situation? Dieppe: I wanted to pick up on your throwaway remark about the bilateral osteophyte. It is quite a common experience in animal models of arthritis to see something on the contralateral side if you look for it. This relates a bit to our earlier discussion about contralateral pain representation and spreading. I’d like to hear from the neurophysiologists and pain people what they think is going on with these bilateral e¡ects on animal models of arthritis. Henry: There are two elements. One is mirror pain, which is not uncommon. The osteophyte on the other side could simply be due to changed gait. It is not necessarily a systemic disease, but we can’t eliminate the possibility of it being a systemic disease. Dieppe: It probably is some sort of systemic e¡ect. If you do an in£ammation model you will see synovitis the other side. Henry: We haven’t seen any evidence of synovitis on the other side in our RA model. Blake: You will if you look carefully enough and in the right genetic strain. Henry: We have been using Sprague-Dawley. Blake: If you use Wistars you will see it. Contralaterality in these systems is almost invariable providing that you have the right tool and you know what you are looking for. It goes back, as Paul is alluding to, to a lot of the basic clinical observations made on selective upper motor neuron lesions. The people who really understand this are the ophthalmic surgeons who deal with sympathetic uveitis. This is where someone injures one eye and in the worst cases they can end up going blind in both. They have techniques where they can show that this mirroring occurs in almost every instance. Blindness is not 100%, but electrophysiological disturbance in the other side is.

ANIMAL MODEL OF OSTEOARTHRITIS

149

Lohmander: I wanted to follow up on Paul Dieppe’s comment and point out that we see similar ¢ndings in our corresponding human post-injury models of OA. In these patients we have a higher than expected rate of OA in the other knee, although patients claim that they haven’t injured it. Of course, there are issues about overloading of the other, un-injured, side, but my gut feeling is that there is more to it than that. Felson: Ted Pincu has tried to improve the e⁄ciency of OA clinical trial outcome measurement by asking about pain in the signal knee, and avoiding having to ask about the other joints. He found that it was more e⁄cient to ask about both knees because patients entering the trials often switched side where the knee was most painful. Having completed a longitudinal study of knees, we had to keep track of which side we got the initial MRI on. Subjects would come back for repeated MRIs and they wouldn’t remember which side they had had the initial MRI on. Grubb: I have a more direct answer to Paul Dieppe’s question about the contralateral issue and the hard wiring of the neurons. There is no doubt that if you record from wide dynamic range neurons in rat spinal cord (these are likely present in the deep dorsal horn) that they can have quite large receptive ¢elds on the ipsilateral limb. But even in some normals, occasionally you can ¢nd a hard wired input from the other side, the contralateral limb. We don’t see many of these far away from the region where we are trying to identify the receptive ¢eld. Most of the neurons we have looked for have the centre of the receptive ¢eld around the ankle joint, because we were interested in the rat ankle joint. But occasionally they had receptive ¢elds running right up the limb, and just occasionally we got a neuron that responded in a normal animal to squeezing the paw on the other side. But when we induced the in£ammation peripherally on this ipsilateral side, the number of neurons responding to contralateral pressure increases to about 18^ 20%. There are ¢bres that go across, into the deep dorsal horn on the other side, and have an in£uence on those deep dorsal horn neurons. The fact that the central neurons become a bit more sensitive and a bit depolarized by the constant a¡erent input from the in£amed region means that they then respond more to these contralateral inputs because they are more sensitive. Central sensitization has occurred, but because the contralateral inputs are there, their in£uence becomes more obvious and you see more cells responding to that stimulus. Henry: However, some but not all of the contralateral e¡ects can be due to crossed projections and sprouting of primary a¡erent or even second-order neurons. It is interesting that in some models there are even changes in expression in DRG cells contralaterally. There are changes in mRNA levels, for example, on the contralateral DRG. Schaible: We have done a similar study. One has to make a distinction between the segmental bilateral increase and the systemic increase. A good control might be to study the ganglion at di¡erent levels.

150

DISCUSSION

Pisetsky: I have a question about the e¡ects of cytokines on nociception in these models. Clinically, if we give people cytokines, they feel sick. How much of this is central? Should we consider this in these models? Schaible: I mentioned earlier Linda Watkins who has talked a lot about sickness response. This implies that interleukins will act in the CNS to organize a coordinated response to a noxious stimulus. So there is some good evidence from her studies that these cytokines can induce a general sickness response. We don’t know whether that is happening in these models. She can evoke such a sickness response with LPS injection and so on, and she can demonstrate pathways involved. Kidd: One of the problems with the cytokine story is that the blood^brain barrier is impermeable to many cytokines. Linda Watkins was hypothesizing the role of the vagus nerve for a long time, but it is now becoming apparent that in situations of in£ammation the blood^brain barrier can break down and it becomes much more permeable to cytokines. It might be that in an in£ammatory situation there is an entirely di¡erent physiology, and under these circumstances IL6, IL1 and TNF may gain access to the brain receptors. Henry: Yes, the blood^brain barrier should not be seen as a ¢xed barrier. It is under functional control. Malcangio: In any case, sensory neurons can produce IL and TNF themselves, so the cytokines don’t need to cross the blood^brain barrier. And in some other models microglia produce cytokines within the spinal cord. Jim Henry, you talk about the ‘good guy’ that might be released by sensory neurons. Do you think that somatostatin could be protective? Henry: We haven’t looked at somatostatin, but it was a surprise to us that neurokinin A (NKA) seems to have opposite e¡ects to substance P. Malcangio: By which receptor? Henry: We haven’t looked at that yet, but it could be the NK-2 receptor which is in peripheral tissues. The reason why this is surprising is because substance P and neurokinin A are derived from the same precursor. So you have one neuron releasing both a good guy and a bad guy from the terminal, which suggests that the regulatory controls
whether the release is having a bene¢cial e¡ect or a damaging e¡ect to peripheral tissues
depend on the enzymes that are present breaking them down. Malcangio: This is in the periphery, not centrally. Henry: That is correct. Mackenzie: There was an intriguing series of experiments published some years ago from Sergio Ferreira’s laboratory (Ferreira et al 1988, Cunha et al 1992) in which minute quantities of TNF and IL1 were injected into rat footpads. He then looked at pressure hyperalgesia. He demonstrated that not only was there hyperalgesia in the injected paw, but also in the contralateral paw. The quantities

ANIMAL MODEL OF OSTEOARTHRITIS

151

used were so tiny that it is very unlikely that this is a systemic e¡ect. The only conclusion that can be drawn is that this was a neuronal e¡ect. The ¢ndings have been reproduced in a number of laboratories, so I think it is a genuine phenomenon that has never been fully explained mechanistically. It might be important in terms of sensitization through the release of cytokines. Henry: Cytokines are obviously one of the areas on which we will be focusing. I can add that we have an animal model of neuropathic pain where the animals become extremely sensitive. We get a contralateral e¡ect there also. Rediske: Your OA model has a lot of promise. It underscores some of the designer nature of OA models, because you build in two instabilities plus chronic loading. Instead of building in those kinds of stimuli, have you ever thought of building in a chronic neurogenic or neuropathic component into a destabilization model? Henry: It is interesting that you ask about a neuropathic element. Part of me as a neurophysiologist wonders how much of the pain of OA is neuropathic pain. I don’t know. As we develop this model we are going to be adding other things. We can address indirectly the issue of neurogenic in£ammation by giving antagonists long-term to some of these chemicals that we believe are participating in neurogenic in£ammation. Substance P is the ¢rst one we will do because it seems to be the most obvious. Fox: Thinking about extrapolating from models to humans, preclinical models regard chronic pain as lasting a day whereas clinicians are looking at months and years. What happens to the wide dynamic range neurons in the spinal cord after the in£ammation has resolved in one of these models? Presumably in OA there is not a persistent in£ammation that might be constantly driving the input to the spinal cord, yet we are talking in terms of a prolonged central sensitization. Henry: As far as we know, our RA model goes on forever. Our model of neuropathic pain also goes on for ever: we have measured that up to 150 days. We haven’t got to this stage with the OA model yet. Evans: We were talking earlier about the contralateral e¡ect and communication through the nervous system. I want to point out that there are other ways that these joints can communicate aside from general systemic means. We have done the reverse contralateral experiment using bilaterally arthritic animals. When we inhibited the arthritis in one joint through gene transfer, the other joint got better. This seems to be connected with the tra⁄cking of cells that we are still trying to identify, but we think they are DCs that migrated from the joint we injected to the joint we didn’t inject. They may be moving under the in£uence of chemokines released from the nervous system. I am not saying that the nervous system is not involved, but it doesn’t always have to be hard wiring. Grubb: Hard wiring does exist, though. Blake: Could you block this by sciatic nerve section?

152

DISCUSSION

Evans: That is on the list of things to do. Felson: Peter Simkin, I want to ask you about the bilaterality of gout. How related is this very in£ammatory but episodic disorder? Simkin: The localization of gout, to me, is much more associated with the presence of OA. The big toe is hit because this is the bunion joint, for instance (Simkin 1977, Simkin et al 1983). OA tends to be bilateral, and I think this is what causes bilaterality when it is seen in gouty arthritis. Hunter: I wanted to get you to comment on structure rather than pain. With your OA model, where you have forced articulation, can you comment on the di¡erences that you saw between the rats that did the weight bearing versus the non-weight bearing exercise? Henry: If we go at a certain distance and time, we see more osteophytes and more bone degradation sclerosis in the articulated animals than in the surgically treated non-articulated animals. This work is so new that I can’t answer directly whether there is more at 2 months in the articulated animal than we ever see in the nonarticulated model. Pisetsky: In RA models in rodents, there is a big impact of genetics. Is there any genetics in OA models so far, in terms of the likelihood that one or another strain will develop these? Henry: My feeling is that we are sadly lacking in good OA models. Pisetsky: Are the rat strains that get adjuvant arthritis more likely to get OA in these models? Henry: There is such a lack of activity in OA models that we haven’t got to that stage yet. Van den Berg: There is a lot of genetics in spontaneous OA. However, when you use an instability model you can cause OA in any animal, and the genetic background is not that relevant. Pisetsky: Unless the in£ammatory component has more of a role in the disease. Lohmander: We have some recent data which are from a repeat of a study that Paul Dieppe was involved in about 20 years ago. This was the ¢rst published study in the area and when we publish ours, it will be the second! We showed that if you have hand OA as a marker of a genetic trait of OA, your post injury OA will be worse and more frequent than if you don’t have hand OA. This shows that genetics does feed into human post-injury OA. Dieppe: It is nice to hear that this study has been repeated. I was worried about it because our early report contained statistical £aws! Jim Henry, you said you worry about how much OA pain might be neuropathic. Perhaps there is another clinical model we should be thinking about which might get us some way to that. We have been doing quite a lot of work with surgical colleagues on joint replacement. For most people, pain disappears after joint replacement, but there are about 10% of people where it doesn’t. This is

ANIMAL MODEL OF OSTEOARTHRITIS

153

fairly consistent across all the studies that the pain stays in this 10% even though the joint has been replaced. Could those be the people who have the neuropathic pain? Henry: That is one of the two possibilities that come to mind. The other is phantom limb pain. Dieppe: They might be quite an interesting group to study in terms of our understanding of pain in OA. Grubb: Have treatments for neuropathic pain been tried in OA? Dieppe: Not that I know of. Blake: There is a third explanation given to me by a knee surgeon, that there was a form of CRPS (complex regional pain syndrome) developing in those knees. Dieppe: I have never seen re£ex sympathetic dystrophy in that situation. Blake: They are much more aware of it than we are, and they are not using the same clinical criteria that we are using. One thing that is for certain is that the pain isn’t necessarily the same as it was before the operation. Fox: Do you know that there hasn’t been some sort of neuronal activation just because of the operation? Blake: That is very rare. You can usually dissect these from a decent history and examination. Ordeberg: It is a multifactorial problem. We see change in pain even in those patients who bene¢t from the operation. Going back to my presentation, there were patients who had sensory aberrations. These are patients where the analgesia is no longer su⁄cient in the treatment. The pain is not purely nociceptive: it may be neuropathic or neurogenic. I am mainly a spine surgeon. Among those patients with failed hip arthroplasties there are quite a few who have L3 or L4 root pain. This seems fairly natural as the distribution of L3 and L4 root pain is very much the same as in pain from hip osteoarthritis. References Cunha FQ, Poole S, Lorenzetti BB, Ferreira SH 1992 The pivotal role of tumour necrosis factor alpha in the development of in£ammatory hyperalgesia. Br J Pharmacol 107:660^664 Ferreira SH, Lorenzetti BB, Bristow AF, Poole S 1988 Interleukin-1 beta as a potent hyperalgesic agent antagonized by a tripeptide analogue. Nature 334:698^700 Millward-Sadler SJ, Mackenzie A, Wright MO et al 2003 Tachykinin expression in cartilage and function in human articular chondrocyte mechanotransduction. Arthritis Rheum 48:146^156 Pap G, Eberhardt R, Sturmer I et al 1998 Development of osteoarthritis in the knee joints of Wistar rats after strenuous running exercise in a running wheel by intracranial selfstimulation. Pathol Res Pract 194:41^47 Simkin PA 1977 The pathogenesis of podagra. Ann Intern Med 86:230^233 Simkin PA, Campbell PM, Larson EB 1983 Gout in Heberden’s nodes. Arthritis Rheum 26: 94^97

Osteoarthritic Joint Pain: Novartis Foundation Symposium 260. Volume 260 Edited by Derek J. Chadwick and Jamie Goode Copyright ¶ Novartis Foundation 2004. ISBN: 0-470-86763-9

Phantoms in rheumatology C. S. McCabe, R. C. Haigh*, N. G. Shenker, J. Lewis and D. R. Blake1 The Royal National Hospital for Rheumatic Diseases, Upper Borough Walls in conjunction with The Department of Medical Sciences and The Department of Pharmacy and Pharmacology, University of Bath, Bath BA1 1RL, and *Royal Devon & Exeter Hospital (Wonford), Exeter EX2 5DW, UK

Abstract. This paper examines rheumatology pain and how it may relate to amputee phantom limb pain (PLP), speci¢cally as experienced in rheumatoid arthritis, ¢bromyalgia and complex regional pain syndrome (CRPS). Clinical ¢ndings, which suggest cortical sensory reorganization, are discussed and illustrated for each condition. It is proposed that this sensory reorganization generates pain and altered body image in rheumatology patients in the same manner as has previously been hypothesized for amputees with PLP; that is via a motor/sensory con£ict. The correction of this con£ict through the provision of appropriate visual sensory input, using a mirror, is tested in a population of patients with CRPS. Its analgesic e⁄cacy is assessed in those with acute, intermediate and chronic disease. Finally, the hypothesis is taken to its natural conclusion whereby motor/sensory con£ict is arti¢cially generated in healthy volunteers and chronic pain patients to establish whether sensory disturbances can be created where no pain symptoms exists and exacerbated when it is already present. The ¢ndings of our studies support the hypothesis that a mismatch between motor output and sensory input creates sensory disturbances, including pain, in rheumatology patients and healthy volunteers. We propose the term ‘ominory’ to describe the central monitoring mechanism and the resultant sensory disturbances as a dissensory state. 2004 Osteoarthritic joint pain. Wiley, Chichester (Novartis Foundation Symposium 260) p 154^178

Pain is the predominant complaint of patients with a rheumatological condition. It may be intermittent or continuous and vary in nature depending on the cause and course of the disease. In the majority of cases clinical ¢ndings provide supporting evidence for the source of this pain such as swollen joints in rheumatoid arthritis (RA) or bony overgrowth in osteoarthritis (OA). However, there are some conditions in rheumatology where a patient’s pain cannot be matched to physical ¢ndings or relieved by traditional therapeutic measures. It is pain of this nature that this paper addresses, speci¢cally the types of pain experienced in rheumatoid 1This paper was presented at the symposium by D. R. Blake to whom correspondence should be

addressed. 154

PHANTOMS

155

arthritis (RA), ¢bromyalgia and complex regional pain syndrome (CRPS), and how these may relate to amputee phantom limb pain.

Amputee phantom limb pain Phantom limb pain (PLP) is a phenomenon that occurs in approximately 70% of patients after amputation (Jensen et al 1985). For the amputee ‘these pain memories are vivid, perceptually integrated experiences which incorporate both emotional and sensory aspects of the pre-amputation pain’ (Hill et al 1996). Tingling is the most common complaint but pins and needles, shooting, burning or crushing pain have all been reported (Melzack 1971). Phantom sensations are also described by 90^100% of amputees (Melzack 1990). Post-operatively an amputee will perceive a phantom limb that has all the same sensations and mobility of the real limb prior to amputation and is so strikingly real to the individual that it feels an integral part of them. The phantom appears to ‘inhabit the body’ (Melzack 1990) when the eyes are open and moves appropriately with other limbs. It initially feels perfectly normal in size and shape but may alter over time so that the phantom gradually becomes less apparent and may eventually fade away (Katz & Melzack 1990). For those people who wear a prosthesis the phantom limb can appear to ¢ll it or telescope up into the remaining stump (Melzack 1990). It has been proposed that it is a combination of the duration and intensity of such pre-operative pain that determines whether long-term central nervous system processes are altered with resulting persistent phantom sensations (Katz & Melzack 1990). A long-lasting mild sensation such as a watch on a wrist or a sock on a foot may be just as e¡ective at developing somatosensory memories as the intense short-term pain of gangrene. Referred sensations (RS) have also been described in amputees. These are somatosensory feelings that are perceived to emanate from a body part other than but in association with the body part being stimulated. They have not only been reported following limb amputation (Ramachandran et al 1992), but also somatosensory dea¡erentation (Clarke et al 1996), local anaesthesia (Gandevia & Phegan 1999), stroke (Turton & Butler 2001) and spinal cord injury (Moore et al 2000). Collectively these studies have shown that the referred sites (the body part not physically touched) are non-random and often closely correspond to the cortical topographical map representing the body structure ¢rst described by Pen¢eld and Rasmussen (1950). In the case of an amputated upper limb, patients report sensation in their phantom when parts of the face are lightly stroked (Ramachandran et al 1992). This is thought to be because the hand is positioned adjacent to the face on Pen¢eld’s map. These aberrant somatosensory, but reliable sensations were interpreted as resulting from central sensory reorganization

156

McCABE ET AL

following disconnection or dysfunction of sensory pathways (Ramachandran et al 1992). In conclusion, amputees report a variety of sensations that are not supported by conventional notions of clinical pathology. The nature of the sensations described, provide the ¢rst link to the rheumatology patient with unexplained pain. Pain in rheumatology Rheumatoid arthritis Rheumatoid Arthritis (RA) a¡ects one per cent of the population and is a chronic disabling disease which occurs two thirds more frequently in women than men (Walker 1995). The peak age of onset is between 40 and 50 years, its aetiology is uncertain and there is, as yet no cure. The main symptoms of this disease are pain, sti¡ness, fatigue and joint swelling but other organs in the body may also be involved (Gordon & Hastings 1995). The pain that these patients experience is ‘chronic, unpredictable and frequently severe’ (Parker et al 1989) and combined with joint destruction results in progressive disability over time. A key feature of RA is a pattern of remissions and £ares that are a result of the £uctuations in disease activity. During these £ares the joints, particularly the small joints of the hands and feet, become swollen and tender. This swelling is due to increased activity in the joint caused by an inappropriate in£ammatory response. As a result of prolonged or frequent episodes of this in£ammation the synovium lining the joint, becomes permanently thickened and bony erosions may occur (Gordon & Hastings 1995). Pain is synonymous with the disease of rheumatoid arthritis and the types of pain that su¡erers of this disease experience are complex and varied. The descriptions that they use may alter depending on the time of day, the duration of their disease, the joints that are involved and whether those joints are moving or at rest (Papageorgiou & Badley 1989). A less well-reported quality of pain that some RA patients describe is where they feel their joints to be excessively more swollen than they look. They describe all the sensations associated with swollen joints but they are clinically not swollen and indeed when the subject looks at the a¡ected joints they too are aware that they are not swollen (Blake et al 2000). Interestingly this perception of swelling is not isolated to the joints, the patient will report that they feel their whole digit to be a¡ected (Fig. 1). These sensations are similar to the e¡ects you may have after an injection in your mouth at the dentist. The anaesthetic leaves you feeling that your lip is huge and yet you look in the mirror and ¢nd that it is actually its normal size.

PHANTOMS

157

FIG. 1. Rheumatoid arthritis patient’s drawing of ‘phantom swelling’ a¡ecting their hands. The shaded areas depict perceived swelling over the joints and the outer lines, perceived swelling of the digits.

The characteristics of this ‘phantom swelling’ and how it di¡ers from routine reports of RA joint swelling, were identi¢ed in a cross-sectional study involving 10 patients with RA (McCabe 1999). Five of the subjects reported ‘phantom swelling’ and ¢ve did not. The two groups did not di¡er signi¢cantly in age, disease duration or disease activity, as measured by in£ammatory markers and joint activity. Using a modi¢ed McGill Pain Questionnaire (MPQ), each subject was asked to describe the sensations they currently experienced in all their joints at rest and on movement. A semi-structured interview was used to collect additional information on duration and severity of disease in each joint and the impact of vision on the sensations that they reported. The subjects with ‘phantom swelling’ reported that their a¡ected joints felt excessively hot (‘burning’, ‘scalding’) and hugely swollen (‘massive’). Their remaining RA-a¡ected joints were described in exactly the same manner as the control group described theirs, ‘warm’ and ‘slightly pu¡y’. When the phantom swollen joints were viewed by the subjects the perception of swelling disappeared but the lesser sensation of ‘slight pu⁄ness’ in their other joints remained on visualization. Phantom swelling was only present in those joints that had been most severely a¡ected by RA and for the longest duration which

158

McCABE ET AL

is very reminiscent of Katz and Melzacks’ theory that it takes a certain duration and intensity of pain to alter central processing resulting in persistent sensations. Interestingly the nature and cause of sti¡ness in RA, another pain related symptom, is not well explained, even though it is a well established and de¢ning symptom of the disease (Arnett 1988). Objective measures of sti¡ness do not relate to the subjective experience and indeed, compared with non-arthritic controls, objective sti¡ness can be reduced in RA joints (Helliwell et al 1988). We therefore hypothesised that the central nervous system is capable of generating a feedbackdependent state which can result in pathological sensations such as pain and sti¡ness in RA, that are to some extent independent of the initial peripheral pathology. We sought clinical evidence to support this proposal by investigating the clinical presentation of perceived sti¡ness in RA patients who had undergone limb amputation but nevertheless retained an experience of a phantom limb (Haigh et al 2003). Three patients with a current diagnosis of RA and lower limb amputation were identi¢ed from the local Arti¢cial Limb Centre database and investigated to determine the nature and pattern of pain and sti¡ness in their phantom and intact limb. In addition to standard physical examination, pain and sti¡ness severity were measured using visual analogue scales (VAS) for both limbs. The duration and timing of sti¡ness was also recorded for each limb. In all three cases, the pattern of perceived RA sti¡ness was similar for the intact and phantom limb. All three patients described sti¡ness in their phantom limb which mirrored that of physical RA joint symptoms in terms of quality, frequency, diurnal variation, location, distribution and response to medication (non-steroidal anti-in£ammatory drugs, corticosteroid, opiate and disease-modifying drugs). Unilateral exercise (or attempted exercise) relieved sti¡ness only in the limb being exercised. Thus, the extent to which the subjective experience of perceived sti¡ness could be dissociated from the assumed original peripheral source was strikingly illustrated in RA patients with phantom limbs. Accordingly, we proposed that the experience of peripherally located sti¡ness results from impairment to central brain processes. Conditions are present in RA to produce inaccurate sensory information which may lead to con£ict with planned output from motor systems. These include peripheral and central proprioceptive abnormality, cortical reorganization, neurogenic in£ammation and circulating cytokines with central e¡ects. Such con£ict of information is ultimately perceived as ‘sti¡ness’ by the patient with RA. RA is not the only rheumatological condition where phantom swelling and sti¡ness are described. Clinical experience has long shown that patients with ¢bromylagia also report sti¡ness and perceive body areas to be subjectively swollen when objectively they are not.

PHANTOMS

159

Fibromylagia Fibromyalgia (FMS) is a chronic pain condition where su¡erers report widespread pain, fatigue and psychological distress all of which have a major impact upon their daily lives (Wolfe et al 1990). Although hyperalgesia and allodynia are commonly reported at speci¢c trigger points these sensations often spread far beyond these areas with su¡erers describing generalized sensitivity (Staud et al 2001). For the majority of patients there is no known initiating event or observable physical pathology and symptoms are frequently resistant to therapeutic initiatives. In addition to the symptoms described above it has long been observed, but only recently systematically recorded, that these patients also experience phantom swelling sensations in the same manner as those with RA (C. McCabe, D. Blake, unpublished work). The sensation most commonly a¡ects the hands, bilaterally from the wrist to the ulna styloid, or the feet, bilaterally from the toes to the ankle joints. The subject is most aware of the perceived swelling when they have their eyes closed and it decreases or disappears completely when they view the a¡ected area. With regular viewing on a daily basis the sensation can be diminished permanently. When phantom swelling is reported it is commonly associated with the patient feeling that they are clumsy or less aware of where their limbs are in space. This reduction in limb position sense will be discussed further towards the end of this paper. Phantom swelling in FMS is a clear example of a sensation being reported without supporting underlying clinical pathology and CRPS is another such condition where the cause of the characteristic symptomology is ambiguous.

Complex regional pain syndrome Complex regional pain syndrome (CRPS) is a painful, debilitating condition. This diagnostic term embraces several syndromes including re£ex sympathetic dystrophy, causalgia, and algodystrophy. The pain that a patient with CRPS will report shares many similar characteristics to amputee phantom limb pain: mislocalized, intense and burning. Clinical features include sensory disturbances such as burning pain with allodynia and hyperanalgesia, motor disturbances such as weakness, tremor and muscle spasms, and changes in vascular tone, temperature and oedema (Scadding 1999). Over time functional loss and trophic changes may occur. The syndrome can occur spontaneously or following trauma (CRPS Type 1) or in association with peripheral nerve damage (CRPS Type 2). A characteristic feature of CRPS is that signs and symptoms spread beyond the site of initial insult. Severe pain may occur seemingly out of proportion to the original pathology. It may persist over long periods and is frequently resistant to a wide range of treatments. Theories abound on the cause of this pain and its

160

McCABE ET AL

underlying pathology. Traditionally, interrupting the sympathetic supply to the painful area was thought to treat such pain. However, the e¡ectiveness of this approach is not supported by randomised controlled trials (Jadad et al 1995). Neural plasticity occurs in a variety of pain syndromes (Harris 1999, Lenz & Byl 1999). We predicted that referred sensations would be present in patients with CRPS type 1 as evidence of sensory cortical reorganization. The resultant sensory mislocalizations could then provide the inappropriate sensory feedback required to create painful sensations (McCabe et al 2003a). Furthermore, we hypothesized that these referred sensations would be perceived to emanate from the body structures immediately adjacent to the stimulated site and in keeping with their topographical location on the Pen¢eld homunculus as in phantom and allied pain states. We speci¢cally selected those patients with CRPS Type 1 as we wished to discover whether central reorganization occurs even where there is no evidence of local peripheral nerve damage. Over two years, 16 subjects (13 female, 3 male) who met the entry criteria were recruited. Five showed evidence of referred sensations (Table 1). There was no di¡erence in age, disease duration, levels of pain, or severity of disease (Table 2) between those who presented with RS and those who did not. All ¢ve patients reported referred sensations during examination with their eyes closed (Fig. 2). They were experienced in real time and disappeared when stimulation ceased or vision was permitted. When the subjects viewed the area being touched the sensations were either diminished (Case 5) or not present and when the symptoms of CRPS resolved (Cases 1, 2 and 4), referred sensations were lost. Sensations were referred in a modality-speci¢c manner with touch referred in all cases and pinprick also referred in two (Cases 1 and 2). Vibration was never referred. All referred sites were located on body parts immediately adjacent, on Pen¢eld’s homunculus, to the stimulated site. The location of the referred sites, in our study population, was consistent with previous reports in other pain conditions (Ramachandran et al 1992, Flor et al 1997) and ¢t particularly well with predicted cortical changes that have been shown to occur within the somatosensory body map in amputees (Halligan et al 1993). Ramachandran (Ramachandran & Hirstein 1998) proposed that due to the location and speed with which referred sensations occur in amputees, such ‘ectopic representations’ following functional remapping were probably due to the unmasking of latent synapses within the cortex, as previously described in primates (DeFelipe et al 1986, Jones 1990). These synapses are suppressed when there is simultaneous input from two connected receptors but with reduced or impaired sensory activation in one area, the connection becomes disinhibited. Recent imaging studies, using magnetoencephalography, in six patients with upper limb CRPS Type 1 have also shown changes in the cortical somatosensory map though it was not reported whether these were associated with referred

PHANTOMS

161

TABLE 1 Location, direction and type of sensations referred in subjects with a diagnosis of CRPS Type 1 (Cases 1^5). Loss of detection of referred sensations is shown in relation to current disease duration and future status (resolved or chronic)

Patient

Pain site

3 weeks L 3rd ¢ngertip (1) L lower jaw (2) Left 8 weeks L forefoot Case 2 (1) 34 years F ankle L patella (2) (Fig. 2) Case 3 Left 3 years L patella (1) 24 years M knee L forefoot (Fig. 3) (2) Case 4 Right 6 years R forefoot 41 years F foot (1) (Fig. 4) R patella (2) Case 5 Left 4 years L shoulder 57 years F hand (1) (Fig. 5) L ear (2) L hand (3) L cheek (1) (Fig. 6) L hand (2) Case 1 28 years F (Fig. 1)

Left hand

Disease duration

Area touched (1) Referral site (2,3)

Loss of referred sensation

Resolution of CRPS (wks)

Direction of referral

Type of sensation

1^2

Light touch 3 weeks and pinprick

6

1^2 and 2^1

Light touch 3 weeks and pinprick

4

1^2 and 2^1

Light touch

No

Chronic

2^1

Light touch

4 weeks

Chronic

1^2 1^3

Pulling, light touch and hand movement

No change

Chronic

1^2 Light touch

sensations (Juottonen et al 2002). There was a signi¢cantly shorter distance between the areas representing the thumb and little ¢nger on the somatosensory cortex contralateral to the a¡ected limb than the ipsilateral side. Interestingly, there was no signi¢cant correlation between the distance of thumb and ¢nger and the level or duration of pain. Hand dominance was also not an in£uencing factor. Alternatively, referral of sensations may occur at the spinal level. A large body of evidence shows that sensitization of wide dynamic range neurons at level V of the dorsal horn results in ipsilateral and contralateral enlarged receptive ¢elds which do not rely on a cortical homunculus (Ji & Woolf 2001). In addition, experimental models of peripheral neuropathic pain demonstrate bilateral spinal cord changes after unilateral nerve damage (Koltzenburg et al 1999). However, all of our patients had CRPS 1, so therefore had no precipitating neural trauma. Their sensations were not referred bilaterally, either from the stimulated site to its

162

McCABE ET AL

TABLE 2 Demographics of total study population to compare di¡erences in age, disease duration and levels of pain between subjects who experienced referred sensations and those who did not

Case

Age

Gender

Disease duration

1** 2** 3** 4** 5** Mean** 6 7 8 9 10 11 12 13 14 15 16 Mean

28 years 34 years 24 years 41 years 57 years 36.8 years 38 years 35 years 40 years 38 years 27 years 51 years 68 years 54 years 38 years 22 years 59 years 42.7 years

F F M F F 4F:1M F F F F M F F M F F F 9F:2M

3 weeks 8 weeks 3 years 6 years 4 years 2.6 years 6 weeks 5 months 1 year 3 years 2 years 2 years 1 year 4 years 7 years 4 years 10 years 3.1 years

A¡ected limb Left hand Left ankle Left knee Right foot Left hand Left ankle Right arm Right arm Left leg Left leg Right arm Left arm Left foot Left foot Left foot Left foot

Pain level on movement at presentation 8 8 8 9 5 7.6 9 5 6 5 8 7.5 5 9 10 9 9.5 8

contralateral partner (i.e. left hand to right hand) or mirrored on the contralateral side (i.e. from stimulated site to referral site on the una¡ected limb). In addition, the speed of referral (both in terms of disease duration, response time on stimulation and resolution as the condition improved), combined with the magnitude of the sensations all detract from a purely spinal route. Contemporary theories suggest that CRPS is a disorder involving both CNS and peripheral nervous system components (Baron et al 2002, Janig & Baron 2002). This is based on the

FIG. 2. Artist’s impression of Cases 1^5 illustrating location of stimulus and direction of referred sensations (area touched ¼ 1, referred site/s ¼ 2, 3). (a) to (d) correspond to Cases 1 to 4, (e) & (f) correspond to Case 5. Shaded area (1) depicts area stimulated by examiner, shaded areas (2) & (3) depict where referred sensations were felt. The arrows illustrate direction of referral. Reprinted with permission from McCabe et al (2003b).

PHANTOMS

163

164

McCABE ET AL

evidence that some patients respond positively to sympathetic blockade, thereby implicating involvement of the sympathetic nervous system but conversely, sympathetically maintained pain involves the deep somatic tissue (as demonstrated by our patients’ report of increased pain on movement) which is the domain of the autonomic system. Therefore isolating one clear route for referred sensations is at present problematic. The power of vision and mirror visual feedback Visual feedback strongly in£uences the experience of referred sensations in patients with CRPS. Recent studies have shown this also to be the case in amputees where stimulation of the intact limb evoked sensory changes in the phantom only when the subjects’ eyes were closed (Hunter et al 2003). Touch and vision are inextricably linked. Touch is known to in£uence vision such as dispelling the visual illusion of a three-dimensional object when it is drawn on a £at surface. Equally, in some clinical conditions such as somatosensory loss after stroke, visual feedback of the a¡ected limb during testing can signi¢cantly improve reported perception (Halligan et al 1997). In addition, recent ¢ndings by Taylor-Clarke et al (2002) showed that the enhancing e¡ect of vision modulated somatosensory cortical processing. Gregory (1998) points out that vision evolved from the simpler processes for touch and that it is possible that the somatosensory map is inverted (the feet above the hand) in order to correspond with the inverted visual image on the retina. This ensures that the link between vision and touch is as short as possible. Consequently, when our subjects viewed their limbs being stimulated it would appear that the more powerful sense of vision overruled the referred sensations. It has already been stated that vision is able to dismiss the sensation of phantom swelling in RA and FMS but in recent studies on PLP vision has been shown to also provide an analgesic bene¢t. Ramachandran & Rogers-Ramachandran (1996) superimposed the image of amputees’ normal limbs, by means of a mirror, on the space that their phantom limbs occupied. Viewing the mirror image of their residual limb, the amputees moved their normal limbs and attempted to move their abnormal side. Subjects reported that sensation in their abnormal limb returned towards normal during the exercises and their pain diminished. Harris (1999) subsequently hypothesized that the reason for this analgesic e¡ect was that PLP is generated by a discordance in motor intention and predicted proprioceptive feedback and that when this mismatch is corrected, through appropriate visual feedback via the mirror, pain is relieved. Objective evidence of the cortical e¡ects of this mismatch was provided by Fink and colleagues (Fink et al 1999) using PET imaging and healthy volunteers. They demonstrated that when congruent and incongruent movements were performed,

PHANTOMS

165

whilst viewing only one limb in a mirror, cortical activity varied depending on the movement. When the limbs moved incongruently and yet were seen, by means of mirror imaging, to move congruently, cortical activity was unilateral, unlike visually observed congruent and actual congruent movement, where bilateral cortical activity was produced. When unilateral cortical activity occurred it was in the right dorsolateral pre-frontal cortex and it was this area that Fink and colleagues concluded was speci¢cally involved in the monitoring of con£ict between motor intention and its sensory/perceptual consequences. The existence of referred sensations in CRPS and evidence of changes in cortical representation (Juottonen et al 2002) suggest that pain in CRPS may also be driven by a mismatch between motor output and sensory input as Harris proposed for PLP. We hypothesized that if this were the case then the provision of appropriate sensory input should correct the mismatch and reduce pain. Modifying Ramachandran’s methodology for the relief of PLP, we too used a mirror to provide congruent visual feedback, from the moving una¡ected limb, to restore the integrity of cortical processing aiming to relieve pain and restore function in the a¡ected limb (McCabe et al 2003a). Eight subjects were recruited aged 24^40 years (mean 33 years) with disease duration of 3 weeks to 3 years (three subjects early disease 48 weeks, two intermediate, 5 months and 1 year and the remaining three long standing disease of 52 years). All presented with a single limb a¡ected by allodynia, hyperalgesia, reduced movement with related pain and sti¡ness, and vasomotor disturbances (Table 3). All subjects reported no relief of pain on movement when both limbs were visualized without a device or when a non-re£ective surface was viewed (Fig. 3). Indeed, movement exacerbated pain. All three subjects with early CRPS (48 weeks) reported a striking reduction in their VAS for pain, during and after visual feedback of their re£ected moving, una¡ected limb as provided by the mirror (Fig. 4). A marked analgesic e¡ect was observed within a few minutes of mirror usage, followed by an abrupt return of pain when the mirror was removed initially. With repeated usage (4^9
daily, week 1), the period of analgesia progressively extended from a few minutes to hours, requiring less mirror use over the six-week study period. At six weeks there was a reversal of vasomotor changes as measured by infrared thermal imaging, a return to normal function and no pain at rest or on movement. All three subjects felt they no longer required analgesic relief from the mirror and had stopped prior to assessment at six weeks (Case 3, week 4, Cases 1 and 2, week 6). The two subjects with intermediate disease duration, 5 months and 1 year (Cases 4 & 5), reported that the mirror immediately eased their movement related sti¡ness but there was no analgesic e¡ect in Case 5. They both reported that this reduction in sti¡ness facilitated movement and the e¡ect lasted for increasing periods after

1.1

2.0

2.7

1.9

0.5

1.4

Not performed***

2.1

6 weeks

3 weeks

8 weeks

5 months

1 year

2 years

3 years

2 years

Symptom duration

*Mean temperature di¡erence (8C) [non-painful painful limb]

7

4

7

4

0

6

7

9

8

5

8

6

5**

8

8

9

Pain Pain VAS at VAS on rest movement

Control Phase 1 Looking at both limbs (No device)

8

5

8

6

5**

8

8

9

Pain VAS on movement

Control Phase 2 Painful limb hidden

8

5

8

6

3**

2

3

2

Pain VAS on movement

Mirror visual feedback

Intervention

4

4

5

5

5

9

4

8

1

Week

4

4

5

4

4

4

4

3

2

0

0

5

5

5

3

3

3

3

0

0

0

4

4

0

3

3

4

0

0

0

4

5

0

2

2

5

0

0

0

3

4

0

0

0

6

Frequent mirror usage
per day (Duration of each treatment 10 min)

Follow up

8

5

8

1

2**

0

0

0

Pain VAS

2.6

Not performed

1.3

0.4

0.3

0.8

0.4

0.2

Unresolved

Unresolved

Unresolved

6

6

4

6

6

Mean Treatment temperature duration di¡erence (8C) (weeks)

At 6 weeks

LA, left arm; LL, left leg; RA, right arm; F, female; M, male; n.d., not done. *Region of interest constant Signi¢cant di¡erence if 40.4 8C13. ** Sti¡ness. ***Case 7 had widespread ulceration on her left leg which made thermal image interpretation impossible.

Case 1 (left leg) 38 years F Case 2 (left arm) 28 years F Case 3 (left leg) 34 years F Case 4 (right arm 35 years F Case 5 (right arm) 40 years F Case 6 (left leg) 24 years M Case 7 (left leg) 38 years F Case 8 (left leg) 27 years M

Subject (painful limb)

At presentation

TABLE 3 Patient characteristics and the e¡ect of the control and intervention phases on their pain at presentation; the frequency of mirror use on follow-up and ¢nal pain scores at 6 weeks with infra-red thermal di¡erences and una¡ected limbs 166 McCABE ET AL

PHANTOMS

167

FIG. 3. Subject viewing non-re£ective surface with painful limb hidden. Reprinted with permission from McCabe et al (2003a).

mirror usage. Although no objective data were collected on function, both subjects felt that by six weeks function had improved to such an extent that they were able to return to their usual manual occupations. Interestingly, despite the lack of analgesic e¡ect during the mirror visual feedback procedure, Case 5 reported reduced pain at the 6 week follow-up (VAS 6/10 at presentation to 1/10 at 6 weeks). Reversal of infrared thermal (IRT) imaging temperature di¡erences were recorded in Case 4 at 6 weeks and Case 5 remained with no signi¢cant di¡erence between the two a¡ected limbs. No subjective relief of pain and sti¡ness or reversal of IRT temperature di¡erences were observed in the three subjects with chronic disease (52 years) and they had all discontinued mirror usage by the end of week 3 due to lack of analgesic e¡ect. These observations suggest that congruent visual feedback of the moving una¡ected limb, via a mirror, signi¢cantly reduces the perception of pain in early CRPS (Type 1) and sti¡ness in the intermediate stages of the disease. This supports the hypothesis that the CNS is capable of generating a feedback dependant state that can produce pathological levels of pain. In CRPS, this might involve a mismatch between di¡erent interdependent modalities, such as a disruption of

168

McCABE ET AL

FIG. 4. Subject viewing non-painful limb in mirror with painful limb hidden. Reprinted with permission from McCabe et al (2003a).

normal interaction between motor intention and sensory feedback. In those with an inherent vulnerability to this incongruence it can lead, in some, to referred, intractable pain following trauma or, in others, promote CRPS with a central nervous system origin. This might explain why some types of CRPS occur without discrete peripheral injury. If the correction of a sensory/motor mismatch produces an analgesic response then the reverse should also be true. That is to say when expected sensory input is deliberately falsi¢ed, sensory abnormalities should be generated in healthy volunteers and exacerbated in patients with chronic pain of unknown aetiology. Generating pain In a recent study we invited healthy volunteers and patients with FMS and CRPS to move their upper and lower limbs whilst undergoing normal and altered visual sensory feedback as provided via a mirror (McCabe et al 2003c,d). Motor/sensory con£ict was at its optimum when the subjects moved their limbs in opposing directions whilst viewing, via the mirror, their limbs apparently moving together. The primary aim of this study was to comprehensively capture, using a

PHANTOMS

169

qualitative methodology, the range of sensory experiences that subjects described as they underwent these manoeuvres. Each assessment was conducted ¢rst with the subjects viewing the control side (a whiteboard) and moving their limbs congruently and incongruently and then repeating the movements whilst viewing the intervention side (a mirror). 41 healthy volunteers were recruited (9 males, 32 females) aged 23^65 years (mean 40.4 years). They reported sensory changes at all stages of the protocol, control (congruent movement n ¼6 [15%], incongruent movement n ¼4 [10%]) and intervention. However, the maximum number of reports occurred when the subjects moved their limbs incongruently but perceived, via mirror imaging, that they were moving them congruently (congruent movement n ¼10 [25%], incongruent movement n ¼23 [56%]). The healthy volunteers reported discomfort (‘pins and needles’, ‘shooting pain’), changes in temperature and/or weight (‘£oaty sensation’ or ‘my arm was so heavy I was unable to lift it’), perceived loss of or additional limbs and disorientation (‘dizzy’, ‘strange’) (Table 4). Altered sensations were described predominantly in the hidden limb though this sometimes automatically conferred sensations on to the visualized limb, such as a hidden limb felt heavier and therefore the visualised limb was perceived as lighter. All altered sensations faded rapidly after limb movement had ceased and the hidden limb was visualized by the subject. Data collection in the patient population is still ongoing with 24 patients (7 CRPS Type 1, 17 FMS) recruited to date (3 males, 21 females) aged 23^73 (mean

TABLE 4 Type and incidence of sensory changes reported by healthy volunteers in hidden limb during congruent and incongruent movement whilst viewing a whiteboard (control) and mirror (intervention)

Type of sensation Discomfort/pain Temperature change Weight change Perceived ‘‘loss’’ of limb Perceived ‘‘extra’’ limb Disorientation Total number of subjects experiencing any sensory disturbances n ¼ 41 (male ¼ 8, female ¼ 33).

