Surgery of the third ventricle

SURGERY OF THE THIRD VENTRICLE EDITED BY Michael L. J. Apuzzo, M.D. Professor Department of Neurological Surgery University of Southern California School of Medicine Los Angeles, California

WILLIAMS & WILKINS Baltimore • London • Los Angeles • Sydney

Foreword

The third ventricle lying at the center of the cerebral hemispheres surrounded by brain, exquisite both in terms of complexity and function, has enticed neurosurgeons for as long as the depths of brain have entered their ken. The ventricle and its neural surround also have interested neuroanatomists, neurophysiologists, endocrinologists, psychiatrists, and psychologists attempting to decipher the anatomical and functional relationships of the area. These functions are as basic as reproduction and emotion and extend to maturation and differentiation of the entire organism. The relationships of this area to cerebral function and differentiation of the organism have resisted study with an almost overwhelming tenacity, in part because of the complexities involved, in part because of the anatomical position of the neural neighborhood, and in part because the techniques used for the studies have, in themselves, altered the functional milieu being investigated. The background of almost unlimited questions, anatomical position, and altered function when the area has been approached, has also caused problems in diagnosing and treating lesions of the third ventricle, as well as being an attraction to basic and clinical neuroscientists. Studies, both basic and clinical, have, however, slowly assembled a matrix of information that allows, at this time, the formation of concepts and hypotheses that can be evaluated and supported or disproved. Not the least of the supporting structures for building the matrix have been changes in diagnostic imaging, surgical techniques, anesthesia, and control of vital functions of the body during surgery that have allowed the clinician both to treat and study problems in the area. As a clinician, I am more aware of the changes in clinical approaches to malfunction of areas adjacent to the third ventricle, particularly neoplasms adjacent to and within the ventricle and how the information gained from this has influenced the study of this central neural area. This manuscript fills a void that has been present in assembling what is known about this area; what experimental approaches are available, what they have revealed, and how modern diagnostic imaging, neuroan-atomical techniques, neurosurgical techniques, endocrine studies and psychological methods have contributed to an understanding of the functional relationships of the neural. The sum of this information is greater than its parts and cannot be appreciated without this collation. When I started in medical school, or even at the time when I finished neurosurgical training, a volume such as this would have been short, inaccurate, and mainly of value in demonstrating what damage was done with surgical approaches, what pathology was found, and perhaps suggesting that other methods of treatment of conditions in this area must evolve. The manuscripts constituting this volume, in contrast, not only summarize what has been accomplished, but also rather than pointing to an area that on the basis of previous work appears fruitful or should be studied, defines those areas and those directions. Lesions in this area are not common in the average physician's practice, are relatively uncommon even in a neurosurgeon's practice, but they are important. Their diagnosis and treatment not only adds to the well-being of the individual suffering from such conditions but also allows investigative clinicians, treating such patients, to add to the general knowledge of the nervous system and neurosurgery. This book defines the bases for such investigations and, in addition, offers the neurosurgeon an up-to-date evaluation and summary of the multiple therapeutic options for treating lesions in and around the third ventricle. William F. Collins, Jr. New Haven, Connecticut

Preface It has been more than 50 years since the publication of Walter Dandy's monograph, Benign Tumors in the Third Ventricle of the Brain, an important work detailing his concepts of pathology and surgery and his experience with the management of tumors affecting the third ventricular chamber. Since that time progress in aspects of surgical techniques, radiological imaging, scientific methodologies related to neurological anesthesia, and comprehension of neurophysiological parameters has developed at a striking pace. As no text or atlas has singularly addressed the topic of third ventricular surgery for over a half century, it seems that a work (text/atlas) focusing on issues attendant to this challenging region is in order. In the development of this project an effort has been made to provide commentary on pertinent topics from recognized authoritative neurologists, neurosurgeons, neuroradiologists, and neuroscientists with special interest in the topic and to develop a broad and practical perspective. The book is particularly addressed to the general neurological surgeon who encounters lesions involving this region infrequently. The work has been arranged in two major portions. The first provides an intellectual substrate related to issues of anatomy, embryology, physiology, pathology, and radiology that is the essential basis for substantive approaches to issues of therapy. The second section focuses upon surgical technique and strategy with a detailed account of the methodology attendant to the multiple surgical corridors applicable for access to the region. Matters associated with intraoperative anatomy, techniques, and decision-making are approached. Issues of preoperative and intraoperative strategies are discussed and developed in detail. To expand perspective, innovative and challenging methodologies are also presented. Considering the number of contributors, each with a wealth of experience in dealing with specific surgical issues, no effort has been made to provide a rigidly unified surgical attitude, and certain elements of controversy related to the individual prejudices of the authors are presented within the work. However, effort has been expended to provide the logic involved in the development of each perspective. Finally, commentary is offered on the role of and expectations for adjuvant therapy (radiotherapy and chemotherapy) for neoplastic lesions in the region. The primary objective of this text/atlas is to provide neurosurgeons, neurologists, and neuroscientists with a single reference that will offer current, substantive, and practical guidance in the comprehension and management of lesions affecting the third ventricular chamber. Michael L. J. Apuzzo Los Angeles

Acknowledgments

Special gratitude must be expressed to Carolyn Soter, who provided a major force in bringing this project to completion. Over a 3-year period she gave attentive and tireless energy to all aspects of the general organization, solicitation of manuscripts, and physical preparation of this work. Janice Jones energetically attended to the complex task of bibliography retrieval and the procurement of resource materials, invaluably contributing to the substance and credibility of the manuscripts. The Department of Photography at the Los Angeles County-University of Southern California Medical Center provided careful attention to the reproduction and generation of many of the illustrations in this volume. In particular Andy Gero, R.B.P., Chief Medical Photographer, Frank Park, Michael Gail, and Tom Meichelbock should be recognized for their patience and care in producing illustrations of the highest quality. The Williams & Wilkins staff has been a flexible, valuable, and understanding partner during the development of this complex work. In particular, Carol-Lynn Brown must be recognized for her guidance, patience, prudence, and support during the evolution of the scope of the volume. Finally, it is most important to recognize our teachers, patients, and predecessors in neurosurgery who have provided the events, stimulation, and experiences generating the intellectual substrate that made this work possible.

Contributors

Diane Abeloff, M.A., A.M.I. Medical Illustrator, Baltimore, Maryland Michael L. J. Apuzzo, M.D. Professor, Department of Neurological Surgery, University of Southern California School of Medicine, Los Angeles, California Erik-Olof Backlund, M.D. Professor and Chairman, Department of Neurological Surgery, University of Bergen, School of Medicine, Bergen, Norway Gilbert A. Block, M.D. Instructor, Department of Neurology, Cornell University Hospital, New York, New York Joseph E. Bogen, M.D. Clinical Professor of Neurological Surgery, University of Southern California School of Medicine, and Adjunct Professor, Department of Psychology, University of California at Los Angeles, Los Angeles, California Adam Borit, M.D. Professor of Pathology, Department of Pathology, The University of Texas System Cancer Center, Houston, Texas Robert E. Breeze. M.D. Clinical Instructor, Department of Neurological Surgery, University of Southern California School of Medicine, Los Angeles, California Parakrama T. Chandrasoma, M.D. Assistant Professor, Department of Pathology, University of Southern California School of Medicine, Los Angeles, California Ivan S. Ciric, M.D. Professor of Clinical Surgery (Neurosurgery), Northwestern University Medical School, Chicago, Illinois

W. Kemp Clark, M.D. Professor, Department of Neurological Surgery, The University of Texas Southwestern Medical School, Dallas, Texas William F. Collins, Jr., M.D. Harvey and Kate Cushing Professor, Division of Neurological Surgery, and Chairman, Department of Surgery, Yale University School of Medicine, New Haven, Connecticut Eric R. Cosman, Ph.D. Professor, Department of Physics, Massachusetts Institute of Technology, Cambridge, Massachusetts Antonio R. Damasio, M.D., Ph.D. Professor and Chief, Division of Behavioral Neurology, The University of Iowa Hospitals and Clinics, Iowa City, Iowa Richard L. Davis, M.D. Professor of Pathology, Department of Pathology, University of California Medical Center, San Francisco, California Michael S. B. Edwards, M.D. Associate Professor, Department of Neurosurgery, University of California Medical Center, San Francisco, California Bruce Ehni, M.D. Clinical Instructor, University of Texas Medical School, Houston, Texas George Ehni, M.D. Professor, Neurological Surgery, University of Texas Medical School, Houston, Texas Craig A. Fredericks, M.D. Clinical Instructor, Department of Neurological Surgery, University of Southern California School of Medicine, Los Angeles, California

Henry D. Garretson, M.D., Ph.D. Professor and Director, Division of Neurosurgery, University of Louisville School of Medicine, Louisville, Kentucky Steven L. Giannotta, M.D. Associate Professor, Department of Neurological Surgery, University of Southern California School of Medicine, Los Angeles, California Philip H. Gutin, M.D. Associate Professor, Department of Neurosurgery, University of California Medical Center, San Francisco, California M. Peter Heilbrun, M.D. Professor and Chairman, Division of Neurosurgery, University of Utah Medical Center, Salt Lake City, Utah Harold J. Hoffman, M.D., F.R.C.S.(C) Professor, Division of Neurological Surgery, University of Toronto School of Medicine, Toronto, Ontario, Canada Kazuhiro Hongo, M.D. Clinical and Research Fellow, Department of Neurosurgery, Shinshu University, School of Medicine, Matsumoto, Japan Patrick J. Kelly, M.D. Associate Professor, Department of Neurosurgery, Mayo School of Medicine, Rochester, Minnesota E. Leon Kier, M.D. Professor and Chief, Division of Neuroradiology, Yale University School of Medicine, New Haven, Connecticut Alexander N. Konovalov, M.D. Professor and Director, N. N. Burdenko Neurosurgical Institute, Moscow, U.S.S.R. Eddie Kwan, M.D. Assistant Professor, Department of Radiology, Tufts University School of Medicine, Boston, Massachusetts Claude Lapras, M.D. Professeur a la Faculte, Neurochirurgien des Hopitaux, Lyon, France Michael H. Lavyne, M.D. Associate Professor of Surgery, Division of Neurological Surgery, Cornell University Medical Center, New York, New York Victor A. Levin, M.D. Professor, Department of Neurosurgery, University of California Medical Center, San Francisco, California F. Miles Little, M.D. Assistant Professor, Department of Neurosurgery, University of Southern California School of Medicine, Los Angeles, California

Masao Matsutani, M.D. Assistant Professor, Department of Neurosurgery, Tokyo University, Tokyo, Japan J. Gordon McComb, M.D. Professor, Department of Neurological Surgery, University of Southern California School of Medicine, Los Angeles, California Michael T. Modic, M.D. Professor, Department of Radiology, University Hospital, Case Western Reserve School of Medicine, Cleveland, Ohio Robert B. Page, M.D. Professor, Division of Neurosurgery, Department of Anatomy, Hershey Medical Center, Hershey, Pennsylvania J. D. Patet, M.D. Neurochirurgien des Hopitaux, Lyon, France Russel H. Patterson, Jr., M.D. Professor and Chairman, Department of Neurological Surgery, Cornell University Medical Center, New York, New York Jerome B. Posner, M.D. Professor and Chairman, Department of Neurology, Cornell University Hospital, New York, New York Albert L. Rhoton, Jr., M.D. R. D. Keene Family Professor and Chairman, Department of Neurosurgery, University of Florida College of Medicine, Gainesville, Florida Peter Roth Scientific Artist, Neurosurgical Department, University Hospital, Zurich, Switzerland Madjid Samii, Prof. Dr. med. Neurochirugische Klinik, Krankaenhaus Nor-stadt, Hannover, West Germany Robert A. Sanford, M.D. Associate Professor, Department of Neurosurgery, University of Tennessee, Memphis, Tennessee Keiji Sano, M.D., D.M.Sc., F.A.C.S.(hon.) Professor and Director, Department of Neurosurgery, Teikyo University Hospital, Tokyo, Japan Henry H. Schmidek, M.D. Professor of Neurosurgery, Department of Neu-robiology, Harvard Medical School, Boston, Massachusetts William Shucart, M.D. Professor and Chairman, Department of Neurosurgery, New England Medical Center, Boston, Massachusetts Robert R. Smith, M.D. Professor and Chairman, Department of Neurological Surgery, University of Mississippi Medical Center, Jackson, Mississippi

Steven P. Smith, M.D. Fellow, Neuroradiology, New England Medical Center, Boston, Massachusetts Bennett M. Stein, M.D. Byron Stookey Professor of Neurosurgery, Neurological Institute, New York, New York Kenichiro Sugita, M.D. Professor and Chairman, Department of Neurosurgery, Shinshu University, Matsumoto, Japan Jiro Suzuki, M.D. Professor and Chairman, Department of Neurosurgery, Tohoku University, Sendai, Japan Kintomo Takakura, M.D. Professor and Chairman, Department of Neurosurgery, Tokyo University, Tokyo, Japan Peter J. Teddy, D.Phil, F.R.C.S. Consultant Neurosurgeon, Department of Neurological Surgery, Oxforshire Area Health Authority, The Radcliffe Infirmary, Oxford, England George T. Tindall, M.D. Professor of Surgery, Chief, Division of Neurosurgery, The Emory University School of Medicine, Atlanta, Georgia Suzie C. Tindall, M.D. Assistant Professor of Surgery, Division of Neurosurgery, The Emory University School of Medicine, Atlanta, Georgia Gary W. Van Hoesen, Ph.D. Professor, Anatomy and Neurology, Department of Anatomy, The University of Iowa Hospitals and Clinics, Iowa City, Iowa Roger I. von Hanwehr, M.D. Clinical Instructor, Department of Neurological Surgery, University of Southern California School of Medicine, Los Angeles, California

Steven L. Wald, M.D. Assistant Professor of Neurosurgery, Division of Neurosurgery, University of Vermont College of Medicine, Burlington, Vermont William M. Wara, M.D. Professor and Vice Chairman, Department of Radiation Oncology, University of California, San Francisco, San Francisco, California Martin H. Weiss, M.D. Professor and Chairman, Department of Neurological Surgery, University of Southern California School of Medicine, Los Angeles, California Trent H. Wells, Jr. President, Trentwells, Inc., South Gate, California Robert H. Wilkins, M.D. Professor and Chief, Division of Neurological Surgery, Duke University Medical Center, Durham, North Carolina Samuel M. Wolpert, M.B., B.Ch. Professor of Radiology and Neurology, Tufts University School of Medicine, and Chief, Neuroradiology, New England Medical Center, Boston, Massachusetts M. Gazi Yasargil, Prof. Dr. med. Director der Neurochirurg, Universitatsklinik Zurich, Zurich, Switzerland Chi-Shing Zee, M.D. Assistant Professor, Department of Radiology, University of Southern California School of Medicine, Los Angeles, California Vladimir Zelman, M.D., Ph.D. Associate Professor, Department of Anesthesiol-ogy and Department of Neurosurgery, University of Southern California School of Medicine, Los Angeles, California

Contents

обложка

Foreword .................................................................................................................. Preface ................................................................................................................... Acknowledgments ................................................................................................. Contributors .............................................................................................................

vii ix xi xiii

I. HISTORICAL PERSPECTIVE _____________________________________ 1 . History of Surgery of the Third Ventricular Region. Robert H. Wilkins, M.D .......................................................................................................

3

II. FUNDAMENTAL CONSIDERATIONS ______________________________ 2. Comparative Anatomy of the Third Ventricular Region. E. Leon Kier, M.D....................................................................................................

37

3. Microsurgical Anatomy of the Third Ventricular Region. Albert L. Rhoton, Jr., M.D .......................................................................................

92

4. Surgery of the Third Ventricle: Regional Embryology. Ivan S. Ciric, M.D .................................................................................................... 167 5. Physiological Consequences of Complete or Partial Commissural Section. Joseph E. Bogen, M.D

175

6. Pathological Correlates of Amnesia and the Anatomical Basis of Memory. Antonio R. Damasio, M.D., Ph.D., and Gary W. Van Hoesen, Ph.D

195

Commentary A. Memory in Man: A Neurosurgeon's Perspective. Henry D. Garretson, M.D., Ph.D

209

7. Anatomy and Physiology of Consciousness: Syndromes of Altered Consciousness Related to Third Ventricular Surgery. Gilbert A. Block, M.D., and Jerome B. Posner, M.D

213

8. Deep Veins. Robert R. Smith, M.D., Robert A. Sanford, M.D., and Henry H. Schmidek, M.D

224

9. Pathological Lesions of the Third Ventricle and Adjacent Structures. Richard L. Davis, M.D

235

10. Tumor Markers in Third Ventricular Neoplasms. Henry H. Schmidek, M.D., Adam Borit, M.D., and Steven L. Wald, M.D. . . 253 1 1 . Radiology of Third Ventricular Lesions. Eddie Kwan, M.D., Samuel M. Wolpert, M.B., B.Ch., Steven P. Smith, M.D., and Michael T. Modic, M.D

262

III. SURGICAL APPROACHES, TECHNIQUES, AND STRATEGIES ANTERIOR APPROACHES 12. Anterior Transcallosal and Transcortical Approaches. William Shucart, M.D

303

13. Considerations in Transforaminal Entry. George Ehni, M.D., and Bruce Ehni, M.D

326

14. Transcallosal Interforniceal Approach. Michael L. J. Apuzzo, M.D., and Steven L. Giannotta, M.D

354

15. Subchoroidal Trans-Velum Interpositum Approach. Michael H. Lavyne, M.D., and Russel H. Patterson, Jr., M.D. . . . 381 16. Subfrontal Transsphenoidal and Trans-Lamina Terminalis Approaches. Russel H. Patterson, Jr., M.D

398

17. Bifrontal Anterior Interhemispheric Approach. Jiro Suzuki, M.D

413

18. Pterional Approach. George T. Tindall, M.D., and Suzie C. Tindall, M.D

440

19. Combined Approaches. M. Gazi Yasargil, Prof. Dr. med., Peter J. Teddy, D. Phil., F.R.C.S., and Peter Roth

462

20. Transnasal Transsphenoidal Approach. Martin H. Weiss, M.D

476

21. Anterior and Mid-Third Ventricular Lesions: A Surgical Overview. Michael L. J. Apuzzo, M.D., Chi-Shing Zee, M.D., and Robert E. Breeze, M.D

495

Commentary B. Technique and Strategies of Direct Surgical Management of Craniopharyngioma. Alexander N. Konovalov, M.D

542

Commentary C. Diencephalic Ventricular Surgery. Robert B. Page, M.D

Structures

at

Risk

in

Third 553

POSTERIOR APPROACHES 22. Posterior Transcortical Approach. Kenichiro Sugita, M.D., and Kazuhiro Hongo, M.D

557

23. Infratentorial Supracerebellar Approach. Bennett M. Stein, M.D

570

24. Occipital Transtentorial Approach. W. Kemp Clark, M.D. 591 25. Posterior Intrahemispheric Retrocallosal and Transcallosal Approaches. J. Gordon McComb, M.D., and Michael L. J. Apuzzo, M.D

611

Commentary D. Operative Management of Malformations of the Vein of Galen. J. Gordon McComb, M.D., and Michael L. J. Apuzzo, M.D

641

26. Controversies, Techniques, and Strategies for Pineal Tumor Surgery. Claude Lapras, M.D., and J. D. Patet, M.D. . . . 649 27. Pineal Region and Posterior Third Ventricular Tumors: A Surgical Overview. Keiji Sano, M.D., D.M.Sc., F.A.C.S.(hon.) . . . 663 SPECIALIZED ISSUES 28. Technical Aspects of Excision of Giant Basal Tumors with Third Ventricular Involvement. Madjid Samii, Prof. Dr. med. . . 684 29. Cerebrospinal Fluid Diversion. J. Gordon McComb, M.D., and F. Miles Little, M.D

699

30. Considerations and Techniques in the Pediatric Age Group. Harold J. Hoffman, M.D., F.R.C.S.(C)

727

31. Applications of Computerized Tomographic Guidance Stereotaxis. Michael L. J. Apuzzo, M.D., Parakrama T. Chandrasoma, M.D., Vladimir Zelman, M.D., Roger I. von Hanwehr, M.D., and Craig A. Fredericks, M.D

751

Commentary E. Contemporary European Contributions to Neurosurgical Stereotaxy. Roger I. von Hanwehr, M.D

793

Commentary F. Role of Stereotaxis in the Management of Midline Cerebral Lesions. Erik-Olof Backlund, M.D., Ph.D

802

Commentary G. Magnetic Resonance Stereotaxy. Eric R. Cosman, Ph.D., M. Peter Heibrun, M.D., and Trent H. Wells, Jr

806

32. Computer-assisted Stereotaxic Laser Microsurgery. Patrick J. Kelly, M.D

811

IV. ADJUVANT THERAPIES 33. Radiotherapy of Pineal and Suprasellar Tumors. William M. Wara, M.D., and Philip H. Gutin, M.D

831

34. Chemotherapy of Tumors of the Third Ventricular Region. Michael S. B. Edwards, M.D., and Victor A. Levin, M.D

838

35. Therapeutic Modality Selection in Management of Germ Cell Tumors. Kintomo Takakura, M.D., and Masao Matsutani, M.D.

843

1 History of Surgery of the Third Ventricular Region

Robert H.Wilkins,M.D.

This chapter deals with the history of surgery of the third ventricular region. I have arbitrarily limited my considerations to the surgical treatment of abnormalities of or in the third ventricle itself, of lesions of the roof of the third ventricle, and of tumors in the region of the pineal gland. I have not included the treatment of sellar or suprasellar lesions (even those that invaginate the floor of the third ventricle), of arteriovenous malformations draining into the vein of Galen, or of tumors of the thalamus, hypothalamus, or brain stem. I have tried to avoid undue emphasis on the establishment of priority because some uncertainty always exists about who was the first to describe a pathological entity or an approach to treatment. As Palmer has noted, "... an author can never tell how many cases . . . have previously been reported: no amount of library research will permit more than a rough guess" (106). The reason for this is that even after a thorough search in a medical library, the author will have missed unindexed cases reported in sources such as textbooks and monographs, governmental reports, doctoral theses, and the letters to the editor section of many journals. Furthermore, "... the author also must understand that, as important as the entity seems to him, others who have dealt with similar cases [may not] . . . have bothered to write them up" (106). With that caveat in mind, I focus attention in this chapter on the contributions of Walter Dandy, who had a unique interest in third ven-

tricular lesions and an unprecedented experience in dealing with them (36, 55, 62). I begin each section of this chapter with an account of his work and then consider the other historical developments, most of which came after Dandy's pioneering efforts. Cerebrospinal Fluid and the Ventricular System; Hydrocephalus During his senior year and then after his graduation from medical school in 1910, Walter Dandy worked in the Hunterian Laboratory of Experimental Medicine of Johns Hopkins University, where he studied the blood supply and nerve supply of the pituitary gland (24, 46). Subsequently, in association with Kenneth Black-fan, he performed a series of animal experiments and human investigations that clarified where cerebrospinal fluid (CSF) is produced, how it circulates, where it is absorbed, and what types of hydrocephalus result from the various abnormalities that alter its production, circulation, and absorption (43-45). Then during his last year of residency (1918) and first year of practice (1919), Dandy introduced ventriculography and pneu-moencephalography (26, 28), thus permitting visualization of the third ventricle for diagnostic purposes. Based on these early investigations, Dandy maintained for the rest of his career an interest in hydrocephalus and its treatment and in the diagnosis and treatment of tumors and other

lesions in and around the third ventricle (36, 55, 62, 158). He introduced the operations of choroid plexectomy (27, 39), cannulation of the aqueduct of Sylvius (29), and third ventriculostomy (33, 41) to treat hydrocephalus and ventriculoscopy (32) for inspection of the ventricular system and as an aid to choroid plexectomy. In his 1918 paper on extirpation of the choroid plexus of the lateral ventricles for treatment of communicating hydrocephalus, Dandy wrote, "A remarkable exposure is obtained during the operation in the ventricle. One can see the third and opposite lateral ventricle and the septum lucidum which is frequently perforated in many places owing to pressure atrophy" (27). In a report in 1922 on ventriculoscopy, he stated, On two occasions, it has seemed advisable to inspect a lateral ventricle. This was done in one instance through a small cystoscope and in a second an attempt was made with the help of a small operating ventriculoscope to remove and fulgurate the

choroid plexus. ... It was possible to see practically the entire extent of the lateral ventricle, the foramen of Monro, the septum lucidum with numerous perforations in it, and the entire extent of the choroid plexus. . . (32). The procedure of excision of the choroid plexus had a high mortality rate and was abandoned early (59, 113, 131, 132). However, endoscopic cauterization of the choroid plexus, which apparently had also been tried by Lespinasse in 1910 (95), was developed and used by Dandy (39) (Fig. 1.1), Putnam (115), Scarff (129, 132), Feld (57, 58), and others for 30 to 40 years, when the combination of the equivocal results of this procedure and the introduction of successful valved shunt systems led to its abandonment as well (113). Likewise, direct cannulation of the aqueduct of Sylvius was tried by Dandy (29) (Fig. 1.2), and subsequently by Leksell (94), Norlen (104), Elvidge (53), and a few other neurosurgeons (145, 148), but the procedure did not prove to be of value.

Figure 1.1. Composite drawing. Center. Sites at which the choroid plexus is cauterized and removed. A and a. The glomus of the choroid plexus on each side. В and b. The choroid plexus in the posterior cranial fossa. С, с, and c'. The choroid plexus from the body of each lateral ventricle. (From Dandy WE: The operative treatment of communicating hydrocephalus. Ann Surg 108:194-202, 1938.)

Figure 1.2. Sagittal view of the brain showing a shunt tube in position in the aqueduct of Sylvius. (From Dandy WE: The diagnosis and treatment of hydro-cephalus resulting from strictures of the aqueduct of Sylvius. Surg Gynecol Obstet 31:340-358, 1920.) In contrast, neurosurgeons have maintained some interest in third ventriculostomy over the years because of its promise of relieving hydro-cephalus without the insertion of a shunt system (59, 113, 131). In 1922, Dandy wrote,

space where the absorption is slightly, if at all, greater than in the ventricles. Moreover, the opening is through cerebral tissue which proliferates and closes it, unless a good deal of the brain has been destroyed. We have employed this method 6 times. No claim is made for its success. Time alone will decide . . . (33).

Strictures of the aqueduct of Sylvius recur after any attempt to restore the lumen. For this reason, if treatment is to be successful, the fluid must be sidetracked into its normal channels. With this in mind, a procedure which apparently is anatomically correct has been devised, to supersede any direct attack on the aqueduct. This consists in removing the floor of the third ventricle. A small opening is made in the skull and dura in the frontal region, the frontal lobe is elevated until the bulging third ventricle is well exposed. . . . This opening in the floor of the third ventricle affords an exit from the dilated ventricles, so that the fluid can now pass directly into the cisterna chiasmatis and interpeduncu-laris—the normal distributing centers for cerebro-spinal fluid.. . . The ventricular wall is a very thin membrane and offers a minimum of glia tissue to repair the defect. This procedure is by no means analogous to making an opening in the roof of the third ventricle. The latter can have no beneficial result because the fluid escapes into the subdural

When the pathogenesis of obstructive hydrocephalus was fully appreciated, operations to reroute the ventricular fluid were devised. Perhaps, the forerunner of these was the "Balkenstich" operation of Anton and von Bramann [ ], which consisted of passing a brain cannula through a midline frontal trephine opening, along the falx through the corpus callosum and into the ... ventricle. It was presumed that the ventricular fluid would pass through the fenestrated corpus callosum to the cal-losal subarachnoid pathways, but the procedure lost its popularity when it was found that the passageway did not remain open (59). As mentioned, Dandy thought that ventriculostomy through the floor of the third ventricle held advantages over ventriculostomy through its roof. His initial subfrontal approach to third ventriculostomy involved the sectioning of one optic

nerve to gain exposure (33), but in 1932 he reported on a lateral (transtemporal) third ven-triculostomy that did not require division of the optic nerve (36). The steps in this procedure, as listed by Dandy, were: 1 . A plaster cast was molded to the infant's head (Fig. 1.3). A defect was made in the tem poral region of the side to be operated upon. 2. Small curved skin and muscle incisions were made in the temporal region (Fig. 1.3). 3. A small area of bone was removed above the base of the skull and a dural flap was reflected toward the base. 4. The head was lowered about 50° and the temporal horn of the lateral ventricle was tapped, with the evacuation of 60 to 80 ml of CSF. A short flanged ventricular needle was left in place during the operation. 5. The temporal lobe was retracted to expose the lateral wall of the interpeduncular cistern and this wall was opened behind the oculomotor nerve or between the carotid artery and the ocu lomotor nerve (Fig. 1.4). 6. The bulging floor of the third ventricle was then identified and opened, with the opening preferably being made just posterior to the hypophyseal stalk (Fig. 1.5). 7. The ventricles were then filled with the pre viously aspirated CSF or with an isotonic solution and the wound was closed in layers (Fig. 1.5). In 1945, Dandy reported 92 patients that he had treated in this fashion, with a 12% mortality rate and with arrest of the hydrocephalus in 50% of the patients for periods averaging more than 7 years (Table 1.1) (131). Reoperation had been necessary in 7 of the 92 patients. After Dandy's 1922 report of subfrontal third ventriculostomy, several surgeons pursued the idea of venting the third ventricle into an adjacent subarachnoid cistern. The following year, Mixter reported treating an infant with obstructive hydrocephalus by approaching the floor of the third ventricle with a urethroscope inserted through the lateral ventricle and foramen of Monro and then using a flexible sound to puncture an opening into the interpeduncular cistern (102). Scarff subsequently performed a similar procedure, which he reported in 1936 (129). In 1936, Stookey and Scarff described a subfrontal approach with puncturing of the lamina terminalis and then the floor of the third ventricle (130, 141) (Fig. 1.6). In 1963, Scarff gave the results in 527 hydrocephalic patients treated by 12 surgical groups by puncture of the lamina terminalis alone or combined with puncture of the floor of the third ventricle, and/or by

puncture of the lateral wall of the hypothalamus (131). The patients were followed for an average of 5 years. The operative mortality was approximately 15% and the initial success rate was about 70%. Various surgeons have pursued the operation of anterior, inferior, lateral, or superoposterior third ventriculostomy by direct or indirect techniques, as detailed in the excellent publications of Scarff (131) and of Pudenz (113). The procedure has been aided by the development of stereotaxic techniques and positive contrast ventric-ulography (Fig. 1.7) (74, 80, 128). In 1980, Hoffman et al. summarized the results of third ventriculostomy by the open technique that had been performed by 13 surgical groups (80). There were 569 patients; the operative mortality was 10.3% and the success rate was 53.6% (Table 1.2). These authors also showed the results of third ventriculostomy by the percutaneous technique as performed by 12 groups of surgeons. The success rate was approximately the same (53%), but the operative mortality was lower (3.5%) (Table 1.3). During the period when the procedures of cauterization of the choroid plexus and of third ventriculostomy were being tried, the techniques of ventriculoscopy and ventriculography were also being developed. Ventriculoscopy was of some help in these operations initially, but despite clever technical advances such as flexibility, fiberoptic lighting, and video attachments, ventriculoscopy never has become an important diagnostic or therapeutic tool (32, 36, 39, 56-58, 64, 65, 75, 80, 95, 102, 115, 129, 132, 144, 155, 156). In comparison, after their introduction in 1918 (26) and 1919 (28), pneumoventriculography and pneumoencephalography rapidly became the keys to the diagnosis of third ventricular lesions and subsequently to the performance of stereo-taxic operations adjacent to the third ventricle. This has changed only in recent years because of the advent of computerized tomographic (CT) scanning and magnetic resonance imaging. For the contrast media, Dandy used ordinary room air. Bingel ... suggested the use of carbon dioxide in 1922, because it was absorbed rapidly. In the same year, Jungling . . . recommended the use of oxygen. . . . Although gaseous media demonstrated the lateral ventricles clearly, the third and fourth were often not well visualized even with special positioning techniques. To obtain better contrast various other media were tried. After the introduction of lipiodol, . . . Jacobaeus and Nord ... in 1924 experimented with it for ventriculography. The following year Schuster . . . used it. About the same

Figure 1.3. Dandy's operative approach for third ventriculostomy. (From Dandy WE: The brain. In Lewis D (ed): Practice of Surgery. Hagerstown, MD, WF Prior, 1932, pp 1-682.)

Figure 1.4. The bulging third ventricle exposed through the lateral wall of the interpeduncular cistern between the carotid artery and the oculomotor nerve. (From Dandy WE: The brain. In Lewis D (ed): Practice of Surgery. Hagerstown, MD, WF Prior, 1932, pp 1-682.)

a From Scarff JE: Treatment of hydrocephalus: An historical and critical review of methods and results. J Neurol Neurosurg Psychiatry 26:1-26, 1963.

time Balado, Morea and Donovan ... in the Argentine tested the oil in the ventricles of experimental animals and finding no untoward reactions began using it routinely, particularly for demonstration of the third and fourth ventricles (157).

Figure 1.5. The upper composite drawing indicates all of the steps in Dandy's third ventriculostomy. The lower drawing shows the method of filling the ventricles before closure. (From Dandy WE: Diagnosis and treatment of strictures of the aqueduct of Sylvius [causing hydrocephalus]. Arch Sarg 51:1-14, 1945.)

The use of lipiodol waned, and trials of other positive contrast agents were performed (124, 157). These included thorium dioxide, diiodoty-rosine-gelatin, and abrodil. Then, when ethyl iodophenylundecylate was introduced as a mye-lographic agent, it found use in ventriculography as well. More recent attempts to use water-soluble contrast media for ventriculography did not meet with much success until the introduction of metrizamide, and this has been combined with CT scanning to give excellent visualization of the cisterns and ventricles (49).

Table 1.2. Summary of Experiences with Third Ventriculostomy by Open Operationa Operative Success Mortality Rate (%) (%) Wertheimer & Mansuy 1938 3 33 66 Pennybacker 1940 5 20 80 White & Michelsen 1942 11 18 46 Dandy 1945 92 17 39 Tolosa 1948 26 25 50 Guillaume & Mazars 1950 230 3.4 54 Krayenbuhl et al. 1950 17 17 60 Fasiani 1951 72 20 70 Scarff 1951 34 15 56 Morello & Migliavacca 1959 28 3 75 Volkel & Voris 1966 12 0 33 Patterson & Bergland 1968 29 6 34 Brocklehurst 1974 10 0 70 Summary 569 10.3 53.6 a From Hoffman HJ, Harwood-Nash D, Gilday DL: Percutaneous third ventriculostomy in the management of noncommunicating hydrocephalus. Neurosargery 7:313-321, 1980. Author

Year

No. Cases

Table 1.3. Summary of Experiences with Third Ventriculostomy by Percutaneous Techniquea Author McNickle Forjaz et al. Perlman Raimondi Guiot Plerre-Kahn et al. Poblete & Zamboni Hoffman Sayers & Kosnick des Plantes & Crezee Vries Hoffman et al. Summary

Year 1947 1968 1968 1972 1973 1975 1975 1976 1976 1978 1978 1980

No. Cases 7 15 1 3 14 44 10 11 46 61 5 22 228

Operative Mortality (%) 0 20 0 0 0 7 0 0 4 0 0 0 3.5

Success Rate (%) 71 67 100 0 64 64 70 27b 89 15 20 45 53

a From Hoffman HJ, Harwood-Nash D, Gilday DL: Percutaneous third ventriculostomy in the management of noncommunicating hydrocephalus. Neurosurgery 7:313-321, 1980. b Originally reported as 64%; subsequent obstruction largely due to shunt placement has reduced this success rate to 27%.

Figure 1.6. Third ventriculostomy by the subfrontal route, with puncture of both the lamina termlnalis and the floor of the third ventricle. (From Scarff JE: Treatment of obstructive hydrocephalus by puncture of the lamina terminalis and floor of the third ventricle. J Neurosurg 8:204-213, 1951.)

contents of the fifth and sixth ventricles, Dandy then presented two patients with symptomatic cysts of these structures (Fig. 1.8) that he had treated by a transcallosal exposure and fenestra-tion of the cyst into the lateral ventricles (Fig. 1.9). He ended his paper as follows: 1 . Two cases of cysts of the cavum septi pellucidi and cavum vergae are reported. In each case the two cavities were continuous. 2. The cysts acted as tumors and caused compression of the motor tracts on both sides. Men tal symptoms were decided in both cases. One pa tient had peculiar epileptic attacks.... Suggestive evidence of intermittent intracranial pressure ex isted in both instances. . . . 4. The diagnosis is easily made by ventriculography. . . . 5. An operation is offered for cysts of this char acter (35).

Figure 1.7. Diagrammatic sketch of a percutaneous method of third ventriculostomy. (From Hoffman HJ, HarwoodNash D, Gilday DL: Percutaneous third ventriculostomy in the management of noncommunicat-ing hydrocephalus. Neurosurgery 7:313-321, 1980.) Lesions of the Roof of the Third Ventricle In 1931, Walter Dandy published an unusual paper entitled, "Congenital Cerebral Cysts of the Cavum Septi Pellucidi (Fifth Ventricle) and Ca-vum Vergae (Sixth Ventricle): Diagnosis and Treatment" (35). He began the paper as follows: "In the midline of the brain and within the confines of the corpus callosum either or both of the cavum septi pellucidi and cavum vergae are not infrequently found. Neither cavity has excited much interest either anatomically or clinically. The two cases here reported are, I believe, the first instances in which a diagnosis of these spaces, dilated in abnormal degree, has been made during life, the first in which the lesion has been found at or treated by operation and the first in which clinical symptoms are shown to be related to the lesions (35). After discussing the anatomy, occurrence, and

In 1967, Kempe and Busch reported a symptomatic cyst of the cavum septum interpositum that had caused head enlargement during the first 4 months of life (90). This cyst was exposed through the corpus callosum and psalterium, and a communication was established with the cis-terna venae magnae Galeni and ambiens. The patient then developed normally over the follow-up period of 4 years. Symptomatic cysts of this sort have been very rare (152), as have neuroepithelial (colloid) cysts of the septum pellucidum (18) and diencephalic cysts that extend up from the third ventricle through the corpus callosum (11). Somewhat less rare are meningiomas arising from the velum interpositum and confined predominantly to the third ventricle and pineal region (without dural attachment) (111, 123). Among 16 cases of this nature reviewed by Rozario et al. in 1979, 8 were located in the pineal/posterior third ventricular region, whereas only 3 were located in the anterior third ventricle (123). Various surgical approaches were used for the removal of these tumors. Most were operated upon via a transcorti-cal or occipital approach before the advent of the operating microscope. In 50%, total removal was achieved. The operative mortality was 37% (before 1970 it was 60% and after 1970 it was 0%).

Figure 1.8. Position of a congenital cyst of the cavum septi pellucidi and cavum vergae in relationship to the ventricular system. (From Dandy WE: Congenital cerebral cysts of the cavum septi pellucidi [fifth ventricle] and cavum vergae [sixth ventricle): Diagnosis and treatment. Arch Neurol Psychiatry 25:44-66, 1931.)

Figure 1.9. Dandy's method of producing an opening between each lateral ventricle and the cyst of the cavum septi pellucidi. (From Dandy WE: Congenital cerebral cysts of the cavum septi pellucidi [fifth ventricle] and cavum vergae [sixth ventricle]: Diagnosis and treatment. Arch Neurol Psychiatry 25:44-66, 1931.) Third Ventricular Tumors In 1922, Walter Dandy published the following brief account in the Johns Hopkins Hospital Bulletin: In one patient, a woman 24 years of age, the only symptoms were those referable to intracranial pressure. By cerebral pneumography, it was determined that each lateral ventricle was greatly enlarged, but there was no communication between them. Hence we concluded that there must be a tumor in the third ventricle and occluding each foramen of Monro; that the tumor must be small because neither ventricle was dislocated away from it. A large bone flap was turned down as in a pineal approach, and the corpus callosum split posteriorly for about 5 to 6 cm. The right lateral ventricle was then opened through the mesial wall at the septum lucidum. No tumor could be seen; the right foramen of Monro seemed normal on inspection from this lateral ventricle, but a probe would not pass through it, an obstruction being encountered in the third ventricle. The foramen of Monro was then widened and a small encapsulated, spherical tumor, about 1 cm. in diameter, easily shelled out in toto. The tumor was of ependymal origin. The patient recovered. Without cerebral

pneumography localization of the tumor and consequently its removal would have been impossible (31). Then in a monograph published 11 years later, Walter Dandy summarized his experience with the diagnosis and treatment of benign third ventricular tumors (37). He purposely omitted from this work a consideration of cerebral gliomas, pineal tumors, "hypophyseal duct tumors," and large pituitary tumors with upward extension into the third ventricle. The tumors in his 21 patients included 5 colloid cysts, 5 "ependymal gliomata," 1 choroid plexus papilloma, 1 epider-moid tumor, and 1 "dermoid." The other 8 tumors were more difficult to classify. In addition to his own 21 cases, Dandy also considered in his discussion "all cases of similar character that could be assembled from the literature" (37). These latter tumors were from autopsy material, "... there being no instance of a tumor having been disclosed at operation" (37). This last statement by Dandy is of interest be-

cause Harvey Cushing had removed a third ventricular meningioma from one patient in 1927 and from another in 1931 (23, 123) and had resected an astroblastoma of the third ventricle at the first meeting of the Harvey Cushing Society in 1932 (66). The operation for astroblastoma was memorable. At 10 a.m. on May 6, 1932, Dr. Cushing welcomed the members of the newly formed organization that had adopted his name. "He then operated in the large amphitheater before the entire group, exposing a thirdventricle tumor through a transcortical incision and removing a large part of it. [The patient, who was shown at the evening clinic the day of her operation turned out to have a most favorable tumor. . . . She was married a short time later and is still living and well, with a family of two children (Nov. 1945).]" (66). In Dandy's defense, Cushing did not publish the first two cases until 1938 (23), and Dandy was not a member of the Harvey Cushing Society. Dandy began his monograph as follows: A most interesting and important group of tumors lying within the third ventricle has, until the past few years, remained without the fields of diagnosis and treatment. Of benign type, slow growth and

Figure 1.10. Diagram to show how air injected into a dilated lateral ventricle fails to fill the third ventricle, which is occupied by tumor, and the opposite lateral ventricle (unless there are f enestrations in the septum pellucidum). (From Dandy WE: Benign Tumors in the Third Ventricle of the Brain: Diagnosis and Treatment. Springfield, IL, Charles С Thomas, 1933.)

small size, their removal by surgical treatment is relatively simple and affords a permanent cure if only a correct diagnosis and localization can be made. They produce fulminating signs and symptoms of intracranial pressure owing to their strategic position along the channels through which cerebro-spinal fluid must leave the brain, and yet localization of the tumors by neurological signs and symptoms is almost impossible. The advent of ventricu-lography has, however, by its mechanical evidence made the diagnosis and localization one of the greatest accuracy and certainty, and has, therefore, provided the precision that is an absolute prerequisite for their surgical attack (37). Dandy then went on to present his 21 cases individually and then to summarize the various aspects of these cases plus similar ones from the literature. He pointed out the characteristic ven-triculographic findings of third ventricular tumors (Fig. 1.10) and described various operative techniques for their removal (Figs. 1.11 to 1.15). Concerning these operative approaches, Dandy stated: Four operative attacks are used to expose and remove tumors of the third ventricle: (1) Posterior (pineal) approach . . . ; (2) Anterior (a) Frontal (hy-pophyseal) approach with removal of a round or oval area of the frontal lobe (usually right); (b) Frontal (hypophyseal) with transection of the frontal lobe (usually right); (c) Mid-sagittal approach through corpus callosum (used only for cysts of the septum pellucidum). .. . Since tumors within the third ventricle are quite

Figure 1.11. Sketches concerning a posterior trans-callosal operation for the removal of a colloid cyst of the third ventricle. (From Dandy WE: Benign Tumors in the Third Ventricle of the Brain: Diagnosis and Treatment. Springfield, IL, Charles С Thomas, 1933.)

Figure 1.12. Another posterior transcallosal operation for the removal of a colloid cyst, showing the method of incising the roof of the ventricle to expose the tumor. (From Dandy WE: Benign Tumors in the Third Ventricle of the Brain: Diagnosis and Treatment. Springfield, IL, Charles С Thomas, 1933.)

Figure 1.14. Transfrontal operation for the removal of a colloid cyst of the third ventricle. These drawings show the intraventricular exposure and removal of the cyst. (From Dandy WE: Benign Tumors in the Third Ventricle of the Brain: Diagnosis and Treatment. Springfield, IL, Charles С Thomas, 1933.)

Figure 1.13. Transfrontal operation for the removal of a colloid cyst of the third ventricle. These drawings show the approach. (From Dandy WE: Benign Tumors in the Third Ventricle of the Brain: Diagnosis and Treatment. Springfield, IL, Charles С Thomas, 1933.)

Figure 1.15. A posterior transcallosal operation for the removal of a posterior third ventricular tumor. (From Dandy WE: Benign Tumors in the Third Ventricle of the Brain: Diagnosis and Treatment. Springfield, IL, Charles С Thomas, 1933.)

small when compared with other intracranial tumors, it might seem that any approach that exposes the growth would be satisfactory. On the contrary, the determination of the approach is a matter of the utmost importance. The choice of methods is almost entirely dependent upon the ventriculographic information concerning the third ventricle. As previously noted, it is, therefore, necessary to fill with air and roentgenographically disclose, as far as possible, every part of the third ventricle that is not actually filled by tumor. The choice between either anterior and the posterior (pineal) approach is decided entirely upon the position of the tumor, i.e., whether it lies in the anterior or posterior part of the third ventricle. If the tumor occupies only the posterior part of the third ventricle the pineal approach is absolutely necessary. If the tumor occupies only the anterior part of the third ventricle the frontal (hypophyseal) approach is much superior, although the pineal approach is very satisfactory. The greatest diagnostic difficulty lies in separating the small tumors which occupy only the anterior part of the third ventricle from the larger growths that occupy the entire ventricle. Obviously the only instances in which this differential diagnosis can be made by ventriculog-raphy are afforded by tumors causing only a partial block and allowing a small amount of air to fill the small posterior remains of the third ventricle. . .. When the tumor is known to block the foramina of Monro and no information is obtainable, either by ventriculographic or neurological findings, of the posterior extent of the tumor, the choice of approach is difficult. Formerly I used the pineal approach exclusively, but more recently the frontal approach in increasing numbers, because of the possibility of encountering tumors that are not primary in the third ventricle but have grown into the ventricle from without. Then too the colloid cysts which usually give signs of complete obstruction of the foramina of Monro are much easier to remove by the anterior approach. Other things being equal, I prefer the frontal approach with oval resection or transverse section of the frontal lobe to the pineal approach. It is perhaps less dangerous and there is a much better exposure of the tumor area. The danger is less because the attack upon the tumor is at a distance from the veins of Galen. On the other hand, if the growth extends back to the great vein of Galen, the danger is greater by the frontal route, for the tumor must be blindly separated from the venous trunks. There is practically no risk in either type of section of the frontal lobe and no loss of function, mental or physical. One may ask why it is not possible to remove such tumors through a speculum introduced into the anterior horn of the lateral ventricle. Such a procedure is indeed possible with some of the simpler tumors, but should the operator encounter bleeding, he would be greatly handicapped for room and might well lose the patient through the effects of trauma and hemorrhage, when it is too late to provide the adequate exposure except through a heroic effort. There is little to lose and everything to gain in safety of life and function by providing an adequate exposure through the preliminary removal of cerebral tissue. There is, of course, always the

thought in mind that epilepsy may follow in the wake of the cerebral defects created for the purpose of exposing the tumor. As yet no patients so treated have had convulsions following the operation, if they were absent before. It is not always certain that the tumor is confined entirely to the third ventricle. For example, in Case 8, Group II, the tumor had passed through the foramen of Monro and the extension in the lateral ventricle was much larger than the tumor in the third ventricle. For the removal of such a growth, the pineal approach would be very embarrassing, perhaps impossible. In this instance the filling defect in the lateral ventricle left no doubt of the exact location of the tumor and of the character of the operative approach. For the anterior approach either the right or left side may be chosen. Although there is little difference in the choice, the right side is chosen in right-handed individuals, and the left side in those who are lefthanded, unless there are reasons to believe that the tumor may have extended to one side as in the above case. If the tumor is suspected or known to have extended to one side, the operative approach is made upon that side. . . . The size of the lateral ventricle is another very important factor in determining the site of the operative approach. In all of our cases there have been very large lateral ventricles, such as would be expected with tumors obstructing either the foramina of Monro or the aqueduct of Sylvius. Large ventricles are absolutely essential to a posterior (pineal) approach because room for the operative attack is provided only by releasing fluid from one or both lateral ventricles. It is conceivable that in the early stages of primary tumors of the third ventricle the lateral ventricles might not yet be sufficiently enlarged to permit of this approach, but in the present state of our diagnostic ability it is most unlikely that such an early diagnosis would be made. But in the event that a tumor should be diagnosed when the ventricles are small, the anterior approach with resection of the frontal lobe would afford ample room. The larger tumors in the region of the third ventricle and filling it secondarily usually show small lateral ventricles and, therefore, practically preclude the posterior approach, except perhaps to explore a tumor along the falx. At first glance the statement that in the larger tumors the lateral ventricles are smaller may appear paradoxical. Although they cause the same ventricular obstruction and should, therefore, produce the same degree of hydrocephalus, the ventricles are actually smaller because the bulk of the tumor encroaches upon and fills or obliterates both lateral ventricles. The primary tumors of the third ventricle are usually so small that their actual volume is of little or no importance in the creation of intracranial pressure, which is due solely to the obstruction to the channels of exit of the cerebrospinal fluid. Small ventricles with a tumor of the third ventricle, therefore, strongly suggest a large tumor and one arising presumably outside the third ventricle. Moreover, such tumors-usually gliomata—are much more common than the primary tumors in the third ventricle. In the treatment of such large growths the anterior approach, with cerebral resection, is necessary (37).

Dandy ended his monograph with the following statement: "The mortality from the entire series of extirpations, covering a period of eleven years, has been 33.3 per cent. However, in the past two and one-half years fourteen of these tumors have been removed with a mortality of 14.3 per cent. Except for occasional minor effects, these patients make almost perfect recovery of all functions (37). After the publication of Dandy's book, the direct operative removal of colloid cysts of the third ventricle became the standard approach to the treatment of these lesions (3) until the development of CT scanning and of stereotaxic and ven-triculoscopic techniques offered other possibilities for management (depending upon symptoms and ventricular size), such as simply following the patient with serial CT scans, shunting the lateral ventricle(s), shunting the cyst, aspirating the cyst percutaneously, or aspirating or removing the cyst endoscopically (10, 76, 112, 121). Direct operative excision also became the preferred treatment of other types of benign tumors of the third ventricle such as epidermoid and dermoid cysts, meningiomas (23, 111, 123), symptomatic xanthogranulomas (68, 125), cho-roid plexus papillomas (70), intraventricular craniopharyngiomas (69, 91, 96, 97, 99, 107, 137, 143, 149), and the subependymal giant cell astrocytomas of tuberous sclerosis (12, 20, 86). However, the removal of such tumors sometimes proved difficult, especially when they were located primarily in the posterior portion of the third ventricle or were adherent to the hypothal-amus. It was found that operative injury to the hypothalamus might result in diabetes insipidus. (Incidentally, another condition studied by Walter Dandy during his career was diabetes insipidus (40, 117)). Problems of this sort were especially true before the development of microsurgery and prompted Torkildsen in 1948 to publish a paper entitled, "Should Extirpation be Attempted in Cases of Neoplasm in or near the Third Ventricle of the Brain? Experiences with a Palliative Method" (146). Torkildsen concluded as follows: Attempts at the removal of neoplasms in regions of the pineal gland and the 3rd ventricle of the brain are associated with such a grave mortality rate that, if possible, such operations should be avoided. It is possible to treat patients with such lesions by ventriculocisternostomy, with satisfactory results. I have performed the operation in 8 cases of tumor in the pineal region, with 1 postoperative death, and in 11 cases of tumor in or near the 3rd ventricle outside the pineal region, with 2 fatalities. The above results show that the mortality rate by

this operation is far below that following attempts at surgical extirpation of the tumors. The fate of the patients depends primarily on the malignancy of the neoplasm. Tumors of high expanding energy infiltrating this region of the brain will in any case cause the death of the patient within limited time. In such cases no other operation can possibly save the life of the patient. Most of the tumors, however, are of very slow growth, and the patients may live comfortably with the neoplasm if the symptoms due to the obstruction of the flow of the cerebrospinal fluid can be relieved. In my series there are several cases in which no visible growth of the neoplasm has taken place after intervals of 7 or 8 years. In any case of tumor in the pineal region, neoplasm in or near the 3rd ventricle, craniopharyn-gioma, and stenosis of the Sylvian aqueduct whether due to neoplasm or not, ventriculocisternostomy should be carried out before any other surgical procedures are undertaken (146). Most neurosurgeons at that time took a similar approach to third ventricular masses, namely, diagnosis by air study and (unless a colloid cyst, choroid plexus papilloma, epidermoid or dermoid cyst, meningioma, craniopharyngioma, or subependymal giant cell astrocytoma was suspected) treatment by radiation therapy from an external source. Significant hydrocephalus was treated by ventriculocisternostomy. Yet without a tissue diagnosis the neurosurgeon took the chance that the lesion was benign and the radiation therapy inappropriate. With the advent of microneurosurgery, the pendulum then swung toward open operative exposure of the tumor for diagnosis and, frequently, resection. Stein stated, "In our experience with 46 patients encompassing all types of third ventricular tumors, the surgical mortality has been under 5% and approximately 30% of the tumors have been benign, encapsulated, and resectable" (139). But with such a direct approach the surgeon did add to the patient's immediate morbidity and mortality risks and took the chance of opening up a tumor that would be more appropriately treated with radiotherapy or chemotherapy. Fortunately, the subsequent development of CT-guided stereotaxic biopsy techniques (5, 87, 109, 118), the discovery of tumor markers (126, 134), and the improvement of CSF cytodiagnosis (9) have provided the neurosurgeon the means of being relatively certain of the diagnosis before the decision is made about definitive treatment. In addition, if surgical resection is chosen, the neurosurgeon of today has more tools than ever before to accomplish this goal, such as micro-surgical instruments, the ultrasonic aspirator,

and various surgical lasers. Kelly and his colleagues are currently perfecting a CT-guided stereotaxic system that uses a computer to direct and monitor tumor removal by laser energy (88, 89). Cysticercosis of the Third Ventricle Cysticercosis can involve the central nervous system in various ways (54, 98, 100, 140). One of these is by the development of cysticercus cysts within the ventricular system. The larval forms pass through the choroid plexus, develop in the ventricles as cysts, and migrate caudally through the ventricular system. When these cysts are within the ventricles, they usually do not incite an inflammatory response, but when they reach the basilar cisterns they usually evoke a significant adhesive basilar arachnoiditis. Thus, over the years, emphasis has been placed on the surgical removal of such cysts while they are still within the ventricular system. The cysts are ordinarily diagnosed by positive contast ventriculography, now with CT imaging. They frequently are multiple, and the exact route of operative attack is determined by their locations and sizes. When such cysts have been discovered within the third ventricle, the same surgical approaches have been used as for third ventricular tumors. Ventriculoscopy and cyst removal have been described, but the latter has been reported to be associated with breakage of the cyst with intraventric-ular spillage, which can evoke ventricular inflammation. Exposure . . . permits gradual removal of the lesion into the aperture and complete delivery of the cysts... . The lesions seem to conform to the surgical egress route but may need to be assisted in their delivery by the use of hydraulic dissection, a mild Valsalva maneuver by the anesthesiologist, and gentle teasing with nonpenetrating instruments. When the cysts are approached surgically in the lateral and third ventricles, the use of a slender Pasteur pipette with gentle suction is reported to extract the lesions effectively. . . . Patients are best treated intraoperatively by corticosteroid medication. Under these circumstances, rupture has not excited adverse reactions. . . . Results in the management of exclusively intraventricular cysts are excellent, especially when these are isolated or solitary. The surgical management of the other forms appears to be primarily palliative (140). Pineal Tumors In 1915, Walter Dandy published a paper, "Extirpation of the Pineal Body," that dealt with pinealectomy in the dog (25). Dandy concluded: 1. Following the removal of the pineal I have observed no sexual precocity or indolence, no adi-

oposity or emaciation, no somatic or mental precocity or retardation. 2. Our experiments seem to have yielded nothing to sustain the view that the pineal gland has an active endocrine function of importance either in the very young or adult dogs. 3. The pineal is apparently not essential to life and seems to have no influence upon the animal's well being. [25] Of more importance to the subsequent development of the surgical exposure and resection of pineal tumors in humans, Dandy noted, in comparing his new method to his previous methods: To be of practical value the pineal body must be more easily reached and removed with greater certainty and less mortality. Consequently a new and simple method of attack has been evolved. Though more delicate and requiring more painstaking care, it can be done almost as easily as a canine hypoph-ysectomy. The new operation can be done in less than one hour. It differs from the preceding operation in that the pineal is reached from in front through the third ventricle rather than from behind. In this way the extensive bleeding consequent to liberation of the vein of Galen is obviated, sidetracked as it were, and the operation can be performed almost bloodlessly. This is accomplished by dividing the splenium of the corpus callosum in the midline for a distance of about 2 cm. from its posterior terminus. This exposes the transparent roof of the third ventricle which is distended by the contained cerebrospinal fluid. A large anemic area is visible in the midline of the roof of the ventricle, between the two small veins of Galen. This is perforated and the opening enlarged backward to the origin of the vena Galena magna by releasing the blades of the forceps. The entire third ventricle is thus brought in full view and the pineal body is readily seen under the origin of this vein, in the median quadrigeminal groove. The pineal body can easily be grasped in the jaws of the cupped biting forceps and completely removed (Figs. 1.16 and 1.17) (25). Six years later, in 1921, Dandy published an account entitled, "An Operation for the Removal of Pineal Tumors" (30), a subject to which he returned in 1929 (34) and again in 1936 (38). He began his 1921 paper in this way: "Tumors of the pineal body have rarely been diagnosed and substantiated. The total number of authenticated pineal tumors is less than one hundred and almost all have been accidental findings at necropsy" (30). In his thorough review of the history of the surgery of pineal lesions, Pendl (110) gives credit to Victor Horsley (81, 82) as being the first to attempt to remove a pineal tumor by direct surgical attack. Two such procedures (in 1905 and 1909), done through an infratentorial approach, were unsatisfactory, prompting Horsley in 1910

Figure 1.16. Canine pinealectomy. The upper drawing demonstrates the sple-nium divided, exposing the roof of the third ventricle with its veins. The lower drawing shows the method of perforation of the roof between the internal cerebral veins. (From Dandy WE: Extirpation of the pineal body. J Exp Med 22:237-247, 1915.) to recommend a supratentorial exposure with splitting of the tentorium (81). Others also reported operations for palliation of pineal tumors in 1908, 1910, and 1911 (110). * According to Pendl (110), Ludwig Pussep (114) in 1910 was probably the first to reach a pineal tumor by a direct surgical approach. Pussep exposed the cystic tumor of a 10year-old boy through an occipital approach by transecting the transverse sinus and splitting the tentorium. He was able to remove parts of the tumor, but the child died on the third postoperative day. In his neurosurgical textbook, Fedor Krause (92) "... suggested in 1911 an infratentorial and supracerebellar route for surgical exploration of the upper vermis and quadrigeminal region. In 1913 he presented a 10-year-old boy to the Medical Society in Berlin, from whom, 6 weeks prior, he had resected a 4 cm pineal tumor. This was achieved in the sitting position using an inf raten-torialsupracerebellar approach. ... As was pointed out by Zulch. . ., this was the first report

of a successful extirpation of a tumor from the pineal region" (110). Pendl further reports that Brunner unsuccessfully explored the posterior fossa of a patient with a quadrigeminal tumor in 1911 and then the following year decided to approach a pineal tumor in a cousin of the first patient by a transcallosal route. "Because of severe venous hemorrhage and difficulties in exposure, the tumor was not reached and the operation abandoned" (110). Other operations were proposed or attempted by various surgeons, but without success (110). Thus, the stage was set for Dandy's 1921 report of his method of posterior transcallosal exposure of a pineal tumor (Figs. 1.18 to 1.22): The operation has been performed on three patients. In the first instance a silent cerebellar tumor had secondarily involved the region of the pineal body and the corpora quadrigemina; after exposure of the tumor, no attempt was made to remove it, because of its infiltrating character. This case, how-

Figure 1.17. Canine pinealectomy. The upper drawing shows that, after the defect in the roof of the third ventricle is opened and widened, the pineal gland is well exposed. The lower drawing illustrates the way the pineal gland is grasped with the biting forceps, the jaws of which are also shown separately. (From Dandy WE: Extirpation of the pineal body. J Exp Med 22:237-247, 1915.) ever, showed that a good exposure of this region is possible. On two subsequent occasions, tumors of the pineal body have been completely removed. In one case an encapsulated tubercle of the pineal body was extirpated, the patient recovering only to die 8 months later, presumably of the effects of other tubercles of the brain. . . . The results of this case demonstrated not only the feasibility of the removal of tumors of the pineal body but also the absence (in this case at least) of any injurious mental or physical effects due to the operation. The extirpation of the second pineal tumor was very much more difficult. The tumor was much larger and since the vena Galena magna and both small veins of Galen passed directly through the tumor and could not be dissected from it, the removal of practically all of these veins was a prerequisite to enucleation of the tumor. . . . The patient lived 48 hours, dying presumably of the shock due to the magnitude of the operation. His vitality was doubtless impaired by a cerebellar operation which was performed 10 days previously and at which the

tumor was found; but it was entirely inaccessible for removal by this approach. . . . One must also consider whether the complete removal of all the main trunks of the intracranial venous system— the large vein of Galen and both small veins of Galen—would be compatible with life. I know of no previous instance in which these veins have been removed or even ligated. In dogs I have ligated the vena Galena magna without effect when the ligation is distally placed but when proximally placed a mild grade of hydrocephalus has resulted. . . . Whether the collateral venous system in the human brain could be developed to compensate for this tremendous loss can only be surmised; it does seem doubtful, but there was no alternative to the removal of these veins with the tumor. [30] In 1928, Otfrid Foerster published his experience with the exposure of the quadrigeminal plate and pineal area in three patients (60, 61). A large occipital craniotomy, close to the superior

Figure 1.18. Cross sectional diagram showing the various steps in the transcal-losal removal of a pineal tumor. (From Dandy WE: An operation for the removal of pineal tumors. Surg Gynecol Obstet 33:113-119, 1921.) longitudinal sinus and the right transverse sinus, exposed the occipital lobe; the dura was opened and reflected as well over the sagittal and transverse sinuses; the bridging veins were ligated and transected. The posterior horn of the lateral ventricle was punctured by a cannula with sufficient decompression to allow retraction of the occipital lobe from the falx and tentorium. The splenium corporis callosi was thus exposed. For better access to the quadrigeminal area, the tentorium along the sinus rectus could be cut and reflected laterally, and a few centimeters of the splenium split. If the tumor on the left was not sufficiently exposed, the falx could be incised to better visualize the opposite side. Foerster's first case, a 25-year-old man, was operated on in 1927 with a total removal of a glioma and subsequent full recovery. The second reported case, operated on initially in 1923 with only a partial evacuation of a cystic tumor, required a second approach. However, the well-exposed tumor could not be removed from dense adhesions to the midbrain and the patient died on the second postoperative day. In the third case, the clinical signs indicated a pineal tumor; the whole quadrigeminal area was exposed, but no tumor and also no pineal gland was found (110). In 1931, William Van Wagenen introduced an entirely different approach to pineal tumor sur-

gery (150). He had exposed the tumor in his patient through the dilated right lateral ventricle. The lateral cortical incision had extended superiorly and posteriorly from the posterior end of the superior temporal gyrus, and the ventricle had been entered at the atrium. The spongioblas-tic tumor had then been exposed through the thinned medial ventricular wall and had been almost completely removed. Harris and Cairns in 1932 reported an unsuccessful attempt to remove a pineal tumor through a suboccipital exposure in 1930 (77). Six months after the initial attempt, the tumor was successfully removed through a right parietooccipital transcallosal approach. In his 1932 monograph on intracranial tumors, Cushing wrote: "Personally I have never succeeded in exposing a pineal tumour sufficiently well to justify an attempt to remove it. Encouraging reports of such procedures, however, have been made by W. E. Dandy of Baltimore (1921), by Otrid Foers-ter of Breslau (1928), and I have personal knowledge of two highly successful operations per-

Figure 1.19. Exposure of the medial cerebral cortex before exposure of the corpus callosum. A ventricular needle has been inserted to drain CSF from the lateral ventricle. (From Dandy WE: An operation for the removal of pineal tumors. Surg Gynecol Obstet 33:113-119, 1921.) formed by my former assistants, Dr. W. P. Van Wagenen of Rochester, New York, and Mr. Hugh Cairns of London" (22). Pendl cites various sources to indicate that Ernest Sachs removed a large cystic tumor from the pineal region in 1926 and Max Peet removed a pineal tumor through a right parietooccipital transcallosal approach in 1927 (110). By 1936, Dandy was able to comment on 10 personal cases (Figs. 1.23 to 1.25) (38), but he was pessimistic: Although an operative approach to the pineal region was proposed by me in 1921 [30], it was not until a decade later (1931) that the first pineal tumor was successfully extirpated [37]. A disastrous toll of seven fatal issues during this long period seemed almost to indicate the futility of further efforts. Yet the same approach was used successfully in a number of cases of tumor of the third ventricle in which, though the lesion was in the same general region, less serious difficulties were offered. In addition to the aforementioned case, successful extirpation has since been performed in two others, thus indicating that the lesion is not entirely hopeless, although it remains one of the most dangerous of all intracra-nial growths. This seeming success, however, is still further tempered by the fact that in two of these

cases the tumor has since recurred, in one after two and one-half years and in the other after three months. Only the last of the three patients remains well, and the period after operation has been only four months. Since the growth in the first case was solid and encapsulated and was extirpated intact, I expected a permanent cure. However, the growth was teratomatous, containing, in addition to the pineal element, ciliated columnar cells of ependymal origin, cartilage and tissue resembling the salivary gland. The tumor in the second case was much larger; it was so soft and cellular and its capsule so thin that it could not be shelled out intact but had to be removed in fragments. The patient died at her home three months later, but permission for necropsy was not obtained. It is, perhaps, more conceivable that cicatrization at the mouth of the aqueduct caused her death than that such rapid recurrence was responsible. The tumor of the third patient was a pure pinealoma and was perfectly encapsulated. It is unbelievable that it can recur, although I felt equally secure in such a prediction for the first case. In that instance, however, the tumor was a teratoma, whereas in this case the growth was a pure pinealoma. Reports of three cases of pineal tumor in which the patient survived the operation are appended. Particularly of interest are the remarkable transient postoperative sequelae and the absence of recognizable effects arising from the removal of part or all

Figure 1.20. After the bridging veins have been ligated and divided, the cerebral hemisphere is retracted to expose the falx and the corpus callosum. (From Dandy WE: An operation for the removal of pineal tumors. Sura Gynecol Obstet 33-113-119, 1921.)

Figure 1.21. The inferior longitudinal sinus has been clipped and the falx has been divided inferiorly. The corpus callosum has been split longitudinally to expose the pineal tumor, the internal cerebral veins, and the vein of Galen. (From Dandy WE: An operation for the removal of pineal tumors. Surg Gynecol Obstet 33:113-119, 1921.)

Figure 1.22. After removal of the pineal tumor, the intact roof of the third ventricle is seen. (From Dandy WE: An operation for the removal of pineal tumors. Surg Gynecol Obstet 33:113-119, 1921.)

Figure 1.23. Drawings showing the operative procedure in the removal of a teratoma from the pineal region. (From Dandy WE: Operative experience in cases of pineal tumor. Arch Surg 33:19-46, 1936.)

Figure 1.24. Drawings illustrating the method of producing more adequate exposure of a large tumor in the pineal region by resecting the occipital lobe and, when necessary, the lower part of the falx, parts of the right and left sides of the tentorium, the vein of Galen, and the straight sinus. (From Dandy WE: Operative experience in cases of pineal tumor. Arch Surg 33:19-46, 1936.)

Figure 1.25. Operative sketches of the removal of a pineal tumor, showing that the right internal cerebral vein had to be divided. (From Dandy WE: Operative experience in cases of pineal tumor. Arch Surg 33:19-46, 1936.) of the main trunks of the internal venous system of the brain (38). Many other surgeons have been involved in the development of the surgical treatment of pineal tumors since 1930, as mentioned in the publications of Thomson (145), Schmidek (133), Zulch (162), Sano (126), and Pendl (110). As with tumors of the third ventricle, the high operative morbidity and mortality of direct exposure and resection of pineal tumors led most neurosur-geons for the next 40 years to use the more conservative course of CSF shunting and radiation therapy (1, 48, 52, 79, 116, 146). This ap-

proach to treatment was strengthened by the development of better techniques of CSF shunting and of radiotherapy (7, 59, 113, 131). Then, after the advent of microneurosurgery, interest in direct surgical treatment began to return (122, 138, 153, 160). However, the need for operative exposure and resection of these lesions has been modified by the development of CT-guided stereotaxic biopsy techniques, the increase in reliance on biochemical tumor markers and CSF cytology for the diagnosis of pineal tumors, and a revival of interest in the implantation of radiation sources for the treatment of intracranial neoplasms (5, 9, 63, 85, 103, 109, 118, 126, 134).

Operative Approaches to the Third Ventricle As noted previously, Dandy exposed the third ventricle for ventriculostomy by a subfrontal and later by a transtemporal approach. For anterior third ventricular tumors, he preferred a frontal transcortical route, and for posterior third ventricular and pineal tumors, a parietooccipital transcallosal approach. Dandy thought that no neurological deficit occurred as a result of either approach (37, 71). However, "if the splenium of the corpus callosum is completely sectioned, appropriate testing will reveal that most postoperative patients cannot read in their left visual fields, even when all other neurological functions have returned to normal. This phenomenon, appropriately called hemialexia, was first reported by Trescher and Ford in 1937 [147]. Their patient had the posterior corpus callosum split by Walter Dandy for removal of a colloid cyst of the 3rd ventricle. . . . Since that time the reality of the deficit has been proven repeatedly. . ." (71).

These and various other operative approaches to the region of the third ventricle have been developed, primarily since Dandy's time. Many of these have been discussed by Rhoton and Yamamoto (Fig. 1.26) (119, 120, 161) and by Antunes et al. (4), as well as at a recent international symposium organized by Madjid Samii and held in Hannover, West Germany (83). The transsphenoidal and certain subfrontal (i.e., subfrontal subchiasmatic or subfrontal transsphenoidal) approaches permit access to the sella turcica and chiasmatic area but do not ordinarily permit visualization within the third ventricle. The same is true of the more lateral subfrontal and subtemporal approaches. Tumors can be drawn down from the third ventricle but entry into the ventricle is quite limited by the angle of entry and the important structures that are interposed (optic nerves, chiasm, and tracts; arteries of the circle of Willis and their branches; hypothalamus; pituitary stalk and gland; etc.). The occipital transtentorial and infratentorial supracerebellar approaches have been used pri-

Figure 1.26. A midsagittal view of the head, showing the basic operative approaches to the third ventricle. (From Rhoton AL Jr, Yamamoto I, Peace DA: Microsurgery of the third ventricle: Part 2. Operative approaches. Neurosurgery 8:357-373, 1981.)

marily to expose tumors in the region of the pineal gland. The early history of these approaches has been considered, and the reader is referred to the publications of Thomson (145), Schmidek (133), Zulch (162), Sano (126), and Pendl (110) for details about the subsequent development of these procedures. Among the subf rental approaches, the one that has been somewhat helpful in dealing with low anterior third ventricular tumors has been the one that involves entry into the ventricle through the lamina terminalis. Such entry was used for third ventriculostomy, as previously discussed. King (91) and Suzuki et al. (142) have outlined the history of its use for tumors such as intra-ventricular craniopharyngiomas. The posterior transventricular approach has been of limited value. However, the anterior transventricular route through a dilated lateral ventricle to high anterior third ventricular tumors such as colloid cysts has remained useful (52). Transcallosal operations were done occasionally after Dandy developed the posterior transcallosal approach (16, 17, 50-52, 72, 73, 108, 151), but it was not until the introduction of microsurgical techniques (101) that the transcallosal route became popular (52, 96, 97, 136). The anterior transcallosal approach has become a preferred method of access to the third ventricle (Sano even uses it for pineal region neoplasms (126, 127)), especially since it has been found that anterior or middle transcallosal operations may not cause behavioral or neurological abnormalities (67, 159). The emphasis in recent years has been on the refinement of techniques for opening the roof of the third ventricle (4, 6, 13, 21,47,78,83,93, 154). Concluding Remarks The central location of the third ventricle and the pineal gland within the brain have lent them a certain mystery. Descartes thought of the pineal gland as "the seat of the soul" (19). Surgeons dealing with lesions of the third ventricle have noted that such lesions (or injuries resulting from the attempts to remove them) may cause alterations in consciousness (14, 15, 42). In fact, Walter Dandy's last paper, published 3 months after his death, contained his speculations about the location of the "conscious center" in the brain, based on a study of 10 of his patients, 2 of whom were rendered unconscious by the removal of tumors from the third ventricle (42). The mysteries surrounding lesions of the third

ventricular region are slowly being solved. The pioneering work of Walter Dandy and others has laid the foundation for our present methods of diagnosing and treating these conditions. These current methods and the future advancements they may permit are detailed in the rest of this book. References 1 . Albay EO II, Laws ER Jr, Grado GL, Bruckman JE, Forbes GS, Gomez MR, Scott M: Pineal tu mors in children and adolescents: Treatment by CSF shunting and radiotherapy. J Neurosurg 55:889-895, 1981. 2. Anton, von Bramann: Balkenstich bei Hydrozephalien, Tumoren und bei Epilepsie. Munch Med Wochenschr 55:1673-1677, 1908. 3. Antunes JL, Louis KM, Ganti SR: Colloid cysts of the third ventricle. Neurosurgery 7:450-455, 1980. 4. Antunes JL, Muraszko K, Quest DO, Carmel PW: Surgical strategies in the management of tu mours of the anterior third ventricle, in Brock M (ed): Modern Neurosurgery 1. Berlin, SpringerVerlag, 1982, pp 215-224. 5. Apuzzo MLJ, Chandrasoma PT, Zelman V, Giannotta SL, Weiss MH: Computed tomographic guidance stereotaxis in the management of le sions of the third ventricular region. Neurosur gery 15:502-508, 1984. 6. Apuzzo MLJ, Chikovani OK, Gott PS, Teng EL, Zee CS, Giannotta SL, Weiss MH: Transcallo sal, interfornicial approaches for lesions affect ing the third ventricle: Surgical considerations and consequences. Neurosurgery 10:547-554, 1982. 7. Backlund EO, Rahn T, Sarby B: Treatment of pinealomas by stereotaxic radiation surgery. ActaRadiol Ther 13:368-376, 1974. 8. Balado M, Morea R, Donovan C: La radiografia del tercer ventriculo. Bol Inst Clin Quirurg 2:603-610, 1926 (cited by Walker (157)). 9. Bigner SH, Johnston WW: Cytopathology of the Central Nervous System. New York, Masson, 1983. 10. Bosch DA, Rahn T, Backlund EO: Treatment of colloid cysts of the third ventricle by stereotactic aspiration. Surg Neurol 9:15-18, 1978. 1 1 . Brocklehurst G: Diencephalic cysts. J Neurosurg 38:47-51, 1973. 12. Burger PC, Vogel FS: Surgical Pathology of the Nervous System and Its Coverings. New York, John Wiley & Sons, 1982, ed 2, p 279. 13. Busch E: A new approach for the removal of tumors of the third ventricle. Acta Psychiatr Neurol 19:57-60, 1944. 14. Cairns H: Disturbances of consciousness with lesions of the brain-stem and diencephalon. Brain 75:109-146, 1952. 15. Cairns H, Oldfield RC, Pennybacker JB, Whitteridge D: Akinetic mutism with an epidermoid cyst of the 3rd ventricle. (With a report on the asso ciated disturbance of brain potentials.) Brain 64:273-290, 1941.

16. Cassinari V, Bernasconi V: Tumori della parte anteriore del terzo ventricolo. Acta Neurochir (Wien) 11:236-271, 1964. 17. Christensen JC: Third ventricle tumours: Clini cal findings and surgical results in 28 consecu tive cases. Excerpta Med Int Congr Series 36:E61-E62, 1961. 18. Ciric I, Zivin I: Neuroepithelial (colloid) cysts of the septum pellucidum. J Neurosurg 43:69-73, 1975. 19. Clarke E, O'Malley CD: The Human Brain and Spinal Cord: A Historical Study Illustrated by Writings from Antiquity to the Twentieth Cen tury. Berkeley, CA, University of California Press, 1968, pp 469-472. 20. Cooper JR: Brain tumors in hereditary multiple system hamartomatosis (tuberous sclerosis). J Neurosurg 34:194-202, 1971. 21. Cossu M, Lubinu F, Orunesu G, Pau A, Sehrbundt Viale E, Sini MG, Turtas S: Subchoroidal approach to the third ventricle: Microsurgical anatomy. Surg Neurol 21:325-331, 1984. 22. Cushing H: Intracranial Tumours: Notes Upon a Series of Two Thousand Verified Cases with Surgical-Mortality Percentages Pertaining Thereto. Springfield, IL, Charles С Thomas, 1932, p 64. 23. Cushing H, Eisenhardt L: Meningiomas: Their Classification, Regional Behaviour, Life His tory, and Surgical End Results. Springfield, IL, Charles С Thomas, 1938 (New York, Hafner, 1969), pp 125-126, 136-138. 24. Dandy WE: The nerve supply to the pituitary body. Am J Anat 15:333-343, 1913. » 25. Dandy WE: Extirpation of the pineal body. J Exp Med 22:237-247, 1915. 26. Dandy WE: Ventriculography following the in jection of air into the cerebral ventricles. Ann Surg 68:5-11, 1918. 27. Dandy WE: Extirpation of the choroid plexus of the lateral ventricles in communicating hydrocephalus. Ann Surg 68:569-579, 1918. 28. Dandy WE: Rontgenography of the brain after the injection of air into the spinal canal. Ann Surg 70:397-403, 1919. 29. Dandy WE: The diagnosis and treatment of hydrocephalus resulting from strictures of the aq ueduct of Sylvius. Surg Gynecol Obstet 31:340358, 1920. 30. Dandy WE: An operation for the removal of pineal tumors. Surg Gynecol Obstet 33:113119, 1921. 31. Dandy WE: II. Diagnosis, localization and re moval of tumors of the third ventricle. Johns Hopkins Hosp Bull 33:188-189, 1922. 32. Dandy WE: III. Cerebral ventriculoscopy. Johns Hopkins Hosp Bull 33:189, 1922. 33. Dandy WE: V. An operative procedure for hydrocephalus. Johns Hopkins Hosp Bull 33:189190, 1922. 34. Dandy WE: Mechanism and symptoms of tumors of the third ventricle and pineal body. Res Publ Assoc Res Nerv Ment Dis 8:375-385, 1929. 35. Dandy WE: Congenital cerebral cysts of the cavum septi pellucidi (fifth ventricle) and cavum vergae (sixth ventricle): Diagnosis and treat ment. Arch Neurol Psychiatry 22:44-66, 1931.

36. Dandy WE: The brain, in Lewis D (ed): Practice of Surgery. Hagerstown, MD, WF Prior, 1932, pp 1-682. 37. Dandy WE: Benign Tumors in the Third Ven tricle of the Brain: Diagnosis and Treatment. Springfield, IL, Charles С Thomas, 1933. 38. Dandy WE: Operative experience in cases of pineal tumor. Arch Surg 33:19-46, 1936. 39. Dandy WE: The operative treatment of commu nicating hydrocephalus. Ann Surg 108:194202, 1938. 40. Dandy WE: Section of the human hypophysial stalk: Its relation to diabetes insipidus and hy pophysial functions. JAMA 114:312-314, 1940. 41. Dandy WE: Diagnosis and treatment of stric tures of the aqueduct of Sylvius (causing hydro cephalus). Arch Surg 51:1-14, 1945. 42. Dandy WE: The location of the conscious center in the brain—the corpus striatum. Bull Johns Hopkins Hosp 79:34-58, 1946. 43. Dandy WE, Blackfan KD: An experimental and clinical study of internal hydrocephalus. JAMA 61:2216-2217, 1913. 44. Dandy WE, Blackfan KD: Internal hydrocepha lus: An experimental, clinical and pathological study. Am J Dis Child 8:406-482, 1914. 45. Dandy WE, Blackfan KD: Internal hydrocepha lus: Second paper. Am J Dis Child 14:424-443, 1917. 46. Dandy WE, Goetsch E: The blood supply of the pituitary body. Am J Anat 11:137-150, 1911. 47. Delandsheer JM, Guyot JF, Jomin M, Scherpereel B, Laine E: Acces au troisieme ventricule par voie inter-thalamo-trigonale. Neurochirurgie 24:419-422, 1978. 48. Donat JF, Okazaki H, Gomez MR, Reagan TJ, Baker HL Jr, Laws ER Jr: Pineal tumors: A 53year experience. Arch Neurol 35:736-740, 1978. 49. Drayer BP, Rosenbaum AE: Studies of the third circulation; Amipaque CT cisternography and ventriculography. J Neurosurg 48:946-956, 1978. 50. Ehni G: Pineal teratoma: Report of case. J Neu rosurg 3:86-94, 1946. 51. Ehni G: Transcallosal Removal of Brain Tumors. Motion picture available from American College of Surgeons Film Library or Baylor College of Medicine, 1969. 52. Ehni G: Interhemispheric and percallosal (transcallosal) approach to the cingulate gyri, intraventricular shunt tubes, and certain deeply placed brain lesions. Neurosurgery 14:99-110, 1984. 53. Elvidge AR: Treatment of obstructive lesions of the aqueduct of Sylvius and the fourth ventricle by interventriculostomy. J Neurosurg 24:1123, 1966. 54. Estanol B, Kleriga E, Loyo M, Mateos H, Lombardo L, Gordon F, Saguchi AF: Mechanisms of hydrocephalus in cerebral cysticercosis: Impli cations for therapy. Neurosurgery 13:119-123, 1983. 55. Fairman D: Evolution of neurosurgery through Walter E. Dandy's work. Surgery 19:581-604, 1946. 56. Fay T, Grant FC: Ventriculoscopy and intraven-

57. 58.

59.

60. 61. 62. 63.

64. 65.

66. 67.

68. 69. 70.

71. 72. 73. 74.

75.

tricular photography in internal hydrocephalus: Report of case. JAMA 80:461-463, 1923. Feld M: Ventriculoscope coagulant utilise dans les hydrocephalies communicantes. Presse Med 64:632, 1956. Feld M: La coagulation des plexus choroides par ventriculoscopie directe dans l'hydrocephalie non obstructive du nourrisson. Neurochirurgie 3:70-79, 1957. Fisher RG: Surgery of the congenital anomalies, in Walker AE (ed): A History of Neurological Surgery. Baltimore, Williams & Wilkins, 1951, pp 334-361. Foerster O: Das operative Vorgehen bei Tumor der Vierhugelgegend. Wien Klin Wochenschr 41:986-990, 1928. Foerster O: Uber das operative Vorgehen bei Operationen der Vierhugelgegend. Zbl Ges Neurol Psychiat 61:457-459, 1932. Fox WL: Dandy of Johns Hopkins. Baltimore, Williams & Wilkins, 1984. Frank F, Gaist G, Piazza G, Ricci RF, Sturiale C, Galassi E: Stereotaxic biopsy and radioactive im plantation for interstitial therapy of tumors of the pineal region. Surg Neurol 23:275-280, 1985. Fukushima T: Endoscopic biopsy of intraventricular tumors with the use of a ventriculof iberscope. Neurosurgery 2:110-113, 1978. Fukushima T, Ishijima B, Hirakawa K, Nakamura N, Sano K: Ventriculofiberscope: A new technique for endoscopic diagnosis and opera tion. Technical note. J Neurosurg 38:251-256, 1973. Fulton JF: Harvey Cushing: A Biography. Springfield, IL, Charles С Thomas, 1946, p 618. Geffen G, Walsh A, Simpson D, Jeeves M: Com parison of the effects of transcortical and transcallosal removal of intraventricular tu mours. Brain 103:773-788, 1980. Godersky JC, Rockswold G, Larson DA: Xanthogranuloma of the third ventricle producing hy drocephalus. Neurosurgery 7:68-70, 1980. Goldstein SJ, Wilson DD, Young AB, Guidry GJ: Craniopharyngioma intrinsic to the third ventri cle. Surg Neurol 20:249-253, 1983. Gradin WC, Taylon C, Fruin AH: Choroid plexus papilloma of the third ventricle: Case report and review of the literature. Neurosurgery 12:217220, 1983. Greenblatt SH: Neurosurgery and the anatomy of reading: A practical review. Neurosurgery 1:6-15, 1977. Greenwood J Jr: Paraphyseal cysts of the third ventricle: With report of eight cases. J Neuro surg 6:153-159, 1949. Greenwood J Jr: Radical surgery of tumors of the thalamus, hypothalamus, and third ventricle area. Surg Neurol 1:29-33, 1973. Guiot G, Derome P, Hertzog E, Bamberger C, Akerman M: Ventriculo-cisternostomie sous controle radioscopique pour hydrocephalie ob structive. Presse Med 76:1923-1926, 1968. Guiot G, Rougerie J, Fourestier M, Fournier A, Comoy C, Vulmiere J, Groux R: Explorations endoscopiques intracraniennes. Presse Med 71:1225-1228, 1963.

76. Gutierrez-Lara F, Patino R, Hakim S: Treatment of tumors of the third ventricle: A new and sim ple technique. Surg Neurol 3:323-325, 1975. 77. Harris W, Cairns H: Diagnosis and treatment of pineal tumours: With report of a case. Lancet 1:3-9, 1932. 78. Hirsch JF, Zouaoui A, Renier D, Pierre-Kahn A: A new surgical approach to the third ventricle with interruption of the striothalamic vein. Acta Neurochir (Wien) 47: 135-147, 1979. 79. Hitchon PW, Abu-Yousef MM, Graf CJ, Turner DM, VanGilder JC: Management and outcome of pineal region tumors. Neurosurgery 13:248253, 1983. 80. Hoffman HJ, Harwood-Nash D, Gilday DL: Per cutaneous third ventriculostomy in the manage ment of noncommunicating hydrocephalus. Neurosurgery 7:313-321, 1980. 81. Horsley V: Discussion of paper by Howell (82). Proc R Soc Med 3:77, 1910. 82. Howell CMH: Tumours of the pineal body. Proc R Soc Med 3:65-77, 1910. 83. International Symposium on Surgery in and around the Brain Stem and the Third Ventricle. Hannover, West Germany, February 18-23, 1985. 84. Jacobaeus HC, Nord F: Air and lipiodol as con trast agents for roentgen diagnosis within the central nervous system. Acta Radiol 3:367-382, 1924. 85. Jooma R, Kendal BE: Diagnosis and manage ment of pineal tumors. J Neurosurg 58:654665, 1983. 86. Kapp JP, Paulson GW, Odom GL: Brain tumors with tuberous sclerosis. J Neurosurg 26:191202, 1967. 87. Kelly PJ, Earnest F IV, Kall BA, Goerss SJ, Scheithauer B: Surgical options for patients with deepseated brain tumors: Computer-assisted stereotactic biopsy. Mayo Clin Proc 60:223-229, 1985. 88. Kelly PJ, Kall BA, Goerss S: Transposition of volumetric information derived from computed tomography scanning into sterotactic space. Surg Neurol 21:465-471, 1984. 89. Kelly PJ, Kall B, Goerss S, Alker GJ Jr: Precision resection of intra-axial CNS lesions by CT-based stereotactic craniotomy and computer monitored CO2 laser. Acta Neurochir (Wien) 68:1-9, 1983. 90. Kempe LG, Busch E: Clinical significance of the cisterna veli interpositi. Acta Neurochir (Wien) 16:241-248, 1967. 91. King TT: Removal of intraventricular craniopharyngiomas through the lamina terminalis. Acta Neurochir (Wien) 45:277-286, 1979. 92. Krause F: Chirurgie des Gehirns und Ruckenmarks, Vol 1. Berlin, Urban and Schwarzenberg, 1911 (cited by Pendl (110)). 93. Lavyne MH, Patterson RH Jr: Subchoroidal trans-velum interpositum approach to mid-third ventricular tumors. Neurosurgery 12:86-94, 1983. 94. Leksell L: A surgical procedure for atresia of the aqueduct of Sylvius. Acta Psychiatr Neurol 24:559-568, 1949. 95. Lespinasse VL: Cited by Davis L: Neurological

Surgery. Philadelphia, Lea & Febiger, 1936, p 405. 96. Long DM, Chou SN: Transcallosal removal of craniopharyngiomas within the third ventricle. J Neurosurg 39:563-567, 1973. 97. Long DM, Leibrock L: The transcallosal ap proach to the anterior ventricular system and its application in the therapy of craniopharyngioma. Clin Neurosurg 27:160-168, 1980. 98. Madrazo NI, Garcia-Renteria JA, Sandoval BM, Lopez Vega FJ: Intraventricular cysticercosis. Neurosurgery 12:148-152, 1983. 99. Matthews FD: Intraventricular craniopharyngioma. AJNR 4:984-985, 1983. 100. McCormick GF, Zee CS, Heiden J: Cysticercosis cerebri: Review of 127 cases. Arch Neurol 39: 534-539, 1982. 101. Milhorat TH, Baldwin M: Technique for surgical exposure of the cerebral midline: Experimental transcallosal microdissection. J Neurosurg 24:687-691, 1966. 102. Mixter WJ: Ventriculoscopy and puncture of the floor of the third ventricle. Boston Med Surg J 188:277-278, 1923. 103. Neuwelt EA (ed): Diagnosis and Treatment of Pineal Region Tumors. Baltimore, Williams & Wilkins, 1984. 104. Norlen G: Contribution to the surgical treatment of inoperable tumours causing obstruction of the sylvian aqueduct. Acta Psychiatr Neurol 24: 629-637, 1949. 105. Oppenheim H, Krause F: Operative Erfolge bei Geschwulsten der Sehhugel-und Vierhugelgegend. Berl Klin Wochenschr 50:2316-2322, 1913. 106. Palmer ED: Who was first? JAMA 239:16091610, 1978. 107. Papo I, Scarpelli M, Caruselli G: Intrinsic third ventricle craniopharyngiomas with normal pres sure hydrocephalus. Acta Neurochir (Wien) 23: 80-88, 1980. 108. Pecker J, Ferrand B, Javalet A: Tumeurs du troisieme ventricule. Neurochirurgie 12:7-136, 1966. 109. Pecker J, Scarabin JM, Vallee B, Brucher JM: Treatment in tumours of the pineal region: Value of stereotaxic biopsy. Surg Neurol 12:341- 348, 1979. 110. Pendl G: The surgery of pineal lesions—histori cal perspective, in Neuwelt EA (ed): Diagnosis and Treatment of Pineal Region Tumors. Bal timore, Williams & Wilkins, 1984, pp 139-154. 1 1 1 . Piatt JH Jr, Campbell GA: Pineal region meningioma: Report of two cases and literature review. Neurosurgery 12:369-376, 1983. 112. Powell MP, Torrens MJ, Thomson JLG, Horgan JG: Isodense colloid cysts of the third ventricle: A diagnostic and therapeutic problem resolved by ventriculoscopy. Neurosurgery 13:234-237, 1983. 113. Pudenz RH: The surgical treatment of hydro cephalus—an historical review. Surg Neurol 15:15-26, 1981. 114. Pussep L: Die operative Entfernung einer Zyste der Glandula pinealis. Neurol Zentralbl 33:560563, 1914. 115. Putnam TJ: Treatment of hydrocephalus by endoscopic coagulation of the choroid plexus: De-

116. 117.

118.

119.

120. 121. 122.

123. 124.

125. 126. 127. 128. 129.

130.

131.

132.

133. 134.

scription of a new instrument and preliminary report of results. N Engl J Med 210:1373-1376, 1934. Rand RW, Lemmen LJ: Tumors of the posterior portion of the third ventricle. J Neurosurg 10:118, 1953. Reichert FL, Dandy WE: Polyuria and polydipsia (diabetes insipidus) and glycosuria resulting from animal experiments on the hypophysis and its environs. Bull Johns Hopkins Hosp 58:418427, 1936. Rekate HL, Ruch T, Nulsen FE, Roessmann U, Spence J: Needle biopsy of tumors in the region of the third ventricle. J Neurosurg 54:338-341, 1981. Rhoton AL Jr, Yamamoto I: Operative ap proaches to the third ventricle, in Wilkins RH, Rengachary SS (eds): Neurosurgery. New York, McGraw-Hill, 1985, pp 938-951. Rhoton AL Jr, Yamamoto I, Peace DA: Microsur gery of the third ventricle: Part 2. Operative approaches. Neurosurgery 8:357-373, 1981. Rivas JJ, Lobato RD: CT-assisted stereotaxic as piration of colloid cysts of the third ventricle. J Neurosurg 62:238-242, 1985. Rout D, Sharma A, Radhakrishnan VV, Rao VRK: Exploration of the pineal region: Observa tions and results. Surg Neurol 21:135-140, 1984. Rozario R, Adelman L, Prager RJ, Stein BM: Meningiomas of the pineal region and third ven tricle. Neurosurgery 5:489-495, 1979. Ruggiero G, Bories J, Calabro A, Cristi G, Scialfa G, Smaltino F, Thibaut A: Radiological Explo ration of the Ventricles and Subarachnoid Space. New York, Springer-Verlag, 1974. Rush JL, Kusske JA, Porter RW, Pribram HW: Xanthogranulomas of the third ventricle. Neu rosurgery 4:329-333, 1979. Sano K: Pineal region tumors: Problems in pa thology and treatment. Clin Neurosurg 30:5991, 1983. Sano K: Microsurgery of tumors in the pineal region, in Rand RW (ed): Microneurosurgery. St Louis, CV Mosby, 1985, ed 3, pp 391-397. Sayers MP, Kosnik EJ: Percutaneous third ventriculostomy: Experience and technique. Childs Brain 2:24-30, 1976. Scarff JE: Endoscopic treatment of hydroceph alus: Description of a ventriculoscope and pre liminary report of cases. Arch Neurol Psychia try 35:853-860, 1936. Scarff JE: Treatment of obstructive hydroceph alus by puncture of the lamina terminalis and floor of the third ventricle. J Neurosurg 8:204213, 1951. Scarff JE: Treatment of hydrocephalus: An his torical and critical review of methods and re sults. J Neurol Neurosurg Psychiatry 26:1-26, 1963. Scarff JE: The treatment of nonobstructive (communicating) hydrocephalus by endoscopic cauterization of the choroid plexuses. J Neuro surg 38:1-18, 1970. Schmidek HH: Surgical management of pineal region tumors, in Schmidek HH (ed): Pineal Tu mors. New York, Masson, 1977, pp 99-113. Schmidek HH, Waters A: Pineal masses: Clinical

135. 136. 137.

138. 139. 140. 141.

142.

143. 144.

145.

146.

147.

features and management, in Wilkins RH, Rengachary SS (eds): Neurosurgery. New York, McGraw-Hill, 1985, pp 688-693. Schuster J: Ventriculographie mit Lipiodol Ascendens und Descendens. Klin Wochenschr 4:2064-2068, 1925. Shucart WA, Stein BM: Transcallosal approach to the anterior ventricular system. Neurosur gery 3:339-343, 1978. Sole-Llenas J, Royo Salvador M, Llovet J, Sanchez-Larrea R, Rovira RR: Craniopharyngioma of the third ventricle. Neurochirurgia 26:93-94, 1983. Stein BM: Surgical treatment of pineal tumors, Clin Neurosurg 26:490-510, 1979. Stein BM: Third ventricular tumors, din Neurosurg 27:315-331, 1980. Stern WE: Parasitic infestations, in Wilkins RH, Rengachary SS (eds): Neurosurgery. New York, McGraw-Hill, 1985, pp 2010-2015. Stookey B, Scarff J: Occlusion of the aqueduct of Sylvius by neoplastic and non-neoplastic processes with a rational surgical treatment for relief of the resultant obstructive hydrocephalus. Bull Neurol Inst NY 5:348-377, 1936. Suzuki J, Katakura R, Mori T: Interhemispheric approach through the lamina terminalis to tu mors of the anterior part of the third ventricle. Surg Neurol 22:157-163, 1984. Sweet WH: Radical surgical treatment of craniopharyngioma. Clin Neurosurg 23:52-79, 1976. Takei Y, Pearl GS, Fresh CB, Fleischer AS, Payne NS, Tindall GT: Intraventricular fiberscopic ob servations in adult hydrocephalic patients. Neu rosurgery 5:437-440, 1979. Thomson AF: Surgery of the third ventricle, in Walker AE (ed): A History of Neurological Sur gery. Baltimore, Williams & Wilkins, 1951, pp 136-149. Torkildsen A: Should extirpation be attempted in cases of neoplasm in or near the third ventri cle of the brain? Experiences with a palliative method. J Neurosurg 5:249-275, 1948. Trescher JH, Ford FR: Colloid cyst of the third ventricle. Arch Neurol Psychiatry 37:959-973, 1937.

148. Turnbull IM, Drake CG: Membranous occlusion of the aqueduct of Sylvius. J Neurosurg 24:2433, 1966. 149. Van Den Bergh R, Brucher JM: L'abord transventriculaire dans les cranio-pharyngiomes du troisieme ventricule: Aspects neuro-chirurgicaux et neuro-pathologiques. Neurochirurgie 16:51-65, 1970. 150. Van Wagenen WP: A surgical approach for the removal of certain pineal tumors. Surg Gynecol Obstet 53:216-220, 1931. 151. Van Wagenen WP: Surgery of the hypothalamic region. Res Publ Assoc Res Nerv Ment Dis 20:841-853, 1940. 152. Van Wagenen WP, Aird RB: Dilatations of the cavity of the septum pellucidum and cavum vergae. Am J Cancer 20:539-557, 1934. 153. Ventureyra ECG: Pineal region: Surgical man agement of tumours and vascular malforma tions. Surg Neurol 16:77-84, 1981. 154. Viale GL, Turtas S: The subchoroid approach to the third ventricle. Surg Neurol 14:71-76, 1980. 155. Volkmann J: Ueber Versuche zur unmittelbaren Besichtigung der Gehirnkammern (Enzephaloskopie). Munch Med Wochenschr 70:1382, 1923. 156. Volkmann J: Das Encephaloskop. Zbl Chir 51:1233-1234, 1924. 157. Walker AE: Diagnostic procedures, in Walker AE (ed): A History of Neurological Surgery. Balti more, Williams & Wilkins, 1951, pp 23-39. 158. Wilkins RH: Neurosurgical classic—XX. J Neu rosurg 21:586-635, 1964. 159. Winston KR, Cavazzuti V, Arkins T: Absence of neurological and behavioral abnormalities after anterior transcallosal operation for third ven tricular lesions. Neurosurgery 4:386-393, 1979. 160. Wood JH, Zimmerman RA, Bruce DA, Bilaniuk LT, Norris DG, Schut L: Assessment and man agement of pineal-region and related tumors. Surg Neurol 16:192-210, 1981. 161. Yamamoto I, Rhoton AL Jr, Peace DA: Microsurgical anatomy of the third ventricle, in Brock M (ed): Modern Neurosurgery 1. Berlin, Sprin ger-Verlag, 1982, pp 205-214. 162. Zulch KJ: Reflections on the surgery of the pineal gland (a glimpse into the past). Neurosurg Rev 4:159-162, 1981.

2 Comparative Anatomy of the Third Ventricular Region E.Leon Kier,M.D. What is often more condensed or more concealed in one species Nature displays more clearly and openly in another. De Graaf (1664) quoted by Cole (1944)

Lateral Walls Hypothalamus Thalamus (Massa Intermedia) Intraventricular Foramen (Monro) Floor Optic Chiasm Optic Recess Infundibular Recess Saccus Vasculosus Tuber Cinereum Hypophyseal Recess Hypophysis Mammillary Bodies Premamillary Recess Postmamillary Recess Midbrain Roof and Posterior Wall Paraphysis Velum Transversum Dorsal Sac (Saccus Dorsalis) Suprapineal Recess Parencephalon Habenulae Pineal (epiphysis) Parietal Eye Pineal Recess Parapineal Posterior Commissure Tela choroidea (Velum Interpositum) Optic Lobes (Corpora Bigemina) Auditory Lobes Superior Colliculi Inferior Colliculi Quadrigeminal Plate (Corpora Quadrigemina) Mesencephalic Ventricle Optic Ventricle Aqueduct Aqueductal Ampulla

Commissures Limbic System Lamina Terminalis Anterior Commissure Paraterminal Body Septal Area Hippocampal Commissure Hippocampus Fornix Corpus Callosum Subiculum Indusium Griseum Longitudinal striae Septum Pellucidum Cavum Septi Pellucidi Arteries Principles of Cerebral Vascular Development Delivering Arteries End Arteries Recurrent Artery of Heubner Perforating end arteries of Hypothalamus and Thalamus Arterial Circle (Willis) Anterior Cerebral Anterior Choroidal Posterior Cerebral Posterior Choroidal Veins

Comparative anatomical knowledge can enhance and enliven the study of neuroanatomy and neuroradiological anatomy (11). The study of developmental and adult neuroanatomy is thus transformed from memorizing a large number of anatomical structures to a meaningful understanding of many features of the human brain (32, 33). This meaningful understanding is based on the slowly evolving needs of different vertebrates manifested in various evolutionary modifications of the central nervous system. The study of comparative anatomy permits a relatively slow longitudinal study of the developing human central nervous system. This slower pace is not available in human embryos that "recapitulate" many stages very quickly and transiently (13). In addition, many congenital, metabolic, toxic, and degenerative disorders may be viewed through the perspective of evolution (67). Certain disorders such as developmental hypoplasias involve new phylogenetic structures while sparing the older phylogenetic section of the nervous

Figure 2.2. Midsagittal section of the third ventricle. The floor extends from the optic chiasm to the aqueduct. The anterior wall extends from the optic chiasm to the foramen of Monro (interventricular foramen). The roof extends from the foramen of Monro to the suprapineal recess. The posterior wall extends from the suprapineal recess to the aqueduct. (From Yama-moto I, Rhoton AL, Pierce RA: Microsurgery of the third ventricle: Part 1. Microsurgical anatomy. Neu-rosurgery 8:333-356, 1981.)

Figure 2.1. Diagram of the medial aspect of the left cerebral hemisphere demonstrating the majority of the structures of the limbic system. Note that many of these complex structures are part of the anatomy of the third ventricular region. From Warwick R, Williams PL (eds): Gray's Anatomy. Philadelphia, WB Saunders Co, 1980, British ed 36.

system. In contrast, anomalies of the central nervous system such as holoprosencephaly and Dandy-Walker selectively affect the oldest phy-logenetic structures while sparing the newer ones. The merit of a comparative anatomical study is particularly apparent in the elucidation of the complex anatomy of the commissures in the region of the third ventricle (Fig. 2.1). In addition the various forms of corpus callosum dysgenesis are clarified by the knowledge of its evolutionary changes. In this chapter an effort was made to group the various structures according to the usual anatomical and neurosurgical landmarks of the third ventricle (83, 89) (Fig. 2.2). However, the roof and posterior wall structures are grouped together because in many vertebrates the posterior wall cannot be defined as a separate structure. The complex evolution of the commissures and cer-

tain principles governing vascular evolutionary changes are described in separate sections. Lateral Walls Phylogenetically the hypothalamus is probably the oldest cerebral structure with remarkably similar nuclear differentiation and fiber connections in all vertebrates (67). Evidence of neurose-cretory activity is present even in the simplest invertebrates. In certain worms the entire brain could be considered to be a form of the hypothalamus. Internal neurosecretory control preceded the evolution of special sensory and locomotor systems relating the organism to the outside world. Thus the phylogenetic variability of the evolving hypothalamus is manifested in the variability of the floor of the third ventricle. The evolving thalamus affects the topography of the phylogenetically changing lateral ventricles and commissures (33). In addition, the evolv-

Figure 2.3. Ventrodorsal roentgenograms of a shark (A) and a frog (B) and the brain of an iguana (C), rabbit (D), dog (E), and monkey (F) with barium-cast ventricles. Except the shark, all other specimens demonstrate a small interven-tricular foramen. The wider thalamic part is superimposed on the narrower hypothalamic part of the third ventricle. In the shark and the frog the lateral ventricles are anterior to the interventricular foramen. In the iguana the posterior pole of the lateral ventricle is now posterior to the interventricular foramen (arrow). In the rabbit the major part of the lateral ventricle is posterior to the interventricular foramen. From Kier EL: The cerebral ventricles: a phylogenetic and ontogenetic study. In Newton TH, Potts DG (eds): Radiology of the Skull and Brain, vol. 3, Anatomy and Pathology, St. Louis, The CV Mosby Co, 1977, pp 2787-2914.

ing thalamus and lateral geniculate body contribute to the phylogenetic changes involving the anterior and posterior choroidal arteries (32). The interthalamic adhesion (massa intermedia) varies in size, and the hypothalamus and thalamus create minor differences in the lateral walls of the third ventricle. Except for the frog and iguana, the vertebrates studied demonstrate that the hypothalamic part of the third ventricle is narrower than the thalamic part (Fig. 2.3). In addition, the evolution of the cerebrum changes the topographic relationship of the lateral ventricles and the intraventricular foramen (Fig. 2.3).

In the shark and the frog the lateral ventricles are anterior to the intraventricular foramen. In the iguana the posterior pole of the lateral ventricle is posterior to the intraventricular foramen. In the rabbit the major part of the lateral ventricle is posterior to the intraventricular foramen. The comparative study demonstrated that all specimens except the shark had small interven-ticular foramina (Fig. 2.3). It seems that the corpus striatum is the major structure reducing the size of the interventricular foramen and not the evolving fornix and thalamus. The latter structures are not present in the frog or iguana.

Floor An optic chiasm is present in the floor of the diencephalon of all vertebrates except bony fish. In the latter the decussation of the optic nerves is between the eyes and the diencephalon (25). The optic chiasm is a major modifier of the floor of the third ventricle because its size and location affect the size and configuration of the optic and infundibular recesses (Figs. 2.4 to 2.11). The number of optic nerve fibers, the extent of myelinization, and the degree of decussation in the optic chiasm vary among the different vertebrates (59, 67). The following examples demonstrate the extensive variability that exists in the number of optic nerve fibers present and the importance and development of the visual system: lamprey, 5,217; frog, 29,000; turtle and alligator, 105,000; shark, 113,000; cat, 119,000; dog, 154,000; rabbit, 265,000; sheep, 649,000. In some primates and birds in which sight is the dominant sense, the number of optic nerve fibers reaches or exceeds 1 million. In the less developed vertebrates the optic nerve fibers are poorly myelinated. In advanced teleosts, birds, and most

terrestrial vertebrates the majority of fibers are heavily myelinated. The presence of myelin seems to be associated with improved visual function. Nearly all optic nerve fibers in nonmammalian vertebrates and in some lower vertebrates decussate in the optic chiasm. The crossing of the optic fibers at a primoridal chiasm was an early evolutionary development. The presence of axons that crossed between the epithelium of the primitive photoreceptor—the precursor of the eye— and the effector system of the opposite side of the body permitted the primitive prevertebrate to move away from a threatening photic stimulus. The evolution of the two eyes in the non-mammalian vertebrates, each with its own monocular field of vision, is postulated to have necessitated the decussation of the optic fibers. Without the decussation the segments of a visualized object would be incompatibly recombined in the brain. In mammals the number of crossing optic nerve fibers is reduced progressively. This reduction is associated with the changing axis of vision and the development of binocular fields of view. In rodents, with laterally placed eyes and

Figure 2.4. Sagittal section of the brain of a shark with barium-cast ventricles. The roof of the third ventricle and its recesses are not covered by the cerebrum. The anterior commissure is present. The dorsal sac is the homologue of the suprapineal recess in the human. The infundibular recess in the shark is the ventricular extension into the inferior lobes of the hypothalamus. Note the large optic ventricle. Only the proximal end of the pineal recess is seen.

minimal overlapping of the monocular fields of vision, only a small number of fibers do not decussate. At least one-third of the fibers are nondecussating in carnivores and primates, with frontally placed eyes and overlapping fields of view. In primates nearly half the fibers do not decussate. This degree of nondecussation is one of the important factors that allow two identical retinal images to be transferred to the brain, producing the highly evolved stereoscopic vision of primates. All of these factors affect the size and shape of the optic chiasm and result in variation of the optic and infundibular recess. The optic recess seems to be a residual space on the floor of the

third ventricle whose formation, shape, and size are the result of the varying indentation of the ventricular floor by the chiasm (Figs. 2.5, 2.7 to 2.10, and 2.13 to 2.17). The infundibular recess, however, is a true extension of the third ventricle that undergoes evolutionary modifications, ranging from that seen in the shark to that seen in mammals. The infundibular recess is a funnelshaped extension of the third ventricle extending through the hypothalamus to end in the hypophysis. The infundibular portion of the hypothalamus and its lumen, the infundibular recess, vary greatly among the different vertebrates (Figs. 2.4 to 2.20). In the lamprey, which is without an infun-

Figure 2.5. Sagittal section of the brain of a frog with barium-cast ventricles. The roof of the third ventricle and its recesses are not covered by the cerebrum. The anterior and hippocampal commissures have developed in the paraterminal body. The infundibular recess is incompletely filled, and its full extent is seen in Figure 2.13. The mesencephalic and optic ventricles are no longer a single large cavity.

Figure 2.6. Sagittal section of the brain of an iguana with barium-cast ventricles. The posterior pole of the cerebrum overlaps the roof of the third ventricle except for the dorsal sac—the homologue of the suprapineal recess. As a result the paraphysis, velum transversum, and dorsal sac are angled posteriorly before reaching the dorsal surface of the brain. The pineal body is large. Note the hippocampal commissure dorsal to the anterior commissure. The cast of the large infundibular recess was partially damaged during dissection. It extended to fill the area outlined by the arrows. 45

Figure 2.7. Sagittal section of the brain of a rabbit with barium-cast ventricles. The enlarged cerebrum covers the entire third ventricle and superior col-liculus. A new commissure, the corpus callosum, is present. Note the junctional zone (crossed arrow) where the posterior margin of the corpus blends into the subiculum dorsal to the hippocampus. The septum pellucidum area is still small. The large suprapineal recess was damaged during dissection. The direction of the long axis of the suprapineal recess changes from the iguana to the rabbit. Note the recesses of the superior and inferior colliculi, the remnants of the optic ventricle. dibulum, the homologue of the neural lobe of the hypophysis is a simple plate of tissue that lies on the floor of the diencephalon (64). In the various fishes the hypothalamus is highly developed and extends ventrally to the midbrain to form two distinct structures, the inferior lobes of the hypothalamus (Fig. 2.21). The prominent inferior lobes of the hypothalamus in the shark surround a large ependymal-linked infundibular recess (Figs. 2.4 and 2.12). An additional unique structure in fishes is the saccus vasculosus, a highly vascularized vesicle with neurosensory cells whose lumen is continuous with the infundibular recess (Figs. 2.4 and 2.12). The saccus is largest in deep sea fishes,

small in shallow water fishes, and absent in amphibians and other terrestrial vertebrates. Its function probably relates to perception of the pressure of intraventricular fluid relative to the depth of water and influences the size of the swim bladder (67). In amphibians and reptiles the hypothalamus is a relatively small structure consisting largely of the tuber cinereum (55) (Figs. 2.22 and 2.23). The small size is a result of the absence of the saccus vasculosus and the limited development of gustatory impulses (30). The largest hypothalamus is present in teleosts and mammals. In mammals the tuber cinereum may be divided into median and lateral eminences. Of note is

Figure 2.8. Sagittal section of the brain of a cat with barium-cast ventricles. The corpus callosum is further developed, especially posteriorly, where the sple-nium overlaps the suprapineal recess. Note the anteroposterior axis of the supra-pineal recess and its curvature around the splenium (arrouj). The fornix is elongated and arched under the corpus callosum. The corpus callosum is thin. The presence of the premamillary and postmamillary recesses relates to the variable development of the mamillary bodies.

the prominent size of the infundibular recess in many nonhuman vertebrates (Figs. 2.4 to 2.7). The hypophyseal recess in the cat (Fig. 2.16) is of interest because it represents the residual cavity of the infundibular recess within the developing neurohypophysis. The hypophyseal recess has been observed by Levinger and Edery (42) and apparently is a frequent finding in adult cats (35). Only two cases of persistence of the infundibular recess in the adult human have been reported (10, 35). The varied appearance of the mammalian infundibular recess and the presence of premamillary and postmamillary recesses are manifestations of the varied development of the components of the hypophysis and the size of the mamillary bodies (Fig. 2.9). In the cat and the dog the pars distalis of the hypo-

physis, originating from Rathke's pouch, extends much more caudally around the pars neur-alis than in the human (56). The pars intermedia, which in the human is limited to the rostral aspect, surrounds the pars neuralis completely in the dog and cat. The mamillary body is probably premordial in teleosts and amphibians. It can be identified in reptiles, birds, and mammals. In the rat it is a band of gray matter continuous in the midline. In humans the mamillary body is relatively larger than in other primates. The posterior inferior recess—demonstrated by Westergaard (84) in the guinea pig, the hamster (85), and the rat (86)—seems to be a manifestation of the previously mentioned varied development of the hypophysis and mamillary body.

Figure 2.9. Sagittal section of the brain of a dog with barium-cast ventricles. The corpus callosum is larger. Note the hippocampus, which is present ventral to the splenium and fornix. The suprapineal recess, the pineal body, and the colliculi are in a more ventral position.

Figure 2.10. Sagittal section of brain of a monkey with barium-cast ventricles. The corpus callosum is increased in size. The septum pellucidum is enlarged. The cavum septi pellucidi is partially filled with barium. The pineal body is wedged between the splenium and the corpora quadrigemina. The smaller suprapineal recess is associated with the larger splenium. The optic chiasm is large.

Figure 2.11. Sagittal section of the brain of adult human. The corpus callosum is markedly enlarged and arched. The splenium is posterior to the thalamus. Note the relative reduction in the size of the quadrigeminal plate.

Figure 2.12. Lateral roentgenograms of a shark (A) and of the brain of a shark (B) with barium-cast ventricles. The ventricular system has a linear arrangement. The intraventricular foramen is large. Contrast (arrow) has leaked into the ventral aspect of the medulla.