Whiteboard congruent movement

Whiteboard incongruent movement

Mirror congruent movement

Mirror incongruent movement

1 (2%) 0 2 (5%) 4 (10%) 0 3 (7%) 6 (15%)

1 (2%) 0 0 2 (4.9%) 0 4 (10%) 4 (10%)

4 (10%) 0 3 (7%) 8 (20%) 1 (2%) 15 (37%) 10 (24%)

7 (17%) 2 (5%) 6 (15%) 11 (27%) 9 (22%) 13 (32%) 23 (56%)

170

McCABE ET AL

TABLE 5 Type and incidence of sensory changes reported by patients with CRPS type 1 and ¢bromyalgia in hidden limb during congruent and incongruent movement whilst viewing a whiteboard (control) and mirror (intervention)

Type of sensation Discomfort/pain Temperature change Weight change Perceived ‘‘loss’’ of limb Perceived ‘‘extra’’ limb Disorientation Total number of subjects experiencing any sensory disturbances

Whiteboard congruent movement

Whiteboard incongruent movement

Mirror congruent movement

Mirror incongruent movement

11 (45.8%) 2 (8.3%) 7 (29.2%) 5 (20.8%) 0 0 20 (83.3%)

13 (54.1%) 3 (12.5%) 8 (33.3%) 8 (33.3%) 0 6 (25%) 15 (62.5%)

10 (41.6%) 4 (16.6%) 5 (20.8%) 13 (54.2%) 0 7 (29.2%) 15 (62.5%)

11 (45.8%) 5 (20.8%) 10 (41.7%) 14 (58.3%) 0 7 (29.2%) 16 (66.7%)

n ¼ 24 (male ¼ 3, female ¼ 21, ¢bromyalgia ¼ 17, CRPS ¼ 7).

47.5). Preliminary ¢ndings suggest patients perceive the same sensations as the healthy controls but the intensity and frequency of these sensations is greater. For example discomfort is reported as ‘crampy’, ‘sharp’ and ‘extremely painful’, temperature changes as ‘very hot’, ‘burning’. Importantly these sensory changes are described in addition to the subjects’ current symptoms at all stages of the protocol and for those with CRPS in their a¡ected and una¡ected limbs. The other striking di¡erence between the two study populations is that the patients report far more sensory disturbances than the healthy volunteers during the control stages (Table 5). It would appear that when sensory disturbances are already present simply hiding a limb from view is su⁄cient to exacerbate existing symptoms and generate new ones. Summary Our clinical observations and research studies support the conjecture put forward by Harris (1999) that when motor intentions to move a limb or series of joints no longer matches the corresponding sensory feedback then the subsequent ‘misrouting of information’ activates a central monitoring mechanism that £ags up such incongruity as pain. However, we would now like to extend Harris’ theory and propose that this monitoring mechanism is one of many monitoring mechanisms that act as alerts to warn the body that there is a problem with information processing and that pain may be only one of a broad range of sensory disturbances that subsequently occur. These central mechanisms we have

PHANTOMS

171

termed ‘ominory’ from the Latin word ominor meaning to prophesy, predict, foreboding. Our studies have focused on the mechanism that monitors motor/ sensory con£ict but a separate ominory mechanism could generate motion sickness when there is discordance between body position, balance and equilibrium. These mechanisms may be triggered by externally induced con£ict (e.g. incongruent movement whilst viewing the mirror) or internally (e.g. disease damage in RA leading to inaccurate execution of movement and/or altered proprioception). The key feature of these mechanisms is that when they are triggered they generate sensory disturbances such as nausea with motion sickness, pain in a phantom limb, phantom swelling and sti¡ness in RA and FMS. These resultant states we have termed dissensory from the Latin word dissensio meaning con£ict, disagreement. These are feedback dependent states in that the sensory/motor con£ict will continue to trigger the ominory mechanism and ultimately either via duration or intensity of this state the subject will su¡er pain. If however, an intervention is targeted to correct the initial source of con£ict, the ominory mechanism is suppressed and ideally pain is prevented or alleviated as with mirror visual feedback in early CRPS or the individual visualizing their phantom swollen joints in RA and FMS. We propose that the threshold at which a person either triggers the ominory mechanism or becomes aware of the subsequent sensory disturbances is individually determined but there will be some who are more sensitive than others. This we assume will relate to the standard variables of genetic factors, age, gender and sex hormone state. This was demonstrated by our healthy volunteer study; not all subjects experienced sensory disturbances. In addition, the preliminary patient data show that where sensory disturbances are already present a far lower stimulus is required to intensify the problem. Simply hiding a limb from view was su⁄cient to exacerbate sensory disturbances. This paper has only addressed three rheumatological conditions, RA, FMS and CRPS, but the same ominory mechanism may apply to the pain of osteoarthritis (OA) and indeed the development of pathological changes in the joint. Sharma et al (1997) and Pai et al (1997) have both shown that patients with unilateral knee OA have worse proprioception in their a¡ected and una¡ected joints than elderly controls without OA. The fact that both knees have reduced proprioception even when only one is diseased supports the theory of ‘mirror imaging’ across the body (Shenker et al 2004, this volume). The abnormal proprioception in the contralateral knee will be su⁄cient to continually trigger the ominory mechanism and perpetuate the problem. The subsequent dissensory state may explain the clinical observation that some individuals report high levels of pain when only minimal changes suggestive of OA are seen on X-ray imaging. This continuous sensory imbalance in the contralateral knee may increase the risk of injury and ultimately of generating OA (Hurley 1997). If targeted exercise is used to

172

McCABE ET AL

improve proprioception the initial trigger is removed and the ominory mechanism suppressed thereby perhaps preventing the onset of pain. Interestingly, patients with OA often report in clinic that their pain is worse at night and this may be a direct result of reduced corrective sensory input exacerbating the dissensory state. A darkened room diminishes visual feedback and immobilised limbs reduce proprioceptive input. In conclusion, a mismatch between motor output and sensory input triggers a warning, ominory mechanism in rheumatology patients and healthy volunteers. This generates the dissensory state and the individual will experience sensory disturbances that may include pain.

Acknowledgements C. S. McCabe is supported as an Arthritis Research Campaign Lecturer in Rheumatological Nursing. N. G. Shenker is supported as an Arthritis Research Campaign Clinical Research Fellow. J. M. Lewis is supported on an Arthritis Research Campaign ICAC award. D. R. Blake holds an endowed Chair
‘The Glaxo Wellcome Chair in Locomotor Sciences’. An Arthritis Research Campaign ICAC award supports the Royal National Hospital for Rheumatic Diseases, Bath.

References Arnett FC, Edworthy SM, Bloch DA et al 1988 The American Rheumatism Association 1987 revised criteria for the classi¢cation of rheumatoid arthritis. Arthritis Rheum 31:315^324 Baron R, Fields HL, Janig W, Kitt C, Levine JD 2002 National Institutes of Health Workshop: re£ex sympathetic dystrophy/complex regional pain syndromes
state of the science. Anaesth Analg 95:1812^1816 Blake DR, McCabe CS, Skevington SM, Haigh R 2000 Cortical origins of pathological pain. Lancet 355:318^319 Clarke S, Regli L, Janzer RC, Assal G, de Tribolet N 1996 Phantom face: conscious correlate of neural reorganization after removal of primary sensory neurones. Neuroreport 7:2853^2857 DeFelipe J, Conley M, Jones EG 1986 Long-range focal collateralisation of axons arising from corticocortical cells in monkey sensory-motor cortex. J Neurosci 6:3749^3766 Flor H, Braun C, Elbert T, Birbaumer N 1997 Extensive reorganization of primary somatosensory cortex in chronic back pain patients. Neurosci Lett 224:5^8 Fink GR, Marshall JC, Halligan PW et al 1999 The neural consequences of con£ict between intention and the senses. Brain 122:497^512 Gandevia SC, Phegan CML 1999 Perceptual distortions of the human body image produced by local anaesthesia, pain and cutaneous stimulation. J Physiol 514:609^616 Gordon DA, Hastings DE 1995 Clinical features of rheumatoid arthritis: early, progressive and late disease. In: Klippel JH, Dieppe, PA (eds) Practical rheumatology. Times Mirror International Publishers, London, p 169^182 Gregory RL 1998 Eye and brain. The psychology of seeing. Oxford University Press, p 53 Haigh RC, McCabe CS, Halligan P, Blake DR 2003 Joint sti¡ness in a phantom limb: evidence of central nervous system involvement in rheumatoid arthritis. Rheumatology (Oxford) 42:888^892

PHANTOMS

173

Halligan PW, Marshall JC, Wade DT, Davey J, Morrison D 1993 Thumb in cheek? Sensory reorganisation and perceptual plasticity after limb amputation. Neuroreport 4:233^236 Halligan PW, Marshall JC, Hunt M, Wade DT 1997 Somatosensory assessment: can seeing produce feeling? J Neurol 244:199^203 Harris AJ 1999 Cortical origins of pathological pain. Lancet 354:1464^1466 Helliwell PS, Howe A, Wright V 1988 Lack of objective sti¡ness in rheumatoid arthritis. Ann Rheum Dis 47:754^758 Hill A, Niven CA, Knussen C 1996 Pain memories in phantom limbs: a case study. Pain 66: 381^384 Hunter JP, Katz J, Davis KD 2003 The e¡ect of tactile and visual sensory inputs on phantom limb awareness. Brain 126:579^589 Hurley MV 1997 The e¡ects of joint damage on muscle function, proprioception and rehabilitation. Man Ther 2:11^17 Jadad AR, Carroll D, Glynn CJ, McQuay HJ 1995 Intravenous regional sympathetic blockade for pain relief in re£ex sympathetic dystrophy: a systematic review and a randomized, doubleblind crossover study. J Pain Symptom Manage 10:13^20 Janig W, Baron R 2002 Complex regional pain syndrome is a disease of the central nervous system. Clin Auton Res 12:150^164 Jensen TS, Krebs B, Nielsen J Rasmussen P 1985 Immediate and long-term phantom limb pain in amputees: incidence, clinical characteristics and relationships to pre-amputation limb pain. Pain 21:267^278 Ji RR, Woolf CJ 2001 Neuronal plasticity and signal transduction in nociceptive neurons: implications for the initiation and maintenance of pathological pain. Neurobiol Dis 8:1^10 Jones EG 1990 The role of a¡erent activity in the maintenance of primate neocortical function. J Exp Biol 153:155^176 Juottonen K, Gockel M, Silen T, Hurrir H, Hari R, Forss N 2002 Altered central sensorimotor processing in patients with complex regional pain syndrome. Pain 98:315^323 Katz J, Melzack R 1990 Pain memories in phantom limbs: review and clinical observations. Pain 43:319^336 Koltzenburg M, Wall PD, McMahon SB 1999 Does the right side know what the left is doing? Trends Neurosci 22:122^127 Lenz F, Byl NN 1999 Reorganisation in the cutaneous core of the human thalamic principal somatic sensory nucleus (Ventral caudal) in patients with dystonia. J Neurophysiol 82:3204^3212 McCabe CS 1999 An exploratory study into the experience of pain in rheumatoid arthritis. MSc thesis, University of Bath, Bath, UK McCabe CS, Haigh RC, Ring EFR, Halligan PW, Wall PD, Blake DR 2003a A controlled pilot study of the utility of mirror visual feedback in the treatment of Complex Regional Pain Syndrome (Type 1). Rheumatology (Oxford) 42:97^101 McCabe CS, Haigh RC, Halligan PW, Blake DR 2003b Referred sensations in patients with complex regional pain syndrome type 1. Rheumatology (Oxford) 42:1067^1073 McCabe CS, Haigh RC, Halligan PD, Blake DR 2003c Generating sensory disturbance in healthy controls. Rheumatology (Oxford) 42:63 McCabe CS, Haigh RC, Halligan PD, Blake DR 2003d Distorting proprioception in chronic pain patients exacerbates sensory disturbances
implications for pathology. Rheumatology (Oxford) 242:145 Melzack R 1971 Phantom limb pain. Anesthesiology 35:409^419 Melzack R 1990 Phantom limbs and the concept of a neuromatrix. Trends Neurosci 13:88^92 Merskey H, Bogduk N 1994 Classi¢cation of chronic pain. IASP Press, Seattle

174

DISCUSSION

Moore CI, Stern CE, Dunbar C, Kostyk SK, Gehi A, Corkin S 2000 Referred phantom sensations and cortical reorganisation after spinal cord injury in humans. Proc Natl Acad Sci USA 97:14703^14708 Pai YC, Rymer WZ, Chang RW, Sharma L 1997 E¡ect of age and osteoarthritis on knee proprioception. Arthritis Rheum 40:2260^2265 Papageorgiou AC, Badley EM 1989 The quality of pain in arthritis: the words patients use to describe overall pain and pain in individual joints at rest and on movement. J Rheumatol 16:106^112 Parker JC, Smarr KL, Buescher KL et al 1989 Pain control and rational thinking. Implications for rheumatoid arthritis. Arthritis Rheum 32:984^990 Pen¢eld W, Rasmussen TL 1950 The cerebral cortex of man: a clinical study of localization of function. Macmillan, New York Ramachandran VS, Rogers-Ramachandran D 1996 Synaesthesia in phantom limbs induced with mirrors. Proc R Soc Lond B Biol Sci 263:377^386 Ramachandran VS, Hirstein W 1998 The perception of phantom limbs. The D.O. Hebb lecture. Brain 121:1603^1630 Ramachandran VS, Stewart M, Rogers-Ramachandran D 1992 Perceptual correlates of massive cortical reorganisation. Neuroreport 3:583^586 Scadding JW 1999 Complex regional pain syndrome. In: Wall PD, Melzack R (eds) Textbook of pain. 4th edn. Churchill Livingston, Edinburgh, p 835^850 Sharma L, Pai YC, Holkamp K, Rymer WZ 1997 Is knee joint proprioception worse in the arthritic knee versus the una¡ected knee in unilateral knee osteoarthritis? Arthritis Rheum 40:1518^1525 Shenker NG, Blake DR, McCabe C, Haigh R, Mapp P 2003 Symmetry, T cells and neurogenic arthritis. In: Osteoarthritic joint pain. Wiley, Chichester (Novartis Found Symp 260) p 241^257 Staud R, Vierck CJ, Cannon RL, Mauderli AP, Price DD 2001 Abnormal sensitization and temporal summation of second pain (wind-up) in patients with ¢bromyalgia syndrome. Pain 91:165^175 Taylor-Clarke M, Kennett S, Haggard P 2002 Vision modulates somatosensory cortical processing. Curr Biol 12:233^236 Turton AJ, Butler SR 2001 Referred sensations following stroke. Neurocase 7:397^405 Walker DJ 1995 Rheumatoid arthritis. In: Butler RC, Jayson MIV (eds) Collected reports on the rheumatic diseases. Arthritis and Rheumatism Council, p 39^44 Wolfe F, Smythe HA, Yunus MB et al 1990 The American College of Rheumatology 1990 criteria for the classi¢cation of ¢bromyalgia. Report of the Multicenter Criteria Committee. Arthritis Rheum 33:160^172

DISCUSSION Pisetsky:What is the current working knowledge of referred pain in general? Is it common in OA? Blake: I’m not sure I’m the best person to answer that question. Clearly you can get referral at various levels: spinal, midbrain, brainstem or cortical. It is not just a sensation. Pisetsky: Does it re£ect sensitization, or is it neural spread? Why do some people have it?

PHANTOMS

175

Blake: Over and above some of the explanations that have been given where you can have two peripheral events that look like referred sensation, like back pain and hip pain, but which aren’t, I would imagine that practically all of it re£ects some degree of central sensitization. But since I am saying that every single pain centrally sensitizes if you know how to look for it, the distinction is not so important to me. Felson: The common clinical explanation is that a given joint is innervated by several di¡erent routes, and referred pain can go in that distribution. Blake: It has to be more complex than that. This is the simplest pathway, but there are clearly explanations in Halligan’s work that are over and above that system. Felson: There is clashing nomenclature here, and I wondered if you would help us a bit. The word ‘proprioception’ is the problem. Yesterday there was discussion of proprioceptive inaccuracy in OA. The last thing you talked about with ‘arms not conjugate if you try to make them conjugate in ¢bromyalgia’ is similar to the proprioceptive inaccuracy discussed yesterday. But you seemed to be using the term proprioceptive in several di¡erent ways. Could you de¢ne it for us? Blake: I’m embracing the entire concept of your appreciation of yourself in space. This breaks down into lots of di¡erent contributory factors which vary with time of day and so on. This would involve ocular and vestibular inputs and would relate much more to so-called joint position sense, which is what we have traditionally de¢ned as proprioception. Quite clearly, proprioception is vastly more than knowing whether your toe is pointing up or down in an extraordinarily crude examination by a clinician who doesn’t understand what proprioception is. My de¢nition at the moment encompasses all those factors that relate to localization of the body in space. What is good about the trick we have learned directly from Ramachandran’s work is that by throwing out one that we don’t tend to use very much
a visual one
we can magnify inherent defects or what we presume were inherent defects. Koltzenburg: What if you just ask people to close their eyes and then tell them to mimic one movement that you show them how to do on one side, on the other side? The question is, if you do this for 10 minutes, will they have weakness or pain that will interfere with that? You could be causing that relationship. Blake: Possibly, except that closing your eyes and looking at the mirror when the movement is coming in backward at you is not the same thing. We have done closed eye experiments to show this. Likewise, the di¡erent referred sensations that we have reported vary between eyes open and eyes closed. The test that we are doing is not mimicked by just closing your eyes. Koltzenburg: I am not talking about the defect, but rather the inference that these are proprioceptive de¢cits, and if there is a structural reason for such a proprioceptive de¢cit this should be evidenced in the ¢rst test that you do. The

176

DISCUSSION

fact that you have to do it several times suggests to me there may be other factors that need to be activated. Blake: I have no doubt that this is the case. Hunter: For those people who have chronic pain states, is there any biological rationale in terms of their cortical distribution for pain sensation, in terms of what psychological disturbances they may subsequently develop? Blake: I can’t answer that. A major cognitive disturbance can create an environment where someone is sensitized at a point in time. We are right at the edges of our knowledge of this. It appears to me that since we repeatedly hear that a major cognitive in£uence appears to precipitate something that it is more likely that they are telling the truth than they are fabricating it. And all of those of us who specialize in RA are aware that the patients can develop bizarre £ares in their arthritis which are described di¡erently from more classical in£ammatory £ares, particularly following grief reactions. The delay seems to be extremely variable. Koltzenburg: Is it possible that this is something congruent with a patient’s desire to anchor an event? It is correlative. If you ask most patients what you thought brought their £are on, they will come up with an explanation. Blake: That is the traditional approach that medicine has taken to these problems. I am suggesting that we need to be more tolerant than that and embrace these cognitive events as likely inciting factors. For any individual, you could easily statistically dismiss the association by denying it in many other people. This is how medicolegal practice is conducted in this country. But it doesn’t make the individual observation incorrect, particularly when you have an explanation for it. Felson: There were a couple of large-scale longitudinal studies in RA. One was by Alex Zautra, a psychologist at the University of Connecticut, and another was by John Mason in our group. These studies looked at stressful life activities and £ares of RA, and both were well done and well powered but they showed nothing. Why is that? Blake: I think it is because we are not adopting the right models. It is similar to the way that doctors dismissed the ability of patients to relate £ares in their arthritis to changes in the weather. Exactly the same kind of cross-sectional study was done, and found negative results. Felson: These were longitudinal diary studies. Blake: This was also done in the study I’m referring to. You can use this to say that this event doesn’t occur. But I am not so sure that the fault isn’t with our modelling system. As anyone with OA of the knee knows, you can detect weather change because the joint becomes barometric. The problem is that the tools we are using to study it are poor. It is a very hard position to argue from. It certainly allows the traditionalist view that we have just had, which then gains further support when it becomes a medical legal issue, but I am not sure it is correct.

PHANTOMS

177

Dieppe: You said that some people undergoing these clever incongruity experiments will experience sti¡ness. Do you regard the sensation of sti¡ness as part of the pain experience? Blake: Yes. It is clearly not a peripheral event. There is a paper coming out by my colleague Richard Hague on this. When we had our ¢rst phantom pain patient, the thing that distressed her most was her phantom sti¡ness. Each morning she would wake up and mentally engage her phantom limb. Then she would exercise her phantom limb in parallel with her normal limb. What we were interested in was exactly the question you are asking, which is totally pivotal to what people think sti¡ness is. The general feeling is that it is a peripheral event. This is rubbish. It may start that way but it doesn’t end up that way. We did a whole series of similar patients and they all said exactly the same thing. Then we started taking histories of sti¡ness very di¡erently. We ask people when they get up in the night to use the toilet whether they have a full quantum of sti¡ness or do they have to wait for a certain amount of rest ¢rst. They get a full quantum ever so quickly once they have hit a certain stage of sleep. They can get a full quantum of sti¡ness within an hour. This leads us on to the issues of night pain and rest pain. Is it rest, or is it changes in sensory perception which are altered when people down-regulate their brains? Is that what changes the quality of OA pain at night? It is not there all night but comes at speci¢c times and is extraordinarily distressing. Herzog: I was very intrigued by the pain you induced in normal people with incongruent movements. I was wondering if you had ever tried to measure in people with incongruent movements without the mirror and with the mirror whether the muscular control of the movement is the same. My hunch would be that since you get the wrong feedback, it might be quite an exercise for these people to do what you are asking them and there could be a lot of co-contraction. This could be measured by electromyogram. Some of the pain might be due to this. Blake: It is not too much exercise. We have done it on ourselves thousands of times. I used 10 min as a simple system. You can get this in some people in one or two minutes. The ones that are going to go, go quite quickly. Herzog: Have you ever measured electromyographic activity? Blake: No. There are a lot of experiments we need to do, linking these observations back to monitors of the incongruence centre. Are the ones that are experiencing pain the ones that are ¢ring o¡ in the right hemisphere? We presume they are, but we don’t know. In some people we can do this to them in seconds. It is not a fatigue event. Herzog: I just assume that the whole motor control may get the wrong feedback. I am not sure what this would do to the mechanics of the system. Blake: You are alluding now to phenomena such as repetitive strain injury. We have many more subtle tests where we are making people do false movements that are essentially seen as incongruent when viewed on a computer screen. These are

178

DISCUSSION

very much more tiny, ¢ne movements. This allows us to look at the position of the axial skeleton, which seems to impinge on this. Herzog: If you just left the muscle on because of wrong control at quite a high level, which would prevent it having an adequate blood supply, within 20^30 s you would feel a certain tingling sensation and pain. Blake: In terms of where they are feeling the pain, it seems to me to be much wider than this. It becomes a very di¡use type of pain, very much akin to ¢bromyalgia. Herzog: If I understood correctly, you said that in normal people you made your observations very systematically, that is, all normal people undergoing your exercise regime would end up with pain. Blake: There certainly was a self-similarity in those that reported it with what people experience in repetitive strain injury. It seemed to cross multiple muscle groups. We have lots of work to do with this system. This is a very crude way of creating this incongruent input and I am sure there will be many better ways. We are working on this. Conaghan: How did you come to the four month CRPS rule? Blake: In our series of 15, it is roughly about four months. SPET scanning of CRPS localizes where uptake is maximal. There are uptake changes in the midbrain and they change at about 4^6 months. There is a threshold change in the population at around that time. It is not that we don’t make the chronic ones better: we actually make them worse, because they are incapable of doing congruent movement. Felson: Another issue you raised was the denervated synovium of RA. This was a surprise to me. Why, then, is synovectomy such an e¡ective surgery for people with RA? Blake: I am talking microns, whereas surgeons are talking feet and inches! It is a very di¡erent process. I am referring to the very super¢cial synovium. In patients with palindromic RA, the joints are fantastically painful. This is when the synovium is innervated, and this goes. Grubb: What is the evidence that it is truly denervated and you haven’t simply lost the peptide content? There are non-peptidergic a¡erents that would be a lot harder to see. Blake: We did this in Julia Pollock’s lab, who is an expert on this. As far as we could see, in terms of what we have marked for, the nerves have gone.

Osteoarthritic Joint Pain: Novartis Foundation Symposium 260. Volume 260 Edited by Derek J. Chadwick and Jamie Goode Copyright ¶ Novartis Foundation 2004. ISBN: 0-470-86763-9

Bone pain and pressure in osteoarthritic joints Peter A. Simkin Professor of Medicine, Division of Rheumatology, University of Washington, Box 356428, Seattle WA 98195, USA

Abstract. Intraosseous hypertension has been associated with a deep aching bone pain, particularly at rest, in subsets of patients with osteoarthritis of the hip and knee. The pathophysiology of this problem remains uncertain, but intraosseous phlebography implicates out£ow impairment at relatively distal venous sites. Although the issue has been controversial, intraosseous pressures rise normally, and painlessly, when epiphyseal bone is loaded and these pulses may be mechanically meaningful in the distribution and transmission of impact energy. Increased out£ow resistance may amplify the episodic pressure response with subsequent intravasation of epiphyseal fat leading to ‘marrow oedema’ and altered mechanics. The relationship between persisting pain and pressure is an old but convincing association. Its precise mechanism in osteoarthritis remains in need of an adequate explanation. 2004 Osteoarthritic joint pain. Wiley, Chichester (Novartis Foundation Symposium 260) p 179^190

This symposium was convened to consider the problem of pain in patients with osteoarthritis (OA). Where are the relevant receptors, and what triggers them, do the patterns di¡er in a meaningful (and therefore interpretable) way, and why is there such variance in the pain perceived by individuals who appear to have comparable degrees of degradation in articular cartilage and in the hard and soft tissues of the underlying subchondrium? For the most part, these questions remain unanswered and our deliberations focus properly more on opportunities for investigation than on meaningful current data. A conspicuous exception, however, lies in the extensive, older studies of hypertension in bones adjacent to large, painful OA human joints. This paper will review some of those ¢ndings and will consider further the possible role of intraosseous pressure in articular physiology and pathophysiology. Intraosseous hypertension in osteoarthritis The foremost champion of the pressure/pain relationship in OA has been Carl Arnoldi of Copenhagen. In a series of papers on the subject, Arnoldi and his 179

180

SIMKIN

colleagues focused on rest pain, especially that occurring at night (Arnoldi 1991, Lemperg & Arnoldi 1978, Arnoldi et al 1980). They reported that this symptom was best described as a deep, throbbing ache; that it was usually accompanied by an increase in intraosseous pressure; and that decompressive surgical procedures such as simple fenestration or a tibial osteotomy led to e¡ective pain relief along with abolition of the intraosseous hypertension (Dey et al 1989). When intraosseous phlebography was done, there was impaired clearance of the radiopaque dye as well as impressive dilation of sinusoids and collecting veins. Further evidence of obstructed out£ow was provided by impaired clearance of injected radio-iodide (Hernborg 1969). Most of these basic ¢ndings have been shown both in hips and in knees and were perhaps demonstrated most convincingly in the patellae of patients with anterior knee pain (Bj˛rkstr˛m et al 1980, Schneider et al 2000, Waisbrod & Treiman 1980, Ficat & Hungerford 1977). Osteoarthritis is not unique in its disposition toward intraosseous hypertension. Avascular necrosis of bone is perhaps the best studied entity with persuasive evidence of obstructed venous out£ow in each of the disparate entities associated with this lesion. Thus steroid therapy, alcoholism and Gaucher’s disease lead to hyperplasia and or hypertrophy of intraosseous cells with resulting compression of out£ow vessels while sickle cell disease and disbaric decompression obstruct the same vessels through intravascular processes. All of these patients are then subject to the same ischaemic catastrophes in the £exible convex members of large, human joints (Ficat & Arlet 1980). Like OA, the remaining entities with bony hypertension (re£ex sympathetic dystrophy, fatigue fracture, regional osteoporosis, etc.) have less logical vascular pathophysiology. It is intriguing to note, however, that each of them is associated with ‘marrow oedema’ by magnetic resonance imaging (MRI). Since the volume of each bone is ¢xed by its mineralized shell, it seems axiomatic that its £uid contents must obey zero sum principles. This means that more water (bony oedema) means less fat. It seems reasonable to hypothesize that increased local pressure in some way promotes intravasation of bone fat cells, with resultant vascular clearance that parallels the increase in intraosseous water. Parenthetically, it is also reasonable to examine the extent to which intraosseous hypertension is also present, and a contributing cause of pain, in ankylosing spondylitis, rheumatoid arthritis, and other in£ammatory and traumatic conditions associated with marrow oedema (Arndt et al 1996, Visuri 1997, Yamamoto & Bullough 1999, McGonagle et al 1999, 2003). Thus, intraosseous hypertension may be a more pervasive mechanism in human disease than rheumatologists now recognize, and it is appropriate to consider some of the outstanding questions that must be addressed before we can understand and interpret this ¢nding.

PRESSURE AND BONE PAIN

TABLE 1

181

Mean tibial intramedullary pressures

Epiphysis Metaphysis Diaphysis

Child

Adult

26.1 (24)* 22.3 (69)* 13.6 (124)*


19.9 (14) 11.8 (26)

Pressures are in mmHg with the number of observations in parenthesis. Children were 3^17 years old. *Single, high outlying values were deleted from each of the three bony regions in children.

What is the normal intraosseous pressure? This question has been addressed in a number of studies but there is considerable variation in the ¢ndings both in humans and in experimental animals (Stein et al 1957). Virtually all data have been obtained under anaesthesia and this practice as well as the speci¢c agent used, could be important factors in the observed variation. One study in vertebral bodies found the intraosseous pressure to be comparable to that in adjacent veins, and similar equilibrium pertains in the sternum, skull and other ‘£at bones’. Almost all studies in appendicular bones, however, have found higher pressures, usually in the 15^20 mmHg range that prevails in the normal human eye
an unquestionably turgid tissue. Ficat and Arlet, in their book Ischemia and Necrosis of Bone (Ficat & Arlet 1980) cite an extensive study of young African subjects (Table 1) (Kabakele 1972, cited by Ficat & Arlet). In accord with their own data from the proximal femur, these ¢ndings indicate that intraosseous pressure rises as the points of observation move away from the shaft and into the subchondral regions of the epiphyses. They then point out that ‘notwithstanding the direction of this gradient, its existence presupposes the presence of adaptable barriers and permanent di¡erences in neurovascular control.’

What regulates intraosseous pressure? Overall, the resting pressure in each bony compartment must be determined by vascular factors a¡ecting the balance between arterial in£ow and venous out£ow. Speci¢cally, osseous pressures greater than these in adjacent veins means that intraosseous vascular resistances must be in action (Wilkes & Visscher 1975). To investigate possible controlling factors, infusions of epinephrine and norepinephrine have been studied by many investigators (Stein et al 1958, Schneider et al 1998, Shim 1968). The consensus ¢nding of such work is an increase in vascular resistance with a concomitant fall in intraosseous pressure. Thus, in this case, arteriolar, a¡erent constriction explains the results well.

182

SIMKIN

In OA however, the compartmental pressure increases with decreased £ow and concomitant dilatation of sinusoids. This combination ¢ts only with an e¡erent constrictor analogous to those present in erectile tissues. At present, no such constrictor has been identi¢ed clearly, although at least two electron microscopic studies have identi¢ed apparent sphincters at the out£ow site where sinusoids empty into collecting venules (Ohtani et al 1982, Kessel & Kardon 1979). We know almost nothing about the controls of these putative sphincters but one study has found an increase in pressure with a decrease in £ow with ephedrine infusion, and re£ex neurological and/or metabolic controls deserve consideration (Bˇnger et al 1982, Holm et al 1990). Somewhat surprisingly, exercise leads to increased resistance with decreased £ow while muscle contraction (in di¡erent studies) leads to a modest increase in intraosseous pressure (Gross et al 1979). These ¢ndings are consistent with an epiphyseal pressurization during active exercise, but more experimental con¢rmation is clearly needed. Finally, an intriguing ¢nding from the clinic may be relevant. Transplant patients taking cyclosporine, a known vasoconstrictor, are subject to severe, episodic knee pain in the absence of any apparent articular pathology. This phenomenon is readily controlled, however, by administration of a vasodilator: nifedipine (Barbosa et al 1995, Kart-Koseoglu et al 2002). To date, there is no direct con¢rmation of increased intraosseous pressure in this syndrome, but the ¢ndings are considered most consistent with the possibility of drug-induced vasoconstriction in venous, e¡erent vessels. It may be relevant to note that this pain is perceived in the distal femoral epiphysis, rather than the shaft of the femur, and the putative vasoconstriction would presumably be localized there. What happens with joint use? Within narrow limits, the volume of each normal bone remains constant at a value determined by its enveloping walls of semi-rigid, calci¢ed tissue. Under load, however, those walls, and the trabecular framework within them, will £ex. In so doing, the bone serves as a complex series of springs which are capable of storing and distributing loading energy which then can be recovered when each compressed spring re-expands to resume its resting position. Most human joints are comprised of a sti¡, in£exible, concave component such as the glenoid fossa opposed by a £exible convex member such as the humeral head (Simkin et al 1980). In vitro, and presumably in vivo, this £exion causes a signi¢cant boost in the humeral intraosseous pressure (Downey et al 1988). This process of compression under load with subsequent re-expansion must necessarily impact on the cells, interstitial £uid, and other soft tissue elements contained within the intertrabecular space, and the hydrostatic pressure must rise accordingly. This rise will be greatest in the bony compartments that are immediately subchondral,

PRESSURE AND BONE PAIN

183

but as their walls and £oors then £ex as well, a descending pressure gradient develops in which each trabeculum is burdened primarily by the pressure di¡erence between adjoining soft tissue spaces. There has been long-standing controversy regarding the possibility that articular loading might induce su⁄cient pressure to provide a hydraulic contribution to the mechanics of trabecular bone, but it is necessary to acknowledge that most bioengineers currently feel this possibility has been excluded. Most of the available, mechanical data are derived, however, from testing, in vitro, of small pieces of excised bone, often obtained from elderly subjects and/or from bones (such as vertebrae) or animals (such as cows) where hydraulic support is less likely. The results may di¡er with more appropriate (but more di⁄cult) testing in vivo.

Does trabecular tension cause pain? It is not controversial to say that impact events may raise subchondral intraosseous pressure by amounts well in excess of 100 mmHg (Downey et al 1988). Injection of 5 ml of saline into human femoral necks is reported to be highly painful when the pressure rises just as pressure causes pain in an obstructed bowel, a blocked ureter, a gouty great toe, or a simple old fashioned boil. Such a distension-induced pain can not be excluded in OA, particularly in the use-pain that many patients report. This explanation would not seem to ¢t, however, with the nocturnal rest-pain that has been most closely related to the ¢nding of intraosseous hypertension. It also seems unlikely in the epiphyses where I believe that intermittent, load-induced pressure pulses are a painless part of normal articular physiology.

Does ischaemia cause intraosseous pain? This important question must be asked although there has been little work that addresses the possibility directly. Metabolic evidence of ischaemia (decreased pO2 and increased venous lactate) was found to accompany increased pressure in one study of OA femoral heads (Kiaer et al 1988). Further intraoperative studies could provide important support for or against the presence of this mechanism. Indirect evidence may be found, however, in studies of post-traumatic compartment syndromes in human extremities (Mars & Hadley 1998). There, a persisting pressure greater than 35 mmHg will regularly induce pain and poses a signi¢cant risk of ischemic infarction for motor nerves traversing such a space. Intraosseous pressures in OA regularly exceed this level, and it seems logical to attribute the attendant pain to a bony compartment syndrome (Fricker et al 1995). The prompt pain relief provided by decompressive fenestration of bone would be

184

SIMKIN

entirely consistent with this interpretation just as decompressive fasciotomy relieves the pain of a soft tissue compartment syndrome.

Is intraosseous in£ammation important? Although in£ammation has become increasingly important in studies of OA, most of this work has focused on the synovium and the chonodrocyte. In£ammatory cells are present in the subchondral bone of OA patients, however, and it is plausible (in the absence of evidence to the contrary) to think that proin£ammatory products of these cells could play a role in bone pain. The unusual human syndrome of pancreatitic arthritis provides an example of a speci¢c mechanism that could be relevant (Simkin et al 1983). There, the presence of circulating pancreatic enzymes appears to activate and to amplify the e¡ects of local tissue lipases with a resultant £ood of fatty acids into the involved tissue. When the fatty acid release exceeds the binding capacity of available albumin, the excess of free fatty acids causes intense local in£ammation. This phenomenon is known best, and recognized most readily when it occurs focally in subcutaneous fat. The same events may occur, however, in the fatty subchondrium with devastating pain and rapid osteolysis. Were a similar activation of tissue lipase to be triggered by ischaemic infarction in hypertensive bone, the resultant fatty acid-induced in£ammation could result in substantial local pain and marrow oedema.

Conclusion The organizers of this symposium advised each contributor that ‘your talk should form a provocative basis for the discussion’ and that it was appropriate ‘to present new or preliminary results and to speculate on their signi¢cance.’ In fact, most of the information I’ve reviewed here is old, but I’d like to exercise the licence I’ve been given to speculate anyway and to present the following working hypotheses: (1) Appendicular bones are normally pressurized through control systems that remain unknown. That pressure is greatest in the subchondral regions where pressurization enhances trabecular elasticity just as in£ation enhances the bounce of a basketball or a rubber tyre. (2) The baseline pressure in intermittently ampli¢ed during loading, and this process serves to distribute impact energy down a steep pressure gradient throughout the extensive system of subchondral trabeculae. In so doing, it enlists the entire trabecular framework to serve as one, extensive spring.

PRESSURE AND BONE PAIN

185

(3) This process serves to protect both the overlying cartilage and the supporting cortical bone against excessive energy burdens while also permitting recovery of the stored energy to perform useful work. (4) Under a variety of clinical circumstances, including OA, the neurovascular control of this system breaks down with a resultant rise in resting pressure that causes pain, most probably on an ischaemic basis. (5) A more e¡ective understanding of the normal system may help us to understand how and why it fails and may lead ultimately to newer and more e¡ective strategies for controlling this part of the arthritis pain spectrum. References Arndt WF 3rd, Truax AL, Barnett FM, Simmons GE, Brown DC 1996 MR diagnosis of bone contusions of the knee: comparison of coronal T2-weighted fast spin-echo with fat saturation and fast spin-echo STIR images with conventional STIR images. Am J Roentgenol 166: 119^124 Arnoldi CC 1991 Patellar pain. Acta Orthopaedica Scandinavica 62(suppl 244):1^29 Arnoldi CC, Djurhuus JC, Heerfordt J, Karle A 1980 Intraosseous phlebography, intraosseous pressure measurements and 99mTC-polyphosphate scintigraphy in patients with various painful conditions in the hip and knee. Acta Orthop Scand 51:19^28 Barbosa LM, Gauthier VJ, Davis CL 1995 Bone pain that responds to calcium channel blockers. A retrospective and prospective study of transplant recipients. Transplantation 59:541^524 Bj˛rkstr˛m S, Goldie IF, Wetterqvist H 1980 Intramedullary pressure of the patella in Chondromalacia. Arch Orthop Traumat Surg 97:81^85 Bˇnger C, Harving S, Bˇnger EH 1982 Intraosseous pressure in the patella in relation to simulated joint e¡usion and knee position: an experimental study in puppies. Acta Orthop Scand 53:745^751 Dey A, Sarma UC, Dave PK 1989 E¡ect of high tibial osteotomy on upper tibial venous drainage: study by intraosseous phlebography in primary osteoarthritis of knee joint. Ann Rheum Dis 48:188^193 Downey DJ, Simkin PA, Taggart R 1988 The e¡ect of compressive loading on intraosseous pressure in the femoral head in vitro. J Bone Joint Surg Am 70:871^877 Ficat RP, Hungerford DS 1977 Disorders of the patello-femoral joint. 1st edn. Williams & Wilkins, Baltimore Ficat RP, Arlet J 1980 Ischemia and necroses of bone. Williams & Wilkins, Baltimore Fricker C, Bucher K, Stuker G 1995 Are degenerative joint diseases chronical compartment syndromes? (German) Schweiz Arch Tierheilkd 137:137^140 Gross PM, Heistad DD, Marcus ML 1979 Neurohumoral regulation of blood £ow to bones and marrow. Am J Physiol 237:H440^H448 Hernborg J 1969 Elimination of Na-131-I from the head and the neck of the femur in una¡ected and osteoarthritic hip joints. Arthritis Rheum 12:30^33 Holm IE, Ewald H, Bulow J, Bunger C 1990 Vasoactive substances in subchondral bone of the dog knee. J Orthop Res 8:205^212 Kabakele M 1972 Contribution au diagnostic pre¤ coce de l’oste¤ one¤ crose dre¤ panocytaire. The' se d’agre¤ gation, Kinshasa, Zaire Kart-Koseoglu H, Yucel AE, Isyklar I, Turker I, Akcaly Z, Haberal M 2003 Joint pain and arthritis in renal transplant recipients, and correlation with cyclosporine therapy. Rheumatol Int 23:159^162

186

DISCUSSION

Kessel RG, Kardon RH 1979 Tissues and organs: a text-atlas of scanning electron microscopy. W.H. Freeman, San Francisco Kiaer T, Gronlund J, Sorensen KH 1988 Subchondral pO2, pCO2, pressure, pH, and lactate in human osteoarthritis of the hip. Clin Orthop 149^155 Lemperg RK, Arnoldi CC 1978 The signi¢cance of intraosseous pressure in normal and diseased states with special reference to the intraosseous engorgement-pain syndrome. Clin Orthop Rel Res 136:143^156 Mars M, Hadley GP 1998 Raised intracompartmental pressure and compartment syndromes. Injury 29:403^411 McGonagle D, Conaghan PG, O’Connor P et al 1999 The relationship between synovitis and bone changes in early untreated rheumatoid arthritis: a controlled magnetic resonance imaging study. Arthritis Rheum 42:1706^1711 McGonagle D, Marzo-Ortega H, Benjamin M, Emery P 2003 Report on the Second international Enthesitis Workshop. Arthritis Rheum 48:896^905 Ohtani O, Gannon B, Ohtsuka A, Murakami T 1982 The microvasculature of bone and especially of bone marrow as studied by scanning electron microscopy of vascular casts
a review. Scan Electron Microsc (Pt 1):427^434 Schneider T, Drescher W, Becker C et al 1998 The impact of vasoactive substances on intraosseous pressure and blood £ow alterations in the femoral head: a study based on magnetic resonance imaging. Arch Orthop Trauma Surg 118:45^49 Schneider U, Breusch SJ, Thomsen M, Wenz W, Graf J, Neithard FU 2000 A new concept in the treatment of anterior knee pain: patellar hypertension syndrome. Orthopedics 23:581^586 Shim SS 1968 Physiology of blood circulation of bone. J Bone Joint Surg Am 50:812^824 Simkin PA, Graney DO, Fiechtner JJ 1980 Roman arches, human joints, and disease: di¡erences between convex and concave sides of joints. Arthritis Rheum 23:1308^1311 Simkin PA, Brunzell JD, Wisner D, Fiechtner JJ, Carlin JS, Willkens RF 1983 Free fatty acids in the pancreatitic arthritis syndrome. Arthritis Rheum 26:127^132 Stein AH Jr, Morgan HC, Reynolds FC 1957 Variations in normal bone-marrow pressures. J Bone Joint Surg Am 39:1129^1134 Stein AH Jr, Morgan HC, Porras RF 1958 The e¡ect of pressor and depressor drugs on intramedullary bone-marrow pressure. J Bone Joint Surg Am 40:1103^1110 Visuri T 1997 Stress osteopathy of the femoral head. 10 military recruits followed for 5^11 years. Acta Orthop Scand 68:138^141 Waisbrod H, Treiman N 1980 Intra-osseous venography in patellofemoral disorders: a preliminary report. J Bone Joint Surg Br 62:454^456 Wilkes CH, Visscher MB 1975 Some physiological aspects of bone marrow pressure. J Bone Joint Surg Am 57:49^57 Yamamoto T, Bullough PG 1999 Subchondral insu⁄ciency fracture of the femoral head: a di¡erential diagnosis in acute onset of coxarthrosis in the elderly. Arthritis Rheum 42:2719^ 2723

DISCUSSION Pisetsky: I’d like to ask about the relationship between OA and osteoporosis. Where does bone loss ¢t into your model in terms of the ability to handle pressure? Simkin: The inverse relationship that you are speaking of is intriguing. It would support my position if osteoporosis does not lead to perforations of the trabecular walls in the convex areas that I think are hydraulically supported. The fact that osteoporotic bone is more compliant under load could actually facilitate

PRESSURE AND BONE PAIN

187

load-bearing by the marrow content. In our studies of femoral head loading in vitro the highest intraosseous pressures were found in two very osteoporotic specimens (Downey et al 1988). Pisetsky: What about women and their post-menopausal bone loss problems? How does this ¢t in? Simkin: It is the same issue: if someone has lost bone, I would think it would be lost in a way that preserves the ba¥e system. Those women are at risk of fracturing their femoral necks but they don’t crush their femoral heads when they fall (Todd et al 1972, Wicks et al 1982). Fernihough: Earlier we discussed the di⁄culty of getting a good pain model of OA, particularly in rat. Although the rat walks with its knee at 90 degrees of £exion, which would suggest a high intraosseous pressure, it is quite striking that their growth plate is not fused and remains cartilagenous. How do you think this would a¡ect intraosseous pressure, and how is this related to the pain response? Simkin: I don’t have any data on intraosseous pressure in animal models of OA. In humans, we don’t expect to see changes of OA in children with open growth plates. Perhaps their intraosseous springs are more e¡ective. As I’ve listened to the papers over the last two days I have wondered whether there is intraosseous hypertension in the cat and also the human elbow. It may be very useful to know this in sites where there are degenerative changes without accompanying pain. Lohmander: You have done some seminal work in joint £uid physiology, and on the turnover of joint £uid. Listening to your talk on the hydraulics of bone makes me want to ask you about your opinion on the communication between the subchondral hydraulic space and the joint space. What are the possibilities for communication between subchondral cells and cartilage cells? Simkin: There is no disagreement that in immature individuals that pathway is open. We have done some work on this in mature sheep, and when we use high pressures analogous to those that occur within normal joints we can drive £uid across (Simkin & Peterson 2001). The evidence against osteochondral communication is largely based on di¡usion experiments rather than convection experiments. Felson: What Stefan Lohmander is implying is that your view is unconventional: it is generally felt that there is no communication between bone marrow and the cartilagenous and articular surface, except in pathology where there might be cysts that communicate. Would this be a fair statement? Simkin: Yes, it’s fair. In the child though, there is clearly di¡usion across the cartilage/bone interface. In the adult, it is thought there is no such di¡usion. The experiments that I am alluding to indicate that in fact there is convection when the pressure is high. Kuettner: Cartilage is as hydraulic as your bone. There are two hydraulic systems sitting on top of each other under normal conditions. What you presented was the

188

DISCUSSION

bone excluding the cartilage, but your lines could have continued into the cartilage. There is a connection. It may not be a liquid £ow, but a biomechanical supplementation of the two tissues. Would you agree? Simkin: Yes. The other thing I would add is that I suspect the permeability we see with convective £ow, is in fact, ‘semipermeability’. Water will cross; the larger molecules will not. I think the feature that we all know as the tidemark is the debris of apoptotic chondrocytes. These macromolecules may be driven downward by through-£owing water and be sieved out at the border of calci¢ed cartilage. Herzog: I can accept one or the other of your stories, but not both. The dilemma I have is that on the one hand, to have good load distribution you need a viscoelastic system. On the other hand, you are saying that this is very good for the recovery of energy, but for the recovery of energy it would be much better if you had a purely elastic system. If you do the indentation test on an entire joint, you will see that it is viscoelastic, with a tremendous loss of energy. I can believe in your viscoelastic argument, but not so much in your bouncy element. Simkin: The key thing is the time factor. Someone who hops twice a second is operating at the elastic end of the viscoelastic curve. This is very fast and permits recovery of most of the energy. If you hop once every ¢ve seconds you get signi¢cant £ow within the bone, this disspates energy and it can’t be recovered. Herzog: This is correct if we think about one hop or one step, but if we think about someone doing a 2 hour run then the viscoelastic e¡ect will over time dissipate all the energy that is put in. Simkin: You keep recovering with each step. Herzog: We know that you don’t. If we do stress^relaxation testing where we go to the same displacement each time, we lose pressure. If we go to the same pressure we lose displacement. Simkin: Are the experiments you are talking about done in living vascularized bone? Herzog: These are in situ joints from freshly harvested animals. Simkin: So there is no blood £ow: I think this makes a big di¡erence. In our own studies of femoral head loading in vitro, we maintained a slow, constant infusion of saline and observed no decay in the viscoelastic behaviour (Downey et al 1988). Herzog: You started by talking about bone pressure. You talked about arterial pressures, but then towards the end of the talk I had the feeling you were talking about pressure within the £uid of the bone. It wasn’t clear to me whether they were mixed from the beginning to the end. Simkin: The original data I showed were from living bone, but some of the mechanical data were from intact bones studied in vitro. I didn’t really talk about arterial pressures. Herzog: So you are saying that if we did indentation testing in an anaesthetized but fully vascularized animal, that we would not see much loss of energy. In fact,

PRESSURE AND BONE PAIN

189

you would argue that the system would behave virtually elastically and all energy put into the joint would be fully recovered. If that was really the case, that would challenge much of the current thinking on joint mechanics. I think I will try this experiment, as we are perfectly set up to do precisely these types of experiments. Simkin: Yes, but it wouldn’t be perfect. As with a bouncing rubber ball some energy would be lost with each impact. Herzog: That would be a tremendous change in mechanical thinking if that was correct, and it is easily testable. Felson: Walter Herzog, does bone deform during weight bearing? This would be part of this hypothesis. Herzog: From extreme studies, on humans doing the triple jump, I know that bone can deform to quite an extent. I guess the issue that you are bringing up is that the elastic modulus of articular cartilage is about 1000 times smaller than that of bone. So, articular cartilage would deform under a given amount of compression much more than bone. Simkin: But, as Eric Radin pointed out, the cartilage is very thin compared to the subchondrium (Radin et al 1970). When we measured compliance in canine shoulders in vitro, the overall de£ection was much greater in bone than in cartilage (Simkin et al 1985). Lohmander: Accepting your suggestion of convective £ow between cartilage and subchondral space, could that be related to what the magnetic resonance imaging (MRI) people observe as signal changes in bone marrow oedema, which in turn is connected back to pain in OA? Simkin: I wouldn’t think so. The amount of £uid we are talking about is relatively small. The subchondrium is surprisingly vascular and I suspect that its interstitial water turns over quite rapidly. Pisetsky: What happens in core decompression? What would happen to the cartilage and bone when pressure goes? Simkin: I would think that the normal spring would be compromised. But it is not a normal spring that you are decompressing: it is one that is already sti¡ened by disease. This is certainly a procedure that we believe to be useful. Pisetsky: Is there any sense that this makes cartilage worse? Simkin: I can’t say. Dieppe: I wanted to come back to the homage we have been paying to Carlo Arnoldi throughout this meeting concerning decompression experiments and there importance in pain. There is another phenomenon one sees with so-called atrophic OA, when you get massive destruction of bone. You can wipe out most of the femoral head of the hip. These people don’t have a little pain, they have a huge amount of pain. It doesn’t make any sense that this could have anything to do

190

DISCUSSION

with intraosseous pressure because they have lost the whole subchondrium and lots more bone. Simkin: I agree. Clearly, the study that I showed initially in which they selected 50 patellae out of 136 painful knees is re£ective of the idea that I am speaking of which concerns a subset of OA patients. This is a heterogeneous lesion, however, and I believe it makes sense to actively seek out subsets in the hope, and the expectation, that di¡erences in pathophysiology may lead to useful di¡erences in management. Hunter: With those atrophic people there hasn’t been a lot of documentation of their bone density. I would suspect that they may be having a lot of microfractures and underlying remodelling as a result of this. As an aside, I don’t think oedema is a good term to describe these lesions, because there isn’t a great deal of oedema on pathological specimens. There isn’t a great deal of longitudinal improvement in bone marrow lesions, to suggest that this is going to account for variability in pain on its own. Intraosseus blood pressure or pressure changes in addition to those bone marrow lesions may potentially account for some of that variability. I would be interested in your thoughts about the aetiology of some of the structural and remodelling changes, and the relationship that this intraosseous blood£ow may have with those remodelling changes that are occurring in the subchondral bone. Simkin: The key issue is time. It may be that the ‘oedema’ that we recognize happened within a single event and then it persists. Disordered mechanics then persist because of the loss of that substance. We shouldn’t think of these as things that must be concurrent. References Downey DJ, Simkin PA, Taggart R 1988 The e¡ect of compressive loading on intraosseous pressure in the femoral head in vitro. J Bone Joint Surg Am 70:871^877 Radin EL, Paul IL, Lowy M 1970 A comparison of the dynamic force-transmitting properties of subchondral bone and articular cartilage. J Bone Joint Surg Am 52:444^456 Simkin PA, Peterson JR 2001 The cartilage/bone interface is permeable to saline under physiologic pressures. Arthritis Rheum 44:548(abstr) Simkin PA, Houglum SJ, Pickerell CC 1985 Compliance and viscoelasticity of canine shoulders loaded in vitro. J Biomech 18:735^743 Todd RC, Freeman MAR, Pirie CJ 1972 Isolated trabecular fatigue fractures in the femoral head. J Bone Joint Surg Br 54:723^728 Wicks M, Garrett R, Vernon-Roberts B, Fazzalari NL 1982 Absence of metabolic bone disease in the proximal femur in patients with fracture of the femoral neck. J Bone Joint Surg Br 64: 319^322

Osteoarthritic Joint Pain: Novartis Foundation Symposium 260. Volume 260 Edited by Derek J. Chadwick and Jamie Goode Copyright ¶ Novartis Foundation 2004. ISBN: 0-470-86763-9

Structural associations of osteoarthritis pain: lessons from magnetic resonance imaging Philip G Conaghan and David T. Felson* Academic Unit of Musculoskeletal Disease, University of Leeds & Department of Rheumatology, Leeds General In¢rmary, Great George Street, Leeds LS1 3EX, UK and *Multipurpose Arthritis and Muscoskeletal Diseases Center, Boston University School of Medicine, 715 Albany Street, A 203, Boston, MA 02118-2526, USA

Abstract. For many years the search for structural associations of osteoarthritis (OA) pain were based on conventional radiographic imaging that predominantly visualizes bone. As well as being tomographic, magnetic resonance imaging (MRI) has the ability to directly visualize all the structures of a joint, including soft tissue and cartilage. Initial MRI studies focused on cartilage assessment, but recently there has been a growing body of work examining the correlation of structural ¢ndings with pain in OA and their relation to structural progression. Painful OA knees have more MRI-detected abnormalities and these pathologies are often correlated making individual contributions di⁄cult to assess. However, in large cohort studies, both synovial hypertrophy and large synovial e¡usions were demonstrated to be more frequent in patients with OA knee pain. Similarly MRI-determined subchondral bone marrow oedema lesions (BME), particularly large ones, are associated with OA knee pain. Meniscal tears in OA knees, although common, have not been linked with pain. Improved, reliable quanti¢cation of the structural features and the rapid advances in MRI technology can only improve structure^pain understanding. 2004 Osteoarthritic joint pain. Wiley, Chichester (Novartis Foundation Symposium 260) p 191^205

It is currently accepted that the osteoarthritis (OA) process involves the whole joint organ including the cartilage, synovium, subchondral bone, menisci and ligaments (Felson et al 2000). However traditional research into structure^pain associations relied on the conventional radiograph which predominantly images only bone and a surrogate measure of cartilage thickness, the joint space. Although the odds for knee pain generally increase with radiographic severity of OA (Felson et al 1987, Lethbridge-Cejku et al 1995), signi¢cant discordance between clinical and radiographic changes has been described in community based cohorts (Dieppe et al 1997, Creamer & Hochberg 1997, Hannan et al 2000). 191

192

CONAGHAN & FELSON

Magnetic resonance imaging (MRI) provides the ability to visualize not only all the relevant structures within a joint, but its tomographic nature allows for imaging in three dimensions. Hence whole organ evaluation of this complex disease process is possible (Peterfy 2002). Until recently, much of the OA MRI literature focused on cartilage with attention to scoring cartilage defects and analysing cartilage volume. Since the cartilage tissue is not supplied with nerves, it seems an unlikely primary source of joint pain (even though it can produce proin£ammatory molecules); consequently this review will focus on the relatively few, recent OA MRI studies pertaining to non-cartilage structures and pain. Preliminary data from whole OA joint evaluation will be presented before reviewing information on individual structures. Of course, structure^pain studies cannot fully explain the complex and personal pain process, and this presentation should be viewed in the context of the entire Symposium proceedings. Structure^pain study limitations Before proceeding it is worthwhile considering the limitations of all studies on structure^pain associations (and the majority to date concern the knee). How was subject pain identi¢ed and quanti¢ed? The phasic or episodic nature of pain may interfere in detecting associations, so inclusion criteria specifying duration and frequency of pain may in£uence study results. Which pain measure is chosen (examples include global, night or weightbearing pain) may also in£uence the outcome. How was the structure visualized? Even using the conventional radiograph of the knee, many studies have imaged only the tibiofemoral joint and not the patellofemoral joint. Arthroscopy may not fully visualize the posterior portion of a joint. Studies employing MRI bring a host of novel variables to the analysis, including di¡erent magnet strengths, di¡erent sequences, lack of standardized de¢nitions of structures and, for noncartilage structures, no consensus on quanti¢cation of structural abnormalities. This area of MRI is rapidly expanding and improving but care should be taken in interpreting and comparing studies. Whole organ evaluation Recently preliminary data have been reported from large studies employing MRI to assess the whole joint. Although there is no consensus yet on scoring methods

OA PAIN AND MRI

193

for assessing the multiple structures within a joint, these studies have employed a whole-organ semi-quantitative OA knee scoring system (the WORMS score) (Peterfy et al 2004). This complex scoring system assesses 14 anatomical features at multiple intra-articular sites. The features are scored with semi-quantitative scales varying from 0^1 to 0^7. Some important messages are emerging from these novel cohorts. It is clear that the radiographically ‘normal’ knee may be far from normal in structure. A recent report evaluated subjects from the Health ABC study (a biracial community based cohort aged 70^79), many of whom had normal knee radiographs (Taouli et al 2002). This study demonstrated that over 75% had some cartilage abnormalities while 30^60% had meniscal tears, bone marrow oedema (BME), bone cysts and synovitis. There was at least a moderate correlation between the presence of cartilage defects and the other structural abnormalities. Importantly, when painful and contralateral painless knees in the cohort were comparted, higher total WORMS scores were found in painful knees (Wildy et al 2002), especially in the tibiofemoral compartments. Surprisingly, the MRI abnormalities were equally present in both painful and non-painful patellofemoral compartments, suggesting that patellofemoral compartments may not be the source of pain in many persons. Synovitis, e¡usions and synovial cysts Synovitis in OA, although secondary, is common; synovial abnormalities are present from the earliest stages of OA and the severity of synovitis is generally related to the severity of chondropathy in the a¡ected joint (see Fig. 1; Myers et al 1990, Smith et al 1997, Loeuille et al 2002). This synovitis is the source of many pro-in£ammatory cytokines and pain mediators (Pelletier et al 2001). Synovitis is commonly assessed with MRI using the intravenous, paramagnetic-enhancing agent gadolinium (
stergaard et al 1997), although non-gadolinium sequences can be optimized for this purpose (Peterfy et al 1994). Studies have correlated the MRI synovial hypertrophy seen in OA with microscopic synovial in£ammation (Fernandez-Madrid et al 1995,
stergaard et al 1997). One recent cross-sectional MRI studied synovitis, e¡usions and popliteal cysts in a large cohort of OA knee subjects comparing persons with symptomatic knee OA with persons with radiographic OA but without symptoms recruited predominantly from Veterans A¡airs medical centres in Boston, USA (Hill et al 2001). The study employed semi-quantitative scoring systems for assessing e¡usions and cysts which were graded as absent, small, moderate or large and correspondingly scored 0^3. Synovial hypertrophy was assessed as present or absent (scored 0 or 1) at three sites: the infrapatellar fat pad, intercondylar space and anterior horn of the lateral meniscus. There was a signi¢cant increase in the

194

CONAGHAN & FELSON

FIG. 1. Sagittal MRI image of an osteoarthritic knee demonstrating a large amount of synovitis in the supra-patella pouch and a large anterior tibial osteophyte (courtesy of Dr Andrew Grainger, Leeds, UK).

frequency of both e¡usions (moderate or large) and synovial hypertrophy in the painful knees compared to those without pain, after adjustment for the radiographic OA severity. In subjects with knee pain and radiographic OA, there was an association between synovitis and pain severity. Popliteal cysts were common (420%) amongst those people without knee pain and not surprisingly they were associated with e¡usions; they were not associated with pain. Another clue to the importance of e¡usions in OA pain derives from a large, cross-sectional study of painful OA knees studied with ultrasonography (D’Agostino et al 2003). This study demonstrated that the presence of e¡usion

OA PAIN AND MRI

195

correlated with sudden aggravation of pain in the week prior to ultrasonography. However not all imaging studies have demonstrated synovial associations with pain. Another large OA knee study employed arthroscopy of the medial compartment and graded the synovium according to macroscopic appearance as normal, reactive or in£ammatory (Ayral et al 2002). There was no association between this synovial score and subject pain. The di¡erent methods in classi¢cation of synovitis and the di¡erent sites evaluated may account for these discrepant ¢ndings. Bone marrow oedema MRI allows the evaluation of the subchondral bone, for many years considered important in the pain and structural progression of OA (Dieppe 1999, Bollet 2001). The commonest MRI subchondral abnormality is BME, ill-de¢ned high signal areas seen on fat-suppressed T2-weighted or STIR sequences (see Fig 2; Peterfy 2002). It is not known exactly what these lesions represent, and this MRI feature is not speci¢c for OA, with similar appearances seen in trauma, osteomyelitis and rheumatoid arthritis (Adalberth et al 1997, Bollet 2001, Conaghan et al 2003). Two small studies have matched the histological ¢ndings from tibial plateau bone with the site of MRI BME in OA patients (Bergman et al 1994, Zanetti et al 2000). In the larger of these studies (16 patients) abnormal tissue was only seen in half the sites corresponding to BME, with marrow necrosis, ¢brosis and abnormal, remodelled bony trabeculae being the commonest abnormalities and actual oedema being a very uncommon ¢nding (Zanetti et al 2000). A link with structural deterioration has been demonstrated for BME lesions: in 256 OA knees followed for up to 30 months, there was a strong association between progressive radiographic joint space loss and the presence of BME in the same (medial or lateral) compartment of the knee, even when adjusted for a known risk factor, knee alignment (Felson et al 2003). Just as importantly, these BME lesions have been linked to OA pain in recent studies. One interesting study described a range of subchondral tibial abnormalities (using T1-weighted images and therefore not directly comparable to other BME studies) in OA patients selected for recent onset of symptoms (mean duration 6 months) and followed for up to 4.5 years (Lotke et al 2000). The lesions described by widespread MRI subchondral changes extending into the metaphysis and with MRI appearance similar to that of osteonecrosis (which typically demonstrates BME) were predominantly associated with persistence of pain over the follow-up period. The most convincing study on the importance of BME and pain involved 401 radiographic OA knee participants, 50 of whom had no knee pain (Felson et al 2001). Subjects had coronal T2-weighted fat-suppressed MRI scans and BME

196

CONAGHAN & FELSON

FIG. 2. Axial fat-suppressed MRI image of an osteoarthritic knee demonstrating massive bone marrow oedema in a femoral condyle and to a lesser extent in the patella (courtesy of Dr Damien Loeuille, Nancy, France).

lesions were graded 0^3 depending on size. Frequency of BME lesions increased with radiographic grade (Kellgren Lawrence, KL, graded on posteroanterior radiographs only) of OA: 48% of KL grade 0 had BME compared with 100% of those with KL grade 4. The BME lesions were present in 78% of the painful knees compared with 30% of the non-painful group (P 50.001), but remarkably large lesions (graded 2 or 3) were present in 36% versus 2% respectively (P50.001). However, this study did not demonstrate an association of BME with pain severity. Another recent study looked at 120 women divided into 4 groups on the basis of presence or absence of knee pain and presence or absence of radiographic changes (Sowers et al 2003). This study used appropriate proton

OA PAIN AND MRI

197

density, fat-suppressed sequences to identify BME, which was graded by size on a 0^2 (grade 0 absent, grade 1 51 cm2, grade 2 41 cm2) scale. Although BME was common and its frequency similar in painful and non-painful OA groups, larger lesions (41 cm2) were more frequent in the painful OA knee group (36 versus 14% in the painless OA group, P50.05). In line with the radiographic ¢ndings in the previous study (Felson et al 2001), women with these large BME lesions were more likely to have full thickness cartilage defects; the painful radiographic OA group with full thickness cartilage defects, adjacent subchondral cortical bone abnormalities and BME had signi¢cantly greater likelihood of painful OA (OR 3.2). Periarticular lesions It is certainly possible that knee pain in OA subjects may arise from extra-articular structures, such as the bursae located around the joint. This concept may be supported by a unique study using local anaesthetic to examine knee pain (Creamer et al 1996). In this study, 6 out of 10 subjects receiving intra-articular local anaesthetic had complete relief of pain at 1 hour, suggesting that not all structures giving rise to joint pain were in contact with the joint cavity. Hill and colleagues looked at the prevalence and pain relationship of MRI-detected periarticular lesions in their Boston cohort (Hill et al 2003). They categorized abnormalities as being peripatellar (prepatellar, super¢cial or deep infrapatella bursitis) or periarticular (including semimembranosus-tibial collateral ligament bursitis, anserine bursitis, iliotibial band syndrome or tibio¢bular cyst). The frequency of peripatellar lesions was not signi¢cantly di¡erent between participants with radiographic OA with and without knee symptoms (12% versus 21% respectively). However periarticular pathology was seen more frequently in the radiographic OA knee pain group than in the pain-free group (15% versus 4% respectively, P ¼0.004). Neither peripatellar nor periarticular lesions were seen in subjects without pain or radiographic OA. Menisci MRI has long been seen as the best non-invasive test for evaluating meniscal pathology (Cheung et al 1997); meniscal damage, and in particular meniscectomy, has also been associated with subsequent increased risk of symptomatic and radiographic OA (Roos et al 2001, Englund et al 2003). Abnormalities of the menisci are very common in OA: the prevalence of meniscal tears in a large elderly community-based cohort (the Health ABC study referred to above) with radiographic knee OA subjects was 83% for men and 73% for women (Guermazi et al 2002). Another study also demonstrated a high prevalence of

198

CONAGHAN & FELSON

meniscal tears even in elderly patients without symptoms, and that tears were more common in symptomatic OA knees compared with asymptomatic controls (91% versus 76% respectively, P50.005) (Bhattacharyya et al 2003). This latter study also showed that the frequency of meniscal tears increased with higher KL radiographic grade, but that both pain and function were no di¡erent between OA patients with and without meniscal tears. This suggests that meniscal tears are not contributing to pain in OA knees, although they may occasionally contribute to mechanical (locking) symptoms. Ligaments The association of anterior cruciate ligament (ACL) tears with subsequent development and progression of radiographic OA is well described, and combined injuries involving collateral ligaments result in higher incidence of OA (Lundberg & Messner 1997, Gillquist & Messner 1999). Modern MRI cohorts have again suggested a surprisingly high frequency of ligamentous abnormalities in OA patients, with the Health ABC study demonstrating partial or complete ligament tear (cruciate or collateral ligaments) frequencies of 27% in men and 30% in women (Guermazi et al 2002). Preliminary data from a Boston cohort demonstrated a complete ACL tear in 26% of 234 subjects with painful, radiographic OA knee (Amin et al 2003). Analysis of this cohort demonstrated that the ACL tears were not associated with worse pain or disability when compared to those subjects without ACL tears, nor was it associated with greater progression of pain over a 30 month follow-up. Osteophytes Osteophytes are integral to radiographic de¢nitions of OA, and MRI studies have demonstrated correlations between radiographic osteophytes and cartilage defects in all compartments of the knee (Boegrd et al 1998a,b). Not surprisingly, the multi-planar nature of MRI has resulted in a marked sensitivity to osteophyte detection (see Fig. 1) and the Health ABC cohort reported osteophytes in 72% of men and 67% of women (Taouli et al 2002). As yet MRI studies have not suggested an association of osteophytes with joint pain, although radiographic studies have previously demonstrated this association (Cicuttini et al 1996). Summary This presentation could not attempt to address the rapid improvements in MRI technology that will improve joint evaluation. The remarkable ability of MRI to visualize all the structures in the OA joint has already dramatically and rapidly

OA PAIN AND MRI

199

increased our understanding of potential sources of OA pain, although such studies have been limited largely to the knee. Important candidate features include synovitis and e¡usions, BME and periarticular lesions such as anserine bursitis. The knowledge emerging from MRI cohorts will help us better understand pain and structural progression and their complex inter-relationship, ultimately allowing for a rational basis for therapeutic strategies.

References Adalberth T, Roos H, Lauren M et al 1997 Magnetic resonance imaging, scintigraphy, and arthroscopic evaluation of traumatic hemarthrosis of the knee. Am J Sports Med 25:231^237 Amin S, LaValley MP, Niu J et al 2003 Complete anterior cruciate ligament (ACL) tear and risk for radiographic and symptom progression in subjects with knee osteoarthritis. Arthritis Rheum 48:S70 Ayral X, Pickering EH, Woodworth TG, Loose LD, MacKillop N, Dougados M 2002 Synovitis is not correlated with the level of symptomatic severity in painful knee osteoarthritis patients. Ann Rheum Dis 61:S37 Bergman AG, Willen HK, Lindstrand AL, Pettersson HT 1994 Osteoarthritis of the knee: correlation of subchondral MR signal abnormalities with histopathologic and radiographic features. Skeletal Radiol 23:445^448 Bhattacharyya T, Gale D, Dewire P et al 2003 The clinical importance of meniscal tears demonstrated by magnetic resonance imaging in osteoarthritis of the knee. J Bone Joint Surg Am 85:4^9 Boegrd T, Rudling O, Petersson IF, Jonsson K 1998a Correlation between radiographically diagnosed osteophytes and magnetic resonance detected cartilage defects in the patellofemoral joint. Ann Rheum Dis 57:395^400 Boegrd T, Rudling O, Petersson IF, Jonsson K 1998b Correlation between radiographically diagnosed osteophytes and magnetic resonance detected cartilage defects in the tibiofemoral joint. Ann Rheum Dis 57:401^407 Bollet AJ 2001 Edema of the bone marrow can cause pain in osteoarthritis and other diseases of bone and joints. Ann Intern Med 134:591^593 Cheung LP, Li KCP, Hollett MD, Bergman AG, Herfkens RJ 1997 Meniscal tears of the knee: accuracy of detection with fast spin-echo MR imaging and arthroscopic correlation in 293 patients. Radiology 203:508^512 Cicuttini FM, Baker J, Hart DJ, Spector TD 1996 Association of pain with radiological changes in di¡erent compartments and views of the knee joint. Osteoarthritis Cartilage 4:143^147 Conaghan PG, O’Connor P, McGonagle D et al 2003 Elucidation of the relationship between synovitis and bone damage: a randomised MRI study of individual joints in patients with early rheumatoid arthritis. Arthritis Rheum 48:64^71 Creamer P, Hochberg MC 1997 Why does osteoarthritis of the knee hurt  sometimes? Br J Rheumatol 36:726^728 Creamer P, Hunt M, Dieppe P 1996 Pain mechanisms in osteoarthritis of the knee: e¡ect of intraarticular anesthetic. J Rheumatol 23:1031^1036 D’Agostino MA, Le Bars M, Schmidely N et al 2003 Interest of ultrasonography to detect synovitis in painful knee osteoarthritis in daily practice. Arthritis Rheum 48:S80 Dieppe PA, Cushnaghan J, Shepstone L 1997 The Bristol ‘OA500’ Study: progression of osteoarthritis (OA) over 3 years and the relationship between clinical and radiographic changes at the knee joint. Osteoarthritis Cartilage 5:87^97

200

CONAGHAN & FELSON

Dieppe P 1999 Subchondral bone should be the main target for the treatment of pain and disease progression in osteoarthritis. Osteoarthritis Cartilage 7:325^326 Englund M, Roos E, Lohmander LS 2003 Impact of meniscal tear on radiographic and symptomatic knee osteoarthritis: a sixteen year followup of meniscectomy with matched controls. Arthritis Rheum 48:2178^2187 Felson DT, Naimark A, Anderson J, Kazis L, Castelli W, Meenan RF 1987 The prevalence of knee osteoarthritis in the elderly: the Framingham Osteoarthritis Study. Arthritis Rheum 30:914^918 Felson DT, Lawrence RC, Dieppe PA et al 2000 Osteoarthritis: new insights. Part 1: the disease and its risk factors. Ann Intern Med 133:635^646 Felson DT, Chaisson CE, Hill CL et al 2001 The association of bone marrow lesions with pain in knee osteoarthritis. Ann Intern Med 134:541^549 Felson DT, McLaughlin S, Goggins J et al 2003 Bone marrow edema and its relation to progression of knee osteoarthritis. Ann Intern Med 139:330^336 Fernandez-Madrid F, Karvonen RL, Teitge RA et al 1995 Synovial thickening detected by MR imaging in osteoarthritis of the knee con¢rmed by biopsy as synovitis. Magn Reson Imaging 13:177^183 Gillquist J, Messner K 1999 Anterior cruciate ligament reconstruction and the long-term incidence of gonarthrosis. Sports Med 27:143^156 Guermazi A, Taouli B, Lynch JA et al 2002 Prevalence of meniscus and ligament tears and their correlation with cartilage morphology and other MRI features in knee osteoarthritis (OA) in the elderly. The Health ABC Study. Arthritis Rheum 46:S567 Hannan MT, Felson DT, Pincus T 2000 Analysis of the discordance between radiographic changes and knee pain in osteoarthritis of the knee. J Rheumatol 27:1513^1517 Hill CL, Gale DG, Chaisson CE et al 2001 Knee e¡usions, popliteal cysts, and synovial thickening: association with knee pain in osteoarthritis. J Rheumatol 28:1330^1337 Hill CL, Gale DR, Chaisson CE et al 2003 Periarticular lesions detected on magnetic resonance imaging: prevalence in knees with and without symptoms. Arthritis Rheum 48:2836^2844 Lethbridge-Cejku M, Scott WW Jr, Reichle R et al 1995 Association of radiographic features of osteoarthritis of the knee with knee pain: data from the Baltimore Longitudinal Study of Aging. Arthritis Care Res 8:182^188 Loeuille D, Toussaint F, Champigneulles J et al 2002 MR evaluation of synovial in£ammation in knee OA: histological correlation. Arthritis Rheum 46:S566 Lotke PA, Ecker ML, Barth P, Lonner JH 2000 Subchondral magnetic resonance imaging changes in early osteoarthrosis associated with tibial osteonecrosis. Arthroscopy 16:76^81 Lundberg M, Messner K 1997 Ten-year prognosis of isolated and combined medial collateral ligament ruptures. A matched comparison in 40 patients using clinical and radiographic evaluations. Am J Sports Med 25:2^6 Myers SL, Brandt KD, Ehlich JW et al 1990 Synovial in£ammation in patients with early osteoarthritis of the knee. J Rheumatol 17:1662^1669
stergaard M, Stoltenberg M, Lovgreen-Nielsen P, Volck B, Jensen CH, Lorenzen I 1997 Magnetic resonance imaging-determined synovial membrane and joint e¡usion volumes in rheumatoid arthritis and osteoarthritis. Comparison with the macroscopic and microscopic appearance of the synovium. Arthritis Rheum 40:1856^1867 Pelletier JP, Martel-Pelletier J, Abramson SB 2001 Osteoarthritis, an in£ammatory disease. Potential implication for the selection of new therapeutic targets. Arthritis Rheum 44:1237^ 1247 Peterfy CG 2002 Imaging of the disease process. Curr Opin Rheumatol 14:590^596 Peterfy CG, Majumdar S, Lang P, van Dijke CF, Sack K, Genant HK 1994 MR imaging of the arthritic knee: improved discrimination of cartilage, synovium, and e¡usion with pulsed saturation transfer and fat-suppressed T1-weighted sequences. Radiology 191:413^419

OA PAIN AND MRI

201

Peterfy CG, Guermazi A, Zaim S et al 2004 Whole-Organ Magnetic Resonance Imaging Score (WORMS) of the knee in osteoarthritis. Osteoarthritis Cartilage 12:177^190 Roos EM, Ostenberg A, Roos H, Ekdahl C, Lohmander LS 2001 Long-term outcome of meniscectomy: symptoms, function, and performance tests in patients with or without radiographic osteoarthritis compared to matched controls. Osteoarthritis Cartilage 9:316^324 Smith MD, Trianta¢llou S, Parker A, Youssef PP, Coleman M 1997 Synovial membrane in£ammation and cytokine production in patients with early osteoarthritis. J Rheumatol 24:365^371 Sowers MF, Hayes C, Jamadar D et al 2003 Magnetic resonance-detected subchondral bone marrow and cartilage defect characteristics associated with pain and X-ray-de¢ned knee osteoarthritis. Osteoarthritis Cartilage 11:387^393 Taouli B, Guermazi A, Zaim S et al 2002 Prevalence and correlates of knee cartilage defects, meniscal lesions and other abnormalities evaluated by MRI in a population sample of knees with normal x-rays. The Health ABC Study. Arthritis Rheum 46:S148 Wildy KS, Nevitt MC, Kwoh CK et al 2002 MRI ¢ndings associated with knee pain: analysis of discordant knee pairs in Health ABC. Arthritis Rheum 46:S148 Zanetti M, Bruder E, Romero J, Hodler J 2000 Bone marrow edema pattern in osteoarthritic knees: correlation between MR imaging and histologic ¢ndings. Radiology 215:835^840

DISCUSSION Brandt: Are there any studies looking for the presence of e¡usion by both ultrasound and MRI in the same patient? Conaghan: Yes. Back in 1995 Mikkel
stergaard looked at OA or rheumatoid arthritis (RA) knees. He did correlations and e¡usion was the feature that correlated best (
stergaard et al 1995). Ultrasound detected e¡usions 30% more often than clinical examination. Brandt: What about the presence and size of e¡usions, as judged by ultrasound versus MRI of the same knee? Conaghan: When e¡usions are seen by both modalities they correlate very well, but MRI is more sensitive. Creamer: I have a question about osteophytes on MRI: are they all the same? There was a thought that perhaps growing osteophytes might be painful as they irritate the periosteum, whereas more established osteophytes may not be. When MRI ¢rst came out we noticed some osteophytes that were brighter. Conaghan: It would be nice to go back and look at some of the cohorts according to bone oedema in the osteophytes alone. This hasn’t really been looked at. There are a lot of data to be gained out of the cohorts we have. I suspect that while the osteophytes are remodelling we might have some activity. This is why I say that BME isn’t BME, because the bone oedema in a remodelling osteophyte is probably going to be quite di¡erent to the bone oedema of microfractured trabeculae. Felson: I think the radiographic studies are pretty consistent in suggesting a correlation between the presence of osteophytes or their size and pain. Creamer: Is that correlation with pain with the presence or absence of pain?

202

DISCUSSION

Felson: No, it was correlated with severity. Schaible: I think this is an important study. Presumably you had a good reason for doing MRI scans of the patients. Did all of them come because they had some pain or knee problems? Conaghan: There are a number of di¡erent cohorts, but of the two largest I talked about, one of them is a community based cohort, so these were not just symptomatic cohorts. The problems with the smaller cohorts is that they are just symptomatic presentations. I think we learn more from the community-based cohorts. Schaible: In the community cohorts are there people with no pain in which you check whether they have any knee abnormalities? Felson: I was involved in both cohorts that Philip described. The Health ABC cohort is a population-based cohort of people in the USA aged 70^79. Roughly half of the MRIs were from people in that large population-based cohort who happened to have knee pain, and the other half were in people without any knee pain. We designed the pain MRI study so we could control for confounders of pain such as activity and medication. We compared a unilateral painful knee with the other knee that wasn’t painful to see whether the MRI features were di¡erent. In the BOKS study, the comparison was between a person who has symptomatic OA and other people who have radiographic OA without symptoms. The comparisons in the BOKS study were between a large number of people with knee pain and a much smaller number that don’t have any symptoms, but have radiographic disease. Pisetsky: I have a comment about synovitis. We have drugs from RA that work on synovitis. When you see this amount of synovitis in OA, do you do anything therapeutically di¡erent? Conaghan: That is exactly why we are looking at methods of determining this. What clinicians want out of this is an algorithm that enables us to pick the subgroups that we can target appropriate therapies to. Perhaps even hyaluronans could work if we can target them to the right group. Hunter: Another aspect that needs to be taken into account is the way that pain is measured. In the Health ABC study they are looking at WOMAC global pain, not speci¢cally activity-related pain. Koltzenburg: Are there techniques such as voxel-based morphometry that you can apply in longitudinal studies? Can you use gadolinium to gauge the intensity of the synovitis, or other contrast media? Conaghan: There is a lot of work on the technology that I didn’t present. There is also a lot of work on cartilage volume. Cartilage reliability and precision is much worse in areas where there are curved surfaces, so the knee is better than the hip to work on. In terms of quantitating, we have done work on synovial volume. There is quite a lot of published work on this.