Figure 2.13. Lateral roentgenograms of a frog (A) and of the brain of a frog (B) with barium-cast ventricles. The ventricular system is still linear with the entire lateral ventricle anterior to the interventricular foramen, which is small.

Figure 2.14. Lateral roentgenograms of an iguana (A) and of the brain of an iguana (B) with barium-cast ventricles. The linearity of the ventricular system is no longer present and the lateral ventricles are dorsal to the third ventricle. Note the molding of the lateral ventricle with the presence of frontal and temporal horns.

Figure 2.15. Lateral roentgenograms of a rabbit (A) and of the brain of a rabbit (B) with barium-cast ventricles. Note the large suprapineal recess and massa intermedia. In the midbrain of the shark, an acoustico-lateral line and trigeminal fibers, nuclei of the trochlear nerve, and a few fiber tracts are present. In amphibians the midbrain is less devel-

oped than in bony fish. In reptiles, midbrain structures are demonstrated, suggestive of mammalian differentiation. The reticular formation in reptiles has changed to a nuclear mass that

Figure 2.16. Lateral roentgenograms of a cat (A) and of the brain of a cat (B) with barium-cast ventricles. The suprapineal recess is in a more caudal position as a result of enlargement of the splenium of the corpus callosum. The central cavity of the hypophysis is filled via the infundibular recess.

Figure 2.17. Lateral tomographic roentgenogram of a monkey (A) and a roent-genogram of the brain of a monkey (B) with barium-cast ventricles. Note the small size and caudal position of the suprapineal recess. The optic recess is very small because of the large chiasm and was visualized only by tomography.

Figure 2.18. Ventral views of the brains of shark (A), frog (B), rabbit (C), cat (D), and sheep (£) demonstrating evolutionary modifications in the relationship of the chiasm, hypothalamus, and hypophysis. The hypothalamus is no longer visible in the cat and sheep.

constitutes a primordial red nucleus. In mammals the red nucleus becomes a prominent structure. The cerebral peduncles increase progressively as the corticopontine, corticobulbar, and corticospinal tracts enlarge during cerebral evolution. The substantia nigra also enlarges progressively in mammals. The foregoing evolutionary changes are maniFigure 2.20. Diagram of the third ventricle and neighboring structures of a lizard to show the various structures of the

Figure 2.19. Schema to show probable evolutionary changes in the vertebrate pituitary gland. Arrow extending from the median eminence to the pars distalis represents the hypophyseal portal system. CAU, caudal division of the avian interior lobe; CEP, cephalic division of the same; PPD, proximal pars distalis; RPD, rostral pars distalis; VL, ventral lobe of the elasmo-branch pituitary; SV, saccus vasculosus. (From Romer AS: The Vertebrate Body. Philadelphia, WB Saunders Co, 1970, ed 4.)

roof including the dorsal (parietal) eye. (From Romer AS: The Vertebrate Body. Philadelphia, WB Saunders Co, 1970, ed 4).

Figure 2.21. Lateral view of the brain of a shark demonstrates the major subdivisions of the brain. The roof of the diencephalon is visible and is interlaced by a prominent venous plexus.

Figure 2.22. Lateral view of the brain of a frog shows the linearity of the major subdivisions of the brain. The forebrain (cerebrum) demonstrates a small cerebral hemisphere (arrow), which begins to overlap the diencephalon. The relatively small hypothalami of the amphibians and reptiles consist largely of the tuber cinereum.

Figure 2.23. Lateral view of the brain of an iguana. The linear arrangement of the subdivisions of the brain is no longer present. The forebrain (cerebrum) is increased in size and dorsally overlaps the diencephalon. Only the ventral hypo-thalamus and hypophysis are still visible.

fested in the increased size of the midbrain ventral to the mesencephalic ventricle and aqueduct. In the shark and the frog the floor of the midbrain is sagittally smaller than either the optic lobes or the spinal cord (Figs. 2.4 and 2.5). In the iguana

the floor of the midbrain is enlarged and is greater than the optic lobes in sagittal height (Fig. 2.6). The mammalian midbrain is greatly increased in size and is even larger sagittally than the spinal cord (Figs. 2.7 to 2.11).

Roof and Posterior Wall In the shark (Figs. 2.4 and 2.12) the following structures are identifiable: paraphysis, velum transversum, dorsal sac, pineal recess, and ha-benular and posterior commissures. The paraphysis and velum transversum are choroid plexus structures (73). The paraphysis is an extraventricular choroid plexus. In the shark it is a dorsal evagination with origin in the anterior end of the roof of the third ventricle (Fig. 2.4). The paraphysis curves around the posterior pole of the cerebral hemisphere and its cavity communicates freely with the third ventricle (Fig. 2.4 and 2.12). The velum transversum is a modified choroid plexus that originates in the roof of the third ventricle, projects into the ventricle, and serves as an arbitrary border between the diencephalon and the telencephalon (Fig. 2.4). In the frog the paraphysis is a prominent extraventricular structure (Fig. 2.5). It is cone-shaped and protrudes dorsally from the anterior end of the roof of the third ventricle (73). The numerous villi of the intraventricular choroid plexus are attached to the roof of the third ventricle around the narrow orifice of the paraphysis. These villi are seen as filling defects in casts of the ventricles (Figs. 2.5 and 2.13). A distinct velum transversum does not exist in the frog (28). The dorsal sac (saccus dorsalis) is the thin-walled roof of the third ventricle in front of the habenulae. It is also called the parencephalon (30). Whether the term parencephalon includes the velum transversum and paraphysis is not clear. The dorsal sac is a large, distensible structure in the shark (Figs. 2.4 and 2.12). The present investigation did not reveal a dorsal sac in the frog. This absence may relate to the relatively thick roof of the third ventricle, which seems to be secondary to the habenular commissure. Herrick (24) demonstrated the habenular commissure as forming the roof of the third ventricle between the paraphysis and the posterior commissure. Comparative investigation suggested that the dorsal sac in the shark is the homologue of the human suprapineal recess (33). The pineal and the dorsal sac—suprapineal recess—maintain their adjacent relation from the shark to the human. What does change is the direction of the long axis of these two structures because of overlapping of the third ventricle by the cerebrum and corpus callosum. In the shark the dorsal sac has a ventrodorsal axis (Fig. 2.4). In the iguana

the proximal end of the dorsal sac is angled posteriorly by the posterior pole of the enlarged cerebrum (Fig. 2.6). The axis of the large supra-pineal recess in the rabbit is more posteriorly directed (Fig. 2.7 and 2.15). In the cat and the dog, because of the great enlargement of the corpus callosum, the long axis of the suprapineal recess has an anteroposterior direction (Figs. 2.8, 2.9, and 2.16). The shape and size of the suprapineal recess are related to the size of the corpus callosum. The increasing size of the splenium, as seen from the cat to the monkey, is associated with a concomitant reduction of the suprapineal recess (Figs. 2.8 to 2.10, 2.16, and 2.17). The habenulae are relatively large, paired structures in vertebrates with a small forebrain (67). The habenulae are especially developed in vertebrates that rely heavily on olfaction, such as sharks and bloodhounds. In humans, because of the development of the neocortex and the relative unimportance of olf action, the habenulae are the smallest. The habenulae may have an important function in neonatal feeding reactions. An unusual feature in the central nervous system is the asymmetry of the habenulae. In sharks and frogs the left habenula is larger. In birds and mammals, habenular symmetry is variable. The pineal gland in the shark is a long and slender structure whose distal end is situated in a depression in the cartilaginous brain case (Fig. 2.4). In the frog the pineal is a long slender nerve structure extending from the roof of the diencephalon through the parietal foramen in the roof of the skull. It terminates in the frontal organ, situated under the calvarial skin (80). The frontal organ in the frog is a poorly differentiated parietal eye (67). Jolie (28) and van de Kamer (80) demonstrated a small pineal cavity situated within the diencephalon of the frog. This location may explain why the pineal recess was not seen in the studied specimens. In the iguana the striking evolutionary changes of the reptilian brain are demonstrated. The cerebrum has increased to such a degree that it completely overlaps the roof of the third ventricle (Figs. 2.6, 2.24, and 2.25). Only the distal end of the dorsal sac and the dorsal tip of the pineal gland remain visible on the dorsal aspect of the brain. In the shark the paraphysis curves anteriorly and the dorsal sac has a straight dorsal course (Fig. 2.4). In the iguana these two structures are angled posteriorly around the posterior

Figure 2.24. Lateral view of the brain of a rabbit. In the rabbit the neopallium of the cerebrum has enlarged to such a degree that it dorsally overlaps the diencephalon and mesencephalon. None of the diencephalic structures are visible. pole of the cerebral hemisphere (Fig. 2.6). Of note is the massive size of the pineal. The pineal undergoes complex evolutionary changes closely related to changes of the parietal eye (also called the median or dorsal eye). A median eye situated on the forehead and directed dorsally was apparently present in ancestral bony fishes, amphibians, and reptiles (64) (Fig. 2.20). Some of these ancestral forms may have had two median eyes. In modern vertebrates the parietal eye is present in the lamprey, some fishes and lizards, and the sphenodon (a surviving Mesozoic lizard) (64). Although a miniature cornea and retina may be present, the parietal eye—covered by connective tissue and epithelium—is capable of little except detection of light. Romer postulated that the ancestral vertebrates were presumably bottomdwelling mud strainers in which the parietal eye provided vital protective information (64). Vertical vision lost its importance for the more active modern vertebrates. It also has been theorized that the parietal eye-pineal complex was associated with thermoregulation (63).

The parietal eye, when present, develops from the distal end of the pineal body. Since the pineal originates as an evagination from the roof of the diencephalon, the parietal eye, like the lateral eyes, is essentially a structure of the brain. Occasionally the parietal eye develops from another vesicular structure adjacent and similar to the pineal, the parapineal organ. In the sphenodon and modern lizards the parapineal organ is the functional eye. In the lamprey both the pineal and the parapineal form parietal eyes, of which the pineal eye is dominant. The complex and controversial anatomy and terminology of the pineal and parapineal bodies and the parietal eye are reviewed by Oksche (50). The lumen of the pineal gland, the pineal recess, becomes progressively reduced in the ascending vertebrate series. This reduction results from cellular proliferation and leads to the compact glandlike structure in mammals. In some bony fish and turtles the lumen of the pineal does not communicate with the third ventricle. The pineal is absent in certain vertebrates, including the

Figure 2.25. Dorsal views of brains demonstrate evolutionary changes at the roof of the third ventricle. A. Shark. The highly vascular roof of the third ventricle (diencephalon) is uncovered. Note that the veins of the cerebrum are anterior to the diencephalic veins.

crocodile and some whales and dolphins. The external form of the pineal body varies greatly among the different mammals (30, 50, 63, 81, 82). It is elongated in guinea pigs, club-shaped in rabbits, round in goats, conical in carnivores, and pear-shaped in cattle and humans (Figs. 2.6 to 2.10). The mechanism of formation of the double-layered tela choroidea (velum interpositum) in the human (7, 17) is diagrammed in Figure 2.26. This process is clarified when considered in evolutionary terms, as demonstrated by the shark, frog, iguana, and rabbit (Figs. 2.21 to 2.25). In the shark the linear arrangement of the

major subdivisions of the brain is demonstrated (Fig. 2.21). The cerebrum is anterior to the diencephalon. The entire roof of the diencephalon is visible and is comprised of a single layer of pia mater (Figs. 2.21 and 2.25A). The linear arrangement of the brain is still present in the frog (Figs. 2.22 and 2.25B). A rudimentary cerebral hemisphere, however, is beginning to overlap the diencephalon (Fig. 2.22). The pronounced changes of the reptilian brain are shown in Figure 2.23. The linear arrangement of the subdivision of the brain is no longer present. The cerebrum has increased to such a degree that it overlaps the diencephalon, which is no longer

Figure 2.25B. Frog. The roof of the third ventricle is still uncovered. visible (Figs. 2.23 and 2.25C). In the rabbit the cerebrum overlaps the diencephalon and mes-encephalon (Fig. 2.24). This overlapping produces a double-layered tela choroidea. The human embryological changes that result in the formation of a double-layered tela choroidea are an ontogenic "recapitulation" of the phylogenetic process previously described. In their anatomical and radiographic studies, Larroche (36), Larroche and Baudey (37), and Baudey-Pasquier and Larroche (6) considered the

cavum vergae, which is situated below the corpus callosum and above the fornix, as the posterior continuation of the cavum septi pellucidi. In the material of Larroche, a cavum vergae was never present without a cavum septi pellucidi (6, 37). The cavum vergae seemed to develop ontogenetically later than and disappeared before the cavum septi pellucidi. The disappearance of the cavum vergae by the time of birth may be explained by several factors demonstrated in a comparative study (33). The progressive enlargement of the splenium as seen in the

Figure 2.25C. Iguana. The forebrain (cerebrum) is increased in size and overlaps the roof of third ventricle, which is no longer visible. The only diencephalic structures visible are the distal ends of the dorsal sac and pineal gland. rabbit, cat, dog, and monkey (Figs. 2.7 to 2.10) is "recapitulated" in the human fetus (Fig. 2.27). Kappers et al. (30) also noted that the posterior thalamus enlarges progressively in the ascending phylogenetic series. The thalamic enlargement increases the dorsal arching of the fornix. All of these factors would contribute to the obliteration of the cavum vergae in the older human fetus while the cavum septi pellucidi is maintained.

A comparative study demonstrated that the human aqueduct is a residual cavity that results from phylogenetic modifications of the walls surrounding the ventricular system of the midbrain (33). The roof (tectum) of the midbrain is the primary visual center in nonmammalian vertebrates. The presence of the visual center in the tectum would seem to have been the reason for the evolution of this region into a major associa-

Figure 2.25D. Rabbit. The cerebrum covers both the diencephalon and mesen-cephalon.

tion center (64). The tectum in fishes and amphibians is the dominant brain center; it associates visual, olfactory, lateral line, and other somatic sensory stimuli with the motor columns of the brain stem and cord. In the lamprey the nerve structures of the eye are developed and the optic tract ends in a discernible tectum. Although the tectum is still partly ependymal, it already has enlarged into a pair of recognizable optic lobes. The two optic lobes (corpora bigemina) are the homologues of the superior colliculi. In the shark and the frog the optic lobes are prominent structures (Figs. 2.4, 2.5, 2.21, and 2.22). In the iguana the cerebrum is beginning to

become the dominant structure and the optic lobes are relatively reduced in size (Figs. 2.6 and 2.23). In some reptiles a pair of auditory lobes, the homologues of the inferior colliculi, are present caudal to the optic lobes. The auditory lobes, which receive fibers from the cochlea, are not present in fishes that lack a cochlea. In amphibians the cochlea is small and the auditory center is not large enough to form a bulge on the surface of the mesencephalon (31). At some critical evolutionary stage a supraseg-mental area, more elaborate than the optic lobes, became necessary for the integration of refined visual perceptions (79). The evolution of the cerebrum provided a more elaborate suprasegmen-

Figure 2.26. Diagram to demonstrate the formation of the double layer of the tela choroidea (velum inter-positum) in the human. 1. The developing neural tube includes the forebrain (FB), which is surrounded by primitive pia mater (broken line). 2 and 3. The expanding vesicle of the cerebral hemisphere (HV) carries its own pia mater (b), which overlaps the pia mater (a) of the third ventricle. IF, interventricular foramen. 4. A double layer of pia is interposed between the two cerebral vesicles and the roof of the third ventricle (3V). E, ependyma. 5. The two layers of pia (b, a) over the roof of the third ventricle persist after connection of the hemispheres by the commissures. This process "recapitulates" the phylogenetic overlapping of the diencephalon by the cerebrum demonstrated in Figures 2.12 to 2.17 and 2.21 to 2.25. (From Brash JC (ed): Cunningham's Textbook of Anatomy. New York, Oxford University Press, 1951, ed 9. Modified from Frazer JE: A Manual of Embryology. New York, William Wood & Co, 1932.) tal region capable of integrating refined visual perceptions such as color, form, size, and detection of motion and distance. In mammals most of the visual, auditory, and other somatic sensations—instead of being integrated in the mid-brain—are relayed via the thalamus to the cerebral hemispheres. The thalamus in nonmammalian vertebrates is a smaller anterior extension of the sensory association area of the mid-brain. It reaches its maximal development in mammals as a result of its function as a relay center to the association centers in the cerebral hemispheres. In mammals the optic lobes differentiate into the superior and inferior colliculi (corpora quadrigemina). The cephalic shift in mammals is demonstrated in the inverse relationship in the sizes of the cerebrum and corpora quadrigemina. The increase in the size of the cerebrum from the rabbit to the human is associated with a relative reduction in the size of the corpora quadrigemina (Figs. 2.7 to 2.11). Furthermore, the relative sizes of the superior and inferior colliculi also change. In the rabbit the superior colliculi are much larger than the

inferior colliculi (Fig. 2.7), in the cat the difference between collicular sizes is diminished (Fig. 2.8), and in the dog and monkey the superior and inferior colliculi are of approximately equal size. Previously the superior colliculi were considered to be exclusively an optic structure and thus the homologue of the optic lobes. The inferior colliculi were considered to be auditory centers that developed as a result of the evolution of the hearing function of the ear. This interpretation of collicular function has been modified as other connections than purely visual and auditory have been demonstrated in the colliculi (83). The human fetus demonstrates a "recapitulation" of the interrelationship in the sizes of the tectum and cerebrum (Fig. 2.27). As the cerebrum enlarges, particularly the occipital lobe, the large tectum is transformed into the relatively smaller corpora quadrigemina. The ventricular system of the midbrain, from the fish to the primate, undergoes a pronounced transformation. In the shark the walls of each optic lobe surround a large optic ventricle, a dorsal expansion of the mesencephalic ventricle (Figs. 2.4 and 2.12). The mesencephalic and optic ventricles form a large continuous cavity. In the frog the ventricular system of the mid-brain is smaller as a result of increased thickness of the midbrain walls (tegmentum) (Figs. 2.5 and 2.13). A constricted communication demarcates the dorsolateral optic ventricles from the mesencephalic ventricle. In the iguana, as a result of the enlarged tegmentum and peduncles, the mesencephalic ventricle joins the third ventricle at a more acute angle than in the frog (Figs. 2.6 and 2.14). In the rabbit the mesencephalic ventricle demonstrates some of the features of the definitive aqueduct. As a result of the enlargement of the tegmentum and peduncles, the floor assumes a dorsally convex curve. The rostral end of the ventricle, molded by the superior colliculis and tegmentum, has the configuration of the anterior part of the aqueduct (Figs. 2.7 and 2.15). The recesses of the superior and inferior colliculi are the remnants of the optic ventricles. The definitive aqueductal configuration may be seen in the cat (Figs. 2.8 and 2.16). The aqueductal ampulla, the widest segment of the mammalian aqueduct, is the remnant of the mesencephalic and optic ventricles. Further minor modifications in the contours of the mammalian aqueduct are the result of enlargement of the cerebellum, tegmentum, and peduncles associated with reduction in size of the corpora quadrigemina and increased angulation of the brain (46).

Figure 2.27. Sagittal sections of human fetal heads demonstrate the "recapitulating" change in the relative size of the midbrain tectum and cerebrum. As the cerebrum enlarges, particularly the occipital lobe, the large tectum is transformed into the relatively smaller corpora quadrigemina. Also note the enlarging corpus callosum. A. 11 weeks. The posterior margin of the thin occipital lobe does not extend posteriorly to the relatively large midbrain. Note the size of the tectum relative to cerebellar size. From Kier EL: The cerebral ventricles: A phylogenetic and ontogenetic study. In Newton TH, Potts DG (eds): Radiology of the Skull and Brain, Vol. 3, Anatomy and Pathology. St. Louis, The CV Mosby Co, 1977, pp 2787-2914.

Figure 2.27B. 15 weeks. The occipital pole of the cerebrum extends posteriorly between the tectum and the skull. The tectum is still larger than the cerebellum.

Figure 2.27C. 24 weeks. The occipital pole is behind the cerebellum, which is now larger than the tectum.

Figure 2.27D. 30 weeks. Note the relatively small corpora quadrigemina. The splenium is well developed. A cavum vergae is present between the corpus callosum and the fornix. Commissures The literature dealing with the evolution of the commissures is not always easy to understand. Occasionally the descriptions and terminology are conflicting. In this section, I describe aspects relevant to the third ventricle. The commissures are part of the limbic system (lobe). Many of the structures of the limbic system (Fig. 2.1) are phylogenetically old and are topographically interposed between the dien-cephalon and the neocortex of the cerebral hemispheres (83). The fibers of the anterior commissure in the human connect with the olfactory tracts and the piriform, prepiriform, and amygdaloid bodies. The fornix is the efferent fiber system of the hippocampi, and the corpus callosum joins the neopallial regions of the cerebral hemispheres. The location, size, and interrelation of the anterior commissure, fornix, and corpus callosum are understood best in light of the evolutionary modifications of the cerebrum.

The cerebrum first evolved as an olfactory organ. In the lamprey the primordia of the cerebral hemispheres, manifested by the presence of bilateral vesicular outpouchings of the forebrain, are united by the lamina terminalis. Cells from the two adjacent medial portions of the olfactory lobes invaded the lamina terminalis. Thus the two olfactory nuclei were connected by a thickened lamina terminalis through which passed the anterior commissure (27). The term para-terminal body was introduced by Smith (75) to include the septal area and the medial olfactory ganglionic mass. The septal area consists mainly of nuclear masses of gray matter anterior and superior to the lamina terminalis and anterior commissure (83). The septal area superior to the anterior commissure (supracommissural) corresponds to the area of the septum pelluci-dum. During the evolutionary growth of the cerebrum, the hippocampal formation (archipallium) and the neopallium evolved dorsal to the olfactory cortex (Fig. 2.5). At the same time the fornix

(the commissures of the hippocampus) and corpus callosum evolved dorsal to the anterior commissure (Figs. 2.6 and 2.7). In the lamprey the lamina terminalis delineates the anterior end of the brain. An anterior commissure connecting the striatal areas and a primordial hippocampus are already present (26). The rudimentary hippocampus is situated above the interventricular foramen and receives second order olfactory fibers and ascending fibers from the hypothalamus. The entrance of these fiber tracts into the rudimentary hippocampus establishes it as a correlating center for olfactory and visceral impulses. In the shark the primordial hippocampus is larger, occupies the roof of the lateral ventricle, and extends along the upper border of the wall of the third ventricle (Fig. 2.4). According to Johnston (26), the fiber tracts, homologous to the fornix, descend from the primordial hippocampus through the mediorostral and ventral wall of the cerebrum and pass between the lateral ventricles to the hypothalamus.

In the frog the primordial hippocampus forms the entire length of the dorsomedial wall of the cerebral hemisphere (23). Johnston (26) and Her-rick (24) described a hippocampal commissure, dorsal to the anterior commissure, that projected into the floor of the third ventricle of the amphibian salamander (Fig. 2.5). The hippocampal commissure is the homologue of the fornix and in the amphibian passes behind the interventricular foramen. In reptiles most of the primordial hippocampus has been transformed into a specialized cerebral cortex that occupies the dorsomedial aspect of the entire cerebral hemisphere (23, 26, 27). Within the hippocampus, fibers arise to form a tract on its ventricular surface. This tract, the homologue of the mammalian fornix, passes rostral to the interventricular foramen and forms a commissure that crosses the lamina terminalis dorsal to the anterior commissure (Fig. 2.6). In monotremes such as the platypus and spiny anteater, which belong to the lowest subclass of mammals, the hippocampus is still large and

Figure 2.28. Midsagittal sections of the head of a rabbit (A), dog (B), and monkey (C) demonstrating the relationship between the increasing size of the cerebrum and corpus callosum. The genu and splenium are present in the dog and are increased in size in the monkey. Note the changing facial-neurocranium ratio from the rabbit to the monkey. Compare with the predominant neurocranium in the human fetus (Fig. 2.27).

Figure 2.28B

Figure 2.28C

assumes the elongated and arched contours of the cerebral hemisphere. In the platypus the posterior limb of the fornix assumes the curved shape of the hippocampus (74). In the spiny anteater the posterior limb of the fornix is better developed and corresponds to the larger temporal segment of the hippocampus. Thus already in monotremes the fornix extends from the temporal lobe to the mamillary body and is curved by the large mammalian thalamus. The arch of the fornix increases in the ascending mammalian scale as a result of the progressive enlargement of the thalamus. A major transformation of the mammalian brain is the progressive enlargement of the cerebrum. The increase in size of the mammalian cortical neopallium in placental mammals is accompanied by a new commissure, the corpus callosum (Figs. 2.7 to 2.11). Nieto et al. (48) demonstrated a linear correlation between surface area of the corpus callosum and the brain weight of different animals (Fig. 2.28). The only exception was in the dolphin. In this mammal the corpus callosum is comparatively small in spite of relatively massive development of the cerebral hemispheres. They postulated that the lack of extremities in cetaceans may account for the relative smallness of their corpus callosum. The appearance of the corpus callosum first in placental mammals may relate to expansion of the associated areas of the neocortex (67). Smith (74), Johnston (27), and Abbie (4) demonstrated that the evolutionary increase of the corpus is associated with a concomitant decrease of the hippocampal formation in the roof and medial wall of the lateral ventricle (Fig. 2.29). To state that the corpus callosum replaces the hippocampus in the roof of the body of the lateral ventricle is not strictly accurate. The concept of replacement, substitution, or suppression, however, facilitates understanding of the topographic relation of the corpus callosum, fornix, and hippocampus. An example of the "replacement" is seen in the rabbit, where the posterior margin of the evolving corpus blends with the anterior margin of the remaining hippocampus (Fig. 2.7). Smith (75) stated that, "in all mammals the cerebral cortex is fringed with some structure representing a hippocampal formation along the whole extent of its mesial edge—from the olfactory peduncle in front to the region of the uncus behind and below" (Fig. 2.30). Smith thought that the indusium griseum and the longitudinal striae were vestigial hippocampal structures. Kappers et al. (30) further elaborated this theory. Abbie (4) demonstrated that the indusium gri-

seum is not a remnant of the hippocampus but is formed by eversion of a lip of the subiculum. The subiculum is transitional (mesocortex) cerebral tissue at the junction of the hippocampus and the neocortex (3). The evolution of the corpus callosum was divided by Abbie (4) into the following stages (Fig. 2.29): (a) reduction in the size of the hippocampal segment adjacent the lamina terminalis, (b) piercing of the subiculum adjacent to the lamina terminalis by fibers of the corpus callosum, (c) enlargement of the corpus callosum adjacent to the lamina terminalis, (d) formation of the sple-nium by linear caudal expansion, and (e) formation of the genu and septum pellucidum by frontal arciform expansion. These stages are diagrammed in Figure 2.29 and are demonstrated in dissected specimens in Figures 2.7 to 2.10. In monotremes, such as the platypus and the spiny anteater, the paraterminal body contains the lamina terminalis, a large anterior commissure, and dorsal to it a smaller round hippocampal commissure. In the marsupial a large hippocampus on the medial aspect of the cerebral hemisphere extends anterior to the lamina terminalis (precommissural hippocampus) and posteriorly (postcommissural hippocampus) into the temporal lobe (Fig. 2.29A). The anterior commissure is still a large structure. The commissure of the fornix (inferior fornix) has a primordial elongated shape. A postulated intermediate phase demonstrates the first stage in the formation of the corpus callosum (Fig. 2.29B). The hippocampus adjacent to the lamina terminalis is reduced in size secondary to infolding. The hippocampal infolding results in approximation of the subiculum and lamina terminalis. The broken line in the subiculum indicates the point at which the fibers of the corpus callosum pierce the subiculum. In the hedgehog and the bat the fibers of the corpus callosum break through the subiculum (Fig. 2.29C). With the appearance of the corpus callosum, the hippocampus disappears almost completely from the region dorsal to the commissures in the lamina terminalis. The anterior commissure is reduced, the inferior fornix is enlarged, and a new hippocampal fiber tract (the superior fornix) appears in the region of the lamina terminalis. In the rat the corpus callosum is enlarged dorsal to the lamina terminalis (Fig. 2.29D). This enlargement is associated with a further reduction of the precommissural and postcommissural hippocampi, which are ventral to the corpus cal-

Figure 2.29. Diagram of the medial aspect of a cerebral hemisphere to demonstrate the evolution of the corpus callosum and septum pellucidum. A, marsupial; B, postulated intermediate phase; C, hedgehog and bat; D, rat; E, rodent; F, primate. Expansion of the corpus callosum (c.c.) is associated with concomitant reduction in size of the hippocamus (hi.). Elongation of the superior (f.s.) and inferior (fi.) fibers of the fornix is related to growth of the splenium (s.) of the corpus callosum. The septum pellucidum (si.) results from thinning of the lamina terminalis (lt.) secondary to the arched expansion of the genu of the corpus callosum; ca., anterior commissure; pt. b., paraterminal body; hi. p., precommis-sural hippocampus; r., rostrum; su., subiculum. (From Abbie AA: J Comp Neural 70:9-44, 1939.)

Figure 2.30. Diagrammatic across sections of the left cerebral hemisphere show the evolutionary changes in the position of the hippocampus. A. In the primitive stage the cerebrum is essentially an olfactory lobe composed only of the paleopallium. The hippocampus (archipallium) is not present. B. In the amphibian the hippocampus forms the medial wall of the hemisphere. С and D. In early and late reptilian stages the hippocampus is medial and superior to the lateral ventricle. In the late reptilian stage (D), the neopallium is still rudimentary. E. In the primitive mamallian stage, the neopallium has expanded and the hippocampus is rolled in on the medial surface of the hemisphere, adjacent to the lateral ventricles. F. In the higher mammals the neopallium has markedly expanded. The rolled-in hippocampus is separated from the ventricle by the corpus callosum. a, archipallium (hippocampus); b, basal nuclei; cc, corpus callosum; n, neopallium; p, paleopallium; v, ventricle. (From Romer AS: The Vertebrate Body. Philadelphia, WB Saunders Co, 1970, ed 4.) losum. The inferior and superior fornices are elongated. Similar changes are seen in the rabbit (Fig. 2.7). In another rodent the stage of caudal expansion of the corpus callosum is demonstrated (Fig. 2.29E). This expansion leads to formation of the splenium. The caudal expansion of the corpus callosum is a simple linear extension and carries with it the fibers of the fornix, which become increasingly elongated. Both the splenium and the commissure of the fornix remain dorsal to

the hippocampus. These changes are demonstrated in the cat and dog (Figs. 2.8 and 2.9). The last stage of development of the corpus callosum involves the region of the genu. Progressive expansion of the genu associated with formation of a septum pellucidum is seen in the cat and the dog (Figs. 2.8 and 2.9). The frontal expansion of the corpus callosum reaches its maximal size in primates (Figs. 2.10 and 2.11). The development of large frontal lobes in primates is associated with the development of the large arched genu and a large septum pellucidum (Figs. 2.29F). Thompson (78) reviewed the controversial literature dealing with the presence of a cavum septi pellucidi and its open status in a large number of mammals such as ungulates, carnivores, and primates. Johnston (27) postulated that the cavum septi pellucidi was the result of trapping and roofing by the genu of a cavum formed by lamina of the paraterminal body and open dorsally into the superior longitudinal fissure (Fig. 2.31). Abbie (4) pointed out that a cavum opening dorsally into the superior longitudinal fissure has never been demonstrated in vertebrates without a corpus callosum. Abbie (4) attributed the formation of the septum pellucidum to the arched expansion of the genu (Fig. 2.31). The upward expansion of the genu draws up the lamina terminalis and the superior fornix into the concavity of the arch, thus forming the septum pellucidum. In the cat the septum is still solid. The tension produced by further arched expansion of the genu results in condensation of the septum into two laminae with a cavum between them. Abbie (4) postulated that similar upward expansion in the region of the splenium cannot take place because of the restraining fibers of the columns of the fornix (62). According to Sarnat and Netsky (67) the leaves of the septum pellucidum are composed of neural tissues and the cavum septi pellucidum is an ependyma-linked cavity deriving from the lamina terminalis. Rakic and Yakovlev (62) considered the cavum septi pellucidi as a dorsally open pocket that is open at birth. Later it is sealed by the rostrum of the corpus callosum. The pocket stays open in rodents, carnivores, and monkeys (rhesus). Loeser and Alvord (43, 44), Probst (60, 61), and Lemire et al. (41) reviewed the contro-versial development of the human septum pellucidum.

Figure 2.31. Diagrams of the brain of the rat to illustrate two different theories of the formation of the septum pellucidum, the indusium griseum, and the longitudinal striae. A. Johnston demonstrates a hippocampal remnant above the corpus callosum (c.c.). The septum pellucidum is formed within the hippocampal primordium (spindle-shaped spots). The paraterminal body is shaded with horizontal lines. c. h., hippocampal commissure; c.a., anterior commissure. B. Abbie demonstrates no hippocampal (black) tissue above the corpus callosum (c.c.). Instead, the subiculum (su), a mesocortical tissue, lies above the corpus callosum. The septum pellucidum area is formed by the drawing up of the lamina terminalis (l.t.).f.s., superior fornix;f.i., inferior fornix; hi.p., precommissural hippocampus; c.a., anterior commissure;f.l. (f.s.), long fornix; pt.b., paraterminal body; ol.t., olfactory tubercle. (A. From Johnston; J Comp Neurol 23:371478, 1913.) (B. From Abbie: J Comp Neurol 70:9-44, 1939.)

Arteries Neuroradiologists and neurosurgeons are constantly aware of the marked variability in angio-graphic and neurosurgical vascular anatomy. A

study of the general principles that govern the vascularization of the brain and of certain evolutionary modifications of the arterial system is helpful in understanding the vascular variability of the arteries supplying the hypothalamus and the thalamus (32). Streeter (76, 77) pointed out the following important principles of cerebrovascular develop-

ment. (a) The development of the cerebrovascu-lar system is not an independent phenomenon. (b) The vascular system constantly adapts to the changing brain. In comparative anatomical studies Shellshear (69-72) repeatedly stressed the unity of the brain and its vascular supply. He expressed the belief that earlier workers had been misled by overemphasis on the variation of the larger vessels. Because the size of the arteries served as an index of their importance, the fundamental characteristic of arterial distribution was overlooked. An analysis of arteries, in terms of their final areas of supply, showed a remarkable constancy. Shellshear divided the cerebral arteries into two major divisions: the end arteries and the blood vessels that lead to the end arteries (32). Abbie (1, 2), in studies of the comparative anatomy of the circulation of the brain, confirmed the postulates of Shellshear in regard to constancy of the cerebral arteries in their terminal supply. In his studies he reemphasized the concept that the brain and its blood supply are not independent and that both evolve simultaneously. Abbie, as well as Shellshear, pointed out that the variation in the origin and course of the vascular trunks leading to the functional end arteries is of secondary importance. This variation depends on the evolutionary direction of growth of an organ and on hemodynamic factors such as the closeness, economy of distribution, and convenience of the source of blood (Fig. 2.38). This last concept was alluded to by Evans (15), who stated, "The larger vessels are to be considered in the light of servants of the capillaries, for which they are but the delivering and draining pipes." Roofe (65), in his comprehensive study of the endocranial circulation of the salamander, emphasized the great variability in the supply of various territories of the brain. Areas of the brain that are well defined morphologically and functionally are supplied by delivering arteries whose course varies greatly. The terms nutrient, applied by Gillilan (19) to the end arteries, and conducting, applied to the major trunks that lead to the end arteries, convey the physiological aspects of the cerebral circulation. The end arteries supply a constant functional territory. This constancy is mediated through elements of the sympathetic nerves that accompany the vessels. These end arteries develop synchronously with the structures that they supply, and their evolution parallels the evolutionary changes of their territories of supply.

The central artery of the retina is the best example of an end artery. The cerebral perforating arteries of the brain communicate with each other only through the capillary bed and from a practical point of view are end arteries (83). Neuroangiography and microneurosurgery visualize only the variable "delivering (conducting) pipes" of the hypothalamus and thalamus. These are the circle of Willis and the anterior cerebral, anterior communicating, internal carotid, posterior cerebral, posterior communicating, and anterior and posterior choroidal arteries (38, 89). The perforating branches of these arteries are the end arteries supplying fairly specific territories in the third ventricular region (Fig. 2.32) (18, 34, 39, 40, 57, 58, 66). Although the hypo-thalamic branches of the superior hypophyseal artery are end arteries, the branches supplying the hypophysis and its stalk are not end arteries (47). Padget (51, 52), in her studies of the early fetal arterial system, noted that, in accordance with the five stages of cerebral blood vessel development of Streeter (77), the early process of vascular development involves changes in certain channels of the primitive vascular plexus. These changes depend on the specific needs of areas to be supplied. The needs may be permanent or transitory and result in the elaboration of permanent or transitory arteries. The great variability in the region of the anterior communicating artery is a function of these factors (38). The recurrent artery of Heubner is an example of evolutionary change of source of supply for reasons of economy of blood distribution (2). Because of the bending of the rhinal fissure, the anterior cerebral artery takes over some of the territory previously supplied by the middle cerebral artery. This territory includes a segment of the basal ganglia. The terminal arterioles of Heubner's artery never change their territory of supply; rather the conducting arteries to these arterioles are changed (32). Earlier in the evolutionary phase, the terminal arterioles of Heub-ner's artery are supplied by the middle cerebral artery. In primates they are supplied by a new conducting artery, which, because of its origin from the anterior cerebral artery, must pass laterally to reach its arterioles. The great variability in the origin of the artery of Heubner depends on the particular anastomotic channel that becomes the stem of the artery (Fig. 2.33). Other examples of variability in the vascular trunks supplying the same end arterioles are the inverse relationship between the size of the anterior and posterior choroidal arteries and the

Figure 2.32. Diagrams of the arterial territories in the third ventricular region. A. Sagittal: 1, anterial cerebral; 2, posterior cerebral; 3, anterior and posterior choroidal; 4, posterior communicating; 5, internal carotid. B. Floor: 1, anterior cerebral; 2, posterior cerebral; 3, posterior communicating; 4, internal carotid. С and D. Transverse and coronal: 1, anterior cerebral; 2, middle cerebral; 3, posterior cerebral; 4, anterior choroidal; 5, posterior choroidal; 6, posterior communicating. (From Krayenbuhl HA, Yasargil MG: Cerebral Angiography. Philadelphia, JB Lippincott, 1968. After Lazorthes G: Vascularisation et Circulation Cere-brales. Paris, Masson & Cie, 1961.)

Figure 2.33. Dissected human fetal arteries: 16-week fetus (A); 24-week fetus (B). Note the variability in the origin, size, and course of the recurrent arteries of Heubner. From Kier EL: The fetal cerebral arteries: a phylogenetic and ontogenetic study. In Newton TH, Potts DG (eds): Radiology of the Skull and Brain, Vol. 2, Book 1, Angiography. St. Louis, The CV Mosby Co, 1974. large anterior inferior cerebellar artery supplying the territory of the posterior inferior cerebellar artery or the reverse anomaly. The human circle of Willis is but one of many cerebral arterial circles that evolved in the various vertebrate groups to assure an adequate supply of blood to the brain (87). An additional possible explanation of its function in the evolve-ment of the cerebral arterial circle is to assure equal pressure in the three major cerebral arteries (88). In fish, amphibians, reptiles, and birds, the internal carotid arteries are the sole source of blood to the brain (14). In these four classes of vertebrates, the internal carotid arteries divide into cranial and caudal divisions. The caudal division unites to form the basilar artery in all four classes. The cranial divisions remain separate in fish, amphibians, and birds. In most rep-

tiles the cranial division unites to form a single artery that passes anteriorly to supply the olfactory lobes. Thus in fish, amphibians, and birds the arterial circle is open anteriorly, whereas in the majority of reptiles it is completely closed. Occasionally an anterior communicating artery may exist in certain amphibians (14). The monotremes possess definite mammalian features, such as the presence of an internal capsule and a crus cerebri and pons (22). The cerebral arterial circle of monotremes is the earliest true homologue of the human circle of Willis, for it unites the carotid and vertebral circulations and is completely closed. The term circle of Willis should be limited to the human cerebral arterial circle (22). The cerebral arterial circle of the different mammals varies greatly and can be grouped into three fundamental groups (14). These groups are

distinguished by the relative contribution of the carotid and vertebral arteries to the arterial circle. Group I The carotid and vertebral circulations participate in approximately equal manner. To this group belong humans, anthropoid apes, monkeys, insectivores (shrew, mole, hedgehog), and certain carnivores, rodents, and toothless mammals. Group II The carotid arteries are the major source of supply to the cerebral arterial circle. In this group the vertebral and basilar arteries are greatly attenuated. The monotremes, marsupials, whales, odd-toed ungulates (horse, zebra), even-toed ungulates (cattle, deer, camel, giraffe), and many land and marine (seal, sea lion, walrus) carnivores belong to this group. Group III The vertebral and basilar arteries are the major source of supply to the arterial circle. The inter-

nal carotid arteries are attenuated. To this group belong many rodents and some lemurs, bats, and toothless mammals. According to Nilges (49) the arterial circle of the guinea pig is derived entirely from the vertebral arteries. The internal carotid artery leaves the cranial cavity as the ophthalmic artery without participating in the formation of the cerebral arterial circle. In some ruminants (ox, goat, sheep, pig) and in the cat, the internal carotid artery has been replaced by a vascular plexus—the rete mirabile— which connects the cerebral arterial circle with branches of the external carotid artery (12, 16, 20). The fish brain is supplied entirely by the internal carotid artery. The internal carotid artery divides into two branches, a rostral and a caudal division, whose sizes relate to the functional preponderance of the various brain components. The rostral branch is small and divides into a medial and a lateral olfactory artery. The medial olfactory artery supplies the medial aspect of the primitive forebrain, the septal region, and the medial part of the primordial hippocampi. It is

Figure 2.34. Diagrams demonstrating the variable evolutionary origin of the posterior cerebral artery. A. In the chicken the posterior cerebral artery (a. cer. post.) originates from the rostral branch (r. rost.) of the internal carotid artery. B. In the marsupial the origin of the posterior cerebral artery is now more posterior, originating from the basilar artery. (A. From Gillilan LA: J Comp Neurol 130:175-196, 1967. B. From Abbie AA: J Anat 67:491-521, 1933.)

the homologue of the anterior cerebral artery of the higher forms. The lateral olfactory artery supplies the region of the primordia of the hy-popallium, piriform lobe, most of the hippocampus, and the paleostriatum. The caudal branch is larger, and its branches supply the dien-cephalon, optic lobes, cerebellum, and medullary centers. In amphibians the internal carotid artery divides into a cranial and a caudal division of approximately equal calibers. The medial olfactory artery of amphibians is similar to that of the fish in its course and territory of supply and is the homologue of the anterior cerebral artery. The anterior cerebral artery of reptiles is the homologue of the medial olfactory artery of fish and amphibians. In reptiles the main stream of blood has been deflected to the medial wall of the telencephalic hemisphere, probably in response to the increased demands of the new reptilian cortex. The anterior cerebral artery in birds contributes little to the supply of the forebrain. A few small branches to the medial and central aspect of the forebrain originate from the anterior cerebral artery before this vessel leaves the cranial cavity to become the ethmoidal artery. The original course of the anterior cerebral artery homologue in the lower forms is indicated in humans by a few small branches that run forward toward the olfactory bulb. The homologue of the anterior choroldal artery first appears as an entity in the reptilian class. This vessel supplies branches to the optic tract, lateral geniculate body, and amygdaloid. In marsupials the anterior choroidal artery has evolved into a complete homologue of the primate anterior choroidal artery. It arises from the internal carotid artery, courses posteriorly along the optic tract and lateral geniculate body, and terminates in the posterior cerebral artery. Along its course it sends a few small branches that reach the choroid plexus through the choroid fissure. In the sheep the course of the artery is similar to that in marsupials. The choroidal branches in the sheep, however, are much more prominent than in the lower forms. The branches of the anterior choroidal artery that supply the choroid plexus in the ape equal the size of those that supply the optic tract and lateral geniculate body. From reptiles to humans, the anterior choroidal artery maintains a close topographic relation with the optic tract and lateral geniculate body. The artery undergoes a functional change, however. From mainly a visual function in the

reptiles, its contribution to the choroid plexus becomes greater as the cerebrum progressively enlarges in the ascending evolutionary scale. The posterior cerebral artery and its homo-logues undergo significant evolutionary changes in their origin, course, and territory of supply. The diencephalic and tectal arteries of fish and amphibians (21) are homologues of the posterior cerebral artery of higher forms. They arise from the caudal division of the internal carotid artery and supply the optic lobes and the dorsomedial aspect of the primitive forebrain. The reptilian posterior cerebral artery is the true forerunner of that artery in mammals. In the various reptiles it may arise as a branch of either the cranial division of the internal carotid artery (21) or the caudal division (8); it may also arise at the bifurcation of the internal carotid artery (9), or it may represent the entire caudal division of lower forms (2). In the mammalian class the origin or source of blood for the posterior cerebral artery shifts from the carotid to the basilar artery. As the cerebral hemisphere enlarges in the ascending mammalian series, the cerebral territory supplied by the posterior cerebral artery is gradually shifted more posteriorly and overhangs the midbrain. The distance between the origin of the posterior cerebral artery and the final areas of supply is increased. For economy of distribution, the posterior cerebral artery shifts its stem progressively posteriorly (Fig. 2.34). Successive posterior anastomotic channels enlarge to convey the main stream of blood to the posterior cerebral artery. These anastomotic channels, which successively enlarge to form the stem of the posterior cerebral artery, revert to their original small size when their need as the main conduit of blood to the posterior cerebral artery disappears. In its evolutionary course posteriorly the main trunk of the posterior cerebral artery incorporates the territory of the various vessels that have become its temporary or permanent stem. In the higher mammals the posterior cerebral artery uses the anterior midbrain channels as a stem. These channels are supplied by a greatly enlarged basilar artery; eventually the posterior cerebral artery seems to arise directly from the basilar artery. The posterior choroldal arteries join the anterior choroidal artery in a rich network that supplies the optic tract, lateral geniculate body, and choroid plexus of the third ventricle. The posterior choroidal vessels increase in number and complexity in the ascending phylo-genetic scale. According to Abbie (1) the variabil-

Figure 2.35. Lateral view of arterially injected brain of an iguana. The posterior cerebral artery is seen on the surface of the brain supplying the optic lobe. Note the artery supplying the chiasm and hypophysis (arrow).

Figure 2.36. Lateral view of a dissected brain of a rabbit. The posterior cerebral artery can only be demonstrated by dissection because of the marked expansion of the neopallium. Note the relatively large size of the superior colliculus and the large homologues of the medial and lateral posterior choroidal arteries. The posterior pericallosal artery is curving around the corpus callosum to supply the roof of the third ventricle.

Figure 2.37. Lateral view of dissected brain of a cat. The colliculi and posterior choroidal arteries are relatively reduced in size. The enlarged occipital lobe is supplied by a large branch of the posterior cerebral artery (crossed arrow).

Figure 2.38. Dorsal view of the roof of the third ventricle of a dissected 24-week human fetus. Note the large posterior pericallosal artery curving around the rudimentary splenium of the corpus callosum.

ity of the major posterior choroidal branches is a function of the rich vascular network that supplies the choroid plexus. The functional end arteries of the posterior choroidal arteries remain constant; only their larger conducting trunks vary. The constancy of the end arteries is manifested in the precise correlation of their distribution in the various laminae of the lateral ge-niculate body and in the areas of the functional representation of the retina. In the iguana the posterior cerebral artery supplies the optic lobe (Fig. 2.35). In the rabbit significant evolutionary change has occurred. The posterior cerebral artery supplies not only the optic lobe but the new visual centers, the geniculate body and pulvinar. The homologues of the medial and lateral posterior choroidal arteries can be seen only by dissection because of the marked enlargement of the cerebrum (Fig. 2.36). In the cat there are further changes. The optic lobe is reduced in relative size. A large posterior cerebral artery branch to the occopital lobe is seen (Fig. 2.37). In higher vertebrates the choroidal branches will relatively diminish in size, while the cortical branches increase in size. The posterior growth of the cerebrum and corpus callosum elongates the course of the anterior cerebral artery to the roof of the third ventricle. This long course results in the inconsistent presence of the posterior pericallosal artery in the human adult. The medial posterior choroidal artery has a shorter course to the roof of the third ventricle and is rarely absent (Fig. 2.38).

Veins There are few studies dealing with the cerebral veins in nonhuman vertebrates. In the shark the forebrain veins are anterior to the diencephalic veins of the uncovered roof of the third ventricle (Fig. 2.25A). In the salamander (65) the veins of the brain are superficial and their course is exceedingly variable. The veins of the choroid plexus are wide sinusoidal channels that drain into a rete of the nodus vasculoses that permeates the paraphysis and dorsal sac. The dorsal hemispheric veins are pial and also drain into the nodus vasculoses. In the turtle (68) the venous channels are situated on the superior surface of the brain. The largest one is a median dorsal longitudinal vein that commences anteriorly near the interventricular foramen and passes posteriorly on the surface of the brain to exit from the skull near the fourth ventricle. There are no venous channels within the brain. In the dog the great cerebral vein is the main vessel of the velum interpositum. It passes ven-

tral to the corpus callosum (5). Review of their Figure 5 demonstrates a straight great cerebral vein secondary to the small splenium in the dog. A diagram of the intraventricular venous drainage of the capuchin monkey also demonstrates a straight great cerebral vein (45). These features seem to be "recapitulated" in the human embryo (29, 54). In the 40-mm embryo (21/2 months) the arterial circle of Willis and its branches are clearly defined. However, the deep venous system is still rudimentary, consisting of the primitive great cerebral vein (Galen), which leaves the membranous diencephalic roof near the pineal primordium (53). The primitive internal cerebral vein is still extracer-ebral, lying in dorsal contact with the diencephalic roof. The internal cerebral vein will become enclosed within the two leaves of the tela choroidea of the third ventricle during further growth of the cerebrum. The great cerebral vein derived its curvature during the growth of the splenium of the corpus callosum. References 1 . Abbie AA: The blood supply of the lateral geniculate body, with a note on the morphology of the choroidal arteries. J Anat 67:491-521, 1933. 2. Abbie A A: The morphology of the fore-brain ar teries, with special reference to the evolution of the basal ganglia. J Anat 68:433-470, 1934. 3. Abbie AA: The relations of the fascia dentata, hippocampus and neocortex and the nature of the subiculum. J Comp Neurol 68:307-323, 1938. 4. Abbie AA: The origin of the corpus callosum and the fate of the structures related to it. J Comp Neurol 70:9-44, 1939. 5. Armstrong LD, Horowitz A: The brain venous sys tem of the dog. Am J Anat 132:479-490, 1971. 6. Baudey-Pasquier J, Larroche JC: Les cavites cerebrales supplementaires pendant le developpement foetal et neofoetal. Acta Radiol [Diagn] (Stockh) 1:533-540, 1963. 7. Brash JC (ed): Cunningham's Textbook of Anat omy. New York, Oxford University Press, Inc, 1951, ed 9. 8. Burda DJ: Development of intracranial arterial patterns in turtles. J Morphol 116:171-188, 1965. 9. Burda DJ: Developmental aspects of intracranial arterial supply in the alligator brain. J Comp Neu rol 135:369-380, 1969. 10. Cabanes MD: Asymptomatic persistence of indundibularis recessus. J Neurosurg 49:769-772, 1978. 1 1 . Cole FJ: A History of Comparative Anatomy. London, Macmillan & Co, Ltd, 1944. 12. Daniel PM, Dawes JDK, Prichard ML: Studies of the carotid rete and its associated arteries. Philos Trans R Sac Lond Ser В 237:173-208, 1953. 13. De Beer GR: Embryos and Ancestors. New York, Oxford University, 1958, ed 3. 14. De Vriese B: Sur la signification morphologique des arteres cerebrales. Arch Biol 21:357-457, 1905.

15. Evans HM: The development of the vascular sys tem. In Kiebel F, Mall FP (eds): Manual of Human Embryology. Philadalphia, JB Lippincott Co, 1912. 16. Francke JP, Clarisse J, Dhellemmes P, Bousquet C, Francke-Mauroy B: The arterial circle of the base of the brain and its feeding vessels in certain mammals. J Neuroradiol 4:271-289, 1977. 17. Frazer JE: A Manual of Embryology. New York, William Wood & Co, 1932. 18. George AE, Raybaud C, Salamon G, Kricheff I: Anatomy of the thalamoperforating arteries with special emphasis on arteriography of the third ventricle: Part 1. AJR 124:220-230, 1975. 19. Gillilan LA: Significant superficial anastomoses in the arterial blood supply to the human brain. J Comp Neurol 112:55-74, 1959. 20. Gillilan LA, Markesbery WR: Arteriovenous shunts in the blood supply to the brains of some common laboratory animals—with special atten tion to the rete mirabile conjugatum in the cat. J Comp Neurol 121:305-311, 1963. 21. Gillilan LA: A comparative study of the extrinsic and intrinsic arterial blood supply to brains of submammalian vertebrates. J Comp Neurol 130:175-196, 1967. 22. Gillilan LA: Blood supply to primitive mammalian brains. J Comp Neurol 145:209-221, 1972. 23. Herrick CJ: The morphology of the forebrain in amphibia and reptilia. J Comp Neurol 20:413547, 1910. 24. Herrick CJ: The Brain of the Tiger Salamander. Chicago, University of Chicago Press, 1948. 25. Johnston JB: The Nervous System of Verte brates. Philadelphia, P Blakiston's Son & Co, 1906. 26. Johnston JB: The evolution of the cerebral cortex. Anat Rec 4:143-166, 1910. 27. Johnston JB: The morphology of the septum, hip pocampus, and pallial commissures in reptiles and mammals. J Comp Neurol 23:371-478, 1913. 28. Jollie MT: Chordate morphology. New York, Reinhold Publishing Corp, 1962. 29. Kaplan HA, Ford DH: The Brain Vascular Sys tem. New York, American Elsevier Publishing Co, Inc. 1966. 30. Kappers CUA, Huber GC, Crosby EC: The Com parative Anatomy of the Nervous System of Ver tebrates, including Man. New York, The Macmillan Co, 1936. (Reprinted by Hafner Publishing Co, Inc, New York, 1967.) 31. Kent GC: Comparative Anatomy of the Verte brates. St Louis, The CV Mosby Co, 1973, ed 3. 32. Kier EL: The fetal cerebral arteries, a phylogenetic and ontogenetic study. In Newton TH, Potts DG (eds): Radiology of the Skull and Brain, vol 2, book 1, Angiography. St Louis, The CV Mosby Co, 1974, pp 1089-1130. 33. Kier EL: The cerebral ventricles: A phylogenetic and ontogenetic study. In Newton TH, Potts DG (eds): Radiology of the Skull and Brain, vol 3, Anatomy and Pathology. St Louis, The CV Mosby Co, 1977, pp 2787-2914. 34. Krayenbuhl HA, Yasargil MG: Cerebral Angiog raphy. Philadelphia, JB Lippincott, 1968.

35. Kuhne D, Schwartz RB: Persisting intrapituitary recessus infundibuli. Neuroradiology 10:177178, 1975. 36. Larroche JC: The development of the central nerv ous system during intrauterine life. In Falkner F (ed): Human development. Philadelphia, WB Saunders Co, 1966. 37. Larroche JC, Baudey J: Cavum septi lucidi, cavum vergae, cavum veli interpositi: Cavites de la linge mediane. Biol Neonat 3:193-236, 1961. 38. Lazorthes G: Vascularisation et Circulation Cerebrales. Paris, Masson & Cie, 1961. 39. Lazorthes G, Salamon G: Etudes anatomique et radio-anatomique de la vascularisation arterielle du thalamus. Ann Radiol 14:11-12, 905-924, 1971. 40. Lazorthes G, Salamon G: The arteries of the thal amus: An anatomical and radiological study. J Neurosurg 34:23-26, 1971. 41. Lemire RJ, Loeser JD, Leech RW, Alvord EC Jr: Normal and Abnormal Development of the Hu man Nervous System. Hagerstown, MD, Harper & Row, 1975. 42. Levinger IM, Edery H: Casts of the cat cerebroventricular system. Brain Res 11:294-304, 1968. 43. Loeser JD, Alvord EC Jr: Agenesis of the corpus callosum. Brain 91:553-570, 1968. 44. Loeser JD, Alvord EC Jr: Clinicopathological cor relations in agenesis of the corpus callosum. Neu rology 18:745-756, 1968. 45. Madeira MC, Watanbe S: Anatomical data on the intracranial venous drainage of the tufted capu chin. Anat Rec 183:589-598, 1975. 46. Millen JW, Wollam DHM: The Anatomy of the Cerebrospinal Fluid. London, Oxford University Press, 1962. 47. Netter FH, Crelin ES: Nervous System, vol 1, part 1, Anatomy and Physiology. West Caldwell, NJ, CIBA, 1983. 48. Nieto A, Nieto D, Pacheco A: Possible phylogenetical significance of the corpus callosum with spe cial reference to the dolphin brain (Stenella graffmani). Acta Anat 94:397-402, 1976. 49. Nilges RG: The arteries of the mammalian cornu ammonis. J Comp Neurol 89:177-190, 1944. 50. Oksche A: Survey of the development and com parative morphology of the pineal organ. Prog Brain Res 10:3-29, 1965. 51. Padget DH: The circle of Willis, its embryology and anatomy. In Dandy WE (ed): Intracranial Ar terial Aneurysms. Ithaca, NY, Comstock Publish ing Co, Inc, 1944, pp 67-147. 52. Padget DH: The development of the cranial arter ies in the human embryo. Contrib Embryol 32:205-262, 1948. 53. Padget DH: The cranial venous system in man in reference to development, adult configuration, and relation to the arteries. Am J Anat 98:307356, 1956. 54. Padget DH: The development of the cranial venous system in man, from the viewpoint of comparative anatomy. Contrib Embryol 36:79-140, 1957. 55. Papez JW: Comparative Neurology. New York, Thomas Y Crowell Co, 1929 (Reprinted by Hafner Publishing Co, Inc, New York, 1967.) 56. Patten BM: Human Embryology. New York,

McGraw-Hill Book Co, 1968, ed 3. 57. Percheron G: Les arteres du thalamus humain. Rev Neurol (Paris) 132:297-307, 1976. 58. Plets C: The arterial blood supply and angioarchitecture of the posterior wall of the third ventricle. Acta Neurochir (Wien) 21:309-317, 1969. 59. Polyak S: The Vertebrate Visual System. Chi cago, University of Chicago Press, 1957. 60. Probst FP: Congenital defects of the corpus callosum: Morphology and encephalographic appear ances. Acta Radiol (Suppl) 331:1-152, 1973. 61. Probst FP: The Prosencephalies: Morphology, Neuroradiological Appearances and Differen tial Diagnosis. New York, Springer, 1979. 62. Rakic P, Yakovlev PI: Development of the corpus callosum and cavum septi in man. J Comp Neurol 132:45-72, 1968. 63. Reiter RR: The mammalian pineal gland, struc ture and function. Am J Anat 162:287-313, 1981. 64. Romer AS: The Vertebrate Body. Philadelphia, WB Saunders Co, 1970, ed 4. 65. Roofe PG: The endocranial blood vessels of Amblystoma tigrinum. J Comp Neurol 61:257-293, 1935. 66. Rosner SS, Rhoton AL Jr, Ono M, Barry M: Microsurgical anatomy of the anterior perforating ar teries. J Neurosurg 61:468-485, 1984. 67. Sarnat HB, Netsky MG: Evolution of the Nervous System. New York, Oxford University Press, Inc, 1981, ed 2. 68. Schepers GWH: Evolution of the Forebrain. Cape Town, Maskew, Miller Ltd., 1948. 69. Shellshear JL: The basal arteries of the forebrain and their functional significance. J Anat 55:2735, 1920. 70. Shellshear JL: The arteries of the brain of the orangutan. J Anat 61:167-197, 1927. 71. Shellshear JL: The arterial supply of the cerebral cortex in the chimpanzee (Anthropopithecus troglodytes). J Anat 65:45-87, 1930. 72. Shellshear JL: A study of arteries of the brain of the spiny anteater (Echidna aculeata) to illus trate the principle of arterial distribution. Philos Trans R Soc Lond Ser В 218:1036, 1930. 73. Shuangshoti S, Netsky MG: Choroid plexus and paraphysis in lower vertebrates. J Morphol 120:157-188, 1966.

74. Smith GE: The morphology of the true "limbic lobe" corpus callosum, septum pellucidum, and fornix: A preliminary communication. J Anat Physiol 30:157-205, 1895. 75. Smith GE: Some problems relating to the evolu tion of the brain. Lancet 1-6, 147-155, 221-227, 1910. 76. Streeter GL: The development of the nervous sys tem. In Keibal F, Mall FP (eds): Manual of Human Embryology. Philadelphia, JB Lippincott Co, 1912. 77. Streeter GL: The developmental alterations in the vascular system of the brain of the human em bryo. Contrib Embryol 8:5-38, 1918. 78. Thompson IM: On the cavum septi pellucidi. J Anat 67:59-77, 1932. 79. Tilney F, Riley HA: The form and function of the central nervous system. New York, Paul В Hoeber, Inc, 1938, ed 3. 80. van de Kamer JC: Histological structure and cy tology of the pineal complex in fishes, amphibi ans, and reptiles. Prog Brain Res 10:30-48, 1965. 81. Vollrath L: Comparative morphology of the verte brate pineal complex. In Kappers JA, Peret P (eds): The Pineal Gland of Vertebrates Including Man. Prog Brain Res 52: 1979, pp 25-38. 82. Vollrath L: The Pineal Organ. Berlin, Springer, 1981. 83. Warwick R, Williams PL (eds): Gray's Anatomy. Philadelphia, WB Saunders Co, 1980, British ed 36. 84. Westergaard E: The cerebral ventricles of the guinea pig during growth. Acta Anat 70:382-402, 1958. 85. Westergaard E: The cerebral ventricles of the golden hamster during growth. Acta Anat 72:533-548, 1969. 86. Westergaard E: The cerebral ventricles of the rat during growth. Acta Anat 74:405-423, 1969. 87. Wolff HG: The cerebral blood vessels—anatomi cal principles. Assoc Res Nerv Ment Dis Proc 18:29-68, 1938. 88. Woollard HH, Weddell G: Arterial vascular pat terns. J Anat 69:25-37, 1934. 89. Yamamoto I, Rhoton AL, Peace RA: Microsurgery of the third venticle: Part 1. Microsurgical anat omy. Neurosurgery 8:333-356, 1981.

3 Microsurgical Anatomy of the Third Ventricular Region Albert L. Rhoton, Jr., M.D.

The third ventricle is located in the center of the head, below the corpus callosum and the body of the lateral ventricle; above the sella turcica, pituitary gland, and midbrain; and between the cerebral hemispheres, thalami, and the walls of the hypothalamus. It is intimately related to the circle of Willis and its branches and the vein of Galen and its tributaries. Selecting an operative approach to a lesion involving the third ventricle requires an understanding of the anatomy of the sella turcica, sphenoid sinus, lateral ventricles, and basal cisterns. This chapter is divided into four sections. The first three sections deal with the neural, arterial, and venous relationships around the third ventricle. The fourth section on the sellar region deals with the structures in the cranial base that are important in performing the subcranial operative approaches to lesions around the third ventricle. The section on neural relationships includes a review of the anatomy of the lateral ventricles and the cisterns surrounding the ten-torium incisura because many lesions involving the third ventricle are approached through these spaces (8, 25, 26, 43, 55). Neural Relationships Third Ventricle The third ventricle is a narrow, funnel-shaped, unilocular, midline cavity. It communicates at its anterosuperior margin with each lateral ventricle through the foramen of Monro and posteriorly with the fourth ventricle through the aqueduct

of Sylvius. It has a roof, a floor, and an anterior, posterior, and two lateral walls (Fig. 3.1). Roof The roof of the third ventricle forms a gentle upward arch, extending from the foramen of Monro anteriorly to the suprapineal recess posteriorly. The roof has four layers: one neural layer formed by the fornix, two thin membranous layers of tela choroidea, and a layer of blood vessels between the sheets of tela choroidea (Figs. 3.2 to 3.4). The upper, or neural, layer is formed by the fornix. The fornix is formed by axons that arise in the floor of the temporal horns from the hippocampi and extend around the thalami to reach the mamillary bodies. The fornix is composed of a body, two anterior limbs called columns, two posterior limbs called crura, and two inferior limbs called fimbria (Fig. 3.2). The initial part of the fornix, the fimbria, arises in the floor of the temporal horn on the ventricular surface of the hippocampal formation and passes posteriorly to become the crus of the fornix. The crus wraps around the posterior surface of the pulvinar of the thalamus and arches superomedial toward the lower surface of the splenium of the corpus callosum. At the junction of the atrium and the body of the lateral ventricle, the paired crura meet to form the body of the fornix, which runs forward along the superomedial border of the thalami in the medial wall of the body of the lateral ventricle. The body forms a gentle arch located between the roof of the third ventricle

Figure 3.1. Midsagittal section of the third ventricle. The floor extends from the optic chiasm (O.Ch.) to the aqueduct of Sylvius and includes the lower surface of the optic chiasm, the infundibulum (Infund.), the infundibular recess (Infund. Recess), the pituitary gland (Pit. Gland), the tuber cinereum (Tuber Cin.), the mamillary bodies (Mam. В.), the posterior perforated substance (Post. Perf. Subst.), and the part of the midbrain anterior to the aqueduct. The anterior wall extends from the optic chiasm to the foramen of Monro (F. Monro) and includes the upper surface of the optic chiasm, the optic recess (O. Recess), the lamina terminalis (Lam. Ter.), the anterior commissure (Ant. Comm.), and the foramen of Monro. The roof extends from the foramen of Monro to the suprapineal recess and is formed by the fornix and the layers of the tela choroidea (Tela), between which course the internal cerebral vein and the medial posterior choroidal artery. The hippocampal commissure (Hippo. Comm.), corpus callosum (Corp. Call.), and septum pellucidum (Sept. Pel.) are above the roof. The posterior wall extends from the suprapineal recess to the aqueduct and includes the habenular commissure (Hab. Comm.), pineal gland, pineal recess, and posterior commissure (Post. Comm.). The oculomotor nerve (HI) exits from the midbrain. The hypothalamic sulcus (Hypothal. Sulc.) forms a groove between the thalamic and hypothalamic (Hypothal.) surfaces of the third ventricle. (From Yamamoto I, Rhoton AL Jr, Peace DA: Microsurgery of the third ventricle: Part 1. Microsurgical anatomy. Neurosurgery 8:334-356, 1981.)

and the floor of the body of each lateral ventricle. The body of the fornix splits into two columns at the anterior margin of the opening of each foramen of Monro into the ipsilateral lateral ventricle. The columns terminate in the mamillary bodies. The crura, just posterior to where they join to form the body of the fornix, are interconnected by a sheet of white matter called the hippocampal commissure. The upper layer of the anterior part of the roof of the third ventricle is formed by the body of the fornix, and the posterior part of the roof is formed by the crura and the hippocampal commissure. The septum pellucidum is attached to the upper surface of the body of the fornix. The septum pellucidum is tallest anteriorly and shortest posteriorly, disappearing near the junction of the body and crura of the fornix. In our studies, the anteroposterior length of the septum pellucidum varied from 28 to 50 mm (55). At the posterior end of the septum pellucidum, the crura and hippocampal commissure fuse with the lower surface of the corpus callosum. The tela choroidea forms two of the three layers in the roof below the layer formed by the fornix (Fig. 3.3). The tela choroidea consists of two thin, semiopaque membranes derived from pia mater, which are interconnected by loosely organized trabeculae. The final layer in the roof is a vascular layer located between the two layers of tela choroidea. The vascular layer consists of the medial posterior choroidal arteries and their branches and the internal cerebral veins and their tributaries. Parallel strands of choroid plexus project downward on each side of the midline from the inferior layer of tela choroidea into the superior part of the third ventricle. The velum interpositum is the space between the two layers of the tela choroidea in the roof of the third ventricle. The velum interpositum is usually a closed space that tapers to a narrow apex just behind the foramen of Monro, but it may infrequently have an opening situated between the splenium and the pineal body that communicates with the quadrigeminal cistern to form the cisterna velum interpositum. The upper layer of tela choroidea is attached to the lower surface of the fornix and the hippocampal commissure (Figs. 3.2 and 3.3). The anterior part of the lower wall is attached to the teniae thalami, small ridges on the free edge of a fiber tract, the striae medullaris thalami, which extends along the superomedial border of the thalamus from the foramen of Monro to the habenular commissure. The posterior part of the lower wall is attached to the superior surface of the pineal body. The internal cerebral veins arise in the anterior part of the velum interpositum, just behind the foramen of Monro, and they exit the velum inter-

positum above the pineal body to enter the quadrigeminal cistern and join the great vein. The lateral margin of the roof is formed by the cleft between the lateral edge of the fornix and the superomedial surface of the thalamus (Fig. 3.2). This narrow cleft between the fornix and the thalamus, which in its entirety is C-shaped, is called the choroidal fissure. The choroid plexus of the lateral ventricle is attached along this fissure. The fornix forms the outer margin of the C-shaped fissure, and the thalamus forms the inner margin. The choroidal fissure is limited in the body of the ventricle by the body of the fornix superiorly and by the thalamus inferiorly, in the atrium by the crus of the fornix posteriorly and the pulvinar anteriorly, and in the inferior horn by the fimbria of the fornix below and the striae terminalis and the thalamus above. The tela cho-roidea forming the pedicle of the choroid plexus of the third ventricle is continuous through the choroidal fissure with the choroid plexus in the lateral ventricle.

Figure 3.2. A-D. Neural relationships. A. Superior view. The upper part of the cerebral hemispheres has been removed to expose the lateral ventricles. The frontal horn (Front. Horn) extends into the frontal lobe (Front. Lobe) anterior to the foramen of Monro (For. Monro). The body of the lateral ventricle (Body Lat. Vent.) is behind the foramen of Monro and above the thalamus. The atrium is located posterior to the thalamus deep to the parietal lobe (Par. Lobe). The occipital horn (Occip. Horn) extends backward from the atrium into the occipital lobe (Occip. Lobe). The temporal horn extends forward from the atrium into the temporal lobe (Temp. Lobe). The lateral wall of the frontal horn is formed by the head of the caudate nucleus (Caudate Nucl. Head); the floor medial to the caudate nucleus is formed by the rostrum of the corpus callosum; the medial wall is formed by the septum pellucidum (Sept. Pell.); the anterior wall is formed by the forceps minor, a bundle composed of fibers crossing in the genu of the corpus callosum; and the roof is formed by the anterior part of the body and the posterior part of the genu of the corpus callosum. The lateral wall of the body of the lateral ventricle is formed by the body of the caudate nucleus: the floor is formed by the superior surface of the thalamus; the roof is formed by the body of the corpus callosum; and the medial wall is formed by the septum pellucidum above and the body of the fornix below. The fornix passes along and is separated from the superomedial margin of the thalamus by a narrow cleft, the choroidal fissure (Chor. Fiss.), along which the choroid plexus (Chor. Plex.) is attached. The choroid plexus has been removed along the posterior part of the choroidal fissure on the right side. The atrium and occipital horn have the same structures in their walls. The lateral part of the anterior wall of the atrium is formed by the pulvinar, and the medial part is formed by the cms of the fornix. The superior part of the medial wall is formed by a large fiber tract from the splenium, the forceps major, which interconnects the occipital lobes and produces a prominent bulge in the medial wall called the bulb of the corpus callosum. The inferior part of the medial wall is formed by the calcar avis, a prominence overlying the depths of the calcarine sulcus. The medial part of the floor of the atrium is formed by the posterior part of the hippocampus, the smooth prominence overlying the hippocampal formation, and the lateral part of the floor is formed by the collateral trigone (Collat. Trig.), a prominence that overlies the collateral sulcus on the inferior surface of the temporal lobe. The lateral wall just posterior to the pulvinar is formed by the tail of the caudate nucleus as it wraps around the lateral edge of the pulvinar, and the part behind the caudate nucleus is formed by the tapetum of the corpus callosum, a fiber tract from the body and splenium that passes around the lateral margin of the atrium and

the temporal horn. The parietooccipital sulcus (Par. Occip. Sufc.) extends deeply into the medial surface of the hemisphere above the cuneus. The superior sagittal sinus (Sup. Sag. Sinus) courses in the outer periphery of the falx. B. Enlarged view of the region of the foramen of Monro. The choroid plexus has been removed from its attachment along the choroidal fissure on the right side. The anterior thalamic tubercle (Ant. Thal. Tuber.), which overlies the anterior nucleus of the thalamus, bulges upward at the posterior margin of the foramen of Monro. The columns of the fornix pass anterior and superior to the foramen of Monro. The striothalamic sulcus (Str. Thal. Sufc.) separates the caudate nucleus and the thalamus. C. The level of the cross section has been extended deeper into the right cerebral hemisphere through the caudate nucleus and the thalamus. The posterior part of the body and the superior part of the crus of the fornix have been removed on the right side to expose the tela choroidea (Tela) in the roof of the third ventricle. The anterior limb of the internal capsule (Int. Cap. Ant. Limb) is located between the head of the caudate nucleus and the lentiform nuclei (Lent. Nucl.). The posterior limb of the internal capsule (Int. Cap. Post. Limb) is located between the thalamus and the lentiform nucleus. The genu of the internal capsule (Genu Int. Cap.) touches the lateral wall of the ventricle between the caudate nucleus and the thalamus. The calcar avis and the bulb of the corpus callosum (Bulb Corp. Call.) bulge into the medial wall of the atrium and the occipital horn. The collateral trigone bulges in the floor of the atrium. The glomus, a large tuft of choroid plexus, is attached in the atrium between the fornix and the pulvinar. D. The fornix has been divided at the junction of its body and the columns above the foramen of Monro and reflected backward to expose the velum interpositum (Vel. Interpos.) located between the layers of tela choroidea in the roof of the third ventricle. The medial part of the floor of the temporal horn (Temp. Horn) is formed by the hippocampus, the smooth prominence overlying the hippocampal formation, and the lateral part of the floor is formed by the collateral eminence (Coff. Eminence), which overlies the deep part of the collateral sulcus. The pes hippocampus (Pes Hipp.) is the bulbous, digitated anterior part of the hippocampal formation. The hippocampal commissure (Hipp. Comm.) interconnects the medial margins of the crura of the fornix.

Figure 3.2. E-H. Neural relationships (continued). E. The body and anterior part of the crura of the fornix have been removed to expose the third ventricle (3 Vent.) and the quadrigeminal plate and pineal regions. The massa intermedia (Massa Inter.) extends into the third ventricle anteriorly, and the habenular (Hab. Comm.) and posterior (Post. Comm.) commissures cross the posterior part of the third ventricle. The habenular commissure is connected anteriorly with the habenular trigones (Hab. Trig.) and the striae medullaris thalami (Str. Med. Thal.). The pineal body projects posteriorly above the superior (Sup. Coll.) and inferior colliculi (Inf. Coll.). The subiculum, the rounded medial edge of the parahippocampal gyrus, is exposed lateral to the colliculi. The inferior sagittal sinus (Inf. Sag. Sinus) courses in the inferior margin of the falx and joins the straight sinus at the tentorial apex (Apex. Tent.). F. Right superolateral view. The frontal and parietal operculae have been removed to expose the sylvian fissure (Sylvian Fiss.) and the transverse temporal gyri (Transv. Temp. Gyr.). G. Further removal of deep cerebral tissue exposes the hippocampal formation and the collateral eminence in the floor of the temporal horn. The collateral eminence is continuous posteriorly with the collateral trigone in the floor of the atrium. H. Enlarged view to show structures in the medial wall of the atrium and the floor of the temporal horn. The fimbria of the fornix arises on the surface of the hippocampus and is continuous posteriorly with the crus of the fornix. (From Ono M, Rhoton AL Jr, Peace DA, Rodriguez R: Microsurgical anatomy of the deep venous system of the brain. Neurosurgery 15:621-657, 1984.)

Figure 3.3. Superior view of stepwise dissection of layers in the roof of the third ventricle. A. Superior part of cerebral hemispheres, with the corpus callosum (Corp. Call.) including a part of the splenium removed to expose the choroid plexus (Ch. PL) of the lateral ventricle. A superior choroidal vein (S. Ch. V.) passes along the free edge of the choroid plexus. The glomus of the choroid plexus projects into the atrium. The caudate nucleus projects into the anterior horn and body of the lateral ventricle. The septum pellucidum (Sept. Pel.) forms the medial walls of the lateral ventricles. The bodies of the fornix are medial to the choroid plexus in the bodies of the lateral ventricle. B. The body of the fornix and septum pellucidum are divided and lifted up to expose the tela choroidea (Tela) in the roof of the third ventricle. C. The body of the fornix removed at the junction with the columns of the fornix (Columns). The medial posterior choroidal arteries (M. P. Ch. A., broken line) course between the upper and lower layers of the tela choroidea. D. The tela choroidea was removed to expose the parallel strands of choroid plexus in the roof of the third ventricle (Ch. PI.-3V) and the interior third ventricle (3V). Part of the splenium of the corpus callosum was removed to show the pineal body. Right and left thalami form the lateral walls of the third ventricle. (From Fujii K, Lenkey C, Rhoton AL Jr: Microsurgical anatomy of the choroidal arteries: Lateral and third ventricles. J Neurosurg 52:165-188, 1980.)

Figure 3.4. Inferior views of the roof of the third ventricle. A. The portion of the third ventricle and cerebral hemispheres below the roof of the third ventricle has been removed to provide an inferior view of the roof. The choroid plexus (Ch. Pl.) hangs downward from the tela choroidea into the superior part of the third ventricle. The internal cerebral veins (Int. Сет. V) and the medial posterior choroidal arteries (Med. Post. Ch. A.) course below the fornix and between the two halves of the thalamus. The choroid plexus in the roof of the third ventricle is continuous through the foramen of Monro (F. Monro) with the choroid plexus in the lateral ventricle. The body of the fornix splits into two columns at the anterior margin of the foramen of Monro. The anterior commissure (Ant. Comm.) is anterior to the columns of the fornix. The massa intermedia (Massa Inter.) projects into the upper partof the third ventricle. The internal cerebral veins join to form the vein of Galen (V. Galen) in the posterior part of the roof. B. The medial edges of the thalamus on each side have been removed to provide a wider view of the body and crura of the fornix, the medial posterior choroidal arteries, and the internal cerebral veins. The crura of the fornix pass posteriorly below the sple-nium of the corpus callosum (Splenium). The internal cerebral veins pass posteriorly above the pineal gland (Pineal) and join to form the vein of Galen. The choroidal fissure (Chor. Fiss.) is a groove on each side between the fornix and the thalamus. The upper part of the stalk of the pineal gland formed by the habenular commissure (Hab. Comm.) has been preserved. C. The choroidal fissure on each side has been opened wider by the removal of more of the medial surface of the thalamus. The lateral ventricles (Lat. Vent.) are seen through the enlarged choroidal fissures. The choroid plexus of the lateral ventricles is exposed and branches of the lateral posterior choroidal arteries (Lat. Post. Ch. A.) and the superior choroidal veins (Sup. Ch. V.) are seen on the surface of the choroid plexus. The hippocampal commissure (Hippo. Comm.) interconnects the medial margins of the crura of the fornix above the pineal gland. The pineal gland is below the splenium and the hippocampal commissure. The posterior part of the crura of the fornix and the hippocampal commissure are adherent to the lower margin of the splenium of the corpus callosum. (From Yamamoto I, Rhoton AL Jr, Peace DA: Microsurgery of the third ventricle: Part 1. Microsurgical anatomy. Neurosurgery 8:334-356, 1981.)

Floor The floor extends from the optic chiasm anteriorly to the orifice of the aqueduct of Sylvius posteriorly (Figs. 3.1 and 3.5). The anterior half of the floor is formed by diencephalic structures and the posterior half is formed by mesence-phalic structures. When viewed from inferiorly, the structures forming the floor from anterior to posterior include the optic chiasm, the infundibulum of the hypothalamus, the tuber cinereum, the mamil-lary bodies, the posterior perforated substance, and (most posteriorly) the part of the tegmentum of the midbrain located above the medial aspect of the cerebral peduncles. The optic chiasm is located at the junction of the floor and the anterior wall of the third ventricle. The chiasm slopes posteriorly and superiorly from its junction with the optic nerves. The inferior surface of the chiasm forms the anterior part of the floor and the superior surface forms the lower part of the anterior wall. The optic tracts arise from the posterolateral margin of the chiasm and course obliquely away from the floor toward the lateral margin of the midbrain. The infundibulum, tuber cinereum, mamillary bodies, and posterior perforated substance are located in the space limited anteriorly and laterally by the optic chiasm and tracts and posteriorly by the cerebral peduncles. The infundibulum of the hypothalamus is a hollow, funnel-shaped structure located between the optic chiasm and the tuber cinereum. The pituitary gland (hypophysis) is attached to the infundibulum, and the axons in the infundibulum extend to the posterior lobe of the hypophysis. The tuber cinereum is a prominent mass of hypothlamic gray matter located anterior to the mamillary bodies. The tuber cinereum merges anteriorly into the infundibulum. The tuber cinereum around the base of the infundibulum is raised to form a prominence called the median eminence. The mamillary bodies form paired, round prominences posterior to the tuber cinereum. The posterior perforated substance is a depressed, punctuated area of gray matter located in the interval between the mamillary bodies anteriorly and the medial surface of the cerebral peduncles posteriorly. The posterior part of the floor extends posterior and superior to the medial part of the cerebral peduncles and superior to the tegmentum of the midbrain. When viewed from above and inside the third ventricle, the optic chiasm forms a prominence at the anterior margin of the floor (Fig. 3.5). The infundibular recess extends into the infundibu-lum behind the optic chiasm. The mamillary bodies form paired prominences on the inner surface of the floor posterior to the infundibular recess. The part of the floor between the mamillary bodies and the aqueduct of Sylvius has a smooth surface, which is concave from side to side. This smooth surface lies above the posterior perforated substance anteriorly and the medial part of the cerebral peduncles and the tegmentum of the midbrain posteriorly. Anterior Wall The anterior margin of the third ventricle extends from the foramina of Monro above to the optic chiasm below (Figs. 3.1, 3.5, and 3.6). Only the lower two-thirds of the anterior surface is seen on the external surface of the brain;

the upper one-third is hidden posterior to the rostrum of the corpus callosum. The part of the anterior wall visible on the surface is formed by the optic chiasm and the lamina terminalis. The lamina terminalis is a thin sheet of gray matter and pia mater that attaches to the upper surface of the chiasm and stretches upward to fill the interval between the optic chiasm and the rostrum of the corpus callosum. When viewed from within, the boundaries of the anterior wall from superiorly to inferiorly are formed by the columns of the fornix, foramina of Monro, anterior commissure, lamina terminalis, optic recess, and optic chiasm. The foramen of Monro on each side is located at the junction of the roof and the anterior wall (Figs. 3.1 and 3.5). The foramen is a ductlike canal that opens between the fornix and the thalamus into the lateral ventricle and extends inf eriorly below the fornix into the third ventricle as a single channel. The foramen of Monro is bounded anteriorly by the junction of the body and the columns of the fornix and posteriorly by the anterior pole of the thalamus (Figs. 3.1, 3.2, and 3.5). The size and shape of the foramina of Monro depend on the size of the ventricles: if the ventricles are small, each foramen is a crescent-shaped opening bounded anteriorly by the concave curve of the fornix and posteriorly by the convex anterior tubercle of the thalamus. As the ventricles enlarge, the foramen on each side becomes rounder. The structures that pass through the foramen are the choroid plexus, the distal branches of the medial posterior choroidal arteries, and the internal cerebral, thalamostriate, superior choroidal, and septal veins. The anterior commissure is a compact bundle of myelinated nerve fibers that cross the midline in front of the columns of the fornix. The anter-oposterior diameter of the anterior commissure varies from 1.5 to 6.0 mm (55). The distance from the posterior end of the anterior commissure to the anterior border of the foramen of Monro ranges from 1.0 to 3.5 mm (average, 2.2 mm). The distance from the upper edge of the optic chiasm to the anterior border of the anterior commissure ranges from 8 to 12 mm (average, 10 mm). The lamina terminalis fills the interval between the anterior commissure and the optic chiasm (Fig. 3.1). The lamina attaches to the midportion of the superior surface of the chiasm, leaving a small cleft between the upper half of the chiasm and the lamina, called the optic recess.

Figure 3.5. Anterosuperior views of the third ventricle. A. Part of the anterior wall of the third ventricle and the anterior part of the cerebral hemispheres have been removed. The septum pellucidum (Sept. Pel.) is attached to the upper margin of the fornix. The optic chiasm (O. Ch.) and nerves (O. N.) are at the lower margin of the anterior wall, and the optic tracts (O. Tr.) extend laterally below the floor of the third ventricle (3V). The choroid plexus (Ch. Pi.) is attached over the surface of the thalamus on each side. The thalamostriate vein on each side (Thal. Sir. V.) courses forward between the caudate nucleus (Caudate Nucl.) and the thalamus. The infundibular recess (Infund. Recess) extends downward posterior to the optic chiasm and anterior to the mamillary bodies (Mam. В.). The midportion of the anterior commissure (Ant. Comm.) has been removed to expose the columns of the fornix anterior to the foramina of Monro. The body and columns of the fornix join anterior to the foramina of Monro. The corpus callosum (Corp. Call.) is above the lateral ventricles. B. The septum pellucidum and the medial part of the body of the fornix have been removed to expose each foramen of Monro (F. Monro) and the full length of the floor of the third ventricle. The floor extends from the aqueduct of Sylvius (Aqueduct) posteriorly to the optic chiasm anteriorly. The habenular commissure (Hab. Comm.) forms the upper margin of the stalk of the pineal gland (Pineal), and the posterior commissure (Post. Comm.) forms the lower stalk of the pineal gland. The pineal recess is between the two commissures. A branch of the medial posterior choroidal artery (Med. Post. Ch. A.) passes under the fornix to join the choroid plexus. The superior choroidal veins (Sup. Ch. V.) course on the surface of the choroid plexus. The crus of the fornix is above the pineal gland. The optic recess (O. Recess) extends anterior to the upper one-half of the optic chiasm. C. The body of the fornix has been removed to show the internal cerebral veins (Int. Сет. V.) and the medial posterior choroidal arteries in the roof of the third ventricle. A ventricular vein drains into the internal cerebral vein. The internal cerebral veins join to form the vein of Galen (V. Galen). D. Enlarged view shows the habenular commissure above the posterior commissure below the pineal recess. The internal cerebral veins and the medial posterior choroidal arteries pass forward and intermingle along the superolateral margin of the pineal gland. The columns of the fornix are anterior to the foramen of Monro. The crura of the fornix are above the pineal gland and below the splenium (Splenium) of the corpus callosum. (From Yamamoto I, Rhoton AL Jr, Peace DA: Microsurgery of the third ventricle: Part 1. Microsurgical anatomy. Neurosurgery 8:334-356, 1981.)

Figure 3.6. Neural relationships. Posterior and midsagittal views. A. Posterior view with orientation as shown in the insert. The right occipital lobe (Occip. Lobe) has been retracted away from the tentorium (Tent.) and the straight (Str. Sinus) and inferior sagittal (Inf. Sag. Sinus) sinuses to expose the tentorial apex (Apex Tent.). The tentorium behind its free edge (Tent. Edge) has been opened to expose the superior surface of the cerebellum. The pineal body projects backward above the superior colliculi (Sup. Coll.). The habenular commissure (Hab. Comm.) crosses the posterior end of the third ventricle. The suprapineal recess (Sup. Pineal Recess) of the third ventricle extends posteriorly between the superior surface of the pineal body and the tela choroidea (Tela) in the roof of the third ventricle. The isthmus of the cingulate gyrus (Isthmus) is a narrow cortical bridge between the cingulate (Cing. Gyr.) and parahippocampal (Parahippo. Gyr.) gyri. The medial geniculate body (Med. Gen. Body) and brachium of the inferior colliculus (Brach. Inf. Coll.) are posterior to the cerebral peduncle (Ped.). B. Posterior view. The posterior part of the cerebral hemispheres has been removed to expose the posterior part of the third ventricle (3rd Vent.) and the pineal region, where the deep veins converge on the vein of Galen. The atria and the bodies of the lateral ventricles (Body Lat. Vent.) are deep within the parietal lobe (Par. Lobe), below the corpus callosum (Corp. Call.). The caudate nuclei (Caudate Nucl.) are lateral to the thalami. The crura of the fornix pass around the pulvinar. Choroid plexus (Chor. Plex.) is attached along the choroidal fissure, the cleft between the fornix and the thalamus. The subiculum is the rounded medial edge of the parahippocampal gyrus of the temporal lobe (Temp. Lobe). The collateral sulcus (Coll. Sulc.) is lateral to the parahippocampal gyrus. C. Posterior view below the tentorium. The cerebellum was removed by dividing the superior (Sup. Cer. Ped.), middle (Mid. Cer. Ped.), and inferior (Inf. Сет. Ped.) cerebellar peduncles just dorsal to the floor of the fourth ventricle (4 Vent.}. The structures just above the posterior part of the incisura include the splenium, the crura of the fornix, the pineal body, and the superior and inferior colliculi (Inf. Coll.). The trochlear nerves (V) arise below the inferior colliculi and pass around the brain stem. The superior medullary velum (Sup. Med. Velum) forms the medial part of the roof of the fourth ventricle. The flocculus and the rhomboid lip border the lateral recess (Lat. Recess) of the fourth ventricle and extend laterally above the glossopharyngeal (IX), vagus (X), and accessory (XI) nerves. The hippocampal commissure (Hippo. Comm.) is a fiber tract connecting the medial margins of the crura of the fornix. D. Midsagittal section. The rostrum of the corpus callosum is continuous below the anterior commissure (Ant. Comm.) with the lamina termin-alis (Lam. Term.). The massa intermedia (Massa Inter.) protrudes into the central part of the third ventricle. The pineal body is attached to the posterior part of the third ventricle by a stalk composed of the habenular and posterior commissures (Post. Comm.). The striae medullaris thalami (Str. Med. Thal.) course on the medial surface of the thalamus from the foramen of Monro to the habenular commissure. The septum pellucidum and the part of the fornix above the thalamus have been removed. The choroid plexus is attached along the choroidal fissure (Chor. Fiss.), the cleft between the fornix and thalamus. The caudate nucleus (Caudate Nucl.) is exposed in the far wall of the frontal horn (Front. Horn) and the body of the right lateral ventricle. The cms of the fornix is fused for a short distance to the lower surface of the splenium. The brain stem has been sectioned at the level of the cerebral peduncle. The dentate gyrus (Dentate Gyr.) forms a beaded longitudinal cortical strip above the parahippocampal gyrus. The hippocampal sulcus (Hippo. Sulc.) separates the dentate and parahippocampal gyri. The fimbria of the fornix arises on the surface of the hippocampal formation in the floor of the temporal horn and forms a longitudinal strip that is separated from the dentate gyrus by the fimbriodentate sulcus (Fimb. Dentate Sulc.). The fimbria is continuous posteriorly with the crus of the fornix. The column of the fornix passes anterior and superior to the foramen of Monro (For. Monro) to terminate in the mamillary bodies (Mam. Body). The lateral geniculate body (Lat. Gen. Body) protrudes from the inferior surface of the thalamus. The infundibular recess (Infund. Recess) extends inferiorly into the base of the infundibulum. The hypothalamic sulcus (Hypothal. Sulc.) separates the thalamus and hypothalamus. (From Ono M, Rhoton AL Jr, Peace DA, Rodriguez R: Micro-surgical anatomy of the deep venous system of the brain. Neurosurgery 15:621-657, 1984.)

Figure 3.7. A-D. Tentorial incisura: stepwise dissection, anterior-superior view. A. The anterior part of the frontal lobe has been removed to expose the anterior incisural space. The section of the frontal lobe passes adjacent to the septum pellucidum (Sept. Pell.) and through the rostrum and genu of the corpus callosum (Corp. Call.), the caudate nucleus (Caudate Nucl.), the putamen, and the frontal horn of the lateral ventricle (Lat. Vent.). The choroid plexus (Chor. Plex.) extends through the foramen of Monro (For. Monro). The anterior incisural space is located anterior to the midbrain and extends upward around the optic chiasm, lamina terminalis (Lam. Term.), and anterior part of the third ventricle (3 Vent). The optic tract (Optic Tr.) extends posteriorly above the oculomotor nerve (HI). The infundibulum (Infund.) of the pituitary gland passes through the diaphragma sellae (Diaph.). The carotid artery (Car. A.) has been divided below the optic nerve (Optic N.). The cingulate gyri (Cing. Gyr.) lie above the corpus callosum. A septal vein (Septal V.) courses on the septum pellucidum. The paraterminal (Paraterm. Gyr.) and paraolfactory gyri (Paraolf. Gyr.) are above the gyrus rectus (Gyr. Rectus). The semilunar (Semilunar Gyr.) and ambient gyri (Ambient Gyr.) are on the superior surface of the temporal lobe bordering the sylvian fissure (Sylvian Fiss.). The anterior limb of the internal capsule (Int. Cap. Ant. Limb) is situated between the caudate nucleus and the putamen. The oculomotor nerve enters the dura mater posterior to the anterior clinoid process (Ant. Ciinoid). B. The transverse section has been extended behind the foramen of Monro to include part of the cerebral peduncle (Ped.). The posterior part of the right optic nerve and the right half of the optic chiasm have been removed to expose the posterior part of the anterior incisural space. The substantia nigra (Subst. Nigra) and red nucleus (Red. NucL) are at the level of the transection of the midbrain. The thalamus and internal capsule are located directly above the cerebral peduncle. The middle incisural space is between the uncus and the midbrain. The parahippocampal gyrus (Parahippo. Gyr.) and uncus protrude medial to the tentorial edge (Tent. Edge). The caudate nucleus extends from the lateral part of the frontal horn around the atrium into the roof of the temporal horn (Temp. Horn), where the tail of the caudate nucleus is located. The lentiform nucleus formed by the globus pallidus (Globus Pall.) and the putamen are lateral to the internal capsule. The collateral sulcus (Coll. Sulc.) extends into the temporal lobe near the lateral part of the temporal horn. The anterior hippocampal sulcus (Ant. Hippo. Sulc.) separates the uncus and parahippocampal gyrus. The anterior commissure (Ant. Comm.) is in the anterior, the mamillary bodies (Mam. Body) are in the inferior, and the massa intermedia (Massa Int.) is in the central part of the third ventricle. The choroid plexus is attached in the lateral and third ventricle along the choroidal fissure (Chor. Fiss.) located between the fornix and the thalamus. The callosal sulcus (Call. Sulc.) separates the corpus callosum and the cingulate gyrus. C. Enlarged view. The anterior choroidal (Ant. Chor. A.) and posterior cerebral arteries (P. C. A.) pass around the cerebral peduncle medial to the uncus and the parahippocampal gyrus. The tela choroidea and choroid plexus in the temporal horn are attached to the tenia choroidea and the tenia fimbriae along the choroidal fissure in the medial part of the temporal horn. The pontomesencephalic sulcus (Pon. Mes. Sulc.) marks the junction of the midbrain and pons. The oculomotor nerve enters the dura lateral to the posterior clinoid process (Post. Clinoid). The trochlear nerve (IV) passes below the uncus. D. The coronal section extends just behind the posterior (Post. Comm.) and habenular (Hab. Comm.) commissures into the base of the pineal body (Pineal). The internal cerebral veins (Int. Cer. V.) course in the velum interpositum located between the two layers of the tela choroidea (Tela) in the roof of the third ventricle. The upper layer of tela is attached to the lower surface of the fornix. The mesial temporal structures that form the lateral wall of the middle incisural space include the subiculum, the rounded medial surface of the parahippocampal gyrus, which blends into the Ammon's horn (a curled horn-shaped collection of gray matter in the hippocampal formation), the dentate gyrus (Dentate Gyr.), a strip of cortex located above the subiculum, and the fimbria of the fornix, a fiber tract that arises in the floor of the temporal horn on the surface of the hippocampal formation. The pineal body projects into the posterior incisural space below the suprapineal recess and the internal cerebral veins. The medial (Med. Gen. Body) and lateral geniculate bodies (Lat. Gen. Body) are dorsolateral to the cerebral peduncle.

midbrain anteriorly, the splenium above, and the cerebellum below. The vein of Galen (V. of Galen) enters the straight sinus. The trochlear nerve arises below the inferior colliculus (Inf. Coll.) and passes forward in the cerebellomesencephalic fissure (Cer. Mes. Fiss.) situated between the cerebellum and midbrain. The paraterminal and paraolfactory gyri, the gyrus rectus, and the anterior (Ant. Paraolf. Sulc.) and posterior paraolfactory (Post. Paraolf. Sulc.) sulci are below the rostrum. The optic recess extends inferiorly between the optic chiasm and the lamina terminalis, and the infundibular recess (Infund. Recess) extends into the infundibulum behind the chiasm. The layer of tela choroidea that forms the upper wall of the velum interpositum is adherent to the lower margin of the body and cms of the fornix. The layer of tela choroidea that forms the lower wall of the velum interpositum is attached anteriorly to the striae medullaris thalami (Str. Med. Thal.) and posteriorly it adheres to the superior margin of the pineal body. The striae medullaris thalami extend forward from the habenular commissure along the superomedial margin of the thalamus. The lateral mesencephalic sulcus (Lat. Mes. Sulc.) extends along the posterior edge of the cerebral peduncle, and the interpeduncular fossa (Interped. Fossa) is between the cerebral peduncles. G. Enlarged view. The tela choroidea and the deep veins have been removed. The choroidal fissure is located between the fornix and the thalamus. The isthmus of the cingulate gyrus (Isthmus) is a cortical bridge between the parahippocampal and cingulate gyri. The fasciolar gyrus and the gyri Andreae Retzii form a beaded collection of gray matter that is continuous below with the dentate gyrus and above with the indusium griseum, a thin layer of gray matter on the surface of the corpus callosum. The quadrangular lobule (Quad. Lobule) of the cerebellum is behind the colliculi. (From Ono M, Ono M, Rhoton AL Jr, Barry M: Microsurgical anatomy of the region of the tentorial incisura. J Neurosurg 60:365-399, 1984.)

Figure 3.7. E-G. Tentorial incisura (continued). E. The level of the section has been extended behind the pineal body to expose the posterior incisural space and the vein of Galen. The pineal body projects into the posterior incisural space. The parahippocampal gyrus overlies the free edge and forms part of the lateral wall of the posterior incisural space. The fibers arising on the surface of the hippocampal formation pass through the fimbria to the crus of the fornix, which extends upward toward the splenium. The parahippocampal gyrus is separated from the dentate gyrus by the hippocampal sulcus (Hipp. Sulc.), and the fornix is separated from the dentate gyrus by the fim-briodentate sulcus (Fimb. Dent. Sulc.). F. The right cerebral hemisphere has been removed to expose all of the posterior incisural space located between the apex of the tentorium (Apex. Tent.) posteriorly, the

Posterior Wall The posterior wall of the third ventricle extends from the suprapineal recess above to the aqueduct of Sylvius below (Figs. 3.1, 3.5, and 3.6). When viewed from anteriorly and within the third ventricle, it consists, from above to below, of the suprapineal recess, the habenular commissure, the pineal body and its recess, the posterior commissure, and the aqueduct of Sylvius (Fig. 3.5). The suprapineal recess projects posteriorly between the upper surface of the pineal gland and the lower layer of tela choroidea in the roof. The pineal gland extends posteriorly into the quadrigeminal cistern from its stalk. The stalk of the pineal gland has a cranial and a caudal lamina. The habenular commissure, which interconnects the habenulae, crosses the midline in the cranial lamina, and the posterior commissure crosses in the caudal lamina. The pineal recess projects posteriorly into the pineal body between the two laminae. The shape of the orifice of the aqueduct of Sylvius is triangular: the base of the triangle is on the posterior commissure and the other two limbs are formed by the central gray matter of the midbrain. When viewed from posteriorly, the only structure in the posterior wall is the pineal body (Fig. 3.6). The pineal gland projects posteriorly into the quadrigeminal cisterns and is concealed by the splenium of the corpus callosum above, the thalamus laterally, and the quadrigeminal plate and the vermis of the cerebellum inferiorly. Lateral Wall The lateral walls are not visible on the external surface of the brain, but are hidden between the cerebral hemispheres (Figs. 3.1, 3.6, and 3.7). They are formed by the hypothalamus inferiorly and the thalamus superiorly. The lateral walls have an outline like the lateral silhouette of a bird's head with an open beak. The head is formed by the oval medial surface of the thalamus; the open beaks, which project anteriorly and inferiorly, are represented by the recesses in the hypothalamus: the pointed upper beak is formed by the optic recess and the lower beak is formed by the infundibular recess. The hypo-thalamic and thalamic surfaces are separated by the hypothalamic sulcus, a groove that is often

ill-defined and extends from the foramen of Monro to the aqueduct of Sylvius. The superior limit of the thalamic surfaces of the third ventricle is marked by narrow, raised ridges, known as the striae medullaris thalami. These striae extend forward from the habenulae along the superomedial surface of the thalamus at the site of the attachment of the lower layer of the tela choroidea. The habenulae are small eminences on the dorsomedial surfaces of the thalamus just in front of the pineal gland. The habenulae are connected across the midline in the rostral half of the stalk of the pineal gland by the habenular commissure. The massa intermedia projects into the upper one-half of the third ventricle and often connects the opposing surfaces of the thalamus. The massa intermedia was present in 76% of the brains examined and was located 2.5 to 6.0 mm (average, 3.9 mm) posterior to the foramen of Monro (55). The columns of the fornix form distinct prominences in the lateral walls of the third ventricle just below the foramen of Monro, but inferiorly they sink below the surface. Lateral Ventricles An understanding of the anatomy of the lateral ventricles is important because third ventricular tumors are commonly approached through the lateral ventricles. Each of the paired lateral ventricles is a C-shaped cavity that wraps around the thalamus and is situated deep within the cerebrum (Figs. 3.2, 3.3, and 3.7). Each ventricle has five parts: the frontal, temporal, and occipital horns and the body and atrium. Each of these five parts has medial and lateral walls, a roof, and a floor; in addition, the frontal and temporal horns and the atrium have anterior walls. These walls are formed predominantly by the thalamus, septum pellucidum, deep cerebral white matter, corpus callosum, and two C-shaped structures, the caudate nucleus and fornix, which wrap around the thalamus. The thalamus is located in the center of the lateral ventricle. Each lateral ventricle wraps around the superior, inferior, and posterior surfaces of the thalamus. The body of the lateral ventricle is above the thalamus, the atrium and occipital horn are posterior to the thalamus, and the temporal horn is inferior to the thalamus. The superior surface of the thalamus forms the floor of the body, the posterior surface of the thalamus forms the anterior wall of the atrium, and the inferior surface of the thalamus forms the medial part of the roof of the temporal horn. The caudate nucleus is an arched, C-shaped,

cellular mass that wraps around the thalamus and constitutes an important part of the wall of the lateral ventricle. It has a head, body, and tail. The head bulges into the lateral wall of the frontal horn and the body of the lateral ventricle. Its body forms part of the lateral wall of the atrium, and its tail extends from the atrium into the roof of the temporal horn and is continuous with the amygdaloid nucleus near the tip of the temporal horn. In the body of the lateral ventricle, the caudate nucleus is superolateral to the thalamus; in the atrium, it is posterolateral to the thalamus; and, in the temporal horn, it is inferolateral to the thalamus. The striae terminalis, a fiber tract that runs parallel and deep to the thalamostriate vein, arises in the amygdaloid nucleus and courses along the border between the caudate nucleus and the thalamus in the wall of the lateral ventricle. The fornix is another C-shaped structure that wraps around the thalamus in the wall of the ventricle. The fornix has four parts (the fimbria, crus, body, and columns) and extends from the floor of the temporal horn around the thalamus to the mamillary bodies. In the body of the lateral ventricle, the body of the fornix is in the lower part of the medial wall; in the atrium, the crus of the fornix is in the medial part of the anterior wall; and, in the temporal horn, the fimbria of the fornix is in the medial part of the floor. The corpus callosum, which forms the largest part of the ventricular walls, contributes to the wall of each of the five parts of the lateral ventricle. The corpus callosum has two anterior parts, the rostrum and genu; a central part, the body; and a posterior part, the splenium. The rostrum is situated below and forms the floor of the frontal horn. The genu gives rise to a large fiber tract, the forceps minor, which forms the anterior wall of the frontal horn as it sweeps obliquely forward and lateral to connect the frontal lobes. The genu and the body of the corpus callosum form the roof of both the frontal horn and the body of the lateral ventricle. The splenium gives rise to a large tract, the forceps major, which forms a prominence, the bulb, in the upper medial wall of the atrium and the occipital horn as it sweeps posteriorly to connect the occipital lobes. Another fiber tract, the tapetum, which arises in the posterior part of the body of the corpus callosum, sweeps laterally and inferiorly to form the roof and lateral walls of the atrium and the temporal and occipital horns. The septum pellucidum, which is composed of paired laminae, separates the frontal horns and bodies of the lateral ventricles in the midline. In

the frontal horn, the septum pellucidum is attached to the rostrum of the corpus callosum below, the genu anteriorly, and the body above. In the body of the lateral ventricle, the septum is attached to the body of the corpus callosum above and the body of the fornix below. The septum pellucidum disappears posteriorly where the body of the fornix meets the splenium. There may be a cavity, the cavum septum pellucidum, in the midline between the laminae of the septum pellucidum. The choroid plexus in the lateral ventricle has a Cshaped configuration that parallels the fornix (Figs. 3.1 to 3.3) (8). It is attached along the choroidal fissure, a narrow cleft between the fornix and the thalamus, in the medial part of the body, atrium, and temporal horn. The choroid plexus extends through the foramen of Monro into the roof of the third ventricle. In the atrium, the choroid plexus has a prominent triangular tuft called the glomus. The edges of the thalamus and fornix bordering this fissure have small ridges, the teniae, along which the tela choroidea, the membrane in which the choroid plexus arises, is attached. The choroidal fissure extends from the foramen of Monro along the medial wall of the body, atrium, and temporal horn to its inferior termination, the inferior choroidal point, located just behind the tip of the temporal lobe and uncus. The veins coursing in the walls of the ventricles exit the ventricles by passing, in a subependymal location, through the margin of this fissure to reach the internal cerebral, basal, or great veins. The frontal horn, the part of the lateral ventricle located anterior to the foramen of Monro, has a medial wall formed by the septum pellucidum, an anterior wall formed by the genu of the corpus callosum, a lateral wall composed of the head of the caudate nucleus, and a narrow floor formed by the rostrum of the corpus callosum (Fig. 3.2). The columns of the fornix, as they pass anterior to the foramen of Monro, are in the posteroinfer-ior part of the medial wall. The body of the lateral ventricle extends from the posterior edge of the foramen of Monro to the point where the septum pellucidum disappears and the corpus callosum and fornix meet. The roof is formed by the body of the corpus callosum, the medial wall by the septum pellucidum above and the body of the fornix below, the lateral wall by the body of the caudate nucleus, and the floor by the thalamus (Fig. 3.2). The atrium and occipital horn together form a roughly triangular cavity, with the apex posteriorly in the occipital lobe and the base anteriorly

on the pulvinar (Figs. 3.2, 3.6, and 3.7). The atrium opens anteriorly above the thalamus into the body, anteriorly below the thalamus into the temporal horn, and posteriorly into the occipital horn. The roof of the atrium is formed by the body and tapetum of the corpus callosum. The medial wall has an upper part formed by the bulb of the corpus callosum and a lower part formed by the calcar avis, a prominence overlying the deepest part of the calcarine sulcus. The lateral wall has an anterior part formed by the caudate nucleus as it wraps around the lateral margin of the pulvinar and a posterior part formed by the fibers of the tapetum as they sweep anteroinfer-iorly along the lateral margin of the ventricle. The anterior wall has a medial part composed of the crus of the fornix as it wraps around the posterior part of the pulvinar and a lateral part formed by the pulvinar of the thalamus. The floor has a medial part composed of the posterior part of the hippocampus, the prominence overlying the hippocampal formation, and a lateral part formed by the collateral trigone, the triangular prominence deep to the posterior end of the collateral sulcus. In the atrium, the choroid plexus has a prominent tuft called the glomus. The occipital horn extends posteriorly into the occipital lobe from the atrium. Its medial wall is composed of the bulb and the calcar avis, the roof and lateral wall are formed by the tapetum, and the floor is formed by the collateral trigone. The temporal horn extends forward from the atrium below the pulvinar into the medial part of the temporal lobe and ends blindly in the anterior wall situated immediately behind the amygdaloid nucleus (Figs. 3.2 and 3.7). The floor of the temporal horn is formed medially by the hippocampus and laterally by the smooth white prominence overlying the collateral sulcus. The roof is formed medially by the inferior surface of the thalamus and the tail of the caudate nucleus, which are separated by the striothalamic sulcus, and laterally by the tapetum of the corpus callosum, which also sweeps inferiorly to form the lateral wall of the temporal horn. The medial wall is little more than a narrow cleft, the choroidal fissure, between the inferolateral part of the thalamus and the fimbria of the fornix. The inferior end of the choroid fissure, the inferior choroidal point, is located just behind the amygdaloid nucleus and the uncus. Basal Cisterns in Tentorial Incisura The third ventricle is commonly approached through the cisterns surrounding the tentorial incisura (25). These cisterns, which are inti-

mately related to the third ventricle, contain the bifurcation of the carotid and basilar arteries, the circle of Willis, the convergence of the deep intracranial venous system on the great vein, and parts of the first six cranial nerves. The tentorial incisura is a triangular space situated between the dorsum sellae and the free edges of the tentorium. Its apex is dorsal to the midbrain, just posterior to the pineal gland. The upper brain stem sits in the center of the incisura. The area between the brain stem and the free edges is divided into: (a) an anterior incisural space located in front of the brain stem; (b) paired middle incisural spaces situated lateral to the brain stem; and (c) a posterior incisural space located behind the brain stem (Figs. 3.7 to 3.9). Anterior Incisural Space The anterior incisural space is located anterior to the midbrain. It extends obliquely forward and upward around the optic chiasm to the subcal-losal area. It opens laterally into the medial part of the sylvian fissure and posteriorly between the uncus and the brain stem into the middle incisural space (Figs. 3.7 to 3.9). The part of the anterior incisural space located below the optic chiasm has posterolateral and posterior walls. The posterolateral wall is formed by the anterior one-third of the uncus, which hangs over the free edge above the oculomotor trigone. The posterior wall is formed by the cerebral peduncles. The infundibulum of the pituitary gland crosses the anterior incisural space to reach the opening in the diaphragma sellae. The part of the anterior incisural space situated above the optic chiasm is limited superiorly by the rostrum of the corpus callosum, posteriorly by the lamina terminalis, and laterally by the part of the medial surfaces of the frontal lobes located below the rostrum. The anterior incisural space opens laterally into the part of the sylvian fissure situated below the anterior perforated substance. The anterior limb of the internal capsule, the head of the caudate nucleus, and the anterior part of the lentiform nucleus are located above the anterior perforated substance. The interpeduncular cistern, which sits in the posterior part of the anterior incisural space between the cerebral peduncles and the dorsum sellae, communicates laterally with the sylvian cistern below the anterior perforated substance and anteriorly with the chiasmatic cistern located below the optic chiasm. The interpeduncular and chiasmatic cisterns are separated by Liliequist's membrane, an arachnoidal sheet ex-

Figure 3.8. A-C. Tentorium and inclsura: stepwise dissection. A. Right super-olateral view. The right cerebral hemisphere has been removed to expose the tentorium (Tent.), falx, and medial surface of the right cerebral hemisphere. The tentorial edge (Tent. Edge) slopes downward from the apex (Apex Tent.) along the side of the brain stem to the area lateral to the dorsum sellae (Dorsum). The third ventricle (3 Vent.) has been divided in the midline and the right half of the upper midbrain has been divided transversely to expose the cerebral peduncle (Ped.), red nucleus (Red Nucl.), substantia nigra (Subst. Nigra), and inferior colliculi (Inf. Coll.). The midbrain is in the center of the incisura. The anterior incisural space is located anterior to the midbrain and extends downward between the dorsum sellae and brain stem and upward around the optic chiasm to the area anterior to the lamina terminalis (Lam. Term.). The optic nerves (Optic N.), carotid arteries (Car. A.), and infundibulum (Infund.) of the pituitary gland cross the anterior incisural space. The middle incisural space is located lateral to the midbrain. It extends into the supratentorial area between the temporal lobe and the midbrain and into the infratentorial area between the cerebellum and the brain stem. The posterior incisural space is located between the apex of the tentorium posteriorly, the midbrain anteriorly, the splenium above, and the cerebellum below. The corpus callosum (Corp. Call.), septum pellucidum (Sept. Pell.), and fornix are located above the third ventricle. The cingulate gyrus (Cing. Gyr.) wraps around the corpus callosum and has the cingulate sulcus (Cing. Sulc.) and medial frontal gyrus (Med. Front. Gyr.) on its outer margin. The tentorium is attached anteriorly to the petrous ridge and temporal bone and posteriorly to the occipital bone. The posterior part of the falx has been removed to expose the medial surface of the parietal and occipital lobes, including the precuneus and cuneus, the parietooccipital (Par. Occip. Sulc.), and calcarine (Calc. Sulc.) sulci and the lingual gyrus (Lingual Gyr.). The olfactory tract (Olf. Tr.) passes posteriorly along the lateral margin of the gyrus rectus (Gyr. Rectus). The oculomotor nerve (III) arises on the medial surface of the cerebral peduncle and enters the dura mater lateral to the posterior clinoid process. The trochlear nerve (IV) arises below the inferior colliculus and passes forward in the cerebellomesencephalic fissure (Cer. Mes. Fiss.) between the cerebellum and midbrain. B. Superior view. The falx and the dura covering the left hemisphere have been removed. The left insula has been exposed by removing the operculum of the frontal lobe (Front. Lobe). The central sulcus separates the precentral and postcentral gyri and the frontal (Front. Lobe) and parietal (Par. Lobe) lobes. The straight sinus (Str. Sinus) passes backward between the occipital lobes (Occip. Lobe). The cerebellum forms the floor of the posterior incisural space. The trochlear nerve enters the anterior part of the free edge. The oculomotor nerve enters the roof of the cavernous sinus. The trigeminal nerve (V) arises in the infratentorial part of the middle incisural space. The frontal sinus (Front Sinus) is anterior to the frontal lobe and the floor of the anterior cranial fossa (Ant. Fossa). C. The superior part of the left cerebral hemisphere has been removed to expose the lateral ventricle (Lat. Vent.). The thalamus and the body of the lateral ventricle (Body) are located above the central part of the incisura; the frontal horn (Front. Horn) is above the anterior incisural space; and the atrium is above the posterior incisural space. The culmen and the quadrangular lobule (Quad. Lobule) of the cerebellum form the floor of the posterior incisural space. The forceps major, a bundle of white matter emanating from the splenium of the corpus callosum, sweeps posteriorly in the medial wall of the atrium and occipital horn (Occip. Horn) deep to the parietooccipital sulcus. The forceps minor, another bundle of white matter, radiates from the genu of the corpus callosum along the medial margin of the frontal horn. The caudate nucleus (Caudate Nucl.) and corona radiata are lateral to the frontal horn and body of the lateral ventricle. The choroid plexus (Chor. Plex.) is attached along the choroidal fissure, the cleft between the fornix and thalamus. The oculomotor nerve enters the roof of the cavernous sinus through the oculomotor trigone (Oculomotor Trig.), which is situated between three dural folds: the interclinoid fold (Interclin. Fold), which extends from the posterior (Post. Clinoid) to the anterior (Ant. Ca'noid) clinoid process, forms the medial margin; the anterior petroclinoid fold (Ant. Petroclin. Fold), which extends from the petrous apex to the anterior clinoid process, forms the lateral margin; and the posterior petroclinoid fold (Post. Petroclin. Fold), which extends from the petrous apex to the posterior clinoid process, forms the posterior margin. The septum pellucidum is in the medial wall of the lateral ventricle. The infundibulum passes through the diaphragma sellae (Diaph.).