OA PAIN AND MRI

203

Koltzenburg: Have people looked at the e¡ects of non-steroidal antiin£ammatory drugs (NSAIDs) on synovial thickness? Conaghan: Not at that sort of level in OA. In RA we have looked at this, but we haven’t published this yet. We are able to see di¡erences in synovial volume. Hunter: There are a number of studies including a recent one from Canada looking at the use of intra-articular steroids (Raynauld et al 2003). This suggested that there was little if any measurable change on synovitis scored semiquantitatively. Conaghan: Yes, there are six or seven studies on intra-articular steroid studies using MRI as an outcome. But there are no published studies for NSAIDs. Brandt: We have performed a pilot study in patients with knee OA who were on NSAIDs, whom we studied with a battery of imaging procedures. The things that were most closely related to the increase in joint pain after they were washed out of their NSAIDs were changes in synovial e¡usion volume and in synovial volume. Van den Berg: I was intrigued with what you said about the patellar^femoral lesions not relating to pain. Let me try to make a parallel with animal model studies. If you make an OA model through direct enzymatic injury, for example, you see damage at multiple sites in the joint. If you make a model with biomechanical insults, inducing joint instability, you will normally only see lesions in the tibial^femoral area. Your selective pain story might suggest that only such biomechanically induced lesions will give an e¡usion at load bearing areas that will relate to pain, which you don’t have if you make OA in a nonbiomechanical way. There might be a relationship between biomechanical instability and joint pain, and having lesions of the tibial^femoral area. Conaghan: I think that is right: position is important and the location of bone oedema is giving us a clue to what is important. But I think if you have patellar^ femoral chondropathy, you don’t get synovitis there but somewhere else. We have to look at the whole joint. I ¢nd these data really interesting. The worst group for a clinician to deal with in the knee is patellar^femoral arthritis. This is the area we can do the least for. Herzog: If you do make instabilities in the tibial^femoral joint in some animal models you can clearly see an e¡ect on the patella^femoral joint. Van den Berg: The patella^femoral area is not showing the dominant lesions, though. I am not saying that it is not there. However, if you make OA lesions without an instability, then normally the lesions in the tibial^femoral joint are less dominant compared with the patella^femoral area. This is what we see in mice, which might of course be di¡erent in other species. I have a comment about synovitis. It is interesting to look at this in people and induced models of OA, but to be honest, if you look in spontaneous OA in animals, there is hardly any synovitis. Why does it have such prominence in human disease,

204

DISCUSSION

while it is not necessarily involved in the animal lesions which go to full destruction? Conaghan: I have done some work with animal models and I used to think this was because of weight bearing. But even the big animal models don’t seem to get the degree of synovitis that humans get later on. I can’t explain this. Creamer: With regard to the common appearance of synovitis in human OA whereas it is rarely seen in animal OA, this may be a presentation bias. If it is causing symptoms, it may be that these are the people that we are seeing in human OA, whereas there are lots of people out there with OA who we never get to see. Pisetsky: Have you correlated those with synovitis with any biomarker such as C reactive protein (CRP)? Conaghan: No. Felson: There were two very nice large-scale papers on CRP and its association with synovitis and OA at the 2002 American College of Rheumatology (ACR) meeting. One was from the Health ABC study (Nevitt et al 2002) with a very large n and lots of synovitis on MRI. There was a univariate correlation with CRP, but CRP is very strongly correlated with weight, as is OA. Once weight was adjusted for there was nothing left. This was also seen in a radiographic study from North Carolina (Jordan et al 2002). The best current data therefore suggest that CRP and OA are not on their own correlated with one another, but that this is mediated through a relationship with obesity. Pisetsky: In that study, were those who had synovitis any di¡erent to those who did not? Felson: The CRP study from Health ABC was a synovitisCRP correlative study. Van den Berg: If you want to push synovitis as being something related to pain, then why don’t painkillers work in OA? There is a tendency now to believe that synovitis is not only important in RA but also OA. Perhaps we should even treat OA with anti-tumour necrosis factor (TNF) and anti-interleukin (IL)1 if this is the case. But then if you think the synovitis is really contributing, you would expect that the pain would be rather similar in RA and OA, but it is not. The OA pain is probably more related to oedema in the bone as compared with things taking place in the synovial tissue. Conaghan: I think they are both important. I didn’t present the data on synovitis and progression of chondropathy: this is now pretty clear cut. Two large studies have shown that synovitis is related to progression of chondropathy both at the patellar femoral joint and in the medial tibiofemoral joint. For progression alone, synovitis is worth treating. For pain, I think there are some parameters we need to tweak. One of them is how we measure pain, and secondly our measures of synovitis so far have not been very thorough. They have been global measures of

OA PAIN AND MRI

205

synovitis rather than compartment speci¢c measures. I think this literature will come out in the next year. Dieppe: I sense there are dangerous concepts of causality creeping into this conversation. You could interpret all these data as saying that bad OA does badly: everything is worse in the worse people. Let’s beware of linking these causally in casual conversation. References Jordan JM, Luta G, Stabler T 2002 Serum C-reactive protein (CRP) and osteoarthritis (OA). Arthritis Rheum 46(suppl):S148 Nevitt M, Felson D, Peterfy C et al 2002 In£ammation markers (CRP, TNF-a, IL-6) are not associated with radiographic or MRI ¢ndings of knee OA in the elderly: the Health ABC Study. Arthritis Rheum 46(suppl):S148
stergaard M, Court-Payen M, Gideon P et al 1995 Ultrasonography in arthritis of the knee. A comparison with MR imaging. Acta Radiologica 36:19^26 Raynauld JP, Buckland-Wright C, Ward R et al 2003 Safety and e⁄cacy of long-term intraarticular steroid injections in osteoarthritis of the knee: a randomized, double-blind, placebo-controlled trial. Arthritis Rheum 48:370^377 [Erratum in Arthritis Rheum 48:3300]

Osteoarthritic Joint Pain: Novartis Foundation Symposium 260. Volume 260 Edited by Derek J. Chadwick and Jamie Goode Copyright ¶ Novartis Foundation 2004. ISBN: 0-470-86763-9

The role of TRP channels in sensory neurons Martin Koltzenburg Neural Plasticity Unit, Institute of Child Health, 30 Guildford Street, London WC1N 1EH, UK

Abstract. Two parallel processes characterize the contemporary pain ¢eld. Firstly, enormous progress is being made in the discovery of the cellular and molecular mechanisms responsible for the pathogenesis of pain and secondly, there is a growing appreciation that multiple mechanisms contribute to common clinical pain syndromes. The aim of this chapter is to provide a short overview how transient receptor potential (TRP) channels could contribute to acute and chronic pain states. TRP channels of the vanilloid family (TRPV1, TRPV2, TRPV3, TRPV4) are excited by heat stimuli whereas TRPM8 and ANKTM1 are cold responsive. TRPV1 and ANKTM1 are mediating the pungency of nociceptor-speci¢c chemicals such as capsaicin or mustard oil. Sensitization of TRPV1 is an important mechanisms for heat hyperalgesia and thus the generation of chronic pain symptoms. 2004 Osteoarthritic joint pain. Wiley, Chichester (Novartis Foundation Symposium 260) p 206^220

Several TRP channels are expressed in primary sensory neurons Transient receptor potential (TRP) channels were ¢rst described in the photoreceptors of the fruit £y Drosophila. In mammals TRP channels ful¢l diverse functions and to date six subfamilies have been recognized on the basis of their structural homology (Minke & Cook 2002, Clapham 2003). Following the seminal discovery of the capsaicin receptor TRPV1 (Caterina et al 1997) it has been recognized that several other TRP channels are expressed in dorsal root ganglia or peripheral tissues innervated by sensory neurons, including TRPV2 (Caterina et al 1999), TRPV3 (Peier et al 2002a, Smith et al 2002, Xu et al 2002 ), TRPV4 (Trost et al 2001, Guler et al 2002, Nilius et al 2003, Vriens et al 2004), TRPM8 (McKemy et al 2002, Peier et al 2002b) and ANKTM1 (Story et al 2003, Jordt et al 2004). The sensitivity of nociceptors in normal tissues There is broad consensus that acute pains as studied under laboratory conditions are evoked by excitation of nociceptors with thin myelinated or unmyelinated 206

TRP CHANNELS IN SENSORY NEURONS

207

axons (Meyer et al 1994, Handwerker 1996, Raja et al 1999). Using microneurographic recordings of primary a¡erents and psychophysical magnitude estimation techniques simultaneously in conscious humans, investigators have shown that cutaneous nociceptors can encode the intensity of painful heat (Gybels et al 1979), mechanical (Koltzenburg & Handwerker 1994, Schmidt et al 2000, Schmelz et al 2003) or chemical stimuli (Schmelz et al 2003). However, it is also agreed that ¢ring of a single nociceptor cannot necessarily be equated with the perception of pain. In human microneurographic experiments, it is not uncommon to activate individual nociceptors with painless stimuli. Moreover, mechano-heat sensitive C-¢bres (also known as polymodal nociceptors) innervating the hairy skin have thresholds around 41^43 8C, but the psychophysical heat pain threshold of individuals is often considerably higher (LaMotte et al 1984). Correspondingly, increases of skin temperature that evoke a low discharge rate of less than 0.3 impulses per second over short periods are usually painless (Van Hees & Gybels 1981). Furthermore, brief mechanical impact stimuli can elicit bursts of activity with instantaneous frequencies exceeding 10 Hz without being called painful (Koltzenburg & Handwerker 1994). Moreover, there is also often a considerable time lag between the ¢ring of nociceptors and the appearance of pain following application of algesic chemicals (Adriaensen et al 1980). In aggregate, these results led to the conclusion that both temporal and spatial summation in a population of nociceptors is important for encoding the magnitude of pain (Raja et al 1999). Apart from anatomical studies that have localized many TRP channels to sensory neurons there is now a large body of evidence that TRP channels are the molecular correlate for many of the receptive properties of nociceptors. Heterologous expression studies have shown that TRPV1 (Caterina et al 1997), TRPV2 (Caterina et al 1999), TRPV3 (Peier et al 2002a, Smith et al 2002, Xu et al 2002) and TRPV4 (Guler et al 2002) respond to heat. Mutant mice lacking TRP channels have been very informative in revealing the contribution of TRP channels for the functional properties of sensory neurons. Capsaicin has long been known to speci¢cally excite nociceptors and mice lacking TRPV1 are completely insensitive to this and related irritants and display a strongly reduced response to protons indicating that TRPV1 is a main transducer in the peripheral pain pathway (Caterina et al 2000, Davis et al 2000). TRPV1 knockout mice also show a reduced sensitivity to strong noxious heat stimuli. However, while the heatinduced currents of dorsal root ganglia in culture are completely abolished there remains a signi¢cant heat sensitivity in sensory neurons recorded in situ (Caterina et al 2000). The discrepancy for this observation is not fully understood. However, one possible explanation that would be compatible with these ¢nding is that there are other heat transducing receptors in the target tissue that could release mediators and thereby indirectly excite heat-sensitive nociceptors by a TRPV1-independent

208

KOLTZENBURG

mechanism. Indeed, other heat-sensitive TRP channels such as TRPV3 and TRPV4 are found in keratinocytes or other epithelial cells, but although well positioned for such a task their relatively low heat thresholds are not compatible with a role in nociception (Caterina 2003, Chung et al 2003). TRPV2 is a very strong candidate to mediate the heat sensitivity of thin myelinated nociceptors. TRPV2 is found in neurons that do not express TRPV1 with larger cell sizes than those of TRPV1-containing neurons, and it is coexpressed with markers for myelinated ¢bres (Caterina et al 1999). Importantly, the properties in the heterologous system resemble the heat currents in some capsaicin-insensitive sensory neurons recorded in culture (Nagy & Rang 1999). This has led to the suggestion that TRPV2 is the key heat transducing molecule on myelinated capsaicin-insensitive nociceptors whereas TRPV1 is the important heat sensor on thin myelinated and unmyelinated capsaicin-sensitive nociceptors (Caterina & Julius 2001). Because both TRPV3 and TRPV4 are readily activated by innocuous temperatures they are unlikely to play a signi¢cant role in nociception in isolation. However, because TRPV3 is found in many capsaicin-sensitive neurons it is possible that it modulates the sensitivity of TRPV1, possibly by forming heteromultimers (Smith et al 2002). TRPV4 shows a wide spectrum of sensitivity including heat, arachidonic acid derivates, endocannabinoids and hypo-osmolar stimuli (Guler et al 2002, Nilius et al 2003, Suzuki et al 2003, Watanabe et al 2003) and animals lacking TRPV4 have reportedly a mild impairment in their mechanical nociception (Suzuki et al 2003). Interestingly, hypo-osmolar stimuli, leading to swelling of neurons, have been used in culture to simulate mechanical stimuli (Viana et al 2001) and cells responding to this stimulus express TRPV4 (Alessandri-Haber et al 2003). TRP channels have also been identi¢ed as the crucial molecules in cold sensation (Jordt et al 2003). TRPM8 is found in a small percentage of sensory neurons in the trigeminal or dorsal root ganglia. In heterologous expression studies TRPM8 is activated by small reduction of temperatures and is sensitized by the cooling compound menthol (McKemy et al 2002, Peier et al 2002b). The properties of this channel suggest that it is mediating the excitation of non-nociceptive cold sensitive neurons that signal innocuous cold (Konietzny 1984). However, many sensory neurons in the skin are also excited by noxious cold stimuli and it is generally thought that these neurons do not respond to menthol. Another coldsensitive TRP channel, ANKTM1, has many properties that make it a strong candidate for mediating the excitation of cold sensitive nociceptors. This includes the restricted distribution amongst dorsal root ganglion neurons that also express markers for nociceptors (Story et al 2003), a cold activation threshold in the noxious temperature range (Story et al 2003) and the response to

TRP CHANNELS IN SENSORY NEURONS

209

the irritant mustard oil (Jordt et al 2004) which is known to speci¢cally excite nociceptors (Reeh et al 1986). The sensitivity of nociceptor changes during tissue in£ammation The properties of nociceptors change profoundly following tissue injury and in£ammation. The release of in£ammatory mediators usually activates nociceptors and protons, bradykinin, serotonin and prostaglandins (Reeh & Kress 1995) are amongst the most potent substances that excite nociceptive terminals whereas non-nociceptive a¡erents are typically not a¡ected (Reeh & Kress 1995, Handwerker 1996, Raja et al 1999). While part of a persistent change in excitability is undoubtedly the consequence of the maintained availability of mediators, there is also evidence that chronic in£ammation leads to long-lasting changes in the receptive properties of nociceptors. Nerve growth factor (NGF) appears to play a prominent role in this process of acute and long-term sensitization. NGF tissue concentration increases rapidly in in£ammatory lesions (Donnerer et al 1993, McMahon & Bennett 1997) or after application of proin£ammatory cytokines notably interleukin 1b or tumour necrosis factor a (Sa¢eh-Garabedian et al 1995, Woolf et al 1997). Moreover, application of NGF to rodents results in profound hyperalgesia (Lewin & Mendell 1993). In the adult nervous system many nociceptors express receptors for NGF. While all small-diameter sensory neurons require NGF during early development (Crowley et al 1994), it is only the subpopulation of peptidergic neurones which continues to express the NGF receptors TrkA and p75 throughout adult life. Non-peptidergic neurones express the receptor elements for the transforming growth factor (TGF)b-related neurotrophic factor glial cell linederived neurotrophic factor (GDNF), ret and GFRa (Snider & McMahon 1998, Airaksinen et al 1999). This suggests that NGF could sensitize the subpopulation of peptidergic nociceptors. During tissue in£ammation the hallmark of an altered excitability of polymodal nociceptors is the ongoing activity and the strong sensitization to thermal, but usually not to mechanical stimuli. At the onset of an in£ammation there is also the excitation of mechanically insensitive nociceptors that appear to be particularly important to signal some aspects of mechanical hyperalgesia (Schmidt et al 1994, Schmelz et al 2000, Koltzenburg et al 2002). The intensity of the ongoing discharge of these nociceptors correlates with the magnitude of persistent pain and hyperalgesia to heat in humans (Treede et al 1992, Koltzenburg 1995). Electrophysiological recordings of the receptive properties of thin myelinated and unmyelinated nociceptors innervating normal hairy skin or carrageenan-in£amed skin have shown a close correlation between nociceptor excitability and NGF (Koltzenburg et al 1999). Following carrageenan

210

KOLTZENBURG

in£ammation, about half of the nociceptors displayed ongoing activity that was only rarely observed in nociceptors innervating non-in£amed skin. Spontaneously active ¢bres were sensitized to heat and displayed a more than twofold increase in their discharge to a standard noxious heat stimulus, but there were no changes in the same ¢bres to mechanical stimuli. Furthermore, the number of nociceptors responding to the algesic mediator bradykinin increased signi¢cantly. When the NGF-neutralizing molecule TrkA^IgG was coadministered with carrageenan at the onset of the in£ammation, primary a¡erent nociceptors did not sensitize and displayed essentially normal response properties although the in£ammation as evidenced by tissue oedema developed normally. Thus, endogenous NGF is an important factor in the sensitization of nociceptors and it is tempting to suggest that the nociceptors that developed NGF-mediated ongoing activity are peptidergic neurons that express the TrkA receptor. Studies of cultured dorsal root ganglion cells have extended these ¢ndings and have shown that NGF regulates the capsaicin and bradykinin sensitivity (Bevan & Winter 1995, Petersen et al 1998, Nicholas et al 1999). Several studies also investigated the relative contribution of both NGF receptors and have shown that the relative importance of TrkA or p75 may depend on the functional context. Whereas p75 knockout mice develop a normal acute heat hyperalgesia to systemic injection of recombinant human NGF (Bergmann et al 1998), dorsal root ganglion cells from these animals do not show the up-regulation of bradykinin binding sites that can be normally induced by NGF (Petersen et al 1998). In agreement, several studies suggest that the acute e¡ects of NGF resulting in heat hyperalgesia are mediated by TrkA (Lewin & Barde 1996). TRPV1 is an essential component for the behavioural manifestation of heat hyperalgesia, because mice lacking TRPV1 show a complete abolition of the behavioural correlates of hyperalgesia to heat, but not to mechanical stimuli (Caterina et al 2000, Davis et al 2000). The strong link between TRPV1 and in£ammatory mediators for the generation of heat hyperalgesia is also apparent on a cellular level and there are at least two possible explanations for this interaction that are not mutually exclusive. In the model of bradykinin-induced hyperalgesia it appears that the activation of bradykinin type 2 receptors leads to the membrane translocation of protein kinase C epsilon (PKCe) (Cesare et al 1999) and subsequent phosporylation of TRPV1. Another possibility for TRPV1 sensitization is the diminution of phosphatidylinositol-bisphosphate through phospholipase C (PLC)-mediated hydrolysis. Thus, the excitatory PKCe e¡ects and the disinhibition mediated by PLC would in aggregate result in the sensitization of TRPV1 and hence of nociceptors. In conclusion there is an increasing body of evidence implicating TRP channels as a key foundation for the di¡erent functional properties seen in subpopulations of sensory neurons. While several TRP channels of the vanilloid family appear to be

TRP CHANNELS IN SENSORY NEURONS

211

important for the signalling of normal heat, TRPM8 and ANKTM1 are cold sensors. TRPV1 and ANKTM1 mediate the pungency of many irritant chemicals including capsaicin and mustard oil. The role of TRP channels for mechanotransduction is currently less clear although some studies have implicated TRPV4 for this important function. The important role of TRPs under pathological conditions leading to chronic pain channels is demonstrated by the crucial speci¢c involvement of TRPV1 in the generation of heat hyperalgesia. References Adriaensen H, Gybels J, Handwerker HO, Van Hees J 1980 Latencies of chemically evoked discharges in human cutaneous nociceptors and of the concurrent subjective sensations. Neurosci Lett 20:55^59 Airaksinen MS, Titievsky A, Saarma M 1999 GDNF family neurotrophic factor signaling: four masters, one servant? Mol Cell Neurosci 13:313^325 Alessandri-Haber N, Yeh JJ, Boyd AE et al 2003 Hypotonicity induces TRPV4-mediated nociception in rat. Neuron 39:497^511 Bergmann I, Reiter R, Toyka KV, Koltzenburg M 1998 Nerve growth factor evokes hyperalgesia in mice lacking the low-a⁄nity neurotrophin receptor p75. Neurosci Lett 255:87^90 Bevan S, Winter J 1995 Nerve growth factor (NGF) di¡erentially regulates the chemosensitivity of adult rat cultured sensory neurons. J Neurosci 15:4918^4926 Caterina MJ 2003 Vanilloid receptors take a TRP beyond the sensory a¡erent. Pain 105:5^9 Caterina MJ, Julius D 2001 The vanilloid receptor: a molecular gateway to the pain pathway. Annu Rev Neurosci 24:487^517 Caterina MJ, Le¥er A, Malmberg AB et al 2000 Impaired nociception and pain sensation in mice lacking the capsaicin receptor. Science 288:306^313 Caterina MJ, Rosen TA, Tominaga M, Brake AJ, Julius D 1999 A capsaicin-receptor homologue with a high threshold for noxious heat. Nature 398:436^441 Caterina MJ, Schumacher MA, Tominaga M, Rosen TA, Levine JD, Julius D 1997 The capsaicin receptor: a heat-activated ion channel in the pain pathway. Nature 389:816^824 Cesare P, Dekker LV, Sardini A, Parker PJ, McNaughton PA 1999 Speci¢c involvement of PKC-epsilon in sensitization of the neuronal response to painful heat. Neuron 23:617^624 Chung MK, Lee H, Caterina MJ 2003 Warm temperatures activate TRPV4 in mouse 308 keratinocytes. J Biol Chem 278:32037^32046 Clapham DE 2003 TRP channels as cellular sensors. Nature 426:517^524 Crowley C, Spencer SD, Nishimura MC et al 1994 Mice lacking nerve growth factor display perinatal loss of sensory and sympathetic neurons yet develop basal forebrain cholinergic neurons. Cell 76:1001^1011 Davis JB, Gray J, Gunthorpe MJ et al 2000 Vanilloid receptor-1 is essential for in£ammatory thermal hyperalgesia. Nature 405:183^187 Donnerer J, Schuligoi R, Stein C, Amann R 1993 Upregulation, release and axonal transport of substance P and calcitonin gene-related peptide in adjuvant in£ammation and regulatory function of nerve growth factor. Regul Pept 46:150^154 Guler AD, Lee H, Iida T, Shimizu I, Tominaga M, Caterina M 2002 Heat-evoked activation of the ion channel, TRPV4. J Neurosci 22:6408^6414 Gybels J, Handwerker HO, Van Hees J 1979 A comparison between the discharges of human nociceptive nerve ¢bres and the subject’s ratings of his sensations. J Physiol (Lond) 292: 193^206

212

KOLTZENBURG

Handwerker HO 1996 Sixty years of C-¢ber recordings from animal and human skin nerves: historical notes. Prog Brain Res 113:39^51 Jordt SE, McKemy DD, Julius D 2003 Lessons from peppers and peppermint: the molecular logic of thermosensation. Curr Opin Neurobiol 13:487^492 Jordt SE, Bautista DM, Chuang HH et al 2004 Mustard oils and cannabinoids excite sensory nerve ¢bres through the TRP channel ANKTM1. Nature 427:260^265 Koltzenburg M 1995 The stability and plasticity of the encoding properties of peripheral nerve ¢bres and their relationship to provoked and ongoing pain. Semin Neurosci 7:199^210 Koltzenburg M, Handwerker HO 1994 Di¡erential ability of human cutaneous nociceptors to signal mechanical pain and to produce vasodilatation. J Neurosci 14:1756^1765 Koltzenburg M, Bennett DL, Shelton DL, McMahon SB 1999 Neutralization of endogenous NGF prevents the sensitization of nociceptors supplying in£amed skin. Eur J Neurosci 11:1698^1704 Koltzenburg M, Handwerker HO, Koerber HR 2002 Diversity of nociceptors
which are important for the generation of clinically relevant persistent pains? 10th World Congress of Pain (abstr 710) Konietzny F 1984 Peripheral neural correlates of temperature sensation in man. Hum Neurobiol 3:21^32 LaMotte RH, Torebj˛rk HE, Robinson CJ, Thalhammer JG 1984 Time-intensity pro¢les of cutaneous pain in normal and hyperalgesic skin: a comparison with C-¢ber nociceptor activities in monkey and human. J Neurophysiol 51:1434^1450 Lewin GR, Mendell LM 1993 Nerve growth factor and nociception. Trends Neurosci 16:353^ 359 Lewin GR, Barde Y-A 1996 Physiology of the neurotrophins. Ann Rev Neurosci 19:289^317 McKemy DD, Neuhausser WM, Julius D 2002 Identi¢cation of a cold receptor reveals a general role for TRP channels in thermosensation. Nature 416:52^58 McMahon SB, Bennett DLH 1997 Growth factors and pain. In: Dickenson A, Besson J-M (eds) The pharmacology of pain. Springer Verlag, Berlin, p 135^165 Meyer RA, Campbell JN, Raja SN 1994 Peripheral neural mechanisms of nociception. In: Wall PD, Melzack R (eds) Textbook of pain. Churchill Livingstone, Edinburgh, p 13^44 Minke B, Cook B 2002 TRP channel proteins and signal transduction. Physiol Rev 82:429^472 Nagy I, Rang H 1999 Noxious heat activates all capsaicin-sensitive and also a sub-population of capsaicin-insensitive dorsal root ganglion neurons. Neuroscience 88:995^997 Nicholas RS, Winter J, Wren P, Bergmann R, Woolf CJ 1999 Peripheral in£ammation increases the capsaicin sensitivity of dorsal root ganglion neurons in a nerve growth factor-dependent manner. Neuroscience 91:1425^1433 Nilius B, Watanabe H, Vriens J 2003 The TRPV4 channel: structure-function relationship and promiscuous gating behaviour. P£ˇgers Arch 446:298^303 Peier AM, Reeve AJ, Andersson DA 2002a A heat-sensitive TRP channel expressed in keratinocytes. Science 296:2046^2049 Peier AM, Moqrich A, Hergarden AC et al 2002b A TRP channel that senses cold stimuli and menthol. Cell 108:705^715 Petersen M, Segond vB, Heppelmann B, Koltzenburg M 1998 Nerve growth factor regulates the expression of bradykinin binding sites on adult sensory neurons via the neurotrophin receptor p75. Neuroscience 83:161^168 Raja SN, Meyer RA, Ringkamp M, Campbell JN 1999 Peripheral neural mechanisms of nociception. In: Wall PD, Melzack R (eds) Textbook of pain. Churchill Livingstone, Edinburgh, p 11^57 Reeh PW, Kress M 1995 E¡ect of classic algogens. Semin Neurosci 7:221^226 Reeh PW, Kocher L, Jung S 1986 Does neurogenic in£ammation alter the sensitivity of unmyelinated nociceptors in the rat? Brain Res 384:42^50

TRP CHANNELS IN SENSORY NEURONS

213

Sa¢eh-Garabedian B, Poole S, Allchorne A, Winter J, Woolf CJ 1995 Contribution of interleukin-1 beta to the in£ammation-induced increase in nerve growth factor levels and in£ammatory hyperalgesia. Br J Pharmacol 115:1265^1275 Schmelz M, Schmid R, Handwerker HO, Torebj˛rk HE 2000 Encoding of burning pain from capsaicin-treated human skin in two categories of unmyelinated nerve ¢bres. Brain 123:560^571 Schmelz M, Schmidt R, Weidner C, Hilliges M, Torebjork HE, Handwerker HO 2003 Chemical response pattern of di¡erent classes of C-nociceptors to pruritogens and algogens. J Neurophysiol 89:2441^2448 Schmidt R, Schmelz M, Torebjork HE, Handwerker HO 2000 Mechano-insensitive nociceptors encode pain evoked by tonic pressure to human skin. Neuroscience 98:793^800 Schmidt RF, Schaible H-G, Messlinger K, Heppelmann B, Hanesch U, Pawlak M 1994 Silent and active nociceptors: structure, functions and clinical implications. In: Gebhart GF, Hammond DL, Jensen TL (eds) Proceedings of the 7th World Congress on Pain. IASP Press, Seattle, p 213^250 Smith GD, Gunthorpe MJ, Kelsell RE et al 2002 TRPV3 is a temperature-sensitive vanilloid receptor-like protein. Nature 418:186^190 Snider WD, McMahon SB 1998 Tackling pain at source: new ideas about nociceptors. Neuron 20:629^632 Story GM, Peier AM, Reeve AJ et al 2003 ANKTM1, a TRP-like channel expressed in nociceptive neurons, is activated by cold temperatures. Cell 112:819^829 Suzuki M, Mizuno A, Kodaira K, Imai M 2003 Impaired pressure sensation in mice lacking TRPV4. J Biol Chem 278:22664^22668 Treede R-D, Meyer RA, Raja SN, Campbell JN 1992 Peripheral and central mechanisms of cutaneous hyperalgesia. Prog Neurobiol 38:397^421 Trost C, Bergs C, Himmerkus N, Flockerzi V 2001 The transient receptor potential, TRP4, cation channel is a novel member of the family of calmodulin binding proteins. Biochem J 355:663^670 Van Hees J, Gybels J 1981 C nociceptor activity in human nerve during painful and non painful skin stimulation. J Neurol Neurosurg Psychiat 44:600^607 Viana F, de la PE, Pecson B, Schmidt RF, Belmonte C 2001 Swelling-activated calcium signalling in cultured mouse primary sensory neurons. Eur J Neurosci 13:722^734 Vriens J, Watanabe H, Janssens A, Droogmans G, Voets T, Nilius B 2004 Cell swelling, heat, and chemical agonists use distinct pathways for the activation of the cation channel TRPV4. Proc Natl Acad Sci USA 101:396^401 Watanabe H, Vriens J, Prenen J, Droogmans G, Voets T, Nilius B 2003 Anandamide and arachidonic acid use epoxyeicosatrienoic acids to activate TRPV4 channels. Nature 424:434^ 438 Woolf CJ, Allchorne A, Sa¢eh-Garabedian B, Poole S 1997 Cytokines, nerve growth factor and in£ammatory hyperalgesia: the contribution of tumour necrosis factor alpha. Br J Pharmacol 121:417^424 Xu H, Ramsey IS, Kotecha SA et al 2002 TRPV3 is a calcium-permeable temperature-sensitive cation channel. Nature 418:181^186

DISCUSSION Schaible: I have looked at the literature on the TTX-resistant Na+ channels. Many people say they are only expressed in primary a¡erent neurons that are

214

DISCUSSION

nociceptive, but there is also a big literature on TTX-resistant Na+ channels in the CNS. What is your view? Koltzenburg: Some of these ion channels, in particular the SNS channels, were initially cloned only from dorsal root ganglion (DRG) neurons. Of course, there are TTX-resistant Na+ channels outside the ganglion neurons, but they are not SNS. There are studies that show that some ion channels are up-regulated in pathological conditions in the CNS, and there are certainly also TTX-resistant Na+ channels in the heart but these are not identical to those neuronal ion channels. Grubb: I have been through the arguments before about the acid sensitivity of ion channels. How much acid is required to activate many of these ion channels? Is it within the physiological range? Koltzenburg: There is quite an important species di¡erence. The pH sensitivity in the rat is much stronger than in the mouse. This could suggest that there may be di¡erent stoichiometries of ion channels expressed in sensory neurons in the rat and mouse. There is consensus that in the mouse the sustained proton current is mainly mediated by TRPV1. Grubb: You showed very elegantly the B2 receptor modulation of the vanilloid receptor being mediated by PKC. For one of the inward rectifying K+ channels that we are interested in, people have shown in heterologous expression systems that this can be modulated not just by PKC, but also by bg subunits of the G protein-coupled receptor and phosphatidylinositol-4,5-bisphosphate (PIP2). This channel regulates membrane potential and we think it is present in nociceptors. So there are actually three ways of modulating this ion channel. Has this been shown for the TRPV1 receptor? Koltzenburg: Peter McNaughton and David Julius would probably argue that the two hypotheses that I put forward are exclusive. I don’t see them necessarily being exclusive, but I ¢nd it di⁄cult to see why they come to such di¡erent conclusions. David Julius works primarily on heterologous expression systems and this may be one reason for discrepancies. The mechanisms he described are clearly possible for cellular activation, and phosphorylation and sensitization of the TRPV1 channel. I would be surprised if there were not more than one mechanism capable of modulating the function of this channel. Grubb: If you inhibit PKCe, does this completely abolish the B2 sensitization of the TRPV1 receptor? Koltzenburg: That is an interesting question. To my knowledge this has never been systematically investigated. One of the problems is that some of the phorbol esters that are used for PKC activation actually appear to bind and displace RTX, suggesting that they can compete with the vanilloid binding site. This is, of course, a confounding problem. Pisetsky: Can anyone clarify the di¡erence between acute and chronic in£ammation in terms of this pattern? On the one hand, we hear that acute

TRP CHANNELS IN SENSORY NEURONS

215

in£ammation sensitizes; on the other hand, the histology shows that it causes denervation, at least in the joint or synovium. Is the synovium special, or should we look in another tissue? Grubb: I ¢nd this idea that the tissue becomes denervated in in£ammation very unusual. All the evidence that Hans-Georg Schaible has provided in the joint and other people have provided in the skin shows that there is an increased sensitivity and there is no loss. Pisetsky: Do you see new nerves come into those sites? Koltzenburg: We looked in skin biopsies in patients with acute contact dermatitis. There we see a mild increase of GAP43, which is a protein associated with axonal sprouting. Schaible: Both phenomena have been reported, which makes it di⁄cult to sort out. Kidd: I would not use the word denervation. I think what happens is that there is minor retraction for a very small area. It could be a volume e¡ect related to oedema. I don’t believe that there is denervation in synovium. I think there is retraction or oedema to reduce the density. We do not see a neuropathic picture. It is very easy to look for damaged nerves by looking in the DRG. If you have denervation you would expect to see a neuropathic picture in the DRG, which we don’t see. Schaible: The only receptor which is missing now is the warmth receptor, or would you say that there is already a candidate? Koltzenburg: I think there are two receptors, TRPV3 and TRPV4. The Novartis group says that TRPV3 is not expressed on DRG neurons, but there are two other publications that suggest this is the case. If I understand correctly, the Novartis group originally cloned it from DRG libraries, so it is probably in DRG, but possibly in low amounts. The interesting question is why this channel hasn’t been seen in animals lacking TRPV1. In the heterologous expression system for TRPV3, repetition of the stimulus often sensitized the current and it also has very strong outward rectifying properties. When we retrospectively looked at some of our patch clamp recordings in the TRPV1 knockout animal, we sometimes see that there are small inward currents, but they are so small, in both capsaicin-sensitive and -insensitive neurons that we think they are caused by non-speci¢c e¡ects of the heating rather than re£ecting activation of TRPV3. Henry: Your HCN seems to be correlated with IB4, which is the presumed nonnociceptive C ¢bre. All three seem to be more highly correlated with IB4 in the capsaicin. This would suggest that if you are able to nullify the activation of that channel, you might modify alterations in non-nociceptive mechanisms. Koltzenburg: Recent reports have shown that there is an increase of the Ih current in myelinated ¢bres after peripheral nerve damage (Yao et al 2003, Chaplan et al 2003). To my knowledge this hasn’t been investigated in in£ammation and these two studies focused on non-nociceptive a¡erents. Chaplan et al (2003) proposed

216

DISCUSSION

that the increased Ih current contributed to the ectopic ¢ring of large myelinated ¢bres after nerve injury. Speci¢c blockers of Ih had some antihyperalgesic properties in the nerve-injury model. Henry: Do those channels shift in animal models or human tissues? Koltzenburg: I think they will shift. There is only one publication out that has looked speci¢cally in the rat at the expression of HCN isoforms after nerve injury using quantitative PCR (Chaplan et al 2003). They actually found a reduction in mRNA levels, but they showed an increase in functional properties of the channels as well as an increase of the protein, implying signi¢cant post-transcriptional regulation. Henry: In relation to types of pain, acute pain is not a big hurdle for medical treatment
the real problem is chronic, non-cancer pain. It would be interesting to see over time how many of these ion channels start to correlate with di¡erent types of pain. Koltzenburg: If one looks at the TTX-resistant Na+ channel Nav1.8, there is some indication that animals lacking this channel exhibit only a mild phenotype for acute cutaneous nociceptive pain. In models of visceral pain, the contribution of Nav1.8 may be stronger. There is a recent publication indicating that spontaneous activity developing in axotomized neurons is actually much lower in animals lacking this TTX-resistant Na+ channel (Roza et al 2003). Fox: We have also found that TRPV1 blockers are e¡ective in all sorts of models of pain. What is interesting is that this includes in£ammatory pain and neuropathic pain: this was a bit of a surprise to us. The end point we used was mechanical hyperalgesia as well as thermal hyperalgesia. It is not just the neurons that are polymodal; these ion channels are responding to all sorts of things, perhaps in ways that we don’t yet understand. Koltzenburg: I’d have thought that TRPV1 was not directly involved in mechanotransduction. Fox: Somehow the ligand it is responding to endogenously must be, in some way, doing something to the neuron that is responsive to it. Henry: We have also found that TRPV1 receptor inhibitor or antagonist is e¡ective in our rheumatoid arthritis models. Schaible: A potentially important question is whether we should block ion channels or mediators and mediator receptors. Koltzenburg: Blockade of conduction in a nociceptive neuron would prevent signalling, regardless of which receptors it expresses at the receptive terminal. Thus, this would provide analgesia for a large number of stimuli or algesic mediators. On the other hand, if you have an in£ammatory cascade involving molecules such as TNFa, blockade of a key component upstream of the synthesis of other in£ammatory mediators may also be very e¡ective. In rheumatoid arthritis

TRP CHANNELS IN SENSORY NEURONS

217

it appears that if you block the crucial molecule, you also inhibit all the downstream signalling and the in£ammation. It probably depends on the pathophysiology. It would be ideal if we had a mechanism that blocked exclusively the action potential propagation in nociceptors. We all know that local anaesthetics are the best analgesics. One of the reasons we don’t give them more often is because of motor side e¡ects and complete sensory loss. But if you have something that would exclusively block nociceptors you would have an ideal analgesic, especially if you could modify it so that it just reduces the frequency of ¢ring. In fact ideally you want to have a reduction in ¢ring that results in anti-hyperalgesia, but not in anaesthesia. Grubb: Regulating the frequency is key. In a C ¢bre, when we elicit an action potential, the cell after-hyperpolarizes and then it takes ages for its resting membrane potential to return to normal
longer than a second. We are interested in which ion channels regulate that repolarization. In repetitive ¢ring this after-hyperpolarization gets deeper and longer. This is regulated by prostaglandins, histamine and bradykinin. One of the candidate channels is the small-conductance Ca2+-activated K+ channel. Rediske: Do you think this TRPV family could play a role as noxious stimuli receptors in non-neuronal cells? Koltzenburg: They certainly produce an inward current in keratinocytes. It is possible. The reason why I am not too keen on that particular hypothesis is because heat sensitivity is restricted to those neurons that express the capsaicin receptor. If there were a generalized secondary e¡ect of TRPV channel activation by heat-sensitive keratinocytes and a release of a mediator, I would assume a more widespread heat activation of nociceptors unless one assumes that TRPV1 and a receptor for the elusive keratinocyte mediator were expressed on the same neuron. Pisetsky: Many of the agents that work in the animal models of rheumatoid arthritis are anti-in£ammatory and analgesic. Are there any data on the e¡ects of just analgesics in these models, in terms of the in£ammation and structural damage? Is there a way that this may make things worse? Koltzenburg: In some conditions there is a discrepancy between oedema and nociceptor in£ammation. This can be seen in models of carrageenan where there is frequently profound oedema, but sometimes no sensitization. It may also depend on which tissue you study. Many studies have focused on hairy skin and there is room for the tissues to expand, whereas in more con¢ned spaces it would be di⁄cult and the resulting pressure would contribute to nociceptor activation. Conversely, if NGF is sequestered in the carrageenan model there is reduction of in£ammation, but the increased excitability of the nociceptors is strongly reduced. Thus both phenomena of nociceptor sensitization and oedema can be dissociated.

218

DISCUSSION

Schaible: Most of these mediators act on nerve cells and other cells. What is your opinion of the importance of the opioid receptors? Koltzenburg: The opioid receptors and other potentially analgesic molecules such as cannabinoid receptors are expressed there. It seems that the opioid receptors become primarily functional under in£ammatory conditions. Pisetsky: Do opioids make arthritis worse? Grubb: There are studies where opiates have been injected intra-articularly. In humans they produce a degree of analgesia, but I have no idea whether this a¡ects disease progression. Mackintosh: There is evidence that they reduce cellular in£ux. If you count that as a component of in£ammation then they are antiin£ammatory. Schaible: Theoretically they would be ideal for treating pain with because it doesn’t matter how the pain is produced. Felson: It is probably time to bring in an entity called the analgesic hip. There was a study looking at people on the waiting list for total hip replacement, comparing people treated with potent anti-in£ammatories with those who weren’t. The people treated with anti-in£ammatories had more degradation and degeneration of their hip and worse pathology at surgery. This suggested that pain is protective and functions to help us know when to protect our joints. Too much suppression of pain may not be healthy. I guess this calls into question this whole symposium! One wonders whether, by messing around too much with nature, we might not be helping people. Koltzenburg: It is very clear that we don’t want to have an anaesthetic condition. This is probably best illustrated in people with congenital analgesia. The organs that are most e¡ectively traumatized tend to be joints. Felson: We talked about Charcot arthropathy yesterday. This is probably more than just pain and temperature loss. It is almost certainly vibration and position sense loss. This would be an example of pharmacological-induced Charcot arthropathy. Kidd: I think we might be unnecessarily damning analgesia here. I think you are confusing analgesia with anti-in£ammatory. Non-steroidal anti-in£ammatory drugs (NSAIDs) do more than just act as antihyperalgesics. What we need is a comparable study with comparable algesia to see whether the same e¡ect is achieved. We must also remember that the NSAIDs are not giving you an anaesthetic joint, they are just reducing the sensitization. They are antihyperalgesic as opposed to analgesic in the true sense of the word. If we are going to incriminate NSAIDs let’s look at alternative mechanisms for e¡ects on blood £ow and so on, not focusing exclusively on the analgesic e¡ect. There is more than one explanation of that study.

TRP CHANNELS IN SENSORY NEURONS

219

Dieppe: I agree with that. The NSAIDs also have a signi¢cant e¡ect on bone. There is some other evidence that this so-called indomethicin or analgesic hip is actually a bone problem. Grubb: In a way, the holy grail is to leave the in£ammation alone and let it try to do the repair process that it is trying to do, and block one of these important ion channels that Martin Koltzenburg has been talking about. We don’t know what the mechanical transducer is. If we can ¢nd this, and it turns out to be selectively expressed in small C ¢bres, then we are on to a winner. Fernihough: Do we know the expression pattern of TRPV1 in structures within the joint? Koltzenburg: There is one anatomical study that found TRPV1 in a¡erents innervating the facet joint of the rat. The percentage of joint a¡erents expressing TRPV1 was reported to be much lower than for the DRG population at large (Ohtori et al 2001). TRPV1 has also been found in non-neuronal cells such as urothelium. However, if one studies the expression level with immunohistochemistry it is not very high (Birder et al 2002). Schaible: In our previous recordings in the cat we could block a lot of these joint a¡erents with a higher dose of capsaicin (He et al 1990). At that time it was more fashionable to study desensitization. I would assume that many of them are capsaicin sensitive, but this has not been shown histologically. Kidd: From my rheumatological rather than pain perspective, listening to this debate on TRPV1 in which you have told us that it is a promiscuous receptor in that it responds to heat but also protons and other things, in the joint heat is not such an important thing so could it be that protons are the most important activator? When I get a sore back I get a burning sort of pain, so do you think TRPV1 could explain a lot of the symptoms that people get? Acidic environments might be triggering this. Koltzenburg: If you induce spontaneous activity in capsaicin-sensitive nociceptors, this results in pain and also central sensitization. Depending on how strong this particular component is in deep somatic tissue, this would then be an important aspect. If one looks in skeletal muscle, it appears that the responses to capsaicin and the proton responses via TRPV1 are rather small, compared with some of the acid-sensing ion channels (Benson et al 1999). At this moment in time we have little information about the relative importance of TRPV1 and ASICs in skeletal muscle and joint. Alyson Fox referred to some of the work done at Novartis where they used antagonists. They found a much wider e⁄cacy than you would expect from the studies on mice lacking TRPV1. The knockout studies are quite clear that you would not predict an involvement in cutaneous mechanical hyperalgesia and neuropathic pain. However, from what Alison has said it appears that if you use TRPV1 antagonists you observe a wider-spread e¡ectiveness. This can be

220

DISCUSSION

explained in a couple of ways. It could be that compensatory changes occur in the transgenic model. It is also possible that there are other non-TRPV1 e¡ects of these drugs. References Benson CJ, Eckert SP, McCleskey EW 1999 Acid-evoked currents in cardiac sensory neurons: a possible mediator of myocardial ischemic sensation. Circ Res 84:921^928 Birder LA, Nakamura Y, Kiss S et al 2002 Altered urinary bladder function in mice lacking the vanilloid receptor TRPV1. Nat Neurosci 5:856^860 Chaplan SR, Guo HQ, Lee DH et al 2003 Neuronal hyperpolarization-activated pacemaker channels drive neuropathic pain. J Neurosci 23:1169^1178 He X, Schepelmann K, Schaible HG, Schmidt RF 1990 Capsaicin inhibits responses of ¢ne a¡erents from the knee joint of the cat to mechanical and chemical stimuli. Brain Res 530:147^150 Ohtori S, Takahashi K, Chiba T, Yamagata M, Sameda H, Moriya H 2001 Brain-derived neurotrophic factor and vanilloid receptor subtype 1 immunoreactive sensory DRG neurons innervating the lumbar facet joints in rats. Auton Neurosci 94:132^135 Roza C, Laird JM, Souslova V, Wood JN, Cervero F 2003 The tetrodotoxin-resistant Na+ channel Nav1.8 is essential for the expression of spontaneous activity in damaged sensory axons of mice. J Physiol 550:921^926 Yao H, Donnelly DF, Ma C, LaMotte RH 2003 Upregulation of the hyperpolarization-activated cation current after chronic compression of the dorsal root ganglion. J Neurosci 23: 2069^2074

Osteoarthritic Joint Pain: Novartis Foundation Symposium 260. Volume 260 Edited by Derek J. Chadwick and Jamie Goode Copyright ¶ Novartis Foundation 2004. ISBN: 0-470-86763-9

Mechanisms that generate and maintain bone cancer pain Patrick W. Mantyh and Stephen P. Hunt*1 Departments of Preventive Sciences, Neuroscience, Psychiatry, and the Cancer Center, University of Minnesota, 18-208 Moos Tower, 515 Delaware Street SE, Minneapolis, MN 55455, USA and *Department of Anatomy and Developmental Biology, Medawar Building, University College London, Gower Street, London WC1E 6BT, UK

Abstract. Although bone cancer pain can be severe and is relatively common, as it frequently arises from metastases from breast, prostate, and lung tumours, very little is known about the basic mechanisms that generate and maintain this chronic pain. To begin to de¢ne the mechanisms that give rise to bone cancer pain, we have developed mouse and rat models using the intramedullary injection and containment of tumour cells into the femur. These tumour cells induced bone remodelling as well as ongoing and movement evoked pain behaviours similar to that found in patients with bone cancer pain. In addition there is a signi¢cant reorganization of the spinal cord that received sensory input from the cancerous bone and this reorganization generated a neurochemical signature of bone cancer pain that is both dramatic and signi¢cantly di¡erent from that observed in mouse and rat models of chronic neuropathic or in£ammatory pain. These models have provided insight into the mechanisms that drive cancer pain and have begun to allow the development of mechanism-based therapies. Together these advances should reduce tumour-induced pain and su¡ering and signi¢cantly improve the quality of life of cancer patients. 2004 Osteoarthritic joint pain. Wiley, Chichester (Novartis Foundation Symposium 260) p 221^240

The negative impact that cancer pain has on quality of life cannot be overestimated. As advances in cancer detection and therapy are extending the life expectancy of cancer patients, there is increasing focus on improving the patients’ quality of life. Many patients present with pain as the ¢rst sign of cancer and 30^50% of all cancer patients will experience moderate to severe pain (Mercadante 1997, Mercadante & Arcuri 1998, Portenoy & Lesage 1999, Portenoy et al 1999). Cancer-associated pain can be present at any time during the evolution of the disease, but the frequency and intensity of cancer pain tends to increase with advancing stages of 1This

paper was presented at the symposium by Stephen P. Hunt. 221

222

MANTYH & HUNT

cancer. 75^95% of patients with metastatic or advanced cancer will experience signi¢cant, life-altering cancer-induced pain (Mercadante 1997, Mercadante & Arcuri 1998, Portenoy & Lesage 1999, Portenoy et al 1999). The treatment of cancer pain can involve a variety of modalities. Therapies targeted at decreasing tumour size are often e¡ective and include radiation, chemotherapy, and/or surgery, but these can be burdensome to administer and are accompanied by signi¢cant unwanted side e¡ects. Medications targeted at decreasing in£ammation and pain such as non-steroidal anti-in£ammatory drugs (NSAIDs) or opiates can also be very useful, but these too are accompanied by unwanted side e¡ects. The relative ine¡ectiveness of current treatments re£ects the fact that therapies have not changed for decades (Coyle et al 1990, Payne 1997, Payne et al 1998, Hoskin 2000). Largely because of treatment-associated side e¡ects, it has been reported that 45% of cancer patients have inadequate and under-managed pain control (de Wit et al 2001, Meuser et al 2001). A formidable obstacle to the development of new therapies is the fact that the current neurobiological basis for pharmacological treatments is largely empirical, and is based on scienti¢c advances in painful conditions other than those induced by cancer. Recently, the ¢rst animal models of cancer pain have been developed. In the mouse femur model, bone cancer pain is induced by injecting murine osteolytic sarcoma cells into the intramedullary space of the murine femur (Fig. 1) (Schwei et al 1999). Critical to this model is that the tumour cells are con¢ned within the marrow space of the injected femur and the tumour cells do not invade adjacent soft tissues which would directly impact the joints of the muscle making behavioural analysis problematic (Schwei et al 1999, Honore et al 2000, Luger et al 2001). Following injection the tumour cells proliferate, and ongoing, movement, and touch-evoked pain related behaviours develop that increase in severity with time. These pain behaviours correlate with the progressive tumour-induced bone destruction that ensues and that appears to mimic the condition in patients with primary or metastatic bone cancer (Fig. 1). These models have allowed us to gain mechanistic insights into how cancer pain is generated and how the sensory information it initiates is processed as it moves from sense organ to the cerebral cortex under a constantly changing molecular architecture. As detailed below, these insights promise to fundamentally change the way cancer pain is controlled. Primary a¡erent sensory neurons Primary a¡erent sensory neurons are the gateway by which sensory information from peripheral tissues is transmitted to the spinal cord and brain (Fig. 2), and these neurons innervate the skin and every internal organ of the body including mineralized bone, marrow, and periosteum. The cell bodies of sensory ¢bres that

BONE CANCER PAIN

223

FIG. 1. Progressive destruction of mineralized bone in mice with bone cancer. (A) Low power anterior^posterior radiograph of mouse pelvis and hindlimbs after a unilateral injection of sarcoma cells into the distal part of the femur and closure of the injection site with an amalgam plug (arrow) which prevents the tumour cells from growing outside the bone (Honore et al 2000). Radiographs of murine femora (B) show the progressive loss of mineralized bone caused by tumour growth. These images are representative of the stages of bone destruction in the murine femur. At week 1 there is a minor loss of bone near the distal head (arrow); at week 2 substantial loss of mineralized bone at both the proximal and distal (arrow) head; and at week 3 loss of mineralized bone throughout the entire bone and fracture of distal head of the femur (arrow). Scale bar ¼ 2 mm. (Modi¢ed from Schwei et al 1999.)

224

MANTYH & HUNT

FIG. 2. Sensory neurons and detection of noxious stimuli due to tumour cells. Nociceptors use a diversity of signal transduction mechanisms to detect noxious physiological stimuli and many of these signal transduction mechanisms may be involved in driving cancer pain. Thus, when nociceptors are exposed to products of tumour cells, tissue injury, or in£ammation the excitability is altered and this nociceptive information is relayed to the spinal cord and then onto higher centres of the brain. Some of the mechanisms that appear to be involved in generating and maintaining cancer pain include: activation of nociceptors by factors such as extracellular protons (+), endothelin 1 (ET-1), interleukins (ILs), prostaglandins (PG) and tumour necrosis factor (TNF).

innervate the head and body are housed in the trigeminal and dorsal root ganglia respectively, and can be divided into to two major categories: myelinated A ¢bres and smaller diameter unmyelinated C ¢bres. Nearly all large diameter myelinated Ab ¢bres normally conduct non-noxious stimuli applied to the skin, joints, and muscles and thus these large sensory neurons usually do not conduct noxious stimuli (Djouhri et al 1998). In contrast, most small-diameter sensory ¢bres
unmyelinated C ¢bres and ¢nely myelinated myelinated A ¢bres
are specialized sensory neurons known as nociceptors whose major function is to detect and convert environmental stimuli that are perceived as harmful into electrochemical signals that are then transmitted to the central nervous system. Unlike primary sensory neurons involved in vision or olfaction which are required to detect only one type of sensory stimulus (light or chemical odorants, respectively), individual primary sensory neurons of the pain pathway have the remarkable ability to detect a

BONE CANCER PAIN

225

wide range of stimulus modalities including those of physical and chemical nature (Basbaum & Jessel 2000, Julius & Basbaum 2001). To accomplish this, nociceptors express an extremely diverse repertoire of transduction molecules that can sense forms of noxious stimulation (thermal, mechanical, and chemical), albeit with varying degrees of sensitivity. In the past few years remarkable progress has been made toward understanding the signalling mechanisms and speci¢c molecules that nociceptors use to detect noxious stimuli. For example, the vanilloid receptor (VR-1, or TRPV1) that is expressed by most nociceptors detects heat (Kirschstein et al 1999) and also appears to detect acid (Welch et al 2000), extracellular protons (Bevan and Geppetti 1994, Caterina et al 2000) and lipid metabolites (Tominaga et al 1998, Nagy & Rang 1999). In order to detect noxious mechanical stimuli, nociceptors express mechanically gated channels which initiate a signalling cascade upon excessive stretch (Price et al 2001). The cells also express several purinergic receptors capable of sensing ATP, which is released from cells upon excessive mechanical stimulation (Krishtal et al 1988, Xu & Huang 2002). To sense noxious chemical stimuli, nociceptors express a complex array of receptors capable of detecting in£ammation-associated factors released from damaged tissue including protons (Bevan & Geppetti 1994, Caterina et al 2000), endothelins (Nelson & Carducci 2000), prostaglandins (Alvarez & Fy¡e 2000), bradykinin (Alvarez & Fy¡e 2000) and nerve growth factor (McMahon 1996). Aside from providing promising targets for the development of more selective analgesics, identi¢cation of receptors expressed on the nociceptor surface has increased our understanding how di¡erent tumours generate cancer pain in the peripheral tissues they invade and destroy. In addition to expressing channels and receptors that detect tissue injury, sensory neurons are highly ‘plastic’, in that they can change their phenotype in the face of a sustained peripheral injury. Following tissue injury, sensory neuron sub-populations alter patterns of signalling peptide and growth factor expression (Woolf & Salter 2000). This change in phenotype of the sensory neuron in part underlies peripheral sensitization, whereby the activation threshold of nociceptors is lowered so that what would normally be perceived as a mild noxious stimulus is perceived as highly noxious (hyperalgesia). Damage to a peripheral tissue has also been shown to activate previously ‘silent’ or ‘sleeping’ nociceptors which then become highly responsive to both normally non-noxious (allodynia) and noxious (hyperalgesia) stimuli. There are several examples of nociceptors that undergo peripheral sensitization in experimental cancer models (Schwei et al 1999, Honore et al 2000a,b,c, Luger et al 2001). In normal mice, the neurotransmitter substance P is synthesized by nociceptors and released in the spinal cord in response to a noxious but not a non-noxious palpation of the femur. In mice with bone cancer, normally non-

226

MANTYH & HUNT

FIG. 3. Development of chronic pain in mice with bone cancer. Time course of the development of ongoing, ambulatory, and touch-evoked pain behaviours in na|« ve, sham and sarcoma-injected animals. Note that the development of pain behaviours closely follows the time course of tumour growth and tumour-induced bone remodelling seen in Fig. 1. Grey shading indicates a signi¢cance of P50.05 vs. sham.

painful palpation of the a¡ected femur now induces the release of substance P from primary a¡erent ¢bres that terminate in the spinal cord. Substance P in turn binds to and activates the neurokinin 1 receptor that is expressed by a subset of spinal cord neurons (Mantyh et al 1995a,b, Hunt & Mantyh 2001). Similarly, normally non-noxious palpation of tumour-bearing limbs of mice with bone cancer also induces the expression of c-Fos protein in spinal cord neurons. In normal animals that do not have cancer, only noxious stimuli will induce the expression of c-Fos in the spinal cord (Hunt et al 1987). Peripheral sensitization of nociceptors may be involved in the generation and maintenance of bone cancer pain. Properties of tumours that excite nociceptors Tumour cells and tumour-associated cells that include macrophages, neutrophils, and T lymphocytes secrete a wide variety of factors that have been shown to

BONE CANCER PAIN

227

sensitize or directly excite primary a¡erent neurons (Fig. 3). These include prostaglandins (Nielsen et al 1991, Galasko 1995), endothelins (Nelson & Carducci 2000, Davar 2001), interleukins 1 and 6 (Watkins et al 1995, Opree & Kress 2000, DeLeo & Yezierski 2001), epidermal growth factor (Stoscheck & King 1986), transforming growth factor (Poon et al 2001, Roman et al 2001), and platelet-derived growth factor (Daughaday & Deuel 1991, Radinsky 1991, Silver 1992) and receptors for many of these factors have been shown to be expressed by primary a¡erent neurons. While each of these factors may play an important role in the generation of pain in particular forms of cancer, two that are currently approved for use in humans with other non-cancer indications are the prostaglandins and the endothelins. Prostaglandins are pro-in£ammatory lipids that are formed from arachidonic acid by the action of cyclooxygenase (COX) and other downstream synthetases. There are two distinct forms of the COX enzyme, COX-1 and COX-2. Prostaglandins have been shown to be involved in the sensitization and/or direct excitation of nociceptors by binding to several prostanoid receptors expressed by nociceptors which sensitize or directly excite nociceptors (Vasko 1995). Several tumour cells and tumour-associated macrophages have been shown to express high levels of COX-2 and produce high levels of prostaglandins (Dubois et al 1996, Molina et al 1999, Kundu et al 2001, Ohno et al 2001, Shappell et al 2001). The COX enzymes are a major target of current medications, and COX inhibitors such as aspirin or ibuprofen are commonly administered for reducing both in£ammation and pain. A major problem with mixed COX inhibitors such as aspirin or ibuprofen to block cancer pain is that these compounds inhibit both COX-1 and COX-2, and inhibition of the constitutively expressed COX-1 can cause bleeding and ulcers. In contrast, the new COX-2 inhibitors preferentially inhibit COX-2 and avoid the side-e¡ects of COX-1 inhibition, which may allow their use in treating cancer pain. Other experiments have suggested that COX-2 is involved in angiogenesis and tumour growth (Masferrer et al 2000, Moore & Simmons 2000, Sabino et al 2002), so in cancer patients COX-2 inhibitors may have the added advantage that in addition to blocking cancer pain, a COX-2 inhibitor may also reduce the growth and metastasis of the tumour (Sabino et al 2002). COX-2 antagonists show signi¢cant promise for alleviating at least some aspects of cancer pain, although clearly more research is required to fully de¢ne the actions of COX-2 in di¡erent types of cancer. A second pharmacological target for treating cancer pain, this one a peptide possibly responsible for sensitizing and/or activating primary a¡erent neurons, is endothelin 1. Several tumours including prostate cancer express very high levels of endothelins (Shankar et al 1998, Kurbel et al 1999, Nelson & Carducci 2000) and clinical studies have reported a correlation between the severity of the pain in patients with prostate cancer and plasma levels of endothelins (Nelson et al 1995).

228

MANTYH & HUNT

Endothelins could contribute to cancer pain by directly sensitizing or exciting nociceptors, as a subset of small unmyelinated primary a¡erent neurons express receptors for endothelin (Pomonis et al 2001) and direct application of endothelin to peripheral nerves induces activation of primary a¡erent ¢bres and an induction of pain behaviours (Davar et al 1998). Like prostaglandins, endothelins that are released from tumour cells are also thought be involved in regulating angiogenesis (Dawas et al 1999) and tumour growth (Asham et al 1998), suggesting again that use of endothelin antagonists may be useful not only in inhibiting cancer pain but in reducing the growth and metastasis of the tumour. Tumour-induced release of protons and acidosis Tumour cells become ischaemic and undergo apoptosis as the tumour burden exceeds its vascular supply (Helmlinger et al 2002). Local acidosis, a state where an accumulation of acid metabolites is present, is a hallmark of tissue injury (Reeh & Steen 1996, Julius & Basbaum 2001). In the past few years, the concept that sensory neurons can be directly excited by protons or acidosis has generated intense research and clinical interest. Studies have shown that subsets of sensory neurons express di¡erent acid-sensing ion channels (Olson et al 1998, Julius & Basbaum 2001). The two major classes of acid-sensing ion channels expressed by nociceptors are VR-1 (TRPV1)(Caterina et al 1997, Tominaga et al 1998) and the acid-sensing ion channel 3 (ASIC-3) (Bassilana et al 1997, Olson et al 1998, Sutherland et al 2000). Both of these channels are sensitized and excited by a decrease in pH. More speci¢cally, the VR-1 is activated when the pH falls below 6.0, while the pH that activates ASIC-3 appears to be highly dependent on the coexpression of other ASIC channels in the same nociceptor (Lingueglia et al 1997). There are several mechanisms by which a decrease in pH could be involved in generating and maintaining cancer pain. As tumours grow, tumour-associated in£ammatory cells invade the neoplastic tissue and release of protons that generate local acidosis (Helmlinger et al 2002). A second mechanism by which acidosis may occur is apoptosis of the tumour cells. Release of intracellular ions may generate an acidic environment that activates signalling by acid-sensing channels expressed by nociceptors. Tumour-induced release of protons and acidosis may be particularly important in the generation of bone cancer pain. In both osteolytic (bone destroying) and osteoblastic (bone forming) cancers there is a signi¢cant proliferation and hypertrophy of osteoclasts (Clohisy et al 2000a,b, 2001). Osteoclasts are terminally di¡erentiated, multinucleated, monocyte lineage cells that are uniquely designed to resorb bone by maintaining an extracellular microenvironment of acidic pH (4.0^5.0) at the osteoclast-mineralized bone interface (Delaisse & Vaes 1992). Most sensory neurons that innervate bone express the VR-1 (Guo et al 1999)

BONE CANCER PAIN

229

and/or ASICs (Olson et al 1998), and these sensory neurons will be depolarized and transmit pain signals to the spinal cord if exposed to the osteoclast’s acidic extracellular microenvironment. Recent experiments in a murine model of bone cancer pain reported that osteoclasts play an essential role in cancer-induced bone loss, and that osteoclasts contribute to the aetiology of bone cancer pain (Honore et al 2000, Luger et al 2001). Based on these reports it has recently been shown that osteoprotegerin (Honore et al 2000) or a bisphosphonate (Fulfaro et al 1998, Mannix et al 2000), both of which are known to induce osteoclast apoptosis, are e¡ective at decreasing osteoclast-induced bone cancer pain. Similarly, VR-1 or ASIC antagonists may be used to reduce pain in patients with soft tumours or bone cancer by blocking excitation of the acid sensitive channels on sensory neurons. Release of growth factors by tumour cells One of the most important discoveries in the past decade has been the demonstration that the biochemical and physiological status of sensory neurons is maintained and modi¢ed by factors derived from the innervated tissue. Changes in the periphery associated with in£ammation, nerve injury, or tissue injury are mirrored by changes in the phenotype of sensory neurons (Honore et al 2000). After peripheral nerve injury, expression of a subset of neurotransmitters and receptors by damaged sensory neurons is altered in a highly predictable fashion. These changes are caused, in part, by a change in the tissue level of several growth factors, including nerve growth factor, bradykinin and tumour necrosis factor (Hunt & Mantyh 2002). While tumour invasion alters the invaded tissue, it is also clear that the tissue also in£uences the phenotype of the invading tumour cell (Mundy 2002). Because the local environment can in£uence what tumour cells express and release, it follows that the same tumour in the same individual may be painful at one site of metastasis but not at another. Based on clinical observations, pain from cancer can be quite perplexing, as the size, location, or type of cancer tumour does not necessarily predict symptoms. Di¡erent patients with the same cancer may have vastly di¡erent symptoms. Kidney cancer may be painful in one person and asymptomatic in another. Metastases to bone in the same individual may cause pain at the site of a rib lesion, but not at the site of a humeral lesion. Small cancer deposits in bone may be very painful, while very large soft tissue cancers may be painless (Mantyh et al 2002). Important areas for future research include identi¢cation of tissue-speci¢c mechanisms of cancer pain, i.e. soft tissue vs. bone; site-speci¢c mechanisms of cancer pain, i.e. £at bones (rib) vs. tubular bones (femur); and patient-speci¢c factors that in£uence disease progression and its relationship to pain perception.

230

MANTYH & HUNT

Tumour-induced distention and destruction of sensory ¢bres In reports, data have suggested that tumours are not highly innervated by sensory or sympathetic neurons (O’Connell et al 1998, Seifert & Spitznas 2001, Terada & Matsunaga 2001). However, in a large number of cancers rapid tumour growth frequently entraps and injures nerves, causing mechanical injury, compression, ischaemia or direct proteolysis (Mercadante 1997). Proteolytic enzymes produced by the tumour can also cause injury to sensory and sympathetic ¢bres, causing neuropathic pain. The capacity of tumour to injure and destroy peripheral nerve ¢bres has been directly observed in an experimental model of bone cancer. Following injection and containment of lytic murine sarcoma cells in the intramedullary of the mouse femur, tumour cells grow in the marrow space and disrupt innervating sensory ¢bres. As the tumour cells grow they ¢rst compress and then destroy both the haematopoietic cells that normally comprise the marrow as well as the sensory ¢bres that normally innervate the marrow, mineralized bone and periosteum (Schwei et al 1999). While the mechanisms by which any neuropathic pain is generated and maintained are still not well understood, several therapies that have proved useful in the control of other types of neuropathic pain may also be useful in treating tumour-induced neuropathic pain. For example gabapentin, originally developed as an anticonvulsant but whose mechanism of action still remains unknown, has been shown to be e¡ective in treating several forms of neuropathic pain and may also be useful in treating cancer pain of neuropathic origin (Ripamonti & Dickerson 2001). Central sensitization in cancer pain A critical question is whether the spinal cord and forebrain also undergo signi¢cant neurochemical changes as a chronic cancer pain state develops. In the murine cancer pain model it was observed that there was extensive neurochemical reorganization within spinal cord segments that receive input from primary a¡erent neurons innervating the cancerous bone (Honore et al 2000, Luger et al 2001). These changes included astrocyte hypertrophy (Fig. 4) and up-regulation of the pro-hyperalgesic peptide dynorphin. Spinal cord neurons that normally would only be activated by noxious stimuli were activated by normally nonnoxious stimuli. These spinal cord changes were attenuated by blocking the tumour-induced tissue destruction and pain (Honore et al 2000, Luger et al 2001). Together these neurochemical changes suggest that cancer pain induces and is at least partially maintained by a state of central sensitization in which an increased transmission of nociceptive information allows normally non-noxious input to be ampli¢ed and perceived as noxious stimuli.

BONE CANCER PAIN

231

FIG. 4. Cancer-induced reorganization of the central nervous system. Chronic cancer pain not only sensitizes peripheral nociceptors, but also can induce signi¢cant neurochemical reorganization of the spinal cord which may participate in the phenomenon of central sensitization i.e. an increased responsiveness of spinal cord neurons involved in transmission of pain. Confocal image of a coronal section of the mouse L4 spinal cord showing glial ¢brillary acidic protein (GFAP)-positive astrocytes (white) which have undergone hypertrophy on the side ipsilateral to the tumour-bearing bone. (B, C) Higher magni¢cation photos ipsilateral (B) and contralateral (C) to the dorsal horn seen in (A) where the GFAP staining is present with staining for the neuron speci¢c antibody NeuN. Note that while the astrocytes (spindle-shaped cells) have undergone a massive hypertrophy, there does not appear to be any signi¢cant loss of NeuN-positive neurons (modi¢ed from Schwei et al 1999). Scale bars ¼ A, 200 mm; B, C, 30 mm.

232

MANTYH & HUNT

Once nociceptive information has been transmitted to the spinal cord by primary a¡erent neurons there are multiple ascending ‘pain’ pathways that project from the spinal cord to higher centres of the brain. Classically, main emphasis in examining the ascending conduction of pain has been placed on spinothalamic tract neurons. However, recent data have necessitated a reassessment of this position, for recent clinical studies have shown that signi¢cant attenuation of some forms of di⁄cult to control visceral cancer pain can be achieved following lesion of the axons non-spinothalamic tract neurons (Willis et al 1999, Nauta et al 2000). Together these data suggest that one reason that cancer pain is frequently perceived as such an intense and disturbing pain is that it ascends to higher centres of the brain via multiple parallel neuronal pathways (Hunt & Mantyh 2001). Importantly for cancer patients, it is clear that higher centres of the brain can modulate the ascending conduction of pain so that the general mood and attention of the patient may in itself be a signi¢cant factor in determining the intensity and degree of unpleasantness of the pain. Since many cancer patients are frequently anxious and/or depressed, these descending pathways, which modulate the ascending conduction of cancer pain, may themselves play an important role either enhancing or inhibiting the perception of cancer pain by the patient. A changing set of factors may drive cancer pain as the disease progresses Cancer pain frequently becomes more severe as the disease progresses and control of cancer pain is more di⁄cult to fully achieve without encountering signi¢cant unwanted side e¡ects (Payne 1998, Foley 1999, Portenoy & Lesage 1999). While tolerance may contribute to the escalation of the dose of analgesics required to control cancer pain, a compatible possibility is that with the progression of the disease di¡erent factors assume a greater importance in driving cancer pain. For example, in the mouse model of bone cancer, tumour cells ¢rst begin to proliferate pain-related behaviours that are present long before any signi¢cant bone destruction is evident and this pain may be due to pro-hyperalgesic factors such as prostaglandins and endothelin that are released by the growing tumour cells that activate nociceptors in the marrow and this pain could be attenuated by COX-2 inhibitors and endothelin antagonists. As the tumour continues to grow, sensory neurons innervating the marrow are compressed and destroyed causing a neuropathic pain to develop which may best respond to treatment with drugs which attenuate neuropathic pain such as gabapentin. When the tumour begins to induce proliferation and hypertrophy of osteoclasts, the pain due to excessive osteoclast activity could be largely blocked by anti-osteoclastogenic drugs such as bisphosphonates or osteoprotegerin (Fig. 5). As the tumour cells completely ¢ll

BONE CANCER PAIN

233

FIG. 5. Attenuation of bone cancer pain by osteoprotegerin (OPG). Histograms showing that administration of OPG which was started 6 days following tumour implantation attenuated both spontaneous (A) and palpation-evoked (B) pain in mice at day 17 following tumour implantation (modi¢ed from Honore et al 2000). OPG is a naturally occurring protein that is a secreted decoy receptor that inhibits osteoclast di¡erentiation, proliferation, and hypertrophy resulting in reduced osteoclast activity and bone resorption.

the intramedullary space they begin to die, generating an acidic environment, antagonists to VR-1 or ASICs should attenuate the pain induced by this acidosis. Finally, as bone destruction ensues and the mechanical strength of the bone is compromised, antagonists that block the mechanically gated channels and/or ATP receptors in the richly innervated periosteum may attenuate the movement evoked pain. While the above pattern of tumour-induced tissue destruction and nociceptor activation may be unique to bone cancer, an evolving set of nociceptive events probably occurs in other cancers. This may, in part, explain why cancer pain is frequently di⁄cult to treat and why it is so heterogeneous in nature and severity. That the tumour-induced tissue injury, nociceptor activation, and the CNS brain areas involved in transmitting these nociceptive signals are changing as the disease progresses suggests that di¡erent therapies will be e⁄cacious at particular stages of the disease. Understanding how tumour cells di¡erentially excite nociceptors at di¡erent stages of the disease, and how the phenotype of nociceptors and CNS neurons involved in nociceptive transmission change as the disease progresses should allow a mechanistic approach to designing more e¡ective therapies to treat cancer pain. Progress and future directions For the ¢rst time, animal models of cancer pain are now available that mirror the clinical picture of human patients with cancer pain. Information generated from

234

MANTYH & HUNT

these models should provide insight into the mechanisms that generate and maintain the di¡erent types of cancer pain. Since many of the cancer models have been developed in mice, human tumours can be implanted in immunocompromised mouse strains, which should allow examination of the pain that di¡erent human tumours generate. These animal models may also o¡er insight into one of the major conundrums of cancer pain: that the severity of cancer pain is so variable from patient to patient, tumour to tumour, and even site to site. Newer molecular techniques using microarrays and proteomics should provide data about what speci¢c features of di¡erent tumours are important in inducing cancer pain. Once the mechanisms by which the di¡erent types of cancer induce pain have been elucidated, molecular targets can be identi¢ed and mechanism based therapies developed. Ultimately, the key will be to integrate tumour biology and the host’s response to neoplasia with our understanding of how chronic pain is generated and maintained. These studies should lead to improvement in the quality of life for all those who su¡er from cancer pain.

References Alvarez FJ, Fy¡e RE 2000 Nociceptors for the 21st century. Curr Rev Pain 4:451^458 Asham EH, Loizidou M, Taylor I 1998 Endothelin-1 and tumour development. Eur J Surg Oncol 24:57^60 Basbaum AI, Jessel TM 2000 The perception of pain. In: Kandel ER, Schwartz JH, Jessell TM (eds) Principles of neural science. McGraw-Hill, New York, p 472^490 Bassilana F, Champigny G, Waldmann R et al 1997 The acid-sensitive ionic channel subunit ASIC and the mammalian degenerin MDEG form a heteromultimeric H+-gated Na+ channel with novel properties. J Biol Chem 272:28819^28822 Bevan S, Geppetti P 1994 Protons: small stimulants of capsaicin-sensitive sensory nerves. Trends Neurosci 17:509^512 Caterina MJ, Schumacher MA, Tominaga M et al 1997 The capsaicin receptor: a heat-activated ion channel in the pain pathway. Nature 389:816^824 Caterina MJ, Le¥er A, Malmberg AB et al 2000 Impaired nociception and pain sensation in mice lacking the capsaicin receptor. Science 288:306^313 Clohisy DR, Perkins SL, Ramnaraine ML 2000a Review of cellular mechanisms of tumor osteolysis. Clin Orthop Rel Res 373:104^114 Clohisy DR, Ramnaraine ML, Scully S et al 2000b Osteoprotegerin inhibits tumor-induced osteoclastogenesis and bone growth in osteopetrotic mice. J Orthop Res 18:967^976 Clohisy DR, O’Keefe PF, Ramnaraine ML 2001 Pamidronate decreases tumor-induced osteoclastogenesis in mice. J Orthop Res 19:554^558 Coyle N, Adelhardt J, Foley KM, Portenoy RK 1990 Character of terminal illness in the advanced cancer patient: pain and other symptoms during the last four weeks of life [see comments]. J Pain Symptom Manage 5:83^93 Daughaday WH, Deuel DF 1991 Tumor secretion of growth factors. Endocrinol Metab Clin N Am 20:539^563 Davar G 2001 Endothelin-1 and metastatic cancer pain. Pain Med 2:24^27 Davar G, Hans G, Fareed MU et al 1998 Behavioral signs of acute pain produced by application of endothelin-1 to rat sciatic nerve. Neuroreport 9:2279^2283

BONE CANCER PAIN

235

Dawas K, Laizidou M, Shankar A et al 1999 Angiogenesis in cancer: the role of endothelin-1. Ann R Coll Surg Engl 81:306^310 de Wit R, van Dam F, Loonstra S et al 2001 The Amsterdam Pain Management Index compared to eight frequently used outcome measures to evaluate the adequacy of pain treatment in cancer patients with chronic pain. Pain 91:339^349 Delaisse J-M, Vaes G 1992 Mechanism of mineral solubilization and matrix degradation in osteoclastic bone resorption. In: Rifkin BR, Gay CV (eds) Biology and physiology of the osteoclast. CRC, Ann Arbor, p 289^314 DeLeo JA, Yezierski RP The role of neuroin£ammation and neuroimmune activation in persistent pain. Pain 90:1^6 Djouhri L, Bleazard L, Lawson SN 1998 Association of somatic action potential shape with sensory receptive properties in guinea-pig dorsal root ganglion neurones. J Physiol 513:857^872 Dubois RN, Radhika A, Reddy BS, Entingh AJ 1996 Increased cyclooxygenase-2 levels in carcinogen-induced rat colonic tumors. Gastroenterology 110:1259^1262 Foley KM 1999 Advances in cancer pain. Arch Neurol 56:413^417 Fulfaro F, Casuccio A, Ticozzi C, Ripamonti C 1998 The role of bisphosphonates in the treatment of painful metastatic bone disease: a review of phase III trials. Pain 78:157^169 Galasko CS 1995 Diagnosis of skeletal metastases and assessment of response to treatment. Clin Orthop Rel Res 312:64^75 Guo A, Vulchanova L, Wang J et al 1999 Immunocytochemical localization of the vanilloid receptor 1 (VR1): relationship to neuropeptides, the P2X3 purinoceptor and IB4 binding sites. Eur J Neurosci 11:946^958 Helmlinger G, Sckell A, Dellian M et al 2002 Acid production in glycolysis-impaired tumors provides new insights into tumor metabolism. Clin Cancer Res 8:1284^1291 Honore P, Luger NM, Sabino MA et al 2000a Osteoprotegerin blocks bone cancer-induced skeletal destruction, skeletal pain and pain-related neurochemical reorganization of the spinal cord. Nat Med 6:521^528 Honore P, Menning PM, Rogers SD et al 2000b Neurochemical plasticity in persistent in£ammatory pain. Prog Brain Res 129:357^363 Honore P, Rogers SD, Schwei MJ et al 2000c Murine models of in£ammatory, neuropathic and cancer pain each generates a unique set of neurochemical changes in the spinal cord and sensory neurons. Neuroscience 98:585^598 Honore P, Schwei J, Rogers SD et al 2000 Cellular and neurochemical remodeling of the spinal cord in bone cancer pain. Prog Brain Res 129:389^397 Hoskin P 2000 Radiotherapy. In: Body J-J (ed) Tumor bone diseases and osteoporosis in cancer patients. Marcel Dekker, New York, p 263^286 Hunt SP, Mantyh PW 2001 The molecular dynamics of pain control. Nat Rev Neurosci 2:83^91 Hunt SP, Pini A, Evan G 1987 Induction of c-fos-like protein in spinal cord neurons following sensory stimulation. Nature 328:632^634 Julius D, Basbaum AI 2001 Molecular mechanisms of nociception. Nature 413:203^210 Kirschstein T, Gre¡rath W, Busselberg D, Treede RD 1999 Inhibition of rapid heat responses in nociceptive primary sensory of rats by vanilloid receptor antagonists. J Neurophysiol 82:2853^2860 Krishtal OA, Marchenko SM, Obukhov AG 1988 Cationic channels activated by extracellular ATP in rat sensory neurons. Neuroscience 27:995^1000 Kundu N, Yang QY, Dorsey R, Fulton AM 2001 Increased cyclooxygenase-2 (cox-2) expression and activity in a murine model of metastatic breast cancer. Int J Cancer 93:681^686 Kurbel S, Kurbel B, Kovacic D et al 1999 Endothelin-secreting tumors and the idea of the pseudoectopic hormone secretion in tumors. Med Hypotheses 52:329^333

236

MANTYH & HUNT

Lingueglia E, Weille JR, Bassilana F et al 1997 A modulatory subunit of acid sensing ion channels in brain and dorsal root ganglion cells. J Biol Chem 272:29778^29783 Luger NM, Honore P, Sabino MAC et al 2001 Osteoprotegerin diminishes advanced bone cancer pain. Cancer Res 61:4038^4047 Luger NM, Sabino MAC, Schwei MJ et al 2002 E⁄cacy of systemic morphine suggests a fundamental di¡erence in the mechanisms that generate bone cancer vs. in£ammatory pain. Pain 99:397^406 Mannix K, Ahmedzai SH, Anderson H, Bennet M, Lloyd-Williams M, Wilcock A 2000 Using bisphosphonates to control the pain of bone metastases: evidence based guidelines for palliative care. Palliat Med 14:455^461 Mantyh PW, Allen CJ, Ghilardi JR et al 1995a Rapid endocytosis of a G protein-coupled receptor: substance P evoked internalization of its receptor in the rat striatum in vivo. Proc Natl Acad Sci USA 92:2622^2626 Mantyh PW, DeMaster E, Malhotra A et al 1995b Receptor endocytosis and dendrite reshaping in spinal neurons after somatosensory stimulation. Science 268:1629^1632 Mantyh PW, Clohisy DR, Koltzenburg M, Hunt SP 2002 Molecular mechanisms of cancer pain. Nat Rev Cancer 2:201^209 Masferrer JL, Leahy KM, Koki AT et al 2000 Antiangiogenic and antitumor activities of cyclooxygenase-2 inhibitors. Cancer Res 60:1306^1311 McMahon SB 1996 NGF as a mediator of in£ammatory pain. Phil Trans R Soc Lond B Biol Sci 351:431^440 Mercadante S 1997 Malignant bone pain: pathophysiology and treatment. Pain 69:1^18 Mercadante S, Arcuri E 1998 Breakthrough pain in cancer patients: pathophysiology and treatment. Cancer Treat Rev 24:425^432 Meuser T, Pietruck C, Radbruch L et al 2001 Symptoms during cancer pain treatment following WHO-guidelines: a longitudinal follow-up study of symptom prevalence, severity and etiology. Pain 93:247^257 Molina MA, Sitja-Arnau M, Lemoine MG et al 1999 Increased cyclooxygenase-2 expression in human pancreatic carcinomas and cell lines: growth inhibition by nonsteroidal antiin£ammatory drugs. Cancer Res 59:4356^4362 Moore BC, Simmons DL 2000 COX-2 inhibition, apoptosis, and chemoprevention by nonsteroidal anti-in£ammatory drugs. Curr Med Chem 7:1131^1344 Mundy GR 2002 Metastases to bone: causes, consequences, and therapeutic opportunities. Nat Rev Cancer 2:584^593 Nagy I, Rang H 1999 Noxious heat activates all capsaicin-sensitive and also a sub-population of capsaicin-insensitive dorsal root ganglion neurons. Neuroscience 88:995^997 Nauta HJW, Soukup VM, Fabian RH et al 2000 Punctate midline myelotomy for the relief of visceral cancer pain. J Neurosurg 92:S125^S130 Nelson JB, Carducci MA 2000 The role of endothelin-1 and endothelin receptor antagonists in prostate cancer. BJU Int 85:S45^S48 Nelson JB, Hedican SP, George DJ et al 1995 Identi¢cation of endothelin-1 in the pathophysiology of metastatic adenocarcinoma of the prostate. Nat Med 1:944^949 Nielsen OS, Munro AJ, Tannock IF 1991 Bone metastases: pathophysiology and management policy. J Clin Oncol 9:509^524 O’Connell JX, Nanthakumar SS, Nielsen GP, Rosenberg AE 1998 Osteoid osteoma: the uniquely innervated bone tumor. Mod Pathol 11:175^180 Ohno R, Yoshinaga K, Fujita T et al 2001 Depth of invasion parallels increased cyclooxygenase2 levels in patients with gastric carcinoma. Cancer 91:1876^1881 Olson TH, Riedl MS, Vulchanova L et al 1998 An acid sensing ion channel (ASIC) localizes to small primary a¡erent neurons in rats. Neuroreport 9:1109^1113

BONE CANCER PAIN

237

Opree A, Kress M 2000 Involvement of the proin£ammatory cytokines tumor necrosis factoralpha, IL-1 beta, and IL-6 but not IL-8 in the development of heat hyperalgesia: e¡ects on heat-evoked calcitonin gene-related peptide release from rat skin. J Neurosci 20:6289^6293 Payne R 1997 Mechanisms and management of bone pain. Cancer 80:S1608^S1613 Payne R 1998 Practice guidelines for cancer pain therapy. Issues pertinent to the revision of national guidelines. Oncology 12:169^175 Payne R, Mathias SD, Pasta DJ et al 1998 Quality of life and cancer pain: satisfaction and side e¡ects with transdermal fentanyl versus oral morphine. J Clin Oncol 16:1588^1593 Pomonis JD, Rogers SD, Peters CM et al 2001 Expression and localization of endothelin receptors: implication for the involvement of peripheral glia in nociception. J Neurosci 21:999^1006 Poon RT, Fan ST, Wong J 2001 Clinical implications of circulating angiogenic factors in cancer patients. J Clin Oncol 19:1207^1225 Portenoy RK, Lesage P 1999 Management of cancer pain. Lancet 353:1695^1700 Portenoy RK, Payne D, Jacobsen P 1999 Breakthrough pain: characteristics and impact in patients with cancer pain. Pain 81:129^134 Price MP, McIlwrath SL, Xie JH et al 2001 The DRASIC cation channel contributes to the detection of cutaneous touch and acid stimuli in mice. Neuron 32:1071^1083 Radinsky R 1991 Growth factors and their receptors in metastasis. Semin Cancer Biol 2: 169^177 Reeh PW, Steen KH 1996 Tissue acidosis in nociception and pain. Prog Brain Res 113:143^151 Ripamonti C, Dickerson ED 2001 Strategies for the treatment of cancer pain in the new millennium. Drugs 61:955^977 Roman C, Saha D, Beauchamp R 2001 TGF-beta and colorectal carcinogenesis. Microsc Res Tech 52:450^457 Sabino MAC, Ghilardi JR, Jongen JLM et al 2002 Simultaneous reduction of cancer pain, bone destruction, and tumor growth by selective inhibition of cyclooxygenase 2. Cancer Res 62:7343^7349 Schwei MJ, Honore P, Rogers SD et al 1999 Neurochemical and cellular reorganization of the spinal cord in a murine model of bone cancer pain. J Neurosci 19: 10886^10897 Seifert P, Spitznas M 1998 Tumours may be innervated. Virchows Archiv 438:228^231 Shankar A, Loizidou M, Aliev G et al 1998 Raised endothelin 1 levels in patients with colorectal liver metastases. Br J Surg 85:502^506 Shappell SB, Manning S, Boeglin WE et al 2001 Alterations in lipoxygenase and cyclooxygenase-2 catalytic activity and mRNA expression in prostate carcinoma. Neoplasia 3:287^303 Silver BJ 1992 Platelet-derived growth factor in human malignancy. Biofactors 3:217^227 Stoscheck CM, King Jr LE 1986 Role of epidermal growth factor in carcinogenesis. Cancer Res 46:1030^1037 Sutherland SP, Cook SP, and McCleskey EW 2000 Chemical mediators of pain due to tissue damage and ischemia. Prog Brain Res 129:21^38 Terada T, Matsunaga Y 2001 S-100-positive nerve ¢bers in hepatocellular carcinoma and intrahepatic cholangiocarcinoma: an immunohistochemical study. Pathol Int 51:89^93 Tominaga M, Caterina MJ, Malmberg AB et al 1998 The cloned capsaicin receptor integrates multiple pain-producing stimuli. Neuron 21:531^543 Vasko MR 1995 Prostaglandin-induced neuropeptide release from spinal cord. Prog Brain Res 104:367^380 Watkins LR, Goehler LE, Relton J et al 1995 Mechanisms of tumor necrosis factor-alpha (TNFa) hyperalgesia. Brain Res 692:244^250