Figure 3.8. D and E. Tentorium and incisura (continued). D. Another layer of the left cerebral hemisphere has been removed. The anterior limb (Ant. Limb) of the internal capsule (Int. Cap.) is between the head of the caudate nucleus and the putamen, and the posterior limb (Post. Limb) is between the thalamus and the putamen. The thalamostriate vein (Thal. Str. V.) courses between the caudate nucleus and the thalamus. The column of the fornix passes anterior to the foramen of Monro (For. Monro). Structures in the anterior wall of the atrium include the crus of the fornix, the pulvinar of the thalamus, and the tail of the caudate nucleus. The choroid plexus in the atrium is attached between the thalamus and the fornix. E. Deeper section through the left cerebral hemisphere. The caudate and lentiform nuclei, the thalamus, the internal capsule, and the body, frontal horn, and atrium of the lateral ventricle are above the tentorial incisura. The junction of the anterior and posterior limbs of the internal capsule abuts on the wall of the lateral ventricle between the caudate nucleus and thalamus in the area lateral to the foramen of Monro. The thalamus lies directly above the midbrain and the central part of the incisura. The head of the caudate nucleus, the anterior limb of the internal capsule, and the anterior part of the lentiform nucleus, formed by the putamen and globus pallidus (Globus Pall.), are above the anterior incisural space. The posterior limb of the internal capsule and the posterior part of the lentiform nucleus are above the middle incisural space. The atrium and splenium are directly above the posterior incisural space. (From Ono M, Ono M, Rhoton AL Jr, Barry M: Microsurgical anatomy of the region of the tentorial incisura. J Neurosurg 60:365-399, 1984.) tending from the dorsurn sellae to the anterior edge of the mamillary bodies. The chiasmatic cistern communicates around the optic chiasm with the cisterna laminae terminalis, which lies anterior to the lamina terminalis. The optic and oculomotor nerves and the posterior part of the olfactory tracts pass through the anterior incisural space (Figs. 3.7 to 3.9). Each olfactory tract runs posteriorly and splits just above the anterior clinoid process to form the medial and the lateral olfactory striae, which course along the anterior margin of the anterior perforated substance. The optic nerves and chiasm and the anterior part of the optic tracts cross the anterior incisural space. The optic nerves emerge from the optic canal medial to the attachment of the free edge to the anterior clinoid processes and are directed posterior, superior, and medial toward

the optic chiasm. From the chiasm, the optic tract continues in a posterolateral direction around the cerebral peduncle to enter the middle incisural space. The optic nerve proximal to its entrance into the optic canal is covered by a reflected leaf of dura mater, the falciform process, which extends medially from the anterior clinoid process across the top of the optic nerve (36). The length of nerve covered by dura only at the intracranial end of the optic canal may vary from less than 1 mm to as great as 1 cm. Coagulation of the dura above the optic nerve just proximal to the optic canal on the assumption that bone separates the dura mater from the nerve could lead to nerve injury. Compression of the optic nerves against the sharp edge of the falciform process may result in a visual field deficit even if the compressing lesion does not damage the nerve enough to cause visual loss.

The ophthalmic artery is found inferolateral to the optic nerve when the periosteum lining the optic canal is opened. Normally the optic nerve is separated medially from the sphenoid sinus by a thin plate of bone, but in a few cases this bone is absent and the optic nerves may protrude directly into the sphenoid sinus, separated from the sinus by only mucosa and the dural sheath of the nerve (7, 36). The relationship of the chiasm to the sella (Fig. 3.10) is an important determinant of the ease with which the pituitary fossa may be exposed by the transfrontal surgical route. The normal chiasm overlies the diaphragma sellae and the pituitary, the prefixed chiasm overlies the tuberculum sellae, and the postfixed chiasm overlies the dorsum sellae. In approximately 70% of cases the chiasm is in the normal position. Of the remaining 30%, about half are "prefixed" and half "postfixed" (36). A prominent tuberculum sellae may restrict access to the sellae even in the presence of a normal chiasm. The tuberculum may vary from being almost flat to protruding upward as much as 3 mm, and it may project posteriorly to the margin of a normal chiasm (36). The limited transfrontal approach exposure provided by a prefixed chiasm, a normal chiasm with a small area between the tuberculum and the chiasm, and a superior protruding tuberculum sellae may be enlarged by using the trans-frontal-transsphenoidal approach advocated by Rand (35). In this approach, access is obtained by opening into the sphenoid sinus from above by drilling through the tuberculum and planum sphenoidale, thus converting the approach to a "transfrontal-transsphenoidal" one. If the chiasm is normal or "postfixed," a subchiasmatic approach below the chiasm is possible. If the chiasm is prefixed and the tumor is seen through a thin, stretched anterior wall of the third ventricle, the lamina terminalis may be opened to expose the tumor. If the space between the carotid artery and the optic nerve has been enlarged (e.g., by a lateral or parasellar extension of tumor), the tumor may removed through this space (43). If more space is needed in the interval between the carotid artery and optic nerve, the carotid artery, rather than the optic nerve, may be gently retracted. An understanding of the relationship of the carotid artery, optic nerve, and anterior clinoid process is fundamental to all surgical approaches to the sellar and parasellar areas (Fig. 3.11). The carotid artery and the optic nerve are medial to the anterior clinoid process. The artery exits the

cavernous sinus beneath and slightly lateral to the optic nerve. The optic nerve pursues a pos-teromedial course toward the chiasm and the carotid artery pursues a posterolateral course toward its bifurcation into the anterior and middle cerebral arteries. The oculomotor nerve emerges from the mid-brain on the medial surface of the cerebral peduncle. It crosses the anterior incisural space between the posterior cerebral and the superior cerebellar arteries and passes inferomedial to the uncus to enter the roof of the cavernous sinus. Middle Incisural Space The middle incisural space is located lateral to the brain stem (Figs. 3.6 to 3.9). This narrow space, which extends upward between the temporal lobe and the midbrain, has medial and lateral walls and a roof. The medial wall is formed by the lateral surface of the midbrain. This surface of the midbrain is divided into a larger anterior part formed by the cerebral peduncle and a smaller posterior part formed by the tegmental surface. The tegmental surface is formed by the lemniscal trigone, a triangular area situated just posterior to the peduncle, and by the brachium of the inferior col-liculus, a prominence posterior to the lemniscal triangle that is directed upward from the inferior colliculus toward the medial geniculate body. The roof of the middle incisural space has a narrow anterior part formed by the posterior part of the optic tract, which is flattened between the cerebral peduncle and the uncus, and a wider posterior part formed by the inferior surface of the pulvinar. The lateral geniculate body protrudes from the inferior surface of the pulvinar just lateral to the posterior edge of the cerebral peduncle. The medial geniculate body bulges into the roof posteromedial to the lateral geniculate body just behind the lateral mesencephalic sul-cus. The lateral wall of the supratentorial part of the middle incisural space is composed of the hippocampal formation on the medial surface of the temporal lobe. The uncus and parahippocam-pal gyri, the most inferior structures in this part of the lateral wall, form a curved border around the middle incisural space. The uncus bulges medially at the anterior end of the parahippo-campal gyrus. Posterior to the uncus, the surface of the temporal lobe facing the middle incisural space is formed by three longitudinal strips of neural tissue, one located above the other, which are interlocked with the hippocampal formation to

Figure 3.9. A-C. Neural relationships. A. Lateral view. The temporal lobe (Temp. Lobe) has been elevated from the tentorium (Tent.) and the floor of the middle cranial fossa. The free edge (Tent. Edge) has grooved (Tent. Groove) the lower surface of the uncus. The rhinal (Rhinal Sulc.) and collateral (Coll. Sulc.) sulci separate the occipitotemporal (Occip. Temp. Gyr.) and the parahippocampal (Parahippo. Gyr.) gyri. The gyrus rectus (Gyr. Rectus) is medial and the orbital gyri (Orb. Gyr.) are lateral to the olfactory tract (Off. Tr.). The anterior perforated substance (Ant. Perf. Subst.) extends into the medial part of the sylvian fissure (Sylvian Fiss.). The calcarine sulcus (Calc. Sulc.) divides the medial surface of the occipital lobe. The optic tract (Optic Tr.) passes backward between the uncus and the cerebral peduncle (Ped.). The brachium of the inferior colliculus (Brach. Inf. Coll.) extends from the inferior colliculus (Inf. Coll.) along the side of the superior colliculus (Sup. Coll.) toward the medial geniculate body (Med. Gen. Body). The lemniscal trigone (Lemniscal Trig.) is located between the brachium of the inferior colliculus and the lateral mesencephalic sulcus (Lat. Mes. Sulc.). The oculomotor nerve (III) arises below the mamillary bodies (Mam. Body) just above the pontomesencephalic sulcus (Pon. Mes. Sulc.). B. Enlarged view. The temporal lobe has been elevated further to show the lower surface of the pulvinar and the termination of the optic tract in the lateral geniculate body (Lat. Gen. Body). The optic nerve (Optic N.) passes above the divided end of the carotid artery (Car. A.). The anterior hippocampal sulcus (Ant. Hippo. Sulc.) separates the uncus and the parahippocampal gyrus. The calcarine sulcus separates the lingual gyrus (Lingual Gyr.) and the cuneus. C. The anteromedial part of the temporal lobe has been removed to expose the temporal horn (Temp. Horn). The removal crosses the amygdaloid nucleus (Amygd. Nucl.), which lies deep to the uncus. The choroid plexus (Chor. Plex.) is attached along the choroidal fissure (Chor. Fiss.), a cleft between the temporal lobe and pulvinar. The olfactory tract splits into the medial (Med. Olf. Stria) and lateral olfactory striae (Lat. Olf. Stria).

Figure 3.9. D and E. Neural relationships (continued). D. Inferior view of basal structures surrounding the course of the basal vein. The cerebrum was removed by dividing the midbrain at the level of the red nucleus (Red Nad), substantia nigra (Subst. Nigra), and posterior perforated substance (Post. Perf. Subst.). The infundibulum (Infund.) arises from the tuber cinereum (Tuber). The uncus extends medial to and is grooved by the tentorial edge in the area anterolateral to the cerebral peduncle. The subiculum, the rounded medial edge of the parahip-pocampal gyrus, has been retracted to expose the dentate gyrus (Dentate Gyr.) and the fimbria of the fornix on the medial surface of the temporal lobe. The fimbria and dentate gyrus are separated by the fimbriodentate sulcus (Fimb. Dentate Sulc.), and the dentate gyrus and the subiculum are separated by the hippocampal sulcus (Hippo. Sulc.). The choroidal fissure opens into the temporal horn between the pulvinar and fimbria. The fibers in the fimbria are continuous with those in the crus of the fornix. The dentate gyrus is continuous posteriorly with the fasciolar gyrus (Fasciolar Gyr.), located below the splenium. E. The temporal tip has been retracted to expose the anterior perforated substance, which is bounded anteriorly by the medial and lateral olfactory striae, medially by the optic tract, posteriorly by the temporal lobe, and laterally by the limen insula. The semilunar (Semilunar Gyr.) and ambient (Ambient Gyr.) gyri, which are continuous posteriorly with the uncus, are on the superior surface of the temporal tip immediately posterior to the anterior perforated substance. (From Ono M, Rhoton AL Jr, Peace DA, Rodriguez R: Microsurgical anatomy of the deep venous system of the brain. Neurosurgery 15:621-657, 1984.)

Figure 3.10. Three sagittal sections of the sellar region showing neural structures, chiasm, posterior lobe pituitary and hypothalamus, anterior lobe pituitary, dura, and intracavernous venous connections. Prefixed chiasm above tuberculum (left), normal chiasm above diaphragma (center), and postfixed chiasm above dorsum (right). (From Rhoton AL Jr, Maniscalco J: Microsurgery of the sellar region. Neuroophthalmology 9:106-127, 1977.)

Figure 3.11. Right anterior and lateral views of the suprasellar area. A. Right anterolateral view. Anterior clinoid process (Ant. Clinoid), lateral to optic nerve (O.N.) and carotid artery (C.A.). Perforating arteries pass from the carotid to terminate in the optic chiasm and tract (O.Tr.) arid the hypothalamus anterior to the mamillary body (Mam. В.). Right posterior communicating artery (P.Co.A.) is medial to the carotid artery. Anterior choroidal artery (A.Ch.A.) arises from the carotid and passes above the posterior cerebral artery (P.C.A.). The third nerve (HI) lies below the right posterior cerebral artery. The right recurrent artery of Huebner (Rec. A.) arises from the anterior cerebral artery (A.C.A.) proximal to the anterior communicating artery (A.Co.A.). M.C.A., middle cerebral artery; A-2, anterior cerebral artery distal to the anterior communicating artery. B. Right lateral view with temporal lobe removed. Anterior clinoid (din.) is lateral to the carotid artery. Premamillary (Premam. A.) arises from the posterior communicating artery. Thalamoperfor-ating arteries (Th. Pe. A.) arise from the proximal posterior cerebral artery. The optic tract passes around the peduncle to the lateral geniculate body. Arterioles pass from the carotid artery to the optic nerve. The basilar artery (B.A.) is medial to the third and fourth (IV) cranial nerves. The trigeminal nerve (V) is below the superior cerebellar artery (S.C.A.). (From Saiki N, Rhoton AL Jr: Microsurgical anatomy of the upper basilar artery and the posterior circle of Willis. J Neu-rosurg 46:563-577, 1977.)

make an important part of the limbic system (Figs. 3.7 and 3.9). The most inferior strip is formed by the subiculum, the rounded medial edge of the parahippocampal gyrus; the middle strip is formed by the dentate gyrus, a serrated or beaded strip of gray matter located on the medial surface of the hippocampal formation; and the superior strip is formed by the fimbria of the fornix, a white band formed by the fibers emanating from the hippocampal formation that are directed posteriorly into the crus of the fornix. The choroidal fissure, the cleft along which the choroid plexus of the temporal horn is attached, is situated at the junction of the roof and lateral wall of the middle incisural space between the fimbria below and the pulvinar above. The inferior choroidal point, at the inferior end of the choroidal fissure, is located immediately lateral to the lateral geniculate body. The supratentorial part of the middle incisural space contains the crural and ambient cisterns. The crural cistern, located between the cerebral peduncle and the uncus, is a posterolateral extension of the interpeduncular cistern. The crural cistern opens posteriorly into the ambient cistern, a narrow communicating channel demarcated medially by the midbrain, above by the pulvinar, and laterally by the subiculum, dentate gyrus, and fimbria of the fornix. The ambient cistern is continuous posteriorly with the quadrigeminal cistern. The trochlear nerve is related predominatly to the middle incisural space and is the cranial nerve most intimately related to the free edge. The trochlear nerve arises below the inferior colliculus in the posterior incisural space and passes forward through the middle incisural space between the posterior cerebral and superior cerebellar arteries. Its initial course around the midbrain is medial to the free edge in the space between the tectum and cerebellum. It reaches the lower margin of the free edge at the posterior edge of the cerebral peduncle. It pierces the anterior part of the free edge before entering the cavernous sinus. Posterior Incisural Space The posterior incisural space lies posterior to the midbrain and corresponds to the pineal region (Figs. 3.6 to 3.8). It has a roof, floor, and anterior and lateral walls and extends backward to the level of the apex of the tentorium. The quadrigeminal plate is located at the center of the anterior wall. The anterior wall rostral to

the colliculi is formed by the pineal body and the habenular trigones and commissure. The habenular trigones overlie the habenular nuclei on the posteromedial surface of the pulvinar and are connected across the midline by the habenular commissure. The habenular trigones are continuous anteriorly with the striae medullaris thai-ami, which course along the medial surface of the thalamus. The habenular commissure forms the upper half and the posterior commissure forms the lower half of the attachment of the pineal body to the posterior part of the third ventricle. The part of the anterior wall below the colliculi is formed in the midline by the lingula of the vermis and laterally by the superior cere-bellar peduncles as they ascend beside the lingula. The roof of the posterior incisural space is formed by the lower surface of the splenium, the terminal part of the crura of the fornix, and the hippocampal commissure. The floor of the posterior incisural space is formed by the anterior superior part of the cerebellum. This space extends inferiorly into the cerebellomesencephalic fissure, the cleft opening inferiorly between the anterior superior part of the vermis and the colliculi. The anterior wall of the fissure is formed by the lingula and superior cerebellar peduncles, and the posterior wall is formed by the cerebellum. Each lateral wall is formed by the pulvinar, crus of the fornix, and medial surface of the cerebral hemisphere. The anterior part of the lateral wall is formed by the part of the pulvinar located just lateral to the pineal body. The lateral wall, posterior to the pulvinar, is formed by the segment of the crus of the fornix that wraps around the posterior margin of the pulvinar. The posterior part of the lateral walls is formed by the cortical areas located below the splenium on the medial surface of the hemisphere. These areas include the posterior part of the parahippocampal and dentate gyri and several small nodular collections of gray matter, the fasciolar gyri and Retzius' gyri, which are located at the posterior end of the dentate gyrus (Fig. 3.7). The quadrigeminal cistern, situated posterior to the quadrigeminal plate, is the major cistern in the posterior incisural space. It has also been referred to as the cisterna venae magnae cerebri because the vein of Galen passes through it. The quadrigeminal cistern communicates above with the posterior pericallosal cistern, which extends around the splenium between the cerebral hemispheres; inferiorly into the cerebellomesencephalic cistern, also called the precentral cerebel-

lar cistern, which extends into the cerebellomesencephalic fissure; inferolaterally into the posterior part of the ambient cistern located between the midbrain and the parahippocampal gyrus; and laterally into the retrothalamic cistern, which curves around the posterior margin of the pulvinar medial to the crus of the fornix. The quadrigeminal cistern may communicate with the velum interpositum. Another cavity, the cavum vergae, is located just above the velum interpositum between the hippocampal commissure and the splenium: its roof is formed by the lower surface of the splenium, its floor is formed by the hippocampal commissure, and the lateral walls are formed by the crura. Substances instilled in the subarachnoid space only infrequently enter a cavum vergae. Arterial Relationships Each wall of the third ventricle has surgically important arterial relationships: the posterior part of the circle of Willis and the apex of the basilar artery are below the floor; the anterior part of the circle of Willis and the anterior cerebral and anterior communicating arteries are intimately related to the anterior wall; the posterior cerebral, pericallosal, superior cerebellar, and posterior choroidal arteries pass adjacent to the posterior wall; both the anterior and posterior cerebral arteries send branches into the roof; and the internal carotid, anterior choroidal, anterior and posterior cerebral, and anterior and posterior communicating arteries give rise to perforating branches that reach the walls of the third ventricle. All arterial components of the circle of Willis and the adjacent carotid and basilar artery and their perforating branches may become stretched around third ventricular and selar tumors. The arterial relationships in the anterior incisural space are among the most complex in the brain because it contains all the components of the circle of Willis (Figs. 3.11 to 3.13). Internal Carotid Artery The internal carotid artery exits the cavernous sinus along the medial surface of the anterior clinoid process to reach the anterior incisural space. After entering this space it courses posterior, superior, and lateral to reach the site of its bifurcation below the anterior perforated substance. It is first below and then lateral to the optic nerve and chiasm (Figs. 3.11, 3.13, and 3.14). It sends perforating branches to the optic nerve, chiasm, and tract and to the floor of the third ventricle. These branches pass across the interval between the internal carotid artery and

Figure 3.12. Arterial relationships of the third ventricle. A and С are inferior views of the floor of the third ventricle, and В and D are midsagittal sections through the third ventricle. A and В show the relationship of the main trunks and perforating branches of the following arteries to the third ventricle: internal carotid (C.A.), anterior choroidal (A.Ch.A.}, basilar apex (B.A.), posterior cerebral (P.C.A.), medial posterior choroidal (Med.Post.Ch.A.), lateral posterior choroidal (Lat.Post.Ch.A.), thalamoperforating (Thal. Perf.A.), and thalamogeniculate (Thal.Gen.A.) arteries. С and D show the relationships of the main trunks and perforating branches of the following arteries to the third ventricle: anterior cerebral (A.C.A.), anterior communicating (A..Co.A.), and posterior communicating arteries. The olfactory (Olf.N.) and optic (O.N.) nerves are anterior to the floor of the third ventricle. The structures in the floor are the optic chiasm (O.Ch.), optic tracts (O.Tr.), infundibulum (Infund.), tuber ci-nereum (Tuber Cin.), and mamillary bodies (Mam. В.). The midbrain and cerebral peduncles (Ped.) are inferior to the posterior half of the floor. The anterior perforated substance (Ant. Perf. Subst.) is lateral to the optic tracts. The lateral geniculate (Lat. Gen. B.) and medial geniculate (Med. Gen. B.) bodies are attached to the lower margin of the thalamus near the pulvinar (Pulv.}, lateral to the midbrain. The structures in the anterior wall of the third ventricle are the anterior commissure (Ant. Comm.), lamina terminalis (Lam. Теr.), and optic chiasm. The corpus callosum (Corp. Call.) and septum pellucidum (Sept. Pel.) are above the roof of the third ventricle. The roof is formed of the two layers of tela choroidea (Tela), the fornix, and a vascular layer composed of the internal cerebral veins and the medial posterior choroidal arteries. The oculomotor nerve (HI) exits from the midbrain. (From Yamamoto I, Rhoton AL Jr, Peace DA: Microsurgery of the third ventricle: Part 1. Microsurgical anatomy. Neurosurgery 8:334-356, 1981.)

the optic nerve and may serve as an obstacle to the operative approaches directed through the triangular space between the internal carotid artery, the optic nerve, and the anterior cerebral artery. The internal carotid artery also gives off the superior hypophyseal artery, which runs medially below the floor of the third ventricle to reach the tuber cinereum and joins its mate of the opposite side to form a ring around the infundibulum (10). The supraclinoid (C4) portion of the internal carotid artery is divided into three segments based on the origin of its major branches: the ophthalmic segment extends from the origin of the ophthalmic artery to the origin of the posterior communicating artery; the communicating segment extends from the origin of the posterior communicating artery to the origin of the anterior choroidal artery; and the choroidal segment extends from the origin of the anterior choroidal artery to the bifurcation (Figs. 3.15 to 3.17) (10). Each segment gives off a series of perforating branches with a relatively constant site of termination. The branches arising from the ophthalmic segment pass to the optic nerve and chiasm, infundibulum, and the floor of the third ventricle. The branches arising from the communicating segment pass to the optic tract and the floor of the third ventricle. The branches arising from the choroidal segment pass upward and enter the brain through the anterior perforated substance. Ophthalmic Artery The ophthalmic artery is the first branch of the internal carotid artery above the cavernous sinus (Figs. 3.14 and 3.16). Its origin and proximal segment may be visible below the optic nerve without retracting the nerve, although elevation of the optic nerve away from the carotid artery is usually required to see the segment proximal to the optic foramen. The artery arises from the supraclinoid segment at the carotid artery in most cases, but it may also arise within the cavernous sinus or be absent in a few cases (14, 36). The ophthalmic artery origin has a variable relationship to the tip of the anterior clinoid process. Its origin may vary from as far as 5 mm anterior to 7 mm posterior to the clinoid tip and 2 to 10 mm medial to the clinoid process (14). Most arise anterior and approximately 5 mm medial to the tip. In our studies, 14% of ophthalmic arteries exited from the carotid artery and immediately entered the optic canal; in the remaining 86% the maximal preforaminal length was 7 mm and the mean was 3 mm (14).

Posterior Communicating Artery The posterior communicating artery arises from the posterior wall of the internal carotid artery and courses posteromedially below the optic tracts and the floor of the third ventricle to join the posterior cerebral artery (Figs. 3.11 to 3.19). Its branches penetrate the floor between the optic chiasm and the cerebral peduncle and reach the thalamus, hypothalamus, subthala-mus, and internal capsule. Its posterior course varies depending on whether the artery provides the major supply to the distal posterior cerebral artery. If it is normal, with the posterior cerebral artery arising predominantly from the basilar artery, it is directed posteromedially above the oculomotor nerve toward the interpeduncular fossa. If the posterior cerebral artery has a fetal type configuration in which it arises from the carotid artery, the posterior communicating artery is directed posterolaterally below the optic tract. The oculomotor nerve pierces the dura mater of the roof of the cavernous sinus 2 to 7 mm (average, 5 mm) posterior to the initial supracli-noid segment of the carotid artery (14, 42). Anterior Choroidal Artery The anterior choroidal artery arises from the posterior surface of the internal carotid artery 0.1 to 3.0 mm above the origin of the posterior communicating artery (Figs. 3.11 to 3.20). It is directed posterolaterally below the optic tract between the uncus and cerebral peduncle. It passes through the choroidal fissure near the inferior choroidal point to supply the choroid plexus in the temporal horn. It sends branches into the optic tract and posterior part of the floor that reach the optic radiations, globus pallidus, internal capsule, midbrain, and thalamus (8, 38). Anterior Cerebral and Anterior Communicating Arteries The anterior cerebral artery arises from the internal carotid artery below the anterior perforated substance and courses anteromedially above the optic nerve and chiasm to reach the interhemispheric fissure, where it is joined to the opposite anterior cerebral artery by the anterior communicating artery (Figs. 3.11 to 3.20) (30, 31, 37). The junction of the anterior communicating artery with the right and left A1 segments is usually above the chiasm rather than above the optic nerves. In our studies 70% were above the chiasm and 30% were in a prefixed position above the optic nerves (Fig. 3.21). The shorter A1

segments are stretched tightly over the chiasm, and the longer ones pass anteriorly over the nerves. The arteries with a more forward course are often tortuous and elongated, and some may course forward and rest on the tuberculum sellae or planum sphenoidale (Fig. 3.21). The anterior cerebral artery ascends in front of the lamina terminalis and the anterior wall of the third ventricle and passes around the corpus callosum. It gives off the orbitofrontal artery before reaching the corpus callosum. Often the anterior cerebral artery on one side sends branches across the interhemispheric fissure to supply the medial part of the opposite cerebral hemisphere (31). The distal part of the anterior cerebral artery may be exposed not only above, but also below the corpus callosum because the terminal branch of the pericallosal artery may pass around the splenium and course forward in the roof of the third ventricle, reaching as far anterior as the foramen of Monro (31). The anterior cerebral and anterior communicating arteries give rise to perforating branches that terminate in the whole anterior wall of the third ventricle and reach the adjacent parts of the hypothalamus, fornix, septum pellucidum, and striatum (Figs. 3.22 and 3.23). A precallosal artery may originate from the anterior cerebral or the anterior communicating artery, run upward across the lamina terminalis, and send branches into the anterior wall (Fig. 3.22D). The pericallosal arteries send short and long callosal arteries into the corpus callosum. The short callosal arteries pass directly into the corpus callosum. The long callosal arteries course parallel to the pericallosal arteries, between them and the corpus callosum. The callosal arteries send branches into the corpus callosum that reach the septum pellucidum and the fornix. The recurrent branch of the anterior cerebral artery, which is referred to as the recurrent artery of Heubner, is encountered frequently in approaches to the anterior part of the third ventricle (Figs. 3.11, 3.22, and 3.23). It arises from the anterior cerebral artery in the region of the anterior communicating artery, courses laterally above the bifurcation of the internal carotid artery, and enters the anterior perforated substance (30). The recurrent artery courses anterior to the A1 segment of the anterior cerebral artery and would have been seen when elevating the frontal lobe before visualizing the A1 segment in about two-thirds of cases. Of the remaining one-third, most coursed superior to A1 Some of its branches reach the anterior limb and genu of the internal capsule.

Middle Cerebral Artery The middle cerebral artery courses laterally from its origin below the anterior perforated substance (Figs. 3.11, 3.13, 3.15, 3.18, and 3.19) (9). The major bifurcation of the middle cerebral artery is usually located in the lateral part of the anterior incisural space. The middle cerebral artery has been divided into four segments, M1 through M4 (9). The M1 segment begins at the origin of the middle cerebral artery lateral to the optic chiasm, at the medial end of the sylvian fissure, and extends laterally below the anterior perforated substance, toward the insula. The M1 segment terminates at the limen insulae, the site of a 90° turn, the genu, where the artery curves sharply posterosuperior to form the M2 segment, composed of the trunks coursing over the insula. The M1 segment is subdivided into a prebifurcation and a postbifurcation part. The major division of the artery is a bifurcation in 86% of hemispheres and a trifurcation in the remaining 14% of hemispheres (9). The bifurcation occurs proximal to the genu in most hemispheres.

The middle cerebral artery branches to the anterior perforated substance are called the len-ticulostriate arteries (44). The lenticulostriate arteries are divided into medial, intermediate, and lateral groups, each of which has a unique origin, composition, morphology, and characteristic distribution in the anterior perforated substance. The medial group of lenticulostriate arteries arises on the medial part of the M1 segment near the carotid bifurcation and pursues a relatively direct course to enter the anterior perforated substance. The intermediate lenticulostriate arteries form a complex array of branches resembling a candelabra before entering the anterior perforated substance between the medial and lateral lenticulostriate arteries. The lateral lenticulostriate arteries originate predominantly on the lateral part of the M1 portion, pursue an S-shaped course, and enter the lateral part of the anterior perforated substance. They arise from the parent trunks, travel medially with the parent trunks, then loop sharply posterior, lateral, and superior, and finally turn posteromedial just before penetrating the anterior perforated substance.

Figure 3.13. A and B. Superior view of the arteries in the suprasellar area. A. The posterior part of the optic chiasm is split at the junction with the optic tracts (O. Tr.) to give this view. The proximal portions of the anterior cerebral arteries (A-1) arise from the carotid arteries (C.A.) and pass above the optic nerves (O. N.) and chiasm; the anterior third ventricle lies above the mamillary bodies (Mam. В.). The thalamoperforating arteries (Th.Pe.A.) terminate in the retromamillary area and the interpeduncular fossa. The right posterior choroidal artery (P.Ch.A.) originates from the posterior cerebral artery (P.C.A.). Small branches arise from the posterior cerebral artery and terminate in the peduncle. The left third nerve (III) courses medial to the posterior communicating artery (P.Co.A.). The right posterior communicating artery joins the posterior cerebral artery medial to the third nerve. The left anterior choroidal artery (A.Ch.A.) passes to the peduncle and optic tract. Trigeminal nerves (V) are lateral to the carotid artery. M.C.A., middle cerebral artery. B. Inferior view of the circle of Willis. Fetal type of left posterior cerebral artery. The left third nerve courses under the posterior cerebral artery distal to its junction with the posterior communicating artery and the right nerve courses under the part proximal to the communicating artery. The left thalamoperforating and the posterior choroidal (P.Ch.A.) arteries originate from the proximal part of the posterior cerebral artery. The right premamillary artery (Premam. A.) emerges from the anterior and the left from the middle third of the posterior communicating artery. Small posterior communicating branches course superiorly and medially, terminating in the premamillary area. The left thalamoperforating and right premamillary arteries are well developed in spite of the small trunk of origin. The superior cerebellar artery (S. C. A.) arises below the posterior cerebral artery. No branches arise on the anterior surface of the basilar artery (B.A.) A-2, distal part of the anterior cerebral artery. C. Superior view of the suprasellar area. Arterial branches stretch around the superior extension of a pituitary tumor. Anterior cerebral arteries send branches to the superior surface of the optic nerves and chiasm. The posterior communicating, internal carotid, and posterior cerebral arteries send branches into the area below and behind the chiasm. Recurrent arteries (Rec. A.) arise just distal to the anterior communicating artery. IV, trochlear nerve. (From Saeki N, Rhoton AL Jr: Microsurgical anatomy of the upper basilar artery and the posterior circle of Willis. J Neurosurg 46:563-577, 1977.)

Figure 3.14. Inferior views of the perforating branches of the supraclinoid portion of the internal carotid artery (ICA). A. The supraclinoid portion of the ICA gives rise to the ophthalmic (Ophth. A.), posterior communicating (P.Co.A.), and anterior choroidal (A.Ch.A.) arteries. The supraclinoid portion of the artery is divided into three segments based on the site of origin of these three branches: the ophthalmic segment (C4-Op.) extends from the origin of the ophthalmic artery to the Origin of the PCoA; the communicating segment (C4-Co.) extends from the origin of the PCoA to the origin of the AChA; and the choroidal segment (C4-Ch.) extends from the origin of the AChA to the bifurcation of the carotid artery into the anterior and middle cerebral arteries. The optic nerves (O.N.) are above the ophthalmic arteries. The optic chiasm and optic tracts (O.Tr.) are above the anterior (Ant.Lobe) and posterior (Post. Lobe) lobes of the pituitary gland. The tuber cinereum (Tuber Cin.) is anterior to the apex of the basilar artery (B.A.). The posterior cerebral arteries (P.C.A.) pass around the cerebral peduncles (Cer. Ped.) above the oculomotor nerves (III). The perforating branches arising from the ophthalmic segment pass superior to the anterior lobe to the optic nerve and chiasm and to the anterior part of the tuber cinereum. A single perforating branch arises from the communicating segment on each side and passes upward to the optic tract and the floor of the third ventricle. B. The anterior lobe of the pituitary gland has been reflected backward to show the superior hypophyseal arteries (Sup. Hyp. A.) passing from the ophthalmic segment to the infundibulum (Infund.). The anterior cerebral (A.C.A.) and the anterior communicating (A.Co.A.) arteries are seen above the optic chiasm (O.Ch.). (From Gibo H, Lenkey C, Rhoton AL

Posterior Cerebral Artery The bifurcation of the basilar artery into the posterior cerebral arteries is located in the posterior part of the anterior incisural space below the posterior half of the floor of the third ventricle (Figs. 3.11 to 3.20) (45, 57). A high basilar bifurcation may indent the floor. The posterior cerebral artery encircles the mesencephalon to reach the posterior incisural space and the quad-rigeminal cistern. It courses laterally around the cerebral peduncle, above the oculomotor nerve, and exits the anterior and enters the middle incisural space by coursing between the uncus and the cerebral peduncle. It enters the posterior incisural space from anteriorly and courses through the lateral part of the posterior incisural space. It bifurcates into the parietooccipital and calcarine arteries as it crosses above the free edge of the tentorium. It sends cortical branches to the temporal, occipital, and parietal lobes and callosal branches to the splenium of the corpus callosum. Its callosal branch, the posterior peri-callosal artery, may anastomose with the callosal branches of the anterior cerebral artery. Its branches reach the floor, roof, and posterior and lateral walls of the third ventricle. The thalamogeniculate and the thalamoperfor-ating arteries are two of the larger perforating branches (Figs. 3.13 and 3.19). The thalamoper-forating arteries arise from the proximal part of the posterior cerebral arteries and the posterior part of the posterior communicating arteries, enter the brain through the posterior part of the floor and the lateral walls. The thalamogeniculate arteries arise from the posterior cerebral arteries in the ambient cisterns below the pulvi-nar and course superiorly through the geniculate bodies to reach the adjacent parts of the thala-mus and internal capsule. The medial posterior choroidal arteries arise from the proximal portions of the posterior cerebral arteries in the anterior incisural space and course around the midbrain in the middle incisural space to reach the quadrigeminal cistern in the posterior incisural space (Figs. 3.12 and 3.18 to 3.20) (8, 38). They turn forward at the side of the pineal body and pass above the habenular trigone, course between the layers of the tela choroidea in the velum interpositum, and supply the choroid plexus in the roof of the third ventricle and the body of the lateral ventricle. The

Jr: Microsurgical anatomy of the supraclinoid portion of the internal carotid artery. J Neurosurg 55:560-574, 1981.)

lateral posterior choroidal arteries arise from each posterior cerebral artery in the area lateral to the midbrain and pass superolaterally through the choroidal fissure and around the pulvinar to supply the choroid plexus in the temporal horn and atrium. The posterior choroidal arteries send small branches to the surface of the thalamus along their course. The perforating branches of the posterior cerebral and superior cerebellar arteries and the medial posterior choroidal arteries supply the walls of the posterior incisural space. The posterior cerebral arteries supply the structures above the level of the lower margin of the superior colliculi, and the superior cerebellar arteries supply the structures below the upper margin of the inferior colliculus. There are numerous anastomoses between the two arteries over the surface of the colliculi. The pineal gland and the habenula are supplied by the branches of the medial posterior choroidal arteries.

Figure 3.15. Inferior views of the perforating branches of the supraclinoid portion of the internal carotid artery (ICA). A. The supraclinoid portion of the artery gives rise to the posterior communicating (P.Co.A.), anterior choroidal (A.Ch.A.), middle cerebral (M.C.A.), and anterior cerebral arteries (A.C.A.). The supraclinoid portion of the artery is divided into three segments based on the site of origin of these branches. An ophthalmic segment (C4-Op.) that extends from the origin of the ophthalmic artery (not shown because the ICA was divided above the level of origin of the ophthalmic artery) to the origin of the PCoA; a communicating segment (C4Co.) that extends from the origin of the PCoA to the origin of the AChA; and a choroidal segment (C4-Ch.) that extends from the origin of the AChA to the level of the bifurcation of the ICA into the anterior cerebral and middle cerebral arteries. The ophthalmic segment sends perforating branches to the optic nerves (O.N.), optic chiasm (O.Ch.), and tuber cinereum (Tuber Cin.). The superior hypophyseal arteries (Sup. Hyp. A.) pass to the infundibulum of the hypophysis (Infund.). The communicating segment sends one perforating branch on each side to the optic tracts (O.Tr.) and the region around the mamillary bodies (Mam. Body). The choroidal segment sends its perforating branches into the anterior perforated substance (Anr. Perf. Subst.). The posterior cerebral arteries (P.C.A.) arise from the basilar artery (B.A.) and pass laterally around the cerebral peduncles (Cer. Ped.). The temporal lobe (Temp. Lobe) is lateral to the carotid artery. The middle cerebral arteries pass laterally into the sylvian fissure below the anterior perforated substance. The frontal lobes (Fr. Lobe), gyrus rectus (Gyr. Rectus), and olfactory nerves (Olf. N.) are above the optic nerves. The thalamoperforating

Superior Cerebellar Artery This artery arises from the basilar artery in the posterior part of the anterior incisural space, encircles the midbrain below the posterior cerebral artery, and passes through the quadri-geminal cistern to reach the superior surface of the cerebellum (Figs. 3.18 and 3.19) (13). It dips below the tentorium to reach the superior surface of the cerebellum at the junction of the anterior and middle incisural spaces. It bifurcates into a rostral trunk that supplies the vermis and a caudal trunk that supplies the cerebellar hemisphere as it passes around the lateral margin of the cerebral peduncle to enter the middle incisural spaces. The segment of the artery in the quadrigeminal cistern is exposed in the supra- and infratento-rial operative approaches to the posterior part of the third ventricle, and its cortical branches are exposed in the infratentorial approaches. Perforating branches arising in the quadrigeminal region supply the inferior colliculi.

arteries (Thal. Perf. A.) pass posteriorly between the oculomotor nerves (Ш). B. Inferior view of another specimen. The ophthalmic segment gives rise to perforating branches that pass to the optic nerves, chiasm, and tracts and to the infundibulum and tuber cinereum. No perforating branches arise from the communicating segment on either side in this specimen. The perforating branches from the choroidal segment pass into the anterior perforated substance. (From Gibo H, Lenkey C, Rhoton AL Jr: Microsurgical anatomy of the supraclinoid portion of the internal carotid artery. J Neurosurg 55:560-574, 1981.)

Figure 3.16. Anterior and anteroinferior views of the supraclinoid portion of the internal carotid artery (ICA). A. Anterior view. The optic nerves (O.N.) enter the optic canals medial to the anterior clinoid processes (Ant Clinoid). The infundi-bulum (Infand.) passes inferiorly below the optic chiasm (O.Ch.) to the pituitary gland. The carotid arteries (C.A.) are posterior to the optic nerves. The planum sphenoidale (Planum) is anterior to the chiasmatic sulcus (Ch. Sulc.) and the tuberculum sellae (Tuberculum). The perforating branches of the carotid artery pass medially in the subchiasmatic space. The superior hypophyseal arteries (Sup. Hyp. A.) arise from the carotid artery and pass to the infundibulum. The falciform process (Falc. Process) is a fold of dura mater that passes above the optic nerve proximal to the optic foramen. B. The right optic nerve has been divided at the optic foramen and elevated to show the perforating branches of the supraclinoid portion of the carotid arteries. The right anterior cerebral artery (A.C.A.) was divided at its origin so that the optic nerve and chiasm could be elevated. The carotid artery gives rise to multiple perforating branches as well as the ophthalmic (Ophth. A.), posterior communicating (P.Co.A.), anterior choroidal (A.Ch.A.), and middle cerebral arteries (M.C.A.). The supraclinoid portion of the ICA is divided into three segments based on the origin of its major branches: the ophthalmic segment (C4-Op.) extends from the origin of the ophthalmic artery to the origin of the PCoA, the communicating segment (C4-Co.) extends from the origin of the PCoA to the origin of the AChA, and the choroidal segment (C4-Ch.) extends from the origin of the AChA to the bifurcation of the carotid artery. The perforating branches arising from the ophthalmic segment pass to the optic nerve, chiasm, infundibulum, and floor of the third ventricle. The perforating branches arising from the communicating segment pass to the optic tract and the floor of the third ventricle. The perforating branches arising from the choroidal segment pass upward and enter the brain through the anterior perforated substance (Ant. Perf. Subst.) The diaphragma sellae (Diaph.) surrounds the infundibulum above the pituitary gland. The temporal lobe (Temp. Lobe) is below the middle cerebral artery. C. The left optic nerve has been divided at the optic foramen and the anterior cerebral artery has been divided near its origin so that both optic nerves and the chiasm and tract could be elevated to show the perforating branches of the carotid artery. The Liliequist membrane (Lilieq. Memb.) is posterior to the infundibulum and hides the basilar artery but not the posterior cerebral artery (P.C.A.). The perforating branches of the ophthalmic segment pass upward to the infundibulum and the optic nerve, chiasm, and tract. D. Both optic nerves and both anterior cerebral arteries and the infundibulum have been divided to permit the optic nerves and chiasm to be elevated with a forceps for this view under the optic chiasm and across the diaphragma sellae and dorsum to the upper part of the basilar artery (B.A.) and the oculomotor nerves (III). The oculomotor nerves pass forward below the posterior cerebral arteries. The perforating branches of the supraclinoid segment of the carotid artery pass upward to supply the infundibulum, the optic chiasm and tracts, and the floor of the third ventricle; some enter the brain through the anterior perforated substance. The right AChA is very large. (From Gibo H, Lenkey C, Rhoton AL Jr: Microsurgical anatomy of the supraclinoid portion of the internal carotid artery. J Neurosurg 55:560-574, 1981.1

Figure 3.17. Posterior views of the perforating branches of the supraclinoid portion of the internal carotid artery (ICA). A. The basilar artery (B.A.) and brain stem have been divided and the floor of the third ventricle has been elevated to provide this posterior view of the supraclinoid portion of the ICA. The optic nerves (O.N.) can be followed back to the optic chiasm and the optic tracts (O.Tr.). The infundibulum of the pituitary gland (Infund.) arises from the tuber ciner-eum (Tuber Cin.). The mamillary bodies (Mam. Bodies) lie posterior to the tuber cinereum. The supraclinoid portion of the ICA gives rise to the posterior communicating (P.CO.A.), anterior choroidal (A.Ch.A.), middle cerebral (M.C.A.), and anterior cerebral (A.C.A.) arteries. The PCoA can be followed backward to its junction with its posterior cerebral arteries (P.C.A.). The supraclinoid portion of the ICA is divided into three segments on the basis of the origin of its major branches: the ophthalmic segment (C4-Op.) extends from the origin of the ophthalmic artery to the origin of the PCoA, the communicating segment (Cr-Co.) extends from the origin of the PCoA to the origin of the AChA, and the choroidal segment (C4-Ch.) extends from the origin of the AChA to the bifurcation into the anterior and middle cerebral arteries. The perforating branches arising from the ophthalmic segment in this specimen enter the optic tract, the floor of the third ventricle, and the infundibulum. The superior hypoph-yseal arteries (Sup. Hyp. A.) pass to the infundibulum. A single perforating artery arises from the communicating segment on each side. The posterior wall of the choroidal segment is the source of numerous perforating branches. The pituitary stalk passes through the diaphragma sellae (Diaph.) anterior to the dorsum sel-lae (Dorsum). B. The right half of the dorsum sellae and the right posterior clinoid process (Post. Clinoid) have been removed to expose the pituitary gland and its anterior (Ant. Lobe) and posterior (Post. Lobe) lobes. The infraclinoid segment of the carotid artery (C.A.) is exposed to the right of the pituitary gland. The apex of the basilar artery and the proximal parts of the posterior cerebral and the superior cerebellar (S.C.A.) arteries have been elevated to expose the pituitary stalk and floor of the third ventricle. The intracavernous portion of the carotid artery gives rise to the inferior hypophyseal (Inf. Hyp. A.) and the tentorial (Tent. A.) arteries. The posterior half of the right PCoA is duplicated. The ophthalmic segments give rise to the superior hypophyseal arteries. The left carotid artery has been divided just distal to the origin of the PCoA. C. Posterior views of the anterior part of the circle of Willis. The optic chiasm has been divided posterior to its junction with the optic nerves and anterior to where the infundibulum arises from the floor of the third ventricle. The superior hypophyseal arteries pass medially from the carotid artery to the infundibulum and also give rise to branches coursing to the lower surface of the optic chiasm. The communicating segment gives rise to one perforating branch on the right and two on the left. The choroidal segment gives rise to more perforating branches than the other segments, but they have been divided near their origin. The anterior cerebral artery gives rise to perforating branches coursing to the upper surface of the optic chiasm and nerves. The anterior communicating artery (A.Co.A.) is seen above the optic chiasm. (From Gibo H, Lenkey C, Rhoton AL Jr: Microsurgical anatomy of the supraclinoid portion of the internal carotid artery. J Neuro-surg 55:560574, 1981.)

Figure 3.18. Arterial relationships. A. Right lateral view. The temporal lobe (Temp. Lobe) has been retracted to expose the upper surface of the tentorium (Tent.) and the tentorial edge (Tent. Edge). The uncus, when not retracted, hangs over the free edge antero-lateral to the cerebral peduncle (Ped.). The collateral (Coll. Sulc.) and rhinal sulci (Rhinal Sulc.) separate the parahippocampal (Parahipp. Gyr.) and occipito-temporal gyri (Occip. Temp. Gyr.). The carotid artery (Car. A.) bifurcates below the anterior perforated substance (Ant. Perf. Subst.). The middle cerebral artery (M. C. A.) courses in the sylvian fissure (Sylvian Fiss.) between the temporal and frontal lobes (Front Lobe) and anterior cerebral artery (A.C.A.), which passes between the optic nerve (Optic N.) and olfactory tract (Olf. Tr.). The posterior communicating artery (Post Comm. A.) lies above the oculomotor nerve (III). The anterior choroidal artery (Ant Chor. A.) courses below the optic tract (Optic Tr.). The basilar artery (Bas. A.) gives off the superior cerebellar (S.C.A.) and posterior cerebral arteries (P.C.A.) in the anterior incisural space. The PCA gives off long (Long Circ. A.) and short circumflex (Short Circ. A.) arteries, anterior (Ant Temp. A.), middle (Mid. Temp. A.), and posterior temporal (Ant Temp. A.) middle (Mid. Temp. A.), and posterior temporal (Post Temp. A.) arteries, and medial posterior choroidal (Med. Post. Chor. A.) and cal-carine (Calc. A.) arteries. The SCA bifurcates into rostral (Ro. Tr.) and caudal trunks (Ca. Tr.). The superior (Sup. Coll.) and interior (Inf. Coll.) colliculi are supplied by branches of the SCA and PCA. The lateral mesencephalic sulcus (Lat Mes. Sulc.) separates the peduncular and tegmental parts of the midbrain. The pontomesencephalic sulcus (Pon. Mes. Sulc.) separates the midbrain and pons. B. Enlarged view. The carotid and basilar bifurcations are in the anterior incisural space. The anterior clinoid process (Ant Cli-noid) is lateral to the carotid artery. The lenticulos-triate branches (Lent. Str. A.) arise from the proximal part of the MCA. The anterior choroidal artery passes posteriorly below the optic tract to enter the middle incisural space. The oculomotor nerve enters the dura lateral to the posterior clinoid process (Post Clinoid). C. The temporal lobe has been removed to expose the choroid plexus (Chor. Plex.) within the temporal horn. The mamillary bodies (Mam. Body) lie above the basilar bifurcation. The recurrent artery (Rec. A.) arises from the АСА. (From Ono M, Ono M, Rhoton AL Jr, Barry M: Microsurgical anatomy of the region of the tentorial incisura. J Neurosurg 60:365-99, 1984.)

Figure 3.19. Tentorial incisura: arterial relationships. A. Anterior view. The anterior part of the left frontal lobe (Front. Lobe) has been removed to expose the anterior incisural space, which extends upward around the optic nerves (Optic N.) and chiasm and laterally along the sylvian fissure (Sylvian Fiss.). The internal capsule (Int. Cap.) is located above the anterior incisural space between the caudate nucleus (Caudate Nucl.) and the putamen. The rostrum, genu, and body of the corpus callosum (Corp. Call.) surround the anterior part of the lateral ventricle (Lat. Vent). The fornix extends along the lower margin of the septum

pellucidum (Sept. Pell.) and forms the anterior and superior margin of the foramen of Monro (For. Monro). The middle cerebral artery (M.C.A.) passes laterally in the sylvian fissure above the temporal lobe (Temp. Lobe) and gives rise to the lenticulostriate arteries (Lent. Str. A.). The anterior cerebral artery (A.C.A.) gives rise to the frontopolar (Front. Pol. A.), pericallosal (Pericall. A.), and recurrent (Rec. A.) arteries. Perforating branches (Perf. A.) from the АСА reach the lamina terminalis (Lam. Term.). The olfactory tract (Olf. Tr.) has been divided. The oculomotor nerve (Ш) is exposed behind the carotid artery (Car.A.). Choroid plexus (Chor.Plex.) is attached along the choroidal fissue (Chor.Fiss.) situated between the fornix and thalamus. B. Lateral view. The inferior part of the temporal lobe has been removed. The choroid plexus in the temporal horn (Temp. Horn) has been preserved. The branches of the MCA pass over the insula and loop around the frontal operculum (Front. Operc.). The posterior communicating artery (Post. Comm. A.) passes lateral to the infundibulum (Infiind.) and joins the proximal part of the posterior cerebral artery (P.C.A.). The anterior choroidal artery (Ant. Chor. A.) enters the choroid plexus in the temporal horn. The PCA arises at the bifurcation of the basilar artery (Bas. A.), passes posteriorly around the cerebral peduncle (Ped.), and gives rise to long (Long Circ. A.) and short circumflex (Short Circ. A.) arteries, lateral (Lat. Post. Chor. A.) and medial posterior choroidal (Med. Post. Chor. A.) arteries, and posterior pericallosal (Post. Pericall. A.), posterior temporal (Post. Temp. A.), calcarine (Calc. A.), and parietoocciptal (Par. Occip. A.) arteries. The superior cerebellar artery (S.C.A.) arises from the basilar artery and dips below the tentorial edge (Tent. Edge). The vein of Galen passes above the superior (Sup. Coll.) and inferior colliculi (Inf. Coll.). The lateral mesencephalic sulcus (Lat. Mes. Sulc.) is dorsal to the cerebral peduncle. C. Superolateral view. All of the tentorium except the free edge and the attachment along the petrous ridge has been removed. The basilar artery bifurcates in the anterior incisural space. The trochlear nerve (IV) encircles the brain stem between the PCA and SCA. The abducent nerve (VI) passes above the anterior inferior cerebellar artery (A.I.C.A.). The SCA bifurcates into a rostral (Ro. Tr.) and a caudal trunk (Ca. Tr.) above the trigeminal nerve (V). A bridging vein (Bridg. V.) enters the lateral wall of the cavernous sinus. The oculomotor nerve enters the dura lateral to the posterior clinoid process (Post. Clinoid). D. Another specimen with the temporal horn exposed, lateral view. The anterior choroidal artery gives rise to peduncular perforating branches (Ped. Perf. A.) before entering the temporal horn. The lateral posterior choroidal arteries pass through the choroidal f issue between the f imbria and pulvinar to enter the choroid plexus. The thalamogeniculate arteries (Thal. Gen. A.) enter the medial (Med. Gen. Body) and lateral (Lat. Gen. Body) geniculate bodies and the thalamus. The hippocampal formation on the medial surface of the temporal lobe includes the subiculum of the parahippocampal gyrus (Para-hipp. Gyr.), the dentate gyrus (Dentate Gyr.), and the fimbria. The amygdaloid nucleus (Amygd. Nucl.) is in the anteromedial margin of the temporal horn. E. Posterior incisural space, posterior-superior view. The occipital lobe, splenium, and the posterior part of the fornix have been removed to expose the SCA and PCA and their branches in the posterior incisural space. The SCA gives rise to vermian arteries (Ve. A.), which pass over the culmen just below the tentorial apex. The lateral posterior choroidal arteries pass around the pulvinar to reach the temporal horn and atrium. The medial posterior choroidal arteries encircle the brain stem to enter the roof of the third ventricle. The choroid plexus in the lateral ventricle is attached along the choroidal fissure. The striae terminalis (Str. Term.) courses along the groove between the caudate nucleus and the thalamus. The posterior part of the hippocampal formation (Hippocamp.) is exposed in the floor of the atrium. The tela choroidea (Tela), forming the upper wall of the velum interpositum (Vel. Interpos.), has been removed, but the lower layer, which is attached to the medial surface of the thalamus along the striae medullaris thalami, has been preserved. The suprapineal recess is located between the pineal body and the lower layer of the tela. F. The tentorium, except for the free edge and the strip along the straight sinus (Str. Sinus), and the tela choroidea in the roof of the third ventricle (3 Vent.) have been removed. The lateral posterior choroidal arteries ascend to reach the glomus of the choroid plexus. The medial posterior choroidal arteries pass forward in the roof of the third ventricle above the habenula and the massa intermedia (Massa Inter.). The SCA gives rise to the vermian and hemispheric arteries (He. A.) in the posterior incisural space. The quadrigeminal branches (Quad. A.) of the PCA and SCA pass to the colliculi. (From Ono M, Ono M, Rhoton AL Jr, Barry M: Microsurgical anatomy of the region of the tentorial incisura. J Neurosurg 60:365-399, 1984.)

Figure 3.21. Superior views of the sellar region. A. Carotid arteries are lateral to the optic nerves, anterior cerebral arteries are stretched across the superior surface of the optic chiasm, and the right optic nerve is shorter than the left. B. Long, tortuous anterior cerebral arteries are above the optic chiasm. The A1 segment of the right anterior cerebral artery loops in a complete circle between its origin from the carotid artery and its junction with its mate from the opposite side at the anterior communicating artery. A tortuous left anterior cerebral artery rests against the tubercu-lum sellae covering the prechiasmatic space. (From Renn WH, Rhoton AL Jr: Microsurgical anatomy of the sellar region. J Neurosurg 43:288-298, 1975.)

Figure 3.20. Inferior views with part of the temporal lobes removed to show the choroidal arteries supplying the choroid plexus of the inferior horn and atrium. A. The anterior choroidal arteries (A.Ch.A.) arise from the carotid arteries (C.A.), course posteriorly below the optic tracts (O.Tr.), and pass through the choroidal fissure to enter the choroid plexus (Ch. PL) of the temporal horn (Temp. Horn). The medial posterior choroidal arteries (M.P.Ch.A.) arise from the proximal or P1 segment (P-1) of the posterior cerebral artery (PCA), encircle the brain stem, and pass forward beside the pineal body and below the splenium of the corpus callosum to enter the choroid plexus in the roof of the third ventricle. The lateral posterior choroidal arteries (L.P.Ch.A.) arise from the P2 segment of the PCA (P-2) and course below the pulvinar of the thala-mus (Pulv.) and lateral geniculate bodies (L. Gen. Bo.) to enter the choroid plexus in the temporal horn and atrium. The carotid arteries are lateral to the optic nerves (O.N.) and optic chiasm (O.Ch.). The superior hypophyseal arteries (S. Ну. A.) pass medially from the carotid artery below the optic chiasm to the pituitary stalk (Pit. Stalk). The premamillary artery (Pre-mam. A.) arises from the posterior communicating artery (P.Co.A.) and enters the hypothalamus anterior to the mamillary bodies (Мат. В.). The calcarine (Cal.

A.), parietooccipital (P-O.A.), and posterior temporal arteries (P.T.A.) arise from the PCAs. The cut end of the basilar artery (B. A.) is between the cerebral peduncles (Ped.). B. Inferior view of temporal horn and atrium of right lateral ventricle. Enlarged view of the AChA, LPChA, and MPChA, showing their relationship to the optic tract, cerebral peduncles, brain stem, and pulvinar. The glomus of the choroid plexus projects into the atrium of the ventricle. The lenticulos-triate arteries (Len. Str. A.) arise from the middle cerebral artery (M.C.A.), and the anterior cerebral arteries (A.C.A.) pass above the optic chiasm. (From Fujii K, Lenkey C, Rhoton AL Jr: Microsurgical anatomy of the choroidal arteries: Lateral and third ventricles. J Neurosurg 52:165-188, 1980.)

Figure 3.22. Anterosuperior view of the perforating arteries entering the anterior wall of the third ventricle. A. The optic nerves (O.N.), chiasm (O.Ch.), and tracts (O.Tr.) are medial to the carotid arteries (C.A.) and below the anterior cerebral (A.C.A.) and anterior communicating (A.Co.A.) arteries. The frontal lobes (Fr. Lobe) and anterior perforated substance (Ant. Perf. Subst.) are above the anterior cerebral arteries. The anterior cerebral arteries give rise to a group of perforating arteries that enter the suprachiasmatic area and the upper surface of the optic chiasm. The recurrent artery (Rec. A.) arises from the anterior cerebral artery near the level of the anterior communicating artery. The olfactory nerves (Olf.N.) are above the anterior cerebral arteries. B. The anterior communicating artery gives rise to a series of perforating arteries (Perf. A.) that enter the area around the lamina terminalis (Lam. Теr.). С. A probe elevates the anterior communicating artery to expose two perforating arteries that pass through the lamina terminalis to enter the walls of the third ventricle (3V). The left recurrent artery arises from the anterior communicating artery in a common trunk with a branch to the frontal lobe (Fr. Br.). D. A precallosal artery (Pre.Cal.A.) arises from the anterior communicating artery and passes upward in front of the lamina terminalis to supply the rostrum (Rostrum) of the corpus callosum. (From Yamamoto I, Rhoton AL Jr, Peace DA: Microsurgery of the third ventricle: Part 1. Microsurgical anatomy. Neurosurgery 8:334-356, 1981.)

Internal Cerebral Vein The paired internal cerebral veins originate just behind the foramen of Monro and course posteriorly within the velum interpositum (Figs. 3.24, 3.26, and 3.27). Initially, they follow the gentle convex upward curve of the striae medul-laris thalami and, further distally, as they course along the superolateral surface of the pineal body, they follow the concave upward curve of the inferior surface of the splenium. The union of the paired veins to form the great vein may be located above or posterior to the pineal body and inferior or posterior to the splenium. The length of the internal cerebral vein varies from 19 to 35 mm (average, 30.2) (26).

Figure 3.23. Anterosuperior view of the lamina terminalis. A. The optic nerves (O.N.), chiasm (O.Ch.), and tracts (O.Tr.) are below the lamina terminalis (Lam. Теr.). The carotid arteries (C.A.) pass laterally below the optic nerves. The anterior cerebral (A.C.A.) and anterior communicating (A.Co.A.) arteries are above the optic chiasm. The recurrent arteries (Rec. A.) arise at the level of the anterior communicating artery. The olfactory nerves (Olf. N.) have been divided near their junction with the brain. B. The lamina terminalis has been opened along the dotted line shown in A to expose the cavity of the third ventricle (3V). The optic recess (O. Recess) extends downward between the lamina terminalis and the superior surface of the optic chiasm. The right anterior cerebral artery is hypoplastic, and the left anterior cerebral artery gives rise to both distal segments of the anterior cerebral artery. (From Yamamoto I, Rhoton AL Jr, Peace DA: Microsurgery of the third ventricle: Part 1. Microsurgical anatomy. Neurosurgery 8:334-356, 1981.) Venous Relationships The deep cerebral venous system is intimately related to the walls of the third ventricle (Figs. 3.24 to 3.26) (25, 26). It represents a formidable obstacle to the operative approaches to the third ventricle, especially in the region of the pineal gland, where the internal cerebral vein and the basal vein on each side converge on the great vein.

Great Vein and Straight Sinus The great vein, after being formed by the union of the internal cerebral veins, curves posteriorly and superiorly around the splenium to join the straight sinus at the anterior end of the junction of the falx and tentorium (Figs. 3.24, 3.26, and 3.28). The straight sinus courses posteroinferiorly along the falcotentorial junction to the tor-cular. The junction of the great vein with the straight sinus varies from being nearly flat if the tentorial apex is located below the splenium to forming a sharp angle if the apex is located above the splenium, so that the great vein must turn sharply upward to reach the straight sinus at the apex. The average length of the great vein is 12 mm (range, 8 to 25 mm) (26). The major tributaries of the great vein are the internal cerebral and basal veins, but it also receives blood from numerous other veins in the region. The convergence of multiple veins at the great vein results in a high density of veins in the area. Basal Vein The basal vein is formed below the anterior perforated substance in the anterior incisural space by the union of veins draining the walls of this space. It proceeds posteriorly between the midbrain and the temporal lobe to drain the walls of the middle incisural space before emptying into the internal cerebral or great vein (Figs. 3.24, 3.26, 3.28, and 3.29). The basal vein is divided into anterior, middle, and posterior segments that correspond to the parts of the vein coursing within the anterior, middle, and posterior incisural regions. The anterior and middle incisural regions are drained, almost totally, by tributaries of the basal vein. The veins in the posterior incisural region join the internal cerebral and great veins, as well as the basal vein.

Tributaries of the Deep Veins The tributaries of the internal cerebral, basal, and great veins drain the walls of the lateral and third ventricles, the periventricular white and gray matter, the corpus callosum, the thalamus, the septum pellucidum, the upper midbrain, the choroid plexus of the lateral and third ventricles, and the superior part of the cerebellum.

Figure 3.24. Schematic drawing of the ventricular veins. Lateral (top), anterior (middle), and superior (lower) views. The ventricular veins are divided into a medial and a lateral group. The ventricular veins drain into the internal cerebral (Int. Cer. V.}, basal (Basal V.), and great (V. Galen) veins. The lateral group consists of the anterior caudate vein (Ant. Caud. V.) in the frontal horn; the thalamostriate (Thal. Str. V.), posterior caudate (Post. Caud. V.), and thalamocaudate (Thal. Caud. V.) veins in the body; the lateral atrial vein (Lat. Atr. V.) in the atrium; and the inferior ventricular (Inf. Vent. V.) and amygdalar veins (Amygd. V.) in the temporal horn. The medial group is formed by the anterior septal vein (Ant. Sept. V.) in the frontal horn, the posterior septal veins (Post. Sept. V.) in the body, the medial atrial vein (Med. Atr. V.) in the atrium, and the transverse hippocampal veins (Trans. Hippo. V.) in the temporal horn. The transverse hippocampal veins drain into the anterior (Ant. Long. Hippo. V.) and posterior longitudinal hippocampal (Post. Long. Hippo. V.) veins. The superior cho-roidal veins (Sup. Chor. V.) drain into the thalamostriate and internal cerebral veins, and the inferior choroidal vein (Inf. Chor. V.) drains into the inferior

Ventricular Drainage The veins draining the deep gray and white matter of the cerebrum converge on the large veins coursing in and near the walls of the third ventricle (Figs. 3.24 to 3.29) (26). The veins from the frontal horn, the body of the lateral ventricle, and the surrounding gray and white matter drain into the internal cerebral vein; the veins from the temporal horn and the adjacent periventricular structures drain into the basal veins; and those draining the atrium and adjacent parts of the brain drain into the basal, internal cerebral, and great veins. The veins collecting blood from the periventricular white and gray matter join to form subepedymal channels in the walls of the lateral ventricles. These subependymal channels in the walls of the ventricle are divided into a medial and a lateral group. Both the lateral and the medial group pass through or adjacent to the choroidal fissure to reach the internal cerebral, basal, or great vein. The medial group consists of those veins that traverse the medial wall and roof of the frontal horn and body, the medial wall of the atrium, and the hippocompal surface of the floor of the temporal horn. The lateral group consists of those veins that traverse the lateral wall and floor of the frontal horn and body, the lateral wall of the atrium, and the roof of the temporal horn. In the temporal horn, the roof corresponds to the lateral group and the floor corresponds to the medial group. In the frontal horn, the medial group is formed by the anterior septal veins, which cross the septum pellucidum and fornix to join the internal cerebral vein, and the lateral group is formed by the anterior caudate veins, which cross the ventricular surface of the caudate nucleus and ter-

ventricular vein. The vein of Galen drains into the straight sinus (Str. Sinus). The anterior (Ant. Cer. V.) and deep middle cerebral (Deep. Mid. Cer. V.) veins join to form the basal vein. (From Ono M, Rhoton AL Jr, Peace D, Rodriguez R: Microsurgical anatomy of the deep venous system of the brain. Neurosurgery 15:621-657, 1984.)

Figure 3.25. A and B. Ventricular veins. A. Anterior view (along the arrow in the insert) into the frontal horn (Front. Horn) and body of the lateral ventricle (Body Lat. Vent). The frontal horn is located anterior to the foramen of Monro (For. Monro) and has the septum pellucidum (Sept Pell.) in the medial wall, the genu and body of the corpus callosum (Corp. Call.) in the roof, the caudate nucleus (Caudate Nucl.) in the lateral wall, the genu of the corpus callosum in the anterior wall, and the rostrum of the corpus callosum in the floor. The body of the lateral ventricle has the thalamus in its floor, the caudate nucleus in the lateral wall, the body of the fornix and septum pellucidum in the medial wall, and the corpus

minate in the thalamostriate or internal cerebral vein. In the body, the medial group is formed by the posterior septal veins, and the lateral group is formed by the thalamostriate, thalamocau-date, and posterior caudate veins. The thalamostriate vein is the largest tributary of the internal cerebral vein. It courses anteriorly and medially in the groove between the caudate nucleus and the thalamus beneath the striae terminalis. At the level of the foramen of Monro, it curves around the anterior tubercle of the thalamus and joins the internal cerebral vein. The thalamocau-date originates at the lateral angle of the body of the lateral ventricle and runs across the caudate nucleus and thalamus to join the internal cerebral vein more posteriorly than does the thalamostriate vein. The posterior caudate veins cross the lateral wall of the body of the ventricle and empty into the thalamostriate or thalamocaudate vein. The posterior septal veins cross the posterior part of the septum pellucidum and pass around the fornix to enter the internal cerebral vein. In the atrium, the medial group is formed by the medial atrial veins, and the lateral group is formed by the lateral atrial veins. The atrial veins pass through the choroidal fissure to enter the basal, internal cerebral, or great vein. In the temporal horn, the veins draining the roof correspond to the lateral group, and the veins draining the floor correspond to the medial group. The medial group is formed from the transverse hippocampal veins that run in the floor on the sur-

face of the hippocampus and drain into the basal vein. The lateral group is formed by the inferior ventricular vein. It runs in the roof of the temporal horn and passes through the choroidal fissure to join the basal vein. Incisural and Cisternal Veins The main venous trunk related to the anterior incisural space is the basal vein (Figs. 3.28 to 3.30) (25, 26). The veins joining below the anterior perforated substance to form the basal vein include the olfactory vein, which runs posteriorly in the olfactory sulcus; the frontoorbital vein, which courses along the orbital surface of the frontal lobe; the deep middle cerebral vein, which receives the veins from the insula and passes medially across the limen insulae; the uncal veins, which course medially from the uncus; and the anterior cerebral vein, which descends on the lamina terminalis and crosses the optic chiasm to reach the basal vein. The paired anterior cerebral veins are joined across the midline above the optic chiasm by the anterior communicating vein and receive the paraterminal vein from the paraterminal and parolfactory gyri and the anterior pericallosal veins from the rostrum and genu of the corpus callosum. The veins on the surface of the brain stem that form the posterior wall of the anterior incisural space are divided into transversely or vertically oriented groups (21, 22, 25, 26). The transverse veins are the peduncular vein, which passes hor-

callosum in the roof. The choroid plexus (Chor. Plex.) is attached along the choroidal fissure (Chor. Fiss.), the cleft between the fornix and thalamus. The anterior septal veins (Ant. Sept. V.) cross the roof and medial wall of the frontal horn and pass posteriorly toward the foramen of Monro, where they join the anterior end of the internal cerebral veins. The anterior caudate veins (Ant. Caud. V.) cross the lateral wall of the frontal horn and join the thalamostriate vein (Thal. Str. V.), which passes through the foramen of Monro. The superior choroidal vein (Sup. Chor. V.) courses on the choroid plexus in the body. The posterior septal veins (Post. Sept. V.) cross the roof and medial wall of the body and pass through the margin of the choroidal fissure. The posterior caudate veins cross the lateral wall of the body and join the thalamostriate vein, which courses along the striothalamic sulcus (Str. Thal. Sulc.). Anterior (Ant. Superf.. Thal. V.) and superior (Sup. Superf. Thal. V.) superficial thalamic veins cross the surface of the thalamus. The anterior thalamic vein (Ant. Thal. V.) drains the nuclei in the anterior-superior part of the thalamus. B. Anterior-superior view (along the arrow in the insert) into the body, atrium, and occiptal horn (Occip. Horn) of the lateral ventricle. The calcar avis and bulb of the corpus callosum (Bulb Corp. Call.) form the medial wall of the atrium and occipital horn. The floor of the atrium is formed by the collateral trigone (Coll. Trig.). The roof and posterior part of the lateral walls are formed by the tapetum of the corpus callosum. The caudate nucleus is in the anterior part of the lateral wall of the atrium. The medial (Med. Atr. V.) and lateral (Lat. Atr. V.) atrial veins pass forward on the medial and lateral walls of the atrium toward the choroidal fissure. A thalamocaudate vein (Thal. Caud. V.) crosses the lateral wall posterior to the thalamostriate vein. The superior choroidal vein courses toward the foramen of Monro.

Figure 3.25. С and D. Ventricular veins (continued). C. Posterior view (along the arrow in the insert) into the atrium and temporal horn. The atrium has the tapetum of the corpus callosum in the roof, the bulb of the corpus callosum and the calcar avis in its medial wall, the hippocampal formation and the collateral trigone (Coll. Trig.) in the floor, the caudate nucleus and tapetum in the lateral wall, and the crus of the fornix and the pulvinar and choroid plexus in the anterior wall. The floor of the temporal horn (Temp. Horn) is formed by the hippocampus and collateral eminence (Coll. Eminence). The inferior choroidal vein (Inf. Chor. V.) courses on the choroid plexus in the temporal horn. The lateral atrial veins

izontally around the anterior surface of the cerebral peduncle and terminates in the basal vein at the junction of the anterior and middle inci-sural spaces, and the vein of the pontomesence-phalic sulcus, which courses below the peduncular vein in the pontomesencephalic sulcus. The vertically oriented veins on the posterior wall of the anterior incisural space are the median anterior pontomesencephalic vein, which courses in the midline and connects the peduncular veins above with the pontine veins below, and the lateral anterior pontomesencephalic veins, which course on the anterolateral surface of the cerebral peduncle and the pons and join the basal vein superiorly and the vein of the pontomesencephalic sulcus or a transverse pontine vein inferiorly. The venous relationshps in the middle incisural space are relatively simple (Figs. 3.28 to 3.30). The basal vein courses along the upper part of the cerebral peduncle and below the pul-vinar to reach the posterior incisural space. It may infrequently terminate in a tentorial sinus in the free edge at this level. The veins draining the medial wall, from anterior to posterior, are the lateral pontomesencephalic vein, which runs vertically on the lateral surface of the cerebral peduncle and pons and terminates in the basal vein superiorly and the vein of the pontomesencephalic sulcus inferiorly, and the lateral mesencephalic vein, which runs vertically along the

lateral mesencephalic sulcus and terminates above in the basal vein near the medial genicu-late body and below in a petrosal vein. Numerous veins from the lateral wall of the middle incisural space converge on the basal vein near the inferior choroidal point. These veins are the anterior hippocampal vein, which courses posteriorly from the anterior hippocampal sulcus; the uncal veins, which drain the medial surface of the uncus; and the anterior longitudinal hippocampal vein, which courses anteriorly along the dentate gyrus. The venous relationships in the posterior incisural space are the most complex in the cranium because the internal cerebral and basal veins and many of their tributaries converge on the great vein within this area (Figs. 3.38 to 3.30). The internal cerebral veins exit the velum interpositum and the basal veins exit the ambient cistern to reach the posterior incisural space, where they empty into the great vein. The great vein passes below the splenium to enter the straight sinus at the tentorial apex. Other veins converging on this area include the posterior pericallosal veins, which course posteriorly around the splenium to terminate in the great vein or the internal occipital vein; the lateral atrial veins, which drain the lateral wall and roof of the atrium and course medially through the choroidal fissure to terminate in the basal, internal cerebral, or great vein either directly or after

arise on the lateral wall and cross the tail of the caudate nucleus and the pulvinar to pass through the choroidal fissure. The medial atrial veins pass forward and penetrate the crus of the fornix near the choroidal fissure to reach the quadri-geminal cistern. Some of the medial atrial veins also drain the roof and floor. Transverse hippocampal veins (Transv. Hippo. V.) cross the floor of the atrium and temporal horn. Posterior superficial thalamic veins (Post. Superf. Thal. V.) cross the atrial surface of the thalamus. D. Anterior view (along the arrow in the insert) into the temporal horn. The floor of the temporal horn is formed by the collateral eminence and the hippocampal formation. The roof and lateral wall, from medial to lateral, are formed by the thalamus, the tail of the caudate nucleus, and the tapetum of the corpus callosum. The medial wall is little more than the cleft between the inferior surface of the thalamus and the fimbria. The amygdaloid nucleus (Amygd. Nucl.) bulges into the anteromedial part of the temporal horn. The pes hippocampus (Pes Hipp.), the bulbous digitated anterior end of the hippocampal formation, is in the anterior part of the floor. The fimbria of the fornix arises on the surface of the hippocampal formation and passes posteriorly to become the crus of the fornix. The choroid plexus is attached along the choroidal fissure. The inferior ventricular vein (Inf. Vent. V.) drains the roof of the temporal horn and receives the amygdalar vein (Amygd. V.) from the ventricular surface of the amygdaloid nucleus. The inferior choroidal vein joins the inferior ventricular vein. The transverse hippocampal veins draining the floor of the temporal horn pass medially through the choroidal fissure to enter the basal vein or its tributaries. (From Ono M, Rhoton AL Jr, Peace D, Rodiguez R: Microsurgical anatomy of the deep venous system of the brain. Neurosurgery 15:621657, 1984.)

Figure 3.26. Midsagittal sections. A. The brain stem has been sectioned at the level of the midbrain. The parahippocampal gyrus (Parahippo. Gyr.) has been depressed to expose the basal vein (Basal V.) between the uncus and the cerebral peduncle (Ped.) and the choroidal fissure (Chor. Fiss.), the cleft between the fimbria and the pulvinar. The anterior septal vein (Ant. Sept. V.) passes posteriorly on the septum pel-lucidum (Sep. Pell.) toward the foramen of Monro (For. Monro) and joins the internal cerebral vein (Int. Cer. V.) in the roof of the third ventricle (3rd Vent). The internal cerebral vein passes posteriorly above the massa intermedia (Massa Inter.) and the choroid

plexus (Chor. Plex.) in the roof of the third ventricle and exits the velum interpositum (Vel. Interpos.), situated between the layers of the tela choroidea (Tela) in the roof of the third ventricle, by passing below the splenium to join the vein of Galen (V. of Galen). The basal vein passes below the lateral geniculate body (Lat. Gen. Body) and pulvinar. The anterior longitudinal hippocampal vein (Ant. Long. Hippo. V.) joins the anterior end of the basal vein. The posterior longitudinal hippocampal (Post. Long. Hippo. V.), internal occipital (Int. Occip. V.), and lateral atrial (Lat. Atr. V.) veins join the posterior end of the basal vein. The posterior pericallosal vein (Post. Pericall. V.) passes around the splenium to join the great vein. The paraterminal vein (Paraterm. V.) passes in front of the lamina terminalis (Lam. Ter.) and anterior commissure (Ant. Comm.) on the paraterminal gyrus (Paraterm. Gyr.) and courses toward the anterior end of the basal vein. The habenular (Hab. Comm.) and posterior (Post. Comm.) commissures form the stalk of the pineal body. The fornix fuses to the lower surface of the splenium posteriorly and passes superior and anterior to the foramen of Monro anteriorly. The cin-gulate gyrus (Cing. Gyr.) is above the corpus callosum (Corp. Call.). The calcarine sulcus (Calc. Sulc.) divides the medial surface of the occipital lobe above the lingual gyrus (Lingual Gyr.). The infundibular (Infund.) and mamillary (Mam. Body) bodies are in the floor of the third ventricle. B. Enlarged view of the inferior part of the choroidal fissure. The veins exiting the temporal horn pass through the choroidal fissure, the cleft between the fimbria of the fornix and the pulvinar. The inferior end of the choroidal fissure, called the inferior choroidal point (Inf. Chor. Point), is located just behind the uncus. The basal vein passes below the medial (Med. Gen. Body) and lateral geniculate bodies. The anterior longitudinal hippocampal vein passes forward and the posterior longitudinal hippocampal vein passes backward on the dentate gyrus. The hippocampal sulcus (Hippo. Sulc.) separates the dentate and parahippocampal gyri, and the fimbriod-entate sulcus (Fimb. Dentate Sulc.) separates the fimbria and the dentate gyrus. The section of the brain stem crosses the red nucleus (Red. Nucl.). C. Enlarged view. The septum pellucidum and body of the fornix have been removed to expose the right lateral ventricle (Lat. Vent). The thalamostriate vein (Thal. Str. V.) passes forward between the caudate nucleus (Caudate Nucl.) and the thalamus and enters the internal cerebral vein. The anterior septal vein passes through the posterior margin of the foramen of Monro and enters the internal cerebral vein in the posterior part of the roof of the third ventricle. The stria medullaris thalami (Str. Med. Thal.) passes posteriorly in the wall of the third ventricle toward the habenular commissure. A superior choroidal vein (Sup. Chor. V.) courses on the surface of the choroid plexus. (From Ono M, Rhoton AL Jr, Peace D, Rodriguez R: Microsurgical anatomy of the deep venous system of the brain. Neurosurgery 15:621-657, 1984.)

forming a common stem with the medial atrial veins; the medial atrial veins, which drain the medial wall and roof of the atrium and occipital horn and course anteromedially through the cho-roidal fissure before terminating in the internal cerebral vein; the posterior longitudinal hippocampal vein, which courses posteriorly along the dentate gyrus and terminates in the internal cerebral, basal, or great vein or one of the atrial veins; and the internal occipital veins, which originate in the calcarine and parietooccipital sulci and course anteromedially along the calcarine sulcus to terminate in the great vein. The largest vein from the infratentorial part of the posterior incisural space is the vein of the cerebellomesencephalic fissure, which is also called the precentral cerebellar vein (Figs. 3.28 and 3.30). It originates from the union of the paired veins of the superior cerebellular peduncle and courses anterosuperiorly through the cerebellomesencephalic fissure to terminate with the superior vermain vein in the great vein (21, 26). The veins of the superior cerebellular peduncle arise below the inferior colliculi on the external surface of the superior cerebellar peduncles. The superior vermian and the superior hemispheric veins from the anterosuperior part of the cerebellum arise from the infratentorial part of the posterior incisural space and pass forward under the free edge of the tentorium to empty into the great vein either directly or after joining with the vein of the cerebellomesencephalic fissure. Smaller thalamic, epithalamic, and tectal veins emerge from the walls of the posterior incisural space and terminate in the internal cerebral, basal, or great veins or the vein of the cerebellomesencephalic fissure. Choroidal Veins Two other veins that course near the walls of the third ventricle are the superior and inferior choroidal veins (26). Both lie within the choroid plexus of the lateral ventricles and are covered by ependyma (Figs. 3.24, 3.25, and 3.27 to 3.29). The superior choroidal vein runs in the choroid plexus of the atrium and body of the lateral ventricle. It terminates in either the thalamostriate

or the internal cerebral vein near the foramen of Monro. The inferior choroidal vein drains the choroid plexus of the temporal horn and atrium. It courses downward and forward in the roof of the temporal horn and joins the inferior ventricular vein near where the latter passes through the choroidal fissure or passes directly to the basal vein. Bridging Veins The veins coursing from the surface of the brain to the venous sinuses in the dura mater that are encountered in retracting the brain to reach the walls of the third ventricle include the superficial superior cerebral veins that cross from the superior margin of the cerebrum to the superior sagittal sinus; the cingulate and callosal veins that drain the cingulate gyrus and upper surface of the corpus callosum and pass into the inferior sagittal sinus; the superficial middle cerebral veins that cross the sylvian fissure to enter the sphenoparietal sinus on the inferior margin of the sphenoid ridge; and the inferior cerebral veins that drain the lower portions of the lateral surface of the temporal lobe and the undersur-faces of the occipital and temporal lobes and empty into the transverse and tentorial sinuses. Of the 3 to 12 (average, 6.8) superficial superior cerebral veins per hemisphere, the number anterior and posterior to the central or rolandic vein are close to the same (55). The veins draining the corresponding part of the medial and lateral surfaces of the superior margin of the cerebral hemisphere usually join before entering the superior sagittal sinus. The vein of Labbe, which extends from the sylvian fissure to the transverse sinus, is the largest inferior cerebral vein. There are usually no veins between the medial aspect of the occipital pole and the transverse and straight sinuses, but there are frequently several bridging veins (range, 2 to 7) between the lateral surface of the occipital pole and the transverse sinus (55). There are also numerous bridging veins crossing the interval between the superior surface of the cerebellum and the sinuses in the tentorium. They are more numerous medially than laterally (21).

Figure 3.27. A-F. Superior view. A. The superior part of the cerebral hemisphere has been removed to expose the lateral ventricles. The frontal horn (Front. Horn) extends into the frontal lobe (Front. Lobe). The body of the lateral ventricle (Body Lat. Vent.) is deep to the frontal and parietal lobes (Par. Lobe). The atrium extends posteriorly toward the occipital lobe (Occip. Lobe). The thalamostriate vein (Thal. Str. V.) courses along the junction of the caudate nucleus (Caudate Nucl.) and the thalamus toward the foramen of Monro (For. Monro). The anterior caudate veins (Ant. Caud. V.) course on the caudate nucleus in the lateral wall of the frontal horn. The superior choroidal veins (Sup. Chor. V.) course on the choroid plexus (Chor. Plex.) and join the thalamostriate vein near the foramen of Monro. Medial atrial veins (Med. Atr. V.) course along the medial wall of the atrium. The superior sagittal sinus (Sup. Sag. Sinus) courses in the edge of the falx. B. Enlarged view. The septum pellucidum (Sept. Pell.) separates the lateral ventricles. The anterior caudate and thalamostriate veins pass toward the foramen of Monro. The small anterior superficial thalamic vein (Ant. Superf. Thal. V.) crosses the thalamus near the anterior thalamic tubercle (Ant. Thal. Tuber.). The choroidal fissure (Chor. Fiss.) is located between the fornix and thalamus. The bulb of the corpus callosum (Bulb Corp. Call.) is in the medial wall of the atrium. The cingulate gyrus (Cing. Gyr.) wraps around the corpus callosum (Corp. Call). C. Enlarged view of the frontal horns. The anterior septal veins (Ant. Sept. V.) pass medially on the posterior surface of the genu of the corpus callosum (Genu Corp. Call.) and turn posteriorly on the septum pellucidum. The anterior caudate veins cross the head of the caudate nucleus. The posterior septal veins (Post. Sept. V.) cross the medial wall of the body of the lateral ventricle. On the left side, the anterior septal, thalamostriate, and superior choroidal veins course toward the foramen of Monro. On the right side, the thalamostriate vein is absent and most of the lateral wall of the body is drained by a large thalamocaudate vein

Figure 3.27. G-I. Superior view (continued). G. Another specimen. The thala-mostriate veins receive the superior choroidal, anterior septal, anterior thalamic, and anterior superficial thalamic veins and pass through the choroidal fissure, just behind the foramen of Monro. The striae medullaris thalamic (Str. Med. Thal.) pass along the superomedial margins of the thalami. H. Another specimen. The tela choroidea (Tela) extends across the roof of the third ventricle above the internal cerebral veins. A large superior superficial thalamic vein on the left side receives several small choroidal veins (Chor. V.) and joins the anterior part of the internal cerebral vein. The anterior caudate, anterior septal, anterior thalamic, and superior choroidal veins converge on the anterior end of the internal cerebral vein. I. Another specimen. The right internal cerebral vein receives duplicate thalamostrlate veins, called the anterior (Ant. Thal. Str. V.) and posterior (Post. Thal Str. V.) thalamostriate veins. The right lateral atrial vein (Lat. Atr. V.) joins the posterior thalamostriate vein. On the left side, a single large thalamostriate vein passes through the foramen of Monro to join the internal cerebral vein. The inferior sagittal sinus (Inf. Sag. Sinus) courses in the lower margin of the falx. (From Ono M, Rhoton AL Jr, Peace D, Rodriguez R: Microsurgical anatomy of the deep venous system of the brain. Neurosurgery 15:621-657, 1984.)

(Thal. Caud. V.), which receives the posterior caudate vein (Post. Caud. V.) and passes posteriorly in the striothalamic sulcus (Str. Thal. Sulc.). D. The body of the fornix has been removed at the junction of its attachment to the crura posteriorly and the columns anteriorly. The removal of the tela choroidea in the roof of the third ventricle (3 Vent.) exposes the internal cerebral veins (Int. Сет. V.) above the massa intermedia (Massa Inter.). The choroid plexus on the right side has been removed along its attachment to the tenia choroidea (Tenia Chor.). The anterior caudate, anterior septal, anterior superficial thalamic, and superior choroidal veins converge on the left thalamostriate vein at the foramen of Monro. The medial atrial veins unite to form a common trunk. The posterior longitudinal hippocampal veins (Post. Long. Hippo. V.) course on the dentate gyri (Dentate Gur.) and join the internal cerebral veins. The right posterior longitudinal hippocampal vein unites with the thalamocaudate vein before entering the internal cerebral vein. E. Another specimen. The fornix has been removed. The anterior caudate veins pass along the lateral wall of the frontal horn and join the thalamostriate vein on the left side and a thalamocaudate vein on the right side. The superior choroidal, superior superficial thalamic (Sup. Superf. Thal. V.), and anterior septal veins join the internal cerebral veins near the foramen of Monro. On the right side, the lateral wall of the body is drained by a large thalamocaudate vein. A large anterior caudate vein passes along the superolateral angle of the right lateral ventricle to join the thalamocaudate vein. F. Enlarged view. The tributaries converging on the anterior end of the internal cerebral vein at the foramen of Monro include the anterior septal, anterior caudate, thalamostriate, superior choroidal, anterior thalamic (Ant Thal. V.), and anterior superficial thalamic veins.

Figure 3.28. A and B. Cisternal veins. A. Anterolateral view. The insert shows the direction of view. The frontal (Front, Lobe) and temporal (Temp. Lobe) lobes have been retracted away from the floor of the anterior (Ant. Fossa) and middle cranial (Mid. Fossa) fossae. The veins converging on the anterior end of the basal vein (Basal V.) below the anterior perforated substance (Ant. Perf. Subst.) are the deep middle cerebral veins (Deep Mid. Cer. V.) from the sylvian fissure (Sylvian Fiss.); the olfactory vein (Olf. V.), which drains posteriorly along the olfactory tract (Olf. Tr.) near the gyrus rectus (Gyr. Rectus); the frontoorbital veins (Front. Orb. V.), which drain the orbital gyri (Orb. Gyr.); the inferior striate veins (Inf. Str. V.), which exit the anterior perforated substance; and the anterior cerebral veins (Ant. Cer. V.), which are joined above the optic chiasm by the anterior communicating vein (Ant. Comm. V.). The peduncular vein (Ped. V.} passes around the cerebral peduncle (Ped.) above the oculomotor nerve (III) and joins the median anterior pontomesencephalic vein (Med. Ant. Pon. Mes. V.) in the midline and the basal vein laterally. The infundibulum (Infund.) passes inferiorly behind the anterior clinoid process (Ant. Clinoid), optic nerve (Optic N.), and internal carotid artery (Int. Car. A.). The lateral anterior pontomesencephalic vein (Lat. Ant. Pon. Mes. V.) joins the vein of the pontomesencephalic sulcus (V. of Pon. Mes. Sulc.) below and the basal vein above. The inferior thalamic veins (Inf. Thal. V.) arise behind and the premamillary veins (Preman. V.) arise in front of the mamillary bodies. The inferior ventricular vein (Inf. Vent. V.) exits the temporal horn above the parahippocampal gyrus (Parahippo. Gyr.) and enters the basal vein. An uncal vein (Uncal V.) passes medially from the uncus. The trochlear nerve (IV) courses near the tentorial edge (Tent. Edge). B. Lateral view, right side. The temporal lobe has been elevated, as shown in the insert. The tentorium (Tent.) extends along the side of the brain stem. The basal vein passes around the brain stem and joins the vein of Galen (V. of Galen). The tributaries of the basal veins lateral to the brain stem include the lateral mesen-cephalic vein (Lat. Mes. V.), which courses in the lateral mesencephalic sulcus; the inferior ventricular vein, which drains the roof of the temporal horn; the anterior hippocampal vein (Ant. Hippo. V.), which courses along the sulcus between the uncus and the parahippocampal gyrus; the anterior longitudinal hippocampal vein (Ant. Long. Hippo. V.), which courses along the dentate gyrus; and the medial temporal veins (Med. Temp. V.) from the inferomedial surface of the temporal lobe. In the pineal region, the basal vein receives the lateral atrial vein (Lat. Atr. V.) from the lateral wall of the atrium. The internal cerebral veins (Int. Cer. V.) pass above the pineal body. The superior vermian (Sup. Ve. V.) and superior hemispheric (Sup. He. V.) veins from the cerebellum and the vein of the cerebellomesencephalic fissure (V. of Cer. Mes. Fiss.) from the fissure between the midbrain and cerebellum ascend to join the vein of Galen. Tectal veins (Tectal V.) drain the colliculi. A transverse pontine vein (Trans. Pon. V.) crosses the pons.

Figure 3.28. С and D. Cisternal veins (continued). C. Posterior view. The insert shows the direction of view. The occipital (Occip. Lobe) and parietal (Par. Lobe) lobes have been retracted to expose the termination of the internal cerebral and basal veins in the vein of Galen. The internal occipital (Int. Occip. V.) and posterior pericallosal (Post. Pericall. V.) veins join the internal cerebral vein. The posterior longitudinal hippocampal vein (Post. Long. Hippo. V.) passes along the dentate gyrus (Dentate Gyr.) and joins the medial atrial vein (Med. Atr. V.). The lateral mesencephalic, posterior thalamic (Post. Thal. V.), and inferior ventricular veins join the basal vein. Tectal veins pass from the superior (Sup/. Coll.) and inferior colliculi (Inf. Coll.). The medial (Med. Gen. Body) and lateral (Lat. Gen. Body) geniculate bodies are below the pulvinar. The inferior sagittal sinus (Inf. Sag. Sinus) and the vein of Galen join the straight sinus (Str. Sinus). D. Right anterolateral view with the anterior portion of the right cerebral hemisphere removed to expose the upper brain stem and the third ventricle (3 Vent.) in the midline. The brain stem was sectioned at the level of the cerebral peduncle. The anterior cerebral veins join the deep middle cerebral vein to form the basal vein. The basal vein encircles the brain stem and along its course receives the peduncular, inferior ventricular, anterior hippocampal, anterior longitudinal hippocampal, posterior thalamic, lateral atrial, lateral anterior pontomesencephalic, and lateral mesencephalic veins. The superior vermian vein receives the superior hemispheric and tectal veins and the vein of the cerebellomesencephalic fissure. The paraterminal (Paraterm. V.) and anterior pericallosal (Ant. Pericall. V.) veins join the anterior cerebral vein. The internal cerebral vein courses in the velum interpositum (Vel. Interpos.) in the roof of the third ventricle (3 Vent). The collateral eminence (Coll. Eminence) sits above the collateral sulcus (Coll. Sulc.) in the floor of the temporal horn. Septal veins (Sept. V.) cross the septum pellucidum (Sept. Pell.). The choroid plexus (Chor. Plex.) passes through the foramen of Monro (For. Monro) to reach the roof of the third ventricle. (From Ono M, Rhoton AL Jr, Peace D, Rodriguez R: Microsurgical anatomy of the deep venous system of the brain. Neurosurgery 15:621-657, 1984.)

Figure 3.29. A-D. A. Lateral view of the left side. The temporal (Temp. Lobe) and frontal (Front. Lobe) lobes have been elevated, as shown in the insert. The tentorium (Tent.) has been opened above the cerebellum. The basal vein (Basal V.) arises below the anterior perforated substance (Ant. Perf. Subst.), passes around the cerebral peduncle (Ped.), and joins the vein of Galen (V. of Galen; dashed lines). The veins converging on the anterior end of the basal vein are the olfactory veins (Olf. V.), which pass along the olfactory tract (Olf. Tr.); the inferior striate veins (Inf. Str. V.), which descend through the anterior perforated substance; the deep middle cerebral veins (Deep Mid. Cer. V.), which pass medially from the sylvian fissure (Sylvian Fiss.); and the anterior cerebral veins (Ant. Cer. V.), which pass across the optic chiasm. The tributaries of the basal veins lateral to the brain stem include the peduncular vein (Ped. V.), which passes around the cerebral peduncle; the lateral mesencephalic vein (Lat. Mes. V.), which courses in the lateral mesencephalic sulcus; and the inferior ventricular vein (Inf. Vent. V.), which drains the roof of the temporal horn. Tributaries of the inferior ventricular vein are the anterior hippocampal vein (Ant. Hippo. V.), which courses along the sulcus lateral to the uncus; the anterior longitudinal hippocampal vein (Ant. Long. Hippo. V.), which courses along the dentate gyrus; and the medial temporal veins (Med. Temp. V.) from the inferomedial surface of the temporal lobe. In the pineal region, the basal vein receives the lateral atrial vein (Lat. Atr. V.) from the lateral wall of the atrium. The internal occipital vein (Int. Occip. V.) from the medial surface of the occipital lobe joins the internal cerebral vein. The superior vermian (Sup. Ve. V.) and superior hemispheric (Sup. He. V.) veins from the cerebellum and the vein of the cerebellomesencephalic fissure (V. of Cer. Mes. Fiss.) ascend to join the vein of Galen. The optic nerve (Optic N.) passes above the carotid artery (Car. A.). The optic tract (Optic Tr.) terminates in the lateral geniculate body (Lat. Gen. Body). The infundibulum (Infund.) passes inferiorly in front of the mamillary bodies (Mam. Body). The uncus protrudes over the free edge (Tent. Edge) of the tentorium above the oculomotor nerve (III). The parahip-pocampal gyrus (Parahippo. Gyr.) is medial to the collateral (Coll. Sulc.) and rhinal (Rhinal Sulc.) sulci. The medial geniculate body (Med. Gen. Body) is superolateral to the inferior (Inf. Coll.) and superior (Sup. Coll.) colliculi. The pontomesencephalic sulcus (Pon. Mes. Sulc.) separates the pons and midbrain. A transverse pontine vein (Trans. Pon. V.) crosses the pons. B. The tip of the temporal lobe has been displaced posteriorly. Tributaries converging on the basal vein below the anterior perforated substance include the deep middle cerebral, insular (Insular V.), inferior striate, olfactory, and anterior cerebral veins. In its course around the midbrain, the basal vein receives the peduncular, lateral mesencephalic, and inferior ventricular veins. The uncal veins (Uncal. V.) drain the semilunar (Semilunar Gyr.) and ambient (Ambient Gyr.) gyri. The premam-illary vein (Premam. V.) joins the peduncular vein. The anterior communicating vein (Ant. Comm. V.) connects the paired anterior cerebral veins above the optic chiasm. C. Enlarged view of the region of the inferior ventricular vein. It exits the ventricle through the choroidal fissure, passes near the medial and lateral geniculate bodies, and receives the uncal, anterior hippocampal, anterior longitudinal hippocampal, and medial temporal veins. The anterior hippocapmal vein arises in the anterior hippocampal sulcus (Ant. Hippo. Sulc.). The brachium of the inferior colliculus (Brach. Inf. Coll.) is directed from the inferior colliculus toward the medial geniculate body. D. Enlarged view. The veins converging on the region of the great vein are the basal, internal cerebral (Int. Cer. V.), lateral atrial, superior vermian, internal occipital, and tectal (Tectal V.) veins and the vein of the cerebellomesencephalic fissure. The lemniscal trigone (Lemniscal Trig.) is posterior to the cerebral peduncle.

Figure 3.29. E-H. E. The posterior part of the basal and lateral atrial veins has been removed (dashed lines) to expose the veins in the quadrigeminal cistern. The tectal veins arise from the region of the quadrigeminal plate and pass posteriorly to join the basal vein and the vein of the cerebellomesencephalic fissure. The veins on the superior surface of the cerebellum converge above the apex of the vermis to form the superior vermian vein. The internal occipital vein has been divided near its origin. The epithalamic veins (Epithal. V.) arise in the region of the pineal body. The vein of Galen enters the straight sinus at the apex of the tentorium (Apex. Tent.). F. The orientation is shown in the insert. The temporal horn (Temp. Horn) has been opened between the superior (Sup. Temp. Gyr.) and the middle temporal (Mid. Temp. Gyr.) gyri to expose the veins crossing the roof and floor of the temporal horn. The inferior ventricular vein drains the anterior part of the roof and receives the amygdalar vein (Amygd. V.) from the ventricular surface of the amygdaloid nucleus (Amygd. Nucl.). The lateral atrial vein drains the posterior part of the roof. The transverse hippocampal veins (Trans. Hippo. V.) pass medially across the hippocampus in the floor of the temporal horn and through the junction of the fimbria of the fornix and hippocampus to reach the basal vein. The inferior choroidal vein (Inf. Chor. V.) courses on the choroid plexus. The medial atrial vein (Med. Atr. V.) drains the medial wall of the atrium. G. Enlarged view. The tenia fimbria is a small ridge on the edge of the fimbria along which the tela choroidea and choroid plexus are attached. The choroid plexus has been detached and reflected upward. The inferior ventricular and inferior choroidal veins pass through the choroidal fissure near the inferior choroid point (Inf. Chor. Point), which is defined as the inferior end of the choroidal fissure. The transverse hippocampal veins cross the floor of the temporal horn. Я. The neural structures forming the roof of the temporal horn have been removed, the choroid plexus has been reflected upward, and the choroidal fissure has been opened to expose the surface of the subiculum and the dentate gyms (Dentate Gyr.) facing the ambient cistern. A posterior superficial thalamic vein (Post. Superf. Thal. V.) joins the posterior longitudinal hippocampal vein (Post. Long. Hippo. V.) on the dentate gyrus. The calcar avis forms the lower half of the medial wall of the atrium. Both the collateral eminence (Coll. Eminence) in the floor of the temporal horn and the collateral trigone (Coll. Trig.) in the floor of the atrium overlie the collateral sulcus on the inferior surface of the temporal lobe. The pes hippocampus (Pes. Hippo.) is the bulbous, digitated anterior part of the hippocampus. (From Ono M, Rhoton AL Jr, Peace D, Rodriguez R: Microsurg-ical anatomy of the deep venous system of the brain. Neurosurgery 15:621-657, 1984.)

Figure 3.30. A. Posterior view. Insert shows the orientation. The occipital lobe (Occip. Lobe) has been retracted, and the tentorium (Tent.) has been opened to expose the termination of the internal cerebral (Int. Cer. V.) and basal (Basal V.) veins in the vein of Galen (V. of Galen.). The internal occipital (Int. Occip. V.), posterior pericallosal (Post. Pericall. V.), and posterior longitudinal hippocampal (Post. Long. Hippo. V.) veins drain into the internal cerebral vein. The lateral mesencephalic (Lat. Mes. V.), posterior thalamic (Post. Thal. V.), and inferior ventricular (Inf. Vent. V.) veins join the basal vein. The lingual gyrus (Lingual Gyr.) is below the calcarine sulcus (Calc. Sulc.). The anterior part of the parahip-pocampal gyrus (Parahippo. Gyr.) is lateral to the uncus. The medial geniculate body (Med. Gen. Body) is below the pulvinar. B. Another specimen. The occiptal lobe has been removed to expose the atrium and the glomus of the choroid plexus (Chor. Plex.). The basal vein courses below the pulvinar. A posterior septal vein (Post. Sep. V.) crosses the roof and medial wall of the body of the lateral ventricle (Body Lat. Vent.). A thalamocaudate vein (Thal. Caud. V.) joins the internal cerebral veins above the pineal body. The lower layer of tela choroidea (Tela) in the roof of the third ventricle is attached to the habenular trigones (Hab. Trig.). The superior choroidal vein (Sup. Chor. V.) drains into the internal cerebral vein, and the inferior choroidal vein (Inf. Chor. V.) drains into the basal vein. The oculomotor nerve (Ш) passes along the medial margin of the cerebral peduncle (Ped.). The lateral atrial vein (Lat. Atr. V.) joins the basal vein above the superior colliculus (Sup. Coll.). The lateral mesencephalic vein ascends anterior to the brachium of the inferior colliculus (Brach. Inf. Coll.). The caudate nucleus (Caudate Nucl.) is lateral to the thalamus. C. Right posterolateral view of another specimen. The superior vermian vein (Sup. Ve. V.) receives the vein of the cerebellomesencephalic fissure (V. of Cer. Mes. Fiss.) and the superior hemispheric veins (Sup. H. V.). The tectal veins (Tectal V.) drain the region of the superior and inferior colliculi (Inf. Coll.). D. Another specimen. The internal cerebral vein has been elevated to show the tectal veins joining the superior vermian vein and the vein of the cerebellomesencephalic fissure. The trochlear nerve (IV) arises below the inferior colliculus. E. Midsagittal section of the pineal region. The internal cerebral vein courses in the velum interpositum (Velum Interpos.) above the massa intermedia (Massa Inter.). The velum interpositum is located between the layers of tela choroidea in the roof of the third ventricle (3 Vent.). The internal cerebral vein receives the medial atrial vein (Med. Atr. V.) from the lateral ventricle (Lat. Vent.) and the epithalamic veins (Inf. Epith. V.) from the region of the pineal body. The tectal veins join the superior vermian vein and the vein of the cerebellomesencephalic fissure. The vein of Galen and the inferior sagittal sinus (Inf. Sag. Sinus) drain into the straight sinus. The fourth ventricle (4 Vent.) is behind the pons. F. Inferior view. The cerebellum has been removed by dividing the superior (Sup. Cer. Ped.), middle (Mid. Cer. Ped.), and inferior cerebellar peduncles just posterior to the floor of the fourth ventricle. The colliculi are above the superior medullary velum (Sup. Med. Vel.). The right basal vein enters a sinus in the tentorium (Tent. Sinus), rather than joining the vein of Galen. The posterior mesencephalic (Post. Mes. V.), internal occipital, posterior pericallosal, and posterior longitudinal hippocampal veins drain into the internal cerebral veins. The epithalamic veins arise in the region of the pineal gland. The vein of the cerebellomesencephalic fissure drains into the superior vermian vein. (From Ono M, Rhoton AL Jr, Peace D, Rodriguez R: Microsurgical anatomy of the deep venous system of the brain. Neurosurgery 15:621-657, 1984.)

Sellar Region The purpose of this section is to review the anatomy of the cranial base important to the various operative approaches to the region of the sella turcica and third ventricle. It deals with the sella turcica, sphenoid bone, sphenoid sinus, cavernous sinus, and pituitary gland. Sphenoid Bone The sphenoid bone is located in the center of the cranial base (28, 40-42). Some part of it is exposed in transcranial operations through the anterior, middle, and posterior cranial fossae and in subcranial approaches through the nose and orbit (Figs. 3.31 and 3.32). The intimate contact of the body of the sphenoid bone with the nasal cavity below and the pituitary gland above has led to the transsphenoidal route being the operative approach of choice for selected tumors involving the third ventricle. The neural relationships of the sphenoid bone

are among the most complex of any bone: the olfactory tracts, gyrus rectus, and posterior part of the frontal lobe rest against the smooth upper surface of the lesser wing; the pons and mesen-cephalon lie posterior to the clival portion; the optic chiasm lies posterior to the chiasmatic sul-cus; and the second through sixth cranial nerves are intimately related to the sphenoid bone. All exit the skull through the optic canal, superior orbital fissure, foramen rotundum, or foramen ovale, all foramina located in the sphenoid bone. The sphenoid bone has many important arterial and venous relationships: the carotid arteries groove each side of the sphenoid bone and may bulge into the sphenoid sinus; the basilar artery rests against its posterior surface; the circle of Willis is located above its central portion; and the middle cerebral artery courses parallel to the sphenoid ridge of the lesser wing. The cavernous sinuses rest against the sphenoid bone, and in-tercavernous venous connections line the walls of the sella turcica and dorsum sellae.

Figure 3.31. Osseous relationships of the sphenoid bone. The sphenoid bone is outlined in each view. A. Anterior view. B. Superior view. C. Lateral view. D. Inferior view. (From Rhoton AL Jr, Hardy DG, Chambers SM: Microsurgical anatomy and dissection of the sphenoid bone, cavernous sinus and sellar region. SurgNeurol 12:63-104, 1979.)

Figure 3.32. A to D. Sphenoid bone. Anterior views. A. Conchal type sphenoid bone. B. Bone with presellar type sphenoid sinus. C. Bone with sellar type sphenoid sinus and well-defined sphenoid ostia. D. Bone with sellar type sphenoid sinus with poorly defined sphenoid ostia and obliquely oriented sphenoidal septae. In the anterior view of the skull the sphenoid bone forms part of the posterior wall and roof of the orbit. In the lateral view the greater wing of the sphenoid bone projects above the zygoma, and the pterygoid plates project below in the lateral view. In the anterior view the sphenoid bone resembles a bat with wings outstretched (Figs. 3.31 and 3.32). It has a central portion called the body; the lesser wings, which spread outward from the superolateral part of the body; the two greater wings, which spread upward from the lower part of the body; and the superior orbital fissure, which is situated between the greater and lesser wings. The vomer, the pterygoid processes, and the medial and lateral pterygoid plates are directed downward from the body. The body of the sphenoid bone is more or less cubical and contains the sphenoid sinus. The superior orbital fissure, through which the oculomotor, troch-lear, and abducens nerves and the ophthalmic division of the trigeminal nerve pass, is formed

on its inferior and lateral margins by the greater wing and on its superior margin by the lesser wing. The inferior surface of the lesser wing forms the posterior part of the roof of each orbit and the exposed surface of the greater wing forms the posterior wall of the orbit. The optic canals are situated above the superomedial margin of the superior orbital fissure. The sphenoid ostia open into the sinus. In the superior view the pituitary fossa occupies the central part of the body and is bounded anteriorly by the tuberculum sellae and posteriorly by the dorsum sellae. The chiasmatic groove, a shallow depression between the optic foramina, is bounded posteriorly by the tuberculum sellae and anteriorly by the planum sphenoidale. The frontal lobes and the olfactory tracts rest against the smooth upper surface of the lesser wing and the planum sphenoidale. The posterior margin of the lesser wing forms a free edge called the sphenoid ridge, which projects into the lateral cerebral fissure to separate the

Figure 3.32 E-H. Sphenoid bone (continued). E. Posterior view. F. Lateral view. G. Superior view. H. Inferior view. (From Rhoton AL Jr, Hardy DG, Chambers SM: Microsurgical anatomy and dissection of the sphenoid bone, cavernous sinus and sellar region. Surg Neural 12:63-104, 1979.) frontal and temporal lobes. The anterior clinoid processes are located at the medial end of the lesser wings, the middle clinoid processes are lateral to the tuberculum sellae, and the posterior clinoid processes are situated at the superolateral margin of the dorsum sellae. The dorsum sellae is continuous with the clivus. The upper part of the clivus is formed by the sphenoid bone and the lower part by the occipital bone. The carotid sulcus is lateral to the body of the sphenoid bone. The superior aspect of each greater wing is concave upward and is filled by the tip of each temporal lobe. The foramen rotundum, through which the maxillary division of the trigeminal nerve passes, is located at the junction of the body and greater wing. The foramen ovale transmits the mandibular division of the trigeminal nerve and the foramen spinosum transmits the middle meningeal artery. When viewed from in-feriorly, the vomer, a separate bone, frequently remains attached to the anterior half of the body

of the sphenoid, and its most anterior portion separates the sphenoid ostia. The pterion and the "keyhole" are two important anatomical landmarks in the region of the greater wing in the lateral view (Fig. 3.31). The pterion is located over the upper part of the greater wing. The "keyhole" is located just behind the junction of the temporal line and the zygo-matic process of the frontal bone several centimeters anterior to the pterion. A burr hole placed over the pterion will be located at the lateral end of the sphenoid ridge. Bone in the region of the pterion is often removed to improve the operative exposure along the sylvian fissure. A burr hole placed at the keyhole will expose the orbit at its lower margin and dura over the frontal lobe at its upper margin. Sphenoid Sinus The sphenoid sinus is subject to considerable variation in size and shape and to variation in

the degree of pneumatization (Figs. 3.32 to 3.34) (3, 7, 12). It is present as minute cavities at birth, but its main development takes place after puberty. In early life, it extends backward into the presellar area and subsequently expands into the area below and behind the sella turcica, reaching its full size during adolescence. As the sinus enlarges, it may partially encircle the optic canals. When the sinus is exceptionally large, it extends into the roots of the pterygoid processes or greater wing of the sphenoid bone and may even extend into the basilar part of the occipital bone. As age advances, the sinus frequently undergoes further enlargement associated with absorption of its bony walls. Occasionally there are gaps in its bone with the mucous membrane lying directly against the dura mater. There are three types of sphenoid sinus in the adult: conchal, presellar, and sellar types, depending on the extent to which the sphenoid bone is pneumatized. In the conchal type the area below the sella is a solid block of bone without an air cavity. In the presellar type of sphenoid sinus, the air cavity does not penetrate beyond a vertical plane parallel to the anterior sellar wall. The sellar type of sphenoid sinus is the most common, and here the air cavity extends into the body of the sphenoid below the sella and as far posteriorly as the clivus. In our previous study in adult cadavers, this sinus was of a presellar type in 24% and of the sellar type in 75% (36). The conchal type is infrequent in the adult. The septae within the sphenoid sinus vary greatly in size, shape, thickness, location, completeness, and relation to the sellar floor (Fig. 3.35). The cavities within the sinus are seldom symmetrical from side to side and are often subdivided by irregular minor septae. The septae are often located off the midline as they cross the floor of the sella. In our previous study a single major septum separated the sinus into two large cavities in only 68% of specimens, and even in these cases the septae were often located off the midline or were deflected to one side (36). The most common type of sphenoid sinus has multiple small cavities in the large paired sinuses. The smaller cavities are separated by septae oriented in all directions. Computerized tomograms of the sella provide the definition of the relationship of the septae to the floor of the sella needed for transsphenoidal surgery. Major septae may be found as far as 8 mm off the midline (36). The carotid artery frequently produces a ser-pigenous prominence into the sinus wall below the floor and along the anterior margin of the sella (Figs. 3.33 and 3.34) (7, 36). Usually, the

optic canals protrude into the superolateral portion of the sinus, and the second division of the trigeminal nerve protrudes into the inf erolateral part. A diverticulum of the sinus, called the op-ticocarotid recess, often projects laterally between the optic canal and the carotid prominence. Removing the mucosa and bone from the lateral wall of the sinus exposes the dura mater covering the medial surface of the cavernous sinus and optic canals. Opening this dura exposes the carotid arteries and optic and trigeminal nerves within the sinus. The sixth cranial nerve is located between the lateral side of the carotid artery and the medial side of the first trigeminal division. The second and third trigeminal divisions are seen in the lower margin of the opening through the lateral wall of the sphenoid sinus. In half of the cases the optic and trigeminal nerves and the carotid arteries have areas where bone 0.5 mm or less in thickness separates them from the mucosa of the sphenoid sinus, and in a few cases the bone separating these structures from the sinus is absent (8, 36). The absence of such bony protection within the walls of the sinus may explain some of the cases of cranial nerve deficits and carotid artery injury after transsphenoidal operations (18). The bone is often thinner over the carotid arteries than over the anterior margin of the pituitary gland. Diaphragma Sellae The diaphragma sellae forms the roof of the sella turcica. It covers the pituitary gland, except for a small central opening in its center, which transmits the pituitary stalk (Fig. 3.36). The diaphragma is more rectangular than circular, tends to be convex or concave rather than flat, and is thinner around the infundibulum and somewhat thicker at the periphery; the opening in its center is large when compared to the size of the pituitary stalk. It frequently is a thin, tenuous structure that would not be an adequate barrier for protecting the suprasellar structures during transsphenoidal operation. An outpouching of the arachnoid protrudes through the central opening in the diaphragma into the sella turcica in about half of the patients. This outpouching represents a potential source of postoperative cerebrospinal fluid leakage (18). Pituitary Gland When exposed from above by opening the diaphragma, the superior surface of the posterior lobe of the pituitary gland is lighter in color than the anterior lobe, and the anterior lobe wraps

Figure 3.33. Stepwise dissection of the lateral wall of the right half of a sellar type sphenoid sinus and adjacent structures (see text under "Incisural and Cister-nal Veins"). A. The sphenoid sinus and sellar area are divided in the midsagittal plane. The optic nerve (Optic N.) is seen proximal to the optic canal. The optico-carotid recess separates the carotid prominence and optic canal. The septum in the posterior part of the sinus is incomplete. B. The sinus mucosa and thin bone of the lateral sinus wall are removed to expose dura mater covering the carotid artery (Carotid A.), the second trigeminal division (V2) just distal to the trigeminal ganglion, and the optic nerve. C. The dura is opened to expose the carotid artery, the optic nerve in the optic canal, the second trigeminal division below the carotid artery, and the abducens nerve (VI) between the first trigeminal division (V1) and the carotid artery. D. Lateral view of the specimen showing area of the cavernous sinus. The oculomotor (HI) and trochlear (IV) nerves are seen above. The intra-cavernous portion of the carotid artery is seen medial to the trigeminal root (V) and the ophthalmic, maxillary, and mandibular divisions (V3) of the trigeminal nerve. The petrous portion of the carotid artery is seen in cross section below the trigeminal nerve. The opening into the sphenoid sinus is located between the first and second trigeminal divisions. E. Trigeminal nerve is reflected forward to expose the carotid artery, the trigeminal impression, the artery of the inferior cavernous sinus (Art. Inf. Cav. Sinus), and the abducens nerve, which splits into three bundles as it passes around the carotid artery. (From Fujii K, Chambers SM, Rhoton AL Jr: Neurovascular relationships of the sphenoid sinus: A microsurgical study. J Neurosurg 50:31-39, 1978.)

Figure 3.34. A. Anterior views of a sellar type sphenoid sinus. The anterior wall of the sella is removed to expose the pituitary gland. The specimen was split at the midline. The air cavity is wider below than above, as is typical in a well-pneumatized specimen. The optic canals are above. Carotid prominences are lateral to the sella. The trigeminal prominence is below the carotid prominence. B. Anterior views of a sellar type sphenoid sinus. The specimen is opened slightly to provide a better view of the lateral wall and the carotid and trigeminal prominences. C. Anterior view of a sellar type sphenoid sinus. Mucosa, dura, and bone of the lateral wall of the sinus are removed to expose parasphenoidal segments of the carotid artery. Sympathetic nerves (Symp. N.) are anterior to the left carotid artery. The orbital contents appear laterally. D. Anterior views of a sellar type sphenoid sinus. Specimen is spread slightly to show the abducens nerve (VI) and the ophthalmic (V1), maxillary (V2), and mandibular (V3) divisions of the trigeminal nerve (V). (From Fujii K, Chambers SM, Rhoton AL Jr: Neurovascular relationships of the sphenoid sinus: A microsurgical study. J Neurosurg 50:31-39, 1978.) around the lower part of the pituitary stalk to form the pars tuberalis (Fig. 3.37). The posterior lobe is more densely adherent to the sellar wall than the anterior lobe. The gland's width is equal to or greater than either its depth or its length in most patients. Its inferior surface usually conforms to the shape of the sellar floor, but its lateral and superior margins vary in shape because these walls are composed of soft tissue rather than bone. If there is a large opening in the diaphragma, the glands tends to be concave superiorly in the area around the stalk. The superior surface may become triangular as a result of being compressed laterally and posteriorly by the carotid arteries (Fig. 3.36). As the anterior lobe is separated from the posterior lobe, there is

a tendency for the pars tuberalis to be retained with the posterior lobe. Intermediate lobe cysts are frequently encountered during separation of the anterior and posterior lobes. Pituitary Gland and Carotid Artery The distance separating the medial margin of the carotid artery and the lateral surface of the pituitary gland usually varies from 1 to 3 mm; however, in some cases the artery will protrude through the medial wall of the cavernous sinus to indent the gland (Fig. 3.37) (3, 14, 36). Heavy arterial bleeding during transsphenoidal hy-pophysectomy has been reported to be caused by carotid artery injury, but may also be caused by a tear in an arterial branch of the carotid artery

Figure 3.35. Septa in the sphenoid sinus. The broken line on the central diagram shows the plane of the section of each specimen from which the drawings were taken, and the large arrow shows the direction of view of the specimens. The planum is above, the dorsum and clivus are below, and the sella is in an intermediate position on each diagram. The heavy dark lines on the drawings show the location of the septae in the sphenoid sinus. A wide variety of septae separate the sinus into cavities that vary in size and shape, seldom being symmetrical from side to side. (From Renn WH, Rhoton AL Jr: Microsurgical anatomy of the sellar region. J Neurosurg 43:288-298, 1975.)

(e.g., the inferior hypophyseal artery) or by avulsion of a small capsular branch from the carotid artery (18). If the carotid arteries indent the lateral surfaces of the gland, the gland does lose its rounded shape and conforms to the wall of the artery, often developing protrusions above or below the artery. Intrasellar tumors are subjected to the same forces, which prevent them from being spherical, and the increased pressure within the tumor increases the degree to which the tumor insinuates into surrounding crevices and tissue planes. Separation of these extensions from the main mass of gland or tumor may explain cases in which the tumor and elevated pituitary hormone levels recur although the hormone levels fell to zero immediately after adenoma removal. Intercavernous Venous Connections Venous sinuses may be found in the margins of the diaphragma and around the gland (36). The intercavernous connections (Fig. 3.38) within the sella are named on the basis of their relationship to the pituitary gland; the anterior

intercavernous sinuses pass anterior to the hypophysis, and the posterior intercavernous sinuses pass behind the gland. Actually, these intercavernous connections can occur at any site along the anterior, inferior, or posterior surface of the gland. The anterior sinus is usually larger than the posterior sinus, but either or both may be absent. If the anterior and posterior connections coexist, the whole structure constitutes the "circular sinus." Entering an anterior intercavernous connection that extends downward in front of the gland during transsphenoidal operation may produce brisk bleeding. However, this usually stops with temporary compression of the channel or with light monopolar diathermy, which serves to glue the walls of the channel together. A large intercavernous venous connection called the basilar sinus often passes posterior to the dorsum sellae and upper clivus. The basilar sinus connects the posterior aspect of both cavernous sinuses and is usually the largest and most constant intercavernous connection across the midline. The superior and inferior petrosal sinuses joint the basilar sinus. The abducens nerve often enters the posterior part of the cavernous sinus by passing through the basilar sinus. Cavernous Sinus The cavernous sinus surrounds the horizontal portion of the carotid artery and a segment of the abducens nerve (Figs. 3.33, 3.36, and 3.39). The oculomotor, trochlear, and ophthalamic divisions of the trigeminal nerve are found in the roof and lateral wall of the sinus (11, 14, 29, 36). The lateral wall of the cavernous sinus extends from the superior orbital fissure in front to the apex of the petrous portion of the temporal bone behind. The oculomotor nerve enters the roof of the sinus lateral to the dorsum sellae. The trochlear nerve enters the roof of the sinus posterolateral to the third nerve, and both nerves enter the dura mater immediately below and medial to the free edge of the tentorium. The ophthalamic division of the trigeminal nerve enters the low part of the lateral wall of the sinus and runs obliquely upward to exit through the superior orbital fissure. The abducens nerve enters the posterior wall of the sinus by passing through the dura lining the upper clivus and courses forward between the intracavernous carotid artery medially and the ophthalmic division laterally. It frequently splits into multiple rootlets in its course lateral to the carotid artery.

Figure 3.36. A. Superior view of the cavernous sinus, anterior and posterior lobe of the pituitary, intracavernous portion of the carotid artery, and meningo-hypophyseal trunk with its three branches: the inferior hypophyseal, tentorial, and dorsal meningeal arteries. The ophthalmic artery arises from the anterior surface of the carotid artery above the clinoid and enters the optic canal under the optic nerve. The dorsum was removed to expose the posterior lobe of the pituitary. The sixth cranial nerve (CN VI) receives a branch from the dorsal meningeal artery and courses laterally around the carotid artery. The dural ostium of the third cranial nerve (CN III) (left) is in the roof of the cavernous sinus, and the right third cranial nerve enters its dural ostium. The carotid artery is exposed in the medial part of the foramen lacerum (left). B. Superior view of the sellar region showing optic nerves, carotid arteries, and the third cranial nerve (CN Ш). Carotid arteries bulge into the pituitary fossa. C. Superior view of the sellar region and both cavernous sinuses. The carotid artery indents the lateral margin of the pituitary gland, and a tongue of pituitary extends over the top of the artery. The sixth cranial nerve (CN VI) passes through the duramater and around the lateral margin of the carotid artery. The third cranial nerve enters the roof of the cavernous sinus lateral to the dorsum sellae. The third through sixth cranial nerves are seen in the lateral wall of the cavernous sinus. D. Superior view of the sellar region. The optical chiasm is reflected forward. A congenitally absent diaphragma exposes the superior surface of the gland. The third cranial nerve is posterior to the carotid arteries. The right ophthalmic artery arises below the optic nerve. (A and С are from Harris FS, Rhoton AL Jr: Anatomy of the cavernous sinus: A microsurgical study. J Neurosurg 45:169-180, 1976. В and С are from Renn WH, Rhoton AL Jr: Microsurgical anatomy of the sellar region. J Neurosurg 43:288-298, 1975.) The branches of the intracavernous portion of the carotid artery are the meningohypophyseal trunk, the artery of the inferior cavernous sinus, and McConnell's capsular arteries (Figs. 3.32, 3.36, and 3.39). The ophthalmic artery may also take origin from the carotid artery within the sinus in few cases (36). The most proximal branch of the intracavernous carotid artery, the

meningohypophyseal trunk, usually arises below the level of the dorsum sellae near the apex of the curve between the petrous and intracavernous segments of the artery. The three branches of the meningohypophyseal artery are the tentorial artery (of Bernasconi-Cassinari), which courses toward the tentorium; the inferior hypophyseal artery, which courses medially to sup-

Figure 3.38. Six sagittal sections of the sellar region showing variations in the intercavernous venous connections within the dura. The variations shown include combinations of anterior, posterior, and inferior intercavernous connections and the frequent presence of a basilar sinus posterior to the dorsum. Either the anterior (lower center) or posterior (lower left) intercavernous connection or both (top center) may be absent. The anterior intercavernous sinus may extend along the whole anterior margin of the gland (lower left). The basilar sinus may be absent (lower right). (From Renn WH, Rhoton AL Jr: Microsurgical anatomy of the sellar region. J. Neurosurg 43:288-298, 1975.)

Figure 3.37. A. Pituitary gland: superolateral view. The posterior lobe is a lighter color and has a different consistency than the anterior lobe. The pars tuberalis partially encircles the stalk. The gland is concave around the stalk. B. Pituitary gland: inferior view. Note the relationship of the anterior and posterior lobes. C. Pituitary gland: anterior and posterior lobe separated. The pars tuberalis partially encircles the stalk. (From Rhoton AL Jr, Hardy DG, Chambers SM: Microsurgical anatomy and dissection of the sphenoid bone, cavernous sinus and sellar region. Surg Neurol 12:63-104, 1979.) ply the posterior part of the capsule of the pituitary gland; and the dorsal meningeal artery, which perforates the dura of the posterior wall of the sinus to supply the region of the clivus and the sixth nerve.

The artery of the inferior cavernous sinus originates from the lateral side of the horizontal segment of the carotid artery distal to the origin of the meningohypophyseal trunk (14). It passes above the abducens nerve and downward medially to the first trigeminal division to supply the dura of the lateral wall of the sinus. In a few cases it arises from the meningohypophyseal trunk. McConnell's capsular arteries, if present, arise from the medial side of the carotid artery and pass to the capsule of the gland, distal to the point of origin of the artery of the inferior cavernous sinus (14). Parkinson's Triangle and the Intracavernous Portion of the Carotid Artery Parkinson described a triangle within the lateral wall of the cavernous sinus through which the intracavernous portion of the carotid artery and its branches might be exposed for the surgical treatment of carotid-cavernous fistulas

(29). The boundaries of the triangle are the lower border of the fourth nerve superiorly, the upper margin of the fifth nerve inferiorly, and the slope along the dorsum sellae and clivus posteriorly (Fig. 3.39). Parkinson opened into the cavernous sinus by incising the dura within Parkinson's triangle starting approximately 6 mm below the dural entrance of the third nerve and extending anteriorly for 2 cm parallel to the slope of the third and fourth nerves and retractng the leaves of the dura. The intracavernous course of the extra-ocular nerves can be located through this opening. Retracting the ophthalmic division of the trigeminal nerve inferolaterally exposes the ab-ducens nerve between the trigeminal nerve and the carotid artery. We found in previous studies that it may be difficult to expose all branches of the intracavernous portion of the carotid artery through this opening (14). Discussion The surgical approaches to the third ventricle may require transcortical or transcallosal incisions; division or displacement of the fornix; opening of the lamina terminalis, septum pellu-cidum, and floor of the third ventricle; and dissection and separation of the tumor from the quadrigeminal plate, the optic nerves, chiasm, and tracts, the pituitary gland and its stalk, the cerebral peduncles, and all of the walls of the third ventricle. Injury of some structures exposed in approaching the third ventricle, such as the olfactory and oculomotor nerves, the optic pathways, and the quadrigeminal plate, causes deficits that are well defined and that correspond to the area injured. The sacrifice of other neural structures has produced variable results: in some cases there was no deficit, and in others the deficit was transient or permanent or resulted in the loss of life. Struc-

Figure 3.39. Superolateral view of the pituitary gland, right cavernous sinus, and intracavernous structures including the carotid artery and the third (CN III), fourth (CN IV), and sixth (CN VI) cranial nerves. A. The lateral dural wall of the cavernous sinus is removed. The tortuous carotid artery bulges superiorly, pushing the interclinoid ligament and cavernous sinus roof upward and indenting the lateral margin of the pituitary gland. The inferior hypophy-seal artery passes to the pituitary gland. The third and fourth cranial nerves enter the roof of the cavernous sinus by passing through the interclinoid ligament. The sixth cranial nerve on the left side enters the dura posterior to the dorsum and passes laterally around the left carotid artery. The sixth cranial nerve is above the superior margin of the trigeminal sensory (CN Vs) and motor (CN VM) roots, and only the first division (CN V1) is exposed distal to the ganglion. B. Further dural removal exposes the trigeminal root and its second (CN V2) and third (CN V3) divisions. The foramen lacerum is exposed lateral to the gasserian ganglion.

The trigeminal root is reflected laterally to show a second branch of the sixth cranial nerve as it passes lateral to the carotid artery. C. The trigeminal root is reflected forward, exposing the carotid artery in the foramen lacerum. A sympathetic nerve bundle is on the surface of the carotid artery in the foramen lacerum. Three rootlets of the sixth cranial nerve are seen passing around the carotid artery. The outline of the carotid artery is marked with a broken line in the areas where it is out of view in the petrous bone and in the cavernous sinus. (From Harris FS, Rhoton AL Jr: Anatomy of the cavernous sinus: A microsurgical study. J Neurosurg 45:169-180, 1976.)

tures sacrificed with variable results include the anterior and posterior parts of the corpus cal-losum and one or both of the fornices. Callosal incisions have resulted in disorders of the inter-hemispheric transfer of information, visuospatial transfer, the learning of bimanual motor tasks, and memory and also in deficits including alexia, apraxia, and astereognosis (20, 34, 46, 52, 56). Division of the fornix on both sides may cause a memory loss (16, 19, 49). The cerebral retraction needed for the anterior and posterior interhemispheric approaches and the cortical incisions for the transventricular surgical approaches have caused convulsions, hemiplegia, mutism, impairment of consciousness, and visual field loss (16, 23, 33). Manipulation of the walls of the third ventricle may cause the following disorders: hypothalamic dysfunction as manifested by disturbances of temperature control, respiration, consciousness, and hypophyseal secretion; visual loss due to damage of the optic chiasm and tracts; and memory loss due to injury to the columns of the fornix in the walls of the third ventricle (23, 24). Dissection in the area of the quadrigeminal plate may cause disorders of eye movement, edematous closure of the aqueduct of Sylvius, blindness from edema in the colliculi or geniculate bodies, and extraocular palsies due to edema of the nuclei of the nerves or the central pathways in the brain stem (6). Splitting the superior part of the cerebellar ver-mis may cause cerebellar edema and brain stem injury (27). Sacrifice of branches of the superficial and deep venous systems has produced inconstant deficits. Before sacrificing these veins, one should try to work around them, displace them out of the operative route, or place them under moderate or even severe stretch (accepting the fact that they may be torn) if it will yield some possibility of their being saved. Another option is to divide only a few of their small branches, which may allow the displacement of the main trunk out of the operative field. Dandy noted that not infrequently one internal cerebral vein had been sacrificed without effect, and on a few occasions both small veins and even the great vein of Galen had been ligated with recovery without any apparent disturbance of function (4-6). On the other hand, it seems that injury to this complicated venous network may cause diencephalic edema, mental symptoms, coma, hyperpyrexia, tachycardia, tachypnea, miosis, rigidity of limbs, and exaggeration of deep tendon reflexes (2, 17, 47, 48). Occlusion of the thalamostriate and other veins at the foramen of Monro may cause

drowsiness, hemiplegia, mutism, and hemor-rhagic infarction of the basal ganglia (16). Obliteration of veins coursing between the cerebrum and the superior sagittal sinus anterior or posterior to the rolandic vein or occlusion of the anterior part of the sagittal sinus, although usually not causing a deficit, may be accompanied by hemiplegia (46). Sacrificing the internal occipital vein or the bridging veins from the occipital pole to the superior sagittal or transverse sinuses may cause hemianopsia (47, 54). Cerebellar swelling after transection of the bridging vein between the cerebellum and tentorium has been reported (27). Numerous arteries including the internal carotid and basilar arteries and the circle of Willis and its branches are frequently exposed during the removal of tumors of the third ventricle, but only infrequently are any of these sacrificed. Symon reported that perforating branches of the posterior communicating artery might be sacrificed in the subtemporal approach to cranio-pharyngioma with a resulting infarct in the basal ganglia, but without obvious deficit (50). Infarction in the distribution of the thalamoperforating branches of the posterior cerebral artery may cause coma and death after the removal of a suprasellar tumor (55). The choroidal arteries are frequently exposed in the approaches through the roof of the third ventricle, but at that point they have given off most of their neural branches. Arterial lesions at the anterior part of the circle of Willis are more likely to result in disturbances in memory and personality, and those at the posterior part of the circle are more likely to result in disorders of the level of consciousness and are frequently combined with disorders of extraocular motion (32, 45, 49). Injuries to the superior cerebellar artery in approaches to the posterior part of the third ventricle may cause a cerebellar deficit. Lesions in the anterior incisural space and in the anterior part of the third ventricle may be approached by the unilateral subfrontal, bifron-tal, frontal-interhemispheric, frontotemporal, subtemporal, transsphenoidal, anterior trans-ventricular, and anterior transcallosal routes. Tumors located anterior to Liliequist's membrane between the optic chiasm and the dia-phragma sella are commonly operated on by the transsphenoidal or subfrontal approach. The transsphenoidal approach is preferred if the tumor extends upward out of an enlarged sella turcica and is located above a pneumatized sphenoid sinus. The subfrontal intracranial approach is reserved for those tumors in the chias-

matic cistern that are not accessible by the transsphenoidal route because they are located entirely above the diaphragma sellae, extend upward out of a normal or small sella, or are located above a nonpneumatized (conchal) type of sphenoid sinus. The subfrontal approach permits exposure of the tumor within the anterior incisural space by four routes: (a) the subchiasmatic approach between the optic nerves and below the optic chiasm; (b) the opticocarotid route directed between the optic nerve and carotid artery; (c) the lamina terminalis approach directed above the optic chiasm through a thined lamina terminalis; and (d) the transfrontal-transsphenoidal approach obtained by entering the sphenoid sinus and sella through the transfrontal craniotomy (43). The subchiasmatic approach is used if the subchiasmatic opening is enlarged by the tumor. The opticocarotid route is selected if parasellar extension of the tumor widens the space between the carotid artery and the optic nerve and the tumor cannot be reached by the subchiasmatic approach. The lamina terminalis approach is selected if the tumor has pushed the chiasm into a prefixed position and extends into the third ventricle to stretch the lamina terminalis so that the tumor is visible through it. The transfrontal-transsphenoidal approach is selected if the tumor grows upward out of the sella, the sphenoid sinus is pneumatized and the tumor does not stretch the lamina terminalis or widen the opticocarotid space, and a prefixed chiasm blocks the subchiasmatic exposure. A bifrontal craniotomy may be used if the tumor extends forward in both anterior cranial fossae and cannot be removed by a unilateral subfrontal exposure. The frontotemporal (pterional) approach is used for a tumor arising from the sphenoid ridge or anterior clinoid process, one arising above the diaphragma and extending along the sphenoid ridge or into the middle cranial fossa, or a lesion accessible through the spaces between the optic nerve and carotid artery or between the carotid artery and the oculomotor nerve. The frontotemporal approach may be combined with a temporal lobectomy to expose lesions extending into the middle incisural space. The anterior transcallosal approach is suitable for lesions located in the anterosuperior part of the third ventricle or extending out of the superior part of the third ventricle into one or both lateral ventricles near the foramen of Monro. The transcallosal approach is easier to perform than the transventricular approach if the ventricles are of a normal size or are minimally enlarged. The anterior transventricular approach is suita-

ble for approaching tumors in the anterosuperior part of the third ventricle, especially if the tumor has a major extension into the anterior part of the lateral ventricle on the side of the approach. It is more difficult to expose the anterior part of the lateral ventricle on the side opposite the craniotomy through the transventricular than through the transcallosal approach. The trans-ventricular approach enters the lateral ventricle in a favorable location for exposure of the superior half of the anterior part of the third ventricle by opening the choroidal fissure or for biopsy or removal of the tumor through an enlarged foramen of Monro, and the transcallosal approach enters the lateral ventricle in a satisfactory position for exposing the superior part of the third ventricle by separating the halves of the body of the fornix in the midline. The ventricular veins provide valuable landmarks in directing one to the foramen of Monro and the choroidal fissure during operations directed through the lateral ventricles. This is especially true if hydrocephalus, a common result of third ventricular tumors, is present because the borders between the neural structures in the ventricular walls become less distinct as the ventricles dilate. The thalamostriate vein is helpful in delimiting the junction of the caudate nucleus and the thalamus because it usually courses along the sulcus separating these structures. The fact that the ventricular veins converge on the choroidal fissure assists in identifying this fissure, which is situated on the periphery of the thalamus. Opening through the fissure in the body of the ventricle will expose the velum inter-positum and the roof of the third ventricle, opening through it in the atrium will expose the quadrigeminal cistern and the pineal region, and opening through it in the temporal horn will expose the crural and ambient cisterns. In the transcortical operative approach through the middle frontal gyrus or the transcallosal approach through the anterior part of the corpus callosum, the veins in the frontal horn are seen to drain posteriorly toward the foramen of Monro because the choroidal fissure does not extend into this area (26). The anterior caudate, anterior septal, superior choroidal, and thalamostriate veins usually join the internal cerebral veins at or near the foramen of Monro. However, these veins may pass through the choroidal fissure behind the foramen of Monro to enter the velum interpositum and course adjacent to the internal cerebral vein for a considerable distance before joining the internal cerebral vein. The junction of the thalamostriate vein with the in-

ternal cerebral vein as seen on the lateral angio-gram usually forms an acute angle at the posterior margin of the foramen of Monro; however, the thalamostriate vein may pass through the choroidal fissure and join the internal cerebral vein posterior to the foramen of Monro. The internal cerebral vein is not seen upon opening into the frontal horn because it courses in the roof of the third ventricle below the body of the fornix. The anterior part of the internal cerebral vein can only be exposed by opening through or displacing the structures forming the roof of the third ventricle. One method of increasing the exposure of the roof of the third ventricle is to section a column of the fornix anterosuperior to the foramen on one side, but this will permit the exposure of no more than a short anterior segment of the internal cerebral vein. To prevent the complications associated with sectioning the fornix, Hirsch et al. sectioned the thalamostriate vein at the posterior margin of the foramen of Monro, rather than damaging the fornix to enlarge the opening in the roof of the third ventricle (16). They stressed that interruption of this vein was harmless; however, some of their patients developed drowsiness, hemiplegia, and mutism, and occlusion of the veins at the foramen of Monro has caused hemorrhagic infarction of the basal ganglia. Other routes to the anterior part of the third ventricle are by the interforniceal approach, in which the body of the fornix is split in the midline and the tela choroidea below the fornix is opened to expose the internal cerebral veins, or the sub-choroidal approach, in which the choroidal fissure is opened between the fornix and thalamus, thus allowing the fornix to be pushed to the opposite side to expose the structures in the roof of the third ventricle (1, 53). The subchoroidal and interforniceal approaches have the advantage of giving access to the central portion of the third ventricle by displacing, rather than dividing, the fibers in the fornix. They also provide a more favorable angle for visualization of the third ventricle than the transforaminal approach. These interforniceal and subchoroidal approaches are uniquely suited to lesions in the superior half of the third ventricle below the roof and behind the foramen of Monro. In both approaches, the tela choroidea below the fornix must be opened to expose the internal cerebral vein and the third ventricle. In the subchoroidal approach, the right half of the fornix and the choroid plexus are displaced to the left side and the third ventricle is entered through the resulting longitudinal opening. A wider opening of the

third ventricle may be provided by displacing the thalamostriate vein and extending the opening into the foramen of Monro. However, occlusion of this vein at the foramen of Monro has caused hemorrhagic infarction of the basal ganglia (16). The maximal limits of the opening are determined by the limits of displacement of these vessels and their branches. The close relationship of the genu of the internal capsule to the foramen of Monro should be kept in mind when operating in this area. The genu of the internal capsule nearly touches the lateral wall of the ventricle in the area lateral to the foramen of Monro near the anterior pole of the thalamus. Approaches to the middle incisural space include the posterior frontotemporal, subtemporal, temporaltransventricular, and lateral suboccip-ital routes. The subtemporal approach with elevation of the temporal lobe is commonly used to expose lesions in the cisterns around the inci-sura. Hemorrhage, venous infarction, and edema after retraction of the temporal lobe during this approach are minimized by placing the lower margin of the craniotomy and the dural exposure at the cranial base so as to reduce the need for retraction and by avoiding occlusion of the bridging veins, especially the vein of Labbe. The ten-torium is frequently divided to increase the exposure or to decompress the brain stem when mass lesions are impacted in the incisura. Resection of part of the parahippocampal gyrus facilitates exposure of the upper part of the middle incisural space. A transventricular approach using a cortical incision in the nondominant inferior temporal gyrus may be used if the lesion involves the temporal horn, choroidal fissure, hippocampal formation, or upper part of the middle incisural space (15). A cortical incision in the occipitotemporal gyrus on the inferior surface of the temporal lobe has been used to minimize visual and speech deficits in exposing the temporal horn of the dominant hemisphere (15). After entering the temporal horn, the surgeon opens the choroidal fissure to expose the middle incisural space. The trochlear nerve is the cranial nerve most frequently injured in the middle incisural space. It can be injured when the free edge is divided and is so thin and friable that it may rupture from gentle retraction on the leaves formed by dividing the tentorium. Lesions in the posterior part of the third ventricle and the posterior incisural space may be approached from above the tentorium along the medial surface of the occipital lobe using an occipital transtentorial approach, through the posterior part of the lateral ventricle using a poste-

rior transventricular approach, and through the corpus callosum using a posterior interhemi-spheric transcallosal approach, or from below the tentorium through the supracerebellar space using an infratentorial supracerebellar approach. The infratentorial supracerebellar and occipital transtentorial approaches, which are most commonly selected for pineal tumors, may be combined with incision of the tentorium approximately 1 cm lateral to the straight sinus. The infratentorial supracerebellar approach is preferred for most lesions in the pineal region because the deep venous system that caps the dorsal aspect of pineal tumors does not obstruct access to the tumor. The approach is best suited to tumors in the midline that grow into the lower half of the posterior incisural space, displacing the quadrigeminal plate and the anterior lobe of the cerebellum. The occipital transtentorial approach is preferred for lesions centered at or above the tentorial edge if there is not a major extension to the opposite side or into the posterior fossa, and for those lesions located above the vein of Galen. The posterior transcallosal approach, in which the splenium is divided, would be used only if the lesion seems to arise in the splenium above the vein of Galen and extends into the posterior incisural space (6). The posterior transventricular approach provides adequate exposure of the atrium and posterior portion of the body of the lateral ventricle and would be the preferred approach to a tumor involving the posterior incisural space if the tumor extends into the pulvinar or involves the atrium or the glomus of the choroid plexus. The preferable approach to the ventricle is through the superior parietal lobule, although Van Wagenen (51) approached the pineal region through the atrium of the lateral ventricle using a cortical incision in the superior temporal gyrus.

6. 7. 8. 9. 10.

1 1

12. 13. 14. 15.

16.

17. 18. 19.

References 1 . Apuzzo ML, Chikovani OK, Gott PS, Teng EL, Zee C, Giannotta SL, Weiss MH: Transcallosal, interfornicial approaches for lesions affecting the third ventricle: Surgical considerations and conse quences. Neurosurgery 10:547-554, 1982. 2. Bailey P: Peculiarities of the intracranial venous system and their clinical significance. Arch Neurol Psychiatry 32:1105, 1934. 3. Bergland RM, Ray BS, Torack RM: Anatomical variations in the pituitary gland and adjacent structures in 225 human autopsy cases. J Neurosurg 28:93-99, 1968. 4. Dandy WE: An operation for the removal of pineal tumors. Surg Gynecol Obstet 33:113-119, 1921. 5. Dandy WE: Benign Tumors in the Third Ventri-

20. 21. 22. 23. 24.

25. 26. 27.

cle of the Brain: Diagnosis and Treatment. Springfield, IL, Charles С Thomas, 1933. Dandy WE: Operative experience in cases of pin eal tumor. Arch Surg 33:19-46, 1936. Fujii K, Chambers SM, Rhoton AL Jr: Neurovascular relationships of the sphenoid sinus: A microsurgical study. J Neurosurg 50:31-39, 1978. Fujii K, Lenkey C, Rhoton AL Jr: Microsurgical anatomy of the choroidal arteries: Lateral and third ventricles. J Neurosurg 52:165-188, 1980. Gibo H, Carver CC, Rhoton AL Jr, et al: Microsurgical anatomy of the middle cerebral artery. J Neurosurg 54:151-169, 1981. Gibo H, Lenkey C, Rhoton AL Jr: Microsurgical anatomy of the supraclinoid portion of the inter nal carotid artery. J Neurosurg 55:560-574, 1981. . Gudmundsson K, Rhoton AL Jr, Rushton JG: De tailed anatomy of the intracranial portion of the trigeminal nerve. J Neurosurg 35:592-600, 1971. Hardy J: Transsphenoidal hypophysectomy. J Neurosurg 34:582-594, 1971. Hardy DG, Peace DA, Rhoton AL Jr: Microsurgical anatomy of the superior cerebellar artery. Neuro surgery 6:10-28, 1980. Harris FS, Rhoton AL Jr: Anatomy of the cavern ous sinus: A microsurgical study. J Neurosurg 45:169-180, 1976. Heros RC: Arteriovenous malformations of the medial temporal lobe: Surgical approach and neuroradiological characterization. J Neurosurg 56:44-52, 1982. Hirsch JF, Zouaoui A, Renier D, Pierre-Kahn A: A new surgical approach to the third ventricle with interruption of the striothalamic vein. Acta Neurochir(Wien) 47:135-147, 1979. Kunicki, A: Operative experiences in eight cases of pineal tumor. J Neurosurg 17:815-823, 1960. Laws ER, Kern EB: Complications of transsphenoidal surgery. Clin Neurosurg 23:401-416, 1976. Little JR, MacCarty CS: Colloid cysts of the third ventricle. J Neurosurg 40:230-235, 1974. Long DM, Chou SN: Transcallosal removal of craniopharyngiomas within the third ventricle. J Neurosurg 39:563-567, 1973. Matsushima T, Rhoton AL Jr, de Oliveria E, et al: Microsurgical anatomy of the veins of the poste rior fossa. J Neurosurg 59:63-105, 1983. Matsushima T, Rhoton AL Jr, Lenkey C: Micro surgery of the fourth ventricle: Part 1. Neurosur gery 11:631-667, 1982. McKissock W. The surgical treatment of colloid cyst of the third ventricle. Brain 74:1-19, 1951. Olivecrona H: Colloidal cysts of the third ventricle, in Olivecrona H, Tonnix W (eds): Handbuch der Neurochirurgie. Berin, Springer-Verlag, 1967, vol 4, pp 101-109. Ono M, Ono M, Rhoton AL Jr, Barry M: Microsurgical anatomy of the region of the tentorial incisura. J Neurosurg 60:365-399, 1984. Ono M, Rhoton AL Jr, Peace DA, Rodriguez R: Microsurgical anatomy of the deep venous system of the brain. Neurosurgery 15:621-657, 1984. Page LK: The infratentorial-supracerebellar ex posure of tumors in the pineal area. Neurosurgery

1:36-40, 1977. 28. Pait TG, Zeal A, Harris FS, Paullus WS, Rhoton AL Jr: Microsurgical anatomy and dissection of the temporal bone. Surg Neural 8:363-39-1, 1977. 29. Parkinson D: Transcavernous repair of carotidcavernous fistula: A case report. J Neurosurg 26:420-424, 1967. 30. Perlmutter D, Rhoton AL Jr: Microsurgical anat omy of the anterior cerebral-anterior communi cating-recurrent artery complex. J Neurosurg 45:259-271, 1976. 31. Perlmutter D, Rhoton AL Jr: Microsurgical anat omy of the distal anterior cerebral artery. J Neu rosurg 49:204-228, 1978. 32. Plum F, Posner JB: The Diagnosis of Stupor and Coma. Philadelphia, FA Davis, 1972, pp 64-118. 33. Poppen JL, Reyes V, Horrax G: Colloid cysts of the third ventricle. J Neurosurg 10:242-263, 1953. 34. Preilowski BFB: Possible contribution of the an terior forebrain commissures to bilateral motor coordination. Neuropsychologia 10:267-277, 1972. 35. Rand RW: Transfrontal transsphenoidal craniotomy in pituitary and related tumors using microneurosurgical techniques. In Rand RW (ed): Microneurosurgery. Mosby, St Louis, 1969, pp 74-86. 36. Renn WH, Rhoton AL Jr: Microsurgical anatomy of the sellar region. J Neurosurg 43:288-298, 1975. 37. Rhoton AL Jr: Anatomy of saccular aneurysms. Surg Neurol 14:59-66, 1980. 38. Rhoton AL Jr, Fujii K, Fradd B: Microsurgical anatomy of the anterior choroidal artery. Surg Neurol 12:171-187, 1979. 39. Rhoton AL Jr, Fujii K, Saeki N, et al: Microsurgical anatomy of aneurysms: Part 1. In Hopkins LN, Long DM (eds): Clinical Management oflntracranial Aneurysms. New York, Raven Press, 1982, pp 201-243. 40. Rhoton AL Jr, Harris FS, Renn WH: Microsurgical anatomy of the sellar region and cavernous sinus. Clin Neurosurg 24:54-85, 1977. 41. Rhoton AL Jr, Maniscalco J: Microsurgery of the sellar region. Neuroophthalmology 9:106-127, 1977. 42. Rhoton AL Jr, Hardy DG, Chambers SM: Microsurgical anatomy and dissection of the sphenoid

49. Sweet WH, Talland GA, Ervin FR: Loss of recent bone, cavernous sinus and sellar region. Surg Neurol 12:63-104, 1979. 43. Rhoton AL Jr, Yamamoto I, Peace DA: Microsur gery of the third ventricle: Part 2. Operative ap proaches. Neurosurgery 8:357-373, 1981. 44. Rosner SS, Rhoton AL Jr, Ono M, Barry M: Microsurgical anatomy of the anterior perforating ar teries. J Neurosurg 61:468-485, 1984. 45. Saeki N, Rhoton AL Jr: Microsurgical anatomy of the upper basilar artery and the posterior circle of Willis. J Neurosurg 46:563-577, 1977. 46. Shucart WA, Stein BM: Transcallosal approach to the anterior ventricular system. Neurosurgery 3:339-343, 1978. 47. Stern WE, Batzdorf U, Rich JR: Challenges of surgical excision of tumors in the pineal region. Bull Los Angeles Neurol Soc 36:105-118,1971. 48. Suzuki J, Iwabuchi T: Surgical removal of pineal tumors (pinealomas and teratomas): Experience in a series of 19 cases. J Neurosurg 23:565-571, 1965. memory following section of fornix. Trans Am Neurol Assoc 84:76-82, 1959. 50. Symon L: The intracranial approach to tumours in the area of the sella turcica, in Symon L (ed): Operative Surgery: Neurosurgery. London, Butterworth, 1979, pp 181-186. 51. Van Wagenen WP: A surgical approach for the removal of certain pineal tumors. Surg Gynecol Obstet 53:216-220, 1931. 52. Van Wagenen WP, Herren RY: Surgical division of commissural pathways in the corpus callosum. Arch Neurol Psychiatry 44:740-759, 1940. 53. Viale GL, Turtas S: The subchoroid approach to the third ventricle. Surg Neurol 14:71-76, 1980. 54. Yamamoto I, Kageyama N: Microsurgical anatomy of the pineal region. J Neurosurg 53:205-221, 1980. 55. Yamamoto I, Rhoton AL Jr, Peace DA: Microsur gery of the third ventricle: Part 1. Microsurgical anatomy. Neurosurgery 8:334-356, 1981. 56. Zaidel D, Sperry RW: Memory impairment after commissurotomy in man. Brain 97:263-272, 1974. 57. Zeal AA, Rhoton AL Jr: Microsurgical anatomy of the posterior cerebral artery. J Neurosurg 48:534-551, 1978.

4 Surgery of the Third Ventricle: Regional Embryology IvanS.Ciric,M.D.

The success of a neurosurgical procedure depends among other factors on the operative approach chosen and the surgical technique used. The development of modern neurosurgical techniques (including microsurgery, laser vaporization techniques, and more recently, stereotacticendoscopic procedures) has made easier and safer the complete removal of tumors in sites previously deemed inaccessible or associated with a high incidence of morbidity and mortality. As for the surgical approaches chosen, they have been modified over the years in accordance with the neurosurgeon's ever increasing knowledge of anatomy and physiology of the area under consideration. Understanding of the embryological planes of development should also enter into consideration in choosing the surgical approach to an intracranial tumor, especially in the region of the third ventricle.

cerebral vesicle (32). Posterior to Rathke's pouch and sac, the cerebral vesicle evaginates to form the pituitary stalk and posterior lobe. Rathke's sac eventually gives rise to the anterior pituitary, which remains separated from the surrounding cerebral vesicle, the stalk, and the posterior lobe by a layer of mesoderm that eventually differentiates into the structures of the diaphragma sellae and the pituitary gland capsule (9). It is not surprising, therefore, that a variety of lesions of different embryological layers can occur in and around the third ventricle. The peculiarities of the embryological development of various lesions will reflect in how they relate not only to the

Third Ventricle and the Pituitary Region The third ventricle develops from the dience-phalic vesicle at the rostral end of the neural tube during the third week of gestation. Soon thereafter, the third ventricle becomes surrounded by the rapidly developing cerebral vesicles except along its roof, which begins to invag-inate, trapping the surrounding mesoderm (Fig. 4.1). Pia mater that eventually invests the floor of the third ventricle forms from the layer of mesoderm adjacent to the ventral surface of the diencephalic vesicle (3). Rathke's pouch and sac form from an invagination of the distal end of the stomodeum into the overlying mesoderm and Figure 4.1. Development of the third ventricle (see text).

third ventricle, but also to the arachnoid membrane. A text on surgery of lesions in and around the third ventricle should include a discussion of the regional embryology of both, lesions abutting against the outer boundaries of the third ventricle as well as those occupying its interior. This chapter includes discussion of the developmental anatomy of pituitary adenomas, craniopharyn-giomas, and neuroepithelial cysts of the third ventricle and of pineal region tumors. In addition, in each instance an attempt is made to explain the neuroanatomy of the lesion with the developmental anatomy of the area under consideration and to correlate these with the choice of surgical approach to be used. Pituitary Adenomas Pituitary adenomas develop as intraparenchy-mal lesions of the anterior lobe (Fig. 4.2) (7). Initially, they consist of a microscopic conglomerate of tumor cells arranged in sheets of uniform, polygonal cells with indistinct cell boundaries. These microscopic adenomas can be found in approximately 20% of routine, postmortem examinations of the pituitary (13). Immunoper-oxidase staining will show that these microad-enomas frequently consist of a variety of hyper-secreting cells with one cell type predominant (27). The initial impetus for the development of a microadenoma is not known. It is conceivable that prolonged use of estrogen-containing contraceptives will stimulate prolactin secretion in a previously hormonally inactive microadenoma (37). Approximately 10% of functioning prolac-tin-secreting microadenomas will show substantive growth over a period of 5 years (45). As they

Figure 4.2. Intraparenchymal stage of pituitary tumor development.

grow, microadenomas compress the surrounding normal anterior pituitary tissue into a relatively tough pseudocapsular layer that blends imperceptibly with the periphery of the usually softer and, in the central part, more necrotic, and at times hemorrhagic adenoma (7). To assure a complete removal of a microadenoma, one should also resect this pseudocapsule along a cleavage plane between it and the surrounding normal, yellow-orange anterior pituitary tissue. A surgical cure is therefore possible in this first, intra-parenchymal stage of pituitary (micro)adenoma development. Depending on the direction of its growth, a pituitary microadenoma will eventually reach the pituitary gland capsule along its superior, lateral, or inferior surface (Fig. 4.3). The pituitary gland capsule will become distended and fragmented. Thus, it is common to find, after opening the dura mater, a pituitary macroadenoma still enveloped by an attenuated and partially fragmented pituitary gland capsule. It is only after this thin layer is incised that a spontaneous extrusion of tumor tissue occurs. During this second or sellar stage of their development, pituitary adenomas can still be removed completely and selectively along the pseudocapsule cleavage plane between the tumor and the surrounding, normal anterior pituitary. In about 10% of cases, the tumor will be locally invasive into the surrounding dural structures (11). As pituitary adenomas continue to grow, they do so in the direction of least resistance, namely, toward the diaphragma sellae, which is covered by the arachnoid membrane. As they extend

Figure 4.3. Sellar stage of pituitary tumor development.

through the diaphragma sellae, pituitary adenomas displace, rather than penetrate, the arachnoid membrane in a superior direction (11). Thus, they preserve their original extraarach-noid location in relationship to all of the important intraarachnoid neurovascular structures. Indeed, even pituitary adenomas that invaginate the floor of the third ventricle and reach as high as the foramen of Monro are in fact separated from these structures by a layer of arachnoid membrane (Fig. 4.4). It is primarily for this reason that an extraarachnoid, transsphenoidal approach to pituitary adenomas is preferable in the majority of patients (Fig. 4.5) (20). A transcranial approach, on the other hand, requires transgression of the subarachnoid space with all of its important neurovascular structures (Fig. 4.5). When the adenoma has reached the suprasellar space it is classified as a pituitary macroade-noma with suprasellar extension. Utilizing the transsphenoidal approach, a gross total removal of a macroadenoma, as evident from the absence of any demonstrable tumor tissue on the postoperative computerized tomographic (CT) scan, can be achieved in about 40% of cases. On the other hand, a complete removal of a macroadenoma, as evident from the endocrine cures achieved, will be possible in only onequarter of such cases (11). A gross total removal of a macroadenoma will depend on a number of factors such as the direction of its growth, the height of the suprasellar extension, the width of the diaphragma sellae, the adenoma consistency, and the presence or absence of any dural invasion. Utilizing the transsphenoidal approach, a mac-

Figure 4.5. The transsphenoidal approach (large arrow) to pituitary tumors is confined to the extraarach-noidal space. The transcranial approach (small arrows) requires transgression of a double layer of arachnoid and of the intervening subarachnoid space. roadenoma with a straight suprasellar extension of less than 2 cm is still amenable to a gross total tumor removal. In contrast, an anterior, posterior, or lateral tumor growth and pituitary adenomas with suprasellar extension of more than 2 cm can rarely be removed in a gross total fashion. A constricted diaphragma sellae usually constitutes a contraindication for a transsphenoidal approach. Soft, pulpy, and especially hem-orrhagic and cystic lesions are easier to remove in a gross total fashion as compared to hard, septate lesions. Finally, dural invasion precludes a gross total tumor removal. Craniopharyngiomas During the development of Rathke's sac and its subsequent anterior-superior rotation, columnar epithelium from the hypophyseal duct will be brought into contact with the infundibular area (Fig. 4.6) (43). When this occurs before the development of the pia mater from the mesoderm that separates Rathke's sac from the diencephalic vesicle, these cell nests will be embedded into the neuroepithelium of the diencephalic floor. As the pia mater develops about the fifth week of gestation, these cell nests will be incorporated into the subpial space (Fig. 4.7) (10) where they may undergo squamous metaplasia (17). Intraventricular Craniopharyngiomas develop from an endophytic growth of the squamous epithelium directed toward the third ventricular cavity (10). Proliferation of the squamous

Figure 4.4. Suprasellar stage of pituitary tumor development. Note displacement rather than penetration of the arachnoid membrane by the tumor.

tuitary, they will be excluded from the subpial space by the formation of the pia mater (Fig. 4.9). Depending on their initial position in relationship to the stalk, these extrapial cell nests may give rise to either a completely intraarachnoid (Fig. 4.10) or a partially intra- and partially ex-traarachnoid craniopharyngioma (10). When the lesion is completely extrapial, a gross total sur-

Figure 4.6. Anterior-superior rotation of the anterior pituitary with implanted epithelial cell nests from the hypophyseal duct.

Figure 4.8. Partial intraparenchymal, subpial craniopharyngioma.

Figure 4.7. Epithelial cell nests from a hypophyseal duct implanted into the neuroepithelium and covered by the developing pia mater. epithelium, however, usually proceeds in an ex-ophytic fashion toward the subarachnoid space. In that case, the lesion will be partially intrapar-enchymal, subpial, and adherent to the neural structures of the hypothalamic floor (Fig. 4.8) (10, 18). The partially intraparenchymal craniopharyngiomas cannot be removed completely. These patients have a normal sella turcica, con-traindicating the transsphenoidal approach. When the cell nests from the hypophyseal duct remain adherent to Rathke's sac or are suspended in the mesenchyma of the infundibular area, separated from the developing anterior pi-

Figure 4.9. Epithelial cell nests from the hypophyseal duct suspended in the mesenchyma surrounding the neuroepithelium.

Figure 4.10. Extrapial, subarachnoid craniopharyngioma. gical removal should be possible utilizing micro-surgical and laser vaporization techniques. In patients with extrapial intraarachnoid cranio-pharyngiomas, the sella turcica is also normal, precluding the transsphenoidal approach. A partially intraarachnoid and partially extraarachnoid craniopharyngioma, a rather frequent occurrence, is usually associated with a large sella, making a transsphenoidal approach possible. Because of their partial intraarachnoid location, a removal of these lesions will inevitably result in a breach of the arachnoid membrane and a per-ioperative cerebrospinal fluid fistula requiring repair. When the hypophyseal duct cell nests remain attached to the upper surface of the developing anterior pituitary or are brought into contact with the lowermost part of the stalk, a craniopharyngioma arising in this location will grow within the sella and thus behave as a pituitary adenoma in that it will continue to grow outside the arachnoid membrane (Fig. 4.11) (10). These tumors are eminently resectable utilizing the transsphenoidal approach (21). Because of their high cholesterol content, intrasellar craniophar-yngiomas are easily distinguishable from pituitary adenomas on a magnetic resonance imaging (MRI) study. It is obvious, therefore, that craniopharyn-

Figure 4.11. Extraarachnoidal craniopharyngioma. giomas have a different relationship to the arachnoid membrane from pituitary adenomas, which remain outside the arachnoid membrane regardless of their size. Craniopharyngiomas can be completely within the arachnoid membrane, partially intra- and partially extraarachnoid, or completely extraarachnoid. Neuroepithelial (Colloid) Cyst of the Third Ventricle By the fourth to fifth week of gestation, the rostral end of the neural tube has enlarged into the diencephalic and the two hemispheric vesicles (3). The initial slitlike communication between the diencephalic and the cerebral vesicles is the early choroidal fissure. It enlarges anteriorly to form the foramen of Monro. While the lateral diencephalic wall and the opposing walls of the hemispheric vesicles undergo rapid proliferation to form the thalamus and hypothalamus, the basal ganglia and the lobes of the cerebral hemispheres, the roof of the diencephalon, and the immediately adjacent dorsal wall of the hemispheric vesicles fail to proliferate. Instead, they undergo a U-shaped infolding resulting in the duplication of the overlying mesoderm, which thus gives rise to the velum interpositum. Further infolding of the fibrovascular stroma of the

early velum interpositum along the medial wall of the lateral ventricles gives rise to the choroid plexus (Fig. 4.1) (12). A similar process of invag-ination of the fibrovascular stroma from the velum interpositum into the diencephalic vesicle leads to the formation of the choroid plexus in the roof of the third ventricle. The infolded fibrovascular stroma of the choroid plexus remains covered, however, by a thin layer of neuroepi-thelium that merges imperceptibly with the surrounding ependyma at the base of the choroid plexus. Thus, the mesenchymal stroma of the choroid plexus is in a strict anatomical sense extraventricular. The corpus callosum develops as a connection between the two cerebral vesicles. The fornix develops as a bundle of fibers between the temporal lobe and the hypothala-mus. The segment of the medial wall of the cerebral vesicles between the corpus callosum and the fornix fails to proliferate and persists as the septum pellucidum (12). Neuroepithelial (colloid) cysts develop when the neuroepithelium of the diencephalic roof begins to invaginate into the third ventricle with a subsequent partial sequestration of a cystlike structure. In the majority of cases, the cyst remains firmly attached to the choroid plexus (Fig. 4.12). According to this concept of their origin, the so-called colloid cysts of the third ventricle are best described as neuroepithelial cysts (4, 12, 38). The surgical approaches to lesions in the anterior and middle sections of the third ventricle should follow the planes of embryological development of the diencephalic area (5, 33). This is best executed either along the midline through

Figure 4.12. Development of neuroepithelial (colloid) cyst.

the corpus callosum (14, 29, 39) or from a lateral direction through the lateral ventricle (28). The third ventricle can then be reached through a distended foramen of Monro (14, 28); through the septum pellucidum and the velum interpositum, reaching into the third ventricle between the two internal cerebral veins (2, 8, 31); or through the choroid fissure itself (44). Although sectioning of the fornix anterior to the foramen of Monro, division of the thalamus striate vein (23), and removal of the anterior tubercle of the thalamus have been advocated to enlarge the foramen of Monro, these maneuvers are usually not necessary and should probably be avoided altogether.

Pineal Region Tumors Pineal gland contains a variety of cells of different origin. The predominant cell type is the pineocytoma, a modified neuroepithelial cell that relates closely to interlobular vessels and contains melanin (22). Other component cell types include glial cells, rarely true ganglion cells, neu-roblast-like cells, fibroblasts, lymphocytes, and related cells, connective tissue, and vessels (22). The development of the pineal gland continues through the first decade of life when connective tissue septi containing blood vessels penetrate the parenchyma of the pineal gland and separate it into anastomosing cords and, indeed, into the first half of the second decade, when laminated calcified bodies, so-called calcospherites, develop in the pineal in their adult form. The classification of pineal neoplasms has undergone an evolutionary change since del Rio-Hortega described in 1932 two main cell types in the human pineal gland (16). According to Russell and Rubinstein (35), pineal region tumors can be classified into four main categories. The most common type are tumors of germ cell origin. The germ cells derive from the yolk sac endoderm and can migrate through the embryo. Thus, they can give rise to tumors in diverse anatomical locations, although usually in such midline structures as the brain, mediastinum, retroperi-toneum, and gonads. The most common germ cell tumor is germinoma (dysgerminoma, atypical teratoma). Less common are teratomas, embryonal carcinomas, endodermal sinus tumors, and choriocarcinomas (listed in order of increasing malignancy). Tumors of the pineal parenchymal cells and pineocytomas constitute approximately 20 to 25% (15, 30). Pineal cells have the capability to differentiate into tumors with divergent histolog-ical makeups. Thus, these tumors may assume

the appearance of ganglionic tumors, astrocyto-mas, or retinoblastomas (34). Indeed, all three components have been found in human pineal neoplasms (40). This is in keeping with the phy-logenetic relationship of the pineal cell with both the neurons and the photoreceptor cells of fish and amphibians (22). Glial tumors and tumors of other cell types, especially meningiomas and benign cysts, are of particular interest to neurosur-geons and should always be included in a differential diagnosis. From this brief embryological discussion, it is obvious that surgical approaches to pineal tumors depend not only on their location, but also on their histological type. A detailed discussion of the various surgical approaches to the pineal region is beyond the scope of this chapter. Although some controversy exists as to whether "pineal tumors" should be operated upon or only radiated (1, 25), it seems more appropriate to consider this issue in light of the histological nature of the lesion. Thus, a correct pretreatment diagnosis arrived at by CT scanning (meningiomas), MRI (cholesterol, fat-containing lesions such as dermoid tumors, teratomas), or stereo-tactic biopsy (2) (germinomas, gliomas) is a prerequisite for choosing the optimal treatment. Although hormonal markers may give a hint as to the histological nature of the lesion, they still fall short of providing an accurate histological diagnosis. In conclusion, it could probably be said that treatment of pineal region tumors should be tailored not only to the location of the lesion and its relationship to the surrounding neurovascular structures, but also to the presumed histological nature of the lesion. Thus, it seems reasonable to suggest that such benign lesions as some teratomas, dermoid cysts, meningiomas, and cystic formations can be cured with surgical excision (41). In contrast, malignant gliomas are best treated with stereotactic biopsy and radiation therapy. Whether tumors of germ cell origin should be treated with radiation therapy and shunting (19, 24, 25, 42), with adjuvant chemotherapy (25) in some instances (germinomas, choriocarcinomas), or with surgical excision (6, 26, 36) followed by radiation therapy and chemotherapy has not been settled. The trend has been toward a more aggressive approach that favors partial resection followed by radiation therapy (25). References 1. Abay EO III, Laws ER Jr, Grado GL, et al: Pineal tumors in children and adolescents: Treatment by

2.

3. 4. 5. 6. 7.

8. 9. 10. 11

12. 13. 14. 15. 16.

17. 18. 19. 20. 21. 22.

23.

CSF shunting and radiotherapy. J Neurosurg 55:889895, 1981. Apuzzo MLJ, Chikovani OK, Gott PS, et al: Transcallosal, interforniceal approaches for lesions af fecting the third ventricle: Surgical considerations and consequences. Neurosurgery 10:547-554, 1982. Arey LB: Developmental Anatomy. A Textbook and Laboratory Manual of Embryology. Phila delphia, W В Saunders, 1965, ed 7. Atunes JL, Louise KM, Ganti SR: Colloid cysts of the third ventricle. Neurosurgery 7:450-455, 1980. Bosch DA, Rahn T, Backlund EO: Treatment of colloid cysts of the third ventricle by stereotactic aspiration. Surg Neurol 9:15-18, 1978. Chapman PH, Lindggood RM: The management of pineal area tumors: A recent reappraisal. Cancer 46:1253-1257, 1980. Ciric I: Microsurgical anatomy of pituitary adeno mas in various stages of their development. Pre sented at the Meeting of the Congress of Neurolog ical Surgeons, Washington, DC, October 1978. Ciric I: Neuroepithelial cysts. J Neurosurg 44:134,1976. Ciric I: On the origin and nature of the pituitary gland capsule. J Neurosurg 46:596-600, 1977. Ciric IS, Cozzens JW: Craniopharyngiomas: Transsphenoidal method of approach—for the virtuoso only? Clin Neurosurg 27:169-187, 1980. . Ciric I, Mikhael M, Stafford T, et al: Transsphe noidal microsurgery of pituitary macroadenomas with long-term follow-up results. J Neurosurg 59:395-401, 1983. Ciric IS, Zivin I: Neuroepithelial (colloid) cysts of the septum pellucidum. J Neurosurg 43:69-73, 1975. Costello RT: Subclinical adenoma of the pituitary gland. Am J Pathol 12:205, 1936. Dandy WE: Benign Tumors in the Third Ventri cle of the Brain: Diagnosis and Treatment. Springfield, Charles С Thomas, 1933. DeGirolami U: Pathology of tumors of the pineal region. In Schmidek HH (ed): Pineal Tumors. New York, Masson Publishing, 1977, pp 1-19. del Rio-Hortega P: Pineal gland. In Penf ield W (ed): Cytology and Cellular Pathology of the Nervous System. New York, Paul В Hoeber, 1932, vol 2, pp 635-703. Erdheim J: Uber Hypophysengangsgeschwulste und Hirmcholesteatome. Sitzungsb Kais Akad Wissen Math Naturw Klin 113:537-726, 1904. Grover WD, Rorke LB: Invasive craniopharyngioma. J Neurol Neurosurg Psychiatry 31:580582, 1968. Handa H, Yamashista J: Summary: Current treat ment of pineal tumors. Surg Neurol 16:279, 1981. Hardy J: Transsphenoidal microsurgery of the normal and pathological pituitary. Clin Neuro surg 16:185-217, 1969. Hardy J, Vezina JL: Transsphenoidal neurosurgery of intracranial neoplasm. Adv Neurol 15:261-274, 1976. Herrick MK: Pathology of pineal tumors. In Neuwelt EA (ed): Diagnosis and Treatment of Pineal Region Tumors. Baltimore, Williams & Wilkins, 1984, pp31-60. Hirsch JF, Zouaoui A, Renier D, et al: A new

24. 25. 26. 27.

28. 29. 30. 31.

32. 33.

surgical approach to the third ventricle with interruption of the striothalamic vein. Acta Neurochir(Wien) 47: 135-147, 1979. Jenkins RDT, Simpson WJK, Keen CW: Pineal and suprasellar geminomas: Results of radiation treatment. J Neurosurg 48:99-107, 1978. Jennings S, Gelman R, Hochberg F: Intracranial germcell tumors: Natural history and pathogenesis. J Neurosurg 63:155-167, 1985. Jooma R, Kendall BE: Diagnosis and management of pineal tumors. J Neurosurg 58:654-665, 1983. Kovacs K, Horvath E: Morphology adrenohypophyseal cells and pituitary adenomas. In Imura H (ed): The Pituitary Gland. New York, Raven Press, 1985, pp 25-56. Little JR, MacCarty CS: Colloid cysts of the third ventricle. J Neurosurg 39:230-235, 1974. Long DM, Chou SN: Transcallosal removal of craniopharyngiomas within the third ventricle. J Neurosurg 39:563-567, 1973. McGovern VJ: Tumors of the epiphysis cerebri. J Pathol Bacteriol 61:1-9, 1949. Pecker J, Guy G, Scarabin JM: Third ventricle tumours: Including tumours of the septum pellicidum, colloid cysts and subependymal glomerate astrocytoma. In Vincken PJ, Bruyn GW (eds): Handbook of Clinical Neurology. Amsterdam, North Holland Publishing Co, 1974, vol 17, pp 440-489. Rathke H: Ueber die entstchung der glandula pituitaria. Arch Anat Physiol. Wissen Med 5:482486, 1838. Rhoton AL Jr, Yamamoto I, Peace DA: Microsur gery of the third ventricle: Part 2. Operative ap proaches. Neurosurgery 8:357-373, 1981.

34. Rubinstein LJ: Tumors of the central nervous system. In Atlas of Tumor Pathology, 2nd series, fascicle 6. Washington, DC, Armed Forces Insti tute of Pathology, 1982. 35. Russell DS, Rubinstein LJ: Pathology of Tumors of the Nervous System. Baltimore, Williams & Wilkins, 1977, ed 4, pp 287-290. 36. Sano K, Matsutani M: Pinealoma (germinoma) treated by direct surgery and postoperative irra diation: A long-term follow-up. Childs Brain 8:81-97, 1981. 37. Sherman BM, Harris CE, Schlechte J, et al: Pathogenesis of prolactin-secreting pituitary adeno mas. Lancet 2:1019, 1978. 38. Shuangshoti S, Robert MP, Netsky MG: Neuroepithelial (colloid) cysts: Pathogenesis and relation to choroid plexus ependyma. Arch Pathol Lab Med 80:214-224, 1965. 39. Shucart WA, Stein BM: Transcallosal approach to the anterior ventricular system. Neurosurgery 3:339-343, 1978. 40. Sobel RA, Trice JE, Neilsen SL, et al: Pineoblastoma with ganglionic and glial differentiation. Acta Neuropathol 55:243-246, 1981. 41. Stein BM: Third ventricular tumors. Clin Neuro surg 27:315-331, 1980. 42. Takeuchi J, Handa H, Nagata I; Suprasellar geminoma. J Neurosurg 49:41-48, 1978. 43. von Mihalkovics V: Wirbesaite und Hirnanhang. Arch Mikrosc Anat 11:389-439, 1875. 44. Viale GL, Turtas S: The subchoroid approach to the third ventricle. Surg Neurol 14:71-76, 1980. 45. Weiss MH, Teal J, Gott P, et al: Natural history of microadenomas: Six year follow-up. Neurosur gery 12:180-183, 1983.

5 Physiological Consequences of Complete or Partial Commissural Section Joseph E. Bogen, M.D.

Complete section of the corpus callosum, which has come to be used increasingly as a treatment for medically intractable epilepsy (88), is regularly followed by a wide variety of neurological and neuropsychological deficits. The deficits that are most apparent immediately after operation make up the acute disconnection syndrome (110). Those deficits that persist make up the stabilized syndrome of hemisphere disconnection (12, 14, 15, 100). Partial section of the corpus callosum to facilitate the operative approach to deep midline lesions may produce a fraction of the full syndrome, depending upon the part of the corpus callosum that has been cut as well as the amount and nature of any associated extracallosal damage. In general, most of the disconnection deficits, both acute and chronic, are not present as long as the splenium is spared (4, 52). But certain partial sections do entail a few inevitable deficits, as well as some common complications, whose recognition is important for the surgeon utilizing the transcallosal approaches. In this chapter I present first the syndromes, acute and chronic, consequent to a complete cerebral commissurot-omy (including the corpus callosum and the underlying dorsal hippocampal commissure, the ventral hippocampal commissure between the fornices, the massa intermedia, and the anterior commissure). This is followed by comments on those particular aspects of the full-blown syndrome that are apt to appear with different partial sections. One should keep in mind that cal-losal lesions, surgical as well as naturally occur-

ring, are often accompanied by damage to or pressure upon neighboring structures, resulting in several distinct types of signs: (a) signs of hemisphere disconnection, (b) neighborhood signs, and (c) nonlocalizing signs such as men-ingismus or signs of increased intracranial pressure. In this chapter, attention is given mainly to signs of the first type. A recent history of studies of the corpus callosum explains how the disconnection signs have come to be understood and emphasized (12). Other extensive reviews are available (8, 16, 27, 28, 37, 42, 70, 80, 81, 106). Historically Salient Signs Before callosal syndromes are considered in detail, it is useful to have an overall view of those signs of hemispheric disconnection that have been emphasized historically and whose reality has been confirmed by more recent investigations. After hemisphere disconnection in the human, unilateral tactile anomia, left hemialexia, and unilateral apraxia are typical. That is, a righthander with complete cerebral commissurotomy cannot name aloud objects correctly manipulated (hence, recognized) with the left hand, cannot read aloud written material presented solely to the left halffield of vision, and cannot execute with the left hand actions verbally named or described by the examiner, although these actions are readily imitated when demonstrated. The apraxia usually recedes within a few months, whereas the hemialexia and unilateral anomia can persist for years.

The relationship of apraxia to the corpus cal-losum was pointed out by Liepmann and Mass (71). They considered the corpus callosum instrumental in left hand responses to verbal command: the verbal instruction was comprehended only by the left hemisphere and the left hand followed instructions delivered by a route involving callosal interhemispheric transfer from left to right and then by what we now call "contra-lateral control," that is, by right hemisphere control of the left hand. Necessarily then, callosal interruption would result in an inability to follow verbal commands with the left hand although there would be no loss of comprehension (as expected from a left hemisphere lesion) and there would be no weakness or incoordination of the left hand (as expected from a right hemisphere lesion). We now recognize the notion that spatial or pictorial instructions understood by the right hemisphere may require callosally mediated interhemispheric communication for correct right hand execution. This right-to-left aspect of callosal function was not part of Liepmann's original callosal concept although, in retrospect, it seems a natural corollary. Liepmann himself considered the left hemisphere to be the organizer of complex (particularly learned) motor behavior. Whether and in what way the left hemisphere is dominant for skilled movements generally (and not just those linguistically related) are currently matters of active controversy (56, 58, 63, 66, 115). Acute Disconnection Syndrome (As Observed in RightHanders Operated from the Right Side) During the first few days after complete cerebral commissurotomy, the patients commonly respond reasonably well, with their right limbs, to simple commands. But they are easily confused by three- or even two-part commands, each part of which is obviously understood. The patients often lie quietly and may seem mildly "akinetic," although cooperating when stimulated. There is sometimes an "imperviousness" resembling that often seen with naturally occurring genu lesions (2, 12). The patients are often mute even when willing to write short (usually one-word) answers. This mutism is of special interest and is discussed in more detail under "Postcallosotomy Mutism." The left-sided apraxia to verbal command is usually severe and can be mistaken for hemiple-gia. Similarly, a disregard for the left half-field of vision can be mistaken for a hemianopsia. Left side weakness in the first week or so due to

retraction edema in the right hemisphere sometimes confounds the picture. The neurological status is difficult to evaluate not only because the seemingly flaccid and unresponsive left extremities may exhibit coordinated movements, but also because, as the patient improves, there may be competitive movements between the left and right hands. Moreover, some patients have focal motor seizures, manifested by clonic contractions on alternating sides of the body and without loss of consciousness, occasionally followed by transient unresponsiveness of whichever limbs were involved. The patients commonly have bilateral Babinski signs as well as bilaterally absent superficial abdominal reflexes. Well-coordinated but repetitive reaching, groping, and grasping with the left hand sometimes resembles a grasp reflex; grasp reflexes may actually be present bilaterally for a day or two. When forced grasping cannot be elicited (by inserting two fingers into the patient's palm from the ulnar side, with some distal stroking), it is nevertheless possible in most cases to demonstrate a proximal traction response (PTR) (i.e. the patient is unable to relax the hand grip when the examiner pulls so as to exert traction on the elbow and shoulder flexors). The left arm hypo-tonia, the left arm PTR, the responses bilaterally to plantar stimulation, and the mutism were regularly observed in our cases; two representative graphic summaries are shown in Figures 5.1 and 5.2. Figure 5.1 indicates the intensity and duration of mutism and other features of the acute disconnection syndrome in our first patient, W.J. (21-23). His commissural section was done without entering the third ventricle; two exposures, before and behind the rolandic veins, made it possible to confirm under direct vision that his callosal section was complete. The patient's condition worsened at the end of the first week, associated with an alkalotic hyponatremia present on Days 5 to 7 (the patient had had diabetes insipidus ever since his previous head injury, and vas-opressin was used throughout his hospital stay). When the hyponatremia was corrected, the patient rapidly improved with respect to alertness and left side coordination, with subsidence of the hypotonia and PTR. But some degree of mutism persisted for another month. This patient's mutism as well as that of several others has been described in detail elsewhere (11). As discussed below, the duration of mutism seems to be related to the extent of extracallosal damage; in this case there was distinct atrophy of the right frontal lobe quite evident at the time of operation, as

Figure 5.1. Features of the acute disconnection syndrome as observed postop-eratively in Patient W.J., a man aged 48 at the time of operation on Feb. 2, 1962. The capital letter N means "none" (i.e., normal). A solid block in black indicates a marked deficit, whereas interrupted black bars mean that the deficit was present but mild, the observations having been made by the author (J.E.B.). Where the solid block or interrupted bars are gray, the observation was made by someone else, usually a resident.

Figure 5.2. Features of the acute disconnection syndrome as observed postop-eratively in Patient N.G., a woman aged 30 at the time of operation on Sept. 5, 1963. The symbols are the same as for Figure 5.1. well as a left parietotemporal electroencephalogram (EEG) focus and bitemporal EEG foci on sphenoid recordings (92). Figure 5.2 shows the daily postoperative observations of key features of the acute disconnection syndrome in Patient N.G. (13, 18, 23). Her anterior commissure and massa intermedia were sectioned under direct vision after opening into the third ventricle between the fornices. Recent magnetic resonance imaging studies have con-

firmed that her callosotomy was complete and that there was a paucity of extracommissural damage. That her extracommissural damage was slight is probably important with respect to the relatively brief duration of her mutism. Her postoperative course was overall quite smooth (small amounts of vasopressin were used on Days 3 to 6). Even so, there was a marked left apraxia to verbal command for many weeks. There was also, as in every one of our 12 right-handers

tested after complete commissurotomy, a persistent anomia in the left hand. The left hand anomia has been such a regularly appearing, easily tested, and persistently present phenomenon that a separate section, "Unilateral (Left) Tactile Anomia," is devoted to its further description. N.G.'s postoperative course was summarized at the end of a month by the author: Two hours after operation the patient did not react to pain and an hour later there was bilateral flexion to sternal rub. An hour later there was a good grip on the right to request, the right toe sign was positive and the entire left side was flaccid. In succeeding hours the left side remained flaccid to passive motion and to verbal request but the patient used both hands together to pull the covers over her chest. Hyperreflexia and positive toe sign on the right side subsided after several days; a tonic grasp and proximal traction response on the right rapidly decreased but could be intermittently elicited as late as a week postoperatively. On the first postoperative day there was occasional mumbling including "yes"and "no," on the second day there was intelligible speech and on the third day the patient was well oriented for place and person although not for time and did not recognize her husband. Disorientation for time and date and intermittent inability to recognize the doctors persisted for nearly two weeks. Complete apraxia, tonic grasp and PTR remained pronounced on the left side for two weeks. By the end of the third week, the left grasping had largely subsided; no verbal instructions were followed with the left hand but poses and gestures were copied excellently. The patient graduated from wheelchair to walker after two weeks but continued to have difficulty controlling the left leg. By the fourth week the patient was walking very well although still apraxic with the left hand. A positive Babinski sign has been present not only contralateral to the retracted hemisphere but also, for at least a day or two, on the ipsilat-eral side. An ipsilateral (as well as contralateral) toe sign was also seen in our one patient operated from the left, Patient C.C. (9, 23). The best explanation for this ipsilateral Babinski sign is probably that the diaschistic shock effect on the unexposed hemisphere after an extensive cal-losal section is sufficient to depress for a few days the corticofugal inhibition of the primitive extensor response (10, 107). In any event, the Babinski signs rapidly subside bilaterally and are not present over the long term (24). An ipsilateral Babinski sign is usually not observed when the callosal section is partial. It is not possible to include here all of the daily progress notes on this patient, who had a rather smooth, almost ideal postoperative course after a complete section, but the following provides some idea of the patient's status at the end of 2l/2 weeks:

Progress Note of Sept. 23, 1963 The patient has progressed rapidly in the past week and is now up walking; she eats by herself without difficulty and discusses things at some length although after a lengthy conversation she tends to become a little bit confused and says, "you people must have taken out my brains." When the patient walks she tends to have a slight shuffle but there is no asymmetry. However, if one asks her to walk on her heels she does this readily with the right foot but does not do it with the left foot. If one asks her to walk standing up on her tiptoes she again does this only with the right foot; she could not rise up on either the heel or the toe of the left foot. The hand is progressing and whereas some days ago there was a PTR on the left the patient now has no difficulty letting go with the left hand when asked to do so, even if considerable stretch is applied to the flexor joints. However, apraxia of the left hand continues in that the patient has considerable difficulty carrying out any command with the left hand and often, while laughing, reaches over with the right hand to make the left hand behave in a more appropriate or precise fashion. When the patient was asked to put her left hand on her head she could not do this, but when she was asked to imitate the examiner when he put his left hand on his head she did it without any difficulty. She was then asked, "now put your left hand on your head" and she did this right away. She was then asked, "put your left hand on your left ear" and the left hand went on top of the head. This could easily be duplicated with any combination of movements; that is, the patient would be shown how to do something and do it excellently with the left hand and then she would be asked to do something else and the hand would do what she had previously just done. . . . Without warning, I put my hand in front of my forehead so that a finger was placed in each supra-orbital notch with the palm of my hand up over my forehead and the little finger sticking out to the side. No sooner had I done this without saying anything, when the patient's left arm went up in complete imitation, to the exact positioning of the fingers. After her arm was lowered, the patient was then asked, "just lift your arm up in the air." The left arm went up with palm on the forehead, in an imitation of the previous peculiar position. Reflex testing showed no asymmetry; the knee jerks were quite hyperactive bilaterally but the reflexes at the ankle, biceps and triceps were normal. Plantar stimulation produced some withdrawal on each side with no suggestion of dorsiflexion. Testing with the pinwheel disclosed that on the left the patient felt the pinwheel to be somewhat less sharp than the corresponding parts of the right side; this hypalgesia is similar to what she had before surgery and is quite distinct from the "shock-like" sensation which W. J. has on the left side of his body. Vibratory sense is apparently quite normal on the right side but the patient is not cooperative enough to give reliable results over the left side. As the severity of the acute disconnection syndrome subsides in a week or so, a phenomenon variously called "intermanual conflict" or "the alien hand" appears. Almost all complete com-

missurotomy patients manifest some degree of intermanual conflict during the early postoperative period. For example, a few weeks after operation, Patient R.Y. (23) was seen buttoning his shirt with one hand while the other hand was following along just behind undoing the buttons. Intermanual conflict was observed after commissurotomy by Wilson et al. (110) and by Ake-laitis (1), who called it "diagnostic dyspraxia." The phenomenon has been described in many case reports of callosal infarcts or tumors. Even our youngest patient (L.B.) with no appreciable apraxia to verbal command in the long term, manifested this "alienation" 3 weeks after operation. While doing the block design test uniman-ually with his right hand, his left hand came up from beneath the table and was reaching for the blocks when he slapped it with his right hand and said, "That will keep it quiet for a while." Such behavior progressively subsides after callosotomy, probably because of other integrative mechanisms supplementing or replacing com-missural function. More recently, examples of intermanual conflict have been reported in detail by Rayport et al. (87) and Ferguson (43). In rare instances, the intermanual conflict may reappear even years later. An example is the following from a follow-up examination done on Feb. 26, 1973. (Patient A.M. was operated on Jul. 7, 1964, at age 31). The most interesting finding in the entire examination is the frequent occurrence of well-coordinated movements of the left arm which are at cross-purposes with whatever else is going on. These sometimes seem to occur spontaneously, but on other occasions are clearly in conflict with the behavior of the right arm. For example, when attempting a Jendrassic reinforcement, the patient reached with his right hand to hold his left, but the left hand actually pushed his right hand away. While testing finger-to-nose test (with the patient sitting), his left hand suddenly started slapping his chest like Tar-zan. The patient follows verbal instructions readily with the right hand, but poorly if at all with the left hand, although he can raise and lower the left arm with considerable reliability, the defect being more obvious the more distal the joints involved in the requested act. After a variety of maneuvers (the left hand often following a visual demonstration fairly well) it was possible to test the grip of the left hand while keeping the thumb extended—so it was apparent that there is a negative Wartenberg thumb sign on the left as well as on the right. In most patients, the intermanual conflict progressively subsides in a few months or even weeks. Patient N.G. had only mild intermanual conflicts in the hospital and none were noted after her return home. Her status 5 months post-operatively was as follows:

Progress Note of Feb. 2, 1964 N.G. is doing excellently. There is still some lack of initiative and an occasional defect in memory, but otherwise her husband and she were both delighted with her progress when I visited them in their home today. The surgical wound has healed well and has a good cosmetic appearance except for a small depression in the most anterior burr hole. She has no headaches. The patient is eating well and has gained weight although she has been doing calisthenics each morning to the TV. She sleeps well except for an occasional nocturia (she needs no help with personal care); she recently had a urethral dilatation and she takes mandelamine 4 times a day along with her Dilantin 100 mg and Mebaral 50 mg. She has had no seizures (in particular no convulsive movements) unless one wishes to suspect as seizures an occasional inattention and some numbness of the left side. This absence of seizures is in spite of the fact that the patient has had two menstrual periods since surgery, the most recent a few weeks ago. The numbness of the left side of the body is apparently continuous at a low level, but once or twice a day it is quite bothersome for 5 to 10 minutes at a time. It involves more the leg (thus resembling her aura before surgery) and also the arm but not the face. In spite of the "numbness" the patient has no disability with the left hand, which she uses quite naturally for dressing, cooking, etc. Although she occasionally has trouble fastening her bra behind her back she says that she dresses with ease and demonstrated for me twohanded tying of shoe laces, this being done in a manner indistinguishable from normal. The patient also shuffled a pack of cards and dealt them deftly. She dealt onehanded with either hand (slightly better with the left) and at my request she unbuttoned and rebuttoned the top of her blouse with her right hand while simultaneously dealing cards with the left hand. The patient wrote well with the right hand but could make no legible characters with the left hand on several trials. However, copying of designs was somewhat better with the left. Her deficit with the right hand appeared to be a deficit of conceptualization whereas the deficit in the left hand appeared to be one of muscular coordination. It is of interest that neither the patient nor her husband could recall any incident of intermanual conflict since her arrival home. On blindfold testing today, the patient followed my verbal commands with the left quite well, her unilateral apraxia seemingly overcome. However, the anomia was still present in the left hand to full extent. I repeatedly demonstrated this with various objects while her husband sat watching in open mouthed silence (Fig. 5.2). The patient's personality seems to me essentially unchanged from before surgery; she laughs easily and appropriately. Her husband says she talks less ("she used to talk my ear off") but that she socializes well with her neighbors. She is perhaps a little more interested in sexual activity than before the operation. Her memory occasionally fails in that she may sometimes accompany her husband somewhere and on the next day ask him what they did. The memory loss occasionally takes the form of absent minded-

ness; while helping her husband prepare lunches for the family, "she sometimes puts 2 or 3 slices of lunch meat in a sandwich instead of one." During my visit the patient was quite interested and alert and not only responded appropriately to questions, but spoke spontaneously on a number of subjects and did not appear to have any memory difficulties at this time. Her husband makes each night a list of housework for her to do the next day, and, "she does everything well; but if I forget to make the list, she will just sit around the next day and do nothing." He makes out a menu and she cooks the meals. "I used to have trouble getting things to come out at the same time," she said, "but the meals are better now." (Her husband nodded agreement.) N.G. remained seizure-free for 8 years and then had status epilepticus a few months after her family physician discontinued her anticon-vulsant medication (13). Her seizures were readily controlled when her medication was reinsti-tuted, and her EEG abnormalities cleared during the next 18 months. When N.G. was examined in November 1984, she continued to be seizure-free, to maintain her household efficiently, and to participate in laboratory studies at Cal Tech. Her left-handed anomia was still readily demonstrable 21 years postoperatively. This persistence of unilateral tactile anomia was also true of all of our other patients (17). Chronic, Stabilized Syndrome of Hemisphere Disconnection Overall Effects Within a few months after operation, the symptoms of acute hemisphere disconnection become compensated to a remarkable degree. In personality and social situations the patient appears much as before. However, when input is lateralized (i.e., presented to only one hemisphere) each hemisphere can be shown to operate independently to a large extent. Each hemisphere seems to have its own learning processes and its own separate memories. Split-brain patients soon accept the idea that they have capacities of which they are not conscious, such as lefthand retrieval of objects not nameable. They may quickly rationalize such acts, sometimes in a transparently erroneous way (46). But even many years after operation the patients will occasionally be quite surprised when some well-coordinated or obviously well-informed act has just been carried out by the left hand. This is particularly common under conditions of continuously lateralized input (116, 118). Visual Effects: Double Hemianopia Visual material can be presented selectively to a single hemisphere by having the patient fix his

gaze on a projection screen onto which pictures of objects or symbols are backprojected to the right, left, or both visual half-fields using exposure times of 0.1 second or less. The patients can read and describe material of various kinds in the right half-field essentially as before operation. When stimuli are presented to the left half-field, however, the patients usually report that they see "nothing" or at most "a flash of light." The disconnection (if it includes the splenium) can sometimes be demonstrated with simple confrontation testing. The patient is allowed to have both eyes open but does not speak and is allowed to use only one hand (sitting on the other hand, for example). Using the free hand, the subject indicates the onset of a stimulus, such as the wiggling of the examiner's fingers. With such testing there may seem to be a homonymous hemianopia in the half-field contralateral to the indicating hand. When the patient is tested with the other hand, there seems to be a homonymous hemianopia in the other half-field. This situation must be distinguished from the much more commonly occurring extinction or hemiinattention deficit from a hemispheric lesion, in which the patient tends to indicate only one stimulus when the stimuli are in fact bilateral (109). An observable difference is that the double hemianopia is a symmetrical phenomenon (the deficit occurs on each side) whereas extinction or hemiinattention is typically onesided, more commonly for the left side. In most patients there eventually appears a condition in which no field defect can be demonstrated by a casual confrontation technique. This apparently depends mainly upon the ability of each hemisphere to direct the head and eyes. For example, if the patient is instructed to point with the right hand and if the examiner then wiggles the fingers in the patient's left visual half-field, the patient's head and eyes quickly turn to the left and then the right hand points to the correct target. If turning of the head is prevented, a leftward glance will suffice for the patient's need for a cue. Or the right hand may point to the left visual half-field as soon as it is apparent that there is no suitable stimulus in the right visual half-field. This "cheating" by the left hemisphere can often be detected by providing no stimulus at all. Furthermore, if stimuli are presented simultaneously in both visual halffields, only the stimulus in the right half-field is described by the patient, that is, by the left hemisphere. There is usually no verbal response to the stimulus in the left visual half-field until the left hemisphere realizes

that the patient's left hand is also in action, pointing to the left half-field stimulus. When a patient has been tested repeatedly, so that the occurrence of bilateral stimuli can be anticipated, both stimuli may be identified by a single hand; if using the right hand, the patient will point first to the right and then to the left half-field. In those patients who have been frequently tested, the appearance of a stimulus in the left visual half-field is occasionally recognized in spite of our attempts to circumvent the various cross cueing strategies. However, even in our patient who does the best (L.B.), performance on confrontation testing of visual fields is still distinguishable from normal. Auditory Suppression After cerebral commissurotomy, the patient readily identifies single words (and other sounds) if they are presented to one ear at a time. But if different words are presented to the two ears simultaneously ("dichotic listening") only the words presented to the right ear will be reliably reported (40, 79, 95). This large right ear advantage is usually considered the result of two concurrent circumstances: (a) the ipsilateral pathway (from the left ear to the left hemisphere) is suppressed by the presence of the simultaneous but differing inputs, as it is in intact individuals during dichotic listening (65, 103). (b) The contralateral pathway from the left ear to the right hemisphere conveys information that ordinarily reaches the left (speaking) hemisphere by the callosal pathway and that has now been severed. Although left ear words are rarely reported, their perception by the right hemisphere is occasionally evidenced by appropriate actions of the left hand (50). Contralateral ear suppression commonly appears after hemispherectomy or the creation of another large hemispheric lesion. Because there is usually suppression of the right ear by left hemisphere lesions, suppression of the left ear by a left hemisphere lesion has been called "paradoxical ipsilateral extinction." Further observations have led to the conclusion that, whether the lesion is in the left or the right hemisphere, if it is close to the midline the suppression of left ear stimuli is probably attributable to interruption of interhemispheric pathways (32, 76, 96). Unilateral Apraxia The degree of left hand dyspraxia is subject to large individual differences. Immediately after operation all of our patients showed some left-sided apraxia to verbal commands such as "Wig-

gle your left toes" or "Make a fist with your left hand." Left limb dyspraxia can be attributed to the simultaneous presence of two deficits: poor comprehension by the right hemisphere (which has good control of the left hand) and poor ipsilateral control by the left hemisphere (which understands the commands). Subsidence of the dyspraxia, therefore, can result from two compensatory mechanisms: increased right hemisphere comprehension of words or increased left hemisphere control of the left hand. The extent of ipsilateral motor control can be tested by flashing to right or left visual half-field sketches of thumb and fingers in different postures for the subject to mimic with one or the other hand. Responses are poor with the hand on the side opposite the visual input, simple postures such as a closed fist or an open hand being attainable after further recovery. As recovery proceeds, good ipsilateral control is first attained for responses carried out by the more proximal musculature. After several months, most of the patients can form a variety of hand and finger postures with either hand to verbal instructions, such as, "Make a circle with your thumb and little finger" and the like. Even many years later, however, some degree of apraxia can be demonstrated (115). Somesthetic Effects The lack of interhemispheric transfer after hemisphere disconnection can be demonstrated with respect to somesthesis (including touch, pressure, and proprioception) in a variety of ways. Cross Retrieval of Small Test Objects Unseen objects in the right hand are handled, named, and described in normal fashion. In contrast, attempts to name or describe the same objects held out of sight in the left hand consistently fail. In spite of the patient's inability to name an unseen object in his left hand, identification of the object by the right hemisphere is evident from appropriate manipulation of the item, showing how it is used, or by retrieval of the same object with the left hand from among a collection of other objects screened from sight. What distinguishes the split-brain patient from normal is that excellent same hand retrieval (with either hand) is not accompanied by the ability to retrieve with one hand objects felt with the other. Cross Replication of Hand Postures Specific postures impressed on one (unseen) hand by the examiner cannot be mimicked in the

opposite hand. Also, if a hand posture in outline form is flashed by tachistoscope to one visual half-field, it can be copied easily by the hand on that side but usually not by the other hand. A convenient way to test for lack of interhem-ispheric transfer of proprioceptive information is as follows: the patient extends both hands beneath the opaque screen (or vision is otherwise excluded) and the examiner impresses a particular posture on one hand. For example, one can put the tip of the thumb against the tip of the little finger and have the other three fingers fully extended and separated. The split-brain patient cannot mimic with the other hand a posture being held by the first hand. This procedure should be repeated with various postures and in both directions. Cross Localization of Finger Tips After complete cerebral commissurotomy there is a partial loss of the ability to name exact points stimulated on the left side of the body. This defect is least apparent, if at all, on the face and it is most apparent on the distal parts, especially the finger tips. This deficit is not dependent upon language; it can be shown in a nonverbal (picture identification) fashion, in which case the deficit is present in both directions (right-to-left and vice versa). An easy way to demonstrate the defect is to have the subject's hands extended, palms up (with vision excluded). One touches the tip of one of the four fingers with the point of a pencil, asking the patient to then touch the same point with the tip of the thumb of the same hand. Repeating this maneuver many times produces a numerical score, about 100% in normals for either hand. In the absence of a parietal lesion, identification of any of the four finger tips by putting the thumb tip upon the particular finger can be done at nearly 100% level by the split-brain patient. One then changes the task so that the finger tip is to be indicated, not by touching it with the thumb of the same hand, but by touching the corresponding finger tip of the other hand with the thumb of that (other) hand. Sometimes the procedure should be demonstrated with the patient's hand in full vision until the patient understands what is required. This cross localization cannot be done by the split-brain patient at much better than chance level (25%), whereas most normal adults do better than 90%. An incompetence to cross localize or cross match has been found in young children (44) possibly because their commissures are not yet fully functioning (113).

Verbal Comprehension by the Right Hemisphere Auditory comprehension of words by the disconnected right hemisphere is suggested by the subject's ability to retrieve with the left hand various objects if they are named aloud by the examiner. Visual comprehension of printed words by the right hemisphere is often present; after a printed word is flashed to the left visual half-field, the subject is often able to retrieve with the left hand the designated item from among an array of hidden objects. Control by the left hemisphere is excluded in these tests because incorrect verbal descriptions given immediately after a correct response by the left hand show that only the right hemisphere knew the answer. While the disconnected right hemisphere's receptive vocabulary can increase considerably over the years, this single word comprehension is rarely accompanied by speech. The most extreme cases (to date) of right hemisphere language ability in right-handed (and left hemisphere-speaking) split-brain subjects are two with right hemisphere speech, both with the anterior commissure uncut (75, 93). Right hemisphere language in the split-brain subject has other limitations, with syntactic ability being rudimentary at best. After studying a few cases in great depth for over 10 years, Zaidel concluded: Whereas phonetic and syntactic analysis seem to specialize heavily in the left hemisphere, there is a rich lexical structure in the right hemisphere. The structure of the right hemisphere lexicon appears to be unique in that it has access to a severely limited short term verbal memory, and it has neither phonetic encoding nor grapheme-to-phoneme correspondence rules . . . [this] represents the limited linguistic competence that can be acquired by a nonlin-guistic, more general purpose (or other purpose) cognitive apparatus (117). Right Hemisphere Dominance After commissurotomy, each hemisphere can be tested separately, demonstrating in a positive way those things that each hemisphere can do better than the other, rather than inferring what a hemisphere does from loss of function when it is injured. Representative reviews are included in the References (16, 69, 83, 99, 118). The rapidly growing literature on hemispheric specialization was thoroughly summarized by Brad-shaw and Nettleton (25) and by Trevarthen (105) and more recently was discussed by Gordon (51). Unilateral (Left) Agraphia Right-handers can write legibly, albeit not fluently, with the left hand. This ability is com-

monly lost with callosal lesions, especially those that cause unilateral apraxia. An inability to write to dictation is common with left hemisphere lesions, almost always affecting both hands. The left hand may be dysgraphic if affected by a right hemispheric lesion, such as a frontal lesion causing forced grasping. That the left dysgraphia after callosal sectioning is not simply attributable to an incoordination or paresis can be established if one can demonstrate some other ability in the left hand requiring as much control as would be required for writing. The left hand may spontaneously doodle or it may copy various designs or diagrams. It is not so much the presence of a deficit but rather the contrast between certain deficits and certain retained abilities that is most informative. Simple or even complex geometric figures can often be copied by a left hand that cannot write or even copy writing previously made with the patient's own right hand (9, 19, 67, 115). Unilateral (Left) Tactile Anomia The most useful single sign of hemisphere disconnection is unilateral tactile anomia: this is an inability to name or to describe an object when it is felt by one hand whereas it is readily named (or well described if the name is unknown) when it is placed into the other hand or when it is presented either to vision or to audition. This unilateral tactile anomia was present in every one of our complete commissurotomy patients, in the left hands of 12 righthanders and in the right hand of our 1 left-handed patient. Others have repeatedly confirmed this finding, and it has also been observed when the patient has had a callosotomy sparing the anterior commissure (46, 74). Not only is unilateral anomia quite regular in its appearance, but it is also quite persistent, whenever appropriately tested, for as long as 20 years (17) (and longer, as shown by my more recent testing). In additon to its regularity and its persistence, the demonstration of this sign requires a minimum of equipment and time, and the interpretation of results is usually quite clear. Of the many maneuvers developed in the laboratory to test split-brain patients, this is the principal one to be adopted as part of a routine neurological examination. Although there are many signs of brain bisection (as described), one of the most convincing ways to demonstrate hemisphere disconnection is to ask the patient to feel with one hand and then name various small, common objects such as a button, coin, safety pin, paper clip, pencil

stub, rubber band, or key. Vision must be excluded. A blindfold is notoriously unreliable; it is better to hold the patient's eyelids closed, put a pillowcase over the patient's head, or use an opaque screen. The patient with a hemisphere disconnection is generally unable to name or describe an object in the left hand although he readily names objects in the right hand. Sometimes the patient will give a vague description of the object although unable to name it, but there is a contrast with the ability to name the object readily when it is placed into the right hand. To establish hemisphere disconnection, other causes of unilateral anomia, particularly aster-eognosis (or a gross sensory deficit) must be excluded. The most certain proof that the object has been identified is for the subject to retrieve it correctly from a collection of similar objects. Such a collection is most conveniently placed in a paper plate about 12 to 15 cm in diameter, around which the subject can shuffle the objects with one hand while exploring for the test object. Even without evidence of correct retrieval, aster-eognosis can be reasonably excluded by observing the rapid, facile, and appropriate manipulation of an object despite an inability to name or verbally describe it. In testing for unilateral anomia, the examiner must be aware of strategies for circumventing the defect. For example, the patient may manipulate it to produce a characteristic noise, or the patient may identify a pipe or some other object by a characteristic smell and thus circumvent the inability of the left hemisphere to identify, by palpation alone, an object placed in the left hand. Syndrome after Corpus Callosotomy When the corpus callosum is sectioned in its entirety at one sitting the acute disconnection syndrome appears in almost all respects as it is seen after a complete cerebral commissurotomy including the anterior commissure and massa intermedia. This state after callosotomy includes transient mutism; hence this symptom when it follows a complete section is not reasonably ascribed to molestation of third ventricular structures (Fig. 5.3). Moreover, mutism does not necessarily follow a frontal commissurotomy that does include molestation of the third ventricular structures but spares the splenium. These two observations taken together suggest that post-commissurotomy mutism requires explanation on some other basis than in terms of third ventricle retraction.

Figure 5.3. Degree of mutism (marked, mild, or none) after operation on 15 patients with medically intractable epilepsy. Thirteen patients underwent complete commissurotomy and 2 patients (D.M. and N.F.) underwent frontal commis-surotomy (52). The symbols are the same as for Figure 5.1. Staying out of the third ventricle does avoid some problems, including transient diabetes in-sipidus. And there may be less likelihood of aseptic meningitis or other complications leading to hydrocephalus, a point that was often advanced by Wilson and colleagues (59, 110). Experiments with monkeys have shown that if the anterior commissure is left intact it can compensate for loss of the splenium with respect to interhemispheric transfer of certain kinds of visual information (57). But the anterior commissure cannot compensate completely for splenial loss in the human because hemialexia usually is present after splenial section. Indeed, most of the stabilized syndrome seen after a complete cerebral commissurotomy is also seen (i.e., has not been compensated) after a callosotomy sparing the anterior commissure (46, 74). This is perhaps not surprising because the anterior commissure is only Vioo the size of the corpus cal-losum. On the other hand, we can appreciate how significant it might be when we consider the wealth of information that is conveyed over one

optic nerve—the diameter of which is about the same as that of the anterior commissure. This question is complicated by the fact that the size of the anterior commissure is quite variable; a diameter difference of 3 or 4 times has been reported (112). The discrepancy between monkeys (transfer of learning by the anterior commissure) and humans (inability of the anterior commissure to compensate for callosotomy) may reflect differences between recently acquired memories as opposed to longstanding ones. Current evidence suggests that memory deficits can be expected after commissurotomy that includes the anterior commissure, even when the splenium is spared (30, 78, 114). Syndrome after Frontal Commissurotomy By "frontal commissurotomy" we refer (52) to section of the anterior two-thirds of the corpus callosum together with anterior commissurotomy; section of the anterior commissure was done under direct vision and the third ventricle was entered between the two fornices so the

Figure 5.4. Frontal commissurotomy: Sketched here are the structures sectioned during a frontal commissurotomy. This includes most of the corpus callosum (sparing the splenium), the ventral hippocampal commissure (between the forn-ices), and the anterior commissure. section also included the ventral hippocampal commissure (Fig. 5.4). The same operation was used on a few occasions by Wilson et al. (110). When the splenium is spared, as it is in the case of frontal commissurotomy, very little of either the acute or the chronic disconnection syndrome is seen. In particular, mutism did not occur in our four operations for intractable seizures with bitemporal foci. Because these patients had retraction within the third ventricle to allow section of the anterior commissure under direct vision, the mutism in cases with more extensive section (i.e., complete commissurotomy) was not likely of third ventricular origin. Moreover, all of these four had retraction of the medial aspect of the right frontal lobe, comparable with that in the complete cases; retraction on the supplementary motor cortex does not therefore seem to be a sufficient explanation for the mutism after callosotomy, an issue discussed in more detail under "Postcallosotomy Mutism." In the long term, Preilowski could show some deficits of motor coordination in two of our pa-

tients with frontal commissurotomy (85). But exhaustive testing for the usual disconnection deficits (as described under "Chronic, Stabilized Syndrome of Hemisphere Disconnection"), has shown that, with retention of the splenium, such deficits should not be expected (17, 47, 52, 55). Memory deficits have been observed after complete commissurotomy and also in lesser degree after frontal commissurotomy sparing the splenium. This is probably in part attributable to deficits in combining linguistic representations with visual or spatial images. We need further research using a wide variety of memory tests before and after operation, particularly with prolonged follow-up. Syndrome after Callosotomy Sparing the Splenium Section of the corpus callosum that spares the splenium (as well as the anterior commissure) but is otherwise complete can be readily accomplished via an exposure anterior to the rolandic

Figure 5.5. Anterior callosotomy. This sketch shows how anterior callosotomy differs from frontal commissurotomy, which is illustrated in Figure 5.4. There are two principal differences: the anterior commissure is spared, and third ventricle is not entered by separating the fornices. bridging veins (Fig. 5.5). This procedure, which avoids entry into the third ventricle, has come to be the most popular version of commissural section for epilepsy. After this operation one does not observe most of the acute disconnection syndrome. This smoother postoperative course is not surprising because the acute disconnection syndrome does not usually occur after frontal commissurotomy, which is essentially the same procedure plus section of the anterior commissure and massa intermedia. Absence of significant mutism with callosotomy sparing the splenium has been reported by Rayport et al. (87) as well as in personal communications to myself from M. Rayport, M. Apuzzo, J. Vries, J. Wada, and G. Yasargil. On the other hand, a recent abstract by Ross et al. (91) reported transient (1 - to 10-day) mutism with this procedure. This difference could depend upon the amount of retraction, particularly if there has been retraction of the left hemisphere below the falx as well as the usual right hemi-

sphere retraction. Or it might reflect differences in the criteria of the observers. Perhaps most important are differences in the patients; even a lesser callosal section may be followed by mutism when it is associated with a concurrent thalamic lesion. Midcallosal Section The most common variant of callosal section is an incision through the trunk sparing not only the splenium but also most of the genu as well. Such an incision affords ready access to both foramina of Monro with practically no obligatory physiological cost as presently assessable (3, 4, 52, 72, 84, 94, 111). With some possible exceptions (occasional brief mutism, inconstant auditory effects, tactile transfer deficits) any impairments after callosal trunk section are typically of the neighborhood type; for example, the language deficits allegedly secondary to callosal trunk section have been reported in only one

case, one in which the callosum was approached in a right-hander by retraction of the left hemisphere (36). Mutism in Special Cases Mutism can be expected, at least for a time, if partial callosal section is done in a right-hander to remove or to biopsy a right thalamic lesion; this conclusion is based on three personal cases and two cases of others on which I was consulted. Mutism for several weeks followed callosal trunk incision for the removal of a colloid cyst in a case in which right thalamic injury had occurred during a previous emergency ventriculostomy (personal communication from Dr. Paul LaPrade). But significant mutism did not appear after callosal trunk section in five other personal cases including a small left thalamic arteriovenous malformation (AVM) (not resected) and two small AVMs of the septum (with resection in both cases). As mentioned under "Syndrome after Frontal Commissurotomy," no mutism appeared after frontal commissurotomy in four of our epilepsy patients and no mutism was seen in two personal tumor patients (both still being followed after 10 and 13 years) with quite extensive mid-callosal section for a third ventricle craniopharyngioma and a third ventricle glioma (84). Inconstant Auditory Effects As pointed out under "Auditory Suppression," a naturally occurring lesion near the callosal trunk can result in suppression of one ear when tested dichotically. Hence, one might except similar changes after section of the callosal trunk. A few changes have sometimes been detected (personal communications from E. Zaidel and from E. Teng); this requires further investigation. In any case, the auditory effects are unlikely to be of clinical importance, although one can anticipate certain patients (for example, a symphony conductor) in whom slight auditory alterations might be quite important. Effects on Tactile Interaction? The somesthetic nontransfer described under "Somesthetic Effects" has not been detected in most cases of midcallosal section. But if the task is made more difficult, so that normal individuals commonly make mistakes, some deficits (as compared with normal subjects) have been elicited (7, 64). Further studies along this line will no doubt be forthcoming; they are clearly needed and the growing popularity of this approach will afford frequent opportunities for both pre- and postoperative testing.

Posterior Callosal (Splenial) Section Whereas more anterior section of the corpus callosum entails minimal long term physiological cost, section of the splenium (Fig. 5.6) commonly causes a left hemialexia (33, 62, 73, 104, 108). This inability to read in the left visual half-field may be accompanied by so-called color anomia, an inability to give the names of colors presented to the patient's view although the colors can be matched and the patient can give (speak or write) the color names of objects, for example, "yellow" when asked the color of a banana (49). When the hemialexia and associated deficits are most severe they can be mistaken for a left homonymous hemianopsia. Although use of a tachistoscope is more precise, it is not necessary because the deficit has often been demonstrated without such equipment. In a recent review (101), hemialexia is considered of two types: an inability to match written words with objects and an inability to read aloud written words or letters. Both of these are said to be mimicked by the condition of visual hemineglect. The foregoing hemideficits may not prove totally disabling for most persons, although they can be so in individuals whose occupations involve seeing large groups of symbols more or less simultaneously, such as a symphonic score. Where the splenial disconnection can be quite crippling for almost any literate person is when the callosal lesion is combined with a left occipital lesion or indeed any lesion causing a right hemianopsia. Such individuals typically have al-exia without agraphia; that is, they can write but are unable to read, even what they have just written correctly to dictation. This remarkable dissociation of reading from writing has been known for nearly a century (6, 35). One explanation is as follows: When the patient has a right homonymous hemianopsia resulting from a left occipital lobe lesion, nothing can be seen, much less read, in the right half-field. Hence, visual information can reach the left hemisphere language zone only from the left half-field via the right occipital cortex and transfer via the splenium. If another (pr confluent) splenial lesion has disconnected the right occipital cortex from the left hemisphere, the left hemisphere retains competence to write to dictation but no longer has access to information arriving in the right occipital lobe from the left visual half-field. As its proponents have recognized, there are some difficulties with this explanation of alexia without agraphia. These patients can often name objects or pictures of objects visualized in the left half-field, showing that information can reach the language zone from the left half-field. More-

Figure 5.6. Posterior callosotomy: In this operation, in addition to the splenium, part of the dorsal hippocampal commissure has been divided as it is usually adherent to the undersurface of the corpus callosum. over, alexia without agraphia can occur without an accompanying loss of the right visual half-field (54, 61). And alexia without agraphia can occur in cases with the splenium largely intact. Reading seems to be a multistage process that can be disturbed in a variety of ways (54, 60, 68). One can readily see that if splenial section were accompanied by left occipital damage the patient might be unable to read altogether and not just be hemialexic. Hence, retraction of the left occipital pole is to be avoided in approaching the splenium, although there may be situations, such as an AVM in the medial aspect of the left occipital lobe, where it seems necessary. In one such personal case, the patient was "cured" of her AVM and was still able to read albeit with difficulty, but she lost entirely her ability, said to have been phenomenal, to sight read music for the piano. Approaching the splenium from the left may be preferable in the rare case where a left-hander has been shown to be a right hemisphere speaker, preferably by use of the Wada carotid amobarbital technique.

Postcallosotomy Mutism After complete section of the cerebral commissures, there was in almost every case a postoperative period of mutism, of varying duration, during which speech was absent or extremely sparse, although comprehension and writing were retained. The patient did not talk even when quite cooperative and having some ability to write. This emissive deficit lasted, in milder form at least, in 11 righthanded patients for nearly 3 months, 3 days, 20 years, 4 weeks, 4 months, 2 days, 4 weeks, 2 weeks, 1 week, 4 weeks, and 4 weeks (see Fig. 5.3). In 1 left-hander (P.D.) mutism lasted 8 months. In all cases, there was little if any paraphasia; when ability to talk returned there was no nominal amnesia; in some cases there was a definite lack of bodily spontaneity and motor initiative for a time, but only partially correlated in duration with the loss of speech. As the mutism subsided, there was a stage of partial recovery that usually included hoarseness or whispering, but without para-

phasia, anomia, or novel semantic or syntactic errors, except for 1 righthander (C.C.) operated from the left. After complete cerebral commissurotomy there was almost always at least some degree of mutism. The notable exception in our series (see Fig. 5.3) was Patient M.K., whose convulsive disorder, beginning on the left side, was associated with a right ventricular dilatation as well as a left footdrop present since childhood. Compensation as a result of this right cerebral atrophic change may have contributed to the absence of postoperative mutism. On the other hand, this was our only case in which a very large bone plate (nearly a hemicraniotomy) was left out and replaced later to minimize pressure effects during postoperative swelling. Clearing of the mutism occurred in all cases but one. This patient (A.M.) continues, even years later, to speak very little—and when he does speak, his speech is poorly phonated and badly articulated. He had severe brain damage before the operation and we believe that he was, to an unusual degree, dependent upon a compensatory function of his corpus callosum for artic-ulatory control. In this case, the prolonged speechlessness was almost certainly not attributable to callosal section alone. This case further illustrates the previously emphasized point that postcommissurotomy deficits depend in large part upon the nature and amount of preexisting, extracallosal damage (11, 52, 100). Postcommissurotomy mutism was originally considered a simple neighborhood sign, a partial akinetic mutism from retraction affecting the anterior end of the third ventricle (29, 90) during section of the anterior commissure. However, a number of patients have now been seen with similar retraction, but a spared splenium, who did not become mute, including N.F. and D.M. (shown in Fig. 5.3). In contrast to the cases of complete commissurotomy, in these (and two other patients, D.B. and B.K.) all of the same structures (including the massa intermedia and the anterior commissure) were severed except the splenium and there was no immediate postoperative mutism (52). Such cases afford evidence not only against a third ventricle origin for the mutism, but also against a right supplementary cortex origin. On the other hand, Ross et al. (91) reported mutism after anterior callosal section but not after posterior callosal section. How to reconcile this apparent conflict is one task of this section, as well as for the future. After complete callosotomy in two stages, Ray-

port et al. (87) observed in three of eight cases a marked decrease in spontaneous speech unaccompanied by paraphasia or comprehension deficit or inability to sing. The authors suggested that, in these most affected cases, mixed hand dominance may have been an important consideration. Of particular importance was the absence of any language or speech problem after the first stage (rostrum, genu, and most of the trunk). The mutism appeared only after the second stage (splenium and remainder of the trunk). This result does not totally eliminate retraction on supplementary motor cortex as a partial contributory cause, but edema probably did not play a role because the second stage followed the first by 2 months, 36 months, and 19 days. Rayport and colleagues have suggested that mutism could result from an interhemispheric conflict (see also Ref. 11). But they emphasize more the aspect of mixed dominance, which might indicate a greater role than usual of the corpus callosum in the production of speech, as was probably the situation in Patient A.M. (Fig. 5.3) and possibly the patient reported by Sussman et al. (102). The role of the corpus callosum in speech may be much greater in certain patients. We can probably be reassured with respect to the risk of severe postoperative mutism by a favorable response to carotid amobarbital testing. This test can give ambiguous results, but it can, we believe, be used to exclude a critical dependence upon the commissures. That is, if the patient continues to speak intelligibly when the hemisphere minor for speech is narcotized, then disconnection of this minor hemisphere will probably not deprive the patient of an essential resource with respect to speech production. Although severe, long term mutism may occur only when callosal section is done in certain patients who are peculiarly dependent for speech on these pathways, there remains to be explained the mutism lasting several weeks in callosotomy patients without either widespread damage or anomalous lateralization from other causes. A diaschisis secondary to deafferentation of speech areas accounts for the deficits in terms of left hemisphere dysfunction even when the left hemisphere is unmolested. This interpretation is consistent with the appearance of a right Babinski sign (a sign of left hemisphere malfunction) and it may help to explain why our patients (and those of Rayport) whose spleniums were spared had no mutism. This means that the diaschisis of the left hemisphere (from a complete section) must affect the speech "centers" or "circuits" more than it affects writing "centers"

or "circuits." This implies, in turn, that the writing function is more robust or resistant in some sense than that for speech, in spite of having developed later in both phylogenesis and ontogenesis. A more speculative possibility is that speech usually requires interhemispheric integration for control of the larynx and other midline structures in the following sense: One can suppose that left hemisphere speech ordinarily includes a corollary discharge (97) to the other hemisphere. When the commissures are completely severed a downstream interhemispheric conflict occurs at the level of the motor nuclei for the larynx, which results in dysphonia. This explanation is consistent with most of the evidence. In particular, it fits the case of one patient (M.K.) who had no mutism at all—this was the patient with right hemisphere atrophy from childhood. It also fits the report of Sussman et al. (102) that their patient uttered several complex sentences shortly after operation although he was otherwise speechless for 16 months. Those concerned with the normal physiology of speech will be more interested in the temporary (several weeks to months) mutism after splenial section (either as part of a total calloso-tomy or as a second stage) than the longlasting mutism (many months to years) of those whose anomalous laterality is associated with longstanding cortical lesions. But for the surgeon using a transcallosal approach to the third ventricle, neither of these is apt to be as important as transient mutism (several days to weeks) after section of callosal segments (such as the trunk) well anterior to the splenium. The mutism reported by Ross et al. (91) lasted 1 to 10 days; hence, it is more relevant to the problems of third ventricle access than the severe, prolonged type of mutism. We have already alluded to the notion that the lack of speech could be considered a mutism of the type often associated with akinesia of either cingulate or subfron-tal origin. This suggests that we think of the postcommissurotomy state as a sort of "forme fruste" of akinetic mutism, in which the mutism is much more evident than is the akinesia. The occurrence of mutism in callosotomy cases without entry into the third ventricle could be explained as the result of subf rontal retraction during section of the rostrum. Arguing against this is the paucity of postoperative mutism in four of our patients (D.M., N.F., D.B., and B.K.) whose splenium was spared but who underwent section of both the rostrum and the anterior commissure under direct vision. The cause for transient mutism may be differ-

ent from one case to the next, and in any given case there may be more than one factor operative, (a) The commissural fibers may have been acting in a compensatory fashion because of previous brain injury, (b) Speech may require some absolute number of relevant brain connections. In a marginal case the callosotomy may be sufficient to bring the number of connections below the minimum required, (c) The callosotomy may produce a diaschisis in the speaking hemisphere, from which it recovers only slowly, (d) The cal-losum may carry a corollary discharge for speech output whose sudden loss results in downstream interhemispheric conflict, (e) Temporary circul-tory alterations (particularly of the internal cerebral veins) may transiently derange the more sensitive functions of the basal ganglia. (f) Damage to one or the other fornix may be contributory, (g) Trauma to the anterior third ventricle during division of the anterior commissure or during division of the rostrum from in front may be relevant in some cases, (h) Supplementary motor cortex dysfunction may be contributory in some cases, (i) Decussating fibers (for example, from medial frontal cortex to opposite basal ganglia) may be relevant so that their interruption plus medial frontal cortex edema combines to produce mutism when associated with a thala-mic or striatal lesion. Concluding Remarks The multifactorial causation of postcalloso-tomy mutism affords a measure of our ignorance, representing as it does the difficulties in predicting the likelihood of this usually transient but nonetheless distressing malfunction. On the other hand, the very existence of this riddle affords an opportunity for someone to unravel it. Likewise, the lack of replicable signs or symptoms from section of the anterior two-thirds of the corpus callosum represents a territory still unknown and awaiting exploration. That the transcallosal approach to the third ventricle is largely without obligatory physiological cost is both a boon to the operator and a challenge to the scientist. Neurosurgeons have historically been both therapeutic and investigative, and we can expect that continued exploitation of the transcallosal approach will benefit our patients both at the time of treatment and by augmenting our understanding of what Bremer (26) once called, "the highest and most elaborate activities of the brain." Acknowledgment I am grateful for library assistance from S. Zeind and staff, including P. Logan, C. DeCicco, L. Walden,

and V. Caullay of the Huntington Memorial Hospital, and for word processing by Sally Johnstone. References 1 . Akelaitis AJ: Studies on the corpus callosum: IV. Diagnostic dyspraxia in epileptics following par tial and complete section of the corpus callosum. Am J Psychiatry 101:594-599, 1944-1945. 2. Alpers BJ, Grant FC: The clinical syndromes of the corpus callosum. Arch Neurol Psychiatry 25:67-86, 1931. 3. Antunes JL, Louis KM, Ganti SR: Colloid cysts of the third ventricle. Neurosurgery 7:450-455, 1980. 4. Apuzzo MLJ, Chikovani OK, Gott PS, Teng EL, Zee CS, Giannotta SL, Weiss MH: Transcallosal, interfornicial approaches for lesions affect ing the third ventricle: Surgical considerations and consequences. Neurosurgery 10:547-554, 1982. 5. Benes V: Sequelae of transcallosal surgery. Childs Brain 9:69-72, 1982. 6. Benson DF: Alexia. In Vinken PJ, Bruyn CW (eds): Handbook of Clinical Neurology. Amster dam, Elsevier-Biomedical, 1985, revised ed vol 1. 7. Bentin S, Sahar A, Moscovitch M: Intermanual information transfer in patients with lesions in the trunk of the corpus callosum. Neuropsychology 22:601 -611, 1984. 8. Berlucchi G: Anatomical and physiological as pects of visual functions of corpus callosum. Brain Res 37:371 -392, 1972. 9. Bogen JE: The other side of the brain: I. Dysgraphia and dyscopia following cerebral commissurotomy. Bull Los Angeles Neurol Soc 34:73105, 1969. 10. Bogen JE: Hemispherectomy and the placing re actions in cats. In Kinsbourne M, Smith WL (eds): Hemispheric Disconnection and Cerebral Function. Springfield, IL, Charles С Thomas, 1974. 1 1 . Bogen JE: Linguistic function in the short term following cerebral commissurotomy. In AvakianWhitaker H, Whitaker HA (eds): Studies in Neurolinguistics, vol 2, New York, Academic Press, 1976, pp 193-224. 12. Bogen JE: The callosal syndrome. In Heilman KM, Valenstein E (eds): Clinical Neuropsychology. New York, Oxford University Press, 1978, pp 308-359. 13. Bogen JE: Concluding overview. In Reeves A (ed): Epilepsy and the Corpus Callosum. New York, Plenum, 1985. 14. Bogen JE: Split-brain syndromes. In Handbook of Clinical Neurology, vol 45, ch 9, pp 1-8. Amsterdam, Elsevier, 1985. 15. Bogen JE: The stabilized syndrome of hemi sphere disconnection. In Benson DF, Zaidel E (eds): The Dual Brain: Hemispheric Speciali zation in the Human. New York, Guilford Press, 1985. 16. Bogen JE, Bogen GM: The other side of the brain: III. The corpus callosum and creativity. Bull Los Angeles Neurol Soc 34:191-220, 1969. 17. Bogen JE, Campbell A, Thompson A: Long-term persistence of unilateral anomia following com-

18. 19.