238

DISCUSSION

Welch JM, Simon SA, Reinhart PH 2000 The activation mechanism of rat vanilloid receptor 1 by capsaicin involves the pore domain and di¡ers from the activation by either acid or heat. Proc Natl Acad Sci USA 97:13889^13894 Willis WD, Al-Chaer ED, Quast MJ, Westlund KN 1999 A visceral pain pathway in the dorsal column of the spinal cord. Proc Natl Acad Sci USA 96:7675^7679 Woolf CJ, Salter MW 2000 Neuronal plasticity: increasing the gain in pain. Science 288:1765^ 1769 Xu GY, Huang LYM 2002 Peripheral in£ammation sensitizes P2X receptor-mediated responses in dorsal root ganglion neurons. J Neurosci 22:93^102

DISCUSSION Blake: The COX-2 inhibition is quite surprising in terms of its extent. Could some of that e¡ect which probably was generated at decent drug levels be attributed to promoting thrombosis in a hypoxic tumour? Hunt: My understanding is that there is some literature on reduced tumour growth anyway from COX-2 inhibitors. We don’t know the mechanism and your suggestion is one possibility. Would prostaglandins drive tumour growth? There is some evidence that this is the case. Blake: This seems the more plausible option. It is just that the extent of the e¡ect you got seemed remarkable. Fox: We see the same e¡ect with COX-2 inhibitors. Blake: Do you see increased thrombosis? Fox: No, we see reduction in tumour size. Felson: How tumour speci¢c are all the mechanisms you talked about? Colon cancer is the one that is supposed to be COX-2 sensitive. Some tumours cause blastic lesions in bone, not lytic lesions, and are associated with pain. Hunt: I can’t answer that. The only experience I have is with bone cancer. Kuettner: How e¡ective is the bisphosphonate treatment? You can theoretically make the bone so hard that it cannot be resorbed by either tumour cells or osteoclasts. Is it an e¡ective treatment? Hunt: I think it is. The advantage of OPG is that it is a drug that has been developed already. It is one of those drugs that could be used if they could be bothered to do the trials. Companies are more interested in the osteoporosis market than they are in the bone cancer market. Unless it is speci¢cally tested on this group of patients it may not be widely used and they’ll carry on using bisphosphonates. OPG is much simpler. Schaible: You mentioned changes in gene expression. Which gene groups are changing? Hunt: I don’t know if you are familiar with the bioinformatics programs. They are elaborate programs with a large do-it-yourself component. If you are interested in a particular pathway, such as the MAP kinase pathway, you have to ¢nd it,

BONE CANCER PAIN

239

annotate it and put it into your databank. Then you can ask which of these pathways is activated. We are in the process of doing this. It produces a bewildering array of data. Kidd: What are the implications for normal bone of the osteoclast^nerve cell interaction you described? What is to prevent an osteoclast just blundering across a nerve cell in normal bone? Hunt: My understanding is that there are very few osteoclasts in normal bone. I would imagine that any slight activation of osteoclasts has quite severe consequences for the nerve supply. Grubb: You raised the issue of markers of neuronal damage being up-regulated in these models of sarcoma. Does this mean that this pain is neuropathic in nature? Are these animals responsive to gabapentin, a drug frequently used to treat neuropathic pain? Hunt: They are responsive to gabapentin. If you make even a fairly severe in£ammation you never see ATF3 in the sensory neurons. It has to be beyond that. I don’t know what the line you have to cross is, whether it is qualitative or quantitative, but you can make a massive in£ammation and not get ATF3. Blake: When you set up rodent models of arthritis you can more or less completely block the so-called T cell-mediated complete Freund’s adjuvant (CFA) model with non-selective or selective COX-2 inhibitors. But when this is translated to rheumatoid arthritis in humans, the e¡ect is very small, apart from a bit of a suppression of in£ammation and a bit of analgesia. Then you look at a whole host of very di¡erent acting drugs in the animal model of arthritis, and they all seem to have quantitatively a very similar e¡ect on the model. Yet when you translate this to human they don’t. So my question, which is probably unanswerable, is what is di¡erent about a rodent who is turning up 131 genes and turning down 73 that allows it to respond to single and very disparate treatment options in a very similar fashion? When we translate this into human we don’t see the same e¡ect. Hunt: I was talking to Bruce Kidd earlier about this. There seems to be a serious problem in that a lot of the experimental work doesn’t seem to translate to human. Think of the time course of these events. In humans it takes years to get a bad knee. In the animals we are doing this acutely. Perhaps it is the rapidity that changes the mechanism. Fox: I think the time course is key. It is unrealistic that these rapid models can be extrapolated exactly into human. The outcome measures we have in animals are also imperfect. All drug e¡ects are being squashed into this window. Mackenzie: One of the big di¡erences is between the di¡erent models of arthritis. In adjuvant arthritis COX-2 inhibitors and non-selective COX inhibitors are extremely e¡ective, but they are much less e¡ective in monoarticular antigeninduced arthritis. One of the di¡erences is that in adjuvant arthritis there is

240

DISCUSSION

massive bone remodelling. It is very much like the picture in the sarcoma model. It may be that the primary thing the COX inhibitors are doing is to inhibit the bone remodelling. Felson: You commented on preventing osteoclast activation, and showed that this prevented pain. There was a decrement of pain but it wasn’t all that impressive. What does this mean? Hunt: I would say that we are talking about components. I think what this work shows is that you can break up pain into di¡erent components, but there is no single pathology that generates the signal. Palpation is a particularly severe stimulus for these animals. If you can get 40% decrease in pain that is pretty acceptable. Felson: Does anyone want to comment on osteoclast activation in osteoarthritis (OA)? There is subchondral bone remodelling in OA, which would involve osteoclasts. One of the most rapid osteoclast activation syndromes is women going through the menopause and there is no obvious bone pain there. The other one which might be relevant is transient regional osteoporosis, which is a disorder in which bone is resorbed quite rapidly associated with pain. This pain may be because of microcracks or microfractures. Pisetsky: One other condition to consider is acute vertebral collapses, which are very painful. A lot of that pain is attributed to the fracture, but you could also say that this is osteoclast activation. I have heard that people use calcitonin for this kind of fracture. Felson: Clinically, the observation is that calcitonin isn’t a very e¡ective osteoporosis medication, but it is an e¡ective pain medication. Blake: Another relevant disease state here is hypertrophic pulmonary osteoarthropathy. This presents clinically in people with lung cancer as a neurogenically mediated osteoclastic resorption at the wrist where there is clearly no tumour at the event. This provides yet another way of looking at some of these phenomena that are less tumour speci¢c in terms of destruction and much more secondary in terms of e¡ects that we don’t understand. This is a fantastic human clinical model. Pisetsky: Is there clinical experience with gabapentin in people with cancer pain that would give credence to this idea that it is neuropathic depending on how well gabapentin works? Are people using it? Hunt: I think so. Ordeberg: We have used it, but our patients with cancer pain (with skeletal metastases) also have bone destruction and spinal metastasis with encroachment on neural structures with a mixture of pain of di¡erent mechanisms.

Osteoarthritic Joint Pain: Novartis Foundation Symposium 260. Volume 260 Edited by Derek J. Chadwick and Jamie Goode Copyright ¶ Novartis Foundation 2004. ISBN: 0-470-86763-9

Symmetry, T cells and neurogenic arthritis N. G. Shenker*{, D. R. Blake*{1, C. S. McCabe*, R. Haigh{ and P. I. Mapp{ *Department of Rheumatology, Royal National Hospital for Rheumatic Diseases, Bath, BA1 1RL, {Department of Medical Sciences, University of Bath, Claverton Down, BA2 7AY and {Department of Rheumatology, Royal Devon & Exeter Hospital, Exeter, UK

Abstract. Symmetry in clinical disease occurs more commonly than expected by chance and is unexplained. In this paper we focus on symmetry in arthritis and describe the neurogenic hypothesis. Neuropeptides are anatomically relevant to systemic arthritis and have been shown to have modulating e¡ects on both the immune and circulatory systems. Neural networks project bilaterally and are involved in the development and propagation of in£ammatory disease. These putative pathological neuro-feedback loops may derive from the existence of biologically protective symmetrical mechanisms. 2004 Osteoarthritic joint pain. Wiley, Chichester (Novartis Foundation Symposium 260) p 241^257

Symmetry in clinical disease Symmetry in clinical disease is common and can be seen in a wide spectrum of conditions. Symmetrical diseases are found within the realms of rheumatology, dermatology, ophthalmology and neurology, as might be expected given the anatomical duplicity of the systems with which these specialities deal. The consistency of the presence of this clinical ¢nding points to an underlying mechanism. Bilateral expression of a unilateral stimulus has best been studied in the neurosciences and Koltzenburg et al (1999) have reviewed this phenomenon. They summarised the work of several teams who had independently described symmetrical, topographically precise, time-dependent changes in response to speci¢c local neurological insults. Neural connections between left and right sides of the spinal cord are hypothesized to account for these changes (Fig. 1). 1This paper was presented at the symposium by D. R. Blake to whom correspondence should be

addressed. 241

242

SHENKER ET AL

FIG. 1. Scheme demonstrating routes identi¢ed from histological studies of sensory ¢bres decussating via the dorsal commissure.

Detailed histological examination in the spinal cords of primates and other vertebrates (Light & Perl 1979, Culberson et al 1979) demonstrate posterior decussation of signi¢cant numbers of neurons at all levels of the spinal cord. These ¢bres terminate in the contralateral substantia gelatinosa (Rexed’s laminae I^II) as well as in the deeper layers (laminae III^IV) and their function has yet to be described. The implication that bilateral neurally mediated pathways exist and determine symmetrical disease patterns is important because conditions that exhibit symmetry are common and their aetiology largely unknown. Skin psoriasis, for example, is much more symmetrical than expected by chance (Farber et al 1986). Other dermatological symmetrical conditions include vitiligo and pityriasis rosacea. The classic in£ammatory example seen in ophthalmology is sympathetic ophthalmia. This is a bilateral uveitis complicating a unilateral perforating wound. Interestingly, acute demyelinating plaques seen in the brains of patients with multiple sclerosis are associated with subtle contralateral changes seen on

NEUROGENIC ARTHRITIS

243

TABLE 1 Reports in the literature of the development of arthritic diseases in partially paralysed patients Disease

Neural injury

Outcome suppressed

OA OA OA RA RA RA RA Gout

Hemiplegia Poliomyelitis Nerve injury Hemiplegia Hemiplegia Hand dominance Poliomyelitis Hemiplegia

Heberden’s nodes OA knee and hip OA and Heberden’s nodes Digital vasculitis, RA and RA nodules Arthritis Radiological changes of arthritis Arthritis Arthritis and tophi deposits

Modi¢ed from Dixon (1989).

di¡usion MRI (Werring et al 2000). Glomerulonephritis and pulmonary ¢brosis never occur unilaterally and this is noteworthy in itself. Symmetry in degenerative arthritis With respect to the musculoskeletal system, both degenerative and in£ammatory arthritides are symmetrical. The Heberden’s nodes of osteoarthritis (OA), the commonest arthritis of them all, are symmetrical. This is emphasised by a recent epidemiological survey from the Framingham population (Niu et al 2003) that demonstrated that symptomatic OA is remarkably symmetrical, especially in women. Proprioceptive feedback is abnormal in a radiologically normal knee that is contralateral to an osteoarthritic knee. This contralateral sensory de¢cit is more abnormal than expected when compared to age-matched patients who have no OA (Sharma et al 1997, Garsden & Bullock-Saxton 1999). These observations suggest that neural de¢cits exist before the symmetrical expression of OA occurs. Further evidence that the neural supply to the joint is fundamental to the expression of this disease is that neural disruption prevents the development of Heberden’s nodes (Table 1). Symmetry in in£ammatory arthritis A symmetrical pattern of involvement is clinically so important in rheumatoid arthritis that it is a feature of its diagnostic criteria (Arnett et al 1988). There have

244

SHENKER ET AL

been recent publications noting that in£ammatory arthritis is symmetrical irrespective of rheumatoid factor (Bukhari et al 2002) and that some of the seronegative arthritides, such as psoriatic (Helliwell et al 2000), are more symmetrical than would be expected by chance. The pathophysiology of in£ammatory arthritis currently centres on T lymphocyte-dominated autoimmune processes. Yet, how could the immune system, e¡ected by blood-borne agents, manifest clinically symmetrical disease? This paper supports a neurogenic hypothesis to explain disease symmetry, and with reference to T cell-mediated disease, presents possible mechanisms for the interface between some neuropeptides and the immune system. Role of the neuropeptides substance P and CGRP in arthritis Hench’s classical report of symmetrical palindromic rheumatism dates back more than 50 years and in it he describes wheals and £ares occurring over a¡ected joints. The similarity between this observation and Lewis’ triple £are-wheal response was quickly made. The £are-wheal response is dependent upon an intact nerve supply and, more speci¢cally, upon the release of the co-localized neuropeptides substance P (SP) and calcitonin gene-related peptide (CGRP) contained within populations of unmyelinated neurons. These are the commonest neuropeptides found in small diameter nerve ¢bres. It is now known that some of the unmyelinated ¢bres that innervate joint structures also have sprouting peripheral terminals that project into the surrounding skin. The skin changes that Hench described could well be attributed to the vasodilatory e¡ects, known to be caused by the release of SP and CGRP, via an antidromic route. The role of these neuropeptides in the in£ammatory processes of arthritis is therefore of interest. Experimental work on a variety of animal species utilizing a number of insults to generate mono- and systemic arthritides have demonstrated that there is a release of SP in response to an arthritic insult (Garrett et al 1992, Holzer 1988). There is a corresponding increase in the production of SP and CGRP in the nerve cell bodies located in the rat dorsal root ganglion (Mapp et al 1993, Hanesch et al 1995). The ¢nding that joints ‘primed’ with substance P demonstrate a more severe subsequent arthritis (Levine et al 1984) supports the importance of the role of substance P in in£ammatory arthritis. Furthermore, it can be shown that the depletion of these neuropeptides through denervation (Levine et al 1985) or exposure to capsaicin (Donaldson et al 1995) attenuates the progression and severity of an acute polyarthritis. Thus the presence of SP worsens, and its absence improves, experimental arthritis. There is a persistent increase in SP in the synovial £uid of patients with RA compared with normal controls (Westermark et al 2001). This is a consistent ¢nding reported by several groups independently and SP levels are raised in the

NEUROGENIC ARTHRITIS

245

synovial £uid from patients with in£ammatory arthritis when compared to both plasma concentrations and patients with non-in£ammatory arthritis (Garrett et al 1992). Histochemical studies support the release of neuropeptides from arthritic synovium by ¢nding a reduction in the presence of neural SP (Garrett et al 1992). Interestingly, both CGRP and SP are produced by peripheral lymphocytes and this route may also contribute to the presence of neuropeptide release in the in£amed synovial joint (Qian et al 2001, Wang et al 2002). Although SP and CGRP are the most prevalent neuropeptides and have been best studied, other neuropeptides such as vasoactive intestinal peptide (VIP) have e¡ects on the course of experimental arthritis (Delgado et al 2001). In£uence of neuropeptides on T cells Neuropeptide-containing nerves are in contact with immune cells throughout their lifespan, from the thymus to the peripheral circulation and also the lymph node. There is a rich neuropeptide innervation to the blood vessels within lymph nodes and there appears to be an especially strong anatomical link to the T cells rather than the B cells (Weihe & Krekel 1991). SP and CGRP are located perivascularly in the synovium (Garrett et al 1992). CGRP has powerful vasodilatory e¡ects (Brain et al 1985) and will therefore increase synovial circulation signi¢cantly. These neuropeptides are also important in the migration of lymphocytes from the circulation to the joint. Capsaicin, a SP- and CGRP-depleting agent, reduces the in£ux of T cells into the acutely arthritic joint of guinea pigs (Hood et al 2001). This is a speci¢c e¡ect on these T cells because macrophage in£ux was una¡ected in the same model. Substance P binds to a particular subset of T lymphocytes via speci¢c receptors and causes proliferation of these lymphocytes (McGillis et al 1990). Furthermore, Levite’s work suggests that neuropeptides can cause T lymphocytes to produce a rather di¡erent and distinctive pattern that is described as ‘forbidden’. Th0 cells are driven by antigenic stimulation to induce rather ¢xed patterns of cytokine secretion characterized by the ‘pro-in£ammatory’ Th1 (e.g. interleukin [IL]2, interferon [IFN]g) and the ‘pro-antibody’ Th2 (e.g. IL4, IL10) cytokine pro¢les. The ‘forbidden’ pattern that is induced by some neuropeptides combines both the characteristic Th1 responses with the additional secretion of IL4 and IL10, and the characteristic Th2 responses with the additional secretion of IL2 and IFNg (Levite 2001). These in vitro studies need con¢rmation in vivo. Neuropeptide-mediated T helper responses however obviate the need for antigen and may well contribute to the immune dysregulation seen in rheumatoid arthritis. The exact role of neuropeptides in in£ammatory arthritis is not known. Substance P and other neuropeptides are anatomically, physiologically and pharmacologically relevant to the mechanisms seen in the pathological processes

246

SHENKER ET AL

of in£ammatory arthritis. Neuropeptides have an important role in both the tra⁄cking and the subsequent immunomodulation of T cells in in£ammatory arthritis. The nervous system in arthritis The role that the nervous system has in rheumatoid arthritis in humans is well documented. There are numerous examples of both lower motor and upper motor neuron lesions ameliorating the progression of this deforming joint disease. Denervation also prevents deformities caused by osteoarthritis, gout and seronegative arthritides. Table 1 summarizes some of the identi¢ed case reports. The presence of an intact nervous system is important to allow the full expression of these arthritides. A putative mechanism can be proposed whereby an in£ammatory insult, such as a monoarthritis, triggers changes in the ipsilateral neural network which are then mirrored to the contralateral side to predispose this side to a similar in£ammatory state. This is an example of an autoregulatory loop and this hypothesis is testable. Contralateral in£uences of neuro-in£ammatory insults Rotshenker & Tal (1985) reported in both frogs and mice that sprouting in the contralateral neuromuscular junction followed nerve section and degeneration of the ipsilateral neuromuscular junctions. These contralateral ¢ne extra sprouts were ¢rst seen 5 days after the sciatic nerve was proximally sectioned and this sprouting occurred much more frequently than in control rats. Associated contralateral increases in GAP43, neuropeptide mRNA and expression, and proliferation of microglia were also seen (Koltzenburg et al 1999). An in£ammatory lesion induced in a rat knee using latex spheres resulted in a thickening bilaterally of the synovial intimal layer and this was accompanied by a bilateral cellular in¢ltrate (Kidd et al 1995). No changes were detectable in any of the other joints. There were also contralateral changes in bradykinin-induced extravasation from the knee joint, as measured using Evans’ blue, which followed the course of this latex monoarthritis. This somatotopic distant response to a local in£ammatory insult can only be mediated through the neural network. Systemic circulatory factors cannot display such topographical precision and biomechanical in£uences fail to explain the lack of response in other loadbearing joints. In further support of a neurogenic mechanism, the expression of substance P in the dorsal root ganglion was increased within 3 days bilaterally in this monoarthritis model. There exists a body of published evidence to corroborate the ¢ndings that a unilateral insult induces contralateral topographically-precise changes associated

NEUROGENIC ARTHRITIS

247

with appropriately located changes at the spinal level. We have recently reviewed this literature and this is summarized in Table 2. It is important to note that although contralateral changes in the periphery can be seen from these models, the earliest changes on the contralateral side occur within the spinal cord. These are seen at the cellular level with changes in neuropeptide expression in the contralateral nerve cell bodies (Kidd et al 1995). It has been documented that unilateral capsaicin injections, the pungent extract of chilli that induces the release of SP and CGRP, alters contralateral depolarization thresholds and re£exes (Gjerstad et al 2000). We were able to demonstrate contralateral central pain responses in human by detecting contralateral allodynia and hyperalgesia within 30 minutes following a unilateral intradermal injection of capsaicin (Fig. 2). In a group of 20 normal control subjects, 9 (45%) demonstrated contralateral sensory changes that were between 5^50% of the hypersensitive area induced by the capsaicin injection. In a group of 21 subjects with rheumatoid arthritis, 11/21 (52%) demonstrated contralateral sensory changes (R. Haigh, unpublished data). The prior development of changes within the central nervous system implies that this is driving any contralateral peripheral change. This physiological loop may well be important in conferring survival bene¢t, as will be discussed later, but ‘physiology begets pathology’. It is therefore possible that the nervous system not only underlies the symmetry of some diseases, but may be principle in their aetiopathogenesis as well.

Charcot’s joints Charcot was the ¢rst to note the possible role of the nervous system in arthritis and his name has become synonymous with the neuroarthropathic joint. It is therefore no coincidence that the clinical history of these joints re£ects the surmised role that neuropeptides have in their aetiology. Sella & Barrette (1999) were able to identify ¢ve stages of neuroarthropathy in the feet of diabetics from the careful evaluation of clinical histories and radiological tests. In stage 0 the patient presents with a locally swollen, warm, and often painful foot. It is this stage that is often missed clinically as patients tend to present later. Radiographs are negative whereas a technetium 99 bone scan is markedly positive. In addition to these clinical ¢ndings there are also periarticular cysts, erosions and localized osteopenia radiologically in stage 1. Stages 2^4 demonstrate the progressive joint subluxations, dislocations and destruction of the joints in the diabetic foot. Clinically in these stages, there is no temperature gradient between the two feet. Radiologically, there is bony trabeculation across joint spaces indicative of mature fusion.

248

TABLE 2

SHENKER ET AL

The contralateral e¡ects of localized unilateral in£ammation

Lesions inducing arthritis

Contralateral e¡ect

CFA (1 mg) in knee Freund’s adjuvant in knee (0.5 ml of 1 mg/ml) Carrageenan 2% (0.05 ml) in knee

Decrease in anabolism of cartilage for 6^72 h Increase in SP, CGRP and NPY in knee for 2^24 h Increase in NK-A in knee at 2 h and at 24 h Increase in CGRP and NPY for 2^24 h Increase in SP at 2 h and at 24 h; and NK-A at 24 h Increase in CGRP and NPY for 2^24 h Increase in NK-A at 2^6 h and SP at 2 h Increase in SP for 2^24 h, CGRP for 6^24 h and NPY for 2^6 h No increase in NK-A Decrease in anabolism of cartilage for 6^72 h Histopathological and biochemical evidence of joint destruction up to 80 d Increased proteoglycan loss and mechanical hyperalgesia up to 9 days Bradykinin-induced plasma extravasation enhanced between 10^21 days. Macrophage in¢ltration noted between 3^10 days

Human recombinant IL1 (0.05 ml of 1 mg/ml) in knee SP (0.05 ml of 10 5 M) in knee SP in knee (0.2, 1, 2, 10, 20 mg in 50 ml) 500 ml of mBSA in knee of rats preimmunized with 1 ml CFA/2 mg MTb 500 mg (50 ml) mBSA in knee of presensitized rats 100 ml of 1% latex spheres in knee

Lesions inducing hindpaw oedema

Contralateral e¡ect

CFA (100 ml) Carrageenan (0.1 ml of 2%)

TNFa levels elevated up to 120 h CGRP levels increase in hindpaw perfusate at 3^4 h Thermal and mechanical withdrawal latencies reduced at 3^4 h Mechanical hyperalgesia between 3^5 days Licking responses occur at 10^60 minutes Heat (not mechanical) withdrawal latency reduced for 2^48 h Reduction in mechanical withdrawal latencies for 3^5 h Increase in paw thickness for 3^5 h Swelling observed, maximal at 5 h

Carrageenan 100 ml of 2% Formaldehyde 50 ml (0.1%, 5%, 10%) Bee venom 100 ml (0.2 mg) Repeated saline injection 150 ml on 3 consecutive days Urate, pyrophosphate and oxalate crystals (3 mg in 150 ml) CGRP 100 ml (300 pmol) NGF injections (3 days of 4 mg/day) in hindpaw or ear or forepaw IL1b (10 ng) Thermal stimuli (55 8C for 15 s) Thermal stimuli (55 8C for 15^20 s) Thermal stimuli to decerebrate rat (75 8C for 60 s)

Oedema induced at 5^24 h Increased expression of preprotachykinin and preproCGRP mRNA in nerve (sciatic; trigeminal; or brachial plexus) B1 receptor-mediated mechanical hyperalgesia for 1^6 h Thermal foot withdrawal latencies reduced at 24 h Thermal foot withdrawal latencies reduced at 24 h Reduction in the £exor re£ex threshold to mechanical and thermal stimuli at 1 h

All of these studies were performed in rats, unless otherwise indicated. CFA, complete Freund’s adjuvant; SP, substance P; NPY, neuropeptide Y; CGRP, calcitonin gene related peptide; NK-A, neurokinin A; IL1, interleukin 1; MBSA, methylated bovine serum albumin; MTb, Mycobacterium tuberculosis; TNF, tumour necrosis factor; NGF, nerve growth factor; RNA, ribonucleic acid.

NEUROGENIC ARTHRITIS

249

FIG. 2. Rheumatoid subject demonstrating mirroring of painful stimulus. Ipsilateral Capsaicin injection in the left forearm. Using Semmes Weinstein ¢laments, the sensory changes are then mapped using di¡erent colours to represent 10 minute intervals between 10^ 60 minutes. Changes in allodynia are marked with an ‘x’. Changes to mechanical hyperalgesia are marked with ‘^’. Note the development of both allodynia and hyperalgesia in a topographically precise contralateral region.

Histological examination of Charcot joints has been di⁄cult because tissue from early disease is rare and most samples have been removed from destroyed joints that are late in the disease process. Most of these specimens show non-speci¢c changes consistent with an in£ammatory process but there are isolated reports that show scattered lymphocytes (Clement et al 1984). As can be seen from the experimental evidence presented above, neuropeptides from damaged nerve terminals increase both the synovial blood supply and the in£ux of activated macrophages and lymphocytes into the joint. These cells alter the complex neurochemical intra-articular matrix substantially leading to massive rapid destruction of the joint. Later in the disease, however, neural damage has su⁄ciently depleted neuropeptides so that their vasodilatory e¡ects are negligible. Interestingly Charcot joints are not uncommonly bilateral (Armstrong et al 1997). Survival hypothesis of an auto-neuroin£ammatory loop There is evolutionary pressure for rapid appropriate responses to be available as part of an organism’s defence against noxious environs (Kidd et al 1989).

250

SHENKER ET AL

Bilaterality has evolved so that there is functional capacity should one part be irreparably damaged. It is therefore vitally important that the contralaterally homonymous area should be protected following focal damage. A neuroin£ammatory loop whereby protective responses are contralaterally rapidly upregulated would therefore be a useful addition to any symmetrical organism’s protection and long-term survival. The advantage that this precise response has over any systemic mechanism is that it is economic with the energy expended and limits widespread self-damage caused by inappropriate in£ammation. Similarly, pain withdrawal re£exes are primed through previously documented electrophysiological changes in the spinal cord so that the contralateral limb can be withdrawn more quickly. There is more chance of damage limitation from a dangerous environment if such responses existed and it is argued that survival is enhanced through use of such pathways. As our control over the environment has reduced the risk of noxious exposure in modern society, these pathways may now unfortunately be more associated with disease states. There is insu⁄cient evolutionary pressure to down-regulate these pathways because these diseases have a limited e¡ect on reproductive capacity as their peak onset is after childbearing years. Conclusion Chronic in£ammatory disease is symmetrical where possible. This symmetry is mediated neurogenically through neuropeptide mechanisms that de¢ne a neuroin£ammatory loop that is insult speci¢c and topographically precise. The immune system can fundamentally be a¡ected by the release of such neuropeptides through their e¡ects on T cells. This interaction may be important not only in the expression of symmetry in diseases but also in their aetiopathogenesis. References Armstrong DG, Todd WF, Lavery LA, Harkless LB, Bushman TR 1997 The natural history of acute Charcot’s arthropathy in a diabetic foot speciality clinic. Diabet Med 14:357^363 Arnett FC, Edworthy SM, Bloch DA et al 1988 The American Rheumatism Association 1987 revised criteria for the classi¢cation of rheumatoid arthritis. Arthritis Rheum 31:315^324 Brain SD, Williams TJ, Tippins JR, Morris HR, MacIntyre I 1985 Calcitonin gene-related peptide is a potent vasodilator. Nature 313:54^56 Bukhari M, Lunt M, Harrison BJ, Scott DG, Symmons DP, Silman AJ 2002 Erosions in in£ammatory polyarthritis are symmetrical regardless of rheumatoid factor status: results from a primary care-based inception cohort of patients. Rheumatology (Oxford) 41:246^252 Clement GB, Grizzard K, Vasey FB, Germain BF, Espinoza LR 1984 Neuropathic arthropathy (Charcot joints) due to cervical osteolysis: a complication of progressive systemic sclerosis. J Rheumatol 11:545^548

NEUROGENIC ARTHRITIS

251

Culberson JL, Haines DE, Kimmel DL, Brown PB 1979 Contralateral projection of primary a¡erent ¢bers to mammalian spinal cord. Exp Neurol 64:83^97 Delgado M, Abad C, Martinez C, Leceta J, Gomariz RP 2001 Vasoactive intestinal peptide prevents experimental arthritis by downregulating both autoimmune and in£ammatory components of the disease. Nat Med 7:563^568 Dixon A 1989 Hemiplegia, hand dominance and asymmetry of expression of arthritis in relation to intra-articular pressure. In: Dixon A, Hawkins C (eds) Raised intra-articular pressure
clinical consequences. Bath Institute for Rheumatic Diseases, Bath, p 65 Donaldson LF, McQueen DS, Seckl JR 1995 Neuropeptide gene expression and capsaicinsensitive primary a¡erents: maintenance and spread of adjuvant arthritis in the rat. J Physiol 486:473^482 Farber EM, Nickolo¡ BJ, Recht B, Fraki JE 1986 Stress, symmetry and psoriasis: possible role of neuropeptides. J Am Acad Dermatol 14:305^311 Garsden LR, Bullock-Saxton JE 1999 Joint reposition sense in subjects with unilateral osteoarthritis of the knee. Clin Rehabil 13:148^155 Garrett NE, Mapp PI, Cruwys SC, Kidd BL, Blake DR 1992 Role of substance P in in£ammatory arthritis. Ann Rheum Dis 51:1014^1018 Gjerstad J, Tjolsen A, Svendsen F, Hole K 2000 Inhibition of spinal nociceptive responses after intramuscular injection of capsaicin involves activation of noradrenergic and opioid systems. Brain Res 859:132^136 Hanesch U, Blecher F, Stiller RU, Emson PC, Schaible HG, Heppelmann B 1995 The e¡ect of a unilateral in£ammation at the rat’s ankle joint on the expression of preprotachykinin-A mRNA and preprosomatostatin mRNA in dorsal root ganglion cells
a study using nonradioactive in situ hybridization. Brain Res 700:279^284 Helliwell PS, Hetthen J, Sokoll K et al 2000 Joint symmetry in early and late rheumatoid and psoriatic arthritis: comparison with a mathematical model. Arthritis Rheum 43:865^871 Herzburg U, Murtaugh MP, Mullet MA, Beitz AJ 1995 Electrical stimulation of sciatic nerve alters neuropeptide content and lymphocyte migration in the subcutaneous tissues of the rat hind paw. Neuroreport 6:1773^1777 Holzer P 1988 Local e¡ector functions of capsaicin-sensitive sensory nerve endings: involvement of tachykinins, calcitonin gene-related peptide and other neuropeptides. Neuroscience 24:739^768 Hood VC, Cruwys SC, Urban L, Kidd BL 2001 The neurogenic contribution to synovial leucocyte in¢ltration and other outcome measures in a guinea pig model of arthritis. Neurosci Lett 299:201^204 Kidd BL, Mapp PI, Gibson SJ et al 1989 A neurogenic mechanism for symmetrical arthritis. Lancet 2:1128^1130 Kidd BL, Cruwys SC, Garrett NE, Mapp PI, Jolli¡e VA, Blake DR 1995 Neurogenic in£uences on contralateral responses during experimental rat monoarthritis. Brain Res 688:72^76 Koltzenburg M, Wall PD, McMahon SB 1999 Does the right side know what the left side is doing? Trends Neurosci 22:122^127 Levine JD, Clark R, Devor M, Helms C, Moskowitz MA, Basbaum AI 1984 Intraneuronal substance P contributes to the severity of experimental arthritis. Science 226:547^549 Levine JD, Dardick SJ, Basbaum AI, Scipio E 1985 Re£ex neurogenic in£ammation. I. Contribution of the peripheral nervous system to spatially remote in£ammatory responses that follow injury. J Neurosci 5:1380^1386 Levite M 2001 Nervous immunity: neurotransmitters, extracellular potassium and T-cell function. Trends Immunol 22:2^5 Light AR, Perl ER 1979 Reexamination of the dorsal root projection to the spinal dorsal horn including observation of the di¡erential termination of coarse and ¢ne ¢bres. J Comp Neurol 186:117^131

252

DISCUSSION

Mapp PI, Terenghi G, Walsh DA et al 1993 Monoarthritis in the rat knee induces bilateral and time-dependent changes in substance P and calcitonin gene-related peptide immunoreactivity in the spinal cord. Neuroscience 57:1091^1096 McGillis JP, Mitsuhashi M, Payan DG 1990 Immunomodulation by tachykinin neuropeptides. Ann NY Acad Sci 594:85^94 Niu J, Zhang Y, LaValley M, Chaisson CE, Aliabadi P, Felson DT 2003 Symmetry and clustering of symptomatic hand osteoarthritis in elderly men and women: the Framingham study. Rheumatology (Oxford) 42:343^348 Qian BF, Zhou GQ, Hammarstrom ML, Danielsson A 2001 Both substance P and its receptor are expressed in mouse intestinal T lymphocytes. Neuroendocrinology 73:358^368 Rotshenker S, Tal M 1985 The transneuronal induction of sprouting and synapse formation in intact mouse muscles. J Physiol 360:387^396 Sella EJ, Barrette C 1999 Staging of Charcot neuroarthropathy along the medial column of the foot in the diabetic patient. J Foot Ankle Surg 38:34^40 Sharma L, Pai YC, Holtkamp K, Rymer WZ 1997 Is knee joint proprioception worse in the arthritic knee versus the una¡ected knee in unilateral knee osteoarthritis? Arthritis Rheum 40:1518^1525 Wang H, Xing L, Li W, Hou L, Guo J, Wang X 2002 Production and secretion of calcitonin gene-related peptide from human lymphocytes. J Neuroimmunol 130:155^162 Weihe E, Krekel J 1991 The neuroimmune connection in human tonsils. Brain Behav Immun 5:41^54 Werring DJ, Brassat D, Droogan AG et al 2000 The pathogenesis of lesions and normalappearing white matter changes in multiple sclerosis: a serial di¡usion MRI study. Brain 123:1667^1676 Westermark T, Rantapaa-Dahlqvist S, Wallberg-Jonsson S, Kjorell U, Forsgren S 2001 Increased content of bombesin/GRP in human synovial £uid in early arthritis: di¡erent pattern compared with substance P. Clin Exp Rheumatol 19:715^720

DISCUSSION Hunt: What does bilaterality tell you about the ipsilateral in the original cause of the disease? Blake: The whole thing is basically a neural discharge event. I’m talking here about rheumatoid arthritis (RA). If someone had taught me immunology saying that the primitive system was neuropeptidergic, the secondary evolution of the system is antigen-driven, and then I had heard Bruce Kidd’s lecture on the nature of the pain response and the axon re£ex in palindromic RA, I would by preference choose to adopt the idea that it is a neuropeptide discharge state that creates this unusual degree of symmetry. Then the subsequent disruption creates a secondary autoimmune reaction sequence that is very diverse. This is what we see. My guess is that arthritis might be better dissected by neurologists. Hunt: We are talking about very precise topography here: one joint on one side and one on the other. Yet the work on placebo by Benedetti involving capsaicin injections, placebo response is very precise (Benedetti et al 1999). So what is happening peripherally can be adjusted precisely. It is not clear how, because the

NEUROGENIC ARTHRITIS

253

descending systems themselves from the brain to the spinal cord don’t look very topographically organized. They appear more di¡use. I don’t know how it is done. Blake: Nevertheless, you have a deep suspicion it is done. That is a powerful argument in terms of the upper control of the whole system. Rediske: Earlier in the meeting there was some discussion about how patients with OA in one knee develop OA in the second knee with time. The argument given was that this was more of a compensatory gait phenomenon. How much could this bilateral phenomenon play a role in evolution of secondary OA? When you compare the location of the cartilage injuries in the two knees, how similar are they? Felson: David Blake was on target when he showed the Herberden’s nodes bilaterally. This is nicely described by Cyrus Cooper, and then we published a paper last year in Rheumatology which corroborated it in Framingham (Niu et al 2003). There is a remarkable symmetry in the occurrence of OA in hands in people. If you have a ¢fth DIP on one side the odds ratio for getting a ¢fth DIP in the exact joint on the other side is 15. There is remarkable symmetry in OA also. The question is the one you raised, John: are there other explanations in OA other than some kind of tra⁄cking or neuropeptide release that crosses sides? We have thought about anatomic similarities between sides. David Hunter might want to comment on the genetics of hand OA and how it might relate to anatomy. Hunter: We have recently done a factor analysis on the symmetry using the Framingham family data. This demonstrates that the predisposition is purely symmetrical. The predisposition to disease is largely heritable. We don’t know whether that symmetry is being predisposed to a heritable element in the brain or an anatomical element at the joint. This has some impressive LOD scores in terms of linkage. Pisetsky: Is the heritability joint speci¢c? It is symmetrical but is it joint speci¢c? A person could inherit some factor that predisposes to base of thumb arthritis but which doesn’t increase the risk of DIP or PIP arthritis. This suggests that this is anatomical. Blake: One does not exclude the other options. We can equally have lots of pathways that create this, and it would be foolish to say this explains everything. Clearly, walking unevenly can create symmetrical events for a completely di¡erent reason. This is a partial explanation of some kinds of pathologies. The fact is that the joint system as I would describe it is built up by serial divisions using fractal-based mathematics. What you have just described is a fractal pattern, and what you are saying is that this is a di¡erent genetic pattern from that. That’s ¢ne: this then impacts upon that. Then you have to say to yourself that nodal OA of the ¢ngers relates to OA of the hip, which it seems to do. Could it be that the DIP joints are actually the same fractal zone that embraces the hips? Then if you start going back in the evolutionary sense, and look at

254

DISCUSSION

amphibians, then what lies behind what now becomes a hand system is the hip. So what you are describing are genetically determined fractal zones. Then it all ¢ts together. Pisetsky: There are some forms of arthritis that are classically asymmetrical. How do you interpret them? The other thing is that we have a variety of e¡ective interventions that are asymmetric, such as joint injections and joint replacement. Blake: You are playing into my hands here. The whole purpose of my paper was to describe asymmetry. For example, look at the ¢sh model where there is overt in£ammation on one gill. But destroy the organism with a gaseous system on the other side and clinically you see asymmetry. The processes are occurring in both gills but you just see it in one because it is working. Asymmetry should be the norm if this system is working. It is pretty pointless us having two kidneys, two eyes and two testes if every time a disease gets one it gets the other. The fact that they are bilaterally located in mirror image spots across the spinal cord seems more than a little coincident. But the whole process from the evolutionary point of view is to say that ‘I’m under attack, don’t go down that route’. Then one presumes like everything else in biology that there is a genetic in£uence that promotes the extent of bilaterality. It will be a small subpopulation that have a very dominant mirror-imaging system that then undergo the process. Exactly the same is true in sympathetic uveitis, where most people get away with a metal object in their eye and 5% go blind. But there are all points in between. So asymmetry is the physiological state. Lohmander: I would argue that we need to accept that all of the mentioned factors would feed into the system with regards to symmetry or asymmetry of OA. Clearly, unilateral OA will change gait and loading on the other side. There are neurogenic mechanisms and there are likely systemic mechanisms by which arthritis could be transferred. Finally, we have the genetic aspects and anatomical aspects. An example which would ¢t into David’s argument is that of the fairly few speci¢c mutations we know about in connective tissue-related genes that are associated with bona ¢de OA. These di¡erent mutations do not generate the same forms of OA. Each mutation is fairly speci¢c in the form of OA that it brings about. Matrilin 3 mutations are di¡erent from collagen 2 mutations and so forth. I think this speaks to the developmental/fractal theory that David is proposing. Dieppe: There is some old work from the Leeds biomechanical group (Swann & Seedhom 1993) showing that there were very symmetrical patterns of deformability between joints. If you have one type of mechanical feature of one knee, you had exactly the same on the other side. This suggests that there is a lot of symmetry of the way that joints are built and the way they respond mechanically. I have a question about CRPS, which you have been looking at. My experience is that this is asymmetrical, but I would have thought from your thesis that it shouldn’t be.

NEUROGENIC ARTHRITIS

255

Blake: As it happens it is not asymmetrical. It is surprisingly symmetrical when you have the right tools to look at it. Simkin: It is relevant to mention that the vascular physiology of the synovium is strikingly symmetrical and di¡ers markedly from one pair of joints to the next. The shoulders are not like the knees which are not like the wrists (Simkin & Benedict 1990a,b). I am not sure what this has to do with the pain of OA, but it may have quite a bit to do with the patterns that we see in symmetrical joint disease. Hunter: Perhaps I am misinterpreting what you are saying. It seems you are suggesting that asymmetry is the rule and only 10^20% of people will ¢t this symmetrical-type pattern. Irrespective of whatever tool I have tried to use to image OA, the symmetry is still striking. Blake: I am just making these ¢gures up, but I am talking about 10% as a total physiological event for all processes in human society. But one can understand that there might be a disease or process like OA where there is an underlying zonal fractal system that determines pattern, and which somehow is also in£uenced by a secondary genetic system that evokes symmetry but which can be disrupted in a stroke model. The onset of this in normality could be a fractal genetic aberration, whereas the mirror-imaging process that is meta and paraceptively modi¢ed is another process that comes into the game. We often end up in these symposia by feeling that one explanation provides all explanations. We only have to realise how many genes are going up and down in that cancer model to realise this is not true. These things have to work together. Hunter: If you want to put a ballpark ¢gure on it, it sounds like your belief for RA is that this whole process accounts for the majority of the disease even on the ipsilateral side. If you want to put a ballpark ¢gure on the in£uence of your metaceptive system for OA, for the majority of the community, what would it be? Blake: When people crack the genes for this fractal patterning then you have the answer
basically, it’s what’s left. Evans: Coming back to the T cell and antigen-driven events, and how the nervous system can in£uence them, we have an interesting observation relating to what I talked about yesterday with this contralateral suppression of disease. We take rabbits and sensitize them to two di¡erent antigens, ovalbumin and BSA. Then we have the choice of giving the rabbits bilateral knee disease with the same antigen or di¡erent antigens. We only see the contralateral suppression of disease if the contralateral knee is su¡ering from the same antigen-induced arthritis. We don’t get contralateral suppression of arthritis when a di¡erent antigen is driving it. I don’t know how to explain this, but it suggests that this whole relationship between the immune system and the nervous system is going to be fascinating to tease apart. Grubb: In neuropathic and in£ammatory pain we see sets of neuropeptides and cell markers going up and down, but we see modest contralateral e¡ects

256

DISCUSSION

quite often. Do you have a feeling for what the mechanism is in the nervous system? Hunt: I’ve always ignored contralateral events as being artefacts. If you cut the sciatic nerve on one side the animal is obviously moving in a di¡erent way. But there is a lot of literature on this, and some people have done some very interesting experiments. But it always comes out that the ¢nal reason for this is so obscure. Grubb: The fact that the capsaicin treatment seems to prevent some of these changes is interesting. We know there is a certain amount of hard wiring between left and right. We discussed the development of contralateral inputs. Neurons that show central sensitization develop receptive ¢elds from the contralateral side. We can see cross spinal connections, but we can’t imagine how that gets back to the DRG on the other side. Fox: How frequent are these changes? The contralateral e¡ects aren’t terribly robust, are they? Blake: They are very robust providing that you look at them in the right models. It depends what you are looking at. For the example I gave on the eye, if your only investigative tool was blindness, this is a very infrequent event. But if you use an ERG it is an incredibly frequent event. Hunt: In general, for chip studies I would never use the other side. I think it does change. Blake: That is an extraordinarily sensitive technique for looking at the phenomenon. Fox: We have done this. I can’t quote the numbers, but the changes were not signi¢cant. Blake: My whole point is that the changes were small. This is a down-regulated system. We are looking for a small subtle change. Felson: I want to bring up another potential use for this contralateral e¡ect. We talked earlier about looking for models of early disease that might correspond to early ways of identifying some of this going-to-develop disease. The contralateral phenomenon is so common in clinical OA that it might not be a bad example, sort of like taking someone from a family with a mutation who hasn’t yet developed disease but who is likely to.

References Benedetti F, Arduino C, Amanzio M 1999 Somatotopic activation of opioid systems by targetdirected expectations of analgesia. J Neurosci 19:3639^3648 Niu J, Zhang Y, LaValley M, Chaisson CE, Aliabadi P, Felson DT 2003 Symmetry and clustering of symptomatic hand osteoarthritis in elderly men and women: the Framingham Study. Rheumatology 42:343^348

NEUROGENIC ARTHRITIS

257

Simkin PA, Benedict RS 1990a Iodide and albumin kinetics in normal canine wrists and knees. Arthritis Rheum 33:73^79 Simkin PA, Benedict RS 1990b Hydrostatic and oncotic determinants of microvascular £uid balance in normal canine joints. Arthritis Rheum 33:80^86 Swann AC, Seedhom BB 1993 The sti¡ness of normal articular cartilage and the predominant acting stress levels: implications for the aetiology of osteoarthrosis. Br J Rheumatol 32:16^25

Osteoarthritic Joint Pain: Novartis Foundation Symposium 260. Volume 260 Edited by Derek J. Chadwick and Jamie Goode Copyright ¶ Novartis Foundation 2004. ISBN: 0-470-86763-9

Lessons from ¢bromyalgia: abnormal pain sensitivity in knee osteoarthritis Laurence A. Bradley, Brian C. Kersh*, Jennifer J. DeBerry, Georg Deutsch, Graciela A. Alarco¤n and David A. McLain{ Division of Clinical Immunology and Rheumatology, University of Alabama at Birmingham, 805 Faculty O⁄ce Tower, 510 20th Street South, Birmingham, AL 35294, *New Mexico Veterans Administration Health Care System, 1501 San Pedro Drive, SE Building 41, Albuquerque, NM 87108 and {Brookwood Medical Center, Homewood, AL 35209, USA

Abstract. Fibromyalgia (FM) is a disorder that is characterized by widespread, musculoskeletal pain and abnormal pain sensitivity at multiple anatomic sites. Laboratory studies involving psychophysical and neuroimaging methods suggest that central augmentation of low intensity stimulation may contribute to abnormal pain sensitivity in FM. Recently, several investigators, using similar laboratory methods, have shown that patients with knee or hip osteoarthritis (OA) exhibit abnormal pain sensitivity or abnormal pain inhibition at anatomic sites distal to a¡ected joints. Consistent with animal models of central sensitization, di¡erences between patients and healthy controls in pain processing and pain inhibition at these distal sites are eliminated after nociceptive input is eliminated following total joint replacement surgery. This paper reviews these ¢ndings from our laboratory and those of independent investigators. It also presents verbal, psychophysical and neuroimaging data concerning ethnic group di¡erences in a¡ective and cognitive pain responses among patients with knee OA. We suggest that central sensitization as well as centrally-mediated cognitive and a¡ective factors in£uence the pain responses of patients with knee OA. In addition, ethnic group di¡erences in pain cognition and a¡ect may contribute to di¡erences among these groups in preferences for healthcare interventions such as total joint replacement. 2004 Osteoarthritic joint pain. Wiley, Chichester (Novartis Foundation Symposium 260) p 258^276

Fibromyalgia (FM) is a disorder that is characterized by (a) persistent, widespread, musculoskeletal pain, and (b) abnormal pain sensitivity (i.e. allodynia) evoked by low intensity stimuli (Table 1). Neither psychiatric morbidity nor healthcare seeking behaviour accounts for abnormal pain sensitivity in FM. For example, pain-free healthy controls exhibit signi¢cantly higher pain thresholds in response to pressure stimulation than both rheumatology clinic patients with FM and community residents who meet criteria for FM but do not seek medical care for their pain (i.e. non-patients) (Aaron et al 1996, Bradley et al 1999). However, the 258

ABNORMAL PAIN SENSITIVITY

259

TABLE 1 The American College of Rheumatology classi¢cation criteria for ¢bromyalgia I. . . .

History of Widespread Pain (53 months) Left and right sides of body Above and below the waist Axial skeletal pain . cervical spine, or . anterior chest, or . thoracic spine, or . low back II.Tender point pain sensitivity . Pain, on digital palpation (44 kg), must be present in at least 11 of 18 speci¢c tender point sites Occiput Lateral epicondyle Lower cervical Gluteal Trapezius Greater trochanter Supraspinatus Knees

FM patients are characterized by signi¢cantly greater lifetime psychiatric morbidity than both the non-patients with FM and the controls (Aaron et al 1996, Bradley et al 1999). Moreover, the FM non-patients do not di¡er from the controls in psychiatric morbidity. Patients with FM, compared to controls, also show abnormal pain thresholds in response to thermal stimuli (e.g. Gibson et al 1994) as well as enhanced wind-up responses to phasic mechanical and thermal stimulation (Staud et al 2001, 2003a). However, several centrally mediated pain inhibition mechanisms in patients with FM are intact. It has been found that the abnormal pain responses described above are inhibited by the NMDA antagonist ketamine (Graven-Nielsen et al 2000) and fentanyl (Price et al 2002). In addition, combining counter-irritation procedures, such as hand immersion in a hot water bath, with distraction attenuates thermal wind-up responses in women with FM (Staud et al 2003b). Central correlates of abnormal pain sensitivity in ¢bromyalgia Neuroimaging studies indicate that abnormal pain sensitivity in patients with FM is associated with altered patterns of regional cerebral blood £ow (rCBF) during rest and in response to phasic pressure stimulation. At rest, patients with FM, compared to controls, show signi¢cantly lower levels of rCBF in the thalamus or caudate nucleus (Mountz et al 1995, Bradley et al 1999, Kwiatek et al 2000). Similar

260

BRADLEY ET AL

resting state abnormalities in brain rCBF are displayed by patients with persistent neuropathic pain and patients with metastatic cancer pain (Iadarola et al 1995, Di Piero et al 1991). In addition, recent investigations indicate that low intensity stimuli evoke abnormal pain perceptions and augmented neural input among patients with FM and those with neuropathic pain syndromes (e.g. Gracely et al 2002, Petrovic et al 1999). For example, Gracely and colleagues (2002) found that it was necessary to apply di¡erent levels of phasic pressure stimulation to the left thumbnail of patients with FM (2.4 kg) and healthy controls (4.2 kg) for both groups to produce a mean rating of 11 on a 20 point scale of pain intensity. Consistent with their verbal pain responses, both patients and controls exhibited signi¢cant increases in functional magnetic resonance imaging (fMRI) signal in the same seven brain regions (e.g. somatosensory cortices, insula, putamen, cerebellum) in response to the stimulation. However, when the controls received pressure stimulation at the same intensity level delivered to patients, they showed signi¢cant fMRI signal increases in only two brain regions; neither of these regions overlapped with those activated within the patient group. These ¢ndings suggest that abnormal pain sensitivity in persons with FM is associated with dysregulation of central processing of sensory input. Mechanisms underlying abnormal pain sensitivity Several investigators have suggested that central sensitization may contribute to the abnormal pain sensitivity associated with FM. The mechanisms underlying central sensitization involve hyper-excitability of spinal dorsal horn neurons that transmit nociceptive input to the brain. Speci¢cally, intense or prolonged nociceptive input from Ad and C a¡erents su⁄ciently depolarizes the dorsal horn neurons that Mg2+ exits NMDA-linked ion channels. This is followed by an in£ux of extracellular Ca2+ and production of nitric oxide which di¡uses out of the dorsal horn neurons. Nitric oxide, in turn, promotes the exaggerated release of excitatory amino acids and substance P from presynaptic a¡erent terminals and causes the dorsal horn neurons to become hyperexcitable (Schaible et al 2002). As a consequence, low intensity stimuli delivered to the skin or deep muscle tissue generate high levels of nociceptive input to the brain and the perception of pain. However, central sensitization generally is diminished following cessation of nociceptive input from Ad and C a¡erents. Recently, Watkins and colleagues (2001) have provided evidence that dorsal horn glia cells also play a role in producing and maintaining abnormal pain sensitivity. Synapses within the central nervous system are encapsulated by glia that normally do not respond to nociceptive input from local sites. However, following the initiation of central sensitization, spinal glial cells are activated by a

ABNORMAL PAIN SENSITIVITY

261

wide array of factors that contribute to hyperalgesia such as immune activation within the spinal cord, substance P, excitatory amino acids, nitric oxide, and prostaglandins. Once activated, glial cells release several pro-in£ammatory cytokines (e.g. tumour necrosis factor [TNF], interleukin [IL]6, IL1), substance P, nitric oxide, prostaglandins, excitatory amino acids, ATP and fractalkine that, in turn, (a) further increase the release of excitatory amino acids and substance P from the Ad and C a¡erents that synapse in the dorsal horn and (b) enhance the hyperexcitability of the dorsal horn neurons (Milligan et al 2003). It must be emphasized, however, that investigators have not been able to reliably identify abnormalities in deep muscle tissue or skin among persons with FM that might produce the prolonged nociceptive input that is necessary to initiate the events underlying the development of central sensitization or spinal glial cell activation (Sprott et al 1998). Abnormal pain sensitivity in knee osteoarthritis Patients with knee osteoarthritis (OA) generally identify pain as the primary symptom of knee or hip OA. Although knee OA is associated with abnormalities such as pressure on exposed subchondral bone or microfractures that might induce nociception and pain, patients with knee OA report pain that often is poorly localized and is not highly correlated with abnormalities in joint structure identi¢ed by radiographs or MR. This intriguing similarity in pain reports between patients with knee OA and those with FM have led several investigators to assess whether abnormalities in central processing of sensory input may also be associated with knee OA. Indeed, recent studies have provided evidence that pain associated with knee or hip OA is associated with abnormal central processing of sensory input. For example, patients with OA of the lower extremities, compared to healthy controls, exhibit longer periods of hyperalgesia and larger areas of the leg with referred pain following infusion of hypertonic saline into the tibialis anterior muscle (Bajaj et al 2001). Similarly, Kosek & Orderberg (2000) have shown that patients with unilateral hip OA, compared to healthy controls, display lower pain thresholds for pressure stimulation both at the a¡ected hip and at the una¡ected, contralateral hip. Unlike the controls, the patients fail to show an increase in pain threshold at the una¡ected, contralateral hip in response to ischaemic counterstimulation. However, 5 months after the patients undergo hip replacement surgery, there is no di¡erence between patients and controls in pain threshold levels at the contralateral hip. Moreover, the abnormality in patients’ pain modulation is reversed. That is, the patients now display a signi¢cant increase in pain threshold at the contralateral hip in response to ischaemic counterstimulation. This suggests that cessation of nociceptive input from the a¡ected

262

BRADLEY ET AL

FIG. 1. Mean pain intensity ratings of phasic pressure stimulation delivered to 4 sites on the right and left shoulders and arms. Patients with knee osteoarthritis (OA) produced signi¢cantly higher ratings than age-matched healthy controls (P ¼ 0.045).

hip allowed the patients’ pain thresholds and central pain inhibitory functions to return to normal. Our laboratory is now completing a series of studies of pain sensitivity in 11 patients with knee OA and 11 age-matched healthy controls, all of whom were right-handed (DeBerry et al 2002, Kersh et al 2001). Given that most of our patients were characterized by bilateral knee OA, we sought to determine whether they would display abnormal pain sensitivity at sites quite distant from the knees, such as the arms and shoulders. Figure 1 shows that the patients, compared to the controls, reported signi¢cantly higher levels of pain intensity in response to a series of phasic pressure stimuli applied to four sites on the left and right shoulders and arms (P ¼0.045). This e¡ect was due primarily to the patients’ responses to the higher stimulus intensities (i.e. 5.5 and 7.0 kg/cm2). However, there was no di¡erence in the responses of patients and controls to thermal heat stimulation at the same sites. This suggests that patients with knee OA may exhibit abnormal, generalized pain sensitivity only in response to stimulation of deep muscle tissue. Moreover, the consistent ¢ndings across laboratories of (a) abnormal, generalized pain sensitivity and (b) normalization of pain sensitivity and pain modulation after termination of nociceptive input strongly suggests that central sensitization and perhaps glial cell activation may contribute to the pain experiences of persons with knee OA.

ABNORMAL PAIN SENSITIVITY

263

Cognitive and a¡ective in£uences on generalized pain sensitivity in knee OA We also have begun to examine the extent to which group di¡erences in painrelated cognitive and a¡ective factors might in£uence the responses of our patients and controls to pressure stimulation of the knee. We assessed these responses with a standardized, self-report measure, the Pain Catastrophizing Scale (PCS; Sullivan et al 1995). The PCS includes three subscales that assess maladaptive cognitive and a¡ective responses to pain: helplessness, magni¢cation, and rumination. Helplessness refers to perceived di⁄culty in coping e¡ectively with pain. Magni¢cation represents a tendency to consistently anticipate that pain will produce highly negative consequences. Rumination re£ects a high level of di⁄culty in distracting oneself from pain. Figure 2 shows that the patients with knee OA, compared to controls, produced signi¢cantly higher scores on each PCS subscale (P ¼0.012^0.015). We then measured the pain thresholds of our patients and controls in response to pressure stimulation applied to 6 sites on the left and right knees. Next, using Xenon133 tracer, we performed single photon emission computed tomographic (SPECT) imaging of subjects’ rCBF responses to a 4 minute period of phasic pressure stimulation delivered to four sites on each knee. There were two stimulation conditions. In the sensory control condition, the stimulus intensities

FIG. 2. Mean (
SEM) scores on the Pain Catastrophizing Scale subscales. Patients with knee osteoarthritis (OA) produced signi¢cantly higher ratings than age-matched healthy controls on all subscales (P’s ¼ 0.012^0.015).

264

BRADLEY ET AL

FIG. 3. Mean (
SEM) pain thresholds for pressure stimulation delivered to 6 sites on the right and left knees. Patients with knee osteoarthritis (OA) displayed signi¢cantly lower pain threshold levels than age-matched healthy controls (P ¼ 0.025).

were kept constant at 1 kg. However, in the experimental condition, the intensities of the pressure stimuli were calibrated to each subject’s respective pain threshold level (3 kg/cm24pain threshold). Immediately following stimulation, each subject rated the intensity of the sensory and a¡ective (i.e. unpleasantness) dimensions of his or her pain using the McGill Pain Questionnaire (MPQ; Melzack 1975). Figure 3 shows that, consistent with our expectations, the patients displayed signi¢cantly lower pain threshold levels than the controls (P ¼0.025). In addition, Table 2 reveals that calibrating the intensity of the pressure stimulation to subjects’ pain threshold levels evoked similar MPQ ratings of sensory intensity at the right and left knee among the patients and controls. Nevertheless, compared TABLE 2 Mean (
SEM) MPQ sensory and a¡ective subscale scores of patients with knee osteoarthritis (OA) and healthy controls

MPQ subscale Right knee Sensory A¡ective Left knee Sensory A¡ective MPQ ¼ McGill Pain Questionnaire.

OA patients (n ¼ 11)

Healthy controls (n ¼ 11)

P value

17.6
2.0 6.95
1.9

15.17
2.4 1.76
0.6

0.45 0.02

18.3
2.0 7.0
2.5

17.3
2.5 1.46
0.7

0.76 0.003

ABNORMAL PAIN SENSITIVITY

265

to the controls, the patients produced signi¢cantly higher MPQ ratings of a¡ective intensity (i.e. unpleasantness) at both the right (P ¼0.02) and left (P ¼0.003) knee (Table 2). Moreover, statistically controlling for the in£uence of subjects’ PCS helplessness scale scores on their MPQ responses eliminated the signi¢cant group di¡erences in a¡ective intensity evoked by stimulation of each knee. This suggests that the di¡erence between patients and controls on this measure of pain-related cognition and a¡ect mediated the di¡erence in their MPQ pain-a¡ect scores. Using statistical parametric mapping, we compared the SPECT images of brain rCBF in the sensory control and experimental conditions within the patient and control groups. We found that, during the experimental condition, the patients and controls generally exhibited reliable increases in rCBF in similar brain regions (e.g. somatosensory cortex) in response to stimulation of both the right and left knee. However, only the patients with knee OA exhibited reliable increases in brain rCBF in the left and right anterior cingulate cortex (ACC) in response to stimulation of each knee (P ¼0.05). Activation of the ACC is signi¢cantly associated with individuals’ reports of pain unpleasantness and other measures of the a¡ective component of pain (e.g. Rainville et al 1997). In addition, among healthy volunteers, more robust activation of the ACC is associated with higher levels of thermal pain sensitivity (Coghill et al 2003). Finally, we performed an exploratory analysis of possible ethnic group di¡erences among our patients on their PCS, MPQ and brain rCBF responses. Figure 4 shows that our ¢ve African-American patients tended to produce higher

FIG. 4. Mean (
SEM) scores on the Pain Catastrophizing Scale subscales. African-American patients with knee osteoarthritis (OA) tended to produce higher ratings than Caucasian patients with knee OA on all subscales, although the largest group di¡erence occurred on the helplessness subscale (P ¼ 0.10).

266

BRADLEY ET AL

TABLE 3 Mean (
SEM) MPQ sensory and a¡ective subscale scores of AfricanAmerican (AA) and Caucasian (C) patients with knee osteoarthritis (OA)

MPQ subscale Right knee Sensory A¡ective Left knee Sensory A¡ective

AA patients (n ¼ 5)

C patients (n ¼ 6)

P value

22.2
2.8 11.3
2.7

13.8
1.7 3.3
1.7

0.03 0.03

21.6
2.4 10.9
1.6

15.5
2.6 3.8
1.5

0.12 0.01

MPQ ¼ McGill Pain Questionnaire.

scores than the six Caucasians on all of the PCS subscales; the largest group di¡erence was found on the helplessness subscale (P ¼0.10). Moreover, Table 3 reveals that the African-American patients produced higher MPQ a¡ective intensity scores than the Caucasian patients in response to stimulation of the right (P ¼0.03) and left (P ¼0.01) knees. The African-American patients, compared to the Caucasians, also produced signi¢cantly higher MPQ sensory scores at the right knee (P ¼0.03). In addition, consistent with the self-report data, we found that the African-American patients, compared to the Caucasians, displayed signi¢cantly greater activation of the ACC in response to stimulation of the right and left knees (P ¼0.05). These ¢ndings are very similar to those reported by Edwards and colleagues (Edwards & Fillingim 1999, Edwards et al 2001). Their studies of both healthy volunteers and patients with persistent pain revealed that African-Americans, compared to Caucasians, perceive noxious thermal and ischaemic stimuli as signi¢cantly more aversive. Implications of ethnic group di¡erences in pain sensitivity on disparities in total joint replacement procedures Variations among ethnic groups in pain-related cognition or a¡ect may have important implications for disparities in their preferences for healthcare procedures. For example, Ibrahim et al (2002) have demonstrated that, in the USA, African-Americans are less willing than Caucasians to undergo knee or hip replacement surgery, even if it is recommended by their physicians. After controlling for demographic, socioeconomic and clinical variables, the factors that mediated this relationship were the patients’ expectations in post-surgical pain, walking ability and length of hospital stay. On all of these variables, the African-Americans’ expectations were signi¢cantly more negative than those of the Caucasians, despite the fact the African-Americans reported they were less

ABNORMAL PAIN SENSITIVITY

267

familiar with total joint replacement surgery than the Caucasians. Our ¢ndings described above, as well as those of Edwards et al (1999, 2001) suggest that the relatively greater levels of negative pain-related cognition and a¡ect among African-Americans may be related to their negative expectations regarding postsurgical pain and reluctance to undergo total joint replacement surgery. It should be noted that Ibrahim’s research group recently attempted to address this issue by comparing the responses of their African-American and Caucasian patients to the WOMAC pain and function scales (Ang et al 2003). They found no di¡erences among the ethnic groups on either of these scales. However, unlike the MPQ, the WOMAC pain scale does not include independent measures of the sensory and a¡ective dimensions of pain. Therefore, we plan to perform new studies of possible associations between ethnic group di¡erences in pain-related cognition and a¡ect and group disparities in preferences for knee or hip replacement surgery. Conclusions and future research directions Both patients with FM and those with knee OA display abnormal pain sensitivity at multiple anatomic sites. Laboratory studies using psychophysical pain measurement procedures or neuroimaging of brain activity suggest that abnormal pain sensitivity in persons with these disorders is associated with central augmentation of sensory input. Although several investigators have noted that central sensitization and/or spinal glial cell activation may contribute to abnormal pain sensitivity in FM and OA, it is very important to emphasize that, until the sources of nociceptive input from deep muscle tissue are identi¢ed in persons with FM, it is not appropriate to suggest that central sensitization underlies the pain experiences of these individuals. In contrast, given that (a) the source of nociceptive input in persons with knee OA is known and (b) there is evidence that abnormalities in pain sensitivity and pain modulation are diminished following total joint replacement surgery, it is reasonable to suggest that central sensitization and perhaps spinal glial cell activation contribute to the pain experiences of individuals with knee OA. Recent evidence from our laboratory suggests that pain-related cognitive and a¡ective factors also in£uence the responses of patients with knee OA to noxious stimulation in the laboratory. Speci¢cally, even when patients with knee OA and age-matched healthy controls receive pressure stimulation of the knees that produces similar perceptions of sensory intensity, the patients report that the stimulation evokes signi¢cantly higher levels of pain a¡ect. This group di¡erence in pain-a¡ect ratings is strongly associated with group di¡erences on the helplessness subscale of the Pain Catastrophizing Scale. Consistent with the ¢ndings on these self-report measures, pressure stimulation also evokes

268

BRADLEY ET AL

signi¢cant increases in brain rCBF in substantially larger areas of the anterior cingulate cortex in patients compared to controls. Moreover, within the patient group, the African-Americans, compared to the Caucasians, tend to produce higher ratings of pain a¡ect and helplessness as well as greater cingulate activation. We believe that future research e¡orts should include prospective clinical or population-based studies of the development of generalized, abnormal pain sensitivity in persons with knee OA and the physiologic and psychosocial factors that in£uence pain sensitivity in these individuals. It also is critical to devote greater e¡ort to the development of pharmacological interventions that may alter central nervous system functions that in£uence pain sensitivity and pain modulation in persons with knee OA. Finally, we believe it is very important to better understand ethnic group di¡erences in pain-related cognition and a¡ect that may contribute to group di¡erences in pain responses and preferences for health care interventions such as total joint replacement surgery.

Acknowledgements Supported by a grant from the Fetzer Institute. We gratefully acknowledge the contributions of the Fetzer Working Group on Pain and Su¡ering to the development of our research studies on knee osteoarthritis as well as the contributions of Nancy L. McKendree-Smith, PhD, Adriana Sotolongo, BA and Ronda Cannon, BA to the preparation of this paper.

References Ang DC, Ibrahim SA, Burant CJ, Kwoh CK 2003 Is there a di¡erence in the perception of symptoms between African Americans and whites with osteoarthritis? J Rheumatol 30:1305^310 Aaron LA, Bradley LA, Alarco¤n GS et al 1996 Psychiatric diagnoses are related to health care seeking behavior rather than illness in ¢bromyalgia. Arthritis Rheum 39:436^445 Bajaj P, Graven-Nielsen T, Arendt-Nielsen L 2001 Osteoarthritis and its association with muscle hyperalgesia: an experimental, controlled study. Pain 93:107^114 Bradley LA, Sotolongo A, Alberts KR et al 1999 Abnormal regional cerebral blood £ow in the caudate nucleus among ¢bromyalgia patients and non-patients is associated with insidious symptom onset J Musculoskel Pain 7:285^292 Coghill RC, McHa⁄e JG, Yen YF 2003 Neural correlates of interindividual di¡erences in the subjective experience of pain. Proc Natl Acad Sci USA 100:8538^8542 DeBerry JJ, Fry RB, Kersh BC et al 2002 Trait-like catastrophizing mediates elevated pain a¡ect ratings in patients with knee osteoarthritis (OA) during suprathreshold pressure stimulation. Arthritis Rheum 46:S408 Di Piero V, Jones AK, Iannotti F et al 1991 Chronic pain: a PET study of the central e¡ects of percutaneous high cervical cordotomy. Pain 46:9^12 Edwards RR, Fillingim RB 1999 Ethnic di¡erences in thermal pain responses. Psychosom Med 61:346^354 Edwards RR, Doleys DM, Fillingim RB, Lowery D 2001 Ethnic di¡erences in pain tolerance: clinical implications in a chronic pain population. Psychosom Med 63:316^323

ABNORMAL PAIN SENSITIVITY

269

Gibson SJ, Littlejohn GO, Gorman MM, Helme RD, Granges G 1994 Altered heat pain thresholds and cerebral event-related potentials following painful CO2 laser stimulation in subjects with ¢bromyalgia syndrome. Pain 58:185^193 Gracely RH, Petzke F, Wolf JM, Clauw DJ 2002 Functional magnetic resonance imaging evidence of augmented pain processing in ¢bromyalgia. Arthritis Rheum 46: 1333^1343 Graven-Nielsen T, Aspegren Kendall S, Henriksson KG et al 2000 Ketamine reduces muscle pain, temporal summation, and referred pain in ¢bromyalgia patients. Pain 85:483^491 Iadarola MJ, Max MB, Berman K et al 1995 Unilateral decrease in thalamic activity observed with positron emission tomography in patients with chronic neuropathic pain. Pain 63: 55^64 Ibrahim SA, Simino¡ LA, Burant CJ, Kwoh CK 2002 Di¡erences in expectations of outcome mediate African-American/white patient di¡erences in ‘‘willingness’’ to consider joint replacement. Arthritis Rheum 46:2429^2435 Kersh BC, Deutsch G, Bradley LA et al 2001 Pain sensitivity in knee osteoarthritis patients is associated with bilateral activation of the brain anterior cingulate cortex. Arthritis Rheum 44:S140 Kosek E, Ordeberg G 2000 Lack of pressure pain modulation by heterotopic noxious conditioning stimulation in patients with painful osteoarthritis before, but not following, surgical pain relief. Pain 88:69^78 Kwiatek R, Barnden L, Tedman R et al 2000 Regional cerebral blood £ow in ¢bromyalgia: single-photon-emission computed tomography evidence of reduction in the pontine tegmentum and thalami. Arthritis Rheum 43:2823^2833 Melzack R 1975 The McGill Pain Questionnaire: major properties and scoring methods. Pain 1:277^299 Milligan ED, Twining C, Chacur M et al 2003 Spinal glia and proin£ammatory cytokines mediate mirror-image neuropathic pain in rats. J Neurosci 23:1026^1040 Mountz JM, Bradley LA, Modell JG et al 1995 Fibromyalgia in women: abnormalities of regional cerebral blood £ow in the thalamus and the caudate nucleus are associated with low pain threshold levels. Arthritis Rheum 38:926^938 Petrovic P, Ingvar M, Stone-Elander S, Petersson KM, Hansson P 1999 A PET activation study of dynamic mechanical allodynia in patients with mononeuropathy. Pain 83:459^470 Price DD, Staud R, Robinson ME, Mauderli AP, Cannon R, Vierck CJ 2002 Enhanced temporal summation of second pain and its central modulation in ¢bromyalgia patients. Pain 99:49^59 Rainville P, Duncan GH, Price DD, Carrier B, Bushnell MC 1997 Pain a¡ect encoded in human anterior cingulate but not somatosensory cortex. Science 27:968^971 Schaible H-G, Ebersberger A, von Banchet GS 2002 Mechanisms of pain in arthritis. Ann NY Acad Sci 966:343^354 Sprott H, Bradley LA, Oh SJ et al 1998 Immunohistochemical and molecular studies of serotonin, substance P, galanin, pituitary adenylyl cyclase-activating polypeptide, and secretoneurin in ¢bromyalgic muscle tissue. Arthritis Rheum 41:1689^1694 Staud R Vierck CJ, Cannon RL, Mauderli AP, Price DD 2001 Abnormal sensitization and temporal summation of second pain (wind-up) in patients with ¢bromyalgia syndrome. Pain 91:165^175 Staud R, Cannon RC, Mauderli AP, Robinson ME, Price DD, Vierck CJ Jr 2003a Temporal summation of pain from mechanical stimulation of muscle tissue in normal controls and subjects with ¢bromyalgia syndrome. Pain 102:87^95 Staud R, Robinson ME, Vierck CJ Jr, Price DD 2003b Di¡use noxious inhibitory controls (DNIC) attenuate temporal summation of second pain in normal males but not in normal females or ¢bromyalgia patients. 101:167^174

270

DISCUSSION

Sullivan MJL, Bishop S, Pivik J 1995 The Pain Catastrophizing Scale: development and validation. Psychol Assess 7:524^532 Watkins LR, Milligan ED, Maier SF 2001 Glial activation: a driving force for pathological pain. Trends Neurosci 24:450^455

DISCUSSION Brandt: There is not only ethnic disparity in the performance of total hip and total knee replacement, but also a gender disparity. Did you look at any of the pre-imaging data in relation to gender? Bradley: No we haven’t done this yet. Brandt: Do you have any sense of the extent to which unwillingness to undergo surgery among African-Americans, relative to Caucasians, may be in£uenced by the experience of friends, relatives and personal contacts? Bradley: That is a good question. One of our colleagues studied this phenomenon among healthy college-age students for her doctoral dissertation. She found the same sort of e¡ect that we saw in these patients: in response to a variety of tasks there were much greater a¡ective responses in the AfricanAmerican college students and much more catastrophizing. Even among these healthy college students, one thing that di¡erentiated the African-Americans from the Caucasians is their experiences with chronic pain in their families. The African-Americans had much more experience with chronic pain in family members. Roger Fillingim also has some data showing that African-American students themselves tend to have more experience with painful problems than the Caucasians (Edwards & Fillingim 1999). We think these family factors have a real in£uence on the way people think and respond. Ibrahim’s data on ethnic group di¡erences in preferences for knee arthroplasty are also relevant here (Ibrahim et al 2002). They asked the patients whether they had friends or relatives who had had total joint replacement surgery, and to what extent the patients were familiar with the procedure. There was less reported familiarity among the AfricanAmericans. This was odd. It goes against most of the other data, and I can’t explain it. Pisetsky: In our hospital, pain is called the ¢fth vital sign. On every patient, we are obliged to ask what is their pain at that moment. If we do this with people with ¢bromyalgia, invariably it is 8, 9 or 10. How do you interpret that? Is that intensity, or unpleasantness? Is there a way to ask this question better so that we don’t have all these patients with 8s, 9s or 10s that we are not doing something for? Bradley: It is hard. In that sort of situation where you are giving people a scale, invariably their responses will be in£uenced by emotions, thoughts and a¡ect. I will tell you the rigours of how we have to train people to use these visual analogue scales. In the ¢bromyalgia study we are using a mechanical visual analogue scale that Don Price developed at the University of Florida. It looks

ABNORMAL PAIN SENSITIVITY

271

like a slide rule. There is a lever on top of the scale, and as you pull this lever to one end it exposes a red bar. The patient is supposed to match their internal perception of pain intensity or unpleasantness to the length of that red bar. Training people to use this requires giving people practice in extending the bar all the way, so they really see what 10 cm looks like. You then have to make sure that they can give you at least some semblance of a linear response between stimulus intensity and response on that scale. You also have to make sure that if you ask them to expose half the red line that they actually get close to exposing half. Then you have to teach them the di¡erence between intensity and unpleasantness. We use the analogy that intensity is the loudness of the music on the radio and unpleasantness is how aversive that music is. It takes anything from 30 min to 2 days to teach people to do this. Some people never get it so we can’t use them in the study. Pisetsky: So is the interpretation that people with ¢bromyalgia are experiencing pain unpleasantly but that the pain is not that intense? Bradley: No, I guess the idea would be that when they talk to you in the clinic, whatever they are reporting is an amalgam of intensity and unpleasantness. In the lab you can do some things to di¡erentiate those dimensions of the pain response, but in the clinic it is hard to do so. A rough rule of thumb would be to assess their life stress and psychological distress to get an idea about this. Pisetsky: When you have them think unpleasant thoughts, do you know what they are? Bradley: Yes. Pisetsky: Is that informative? Bradley: Yes. I don’t know how to do a qualitative analysis, but I know in my gut that the quality of the images are very di¡erent. For some patients the stressor might be family death or illness. Then for positive images, very often, these patients will begin with a statement such as ‘I want to think about my daughter’s graduation from grade school’. Then, sometimes, you’ll see an intense blood pressure response. Later on we’ll ask what they were thinking about and they’ll say something like, ‘I was thinking about my daughter’s graduation and then I ran into my rat former husband’. For many patients there often will be unpleasantness weaved into the pleasantness. In healthy controls the images tend to be much less traumatic. Schaible: I have a question about the fMRI images. What you have shown is an e¡ect quite similar to any response to noxious stimulation. Can you show a di¡erence in these patients between a pain-free stage and a painful stage so that you do not need to apply stimuli? If you apply stimuli it may evoke a pattern which is di¡erent from that which is normally there in this brain. And do you have any evidence for mirror pain in these patients? Bradley: With stimulation in our ¢bromyalgia patients, and also a bit in our knee OA patients, even at these relatively low intensity levels we tend to see ipsilateral as

272

DISCUSSION

well as contralateral activation. We have never tried to study these patients before and after an analgesic treatment. We have not been able to do this yet. However, if we image ¢bromyalgia patients only in resting conditions we tend to see a strong reduction in blood£ow in the thalamus (Mountz et al 1995). This is very similar to what people have observed in patients with neuropathic pain (e.g. Iadarola et al 1995). This is part of what started to drive us and other people to think about central abnormalities in pain processing. Mistakenly, this drove some people to attribute the pain sensitivity to central sensitization without knowing the source of the input. Nevertheless, reduced thalamic activity is what we see in ¢bromyalgia patients at baseline without stimulation. Hunt: Do the patients with ¢bromyalgia show an exaggerated placebo response? Bradley: When we look at the pharmacological studies, the placebo response runs anywhere from 30^40%. They really do show a strong placebo response. Hunt: Being a bit more reductionist, there are certain polymorphisms in genes, such as the CRMT gene that breaks down catecholamines, which seem to predispose people to di¡erent sensitivities to pain. This polymorphism is widespread. Isn’t it important to look at these? Have you thought about screening your patients? Bradley: Yes. We have started a family aggregation study. We are in the process of collecting 80 ¢bromyalgia families and 80 control families. We are doing pain sensitivity testing not only with the probands and controls but also with one brother or one sister of each proband or control. As much as we can, we want to end up with 40 brothers and 40 sisters in each group. We have completed six families already. Our pilot data based on 12 families showed us that the sisters of the patients with ¢bromyalgia were much more pain sensitive than all other sibling groups, including the sisters of the controls. Astoundingly, they also show lower serotonin levels. We are also collecting blood and immortalizing DNA. We want to test hypotheses about a number of polymorphisms in a number of genes, including 5HT. There are at least two other groups working on similar studies. Dieppe: I’m particularly interested in your use of imagery. We have been using similar techniques as an intervention, based on the concept of emotional disclosure as a way of treating chronic disease. I wonder whether you are achieving an intervention with these imagery approaches, and are in fact altering the pain experience in these ¢bromyalgia patients over a longer time period. Bradley: Are you using the writing technique? Dieppe: Yes, mostly, but we have been doing a bit of this in RA patients some of whom don’t like to do writing and prefer to talk. Bradley: We have not actually asked that question, but it is an interesting one. My guess would be that since they are really only disclosing over the period of a day and we never ask them to continue disclosing, this is not an ideal situation for

ABNORMAL PAIN SENSITIVITY

273

improvement. What you are doing over an extended period of time is a much better paradigm for achieving a positive e¡ect. Brandt: I am interested in the group of ¢bromyalgia people you recruited from the community who had not made the decision to become patients. They remind me of a cohort of older people in Indiana whom we studied for OA. We found a substantial number of people with radiographic OA who had WOMAC pain scores as high as those of patients in our clinic, but these people had made the decision that they didn’t need medical attention for this. I don’t know whether there was a qualitative di¡erence between their pain and the pain of our clinic patients, or whether they had better coping skills. Were there di¡erences between your people with ¢bromyalgia in the community and your patients with ¢bromyalgia? Bradley: This is interesting. Even though the patients and non-patients showed the same level of pain sensitivity, the patients reported signi¢cantly higher levels of pain. Their average score on the total McGill Pain Questionnaire was about 30, and the score of the non-patients was about 17. This gets back to David’s point. It may well be that the emotional distress was driving the elevation in those scores. This was the major di¡erence between the groups. Even after we controlled for the psychiatric factors, the patients still reported higher levels of depression and anxiety, and lower self e⁄cacy (Kersh et al 2001). Blake: The word ‘catastrophizing’ obviously has a negative implication. Given the way the normalized data were collected, how could you not say that the Americanized Caucasian population was emotionally blunted? Bradley: We have often been accused of this! You could reinterpret a higher score on that scale as re£ecting a greater willingness to admit, if not a greater actual experience of perceived di⁄culty in coping with pain. I think you are right: it is probably good to cast this ethnic group di¡erence in non-pathological terms. Henry: Many patients with chronic pain have sleep disorders, and they often complain of fatigue. In one of your early slides you listed these as separate entities. Can you comment on your vision of these as separate entities, or variations of the same theme? Bradley: I don’t think that they are really separate entities. I think the fatigue reports and sleep disturbance are so frequent in this group of people it is probably part of whatever this syndrome is. Henry: Is the fatigue due to sleep disturbance or something else? Bradley: I don’t know the complete answer to that. There is a large subgroup of people with ¢bromyalgia who show intermittent or phasic alpha intrusion during non-REM sleep. At least in part, we can probably attribute the fatigue to this sleep disturbance. There may be other factors. This gets back to the issue we raised about Linda Watkins’ construct of the sickness response.