20.

21. 22. 23.

24.

25. 26. 27. 28. 29. 30.

31.

32. 33.

34. 35.

plete cerebral commissurotomy. Proc Soc Neu-rosci 7:945, 1981. Bogen JE, Fisher D, Vogel PJ: Cerebral commis surotomy: A second case report. JAMA 194:1328-1329, 1965. Bogen JE, Gazzaniga MS: Cerebral commissur otomy in man: Minor hemisphere dominance for certain visuospatial functions. J Neurosurg 23:394-399, 1965. Bogen JE, Sperry RW, Vogel PJ: Commissural section and the propagation of seizures. In Jas per HH, Ward AA, Pope A (eds): Basic Mecha nisms of the Epilepsies. Boston, Little, Brown, 1969. Bogen JE, Vogel PJ: Cerebral commissurotomy in man. Bull Los Angeles Neurol Soc 27:169172, 1962. Bogen JE, Vogel PJ: Treatment of generalized seizures by cerebral commissurotomy. Surg Forum 14:431-433, 1963. Bogen JE, Vogel PJ: Neurologic status in the long term following cerebral commissurotomy. In Michel F, Schott В (eds): Les syndromes de Dis connexion Calleuse chez l,Нотте. Lyon, Hopital Neurologique, 1975, pp 227-251. Botez MI, Bogen JE: The grasp reflex of the foot and related phenomena in the absence of other reflex abnormalities following cerebral commis surotomy. Acta Neurol Scand 54:453-463, 1976. Bradshaw JL, Nettleton NC: Human Cerebral Asymmetry. Englewood Cliffs, NJ, Prentice-Hall, 1983. Bremer F: Physiology of the corpus callosum. Res Publ Assoc Res Nerv Ment Dis 36:424-428, 1956. Bremer F, Brihaye J, Andre-Balisaux G: Physio logie et pathologie du corps calleux. Arch Suisses Neurol Psychiat 78:31-87, 1956. Brion S, Jedynak CP: Les Troubles du Transfert Interhemispherique. Paris, Masson, 1975. Cairns HR: Disturbances of consciousness with lesions of the brainstem and diencephalon. Brain 75:109-146, 1952. Campbell AL, Bogen JE, Smith A: Disorganiza tion and reorganization of cognitive and sensorimotor functions in cerebral commissurotomy: Compensatory roles of the forebrain commis sures. Brain 104:493-511, 1981. Cobben A, Seron X, Gillet J, Bonnal J: Absence de signe de deconnexion lors d'un examen neuropsychologique differe dans quatre cas de lesions callosales anterieures et medianes partielles d'origine vasculaire et neurochirurgicale. Acta Neurol Belg 78:207-216, 1978. Damasio H, Damasio AR: "Paradoxic" ear extinc tion in dichotic listening: Possible anatomic sig nificance. Neurology (NY) 29:644-653, 1979. Damasio AR, Chui HC, Corbett J, Kassel N: Pos terior callosal section in a non-epileptic patient. J Neurol Neurosurg Psychiatry 43:351-356, 1980. Dandy WE: Operative experience in cases of pin eal tumor. Arch Surg 33:19-46, 1936. Dejerine J: Contribution a l etude anatomo-pathologique et clinique des differentes varietes de cecite verbale. С R Seances Memoires Soc Biol

36.

37. 38.

39.

40. 41.

42. 43.

44.

45. 46. 47.

48. 49. 50. 51.

52.

53. 54.

44 (vol 4 of Ser 9) (Second section-Memoires):6190, 1892. Dimond SJ, Scammell RE, Brouwers EYM, Weeks R: Functions of the centre section (trunk) of the corpus callosum in man. Brain 100:543562, 1977. Doty RW, Negrao N: Forebrain commissures and vision. In Jung R (ed): Handbook of Sensory Physiology VII/3. Berlin, Springer-Verlag, 1972. Doty RW, Overman WH: Mnemonic role of forebrain commissures in macaques. In Harnad S, et al (eds): Lateralization in the Nervous Sys tem. New York, Academic Press, 1977. Downer JLC: Changes in visual gnostic func tions and emotional behavior following unilat eral temporal pole damage in the split-brain monkey. Nature 191:50-51, 1961. Efron R, Bogen JE, Yund EW: Perception of dichotic chords by normal and commissurotomized human subjects. Cortex 13:137-149, 1977. Escourolle R, Hauw JJ, Gray F, Henin D: Aspects neuropathologiques des lesions du corps calleux. In Michel F, Schott В (eds):Les Syndromes de Disconnexion Calleuse Chez I'Homme. Lyon, Hopital Neurologique, 1975. Ethelberg S: Changes in circulation through the anterior cerebral artery. Ada Psychiatr Neurol Suppl 75:3-211, 1951. Ferguson SM: Some neuropsychiatric observa tions on behavioral consequences of corpus cal losum section. In Reeves A (ed): Epilepsy and the Corpus Callosum. New York, Plenum, 1985. Galin D, Johnstone J, Nakell L, Herron J: Devel opment of the capacity for tactile information transfer between hemispheres in normal chil dren. Science 204:1330-1332, 1979. Gazzaniga MS, Freedman H: Observations on visual processes after posterior callosal section. Neurology (NY) 23:1126-1130, 1973. Gazzaniga MS, LeDoux JE: The Integrated Mind. New York, Plenum, 1978. Gazzaniga MS, Risse GL, Springer SP, Clark E, Wilson DH: Psychologic and neurologic conse quences of partial and complete cerebral commissurotomy. Neurology (NY) 25:10-15, 1975. Geschwind, N: Disconnexion syndromes in ani mals and man. Brain 88:237-294, 585-644, 1965. Geschwind N, Fusillo M: Color-naming defects in association with alexia. Arch Neurol 15:137146, 1966. Gordon HW: Verbal and Nonverbal Cerebral Processing in Manjor Audition. Thesis, Califor nia Institute of Technology, 1973. Gordon HW: The neurobiological basis of hemisphericity. In Trevarthen С (ed): Brain Circuits and Theories of Mind. Cambridge, Cambridge University Press, 1986. Gordon HW, Bogen JE, Sperry RW: Absence of deconnexion syndrome in two patients with par tial section of the neocommissures. Brain 94:327-336, 1971. Gott PS, Saul RE: Agenesis of the corpus cal losum: Limits of functional compensation. Neurology(NY) 28:1272-1279, 1978. Greenblatt SH: Neurosurgery and the anatomy of reading: A practical review. Neurosurgery 1:6-15, 1977.

55. Greenblatt SH, Saunders RL, Culver CM, Bogdanowicz W: Normal interhemispheric transfer with incomplete section of the splenium. Arch Neurol 37:567-571, 1980. 56. Haaland KY, Delaney HD: Motor deficits after left or right hemisphere damage due to stroke or tumor. Neuropsychologia 19:17-27, 1981. 57. Hamilton CR: Mechanisms of interocular equiv alence. In Ingle D, Goodale M, Mansfield R (eds): Advances in the Analysis of Visual Behavior. Cambridge, MA, MIT Press, 1982, pp 693-717. 58. Hampson E, Kimura D: Hand movement asym metries during verbal and nonverbal tasks. Can J Psychol 38:102-125, 1984. 59. Harbaugh RE, Wilson DH, Reeves AG, Gazzaniga MS: Forebrain commissurotomy for epilepsy: Re view of 20 consecutive cases. Acta Neurochir (Wien) 68:263-275, 1983. 60. Hecaen H, Kremin H: Neurolinguistic research on reading disorders resulting from left hemi sphere lesions: Aphasic and pure alexias. In Whitaker H, Whitaker HA (eds): Studies in Neu rolinguistic s. New York, Academic Press, 1976, pp 861-872. 61. Henderson VW, Friedman RB, Teng EL, Weiner JM: Left hemisphere pathways in reading: Infer ences from pure alexia without hemianopia. Neurol (NY) 35:962-968, 1985. 62. Iwata M, Sugishita M, Toyokura Y, Yamada R, Yoshioka M: Etude sur le syndrome de discon nexion visuo-linguale apres la transection du splenium du corps calleux. J Neurol Sci 23:421432, 1974. 63. Jason GW: Hemispheric asymmetries in motor function: I. Left-hemisphere specialization for memory but not performance. Neuropsychologie 21:35-45, 1983. 64. Jeeves MA: Some limits to interhemispheric in tegration in cases of callosal agenesis and partial commissurotomy. In Russell IS, von Hof MW, Berlucchi G (eds): Structure and Function of Cerebral Commissures. Baltimore, University Park Press, 1979, ch 37. 65. Kimura D: Functional asymmetry of the brain in dichotic listening. Cortex 3:163-178, 1967. 66. Kimura D, Archibald Y: Motor functions of the left hemisphere. Brain 97:337-350, 1974. 67. Kumar S: Short-term memory for a non-verbal tactual task after cerebral commissurotomy. Cortex 13:55-61, 1977. 68. Landis T, Regard M, Serrat A: Iconic reading in a case of alexia without agraphia caused by brain tumor: A tachistoscopic study. Brain Lang 11:45-53, 1980. 69. Levy J: Cerebral asymmetries as manifested in split-brain man. In Kinsbourne M, Smith WL (eds): Hemispheric Disconnection and Cerebral Function. Springfield, IL, Thomas, 1974. 70. Levy-Valensi J: Le Corps Calleux (Paris Theses 448). Paris, G Steinheil, 1910. 71. Liepmann H, Mass O: Fall von linksseitiger Agraphie und Apraxie bei rechtsseitiger Lahmung. J Psychol Neurol 10:214-227, 1908. 72. Long DM, Leibrock L: The transcallosal ap proach to the anterior ventricular system and its application in the therapy of cran/opharyngioma. Clin Neurosurg 27:160-168, 1980. 73. Maspes PE: Le syndrome experimental chez

74.

75.

76.

77.

78.

79.

80. 81. 82. 83. 84.

85.

86.

87.

88. 89.

90.

1'homme de la section du splenium du corps calleux: Alexie visuelle pure hemianopsique. Rev Neural 80:100-113, 1948. McKeever WF, Sullivan KF, Ferguson SM, Rayport M: Typical cerebral hemisphere disconnec tion deficits following corpus callosum section despite sparing of the anterior commissure. Neu ropsychologia 19:745-755, 1981. McKeever WF, Sullivan KF, Ferguson SM, Rayport M: Right hemisphere speech development in the anterior commissure-spared commissurotomy patient: A second case. Clin Neuropsychol4:l7-22, 1982. Michel F, Peronnet F: Extinction gauche au test dichotique: Lesion hemispherique ou lesion commissurale? In Michel F, Schott В (eds): Les Syn dromes de Disconnexion Calleuse chez I'Homme. Lyon, Hopital Neurologique, 1975. Milner AD, Jeeves MA: A review of behavioral studies of agenesis of the corpus callosum. In Russell IS, van Hof MW, Berlucchi G (eds): Struc ture and Function of Cerebral Commissures. Baltimore, University Park Press, 1979. Milner B: Analysis of memory disorder after cer ebral commissurotomy. In Trevarthen С (ed): Es says in Honour of R. W. Sperry. Cambridge, Cambridge University Press, 1985. Milner B, Taylor L, Sperry RW: Lateralized sup pression of dichotically presented digits after commissural section in man. Science 161:184186, 1968. Mingazzini G: Der Balken. Berlin, Springer, 1922. Myers JJ: Right hemisphere language: Science or fiction? Am Psychol 39:315-320, 1984. Myers RE: Function of the corpus callosum in interocular transfer. Brain 79:358-363. Nebes RD: Hemispheric specialization in commissurotomized man. Psychol Bull 81:1-14, 1974. Ozgur MH, Johnson T, Smith A, Bogen JE: Transcallosal approach to third ventricle tumor: Case report. Bull Los Angeles Neurol Soc 42:5762, 1977. Preilowski BFB: Possible contribution of the an terior forebrain commissures to bilateral motor coordination. Neuropsychologie 10:267-277, 1972. Rayport M, Ferguson SM, Corrie WS: Outcomes and indications of corpus callosum section for intractable seizure control. Appl Neurophysiol 46:47-51, 1983. Rayport M, Corrie WS, Ferguson SM: Results of two-stage corpus callosum section for seizure control in clinically and electroencephalographically defined cases. In Reeves A (ed): Epilepsy and the Corpus Callosum. New York, Plenum, 1985. Reeves A (ed): Epilepsy and the Corpus Cal losum. New York, Plenum, 1985. Risse GL, LeDoux J, Springer SP, Wilson DH, Gazzaniga MS: The anterior commissure in man: Functional variation in a multisensory system. Neuropsychologia 16:23-31, 1978. Ross ED, Stewart RM: Akinetic mutism from hypothalamic damage: Successful treatment with dopamine agonists. Neurology (NY) 31:1435-1439, 1981.

91. Ross MK, Reeves AG, Roberts DW: Postcommissurotomy mutism. Ann Neurol 16:114, 1984. 92. Rovit RL, Gloor P, Rasmussen T: Sphenoidal electrodes in the electrographic study of patients with temporal lobe epilepsy. J Neurosurg 18:151-158, 1961. 93. Sidtis JJ, Volpe ВТ, Wilson DH, Rayport M, Gaz zaniga MS: Variability in right hemisphere lan guage function after callosal section: Evidence for a continuum of generative capacity. J Neurosci 1:323-331, 1981. 94. Shucart WA, Stein BM: Transcallosal approach to the anterior ventricular system. Neurosurgery 3:339-343, 1978. 95. Sparks R, Geschwind N: Dichotic listening in man after section of neocortical commisures. Cortex 4:3-16, 1968 96. Sparks R, Goodglass H, Nickel B: Ipsilateral ver sus contralateral extinction in dichotic listening from hemispheric lesions. Cortex 6:249-260, 1970. 97. Sperry RW: Neural basis of the spontaneous optokinetic response produced by visual inversion. J Comp Physiol Psychol 43:482-489, 1950. 98. Sperry RW: Cerebral organization and behavior. Science 133:1749-1757, 1961. 99. Sperry RW: Lateral specialization in the sur gically separated hemispheres, in Schmitt FO, Worden FG (eds): Neuroscience 3rd Study Prog. Cambridge, MIT Press, 1974. 100. Sperry RW, Gazzaniga MS, Bogen JE Interhemispheric relationships: The neocortical commis sures. Syndromes of hemisphere disconnection. Handbook Clin Neurol 4:273-290, 1969. 101. Sugishita M, Shinohara A, Shimoji T: Does a posterior lesion of the corpus callosum cause hemialexia? In Reeves A (ed): Epilepsy and the Corpus Callosum. New York, Plenum, 1985. 102. Sussman NM, Gur RC, Gur RE, O'Connor MJ: Mutism as a consequence of callosotomy. J Neu rosurg 59:514-519, 1983. 103. Teng EL: Dichotic ear difference is a poor index for the functional asymmetry between the cere bral hemispheres. Neuropsychologia 19:235240, 1981. 104. Trescher HH, Ford FR: Colloid cyst of the third ventricle: Report of a case. Operative removal with section of posterior half of the corpus cal losum. Arch Neurol Psychiatry 37:959-973, 1937. 105. Trevarthen C: Hemispheric specialization. In Handbook of Physiology—The Nervous Sys tem HI. Washington DC, American Society of Physiology, 1984, pp 1129-1190. 106. Unterharnscheidt F, Jalnik D, Gott H: Der balkenmangel. Monograph Ges Neurol Psychiat 128:1-232, 1968. 107. Van Gijn J: The Plantar Reflex. Rotterdam, Krips Repro-Meppel, 1977. 108. Wechsler AF: Transient left hemialexia. Neurol ogy (NY) 22:628-633, 1972. 109. Weinstein EA, Friedland RP: Hemi-Inattention and Hemisphere Specialization. New York, Raven Press, 1977. 110. Wilson DH, Reeves A, Gazzaniga M, Culver C: Cerebral commissurotomy for control of intract able seizures. Neurology (NY) 27:708-715, 1977.

1 1 1 . Winston KR, Cavazzuti V, Arkins T: Absence of neurological and behavioral abnormalities after anterior transcallosal operation for third ven tricular lesions. Neurosurgery 4:386-393, 1979. 112. Yamamoto I, Rhoton AL, Peace DA: Microsurgery of the third ventricle: Part 1. Neurosurgery 8:334-356, 1981. 113. Yakovlev PI, Lecours AR: The myelogenetic cycles of regional maturation of the brain. In Minkowski (ed): Regional Development of the Brain in Early Life. Edinburgh, Blackwell, 1967. 114. Zaidel D, Sperry RW: Memory impairment after commissurotomy in man. Brain 97:263-272, 1974. 115. Zaidel D, Sperry RW: Some long-term motor ef-

fects of cerebral commissurotomy in man. Neuropsychologia 15:193-204, 1977. 116. Zaidel E: Linguistic competence and related functions in the right hemisphere of man follow ing cerebral commissurotomy and hemispherectomy. Ph.D. thesis, California Institute of Technology. Dissertation Abstr Intern 34: 2350B (University Microfilms 73-26,481). 117. Zaidel E: Lexical organization in the right hemi sphere. In Buser P, Rougeul-Buser A (eds): Cer ebral Correlates of Conscious Experience. Am sterdam, Elsevier, 1978. 118. Zaidel E: Disconnection syndrome as a model for laterality effects in the normal brain. In Hellige J (ed): Cerebral Hemisphere Asymmetry: Meth ods, Theory and Application. New York, Praeger, 1983. ~

6 Pathological Correlates of Amnesia and the Anatomical Basis of Memory Antonio R. Damasio, M.D., Ph.D., and Gary W. Van Hoesen, Ph.D.

The early understanding of the structural basis of human memory derived from the anatomical study of two distinct groups of patients. The first comprised patients with Wernicke-Korsakoff's psychosis. The second was made up of patients with temporal lobectomies performed to alleviate seizures refractory to medical treatment. The current views on the anatomy of memory draw from a much larger base of data, including not only the study of such classical cases, but also recently described partial and global amnestic syndromes associated with focal lesions in (a) the basal forebrain, (b) the lateral and polar temporal cortices, and (c) the association cortices of separate sensory modalities. Additionally, the study of senile dementia of the Alzheimer type is contributing important understanding in this area. In this chapter we review these recent developments and indicate why they enlarge our working concept of the relation between brain structure and memory. We also review the anatomy of the system most frequently involved in memory disorders, the limbic system, and note some fundamental aspects of the neuropsychological testing of the amnesias. Amnesia Associated with Medial Temporal and Diencephalic Damage Decades of histopathological study of alcoholic patients with amnesia have revealed damage in diencephalic and mesencephalic regions. The consistent lesions in the hypothalamus, espe-

cially in mamillary bodies, and the dorsomedial nuclei of the thalamus have been related to the memory impairment. Considerable controversy has surrounded the issue of whether the mamillary bodies or the thalamic lesions are crucial in causing the amnesia (41). Our view is that, if the damage is bilateral, both loci can be associated with memory impairment. Quite commonly, this is not even an issue as both loci are involved. The reasonable conclusion is that these structures are part of a system of cortical and sub-cortical neural units necessary for learning, recall, and recognition (11, 86). Future and more detailed analysis of this type of data may well result in the inclusion, as part of this system, of additional structures in the high, paramidline brain stem. Nonetheless, the currently available correlations are quite unequivocal, so long as it is kept in mind that extensive neurochemical changes accompany this disorder and undoubtedly play a crucial role. The work on temporal lobectomies has also led to unequivocal correlations. Patients with extensive, unilateral, temporal excisions develop a partial amnesia that affects predominantly verbal or nonverbal material depending on whether the lobectomy is placed in the dominant or non-dominant hemisphere (45). The inclusion, in the excision, of parahippocampal cortex, hippocampus, and amygdala seems necessary for the appearance of amnesia (93). On the other hand, patients with bilateral damage of the medial temporal lobes encompassing the hippocampal formation, the overlying parahippocampal gyrus,

and part of the amygdala developed a major an-terograde learning defect (65). The best-studied subject with such a lesion, Patient H.M., was operated on in the 1950s and still gives no evidence of learning, at a conscious level, any material that he has come in contact with since then (15, 73). This anterograde amnesia covers material learned in any sensory modality. The exception is the learning of some visuomotor skills, which he has clearly mastered without being aware that learning has taken place (13). H.M. can neither recall nor recognize anterograde material of contextual and generic types. The terms generic or semantic used in association with memory denotes knowledge about the basic properties of a stimulus, such as class membership and function. Contextual or episodic memory denotes knowledge of a specific stimulus, its relationship to other stimuli in a given situation, and its relationship to the perceiver and may contain a temporal element. The concepts of episodic and semantic memory were proposed by Tulving (74). The largely equivalent terms contextual and generic are our own. This anterograde defect covers both verbal and nonverbal information. In contrast to this pervasive anterograde defect, H.M. has only a small retrograde memory impairment. The patient retains normal memory for persons, objects, and experiences up to the age of 16 years, when he started having seizures. That means that his contextual and generic memories are normal up to that age. For the period beyond the age of 16 years, there is a recently uncovered contextual memory impairment, noticeable when the patient is compared with controls (14). The defect is far less marked than the anterograde impairment. Both H.M. and the patients with alcoholic Kor-sakoff's syndrome have normal perception, language, and motor function and preserved intellectual skills that allow them, both in real situations and in laboratory testing, to exhibit a large range of intelligent behaviors (11, 15). Although some cognitive strategies are impaired, the disorder of recall, recognition, and learning dominates the clinical picture. The only other impairment of significance in these amnestic patients has to do with emotion and affect. Most of these patients have some disturbance of affect, emotional display, or motivation. But these disturbances are mild in comparison to the memory disorder (11, 15). Both H.M. and the patients with alcoholic Kor-sakoff's syndrome preserve their knowledge of skills and are able to acquire new visuomotor skills. Their "procedural" memory and "proce-

dural learning" are spared, i.e., their knowledge of "how" to act in some specific circumstances is preserved. This contrasts with their "declarative" memory and "declarative" learning, i.e., their knowledge of "that," which is impaired (12). We might take this qualification a bit further by saying that, in the retrograde compartment of their memory, only the contextual component of declarative knowledge is disturbed, whereas the generic component of declarative knowledge is intact. None of these patients has difficulty recognizing the myriad separate items that compose their universe in their essential, generic nature or the function of those components, i.e., they can recognize any human face as a face, any dog as a dog, any house or airplane as such, the beauty of a landscape as beauty indeed. Their disturbance pertains only to those memories in which a particular person, animal, object, quality, or experience must be placed in the context of the subject who was previously in contact with them, i.e., assigned a specific relation to the beholder and maybe even a temporal position. The hallmark of these amnestic disorders is the contrast between the vast store of intact generic knowledge and intellectual skills and the gross impairment of contextual knowledge (17). Recent Developments Amnesia after Medial and Lateral Temporal Lobe Damage Recent developments in the understanding of the anatomical basis of amnesia include the description of cases in which the profile of amnesia and the anatomical lesions correlated with it were considerably different. One of the most significant is the description of a global amnestic syndrome, similar to that seen in H.M., but with a far more pervasive retrograde amnesia encompassing all previously acquired contextual memory (18). In addition to the predictable bilateral medial temporal lobe lesions, that syndrome is associated with bilateral lesions in both lateral aspects of the temporal lobe including its pole, bilateral insular cortices, and bilateral lesions in the basal forebrain region. The cause is herpes simplex encephalitis. Further investigations are necessary to establish unequivocal correlations between some features of the syndrome, especially the overwhelming retrograde amnesia, and the different sets of paired lesions seen in such cases. But a reasonable working hypothesis is that damage to the polar and lateral temporal cortices and the insular cortex is the best corre-

late of the retrograde amnesia. Those are the areas conspicuously spared in amnestic patients who have a largely preserved retrograde memory, such as H.M. Amnesia after Basal Forebrain Damage Another important development is the description of an amnestic syndrome encompassing an-terograde and retrograde memory, milder than that reported with temporal lobe damage and related exclusively to contextual material, caused by unilateral or bilateral lesions of the basal forebrain area (18, 85). This is the area that includes the substantia innominata (which contains a large part of the nucleus basalis of Mey-nert), the diagonal band nuclei, the nucleus ac-cumbens and septal nuclei, and part of the for-nix. These lesions are caused by vascular damage in the territories supplied by the anterior communicating artery, anterior cerebral artery, and recurrent artery of Heubner. The immediate cause is rupture of aneurysms in one of these arteries. The exact pathological mechanism for this damage is varied— infarction due to ischemia in a vascular territory subject to sudden loss of blood flow, infarction due to embolism from the aneurysmal sac, or direct damage by hemorrhage from ruptured aneurysm. Not uncommonly, there is additional damage in one of the orbitofrontal regions (especially in gyrus rec-tus) and occasionally in the medial frontal lobe, especially in the subcallosal portion of the cin-gulate gyrus. The amnesia is more or less pure depending on how much additional damage of this type is present. A component of the basal forebrain, the nucleus basalis of Meynert, is the site of some changes in its cholinergic neurons in Alzheimer's disease (91) and these changes might be related, in part, to the amnesia seen in Alzheimer's patients. This area is of obvious special importance to neurosurgeons engaged in the treatment of aneurysms or in the management of third ventricle tumors. It is apparent that damage in this region, whether or not it involves the fornix, is likely to cause memory impairment. Because that impairment may be only partial, the potential for missing its presence or minimizing its importance is considerable. Such impairments may cast a major shadow on what might otherwise be an uneventful recovery from a successful operation. Amnesia in Alzheimer's Disease No less important has been the description of specific and preponderant pathological involvement of the entorhinal cortex and subiculum in

Alzheimer's disease (31). The entorhinal cortex is located along the anterior part of the parahip-pocampal gyrus. The subiculum is a component of the hippocampal formation. Both entorhinal and subicular cortices serve as staging areas for input to and output from the hippocampal formation. The net effect of such lesions is to disconnect the hippocampus from the cerebral cortex and from a variety of subcortical structures in the limbic system and diencephalon. This isolation is incompatible with normal function and must be no less devastating than surgical resection or vascular or encephalitic damage. In practice, after such lesions develop, the hippocampus will no longer be able to operate on the basis of the multimodal cortical inputs that normally reach it via the perforant pathway that originates in the superficial layers of the entorhinal cortex (32). As if that were not impairment enough, however, the lesions of Alzheimer's disease also block the exit of outputs at the level of the subiculum. These selective lesions are likely to be a major cause, perhaps the principal cause, of the contextual amnesia that hallmarks the early phase of Alzheimer's disease. Other Developments Other important new findings in the field of human amnesia include: (a) the realization that syndromes such as that of visual agnosia can be understood as disorders of visually triggered memory that impair the recognition, recall, and learning of certain types of visual stimuli (16, 20, 21); (b) the description of defects in learning related to unilateral lesions capable of compromising the processing of information conveyed from one sector of the body alone (i.e., a learning defect of material presented in one hand only (61, 62)); (c) the description of cases of amnesia with lesions of the medial thalamic nuclei caused by stroke or by direct injury (25, 70); and (d) the clarification of a question regarding the role of the temporal stem in the development of amnesia caused by medial temporal lobe damage. The hypothesis that the amnesia associated with medial temporal lesions might be related to section of the temporal stem rather than damage to the hippocampus proper has been posited (30). Recent experimental work in the monkey, specifically designed to test the temporal stem hypothesis, failed to support the claim (94). Furthermore, because the temporal stem carries some of the outputs from the hippocampus, it would be only natural that its damage might lead to amnesia. That certainly does not disqualify the hippocampus from a role in memory.

Anatomical Underpinnings of Memory Reflection on this vast body of classical and recent anatomical and cognitive findings permits the elaboration of a more comprehensive list of neural systems crucial to memory processing. It also permits some hypotheses about the way in which they integrate in a network. Finally, it is possible to venture a relatively specific functional role for some key components of the network. We start this section by reviewing the anatomy of the limbic system. Limbic System Anatomy The preceding material provides abundant evidence that many parts of the limbic system play a critical role in normal human memory, as well as in other behaviors that play a supportive role in this essential process. A review and update on limbic system anatomy is thus of special importance for neurosurgeons as they are often called to operate in its multiple areas or in its immediate vicinity. Cortical Structures of the Limbic System The cortical areas that form the limbic lobe are included in nearly all definitions of the limbic system. They include such major cerebral entities as the cingulate gyrus dorsally and the para-hippocampal gyrus ventrally. These are demarcated clearly in the human brain by the cingulate and collateral sulci, respectively. Bridging areas such as the subcallosal gyrus and the posterior orbital, anterior insular, temporal polar, retro-splenial, and retrocalcarine cortices link the cingulate and parahippocampal gyri. Together, these form a continuous ring around the corpus callosum, thalamus, and upper brain stem. In general, they are uninterrupted by major sulci. The cortical areas that form the limbic lobe differ widely in their cytoarchitecture and include Brodmann's areas 23, 24, 25, 27, 28, 29, 35, 36, and 38. These are not true isocortical areas. Instead, they fall into the categories of periallocortex and proisocortex (63). They are intermediate in structural complexity (Filimonoff's mesocortex (23)) between the less elaborate allo-cortices and the more elaborate isocortices. They are characterized by an incipient or absent layer IV, an accentuated layer II, and heavy, largely unsegregated, layers V and VI. Subcortical Structures of the Limbic System The subcortical structures of the limbic system are numerous and vary widely among authors. For reasons that will be discussed more fully, it

seems appropriate to include the amygdala, septum, anterior thalamus, habenula, and interpe-duncular nucleus. Basal forebrain areas such as the substantia innominata, and, particularly, the nucleus basalis of Meynert are relatively new additions to lists of subcortical limbic system structures. The latter is a widely scattered collection of large hyperchromatic neurons that extends from the septum to the midbrain in a position anterior and lateral to the hypothalamus and ventral to the basal ganglia. Particularly large clusters of these neurons are found beneath the anterior commissure and along its lateral excursion into the temporal lobe. Many of them are contained within the substantia innominata. They have widespread projections to all of the cerebral cortex (22, 35, 37, 42) and provide the major source of acetylcholine to it (44). Heavy input is directed toward the motor cortices. Interconnecting Pathways of the Limbic System The third and final constituent of the limbic system is a number of interconnecting pathways of different structures and magnitudes. Some of these link various parts of the limbic lobe. Others link the limbic lobe with subcortical limbic structures. Still others interconnect subcortical structures. A final group of pathways link the cortex of the limbic lobe and subcortical limbic structures with the hypothalamus and related structures such as the preoptic area. Few of these are entirely pure pathways. Most are dominated largely by one of these types of connections. A list of the major pathways would include the cingulum, the uncinate fasciculus, the fimbria-fornix, the stria medullaris, the stria terminalis, the ventroamygdalofugal pathway, the medial forebrain bundle, the mamillothalamic tract, the mamillotegmental tract, and the habenulointer-peduncular tract or fasciculus retroflexus of Meynert. Many additional connections of limbic system structures travel in wellknown association and commissural pathways of the cerebral hemisphere such as the internal, external, and extreme capsules and the corpus callosum and anterior commissure. (Fig. 6.1 provides a summary of limbic system anatomy.) Limbic System Efferent Projections A full review of all limbic system connections is beyond the scope of this chapter, so we will highlight only those findings that in recent years have altered the classical thinking on this topic. Historically, it has been the convention to view limbic system anatomy in the context of hypo-

Figure 6.1. These illustrations depict schematically some of the major neuroan-atomical components of the limbic system on views of the medial surface of the cerebral hemisphere. A. Limbic system landmarks and sulcal boundaries. B. The major cortical entities of the limbic system, the so-called limbic lobe. C. The subcortical structures of the limbic system. D. The approximate location of major limbic system pathways that interconnect its various components with each other and with the hypothalamus.

thalamic function (40). There has been good reason for this because major interconnecting lim-bic pathways like the fimbria-fornix and stria terminalis terminate on the neurons of various hypothalamic nuclei. The hypothalamus is an important effector center for the control and regulation of both the endocrine system and the autonomic nervous system. Although the structural basis for hypothalamoendocrine interactions, both humoral and vascular, is more generally understood, it has only been in recent years that the structural basis for hypothalamoautonomic interactions has been elucidated. For example, it is now clear that specific nuclei of the hypothalamus project to sympathetic and parasympathetic centers in both the brain stem and spinal cord (64). These are centered largely in the paraventricular, dorsomedial, and posterior nuclei. Moreover, there is now evidence that certain subcortical components of the limbic system itself, such as the central amygdaloid nucleus, project to brain stem autonomic centers as well (28, 29, 57). Such projections, as well as those mediated via the hypothalamus, validate the classical emphasis on limbichypothalamic interactions in the regulation of autonomic and endocrine effectors. There is now good evidence that subcortical limbic system structures can affect brain areas known to regulate somatic effectors as well. The findings entail direct amygdaloid projections to motor-related areas of the cerebral cortex (6, 38, 56) and to the striatum (36). Additionally, the nucleus basalis of Meynert projects powerfully to the motor and premotor cortices (42). The cortical areas of the limbic system also have extensive projections to key central components of the motor system. For example, the anterior part of the cingulate gyrus projects extensively to the premotor and supplementary motor cortices, and this part of the cingulate gyrus receives extensive input from the remainder of the limbic lobe and several subcortical parts of the limbic system (83). It is a focal point for limbic influence on the motor cortices. Moreover, there is evidence demonstrating that the cortices of the limbic lobe project also to subcortical motor structures such as the caudate nucleus and pu-tamen (7, 81, 88) and pons (87). In total, these constitute a major afferent source for motor-related parts of the cerebral hemisphere. Therefore, it now seems appropriate to view the limbic system in the context of all effectors, autonomic, endocrine, and somatic as well. Limbic system structures also have extensive and widespread projections to associative parts

of the cerebral cortex. These are mediated via limbic lobe and amygdaloid projections to the frontal, parietal, and temporal association cortices (5, 7, 43, 53, 56, 60, 84). Such projections, although largely elucidated in recent years, were a major element of Papez's deductions, when he suggested, nearly a half century ago, that the radiations of the cingulate gyrus project not only back into his circuit, but to the cerebral cortex as well, where he believed the "psychic coloring" of sensations took place (55). Unfortunately, this important and major element of Papez's thinking has received far less emphasis than those relating to his so-called circuit. Limbic System Afferent Connections As discussed previously, it seems clear that there are extensive limbic system efferents that affect endocrine, autonomic, and somatic effectors as well as associative areas of the cerebral hemisphere. The existence of such outputs raises the question of afferent input to limbic structures and makes the understanding of this topic critically important. New information regarding limbic system input has been slow to accrue because progress in this area has been linked solidly to the development of newer and more sensitive neuroanatomical tracing methods. A renewed interest in the organization of the connections of the cerebral cortex in general has also been decisive (34, 52, 67, 77). In higher mammals, limbic system structures receive a large part of their input directly from the cerebral cortex and especially from cortical areas designated as associative in function (76). Such connections are well illustrated by direct cortical projections to the amygdala (1, 27, 75, 82, 92) described in nonhuman primates. Neocortical projections converge on the lateral nucleus of the amygdala. These arise largely from the temporal cortex, but insular and frontal projections have been described as well (24, 46, 75). In general, these projections are highly organized in terms of topography. For example, the visual association cortex of the lateral temporal lobe projects to the dorsolateral part of the lateral nucleus, whereas the auditory association cortex projects to its more ventrolateral parts. Temporal polar cortical projections to the lateral nucleus terminate in the medial parts of the nucleus, as do insular and orbitofrontal projections. The basal amygdaloid nuclei receive extensive input from the cortex of the limbic lobe, including the cingulate, temporal polar, and perirhinal cortices as well as the subiculum (82). The hippocampal formation receives its major

cortical input from the entorhinal periallocortex. The sum of this projection is known as the per-forant pathway and it forms an overlooked, but major fiber system in the temporal lobe (71, 79). Input to the entorhinal cortices is derived from both subcortical and cortical structures. The former include the amygdala and septal area, as well as several midline thalamic nuclei (33). The latter, the cortical projections, are extensive. Major projections originate in the subicular cortex (60) and in the olfactory cortex and cortical amygdaloid nuclei (39). Collectively, these represent allocortical projections. Large periallocortical projections arise from the presubicular and retrosplenial cortices (68, 69). Proisocortical projections arise from the posterior parahippocam-pal, temporal polar, perirhinal, and posterior orbital cortices (78-80). Lastly, isocortical input arises from the cortex of the superior temporal gyrus (4) and the banks of the occipitotemporal sulcus (84). In total, these projections represent a large and diverse source of cortical input to the hippocampal formation. They bring modality-specific and multimodal cortical information to the structures of the hippocampal formation. It is plausible to believe that these sources of cortical input to the entorhinal cortex and hippocampal formation provide these structures with a rich variety of input relating to the sensory environment. For example, posterior parahippo-campal and parietooccipital projections would be expected to carry visual or visuosomatic information because these areas receive input from peristriate visual association cortices (69). Projections that arise from the superior temporal gyrus (66) would be expected to convey auditory input because these areas receive direct projections from cortical areas that surround the primary auditory cortex. Projections that relay through the insular cortex, amygdala, and perirhinal cortices to the entorhinal cortex and hippocampal formation would be expected to convey somatic and perhaps gustatory information. Olfactory input arises directly from the olfactory bulb as well as from the prepiriform cortex that forms the anterior part of the parahippocampal gyrus. Such connections would be expected to provide unimodal channels of influence from all sensory modalities, as well as bimodal and multimodal channels. Some areas of the cortex that send direct projections to entorhinal cortex are themselves multimodal because they receive cortico-cortical association projections from areas related to more than one modality. Moreover, many other areas of the limbic lobe receive direct input

themselves from multimodal and sensory-specific association cortices, and in turn project to the entorhinal cortices, creating the potential for nearly all combinations and permutations of simple and complex sensory inputs. Finally, the association cortices undoubtedly play a dual functional role because they are related, on the one hand, to the analysis and synthesis on sensory information (54) and, on the other, to the preservation of memories. Thus, the input to the entorhinal cortices is likely to be a composite, relating to both past and current sensory and cognitive experiences relative to all modalities. It should be apparent from this discussion that experimental studies have altered classical thinking to a considerable extent. It is no longer tenable to view the limbic system as a group of interconnected structures whose input arises largely from its own components and whose output affects only other components of the system. Indeed, extensive intralimbic system connections do exist and form the substance of many elements of classical neuroanatomy. In line with the early deductions of Papez and MacLean, limbic system structures do receive sensory input and, although there were early difficulties in demonstrating this with experimental methods, it is now recognized that these inputs are extensive and highly complex. Some of them arise directly from sensory association areas adjacent to the primary sensory areas, whereas others arise from multimodal association cortices that, themselves, receive input from two or more sensory association cortices. In some instances, certain limbic structures receive these categories of sensory association input directly, whereas in other instances they are relayed indirectly. There is still insufficient neurophysio-logical and functional data to characterize all of these fully. Nonetheless, given what is available, it is plausible to believe that they convey a rather sophisticated digest of the sensory environment that is linked often to the state of the organism in terms of relevance, motivation, and attention. In short, they arise from areas spread along the chain of intrahemispheric sensory stations where an external stimulus is analyzed elementally and then put together again in appropriate perceptual categories and context. In many instances these are multidimensional. One might conceptualize this as a unit in association cortex that responds only when criteria such as form, color, and orientation are combined appropriately and/or if linked with the appropriate drive. We do not know precisely what limbic system structures do with such information, but func-

tional and clinical studies point clearly to the fact that different areas subserve substantially different functions. For example, damage to the anterior cingulate cortex can lead to a pronounced amotivational and akinetic state, with withdrawal from social and interpersonal interactions and impoverished motor activity. Posterior orbitofrontal damage may alter social behavior but it seldom involves the same social withdrawal. It is characterized instead by disinhibi-tion and social and sexual inappropriateness. Motor activities as well may tend toward disin-hibition. Basal forebrain changes, for example, in the case of ruptured anterior communicating aneurysms, that damage the septum, nucleus basalis, and columns of the fornix, may lead to a distinct memory impairment, but disinhibition and personality disorders may be manifest as well. Temporal polar and amygdaloid pathology are linked solidly with emotional changes of several varieties as well as perceptual and sexual

dysfunction. Finally, hippocampal and parahip-pocampal lesions lead to few of the behavioral changes discussed previously, but cause, instead, irreversible alterations in memory, a global amnestic disorder markedly skewed in the direction of new learning. Thus, it is inappropriate to view the limbic system as having a single function, although an involvement with memory is clearly the most visible of its roles. Without a doubt certain of its structural elements are more important for some behaviors than others, but in general, multiple behaviors than others, but in general, multiple behavioral changes accompany nearly any focal limbic disorder. Additionally, it seems clear that limbic system structures act in consort with many other parts of the nonlimbic brain and have a decisive interaction with those areas subserving sensation, cognition, motor function, endocrine control and autonomic regulation. In short, it forms key elements of a multitude of

Figure 6.2. This is a photomicrograph of a Nissl-stained coronal section from the human brain at the level of the optic chiasm. The neuronal cell groups that form the nucleus basalis of Meynert and surrounding surface and subcortical structures are highlighted.

neural systems that involve many brain areas and which subserve many complex behaviors. Anatomical Concerns Relating to Surgery in the Vicinity of the Third Ventricle Access to operable abnormalities in or around the third ventricle requires, in most instances, a surgical approach that compromises the integrity of certain limbic system structures. Memory impairments are not an unusual postoperative complication andrepresent a matter of substantial concern during the immediate postsurgical management period or even longer afterward. The fimbriafornix system, a major interconnecting limbic system pathway, is frequently the neural structure of concern because it wraps nearly around a major portion of the diencephalon and the fornix has a long interstitial course through it enroute to termination sites in the preoptic and hypothalamic areas. The validity of the concern for fornix damage, in totality, might be challenged for several reasons. Foremost is the fact that there is little agreement in the literature as to whether fornix integrity is essential for normal memory. Indeed, it carries significant input to the hippocampal formation as well as significant output from nearly all parts of this collection of structures (72). However, significant other input and output pathways would remain intact so that disruption of the fornix would not totally disconnect the hippocampal formation. In fact, nearly all of its connections with the association cortices would be structurally patent. It seems more plausible to us to believe that fornix damage in consort with damage to other neural systems critical for normal memory are collectively the culprit. Basal forebrain nuclei, thalamic nuclei, and a major, often overlooked, subcortical limbic pathway could play a role. The first would constitute the nucleus basalis of Meynert that lies largely in the substantia innominata (see Fig. 6.2). It extends beneath the basal ganglia and pallidum from the midline of the hemisphere to the temporal lobe. Its largest component lies relatively close to the midline and only a few millimeters from the third ventricle. As discussed previously, it provides the major cholinergic input to the cortex. Fornix damage could disrupt cholinergic input to the hippocampal formation and, in combination with nucleus basalis pathology, would deprive all cortex from much of its normal cholinergic innvervation. A thalamic nucleus of major concern would be the nucleus reuniens and other associated mid-

line nuclei such as the paraventricular nucleus. They occupy a position between the reticular anterior pole at the thalamus and the dorsome-dial thalamic nucleus, directly over the third ventricle. In fact, the nucleus reuniens is directly dorsal to the ependymal cell lining. Both it and the paraventricular nucleus provide a major source of input to the entorhinal cortex (26, 33) (see Fig. 6.3). Moreover, the nucleus reuniens projects also to the hippocampus (26). Little is known about their input, but nonspecific brain stem sources would be expected. In consort with

Figure 6.3. This is a darkfield photomicrograph of the anterior thalamus of the rhesus monkey showing the location of retrogradely labeled neurons in the anterior thalamus and nucleus reuniens after a horseradish peroxidase injection into the entorhinal cortex of the parahippocampal gyrus. These nuclei provide a strong input to various parts of the ventromedial temporal lobe including the hippocampal formation. Note their close proximity to the third ventricle.

fornix damage, the hippocampal formation would be partially deafferented and deeffer-ented. The inferior thalamic peduncle (49) is seldom classified as a subcortical limbic system pathway, but should be elevated to that status. It carries a major component of amygdaloid output and notably a major connection with the dorso-medial thalamic nucleus (1, 47, 48, 56) (see Fig. 6.4). Its course does not have a large anterior-posterior extent, but it travels only a few millimeters from the third ventricle as it swoops in a dorsal direction beneath and medial to the globus pallidus. In fact, it intermingles, in part, with the ansa lenticularis, a topographic feature that has precluded its identification as an entity. In summary, recent experimental neuroana-

tomical observations suggest that significant dorsal, lateral, and ventral surgical damage around the third ventricle would disrupt major connections of other limbic structures that form neural systems related to certain aspects of normal memory. This may not be heartening to the neurosurgeon. However, selective strategies might be exercised during the surgical procedure, when possible, to lessen the destruction or edema that might compromise these limbic structures. Model Network for Memory The picture that emerges currently from this combination of clinical and basic science findings indicates that: (a) the intactness of modal sensory association cortices is essential for the establishment of individual traces of memory.

Figure 6.4. This darkfield photomicrograph shows major parts of the ventroamygdalofugal pathway and the inferior thalamic peduncle in a coronal plane from the rhesus monkey. Tritium-labeled amino acids were injected into the amygdala and tissue sections were processed for autoradiography. The ventroam-ygdalofugal pathway connects the amygdala with the basal forebrain and hypo-thalamus, as well as with other limbic system structures. The inferior thalamic peduncle branches posteriorly at the approximate level of the anterior commissure. It enters the thalamus from an anterior and ventral location and conveys amygdaloid output to the dorsomedial thalamic nucleus.

The process seems to require input from limbic structures and possibly also from brain stem nuclei responsible for specific neurochemical in-nervation. The latter include the locus coeruleus, nucleus basalis, nucleus of the raphe, and ventral tegmental area, which provide the cerebral cortex with intrinsic norepinephrine, acetylcho-line, serotonin and dopamine, respectively, (b) The modal traces of memory are stored in the modal sensory association cortices, (c) The establishment of multimodal contextual memory is dependent upon a complex anatomical system that includes medial temporal regions, basal forebrain structures, diencephalic structures, and some neurochemically specific brain stem nuclei, as well as subcortical connections interrelating these components. The combined function of these units is: (a) the computation of interrelationships between the several modal stimuli that constitute experiential episodes and are linked by temporal and spatial bonds, among others; (b) the storage of the records of that computation, in the form of a multiple entry, cross indexed catalog, capable of permitting recall and recognition from the standpoint of any of its constituents. The store of the abstract records of contextual relationships is possibly dependent, at least in part, upon lateral temporal lobe structures interacting with medial temporal structures such as the amygdala especially. For some types of contextual memory, frontal lobe structures may be important as well. Preoperative and Postoperative Assessment of Memory The effects of neurosurgical procedures on memory are best determined by comparison between preoperative and postoperative performance in neuropsychological tests and general behavior. When it is possible to complete both preoperative and postoperative neuropsychological assessment, patients can serve as their own controls. The results of such a complete study are useful not only for the management of the patient—especially in terms of planning the rehabilitation procedures—but also for the proper reporting of surgical results and for potential clinical research projects. In fact, such complete information is necessary if further cognitive experimental procedures are contemplated for a patient who will participate in future investigation protocols. Here we outline the fundamental principles and procedures for this assessment

and provide appropriate references for those who may want to pursue details of testing. Memory should be examined in two settings, at the bedside and in the neuropsychological laboratory. At the bedside, patients should be tested for: (a) orientation to time, place, and personal information—inquiry should include birthdate, address, and a statement as to why the patient is in the hospital; (b) recognition of doctors' and nurses' faces; (c) naming of doctors' and nurses' names; (d) recognition and naming of family members; (e) identification of the location of the patient's room within the hospital and within the ward, using both actual displacement and maps (either is applicable in immobilized patients); and (/) recall of recent events. The patients should produce an account of special medical procedures in which they have participated, a list of recent visitors (including the topics of conversation and times of the visits), and an itemization of major news events acquired through TV or newspaper. Behavior in the clinical setting (presence of geographic disorientation, reaction to meals and nursing care) will often provide a good tell-tale sign of memory impairment. More detailed and comprehensive assessment of memory can be accomplished only with special laboratory studies. Neuropsychological assessment aims at scrutinizing memory in comparison to other behaviors such as problem solving (e.g., tests of intelligence (90), visual perception (8), constructional ability (51, 58), and language (10)). Deficits of memory may occur either as isolated impairments or as part of a more pervasive cognitive impairment. This comparison is critical for determining the effects of disease and of neurosurgical procedures and for planning the rehabilitation of the patient. Laboratory assessment of memory must comprehend both anterograde and retrograde memory processes. Within the anterograde compartment, corresponding to the epoch after the onset of disease, three stages of memory should be examined. The first is immediate memory, which is most frequently tested with tasks of memory for digital sequences. In amnestic patients this stage of memory processing is almost invariably intact. The second stage is short term memory for verbal and nonverbal material, which can be assessed with a variety of learning tasks using word lists (59), word pairs (50, 89), reproduction of geometric designs (9), or recognition of new faces (i.e., previously unfamiliar). The final stage of anterograde memory to be examined is the long term and recognition of previously familiar

material. As an example, patients should be required to recall a word list that they were asked to learn 30 minutes before. This can be followed by a recognition task, in which the original target words are mixed in with distractor items. Patients with a specific disability of recall may produce a normal performance on the recognition task. The inability to perform a delayed memory task carries very different implications if it is caused by an inability to retain any information as opposed to an inability just to recall information. The other major area of laboratory study is the assessment of retrograde memory. Every patient should be evaluated for the recall and recognition of personal biographical material (such as wedding date, details on friends and relatives, educational background, graduation dates, special trips and vacations, family pets, etc.). Obviously there are no available standardized tests for this type of process and great care should be given in constructing the appropriate personal assessment of each new subject. Information for these tests can easily be provided by the family and the results of questioning can be confirmed with relatives. Nonpersonal retrograde memory can be assessed with a standardized test such as the Boston remote memory test (3), which quizzes the patient on public events and on recognition of the faces of celebrities (political, entertainment, and sports personalities from the 1920s to the 1970s). Acknowledgments We thank Paul Eslinger, Ph.D., for his help with the section dealing with neuropsychological assessment. Betty Redeker prepared the manuscript. Photographs are by Paul Reimann. References 1 . Aggleton JP, Burton MJ, Passingham RE: Cortical and subcortical afferents to the amygdala of the rhesus monkey. Brain Res 190:347-368, 1980. 2. Aggleton JP, Mishkin JP: Projections of the amyg dala to the thalamus in the cynomolgus monkey. J Comp Neural 222:56-58, 1984. 3. Albert MS, Butters N, Levin J: Temporal gradients in the retrograde amnesia of patients with alco holic Korsakoff's disease. Arch Neurol 36:211216, 1979. 4. Amaral DG, Insausti R, Cowan WM: Evidence for a direct projection from the superior temporal gyrus to the entorhinal cortex in the monkey. Brain Res 275:263-277, 1983. 5. Amaral DG, Price JL: Amygdalo-cortical projec tions in the monkey. J Comp Neurol 230:465496, 1984. 6. Avendano C, Price JL, Amaral DG: Evidence for an amygdaloid projection to premotor cortex but

7.

8. 9. 10. 1 1 12.

13.

14. 15.

16. 17. 18.

19.

20. 21.

22.

23. 24.

25.

not to motor cortex in the monkey. Brain Res 264:111-118, 1983. Baleydier C, Mauguiere F: The duality of the cingulate gyrus in monkey: Neuroanatomical study and functional hypothesis. Brain 103:525-554, 1980. Benton AL, Van Allen MW: Impairment in facial recognition in patients with cerebral disease. Cor tex 4:344-358, 1968. Benton AL: Revised Visual Retention Test (ed 4). New York, The Psychological Corp, 1974. Benton AL: Multilingual Aphasia Examination. Iowa City, Department of Neurology, University of Iowa Hospitals and Clinics, 1976. . Butters N, Cermak LS: Alcoholic Korsakoffs Syn drome. New York, Academic Press, 1980. Cohen NJ, Squire LR: Preserved learning and re tention of pattern-analyzing skill in amnesia: Dis sociation of knowing how from knowing that. Sci ence 210:207-210, 1980. Corkin S: Tactually-guided maze learning in man: Effects of unilateral cortical excisions and bilat eral hippocampal lesions. Neuropsychologia 3:339-351, 1965. Corkin S, Cohen NJ, Sager HJ: Memory for remote and public events after bilateral medial temporal lobectomy. Soc Neurosci 9:28, 1983. Corkin S: Lasting consequences of bilateral me dial temporal lobectomy: Clinical course and ex perimental findings in HM. Semin Neurol 4:249259, 1984. Damasio A, Damasio H, Van Hoesen GW: Prosopagnosia: Anatomical basis and behavioral mech anisms. Neurology (NY) 32:331-351, 1982. Damasio A: The anatomical basis of memory dis orders. Semin Neurol 4:223-225, 1984. Damasio A, Eslinger P, Damasio H, Van Hoesen GW, Cornell S: Multimodal amnesic syndrome fol lowing bilateral temporal and frontal damage: The case of patient DRB. Arch Neurol 42:252-259, 1985. Damasio A, Graff-Radford N, Eslinger P, Damasio H, Kassell N: Amnesia following ventromedial le sions of the frontal lobe. Arch Neurol 42:263271, 1985. Damasio A: Prosopagnosia. Trends Neurosci 8:132-135, 1985. Damasio A: Disorders of complex visual process ing. In Mesulam M-M (ed): (Principals of Behav ioral Neurology (Contemporary Neurology Se ries). Philadelphia, FA Davis, 1985. Divac I: Magnocellular nuclei of the basal forebrain project to neocortex, brainstem and olfac tory bulb: Review of some functional correlates. Brain Res 93:385-398, 1975. Filimonoff IN: A rational subdivision of the cere bral cortex. Arch Neurol Psychiatry 58:296-311, 1947. Goldman-Rakic PS, Selemon LD, Schwartz ML: Dual pathways connecting the dorsolateral prefrontal cortex with the hippocampal formation and parahippocampal cortex in the rhesus mon key. Neuroscience 12:719-743, 1984. Graff-Radford NR, Damasio H, Yamada T, Eslin ger PJ, Damasio AR: Nonhemorrhagic thalamic infarctions: Clinical, neuropsychological and electrophysiological findings in four anatomical groups defined by CT. Brain (in press).

26. Herkenham M: The connections of the nucleus reuniens thalami: Evidence for a direct thalamohippocampal pathway in the rat. J Comp Neural 177:589-610, 1978. 27. HerzogAG, Van Hoesen GW: Temporal neocortical afferent connections to the amygdala in the rhe sus monkey. Brain Res 115:57-69, 1976. 28. Hopkins DA: Amygdalotegmental projections in the rat, cat and monkey. Neurosci Lett 1:263270, 1975. 29. Hopkins DA, Holstege G: Amygdaloid projections to the mesencephalon, pons and medulla oblongata in the cat. Exp Brain Res 32:529-547, 1978. 30. Horel JA: The neuroanatomy of amnesia: A cri tique of the hippocampal memory hypothesis. Brain 101:403-445, 1978. 31. Hyman ВТ, Van Hoesen GW, Damasio AR, Barnes CL: Cell specific pathology isolates the hippocam pal formation in Alzheimer's disease. Science 225:1168-1170, 1984. 32. Hyman B, Van Hoesen GW, Damasio AR: Perforant pathway pathology and the memory impair ment of Alzheimer's disease. Soc Neurosci 10(1):384, 1984. 33. Insausti R, Amaral DG, Cowan WM: A horseradish peroxidase study of the af f erents to the entorhinal cortex of the monkey. Soc Neurosci Abstr 8:215, 1982. 34. Jones EG, Powell TPS: An anatomical study of converging sensory pathways within the cerebral cortex of the monkey. Brain 93:793-820, 1970. 35. Jones EG, Burton H, Saper CB, Swanson LW: Midbrain, diencephalic and cortical relationships of the basal nucleus of Meynert and associated structures in primates. J Comp Neurol 167:385420, 1976. 36. Kelly AE, Domesick VB, Nauta WJH: The amygdalostriatal projection in the rat: An anatomical study by anterograde and retrograde tracing methods. Neuroscience 7:615-630, 1982. 37. Kievit J, Kuypers HGJM: Basal forebrain and hypothalamic connections to frontal and parietal cortex in the rhesus monkey. Science 187:660662, 1975. 38. Krettek JE, Price JL: Projections from the amyg daloid complex to the cerebral cortex and thalamus in the rat and cat. J Comp Neurol 172:687722, 1977. 39. Krettek JE, Price JL: Projections from the amyg daloid complex and adjacent olfactory structures to the entorhinal cortex and to the subiculum in the rat and cat. J Comp Neurol 172:723-752, 1977. 40. MacLean P: Some psychiatric implications of psy chological studies on frontotemporal portions of limbic system (visceral brain). Electroencephalogr Cltn Neurophysiol 4:407-418, 1952. 41. Mair WGP, Warrington EK, Weiskrants L: Mem ory disorders in Korsakoff's psychosis: A neuropathological and neuropsychological investigation of two cases. Brain 102:749-783, 1979. 42. Mesulam M-M, Van Hoesen GW: Acetylcholinesterase containing basal forebrain neurons in the rhesus monkey project to neocortex. Brain Res 109:152-157, 1976. 43. Mesulam M-M, Van Hoesen GW, Pandya DN, Geschwind N: Limbic and sensory connections of

44.

45.

46. 47. 48. 49. 50. 51. 52. 53.

54. 55. 56.

57.

58. 59. 60.

61.

62.

63.

the inferior parietal lobule (area PG) in the rhesus monkey: A study with a new method for horseradish peroxidase histochemistry. Brain Res 136:393-414, 1977. Mesulam M-M, Mufson EJ, Levey AI, Wainer BH: Cholinergic innervation of cortex by the basal forebrain: Cytochemistry and cortical connec tions of the septal area, diagonal band nuclei, nucleus basalis (substantia innominata) and hypothalamus in the rhesus monkey. J Comp Neurol 214:170-197, 1983. Milner B: Visually-guided maze learning in man: Effects of bilateral hippocampal, bilateral frontal, and unilateral cerebral lesions. Neuropsychologia 3:317-338, 1965. Mufson EJ, Mesulam M-M, Pandya DN: Insular interconnections with the amygdala in the rhesus monkey. Neuroscience 6:1231-1248, 1981. Nauta WJH: Fibre degeneration following lesions of the amygdaloid complex in the monkey. J Anat 95:515-530, 1961. Nauta WJH: Neural associations of the complex in the monkey. Brain 85:505-520, 1962. Olszewski J: The Thalamus of the Macaca Mulatta. Basel, S Karger, 1952. Osborne D, Davis LJ: Standard scores for Wechsler memory scale subtests. J Clin Psychol 34:115-116, 1978. Osterrieth PA: Le test de copie d'une figure complexe. Arch Psychol 30:206-356, 1944. Pandya DN, Kuypers HGJM: Cortico-cortical con nections in the rhesus monkey. Brain Res 13:1336, 1969. Pandya DN, Van Hoesen GW, Mesulam M-M: Ef ferent connections of the cingulate gyrus in the rhesus monkey. Exp Brain Res 42:319-330, 1981. Pandya DN, Seltzer B: Association areas of the cerebral cortex. Trends Neurosci 5:386-390, 1982. Papez JW: A proposed mechanism of emotion. Arch Neurol Psychiatry 38:725-743, 1937. Porrino LJ, Crane AM, Goldman-Rakic PS: Direct and indirect pathways from the amgydala to the frontal lobe in rhesus monkeys. J Comp Neurol 198:121-136, 1981. Price JL, Amaral DG: An autoradiographic study of the projections of the central nucleus of the monkey amygdala. J Neurosci 1:1242-1259, 1981. Rey A: L'examen psychologique dans les cases d'encephalopathie traumatique. Arch Psychol 28:286-340, 1941. Rey A: L'examen clinique en psychologic. Paris, Presses Universitaries de France, 1964. Rosene DL, Van Hoesen GW: Hippocampal effer ents reach widespread areas of cerebral cortex and amygdala in the rhesus monkey. Science 198:315-317, 1977. Ross ED: Sensory-specific and fractional disor ders of recent memory in man: I. Isolated loss of visual recent memory. Arch Neurol 37:193-200, 1980. Ross ED: Sensory-specific and fractional disor ders of recent memory in man: II. Unilateral loss of tactile recent memory. Arch Neurol 37:267272, 1980. Sanides F: Comparative architectonics of the neo-

64. 65. 66.

67. 68.

69.

70. 71.

72.

73. 74. 75.

76. 77.

78.

79.

cortex of mammals and their evolutionary interpretation. Ann NYAcad Sci 167:404-423, 1969. Saper CB, Loewy AD, Swanson LW, Cowan WM: Direct hypothalamo-autonomic connections. Brain Res 117:305-312, 1976. Scoville WB, Milner B: Loss of recent memory after bilateral hippocampal lesions. J Neurol Neurosurg Psychiatry 20:11-21, 1957. Seltzer B, Pandya DN: Afferent cortical connec tions and architectonics of the superior temporal sulcus and surrounding cortex in the rhesus mon key. Brain Res 149:1-24, 1978. Seltzer B, Pandya DN: Some cortical projections to the parahippocampal area in the rhesus mon key. Exp Neurol 50:146-160, 1976. Shipley MT: The topographic and laminar orga nization of the presubiculum's projection to the ipsi- and contra-lateral entorhinal cortex in the guinea pig. J Comp Neurol 160:127-146, 1975. Sorensen KE, Shipley MT: Projections from the subiculum to the deep layers of the ipsilateral presubicular and entorhinal cortices in the guinea pig. J Comp Neurol 188:313-334, 1979. Squire LR, Moore RY: Dorsal thalamic lesion in a noted case of chronic memory dysfunction. Ann Neurol 6:503-506, 1979. Steward O: Topographic organization of the pro jections from the entorhinal area to the hippocam pal formation of the rat. J Comp Neurol 167:285314, 1976. Swanson LW, Cowan WM: An autoradiographic study of the organization of the efferent connec tions of the hippocampal formation in the rat. J Comp Neurol 172:49-84, 1977. Teuber H-L, Milner B, Vaughn HG: Persistent anterograde amnesia after stab wound of the basal brain. Neuropsychologia 6:267-282, 1968. Tulving E: Episodic and semantic memory, in Tulving E, Donaldson W (eds): Organization of Memory. New York, Academic Press Inc, 1972. Turner BM, Mishkin M, Knapp M: Organization of the amygdalopetel projections from modalityspecific cortical association areas in the monkey. J Comp Neurol 191:515-543, 1980. Van Hoesen GW, Pandya DN, Butters N: Cortical afferents to the entorhinal cortex of the rhesus monkey. Science 175:1471-1473, 1972. Van Hoesen GW, Pandya DN: Some connections of the entorhinal (area 28) and perirhinal (area 35) cortices of the rhesus monkey: I. Temporal lobe afferents. Brain Res 95:1-24, 1975. Van Hoesen GW, Pandya DN, Butters N: Some connections of the entorhinal (area 28) and per irhinal (area 35) cortices of the rhesus monkey: II. Frontal lobe afferents. Brain Res 95:25-38, 1975. Van Hoesen GW, Pandya DN: Some connections

80. 81.

82.

83. 84.

85.

86. 87. 88.

89. 90. 91.

92. 93.

94.

of the entorhinal (area 28) and perirhinal (area 35) cortices of the rhesus monkey: III. Efferent connections. Brain Res 95:39-59, 1975. Van Hoesen GW, Rosene DL, Mesulam M-M: Subicular input from temporal cortex in the rhesus monkey. Science 205:608-610, 1979. Van Hoesen GW, Yeterian EH, Lauizzo-Mourey R: Widespread corticostriate projections from tem poral cortex of the rhesus monkey. J Comp Neurol 199:205-219, 1981. Van Hoesen GW: The differential distribution, diversity and sprouting of cortical projections to the amygdala in the rhesus monkey. In Ben-Ari Y (ed): The Amygdaloid Complex. Amsterdam, Elsevier/North Holland, 1981, pp 77-90. Van Hoesen GW, Benjamin D, Afifi AK: Limbic cortical input to area 6 in the monkey. Anat Rec 199:262-263, 1981. Van Hoesen GW: The parahippocampal gyrus: New observations regarding its cortical connec tions in the monkey. Trends Neurosci 5:345-350, 1982. Volpe ВТ, Hirst W: Amnesia following the rupture and repair of an anterior communicating artery aneurysm. J Neurol Neurosurg Psychiatry 46:704-709, 1983. Victor M, Adams RD, Collins GH: The WernickeKorsakoJJ Syndrome. Philadelphia, FA Davis Co, 1971. Vilensky JA, Van Hoesen GW: Corticopontine pro jections from the cingulate cortex in the rhesus monkey. Brain Res 205:391-395, 1981. Yeterian EH, Van Hoesen GW: Cortico-striate pro jections in the rhesus monkey: The organization of certain cortico-caudate connections. Brain Res 139:43-63, 1978. Wechsler DA: A standardized memory scale for clinical use. J Psychol 19:87-95, 1945. Wechsler DA: The Wechsler Adult Intelligence Scale—Revised. New York, The Psychological Corp, 1981. Whitehouse PJ, Price DL, Clark AW, Coyle JT, DeLong MR: Alzheimer disease: Evidence for se lective loss of cholinergic neurons in the nucleus basalis. Ann Neurol 10:122-126, 1981. Whitlock DG, Nauta WJH: Subcortical projections from the temporal neocortex in Macaco mulatta. J Comp Neurol 106:183-212, 1956. Wieser HG, Yasargil MG: Selective amygdalohippocampectomy as a surgical treatment of mesiobasal limbic epilepsy. Surg Neurol 7:445-457, 1982. Zola-Morgan S, Squire LR, Mishkin M: The neuroanatomy of amnesia: Amygdala-hippocampus versus temporal stem. Science 218:1337-1339, 1982.

Commentary A Memory in Man: A Neurosurgeon's Perspective Henry D. Garrefson, M.D., Ph.D.

Intact memory function is an essential component of the human condition in coping with everyday life. Inability to retain and utilize new information as a result of an intracranial neu-rosurgical procedure is a complication potentially more devastating to the patient's long term rehabilitation and return to productive life than is the occurrence of a severe hemiparesis. In humans the limbic system together with its major association areas in the neocortex and subcorti-cal nuclei form the primary substrate for normal memory function. During the past decade major advances in our understanding of this anatomical substrate have been made through careful pre-, intra-, and postoperative testing of patients undergoing a variety of neurosurgical procedures involving the limbic system and its association areas. The human brain's division of specialized memory functions between dominant and non-dominant hemispheres for speech has made this study more complex. One consequence has been the lack of an adequate animal model with which to study many aspects of memory function as seen in humans. The anatomy of the limbic system and its association areas have already been discussed in some detail in Chapter 6. The areas of particular concern to the neurosurgeon are the vascular territory of the small perforating branches of the distal A1 segment of the anterior cerebral and the anterior communicating arteries (i.e., the substantia innominata, the anterior septal complex, and the descending columns of the fornix (1, 2); the hippocampal complex of the mesial temporal lobe together with its association areas in the lateral temporal cortex; and finally the parietooccipital association areas of the dominant hemisphere for language. The columns of the fornix in the area of the foramen of Monro play a less certain role in global memory function. Substantia Innominata and Anterior Septal Complex with Adjacent Descending Columns of the Fornix Entrance to the anterior third ventricle through the lamina terminalis for evacuation of

tumor is surprisingly well tolerated. As long as the subsequent tumor removal does not extend beyond the walls of the tumor, disturbance of consciousness and memory does not occur. However, disruption of the very fine perforating vessels arising from the distal portion of the A1 segments of the anterior cerebral artery and from the anterior communicating artery (10, 11, 13) will routinely produce a significant global anter-ograde memory disturbance of the episodic or contextual variety. These vessels can to the naked eye mimic fine arachnoidal strands. It is essential that operation in this region be done under magnification for their adequate protection. These fine perforating branches must be looked for and preserved during anterior communicating artery aneurysm clipping as well as during tumor resection in this same vicinity. The memory abnormality produced is due to bilateral ischemic damage of the immediately adjacent descending columns of the fornix and the mesial portions of the adjacent substantia innominata and nucleus basalis as well. Identification and preservation of the larger and anatomically quite variable recurrent arteries of Huebner, arising anywhere from 1 cm proximal up to 3 mm distal to the anterior communicating artery, are not enough to prevent this memory defect. Lesser and more transient degrees of memory impairment due to ischemic insult from incomplete injury to or spasm of these small perforating vessels may persist for as long as 10 to 12 weeks. Hippocampal Complex and Lateral Temporal Lobe Cortex Attention was sharply focused on the functional role of the mesial temporal lobe structures by the appearance of a devastating memory deficit in a patient, H.M., who underwent bilateral resection of the amygdala and hippocampal structures for intractable epilepsy (14). An anterior subfrontal approach was used in this patient, sparing the remainder of the temporal lobe structures. H.M. showed an almost complete inability to retain and recall, after a brief distraction, any new material presented to him, with the exception of a very slight, subtle retention of

some new motor skills after extended periods of practice. He remains able to read and reread the same literature with enjoyment without ever being aware that he has done so previously and, similarly, rework the same puzzles without showing any practice effect. In humans, memory function is partially segregated between the two cerebral hemispheres on the basis of whether there is a significant language component or verbal label for the material to be remembered. Verbally keyed memory function resides primarily in the dominant hemisphere for speech. The dominant hemisphere for speech is not synonymous with the dominant hemisphere for handedness (12). In a group of 262 patients without evidence of left hemisphere damage, 96% of righthanders and 70% of non-right handers (ambidextrous and left-handed) were left hemisphere-dominant for speech. They were shown to have speech disturbance after amobarbital injection into the carotid circulation of the left hemisphere and not after a similar injection on the opposite right side. Only 15% of the ambidextrous and left-handed group had speech function clearly lateralized to the right hemisphere, with the remaining 15% showing evidence of bilateral speech representation. Although none of these 262 patients showed evidence of gross structural lesions, they were being investigated for the management of intractable epilepsy. The translation of these results to the nonepileptic population must therefore be done cautiously. It is of particular interest and some concern that 4% of the righthanded patients in this study showed definite lateralization of speech function to the right hemisphere. Nonverbal memory function resides primarily in the nondominant temporal lobe for speech. Nonverbal memory deals with material for which there is no ready language label to assist memory function in both recognition and recall of this material. Recognizing a previously seen face out of a group of faces or an abstract pattern of lines out of a group or series of such abstract patterns are examples of nonverbal memory. Memory of spatial relationships seems to be a nondominant mesial temporal lobe-mediated function (15). Recall of deliberately memorized spatial information, such as recall of a specific location and of a learned path through a stylus maze, is impaired after excision of the right but not the left hippo-campal region. Automatic encoding of where we have seen something, even though deliberate attention was not paid to its location at the time, is a common experience that also seems to depend on the right (nondominant) hippocampus.

The absence of an adequate animal model for the study of the human type of temporal lobe memory function suggests that clarification of many aspects of neocortical participation in memory function will continue to be slow in appearing. Although the bilateral lesions in H.M. included the amygdala, uncus, and hippocampus, it has been subsequently shown that bilateral lesions of the amygdala and uncus alone do not produce memory deficit. Similarly, bilateral sectioning of the temporal stem has not produced memory impairment (19). Care must be taken not to duplicate the clinical outcome seen with H.M. by unilateral surgical resection of the amygdala and hippocampal complex in a patient who has sustained significant prior damage to the opposite mesial temporal lobe structures through trauma or some other disease process. Ojemann and coworkers recently showed that the lateral temporal cortex of the dominant temporal lobe plays a significant role in verbal memory (3, 6-9). Specific loci for interference with verbal memory produced by cortical stimulation of the dominant lateral temporal cortex during either the input or the recall phase of memory testing have been found in the mid- and posterior-temporal areas of the first and second temporal convolution, both anterior to and posterior to the primary temporal speech area. Cortex adjacent to the posterior temporoparietal speech areas seems to be involved in the input storage of short term verbal memory, whereas cortex adjacent to the anterior language area of the frontal operculum is involved in retrieval of short term verbal memory. Resection of thus identified lateral temporal cortical areas anterior to the primary speech cortex in the temporal lobe produced definite verbal memory deficits even when the hippocampus was spared. Conversely, tailoring the temporal lobe cortical resection so as to spare these identified areas of verbal memory function permitted anterior temporal lobe resection including the amygdala (but sparing the hippocampus) without producing any measurable verbal memory disturbance. Ojemann and Creutzfeldt have recently documented an increase in the neuronal firing rate in these dominant temporal lobe association areas when naming is an input memory stimulus but not during naming when it is not an input to memory. Cortical stimulation studies in the nondominant hemisphere have shown discrete localization of short term nonverbal memory in the posterior first temporal gyrus, further reinforcing the evidence for a significant lateral temporal cortical role in short term memory. Short term memory for line orientation and

face identification ("visuospatial material") is specifically interfered with by stimulation of the posterior nondominant right first temporal gyrus (3). Spatial functions, including memory, seem to be as discretely organized in the nondominant hemisphere as are verbal functions, again including memory, in the dominant hemisphere. Wieser and Yasargil reported that selective unilateral microsurgical resection of the amygdala and hippocampus with sparing of the lateral temporal lobe structures in the dominant temporal lobe did not produce significant impairment of verbal memory tasks (18). The same type of resection in the nondominant temporal lobe did, however, produce slight impairment of maze learning. These results were considered preliminary by the authors, with more detailed testing of nonverbal and verbal memory functions being planned. Columns of the Fornix at the Level of the Foramen of Monro Reported results of bilateral sectioning of the columns of the fornix at the level of the foramen of Monro are conflicting (5). Sweet and coworkers in 1959 described a patient who sustained a severe contextual or episodic memory deficit after bilateral sectioning of the columns of the fornix during the removal of a colloid cyst of the third ventricle (16). Whitty and Lishman in 1966 quoted Cairns as having indicated that bilateral forniceal sectioning could be carried out without producing memory disturbance (17). A similar observation was made by W.E. le Gros Clark and coworkers in 1938 (4). A group of patients with apparent bilateral lesions of the fornix in the immediate vicinity of the foramen of Monro with no apparent significant memory deficit have been summarized in some detail by Horel (5). It seems probable that additional abnormality must accompany bilateral forniceal damage for a flagrant amnestic syndrome to appear. An adequate study of the effect of unilateral sectioning of the columns of the fornix on memory is not available. Enlargement of the foramen of Monro by posteriorly directed dissection under magnification with appropriate microsurgical instrumentation is benign when increased exposure through the foramen of Monro is absolutely necessary. Sectioning of one or both of the columns of the fornix is rarely required and should be avoided. The unilateral destruction of structures important in memory function usually requires more careful psychometric testing than simple casual bedside observation of memory function to document the often subtle effects of such procedures.

Parietooccipital Association Areas of the Dominant Hemisphere The assignment of a disturbance in memory function as a primary etiological factor in explaining impairment of reading or writing secondary to focal lesions in the dominant parie-tooccipital lobe requires some thought. These deficits have been termed a dyspraxic form of memory impairment. Although the primary motor, tactile, and visual skills seem to be intact, the patient is unable to use these primary modalities for certain forms of visual recognition or written expression. Subtle forms of these deficits may be seen after a dominant hemisphere parietal transcortical approach to the area of the trigone and the dorsal surface of the thalamus. These deficits may be noticed clinically only if these functions are important to the patient and are routinely used in his daily activities. More widespread utilization of pre- and postoperative test batteries for the evaluation of parietooccipi-tal function would help to advance our understanding of these functions in a manner similar to the progress that is now being achieved in understanding the function of the "silent" areas of the temporal lobe. References 1 . Damasio A: The anatomical basis of memory dis orders. Semin Neurol 4:223-225, 1984. 2. Damasio A, Graff-Radford N, Eslinger P, Dimasio H, Kassell N: Amnesia following ventromedial le sions of the frontal lobe. Arch Neurol 42:263271, 1985. 3. Fried I, Mateer C, Ojemann G, Wohns R, Febio P: Organization of visuospatial functions in human cortex: Evidence from electrical stimulation. Brain 105:349-371, 1982. 4. Gros Clark WE le, Beattie J, Riddock G, Dott NM: The Hypothalamus. Edinburgh, Oliver and Boyd, 1938. 5. Horel JA: The neuroanatomy of amnesia: A cri tique of the hippocampal memory hypothesis. Brain 101:403-445, 1978. 6. Ojemann GA: Brain organization for language from the perspective of electrical stimulation mapping. In The Behavioral and Brain Sciences. Cambridge, Cambridge University Press, 1983, pp 189-230. 7. Ojemann GA, Dodrill CB: Intraoperative tech niques for reducing language and memory deficits with left temporal lobectomy. Presented at Epi lepsy International, Hamburg, November 1985. 8. Ojemann G: Organization of short-term verbal memory in language areas of human cortex: Evi dence from electrical stimulation. Brain Lang 5:331-340, 1978. 9. Ojemann GA, Dodrill CB: Verbal memory deficits after temporal lobectomy for epilepsy: Mechanism and intraoperative prediction. J Neurosurg 62:101-107, 1985. 10. Perlmutter D, Rhoton AL: Microsurgical anatomy

1 1 12.

13. 14. 15.

of the anterior cerebral-anterior communicatingrecurrent artery complex. J Neurosurg 45:259-272, 1976. . Perlmutter D, Rhoton AL: Microsurgical anatomy of the distal anterior cerebral artery. J Neurosurg 49:204-228, 1978. Rasmussen T, Milner B: The role of early left brain injury in determining lateralization of cerebral speech functions. Ann NY Acad Sci 299:355369, 1977. Rhoton AL, Perlmutter D: Microsurgical anatomy of anterior communicating artery aneurysms. Neurol Res 2:217-251, 1980. Scoville WB, Milner B: Loss of recent memory after bilateral hippocampal lesions. J Neurol Neu rosurg Psychiatry 20:11-21, 1957. Smith ML, Milner B: The role of the right hippo-

16. 17.

18.

19.

campus in the recall of spatial location. Neuropsychologia 19:781-793, 1981. Sweet WH, Talland GA, Ervin FR: Loss of recent memory following section of fornix. Trans Am Neurol Assoc 84:76-82, 1959. Whitty CWM, Lishman WA: Amnesia in cerebral disease. In Whitty CWM, Zangwill OL (eds): Am nesia. Washington, DC, Butterworths 1966, pp 36-75. Wieser HG, Yasargil MG: Selective amygdala hippocampectomy as a surgical treatment of mesiobasal limbic epilepsy. Surg Neurol 17:445-457, 1982. Zola-Morgan S, Squire LR, Mishkin M: The neuroanatomy of amnesia: Amygdala-hippocampus versus temporal stem. Science 218:1337-1339, 1982.

7 Anatomy and Physiology of Consciousness: Syndromes of Altered Consciousness Related to Third Ventricular Surgery Gilbert A. Block, M.D., and Jerome B. Posner, M.D.

Both clinical and experimental evidence assign an important role to structures surrounding the third ventricle in the regulation of motor function and consciousness. The hypothalamus and its surrounding structures are especially important as a reticular system integrator. The general aspects of the anatomy, physiology, and pathology of consciousness were extensively reviewed by Plum and Posner (67). In this chapter, we discuss the pathophysiology of abnormal states of consciousness, primarily hypersomnia and akinetic mutism, that arise when third ventricular lesions damage the hypothalamic region. We also present an integrated model of neuroanatomical and neurochemical properties as they relate to the syndromes of the third ventricle. Anatomy and Chemistry of the Hypothalamic Region The hypothalamus is the ventral portion of the diencephalon. It surrounds the inferior third ventricle and is separated from the thalamus by the hypothalamic sulcus. Rostrocaudally it extends from the preoptic area to the mamillary bodies, merging caudally with the periventricular gray of the midbrain. The hypothalamus has a highly complex cellular organization with multiple efferent and afferent projections. Traditionally it has been subdivided into anterior, medial, and posterior portions, but it is more easily described functionally

by longitudinal subdivisions (Table 7.1) (10). The major hypothalamic nuclei, their presumed function, and their afferent and efferent connections are listed in Table 7.2. The complex anatomical organization and the vast number of connections and fibers of passage through this region hamper our ability to determine completely the function of individual hypothalamic nuclei. Fibers passing through the hypothalamus are of major importance in the consideration of altered states of consciousness. The majority of these ascending and descending fibers travel in the medial forebrain bundle (MFB), which links the limbic forebrain with the lower reticular regions (Table 7.3). This multidirectional pathway constitutes the major longitudinal fiber system of the hypothalamus. The fibers that compose the MFB are not limited to the preoptic and lateral regions of the hypothalamus but convey information to midline thalamic structures as well. Fibers originating from bed nuclei that comprise the MFB terminate in most of the structures that are believed to regulate consciousness (42, 60, 95). The neuropharmacology of the hypothalamus includes norepinephrine, dopamine, serotonin, and acetylcholine pathways and is at least as complex as its anatomy. Neurotransmitter systems have been carefully studied using modern techniques that include immunocytochemical methods, fluorescence, lesioning, and punch biopsy for enzymatic assay (23, 34, 65, 66). The 213

norepinephrine (NE) distribution in the hypo-thalamus parallels the distribution of dopamine-beta-hydroxylase. The majority of the NE cells are found in Al and A2 cell groups as well as the locus coeruleus. Transection of the ventral nor-adrenergic bundle results in a significant reduction of the norepinephrine content of the preoptic area. Most hypothalamic dopamine is probably syn-

thesized in situ in the soma of the arcuate, ven-tromedial, and periventricular nuclei. The remainder arises from axons of passage arising from the mesotelencephalic cell groups (A8, A9, A10). Serotonin and tryptophan hydroxylase are present in the highest concentrations in the suprachiasmatic nucleus. The majority of serotonin originates from axons arising from the raphe nucleus (79). Hypothalamic acetylcholine has been demonstrated by acetylcholinesterase histochemistry and measurement of choline ace-tyltransferase activity (26). The highest concentrations of choline acetyltransferase are found in the paraventricular, arcuate, and dorsomedial nuclei. Surgical deafferentation of the median eminence significantly reduces the choline acetyltransferase levels in the arcuate and dorsomedial nuclei, indicating the existence of extrahypothalamic cholinergic neurons with axon terminations or axons in passage in these regions. Satoh et al. (80) demonstrated that the only choline acetyltransferasepositive hypothalamic neurons are found in the magnocellular preoptic nucleus. Taken together, the neurochemical data indicate that, with the exception of the dopamine in the tuberoinfundibular tract, the majority of the monoamine and acetylcholine content of the hy-

pothalamus arises from axons in passage or ax-ons terminating in the hypothalamus. These neurotransmitters are carried by fibers comprising the MFB, thus neurochemically linking the limbic-forebrain and limbicmesencephalic areas. Functions of the Hypothalamus Acute and chronic experimental preparations (53, 72, 92) have established the role of the hy-pothalamic region in the regulation of the following functions: neuroendocrine, thermoregu-lation, feeding, emotional output, autonomic control, osmoregulation, and consciousness. Definitive localization of these functions to precise anatomical regions of the hypothalamus is made difficult by the complex overlapping functions of different hypothalamic regions. Despite these limitations, certain gross functional generalizations are possible (for review, see 70). Neuroendocrine Functions Neuroendocrine functions of the hypothalamus have been extensively investigated. Lutein-

izing hormone-releasing hormone is predominantly localized to the preoptic area and the arcuate nucleus, whereas thyrotropin-releasing hormone is found in highest concentration in the dorsomedial nucleus. Corticotropinreleasing factor localizes to the paraventricular nucleus but seems not to colocalize with either oxytocin or vasopressin. Growth hormone-releasing hormone (GHRH) localizes to the caudal portion of the arcuate nucleus, with the majority of GHRH-positive axons projecting directly to the median eminence. Thermoregulation Thermoregulatory functions are encoded in the preopticanterior hypothalamic areas, which regulate heat loss, whereas the posterior hypothalamus contains centers for both heat loss and heat production (4, 17, 20, 21). The preoptic anterior hypothalamus contains neural thermodetectors as well as a comparator-resistance network that integrates thermal signals. Effector signals are sent to the posterior hypothalamic region, which activates heat loss systems. The

premamillary region contains neurons that function as a set point generator. Stimulation of this region activates heat loss mechanisms. Large posterior hypothalamic lesions eliminate the integrating mechanisms, prevent heat loss and heat production, and thus result in poikiloth-ermy. Feeding Feeding experiments have shown that stimulating the ventromedial medial nucleus interrupts feeding behavior, whereas bilateral lesions result in a transient hyperphagia with eventual stabilization of caloric intake at a higher level. In association with the increased intake of food is the decrease in the level of activity. Lateral hypothalamic area lesions result in aphagia and can also cause hypokinesia. Stimulation of this area activates feeding behavior mechanisms. The firing rates in the neurons of both of these areas are modified by a wide variety of factors including glucose, free fatty acids, olfactory and sensory inputs, and various neurotransmitters. Acetylcholine, acting via muscarinic receptors, causes excitation of lateral hypothalamic neurons and either excitation or inhibition of ventromedial neurons, depending on the units recorded. Similarly, norepinephrine has an excitatory effect on lateral hypothalamic neurons yet can have an excitatory or inhibitory effect on ventromedial neurons. The precise role of these neurotransmitters in the regulation of feeding behavior remains to be further elucidated. Lesions in either of these areas have been reported to decrease the level of activity. Electrophysiolog-ically, the decrease in activity is reversible (potentially) by stimulation of the lateral hypothalamic region or the administration of cholinergic or adrenergic agonists. Emotional Output Rage responses occur after lesions are made in the basal hypothalamus, particularly the ventromedial nucleus (74). As the cell bodies of the ventromedial nucleus also lie among the fibers of the MFB, it is unclear whether the production of rage behavior with ventromedial nucleus lesions is secondary to destruction of MFB fibers or intrinsic nuclear damage. Autonomic Control Early research demonstrated that the hypothalamus has a profound influence on multiple organ systems by virtue of its role in integrating autonomic nervous system function. Electrical stimulation of anterior and posterior hypothalamic regions produces changes in autonomic

function by virtue of action on descending telen-cephalic pathways that course through the hypothalamus, as well as on integrating centers in the hypothalamic periventricular regions. Lateral hypothalamic stimulation results in increased parasympathetic tone in both animals and humans, with resulting bradycardia, hypotension, and miosis. Posteromedial and ventral stimulation results in increased sympathetic tone. Osmoregulation Water balance and osmolarity are controlled via hypothalamic integration of signals from os-moreceptors and the neurosecretory response with output of vasopressin to various stimuli. The preoptic region contains osmoreceptors responsive to intracellular dehydration. Stimulation of this area in response to contracted blood volume causes increased fluid intake. Stimulation of lateral hypothalamic osmoreceptors, although resulting in increased thirst and polydyp-sia, may cause this by an effect on increasing arousal via stimulation of fibers in passage in the MFB. Consciousness Sleep and wakefulness are both dynamic processes. Wakefulness alternates cyclically with both slow wave (SWS) and paradoxical sleep (rapid eye movement, REM). SWS can be subdivided into four stages, each characterized by a progressively slower electroencephalogram (EEC) frequency and a greater depth of sleep. In humans, approximately 1.5 hours after the onset of sleep, EEG desynchronization occurs, associated with diffuse sympathetic activation and electromyographic quiescence. During this stage, REM and pontogeniculate-occipital phasic spikes occur and the arousal threshold is greatest. In humans, REM sleep occupies 20 to 25%, stage II SWS occupies 50%, and stage IV SWS occupies 15% of total sleep time. Although sleep phylogeny varies, generalizations can be made in humans from animal experiments because SWS and activated sleep (REM) occur in all mammals. Lesions and transections at various levels of the nervous system affect sleep and consciousness in different ways. The effects of surgical lesions are difficult to interpret because of imprecision in making highly localized lesions in animals and the lack of precise reporting of the specific postmortem pathological findings in humans. Nevertheless, certain classical experiments of investigators and of nature provide important insights into specific ascending and descending pathways required for the maintenance

of normal consciousness and normal sleep-wake cycles: The classical intercollicular high mid-brain transection (cerveau isole) produces hyper-somnia and SWS (6). In chronic preparations, there is an eventual return to a desynchronized EEG and increased activity awakening even if the posterior diencephalon is damaged by the lesion. Recovery of the awake EEG probably is due to intact lateral hypothalamic regions or mesencephalicpontine limbic structures. Isolated lateral hypothalamic lesions result in somnolence in cats, whereas in monkeys they cause both hypokinesia and somnolence (72). Ranson thus proposed that the posterolateral hypothal-amus is necessary in maintaining the normal waking state and that removal of downward-projecting systems results in hypersomnia (73). Nauta showed that lateral-caudal transection of the hypothalamus but not mamillary or rostral lesions caused somnolence (57). Transections at the preopticsuprachiasmatic level resulted not only in persistent wakefulness but also in sham rage. He suggested that media forebrain fibers, terminating in and traversing the lateral hypothalamic area, were necessary for maintaining cortical wakefulness. Bilateral posterior hypothalamic and subthalamic lesions, destroying the ventromedial nucleus of the thalamus, posterior hypothalamic nuclei, MFB, and mamillothalamic tract, cause somnolence and an increased arousal threshold (54). Fifty percent of the animals recover from the somnolent state but remain hypokinetic. Recovery from somnolence indicates that the continued hypokinesis reflects a continued depression of pathways of which the rostral reticular formation is only one part. Studies by Shoham and Teitelbaun (81) have demonstrated that lateral hypothalamic lesions, which destroy the ventromedial nucleus, MFB, and inferior thalamic radiations, also cause somnolence and hypokinesia. Cortical high voltage slow waves persist in all animals, but subcortical recordings indicate the presence of organized states of sleep and waking. These animals demonstrate hypokinesia and decreased endogenous arousal similar to humans in a persistent vegetative state. Even when engaged in automatic motor acts, cortical electrical activity remains slow. Thus, persistent hypokinesia represents continued subcortical activity functionally disconnected from the cortex. A neurochemical basis of sleep was first proposed by Jouvet (27). He noted that monoamines induced and maintained certain sleep periods. Serotonin from the raphe nucleus induced and maintained SWS by acting on the preoptic-basal forebrain areas, whereas norepinephrine, from

the locus coeruleus, was thought to play a similar role in REM sleep. Serotonin was further believed to induce REM sleep by acting on the locus coeruleus. Thus, serotoninergic raphe neurons and noradrenergic cells of the dorsal pontine tegmen-tum regulated sleep directly or indirectly by acting on other neurotransmitter systems. Although there are various data to support these contentions, there are also contrary results in which marked decreases in serotonin after parachloro-phenylalanine administration and MFB lesions did not specifically disrupt sleep (48, 49). A more complex mechanism of interactive reciprocal control mechanisms has been proposed. It implies that differing abnormalities may cause hyper- or hyposomnia. Hernandez-Peon et al. (22) implicated the importance of a cholinergic reticular system in the induction of sleep, and Jouvet (27) believed that cholinergic-monoami-nergic interactions help regulate sleep. There are four major groups of cholinergic (choline acetyltransferase-positive) cells in the central nervous system (64, 73): (a) rostral column (basal forebrain), (b) caudal column (mid-brain/pons), (c) local circuit neurons in dopa-mine-rich forebrain areas, and (d) motor neurons. The rostral portion (medial septal nucleus, preoptic area, rostral substantia innominata) of the basal forebrain cholinergic system primarily innervates the cingulate gyri and other telence-phalic limbic structures (5). The cholinergic soma in the brain stem are most prominent in the dorsolateral pontine tegmentum. Rostrally these cells are seen in the periaqueductal gray matter, which merges with the cholinergic soma of the nucleus parabrachialis dorsalis. Smaller populations of cholinergic cells are scattered in the tegmental reticular formation. A6 noradrenergic cells of the locus coeruleus are cholinoceptive as indicated by the positive acetylcholinesterase histochemistry (33). Similarly the raphe nucleus also seems to be cholinoceptive. Retrograde degeneration of the cholinoceptive raphe neurons is seen after MFB radiofrequency lesioning. The existence of two antagonistic cholinergic systems has been proposed by Jouvet (27). The first consists of a ventral tegmental pathway that follows one course of the MFB from the preoptic area and lateral hypothalamic area to the midbrain. Injection of acetylcholine into this area results in the induction and maintenance of sleep. Lesions of this area are followed by insomnia. The cholinergic giant cells of the pontine reticular formation are involved in the regulation of REM sleep onset through interactions with both the cholinoceptive locus coeruleus and raphe nucleus.

The simplistic view that monoamines are the primary sleep generators is not supported by neurochemical evidence indicating more complex regulatory mechanisms. Further, electro-physiological data do not support a simplistic monoamine theory that is based on one-to-one interactions between the raphe and locus areas. The firing rates of locus coeruleus and raphe cell groups do not temporally correlate with specific sleep states, although these cell groups probably play a permissive role in SWS and REM sleep. There is, however, an antagonistic relationship between serotonin and norepinephrine in sleepwake regulation (49). Lesions of the raphe nucleus, which receives dendritic connections from the locus coeruleus, cause insomnia and increased norepinephrine turnover. An increase in SWS and REM sleep is seen with locus coeruleus or dorsal noradrenergic bundle lesions; these lesions result in a concomitant decrease in fore-brain norepinephrine and increase in 5-hydrox-yindoleacetic acid. The increased serotonin turnover is probably secondary to loss of direct or indirect noradrenergic inhibiton of raphe neurons. The circadian rhythm of the sleep-wake cycle is regulated by the suprachiasmatic nucleus (SCN). This nucleus is unique in that it receives direct retinal projections (45, 47). Lesions of this nucleus cause the loss of cortisol, drinking, and motor circadian rhythms (46). Swanson and Cowan have traced SCN afferents to the ventro-medial and arcuate nuclei, and thus this region may serve as a primary clock for neuroendocrine as well as behavioral rhythms (44, 62, 93). Stimulation of this area also alters the oscillatory activity of the lateral hypothalamic neurons (63). Bilateral lesions of the SCN completely eliminate the circadian rhythm of the sleep-wake cycle but have no effect on REM or total sleep time (24), whereas anterior hypothalamic lesions have no effect on circadian rhythms (25). This has been confirmed by Mouret et al. (52), who showed that bilateral SCN lesions do not quantitatively affect SWS or REM sleep. Although the majority of SCN neurons are inhibited by ionto-phoresed serotonin and stimulated by acetylcho-line, there is no definitive effect of raphe lesions on sleep-wake cycle rhythms (61). Clinical Disorders Hypersomnia Hypersomnia can be defined as a subacute or chronic state of prolonged periods of normal sleep. Hypersomnic patients can be roused but

lapse rapidly into sleep. Clinical cases of hyper-somnia have been associated with a wide variety of third ventricular lesions. With the majority of these lesions (Table 7.4) there has been widespread infiltration or destruction of the hypo-thalamus or marked thalamic invasion. In certain instances (12, 37), more focal lesions of ventromedial or lateral hypothalamic nuclei were noted, lesions in regions that interrupt multiple fiber tracts contributing to the MFB. The cases of hypersomnia with encephalitis (12, 75) have shown predominant changes in the caudal and posterior portions of the hypothalamus. It seems, therefore, that the more posterolaterally the lesion occurs, the more likely hypersomnia will result. This is in accord with the experimental data. Akinetic Mutism Akinetic mutism, also termed the persistent vegetative state, is a term originally coined by Cairns (7) to describe the clinical syndrome of apparent wakefulness with the absence of behavioral cortical function in patients with normal sleep cycles. The term is sometimes used to describe patients who are in fact awake but vary in activity from absolute akinesia and mutism to a hypokinetic state without mutism. The syndrome has been reported with a wide variety of lesions, including those in the midbrain/rostral pons tegmentum, the posterior third ventricle (7, 55, 71, 76, 77), the bilateral cingulate gyri (2, 9, 11, 14, 59), and the bilateral basal ganglia (36). Experimental studies on akinetic mutism have yielded conflicting results. Mutism with and without akinesia has been produced in dogs and cats after periaqueductal gray lesions (83, 84). These results were not confirmed in subsequent experiments in which caudal periaqueductal lesions resulted in neither akinesia nor mutism (83). Akinetic mute states reported with anterior cingulate gyri lesions have also been cast into doubt in light of other experimental and clinical data. Bilateral ablation of area 24 in monkeys does not result in akinetic mutism (1, 13, 19, 31, 82, 87, 88, 96). In fact, normal or increased vocalization has been seen in monkeys after bilateral cingulate lesions (13, 19, 31, 41, 85-88, 92). The majority of human studies have failed to demonstrate an effect on speech production, and bilateral anterior cingulate gyral lesions have not produced akinetic mutism (38, 68, 69, 98). The reported cases of akinetic mutism in humans after bilateral cingulate lesions were cases in which the akinetic mutism resulted from

anterior cerebral artery occlusion or aneurysmal hemorrhage and rupture. This indicates that there was more widespread damage and it thus seems unlikely that akinetic mutism results from isolated cingulate lesions. Akinetic mute states have also been reported with lateral hypothalamic lesions, which may

also result in aphagia and adipsia (18, 43, 72). This area is a region of convergence and interconnections between the limbic-forebrain and the limbic-brain stem regions (48, 50, 51). This area contains the lateral and MFBs, which carry the fiber tracts mentioned earlier as well as the pallidohabenular efferents that traverse the lat-

eral hypothalamic area and the ventromedial nucleus (VMN) (58). It has recently been suggested that akinetic mutism is not the result of intrinsic hypothalamic nuclear damage but of ascending A8/A9/A10 dopaminergic fibers in the MFB. In these studies akinetic mutism was experimentally produced by 6-hydroxydopamine injections into the MFB, substantia nigra or ventral teg-mental area and was reversible by apomorphine. However, (a) 6hydroxydopamine is a nonspecific agent affecting all catecholaminergic neurons, (b) MFB injections affect multiple ascending and descending projections (60, 95), (c) reversal of this state may represent nothing more than apomorphine-induced stereotypic behavior, and (d) apomorphine is a mixed dopaminergic agonist-antagonist as well as a serotoninergic agonist in receptor assays. Indeed, the administration of spiroperidol, a dopaminergic antagonist, causes no akinesia or alteration of electromyographic activity in rats (28). Although Ross and Stewart have reported the successful use of bromocryp-tine in akinetic mutism (77), we have been unable to replicate this. Akinetic mutism associated with bilateral basal ganglia lesions is clinically similar in certain respects to akinetic mutism seen in association with third ventricular lesions. These patients demonstrate apparent cognitive impairment with marked hypokinesia or akinesia. Unlike the index case of Cairns (7), they are able to perform complex behavioral tasks after stimulation, and on further careful testing they differ from Cairns' case in that they show multiple deficits of cognitive functions. Laplane et al. (36) have proposed that the apparent "psychic akinesia" in these patients is secondary to interruption of the pallidohabenular afferents. However, their patients had normal visual-spatial activities in spite of isolated pallidal lesions, which have been previously reported to cause deficits in visual-spatial tasks (91). The explanation for this discrepancy is unclear. In summary, akinetic mutism has been reported with a variety of lesions and its association with third ventricular lesions probably results from improper execution of motor planning (40). Thus, perceptual abilities and motor programs are intact. Pathologically, akinetic mutism probably results from the interruption of multiple ascending and descending activating fibers that traverse the lateral and far lateral hypothalamic regions. Clinical Correlations Table 7.4 details all of the cases of akinetic mutism with or without hypersomnia for which

there are sufficient clinical data and either x-ray or pathological data. Lesions of the third ventricle and surrounding structures in humans that cause abnormalities of consciousness fall into three categories: akinetic mutism, hypersomnia, and akinetic mutism with hypersomnia. The table includes only those cases in which there was no evidence of increased intracranial pressure. Further, where patients were said to be apathetic, this was interpreted to indicate a degree of hypokinesia. In a lesser number of patients with third ventricular lesions, other related clinical abnormalities of hypothalamic dysfunction were seen. Hypersomnia or akinetic mutism with hypersomnia are much more common than akinetic mutism alone. A review of the available pathological data does not allow us to determine the localization of discrete lesions causing hypersomnia with or without akinetic mutism. The few cases of pure akinetic mutism with bilateral pallidal lesions were investigated only by computerized tomography, thus not eliminating the possibility of microscopic pathological changes elsewhere in surrounding third ventricular structures. Akinetic mutism alone or in association with hypersomnia is, in general, seen with less extensive lesions, although there is a fair degree of overlap. Therefore, it seems that akinetic mutism and hypersomnia represent a continuum of clinical dysfunction associated with third ventricular lesions. Conclusions The alterations of consciousness associated with lesions of the third ventricle are hypersomnia and akinetic mutism. These conditions constitute a clinical spectrum of abnormalities of normal arousal and attending mechanisms. These deficits result primarily when there has been interruption of multiple ascending and descending fibers in the medial (and perhaps lateral) forebrain bundle. The resulting deficit is probably dependent upon the severity of the insult as well as its localization, i.e., hypersomnia without akinetic mutism is more apt to occur after large posterolateral hypothalamic lesions, whereas more discrete lateral lesions (or far lateral lesions) produce akinetic mutism. The neu-rochemical mechanisms and specific pathways involved in the pathogenesis of these abnormalities are unclear and can only be inferred from the experimental data. It seems that the basal "tone" is such that, without excitatory arousal systems, we would be asleep all the time. Experimental data indicate that acetylcholine is both

inhibitory and excitatory with regard to arousal; the cholinergic cells in the pontine giant cells probably facilitate sleep, whereas descending ax-ons from the basal forebrain cholinergic activating system that terminate in the posterolateral hypothalamus are probably involved in arousal. Similarly, serotoninergic input from the raphe facilitates sleep and cortical synchronization, whereas noradrenergic input from the locus coe-ruleus inhibits sleep. The role of dopaminergic input is unclear, as is the role of the suprachias-matic nucleus. The reported insomnia that is said to occur with anterior lesions may be secondary to interruption of raphe efferents to the suprachiasmatic region; however, this is not at all clear. There should, however, be a rational mechanism for instituting pharmacological therapy for patients with akinetic mutism or hypersomnia in association with third ventricular lesions that are not surgically accessible or are unresponsive to operation. We propose the use, sequentially, of three different classes of pharmacological agents that have a theoretical basis for causing arousal. The first of these is methysergide, a serotoninergic antagonist. The rationale for this is that interruption of raphe outflow results in insomnia with a concomitant increase in norep-inephrine turnover, and methysergide therapy thus may result in arousal. Second, physostigmine, a cholinesterase inhibitor with little tox-icity, can be tried to increase basal forebrain cholinergic drive. From the clinical trials of this drug in Alzheimer's disease (94), there seem to be no "sedating effects" of this drug. This indicates that the sleep-facilitatory role of the pontine giant cells may be minimal. Third, as noradrenergic input seems to have an overall effect of decreasing sleep, it would then be appropriate to use either a monoamine oxidase A inhibitor or another nonspecific agent that increases central noradrenergic drive. Whether more specific alpha or beta adrenergic agonists would be of benefit is unclear as the pharmacology has not been further defined. References 1 . Aiken PG, Wilson WA: Discriminative vocal con ditioning in rhesus monkeys: Evidence for voli tional control. Brain Lang 8:227-240, 1979. 2. Barris RW, Shuman HR: Bilateral anterior cingulate gyrus lesions. Neurology (NY) 3:44-52, 1953. 3. Beal MF, Kleinman GM: Gangliocytoma of third ventricle: Hyperphagia, somnolence, and demen tia. Neurology (NY) 31:1224-1228, 1981. 4. Benzinger TH: Heat production: Homeostasis of central temperature in man. Physiol Rev 49:671759, 1969.

5. Bigl V, Woolf NJ, Butcher LL: Cholinergic projec tions from the basal forebrain to frontal, parietal, temporal, occipital and cingulate cortices: A com bined fluorescent tracer and acetylcholinesterase analysis. Brain Res Bull 8:727-749, 1982. 6. Bremer F: Cerveau 'isole' et physioloqie du sommeil. CR Soc Biol 118:1235-1252, 1935. 7. Cairns H, Oldfield RC, Pennybacker JB, Whitteridge D: Akinetic mutism with an epidermoid cyst of the third ventricle. Brain 64:273-290, 1941. 8. Cairns H: Disturbances of consciousness with le sions of the brain stem and diencephalon. Brain 75:109-146, 1952. 9. Critchley M: The anterior cerebral artery, and its syndromes. Brain 53:120-165, 1930. 10. Crosby E, Woodborne RT: The comparative anat omy of the preoptic area and the hypothalamus. Assoc Res Nerv Ment Dis Prog 20:52-169, 1940. 1 1 . Daly DD, Love JG: Akinetic mutism. Neurology (NY) 8:238-242, 1958. 12. Davison C, Demuth EL: Disturbances in sleep mechanism: A clinicopathologic study. III: Lesions at the diencephalic level (hypothalamus). Arch Neurol Psychiatry 55:111-133, 1946. 13. Franzen EA, Myers RE: Neural control of social behavior, prefrontal and anterior temporal cortex. Neuropsychologia 11:141-157, 1973. 14. Freemon FR: Akinetic mutism and bilateral an terior cerebral artery occlusion. J Neurol Neurosurg Psychiatry 34:693-698, 1971. 15. French JD, Magoun HW: Effects of chronic le sions in central cephalic brain stem of monkeys. Arch Neurol Psychiatry 68:591-604, 1952. 16. Fulton JF, Bailey P: Tumors in the region of the third ventricle: Their diagnosis and relation to pathological sleep. J Nerv Ment Dis 69:1-25, 145-164, 261-277, 1929. 17. Gale CC: Neuroendocrine aspects of thermoregulation. Ann Rev Physiol 35:391-430, 1973. 18. Gladfelter WE, Brobeck JR: Decreased sponta neous locomotor activity in the rat induced by hypothalamic lesions. Am J Physiol 203:811817, 1962. 19. Glees P, Cole J, Whitty CWM, et al: The effects of lesions in the cingulate gyrus and adjacent areas in monkeys. J Neurol Neurosurg Psychiatry 13:178-190, 1950. 20. Hammel HT: Regulation of internal body temper ature. Ann Rev Physiol 30:641-710, 1968. 21. Hensel H: Neural processes in thermoregulation. Physiol Rev 53:948-1017, 1973. 22. Hernandez-Peon R, Chavez-Ibarra G, Morgane PJ, et al: Limbic cholinergic pathways involved in sleep and emotional behavior. Exp. Neurol 8:93111, 1963. 23. Hokfelt T, Elde R, Fuxe K, et al: Aminergic and peptidergic pathways in the nervous system with special reference to the hypothalamus. In Reichlin MS, Baldessarini RJ, Martin JB (eds): The Hypo thalamus. New York, Raven Press, 1978, pp 69126. 24. Ibuka N, Kawamura H: Loss of circadian rhythm in sleep wakefulness in the rat by suprachias matic lesion. Brain Res 96:76-81, 1975. 25. Ibuka U, Inouye S, Kawamura H: Analysis of sleep wakefulness rhythms in male rats after supra chiasmatic nuclear lesion and ocular enucleation. Brain Res 122:33-47, 1977.

26. Jacobowitz DM, Palkovits M: Topographic atlas of catecholamine and acetylcholinesterase-containing neurons in the rat brain: I. Forebrain. J Comp Neurol 157:13-28, 1974. 27. Jouvet M: Cholinergic mechanisms in sleep. In Waser PG (ed): Cholinergic Mechanisms. New York, Raven Press, 1975, pp 455-476. 28. Kafi S, Gaillard J: Brain dopamine receptors and sleep in the rat: Effects of stimulation and block ade. Eur J Pharmacol 38:357-363, 1976. 29. Kahn EC, Crosby EC: Korsakoff's syndrome as sociated with surgical lesions involving the mammillary bodies. Neurology (NY) 22:117-125, 1972. 30. Kelly R: Colloid cysts of the third ventricle. Brain 74:23-65, 1951. 31. Kennard MA: The cingulate gyrus in relation to consciousness. J Nerv Ment Dis 121:34-39, 1955. 32. Killefer FA, Stern WE: Chronic effects of hypothalamic injury: Report of a case of near total hypothalamic destruction resulting from removal of a craniopharyngioma. Arch Neurol 22:419429, 1970. 33. Kimura H, Maeda T: Aminergic and Cholinergic systems in the dorsolateral pontine tegmentum. Brain Res Bull 9:493-499, 1982. 34. Kizer JS, Palkovits M, Brownstein MJ: The projec tions of the A8, A9 and AS 10 dopaminergic cell bodies: Evidence for a nigral-hypothalamic-median eminence dopaminergic pathway. Brain Res 108:363-370, 1976. 35. Klee A: Akinetic mutism: Review of the literature and report of a case. J Nerv Ment Dis 133:536553, 1961. 36. Laplane D, Baulac M, Widlocher D, et al: Pure psychic akinesia with bilateral lesions of basal ganglia. JNNP: 337-385, 1984. 37. Lascelles PT, Lewis PD: Hypodipsia and hypernatremia associated with hypothalamic and suprasellar lesions. Brain 95:249-264, 1972. 38. Lewin W, Whitty CWN: Effects of anterior cingu late stimulation in conscious human subjects. J Neurophysiol 23:445-447, 1960. 39. Lipsett MB, Dreifuss FE, Thomas LB: Hypothelamic syndrome following varicella. Am J Med 32:471-475, 1962. 40. Marsden CD: The mysterious motor function of the basal ganglia: The Robert Wartenberg lecture. Neurology (NY] 32:514-539, 1982. 41. Mettler FA: Anatomy and physiology. In Mettler FA (ed): Selective Partial Ablation of the Frontal Cortex. New York, Hoebner Inc, 1949, pp 477491. 42. Millhouse GE: A Golgi study of the descending medial forebrain bundle. Brain Res 15:341-363, 1969. 43. Mizutani T, Goldberg HI, et al: Cavernous hemangioma in the diencephalon. Arch Neurol 38:379382, 1981. 44. Moore RY: Central neural control of circadian rhythm. In Ganong WF, Martini L (eds): Frontiers in Neuroendocrinology. New York, Raven Press, 1978, ch7, pp 185-206. 45. Moore RY: The suprachiasmatic nucleus and the organization of a circadian rhythm. TINS 5(11):404-407, 1982.

46. Moore RY, Eichler VB: Long circadian adrenal corticosterone rhythm following suprachiasmatic lesion in the rat. Brain Res 42:201-206, 1972. 47. Moore RY, Lenn NJ: A retinohypothalamic projec tion in the rat. J Comp Neurol 146:1-14, 1972. 48. Morgane PJ: Anatomical and neurobiochemical basis of central nervous control of physiological regulations and behavior. In Morgenson GJ, Calaresu F (eds): Neurological Integration of Phys iological Mechanisms and Behavior. Toronto, University of Toronto Press, 1975, pp 24-67. 49. Morgane PJ: Amine pathways and sleep regula tion. Brain Res 9:743-749, 1982. 50. Morgane PJ: The function of limbic and rhinic forebrain limbic midbrain systems and reticular formation in regulation of food and water intake. AnnNYAcadSci 157:806-848, 1969. 51. Morgane PJ: Limbic-hypothalamic-midbrain in teraction in thirst and thirst-motivated behavior. In Wagner MJ (ed): Thirst in the Regulation of Body Water. Elmsford, NY, Pergamon Press, 1964, pp 429-455. 52. Mouret J, Coindet J: Suprachiasmatic nuclei le sions in the rat: Alterations in sleep circadian rhythms. Electroencephalogr Clin Neurophysiol 45:402-408, 1978. 53. Myers RD: Fever secondary to endotoxin injection in the hypothalamus. J Physiol 243:167-193, 1974. 54. McGinty DJ: Somnolence, recovery and hyposomnia following ventro-medial diencephalic lesions in the rat. Electroencephalogr Clin Neurophysiol 26:70-79, 1969. 55. McKissock W: The surgical treatment of colloid cyst of the third ventricle. Brain 74:1-9, 1951. 56. McKissock W, Paine KWE: Primary tumours of the thalamus. Brain 81:41-63, 1958. 57. Nauta WJH: Hypothalamic regulation of sleep in rats. An experimental study. J Neurophysiol 9:285-316, 1946. 58. Nauta WJH, Mehler WIR: Projections of the lentiform nucleus in the monkey. Brain Res 1:3-42, 1966. 59. Nielsen JM, Jacobs LL: Bilateral lesions of the anterior cingulate gyri. Bull Los Angeles Neurol Soc 16:231-234, 1951. 60. Nieuwenhuys R, Geeraedts LMG, Veening JG: The medial forebrain bundle of the rat: I. General in troduction. J Comp Neurol 206:49-81, 1982. 61. Nishino H, Koizumi K: Responses of neurons in the suprachiasmatic nuclei of the hypothalamus to putative transmitters. Brain Res 120:167-172, 1977. 62. Nishino H, Koizumi K, Brooks С МсС: The role of suprachiasmatic nuclei of the hypothalamus in the production of circadian rhythm. Brain Res 112:45-59, 1976. 63. Nishino H, Koizumi K: Rhythmic activity of neu rons in the lateral hypothalamic area and ventromedial nuclei. Fed Proc 34:424-429, 1975. 64. Oomora Y, Ono T, Sugimori M, Wagner MJ: Acetylcholine, an inhibitory transmitter in the rat lateral hypothalamus. Brain Res Bull 1:151-153, 1976. 65. O'Donohue T, Crowley WR, Jacobowitz DM: Bio chemical mapping of the noradrenergic ventral bundle projection sites: Evidence for a noradre-

66.

67. 68. 69. 70.

71. 72. 73. 74. 75.

76. 77.

78. 79.

80.

81.

nergic-dopaminergic interaction. Brain Res 172:87-100, 1979. Palkovits M, Saavedra JM, Kobayashi RM, Brownstein M: Choline acetyltransferese content of limbic nuclei of the rat. Brain Res 79:443-450, 1974. Plum F, Posner JB: The Diagnosis of Stupor and Coma. Philadephia, FA Davis, 1980. Pool JL: The visceral brain in man. J Neurophysiol 11:45-63, 1954. Pool JL, Ransohoff H: Autonomic effects on stim ulating the rostral portion of the cingulate gyri in man. J Neurophysiol 12:385-392, 1949. Plum F, Van Uitert R: Nonendocrine diseases and disorders of the hypothalamus. In Reichlin S, Baldessarini RJ, Martin JB (eds): The Hypothala mus. New York, Raven Press, 1978, pp 415-473. Poppen JL, Marino R Jr: Pinealomas and tumor of the posterior portion of the third ventricle. J Neurosurg 28:357-364, 1968. Ranson SW: Somnolence caused by hypothalamic lesions in the monkey. Arch Neurol Psychiatry 41:1-23, 1939. Ranson SW, Ingram WR: Catalepsy caused by lesions between the mammillary bodies and third nerve in the cat. Brain 55:690-696, 1932. Reeves AG, Plum F: Hyperphagia, rage and de mentia accompanying a ventromedial hypothala mic neoplasm. Arch Neurol 20:616-624, 1969. Richter RB, Traut EF: Chronic encephalitis: Path ologic report of a case with protracted somno lence. Arch Neurol Psychiatry 80:848-866, 1958. Riddoch G: Progressive dementia, without head ache or changes in the optic discs due to tumors of the third ventricle. Brain 59:225-233, 1936. Ross ED, Stewart RH: Akinetic mutism from hy pothalamic damage: successful treatment with dopamine agonists. Neurology [NY] 31:14351439, 1981. Rudelli R, Deck JHN: Selective traumatic infarc tion of the human anterior hypothalamus. J Neu rosurg 50:645-654, 1979. Saavedra JM, Palkovits M, Brownstein MJ, Axelrod J: Serotonin distribution in the nuclei of the rat hypothalamus and preoptic region. Brain Res 77:157-165, 1974. Satoh K, Armstrong DM, Fibiger HC: A compari son of the distribution of central cholinergic neu rons as demonstrated by acetylcholinesterase pharmacohistochemistry and choline acetyltransferase immunohistochemistry. Brain Res Bull 11:693-720, 1982. Shoham S, Teitelbaum P: Subcortical waking and sleep during lateral hypothalamic "somnolence"

in rats. Physiol Behav 28:323-333, 1982. 82. Showers MJ, Crosby EL: Somatic and visceral responses from the cingulate gyrus. Neurology (NY) 8:561-565, 1958. 83. Skultety FM: Stimulation of periaqueductal gray and hypothalamus. Arch Neurol 8:608-620, 1963. 84. Skultety FM: Clinical and experimental aspects of akinetic mutism: Report of a case. Arch Neurol 19:1-14, 1968. 85. Sloan N, Kaada BR: Effects of anterior limbic stimulation on somatomotor and electrocortical activity. J Neurophysiol 16:203-220, 1953. 86. Smith WK: The functional significance of the ros tral cingulate cortex as revealed by its response to electrical stimulation. J Neurophysiol 8:241255, 1945. 87. Smith WK: Vocalization and other responses elic ited by excitation of the regio cingularis in the monkey. Am J Physiol 132:441-452, 1941. 88. Smith WK: The results of ablation of the cingular region of the cerebral cortex. Fed Proc 3:42-43, 1944. 89. Smyth GE, Stern K: Tumours of the thalamus— a clinico-pathological study. Brain 61:339-374, 1938. 90. Soques A, Baruk H, Bertrand I: Un cas de tumeur de 1'infundibulum avec lethargic isolee. Rev Neu rol 1:532-540, 1926. 91. Stern Y: Behavior and the basal ganglia. Adv Neu rol 38:195-209, 1983. 92. Sutton D, Larson C, Lindeman RC: Neurocortical and limbic lesion effects on primate vocalization. Brain Res 71:61-75, 1974. 93. Swanson LW, Cowan, WM: The efferent connec tions of the suprachiasmatic nucleus of the hy pothalamus. J Comp Neurol 160:1-12, 1975. 94. Thai LJ, Fuld PA, Masur MS: Oral physostigmine and lecithin improve memory in Alzheimer dis ease. Ann Neurol 13:491-496, 1983. 95. Veening JG, Swanson LW, Cowan WM, Nieuwenhuys R, Geeraedts LG: The medial forebrain bun dle of the rat: II. An autoradiographic study of the topography of the major descending and ascend ing components. J Comp Neurol 206:82-108, 1982. 96. Ward AA: The cingular gyrus. J Neurophysiol 11:13-23, 1948. 97. Weitzman ED, Triedman MH: Hyperosmolarity as sociated with a hypothalamic lesion. Neurology (NY) 10:584-590, 1960. 98. Whitty CWM, Duffield JE, Tow PM: Anterior cingulectomy in the treatment of mental disease. Lancet 1:475-481, 1952.

8 Deep Veins Robert R. Smith, M.D., Robert A. Sanford, M.D., and Henry H. Schmidek, M.D.

It must be a tribute to humanity's inquisitive nature that Galen and his followers were able to outline the system of veins and arteries on the internal surface of the brain. In a society in which human dissection was banned, it was almost 1500 years later, during the Renaissance, that Vesalius, Michelangelo, and da Vinci were able to confirm and elaborate upon these early observations. William Harvey's work, published in 1628, provided physiological significance. Kaplan (30) and Browder et al. (7) further elucidated this system via stereoscopic examination and, finally, the angiographic methods introduced by Moniz (35) became a powerful diagnostic, as well as scientific, tool for anatomical studies. From the traditional viewpoint, the cerebral venous system can be divided into three primary systems. (a) The deep cerebral veins drain blood away from the midline structures, the deep white matter, the basal ganglia, and the diencephalon toward the superficial venous system and venous sinuses. (b) The superficial veins coalesce on the pial surface and convey blood from the outer zone of 1 to 2 cm of cortex into larger veins that communicate finally with the dural venous sinuses. (c) The cranial dura mater is composed of two layers that split and receive the contribution from both the superficial and deep venous systems. As a result of development of the telen-cephalon in humans and primates, the deep cerebral veins are covered and obscured, although for the most part they also drain major neuronal masses. The deep venous system is, by common

agreement, that system which drains into the great cerebral vein. This structure, the vein of Galen, is a short vessel, about 2 cm in length, that originates under the splenium of the corpus callosum from the union of the two internal cerebral veins. These two veins and those that empty into it are of prime concern to the third ventricular surgeon. The great cerebral vein also receives blood from the basal veins as well as branches from the pericallosal vein, the internal occipital vein, the posterior mesencephalic vein, and the precentral cerebellar and superior ver-mian veins. Thus, venous obstruction in the third ventricle may also be accompanied by venous stasis or venous infarction in the posterior cranial fossa. Anatomy of the Deep Venous System The internal cerebral veins are formed at the level of the interventricular foramen by the union of the anterior septal and the thalamostri-ate veins (Fig. 8.1). The internal cerebral veins pass posteriorly, parallel to each other entering the tela choroidea of the third ventricle. From here, they pass through the velum interpositum to the level of the splenium of the corpus callosum. At this point or slightly below it, they join with the basal veins of Rosenthal to form the great vein of Galen. The vein of Galen passes first inferiorly and then superiorly before entering the straight sinus. The internal cerebral vein has few direct tributaries of its own. However, one or two direct lateral veins may pass from the thalamus, caudate, or choroid plexus of the lat-

Figure 8.1. Structural relationship of the deep veins to the ventricular system. eral ventricles to enter the internal cerebral vein directly. The thalamostriate vein or posterior terminal vein drains the caudate nucleus, internal capsule, and deep white matter of the posterior frontal and parietal lobes and is formed by two main veins, the anterior and posterior terminal veins. The posterior terminal vein, or caudate vein as it may be called, originates at the level of the

atrium, passes in the groove between the caudate nucleus and thalamus anteriorly, and receives transverse caudate veins and medullary veins from the body of the caudate nucleus and deep white matter of the posterior, frontal, and parietal lobes, respectively, and the striate veins from the lentiform nucleus. The thalamostriate vein passes through the intraventricular foramen, creating the angiographic landmark, the

venous angle. At that point, the thalamostriate vein is joined by the anterior terminal vein and the superior choroidal vein. The thalamostriate vein is crossed at the intraventricular foramen by the choroid plexus of the lateral ventricle, which passes posteriorly in the interventricular foramen to enter the third ventricle. The anterior septal vein joins the thalamostriate somewhat more medially and inferiorly, passing usually superior to the choroid plexus (9). The anterior cerebral veins, formed by the union of several intramedullary veins, drain the deep white matter of the frontal lobe. Uniting anterior to the head of the caudate nucleus, these usually form one vein, the anterior septal vein, which runs medially, curving along the septum pellucidum to pass lateral to the columns of the fornix, where it joins the ipsilateral thalamostriate vein forming the internal cerebral vein. The superior choroidal vein courses over the superior aspect of the thalamus beneath the choroid plexus of the lateral ventricle and terminates in the internal cerebral vein or the thalamostriate vein, providing the major venous drainage of the plexus. In the roof of the lateral ventricle, there are several veins. For the most part, these pass across the roof, draining into the internal cerebral vein. Two of these have names: the direct lateral vein and the veins of the posterior horn. The veins of the thalamus drain into the internal cerebral vein, also thalamostriate veins. These enter the internal cerebral vein inferiorly. From the lateral position, the atrial veins course along the floor of the atrium of the lateral ventricle to enter the internal cerebral vein just before the two join. The internal cerebral vein then receives the pineal vein, a precentral cerebellar vein, and a posterior ventricular vein before joining the basal vein of Rosenthal (38). In the peripheral circulatory bed, 70 to 80% of the blood volume at any one time is housed in the venous system. Thus, if the same relationship exists within the intracranial space, min-ute-to-minute changes in intracranial pressure (ICP) might be almost entirely dependent upon cerebral venous capacity. If the veins were only passive tubes, then venous blood volume would change only as related to events on the arterial side. As it turns out, although regulation of arterial blood flow and volume varies considerably, the venous volume remains more or less constant under normal operating conditions. There is some evidence that extremes of ICP, such as that associated with plateau waves, may be due to venous compression. The pressure in cerebral

veins is only slightly above that of normal cere-brospinal fluid (CSF) pressure and yet 70% of the blood volume is found in these structures. Thus, any small change in venous caliber could produce major changes in volume. Little, however, is known about the myogenic response in the cerebral vein. Although a rich adrenergic network is found along the surface of major cerebral veins, at most only a few contractile fibers are present. Smooth muscle cells are occasionally found in large collecting veins but, in smaller cerebral veins, these are nonexistent. It is known, however, that cerebral veins contract spasmodically due to a number of causes, including stroking and manipulation. It may be that contractile elements in the periendothelial cells cause the reaction. Venules on the pial surface contract as much as arteries when noradrenaline is placed upon their surfaces or when the cervical sympathetic chain is stimulated. Auer and colleagues showed that beta receptor blockade did not alter adrenergic venoconstriction but alpha blockade did (24). In normal cats, sympathetic stimulation significantly decreases ICP through venoconstriction of superficial vessels. Little is known about intraparenchymal vessels or the deep venous system. Both substance P and vas-oactive intestinal polypeptide cause dilation of pial veins and arteries (19). By contrast, when serotonin is applied to the outer wall of cerebral veins, these vessels are strongly constricted (3). The endothelial cells of cerebral veins are connected by tight junctions and thus there is a biochemical and enzymatic barrier within the veins like that demonstrated in arterial beds. Under pathological conditions, such as trauma, hypertension, embolism, inflammation, or perhaps neoplasia, the venous barrier may be disturbed. The venous barrier is also influenced at increased ICP levels, beginning at 20 and 30 mm Hg. Venous extravasation occurs during acute ischemia and other conditions in which the arterial barrier is also broken. The clinical significance of venoregulatory function and the blood-brain barrier within the deep cerebral system has yet to be defined. Angiographic Patterns From the diagnostic point of view, the deep cerebral veins are much more important than the superficial veins. The angiographic pattern of the veins that drain into the internal cerebral vein is especially useful in localizing neoplasms involving the basal nuclei and midline corpus callosum. These veins are normally midline, placed one against the other at the roof of the

third ventricle in the tela choroidea. The configuration is such that they outline the roof of the third ventricle in lateral projection (46). At the foramen of Monro, the bend upward slightly and reach their highest point, forming an elliptical curve, the anterior slope of which is approximately equal in length and arc to the posterior slope. The thalamostriate vein, situated in the inferior and lateral aspect of the lateral ventricular wall, outlines the approximate size of the lateral ventricle on the anteroposterior projection. If the ventricles are enlarged, this arc becomes much wider. If the ventricle is narrowed due to medial displacement, the arc is flat and reduced. The angle made by the junction of the thalomostriate vein and the internal cerebral vein has been called the venous angle. At this point, the septal vein, extending backward from the frontal horn of the lateral ventricle on each side, usually joins the internal cerebral vein. In most cases, this is situated at the foramen of Monro; occasionally, however, the junction is much more posterior. This anatomical variant has been referred to as a false venous angle. Several methods have been offered to determine whether the position of the foramen of Monro is normal or abnormal but, because of aberrations in head shape, the internal cerebral vein may differ among individuals. The arc of the internal cerebral vein is also higher in children than in adults. When doubt exists, it should be remembered that the foramen of Monro is situated directly above the dorsum sellae. In the case of a false venous angle, the junction is much more posterior. The internal cerebral vein is humped when a mass displaces the foramen of Monro posteriorly. There are numerous anastomoses between the superficial veins and the deep veins allowing blood to flow from one system to the other. Medullary veins are sometimes seen radiating toward the ventricular wall from the cerebral hemisphere. Ordinarily, they are not visible except under pathological conditions, such as arteriovenous malformations or tumors. Occasionally, a posterior callosal vein is seen at the junction of the vein of Galen; this indicates the inferior margin of the splenium of the corpus callosum. Little can be said about venous filling. Ordinarily, the deep cerebral veins fill later than the superficial ones. However, in about 20% of normal individuals they fill simultaneously. The deep arteriovenous circulation time is about 4.25 seconds whereas the carotid-jugular time has a mean about twice that long. A prolonged circulation time is associated with low carbon dioxide

tension, and neoplasms may produce local slowing of the circulation. Diffuse increased ICP may also slow total circulation time. A regional decrease in the circulation time may be caused by tumors that shunt blood or cerebral infarction associated with the so-called luxury perfusion syndrome, status epilepticus, and, of course, arteriovenous fistulas (46). The position of the internal cerebral vein is a reliable indication of midline shift and hernia-tion. The vein of Galen is less reliable because it is fixed as it joins the straight sinus. The internal cerebral vein can be recognized on the anteroposterior view of the angiogram by tracing the thalamostriate vein to the anterior portion of the internal cerebral vein and then following it upward. If the internal cerebral vein is shifted and the anterior cerebral artery is not, there is a fair indication that a tumor is situated deeply, often in the posterior frontal region or in the thalamic area. If the vein is displaced more than the anterior cerebral artery, the mass is probably situated posteriorly. If the artery is shifted more, the mass is more anterior. The configuration of the deep veins may be helpful in defining the position of tumors near the midline. Those anterior to the venous angle cause humping of the internal cerebral vein and posterior displacement of the venous angle (Fig. 8.2). When there is elevation of the septal vein, the major tumor mass usually occurs anterior to the foramen of Monro (Fig. 8.3). The internal cerebral veins divide the thalamus into an upper one-third and a lower two-thirds. Mass lesions

Figure 8.2. "Humping" of the internal cerebral vein and distortion of the thalamostriate vein due to a large left frontal tumor associated with tuberous sclerosis.

Figure 8.3. Elevation of the internal cerebral vein and septal vein due to a large pituitary adenoma. within the thalamus elevate the thalamostriate vein, causing opening of the venous angle. There may also be separation and midline displacement of the internal cerebral veins. Tumors of the pineal gland cause elevation of the posterior portion of the internal cerebral vein, usually without displacement of the basal vein of Rosenthal (46). The midline veins are useful in delineating and preparing surgical approaches to a third ventricular mass. Colloid cysts of the anterior-superior portion of the third ventricle usually cause flattening of the internal cerebral vein in its posterior portion while the anterior one-third describes a sharp curve concave downward. When the cyst projects predominantly into one lateral ventricle, the internal cerebral vein is flattened throughout its entire length on both sides. Sometimes there is elevation of the septal vein on the side toward which the cyst protrudes. On the contralateral side, the septal vein may be lower. Fortunately, with the current availability of multiple imaging systems, diagnosis of corpus callosal masses does not depend upon angio-graphic data exclusively. However, the chief an-giographic feature of these neoplasms is elevation of the pericallosal artery with depression of the internal cerebral vein. There may be closure of the venous angle (Fig. 8.4). Likewise, there is rarely need today for attention to the deep veins in pituitary neoplasms. However, large pituitary lesions cause elevation of the internal cerebral vein as well as the septal vein (Fig. 8.3). Upward displacement of the basal vein confirms subten-

Figure 8.4. Closure of the venous angle due to a large frontal lobe metastasis. torial extension of these lesions. Angiography is still useful to demonstrate hypertrophied choro-idal vessels, including the choroidal vein, which develops hypervascularity in neoplasms such as intraventricular tumors and papillomas, rarely encountered in the third ventricle. Angiography and venography may also be useful in differentiating intraaxial from extraaxial neoplasms that invade the third ventricle. Angiography is useful in evaluating cranial base brain stem lesions. Posterior displacement of the basilar artery or the anterior pontomesencephalic venous plexus is seen with extraaxial masses that arise from the cranial base. Intrinsic brain stem masses displace these vessels anteriorly. Disorders of the Deep Venous System Although the disorders that afflict the deep venous system are not confined to pediatric age group, these diseases have been recognized more often in infants, neonates, and young children. In neonates and especially among premature infants, intracranial hemorrhage, notably the sub-arachnoid variety, has been one of the most common causes of death and morbidity. In this age group, bleeding from the subependymal matrix at the level of the foramen of Monro and in the head of the caudate nucleus is the most common cause, occurring in perhaps as many as 50% of those with a birth weight of less than 1500 g (14, 15). Hemorrhage from the choroid plexus occurs in premature infants, but much more commonly in full-term infants (18). The predisposing factors leading to hemorrhage in these regions have

been widely debated. At the foramen of Monro, the thalamostriate vein and the choroidal veins join at an acute angle forming the internal cerebral veins. It was reasoned that hypoxemia, hy-percarbia, and the need for mechanical ventilation predispose this angle to torsion, hyperemia, and subsequent hemorrhage (17). Others have implicated the use of respirators and their role in raising intracranial venous pressure. At present, although the cause of the condition is unknown, hypoxia, hypercarbia, and acidosis in premature infants apparently set the stage for the development of the disorder (13, 17). The clinical presentation may fall into one of three categories. A rapid deterioration in neurological function associated with apnea, seizures, and coma is the most devastating variety. The second is characterized by intermittent signs and symptoms with changes of tone and spontaneous movement. The third category is disclosed by screening low birth weight infants with computerized tomography (CT) or ultrasonography and apparently the disorder is asymptomatic in these individuals (21). Most physicians favor ultrasonic scanning of the premature infant in the neonatal unit in preference to CT, but both are useful diagnostic tools. Because the head circumference often does not change with primary infant intraventricular hemorrhage, clinical signs may be lateappearing. The hemorrhages may be graded on the basis of radiographic and ultrasonic findings. In Grade 1, the hemorrhage is purely subependymal. In Grades 2 and 3, intraventricular hemorrhage with and without ventricular dilatation occurs. In Grade 4 hemorrhage, the mass of hematoma dissects into the parenchymal substance of the brain. Mortality is highest in those infants with the lowest birth weight and the most severe hemorrhage (15). Progressive hydrocephalus requiring treatment is associated with the higher grades (15). Angiomas and arteriovenous malformations affect the deep venous system. Nodular high density masses with marked enhancement after intravenous contrast material are hallmarks of this diagnosis. Angiography normally reveals a pyramid-shaped mass of vessels situated peripherally, but feeding toward the deep veins. Occasionally, massive dilatation of midline veins may block the foramen of Monro or aqueduct of Sylvius, leading to ventricular enlargement (Fig. 8.5). In children, venous angiomas and hemato-mas have produced seizures and subarachnoid hemorrhage and obstructed midline CSF pathways. Probst also noted that the deep medullary veins and internal cerebral veins were not opa-

Figure 8.5. A. Obstruction of the foramen of Monro and hydrocephalus due to an enhancing mass in the lateral and third ventricular wall. B. A large dilated internal cerebral vein draining a frontal parietal arteriovenous malformation. Although the diagnosis was suspected on the basis of CT, the operative approach to midline enhancing lesions should not be undertaken without angiography. cified in some cases of the Sturge-Webber syndrome (37), leading to the hypothesis that thrombosis in the deep system may contribute to the diffuse cerebral signs occasionally associated with this disorder. Although aneurysms of the vein of Galen are not contained within the third ventricle, the arteries and veins that make up this malformation affect the third ventricle structurally. The greatly enlarged venous aneurysm draws arterial feeders from the anterior superior border, receiving both anterior cerebral arteries, the lenticulostri-ate arteries, thalamic perforators, and both the

anterior and the posterior choroidal arteries. Occasionally, the superior cerebellar arteries may be involved. In infants, the greatest contributor seems to be the posterior choroidal artery, which is situated inferiorly and laterally. In older children, however, these vessels are located anteriorly and superiorly. Greatly dilated veins resting upon the posterior third ventricle and the cerebral aqueduct vein may obstruct outflow tracts, producing hydrocephalus. In older children, subarachnoid hemorrhage has been a complicating feature. In neonates, congestive heart failure is a common mode of presentation. Angiographic studies confirm the CT diagnosis in all but those in whom the aneurysm has clotted, which presents as an avascular mass. In these cases, the deep venous system, including the internal cerebral veins, vein of Galen, and straight sinus, may not fill on the late phase of the angiogram. It is presumed that these vessels are obstructed by thrombus (49). Age and presentation are important in making therapeutic decisions about vein of Galen malformations. Shunting may be all that is necessary in older children presenting with hydrocephalus. The dilated arteries and veins may also be approached directly interhemispherically, although mortality and morbidity are extremely high (48). After coagulating and dividing perforating branches from the anterior cerebral arteries, the surgeon divides the corpus callosum. In the parietal region, the two choroidal arteries are usually seen entering the large dilated venous channel anterolaterally. These major feeders should be clipped or coagulated. Once nutrient arteries have been occluded, the vein of Galen progressively shrinks although it usually remains quite bulbous (48). Removal of the dilated venous channel, as advocated by Smith and Do-nat, is usually not feasible or warranted. Unfortunately, in many cases, other congenital central nervous system lesions are present and a successful outcome does not necessarily follow. Many of these infants, even though the malformation has been successfully obliterated, are mentally or physically retarded. Isolated thrombosis of veins in the deep cerebral venous system is being more frequently recognized. Information obtained in the operating room correlates poorly with occlusion due to trauma or infection. In the latter instance, there is widespread involvement of several members of the deep venous system and anastomotic channels. This often leads to so-called venous infarction. Spontaneous occlusion of the deep cerebral veins is largely a disorder, again, of

infancy and childhood. In our cases, the disorder was associated with diarrhea and vomiting (40). The usual history is that of a poorly developed, undernourished child appearing in acute distress. The infant is usually dehydrated and febrile, and the anterior fontanelle bulges. Serum electrolytes are often abnormal, and it is not unusual with deep venous occlusion to find cranial nerve abnormalities. Often, the thrombosis extends into the straight sinus, the vein of Galen, and the internal cerebral veins. Intraventricular hemorrhage in the lateral and third ventricles and subarachnoid hemorrhage have been seen in two of our cases. There is often severe cerebral edema. Because the cerebellar system drains into the vein of Galen, widespread deep venous thrombosis often leads to venous infarction in the cerebellum. Isolated internal cerebral vein thrombosis has been described, often without neurological deficit. Embolic occlusion of the internal cerebral veins has also been described in over 20 cases. Infarction in the medial basal ganglia and thalamus with intraventricular hemorrhage has been the usual morphological pattern (47). Thrombosis of the internal cerebral vein is also associated with birth injury. Spontaneous occlusion of the deep cerebral vein is extremely rare. Again, dehydration and vomiting are the usual underlying causes, although Towbin (47) described deep vein occlusion and bilateral sinus thrombosis with occlusion of the confluence in a 74-year-old man with multiple sites of thrombus. He presented with progressive weakness, paralysis, and coma. In that case, congestive heart failure was a predisposing factor along with debility and dehydration. Unaccountably, the disorder affects females with almost twice the frequency as males, possibly because of the use of oral contraceptives. Likewise, occlusion of superficial cerebral veins and sinuses has been associated with systemic cancer. For some unknown reason, these disorders spare the deep system. Surgical Venous Anatomy The pathological process and its anatomical location within the third ventricle largely dictate the operative approach. Because of the essential nature of the central venous structures, each operative method must successfully deal with these veins. Anterior third ventricular lesions may be approached safely by various methods. The lamina terminalis approach, although of historical significance (22, 28, 32, 43), gives extremely limited

exposure and therefore has been abandoned by most surgeons. Recently, Suzuki redescribed this approach and reported 17 cases without operative mortality (44). The transcortical approach, as initially advocated by Dandy in 1933, provides easy access to the lateral ventricle via a small cortical incision (16). In the presence of hydrocephalus, cortical collapse with removal of CSF may complicate surgical efforts due to tearing of the midline draining veins. This approach, however, obviates sacrificing major parasagittal draining veins. Only the small lateral veins of the roof of the lateral ventricle need to be sacrificed, and this is usually well tolerated. Access to the third ventricle is limited, however, via the foramen of Monro. In pathological conditions such as the colloid cyst that penetrates the foramen and is located primarily within the lateral ventricle, the transcortical approach has great appeal. To enter the third ventricle from the lateral ventricle, the foramen usually must be enlarged either at the expense of the columns of the fornix anteriorly (31) or at the expense of the thalamo-striate vein posteriorly (26). Should damage occur to the opposite fornix by surgical manipulation or compression by tumor, severe short term memory deficit may occur (24, 27, 29, 45). Some surgeons, interestingly, report no memory deficit with bilateral injuries to the columns of the fornix (6, 10, 46). Bengochea performed bilateral section of the fornix for control of epilepsy in 12 patients without memory loss (6). Posterior extension of the foramen by section of the thalamostriate vein was without significant sequelae in Hirsch's series of 10 patients (26). In our own series, only 1 venous infarction has occurred among 5 patients in whom the thalamostriate vein was sacrificed (Fig. 8.6). There is a rich anastomotic network between the deep medullary veins and the superficial medullary veins that connect to pial cortical veins (28). In most cases, the thalamostriate vein may be taken safely. It drains a portion of the lenticular nucleus and thalamus but relatively little of the internal capsule (42). Posterior enlargement of the foramen of Monro increases exposure, but the angle of visibility within the third ventricle may severely restrict the surgeon's options. The venous anatomy is also important in approaching the third ventricle through the corpus callosum. The angiogram should be carefully reviewed and the initial interhemispheric exposure planned to avoid sacrifice of major parasagittal draining veins (1). The operating microscope al-

Figure 8.6. Transcallosal approach to the third ventricle showing the thalamostriate vein and internal cerebral vein. In this case, the thalamostriate vein has been coagulated and divided to enlarge the foramen of Monro posteriorly. lows the pericallosal arteries to be separated and the corpus callosum sectioned in a bloodless fashion. At this point, most authors (20, 31, 39, 41) advocate entrance into the lateral ventricle, with access into the third ventricle via the foramen of Monro. The same anatomy is now encountered as previously described. Section of the thalamostriate vein is our preferred method of enlarging the foramen. The advantage of the trans-corpus callosal approach is threefold: either foramen of Monro or both may be inspected by veering left or right through the septum, small lateral ventricles do not hinder exposure, and the angle of surgical exposures may be varied in an anterior or posterior direction, thereby increasing the amount of third ventricle inspected. The choroid plexus overlies the thalamostriate vein in most cases and careful removal or retraction is necessary to prevent troublesome bleeding. Instead of deviating from the midline and entering the lateral ventricle, continued midline

dissection between the f ornices allows direct entry into the third ventricle (1, 5, 8, 12, 34, 36). Although Walter Dandy originally described the transcallosal approach in 1922 (16), Busch suggested the interforniceal approach in 1944 (9). This modification exposes the internal cerbral veins, which must be carefully retracted and spared. Caron et al. described two cases in which the internal cerebral veins were obstructed without ill effects during the removal of pinealomas (11). It is known that extensive venous hemorrhagic infarcts may occur in the infant with spontaneous occlusion of the internal cerebral veins. The extent of callosal section needed to expose the third ventricle varies according to the position of the lesion. Usually 2.5 to 3 cm of incision at the level of the genu is well tolerated. Corpus callosal section may be used to expose the posterior third ventricle, but the vein of Galen must be avoided and the hippocampal commissure should be preserved to prevent major memory loss (25, 52). Yoshii et al. presented the case of a 42-year-old woman who survived spontaneous occlusion of the vein of Galen, inferior sagittal sinus, and straight sinus (51). George occluded the vein of Galen safely in two patients harboring vein of Galen malformations, although the extensive collateral circulation in these patients may have been protective (40). The traditional surgical sanctity of the deep venous system suggests that these veins may not be sacrificed with impunity. There is some evidence to indicate that gradual reduction of the lumen of any major vein may not cause the same neurological dysfunction associated with abrupt occlusion. Obstruction due to a large neoplasm seems better tolerated than surgical occlusion. With reference to the deep venous sinuses, intra-operative deaths have been reported after straight sinus ligation. Merli and Carteri (33), however, resected a large infratentorial menin-gioma that involved a segment of the straight sinus, and there was no postoperative complication. As far as the vein of Galen is concerned, both primates and dogs apparently tolerate abrupt occlusion (23). In a survey of program directors made by us, however, 47% believed that infarction would follow surgical occlusion of this deep vein in humans. There is little information concerning the basal vein of Rosenthal although hemianopsia has been reported after surgical occlusion of the internal occipital vein. Many surgeons question the safety of operative occlusion of the internal cerebral vein. Thirty-five percent of the surgeons surveyed believed

that cerebral infarction would follow surgical ligation of the internal cerebral vein (40). It may be that sacrifice of this vein is justified if a total resection is to be performed for a life-threatening neoplasm. As far as the thalamostriate vein is concerned, an anastomotic system exists between the deep and superficial system. However, 15% of the surgeons surveyed felt that venous infarction was to be the expected sequela of thalamostriate vein occlusion (40). Thus, these experienced neurosurgeons recognize a definite risk for venous infarction. For the septal vein and other communicating veins into the deep system, there is little information and little evidence that obstruction would produce any ill effects. Thus, it should a surgical maxim that the deep veins, unless they must be sacrificed, should be preserved. The taking of the terminal vein, the thalamostriate vein, and even the internal cerebral vein can be accomplished with low risk if the procedure is justified. Surgical occlusion of the great vein of Galen and even the straight sinus can be carried out, but apparently with greater risk. References 1 . Apuzzo MLJ, Chikovani OK, Gott PS, et al: Trans callosal, interfornicial approaches for lesions af fecting the third ventricle: Surgical considerations and consequences. Neurosurgery 10:547-554, 1982. 2. Auer LM, Johansson BB, Lund S: Reaction of pial arteries and veins to sympathetic stimulation in the cat. Stroke 12:528-531, 1981. 3. Auer LM, Johansson BB, MacKenzie ET: Cerebral venous pressure during actively induced hyper tension and hypercapnia in cats. Stroke 11:180183, 1980. 4. Auer LM, Kuschinsky W, Johansson BB, et al: Sympathoadrenergic influence on pial veins and arteries in the cat. In Heistad DD, Marcus ML (eds): Cerebral Blood Flow: Effects of Nerves and Neurotransmitters. New York, Elsevier, North Hol land, 1982, pp 291-300. 5. Baldwin M, Ommaya AK, Farrier R, MacDonald F: Mesial cerebral incision. J Neurosurg 20:679686, 1963. 6. Bengochea FG, De La Torre O, et al: The section of the fornix in the surgical treatment of certain epilepsies. Trans Am Neurol Assoc 79:176-178, 1954. 7. Browder J, Kaplan HA, Krieger AJ: Anatomical features of the straight sinus and its tributaries: Clinical correlations. J Neurosurg 44:55-61, 1976. 8. Busch E: A new approach for the removal of tu mors of the third ventricle. Acta Psychiatr Scand 19:57-60, 1944. 9. Cairns, H, Mosberg WH Jr: Colloid cyst of the third ventricle. Surg Gynecol Obstet 92:545-570, 1951.

10. Capra NF, Anderson KV: Anatomy of the cerebral venous system. In Kapp JP, Schmidek HH (eds): The Cerebral Venous System and Its Disorders. Orlando, FL, Grune and Stratton, Inc, 1984, pp 1-36. 1 1 . Caron JP, Debrun G, Sichez JP, Comoy J, et al: Ligature des veines cerebrales internes et survie: A propos de deux pinealomectomies. Neurochirurgie 20:81-90, 1974. 12. Ciric I, Zivin I: Neuroepithelial (colloid) cysts of the septum pellucidum. J Neurosurg 43:69-73, 1975. 13. Cole FJ: History of anatomical injection. In Singer С (ed): Studies in the History and Method of Science. Oxford, Clarendon Press, 1921, vol 2, pp 285-343. 14. Cooke RWI: Factors associated with peri ventric ular haemorrhage in very low birth weight in fants. ArchDis Child 56:425-431, 1981. 15. Coulon RA Jr: Outcome of intraventricular hem orrhage in the neonate based on CT scan and/or postmortem grading. In Concepts in Pediatric Neurosurgery I. Basel, Karger, 1981, pp 68-173. 16. Dandy WE: Benign Tumors in the Third Ventri cle of the Brain: Diagnosis and Treatment. Springfield, IL, Charles С Thomas, 1933, p 171. 17. DeCourten GM, Rabinowicz T: Intraventricular hemorrhage in premature infants: Reappraisal and new hypothesis. Dev Med Child Neurol 23:389-403, 1981. 18. Donat JF, Okazaki H, Kleinberg F, et al: Intraven tricular hemorrhages in full-term and premature infants. Mayo Clin Proc 53:437-441, 1978. 19. Edvinsson L, Auer LM, Uddman R: Autonomic nerves and morphological organisation of cerebral veins. In Auer LM, Loew F (eds): The Cerebral Veins: An Experimental and Clinical Update. New York, Springer, 1983, pp 73-79. 20. Ehni G: Interhemispheric and percallosal (transcallosal) approach to the cingulate gyri, intraven tricular shunt tubes, and certain deeply placed brain lesions. Neurosurgery 14:99-110, 1984. 21. Fishman MA, Wald SL: Cerebral venous disorders of the neonate and child. In Kapp JP, Schmidek HH (eds): The Cerebral Venous System and Its Disorders. Orlando, FL, Grune and Stratton, Inc, 1984, pp 335-354. 22. Greenwood J: Removal of foreign body (bullet) from the third ventricle. J Neurosurg 7:169-172, 1950. 23. Hammock MK, Milhorat TH, Earle K, et al: Vein of Galen ligation in the primate: Angiographic, gross, and light microscopic evaluation. J Neuro surg 34:77-83, 1971. 24. Hassler O: Deep cerebral venous system in man: A microangiographic study on its area of drainage and its anastomoses with the superficial cerebral veins. Neurology (NY) 16:505-511, 1966. 25. Heilman KM, Sypert GW: Korsakoff's syndrome resulting from bilateral fornix lesions. Neurology (N7)27:490-493, 1977. 26. Hirsch JF, Zouaoui A, Renier D, et al: A new surgical approach to the third ventricle with in terruption of the striothalamic vein. Acta Neurochir(Wien) 47:135-147, 1979. 27. Horel JA: The neuroanatomy of amnesia: A cri tique of the hippocampal memory hypothesis. Brain 101:403-445, 1978.

28. Huang YP, Wolf BS: Veins of the white matter of the cerebral hemispheres (the medullary veins). AJR 4:739-755, 1964. 29. Kahn EA, Crosby EC: Korsakoff's syndrome as sociated with surgical lesions involving mammillary bodies. Neurology (NY) 22:117-125, 1972. 30. Kaplan HA; The transcerebral venous system: An anatomical study. Arch Neurol. 1:148-152, 1959. 31. Long DM, Chou SN: Transcallosal removal of craniopharyngiomas within the third ventricle. J Neurosurg 39:563-566, 1973. 32. Masson CB: Complete removal of two tumors of the third ventricle with complete recovery. Arch Surg 38:527-537, 1934. 33. Merli GA, Carteri A: Su un caso di meningioma della regione della pineale sviluppantesi nel seno retto con ostruzione completa de esso. Sist Nerv 18:62-67, 1966. 34. Milhorat TH, Baldwin M: A technique for surgical exposure of the cerebral midline: Experimental transcallosal microdissection. J Neurosurg 24:687-691, 1966. 35. Moniz E: La visibilite des sinus de la dure-mere par 1'epreuve encephalographique. Presse Med 40:1499-1502, 1932. 36. Ozgur MH, Johnson T, Smith A: Transcallosal approach to third ventricle tumor: Case report. Bull Los Angeles Neurol Soc 42:57-62, 1977. 37. Probst FP: Vascular morphology and angiographic flow patterns in Sture-Webber angiomatosis: Fact, thoughts, and suggestions. Neuroradiology 20:73-78, 1980. 38. Sheldon JJ: Blood vessels of the scalp and brain. CIBA Clin Symp Ser 33(5), 1981. 39. Shucart WA, Stein BM: Transcallosal approach to the anterior ventricular system. Neurosurgery 3:339-343, 1978. 40. Smith RR, Sanford RA: Disorders of the deep cerebral veins. In Kapp JP, Schmidek HH (eds): The Cerebral Venous System and Its Disorders. Orlando, FL, Grune and Stratton, Inc, 1984, pp 547-555. 41. Stein BM: Transcallosal approach to third ventric ular tumors. In Schmidek HH, Sweet HW (eds): Current Techniques in Operative Neurosurgery. New York, Grune and Stratton, 1977, pp 247255. 42. Stein BM: Third ventricular tumors. Clin Neuro surg 27:315-331, 1980. 43. Stookey B: Intermittent obstruction of the fora men of Monro by neuroepithelial cyst of the third ventricle: Symptoms, diagnosis and treatment. Bull Neurol Inst NY 3:446-500, 1934. 44. Suzuki J, Katakura R, Mori T: Interhemispheric approach through the lamina terminalis to tumors of the anterior part of the third ventricle. Surg NeuroZ22:157-163, 1984. 45. Sweet WH, Talland GA, Ervin FR: Loss of recent memory following section of fornix. Trans Am Neurol Assoc 84:76-82, 1959. 46. Taveras JM, Woo EH: Diagnostic Neuroradiol ogy, ed 2. Baltimore, Williams & Wilkins Co, 1976, vol 2. 47. Towbin A: The syndrome of latent cerebral ve nous thrombosis: Its frequency and relation to age and congestive heart failure. Stroke 4:419-430, 1973. 48. Watson DG, Smith RR, Brann AW: AV malfor-

mation of the vein of Galen. Am J Dis Child 130:520525, 1976. 49. Weir BKA, Allen PBR, Miller JDR: Excision of thrombosed vein of Galen aneurysm in an infant: Case report. J Neurosarg 29:619-622, 1968. 50. Woolsey RM, Nelson JS: Asymptomatic destruc tion of the fornix in man. Arch Neurol 32:566-

568, 1975. 51. Yoshii N, Seiki Y, Samejima H, et al: Occlusion of the deep cerebral veins: Case report. Neuroradiology 16:287-288, 1978. 52. Zaidel D, Sperry KW: Memory impairment after commissurotomy in man. Brain 97:263-272, 1974.

9 Pathological Lesions of the Third Ventricle and Adjacent Structures Richard L. Davis, M.D.

The region of the third ventricle has a large variety of tissues and structures and the pathological processes in the area are extremely varied. The menu of possible lesions is thus broad. In this chapter we attempt to group lesions by site, but it must be appreciated that pathological processes in adjacent areas can impinge on a nearby site. Mass Lesions within the Third Ventricle Neoplasms Astrocytomas Tumors of glial, most commonly astrocytic, lineage are the most common neuroepithelial neoplasms in this area. Their symptoms are amazingly variable, and most peculiar symptom complexes have been described (18). Although very difficult to diagnose in the past, the advent of the computerized tomographic (CT) and magnetic resonance imaging scanners has made the diagnosis of mass lesions in this area more attainable in a timely manner, so that some of our modern therapies can be given to the unfortunate patients so afflicted. Juvenile Pilocytic Astrocytoma. One of the most common tumors to fill the third ventricle and involve adjacent structures is the juvenile pilocytic astrocytoma, the astrocytoma most commonly seen in the cerebellum (42). This tumor most commonly arises from the floor of the ventricle, although it also may arise from the optic pathways and impinge on the ventricle (Fig.

9.1) (41). It is a slowly growing lesion and mainly compresses adjacent structures with only a narrow zone of invasion (Fig. 9.2). Only exceptionally have these tumors been reported to undergo malignant transformation (1). These tumors may also arise from the neurohypophysis, and the term "infundibuloma" has been applied to these neoplasms. The tumors may be grossly cystic, although this feature is less common here than in the cerebellar examples. They are often spongy and vary in color from gray to pink. Histologically, the tumors, like their cerebellar counterparts, have a biphasic pattern with compact pilocytic zones, most often around blood vessels, alternating with looser protoplasmic astrocytic areas, often with microcysts. There may be considerable variation from area to area with relation to the prominence of the two component zones. In addition, in some of these tumors there are areas that contain what appear to be oligodendroglia, with cells with a central dark-staining nucleus and a perinuclear halo. These zones rarely contain mineral, as is so commonly observed in typical oligodendrogliomas. There may be very prominent vascular proliferation; this may be so prominent that it leads the unwary to "upgrade" the tumor because of the association of vascular proliferation with increased growth potential. This interpretation is unjustified in the case of this neoplasm. In some examples multinucleate giant cells are seen, often in association with large granular, apparently membrane-bound masses, the nature

Figure 9.1. Gross photograph of a juvenile pilocytic astrocytoma of the third ventricle. The patient had had symptoms for some years and had undergone a negative exploration of the posterior fossa. This was, of course, before modern neurodiagnostic techniques.

Figure 9.2. Low power photomicrograph showing the relatively sharp border of a juvenile pilocytic astrocytoma. Although there is no "capsule," the tumor quite abruptly stops, and normal though compressed tissue is present. Hematoxylin and eosin; original magnification x 10. of which is still uncertain. Mitotic figures are distinctly uncommon, although they have been seen in some of the posterior optic nerve tumors, some of which have shown relatively aggressive biological behavior. There is often involvement of local leptomeninges, although seeding by this route is almost unknown. Fibrillary Astrocytoma. Fibrillary astrocyto-mas may arise from any structure related to the third ventricle. They are primarily infiltrating tumors, although they may form a mass lesion as part of their expression. They are most commonly solid tumors, although cysts may be present and may be prominent. In this area they most commonly arise or at least present as thalamic lesions. Histologically, these tumors show infil-

tration of the preexisting background tissue by neoplastic astrocytes that can be deceptively innocent-appearing. But more commonly, there are varying degrees of anaplasia apparent in the tumor at the time of its diagnosis. There is a gradual merging of this tumor with the glioblastoma multiforme and all degrees of malignancy exist in the spectrum of this group of neoplasms. In the series of infiltrating astrocytomas and glio-blastomas studied at UCSF, there was a clear correlation between the degree of anaplasia and the median survival (11, 26). As biological malignancy increases there is a general increase in the cellularity, the amount of nuclear atypia, the presence of cytoplasmic variability, the nuclear/ cytoplasmic ratio, the number of mitotic figures, the amount of endothelial proliferation, and the presence of necrosis. There is also a general tendency to an increase in the degree of anaplasia with time, although there are significant exceptions to this generalization (10, 11). Thus, although there is often a progression of a tumor from a moderately anaplastic astrocytoma to a glioblastoma multiforme over a period of years, this does not always pertain, and a tumor may retain its original morphology and growth potential for many years. In fact, with some tumors, notably the juvenile pilocytic astrocytomas, the tendency is to retain the original morphology and the attendant slow growth potential for the duration of the neoplasm. Protoplasmic Astrocytoma. Protoplasmic astrocytomas are relatively rare tumors wherever they occur. They theoretically arise in gray matter where protoplasmic astrocytes usually are found, are generally circumscribed at least early in their development, and (although they are invasive) are early a possibly surgically treatable tumor. However, too often at the time of diagnosis they have already become invasive and also have assumed the morphology in some areas of fibrillary astrocytoma and the increased growth potential of this lesion. In fact, the common coexistence of the protoplasmic astrocytoma in a nuclear area with infiltrating fibrillary astrocytoma in adjacent white matter is so frequent that this progression is considered to be expected. Subependymal Giant Cell Astrocytoma. The subependymal giant cell astrocytoma is the tumor associated most commonly with tuberous sclerosis. It is thus far an exclusively intraventricular tumor composed of extremely large astrocytes most often with a minimal background of glial fibers. The cells are very large, 2 to 4 times the size of gemistocytes, and have large vesicular nuclei with prominent nucleoli. Mitotic

figures may be present, but are usually not prominent. Very often the mass of cytoplasm and the nuclear character suggest that the tumor is neu-ronal in origin, and recent data support the often biphasic nature of this neoplasm. These tumors, often confused with gemistocy-tic astrocytomas, have a relatively slowly progressive course. They most often do not recur, and progression to glioblastoma multiforme is unknown or at least unreported. Tuberous sclerosis may present with such a tumor and this tumor is considered by most to be diagnostic of this disease. These tumors most often occur in the lateral ventricles and often slip out easily when encountered, presenting as "sausages" that easily are removed. Third ventricular examples of this tumor have been reported; they are clearly rare, but must be included in the differential diagnosis of third ventricular lesions, especially in the neonate or infant. Patients with tuberous sclerosis occasionally have additional infiltrating brain tumors, so that a progressive course may be due to the additional lesion. Subependymal Glomerate Astrocytoma. Subependymal glomerate astrocytomas, or sub-ependymomas, are rare third ventricular tumors. They most commonly arise from the floor of the ventricle and have been seen in the company of other rest elements, namely squamous epithelial rests. They most commonly grow slowly in other sites, but there are few data relative to their growth rate in this environment. In other areas they are amenable to total surgical extirpation, although it must be doubted that this would be successful in the third ventricle and its surrounds. From the foregoing it must be appreciated that there is little experience with this tumor in this site, so that prognostications based on actual data are not available. Other Astrocytomas. There is little experience with the other astrocytomas occurring in this area. Theoretically any astrocytoma may occur here, but the remaining tumors, the gemis-tocytic astrocytoma, the astroblastoma, and the adult pilocytic astrocytoma are so rare that there is no recorded experience with their biology in this site.

rather fast-growing, and even in vigorously treated cases the median survival is about 1 year. The structure adjacent to the third ventricle most commonly involved is the thalamus, with the tumor extending into the third ventricle from this nuclear mass (Fig. 9.3). As elsewhere, the tumor is multiform both grossly and microscopically, with areas of necrosis, softening, and alternating adjacent areas of increased firmness. These are poorly limited lesions, fading off into the surround usually without a clear margin. In cases where the margin is grossly apparent, histologi-cal examination usually shows this to be false, with infiltration of the adjacent areas by tumor not appreciated grossly. Histologically, the tumors are usually quite classical, with hypercellular areas with nuclear and cytoplasmic pleomorphism, a background of glial processes, and vascular endothelial proliferation. Necrosis with or without the typical pseudopallisading of nuclei may also be present. In the adjacent areas of infiltration there may be tumor showing much less anaplasia, and areas of edema and gliosis may also be present. Mitotic activity may be present, but is not usually very prominent. In fact, extremes of mitotic activity suggest that the tumor may be a metastasis masquerading as a glioblastoma. Primary lung tumors are particularly prone to this mimicry. Quite clearly, glioblastoma multiforme arising in the area of the third ventricle is an almost universally fatal situation. In this site neither operation nor radiation nor any of the adjuvant therapies are likely to be greatly successful, and major physical incapacity and mental impairment are likely as early consequences and as continuing symptoms. Spread across the midline is a frequent and not unexpected occurrence.

Glioblastoma Multiforme Glioblastoma multiforme is the most common primary tumor in the cerebrum of adults and, although it arises most frequently in the white matter, it commonly early involves adjacent gray matter and so may involve the stuctures in and around the third ventricle. This tumor is usually Figure 9.3. Gross photograph of glioblastoma multiforme extending into the third ventricle from the thalamus. Note that the fornix is also infiltrated.

Ependymoma Although ependymomas are relatively rare tumors of the third ventricle, they may arise in the floor, walls, or roof of this structure. As elsewhere, they are usually relatively slowly growing mass lesions, with relatively little invasion of the adjacent tissues. They usually cause symptoms by compressing the adjacent structures and by blocking the ventricular system. As with other neoplasms in this area, removal by surgical means is almost impossible and, even with the best adjuvant therapy, effective control is not to be expected. Grossly, ependymomas are firm to soft, usually delimited lesions, protruding into the ventricle and compressing adjacent structures. They are gray to pink and rarely show either necrosis or hemorrhage. Histologically, they are usually rather typical ependymomas, with prominent gliovascular structuring forming the typical per-ivascular rosettes so commonly seen with this tumor. Less commonly, ependymal rosettes may be seen, and there may be prominent areas with a dense background of glial processes. Other Neuroepithelial Tumors Although conceivably any neuroepithelial neoplasm may arise in this area, all other such tumors are exceedingly rare, except by involvement by extension. Thus, neuroblastomas may extend to involve the third ventricular area, but primary tumors of this type almost never arise here. So, too, for the medulloepitheliomas, ependymoblastomas, and even primitive neuroectodermal tumors. Choroid plexus tumors can be found here, and even gangliocytomas have been reported (5, 15), but they are unusual. Germinal Neoplasms Germinal neoplasms, specifically the germi-noma, were recognized as occasional tumors that often produced a triad of clinical symptoms when located in the floor of the third ventricle (9). This triad includes visual loss, loss of libido, and diabetes insipidus. When this occurs in young adult males germinoma is quite likely. Why germino-mas should involve the floor of the third ventricle is unclear, but they have as good a reason to arise there as in the pineal body. More recently other germinal neoplasms have also been seen in this site, and probably yet others will be described as time goes on. The gross and microscopic features are identical to those of the same tumors as they arise in the pineal body and are described under "Pineal Area Lesions."

Metastatic Tumors Metastatic tumors are the most common tumors found in the brains of adults. Although they may be seen in any site in the central nervous system (CNS), one of the more common sites in the third ventricular area is the floor of the third ventricle and the hypophysis cerebri (Fig. 9.4) (38). Carcinomas of the lung in men and of the breast in women are the most frequently seen in these areas, but neoplasms from other primary sites have also been seen. These tumors are usually found at autopsy, typically as incidental findings, but they may present clinically as pituitary tumors or be found during therapeutic hypophysectomy. Histologically, they are typical metastases, with necrosis, nuclear and cytoplasmic atypia, mitotic activity, and often adjacent vascular endothelial proliferation. However, pituitary adenomas may show some atypia, and it is only by being aware of the possibility that one may make the proper diagnosis if the metastasis is more subtle in its presentation. Craniopharyngiomas and Related Cysts The term "craniopharyngioma" has been used in a variety of different ways by different authors. Some have used it to designate any tumor of supposed Rathke pouch origin, excepting pituitary adenomas, and others have used it in a more restricted sense of the adamantinomatoid craniopharyngioma (29). Thus, care must be used with this term so that correct communication occurs. The adamantinomatoid craniopharyngioma is one of the more common childhood brain tumors. It usually presents as a result of compression of the adjacent pituitary gland, the hypophysis, or the optic pathways. It may arise within the sella or be a suprasellar or even intraventricular mass within the third ventricle (Fig. 9.5). It is usually a slowly growing tumor, but may present abruptly, perhaps as a result of hemorrhage within the tumor. Neuroradiographic studies commonly show mineralization within the mass. Pathologically, the tumor is grossly multicystic with several varieties of cyst contents. Some cysts contain flaky, grumous material, others show liquid or colloid-like material, and yet others show the "motor oil" fluid so characteristic of this disease. The tumor is most often sharply delimited, with a thin, fibrous capsule and firm gliotic tissue in the adjacent compressed brain. Histologically, the tumor is most interesting. There is a delicate fibrovascular stroma, most often with loose connective tissue, and thin-walled vascular channels. On this connective

two directions. There may be maturation to stratified squamous epithelium, with the production of cysts with masses of keratin, often with mineralized masses of large dead keratinized cells (Fig. 9.6), and there may be maturation toward glandular epithelium, with cuboidal or columnar epithelium lining the cysts with a liquid or colloidal cyst contents. Very often both types of cysts are seen. The "motor oil" cysts develop in two different ways. They may arise from cystic change within the stellate layer of epithelium or from cystic change in the loose connective tissue. Although rare, a few reports in the literature describe the presence of partially or fully formed teeth in these tumors. Considering the multiple

Figure 9.5. Gross photograph of an adamantinomatoid craniopharyngioma in the third ventricle. The patient died during pneumography; the needle tracts are visible in the corpus callosum.

Figure 9.4. A. Metastatic carcinoma from the breast infiltrating and enlarging the pituitary stalk. B. Metastatic carcinoma from the lung in a mamillary body. C. Metastatic carcinoma from the lung in the pituitary stalk. tissue is a basal layer of columnar epithelium with transitions to an intermediate stellate layer, which then shows maturation in one or both of

Figure 9.6. Photomicrograph of an adamantinomatoid craniopharyngioma showing the basal "picket fence" layer of epithelium resting on loose connective tissue and showing transitions to a stellate layer of keratinized epithelium. Hematoxylin and eosin; original magnification x 40.

types of epithelium present in most tumors, this may not be as surprising as it initially seems. With recurrence, the relationship of the epithelial components to the connective tissue and to the brain tissue may be markedly altered, and masses of tumor tissue may be present within the ventricles, displacing and apparently invading brain at some distance from the primary site (Fig. 9.7). Because of the multicystic nature of this tumor it may be possible to ameliorate symptoms by aspiration of the cyst contents (16). One of the curiosities of the reaction to compression by this tumor is the frequent dense Rosenthal gliosis that is so commonly seen in the compressed adjacent brain (Fig. 9.8). It is so common a finding that an experienced neuropathologist can predict the presence of an adamantinomatoid craniopharyngioma by this finding. It also can cause some diagnostic concern because the juvenile pilocytic astrocytomas so frequently contain these peculiar degenerating astrocytic fibers. An interesting variant of this tumor (in fact, it may be a quite distinct lesion) is the "papillary" craniopharyngioma. Because it also contains stratified squamous elements, it can be confused with the adamantinomatoid type (14). Colloid Cysts Colloid cysts are the most common intraventricular tumor in this site. They typically arise in the anteriorsuperior zone of the ventricle, characteristically blocking one or both foramina of Monro. The origin of the cysts has been the subject of much controversy and remains unsettled

Figure 9.7. Gross photograph of the brain of a woman who had had several surgical procedures attempting to alleviate her relentlessly progressive craniopharyngioma. Tumor can be seen within the third and lateral ventricles and within the basal ganglia.

(3). Probably most frequently their symptoms are those of intermittent obstruction of the foramina of Monro, with acute hydrocephalus of the lateral ventricles. However, this lesion may be the cause of relatively sudden death; this is well documented in the literature and has been supported by my personal experience of three such cases (27). Grossly the tumors are sharply circumscribed, with a thin wall and a gelatinous or semiliquid center. They may be tiny, measuring only a millimeter or two with the incidental lesions or masses several centimeters in diameter (Fig. 9.9). Even lesions large enough to cause symptoms in many may not do so in all. Histologically, the lining of the cyst wall may

Figure 9.8. Photomicrograph of a recurrent craniopharyngioma showing a mass of keratinized epithelium with a densely gliotic area, with some Rosenthal fibers. Hematoxylin and eosin; original magnification X40.

Figure 9.9. Gross photograph of a colloid cyst of the third ventricle that had caused relatively sudden death by obstruction of the foramina of Monro. The patient's complaints had been of sudden severe headaches, often relieved by minor analgesics. She died leaving the emergency room of the hospital.

vary in its epithelial composition (22). Most of the cysts have a single layer of flattened cuboidal or columnar cells lying on a basement membrane on a thin layer of fibrous connective tissue. Less commonly, the epithelium will be more complex, the least common being a pseudostratified layer of columnar epithelium (Fig. 9.10). At the electron microscopic level there is even more complexity, with the origins of these peculiar lesions not being settled even at high magnification (20, 21, 25). Other Cysts Both epidermoid and dermoid cysts may arise in this area (Fig. 9.11). These masses are often more solid than cystic. Epidermoid cysts are composed of a thin layer of maturing stratified squamous epithelium with much keratinized desquamated epithelium within the cyst (6). The dermoid tumors are more complex, and the contents of the cyst may reflect that complexity. Although the stratified squamous lining is prominent and may vary in thickness and degree of keratiniza-tion markedly from area to area, the appendage structures usually are more eye-catching. Both sweat and sebaceous glands may be quite prominent, hair may be grossly evident, and supporting dermal structures such as piloerector muscles, fat, and even occasionally teeth may be present. Inflammatory Lesions Abscess Inflammatory processes are rarely a cause of symptoms in the area of the third ventricle. However, even if rare, this must be considered when approaching a lesion in this site. The abscesses

Figure 9.10. Photomicrograph of the wall of colloid cyst showing the epithelial wall of pseudostratified columnar ciliated epithelium. Hematoxylin and eosin; original magnification x 40.

Figure 9.11. Gross photograph of an epidermoid cyst compressing the anterior third ventricle.

are almost always of hematogenous origin, and a variety of organisms may be involved. The lesions almost always originate within the adjacent parenchymal structures, although rarely there is an abscess that begins within the ventricle itself. As would be expected the abscess begins with a focus of acute inflammation and extends to involve adjacent structures. In the brain there is a delay in the formation of the fibrous wall around the inflammatory process and thus a delay in the production of an effective barrier to the extension of the inflammatory process. This is because there are no fibroblasts within the substance of the CNS except in and around the blood vessels and in the meninges. Unless a diagnosis is made very early in the course of such a process, the outcome must be expected to be fatal and, even if made early, the prognosis must be grim. If pyogenic organisms have ready access to the ventricular system and subsequently the subarachnoid space, the possibility of rapidly disseminated pyogenic meningitis is apparent and a fatal outcome is not surprising. The rare finding of an apparently healed

(evolved) intraventricular abscess also should engender no surprise, for less virulent organisms do occasionally infect humans. However, such cases must remain conjectural without a history that strongly suggests the possibility and circumstances that support the history. Granulation Tissue Although the possibility of granulation tissue causing a mass lesion in or around the third ventricle must be very small, this possibility must be part of the total differential diagnosis of mass lesions in this area. The causes remain obscure, although hemorrhage and trauma must be high on the list of possible causes. Even very low grade infections must also be considered as possible etiologies in such a situation. However, this should not be a high possibility in any differential diagnosis lacking a suggestive history. Granulomas Most of the granulomatous processes that afflict humans do so in rather a more generalized manner than involvement of the area of the third ventricle specifically. However, the possibility remains that such a lesion might cause symptoms localized to this area. Any of the granuloma-causing diseases might in fact localize to this area and, although rare, might cause symptoms of a mass in the area of the third ventricle. The causes of such a lesion are almost endless, but must include almost any organism that might result in granulomatous inflammation, including tuberculosis, any of the fungi, and the indolent bacteria that are occasionally associated with a granulomatous inflammatory process. Infestations Infestations are relatively rare in the urbanized society of the United States. With increasing travel opportunities there have been more cases of infestations within the U.S. These numbers can be expected to increase as travel to less developed areas becomes more common and the population of the U.S. continues to add immigrants from these areas. It is well known that there is a significant incidence of cysticercosis in the Mexican-American population of the southwest, and the diagnosis of CNS cysticercosis in California and the other states of the para-Mexican tier is well established. However, with the spread of these immigrants to other areas the diagnostic possibility must also be spread. We have seen a remarkable variety of symptoms associated with CNS cysticercosis, ranging

from the classical repeated pyogenic meningitic picture to the child with a mineralized granuloma with seizures to the adult with unilateral intermittent hydrocephalus. Anyone with a history of travel to or residence in an area that is endemic for pig tapeworm infestation must be considered at risk, for if this disease is not considered it is not diagnosed, even by the pathologist, because the tissue is so unfamiliar that the diagnosis may be missed. Should the cyst be found and removed at operation (2), the organism will most often be dead, and the characteristic features are the dense crenelated hyaline lamina that is the residum of the wall of the organism. The internal structures are less characteristic, although the scolex, if found, is again diagnostic. This is one of those circumstances in which knowledge of the possibility of the disease is almost essential to its correct diagnosis because the pathological picture is not well known to the majority of pathol-ogists in the U.S. Choroid Plexus Hemorrhage and " Xanthogranulomas" Hemorrhage into the Choroid plexus is a very common phenomenon, as every neuropathologist knows, although this is not well documented in the literature. Evidence of this hemorrhage is seen in almost every autopsy on individuals over the age of 30 years, in the form of intrachoroidal loose fibrous connective tissue with chronic inflammatory cells, macrophages filled with he-mosiderin and often with focal cholesterol crystals and foreign body giant cells (Fig. 9.12); these

Figure 9.12. Photomicrograph of a mass from a third ventricular Choroid plexus removed after a sudden ictus. Cholesterol clefts, foreign body giant cells, and chronic inflammation are obvious, as is the recent hemorrhage. Hematoxylin and eosin; original magnification X 40.

foci have often been given the name of "Xanthogranulomas," although they are not true neoplasms. They cause symptoms very rarely and most often are an incidental finding at autopsy (36, 37). Occasionally they may be presented by the neurosurgeon for frozen section diagnosis when he has unwittingly gotten into the trigone of the lateral ventricle when searching for a tumor. Even less commonly one sees recent hemorrhage into the Choroid plexus causing symptoms and rarely resulting in death. In the latter circumstance, the hemorrhage has been into the Choroid plexus of the third ventricle, and the mechanism of disability and death has been obstruction of the foramina of Monro (Fig. 9.5). Histologically, in recent hemorrhage there is hematoma in the parenchyma of the Choroid plexus, with a reaction dependent On the age of the lesion. Very early there may be polymorpho-nuclear leukocytes; later, macrophages become prominent, often filled with hemosiderin; later still there may be some transformation of the extravasated blood into cholesterol break-down products with macrophages and even foreign body giant cells, thus, the name "xanthogranu-loma." Although rare, Choroid plexus hemorrhage must be considered when there is a sudden ictus in a younger individual with evidence of hydrocephalus and a dense mass in the anterior third ventricle on CT scan. This is a potentially curable disease if recognized. Mass Lesions in the Walls of the Third Ventricle Neuroepithelial Tumors See the previous discussion. Inflammatory Reactions Sarcoidosis Sarcoidosis is a rare disease that commonly involves the CNS. It most commonly involves the meninges, but localized involvement of the hy-pothalamus, especially the lower walls of the third ventricle, is well known. The reason for this localization is completely cryptic. Grossly, one sees ill-defined areas of firmness with gray discoloration and, histologically, the crisp gran-ulomas without necrosis replace the hypothala-mic nuclei, with gliosis of the adjacent structures (Fig. 9.13). Schaumann bodies and asteroid structures are almost never seen in the CNS (Fig. 9.13, inset). Sarcoidosis is a diagnosis of exclusion, and

Figure 9.13. Photomicrograph of Sarcoidosis in brain, with crisp, well-defined granulomas without necrosis. Hematoxylin and eosin; original magnification x 40. Inset. Shaumann body in CNS sarcoidosis. Hematoxylin and eosin; original magnification x 160. that diagnosis must be repeatedly questioned. One must continue to search for an organism so a treatable disease is not missed. Histiocytosis Involvement of the hypothalamus with histiocytosis X is well known to clinical neuroscien-tists, although it is not a matter of common knowledge among generalists. Any of the pathological processes that bear this name may involve this area, but most frequently it is the less malignant forms that do so. The mixed histiocytic response that is seen in the tissue is nonspecific; the peculiar cells characteristic of Gaucher's disease may be present. Most commonly, the patient will not have an established diagnosis and will present with diabetes insipidus or another hy-pothalamic syndrome. Usually a needle biopsy shows nonspecific pathological findings and then other features of the disease are discovered in other areas of the body. Fortunately, most often the disease responds to radiation therapy, although more malignant courses sometimes ensue. Lymphocytic (Granulomatous) Hypophysitis Although scattered examples had been described in the earlier literature, it was only relatively recently that the entity or entities of lymphocytic (granulomatous) hypophysitis were delineated pathologically and to a lesser extent clinically (4). These lesions, which present as a rapid onset loss of vision, most often in a recently pregnant woman or one in the late third trimester, pathologically show a chronic granulomatous

hypophysitis with prominent plasmacytic infiltration, often prominent lymphoid follicles, and various degrees of fibrosis of the pituitary gland. The appearance is similar to that of Hashimoto's struma of the thyroid, and the obvious possibility of origin in autoimmune phenomena has been proposed. Studies are now under way to test this hypothesis. Until the diagnosis can be made clinically, however, biopsy may be necessary for diagnosis, especially without the "typical" clinical presentation. Cysts See previous and following discussions. Pituitary Tumors and Rathke Pouch Lesions Tumors of the adenohypophysis are among the more common tumors to afflict humans, especially females. We are far more adept now at diagnosing small tumors of this structure and dealing with endocrine-active lesions before they begin to cause symptoms because of pressure on adjacent structures. The advent of CT scans, computerized reconstructions, and radioimmu-noassay has allowed us to detect more lesions at far earlier stages. Now microadenomas are well known and are successfully dealt with using the transsphenoidal approach with its minimal morbidity and rapid recovery. Our much larger experience with a larger number of lesions has confirmed something that was well known to experienced neuropathologists in earlier times, namely, that the classical descriptions of pituitary adenomas by the terms "chrom-ophobic," "eosinophilic," and "basophilic" are almost without meaning and should be abandoned as far as routine diagnostic terminology is concerned. Further, mixed tumors are relatively common, and histologically there is no accurate way to predict whether a given tumor is producing or secreting a hormone without demonstrating the existence of that hormone in the tumor cells and by showing elevated levels of the hormone in the serum. With all that said, there are a number of different patterns in tumors of the anterior pituitary, the pars intermedia, and less commonly the pars tuberalis. Any of the adenohypophyseal portions may give rise to neoplasms and some will occur is less expected sites. Also, the normal pituitary gland is subdivided into lobules of cells, usually of multiple cell types, surrounded by delicate fibrovascular septae. In the normal gland the lobules are quite regular in

size, which may be emphasized by a reticulin stain. Further, most tumors have a thin fibrous capsule, and grossly most have a soft consistency approaching that of toothpaste, although others are much more fibrous and tough. Probably the most common histological appearance is of sheets of relatively uniform tumor cells with a rich vascular supply but without the lobular subdivision seen in the normal gland (Fig. 9.6). Depending on the amount of artifact in the material, this may appear as a very compact cellular tumor or as a looser, even sinusoidal lesion. Because the latter is also seen as a subtype, this can be confusing. In some tumors there is sufficient artifact so that there is the appearance of two types of tumor cells, one with much more granular basophilic cytoplasm and the other with more eosinophilic cytoplasm. The former often stand out as small cords of cells in the background of the other cells. In fact this appearance is most often the result of delayed fixation and, at the immunohistochemical and electron microscopic level, all of the tumor cells are similar. Next most common is the tumor with lobules of varying size, most much larger than the lobules of the normal gland. These can be very confusing because they closely mimic the normal glandular endocrine pattern, but the large size of the lobules and their irregularity betray their neoplastic nature. Most often the cells are all of the same type and in many instances they are larger than normal. The sinusoidal patterns is next most common, the tumor at low magnification appearing as sinuous bands of wavy cords of cells, often separated by almost sinusoidal thinwalled vascular channels. One of the most confusing morphological pictures seen in pituitary adenomas is a pattern of cells with processes extending toward vessel walls, simulating the appearance of ependy-moma. Adding to the confusion is the often-accompanying appearance of round nuclei surrounded by clear cytoplasm mimicking the "halo" of oligodendroglioma. The concurrence of these two cell types is of course well known in glial neoplasms, and thus an erroneous diagnosis of glial tumor can easily be made. This appearance is so common that among the experienced there is a rule to the effect that, when a tumor looks like an ependymoma in some areas and an oligodendroglioma in others, always think of pituitary adenoma. To add to the confusion, contrary to earlier dogma, mineralization in pituitary tumors does

occur and it occurs also in both glial and men-ingeal tumors, so that the pattern of mineralization is of no assistance in differential diagnosis. Necrosis may be a prominent feature of some tumors. It is expected in patients showing the syndrome of pituitary apoplexy, but often confuses the diagnosis of tumors under more routine clinical circumstances and may lead to the erroneous diagnosis of malignancy, either primary or metastatic (38). Metastasis to the pituitary area is not unknown and may present as a mass lesion (38). Hemorrhage, too, may add confusion and yet may be a prominent feature of pituitary tumors. Fibrosis, either focal or rather generalized, is seen with great frequency in pituitary neoplasms and cannot be used as a reliable index of previous operation or radiotherapy. Thus, until a larger experience suggests reliable parameters of malignancy, this diagnosis must be reserved for extraordinary cases and must be given with cautions about indications for aggressive therapy. I have found it useful to describe the features that are of concern and give a concerned although guarded prognostic indication, explaining our lack of good reliable morphological markers for malignancy in this organ. Other Lesions Although extremely rare, other lesions can cause symptoms because of obstruction of or pressure on the third ventricle or the surrounding structures (30). Pineal Area Lesions Teratomatous Lesions Including Cysts It is essential that this section begin with a discussion of some of the semantic traps that lay in wait for the innocent beginning a study of tumors of the pineal region. Even well-experienced pathologists have fallen victim to the clouds of words that fog this area of pathology. Although it is not now possible to prove, it seems likely that del Rio Hortega was the major cause of our modern confusion. According to his translators, he used the term "pinealoma" for the tumor that we today (mostly) recognize as a germinoma occurring in this peculiar site (32). It is also clear that Hortega was enough convinced of the origin of this tumor from the cells of the pineal body that he insisted that this organ contained the two cell types that are present in the tumor. Thus, if you carefully read modern histology texts, even to the last editions, you will find discussion of the two cell types found in the pineal body and references to Hortega to back

this up. Almost all also include a copy of his diagram of the cellularity of the pineal body, including the description of the two cell types. That Hortega was aware of the reality that the pineal gland contains only one morphological cell type at least at the light microscopic level is indicated by the descriptions of the pineal parenchymal tumors elsewhere described; he clearly knew that these tumors, that he called "pineocytomas" and "pineoblastomas," were derived from pineal parenchymal cells or cells that resembled them very closely. Yet, he at least legitimized the term "pinealoma" down to this generation, where it still causes much confusion. In an attempt to end this confusion, I will divide the tumors that arise in the pineal body into two major groups: those that are germinal in type, resembling more the tumors that arise in the testes or ovaries, and those of neuroepithelial origin, resembling the cells of the pineal body itself or cells resembling other neuroepithelial structures (7, 8). In approaching tumors in the pineal area one must be aware of the possible problems and how to deal with them in the best interests of the patient. After much thought I have concluded that a histological diagnosis is a desirable bit of information in dealing with pineal area tumors; I also realize that a significant group oppose that position and believe that, because germinomas compose almost 50% of pineal tumors, this should be the diagnosis of exclusion and that a therapeutic "trial" of radiation will "prove" the diagnosis. Although there is little question that a germinoma will "melt" with radiation, there is also little question that a teratoma will not or at least that the other teratomatous elements will not. Thus, although there may be a miraculous decrease in the mass or even a reversion to normal, this does not mean that the problem is gone. In those patients in whom other teratomatous elements are present, there will be an almost inexorable increase in mass over time, with recurrence of the mass and the ultimate demise of the patient. Additionally, with some of the germinal tumors there is the possibility or probability of spread to other CNS sites or even systemic metastases (12, 13). Having considered these possibilities, I think that biopsy of pineal area tumors is desirable. Although needle or direct biopsy may not demonstrate the total histological type of the lesion, it cannot do less well than no biospy at all. Should the tumor prove to be totally germinoma on biopsy, then radiation can possibly prove curative; if other elements are present, then the poorer

prognosis associated with such lesions can be shared with the patient earlier. Further, being painfully aware of the anatomical isolation of the pineal area, I cannot but urge caution in the surgical approach to this structure. It seems to me that the Creator did not intend frequent neurosurgical intervention in this area. Although I like to think that total removal of neoplasms is desirable, I also think that patient survival is important, and trauma to the midbrain, an almost unavoidable accompaniment of major resection in this area, is not associated with a long, happy, enjoyable, and productive life. Thus, I strongly support whose who espouse biopsy proof of diagnosis, but I do not believe that "total" surgical extirpation is necessarily wise. Germinal Tumors Although the cell of origin of germ cell tumors in the pineal body is not certain, there is no question that tumors of this type are the most frequent in this site (12, 23, 24, 31, 35, 40). Of the germinal tumors arising in the pineal body, the most common is the germinoma, a tumor histologically identical to the seminoma of the testes or the dysgerminoma of the ovary. This is the tumor given the not-to-be-repeated name of "pinealoma" in the older and unfortunate more recent literature. It has two cell types, a large, germinal cell with a large, vesicular nucleus, most often reniform or oval, and a prominent nucleolus. The cytoplasm is usually ill-defined and finely granular. The "other" cells of this tumor are the bystanding lymphocytes that are very often a quite prominent part of this tumor (Fig. 9.14). There is occasionally in the pineal a

Figure 9.14. Photomicrograph of a pineal germinoma showing the lymphocytic areas adjacent to the germinal neoplasm. Hematoxylin and eosin; original magnification x 40.

granulomatous "response" to the tumor, although this is probably less common than in other sites. There is often mitotic activity among the large tumor cells, and there is frequently a good deal of nuclear atypia. Although the microscopic aspects of the germinoma most often get the greatest attention, the gross aspects are equally confusing because they may be so variable. This tumor is usually soft, is often well-demarcated, and may be solid or cystic. There is often invasion of the adjacent structures and spread via cerebrospinal fluid (CSF) pathways; even systemic metastases, presumably via ventricular shunts, have been recorded (17). The nearby fornices may be diffusely involved, and this was thought to be the mode of spread to the third ventricle when that site presented as the "primary" before we were aware that this tumor might arise primarily in the third ventricle. Compression of adjacent structures is the mode of clinical presentation, with Parinaud's syndrome or less commonly such symptoms as precocious puberty. This tumor, once clinically apparent, tends to grow relatively rapidly and seed via CSF pathways. The tumor is sensitive to radiation, but it often spreads widely, and total neuraxis radiation is probably advisable in almost all cases if subarachnoid seeds are not to develop. Because this tumor accounts for almost half of the pineal region tumors, it has given many radiotherapists too much confidence in treating tumors of this area, and the nonresponding tumors have come back to haunt some of our colleagues because of their lack of knowledge of the complete spectrum of tumors that occur in this area. This spectrum of tumors is indeed large, for all of the germinal neoplasms may occur in the pineal region, singly or in combination. Although they will be discussed individually, at least from the histological standpoint, it must be appreciated and will be repeatedly emphasized that all combinations do occur. Mature teratomas are, although less frequent, a common component of the germinal neoplasms in this area. As might be expected they consist of the same tissues seen in these tumors elsewhere. Mature tissues of ectodermal, mesoder-mal, and endodermal origin in various combinations make up these tumors (Fig. 9.15). They commonly also have less well-differentiated components, and combinations of adult tissues with more primitive malignant components are all too common. Thus varying combinations of mature squamous epithelium, glandular tissue of several

Figure 9.15. Low power photomicrograph of a pineal teratoma with a mixture of mesodermal, ectodermal, and endodermal elements. Hematoxylin and eosin; original magnification x 10. types, mixtures of the several connective tissue components, loose fibrous tissue, cartilage, bone, and smooth and striated muscle may be seen (Fig. 9.8). And, perhaps not surprising considering the site, neuroepithelial components are quite common, with even retinal elements being seen rather commonly. Additionally, there may be added the glandular elements of the endo-derm, with recognizable gastric, small intestinal, or colonic epithelium, along with the appendages such as pancreas or liver. Unfortunately, all too often other more malignant and embryonal components are also present. Piled-up zones of primitive columnar cells, forming ill-defined acini, or complex glandular structures with large, polymorphic nuclei with coarse chromatin, many mitotic figures, and granular ill-defined cytoplasm is the histological picture of embryonal carcinoma (Fig. 9.16). In the variant of this tumor called "Teilum" tumor, or yolk sac carcinoma, the masses of epithelium protrude into a glandular center resembling the primitive stage of development of the yolk sac (Fig. 9.17). These are primitive, highly malignant tumors, and unfortunately they tend to spread in the CNS as they do elsewhere, so that CSF seeding is an expected complication of these tumors. Further, they are not as sensitive to radiation therapy and higher radiation doses are necessary for basic control. Another of the variants of teratomatous tumors that occurs rarely in the pineal region is the choriocarcinoma. This tumor is histologically identical (as are the other teratomatous tumors) to the genital tumors of the same name. Unfortunately, it also has similar growth characteristics, with rapid increase in size, and a frequent

Figure 9.16. Photomicrograph showing focal area of embryonal carcinoma in a teratoma. Hematoxylin and eosin; original magnification x 40.

Figure 9.17. Photomicrograph of a focus of yolk sac carcinoma in a pineal teratoma. Hematoxylin and eosin; original magnification x 100. hemorrhagic component. This also is one of the few pineal tumors that may metastasize system-ically, and some, as a result, have doubted that it is really primary in the pineal region, believing that it might have arisen in the gonads, but have gone undetected in that site. This tumor may be seen in association with one or more of the previously described teratomatous tumors, as is the case with all of this group of neoplasms. There has been a great deal of interest in the association of the pineal germinal tumors with some of the immunological markers associated with germinal tumors in other sites (13). Thus, the presence of antibody to carcinoembryonic antigen or chorionic gonadotropin may be helpful not only in diagnosis but also in following the results of treatment and in determining relapse, but to date the results are not definitive, and absolute determiners are not yet available.

Neuroepithelial Tumors The most common neuroepithelial tumors of the pineal area are the tumors derived from pineal parenchymal elements, the pineocytoma and the pineoblastoma. Other pineal parenchyma-derived tumors also occur, but are much rarer. The most common pineal parenchymal tumor is the pineoblastoma. This tumor, identical at the light microscopic level to the medulloblastoma, also shares the ability of the cerebellar tumor to spread widely and early to other sites via CSF pathways (19, 33). Grossly, the tumor is usually soft, somewhat demarcated, pink, and friable. Even at an early stage, there may be evidence of infiltration of adjacent structures, with resultant poor delimitation at the periphery. Histologically, the tumor is highly cellular with small nuclei and little in the way of cytoplasm (Fig. 9.18). Like the medulloblastoma, the pineoblastoma often shows varying degrees of differentiation, more frequently toward the neuroblastic/pineocytic elements than toward the glial elements, as is true of the medulloblastoma. Thus, one sees the exaggerated Homer Wright rosette also seen in the pineocytoma, with a large circlet of dark-staining small nuclei surrounding a mass of tangled processes, much larger than the usual Homer Wright rosette, with 10 to 15 or more nuclei in the circle. This tumor, as suggested previously, also tends to seed the neuraxis widely early in its course, and it must be presumed that CSF spread has occurred at the time of diagnosis. We have used therapy identical to that for medulloblastoma, but thus far have too few patients to assess the results meaningfully. The pineocytoma is the next most common tumor in the pineal area. This tumor mimics the

Figure 9.18. Photomicrograph of a pineoblastoma with small cells resembling those of medulloblastoma. Hematoxylin and eosin; original magnification x 100.

structure of the normal pineal body in an exaggerated way, with cells and nuclei that closely resemble the parent structure. There are usually few mitotic figures, and necrosis is distinctly rare. The tumor is usually grossly defined, although there may be gross evidence of local infiltration. The tumor is usually solid, but cystic examples have been reported. Hemorrhage and necrosis are not usually a gross feature; after therapy, such changes may be seen. Histologically, this tumor grows as sheets and lobules of cells with prominent vesicular nuclei, often with prominent nucleoli, but without evidence of rapid growth, i.e., there are rarely many mitotic figures. Necrosis is also distinctly uncommon. The cytoplasma is as prominent as is the cytoplasm of the pineal cells normally, i.e., there is little to be seen at the light level except for delicate processes forming the background and the mass of processes seen in the center of the rosettes (Fig. 9.19). The vessels may be prominent, but endothelial proliferation is not prominent. There are usually areas in which the exaggerated Homer Wright rosettes previously noted are found with ease, but this may not be the dominant histological feature. In a few cases there are areas with a dense glial mass similar to that found in the core of many normal pineal bodies, with a dense meshwork of background glial processes, with few cell nuclei, usually typically astrocytic, and without evidence of significant cellular proliferation. Rosenthal fibers may be very prominent within these areas. However, these masses are most often several times the mass of the normal pineal body as a whole, and it seems clear that they are part of the neoplastic process. The whole potential of differentiation in these tumors is amazing and, although probably

Figure 9.19. Photomicrograph of a pineocytoma with rosettes with intertwining processes being prominent features. Hematoxylin and eosin; original magnification x 40.

not all types have been appreciated, many have been well described (19, 28, 39). In very rare cases one may see transitions from pineoblastoma through pineocytoma to indolent astrocytoma all in the same tumor. Our experience is insufficient to know the biology of the latter types of tumor, and for the present we are assuming the worst and treating these patients as if they had a pure pineoblastoma. Because all of the basic neuroepithelial cellular elements exist in this area, it is not surprising that tumors derived from these elements arise in this area, although uncommonly. Thus, the spectrum of astrocytic, ependymal, and most rarely oligodendroglial neoplasms may be seen in this area. We have even seen primary lymphomas, at least presenting as pineal neoplasms. Tumors of the meninges, connective tissue, and blood vessels may also be expected sooner or later to give rise to neoplasms of this gland or at least in this area. Metastases Although rarely clinically apparent, metastases to the pineal body, like the other areas of the nervous system, are more common than is generally appreciated. In over 25 years of routine histological examination of pineal tissue, metastatic tumors have been found with more frequency than had been expected with almost no evident symptoms that could be ascertained from the clinical record. The possibility of metastasis must be borne in mind, especially when dealing with biopsy material. When presented with tissue showing only singular differentiation, one must consider the possibility of metastatic carcinoma or even metastatic sarcoma. Although not considered a significant problem, neoplasms, especially glial, may spread from adjacent sites in which they might be more common. Good sense and judgment must prevail. Other As suggested, involvement of the pineal area by tumors from adjacent tissues must be expected. Thus, meningiomas, vascular malformations, and other massincreasing lesions from adjacent areas will sooner or later involve this tissue (34). References 1. Alpers CE, Davis RL, Wilson CB: Persistence and late malignant transformation of childhood cere bellar astrocytoma. J Neurosurg 57:548-551, 1982. 2. Apuzzo ML, Dobkin WR, Zee CS, Chan JC, Giannotta SL, Weiss MH: Surgical considerations in treatment of intraventricular cysticercosis: An

analysis of 45 cases. J Neurosurg 60:400-407, 1984. 3. Ariens Kappers J: The development of the paraphysis cerebri in man with comments on its rela tionship to the intercolumnar tubercle and its significance for the origin of cystic tumors in the third ventricle. J Comp Neurol 102:425-510, 1955. 4. Baskin DS, Townsend JJ, Wilson CB: Lympho cytic adenohypophysitis of pregnancy simulating a pituitary adenoma: A distinct pathological en tity. Report of two cases. J Neurosurg 56:148153, 1982. 5. Beal MF, Kleinman GM, Ojemann RG, Hochberg FH: Gangliocytoma of the third ventricle: Hyperphagia, somnolence, and dementia. Neurology (NY) 31:1224-1228, 1981. 6. Boggan JE, Davis RL, Zorman G, Wilson CB: Intrasellar epidermoid cyst: Case report. J Neuro surg 58:411-415, 1983. 7. Borit A, Schmidek HH: Pineal tumors and their treatment. In Reiter RJ (ed): The Pineal Gland. New York, Raven, 1984. 8. Butler IJ, Borit A: Germ-cell tumors of the diencephalon. Prog Clin Biol Res 92:263-271, 1982. 9. Camins MB, Mount LA: Primary suprasellar atyp ical teratoma. Brain 97:447-456, 1974. 10. Davis RL: Pathologic tumor type and response to treatment. In Modern Concepts in Brain Tumor Therapy: Laboratory and Clinical Investiga tions, National Cancer Institute Monograph 46, 1977. 11. Davis RL, Liu H-C, Vestnys P, Resser KJ, Levin VA: Correlation of survival and diagnosis in supratentorial malignant gliomas. J Neurooncol 2:268, 1984. 12. Donat JF, Okazaki H, Gomez MR, Reagan TJ, Baker HL Jr, Laws ER Jr: Pineal Tumors: A 53year experience. Arch Neurol 35:736-740, 1978. 13. Edwards MSB, Davis RL: Tumor markers and cytology of cerebrospinal fluid: Impact on prog nosis. Cancer 56:1773-1777, 1985. 14. Giangaspero F, Osborne DR, Burger PC, Stein RB: Suprasellar papillary squamous epithelioma ("papillary craniopharyngioma"). Am J Surg Pathol 8:57-64, 1984. 15. Gradin WC, Taylon C, Fruin AH. Choroid plexus papilloma of the third ventricle: Case report and review of the literature. Neurosurgery 12:217220, 1983. 16. Gutin PH, Klemme WM, Lagger RL, MacKay AR, Pitts LH, Hosobuchi Y. Management of the unresectable cystic craniopharyngioma by aspiration through an Ommaya reservoir drainage system. J Neurosurg 52:36-40, 1980. 17. Haimovic 1С, Charer L, Hyman RA, Beresford HR: Metastasis of intracranial germinoma through a ventriculoperitoneal shunt. Cancer 48:10331036, 1981. 18. Haugh RM, Markesbery WR: Hypothalamic glioma and disturbances of behavior and endocrine and autonomic function. Arch Neurol 40:560-563, 1983. 19. Herrick MK, Rubinstein LJ: The cytological differ entiating potential of pineal parenchymal neo plasms (true pinealomas) Brain 102:289-320, 1979. 20. Hirano A, Ghatak NR: The fine structure of colloid

21. 22. 23. 24. 25.

26. 27. 28. 29. 30.

31. 32.

33. 34. 35. 36.

37. 38. 39. 40.

cysts of the third ventricle. J Neuropathol Exp Neurol 33:333-341, 1974. Hirano A, Matsui T, Zimmerman HM: The fine structure of epithelial cysts in the central nervous system. Neurol Surg {Tokyo) 3:639-646, 1975. Holbach K-H, Gullotta F: Zur Formalgenese intrasellarer und intraventrikularer Zysten. Neurochirurgia 20:186-188, 1977. Jellinger K: Primary intracranial germ cell tu mours. Ada Neuropathol [Berl) 25:291-306, 1973. Jooma R, Kendall BE: Diagnosis and management of pineal tumors. J Neurosurg 58:654-665, 1983. Leech RW, Freeman T, Johnson R: Colloid cyst of the third ventricle: A scanning and transmission electron microscopic study. J Neurosurg 57:108113, 1982. Liu H-C, Davis RL, Resser KJ, Vestnys P, Levin VA: Correlation of diagnosis of supratentorial ma lignant gliomas and survival. Cancer (in press). McDonald JA: Colloid cyst of the third ventricle and sudden death. Ann Emerg Med 11:365-367, 1982. Neuwelt EA, Lewy AJ: Disappearance of plasma melatonin after removal of a neoplastic pineal gland. NEnglJ Med 308:1132-1135, 1983. Petito CK, DeGirolami U, Earl KM: Craniophar yngiomas: A clinical and pathological review. Cancer 37:1944-1952, 1976. Pribil S, Boone SC, Waley R: Obstructive hydro cephalus at the anterior third ventricle caused by dilated veins from an arteriovenous malforma tion. Surg Neurol 20:487-492, 1983. Raimondi AJ, Tomita T: Pineal tumors in child hood. Epidemiology, pathophysiology, and sur gical approaches. Childs Brain 9:239-266, 1982. Rio-Hortega P del: The Microscopic Anatomy of Tumors of the Central and Peripheral Nervous System (translated by Pineda A, Russell GV, Earle KM). Springfield, IL, Charles С Thomas, 1962. Rorke LB: The cerebellar medulloblastoma and its relationship to primitive neuroectodermal tumors. J Neuropathol Exp Neurol 42:1-15, 1983. Rozario R, Adelman L, Grager RJ, Stein BM: Men ingiomas of the pineal region and third ventricle. Neurosurgery 5:489-495, 1979. Russell DS, Rubinstein LJ: Pathology of Tumours of the Nervous System, ed 4. Baltimore, Williams & Wilkins, 1977. Shuangshoti S, Phonprasert C, Suwanwela N, Netsky MG: Combined neuroepithelial (colloid) cyst and xanthogranuloma (xanthoma) in the third ventricle. Neurology [NY) 25:547-552, 1975. Szper I, Oi S, Leestma J, Kim KS, Wetzel NE: Xanthogranuloma of the third ventricle: Case re port. J Neurosurg 51:565-568, 1979. Teears RJ, Silverman EM: Clinicopathologic re view of 88 cases of carcinoma metastatic to the pituitary gland. Cancer 36:216-220, 1975. Trojanowski JQ, Tascos NA, Rorke LB: Malignant pineocytoma with prominent papillary features. Cancer 50:1789-1793, 1982. Wara WM, Fellows CF, Sheline GE, Wilson CB, Townsend JJ: Radiation therapy for pineal tu mors and suprasellar germinomas. Radiology 124:221-223, 1977.

41. Yanoff M, Davis RL, Zimmerman LE: Juvenile pilocytic astrocytoma ("glioma") of optic nerve: Clinicopathologic study of sixty-three cases. In Jacobiec FA (ed): Ocular and Adnexal Tumors. Birmingham, Aesculapius, 1978, ch 48. 42. Winston K, Gilles FH, Leviton A, Fulchiero A: Cerebellar gliomas in children. J Natl Cancer Inst 58:833-838, 1977. Additional Readings Adams JH, Graham DI Doyle D: Brain Biopsy: The Smear Technique for Neurosurgical Biopsies. Philadelphia, J.B. Lippincott Co., 1981. Allen JC, Nisselbaum J, Epstein F, et al: Alpha-fetoprotein and human chorionic gonadotropin determination in cerebrospinal fluid. J Neurosurg 51:368374, 1979. Antunes JL, Kvam D, Ganti SR, Louis KM, Good J: Mixed colloid cysts—Xanthogranulomas of the third ventricle. Surg Neurol 16:256-261, 1981. Apuzzo MLJ, Davey LM, Manuelidis ЕЕ: Pineal apoplexy associated with anticoagulant therapy. J Neurosurg 45:223-226, 1976. Arita N, Bitoh S, Ushio Y, et al: Primary pineal endodermal sinus tumor with elevated serum and CSF alphafetoprotein levels. J Neurosurg 53:244-248, 1980. Arita N, Ushio Y, Abekura M, et al: Embryonal carcinoma with teratomatous elements in the region of the pineal gland. Surg Neurol 9:198-203, 1978. Baggenstoss AH, Love JG: Pinealomas. Arch Neurol Psychiatry 41:1187-1206, 1939. Balagura S, Scarff JE: Teratoma of the pineal body: Classification of embryonal tumors of the pineal body. Report of a case of teratoma of the pineal body presenting formed teeth. Arch Surg 36:303-309, 1979. Bloom HJG: Primary intracranial germ cell tumors. Clin Oncol 2:233-257, 1983. Bochner SJ, Scarff JE: Teratoma of the pineal body: Classification of embryonal tumors of the pineal body presenting formed teeth. Arch Surg 36:303-309, 1938. Borit A: History of tumors of the pineal region. Am J Surg Pathol 5:613-619, 1981. Borit A, Blackwood W, Mair WGP: The separation of pineocytoma from pineoblastoma. Cancer 45:14081418, 1960. Borit A, Blackwood W: Pineocytoma with astrocytomatous differentiation. J Neuropathol Exp Neurol 38:253-248, 1979. Borit A: Embryonal carcinoma of the pineal region. J Pathol 97:165-168, 1969. .Britt RH, Silverberg GD, Enzmann DR, Hanbery JW: Third ventricular Choroid plexus arteriovenous malformation simulating a colloid cyst. J Neurosurg 52:246250, 1980. Brun A, Nils E: The pathogenesis of cerebral symptoms in colloid cysts of the third ventricle: A clinical and pathoanatomical study. Ada Neurol Scand 49:525-535, 1973. Buchsbaum HW, Colton RP: Anterior third ventricular cysts in infancy. J Neurosurg 26:264-266, 1967. Brodeur GM, Howarth CB, Pratt CB, Caces J, Hustu HO: Malignant germ cell tumors in 57 children and adolescents. Cancer 48:1890-1898, 1981.

Burger PC, Vogel SF: Neoplasms in the pineal region. In Surgical Pathology of the Nervous System and Its Coverings, ed 1- New York, John Wiley, 1976, pp 398407. Cashion EL, Young JM: Intraventricular craniopharyngioma. J Neurosurg 34:84-87. 1971. Chan RC. Thompson GB: Third ventricular colloid cysts presenting with acute neurological deterioration. Surg Neurol 1 9:358-362, 1983. Ciric I, Zivin I: Neuroepithelial (colloid) cysts of the septum pellucidum. J Neurosurg 43:69-73, 1975. Crain BJ. Bigner SH. Johnston WW: Fine needle aspiration biopsy of deep cerebrum: A comparison of normal and neoplastic morphology. Acta Cytol [Baltimore] 26:772-778. 1982. Dandy WE: Benign Tumors in the Third Ventricle of the Brain: Diagnosis and Treatment. Springfield. Charles С Thomas, 1933. Dayan AD, Marshall AHE, Miller AA, et al: Atypical teratomas of the pineal and hypothalamus. J Pathol Bacteriol 92:1-28. 1966. DeGirolami U, Schmidek H: Clinicopathological study of 53 tumors of the pineal region. J Neurosurg 39:455-462, 1973. Derek R. Jenkin T, Simpson JK, Keen CW: Pineal and suprasellar germinomas. J Neurosurg 48:99-107. 1978. Esteban HM. Villanueva JP, Bueno JG. Puentes MF: Primary intraventricular choriocarcinoma. Surg Neurol 11:21-23, 1979. Fox JL, Al-Mefty O: Suprasellar arachnoid cysts: An extension of the membrane of liliequist. Neurosur-gery 7:615-618. 1980. Fulling KH, Nelson JS: Cerebral astrocytic neoplasms in the adult: Contribution of histologic examination to the assessment of prognosis. Semin Diagnostic Pathol 1:152-163, 1984. Futrell NN. Osborn AG, Cheson BD: Pineal region tumors: Computed tomographic-pathologic spectrum. AJR 137:951-956. 1981. Ghatak NR, Hirano A, Zimmerman HM: Intrasellar germinomas: A form of ectopic pinealoma. J Neurosurg 31:670-675. 1969. ■ Giannotta SL, Pauli FF, Saeed M: Epidermoid cyst of the third ventricle. Surg Neurol 5:164-166, 1976. ■ Goldstein SJ, Wilson DD. Young AB, Guidry GJ: Craniopharyngioma intrinsic to the third ventricle. Surg Neurol 20:249-253, 1983. Grote E, Lorenz R, Vuiz O: Clinical and endocrinological findings in ectopic pinealoma and spongioblas-toma of the hypothalamus. Acta Neurochtr (Wien] 53:89-98. 1980. Herrick MK: Pathology of pineal tumors. In Neuwelt EA [ed): Diagnosis and Treatment of Pineal Region Tumors. Baltimore. Williams & Wilkins, 1984, pp 3160. Hill S. Martin EC, Ellison HWE: Carcinoembryonic antigen in cerebrospinal fluid of adult brain-tumor patients. J Neurosurg 53:627-632, 1980. Hirano A. Lena JF. Chung HD: Some new observations on intracranial germinoma. Acta Neuropathol [Berl] 32:103-113. 1975. Ho K-L, Rassekh ZS: Endodermal sinus tumor of the pineal region: Case report and review of the literature. Cancer 44:1081-1086, 1979. Hosoi K: Teratoma and teratoid tumors of the brain. Arch Pathol 9:1207-1219. 1930. Ito J, Kadekaru T, Hayano M, Kurita I, Okada K, Yosh-

ida Y: Meningioma in the tela choroidea of the third ventricle: CT and angiographic correlations. Neuroradiologu 21:207-211. 1981. Jane JA. Bertrand G: A cytological method for the diagnosis of tumors affecting the central nervous system. J Neuropathol Exp Neurol 21:400-409, 1962. Jenkins RDT. Simpson WJK, Keen CW: Pineal and suprasellar germinomas: Results of radiation treatment. J Neurosurg 48:99-107, 1978. Jennings MT, Gelman R. Hochberg F: Intracranial germ cell tumors: Natural history and pathogenesls. In Neuwelt EA (ed): Diagnosis and Treatment of Pineal Region Tumors. Baltimore, Williams & Wilkins, 1984, pp 116138. Jordan RM. Kendall JW. McClung M. et al: Concentration of human chorionic gonadotropin in the cerebrospinal fluid of patients with germinal cell hypo-thalamic tumors. Pediatrics 65:121-124, 1980. Kitano I, Yoneda K. Yamakawa Y, Fukui M, Kinoshita K: Huge cystic craniopharyngioma with unusual extensions. Neuroradiology 22:39-42, 1981. Kobayashi T, Kageyama N. Kida Y, et al: Unilateral germinomas involving the basal ganglia and thalamus. J Neurosurg 55:55-62, 1981. Koide O, Watanabe Y, Sato K: A pathological survey of intracranial germinoma and pinealoma in Japan. Cancer 45:2119-2130. 1980. Komrower GM: Precocious puberty in association with pineal seminoma. Arch Dis Child 49:822. 1974. Larson CP: Colloid (paraphysial) cyst of third ventricle with rupture into caudate nucleus and internal capsule. JNerv Ment Dis 91:557-565, 1940. Liwnicz BH. Henderson KS, Masukawa T. Smith KD: Needle aspiration cytology of intracranial lesions: A review of 84 cases. Acta Cytol [Baltimore) 26:779-786, 1982. Loftus CM, Marquardt MD, Stein BM: Hemangioblastoma of the third ventricle. Neurosurgy 15:67-72, 1984. Mair WGP, Warrington EK. Weiskrantz L: Memory disorder in Korsakoff's psychosis: A neuropathol-ogical and neuropsychological investigation of two cases. Brain 102:749-783. 1979. Malik GM, Horoupian DS. Boulos RS: Hemorrhagic (colloid) cyst of the third ventricle and episodic neurologic deficits. Surg Neurol 13:73-77. 1980. Markwalder TM, Markwalder RV. Markwalder HM: Meningioma of the anterior part of the third ventricle. J Neurosurg 50:233-235. 1979. McLean AJ: Pineal teratomas: With report of a case of operative removal. Surg Gynecot Obstet 61:523-533. 1935, Morris AA: The use of the smear technique in the rapid histological diagnosis of tumors of the central nervous system: Description of a new staining method. J Neurosurg 4:497-504. 1947. Ortega P, Malamud N. Shimkin MB: Metastasis to the pineal body. Arcfi Pathol 52:518-528. 1951. Pheline C. Engel P, Leguyader B. Tamic P: Meningioma of the third ventricle. J Neurol Neurosurg Psychiatry 40:1024, 1977. Puschett JB. Goldberg M: Endocrinopathy associated with pineal tumor. Ann Intern Med 69:203-219. 1968. Rodrigues EA. Lawson AAH: Meningioma of the third ventricle. Postgrad Med J 58:498-500. 1982. Rubery ED. Wheeler TK: Metastases outside the cen-

tral nervous system from a presumed pineal germinoma. J Neurosurg 53:562-565, 1980. Russell DS: The pinealoma: Its relationship to teratoma. J Pathol Bacteriol 56:146-150, 1944. Russell DS, Krayenbuhl H, Cairns H: The wet film technique in the histological diagnosis of intracra-nial tumors: A rapid method. J Pathol 45:501-505, 1937. Sambasivan M, Nayar A: Epidermoid cyst of the pineal region. J Neurol Neurosurg Psychiatry 37:1333-1335, 1974. Sato M, Ushio Y, Arita N, Mogami H: Hypothalamic hamartoma: Report of two cases. Neurosurgery 16:198206, 1985. Servo A, Porras M, Jaaskinen J: Diagnosis of ependymal intraventricular cysts of the third ventricle by computed tomography. Neuroradiology 24:155-157, 1983. Sheithauer BW: Neuropathology of pineal region tumors. Clin Neurosurg 32:351-383, 1985. Shuangshoti S, Roberts MP, Netsky MG: Neuroepithelial (colloid) cysts. Arch Pathol 80:214-224, 1965. Shuangshoti S, Netsky MG: Neuroepithelial (colloid) cysts of the nervous system. Neurology {NY) 16:887-903, 1966. Smith RR, Sanford RA: Disorders of the deep cerebral veins. In Kapp JP, Schmidek HH (eds): The Cerebral Venous System and Its Disorders. New York, Grune and Stratton, 1984. Solarski A, Panke ES, Panke TW: Craniopharyngioma

in the pineal gland. Arch Pathol Lab Med 102:490-491, 1978. Stachura I, Mendelow H: Endodermal sinus tumor originating in the region of the pineal gland. Cancer 45:2131-2137, 1980. Stefanko SZ, Manschot WA: Pinealoblastoma with retinoblastomatous differentiation. Brain 102:321-332, 1979. Sweet WH: Recurrent Craniopharyngiomas: Therapeutic alternatives. ClinWeurosurg 27:206-229, 1980. Szper I, Oi S, Leestma J, Kimm KS, Wetzel NE: Xanthogranuloma of the third ventricle. J Neurosurg 51:565568, 1979. Takeuchi J, Handa H, Nagata I: Suprasellar germinoma. J Neurosurg 49:41-48, 1978. Tomasello F, Albanese V, Bernini FP, Picozzi P: Choroid plexus papilloma of the third ventricle. Surg Neurol 16:69-71, 1981. Tavcar D, Robbo SJ, Chapman P: Endodermal sinus tumor of the pineal region. Cancer 45:2646-2651, 1980. Vuia O: Embryonic carcinosarcoma (mixed tumour) of the pineal gland. Neurochirurgia (Stuttg) 23:47-54, 1980. Weinberger LM, Boshes B: Paraphysial cysts of the third ventricle. Surgery 13:368-384, 1943. Willems J-S: Aspiration biopsy cytology of tumors of the central nervous system and base of the skull. In Linsk, JA, Franzen S (ed): Clinical Aspiration Cytology. Philadelphia, Lippincott, 1983.