274

DISCUSSION

Fox: Where you have looked at patients and non-patients, is there a di¡erence in how they respond to pharmacological interventions? Bradley: There are no data on this. Ordeberg: I was very interested in this trait of catastrophizing, and the di¡erences between ethnic groups. Have you looked to see whether there was a di¡erence with socioeconomic status (SES)? There might be di¡erences between these groups. And those with less insurance and fewer economic resources have more fear with their disease for this reason. Bradley: That is a very good point. In that particular study we used education as a marker for SES. There was no signi¢cant di¡erence between the groups in terms of education. But, there are various ways of measuring SES, and there is a whole literature on measuring SES using multiple methods. This is something we need to pay more attention to in the pain ¢eld. Felson: Let me change direction a little. I appreciate that the audience is intrigued by the discussion of ¢bromyalgia, but this is a meeting about OA pain. I wanted to change the focus back to OA. The implication of this discussion goes back to the point where we haven’t used the terms OA or ¢bromyalgia in the last ¢ve or 10 minutes because it has been implied that there has been an element of the same going on in OA. I want to raise this question again. Let me throw out a couple of observations. One is that African-Americans seem to have the higher pain reporting you described, but in most epidemiological surveys they have far lower prevalences of ¢bromyalgia than Caucasians. I don’t understand this. Furthermore, your imaging studies suggested that there were qualitative di¡erences in OA brain regions a¡ected, whereas ¢bromyalgia studies show no qualitative di¡erences in general. They show the same areas of the brain being lit up, only by lower levels of the same noxious stimuli. Is OA pain the same sort of entity as ¢bromyalgia pain? Bradley: They are similar. But, there are some di¡erences. When we did our generalized pain sensitivity test in the patients with OA, we not only used the mechanical pressure stimulation applied to the arm and shoulder, but we also used a thermal stimulation protocol applied to the forearm. The patients with OA, unlike what is seen in patients with ¢bromyalgia, did not di¡er from the controls in their response to thermal stimulation. It may be that if we used a more aggressive stimulation technique we might see enhanced thermal responses. At present, however, we can only say that the OA patients don’t show the £orid sensitivity to a variety of stimuli that is seen in patients with ¢bromyalgia. With regards to the imaging study, one reason why I think we saw the great group di¡erence in activation of the cingulate is that we attracted a group of patients that contained roughly the same number of African-Americans and Caucasians. As a result, we saw that the high a¡ective pain response and the high level catastrophizing primarily occurred in the African-Americans. Our studies in

ABNORMAL PAIN SENSITIVITY

275

¢bromyalgia involve primarily Caucasians, just like in the epidemiological data. The mix of subjects has an important impact on the ¢ndings that are produced. There is a recent paper by Bob Coghill (Coghill et al 2003), in which he looked at healthy people whom he de¢ned by quantitative sensory testing as either relatively pain sensitive or pain stoic. He put them through a sensory stimulation task involving the right lower leg and did neuroimaging, ¢nding that if you look at the level of the thalamus and give people the same amount of stimulation, you see essentially the same brain responses. But his pain sensitive people showed much greater activation of the contralateral anterior cingulate cortex in response to the stimulation. Also, the pain sensitive individuals more frequently showed activation of a region of the primary somatosensory cortex that corresponds to the lower leg. These ¢ndings indicate that the cingulate is a good marker for a¡ective responses involved in enhanced pain sensitivity. Hunter: I was interested that you showed a substantial di¡erence in cortisol levels between the ¢bromyalgia and the control patients. You mentioned that serotonin levels are also somewhat di¡erent between these two groups. This potentially could be extrapolated to OA. In terms of the pain sensitization, how much of the variability in pain sensitization can you attribute to potential di¡erences in those hormonal levels? Bradley: I can’t give you an answer in terms of amount of variance accounted for. But the di¡erence in cortisol production in ¢bromyalgia is probably one of the most robust ¢ndings in the ¢bromyalgia literature. In studies where a HPA axis challenge has been produced by injecting a bolus of CRH we also see a diminished cortisol response in patients with ¢bromyalgia (e.g. Cro¡ord et al 1994). This is a very reliable ¢nding among these patients. At the very least I would say that ¢bromyalgia patients show an impaired physiological response to stress. This probably is related to their pain reports or their a¡ective reports as a function of stress. There probably is a relationship but it hasn’t been quanti¢ed yet. Lohmander: This is fascinating stu¡. What do we have in the way of longitudinal data which would answer the question of whether having chronic pain, such as hip OA, would in itself drive you towards catastrophizing? Or does having the personality or trait in itself provide a risk factor for you experiencing pain in this way and then becoming one of these patients? What is the chicken and what is the egg? Do you have a feeling for this at the moment? Bradley: No. That is a very good question. One of our future aims is to be involved in a longitudinal study where we could look at pain/catastrophizing at baseline among patients with early stages of OA, and follow them over time. We would like to look for changes in pain sensitivity and to what extent catastrophizing in£uences those responses.

276

DISCUSSION

Kidd: We looked at this in a group of RSI/work-related upper limb disorder patients, and compared these to a group with carpal tunnel. The hypothesis was that in the RSI group there would be more catastrophizing. Disappointingly, we were unable to show any di¡erences on any of the psychological variables we looked at. It seemed that chronic pain induced the catastrophizing rather than the other way round. Pisetsky: Alan Silman has longitudinal data. His group has de¢ned as widespread body pain followed over a number of years (McBeth et al 2003). This pain is predictive of a number of things including malignancy. References Coghill RC, McHa⁄e JG, Yen YF 2003 Neural correlates of interindividual di¡erences in the subjective experience of pain. Proc Natl Acad Sci USA 100:8538^8542 Cro¡ord LJ, Pillemer SR, Kalogeras KT et al 1994 Hypothalamic-pituitary-adrenal axis perturbations in patients with ¢bromyalgia. Arthritis Rheum 37:1583^1592 Edwards RR, Fillingim RB 1999 Ethnic di¡erences in thermal pain responses. Psychosom Med 61:346^354 Iadarola MJ, Max MB, Berman KF et al 1995 Unilateral decrease in thalamic activity observed with positron emission tomography in patients with chronic neuropathic pain. Pain 63:55^64 Ibrahim SA, Simino¡ LA, Burant CJ, Kwoh CK 2002 Di¡erences in expectations of outcome mediate African-American/White patient di¡erences in ‘willingness’ to consider joint replacement. Arthritis Rheum 46:2429^2435 Kersh BC, Bradley LA, Alarcon GS et al 2001 Psychosocial and health status variables independently predict health care seeking in ¢bromyalgia. Arthritis Rheum 45:362^371 McBeth J, Silman AJ, Macfarlane GJ 2003 Association of widespread body pain with an increased risk of cancer and reduced cancer survival: a prospective, population-based study. Arthritis Rheum 48:1686^1692 Mountz JM, Bradley LA, Modell JG et al 1995 Fibromyalgia in women. Abnormalities of regional cerebral blood £ow in the thalamus and the caudate nucleus are associated with low pain threshold levels. Arthritis Rheum 38:926^938

Osteoarthritic Joint Pain: Novartis Foundation Symposium 260. Volume 260 Edited by Derek J. Chadwick and Jamie Goode Copyright ¶ Novartis Foundation 2004. ISBN: 0-470-86763-9

Chair’s summing-up David Felson Clinical Epidemiology Research and Training Unit,, Boston University School of Medicine, 715 Albany Street, A 203, Boston, MA 02118-2526, USA

I wanted to summarize the questions that have emerged during these discussions. At the beginning of the symposium, I posed a set of questions. These were broad and not as well informed as ones we might think of now. These included: what is the pathophysiology of osteoarthritic (OA) joint pain? Why do some but not all people with OA get joint pain? And are there treatment opportunities that go along with a better understanding of OA joint pain? I have tried to cluster questions that remain by the category of question, and some of them bridge our two scienti¢c communities. First, is there central sensitization in non-in£ammatory mechanical joint disorders such as OA? This question came out in the ¢rst day as we heard about the prominence and importance of central sensitization induced by aggressive in£ammatory provocations. Obviously, OA is not such an entity. The question is, are the same principles relevant in OA? There were many discussions about the role of central sensitization. Also, what is the correlation between nociception seen in animals and human nociception? Then, for clinicians there were another set of questions. What are the relative contributions of structural features to OA pain? Philip Conaghan and Peter Simkin talked a lot about this, suggesting that bone marrow lesions were a frequent source of pain. Paul Creamer cautioned us about using cross-sectional associations to make aetiological inferences. Synovitis is a potential source of pain, as are osteophytes and periarticular lesions. What are their relative frequencies? Which of these causes pain, and is one or more a common source of pain? Do each of the structural contributors to OA pain cause a di¡erent kind of pain? We talked about the di¡erent qualities of pain and what these might re£ect. In knee OA, on the basis of data from the Health ABC study, there has been a speci¢c question as to how important the patellofemoral joint is. Then we go back to neuroscience questions. One of these came up speci¢cally today: what are the changes in gene expression in primary a¡erents in OA pain? Is this just a neurological disease? Should the rheumatologists be thrown out of the room? What part of the nociceptive process should be targeted for interventions? Do we target transduction, neurogenic in£ammation, the other end of the synapse 277

278

CHAIR’S SUMMING-UP

(in the CNS) or post-synaptic events? How similar are nociceptive mechanisms in OA to mechanisms of other pain? And can we generalize from some of the models, such as the bone pain model and in£ammogenic model that we heard about, to OA pain? What is the role of mechanoreceptors here? Jim Henry and Blair Grubb pushed really hard to get the clinicians to de¢ne appropriate models for this entity. This is terribly important. How relevant are models of OA with respect to OA pain? Do we need to ¢nd animal models where the animals have pain with their OA? I think the answer is yes. How do we ¢gure this out? Can clinicians identify the critical elements of a model of OA pain, and what features should it contain? The thought was that it might contain synovium and bone marrow lesions, but does it need to have cartilage loss? Most people would say that cartilage loss is a sine qua non of OA, even though it might not be a source of pain. What about current OA models? Would they work or do we need to develop new models? Then I think there was a lot of back or forth as to how well the acute models we are all dealing with generalize to chronic disease. Are we modelling peripheral or central functions? We heard about in£ammatory mediators and neurogenic mediators being released into joints and spinal cord. One of the questions is, ‘What is the role of substance P and CGRP in the joint?’ It sounds like substance P is a neurologic e¡ector in the joint. Is it important in creating in£ammation in OA that would make OA a similar model to the other ones Hans-Georg told us about? How important is neurogenic in£ammation in general in the overall health of the joint environment? There were parts of talks which dealt with COX-2 inhibitors and their e¡ects on prostaglandins as mediators in the spinal cord. Other models have found that leukotrienes may be as, or more, important than prostaglandins. The question is, what role do all these play in this disease and in models of disease? Next, how can we identify OA at an early enough stage to treat it successfully? This is a human clinical question but it has implications for animal models. This is an important question, especially with some of the emerging work on vulnerable joints showing that once mechanical malalignment or anatomic changes have occurred, it would be di⁄cult to envision an easy pharmacological approach to therapy. Early treatment becomes critical. Then there are the clinical questions. Do di¡erent pain patterns and descriptions by patients correlate with di¡erent sources of pain? Why are some joints painful such as the knee and others almost never painful such as the elbow? This was Paul Dieppe’s question and I think it is a wonderful one. Is it just mechanical loading that does this? A correlated question is why do some people have pain with OA while others don’t? And why do only some animals have pain with OA? It is a parallel set of questions and identifying the answers to these questions may provide answers to the others. Why does exercise ameliorate OA joint pain? What e¡ect does muscle strength have on OA long-term, and is this di¡erent in

CHAIR’S SUMMING-UP

279

vulnerable versus normal joints? How important is the rate of impulse loading and its deceleration in causing OA? Does bone give in OA, or have hydraulic e¡ects in response to loading? Is increased intraosseus pressure a major reason for OA joint pain? Would bone fenestration help this pain? How important are cortical mechanisms, such as discordance between motor intention and sensory feedback, in OA pain? How important is abnormal pain sensitivity in OA pain? Then, there are the questions about the high rate of bilateral disease that we raised earlier. In summary, there are a plethora of questions for clinicians and basic scientists, rheumatologists and neuroscientists to address. Our interactions at this meeting suggest that many of these questions will be most fruitfully addressed by interdisciplinary collaborations, and I hope that this meeting encourages such enterprises.

Osteoarthritic Joint Pain: Novartis Foundation Symposium 260. Volume 260 Edited by Derek J. Chadwick and Jamie Goode Copyright ¶ Novartis Foundation 2004. ISBN: 0-470-86763-9

Index of contributors Non-participating co-authors are indicated by asterisks. Entries in bold indicatepapers; other entries refer to discussion contributions.

187, 189, 191, 201, 202, 204, 218, 238, 240, 253, 256, 274, 277 Fernihough, J. K. 39, 42, 58, 62, 102, 104, 133, 147, 187, 219 Fox, A. 115, 116, 118, 119, 120, 134, 135, 137, 147, 151, 153, 216, 238, 239, 256, 274

A *Alarco¤n, G. A. 258 B Blake, D R. 133, 136, 145, 148, 151, 153, 154, 174, 175, 176, 177, 178, 238, 239, 240, 241, 252, 253, 254, 255, 256, 273 Bradley, L. A. 74, 78, 115, 116, 119, 120, 121, 136, 258, 270, 271, 272, 273, 274, 275 Brandt, K. D. 36, 39, 40, 41, 42, 45, 47, 49, 58, 59, 60, 61, 62, 63, 76, 77, 95, 96, 98, 101, 104, 201, 203, 270, 273

G Grubb, B. D. 23, 24, 25, 27, 28, 37, 38, 39, 40, 41, 42, 43, 44, 46, 47, 62, 76, 77, 97, 99, 100, 115, 118, 119, 135, 146, 149, 151, 153, 178, 214, 215, 217, 218, 219, 239, 255, 256

C *Clark A. 79 Conaghan, P. G. 96, 103, 178, 191, 201, 202, 203, 204 Creamer, P. 23, 42, 44, 64, 74, 75, 76, 78, 103, 117, 137, 201, 204

H *Haigh, R. C. 154, 241 Henry, J. L. 25, 26, 37, 39, 77, 96, 120, 134, 139, 145, 146, 147, 148, 149, 150, 151, 152, 153, 215, 216, 273 Herzog, W. 40, 59, 62, 79, 95, 96, 97, 98, 99, 146, 177, 178, 188, 189, 203 Hunt, S. P. 221, 238, 239, 240, 252, 256, 272 Hunter, D. J. 22, 23, 39, 44, 47, 59, 60, 78, 99, 101, 116, 118, 119, 152, 176, 190, 202, 203, 253, 255, 275

D *DeBerry, J. J. 258 *Deutsch, G. 258 Dieppe, P. 24, 37, 45, 60, 61, 75, 78, 98, 100, 117, 135, 137, 138, 148, 152, 153, 177, 189, 205, 219, 254, 272

I E

*Inglis, J. J. 122

Evans, C. H. 38, 98, 147, 151, 255 K F

*Kersh, B. C. 258 Kidd, B. L. 25, 45, 115, 118, 120, 122, 133, 134, 135, 136, 137, 138, 150, 215, 218, 219, 239, 276

Felson, D. T. 1, 26, 39, 43, 44, 60, 61, 62, 75, 76, 95, 96, 100, 101, 102, 104, 116, 118, 120, 137, 147, 149, 151, 175, 176, 178, 280

INDEX OF CONTRIBUTORS

Koltzenburg, M. 37, 40, 43, 44, 46, 47, 175, 176, 202, 203, 206, 214, 215, 216, 217, 218, 219 Kuettner, K. E. 37, 75, 77, 78, 98, 100, 103, 146, 187, 238 L *Lewis, J. 154 Lohmander, S. 26, 42, 43, 62, 77, 115, 148, 152, 187, 189, 254, 275 *Longino, D. 79 M Mackenzie, A. 27, 78, 134, 146, 150, 218, 239 Malcangio, M. 120, 137, 150 Manning, A. M. 26, 43, 102, 103, 104 *Mantyh, P. W. 221 *Mapp, P. I. 241 *McCabe, C. S. 154, 241 *McLain, D. A. 258 O Ordeberg, G. 105, 115, 117, 118, 119, 121, 153, 240, 274 P *Photiou, A. 122 Pisetsky, D. S. 23, 24, 25, 27, 38, 39, 42, 44, 46, 58, 60, 61, 75, 77, 96, 97, 101, 103,

281

104, 117, 133, 134, 137, 147, 149, 152, 174, 186, 187, 189, 202, 204, 215, 217, 218, 240, 253, 254, 270, 271, 276 R Rediske, J. 26, 60, 61, 102, 120, 137, 146, 151, 217, 253 S Schaible, H.-G. 4, 22, 23, 24, 25, 26, 27, 38, 39, 40, 45, 46, 74, 75, 77, 118, 119, 120, 134, 145, 149, 150, 202, 213, 215, 216, 218, 219, 238, 271 Shen, H. 135 *Shenker, N. G. 154, 241 Simkin, P. A. 25, 41, 44, 62, 152, 179, 186, 187, 188, 189, 190, 255 V van den Berg, W. 148, 152, 203, 204 W Woodworth, T. G. 58 Y Yashpal, K. 137

Osteoarthritic Joint Pain: Novartis Foundation Symposium 260. Volume 260 Edited by Derek J. Chadwick and Jamie Goode Copyright ¶ Novartis Foundation 2004. ISBN: 0-470-86763-9

Subject index aspirin 226 asymmetric arthritis 254 ATP 224 atrophic OA 189^190 attention, bone cancer pain 232 auto-neuro-in£ammatory loop, survival hypothesis 249^250 avascular necrosis intraosseous hypertension 180 in OA 106, 117^118

A A ¢bres 224 Ab ¢bres 29, 224 acid detection 224, 228 acidosis, tumour-induced 228^229 activities of daily living, exercise 55 a¡ect, pain 66, 78, 263^266 African^Americans joint replacement surgery 265^267, 270 pain descriptions 74 alcoholism 180 allodynia 31 capsaicin 124^126 analgesic hip 218, 219 angiogenesis, COX-2 226^227 animal models bone cancer pain 222 joint in£ammation/OA 30^31, 76^77, 81, 95^96, 142 rheumatoid arthritis 100^104 see also canine ACL transection model; cat ACL transection model ANKTM1 206, 208 ankylosing spondylitis 180 anterior cingulate cortex pain perception 265 pain sensitivity marker 275 anterior cruciate ligament (ACL) absence 40 tears 198 anterior knee pain, intraosseous hypertension 180 antidepressants, OA pain 137 anti-epileptics, pain relief 137 anti-TNF, rheumatoid arthritis 103, 128, 133 anxiety 66, 68, 74 arthritic pain 5 arthrogenous re£ex inhibition 40^41, 54, 55 arthroscopic lavage 106 articular cartilage explants, loading 80^81 ASIC-3, pH 228

B balance, pain severity 70 barometric pressure 37^38, 69 bilateral receptive ¢elds 7 bilaterality 147, 148^149, 150^151, 171, 252, 253 see also symmetry bioinformatics 238^239 biomechanics 79 bisphosphonate, bone cancer pain 229, 232, 238 blood^brain barrier 150 Bole animal model 95^96 bone cancer pain 221^238 acidosis 228 animal models 222 attention 232 bisphosphonate 229, 232, 238 c-Fos 225 central sensitization 230^232 COX-2 inhibitors 226, 227 disease progression 232^233 endothelins 227^228 gabapentin 230 growth factors 229 higher brain centres 232 mood 232 osteoclasts 228^229 osteoprotegerin 229, 232, 238 perplexing 229 pH 228 282

SUBJECT INDEX

primary a¡erent sensory neurons 222, 224^225 proton release 228 sensory ¢bre distension and destruction 230 substance P 225 tissue-speci¢c mechanisms 229 bone deformation, weight bearing 189 bone marrow lesions 70^71 bone marrow oedema interosseous hypertension 180 MRI 195^197 bone pain 70, 106 botox model 93, 95, 96, 99 bradykinin articular a¡erent excitation 33 damaged sensory neurons 229 nerve growth factor 210 nociceptors 126, 224 prostaglandin augmentation 33 thermal opening of vanilloid receptor/ channel complex 34 C C ¢bres 29, 119, 207, 224 c-Fos 225 C reactive protein disease progression indicator 71 OA 204 synovitis 204 calcitonin 240 calcitonin gene-related peptide (CGRP) 244^245 dorsal root ganglion neurons 30 nerve growth factor 127, 128 primary a¡erent nerve ¢bres 30 spinal cord neurons 13 synovium 245 cancer-associated pain 221^222, 240 see also bone cancer pain canine ACL transection model 55^56 pain 62^63 capsaicin age 45 allodynia 124^126 contralateral in£uences 247 e⁄cacy in OA 120 hyperalgesia 46, 124^126 nerve growth factor 210 rheumatoid arthritis 124

283

T cells 245 capsular distension 106 carrageenan injections 8, 31 cartilage innervation 25 lesion pain 106 network failure 139 remodelling, substance P 146 smoothing 106 stimulation 4, 5 subchondral cell communication 187 cat ACL transection model 81, 82^83 EMG 87^88 femoral groove cartilage 98 gait 97^98 guarding 97 knee £exor muscles 96^97 mechanics 83^87 muscle forces 87^88 pressure patterns 88^90 catastrophizing 263, 270, 274 caudate nucleus, ¢bromyalgia 259 central pattern generators 54 central sensitization cancer pain 230^232 chronic in£ammation 9 ¢bromyalgia 260^261 glial cells 25 pain sensitivity 5 peripheral component 24 centrally acting analgesics 120 Charcot’s joints 40, 218, 247, 249 chemical stimulation 4, 141, 207, 224 chondrocyte injury 139 substance P 146^147 chondropathy, synovitis 204 cognitive in£uences 176, 263^266 cold sensation 208 collagenase, substance P 129 collateral ligament tears 198 comorbidity, exercise 55 compartment syndromes 183^184 complex regional pain syndrome 159^164 central and peripheral nervous system involvement 162, 164 cortical somatosensory map 160^161 incongruent movements 168^170 joint replacement 153 mirror visual feedback 164^168 pain 159

284

complex regional pain syndrome (cont.) referred sensations 160^162 symmetry 254^255 types 1 and 2 159 contralateral in£uences 246^247, 255 see also bilaterality cortical sensitization 123 cortical somatosensory map, CRPS 160^161 cortisol, ¢bromyalgia 275 COX 32, 33 COX-1 13, 33, 225 COX-2 angiogenesis 226^227 constitutively expressed at spinal level 27 expression 33 in£ammation 20 prostaglandin synthesis 13 tumour growth 226^227 up-regulation 33 COX-2 inhibitors 33 cancer growth and metastasis 227, 238 cancer pain 226, 227 COX-3 27 COX inhibitors cancer pain 226 centrally acting 26 mechanonociceptor sensitization reversal 32 CRMT gene 272 cruciate ligament tears 198 cyclooxygenase see COX headings cyclosporine, knee pain 182 cytokines blood^brain barrier 150 nociception 128 sickness response 149^150 D degeneration see joint degeneration demographics 1 depression exercise 55 healthcare seeking behaviour 67 pain severity 68 dermatological conditions, symmetry 242 diabetic neuropathy, neuropathic joint disease 56 diacerein 2 diagnosis of OA, minimum set of symptoms 74^75

SUBJECT INDEX

di¡use noxious inhibitory controls (DNIC) 7, 107^108, 115, 120^121 dipyrone 120 disability 50 disbaric decompression 180 dissensory state 171, 172 distension-induced pain 183 diurnal variation, pain 68^69 dorsal horn abnormal pain sensitivity 260^261 mechanical stimulation 5 PGE2 20 receptive ¢elds 144 spinal cord neurons 5 dorsal root ganglion neurons, CGRP 30 dorsal root re£exes 20 drug development 2 E early intervention 140^141 educational level, knee pain 66 e¡usions arthrogenous re£ex inhibition 40 joint damage 33 MRI 193^195, 201 ultrasound 201 elbow OA, symptoms 98 EMG, cat ACL transection model 87^88 emotion, pain 78 endothelins cancer pain 227^228 nociceptors 224 endurance, exercise 55 epidermal growth factor 225 ethnicity pain questionnaires 74 pain sensitivity 265^267, 270 excitatory amino acids 12 exercise adherence 52^53 intraosseous pressure 182 knee OA 52^53, 54^55 OA symptoms 60^61 F FAST trial 55 fatigue joint chemicals entering circulation 141 pain 273

SUBJECT INDEX

fatty acid-induced in£ammation 184 femuropatellar joint ¢brillation 106 fentanyl 259 ¢brofatty ankylosis, joint immobilization 51 ¢bromyalgia central correlates of pain sensitivity 259^260 central sensitization 260^261 characteristics 258^259 cortisol 275 incongruent movements 168^170 OA pain compared 274 phantom swelling 159 placebo response 272 pressure pain thresholds 111 psychiatric morbidity 259 sleep disorders 273 substance P-like activity 112 thermal stimuli 259 visual analogue scales 270^271 wind-up response 259 ¢brous cartilage noxious stimulation 4 £are^wheal response 141, 244 forbidden pattern 245 fractals 253^254, 255 free fatty acids, in£ammation 184 Freund’s complete adjuvant 8, 30, 31 Fuji pressure sensitive ¢lm 88 G G protein-coupled receptor 214 gabapentin, pain relief 137, 230, 240 gait cat ACL transection model 97^98 diggers and gliders 59 pain 76 quadriceps weakness 53 Gaucher’s disease 180 genetic factors OA 49^50, 140 pain response 43 polymorphisms 272 post-injury OA 152 GFRa 209 glial cell line-derived neurotrophic factor (GDNF) 209 glial cells 25, 260^261 glomerulonephritis, symmetry 243 glutamate 12 Golgi endings 29

285

gout, bilaterality 151^152 grey matter 5 grip strength 60 ground reaction forces 82, 83, 85^86 growth factors, tumour cells 229 guarding 97 H hand grip strength 60 hand pain, diurnal variation 68^69 hand osteoarthritis grip strength 60 prevalence 1 symmetry 253 health status, knee pain 66 healthcare seeking behaviour 66^68, 75 heat pain threshold 207^208 in£ammation 209^210 see also thermal sensitivity heavy lifting 62 Heberden’s nodes, symmetry 243, 253 heel strike 53, 59^60 hip osteoarthritis abnormal pain sensitivity 261 hyperalgesia 110 osteotomy 106 pain distribution 107 pressure pain thresholds 110 prevalence 1 symptoms 107 hip pain, location 68 hip replacement cement use 118 ethnicity 266^267, 270 pain relief 107, 118 technique 106 hyperalgesia capsaicin 46, 124^126 cytokine-induced 128 hip OA 110 joint in£ammation 5, 11 nerve growth factor 209 neuronal mechanism 9 secondary 9 hypertrophic pulmonary osteoarthropathy 240 hypochondriasis, knee pain 66

286

I Ih current 215^216 ibuprofen 226 incongruent movements 168^170, 177^178 indomethacin 32 in£ammation acute 126 acute/chronic clari¢cation 215 free fatty acids 184 neurogenic 128^131, 141 nociception 126^128, 209^211, 217 opioids 218 pain and 71 see also joint in£ammation in£ammatory mediators 32^33, 126^128 injury-related OA, genetic factors 152 interleukin 1 (IL1) 129, 133^134, 225 inhibition 2 interleukin 1b (IL1b) 127, 128, 209 interleukin 6 (IL6) 128, 129, 225 interleukin 10 (IL10) 134 interoception 123, 133 intraosseous hypertension see intraosseous pressure intraosseous in£ammation 184 intraosseous pain, ischaemia 183^184 intraosseous pressure 179^184 avascular necrosis 180 exercise 182 immobilization 50^52, 61^62 impact events 183 ischaemia 183 joint use 182^183 marrow oedema 180 muscle contraction 182 normal 181 OA 179^180 osteoporosis 187 pain and 41 regulation 181^182 ischaemia, intraosseous pain 183^184 isolectin B4 positive neurons 30 J joint degeneration early detection 87 joint instability 83, 85 joint unloading 87 muscle activation patterns 88

SUBJECT INDEX

muscle weakness 93, 95 joint e¡usion see e¡usions joint immobilization 50^52, 61^62 joint in£ammation articular a¡erents 31 mechanical stimulation 5 pain at rest 5 pain sensitivity 5 silent nociceptors 29 spinal cord neuron hyperexcitability 8^11, 12, 13, 20 joint innervation 28^29 joint instability joint degeneration 83, 85 nociception 40 onset and progression of OA 85 joint loading biological response 90^93 OA association 80 joint mechanics 33, 79^95 joint pain 4^5, 105^115 joint pressure distribution 82 e¡usate removal 33 joint protection 32 joint replacement ethnicity and pain sensitivity 266^267, 270 pain alleviation 44 recovery 46 remaining pain after 152^153 see also hip replacement; knee replacement joint stability see joint instability joint swelling, pain 106 joint unloading 87 K kaolin injections 8, 31 ketamine 12, 259 kinematics 82 knee compartments 106^107 £exor muscles 96^97 radiographically normal 193 knee osteoarthritis abnormal pain sensitivity 261^262 a¡ective in£uences 263^266 arthroscopic lavage/cartilage smoothing 106 cognitive in£uences 263^266 exercise 52^53, 54^55

SUBJECT INDEX

muscle atrophy 52 muscle strength 52 osteotomy 107 pain location 106^107 periarticular pain 45 predicted from knee pain 70 prevalence 1 proprioception 57, 171 quadriceps 52^53, 55 radiographic evidence 1 strength-training, protective role 54 surgery 107 symmetry of pain pattern 24 symptoms 106^107 tenderness 106^107 knee pain bone marrow lesions 70^71 community 65^66 cyclosporine 182 diurnal variation 68^69 interosseous hypertension 180 local anaesthetic 69, 197 location 68, 69 predicting knee OA 70 prevalence 1 psychosocial disability 68 risk factors 66 worsening over time 70 X-rays 65, 66, 68, 75 knee replacement ethnicity 266^267, 270 patellar prosthesis 106 total knee replacement 107 L lateral femurotibial compartment 106 Lequesne Index 68 leukocyte migration, substance P 131 lifting 62 ligaments noxious stimulation 4 tears, MRI 198 light touch perception thresholds 110 lipid metabolites 224 local anaesthetic knee pain 69, 197 pain response 44 low spirits, knee pain 66

287

M magnetic resonance imaging 191^201 bone marrow oedema 195^197 e¡usions 193^195, 201 ligaments 198 menisci 197^198 osteophytes 198, 201 periarticular lesions 197 popliteal cysts 193^195 synovitis 193^195, 202 marrow oedema intraosseous hypertension 180 MRI 195^197 mast cell degranulation 129 McGill Pain Questionnaire 68, 74, 264^265 MCP-1 131 mechanical stimulation articular a¡erents 29 ¢bromyalgia 259 pain 4, 5 painless 207 spinal cord neurons 5^8 mechanonociceptor sensitization 32^33 medial femurotibial compartment 106, 107 medicolegal practice 176 meniscal tears, MRI 197^198 menopause 240 metabolic abnormalities 50 microtrauma 55 mirror visual feedback 164^168 mood, bone cancer pain 232 motor re£exes, sensitization 22^23 multiple sclerosis, symmetry 242^243 muscle, arthrogenous inhibition 40^41, 54, 55 muscle activation patterns 82, 83, 88 muscle atrophy joint immobilization 51 knee OA 52 muscle contraction, intraosseous pressure 182 muscle forces 82, 87^88 muscle mass, obesity 53 muscle weakness, risk factor for joint degeneration 93, 95 N Nav1.8 216 negative a¡ect, knee pain 66 nerve cuts 145^146 nerve ¢bre types, joint physiology 141

288

nerve growth factor damaged sensory neurons 229 nociception 126, 127^128, 209^210, 224 nervous system in arthritis 246 neural plasticity 12, 122^123, 140, 160, 224^225 neurogenic in£ammation 128^131, 141 neurogenic pain 106 neuroin£ammation, contralateral in£uences 246^247 neurokinin-1 (NK-1) 13, 129 neurokinin-2 (NK-2) 13, 129 neurokinin A 150 neurogenic in£ammation 129 spinal cord neurons 13 neuromuscular aspects 49^58 neuropathic arthropathy 55^56 neuropathic pain gabapentin 230 in OA 151, 152^153 pain perception 260 neurotransmitters, disease initiation and progression 141 neurotrophic factors 26 night pain 69, 177 dissensory state 172 intraosseous pressure 180 nitric oxide 260 NK-1 antagonists, antidepressant activity 137 NMDA antagonists 12 side-e¡ects 47 NMDA receptors 12 nociception assessing 123^126 de¢ned 122^123 in£ammatory mediators 32^33, 126^128 joint instability 40 nociceptive-speci¢c neurons 7, 8 nociceptors in£ammation 126^128, 209^211, 217 location 28^29 oedema 217 peripheral sensitization 225 sensitivity 32^33, 206^207 silent 29, 45^46, 225 TRP channels 207^209 tumour excitation 225^228 non-steroidal anti-in£ammatory drugs (NSAIDs) 218^219 mechanonociceptor sensitization reversal 32

SUBJECT INDEX

paracetamol and 47 responders 46^47 synovitis 203 noxious movement, rate of movement 39 noxious stimuli 4, 5, 217, 224 O obesity exercise 55 muscle mass 53 occupational factors 62 oedema, nociceptor in£ammation 217 ominory state 171 opioids 218 osteoarthritis (OA) ageing process 139 atrophic 189^190 cartilage network failure 139 cause 139 chondrocyte injury 139 clinical entity 50 contributing factors 139^140 de¢nition 74 diagnosis 74^75 disability 50 ¢bromyalgia pain compared 274 genetic abnormality 49^50, 140 Heberden node symmetry 243, 253 multifactorial cause 139 organ disease 49 prevalence 1 primary 49 progression 71, 85 secondary 49, 253 signs 2 sociocultural phenomenon 76 source of pain 43^45, 70^71, 106, 141 spectrum of a single end-stage disorder 101 symmetry 243 symptoms 2 when does it start? 37 osteoclasts bone cancer 228^229 menopause 240 normal bone 239 OA 240 osteophytes contralateral 147 MRI 198, 201

SUBJECT INDEX

osteoporosis intraosseous pressure 187 joint immobilization 51 OA and 186^187 transient regional 240 osteoprotegerin, bone cancer pain 229, 232, 238 osteotomy hip OA 106 knee OA 107 rest pain 180 over-use model 147 P p38 MAPK 128 p75 209, 210 Paccinian endings 29 pain acute leading to chronic 140 assessment 141 catastrophizing 263, 270, 274 distribution, behavioural symptom 113 early intervention 140^141 early markers 141 e¡ect of 70 ethnicity 265^267, 270 family of disorders 140 ¢fth vital sign 270 genetic variation 43 importance to patients 67^68 memory 24 nature 68^70 neural plasticity 122^123 patterns 69 process 140 protective function 76, 218 psychological factors 68, 78 quantifying 137 questionnaires 68 referred 174^175 severity assessment 68 sleep disorders 272 source in OA 43^45, 70^71, 106, 141 subsetting 75 surrogate markers 141 treating mechanisms 140 typical 68 variability 45, 68 vascular hypothesis 117 widespread body 276

289

X-rays 68 Pain Catastrophizing Scale (PCS) 263 pancreatic arthritis 184 paracetamol mechanism of action 27 NSAIDs and 47 patella cartilage degeneration 106 prosthesis 106 pathology 77 peptidergic nociceptors 30, 209 periarticular lesions, MRI 197 periarticular pain, knee OA 45 peripheral a¡erent nociceptive ¢bres (PANs) 71 peripheral neurogenic pain 106 peripheral sensitization 5, 123, 225 PGE1, nerve ¢bre sensitization 32 PGE2 bradykinin augmentation 33 dorsal and ventral horn 20 nerve ¢bre sensitization 32 substance P 129 PGI2 bradykinin augmentation 33 IP receptors 26^27 nerve ¢bre sensitization 32 pH, cancer pain 228 phantom limb pain 140, 153, 155^156 phantom swelling 156^158, 159 phospholipase C (PLC) hydrolysis 210 nociception 126 PIP2, nociception 126, 214 placebo response, ¢bromyalgia 272 plasticity 12, 122^123, 140, 160, 224^225 platelet-derived growth factor (PDGF) 225 PN3 128 polymorphisms 272 popliteal cysts, MRI 193^195 post-rehabilitation arthropathy 52 postural stability, pain reduction 70 PPT-A gene 129 pressure algometry 109 pressure pain thresholds 109^110, 111 pressure patterns 88^90 pressure sensation 5 pressure sensitive ¢lm 88 primary a¡erent nerve ¢bres 28 bone cancer pain 222, 224^225 classi¢cation 29

290

primary a¡erent nerve ¢bres (cont.) endings 29 group II 29, 31 group III 29, 31 group IV 29, 31 joint in£ammation 31 locations 28^29 mechanical responses 29 neuropeptide phenotype 30 progression of OA C reactive protein 71 joint instability 85 proprioception 40, 56^57 age 57 exercise 55 improvement 57 knee OA 57, 171 nomenclature 175 pain reduction 70 physical function 57 prostaglandins 26^27 bradykinin augmentation 33 nociception 32^33, 126, 224, 225^226 spinal cord neurons 13, 20, 126^127 tumour cells 226 prostate cancer, endothelins 227^228 protein kinase C epsilon (PKCe) 210 proteoglycan, joint immobilization 51, 52 protons 214, 224 tumour-induced release 228^229 psoriasis, symmetry 242 psychiatric morbidity, ¢bromyalgia 259 psychological status exercise 55 healthcare seeking behaviour 67 knee pain 66 pain 68, 78 psychosocial disability, knee pain 68 pulmonary ¢brosis, symmetry 243 Q quadriceps knee OA 52^53, 55 re£ex inhibition 54 qualitative research 75, 78 quantitative sensory testing (QST) 109, 124 R rat model 142^144 receptive ¢elds 5, 7, 8, 9, 23^24, 144

SUBJECT INDEX

referred pain 174^175 referred sensations 155^156, 160^162 re£ex inhibition 40^41, 54, 55 rehabilitation, early strategies 87 repetitive strain injury 177^178 rest pain 5, 177, 180 ret 209 rheumatoid arthritis animal models 100^104 anti-TNF 103, 128, 133 bony erosions 156 capsaicin 124 contralateral sensory changes 247 intraosseous hypertension 180 nervous system 246 pain 76, 156 phantom swelling 156^158 sti¡ness 158 stroke 146 substance P 244 symmetry 243^244 synovium 156, 178, 203 RT97+ve antibody 30 Ru⁄ni endings 29 S salicylic acid 32 scratch response, £are and wheal 141, 244 seasonal variation 69 secondary OA 49, 253 sensory neurons 28^36 peripheral injury 229 peripheral tissue health 141 tumour-induced changes 230 serotonin 33 sickle cell disease 180 signs of OA 2 silent nociceptors 29, 45^46, 225 sleep disorders 273 SNS 128, 214 sociocultural phenomenon 76 socioeconomic status, catastrophizing 274 spinal cord neurons 4^22 activation thresholds 5, 7 bilateral receptive ¢elds 7 CGRP 13 descending inhibition 7^8, 9 excitatory amino acids 12 glutamate 12 heterotopic inhibition 7^8, 9

SUBJECT INDEX

in£ammation-evoked hyperexcitability 8^11, 12, 13, 20 mechanical stimulation of joint 5^8 neurokinin A 13 neuropeptides 13 nociceptive-speci¢c neurons 7, 8 plasticity 12 projections 7 prostaglandins 13, 20, 26^27 receptive ¢elds 5, 7, 8, 9, 23^24 substance P 13 synaptic activation 12^20 wide-dynamic-range neurons 7, 8 spinal pain 105 spinal sensitization 123 spondyloepiphyseal dysplasia 50 squatting 62 steroid therapy intra-articular 69, 203 intraosseous cells 180 synovitis 203 sti¡ness night 177 rheumatoid arthritis 158 strength training 52^53, 54 stroke, rheumatoid arthritis 146 structure^pain associations 2, 191^201 subchondral bone pain 106 subchondral cells, cartilage cell communication 187 substance P antagonists 47 arthrogenic potential 129, 131 cartilage remodelling 146 chondrocytes 146^147 in£ammatory arthritis 245^246 local level 135 marker role 118^119 nerve growth factor 128 neurogenic in£ammation 129, 131 nociceptors 225 peripheral release 135 rheumatoid arthritis 244 spinal cord neurons 13 T cells 245 substance P-like activity, ¢bromyalgia 112 survival hypothesis 249^250 symmetry clinical disease 241^243 CRPS 254^255 degenerative arthritis 244

291

hand OA 253 in£ammatory arthritis 243^244 knee OA pain 24 neurogenic role 129 sympathetic uveitis 148, 242, 254 symptoms of OA 2 exercise 60^61 minimum set to diagnose OA 74^75 structure and 2, 191^201 synovial tissue CGRP 245 rheumatoid arthritis 156, 178, 203 stimulation 5 vascular physiology symmetry 255 synoviocytes, substance P 129 synovitis C reactive protein correlation 204 chondropathy 204 intra-articular steroids 203 MRI 193^195, 202 NSAIDs 203 prominence in human disease 203^204 T T cells, neuropeptides 245 tachykinins 129 temperature thresholds 109, 110, 207^208 thalamus ¢bromyalgia 259 pain sensation 7 thermal sensitivity 4, 34, 109, 110, 207^208, 209^210, 224, 259 thumb base OA 68 tidemark 188 touch perception 110 trabecular tension, pain 183 transforming growth factor 225 transforming growth factor b 209 TrkA 209, 210 TRP channels 206^213 TRPM8 206, 208 TRPV1 206 acid detection 224, 228 expression pattern 219 heat detection 34, 207^208, 210, 224 lipid metabolite detection 224 mechanosensation 38, 216 nerve growth factor 127 nociception 126 pH 228

292

acid detection (cont.) proton detection 214, 224 skin 134^135 TRPV2 206, 207, 208 TRPV3 206, 207, 208, 215 TRPV4 206, 207, 208, 215 TTX-resistant Na+ channels 213^214 tumour necrosis factor (TNF) damaged sensory neurons 229 inhibition in rheumatoid arthritis 103, 128, 133 substance P stimulation 129 tumour necrosis factor a hyperalgesia 128 nerve growth factor 127, 209 tumours growth factors 229 nociceptor excitation 225^228 proton release and acidosis 228^229 sensory ¢bres 230 U ultrasound, e¡usions 201 urate injections 8

SUBJECT INDEX

vasoactive intestinal peptide 245 ventral horn PGE2 20 spinal cord neurons 5 vertebral collapse 240 visual analogue scales 136^137, 270^271 visual feedback, phantom and referred sensations 157, 159, 164^168 von Frey ¢laments 109 VR-1 see TRPV1 W weather prediction 37, 69 weekend pain 69 weight bearing, bone deformation 189 weight loss, exercise 55 Whole-Organ Magnetic Resonance Imaging Score (WORMS) 193 wide-dynamic-range neurons 7, 8 widespread body pain 276 WOMAC 68, 267 WORMS score 193

V vascular endothelial cell adhesion molecules 129

X X-rays, knee pain 65, 66, 68, 75