10 Tumor Markers in Third Ventricular Neoplasms Henry H. Schmidek, M.D., Adam Borit, M.D., and Steven L Wald, M.D.

The ability to diagnose cancer by a laboratory assay of blood, urine, or cerebrospinal fluid has been a desirable objective for many years. Ideally such a marker would be specific to a disease and positive only in the presence of neoplasia. The myeloma proteins discovered in 1846 are closest to this ideal. Interest in tumor markers was re-

kindled by the discovery of а-fetoprotein (AFP) in 1963 and of carcinoembryonic antigen in 1965. Among primary central nervous system (CNS) tumors a variety of biochemical markers have been identified (Table 10.1). A disease-specific marker is important because it would allow differentiation of the benign from the malignant tumors, help define the tumor's histological subtype, allow monitoring of the efficacy of treatment by changes in marker levels, and permit early detection of asymptomatic recurrences. Tumor markers also have the advantage over, for example, exfoliative cytology in being quantifiable. Once it is possible to identify and quantitatively determine tumor marker signals that will reveal the presence of tumor, the localization and treatment of the neoplasms bearing markers can be explored for diagnosis and potentially for treatment. For example, the in vivo localization of tumor growth can be accomplished using antibody manufactured to the marker, labeled with radioisotope, injected, and then localized using radioscintigraphy or radioimmunoscintigraphy. This technique, which has been applied to the

diagnosis of asymptomatic lymp node metastases, is potentially useful in the diagnosis of germ cell tumors using antibody to AFP or the beta subunit of human chorionic gonadotropin (bHCG) or to malignant neural tumors using the polyamines. An example of this approach is the recent report of monoclonal antibody imaging of normal-sized retroperitoneal lymph nodes for the detection of metastatic colon cancer in lymph nodes judged normal in size and appearance on computerized tomographic (CT) scanning and at operation (36). Therapy may involve conjugation of a chemotherapeutic agent to a tumorspecific antibody to direct chemotherapeutic agents or isotopes to cancer cells in vivo. All cell markers identify one or more specific cell constituents by immunological, enzymatic assay, or analytical chemical methods, although a specific compound detectable by chemical methods may also serve as an antigen detectable by immunological methods. As a practical matter almost all markers are detected by immunological methods and their major application is in tumor diagnosis and in research. Markers discussed in this chapter are those that have established diagnostic implications in neuroncology and are specifically relevant to lesions within or adjacent to the third ventricle. Metastatic tumors will not be discussed; they are rare in this region and would necessitate a fragmentary description of many markers. The primary neoplasms of the pineal region

are germ cell tumors, pineal parenchymal tumors, or (a small number of) mesenchymal tumors and cysts (9, 10, 48, 49). Germ Cell Tumors Human germ cell tumors account for between one-half and two-thirds of tumors in the pineal region and they comprise a group of interrelated neoplasms. The totipotential cells of origin of teratomas may remain undifferentiated, resulting in an embryonal carcinoma, or embark upon lines of differentiation of extraembryonic elements, either vitelline (yolk sac tumor) or tropho-blastic (choriocarcinoma). The entire group of these tumors includes germinoma (atypical teratoma), embryonal carcinoma, choriocarcinoma, endodermal sinus tumor, teratocarcinoma, and mixed tumors consisting of combinations of these elements. The germinoma is the most com-

mon of these tumors arising in the CNS (23-25, 29, 31, 56). All of these tumors resemble the germ cell tumors of the gonads and other extragonadal sites biologically and histopathologically, although they originate within the pineal gland, the suprasellar area ("ectopic pinealoma"), or the anterior and posterior third ventricle. Germ cell tumors originate in diencephalic centers that are synaptically entrained in the regulation of go-nadotropic activity. Gonadotropins are implicated in determining the site of origin of intra-cranial germ cell tumors and seem to function as carcinogenic inducers. The neuroendocrine events of puberty have an activating influence in the expression of malignant behavior among these neoplasms. The question of the multicen-tric origin of the anteriorposterior third ventricle tumors seeding from an anterior or posterior third ventricular germinoma remains unresolved. When the tumor is present in the anterior third ventricle it tends to fill the floor and invade the walls of the ventricle and may extend into the infundibulum, pituitary gland, and optic nerves. Rarely an intracranial germ cell tumor originates outside the diencephalic region, such as in the cerebellar vermis. The last 10 years have seen a marked increase in our knowledge of various tissue-specific antigens. These substances are capable of provoking the synthesis of an antibody by the b cells of the immune system. Compounds vary in their ability to elicit the initial or subsequent production of antibody, and species vary in their ability to respond to the initial or subsequent antigenic challenge. Very few small molecular weight chemicals are able to elicit an antibody response by themselves, but larger compounds (such as poly-peptides) are usually adequate for this purpose. Polypeptides or proteins containing a polysac-charide moiety are especially good antigens. An antigenic site (epitope) of a chemical compound consists of only a few monomers such as amino acids or monosaccharides of macromolecules. The same antigenic site may be present in more than one specific molecule, which can lead to antigenic cross reactivity and misidentification of molecules. Radioimmunoassay (RIA) is the current technique most widely used to measure the concentration of antigens in blood or cerebrospinal fluid. The method combines the immunological method with that of quantitative detection of radioactivity (55, 62). In performing a RIA, the antigen sought is highly purified and radiolabeled usually with iodine-125 or iodine Bl. A known quantity of this labeled antigen is placed in a test tube and used

as the standard. A quantity of the patient's serum or blood that contains an unknown quantity of the (unlabeled) antigen is placed in a test tube, a known quantity of antiserum (antibody) to the sought antigen is then added to the mixture of standard and unknown, and the immunological reaction is allowed to take place. In this reaction, the primary antibody contained in the antiserum reacts with both the labeled and unlabeled antigens. With manipulation of the quantities of the reactants used, it can be arranged that some of the labeled antigen will remain free, i.e., will not react with the primary antibody. Thereafter, the antibody is bound and the free antigens are separated. This separation is usually achieved by adding a secondary antibody that is specific against the primary antibody to the mixture. This secondary antibody is produced in a different species from that in which the primary antibody is produced. The addition of the secondary antibody to the reaction mixture precipitates only the complex of primary antibody with the antigen. Thus, the free antigen remains in the supernatant fluid and the two types of antigen (bound and free) can be separated. Instead of a secondary antibody, other means of effecting this separation can be used, such as staphylococcal protease A, which combines with the FC portion of the immunoglobulin G (IgG). The labeled antigen, either in the precipitate or in the supernatant fluid, can be measured and the equation (total radiolabeled antigen — free labeled antigen = bound labeled antigen) gives the quantities of labeled antigen in free and bound forms. Standard curves can be constructed by systematically varying the amount of unlabeled antigen (unknown sample) to fixed amounts of labeled antigen and primary antiserum (antibody). The ratio of the labeled antigen that is bound by the primary antibody in the presence of varying quantities of unlabeled antigen and in the absence of unlabeled antigen gives the quantity of unlabeled antigen in the unknown sample. The two major specific markers of germ cell tumors are AFP and b HCG. Although present in some germinomas, placental alkaline phospha-tase is not a fully established marker for this group of tumors (40). Both AFP and b HCG are measured by RIA of serum and cerebrospinal fluid. AFP AFP is a 70-kilodalton glycoprotein that is thought to be the fetal equivalent of albumin, initially discovered in the serum of fetal mice and in the serum of adult mice with hepatocellular

cancer. AFP is produced by the yolk sac and the fetal liver and is suppressed around the time of birth but reappears in the serum in the presence of hepatoma or teratocarcinoma. The tumor cells producing AFP are the malignant counterparts of the cell populations producing AFP during development (3, 50). After fertilization and the morula stage of development, differentiation proceeds to the formation of the inner and outer germ cell layers. The inner germ cell layer forms the endoderm, which further differentiates into the visceral and parietal layers. The cells of the visceral endodermal layer are characterized by a thin basement membrane and the synthesis of AFP on Day 7 of mouse development. The gene responsible for directing this synthesis is silent until that time and is then activated, resulting in the production of AFP until birth when it is then turned off, only to be derepressed when certain cells undergo malignant transformation later in life. The visceral endodermal tissues give rise to the fetal yolk sac, the liver, the gastrointestinal epithelium, and their corresponding tumors, the yolk sac (endodermal sinus) tumor, the hepatoma, and gastrointestinal adenocarcinomas (2, 27,41, 54). Occasionally, other tumors express this compound. Of the tumors originating within the CNS, only the yolk sac tumor, embryonal carcinoma, and mixed germ cell tumors produce it in appreciable quantities. AFP and other compounds, such as alpha-1-antitrypsin, are particularly localized in the hyaline globules characteristic of the yolk sac tumor; however, one must be aware of metastatic tumors producing this compound, particularly the extracranial germ cell tumors. Human Chorionic Gonadotropin (HCG) HCG is a hormone produced by cells of the placenta, especially the syncytiotrophoblasts. This marker is used in establishing the diagnosis of pregnancy and in making the diagnosis of trophoblastic disease. HCG is a 30-kilodalton glycoprotein composed of an alpha and a beta subunit. The alpha subunit of HCG is identical to the alpha subunit of the pituitary gonadotropins, follicle-stimulating hormone and luteinizing hormone (LH) and of thyroid-stimulating hormone. The beta subunit of each of these hormones is unique and characteristic. Identification of HCG is performed through RIA of the beta subunit in serum, urine, or cerebrospinal fluid. Aside from its production during pregnancy, choriocarcino-mas, adenocarcinomas of the lung, hepatomas, and malignant melanomas may also occasionally produce this marker. Some neoplasms produce

the alpha, others the beta, and still others both HCG subunits. In the CNS, choriocarcinomas, embryonal carcinomas, certain mixed germ cell tumors, and certain metastatic tumors manufacture b HCG and indicate the existence of these neoplastic cells (17, 18, 51). The clinical experience with the use of tumor markers in the diagnosis and subsequent management of third ventricular neoplasms is limited by the relatively infrequent occurrence of these tumors; however, significant experience exists with the use of these markers in the management of patients with testicular tumors, 95% of which arise from primordial germ cells identical to those found in and around the third ventricle. As with the intracranial tumors these neoplasms occur as germinomas, malignant teratomas, or mixed cell tumors. AFP and b HCG were surveyed in a series of patients with testicular tumors. Even in cases of pure seminomas (germinomas) the b HCG level is occasionally elevated, a finding explained by pathological studies demonstrating that some germinomas contain syncytiotropho-blastic cells. Probably 10 to 20% of "pure" testicular seminomas produce b HCG. Sano found that three of five cases of intracranial germinomas secreted b HCG (47). In a study of 10 patients with intracranial germinomas in whom plasma AFP and b HCG was measured, Pomarede et al. found the AFP elevated in 1 patient, the b HCG elevated in 3 patients, and the LH elevated in 1 patient, probably because of cross reactivity with HCG (42). Assay of the secretion of AFP and b HCG by nonseminomatous testicular tumors has shown that approximately 40% of these cases demonstrate elevations of one of these markers, and that during the course of the disease the two markers, if present together, do not necessarily quantitatively parallel each other. One reason for this observation is the difference in the half-lives of the two markers. The half-life of AFP is 5 days and that of b HCG is 30 hours. AFP may therefore remain elevated for days to weeks after radical tumor removal whereas the b HCG level will fall rapidly. In addition, the production of a marker substance is only one of the biological activities of these cancers, and tumor cells that have ceased to produce these markers are not necessarily dead. The level of marker may drop to normal before the clinical and radiographic disappearance of the tumor (34). The investigation of pineal region tumors and of ectopic pinealomas includes measurement of both AFP and b HCG levels in serum and cerebrospinal fluid. The frequency of occurrence of pos-

itive values among this group of heterogeneous tumors is not known. A germ cell tumor may be present and both of these markers may be negative on repeated examinations. When positive, the presence of AFP is specific for a malignant neoplasm possessing endodermal sinus elements, although it has been reported in association with a "pure" germinomatous tumor (42). b HCG may be positive in some pineal region germinomas, although they are particularly characteristic of one of the other germ cell tumors. In Jooma's series of pineal tumors there is a case of a germinoma with elevated blood and cerebrospinal fluid b HCG levels that then fell in response to radiotherapy and with CT resolution of the lesion (28). Among six male patients with other germ cell tumors, markers were sought in three cases and were present in two of these patients. One patient had elevation of both AFP and b HCG, and the other had elevation of only AFP. In two of these six cases precocious puberty occurred in conjunction with the germ cell tumors probably because the b HCG produced by the malignant teratoma stimulates the interstitial cells of the testis to secrete androgens. In one of these cases the b HCG, testosterone, and AFP levels in the blood were elevated and subsequently fell during radiotherapy. Non-Germ Cell Tumors: Pineal Parenchymal Tumors The non-germ cell tumors of the pineal region include neoplasms that arise from the pineal parenchymal cell, the pinealocyte (pineocytoma, pineoblastoma, and mixed forms of these tumors), or from glial cells (astrocytoma, ependymoma, and mixed forms of these tumors). Pineal parenchymal cells contain the enzymes that mediate the synthesis of norepinephrine, serotonin, and melatonin from tryptophan. N-Acetyltrans-ferase and hydroxyindole-Omethyltransferase (HIOMT) are involved in this process, and N-acetyltransferase is the rate-limiting enzyme for melatonin formation. The enzyme HIOMT is a specific marker for the site of melatonin formation. Some controversy exists concerning the specificity of melatonin as a marker because this hormone is also synthesized by the retina, in red blood cells, peripheral nerves, gastrointestinal tract, the hypothalamus, Choroid plexus, and by the malignant melanoma. Within this limitation there is considerable interest in this subject because of methodological improvements that now allow the accurate assay of melatonin in spite of its small molecular weight and many structural

analogs. Melatonin quantification is performed by bioassay, spectrofluorometric assay, RIA, or the gas chromatography-mass spectrometric technique. Each approach exploits a different biological, chemical, or physical property of melatonin and each assay has a characteristic sensitivity and specificity. Pending the establishment of a definitive assay method, the best documentation of the accuracy of a given assay method is a scatter plot with linear regression analysis of paired data for the test method and a reference method based on an alternative analytical principle. At present RIA methods yield values in agreement with those based on bioassay or on gas chromatography-mass spectrom-etry, although of the different methods the latter technique is probably the most accurate. Melatonin levels in plasma, serum, and cerebrospinal fluid tend to be similar. Melatonin circulates in the blood of humans, is excreted in the urine, and demonstrates a diurnal rhythm, being maximal during darkness (11, 66). Alterations in the pattern of melatonin secretion in the presence of a third ventricle tumor have been described in only a few cases. Neuwelt and Lewy (38) reported the preand postoperative 24-hour melatonin assays of a 17-yearold boy with obstructive hydrocephalus and a partially cystic, calcified pineal tumor. The tumor enhanced on the infusion CT scan, and AFP and b HCG assays were negative. Examination of the cerebrospinal fluid for malignant cells was also negative. After removal of a low grade pineal region astrocytoma, the plasma melatonin level, assayed by gas chromatography-mass spectrom-etry, became unmeasurable. There are also occasional reports of altered (elevated) levels of melatonin in the serum of patients with a pineoblastoma (4). Melatonin has been identified in a few pineoblastomas, in the rare medulloblastoma, and in tumors that may be primary CNS melanomas. Kennaway et al. (30) described a patient with a pathologically verified pineoblastoma in whom melatonin could not be detected

in the serum on several examinations and another patient with a pineal germinoma in whom the melatonin level was elevated (59). HIOMT, the enzyme required for the last step in melatonin synthesis, has been demonstrated in vitro within the cells of a metastatic pineal parenchymal tumor, indicating that melatonin and probably other hormones are produced by some of these tumors. Other Neuroectodermal Tumors The neuroectodermal neoplasms around the third ventricle resemble their counterparts throughout the CNS and their markers are the same as for other neuroectodermal tumors (Table 10.2). The neurofilament proteins and glial fibrillary acidic protein (GFAP) are useful in diagnosis; others, such as neuron-specific enolase (14-3-2 protein) and the SI00 proteins, are less specific and contribute little further diagnostic information (37, 52, 53, 64, 67). Neurofilament Proteins Three proteins with molecular weights of 200, 145, and 68 kilodaltons together form the orga-nelle known as the neurofilament. In vitro the 68-kilodalton component assembles into morphologically normal intermediate-sized filaments by itself. Neurofilament is one of the five, approximately 10-nm-thick, intermediate-sized filaments thought to be specific for neurons and certain neuroendocrine cells. Within neurons it is the axons that stain most consistently, indicating the distribution of most of the neuronal intermediate filaments in these cells. That the Bodian silver stain stains mostly neurofilaments has now been proven using antibodies to the neurofilament proteins on sodium dodecyl sul-fate-polyacrylamide gel electrophoresis of rat whole spinal cord homogenate (15, 16). The predominantly axonal localization of neurofilaments is useful as a more specific method of detecting axons. These proteins are rarely found in primitive neuroectodermal tumors and,

when detected, are seen only in a small percentage of the tumor cells. In tumors in which differentiation toward neurons is well advanced, such as ganglioneuroblastoma, ganglioneuroma, or ganglioglioma (including the hypothalamic gan-glionic hamartoma), the positivity is more extensive and corresponds to the degree of histological differentiation. Luteinizing hormone-releasing hormone (LHRH) antigen has been demonstrated in the neuronal cells (perikarya and axons) of the rare hypothalamic ganglionic hamartoma. The LHRH is manufactured in neurons of various nuclei of especially the infundibular and mamillary regions of the hypothalamus. The axons of these neurons, just as in the case of the other hypothalamic releasing factors, secrete their produce into the portal venous system of the median eminence from whence it reaches the cells of the anteior pituitary. The presence of LHRH in these tumors correlates with the known occurrence of precocious puberty in some of the affected patients. Such a tumor nodule may serve as an "accessory" hypothalamus. This tumor is histopathologically similar to gangliogliomas at other sites and thus contains the neurofilament (neurons) and GFAP (glial) markers (42a). In addition, neurofilaments have also been identified in pheochromocytomas and in the oat cell carcinoma of the lung. Teratomas may contain neu-rofilamentpositive neurons or endocrine cells (15, 39,42a). Neurofilament proteins are specific to the cells and tumors of the neuronal series, as well as to certain cells and tumors of the diffuse endocrine system, and the pathological categorization of these can be important in the study of tumor types. To date assay of these proteins in tissue fluids has not proven of value in the diagnosis of third ventricular neoplasms. Glial Fibrillary Acidic Protein (GFAP) GFAP is a single intracellular protein with a molecular weight of approximately 50 kilodal-tons. It is the specific constituent of the 10-nm-thick, intermediate-sized filaments of the various types of astrocytes and ependymal cells. Its specific function in these cells is unknown. GFAP is detectable in the absence of gliofibrillo-genesis, perhaps in a soluble form. Astrocyte-like cells are also present in the pineal and pituitary and astrocytomas occur there. Some schwannian cells may also contain this protein (5-8, 70). The applicability of the peroxidase-antiperoxi-dase (PAP) method to formalin-fixed and paraffin-embedded tissues has led to its widespread

use to identify neural antigens such as the GFAP. This technique uses a primary antibody against the specific tissue antigen to be identified as well as a link (secondary) antibody generated against the immunoglobulins of the species in which the primary antibody was produced. The link antibody is followed by the PAP complex, the antiperoxidase component of which is generated in the same species in which the primary antibody was produced. In addition, before using the primary antibody, normal serum from the same species from which the link antibody was obtained (to reduce nonspecific staining) and H2O2 (to inactivate endogenous peroxidase) are used sequentially. Tryp-sinization enhances the sensitivity of detection of certain antigens. Because the PAP complex is colorless, certain chromogens are used to visualize the complex. The PAP method is more sensitive than immunofluorescence, and routinely processed tissues can be used in this method; thus it is more useful than immunofluorescence when a frozen section is not available. After immunoperoxidase staining has been completed, the tissue section can be counterstained with hematoxylin, thereby demonstrating the nuclear morphology (12, 13, 20, 26, 33, 60, 61). An alternative method to PAP, the ABC method, is also becoming common (22, 57, 69). GFAP is present in astrocytes, ependymal cells, the normal pineal parenchyma, all types of astrocytomas, and ependymomas. In these situations vimentin, the mesenchymal intermediate filament, is also present. There is a very rough inverse relationship between the quantity of GFAP present and the tumor's histological grade of malignancy. In low grade astrocytomas around the third ventricle many of the tumor cells contain appreciable amounts of GFAP, whereas in glioblastomas many of the cells may be GFAP-negative. In the region of the third ventricle the pilocytic astrocytoma is the most common form and cells in these tumors may contain Rosenthal fibers (8, 70, 71). Two patterns of GFAP positivity have been described in astrocytomas. In the more common pattern, the perikarya as well as the processes contain immunoreactive material; in the other pattern, mostly the cell processes and only a few perikarya stain for GFAP. In ependymomas, cells of all of the structural components may contain GFAP: ependymomatous tubules, gliovascular rosettes, and the diffuse areas. Many of the neural tumors in the vicinity of the third ventricle, aside from the astrocytoma and ependymoma, contain a few GFAP-positive cells. These

include the Choroid plexus papilloma, oligodendroglioma, ganglioglioma, primitive neuroectodermal tumors (medulloblastoma, medulloepi-thelioma, pineoblastoma, neuroblastoma, reti-noblastoma), capillary hemangioblastoma, and craniopharyngioma. In oligodendrogliomas, astrocytes are frequently intermixed with the tumor oligodendrocytes. One of the components of a ganglioglioma is astrocytic. Among the primitive neuroectodermal tumors, differentiation in an astrocytic direction occasionally occurs. The GFAP-positive cells in capillary hemangioblas-tomas and Craniopharyngiomas are thought to be reactive cells included within the tumors. Teratomas may contain GFAP-positive astrocytes and ependymal cells. Studies of GFAP in the cerebrospinal fluid of patients representing a wide spectrum of neurological disease have shown elevations in association with a variety of both neoplastic and non-neoplastic states. No specific disorder is consistently associated with elevated cerebrospinal fluid levels, although the only tumors to show cerebrospinal fluid levels of greater than 500 mg/ ml are glioblastomas and anaplastic astrocytomas (21). Various compounds such as carbonic anhy-drase or galactocerebroside have been suggested as markers of the oligodendroglial cell, but none has been proven clinically useful (32, 43). Current methods allow myelin basic protein (MBP) and other components of myelin to be detected in the cell body of oligodendroglia-synthesizing myelin sheath but not in the adult oligodendro-cyte or in the oligodendroglioma. Few studies have detailed the distribution of myelin basic protein in CNS tumors. Studies in experimentally induced schwannomas, rat cell gliomas, cultures derived from human oligodendrogliomas, and an astrocytoma have shown either no MBP or inconsistent results. Jones and Whitaker found the cerebrospinal fluid of patients with a variety of brain tumors to contain no MBP except in one case when an intratumoral hemorrhage had occurred before lumbar puncture (7, 27, 58, 68). The Polyamines The polyamines, putrescine, spermidine, and spermine, are small cationic molecules with unfortunate names that are related to nucleic acid metabolism and cellular proliferation (Fig. 10.1). Increased polyamine biosynthesis occurs coincident with periods of cell proliferation or the release of polyamines from dead or dying tumor cells. Extensive studies particularly by the Brain Tumor Research Group at the University of Cal-

ifornia, San Francisco, and by others (14, 19, 35, 46) have shown that, although the value of these indicators is limited by the frequent occurrence of false-positive and falsenegative values, they have a distinct and specific usefulness in highly proliferative tumors such as glioblastoma multiforme and medulloblastoma located in or adjacent to the ventricular system. It is among these neoplasms that consistent elevations of cerebrospinal fluid polyamine levels can be demonstrated to parallel the progression or regression of a tumor, frequently preceding other evidence of tumor recurrence. Tumor tissue concentrations of polyamines seem to be indicative of a tumor's malignancy. Even though no data presently exist for third ventricular germ cell tumors or pineal parenchymal tumors, some of these would be expected to produce high levels of polyamines that could then be used to monitor these tumors. Using an amino acid analyzer Marton et al. carried out serial polyamine determinations by the ion-exchange chromatographic separation method (35). In 16 patients with medulloblastoma, 75 cerebrospinal fluid polyamine determinations were performed and gave no false-positive values and only 1 false-negative value. The polyamine values provided an excellent correlation with the patient's condition as evaluated by the neurological examination, the CT and iso-topic scans, cerebrospinal fluid cytology, and myelography. In all cases the putrescine levels were elevated above that of other disease states, although elevated putrescine and sperimidine levels have been reported in the presence of hydrocephalus and encephaloceles. Polyamine determination is of potential value in pineal region tumors because they fulfill the criteria of proximity to the ventricular system, rapid proliferation, and the potential of recurrence within the spinal fluid pathways (1). Recently, radiolabeled putrescine has been used as a probe in an experimental brain tumor model. In a preliminary report by Volkow et al. (65) using transplanted rat gliosarcoma, the in vivo uptake of [14C]putrescine was 35 times greater in tumor than in normal brain and 95% of the injected polyamine cleared

within 5 minutes. These findings demonstrate the potential efficacy of putrescine as a positron emission tomography tracer for brain tumors. References 1. Albright AL, Marton LJ, Lubich WP, Reigel DH: CSF polyamines in childhood. Arch Neurol 40:237-240, 1983. 2. Allen TC, Nisselbaum J, Epstein F, Rosen G, Schwartz MK: Alpha fetoprotein and human cho rionic gonadotrophin determination in cerebro spinal fluid: An aid to the diagnosis and manage ment of intracranial germ-cell tumors. J Neuro surg 51:368-374, 1979. 3. Alpert E, Drysdale JW, Isselbacher KT, Schur PH: Human a-fetoprotein: Isolation, characterization, and demonstration of microheterogeneity. J Biol Chem 247:3792-3801, 1972. 4. Barber SG, Smith JA, Hughes RC: Melatonin as a tumor marker in a patient with pineal tumor. Br MedJ 2:328, 1978. 5. Bignami A, Dahl D, Rueger DC: Glial fibrillary acidic protein (GFA) in normal neural cells and in pathological conditions. Adv Cell Neurobiol 1:285-310, 1980. 6. Bissell MG, Rubinstein LJ, Bignami A, Herman MM: Characteristics of the rat C-6 glioma main tained in organ culture systems: Production of glial fibrillary acid protein in the absence of gliofibrillogenesis. Brain Res 8:77-89, 1974. 7. Borit A, Mclntosh GC: Myelin basic protein and glial fibrillary acidic protein in human fetal brain. Neuropathol Appl Neurobiol 7:279-287, 1981. 8. Borit A, Richardson EP Jr: The biological and clinical behavior of pilocytic astrocytomas of the optic pathways. Brain 105:161-187, 1982. 9. Borit A, Schmidek HH: Pineal tumors and their treatment. In Reiter RJ (ed): The Pineal Gland. New York, Raven Press, 1984, pp 323-344. 10. Borit A: History of tumors of the pineal region. Am JSurgPathol 5:613-620, 1981. 11. Ebadi M: The regulation of the synthesis of mel atonin and its significance to neuroendocrinology. In Reiter RJ (ed): The Pineal Gland. New York, Raven Press, 1984, pp 1-37. 12. Eng LF, DeAmond SJ: Immunochemistry of the glial fibrillary acidic protein. Progr Neuropathol 5:19-39, 1983. 13. Epstein JI, White CL III, Mendelsohn G: Factor VIII related antigen and glial fibrillary acidic pro tein immunoreactivity in the differential diagno sis of central nervous system hemangioblastomas. Am J Clin Pathol 81:285-292, 1984. 14. Fulton DS, Levin VA, Lubich WP, Wilson CB, Marton LJ: Cerebrospinal fluid polyamines in pa tients with glioblastoma multiforme and anaplas tic astrocytoma. Cancer Res 40:3293-3296, 1980. 15. Gambetti P, Autilio-Gambetti L, Papasozomonos SC: Bodian's silver method stains neurofilament Polypeptides. Science 213:1521-1522, 1981. 16. Geisler N, Weber K: Self-assembly in vitro of the 68,000 molecular weight component of the mam malian neurofilament triplet proteins into inter mediate-sized filaments. JMol Biol 151:565-571, 1981. 17. Goldstein DP: Chorionic gonadotropin. Cancer 38:453-459, 1976.

18. Halmi NS, Krieger DT: Immunostaining of human syncytiotrophoblast with anti-B-lipotropin: An il lustration of the limitations of immunocytochemistry. J HistochemCytochem 30:538-540, 1982. 19. Harik SI, Sutton CH: Putrescine as a biochemical marker of malignant brain tumors. Cancer Res 39:5010-5015, 1979. 20. Hautzer NW, Wittkuhn JF, McCaughey WTE: Trypsin digestion in immunoperoxidase staining. J Histochem Cytochem 28:52-53, 1980. 21. Hayakawa T, Morimoto K, Ushio Y, Mori J: Levels of astroprotein in cerebrospinal fluid of patients with brain tumors: An attempt at immunochemical diagnosis of gliomas. J Neurosurg 52:229233, 1980. 22. Hsu SM, Raine L, Fanger H: The use of antiavidin antibody and avidMibiotin-peroxidase complex in immunoperoxidase techniques. Am J Clin Pathol 75:816-821, 198lV 23. Illmensee K, Stevens^LC: Teratomas and chi meras. Scientific Am 240:120-132, 1979. 24. Jellinger K: Primary intracranial germ cell tu mors. Ada Neuropathol 25:291-306, 1973. 25. Jennings MT: Pluripotential germinal tumor tis sue as a source of transplantable graft tissue. Appl Neurophysiol 47:51-56, 1984. 26. Jessen KR, Mirsky R: Glial cells in the enteric nervous system contain flial fibrillary acidic pro tein. Nature 286:736-737, 1980. 27. Jones TR, Bigner DD: Central nervous system tumor markers. In Sell S, Wahren В (eds): Human Cancer Markers. Clifton, NJ, Humana Press, 1982, pp 381-422. 28. Jooma R, Kendall BEW: Diagnosis and manage ment of pineal tumors. J Neurosurg 58:654-665, 1983. 29. Kawakami Y, Yamada O, Tabuchi K, Ohmoto T, Nishimoto A: Primary intracranial choriocarci noma. J Neurosurg 53:369-374, 1980. 30. Kennaway DJ, McCulloch G, Matthews CD, Sea mark RF: Plasma melatonin, lutenizing hormone follicle-stimulating hormone, prolactin, and corticoids in two patients with pinealoma. J Clin Endocrinol Metab 459:144-145, 1979. 31. Koide O, Watanabe Y, Sato K: A pathological sur vey of intracranial germinoma and pinealoma in Japan. Cancer 45:2119-2130, 1980. 32. Kumpulainen T: Immunohistochemical localiza tion of human carbonic anhydrase isoenzymes. Ann NY Acad Sci (USA) 429:359-368, 1984. 33. Kurisaka M, Moriyasu N, Kitajima K: Immuno histochemical studies of brain tumors associated with precocious puberty: A preliminary report of correlation between tumor secreting hormone and Leydig cells in precocious puberty. Neurol Med Chir (Tokyo) 19:675-682, 1979. 34. Lange PH: Testicular cancer markers. In Sell S, Wahren В (eds): Human Cancer Markers. Clifton, NJ, Humana Press, 1982, pp 259-273. 35. Marton LJ, Edwards JS, Levin VA, Lubich WP, Wilson CB: Predictive value of cerebrospinal fluid polyamines in medulloblastoma. Cancer Res 39:993-997, 1979. 36. Moldofsky PJ: Monoclonal antibody imaging of normal sized retroperitoneal lymph nodes for de tection of metastatic tumor. N Engl J Med 311:106-109, 1984. 37. Nakamura Y, Becker LE, Marks A: Distribution of immunoreactive S-100 protein in pediatric brain

tumors. J Neuropathol Exp Neurol 42:136-146, 1983. 38. Neuwelt EA, Lewy AJ: Disappearance of plasma melatonin after removal of a neoplastic pineal gland. NEnglJMed 308(19): 1132-1135, 1983. 39. Osborne M, Altmannsberger M, Shaw G, Schauer A, Weber K: Various sympathetic derived human tumors differ in neurofilament expression: Use in diagnosis of neuroblastoma, ganglioneuroblastoma and pheochromocytoma. Virchows Arch Cell Pathol 40:141-156, 1982. 40. Paiva J, Damjanov I, Lange PH, Harris H: Immu nohistochemical localization of placental-like al kaline phosphatase in testis and germ-cell tumors using monoclonal antibodies. Am J Pathol 111:156-165, 1982. 41. Palmer PE, Safau H, Wolf HJ: Alpha-antitrypsin and alphafetoprotein: Protein markers in endo dermal sinus (yolk sac) tumors. Am J Clin Pathol 65:575-582, 1976. 42. Pomarede R, Czernichow P, Finidori J: Endocrine aspects and tumoral markers in intracranial ger minomas: An attempt to delineate the diagnostic procedure in 14 patients. J Pediatr 101:374-378, 1982. 42a.Price RA, Lee PA, Albright AL, Ronnekleiv OK, Gutai JP: Treatment of sexual precocity by removal of a luteinizing hormone-releasing hormone-secreting hamartoma. JAMA 251:2247-2249, 1984. 43. Raff MD, Mirsky R, Fields KL, Lisak RP, Dorfman SH, Silberberg DH, Gregson NA, Leibowitz S, Ken nedy M: Galactocerebroside is a specific cell-sur face antigenic marker for oligodendrocytes in cul ture. Nature 274:813-816, 1978. 44. Rubinstein LJ, Brucher JM: Focal ependymal dif ferentiation in Choroid plexus papillomas. Ada Neuropathol 53:29-33, 1981. 45. Rubinstein LJ: Tumors of the Central Nervous System. Supplement to Fascicle 6, Second Series, Atlas of Tumor Pathology. Washington DC, Armed Forces Institute of Pathology, 1982, pp 521-530. 46. Russell DH: The role of the polyamines, putrescine, spermidine, and spermine in normal and malignant tissues. Life Sci 13:1635-1647, 1973. 47. Sano K: Pineal region tumors: Problems in pa thology and treatment. Clin Neurosurg 30:59-91, 1982. 48. Schmidek HH (ed): Pineal Tumors. New York, Masson Publishing USA, 1977. 49. Schmidek HH, Waters A: Pineal masses: Clinical features and management. In Wilkins RH (ed): Neurosurgery. New York, McGraw-Hill (in press). 50. Scott RW, Vogt TF, Croke ME, Tilghman SM: Tis sue-specific activation of a cloned fetoprotein gene during differentiation of a transfected em bryonal carcinoma cell live. Nature 310:562-567, 1984. 51. Sell S, Wahren В (eds): Human Cancer Markers. Clifton, NJ, Human Press, 1982. 52. Sharp G, Osborne M, Weber K: Occurrence of two different intermediate filament proteins in the same filament in situ within a human glioma cell line. Exp Cell Res 141:385-395, 1982. 53. Shaw G, Weber K: The distribution of the neuro-

54.

55. 56.

57. 58.

59. 60. 61. 62.

63.

64.

65. 66.

67.

68. 69.

70.

71.

filament triplet proteins within individual neurones. Exp Cell Res 136:119-125, 1981. Shinomiya Y, Toya S, Iwata T, Hosoda Y: Radioim munoassay of alpha-fetoprotein in children with primary intracranial tumors. Childs Brain 5:450-458, 1979. Skelly DS, Brown LP, Besch PK: Radioimmunoas say. ClinChem 19:146-186, 1973. Stachura I, Mendelow H: Endodermal sinus tumor originating in the regions of the pineal gland: Ultrastructural and immunohistochemical study. Cancer 45:2131-2137, 1980. Sternberger LA: Immunocytochemistry, ed 2. New York, Wiley, 1979. Sternberger NH, Quarles RH, Itoyama Y, Webster H: Myelin associated glycoprotein demonstrated immunocytochemically in myelin and myelinforming cells of developing rat. Proc Natl Acad Sci USA 76:1510-1514, 1979. Tapp E: Melatonin as a tumor marker in a patient with pineal tumor. Br MefiJj2:635-636, 1978. Taylor CF: Immunoperoxfidase techniques. Arch Pathol Lab Med 102:113Л121, 1978. Taylor CR, Kledzik G: Iirhnunohistologic tech niques in surgical pathology—spectrum of "new" special stains. Hum Pathol 12:590-596, 1981. Vaitukaitis JL, Braunstein GD, Ross GT: A radioimmunoassay which specifically measures hu man chorionic gonadotropin in the presence of human luteinizing hormone. Am J Obstet Gynecol 113:751-758, 1972. Velasco ME, Dahl D, Koessmann U, Gambett P: Immunohistochemical localization of glial fibril lary acidic protein in human glial neoplasms. Cancer 45:484-494, 1980. Vinores SA, Bonnin JM, Rubinstein LJ, Marangos PJ: Immunohistochemical demonstration of neu ron-specific enolase in neoplasms of the CNS and other tissues. Arch Pathol Lab Med 108:536540, 1984. Volkow N, Goldman SS, Flamm ES, Wolf AP, Brodie JD: Labeled putrescine as a probe in brain tumors. Science 221:673-674, 1983. Waldhauser F, Lynch HJ, Wurtman RJ: Melato nin in human body fluids: Clinical significance. In Reiter RJ (ed): The Pineal Gland. New York, Raven Press, 1984, pp 345-370. Weiss SW, Langloss JM, Enzinger FM: Value of S-100 protein in the diagnosis of soft tissue tu mors with particular reference to benign and ma lignant Schwann cell tumors. Lab Invest 49:299308, 1983. Whitaker JN, Lisak RP, Bashir RM, Fitch OH, Seyer JM, Krance R, Lawrence JA, Ch'ien LT, O'Sullivan P: Ann Neurol 7:58 1980. Wood GS, Warnke R: Suppression of endogenous avidin-binding activity in tissues and its relevance to biotin-avidin detection systems. J Histochem Cytochem 29:1196-1204, 1981. Yen SH, Fields KL: Antibodies to neurofilament, glial filament and fibroblast intermediate fila ment proteins bind to different cell types of the nervous system. J Cell Biol 88:115-126, 1981. Yung WKA, Luna M, Borit A: Vimentin and glial fibrillary acidic protein in human brain tumors. J Clin Neuroncol (in press).

11 Radiology of Third Ventricular Lesions Eddie Kwan. M.D., Samuel M. Wolpert, M.B., S.Ch., Steven P. Smith, M.D., and Michael T. Modie, M.D.

The third ventricle is a narrow funnel-shaped midline cerebrospinal fluid (CSF)-containing compartment lined with ependymal cells in direct communication anterosuperiorly with each lateral ventricle through the foramen of Monro and posteriorly with the aqueduct of Sylvius. It has a roof, a floor, and an anterior, posterior, and two lateral walls. A multitude of disease processes involve the third ventricle, leading to characteristic alterations of the anatomy. Thus, a clear understanding of the normal radiological anatomy is necessary to characterize properly a pathological process and formulate a differential diagnosis. Currently available techniques for examination of the third ventricle include: high resolution computerized tomography (CT), with or without intravenous contrast administration, metriza-mide CT-ventriculography and cisternography, and selective cerebral angiography. In most cases there is no need for metrizamide administration; high resolution CT together with sagittal reformatting are adequate. In problem cases, metrizamide studies can provide excellent detail of the third ventricle. Angiography retains a major role in the diagnosis of suspected aneurysms, vascular malformations, and certain tumors. In neonates, the anterior fontanel serves as an acoustic window for examination with real time ultrasound. Ventricular size and position, the internal sonographic architecture of lesions (i.e., cystic versus solid), and parenchymal hemorrhage are well displayed. Magnetic resonance imaging (MRI) has devel-

oped rapidly since the publication of diagnostic quality images in 1980 (44). Although still undergoing intensive development and investigation, MRI offers several unique features: increased sensitivity and tissue discrimination compared to x-ray CT (7), freedom from potential side effects of ionizing radiation (6), direct mul-tiplanar (e.g., axial, sagittal, or coronal) imaging capability, and lack of the beam-hardening artifact encountered with posterior fossa CT. Radiological Anatomy Roof The roof of the third ventricle, extending from the foramen of Monro anteriorly to the supra-pineal recess posteriorly, has a slightly upward convex contour. It is defined anatomically by the fornix, the tela choroidea, the paired internal cerebral veins and the great vein of Galen (best seen on lateral views in the late venous phase of selective internal carotid angiograms). The body and crura of the fornix and the hip-pocampal commissure (which interconnects the crura) are not well seen on CT scans either in the axial plane or with sagittal reformatting because of the lack of contrast between these structures and the surrounding neural tissue. However, the anterior limbs of the fornix (the columns), highlighted by CSF in the lateral and third ventricles, are well seen on axial scans at the level of the foramen of Monro. MRI displays the fornix in its entirety on sagittal images. The septum pelluci-dum, which is attached inferiorly to the superior

surface of the body of the fornix, is tallest anteriorly and shortest posteriorly. The septum pel-lucidum is best seen in the coronal plane on CT scans. At the posterior end of the septum pellu-cidum, the crura of the fornix and the hippocam-pal commissure fuse with the splenium of the corpus callosum. The medial posterior choroidal artery (MPChA) arises from the peduncular segment of the posterior cerebral artery (PCA), extends posteriorly and medially, and enters the lateral portion of the quadrigeminal cistern. It then contributes to the blood supply of the quadrigeminal plate and pineal gland. Crossing anterior and lateral to the pineal gland, the MPChA penetrates through the layers of the tela choroidea, supplies the Choroid plexus just inferior to the roof of the third ventricle, and lies adjacent to the internal cerebral veins. The roof of the third ventricle is outlined during infancy and childhood by the cistern of the velum interpositum, a triangular or diamond-shaped space formed by infolding of arachnoid over the third ventricular roof. This CSF-con-taining space represents a superior and anterior extension of the quadrigeminal cistern. The cistern of the velum interpositum becomes obliterated by adulthood. The body of the fornix defines the cistern of the velum interpositum superiorly, and the epithelial roof and Choroid plexus of the third ventricle and adjacent tela choroidea define it infe-riorly. The lateral margin of the third ventricular roof is defined by the choroidal fissure, bounded by the lateral margin of the body of the fornix and the superomedial surface of the thalamus. The crus fornix and the pulvinar are posterior. The

choroidal fissure is in direct communication with the retropulvinar cistern and can occasionally be identified in the axial plane during metrizamide CT-cisternography. Anterior Wall The anterior wall of the third ventricle extends from the foramen of Monro above to the optic chiasm below. The superior margin is bounded by the columns of the fornix and the rostrum of the corpus callosum. Immediately below the foramen of Monro, the anterior commissure indents the anterior wall and continues inferiorly for about 10 mm as the lamina terminalis. The anterior wall terminates at the optic recess, just above the optic chiasm. These anatomical landmarks are well seen on either metrizamide ventriculography or cisternography in the lateral projection. CT of the anterior wall region can also be accomplished with thin sections (2 mm or less) in the axial plane, followed by sagittal reformatting (Fig. 11.1 A). MRI is capable of imaging the anatomy of the anterior third ventricle in the sagittal plane without the use of intrathecal contrast or the need for computer reformatting. The thalamostriate vein, septal vein, and superior choroidal vein unite to join the internal cerebral vein; this junction, the venous angle, is U-shaped and angiographically marks the foramen of Monro (Fig. 11.IB and C). In some cases, however, the junction lies behind the foramen— a false venous angle. The Choroid plexus of the lateral ventricle and the distal branches of the MPChA also pass through the foramen of Monro. An inconstant vessel, the precallosal artery, may arise from the anterior communicating artery (ACoA) and cross cephalad in front of the lamina

Figure 11.1. A. Normal third ventricle. Sagittal reformatted image of a metrizamide CTcisternogram outlining the optic (o) and infundibular (i) recesses of the third ventricle. Note the diaphragma sellae (arrowheads) above the pituitary gland (white arrow). The anterior wall of the third ventricle, mamillary bodies (m), and tegmentum of the midbrain (t) are also well visualized.

Figure 11.1. В and С. Normal third ventricle. terminalis to supply the rostrum of the corpus callosum. The ACoA also gives rise to multiple small perforating arteries supplying the lamina terminalis. The precallosal artery and the perforating branches from the ACoA are rarely seen on routine selective internal carotid arteriography. The A1 segments of the anterior cerebral arteries (ACAs) cross over the optic chiasm on either side of the optic recess. These arterial relationships can be best appreciated on thin

section direct coronal CT through the optic chiasm. Floor The floor extends from the chiasm posteriorly to the aqueduct of Sylvius. The external landmarks outlining the floor from anterior to posterior include the optic chiasm, infundibulum, tuber cinereum, mamillary bodies, posterior perforated substance, and tegmentum of the mid-

brain. The infundibular recess of the third ventricle indents the infundibulum of the pituitary gland (Fig. 11.1 A). The chiasm, optic tract, infundibulum, and mamillary bodies are best seen on metrizamide CTcisternography, whereas the optic and infundibular recesses are best imaged with metrizamide ventriculography in the lateral view. The posterior communicating artery (PCoA), the P1 segment of the PCA, and the apex of the basilar artery lie below the optic tract. Multiple small perforating arteries arise from these vessels to supply the structures along the floor of the third ventricle. These perforating branches can be seen routinely on magnification vertebral arteriography. The anterior choroidal artery (AChA) provides a vascular supply to the optic tract and the optic radiations. The hypophysis is supplied by two vessels: the inferior and superior hypophyseal arteries. The inferior hypophyseal artery is a branch of the meningohypophyseal trunk of the precavernous internal carotid artery (IСA), whereas the superior hypophyseal artery is a small branch of the supraclinoid ICA. The terminal branch of the superior hypophyseal artery also supplies the tuber cinereum. The superior and inferior hypophyseal arteries are not seen normally, even on lateral magnified selective internal carotid arteriography, and their visualization indicates a pathological process in the parasellar region or the development of collateral pathways in cases of vascular occlusive disease. The only major venous structure that abuts the third ventricular floor is the anterior pontomesencephalic vein (APMV), coursing along the floor of the ventricle, the anterior surface of the tegmentum, and the belly of the pons, finally draining superiorly and posteriorly into the posterior mesencephalic vein. The venous phase of selective vertebral arteriograms demonstrates the APMV. Posterior Wall The suprapineal recess, extending between the tela choroidea and the superior surface of the pineal gland, forms the posterosuperior boundary of the third ventricle. Below the suprapineal recess, the pineal gland extends posteriorly into the quadrigeminal cistern. The pineal recess indents the gland as a notch dividing the gland into an upper compartment attached to the habenu-lar commissure and a lower compartment attached to the posterior commissure. The aqueduct of Sylvius defines the posteroinferior limits of the third ventricle. Metrizamide ventriculog-

raphy delineates the suprapineal and pineal recesses. If calcified, the habenular commissure and the pineal gland can be demonstrated on plain films or CT scans. The habenular commissure calcification has a characteristic С shape, whereas the pineal gland calcification is usually irregular and speckled. Occasionally, the pineal calcification is shell-like because of the development of a small cyst within the pineal gland. The major vessel supplying the posterior third ventricle is the MPChA. On a lateral vertebral arteriogram, the MPChA has a characteristic M configuration and is located between the thalamoperforating arteries anteriorly and the lateral posterior choroidal artery posteriorly. The deep venous system of the supratentorial compartment is located both anterior and posterior to the pineal gland. The most important venous structures in this region are the paired internal cerebral veins, which course posteriorly along the roof of the third ventricle (Fig. 11.IB and C). These veins deviate slightly from the midline as they approach the pineal gland, joining to form the great vein of Galen beneath the splenium of the corpus callosum. Originating at the lateral angle of the body of the lateral ventricle, the direct lateral vein drains medially into the internal cerebral vein, just anterior to the pineal gland. Another important landmark, the medial atrial vein, traverses the choroidal fissure and drains into the internal cerebral vein just posterior to the pineal gland. Other major venous structures that converge in the pineal region include the precentral cerebellar vein (draining superiorly), the basal vein of Rosenthal (draining posteromedially), and the internal occipital vein (draining anteromedially into the great vein of Galen). Both selective internal carotid and vertebral arteriograms must be obtained to demonstrate all of these venous structures. Contrast-enhanced CT with sagittal reformatting demonstrates the relationship of the pineal gland to the major neighboring venous structures. Subtle lesions in the periaqueductal region may require metrizamide CT-ventriculography and CTcisternography. Lateral Walls The lateral walls of the third ventricle are lined by ependyma and are bordered by the hypothalamus inferiorly and the thalami laterally. Several commissures bridge the third ventricle and form parts of its boundary. The largest of these is the massa intermedia, situated in the upper half of

tion axial CT scanning, the columns of the fornix and the massa intermedia are often seen. Cross Sectional Imaging Techniques In adults, contrast-enhanced CT is performed after the intravenous administration of contrast material containing 42 g of iodine. In children, intravenous contrast is administered at a dose of 2 ml of 30% contrast material per pound of body weight. For metrizamide CT-ventriculography, the volume and concentration of metrizamide depends on the size of the ventricular system. For normal-sized ventricles, a dose of 3 to 5 ml of metrizamide at a concentration of 170 mg of iodine per ml is adequate. Massive hydrocephalus requires the use of a larger volume of contrast in a higher concentration. Contrast may be instilled via ventricular puncture or directly into an indwelling ventricular shunt catheter. For metrizamide CT-cisternography, one performs lumbar puncture with a 22 gauge needle and then 3 to 5 ml of metrizamide at a concentration of 170 mg of iodine per ml are injected. Because metrizamide possesses a higher specific gravity than CSF, the contrast agent flows cephalad into the basal cisterns as the patient assumes the Trendelenberg position. CT scanning may commence within minutes of contrast instillation. Because of the low dose of contrast agent, metrizamide CT-cisternography may be performed safely as an outpatient procedure. Neonatal ultrasound scanning is performed with a portable real time sector scanner through the anterior fontanel, with the use of 5-MHz transducer. The transducer may be swept anter-oposteriorly to obtain optimal coronal sections and is angled laterally to produce parasagittal images. Cranial ultrasound is technically feasible until closure of the anterior fontanel, which generally occurs at 9 to 12 months of age. Several excellent reviews of basic MRI physics are available (6, 23, 64, 72). The proton (1H), by virtue of its abundance and physical characteristics, is best suited for imaging available paramagnetic species found in biological tissues (93). Protons act like spinning tops and tend to line up in the presence of a strong external magnetic

a high signal intensity. B. A T2-weighted spin-echo image demonstrates reversal of the relative signal intensities of CSF, white matter, and gray matter. Figure 11.2. Normal axial MRI scans. A. A heavily Tiweighted inversion recovery image demonstrates excellent discrimination of the lower intensity gray matter and the higher intensity white matter. Lower intensity structures are dark; higher intensity structures are white. CSF and cortical bone signals are low to absent. Fat in the diploic space and scalp produces

Figure 11.3. Normal sagittal and coronal MRI scans. A. A midline sagittal image from a heavily T1-weighted spinecho sequence demonstrates excellent delineation of the aqueduct of Sylvius. Note the detailed depiction of normal structures: corpus callosum, fornix, optic chiasm, mamillary bodies, quadrigeminal plate, brain stem, fat in the marrow space of the clivus, and fourth ventricle. B. A coronal T1-weighted image provides sharp contrast between the high intensity optic chiasm and the adjacent low intensity of CSF. Note the absence of signal in the lumen of the cavernous and supraclinoid portions of the ICAs (arrowheads), with high intensity from the surrounding slowly flowing blood in the cavernous sinuses (particularly on the reader's right). The pituitary gland (white arrows) lies

field, producing within the tissue sample a macroscopic vector called magnetization. A preselected radiofrequency pulse imparts energy to these spinning, aligned protons. The pulse produces two important outcomes: first, the spins of the individual protons become synchronized (i.e., become coherent) and, second, the magnetization vector is tipped away from alignment with the external magnetic field. When the radiofreqeuncy pulse stops, the protons "relax" back toward alignment with the applied external field. In this relaxation phase, some of the absorbed energy is radiated from the tissue as a radiofrequency signal. Protons emit the energy at different radiofrequencies depending on preselected position-encoding spatial gradient fields. By analysis of the emitted radiofrequency spectrum, the intensity and location of the emitted radiofrequency is established and an image is constructed. The intensity of the emitted radiofrequency energy (signal intensity) depends upon four factors: mobile proton density, longitudinal relaxation time (T1), transverse relaxation time (T2), and, in the case of vascular structures, flow rate. Most important in determining final image contrast are the intrinsic tissue characteristics, T1 and T2. T1 denotes the time required to achieve maximal net magnetization and is a function of interactions between individual protons and their surrounding molecular environment, or lattice. T1 is therefore referred to as the "spin-lattice relaxation time." T2 measures decay of the radiofrequency signal from excited tissue and relates closely to loss of coherence within the sample. Because this dephasing of proton spins occurs as a result of interactions with other spinning nuclei, it is referred to as the "spin-spin relaxation time." Regrowth of the net magnetization vector in the direction of the external magnetic field is not equivalent to loss of coherence within the tissue sample and, therefore, knowledge of T1 does not predict the T2 value of the same tissue. Pulsing sequences such as inversion recovery and spinecho are available to accentuate differences in T1 or T2 selectively within the sample (Figs. 11.2 and 11.3). In a "T1-weighted image," for example, regional differences in T1 are highlighted and tissues with short T1 values generate stronger signals than tissues with longer T1 values. Inversion recovery sequences have the

superior to the aerated sphenoid sinus and lamina dura of the sellar floor. Sinus air and compact bone produce little or no signal.

greatest sensitivity to differences in tissue T1 values. By selection of pulse sequence parameters, heavily T1weighted spin-echo imaging is possible. In conventional gray scale display, regions with strong signals appear white and areas of weak or absent signals appear darker. Conversely, structures with a long T2 will generate strong signals (and thereby appear white) on heavily T2-dependent spin-echo sequences, whereas structures with a shorter T2 generate weak signals and appear darker. Lesions along the Roof of the Third Ventricle Congenital Lesions Cavum of the Septum Pellucidum, Cavum Vergae, Cavum of the Velum Interpositum Congenital anomalies represent a major portion of lesions along the roof of the third ventricle. Cavum of the septum pellucidum (CSP), cavum vergae (CV), and cavum of the velum interpositum (CVI) are the most common. These anomalies are typically discovered incidentally on CT scanning (see Fig. 11.5). The CSP and the CV lie in the same plane above the third ventricle and commonly occur together. Embryologically, involution of these two cavae commences posteriorly and extends rostrally; therefore, a CV never occurs in the absence of a CSP. On CT scanning, the cavae have the attenuation of CSF. The CSP is located anterior and the CV posterior to the vertical plane formed by the columns of the fornix. On both axial and coronal scans, the CV always lies superior to the internal cerebral vein. This anatomical relationship differentiates the CV from cystic lesions of the third ventricle, which always lie inferior to the internal cerebral vein. True cysts may arise from the septum pellucidum (22), blocking the foramen of Monro and causing obstructive hydrocephalus. Ventricular obstruction is a differential feature distinguishing these true cysts from a simple CSP. The first two cases of pathological cysts in the septum pellucidum were reported by Dandy in 1931. Since then, other cystic lesions in this region have been described, including neuroepithelial (colloid) cysts, cysticerosis cysts, ependymal cysts, cystic gliomas, superior extension of cystic Craniopharyngiomas, epidermoids, and dermoid cysts. A CVI, as already described, is a normal structure in the infant brain. However, it can be confused with other CSFcontaining lesions in the

pineal region. Dilatation of the suprapineal recess in conjunction with either massive hydrocephalus or partial agenesis of the corpus callosum can mimic a CVI. Metrizamide CT-ventric-ulography defines the exact anatomy. True arachnoid cysts of the velum interpositum also occur, distorting the quadrigeminal cistern and to a lesser degree the ambient cistern. An arachnoid cyst in this region can be differentiated from a CVI by metrizamide CT-cisternography: an arachnoid cyst will generally not fill with contrast, whereas a CVI will. These lesions may even extend superiorly and laterally along the choroidal fissure into the trigone of the lateral ventricle. Agenesis of the Corpus Callosum Agenesis of the corpus callosum is the next most common congenital anomaly that occurs in the roof of the third ventricle. It can be associated with hydrocephalus, interhemispheric cysts, posterior fossa extraaxial cysts, lipomas and other midline craniofacial anomalies (41). Embryologically, the degree of callosal defect depends on the stage at which developmental arrest occurs. Complete agenesis occurs before the 12th gestational week. Partial agenesis occurs later during intrauterine life and is thought to result from a vascular or inflammatory insult. Because the corpus callosum develops in an anteroposter-ior direction, partial agenesis often involves the body and the splenium of the corpus callosum and leaves the rostrum intact. However, isolated agenesis of the rostrum has been reported. The primary radiological method for the diagnosis of agenesis of the corpus callosum is CT in the axial and coronal planes (Fig. 11.4). The frontal horns and bodies of the lateral ventricles are widely separated, and the foramina of Monro are enlarged. Separation of the posterior part of the bodies of the lateral ventricles occurs in cases with isolated posterior callosal agenesis. Enlargement of the occipital horns and suprapineal recess is easily demonstrable. The anterior interhemispheric fissure abuts the anterior wall of the third ventricle because of the absence of the rostrum of the corpus callosum. The medial wall of the lateral ventricle is concave because of protrusion by the cingulate gyrus and Probst's bundle (the latter represents ectopic longitudinal callosal fibers that failed to cross the midline during embryogenesis). The third ventricle is high-riding, situated between rather than below the lateral ventricles. Hypoplasia of the falx is often associated with agenesis of the corpus callosum.

Figure 11.4. Agenesis of the corpus callosum. A. Contrast-enhanced CT scan. Part of the anterior interhemispherie fissure is seen in the usual location of the rostrum of the corpus callosum [arrows). The anterior horns are slightly separated. B. Coronal reformatted image demonstrates the high-riding third ventricle (white arrows), which separates the adjacent frontal horns (black arrowheads). The high density above the third ventricle represents the internal cerebral veins. Lipomas involving the genu of the corpus callosum occur in approximately 50% of cases of agenesis of the corpus callosum. Calcification along the periphery of the lipoma is often associated with this anomaly. Angiography is seldom necessary to diagnose agenesis of the corpus callosum. If carried out in

problem cases the vertically oriented "wandering" pericallosal artery and the high-riding, separated internal cerebral veins are distinctive (97). Diencephalic Cyst If the septum pellucidum is also absent, the roof of the third ventricle may balloon dorsally

and result in an interhemispheric or diencephalic cyst (41) (best imaged on direct coronal CT scans). The CT appearance of a diencephalic cyst associated with agenesis of the corpus callosum may mimic an interhemispheric arachnoid cyst. Septooptic Dysplasia Another rare congenital lesion deserving attention is septooptic dysplasia, first described by deMorsier. Patients with this condition often present clinically with hypothalamic or pituitary dysfunction (predominantly deficiency of growth hormone [GH], thyroid-stimulating hormone, adrenocorticotropic hormone, and antidiuretic hormone). CT findings are combined optic nerve and optic chiasm hypoplasia, absence of the septum pellucidum, squaring of the frontal horns with flattening of the roof of the third ventricle, and inferior pointing of the anteroinferior margin of the frontal horns (70). These changes are best seen in the direct coronal projection. The pituitary stalk may be enlarged when there is associated diabetes insipidus. Enlargement of the optic recess (a remnant of the primitive optic ventricle) can be demonstrated on sagittal reformatting of thin section metrizamide CT-cister-nography. Hypothalamic dysfunction and typical CT findings need not be present in all patients with this syndrome. Absence of the septum pellucidum is found in approximately 75% of the reported cases. Other midline anomalies including harelip, cleft palate, and absence of the olfactory bulb and tract have been reported in patients with septooptic dysplasia. Neoplastic Lesions Tumors along the roof of the third ventricle usually manifest themselves by obstruction of the foramen of Monro. Colloid Cyst Colloid cysts are unsually spherical or ovoid and range in diameter from a few millimeters to 9 cm. Although generally originating from the roof of the third ventricle, colloid cysts arising from the septum pellucidum have been reported, with associated widening of the septum (21). These lesions are usually homogeneous and hy-perdense on noncontrast CT (36) (Fig. 11.5). However, cysts isodense with either the surrounding brain or CSF have also been reported. Most colloid cysts demonstrate minimal enhancement on postcontrast scans. High attenuation on a precontrast scan is presumably due to desquamated secretory products from the cyst wall with high protein content and hemosiderin.

If the suspected colloid cyst is isodense, metrizamide ventriculography will establish the diagnosis. Glioma Glioma involves the roof of the third ventricle, usually as an inferior extension of a primary tumor involving the corpus callosum. The CT appearance is the same as that of gliomas involving the lateral walls of the third ventricle (discussion follows under "Lesions Affecting the Lateral Walls of the Third Ventricle"). Subependymal Giant Cell Astrocytoma Subependymal giant cell astrocytomas usually occur in the walls of the lateral ventricles after malignant degeneration of hamartomas associated with tuberous sclerosis (5). The roof of the third ventricle is involved if the tumor grows inferiorly. Radiological description of this entity is presented under "Intraventricular Lesions." Subependymoma Subependymomas are rare in the septum pellucidum (45), constituting 5% of all reported cases (84). They are histologically benign, lobu-lated, and sharply demarcated from surrounding tissue. Growth is by expansion toward the ventricle, in contradistinction to gliomas, which enlarge by infiltration of surrounding structures. Subependymomas are associated with less edema than are gliomas. Subependymomas are isodense or slightly hyperdense on noncontrast scans. The enhancement pattern of this tumor is rather homogeneous, but not intense. The larger tumors (4 to 5 cm in diameter) frequently demonstrate cyst formation, focal calcification, and hemorrhage due to vessel degeneration. Lymphoma Primary central nervous system (CNS) lymphoma, an uncommon tumor (less than 1.5% of primary brain tumors), most often occurs in patients with primary immunodeficiency or im-munosuppression secondary to organ transplantation. Primary CNS lymphomas involve the roof of the third ventricle by inferior extension of tumor originating in the corpus callosum. In addition, commonly encountered tumor foci occur, particularly in the basal ganglia, as well as in the thalamus, periventricular white matter, and cerebellar vermis (32). The lesions are often mul-tifocal. On the noncontrast CT scan, the lesions are usually isodense or slightly hyperdense. Edema and mass effect are significantly less than with metastases of comparable size. The

Figure 11.5. Colloid cyst. A. A noncontrast axial CT scan demonstrates an intrinsically dense colloid cyst splitting the leaves of the septum pellucidum adjacent to the columns of the fornices. Note the incidental cavum of the septum pellucidum [arrows) immediately anterior to the cyst. B. Sagittal reformatted image depicts the mass near the foramen of Monro (anterior is to the reader's left). enhancement pattern is homogeneous, well circumscribed, and intense (similar to meningiomas). Low density centers within the lesions

are extremely rare, in contradistinction to their frequency in primary gliomas. Angiography often shows a relatively avascular mass.

Epidermoid and Dermoid Cysts Epidermoid and dermoid cysts rarely involve the roof of the third ventricle. Epidermoid cysts outnumber dermoid cysts among intracranial tumors and have a tendency to occur away from the midline structures. The radiological differential diagnostic features are described under

"Lesions Affecting the Anterior Wall of the Third Ventricle." Vascular Lesions Aneurysm of the Vein of Galen Aneurysm of the great vein of Galen is a rare midline arteriovenous malformation. The clini-

Figure 11.6. Vein of Galen aneurysm (with associated dural AV malformation). A. Postcontrast CT scan image demonstrates a large, densely enhancing mass in the posterior third ventricle and straight sinus. Note multiple enhancing arterial feeding vessels. B. Lateral view of the arterial phase of an intraarterial left ICA digital angiogram demonstrates enlarged feeding arteries from the meningohy-pophyseal trunk [three rowheads), cavernous branches of the ICA [single arrowhead), pericallosal artery [single arrow), and dural branches of the posterior cerebral artery [double arrows). C. Lateral view from the late arterial phase of a selective vertebral arteriogram demonstrates a huge vein of Galen aneurysm, continuing inferiorly as a posterior fossa venous varix. The sella and clivus have been outlined in this subtraction image. D. Anteroposterior view from the late arterial phase of the vertebral arteriogram documents the midline location of the vein of Galen aneurysm. Note the tortuous basilar artery [arrow).

cal presentation depends on the age of presentation and on the volume of blood shunted through the malformation (28). In newborns, cardiac failure and cranial bruits are the most common presenting signs. In older children with a lesser degree of arteriovenous (AV) shunting, the clinical picture is that of an enlarging head due to obstructive hydrocephalus secondary to aqueductal compression by the dilated venous varix. Cranial bruits, seizures, and subarachnoid hemorrhage may also be present. Radiologically, this entity can be divided into two groups: (a) malformations supplied by a few large arterial feeders with almost direct shunting between the varix and the enlarged vein of Galen and (b) angiomatous malformations originating in the thalamus and basal ganglia, with a slow-flowing shunt draining into the dilated galenic venous system (59). On a noncontrast CT scan, a fusiform midline structure of homogeneously increased attenuation is located along the superior and posterior aspects of the third ventricle. The lesion connects with an enlarged torcular by a dilated midline channel representing the straight sinus. The arterial feeders to the dilated venous varix prominently enhance in a characteristic serpiginous pattern (Fig. 11.6A). Other CT findings include hydrocephalus caused by compression of the aqueduct by the venous varix or defective CSF re-sorption secondary to intracranial venous hypertension and patchy areas of low attenuation distinct from the malformation, related to ischemia resulting from steal of blood away from the surrounding brain parenchyma by the malformation. Even though contrast-enhanced CT demonstrates both enlarged arterial feeders and draining veins, bilateral selective internal carotid and vertebral arteriography is needed for surgical planning. Vein of Galen malformations are commonly supplied by distal branches of the АСА, AChA, PCA, posterior choroidal arteries, and thalamoperforating arteries (Fig. 11.6B). Spontaneous thrombosis of aneurysms of the great vein of Galen is unusual, potentially mimicking a pineal region tumor on CT. Aneurysms can diminish in size after thrombosis. Inflammatory Lesions Cysticercosis The diagnosis of a cysticercotic cyst involving the septum pellucidum (30) should be considered if (a) there are other calcific lesions within the brain parenchyma, (b) the basal cisterns are dis-

torted because of meningeal involvement, (c) intraventricular cysts are present (99), and (d) communicating hydrocephalus is present. Cisternal and ventricular involvement can be confirmed with metrizamide CT-cisternography or ventric-ulography. Subacute cysticercosis, i.e., dying larvae, can occasionally show ringlike or nodular enhancement after intravenous contrast administration (49). Postinflammatory Gliosis of the Foramen of Monro Rarely, a web of gliosis in response to inflammatory disease can occur in the foramen of Monro, leading to unilateral obstruction of the lateral ventricle. Neither contrast-enhanced CT nor metrizamide CT-ventriculography dependably outline the gliosis in these cases, although the latter test will confirm obstruction of the foramen. Lesions Affecting the Anterior Wall of the Third Ventricle Neoplastic Lesions Meningioma Meningiomas arising from the planum sphenoidale, olfactory groove, medial aspect of the sphenoid ridge, diaphragma sellae, and tubercu-lum sellae can all extend superiorly to deform the anterior wall of the third ventricle. Meningiomas of the planum sphenoidale may cause hyperostosis or blistering of the bone, pneumo-sinus dilatans (an upward bowing of the cortex of the posterior ethmoid or anterior sphenoid air cells) (47), or pneumatization of the anterior cli-noid processes. These findings are easily discernible on plain films or bone window CT images. Meningiomas only rarely cause bony erosion. On noncontrast CT, meningiomas commonly appear hyperdense because of microscopic psammoma body calcifications. Enhancement is homogeneous (Fig. 11.7) except when necrotic, cystic, hemorrhagic, or lipomatous components exist within the tumor. Because of their extraaxial origin, meningiomas characteristically have a long, smooth, broad-based attachment to the planum and the spheroid ridge and superior extension is modest relative to other suprasellar lesions of comparable size. Angiography is indicated in patients with suspected meningiomas to confirm the diagnosis and outline the vascular anatomy before operation. The most characteristic sign of a planum sphenoidale meningioma is elevation of the per-

Figure 11.7. Planum sphenoidale/tuberculum sellae meningioma. A. Postcon-trast axial CT scan demonstrates a homogeneously enhancing midline mass in the floor of the anterior fossa and the suprasellar region. B. Sagittal reformatted image defines the broad-based attachment of the mass to the tuberculum sellae and planum sphenoidale. icallosal and frontopolar arteries above the floor of the anterior cranial fossa. Downward displacement of the supraclinoid ICA and subsequent closing of the carotid siphon may occur. Significant displacement of the АСА and its branches from the midline is not a constant finding. Lesions arising from the planum sphenoidale or the tuberculum region can lead to encasement of the carotid artery; this finding denotes nonresecta-bility. Posterior ethmoidal branches of the ophthalmic artery may enlarge to supply a planum sphenoidale meningioma. In addition, a direct

anastomosis between the middle meningeal and ophthalmic arteries potentially provides additional blood supply to the tumor. If a meningioma erodes through the roof of the orbit, the ophthalmic artery may become stretched and encased. The intermediate phase of angiography demonstrates a homogeneous tumor blush that persists well into the late venous phase. Large planum sphenoidale meningiomas elevate the septal vein and displace the anterior portion of the internal cerebral vein posteriorly. Meningiomas arising from the tuberculum sel-

lae derive their principal blood supply from meningeal branches of the cavernous IСA. Tubercu-lum meningiomas can also elevate the A1 segment of the АСА. Epidermoid Tumors Epidermoid tumors arise from ectodermal cell inclusions within the neural roof at the time of neural tube closure between the third and fifth week of embryogenesis. These lesions comprise less than 1 % of all intracranial tumors. Due to their slow growth rate and soft cheeselike consistency, epidermoids expand initially within CSF cisterns along pathways of least resistance. However, they can later deform and rotate adjacent neural structures. Epidermoid tumors near the anterior third ventricle may extend anteriorly along the floor of the anterior cranial fossa or inferiorly to involve the parasellar region, middle fossa, cerebellopontine angle cistern, interpe-duncular fossa, and prepontine cistern. Epidermoids in the suprasellar region often produce symptoms by compressing the visual pathway and the cranial nerves located within the cavernous sinus. Symptoms are typically longstanding and less severe than with other intracranial tumors of comparable size. On noncontrast CT, epidermoids characteristically have attenuation values approaching those of CSF (25). This finding reflects the variable proportions of low density cholesterol, high density keratin, and desquamated epithelial debris. Capsular calcification rarely occurs. Intrinsically dense epidermoids with attenuation values of 80 to 120 Hounsfield units (HU) on noncontrast CT have been reported in the posterior fossa (12), but not as yet in the floor of the anterior cranial fossa. As a rule, because of their relative avascularity, epidermoids do not enhance after intravenous contrast administration. Metrizamide CT-cisternography is very useful for the diagnosis of epidermoid tumors. The interstices of epidermoids trap metrizamide, and the CT appearance is analogous to the "cauliflower" sign shown on pneumoencephalograms. Epidermoid tumors, even though benign, can recur if not completely excised. Spontaneous rupture of epidermoids may occur, resulting in chronic granulomatous arachnoiditis and communicating hydrocephalus. Lymphoma and Leukemia Lymphomatous and leukemic deposits can occur in the frontal lobe or the surrounding dura

mater with deformity of the anterior third ventricle. The CT appearances of these deposits have already been described and may closely mimic those of meningiomas. Demyelineating Disease Abnormal enhancement of the white matter tracts along the roof and anterior wall of the third ventricle has been reported in Type 2 adrenoleu-kodystrophy (ADL) (27) and in Alexander's disease (52). Enhancement is due to disruption of the blood-brain barrier secondary to an inflammatory response. In the acute phase of Type 2 ADL, prominent enhancement is noted in the rostrum and genu of the corpus callosum. Other areas of white matter tract involvement include the internal capsule, cerebral peduncles, corona radiata, and forceps major. Type 2 ADL is also characterized by the absence of symmetrical decreased attenuation in the peritrigonal white matter and the absence of an advancing margin of contrast enhancement proceeding anteriorly, features characteristic of Type 1 ADL. The CT pattern of Type 2 ADL has not been reported in other types of leukodystrophic, neoplastic, or infectious processes. Type 2 ADL should be considered when insidious personality changes, mental deterioration, and disturbances of gait and vision occur in a young boy. Laboratory evidence of adrenal insufficiency need not be present. Brain biopsy is necessary for definitive diagnosis. Alexander's disease is a rare leukodystrophy of uncertain pathogenesis, also demonstrating abnormal enhancement along the anterior aspect of the third ventricle. All proven cases have been sporadic. Clinical manifestations usually develop during the first year of life, with progressive retardation and an enlarging head circumference. During the acute phase of Alexander's disease, symmetrical, well-demarcated areas of low attenuation involve the deep cerebral white matter and basal ganglia. After intravenous contrast infusion, prominent enhancement appears in the anterior columns of the fornices, caudate nuclei, subependymal white matter, optic radiations, and forceps minor. Again, this constellation of CT findings has not been reported in other leukodystrophic, neoplastic, or infectious processes. Characteristically, megalocephaly without hydrocephalus is present in patients with Alexander's disease. Because there is no known biochemical marker in these patients, a brain biopsy is necessary to establish the diagnosis.

Lesions Affecting the Floor of the Third Ventricle Neoplastic Lesions Optic Nerve and Hypothalamic Glioma Optic nerve glioma occurs infrequently, accounting for 0.6 to 1.2% of primary intracranial tumors. CNS neurofibromatosis is present in 14 to 36% of patients with optic nerve glioma. Lesions confined to the optic nerve occur more frequently in prepubertal children, whereas chiasmal lesions are more common in adolescents. In large series, approximately 30% of the tumors involve the prechiasmatic portion of the optic pathway, with the remainder involving the anterior third ventricle, the optic chiasm, and the optic tract (Fig. 11.8). Lesions involving the posterior optic pathway are considered unrest-able because of invasion of the thalamus and medial temporal lobe. Histologically, two distinct types of glioma arise in the optic pathway, benign and malignant. The benign type occurs more commonly in childhood, tends to be localized to the optic nerve (with or without chiasmal involvement), and usually behaves indolently. However, malignant degeneration of these tumors has been documented. The malignant type is less common, tends to occur in middleaged patients, and may initially mimic optic neuritis clinically (48). Malignant optic nerve glioma may be fatal in a relatively short time. These tumors are often centered in the optic chiasm or optic tract, with infiltration of the hypothalamus and medial temporal lobe. Hypothalamic dysfunction often occurs later in the course of the disease. On CT, the normal optic nerve and optic chiasm are isodense with brain parenchyma and do not enhance with intravenous contrast administration. The normal chiasm has a smooth contour with a U or boomerang shape on axial sections and a dumbbell shape on coronal images. The optic recess of the third ventricle is filled with CSF and has a teardrop configuration above the chiasm on direct coronal scans. The average chiasm normally measures 1.8 cm transversely, 0.8 cm in the anteroposterior direction, and 0.4 cm vertically (24). The CT appearance of optic nerve glioma is variable (83). Uniform or fusiform enlargement of the nerve, as well as irregular thickening, are the most common presentations (Fig. 11.9). Rarely, optic nerve glioma demonstrates calcification before radiotherapy.

Figure 11.8. Optic chiasm glioma. A. Postcontrast axial CT demonstrates an enhancing suprasellar mass with a cyst [white arrow). The tumor extends posteriorly along the optic tract (open arrow). B. A scan at a higher level documents extension posteriorly along the optic tract toward the lateral geniculate body (open arrow).

Figure 11.9. Optic nerve glioma. Top. Fusiform enlargement of the right optic nerve extends back to the optic chiasm (arrow). Bottom. On a scan obtained 2 mm inferior to the upper image, note the unilateral optic canal enlargement and the enlarged optic nerve (arrow). The radiological appearance of optic nerve glioma is sometimes indistinguishable from that of optic nerve meningioma. Both lesions often enlarge the optic canal. The clinical history, sex, and age of the patient are helpful in this regard. Chiasmatic gliomas have a globular appearance with a vertical dimension greater than 0.6 cm (24). Intrathecal metrizamide may be necessary for the diagnosis of small lesions. On non-contrast CT, chiasmatic gliomas are isodense to slightly hyperdense. The majority of tumors enhance. If extension into the anterior third ventricle occurs, obstructive hydrocephalus may develop. Hemorrhage into chiasmatic gliomas and intratumoral cystic degeneration are not uncommon. The major consideration in the differential diagnosis of optic chiasm glioma is hypothalamic glioma. Extraaxial tumor growth from the chiasm indents the inferior third ventricle, resulting in capping of the tumor by the CSF-con-taining ventricle. Conversely, growth of lesions originating in the hypothalamus characteristi-

cally leads to obliteration of the anterior and inferior third ventricle. Early on, hypothalamic glioma causes sideways displacement of the inferior third ventricle, as the tumor arises just lateral to the midline (57). In most cases it is impossible to differentiate an optic chiasm glioma from a hypothalamic glioma (26). After radiotherapy, extensive dystrophic calcification due to mineralizing microangiopathy may involve the entire optic pathway. This diagnosis is easily established by CT. Angiography plays a limited role in the radiological evaluation of lesions of the optic chiasm and should be performed only to exclude an aneurysm or a meningioma. The internal carotid arteriogram often demonstrates lateral displacement of the supraclinoid ICA and upward bowing of the A1 segment of the АСА. With invasion of the optic tract and medial temporal lobe, displacement and encasement of the cisternal segment of the AChA occurs, in keeping with the anatomical proximity of this artery to the optic tract. On vertebral angiography, a chiasmatic glioma may cause displacement and stretching of the thalamoperforating arteries. If the tumor is of sufficient size, venous phase films demonstrate elevation of the internal cerebral vein and evidence of hydrocephalus. The high density on nonenhanced CT and homogeneous blush on angiography characteristic of meningiomas are absent from studies of chiasmatic gliomas. Pituitary Adenomas CT scanning allows direct identification of the pituitary gland, infundibulum, cavernous sinuses, neighboring arteries, and bony sella tur-cica. CT detects intrasellar microadenonas (3 to 4 mm in diameter) and extension beyond the sella (95). Adenomas of any histological types or degree of secretory activity may invade the suprasellar cistern, cavernous sinuses, or sphenoid bone. Prolactinomas are the commonest pituitary adenomas associated with suprasellar extension, followed by endocrine-inactive adenomas and GH-secreting adenomas (94). Normal pituitary tissue possesses no blood-brain barrier and enhances in proportion to the blood-iodine level. Macroadenomas, tumors larger than 10 mm in diameter, usually enhance homogeneously and can produce expansion or erosion of the sella turcica. Central areas of low attenuation from cystic degeneration or prior hemorrhage occur often in macroadenomas (69, 95) (Fig. 11.10). Lateral invasion into the cavernous sinus is difficult to detect radiologically unless there is

sphenoidal surgery. Facial bone and hand films demonstrate bony overgrowth in acromegaly. The first sign of a pituitary adenoma may be pituitary apoplexy, with the acute onset of severe headache and visual field abnormalities. A CT scan reveals a suprasellar mass with high attenuation, indicating acute hemorrhage within the adenoma (76, 95). Although uncommon, a blood-fluid level may be seen within the mass. Mixed density within the tumor is encountered more often, with cystic nonenhancing regions probably representing residua of previous infarction or hemorrhage. Correlation with history and CSF studies is vital, as a pituitary abscess and an uncomplicated pituitary adenoma may also exhibit ring enhancement with central areas of low attenuation.

Figure 11.10. Pituitary macroadenoma. A direct coronal CT image demonstrates a pituitary macroadenoma with suprasellar extension. Note the sloping of the sellar floor (open arrows). A low density area (indicating either a cyst or necrosis) is seen within the tumor [arrow). asymmetrical bulging of the cavernous sinus because the macroadenoma and cavernous sinuses enhance equally. On direct coronal scans, tumor encroachment on the CSF space in the chias-matic cistern indicates supersellar extension. Metrizamide CT-cisternography permits exact delineation of the optic chiasm in problem cases. Angiography is indicated if there is difficulty in distinguishing a macroadenoma from a men-ingioma or an aneurysm. The angiographic features, in addition to major vascular displacements, include a vascular tumor blush and abnormal enlargement of the meningohypophyseal trunk and the superior and inferior hypophyseal branches of the ICA. With extensive lateral growth, the tumor may parasitize supply from the middle meningeal artery. Cavernous sinus venography may provide further information regarding possible cavernous sinus invasion. Calcification is infrequent in pituitary adenomas. Curvilinear calcification within a sellar or parasellar mass obligates the exclusion of an aneurysm or dolichoectasia of arteries in the circle of Willis. Sellar enlargement, sloping and unilateral thinning of the floor, erosion of the dorsum and posterior clinoid processes, and frank extension into the sphenoid sinus are well demonstrated by CT. Anatomical information regarding the sphenoid sinus septa is useful in planning trans-

Craniopharyngioma Craniopharyngioma accounts for 5 to 10% of all childhood brain tumors and 50% of suprasellar tumors in the pediatric age group (34, 95). A second smaller peak occurs during the sixth and seventh decades. Cyst formation and calcification are prominent. Approximately 85% of Craniopharyngiomas are totally or partially cystic (73). Cystic regions are typically of lower attenuation on noncontrast CT than are truly solid components of the tumor, but may be isodense or even hyperdense. Braun et al. attributed increased attenuation in the cystic component, encountered in 4 of 63 surgically resected tumors, to a high intracystic protein concentration (11). Contrast enhancement is limited to the capsule and solid portions of the tumor. Childhood Craniopharyngiomas typically exhibit some degree of contrast enhancement, whereas adult tumors are often poorly enhancing (34, 63). Calcification may be rimlike and peripheral, central, combined peripheral and central, or even present in adjacent brain parenchyma. In children, CT demonstrates calcification in almost all cases. Tumors in adults may calcify but in a smaller percentage of cases. Commonly observed bony changes include sellar enlargement and erosion of the dorsum sellae. CT scanning after contrast administration is virtually diagnostic of craniopharyngioma in a child if at least two of the cardinal features of calcification, cyst formation, and contrast enhancement are present in a suprasellar mass (Fig. 11.11). In adults, however, meningiomas, pituitary adenomas, and aneurysms may demonstrate calcification and enhancement and are indistinguishable from some Craniopharyngiomas.

Figure 11.11. Craniopharyngioma. A sagittal reformatted CT image after contrast administration demonstrates a suprasellar tumor with rimlike calcification superiorly [arrows). The suprasellar component enhances. Anterior is to the reader's left. Craniopharyngiomas are generally large tumors at the time of diagnosis, averaging 3.5 cm in diameter in a large pathological series (73). Modern CT scanners, operating in the direct coronal projection, accurately define intrasellar and suprasellar portions of the tumor. CT detects involvement of the optic chiasm and surrounding vascular structures and encroachment on surrounding neural tissue or bone. Postoperative scans are useful for evaluation of the amount of residual tumor and for possible reaccumulation of cyst fluid. In cases requiring further refinement of anatomical detail, metrizamide CTcis-ternography is a useful adjunct (42). Angiographic findings are limited to displacement of surrounding vessels because Craniopharyngiomas are avascular tumors (63, 69). Meningioma In addition to those arising from the tubercu-lum sellae and medial sphenoid ridge, meningiomas may arise at other sites in the vicinity of the inferior third ventricle: diaphragma sellae, cavernous sinus, optic nerve sheath at the level of the chiasm, or the tentorial incisura, with growth into the posterior suprasellar and prepon-tine cisterns (63, 95). The radiological characteristics of meningiomas were discussed under "Lesions Affecting the Anterior Third Ventricle." Like tuberculum sellae meningiomas, diaphragma sellae tumors present as enhancing masses filling the suprasellar cistern. The characteristic vascular displacement is elevation of the A1 segments of the anterior cerebral arteries.

Diaphragma sellae and tentorial incisural meningiomas cause enlargement of the branches of the meningohypophyseal artery. Suprasellar Germinoma Suprasellar germinoma ("ectopic pinealoma") may be suspected clinically in patients presenting with diabetes insipidus, visual disturbances, and anterior pituitary dysfunction (91). The striking male preponderance in pineal region germinomas is not duplicated with suprasellar germinomas. Patients with this condition generally do not have a concomitant lesion in the posterior third ventricle. Noncontrast CT demonstrates an isodense to slightly hyperdense mass filling the suprasellar cistern. Infiltration of the surrounding brain results in blurred, ill-defined margins. Calcification is absent. Moderate, uniform enhancement is the usual pattern. Germinoma may infiltrate along the infundibulum, walls of the third ventricle, and subependymal regions of the frontal horns and produces abnormal enhancement of these structures. Metrizamide ventriculography reveals upward convexity of the anterior third ventricular floor. The optic and infundibular recesses may be effaced and displaced posterosuperiorly. Cerebral angiography demonstrates vascular displacements characteristic of a suprasellar mass. The mass is usually avascular and does not produce abnormal prominence of the meningohypophyseal trunk, as would often occur with meningiomas or pituitary adenomas.

Dermoid, Epidermoid, Teratoma Epidermoids may closely mimic suprasellar arachnoid cysts, and the CT attenuation may approximate CSF. Metrizamide CT-cisternography confirms the diagnosis when contrast agent outlines the tumor interstices. Dermoid cysts and benign teratomas may occur in the suprasellar region and mimic Craniopharyngiomas (95). The radiological features of dermoids and teratomas are discussed under "Lesions Affecting the Posterior Wall and Pineal Region." Metastases Metastatic disease of the pituitary gland produces radiological signs indistinguishable from those of invasive adenomas (95). An enhancing tumor mass may destroy the bony sella. Primary sites include carcinoma of the breast, lungs, kidney, and colon. Metastases to the infundibulum produce masses within the suprasellar cistern, and hypothalamic involvement is possible. Tuber Cinereum Hamartoma Tuber cinereum (hypothalamic) hamartomas are associated with infantile or early childhood isosexual precocious puberty and seizures (40). These hamartomas appear on CT as nonenhanc-ing masses, 1 to 2 cm in diameter, located in the posterior suprasellar cistern (57). Hamartomas may displace the basilar artery and brain stem posteriorly and occasionally attain sufficient size to distort the anterior third ventricle (57, 63). Calcification within hamartomas is rare. Lack of enhancement is sufficiently characteristic for some authors to recommend diagnosis on purely clinical and neuroradiological grounds. Angiography is helpful if the diagnosis remains in doubt after CT scanning. Abnormal vessels and a tumor blush are not encountered. The distal basilar artery and APMV are displaced posteriorly, and the PCoAs may be displaced laterally. Bony changes are unusual, but local erosion of the dorsum sellae may occur. Roentgenograms of the hands document inappropriately advanced bone age. Metrizamide CT-cisternography detects small masses otherwise not seen on routine CT scans. Other rare causes of suprasellar masses, often with marked contrast enhancement, include choristomas (granular cell myoblastoma) (86, 90) and, in children, idiopathic granulomatous disease. The radiological features of these rare disorders are not specific and the lesions cannot be

distinguished from the more common suprasellar masses. Inflammatory Lesions Chiasmatic Optic Neuritis In cases with isolated symmetrical enlargement of the optic chiasm, chiasmal optic neuritis is indistinguishable from glioma by CT criteria alone (31). Because the median age of presentation of the two is similar, only a therapeutic trial of corticosteroids differentiates neuritis from glioma. Serial CT scans document resolution of chiasmal swelling as patients with chiasmal optic neuritis improve clinically. Histiocytosis X Histiocytosis X produces characteristic bony erosive lesions and often involves the hypothalamus, producing diabetes insipidus (20). The disease affects children more commonly, males outnumbering females. Noncontrast CT reveals a soft tissue mass within the suprasellar cistern, often with an associated low density region within the hypothalamus. Miller et al. reported enhancement of the hypothalamic and suprasellar lesions in three of their five cases of histiocytosis (63). Deformity and flattening of the floor of the third ventricle can occur. Angiography generally reveals an avascular mass. The diagnosis of histiocytosis X can be made in patients with CNS symptoms (diabetes insipidus or panhypopituitarism), a suprasellar mass, and bony lesions. The skull, followed by the femur and pelvis, are the most commonly affected bones. Geographic calvarial lytic lesions are common, with beveled margins produced by unequal involvement of outer and inner tables. Involvement in the periapical regions of the mandible produces an appearance of "floating teeth." Mas-toid destruction and soft tissue changes within the middle ear commonly result, usually sparing facial nerve function. Sellar destruction and sphenoid erosions are uncommon and unrelated to the development of diabetes insipidus. Sarcoidosis CNS sarcoidosis causes a variety of neuroradiological manifestions and may be the first evidence of the systemic disorder (14). Approximately 4 to 7% of all patients with sarcoidosis develop CNS involvement (54). A common pattern of CNS sarcoidosis is granulomatous leptomeningitis, involving the basal meninges, hypothalamic region, floor of the third

ventricle, infundibulum, and optic chiasm. Hydrocephalus is a common complication, resulting from either intraventricular or extraventricular arachnoiditis and adhesions. Extensive contrast enhancement of the basal leptomeninges occurs in these cases, similar to that seen with other granulomatous meningitides (e.g., tuberculous or fungal). Parenchymal granulomas are often widespread and may coalesce to form masses mimicking primary neoplasms. The suprasellar region is the most common intracranial site for mass lesions of sarcoidosis. Hemispheric, subfrontal, intraventricular, and periaqueductal foci have also been reported. The lesions are characteristically hyperdense on noncontrast CT and demonstrate marked, uniform enhancement. Calcifications may occur. Sarcoid masses are typically avascular at an-giography. Sarcoid angiitis is rarely demonstrated. Bony destruction may present as well-defined calvarial lytic lesions, characteristically without marginal sclerosis. Optic foraminal enlargement (due to sarcoidosis involving the optic nerve sheath) and enlargement of the sella turcica have also been reported. Rarely, sarcoidosis produces reactive sclerosis of the orbital plates, sphenoid bone, and pterygoid plates. CT is particularly helpful in the follow-up of patients undergoing corticosteroid treatment. Vascular Lesions Giant Intracranial Aneurysms Giant aneurysms have a diameter greater than 2.5 cm and usually present clinically as space-occupying lesions. Giant aneurysms of the carotid bifurcation and basilar artery may extend superiorly to compress the optic chiasm, deforming the anterior wall and floor of the third ventricle (37). Rarely a giant basilar aneurysm obstructs the posterior third ventricle. Giant aneurysms should be constantly borne in mind in the radiological differential diagnosis of calcified masses in the suprasellar region, and full angio-graphic evaluation should be conducted if any doubt remains. Giant intracranial aneurysms can be divided into three types, partially thrombosed, completely thrombosed, and nonthrombosed. Each of these types has specific CT findings (85). A partially thrombosed giant aneurysm is the most common type in clinical practice (Fig. 11.12). A noncontrast CT scan usually demonstrates curvilinear calcification in an interrupted

pattern. The lumen of the lesion has an inhom-ogeneous attenuation reflecting the thrombosed part (high attenuation) and the portion with free-flowing blood (low attenuation). After intravenous contrast administration the relative density of these two components reverses, indicating homogeneous enhancement of the free-flowing blood within the patent lumen. In addition, a ringlike pattern of contrast enhancement, either continuous or interrupted, may be detected at the periphery of these lesions. This peripheral enhancement represents the rich network of vasa vasorum along the wall of the giant aneurysm. Angiography is necessary to confirm the diagnosis. Often the parent vessels drape over the unopacified portion of the aneurysm. Angiography alone underestimates the size of partially thrombosed aneurysms. Completely thrombosed giant aneurysms are sharply delineated, slightly hyperdense, round lesions on both noncontrast and postcontrast CT. Because the contents are static and thrombosed, the density on noncontrast and contrast-enhanced scans is the same. Because of their extraaxial location, giant aneurysms can deform the suprasellar cistern. Partial calcification and enhancement of the aneurysm wall occur. The walls of nonthrombosed aneurysms often have little or no calcification. These lesions are slightly hyperdense on noncontrast CT and enhance homogeneously. Areas of low attenuation in the brain parenchyma, presumably representing ischemia or atrophy secondary to the mass effect of these lesions, have been described adjacent to giant aneurysms. However, these zones of low attenuation are less extensive than those with intracranial neoplasms of comparable size. Other techniques to confirm the diagnosis of a nonthrombosed giant aneurysm are dynamic CT, MRI, and digital intravenous angiography. Dynamic CT scanning provides rapid sequential measurement of attenuation values. After a rapid intravenous injection of a bolus of 40 ml of iodinated contrast agent, repetitive scanning is obtained through the suspected lesion. Aneurysms characteristically demonstrate a rapid wash-in peak occurring between 6 and 11 seconds postinjection, with washout of the contrast agent by 20 seconds. A recirculation peak appears at 32 to 48 seconds (75). The main differential diagnosis is from a meningioma. Attenuation values of meningiomas also rise rapidly to a peak, but this is followed by a flat washout phase. This plateau characteristically occurs at 30 to 40 seconds and is due to retention of contrast material within the meningioma. There is

no recirculation peak. Thus, the major differences in the time-density curves distinguishing tumor from giant aneurysm are the persistence of contrast enhancement shown by the plateau and the absence of a recirculation peak. MRI characteristics of aneurysms are discussed under "Appearance of Lesions on MRI." Digital intravenous angiography (DIVA) uses an antecubital fossa or preferably superior vena caval catheter for the bolus injection of 40 ml of 76% iodinated contrast material. DIVA produces simultaneous opacification of the giant aneurysm and the arteries of the circle of Willis. Even though dynamic CT and DIVA may confirm the diagnosis of giant aneurysm, conventional in-traarterial angiography remains essential for exact anatomical delineation before operation. For this reason, we advise that the expeditious work-up of an aneurysm bypass dynamic CT and DIVA and commence with selective arteriography. Craniopharyngiomas, planum sphenoidale and diaphragma sellae meningiomas, and suprasellar germinomas can all mimic nonthrombosed giant aneurysms. Associated findings of hy-perostosis and pneumosinus dilatans, if present, provide helpful clues to the diagnosis of menin-gioma. The margins of germinomas are rarely as well defined as those of aneurysms. In addition, subarachnoid and subependymal tumor seeding occurs with germinomas, representing a differential feature of this entity. Basilar Artery Ectasia Atherosclerotic ectasia of the calcified basilar artery may encroach upon the inferior third ventricle (Fig. 11.13). In severe cases the third ventricle may become partially obstructed and hydrocephalus ensues. Chiasmal Apoplexy Chiasmal hemorrhage may be secondary to rupture of small venous angiomas or arteriove-nous malformations or to hemorrhage into a chiasmatic glioma (58). CT scanning reveals a high attenuation mass in the suprasellar region. In Maitland's group of four patients (58), enhancement accompanied hemorrhage into a chiasmatic glioma. Carotid arteriography is often normal but sometimes represents opacified flowing blood within the patent portion of the aneurysm. Note secondary hydrocephalus due to obstruction of the third ventricle. C. Lateral film from a selective ICA angiogram demonstrates opacification of the patent portion of the aneurysm lumen, confirming the diagnosis.

Figure 11.12. Giant supraclinoid carotid artery aneurysm. A. Lateral skull film demonstrates curvilinear calcification in the suprasellar region [arrow). B. Direct coronal enhanced CT displays the calcified rim of the aneurysm enclosing thrombus within its dome [arrows]. A high attenuation area [large white arrow)

diffusion of metrizamide across the cyst wall, in contradistinction to the "cauliflower" appearance in patients with epidermoid tumors. In the appropriate clinical setting, the ventric-ularcisternal form of cysticercosis may mimic a suprasellar arachnoid cyst (18). Other Lesions Deformity and enlargement of the optic chiasm and distortion of the suprasellar cistern can oc-

Figure 11.13. Basilar artery ectasia. A calcified ec-tatic basilar artery [black arrow) deforms the third ventricle [white arrows) and the right cerebral peduncle. demonstrates venous abnormalities in the suprasellar region. Congenital Lesions: Arachnoid Cysts Suprasellar cysts may originate from the epen-dyma of the third ventricle as well as from the basal arachnoid (1). Multiplanar CT is especially valuable in outlining the anatomical relationships. On axial CT, a large rounded or oval CSF density structure fills the suprasellar cistern and third ventricular region. Hydrocephalus involving both lateral ventricles is usual in symptomatic patients. The rounded cyst often lacks the posteroinferior tapering characteristic of the enlarged third ventricle of aqueduct stenosis. However, this appearance alone is not sufficiently reliable for firm diagnosis. Cysts may also cause flattening of the upper brain stem and colliculi. Even with direct coronal CT scanning, delineation of the compressed third ventricle is often impossible. Contrast enhancement and calcification of the cyst wall are characteristically absent. Metrizamide CT-ventriculography is crucial for the evaluation of hydrocephalus associated with a possible arachnoid cyst (1, 96). With aqueduct stenosis, metrizamide fills the enlarged third ventricle; with suprasellar arachnoid cysts, contrast should outline the top of the cyst. Metrizamide CT-cisternography may be necessary in troublesome cases, particularly if the CT appearances mimic those of an epidermoid tumor. The test will demonstrate either immediate entry of contrast agent into the cyst or slow

Figure 11.14. Herniation of the third ventricle and optic chiasm into the sella turcica. A. Normal metrizamide cisternogram of the suprasellar region. Note the normal optic chiasm [arrows). B. Metrizamide cisternogram demonstrates a filling defect in the suprasellar region [arrows). Compare with the normal (A). C. Sagittal reformatted image from the metrizamide cisternogram seen in В illustrating downward herniation of the inferior recesses of the third ventricle into a partially empty sella [arrows). Reformatted images were crucial to the proper characterization of the filling defect seen in the suprasellar cistern on the axial image.

cur in adhesive perichiasmatic arachnoiditis (16, 83) secondary to prior hemorrhage, iophendylate cisternography, operation, or infection. The CT diagnosis is possible given the appropriate clinical history. Finally, downward herniation of the anterior recesses of the third ventricle and the optic chiasm can occur in the empty sella syndrome (16) due to deficiency of the diaphragma sellae (either primarily or secondary to prior transsphenoidal operation or ischemic necrosis of the gland). In these patients, deformity of the optic chiasm in conjunction with an area of low attenuation within the sella presents a confusing picture on the contrast-enhanced CT scan alone. If, however, the infundibulum can be seen extending toward the floor of the sella turcica, the diagnosis is facilitated. Metrizamide CT-cisternog-raphy allows a definitive diagnosis (Fig. 11.14). The third ventricle can also herniate into the sella turcica in patients with obstructive hydrocephalus due to lesions at the level of the aqueduct. Lesions Affecting the Posterior Wall and Pineal Region Neoplastic Lesions The development of high resolution CT scanning in the 1970s combined with widespread utilization of the operating microscope revolutionized the diagnosis and treatment of pineal region mass lesions. Although the histological features of pineal region masses cannot be definitively predicted by noninvasive techniques, the differential diagnosis is usually limited and a single tumor type may often be identified as most probable. Germ cell tumors of the pineal region can be divided into germinomas and embryonal carcinoma and its derivatives, choriocarcinoma, yolk sac tumor (endodermal sinus tumor), and teratoma/teratocarcinoma. Such tumors may be relatively "pure" histologically or comprised of several different subtypes. The incidence of pineal tumors is shown in Table 11.1. Germinoma Germinomas account for approximately half of all pineal region tumors (29, 51, 82). This tumor may originate in the pineal or parapineal region, secondarily engulfing or displacing the intact pineal gland. The majority of cases occur in adolescent boys. Noncontrast CT demonstrates an isodense to slightly hyperdense mass (35, 98, 100). Calcifi-

cation of the tumor itself is rare. Cyst formation is also unusual. Smaller lesions are well defined, but the borders of larger lesions may be indistinct due to infiltration of tumor cells into adjacent neural tissue. In either case, edema of the surrounding brain is unusual. After contrast administration, diffuse uniform enhancement is typical (Fig. 11. 15A), but not invariable. The normal pineal gland calcification appears displaced (Fig. 11.15B) or engulfed by the enhancing tumor mass. Abnormal enhancement in the basal cisterns and walls of the lateral ventricle indicates seeding via CSF pathways. Leptomeningeal spread occurs in approximately 8 to 15% of cases. Although unusual, cases of simultaneous masses in the pineal region and suprasellar region ("ectopic pinealoma") have occurred. CT is useful in the follow-up of these highly radiosensitive tumors. The usual response to external beam radiation is total resolution within 6 weeks. However, tumors with cystic components on the initial study have demonstrated relative resistance to radiotherapy. Embryonal Cell Carcinoma As with the commoner germinoma and teratoma, embryonal cell carcinoma is most often encountered in young adolescent boys (66, 100). Noncontrast CT demonstrates a slightly hyperdense mass, often containing calcium (35, 51). The endodermal sinus variant may develop cysts. Diffuse enhancement after contrast is the usual pattern. The lesions may infiltrate surrounding structures or be well circumscribed. Edema of surrounding brain is scant or absent. Seeding of tumor may occur via CSF pathways, and massive intratumoral hemorrhage is occasionally encountered. Am embryonal carcinoma is highly vascular at angiography.

81). Spontaneous rupture of cystic elements into the subarachnoid space and ventricles has been documented by CT. The CT appearances of dermoid cysts closely mimic those of teratomas. Angiographic manifestations of teratomas are scant, usually limited to the vascular displacements surrounding an avascular mass. A faint tumor blush is encountered occasionally. Glioma Gliomas in the pineal region and posterior third ventricle comprise 10 to 25% of masses in this region (51,100). There is no sex predilection, and most patients are young. The spectrum of glial tumors described in this region includes astrocytoma, glioblastoma, ependymoma, and ganglioglioma. Although the pineal gland contains fibrillary astrocytes, most pineal region gliomas originate in parapineal structures, such as the corpora quadrigemina or the posterior thalamus (81). CT characteristics of gliomas are outlined under "Lesions Affecting the Lateral Walls of the Third Ventricle." Prominent enhancement is encountered occasionally in exophytic components. Angiographic findings are variable, depending on the grade of the tumor and its exact location. Encasement of the basilar artery has been reported.

Figure 11.15. Germinoma (two cases). A. A sagittal reformatted CT scan image demonstrates an inhomogeneously enhancing mass [arrow) deforming the posterior third ventricle [top). This pattern is unusual. The more typical pattern of homogeneous enhancement is seen in the scan of another patient [bottom). B. Postcontrast axial CT scan illustrates displacement of pineal calcification [arrow) by the expanding tumor. Benign Teratomas The pineal region is the most common intracranial location for benign teratomas. A preponderance of cases occurs in young men (81). CT scans reflect the gross pathological features of these tumors. Low attenuation (-70 to —30 HU) regions predominate, representing fatty portions of the tumors (Fig. 11.16). Bone and dental elements are easily seen. Soft tissue components may be isodense to normal brain and demonstrate contrast enhancement. Although often large and irregularly shaped, teratomas are non-invasive and present well-defined borders (51,

Pineal Cell Origin Less than 20% of pineal region tumors represent true pineal cell neoplasms (100). Benign pineocytomas and malignant pineoblastomas may exist in either pure or mixed forms (29, 51). In most series, mixed types and pineoblastomas predominate. Unlike the situation with germ cell tumors, the sex incidence is equal. Mean age at presentation is 10 years. Pineal cell tumors are isodense to slightly hy-perdense before the administration of contrast agent. Abnormal calcification within these tumors is characteristic. Pineocytomas may be cystic. Enhancement is dense and uniform. The tumors may appear either well defined or infil-trative. Pineoblastomas frequently disseminate via CSF pathways (Fig. 11.17) and may also metastasize outside the CNS. At angiography, pineal cell tumors are usually moderately vascular, but avascular masses have been described. Lipoma Although the corpus callosum is the most common site for intracranial lipomas, the pineal region may harbor lipomas originating in the quad-

Figure 11.16. Teratoma. A. A postcontrast axial CT scan demonstrates calcification surrounding a fat-containing tumor (arrows) in the posterior third ventricle. B. Sagittal reformatted CT image depicts the fat-containing tumor in the posterior third ventricle [arrows). Note the marked enlargement of the anterior third ventricle. Anterior is to the reader's left.

lipomas may seem to have significantly higher attenuation values because of the partial volume averaging of surrounding brain with the fat density of the tumor. Enhancement is characteristically absent. The CT differential diagnosis should include dermoid cyst and teratoma.

Figure 11.17. Pineoblastoma. A. Abnormal enhancement in the ambient and sylvian cisterns {arrows) indicates seeding of tumor along the CSF pathways. The tumor (best seen on a higher scan) is a faint area of enhancement in the quadrigeminal plate area. В. А postmetrizamide CT scan of the lumbar spine demonstrates a drop metastasis to a lumbar nerve root [arrow].

rigeminal plate, velum interpositum, or ambient cistern (98, 100). Lipomas occur in children or adults. These tumors exhibit a fat density on nonen-hanced CT (-100 to -40 HU). However, small

Figure 11.18. Arachnoid cyst. A. A quadrigeminal cistern arachnoid cyst [arrow) compresses the posterior third ventricle and aqueduct. The ventricles are shunted. B. A metrizamide CT-ventriculogram demonstrates noncommunication of the cyst with the ventricular system.

Meningioma Pineal region meningiomas share the CT characteristics of meningiomas elsewhere: isodense to slightly hyperdense masses with uniform enhancement. Calcification is variable, as is tumor blush at angiography. These tumors may arise from the pineal gland itself, the velum interpositum, or the falx-tentorium junction (80). Metastases Metastases may involve the pineal region and posterior third ventricle (62). Tumor types include breast, lung, melanoma, and adenocarci-noma (primary site unknown). The CT appearance of pineal region metastases is variable; multiplicity of lesions provides the only distinguishing radiological feature. Individual metastases may be isodense or slightly hyperdense, with variable degrees of edema and enhancement. Epidermoids Epidermoid cysts occur rarely in the pineal region. The radiological features were presented under "Lesions Affecting the Anterior Wall of the Third Ventricle." Congenital Lesions: Arachnoid Cysts Arachnoid cysts and glial cysts of the pineal region share similar CT characteristics. The cysts may be either developmental or acquired secondary to postinflammatory or posthemor-rhagic adhesions. The cysts may involve either the cistern of the velum interpositum, the quadrigeminal plate cistern, or both (62). Cyst outlines may be smooth, lobular, or irregular. Metrizamide CT-cisternography confirms the diagnosis by demonstrating noncommunication with the normal subarachnoid space or a slow diffusion of metrizamide into the cyst (Fig. 11.18). Vascular Lesions Aneurysm of the great vein of Galen was discussed under "Lesions along the Roof of the Third Ventricle." Lesions Affecting the Lateral Walls of the Third Ventricle The most common lesions involving this region are primary brain tumor and hypertensive hemorrhage originating from the basal ganglia and thalamus. Neoplastic Lesions Glioma Primary brain tumors have variable appearances on CT. On noncontrast CT, low grade as-

trocytomas commonly present as an area of ill-defined low attenuation with a subtle mass effect. Although glioblastomas can have increased, decreased, or mixed attenuation on noncontrast CT, the mixed attenuation pattern together with a high degree of white matter edema is most common (65% of cases) (88). Hemorrhagic, necrotic, or cystic components are frequent. Low grade astrocytomas typically do not show contrast enhancement. Glioblastomas have variable appearances on contrastenhanced CT, the most common being a thick, irregular, ring-shaped rim of enhancement surrounding a low density center (Fig. 11.19). Coronal CT scans show a typical anatomical picture (Fig. 11.20). Cystic gliomas may have a single, small, enhancing mural nodule. Although CT scans cannot predict the histological features of primary brain tumors, the degree of contrast enhancement roughly parallels the histological grade. On cerebral angiography, AV shunting and neovascularity characterize a glioblastoma, whereas nonspecific mass effect is the most common finding with a low grade astrocytoma. Unfortunately, other common diseases share many of the same CT characteristics as gliomas. The ring-shaped enhancement associated with primary brain tumor is usually thicker and more irregular than that of an abscess. The thalami and basal ganglia are uncommon sites for metastases. Homogeneous enhancement, extensive surrounding edema, and multiplicity of lesions characterize brain metastases. MRI is more sensitive than CT alone in the detection of low grade astrocytomas (8). MRI findings are discussed under "Appearance of Lesions on MRI." The major limitations of both CT and MRI are an inability to demarcate dependably the tumor margin from peritumoral vasogenic edema or radiation necrosis and an inability to predict the specific histological grading of a primary CNS tumor. Lymphoma Primary CNS lymphomas, although uncommon (less than 1.5% of all intracranial tumors), have a predilection for the corpus callosum, basal ganglia, and thalamic regions. The CT characteristics and anatomical distribution of these lesions were discussed under "Lesions along the Roof of the Third Ventricle." Subependymoma Subependymomas are located most often near the fourth ventricle (84), but can occur along the walls of the third ventricle. Radiological features

Figure 11.19. Thalamic glioblastoma. A and В (В 8 mm above A). A garland pattern of enhancement surrounds a central low attenuation area of necrosis, findings often seen with a glioblastoma. С Lateral magnified view of a selective vertebral arteriogram demonstrates neovascularity [arrows) and stretching of the posterior choroidal arteries. of subependymomas were discussed under "Lesions along the Roof of the Third Ventricle." Germinoma Germinomas most commonly occur in the pineal and suprasellar regions. In a Japanese series,

however, 10% of all intracranial germinomas primarily involved the thalamic and basal ganglia region (53). Although these lesions may be radio-logically difficult to differentiate from high grade gliomas, the clinical features of a slower progression, predominance in the prepubertal boy, and

Figure 11.20. Thalamic glioma. Coronal reformatted CT image illustrates invag-ination of the trigone of the lateral ventricle [small arrows) by a parapineal region glioma. Note the contralateral displacement of the internal cerebral veins [large arrows). favorable response to radiation therapy often distinguish germinoma from glioblastoma. After radiation therapy, germinomas appear as low density areas without mass effect or enhancement. The radiological features of typical pineal germinoma were presented under "Lesions Affecting the Posterior Wall and Pineal Region." Vascular Lesions Hypertensive Hemorrhage The basal ganglia and the thalami are by far the most frequent sites of spontaneous intracranial bleeding (from 40% to approximately 75% of all cases in large series) (74). Isolated thalamic hemorrhages are extremely rare; most cases also involve the internal capsule, basal ganglia, or adjacent deep white matter. Thalamic hemorrhage frequently extends into the ventricles (39). By far the most common cause of thalamic hemorrhage is rupture of a Charcot's aneurysm of the lenticulostriate arteries in a patient with long-standing hypertension. Unusual causes of thalamic hemorrhages include AV malformations, cavernous angiomas, trauma, aneurysms, and hemorrhage into vascular metastases (e.g., choriocarcinoma, melanoma, renal cell carcinoma, or thyroid carcinoma). Resorption of hematomas progresses from the periphery to the center. On noncontrast CT, hematomas evolve from a hyperdense through an isodense to a hypodense phase over 2 to 3 weeks (Fig. 11.21). Contrast-enhanced CT during this period may demonstrate peripheral, thick, ring-like enhancement, potentially mimicking a primary intracranial tumor. This enhancement

Figure 11.21. Hypertensive hemorrhage. A mixed high and low attenuation mass lesion involves the basal ganglia and thalamus. This appearance on non-contrast CT is characteristic of the subacute phase of a hypertensive basal ganglia hemorrhage. Note the compression of the third and lateral ventricles. may remain for 6 to 9 months. Resolved hematomas on noncontrast CT present as well-defined areas of low attenuation with ipsilateral enlargement of the adjacent ventricles. Occasionally, isodense gliosis can develop in place of resolving hematomas.

Cavernous Angioma Hemorrhage from cavernous angiomas may be indistinguishable from primary hypertensive hemorrhage by CT. Unruptured cavernous angiomas are generally hyperdense on noncontrast CT because of microcalcification (77). Mass effect and perifocal edema are characteristically absent. The enhancement pattern is variable. Cavernous angiomas are more common in the cerebral hemispheres than the thalamic region, and the lesions can enlarge slowly on sequential scans. Cavernous angiomas are often missed on routine angiography, but the angiographic detection rate increases with the use of slow injections (16 ml of 60% contrast material injected over 4 seconds) and prolonged filming (68). The high rate of false-negative angiography in cases of cavernous angioma relates to slow flow of blood from the vascular system into the angiomas and to stagnation of blood within the interstices of the lesions. Intraventricular Lesions Neoplastic Lesions A tumor arising from within the third ventricle is rare. Lesions of the Choroid plexus derive from ependymal layers (Choroid plexus papilloma or carcinoma) or from mesenchymal stroma (men-ingioma, vascular malformation, cavernous angioma, hemangioma) that infold during early em-bryogenesis. Giant cell astrocytomas and colloid cysts are the most common intraventricular tumors (65). Choroid Plexus Papillomas/Carcinomas Choroid plexus papillomas occur in patients of all ages. The fourth ventricle is the most frequent location in adults, and the lateral ventricle is the favored site in children, usually before 3 years of age (60). The incidence of Choroid plexus papilloma is 0.5% of intracranial tumors at all ages and 4% of intracranial neoplasms in patients under 12 years of age. Between 4% and 10% of all Choroid plexus papillomas occur in the third ventricle (79, 92). Choroid plexus carcinoma tends to occur in the very young (under 2 years of age). CT is the method of choice for the diagnosis of Choroid plexus papillomas. On noncontrast CT, these lesions are isodense to slightly hyperdense. Enhancement is intense and homogeneous. In benign Choroid plexus papillomas (Fig. 11.22), the margin between the tumor and the ventricular wall is well demarcated, whereas in malig-

Figure 11.22. Choroid plexus papilloma. A and B. Axial (A) and sagittal reformatted (B) images from a postcontrast CT scan demonstrate an enhancing mass containing small cysts filling the third ventricle. Note extension through the foramina of Monro [arrow, B) and associated hydrocephalus. C. Coronal plane ultrasound image depicts the densely echogenic mass within the third ventricle [arrows) and enlargement of the lateral ventricles. nant tumors, margination between the tumor and the ventricular wall is irregular, with extra-ventricular extension and infiltration into the surrounding brain (71). Calcification, cysts, and hemorrhage are rare, occurring in 4% of cases. Cystic and hemorrhagic components occur more frequently in carcinomas than in benign tumors. Malignant degeneration of benign papillomas in children occurs in about 20% of cases (92). Seeding of tumor within the CSF occurs in both benign and malignant Choroid plexus tumors.

Because of the 4- to 5-fold overproduction of CSF, generalized hydrocephalus and enlargement of the basal cisterns characteristically occur in the absence of intraventricular obstructive lesions. Intermittent hemorrhage from this highly vascularized tumor, with convexity block of CSF resorption, further aggravates the existing hydrocephalus. A third ventricular Choroid plexus papilloma, if large enough, can produce hydrocephalus by obstructing the foramina of Monro. Angiography aids in surgical planning. The principal blood supply is by an enlarged MPChA. The angiographic diagnosis of Choroid plexus tumor depends on finding a persistent blush in the presence of enlarged arterial feeders because normal Choroid plexus often exhibits a vascular tangle and a small capillary blush. With Choroid plexus carcinoma, additional findings are neo-vascularity and AV shunting. In the venous phase, elevation and distortion of the internal cerebral vein indicates the presence of the mass within the third ventricle. The diagnostic specificity of the CT and angiographic findings in patients with Choroid plexus lesions is limited because the appearance of a Choroid plexus papilloma is similar to that of intraventricular meningiomas, ependymomas, and cavernous angiomas. Angiography is most useful in the exclusion of an intraventricular AV malformation. Ependymoma Ependymomas constitute 6% of all intracranial gliomas and are predominantly tumors of childhood and adolescence. Ependymomas may arise in any part of the ventricular system, but the fourth ventricle is the most common site (approximately 60%). Supratentorially, ependymomas arise most often near the trigone of the lateral ventricle and less commonly within the third ventricle (15% in Barone's series (2)). Ependymomas are usually slowly growing and symptoms are due to obstructive hydrocephalus and local infiltration of the brain parenchyma. As ependymal cell rests are found far from the ventricular system, these tumors may also arise within the cerebral white matter at a distance from the ependymal surface. Solid tumor components are characteristically isodense. The enhancement pattern is variable. Intratumoral cysts occur commonly. Approximately 50% of ependymomas are calcified (89). After operation and radiation therapy, calcification in the residual tumor is more prevalent. An ependymoma, if totally intraventricular, mimics

a Choroid plexus papilloma on CT. The CT scan also detects local tumor recurrence and seeding of ependymoma within the ventricles and cisterns (33). Giant Cell Astrocytoma Associated with Tuberous Sclerosis Tuberous sclerosis is a neurocutaneous syndrome with dominant inheritance characterized by the triad of mental deficiency, adenoma se-baceum, and seizures. The hamartomas involve multiple sites: skin, brain, kidney, heart, bone, and retina, with the skin and brain being the most common. The tubers may undergo malignant transformation to giant cell astrocytomas (55). These tubers involve the lateral ventricle, third ventricle, foramen of Monro, and septum pellucidum, in decreasing order of frequency (5). In spite of their large size, giant cell astrocytomas are sharply circumscribed both on contrast-enhanced CT scans (3) and gross examination. However, giant glial cells have been detected in adjacent white matter at autopsy. Symptoms develop when the tumor obstructs the CSF circulation through the foramen of Monro with resultant hydrocephalus. When the tumor is large, the exact site of origin of the lesions is difficult to ascertain by CT or at operation (Fig. 11.23). Giant cell astrocytomas, occurring in isolation, may be difficult to differentiate from calcified ependymomas. Tuberous sclerosis may be difficult to distinguish from periventricular calcification associated with toxoplasmosis and cytomegalic virus without an appropriate clinical history. Although angiography is usually not necessary for the diagnosis of giant cell astrocytomas, when carried out the findings vary (46) and can include marked hypervascularity with pooling of contrast agent within the tumor, avascular tumor, arterial ectasia and occlusion of cerebral blood vessels remote from the primary lesion, and aneurysmal change in small cerebral arteries. The last two findings reflect the intrinsic dyspla-sia of the cerebral arterial walls. Highly vascular giant cell astrocytomas can be distinguished from other vascular malignant tumors by the lack of early venous drainage. Astrocytomas and ganglioneuromas not associated with tuberous sclerosis can arise within the third ventricle. Most intraventricular astrocytomas in the third ventricle occur during the first or second decade of life. Hydrocephalus, hypothalamic dysfunction, and memory deficits are the most common presenting symptoms in these patients (4). Astrocytomas commonly originate at the anterior column of the fornix and

may be attached to the ependyma by a pedicle. On CT, these tumors may or may not enhance. Cystic degeneration occurs in some cases. Intraventricular Carniopharyngioma Craniopharyngiomas originating within the anterior third ventricle are rare. The presenting symptoms and signs are due to obstructive hydrocephalus and hypothalamic dysfunction. Diabetes insipidus and visual symptoms are less common with intraventricular Craniopharyngiomas than with those arising from the infundibulum. Cases of intraventricular craniopharyngioma have all occurred in adults. The radiological manifestations of craniopharyngioma were discussed under "Lesions Affecting the Floor of the Third Ventricle." Although CT experience is limited, intraventricular Craniopharyngiomas usually lack calcification and enhance homogeneously (61). Metrizamide CT-ventricu-lography reliably establishes the intraventricular location of these lesions. Intraventricular Meningioma Although very rare, an intra-third ventricular meningioma should be included in the differential diagnosis of any intrinsically dense lesion occurring in the region of the foramen of Monro. Clinical symptoms result from obstruction to the flow of CSF. The major alternative diagnosis to be considered is a colloid cyst. On noncontrast CT, meningioma and colloid cyst are intrinsically dense and well circumscribed. The enhancement pattern of meningiomas is more intense than that of colloid cysts. In addition, the presence of calcium helps to differentiate meningiomas from colloid cysts. On vertebral angiography, a homogeneous tumor blush and hypertrophied distal MPChA have been reported in meningiomas arising from the anterior third ventricle (56). Other tumors arising from within the third ventricle include germinomas, epidermoids, teratomas, and dermoids. Their characteristic CT findings were discussed under "Lesions Affecting the Anterior Wall of the Third Ventricle" and "Lesions Affecting the Posterior Wall and Pineal Region." Hematogenous metastasis to the Choroid plexus should also be considered in the differential diagnosis of an enhancing intraventricular tumor. Another extremely rare lesion that mimics an intraventricular meningioma is an AV malformation originating from the Choroid plexus of the third ventricle. As reported by Britt et al. (13),

Figure 11.23. Giant cell astrocytoma. A. A noncon-trast CT scan of a patient with tuberous sclerosis demonstrates a largely cystic tumor containing calcification centered at the foramen of Monro. B. Contrast enhancement along the anterior margin of the tumor [arrow) indicates malignant degeneration.

this variety of AV malformation appears heterogeneous both before and after contrast infusion. Increased density on nonenhanced CT resulted from hemosiderin deposited in the capsule by an old hemorrhage. Xanthogranuloma Intracranial Xanthogranulomas usually adhere to the Choroid plexus in the trigone of the lateral ventricle and are often discovered as incidental findings. Anterior third ventricular tumors may cause obstructive hydrocephalus because of their strategic location. Radiologically, they appear in-homogeneous and intrinsically dense on noncontrast CT, in contradistinction to the homogeneously dense colloid cyst. Xanthogranulomas have low attenuation centers because of their lipid and cholesterol content. Punctate calcification is often present. The periphery of these lesions may enhance (38). If only a contrast-enhanced scan is performed, ringlike enhancement of a xanthogranuloma might easily be confused with that of an abscess. Chronic chemical meningitis has been associated with Xanthogranulomas. Xanthogranulomas are avascular on angiog-raphy. Even though these lesions appear sharply circumscribed on CT, surgical removal is technically more challenging than is the case with colloid cysts because the fibrous capsule of the xanthogranuloma is often densely adherent to the Choroid plexus, fornix, and wall of the third ventricle. Vascular Lesions Hemorrhage into the third ventricle is a common clinical entity, but seldom occurs as an isolated event. Aneurysm rupture, trauma, and hypertensive hemorrhage originating in the basal ganglia region are often associated with hemorrhage into the third ventricle. Hemorrhage originating from an AV malformation in the basal ganglia/thalamic region can also dissect into the third ventricle. In a series of 68 intraventricular hemorrhages analyzed by Graeb et al. (39), the source of hemorrhage was not identified in 12 patients after neuroradiological evaluation. The cause of intraventricular hemorrhages in this group is presumably rupture of angiographically cryptic AV malformations located along the ependyma. These patients have a better prognosis than patients whose intraventricular hemorrhage is associated with aneurysm, hypertension, or trauma. AV malformations and cavernous angiomas originating from the Choroid plexus are other rare causes of intraventricular hemorrhage.

Obstructive hydrocephalus can be a complication of intraventricular hemorrhage, but more frequently communicating hydrocephalus results from adhesive arachnoiditis and convexity block from associated subarachnoid hemorrhage. Infectious Lesion: Cysticercosis Intracranial cysticercosis may have menin-geal, parenchymal, and ventricular forms. Intraventricular cysticercotic cysts generally cannot be diagnosed directly by CT because their attenuation is similar to that of CSF. The diagnosis should be entertained in the presence of cystic or calcified parenchymal lesions, asymmetrical enlargement of the involved basal cistern, or obstructive or communicating hydrocephalus in patients with appropriate clinical histories. Intraventricular cysticercosis can be confirmed with metrizamide ventriculography. Cysticercotic cysts may be mobile. Entrance of metrizamide into an intraventricular cysticercotic cyst and enhancement of intraventricular cysts have been reported (99). The latter occurs only during the subacute phase of this infection (i.e., shortly after the death of the larvae with the subsequent inflammatory response in the surrounding tissue). Entrance of metrizamide into a third ventricular cyst is not pathognomonic of cysticercosis because it has also been reported with intraventricular arachnoid cysts and ependymal cysts. Intraventricular cysticercosis can be diagnosed with certainty only if there is accompanying radiological evidence of the parenchymal and meningeal forms of the disease. Congenital Lesion: Arnold-Chiari Type II Malformation In patients with Arnold-Chiari Type II malformation, the third ventricle is usually mildly dilated. With associated aqueductal stenosis, however, third ventricular dilatation is more prominent. On CT, the massa intermedia is unusually large in 75% of patients with the Arnold-Chiari Type II malformation. The enlarged massa intermedia frequently lies close to the foramen of Monro. The anterior columns of the fornix may be splayed or the wall of the third ventricle may be tethered, creating a biconcave appearance of the third ventricle (67). Infrequently, the massa intermedia totally obliterates the third ventricular cavity, resulting in fusion of the thalami across the midline. The large massa intermedia does not enhance, a finding sometimes helpful in differentiating the massa from an intraventricular tumor. Associated stigmata of the Ar-

nold-Chiari Type II malformation apparent on CT are an enlarged foramen magnum, a low-lying fourth ventricle, a small posterior fossa, scalloping of the petrous pyramids, hypoplasia of the tentorium and the falx, beaking of the tectum, inferomedial pointing of the floor of the frontal horns, and interdigitation of the gyri along the interhemispheric fissure. Appearance of Lesions on MRI Neoplastic Lesions In general, primary brain tumors demonstrate prolonged T1 and T2 times compared to normal brain (9, 17, 78) (Fig. 11.24). Significant overlap in T1 values exists between benign and malignant tumors, and cystic regions within either type of tumor may cause marked T1 prolongation. Randell et al. (78) reported a focal decrease in T1 within a malignant tumor due to hemorrhage. T1-weighted MRI is particularly helpful in depicting Craniopharyngiomas, lipomas, teratomas, and other lipid-containing masses because of the short T1 of fat (44, 50) (Fig. 11.25). In keeping with their variable lipid content, epidermoid tumors may also have short T1 values or a T1 value longer than that of surrounding brain (23). Calcification within tumors may produce focal areas of signal absence. Analysis of T2-weighted images is often complicated by surrounding edema, which also exhibits a long T2 time relative to brain tissue. Definition of the tumor-edema interface may sometimes be achieved by alteration of the pulse interval between excitation and the arrival of the spin echo. In some cases, particularly meningiomas, tumor T1 and T2 values may be close to those of brain, and the associated mass effect and edema provide the bulk of findings on MRI. Experience with paramagnetic contrast agents is accumulating, and preliminary images indicate that alteration of the blood-brain barrier can be accentuated in a manner similar to contrast-enhanced CT (19). T1- and T2-weighted images provide complementary information. In general, lesions are more conspicuous on T2-weighted images, but may be missed by heavily T2weighted sequences alone if the lesion is surrounded by or adjacent to CSF spaces (Fig. 11.24B). Direct comparisons of the utility of contrast-enhanced CT and MRI in the detection and characterization of tumors are under way. Anecdotal paralleling the occipital horns (arrows) are probably due to interstitial edema.

Figure 11.24. Germinoma. A. A heavily T1-weighted sagittal spin-echo image demonstrates a small mass of low signal intensity involving the pineal region (arrow). Note the compression of the aqueduct. B. A T2-weighted axial spin-echo image of the same case. Note the obscuration of the tumor by the high signal of the surrounding CSF. Linear areas of high signal intensity

reports document the superior detection of low grade astrocytomas, pineal region germinomas, and brain stem metastases by MRI, but no overall increase in supratentorial tumor detection efficiency over CT has been found in larger series (7, 10, 15). Efforts to predict specific histological features of MRI lesions by analysis of T1 and T2 changes have met with little success, as prolongation of T1 and T2 characterizes cerebral infarction, edema, demyelination, inflammatory disease, and many degenerative processes. Hydrocephalus MRI is valuable in the detection of early interstitial edema secondary to obstruction of CSF outflow. On T2weighted images, a high signal intensity develops in the periventricular regions because of transependymal flow of the CSF. Ventricular enlargement is easily detected, but the most helpful contribution of MRI is its excellent, predictable display of the size and configuration of the aqueduct of Sylvius. Vascular Lesions MRI of giant aneurysms has been described (17, 44). Short T1 regions within the aneurysms indicate clotted blood, whereas areas of absent signal indicate either rapidly flowing blood or mural calcifications (Fig. 11.26). Racemose AV malformations present a complex picture: low signal intensities are produced by rapidly flowing blood within feeding arteries and larger draining veins, whereas slower blood flow within the matrix of the malformation gives rise to high signal intensity (64). Hemorrhage presents a variable MRI appearance: very acute bleeding appears dark on T1-weighted images because of the long T1 However, progressive shortening of the T1 of hemorrhagic lesions results in increasing signal intensity from subacute and chronic hematomas (87). The shortened T1 of hemorrhage (intraparenchy-mal or extraaxial) persists long after hematomas become isodense or hypodense to normal brain on CT scans (9). Coronal images may be more accurate in identifying subdural hematoma than are axial images. The signal intensity of hemorrhage is high on T2weighted images. Subarachnoid hemorrhage may be difficult to distinguish on either T1- or T2-weighted images, age at the same level as the CT scan reveals high signal intensity within the mass due to its fat content. Note that the area of calcification (arrow) produces little or no signal.

Figure 11.25. Teratoma. A. A postcontrast CT scan image demonstrates a suprasellar mass with calcification (arrow). Intrinsic high attenuation within the mass was also seen on noncontrast CT scans (not shown). B. A heavily T1weighted axial spin-echo im-

Figure 11.26. Aneurysm. A. A T1-weighted axial inversion recovery image demonstrates absence of signal from rapidly flowing blood within the lumina of the ACoA and bilateral middle cerebral artery aneurysms [arrows). B. Oblique view from a selective ICA angiogram confirms the ACoA aneurysm (arrow). Other aneurysms at the middle cerebral artery bifurcations were also confirmed.

but is usually well demonstrated with conventional CT scanning. Sellar and Parasellar Lesions Several studies of pituitary and juxtasellar lesions have appeared (9, 43, 44). Pituitary adenomas, Craniopharyngiomas, and giant aneu-rysms may be differentiated according to the presence of fat or flowing blood. Cavernous sinus invasion and suprasellar extension of adenomas are well shown. Direct coronal and sagittal imaging is particularly helpful in the diagnosis of sellar and parasellar lesions. Empty sellas, hypothalamic hamartomas, suprasellar teratomas, and colloid cysts of the third ventricle have been demonstrated by MRI. References 1. Armstrong EA, Harwood-Nash DCF, Hoffman H, Fitz CR, Chuang S, Pettersson H: Benign supra sellar cysts: The CT approach. AJNR 4:163166, 1983. 2. Barone BM, Elvidge AR: Ependymomas: A clini cal survey. J Neurosurg 33:428-438, 1970. 3. Barry JF, Harwood-Nash DC, Fitz CR, Byrd SE: Unrecognized atypical tuberous sclerosis diag nosed with CT. Neuroradiology 13:177-180, 1977. 4. Beal MF, Kleinman GM, Ojemann RG, Hochberg FH: Gangliocytoma of third ventricle. Neurology {NY) 31:1224-1228, 1981. 5. Boessel CP, Paulson GW, Kosnik EJ, Earle KM: Brain hamartomas and tumors associated with tuberous sclerosis. Neurosurgery 4:410-417, 1979. 6. Bradley WG, Newton TH, Crooks LE: Physical principles of nuclear magnetic resonance. In Newton TH, Potts DG (eds): Modern Neurora diology, Vol II, Advanced Imaging Techniques. San Francisco, Clavadel Press, 1983, ch 3, pp 15-61. 7. Bradley WG, Waluch V, Yadley RA, Wycoff RR: Comparison of CT and MR in 400 patients with suspected disease of the brain and cervical spinal cord. Radiology 152:695-702, 1984. 8. Brandt-Zawadzki M, Norman D, Newton TH, Kelly WM, Kjos B, Mills CM, Dillon W, Sobel D, Crooks LE: Magnetic resonance of the brain: The optimal screening technique. Radiology 152: 71-77, 1984. 9. Brant-Zawadzki M: Nuclear magnetic resonance imaging: The abnormal brain and spinal cord. In Newton TH, Potts DG (eds): Modern Neuro radiology, Vol II, Advanced Imaging Tech niques. San Francisco, Clavadel Press, 1983, ch 7, pp 159-185. 10. Brant-Zawadzki M, Davis PL, Crooks LE, Mills CM, Norman D, Newton TH, Sheldon P, Kauf man L: NMR demonstration of cerebral abnor malities: Comparison with CT. AJR 140:847854, 1983. 11. Braun IF, Pinto RS, Epstein F: Dense cystic Cra niopharyngiomas. AJNR 3:139-141, 1982. 12. Braun IR, Naidich TP, Leeds NE, Koslow M, Zim-

13.

14.

15.

16. 17.

18.

19.

20.

21. 22.

23.

24. 25.

26.

27. 28.

merman HM, Chase NE: Dense intracranial epidermoid tumors. Radiology 122:717-719, 1977. Britt RH, Silverberg GD, Enzmann DR, Hanberry JW: Third ventricular Choroid plexus arteriovenous malformation simulating a colloid cyst. J Neurosurg 52:246-250, 1980. Brooks BS, El Gammal T, Hungerford GD, Acker J, Trevor RP, Russell W: Radiologic evaluation of neurosarcoidosis: Role of computed tomogra phy. AJNR 3:513-521, 1982. Buonanno FS, Pykett IL, Brady TJ, Black P, New PFJ, Richardson EP, Hinshaw WS, Goldman M, Pohost G, Kistler JP: Clinical relevance of two different nuclear magnetic resonance (NMR) ap proaches to imaging of a low grade astrocytoma. J Comput Assist Tomogr 6:529-535, 1982. Bursztyn EM, Lavyne MH, Aisen M: Empty sella syndrome with intrasellar herniation of the optic chiasm. AJNR 4:167-168, 1983. Bydder GM, Steiner RE, Young IR, Hall AS, Thomas DJ, Marshall J, Pallis CA, Legg NJ: Clinical NMR imaging of the brain: 140 cases. AJR 139:215-236, 1982. Byrd SE, Locke GE, Biggers S, Percy AK: The computed tomographic appearance of cerebral cysticercosis in adults and children. Radiology 144:819-823, 1982. Carr DH, Brown J, Bydder GM, Steiner RE, Weinmann H-J, Speck U, Hall AS, Young IR: Gadolinium-DTPA as a contrast agent in MRI: Initial clinical experience in 20 patients. AJR 143:215-224, 1984. Cinberg JZ: Eosinophilic granuloma in the head and neck: A five year review with report of an instructive case. Laryngoscope 88:1281-1289, 1978. Ciric I, Zivin I: Neuroepithelial (colloid) cysts of the septum pellucidum. J Neurosurg 43:69-73, 1975. Cowley AR, Moody DM, Alexander E, Ball MR, Laster DW: Distinctive CT appearance of cyst of the cavum septi pellucidi. AJR 133:548-550, 1979. Crooks LE, Mills CM, Davis PL, Brant-Zawadzki M, Hoenninger J, ArakawaM, Watts J, Kaufman L: Visualization of cerebral and vascular abnor malities by NMR imaging: The effects of imaging parameters on contrast. Radiology 144:843852, 1982. Daniels DL, Haughton VM, Williams AL, Gager WE, Berns TF: Computed tomography of the optic chiasm. Radiology 137:123-127, 1980. Davis KR, Roberson GH, Taveras JM, New PFJ, Trevor R: Diagnosis of epidermoid tumor by computed tomography. Radiology 119:347353, 1976. Davis PC, Hoffman JC, Weidenheim KM: Large hypothalamic and optic chiasm gliomas in in fants: Difficulties in distinction. AJNR 5:579585, 1984. Di Chiro G, Eiben RM, Manz HJ, Jacobs IB, Schellinger D: A new CT pattern in adrenoleukodystrophy. Radiology 137:687-692, 1980. Diebler C, Dulac O, Renier D, Ernest C, Lalande G: Aneurysm of the vein of Galen in infants aged 2 to 15 months: Diagnosis and natural evolution. Neuroradiology 21:185-197, 1981.

29. Donat JF, Okazaki H, Gomez MR, Reagan TJ, Baker HL, Laws ER: Pineal tumors: A 53-year experience. Arch Neurol 35:736-740, 1978. 30. Dublin AB, French BN: Cysticercotic cyst of the septum pellucidum. AJNR 1:205-206, 1980. 31. Edwards MK, Gilmor RL, Franco JM: Computed tomography of chiasmal optic neuritis. AJNR 4:816-818, 1983. 32. Enzmann DR, Krikorian J, Norman D, Kramer R, Pollock J, Faer M: Computed tomography in primary reticulum cell sarcoma of the brain. Radiology 130:165-170, 1979. 33. Enzmann DR, Norman D, Levin V, Wilson CB, Newton TH: Computed tomography in the fol io wup of medulloblastomas and ependymomas. Radiology 128:57-63, 1978. 34. Fitz CR, Wortzman G, Harwood-Nash DC, Holgate RC, Barry JF, Boldt DW: Computed tomog raphy in Craniopharyngiomas. Radiology 127: 687-691, 1978. 35. Futrell NN, Osborn AG, Cheson BD: Pineal region tumors: Computed tomographic-pathologic spec trum. A JR 137:951-956, 1981. 36. Ganti SR, Antunes JL, Louis KM, Hilal SK: Com puted tomography in the diagnosis of colloid cysts of the third ventricle. Radiology 138:385391, 1981. 37. Ganti SR, Steinberger A, McMurtry JG, Hilal SK: Computed tomographic demonstration of giant aneurysms of the vertebrobasilar system: Re port of eight cases. Neurosurgery 9:261-267, 1981. 38. Godersky JC, Rockswold G, Larson DA: Xanthogranuloma of the third ventricle producing hy drocephalus. Neurosurgery 7:68-70, 1980. 39. Graeb DA, Robertson WD, Lapointe JS, Nugent RA, Harrison PB: Computed tomographic diag nosis of intraventricular hemorrhage: Etiology and prognosis. Radiologу 143:91-96, 1982. 40. Grosvalet A, Ernest C, Diebler C, Sauvegrain J: Computed tomography in precocious puberty of central origin. Ann Radiol 24:32-38, 1981. 41. Guibert-Tranier F, Piton J, Billerey J, Caille JM: Agenesis of corpus callosum. J Neuroradiol 9:135-160, 1982. 42. Hall K, McAllister VL: Metrizamide cisternography in pituitary and juxtapituitary lesions. Ra diology 134:101-108, 1980. 43. Hawkes RC, Holland GN, Moore WS, Corston R, Kean DM, Worthington BS: The application of NMR imaging to the evaluation of pituitary and juxtasellar tumors. AJNR 4:221-222, 1983. 44. Hawkes RC, Holland GN, Moore WS, Worthing ton BS: Nuclear magnetic resonance (NMR) to mography of the brain: A preliminary clinical assessment with demonstration of pathology. J Comput Assist Tomogr 4:577-586, 1980. 45. Hehman K, Norrell H, Howieson J: Subependymomas of the septum pellucidum. J Neurosurg 29:640-644, 1968. 46. Herz DA, Liebeskind A, Rosenthal A, Schechter MM: Cerebral angiographic changes associated with tuberous sclerosis. Radiology 115:647649, 1975. 47. Hirst LW, Miller NR, Allen GS: Sphenoidal pneumosinus dilatans with bilateral optic nerve men ingiomas. J Neurosurg 51:402-407, 1979.

48. Hoyt WF, Meshel LG, Lessell S, Schatz NJ, Suck ling RD: Malignant optic glioma of adulthood. Brain 96:121-132, 1973. 49. Jankowski R, Zimmerman RD, Leeds NE: Cys ticercosis presenting as a mass lesion at fora men of Monro. J Comput Assist Tomogr 3:694696, 1979. 50. Johnson MA, Pennock JM, Bydder GM, Steiner RE, Thomas DJ, Hayward R, Bryant DRT, Payne JA, Levene MI, Whitelaw A, Dubowitz LMS, Dubowitz V: Clinical NMR imaging of the brain in children: Normal and neurologic disease. AJR 141:1005-1018, 1983. • 51. Jooma R, Kendall BE: Diagnosis and management of pineal tumors. J Neurosurg 58:654-665, 1983. 52. Kendall BE, Holland IM: Computed tomography in Alexander's disease. Neuroradiology 20: 103-106, 1980. 53. Kobayashi T, Kageyama N, Kida Y, Yoshida J, Shibuya N, Okamura K: Unilateral germinomas involving the basal ganglia and thalamus. J Neurosurg 55:55-62, 1981. 54. Kumpe DA, Rao KCVG, Garcia JH, Heck AF: Intracranial neurosarcoidosis. J Comput Assist Tomogr 3:324-330, 1979. 55. Lee BCP, Gawler J: Tuberous sclerosis. Radiol ogy 127:403-407, 1978. 56. Lee Y-Y, Lin S-R, Horner FA: Third ventricle meningioma mimicking a colloid cyst in a child. AJR 132:669-671, 1979. 57. Lin S-R, Bryson MM, Gobien RP, Fitz CR, Lee YY: Radiologic findings of hamartomas of the tuber cinereum and hypothalamus. Radiology 127:697-703, 1978. 58. Maitland CG, Abiko S, Hoyt WF, Wilson CB, Okamura T: Chiasmal apoplexy: Report of four cases. J Neurosurg 56:118-122, 1982. 59. Martelli A, Scotti G, Harwood-Nash DC, Fitz CR, Chuang SH: Aneurysm of the vein of Galen in children: CT and angiographic correlation. Neu roradiology 20:123-133, 1980. 60. Matson DD, Crofton FD: Papilloma of the Choroid plexus in childhood. J Neurosurg 17:10021026, 1960. 61. Matthews FD: Intraventricular craniopharyn gioma. AJNR 4:984-985, 1983. 62. Messina AV, Potts DG, Sigel RM, Liebeskind AL: Computed tomography: Evaluation of the pos terior third ventricle. Radiology 119:581-592, 1976. 63. Miller JH, Pena AM, Segall HD: Radiological in vestigation of sellar region masses in children. Radiology 134:81-87, 1980. 64. Mills CM, Brant-Zawadzki M, Crooks LE, Kauf man L, Sheldon P, Norman D, Bank W, Newton TH: Nuclear magnetic resonance: Principles of blood flow imaging. AJR 142:165-170, 1984. . 65. Morrison G, Sobel DF, Kelley WM, Norman D: Intraventricular mass lesions. Radiology 153: 435442, 1984. 66. Murovic JA, Ongley JP, Parker JC, Page LK: Manifestations and therapeutic considerations in pineal yolk-sac tumors: Case report. J Neu rosurg 55:303-307, 1981. 67. Naidich TP, Pudlowski RM, Naidich JB: Com puted tomographic signs of the Chiari II malfor-

68.

69.

70. 71.

72. 73. 74. 75.

76. 77.

78.

79. 80. 81.

82.

83.

mation: Part III. Ventricles and cisterns. Radiology 134:657-663, 1980. Numaguchi Y, Kishikawa T, Fukui M: Prolonged injection angiography for diagnosing intracra nial cavernous hemangiomas. Radiology 131: 137-138, 1979. Numaguchi Y, Kishikawa T, Ikeda J, Fukui M, Kitamura K, Tsukamoto Y, Hasuo K, Matsuura K: Neuroradiological manifestations of suprasel lar pituitary adenomas, meningiomas, and Cra niopharyngiomas. Neuroradiology 21:67-74, 1981. O'Dwyer JA, Newton TH, Hoyt WF: Radiologic features of septooptic dysplasia: deMorsier syn drome. AJNR 1:443-447, 1980. Pascual-Castroviejo I, Villarejo F, Perez-Higueras A, Morales C, Pascual-Pascual SI: Childhood Choroid plexus neoplasms. Eur J Pediatr 140:51-56, 1983. Pavlicek W, Modic M, Weinstein M: Pulse se quence and significance. Radiographics 4:4965, 1984. Petito CK, DeGirolami U, Earle KM: Craniophar yngiomas: A clinical and pathological review. Cancer 37:1944-1952, 1976. Piergras U, Rieger P: Thalamic bleeding: Diag nosis, course, and prognosis. Neuroradiology 22:85-91, 1981. Pinto RS, Cohen WA, Kricheff II, Redington RW, Berninger WH: Giant intracranial aneurysms: Rapid sequential computed tomography. AJR 139:973-977, 1982. Post MJD, David NJ, Glaser JS, Safran A: Pitui tary apoplexy: Diagnosis by computed tomogra phy. Radiology 134:665-670, 1980. Pozzati E, Gaist G, Poppi M, Morrone B, Padovani R: Microsurgical removal of paraventricular cav ernous angiomas. J Neurosurg 55:308-311, 1981. Randell CP, Collins AG, Young IR, Haywood R, Thomas DJ, McDonnell MJ, Orr JS, Bydder GM, Steiner RE: Nuclear magnetic resonance imag ing of posterior fossa tumors. AJR 141:489496, 1983. Rovit RL, Schechter MM, Chodroff P: Choroid plexus papillomas: Observations on radiographic diagnosis. AJR 110:608-617, 1970. Rozario R, Adelman L, Prager RJ, Stein BM: Meningiomas of the pineal region and third ven tricle. Neurosurgery 5:489-495, 1979. Rubenstein LJ: Tumors of the central nervous system. In Rubenstein LJ (ed): Atlas of Tumor Pathology, Series 2, Fascicle 6. Washington, DC, Armed Forces Institute of Pathology, 1972, pp 269-284. Salazar OM, Castro-Vita H, Bakos RS, Feldstein ML, Keller B, Rubin P: Radiation therapy for tumors of the pineal region. Int J Radiat Oncol BiolPhys 5:491-499, 1979. Savoiardo M, Harwood-Nash DC, Tadmor R, Scotti G, Musgrave M: Glioma of the intracranial anterior optic pathways in children. Radiology 138:601-610, 1981.

84. Scheithauer BW: Symptomatic subependymoma. J Neurosurg 49:689-696, 1978. 85. Schubiger O, Valavanis A, Hayek J: Computed tomography in cerebral aneurysm with special emphasis on giant intracranial aneurysms. J Comput Assist Tomogr 4:24-32, 1980. 86. Sholkoff SD, Kerber C, Cramm RE, Silverman L: Parasellar choristoma. AJR 128:1051-1052, 1977. 87. Sipponen JT, Sepponen RE, Sivula A: Nuclear magnetic resonance (NMR) imaging of intracerebral hemorrhage in the acute and resolving phases. J Comput Assist Tomogr 7:954-959, 1983. 88. Steinhoff H, Lanksch W, Kazner E, Grumme T, Meese W, Lange S, Aulich A, Schindler E, Wende S: Computed tomography in the diagnosis and differential diagnosis of glioblastomas: A quali tative study of 295 cases. Neuroradiology 14:193-200, 1977. 89. Swartz JD, Zimmerman RA, Bilaniuk LT: Com puted tomography of intracranial ependymo mas. Radiology 143:97-101, 1982. 90. Symon L, Ganz JC, Burston J: Granular cell myoblastoma of the neurohypophysis: Report of two cases. J Neurosurg 35:82-89, 1971. 91. Takeuchi J, Handa H, Otsuka S, Takebe Y: Neu roradiological aspects of suprasellar germinoma. Neuroradiology 17:153-159, 1979. 92. Thompson JR, Harwood-Nash DC, Fitz CR: The neuroradiology of childhood Choroid plexus neo plasm. A JR 118:116-133, 1973. 93. Wehrli FW, MacFall JR, Newton TH: Parameters determining the appearance of NMR images. In Newton TH, Potts DG (eds): Modern Neurora diology, Vol II, Advanced Imaging Techniques. San Francisco, Clavadel Press, 1983, ch 5, pp 81-117. 94. Wilson CB: A decade of pituitary microsurgery: The Herbert Olivecrona Lecture. J Neurosurg 61:814-833, 1984. 95. Wolpert SM: The radiology of pituitary adeno mas. Semin Roentgenol 19:53-69, 1984. 96. Wolpert SM, Scott RM: The value of metrizamide CT cisternography in the management of cere bral arachnoid cysts. AJNR 2:29-35, 1981. 97. Wolpert SM: Vascular studies of congenital anomalies. In Newton TH, Potts DG (eds): Ra diology of the Skull and Brain, Vol 2, Book 4. St Louis, CV Mosby, 1974, pp 2717-2719. 98. Wood JH, Zimmerman RA, Bruce DA, Bilaniuk LT, Norris DG, Schut L: Assessment and man agement of pineal-region and related tumors. SurgNeurol 16:192-210, 1981. 99. Zee CS, Segall HD, Apuzzo MLJ, Ahmadi J, Dobkin WR: Intraventricular cysticercal cysts: Fur ther neuroradiologic observations and neuro surgical implications. AJNR 5:727-739, 1984. 100. Zimmerman RA, Bilaniuk LT, Wood JH, Bruce DA, Schut L: Computed tomography of pineal, parapineal, and histologically related tumors. Radiology 137:669-677, 1980.

12 Anterior Transcallosal and Transcortical Approaches William Shucart, M.D.

Lesions in the anterior third ventricle are uncommon; therefore, most surgeons have limited experience with lesions of this region. Enthusiasm for operation has often been dampened because surgical access to the anterior third ventricle requires disruption of normal brain tissue and because many of these tumors are not resectable. Over the past several years, however, there has been increasing interest in the diagnosis and treatment of lesions in this area stimulated by increasing sophistication in imaging techniques, refinements in surgical instrumentation, the realization that many lesions are removable, and the development of surgical avenues that do not cause neurological deficits. History Historically, three approaches have been used to reach the anterior third ventricle: the transcortical-transventricular approach going through a frontal lobe, the transcallosaltransventricular approach, and the subfrontal approach through the lamina terminalis (13, 18). More recently, computerized tomography (CT)-assisted stereotaxic biopsies have been used to establish a diagnosis in third ventricular lesions (1). The subfrontal approach has been used primarily for masses arising in the suprasellar area that invaginate into the third ventricle. Both transventricular approaches have been used for tumors arising within the third ventricle (4, 7, 9). The stereotaxic method is generally used for lesions thought to be nonresectable. The most commonly used approach for solely intraventricular lesions is probably the transcortical-transventricular approach initially described by Dandy (5). He also described a transcallosal approach that he used for posterior third ventricular lesions. The anterior transcallosal approach has often been suggested (6, 12), but until recently it was not used widely because of concern about the clinical effect of sectioning a portion of the corpus callosum. There have been suggestions in the literature since the 1940s that

sectioning the more anterior portion of the corpus callosum is not associated with distinct symptoms, whereas complete callosal section has been shown to produce a broad array of definite and persistent behavioral abnormalities (8). Since the 1960s, the surgical literature has contained an increasing number of reports emphasizing the efficacy of the transcallosal approach and confirming the predictions that there would be no specific deficits referable to division of the anterior portion of the body of the corpus callosum (3, 11, 15-17). At present, there are advocates for both the transcallosal and the transcortical approaches. The differences between and the advantages of these two approaches are discussed in this chapter. Surgical Approaches The distinction between masses arising within and expanding the third ventricle and those impinging from outside is important for determining the best operative approach, but can be difficult. There are no certain clinical criteria to ascertain the origin of the mass. Interference with normal functioning of the hypothalamic-pituitary axis and visual field defects are most often seen with lesions arising subfrontally below the third ventricle, but tumors growing within the ventricle may cause similar problems. CT or magnetic resonance imaging with coronal and sagittal reconstructions often define the exact anatomy and origin of the lesion (Fig. 12.1). Angiography may also be helpful in evaluating the origin of the lesion by showing the displacement of the major arteries in that area, particularly the branches of the anterior cerebral arteries. If the lesion in the vicinity of the anterior third ventricle is small, the approach is best determined by the presence or absence of a visual field disturbance; if present, the approach should be subfrontal; if not, the approach should be transventricular. In cases with large lesions and an associated visual field defect, a large craniotomy should be performed to allow access to the subfrontal area and the use of one of the transventricular approaches if needed. If the ventricles are enlarged, which they usually are if a tumor is present, access to a foramen of Monro and ultimately the third ventricle is easily accomplished with either the transcortical or the transcallosal technique. Access to both sides of the third ventricle is obtainable with either exposure but because of the angle of vision the view of the ipsilateral portion of large lesions can be limited using the transcortical route. The line of vision to the depths of the anterior third ventricle is better with the transcallosal method (Fig. 12.2). If the ventricles are small the transcallosal approach is far superior. Use of the transcortical approach to the ventricle in the absence of hydrocephalus requires disruption of a large amount of cortex and white matter, and the maintenance of retraction can be difficult. This route also significantly limits mobility if departure from the initial plane of entry into the lateral ventricle is needed. The advantages of the transcallosal approach to the third ventricle are that the anatomy is constant, the distance to the third ventricle is shorter than transcortically, there is greater flexibility to explore the entire anterior-posterior extent of the third ventricle with no disruption of hemispheral tissue, no cortical incision and therefore no anticonvulsants are required, there is excellent unobstructed vision to the depths of the anterior third ventricle, and ventricular size is irrelevant. Advantages of the transcorticaltransventricular exposure are that there is less chance of compromising an essential draining vein going to the sagittal sinus and less chance of injuring the pericallosal arteries.

Figure 12.1. Л. Standard horizontal CT scan showing large intraventricular mass. B. Sagittal reconstruction showing clearly that the tumor is totally within the third and lateral ventricles. C. Coronal reconstruction showing that the tumor is within the third ventricle and extending primarily into the left lateral ventricle.

Figure 12.2. A. Comparison of angle of approach of transcallosal and transcort-ical techniques. B. Ease of access to entire normal size anterior ventricular system using the transcallosal approach. C. Transcortical approach to a dilated ventricular system; this angle of approach allows less mobility than the transcallosal approach.

As stereotaxic techniques are now so sophisticated, accurate, and safe they should be used to obtain tissue for the diagnosis of third ventricular lesions thought to be nonresectable on the basis of preoperative studies. If the diagnosis shows a lesion that is treatable definitive surgery is performed. Operative Technique Transcallosal-Transventricular Approach The patient is placed in the surgical headholder with the head straight and elevated about 20°. A coronal skin incision provides the most flexibility, best demonstrates the surface landmarks for proper placement of the bone flap, and is the most cosmetically acceptable. A right-sided exposure is used in almost all cases so the coronal incision should extend most inferiorly on the right side to a point approximately 1 cm anterior to the external auditory meatus and approximately 2 cm above the zygoma. A curvilinear incision is made parallel to and about 2 cm behind the coronal suture across the midline toward the front of the left ear. The incision should be carried to the left far enough to allow reflection of the skin flap at least 6 cm anterior to the coronal suture. This usually requires an incision extending to a point approximately 5 cm above the left zygoma. The scalp flap is reflected anteriorly and secured in the usual fashion. The coronal suture and the junction of the sagittal and coronal sutures should be identified on the skull surface (Fig. 12.3). A triangular free bone flap is adequate for most procedures as maximal exposure is required at the midline and not over the surface of the cortex. This is different from a transcortical exposure in which a rectangular bone flap is necessary to expose a greater portion of the cortical surface. The bone flap only need extend 6 to 7 cm laterally because the exposure does not require significant displacement of the hemisphere. The bone flap is made using at least two burr holes. One burr hole should be made with its medial margin at the midline just behind the junction of the right coronal suture and the sagittal suture. The second burr hole should be made with its medial margin at the midline approximately 7 cm anterior to the first burr hole. If the lesion in the third ventricle is very large or occupies a posterior position in the third ventricle, it is sometimes helpful to place the bone flap more posteriorly. The burr holes are each moved posteriorly about 2 cm so that the most posteriorly placed burr hole would be 2 cm behind the coronal suture. This placement allows a little more direct inferior line of vision for the surgeon. The major concern in moving the flap in this fashion is the possibility of compromising a draining vein behind the coronal suture. Greater access to the posterior portion of the third ventricle can be gained by lowering the position of the patient's head on the operating table and angling the operating microscope in a more posterior direction (Fig. 12.4). The dura mater is stripped from the undersurface of the bone and the connection between these two burr holes should be made with a Gigli saw (we avoid using the craniotome in the area of the sagittal sinus). The remainder of the bone work can be done using a craniotome or by placing a third burr hole approximately 6 to 7 cm to the right of the midline equidistant from the two other burr holes. Several surgeons have suggested exposing the entire width of the sagittal sinus, but this serves no useful purpose and only increases the possibility of injury to the sinus. There should be no retraction of the sagittal sinus, and the bony margin at the edge of the sagittal sinus helps prevent inadvertent pressure

Figure 12.3. Relationship of skin incision (A) and burr holes (B) to cranial sutures. against it by retractors. It is important to have the medial aspects of the two midline burr holes at least at the lateral margin of the sagittal sinus so that the exposure goes straight inferiorly and one is not obliged to work under a ridge of bone. If the exposure is not at the midline the rim of bone must be removed using rongeurs. Removing the inner table of bone over the sinus with an angled Kerrison rongeur is helpful. The bone edges are waxed, the holes for wires to secure the bone when it is replaced

Figure 12.4. Possible alterations of microscope angle to provide views of different regions of the third ventricle. are made, and the wires are passed through the holes. After this, the dura mater is tented to the surrounding pericranium in the usual fashion (Fig. 12.5). The dural opening follows the outline of the bone flap with its base being hinged at the sagittal sinus. The dural opening should go to the right lateral edge of the sagittal sinus. At this point, if the ventricles are not enlarged, the patient is given 25 g of mannitol intravenously to shrink the brain and decrease the need for hemispheral retraction. If there is hydrocephalus, no dehydrating agent is given because there will be marked relaxation of the brain when the ventricular system is opened. As the dura mater is being elevated care is taken not to avulse any of the corticodural veins that may be present. Although it is preferable to avoid sacrificing any draining veins from the cortex to the saggittal sinus it is usually necessary to divide one or two to allow retraction of the hemisphere. The veins anterior to the coronal suture can generally be sacrificed with impunity, but if there is a very large draining vein it should be preserved if possible. One or two traction sutures are placed through the base of the dural flap just lateral to the sagittal sinus and are suspended over the crani-otomy to facilitate retraction of the dura. The presence of the medial bony margin prevents too much pressure on the saggittal sinus by these sutures. If the flap has been properly placed, the superior portion of the falx and the medial portion of the cerebral hemisphere will be exposed. Only 3 to 4 cm of longitudinal free space between the hemisphere and falx are required for adequate retraction. Since the medial margin of the

Figure 12.5. A. Reflection of scalp flap to provide room anterior to the coronal suture. В Gigli saw used between the two medial burr holes С: Craniotomy showmg the relationship of the sagittal sinus to the medial bone margin and the dural incision [dashed line). D. Removal of the inner table of bone over the sagittal sinus allowing excellent visualization while leaving the bone edge to prevent undue retraction of or pressure on the sagittal sinus. E. Craniotomy completed showing exposure of medial convexity and falx.

Figure 12.6. A. Relationship of the foramen of Monro to a line between the coronal suture and the external auditory meatus. B. Line of incision through the corpus callosum, well anterior to the motor cortex. bone flap is 6 to 7 cm in length there is opportunity to select an area for retraction that will compromise the smallest amount of venous drainage. Before retraction of the right cerebral hemisphere from the falx, it is imperative to be oriented to the landmarks that lead toward the proper portion of the corpus callosum. The best guide is an imaginary line drawn from the coronal suture at the midline to the external auditory meatus. This path will lead toward the midportion of the corpus callosum (or slightly anteriorly), and if continued would go through the foramen of Monro. The coronal suture is in the operative field and the external auditory meatus is easily palpated through the drapes so this poses no difficulty. If the retractor is placed too anteriorly and the line of retraction is located too anteriorly, abundant adhesions between the frontal lobes will be encountered. There should be very few adhesions between the frontal lobe and the falx. Rarely, one is able to identify the precentral sulcus on the medial aspect of the hemisphere, and this can be used as a guide to the posterior limit of the callosal division (Fig. 12.6). The right hemisphere is gently retracted using a 15-mm-wide malleable retractor. The line of retraction is straight down toward the bottom of the falx without deviation to either side. As the inferior margin of the falx is reached it becomes slightly more difficult to identify the midline, but this is not usually a significant problem. The corpus callosum is identified by its very white color and not by its relationship to the pericallosal arteries. The position of the arteries varies depending on the size of the ventricular system. When retracting the hemisphere, one may see the callosomar-ginal artery as it courses above the cingulate gyrus and mistake it for a pericallosal artery. As retraction is continued and the actual pericallosal artery is seen on the other side of the cingulate gyrus, one may mistake the gyrus for the corpus callosum, thinking that the callosomarginal artery represents a pericallosal artery. The cingulate gyrus, however, looks like typical cortex with its gray color and pial vasculature. It is helpful to use two retractors to reach the corpus callosum, the 15-mm retractor against the right hemisphere and a 10-mm retractor against the left hemisphere. The retractors are gently "walked" down between the hemispheres until the strikingly white corpus

callosum is identified (Fig. 12.7). Neither pericallosal artery may be apparent if the ventricles are enlarged, but there is no need to search for them if there is adequate corpus callosum exposed to gain entry to the ventricular system. If both pericallosal arteries are seen, it is best to go between them to avoid the division of penetrating branches going to either hemisphere. Once the corpus callosum has been identified the hemispheral retractor is removed, a protective covering of cottonoid, rubber, or something similar is placed over the medial surface of the hemisphere, and then the retractor is fixed in place at the depth of the incision, retracting the hemisphere laterally to expose the corpus callosum. A 20- to 25-mm retractor is used to spread the pressure of retraction more widely and to provide a gauge as to the length of the callosal incision. The callosal incision generally need not be more than 2 to 3 cm in length. Before any further retraction is done the corpus callosum is divided (Fig. 12.8). The division of the corpus callosum is confined to the anterior third of the body and the ease of division is related to its thickness. If there is

Figure 12.7. White corpus callosum seen at the depth of field with the right pericallosal artery, right cingulate gyrus, and right callosomarginal artery seen to the right side. The left cingulate gyrus is seen but the left pericallosal artery is not apparent.

Figure 12.8. A. Retraction of the right hemisphere laterally to expose the corpus callosum. B. Retractors against both hemispheres showing both pericallosal arteries and the right callosomarginal artery. C. Corpus callosum being divided after two small pericallosal artery anastomoses have been sacrificed.

hydrocephalus the corpus callosum is usually quite thin with entry into the ventricles being very easily accomplished. If the ventricles are not enlarged the corpus callosum may be a centimetre or so thick and require considerably more dissection to get through. The corpus callosum is avascular and can be divided with a blunt small dissector such as a Penfield #4 or it can be divided using the bipolar cautery and a small suction (#5 French). As soon as the corpus callosum is divided, cerebrospinal fluid escapes from the ventricle and the cerebral hemisphere relaxes in proportion to the ventricular size (i.e., the larger the ventricles the greater the hemispheral relaxation). At this point the hemispheral retractor is repositioned over the edge of the corpus callosum. If further retraction is required, a 15- to 20-mm wide retractor can be placed against the left side of the callosal incision. If the second retractor is used, care must be taken not to create pressure against the sagittal sinus (Fig. 12.9). Once the corpus callosum is divided, orientation is accomplished by identifying the major landmarks within the right lateral ventricle: the Choroid plexus, the thalamostriate vein, and the septal vein. The foramen of Monro is found by following the Choroid plexus and the thalamostriate vein anteriorly. The left lateral ventricle may be entered but the direction of the Choroid plexus as it approaches the foramen of Monro will make this immediately apparent. A cavum septum pellucidum may be entered and is confusing because no intraventricular structures are seen. This usually becomes quickly apparent and generally the presence of a cavum septum pellucidum has been noted on the preoperative CT scan. If the lesion being treated is in the lateral ventricle, the exposure obtained is usually adequate for definitive treatment. If the lesion is in the third ventricle, the operation can occasionally be performed through a dilated foramen of Monro (Fig. 12.10). If greater exposure is needed the interforniceal or subchoroidal approaches should be utilized (2, 10, 14) (Figs. 12.11 and 12.12). The utilization of either of these latter two approaches obviates the need for division of one of the fornices or removal of the anterior portion of the thalamus to widen the foramen of Monro (Fig. 12.13). Utilization of the transcallosal approach allows visualization of the depths of the third ventricle with great ease. Using this approach, exposure can be obtained down to the upper portion of the basilar artery, if the tumor extends that far (Figs. 12.14 and 12.15). If there is hydrocephalus or if there is a probability that one of the foramina of Monro will be compromised, the septum pellucidum should be fenestrated. In the presence of hydrocephalus the septum is stretched quite thin and a fenestration of about 1 cm2 is easily made. It is advisable to make the fenestration through the thinnest part of the septum and avoid dissection near its base. If there is a cavum septum pellucidum both walls should be fenestrated. During dissection of lesions in the lateral ventricle a cottonoid is placed over the foramen of Monro to help keep debris from getting into the cerebrospinal fluid. When working in the ventricular system, one uses Ringer's lactate as an irrigating solution. When the lesion has been removed the ventricular system should be thoroughly irrigated to remove any debris and before closure there should be immaculate hemostasis. At this point the retractors are removed and the dura mater is reapprox-imated in a watertight fashion. The bone flap is replaced and the wound is closed in the usual fashion. Ventricular drainage is rarely, if ever, necessary. Prophylactic antibiotics have not been used unless there are other indications for their use, such as an abscess, and prophylactic anticonvulsants have not been used.

Figure 12.9. A. Retractor placed over the cut edge of the corpus callosum. B. Same exposure as A looking from posterior to anterior. C. View of the right foramen of Monro showing Choroid plexus and the thalamostriate vein approaching the posterior portion of the foramen and the septal vein approaching anteriorly. D. Less magnified view of C.

Figure 12.10. Cystic craniopharyngioma seen through a dilated right foramen of Monro. Choroid plexus is slightly displaced laterally to show the location of the choroidal fissure.

Figure 12.11. A. Choroid plexus displaced medially to show the underlying choroidal fissure. B. View from posterior demonstrating access to the third ventricle via the subchoroidal fissure.

Figure 12.12. A. Incision into the septum pellucidum. B. Leaves of the septum pellucidum being bluntly separated. C. View from posterior showing interforniceal access to the third ventricle after the leaves of the septum pellucidum are separated.

Figure 12.13. Autopsy specimen showing separated leaves of the septum pellucidum and fornices to show normal planes that can be developed either between the fornices or between a fornix and the thalamus (subchoroidal).

Figure 12.14. Apex of the basilar artery (bifurcation) seen through a dilated foramen of Monro after removal of a craniopharyngioma.

Figure 12.15. A. Horizontal and coronal reconstruction of a cystic craniopharyngioma extending from the sella up into the third ventricle. B. Comparable scans after a transcallosal approach and removal of the tumor. The density in the right lateral ventricle is a shunt catheter. Transcortical-Transventricular Approach There are many similarities between the transcallosal and the transcortical approaches. The position of the patient in the surgical headholder and the coronal skin incision are the same. The approach is best performed from the right side (the nondominant hemisphere) unless it is being used for a lesion that is primarily in the left lateral ventricle or is a third ventricular lesion with significant extension into the left lateral ventricle. The bone flap extends approximately the same distance in the anteriorposterior direction but much more exposure is needed over the cortical surface so a rectangular bone flap that extends approximately 8 cm lateral to the midline is required. The two medial burr holes are placed as they are for the transcallosal approach, but the exposure of the lateral edge of the sagittal sinus is not as important and it is adequate if the medial edge of the bone flap is a centimeter or so lateral to the midline. If the lesion is very posterior in the third ventricle, some have advocated placing the bone flap a little more posteriorly so that the coronal suture is roughly at the midportion of the flap. It is preferable to limit the amount of bone removal behind the coronal suture to about 2 cm to allow an adequate margin between the operative area and the motor cortex. When the bone flap is removed, the dura mater is opened parallel to the edges of the bone flap using the most medial portion along the sagittal sinus as the base of the dural flap, which is reflected medially. It is not necessary to expose the lateral edge of the sagittal sinus nor to elevate the dura mater enough to jeopardize the large cortical veins draining into the sagittal sinus. A cortical incision is made paralleling the course of, and in the center of, the middle frontal gyrus for a length of 3 to 4 cm (Fig. 12.16). If this approach is used with normal size ventricles the longer incision is preferable because of the much greater mass of hemispheral tissue that must be retracted to enter the lateral ventricle. Dehydrating agents are used in these cases (usually osmotic diuretics). If the ventricles are significantly enlarged, a smaller incision can be used because the cortical mantle is thinner and the enlarged cavity provides excellent exposure down to the third ventricle. When the incision has been made through the cortex the remaining dissection of white matter to enter the ventricle is done bluntly using malleable retractors or Penfield dissectors (Fig. 12.17). The plane of dissection should generally be directed toward the foramen of Monro on the ipsilateral side. The anterior-posterior guideline is an imaginary line drawn from the coronal suture to the external auditory meatus. The medial-lateral guideline, particularly if the

Figure 12.16. A. Craniotomy landmarks for the transcortical approach. B. Relationship of the craniotomy to the middle frontal gyrus. C. Relationship of the middle frontal gyrus incision to the coronal and sagittal sutures.

Figure 12.17. A. Retractors opening up a gyrus incision. B. Path of retraction to the lateral ventricle.

ventricles are not enlarged, is much more critical. The foramina of Monro are, for practical purposes, midline structures. The plane should be directed from the middle frontal gyrus toward the medial canthus of the contralateral eye. Such a plane will generally cross the foramen of Monro or very close to it (Fig. 12.18). Twenty to twentyfive-millimeter-wide malleable retractors are then used to maintain the exposure after the underlying brain is covered with protective material. The intraventricular anatomy is identified as described previously. This operation, once the ventricles are entered, is identical to the transcallosal approach and the closure is performed in much the same way (Fig. 12.19). Some have advocated making the cortical incision in the superior rather than the middle frontal gyrus. This approach avoids the benefits of the other approaches while retaining the problems. The major advantage of a cortical incision is the decreased chance of compromising the draining veins to the sagittal sinus. The chance of damaging these veins is much greater with the superior frontal gyrus rather than the middle frontal gyrus approach. The slight improvement in angulation for the surgical approach obtained by using the superior frontal gyrus is not as good as that obtained by using the interhemispheric approach and adds the problems of the cortical incision and hemispheral retraction. With either technique, fungating or diffuse gliomas can obscure intraventricular landmarks and immediate orientation can be difficult. Pre-operative imaging techniques so clearly display the anatomy that this usually presents little problem in the operating room. Tumors involving the corpus callosum distort the anatomy and normal white color of that structure, but the tumor is usually very obvious or there is some normal corpus callosum that can be identified. Complications Most of the complications seen with operation of intraventricular lesions have been related to the location and nature of the primary lesion rather than to the approach. Examples of these are diabetes insipidus and akinetic mutism. The latter occurs with tumors in or dissection of the anterior third ventricle, but can be seen transiently with excessive retraction against both cingulate gyri. Aseptic meningitis has occurred in approximately 15% of patients who have undergone intraventricular operations. It is seen more often in patients who had tumors such as malignant astrocytomas, which spill blood and necrotic tissue into the cerebrospinal fluid. All patients are treated with corticosteroids preoper-

Figure 12.18. A. Anterior-posterior orientation to the foramen of Monro using an imaginary line from the coronal suture to the external auditory meatus. B. Mediallateral orientation of the frontal gyrus using a plane from the middle frontal gyrus to the medial canthus of the contralateral eye. С and D. Schematic presentation of the approach to the foramen of Monro using the transcortical route.

Figure 12.19. A. Retractors separating hemispheral white matter after entry into the lateral ventricle. B. Lateral view showing the anterior-posterior plane of hemispheral retractors. atively and for about 1 week postoperatively. The signs of aseptic meningitis (headache, low grade fever, stiff neck) usually become apparent when the corticosteroids are decreased or stopped. Each case has shown cerebrospinal fluid changes consisting of moderate pleocytosis with a decrease in the sugar content, but no organisms have been seen or cultured in any case. The patients responded symptomatically to the reinstitution of corticosteroid therapy and required continued treatment for 5 to 30 days. In about half of the patients undergoing third ventricular operations, a ventriculoperitoneal shunt has been placed before definitive operation because of the severity of symptoms from increased intracranial pressure. With the transcallosal approach the subsequent decrease in ventricular size did not create any problem at the time of operation. In those patients without a shunt placed preoperatively, neither prenor postoperative ventricular drainage was used. It would theoretically be possible to see acute hydrocephalus or an increase in intracranial pressure in nonshunted patients postoperatively if debris from the surgical procedure were to obstruct the aqueduct; we have not seen that complication. The ventriculostomy created by dividing the corpus callosum is probably not permanent but may function temporarily to help alleviate hydrocephalus in the unshunted patient. The cortical incision in the transcorticaltransventricular approach may function in the same manner. In essentially all patients we fenestrate the septum pellucidum to avoid the problem of a trapped or isolated ventricle secondary to unilateral foramen of Monro occlusion.

The two major complications in our experience using the transcallosal approach have been related to compromise of venous drainage. In one case the sagittal sinus was occluded by a retractor and led to venous infarction of the anterior portions of both hemispheres. This is a complication that can and should be avoided by careful attention to retractor placement. The other complication was related to venous infarction of the right frontal lobe secondary to division of a very large vein draining from the brain to the sagittal sinus. Despite the fact that this large draining vein was anterior to the coronal suture, it clearly was an essential vein. We have seen this only once in a large number of cases. The suggestion has been made that preoperative arteriography demonstrating the venous drainage pattern can help in the selection of the proper side for the approach. We have not found this helpful. No significant clinical deficit subsequent to division of the anterior third of the body of the corpus callosum has been seen. The deficit most likely to occur would be some impairment of interhemispheral transfer of information. In our series, as well as others, there has been no demonstrable deficit in any sphere including memory, performance, or affect subsequent to division of this limited portion of the corpus callosum. We have not seen any acute disconnection syndromes related to division of the corpus callosum. The neurological deficits due to the transcortical-transventricular approach have included contralateral hemiplegia and major motor seizures. We have not seen the gastrointestinal hemorrhage others have reported with operation in the third ventricle. Summary At present all lesions in the third ventricle are approachable at least to the extent of obtaining enough tissue to make a definitive diagnosis. In masses thought to be malignant or nonremovable a CT-assisted stereotaxic biopsy should be done. Surgical access to the anterior third ventricle can be obtained using one of several different approaches; the route utilized should be determined by the location of the lesion, ventricular size, and associated signs suggesting involvement or lack of involvement of structures outside the third ventricle. For lesions solely within the anterior third ventricle a transcallosal approach is used, for lesions that seem to arise subfrontally but have a significant third ventricular component a combined subfrontal-transcallosal approach is used, and for practical purposes we no longer use the transcortical approach to the third ventricle. References 1. Apuzzo MLJ, Sabshin JK: Computed tomographic guidance stereotaxis in the management of intracranial mass lesions. Neurosurgery 12:277-258, 1983. 2. Busch E: A new approach for the removal of tumors of the third ventricle. Acta Psychiatr Neurol 19:57-60, 1944. 3. Cassinari V, Bernasconi V: Tumori della parte anteriore del terzo ventricolo. Acta Neurochir (Wien) 11:236-271, 1963. 4. Cristensen JD: Third ventricle tumors: Clinical findings and surgical results in 28 consecutive cases. Excerpta Med Int Congr Series 36:E61-E62, 1961. 5. Dandy WE: Diagnosis, localization and removal of tumors of the third ven tricle. Bull Johns Hopkins Hosp. 33:188-189, 1922. 6. Ehni G: Interhemispheric and percallosal (transcallosal) approach to the cingulate gyri, intraventricular shunt tubes, and certain deeply placed brain lesions. Neurosurgery 14:99-110, 1984. 7. French JD, Bucy PC: Tumors of the spetum pellucidum. J Neurosurg 5:433449, 1948. 8. Gordon HW, Bogen JE, Sperry RW: Absence of deconnexion syndrome in two

patients with partial section of the neocommissures. Brain 94:327-336, 1971. 9. Greenwood J Jr: Radical surgery of tumors of the thalamus, hypothalamus, and third ventricle area. Surg Neurol 1:29-33, 1973. 10. Lavyne MH, Patterson RH Jr: Subchoroidal transvelum interpositum ap proach to mid-third ventricular tumors. Neurosurgery 12:86-94, 1983. 11. Long DM, Chou SN: Transcallosal removal of Craniopharyngiomas within the third ventricle. J Neurosurg 39:563-567, 1973. 12. Milhorat TH, Baldwin M: A technique for surgical exposure of the cerebral midline: Experimental transcallosal microdissection. J Neurosurg 24:687691, 1966. 13. Rhoton AL Jr, Yamamoto I, Peace DA: Microsurgery of the third ventricle: Part 2. Neurosurgery 8:357-373, 1981. 14. Scarff ТВ, Reigel DH, Lyons ТА: Transcallosal approach to the 3rd ventricle in children. Neurosurgery 2:154, 1978 (abstr). 15. Shucart WA, Stein BM: Transcallosal approach to the anterior ventricular system. Neurosurgery 3:339-343, 1978. 16. Van Den Bergh R, Brucher JM: L'abord transventriculaire dans les craniopharyngiomes du triosieme ventricule: Aspects neurochirurgicaux et neuropathologiques. Neurochirurgie 16:51-65, 1970. 17. Winston KR, Cavazzuti U, Arkins T: Absence of neurological and behavioral abnormalities after anterior transcallosal operations for third ventricular lesions. Neurosurgery 4:386-393, 1979. 18. Yamamoto I, Rhoton AL Jr, Peace DA: Microsurgery of the third ventricle: Part 1. Neurosurgery 8:334-356, 1981.

13 Considerations in Transforaminal Entry George Ehni, M.D., and Bruce Ehni, M.D.

Surgical Technique The first consideration in planning transforaminal entry to a lesion in the region of or within the foramen of Monro must be whether such an entry should be made. Foraminal entry with a needle, freehand or with stereotaxic guidance, or by open operation may be appropriate for masses whose nature is accurately perceived, but catastrophically inappropriate for a mass that seems to be third ventricular in origin by computerized tomography (CT) (20) perhaps, but that will bleed uncontrollably if it proves to be a high-lying anterior communicating or basilar tip aneurysm — one of the classical disasters in neurological surgery. Such an outcome results from having insufficient information. Before foraminal entry is begun, vertebral angiography must reinforce CT in case of doubt to avoid making the diagnosis of aneurysm at the moment the knife or needle penetrates the capsule of the presumed colloid cyst (3, 13, 43). Bull and Sutton (6), who gave precise directions for the conduct of pneumoventriculography when CT was unknown and angiography was little used and hazardous, said that aneurysms "cannot possibly be confused" (with colloid cysts), but they have been in the past and may be again. Magnetic resonance imaging (MRI) should be useful in the identification of a mass in the foramen of Monro or anterior third ventricle as an exophytic component of a thalamic glioma for which foraminal entry might be found wanting. The next consideration must be the choice of approach to the foramen. Shall it be stereotaxic or freehand needle biopsy without craniotomy (4, 21, 27, 40)? Should it be unilateral craniotomy and a transcortical approach through one (usually the right) lateral ventricle to the foramen of Monro (5, 7, 13, 19, 25, 32, 38, 39, 50-52, 54)? Or should it be an interhemispheric percallosal approach through the midline of the roof of the third ventricle (2, 5, 7, 13) or percallosal into either the right or the left or both lateral ventricles (1, 2, 15, 22, 27, 47, 49)? Stereotaxic equipment and methods seem more attractive to some than is supported by the relatively sparse literature advocating this method of treatment of colloid cysts. It seems illogical to put a needle into a cyst, permitting aspiration of some of its content, when total removal in a

single stage is feasible. Advantages claimed for the aspiration method of treatment include foreknowledge of the nature of the mass by biopsy (4), negligible morbidity and mortality, reduced time in the hospital, and a presumed permanence of good effect (4, 21, 27, 40). But relatively few procedures of this sort have been reported and the period of postoperative observation is limited to fewer than 10 years. Patients have died of recurrent colloid cysts when the initial extirpations were incomplete (4, 11, 32, 38, 47, 49) and there is little reason to think that aspiration of the cyst content will interrupt continued elaboration of the product of the cyst lining. A cyst wall remnant is said to be responsible for a recurrence rate of 10% (11). Another reason to eschew cyst puncture is that the cyst content cannot be known until after the cyst is punctured, possibly with spillage into the ventricular system. Not all of these cysts contain benign colloid or cellular debris; in some, cholesterin crystals, hemosiderin, and mucus are present (6, 11, 18, 24, 37, 42, 45, 46, 48) and ventriculitis of the type provoked by the fluid of Craniopharyngiomas and dermoid cysts may be anticipated. Shunting of hydrocephalus due to a colloid cyst as definitive treatment is hazardous, although the shunt may function well for a long period. When shunting patency fails on some future date with immediate ventricular decompression unavailable, sudden death may occur. Shunting or ventriculostomy before a definitive operation to effect rescue in a deteriorating situation (8) or to allow for decompression and reduction of ventricular size and a lessened probability of cortical collapse after cortical incision to a ventricle (5) are sometimes useful tactics. Perhaps an aspirated colloid cyst will not refill nor a shunt fail with dramatic and catastrophic suddenness, but such treatments must be considered temporizing, incomplete, and less than wholly satisfactory solutions of a problem best solved by total extirpation of the mass by a safe method (38). Most surgeons base their preferences for the transcortical or the interhemispheric method on safety, considering the level of skill and experience to be applied; ease of access depending on whether the ventricles are large or small; liability to or freedom from postoperative epilepsy, paresis, and other unwanted residues of the operation; and quality of the end result. Another aspect that should be considered in choosing between transcortical and percallosal approaches is whether the operation provides the most versatile and useful access considering the position and nature of the lesion. If it has been determined that the mass is eccentric, perhaps bulging out of the foramen of Monro or bulging the adjacent septum pellucidum over to one side, or is invading the caudate nucleus (9, 11, 30, 54) (and if the lesion has been proven not be an aneurysm), a unilateral approach through the nondominant frontal lobe, particularly if the ventricle is dilated, will provide excellent exposure and an opportunity for total removal with minimal deficit, except for the high probability of subsequent seizures (4, 19, 32, 34, 39) and some risk of contralateral paresis if the incision is carried too far toward the rolandic fissure or retraction is crudely applied. There is also, of course, the risk of cortical collapse (5, 39). This approach gives a unilateral view of the septum pellucidum and the interventricular foramen but, if there is nothing to be gained by looking at the lesion from the side to which it is minimally, if at all, protuberant, the total avoidance of veins bridging to the sagittal sinus and the pericallosal arteries if using the interhemispheric approach may be considered an advantage. Colloid cysts and other lesions within the third ventricle, even if of good size on radiographic study, may not be visible on inspection of the region of the foramen of Monro. The foramen may be slitlike and normal-

looking, although just within it lies a cyst 1 or 2 cm in diameter. Surgeons have used a cystoscope for ventriculoscopy to confirm the existence of a colloid cyst and have seen nothing abnormal. The surgeon who has entered a ventricle through a cortical incision and seen nothing abnormal may think that he should have chosen the opposite hemisphere to wound. If he makes a window in the septum pellucidum he is unlikely to see as well the foramen of Monro on the opposite side because of the unfavorable angle of his line of sight with the axis of the foramen (Fig. 13.1) (2, 15, 22, 32). Despite this occasional problem of not seeing a cyst in the foramen, ventriculoscopic inspection has been recommended to give an opportunity for biopsy (40). Another source of confusion may arise when lateral ventriculomegaly is revealed by CT without the cause being seen— even on repeat CT and MRI. If positive contrast ventriculography is then done and a "lesion" is revealed at the foramen of Monro, the "lesion" may

Figure 13.1. Coronal section through the region of the foramen of Monro. The bony opening is generous to allow retraction of the hemispheres should the ventricles be small. The craniotomy extends across the midline (Fig. 13.2) to permit the dural flap based on the sagittal sinus to be pulled into a direct line with the falx, which may also be retracted, allowing an uncompromised vertical view as well as one angled into either ventricle (A and B). A dural flap based temporalward may or may not be required. Depending on how one makes the opening in the corpus callosum between the pericallosal arteries, either or both ventricles may be entered and the septum pellucidum may be fenestrated as required or the two leaves of the septum may be separated for a direct entry between the fornices to the midline tela choroidea of the third ventricle. The sight line of the transcortical approach is good into the lateral ventricle and foramen on one side (C), but unfavorable (inviting excessive retraction) through a window in the septum pellucidum into the opposite ventricle (D).

turn out to be a void in the contrast shadow if the films were made with the patient in the hanging head position (valuable for pneumoventriculography only) without good mixture of the contrast agent with all of the ventricular fluid. The experienced surgeon should be little handicapped by not seeing anything in the foramen of Monro because he will have anticipated this and will be ready to use an effective tactic that will reveal the lesion and permit its removal. A great virtue of the percallosal method is that it provides lines of sight (Fig. 13.1) directly down the midline and 10 to 15° to the right and to the left, so that either the right or the left lateral ventricle may be entered (15, 22) or the leaves of the septum pellucidum may be separated for a midline entry into the anterior third ventricle (2, 5, 7, 13, 14, 22). The ability to view a lesion from both sides and from above may be of particular value in cases where difficulty is experienced in separating the cyst capsule from the Choroid plexus, the internal cerebral veins, or the tela choroidea of the third ventricle and securing its sometimes vascular pedicle. Occasionally large lesions may be imperfectly understood, as when the third ventricle is obliterated by a possibly cystic mass elevating the floor of one or both lateral ventricles adjacent to the region of the foramen of Monro. Angiography and CT may leave unanswered the question of whether the lesion is a craniopharyngioma with an infradiencephalic component that should be examined first. In such situations, one must consider transforaminal entry with a possible subfrontal approach and the craniotomy must be fashioned to permit a subfrontal extradural avenue for examination of the region of the chiasm and the pituitary stalk and, subsequently, an avenue down the midline if the lesion seems wholly or principally intraventricular (Fig. 13.2) (15). Another consideration is the prevention or avoidance of hemisphere collapse in cases complicated by massive ventriculomegaly (5, 39). Generous ventricular size makes transcortical surgery easier and safer than if the ventricles are small (5, 7, 13, 19, 25, 32, 33, 38, 39, 50-52, 54), but an inability to seal the transcortical incision reliably in a watertight fashion may argue for interhemispheric access because the small opening in the corpus callosum is self-sealing once the retractors are removed and the ventricles are filled with fluid. The options available for dealing with masses in the foramen of Monro and for entry into the third ventricle are several: one can perform freehand penetration of the foramen of Monro with a stiff cannula oriented with reference to external landmarks (21) or use stereotaxic guidance of a cannula or needle using a ventriculoscope (40) or devices in conjunction with CT or more conventional guidance systems using radiographic films (4, 27); one may work to and through the foramen of Monro with magnification after exposure of the foramen through a cortical incision (5, 13, 19, 25, 32, 38, 39, 50-52, 54) or through the corpus callosum (1, 2, 15, 22,27,47,49), utilizing tactics for reducing the bulk of the intraforaminal lesion to facilitate total removal (11, 32, 38, 50); one may enter percallos-ally, using the interforniceal approach down the midline between the internal cerebral veins just back of the foramen of Monro and through the tela choroidea of the third ventricle (2, 5, 7, 13) or the approach through the lateral ventricle, lifting the Choroid plexus from the thalamus (12, 14, 25, 31, 51) or enlarging the foramen of Monro by removing some of the substance of the anterior nucleus of the thalamus, preserving the closely applied internal cerebral vein (11, 15, 19). Similarly the surgeon may cut one or both forniceal columns (13, 32, 39) or divide the thala-

Figure 13.2. A craniotomy appropriate for transcallosal exposure on the right side in a patient with small ventricles or who may require exploration of the chiasm. The low frontal burr hole is directly on the midline and above the frontal sinus. The two holes on either side of the sagittal sinus are staggered and may be safely connected by rongeur. A Gigli saw provides thinner and more bevelled cuts than a high speed router. If the chiasm is to be explored the dura should not be opened at the first opportunity, but only along the sphenoid ridge to the region of the clinoid and then along the midline in the direction of the low frontal burr hole.

mostriate vein to provide improved access under the Choroid plexus and to the roof of the third ventricle (12, 14, 25, 31, 51). Certain of these described procedures, although perhaps technically feasible or even easy, carry unnecessary risks to the complete recovery of the patient. A colloid cyst is invariably benign and should be essentially curable. Certain formerly used approaches are of historical interest only or are little used and largely obsolete. Entrance into the third ventricle via the lamina terminalis (25, 29) gives very limited access because of the tight proximity of the proximal pericallosal arteries, the deep position and shortness of structure of the lamina terminalis between the chiasm below and the anterior commissured superiorly, and the extreme sensitivity to pressure and manipulation of the walls of the hypothalamus, which may cause severe hypothalamic deficits. The Dandy methods currently used include transcortical openings to the right and left ventricles (used once each when his book was written (13)) and the anterior percallosal method, which he did not use for colloid cysts, but did use for cysts of the septum pellucidum and the cavum vergae. The posterior interhemispheric operation with extensive section of the corpus callosum, which he called the "pineal approach," was used by him 3 times, once on the left and twice to the right of the sagittal sinus for colloid cysts. In his Case 2, illustrated by Figure 10 of his book, he kept to the midline and entered the third ventricle between the internal cerebral veins, similar to the method described by Busch in 1944 (7) except that Busch first entered the nondominant lateral ventricle by cortical incision and subsequently gained the midline by splitting the septum pellucidum. The pineal approach of Dandy is no longer used because of the resulting hemisphere disconnection and the almost equal access provided by harmless smaller anterior callosal openings and the use of the microscope (2, 15, 17, 22, 27). For freehand puncture and aspiration of a cyst in the foramen of Monro, one used a flexible shunt tube with an intraluminal obturator introduced 4 cm from the midline in the coronal region, penetrating the frontal lobe on a line (from the frontal view) directed toward the opposite inner canthus and on lateral view toward the external auditory canal (21). A duration of success as long as 4 years has been claimed. The technique is said to be simple, reliable, and safe. We and many others have used this method of entering the foramen of Monro to introduce a catheter into the third ventricle for the instillation of air or another contrast agent to make visible the anterior and posterior commissures of the third ventricle and the ventricular walls preliminary to stereotaxic operations on the thalamus, but have never tried to penetrate a cyst for therapeutic purposes. For ventriculography, one uses a flexible catheter with little or no potential for damaging any of the structures surrounding or inferior to the foramen of Monro in case the aim should be less than perfect, but for the tough walls of colloid cysts an instrument with penetration capability would be necessary. The possibility of wounding the internal cerebral vein, Choroid plexus, fornix, thalamus, and hypothalamus is neither trivial nor remote. Aspiration of the cyst content followed by ventriculoa-trial or peritoneal shunting with the same tube is an instance of using two less than completely satisfactory treatments in hope that, if one does not work, the other will. As already mentioned, spillage of cyst content into the ventricular system is worrisome. Neither would cyst content be suitable for conveyance to the heart and the general circulation. Also, cyst content might easily obstruct the ventricular end of the tube used for shunting. The cyst probably has a higher than realized refill potential,

said by some to be in the neighborhood of 10% (11), which must relate to the duration of follow-up. As long as the cyst wall remains and as long as shunts are fallible, the possibility of sudden death has not been eliminated. Stereotaxic aspiration of a cyst may reduce the risk of damage to an internal cerebral vein and other surrounding structures (4). Although the procedure may be elegant in terms of the technology used and accuracy, one wonders about the true value of this procedure. Little is gained by examining the cyst content because a generally spherical unilocular structure in the interventricular foramen and anterior third ventricle can, for practical purposes, only be a colloid cyst; aneurysms, tumors arising from the septum pellucidum, fornix, and thalamus, and other masses should already have been ruled out by CT, angiography, or MRI. There is danger of spillage of irritating content into the ventricle, imper-manence of the therapeutic result, risk of future sudden death from acute hydrocephalus, or the possible need for another procedure that might better have been extirpative and curative in the first place. Inspection of the foramen of Monro through the nondominant lateral ventricle entered through the corpus callosum or a cortical incision is highly satisfactory and permits colloid cyst extirpation and permanent cure. This should be considered the standard method of dealing with colloid cysts and similar lesions. With smaller lesions up to a centimeter in diameter, the foramen of Monro may be slitlike, with no evidence of a mass within it as first seen. With larger lesions, up to 3 cm and more, the mass will have spread apart the walls of the third ventricle just behind the foramen of Monro and the foramen will be dilated a bit, allowing the pathological surface to be seen. There are reports of cysts as large as 9 cm in greatest dimension (37); cysts presenting above the third ventricle between the fornices and the leaves of the septum pellucidum (9-11); cysts ruptured into the internal capsule and the caudate nucleus (30) or into both lateral ventricles (54); and cysts with other special features (41, 53). The approach used must suit the special circumstances. If the mass has extended upward between the fornices and separated the leaves of the septum pellucidum, perhaps more to the side of the approach so the foramen is overlain by the bulging septum and so unseen, a reliable method of finding the foramen must be used. The best method is to follow forward the Choroid plexus of the lateral ventricle (50). Right where it disappears from view (perhaps under a bulging overhang of the septum pellucidum that will have to be lifted up) will be found the foramen. Entering the foramen and closely applied to the thalamus is the thalamostriate vein (Fig. 13.3). Bounding the foramen superiorly and anteriorly is the fornix. Once the foramen is located, its patency may be tested by catheterization with a Silastic (Dow Corning, Midland, Michigan) shunt tube, using either the manufactured end with multiple apertures or a surplus length cut very obliquely to make a filamentous tip. This may be gently teased into the foramen and past the obstructing mass with no injury to either the Choroid plexus or the vein. One intends not to penetrate the wall of the cyst at this stage, but merely to feel the resistance of catheter passage between the cyst wall and the thalamus. If the cyst is not visible, the closed tip of a forceps should be introduced into the foramen and gently opened, acting as a temporary retractor on the walls of the foramen. This will permit one to see most lesions in the anterior third ventricle. If the catheter passes with no resistance into the slitlike foramen and direct visualization with the spreading forceps shows no evidence of a mass, one should use a #8 French urethral catheter as a soft probe through the foramen; the mass may lie posterior or inferior to the foramen of Monro. In either of these special circumstances the foramen is not the proper route for extirpation of the lesion. Consideration must be given to entry through the roof of the third ventricle directly down the midline behind the foramen of Monro by the methods of Dandy (13) (Case 2; Fig. 10) Busch (7), or Apuzzo et al. (2) or by separating the Choroid plexus just behind the foramen from its attachment to the thalamus and going through the tela choroidea of the third ventricle laterally

Figure 13.3. View into the anterior right lateral ventricle through a high frontal transcortical incision or one in the anterior corpus callosum angled to the right with some retraction of the right cingulate gyrus, the right pericallosal artery, and the right cut edge of the corpus callosum just behind the genu. The foramen of Monro may not be seen beneath the overhang of a septum pellucidum bulging from a colloid cyst that has dissected between and separated the columns of the fornix to lie between the leaves of the septum pellucidum. To find it, follow the Choroid plexus forward to where it meets the thalamostriate vein to enter the foramen. Colloid material may also dissect laterally into the thalamus and caudate nucleus.

Figure 13.4. Diagramatic representation of a coronal section of the brain just posterior to the foramen of Monro (not shown). The fornical columns are closely applied to the rostrum of the corpus callosum and beneath lies the transverse fissure. Below this is the roof of the third ventricle bearing the internal cerebral veins. A. The avenue of Apuzzo et al. directly down the midline. B. The avenue of Busch, which gains the midline beneath the corpus callosum from the lateral ventricle. Dandy's midline incision through the posterior corpus callosum is not shown. D. The avenue of Deladsheer et al. and others (see text) lateral to the Choroid plexus and the internal cerebral veins.

beneath the internal cerebral vein (14, 25, 31) (Fig. 13.4). Although colloid cysts take origin from the tela choroidea just behind the foramen of Monro and in the region of the paraphysis (10, 16, 23, 28, 35, 37, 44, 46, 53), there are circumstances that might cause a cyst to appear in an aberrant position in the ventricle. Neuroepithelial cysts of this sort can develop behind the paraphysis and even in the fourth ventricle (8, 10). Or a cyst only lightly attached to the tela choroidea may be displaced posteriorly by the pressure gradient developing in the lateral ventricles as a consequence of obstruction at the interventricular foramen. Another possibility exists in the patient who has undergone needle aspiration of a cysts or a less than complete extirpation, with the refilled cyst lying in an aberrant location. When the cyst or tumor is reasonably well seen an effort should be made, working through the foramen of Monro, to free its anterior, inferior, lateral, and posterior surfaces with a slender blunt probe such as a curved, ball-ended dissector. One should not stroke superiorly around the mass because at this stage interruption of a possibly vascular attachment or pedicle to the tela choroidea above would be inconvenient. Unless the mass is small and seems inclined to deliver itself, it is unwise to attempt removal of the lesion intact because the pedicle will be the last part to come through the foramen of Monro and, if inadvertently disrupted, bleeding may occur from the roof of the third ventricle — unseen and unsuspected until the third ventricle is filled with blood. A mass too large to squeeze out of the foramen is penetrated with a #20 or #21 spinal needle held near its tip in a hemostat or a slender needle holder. Aspiration may then be made with a 5-ml syringe connected to the needle by a short length of flexible tubing. As material is being obtained the capsule shrinks simultaneously with the appearance of fluid in the transparent tubing. If the material is too thick for aspiration through the needle, a larger needle or even a small caliber suction tip may be used. If the cyst content is semisolid or caseous, a cruciate incision into the capsule will allow, by the use of a small, delicate sucker tip and small ring curettes, fragmentation of the material and fairly complete evacuation. One should never make a pushing force on such a cyst for fear of displacing it posteriorly into the third ventricle and losing sight of it. The end of the sucker tip should always be in contact with the interior of the cyst so as to maintain a slight grip on it, or one may use a microaligator to hold the wall or fasten a length of suture material to the wall of the cyst with a small microclip. A knot tied in the end will prevent slippage through the clip. The objective at this stage is to evacuate the cystic content completely to permit delivery of the capsule by making gentle pulls on it at various sites around the opening already made through the foramen of Monro. The pedicle will be the last part to be seen as it passes through the foramen of Monro, which is then hemostatically divided. The foramen should then be gently irrigated with a small volume of saline delivered through the Silastic catheter. All of the saline return should be bloodless, with no need for hemostatic sponge to be placed in the neighborhood of the foramen of Monro. If bleeding occurs within the foramen of Monro, clips and cautery should be used cautiously because one desires to retain patency of the thalamostriate and internal cerebral veins (4, 11, 50) despite counterclaims of their dispensability (12, 14, 25, 31, 51). One may introduce a #8 French urethral catheter into the foramen and then slide down a very small piece of gelatin sponge between the catheter and the bleeding point, with the catheter serving as a counterforce against the wounded vein. The catheter should be led outside the wound for removal in 12 to 24 hours, by which time the Gelfoam will be firmly adherent to the vein, with little possibility of it obstructing the aqueduct or allowing rebleeding. If the interventricular foramen is too small to permit the tumor to be well seen or if the tumor lies too far posterior to the foramen for removal

through it, the tumor may be approached through the roof of the third ventricle. Most large masses not removable through the foramen will have widened the ventricle and will have spread the tela choroidea and, thereby, the distance between the two internal cerebral veins and fornices. A midline incision through the roof behind the foramen then can be made; this was well shown in Figure 10 (Case 2) of Dandy's monograph on tumors of the third ventricle (13) after what he called his pineal or posterior transcallosal approach. A similar approach was later described by Edward Busch, but he used the transcortical approach to a lateral ventricle and then opened the leaves of the septum pellucidum (7). Apuzzo et al. (2) recently described a modified and refined perfectly midline avenue into the third ventricle. One virtue of midline opening is that it can be extended posteriorly almost without limit, increasing and improving the exposure of just about anything in the third ventricle that arises from or is attached to superior structures. As mentioned previously, a midline opening is not helpful for infradiencephalic tumors invaginating into the third ventricle. The procedure of Dandy (13) minimized injury to the columns of the fornix, which are separated posteriorly, but at the expense of cutting the corpus callosum too far back by modern standards. Dandy's procedure also minimized the possibility of injury to internal cerebral veins. However, in the case he described one of the veins had to be ligated, apparently without complication, because the cyst capsule was tightly attached to it (13). The critical obligation in this approach is to remain so perfectly midline as to be looking down upon the roof of the third ventricle, with the two internal cerebral veins directly visualized on either side, making the incision carefully between them (Fig. 13.4). There is no cross connection between them, but if bleeding occurs it is safer to put a small pledget of Gelfoam under a thin or narrow cottonoid than to use cautery. One must be aware of anatomical variation, e.g., a single internal cerebral vein. The midline interforniceal approach of Busch (7) and Apuzzo et al. (2) requires separation of the closely applied fornices and is somewhat worrisome because of their vulnerability to bilateral injury. Such injury is, however, as yet undescribed. The fornix, similar in construction to the optic nerve, may have similar sensitivity to damage. One may be beguiled into thinking that no damage is done because the consequence of modest forniceal damage must be carefully sought to be found, despite its disability effect. The functions of the fornix and other parts of the limbic system in memory, motivation, emotion, and other hard to quantitate features of human nature suggests that it is prudent to avoid forniceal manipulation if at all possible (26, 36, 49). An excellent method of third ventricular entry is that of Deladsheer et al. (14), Hirsch et al. (25), Viale and Turtas (51), Cossu et al. (12), and Lavyne and Patterson (31), who separate the Choroid plexus from the thalamus and enter the third ventricle beneath it and the internal cerebral vein. The fine illustrations of Deladsheer et al. (14) and Cossu et al. (12) are worth studying before using this method. The method entails manipulation of the Choroid plexus and the thalamostriate vein, and there is some risk of interruption of one of them. This method is not necessarily superior to the midline methods, and all have the same indications. It may be more feasible technically when the two internal cerebral veins are not clearly separate structures or when the effect of the intraventricular neoplasm has been to stretch and broaden the veil of tissue attaching the vein to the thalamus, making it easier to gain entry to the ventricle under and lateral to the rightsided vein rather than medial to it.

Enlargement of the foramen of Monro by removal of tissue from its posteroinferior aspect, the anterior nucleus of the thalamus, has been mentioned as an alternative to forniceal cutting (15, 19, 22). Despite the fact that the anterior nucleus and its contained tract of Vicq d'Azyr is an important part of the limbic system, some of its substance can be removed unilaterally, but the utility of doing so is small unless the thalamostriate vein is also interrupted. The subchoroidal method of entering the ventricle behind the foramen is clearly superior and seems to offer less danger to recent memory. If the foramen needs enlarging only slightly, as when an evacuated cyst capsule remains just too large to come through the foramen easily, the thalamostriate vein may be separated from its attachment to the thalamus where it enters the interventricular foramen. With gentle suction and the use of a micro-Love-Gruenwald forceps or a semi-sharp ring curette, some anterior thalamic substance can be removed from beneath it. Although surgeons who have removed third ventricular lesions with great success in prior years sectioned one or both columns of the fornix (13, 32, 39), the consequence of recent memory impairment and learning disability should forbid the use of this tactic. Perhaps the human brain works just as well with only one fornix, but not knowing everything that a fornix contributes to human behavior and competence makes for uncertainty. At least as important, the lesion being operated upon may, because of features not known when one fornix is cut, have interfered with the function of the other, or the surgeon may damage the opposite fornix as he seeks to gain still better exposure. Reports of the division of both forniceal columns without resulting deficit are highly suspect. Lack of evidence of deficit most likely stems from insufficiency of knowledge of forniceal function and examination during the period when forniceal division was thought acceptable. Thus, many considerations attend third ventricular entry. There are possibilities of postoperative seizures, paresis, cortical collapse, impairment of frontal lobe function due to division of bridging veins, damage to the pericallosal and tributary arteries, and damage to deeper vascular structures such as the Choroid plexus, the thalamostriate vein, the internal cerebral vein, and the choroidal arteries. The perils attending exposure of these structures are immediate and so may seem of greater importance to the surgeon than a depreciated quality of long term outcome that goes on outside his observation once the patient is dismissed and at home. Injury to the limbic system from improper handling may lead to serious functional impairment for the rest of the patient's life. This system, well described elsewhere (26, 36), lies phylogenetically between the oldest parts of the brain having to do with control of visceral function, cardiac, vascular, and respiratory regulation, and crude types of locomotion and the neocortex, which lies atop it and conceals it from view. The limbic system is concerned with motivation, emotion, moods relating to eating and smell, sexual behavior, and behavioral responses to the external and internal environment that far outlast provocative stimuli and contribute immensly to distinctiveness of personality, memory, and mental competence. The parts of the limbic system at risk in this operation include the cingulum and cingulate bundle, the body and columns of the fornix, the termination of the mamilothalamic tract of Vicq d'Ayzr, and the anterior nucleus of the thalamus. Even deliberate lesioning or the excision of lengths of both cingulate gyri and the cingulum bundles 2.5 cm long does not seem to have an unfavorable effect on behavior. On the contrary, it may have an ameliorating effect on chronic pain and obsessive-compul-

sive states (15). The effect of forniceal lesions is, if bilateral, severe impairment of recent memory and of the making of new permanent memory engrams. This is said not to occur if only one fornix column is interrupted, but one can never be certain that the other column is not impaired (15). Undoubtedly, much is to be learned about limbic structures, and it may be surmised that in the future tests will be developed that show handicaps not at present suspected. The appropriate attitude of a surgeon toward such structures should be to inflict as little manipulative damage as possible. Something as gratuitous and avoidable as sectioning the fornix or deep lesioning of the anterior nucleus of the thalamus may change the patient's postoperative personality, drive, attitudes, emotions, and capabilities in ways too subtle to be measured at present, but these changes may significantly reduce his potential as an effective or perhaps superior human being. Limbic System with Special Attention to Structures Involved in the Third Ventricular Approach The purpose of this section is to identify in humans the potential for disability after surgical lesions are made in structures related to the third ventricular approach for the removal of lesions from the diencephalon. We touch upon the anatomy of memory, which has been covered in greater detail in Chapter 6. We also cover in some detail lesions of the fornix, septum, mamillary bodies and mamillothalamic tract, nuclei of the thalamus, and corpus callosum. One facet that concerns us preeminently is disability to the function of memory resulting from diencephalic surgery. We will discuss briefly the anatomy of memory specifically and from there move onto structures in the neighborhood of the third ventricle in greater detail. The anatomy of memory is of some significance in neurosurgery, particularly in surgery of the medial skull base, the diencephalon, and the temporal lobes, but getting to the heart of this subject is difficult. Clini-copathological correlations have never been clear enough to allow perfect determination of the neuroanatomical substrates in the anatomy of memory. Recently utilized techniques such as horseradish peroxidase and neurotransmitter tracing have produced far more information, but little more evidence of the mode of operation, and have failed to answer important clinical questions regarding humans. The search for the functional organ of memory has only proven to indicate that a single entity cannot be found (14, 22, 27, 28, 31). All memories, whether visual, verbal, or tactile, are probably stored in the neocortex in the form of coded engrams holding a restricted number of items of information. Presumably the longer a piece of information is held in memory, the more it is repeated. With repetition its engram becomes more restricted and coded and more compressed. By way of illustrating the importance of transmission of these presumed engrams, the fornix, as small and anatomically insignificant as it seems in the human brain, contains 5 times the number of neurons of the optic tract (19). Weiskrantz (37) has offered a conceptual scheme of the function of memory, not as yet translatable into a neurophysiological process. An event occurs initiating a "short term trace" that lasts about 20 seconds and rapidly decays. The subsequent conversion of a short term trace to a "long term trace" has been termed consolidation. Recall of short term traces is a separate process from consolidation, supported by the observation that the duration of retrograde amnesia (as in head injury) may decrease. Clearly, if a patient can, at a later date, recall an event that he

could not previously remember, the long term trace remained but could not be recalled. Similarly, such patients in the absence of an attention disorder reflect accurate short term memory even during the period of posttraumatic confusion (2). In all likelihood the short term trace and immediate memory processes occur in the cortex. The process requires initial registration, usually some visual or verbal repetition, and short term hold. It seems possibly served by perisylvian language cortex and visual association cortex (32). It is unlikely that much, if any, commissural transmission — i.e., interhemispheric communication for storage, comparison, and retrieval—takes place. The function of immediate recall is not threatened by surgical approaches to the third ventricle (11). As an immediate memory makes the metamorphosis to remote memory, it is presumed to go from relatively uncoded cortical information to a compact engram that can be read out at will. In the intervening time, the subcortical structures of the limbic system and major parts of the thalamus have repeatedly encoded it, triaged it, stored it, retrieved it, deciphered it, and transmitted it. By the time that it is a truly remote memory, the limbic system may be little required. Until it is remote, however, the memory is "recent" and as such is dependent on the subcortical white tracts and nuclei of the limbic system and thalamus. The mechanism of limbic system ciphering seems to be encoding into compact engrams, which with repetition become simpler and more easily recovered (2). Clinical states in which there may be recent memory disturbances include anoxia, bitemporal lobectomy (26), herpes encephalitis, bilateral hippocampal infarction (24), and Korsakoff's syndrome. Alzheimer's disease displays its first pathological changes in the hippocampi; therefore, learning disability is presumably one of its first manifestations. Tumors of the midline involving the fornices have resulted in amnestic states (16, 17), as have tumors of the parasellar and suprasellar region, presumably because of compression of the mamillary bodies, anterior fornices, thalamus, and diencephalic white tracts (9, 20). Finally, trauma can affect the limbic system and thereby memory. The hippocampi and amygdala are both known to have extremely low electrical and epileptogenic thresholds; presumably by their sensitivity and by virtue of their location near the bone and tentorium of the middle fossa, these structures are sensitive to head injury. We will later see that neither the amygdala nor the entorhinal cortex is thought to be important in the registration of recent memory. However, the hippocampus does seem to be, and trauma to it causes the disturbed memory, storage, and retrieval so prominent in traumatic amnesia. One may even go so far as to draw a parallel between the dream states, fugue states, hallucinations, confusion, fear, etc., experienced by temporal lobe epilepsy patients and the same conditions seen in some trauma patients. If, like Horel, one disbelieves the hippocampal theory, one may credit the dysfunction produced in the temporal cortex and stem by trauma. We see, then, that the function of recent memory involves a wide variety of subcortical and particularly limbic structures and that dysfunction of recent memory can be elicited by a variety of processes. The basic limbic circuit was first described by Papez in 1937, and at the time it was thought to be primarily concerned with olfaction and emotions. This view prevailed for the next 20 years without much additional neuroanatomical or functional information. Perhaps the exception to this was Brodal, who in 1947 thought sufficient evidence had accrued to indicate that the circuit had nothing to do with olfaction. In the classical Papez circuit, the hippocampi (Ammon's horn, subiculum, fimbriae, and

dentate gyrus) are connected by way of the fornix to the mamillary bodies. The mamillary bodies are then connected by the tract of Vicq d'Azyr to the anterior hypothalamus and from there to the cingulate gyrus. Since the time of Papez there has been considerable information generated by clinicopathological and animal studies adding to our knowledge and understanding of the limbic system. In Korsakoff,s disease, Victor et al. have attached great importance to pathological conditions in the magnocellular dorsomedial thalamus (36). Similarly, Horel emphasized the dor-somedial thalamus and disbelieved the idea that the hippocampi are central to the function of memory (18). Spiegel and Wycis place orientation in time ("chronotaraxis"), often the first problem experienced in Korsakoff's disease, in the anterior and dorsomedial nuclei of the thalamus (29, 30). The weight of modern information makes the thalamus seem to be more important in effect than the hippocampi so far as discrete lesions are concerned in the production of amnestic states. Diencephalic white fiber tracts such as the anterior commissure (18), inferior thalamic peduncle (18), and thalamocingulate tracts (22) have been added to the list of structures thought crucial to recent memory. The temporal stem to the dorsomedial thalamus, basal ganglia, and orbitofrontal cortex has also been implicated (18). The anterior corpus callosum and anterior commissure (18, 40), neither regarded as "limbic," have been recorded to cause clinical amnesia, perhaps by the loss of interhemispheric transmission. Certain of these structures important in the registration of recent memory are easily damaged in third ventricular approaches and will be more specifically addressed later in the discussion. Remote memory is a function generally spared in all but severe degenerative diseases. Patients with Korsakoff's syndrome or Kluver-Bucy syndrome after bitemporal lobectomy will have surprisingly intact remote memory despite profound new learning disability. The reason, of course, seems to be that the limbic system is not actively used in the retrieval and decoding of remote memory. As currently understood, it takes widespread cortical atrophy to impair remote memory. One never finds intact recent memory in the presence of impaired remote memory. More than 30 years ago in 1955, MacKay said that it was unwise to try to localize memory as a function of any single portion of the nervous system (34). A single lesion will probably not produce a deficit; rather there must be two or more (22, 23, 27-31). Clinical and experimental evidence shows that mnestic derangements are evident with lesions of the temporal cortex, parahippocampal cortex, cingulate gyrus, mamillary bodies, thalamic nuclei, midbrain reticular formation, hippocampal formations, fornices, hippocampal commissures, corpus callosum, thalamocingulate tracts, anterior commissure, and inferior thalamic peduncles. There is no agreement on a "memory site," probably because there is no single site. We see, however, that recent memory disturbances dominate the deficits likely to be produced by approaches to the third ventricle primarily because of the subcortical and limbic nature of the operation. The main problems in dissecting the memory circuit are two: (a) The pathological conditions in human clinicopathological studies are too widespread (tumors, Korsakoff's disease, encephalitis, infarcts, trauma), (b) Animal studies are difficult to design well, execute, and interpret in a meaningful way. A third and probably less well understood problem in clinicopathological and animal studies is that a chronic process such as alcoholism will allow ongoing functional reorganization to proceed at a pace parallel with the destruction of brain, thereby lending the impres-

sion that an individual is actually unimpaired until virtually all tolerance is used up. Regarding memory and the third ventricular approach, certain features bear repeating. The loss of many of the nuclei and fiber tracts in the limbic system will affect memory. The hippocampus, once at the fore of memory theory, has become less eminent as attention has been directed toward the diencephalon. Much of the older literature in which normal mental status reportedly followed limbic injury must be viewed more skeptically. In times past, forniceal injury, for example, was expected to provoke olfactory or emotional disability whereas subtle memory and learning disabilities could have been easily passed over. It is from this older literature that we have gotten certain of our neurosurgical maxims, such as that unilateral limbic injury is well tolerated. Subtle memory deficits may nowadays be expected to follow "minor" unilateral limbic injuries (6, 27, 28, 31). Medial Temporal Structures In an effort to understand the importance of the diencephalic structures involved in third ventricular surgery, one must have an understanding of their afferents. We discuss these briefly before addressing those structures in the vicinity of the third ventricle. Hippocampus The hippocampal formation (Ammon's horn, subiculum, fimbriae, dentate gyrus) receives its afferents from the medial septal nuclei via the dorsal fornix. The medial septal nuclei are thought to drive the theta waves of the hippocampus. The septohippocampal system receives, either indirectly or directly, input from most of the neocortex and amygdala and much of the hypothalamus and from adrenergic, dopaminergic, and serotonergic cell groups in the brain stem (33). The efferents from the hippocampal formation then pass to the contralateral hippocampus via the hippocampal commissure and over the fornix through the mamillary bodies into the septum and into the hypothalamus and habenular nuclei and interpeduncular nuclei. Hippocampal projections can be found as well in substantia nigra, central gray, pontine gray, and corpus striatum. Of particular importance in this discussion, small efferents also pass directly to the indusium griseum, cingulate gyrus, and medial septum via the dorsal fornix, and a small number of efferents pass directly to the anterior thalamus, bypassing the mamillary bodies. Furthermore, a number of fibers pass off the fornix into the dorsomedial nucleus of the thalamus (35). These latter structures may well be involved in tumors and operations of the diencephalon. Acting through the cingulate cortex, habenular nuclei, and hypothalamus, the septohippocampal system in fact may influence activity in most of the systems of the brain (33). Its major efferent system, however, is that to the mamillary body. Because of the involvement of the mamillary bodies in Korsakoff's syndrome, this massive connection of the hippocampus was thought to implicate the fornix in memory. That contention remains controversial to this day. Glees and Griffith in 1952 and Scoville and Milner (26) were the first to implicate lesions of the hippocampus as operant in memory deficits. Scoville and Milner noted in particular a grave loss of recent memory in those cases in which the medial temporal lobe resection involved the hippocampal complex bilaterally. In their noted case, H.M., a full-scale I.Q. minus memory quotient score of 45 was obtained. The hippocampus was specifically suspect because of the lack of memory disturbances in other cases of deliberate removal of the uncus and the lack of memory

deficit after temporal lobectomy without hippocampal removal. Scoville and Milner were particularly impressed by the importance of the anterior hippocampus in the retention of new memory. The idea of a "memory circuit" involving the hippocampi, mamillary bodies, and fornix took root with further clinical evidence, for example, ischemia of the hippocampi responsible for amnesia shown by angiography revealing vertebrobasilar and posterocerebral arterial disease (8, 24). The learning disability present in monkeys with Kluver-Bucy syndrome had already been well established. There have been innumerable animal studies purporting to solidify the central role of the hippocampus in memory. The septohippocampal formation may well be at least as important in spatial orientation as in memory (17, 18). In fact, it must be admitted that the studies in humans implicate a wide area of the medial temporal lobe and the studies in animals are inconclusive as to the hippocampi. Horel (18), for example, implicates the temporal stem and medial temporal neocortex. Reviewing his own and other experimental results in hippocampal lesions, he thinks that the spatial disorientation produced is productive of all of the deficits of purported memory tasks. In a similar vein, Squire and Moore doubt that the hippocampus is a source of mnestic deficits and believe that the dorsomedial nucleus of the thalamus is foremost (31). Despite the known severe memory deficits following bilateral medial temporal lobe injury, then, the evidence implicating the hippocampi specifically is suspect. Reviewing experimental evidence in animals, Horel suggested that learning and memory deficits have not been documented in animals with either hippocampal or mamillary body lesions but that spatial disorientation is common to mamillary, forniceal, and hippocampal lesions (18). The Amygdala Despite its inclusion as a limbic structure and its close ties with many of the same structures as the hippocampus, the amygdala is not thought to play a part in memory. The amygdala has close ties with the hypothalamus and basal ganglion. It claims among its afferents the periamygdalar cortex and each of the sensory association areas such as visual, auditory, olfactory, and tactile association areas. Its efferents pass via the striae terminalis to the hypothalamus and diencephalon and in the latter is included the dorsomedial nucleus of the thalamus. Subsequently, they pass indirectly to the telencephalon. Some efferents pass via the ventral amygdalofugal path to the hypothalamus, and other efferents pass through the uncinate fasciculus to the ventral insular and caudal orbito-frontal cortex (14, 19). The effects produced by amygdala lesions are unlike those produced by other medial temporal lobe injury. The amygdala, a complex organ of multiple nuclei, is active in a large variety of disparate functions such as vegetative, emotional, and orienting. Autonomic activity is changed by lesions of the amygdala, and by electrical stimulation one can produce bradycardia, tachycardia, changes in respiratory rate, and changes in the galvanic skin response. Lesions of the amygdala interfere with the organism's alerting and orienting activity. There is an exaggerated dis-tractability and inability to respond to new experiences. Furthermore, aggression is either facilitated or inhibited by lesions of the amygdala depending on whether the lesion is in the dorsomedial or lateral regions of the amygdala. In animal studies, normal interaction with other animals of the same species is far reduced. Only amygdala

lesions produce a general sluggishness and reduced ability to respond to "subtle social signals" (19). The amygdala comes up in discussions of memory primarily because the reduction in orienting activity, arousal, and attention span impede the learning of new tasks in animals. Furthermore, the expression of formerly learned responses is altered and may surface in memory tests. However, there is little evidence that the amygdala is actually involved in the retrieval of stored memory engrams. One can easily see, then, that poor performance in general is the result of amygdala lesions, and lesions of the amygdala are so general in their effect that it is hard to specifically define the deficits. Nevertheless, it seems clear from the available information that this medial temporal lobe structure is probably not responsible for much of the deficit in lesions of the hippocampal formation-forniceal-mamillary Papez circuit. The Temporal Stem Deep in the medial temporal lobe, inferior to the insular cortex and dorsolateral to the temporal horn of the lateral ventricle, lies the temporal stem, an important outflow path of the temporal neocortex and amygdala. The stem, then, carries white fibers to subcortical structures at the brain base. In particular, one projection goes to the medial magnicellular part (emphasized by Victor) of the dorsal medial nucleus of the thalamus. Other projections go to the pulvinar and midbrain, basal ganglia, and orbitofrontal cortex (18). Temporal neocortex has been shown to be important in memory function by human and animal electrical stimulation and by study of deliberate and natural lesions. If a large amount of temporal lobe neocortex is damaged, a profound learning disability results. Projections Here we discuss the hippocampal-forniceal-mamillary link, as well as other temporodiencephalic connections of importance in lesions of and surgical approaches to the third ventricle. The hippocampal formation projects via the fimbria and fornix to the posterior column of the fornix and the hippocampal commissure; forniceal fibers also arise from the periamygdaloid cortex. The fornix carries fibers in both directions, fibers being received from and projected to cingulate cortex, the septal region, the thalamus, etc., but the main direction is from caudal to rostral. Fornix The fornix is the main efferent bundle from the hippocampal formation (gyrus dentatus, hippocampus, parahippocampal gyrus, subiculum, etc.), as discussed previously. The hippocampal formation projects across to the contralateral formation via the hippocampal commissure, and forniceal fibers then course rostrally above the anterior third ventricle. Along the dorsal part and body of the fornix, fibers go to the cingulate cortex and some fibers pass off into the dorsomedial nucleus of the thalamus (35). Some forniceal fibers also originate in the periamygdaloid cortex, cingulate gyrus, the indusium griseum, septal area, and hypothalamus. Above the anterior commissure, the fornix gives rise to two large bundles. The precommissural fornix carrying fibers mainly from the hippocampus proper (19) goes primarily to the lateral septal areas anterior to the anterior commissure. The second and larger bundle, the postcommissural fornix, carries fibers mainly from the hippocampal formation (subiculum and its divisions) and is bound for the hypothalamus, thalamus, mamil-

lary bodies, rostral midbrain, and subthalamic region (19, 33). The fornix projects, thereby, not only to mamillary bodies, but also to the septal area, preoptic hypothalamus, dorsal and periventricular hypothalamus, anterior nucleus of the thalamus, median and paramedian thalamus, supramamillary region, and central gray of the midbrain. The consequence of section of the fornix is the basis of considerable controversy. At present, one simply cannot say that clinical cases are interpretable as indicating that a forniceal lesion guarantees production of a memory disturbance; other structures besides the fornix are variably damaged by operations and natural lesions. Garcia-Bengochea et al., working with forniceal lesions in monkeys and subsequently therapeutic forniceal section in seizure patients, mentioned no adverse effects of forniceal section (13). They treated 14 patients, presumably severing the body of both fornices in each, producing no disturbance of memory. Such deliberate therapeutic section of the fornix in humans was antedated by Dott (10) and Cairns and Mosberg (4), who sectioned the fornix to remove brain tumor. Cairns and Mosberg (4) presented 11 cases with 9 survivors of third ventricular tumor removal. All patients underwent at least partial forni-cotomy. The result of follow-up in these patients included the impression that forniceal section produced no deficit. In fact, they reported a relief of memory deficit in 1 case. This, however, was done at a time when the fornix was thought to have a primarily emotional and olfactory function and, although these were specifically addressed, memory function was not. Admittedly, there were a few cases in which transient memory disturbance was addressed, but ascribed to the hydrocephalus preoperatively. In a similar vein, Dott (10) presented the use of the transcortical transforaminal approach to third ventricular tumors in which he sectioned the bodies of the fornix and part of the septum in two cases. This was done without deficit, although he may have been looking more for olfaction and emotional disturbance than for amnestic derangements. More recently, in a classical and celebrated case, Woolsey and Nelson described an alcoholic patient with involution of the mamillary bodies who subsequently developed a diencephalic metastatic pulmonary adenocarcinoma (39). This tumor infiltrated the body of the fornices and posterior columns bilaterally without the production of Korsakoff's syndrome. Nevertheless, there are objections to this case on several points. The patient, being alcoholic, may have long before developed other means of establishing recent memory retrieval, as has been implied by Milner (34). Furthermore, the tumor, although largely infiltrating the fornices, did leave a considerable number of fibers intact. Finally, formal neuropsychological evaluation was not undertaken, and testing in such a situation must be careful because the defect may be subtle and largely compensated for. Squire and Moore noted that, at the time of publication of their article, there were 47 cases in the literature in which forniceal section did no harm versus 3 in which memory disturbance could be legitimately ascribed to forniceal injury (31). One must not forget the potential for confusion in the literature, however. Lesions of the far anterior fornix seem to have less potential for memory or other deficits than lesions posterior or near the hippocampal commissure (16, 21). One cannot compare the results of forniceal injury at one level to forniceal injury at another and draw meaningful conclusions (31). Despite the controversy and the presumed lack of detailed neuropsychological testing in many reported cases, one cannot deny that there are instances of forniceal lesions in patients with intact memory. However, we do not have the ability to predict which patient will fare poorly and which will not.

A certain volume of literature supports the contention that section of the fornices produces serious memory disturbance, similar to that reported in hippocampal ablation. Sweet et al. presented a patient in whom transfrontal section of the anterior fornices was produced to ease removal of a large colloid cyst (34). This patient had a disabling Korsakoff's syndrome postoperatively, which was ascribed to the forniceal division. This patient was described as having marked impatience, a significant and lasting memory deficit, a patchy retrograde amnesia, mild cognitive impairment, and general apathy. Milner, in comments following the presentation, stated that neuropsychometrics described a memory deficit not quite as bad as that following bilateral medial temporal lobe section. Section of the anterior fornices in Sweet's patient was productive of a difference in memory quotient to I.Q. of 13, as compared to bilateral hippocampal lesions productive of differences as great as 30 and more (26, 34). One must be cautious in ascribing this patient's recent memory disturbance to forniceal section, however, in that there may well have been damage to the walls of the third ventricle productive of the amnesia. Heilman and Sypert, in a noted case, described a patient with a poorly differentiated glial neoplasm in the quadrigeminal cistern (16). This was surgically removed, but aside from this there was no anatomical confirmation. The patient, both preoperatively and postoperatively, had a severe deficit in memory, with a difference between verbal I.Q. and memory quotient of 47, a severity comparable to that with bilateral medial temporal lobe injury. The authors noted that this severe memory deficit is in contrast to better tolerated more anterior lesions of the fornix and postulated that the difference is due to the fibers coming off the fornix into the dorsomedial nucleus of the thalamus given such importance by Victor et al. This difference has also been noted by Lavyne and Patterson (21). In support of the idea that forniceal fibers course into the dorsomedial thalamus from the posterior fornix, Valenstein and Nauta described their occurrence in monkeys (35). Hassler and Riechert (15) described a female seizure patient initially given an asymptomatic right anterior column forniceal lesion at the level of the anterior commissure. Later a left column lesion was produced and the patient was found to have a severe memory deficit without other behavioral change. Death occurred 8 days later, apparently due to an unrelated cause, and bilateral forniceal injury was confirmed as being productive of the amnestic derangement. Nevertheless, there are points of controversy. The patient had a right medial frontal lobe glioma as well and left hypothalamic softening, possibly including diencephalic fiber tracts and the anterior commissure. The other right-sided lesion was near the inferior thalamic peduncles and perhaps the temporal stem to the dorsomedial nucleus, and the case is thereby open to criticism. Apuzzo described a transcallosal interforniceal approach in 11 patients (1). Of these 11 known to have had forniceal manipulation, 4 had transient memory disturbances on careful postoperative assessment. Although the fornices cannot be implicated exclusively, this can be considered valid evidence of the importance of the forniceal system in memory. In a careful study avoiding the potential for task relearning, Gaffan reported recognition memory impaired in monkeys operated on through a transcallosal approach with bilateral anterior column forniceal tran-section (12). Similarly, Carr (5) described recognition impaired after a transcallosal approach and section of the fornices of baboons. It has already been observed that two lesions are important in memory disturbances, whereas one may not be (7, 23, 27, 28, 31). The controversy surrounding forniceal section, with each school having its evidence, could be considerably cleared were it generally acknowledged that these pa-

tients are not necessarily comparable. For example, forniceal section through frontal lobe cortex may not be as devastating as forniceal section near the origin in the hippocampal formation with possibly other limbic injury. A better question may be, does section of the fornix provide better exposure (11)? Forniceal disturbance should be avoided if at all possible because we cannot predict whether it will cause memory disturbance. Septum The septum reaches it highest degree of development in humans. It consists, classically, of two large groups of nuclei, medial and lateral. The important nuclei in primates and humans are found in front of the anterior commissure at the base of the brain. The septum pellucidum in humans is not entirely comparable to the septal complex of lower animals (19). The lateral group of septal nuclei is essentially composed of fibers of the fornix and the hippocampal formations and commissure. The medial septal nuclei consist of the diagonal band of Broca, the medial septal nucleus, and the septofimbrial nucleus. The medial septal nuclei receive their input primarily from the lateral preoptic and lateral hypothalamic nuclei. The medial septal nuclei send efferents back to the hypothalamus and also to the hippocampus and habenula. The lateral septal nuclei receive the majority of their input from the hippocampus and subiculum in a highly topographical fashion. Some fibers also reach the lateral septal nuclei from the amygdala, striae terminalis, piriform cortex, and other less important areas. Leaving the lateral group, the septal group of nuclei project to the hypothalamus, ventral tegmental area, and habenula. Projections from the cingulate cortex have been noted. A projection from the septal region to the dorsomedial nucleus of the thalamus has also been suggested. Fibers arising in the dorsomedial nucleus of the thalamus terminate in the diagonal band of Broca in the septum. Lesions made in the septal area affecting the important medial and lateral nuclei evoke increased rage reaction and hyperemotionality. In experimental animals these changes dissipate over time (19). Animals with septal lesions may also exhibit reduced aggressiveness. Lesions may also produce an impaired ability to acquire passive avoidance task learning in animals, somewhat similar to hippocampal injury. The septal area, then, plays a role in the ability to interact with the environment, and lesions in the septal area reduce the individual's ability to pay attention to subtle stimulation. A large incision through the septum in association with transection of a fornix of the fornices is to be avoided because these lesions would occur at the septal base and threaten the important nuclei primarily. However, high in the septum pellucidum proper under the corpus callosum, fenestration should not produce deficit. Mamillary Bodies and the Mamilothalamic Tract The mamillary nuclei in the posterior hypothalamus, as is well known, receive the majority of their afferents from the subicular region of the hippocampal formation via the postcommissural fornix. The efferents leave by two large tracts, the mamillothalamic tract to the anterior nucleus of the thalamus and the mamillotegmental tract to the tegmentum of the brain stem. The mamillary bodies, nested as they are in the hypothalamus, are admixed with fibers of the medial forebrain bundle and other important brain stem pathways. Lesions of the bodies, then, may implicate not only the mamillary bodies, but other important rostral systems. Mamillary body lesions and lesions of the mamillothalamic tract

result in impairment of the ability to perform spatial discrimination (18, 19). Animals with mamillothalamic tract lesions are loath to begin new responses to external stimulation. The impairment should not be ascribed to global loss of memory. The mamillary bodies are involuted or hemorrhagic in Wernicke-Kor-sakoff syndrome. Korsakoff's patients are characterized not only by the mnestic loss, but also by profound disturbances in attention, perception, and motivation (23). There is widespread abnormality with cortical damage as well as diffuse periventricular degeneration, however, and one simply cannot ascribe all of the memory and behavioral disturbances to the mamillary injury alone. Nevertheless, this has had historical precedent, essentially starting with Gudden. For example, Kahn and Crosby ascribed mnestic derangements in their tumor patients to mamillary body injury and compression (20). However, there are valid objections to this assumption and similar assumptions elsewhere. One must admit that tumors of the mamillary region such as craniopharyngioma and pituitary adenoma could just as well compress medial thalamic or anterior thalamic structures. For example, one of Kahn and Crosby's patients had poor verbal memory, but excellent tactile memory (overcoming his verbal memory disturbance with the use of braille); this pattern has been ascribed to thalamic lesions (23). Dr. F. R. Ervin, in closing the discussion after Sweet's 1957 article, noted that neither the Indian elephant nor the porpoise, both thought to have excellent behavioral memory and spatial orientation, have mamillary bodies (34). The Thalami: Diencephalic Amnesia The mamillothalamic tract, one of the two main outflow tracts of the mamillary nuclei, projects to the anterior nuclear group of the thalamus. The anterior thalamic nuclei also receive direct input from the septohippocampal complex. The anterior nuclear group then projects to and receives from the cingulum. Many of these projections that end up in the cingulum continue on via the cingulate bundle, terminating in the hippocampal formation (pre- and parasubiculum and entorhinal cortex) (14). Furthermore, the cingulate cortex projects back to several thalamic nuclei. Primarily this is projection back to the anterior nuclear group, but also the dorsomedial and ventromedial nuclei and ventroanterior and lateral dorsal nuclei. Lesions have been placed in the anterior nuclei as a means of treating a variety of neuropsychiatric conditions such as uncontrollable aggression, schizophrenia, and hyperkinesia (29, 30, 38). The neuroanatomical supposition is that the septum and hippocampus may well play a role in the output of the cingulate gyrus via the anterior thalamic nuclei. As pointed out by Speigel and Wycis (29), these connections control emotional reaction, at least in part. Wood and Rowland do not allude to untoward side effects of bilateral anterior thalamic stereotactically produced lesions in hyperactive children; these children are described as virtually untestable in any case (38). Anterior thalamotomy for depression, anxiety, schizophrenia, and pain disorders by Spiegel and Wycis produced reduction in assaultive, aggressive, agitative, and maladaptive behavior with the neurological side effect of "chronotaraxis" (confusion in time). However, most of their lesions were in the dorsomedial nucleus. Although the foramen of Monro may be enlarged by section of the fornix and septum, it may also be enlarged by removal of some of the anterior tubercle of the thalamus. There has been no reported case study

after such a unilateral lesion. In any event, the benefit to exposure is minimal in comparison with subchoroidal entry of the ventricle. Of considerably more importance in a discussion of potential harm to the limbic system and the integration of memory is the dorsomedial thalamus. The dorsomedial thalamus may be the critical lesion in patients with Korsakoff's syndrome (22, 29-31, 36). The dorsomedial nucleus of the thalamus interacts intimately with the frontal cortex. Considerable animal and human evidence has accrued that the medial thalamic lesion produces amnesia and learning deficit even if unilateral and, in fact, the dorsomedial nucleus has become preeminent in modern concepts of memory disturbance. Speedie and Heilman ascribe intention, the focus of attention, and the control of interference to the dorsomedial nucleus of the thalamus (27, 28). The dorsomedial nucleus is the first and most severely affected nucleus in WernickeKorsakoff syndrome. This is supported by the extensive study of the syndrome by Victor et al. (36). The physiological basis for medial thalamic function in memory is not known. The area, however, is richly interconnected with other thalamic nuclei, the temporal stem, the temporal lobes, the amygdala, and the orbital cortex. Amnestic derangement will certainly result from a bilateral lesion of the dorsomedial nuclei; however, a unilateral lesion can produce nmestic derangements whether dominant or nondominant. Markowitsch has offered an extensive review. Neither animal nor human studies can provide an absolutely clear-cut memory-related role of the dorsomedial nucleus, but it may well be an important relay station. It seems to take not only a dorsomedial thalamic lesion, but also another lesion in a memory-related pathway in tandem to produce the memory deficit. Although Victor et al. favor the magnicellular nuclei as operative in the registration of memory, Markowitsch favors the parvicellular part. McEntee et al. studied a man in whom a metastatic pulmonary adeno-carcinoma had invaded and destroyed the medial and posterior thalamus bilaterally with sparing of the mamillary bodies, mamillothalamic tracts, and anterior thalamic nuclei (22). In addition to his Korsakoff's psychosis, the patient had emotional lability, inattention, denial, and visual agnosia for faces. Unfortunately, the fornices were not mentioned. Nevertheless, this case supports the contention that abnormality of the thalami without abnormality of the mamillary bodies can produce amnesia. Spiegel et al. performed dorsomedial thalamotomy upon 30 patients for emotional disturbance or pain (29). Of these, 19 experienced "chronotar-axis" (disorientation in time) and 4 experienced disorientation in place as well. Apparently there was little other learning disability in these patients and the deficiency was transient. Exactly how the dorsomedial nucleus is functional in the registration of memory is unknown. Valenstein and Nauta found that the hippocampal formation communicated with the dorsomedial nucleus via the fornix in monkeys (35). In three other species, the connection did not occur; it seemed to occur only in primates. The dorsomedial nucleus also receives fibers from the amygdala, suggesting a "dual activation" by both the hippocampus and the amygdala, thereby implying a regulating activity by these structures via the dorsomedial nucleus of the prefrontal cortex in primates (19). Speedie and Heilman suggested that a dorsomedial nucleus lesion will disrupt two separate limbic pathways; the basolateral orbitofrontal thalamic amygdalar system and the medial septohippocampal system (27, 28). A dorsomedial thalamic lesion, then, might of itself satisfy the double lesion requirement of a mnestic derangement. A dorsomedial nuclear lesion can be potent. It does not have to be bilateral to

be disabling. In addition to disrupting the two limbic circuits, a unilateral dorsomedial thalamic lesion may also induce frontal lobe signs. A domi-nent dorsomedial lesion will produce verbal memory disturbance, whereas a nondominant lesion will produce visuospatial memory disturbance (6, 27, 28). Nevertheless, these disturbances are probably less severe than the deficits produced by temporal ablation and in patients with Korsakoff's syndrome (27). Additionally, these patients have diminished spontaneity and lack of initiative. Patients with thalamic lesions of the dominant side will evidence not only verbal memory disturbance, but also speech disturbances with perseveration, neologisms, and anomia. It follows, then, that one should exercise caution in manipulation of the thalamus when utilizing the subchoroidal approach (7). Corpus Callosum Short anterior incision of the corpus callosum, 21/2 cm long, will produce no measurable deficit if independent of any other lesion (1). Nevertheless, the frontal two-thirds of the corpus callosum is implicated in memory retrieval and read-out. Memory retrieval is dependent on interhemispheric interaction (40). If the corpus is sectioned, each of the independently acting disconnected hemispheres is capable of storing and retrieving new information, but interhemispheric communication is necessary for other associative processes. A memory disturbance may well result, particularly if there is a requirement for association. Such a task would be the word-paired associate learning task in the Wechsler memory scale in which visual imagery support from the right hemisphere is usually needed. Similarly, a large callosal lesion in a right-handed patient would produce a motor dyspraxia in all tasks with the left hand, assuming organized motor engrams of the left hemisphere. The cases of callosal apraxia reported have been due to large lesions of the corpus callosum, such as large surgical sections for seizure disorders, infarction, or hemorrhage. Other peculiar abnormalities such as unilateral tactile anomia, hemialexia, and apathy seem to occur only in patients with complete, very large, or progressive cerebral commissurotomy. Even if just the splenium is spared, however, these deficits may be transient (3). In some of the commissurotomy patients, callosectomy was so extensive as to include the hippocampal commissure, and memory impairment may well be on that basis (40). It is evident from the literature that rostral callosectomy no larger than 21/2 cm by and of itself will produce no detectable disturbance on formal neuropsychometric testing. One must bear in mind, however, the admonition not to disturb two structures (the corpus callosum and one other), either of which might be operant in emotional and memory registration. Summary The neocortex seems to operate as an aggregation of columnar populations of neurons, each functioning as a unit. Interaction between populations is regulated by subcortical regions and this interaction generates a basic behavioral state and memory. The reticular formation, hypothalamus, thalamus, and basal ganglia, limbic system and related nuclei, mamillary bodies, and anterior and posterior cingulate gyri all seem to operate in this fashion (25). The organization is extremely complex, and attempting to arrive at a unitary concept of the neuroanatomical substrate of memory, for example, is impossible at present. The species difference between animals and humans may well invalidate much of what has been assumed. For one example, the fornix in the mouse and rabbit terminates almost entirely in the mamillary body,

whereas in the guinea pig and cat it terminates almost entirely in the central gray substance (35). There is much conjecture and many assumptions in both the animal and human literature and many times the investigator has not taken into account other limbic areas involved besides the one of his focus. The potential for an incomplete lesion to be collateralized or for functional reorganization to occur apace slower processes is hardly taken into account at all. Reviews of the effect of lesions of the fornix often do not distinguish the apparently high importance of the site of the lesion along the long axis of the fornix. Classically, the hippocampal formation, fornix, mamillary bodies, and possibly the anterior nucleus of the thalamus were implicated in memory. Newer thought including animal studies has added a large number of structures for our consideration. The dorsomedial thalamus, pulvinar, inferior thalamic peduncle, anterior commissure, anterior corpus callosum, temporal stem, diencephalic white fiber tracts, thalamocingulate tract, and parietopetal thalamic nuclei have all been added. Expert reviewers do not agree on a "memory site" because there is no single site of storage of memory. All of the foregoing structures are organs of relay. Furthermore, the literature is replete with evidence that it takes more than a single lesion to produce a deficit in emotion, memory, orientation, or arousal. A transcallosal approach to the third ventricle will have necessitated a callosectomy. To add to this lesion deliberately or accidentally forniceal, basal septal, dorsomedial thalamic, anterior thalamic, mamillothalamic, or hypothalamic lesion is to invite at least transient disturbance of spatial orientation, verbal or visuospatial memory, emotion, or arousal or an inability to interact with the environment. References Surgical Technique 1. Antunes J, Louis K, Ganti S: Colloid cysts of the third ventricle. Neurosurgery 5:450-455, 1980. 2. Apuzzo M, Chikovani O, Gott P, Teng E, Zwee C, Gianotta S, Weiss M: Transcallosal interforniceal approaches for lesions affecting the third ven tricle: Surgical considerations and consequences. Neurosurgery 10:547554, 1982. 3. Batnitzky S, Sarwar M, Leeds NE, Schechter MM, Behroz AK: Colloid cysts of the third ventricle. Radiology 112:327-341, 1974. 4. Bosch DA, Rahn T, Backlund EO: Treatment of colloid cysts of the third ventricle by stereotactic aspiration. Surg Neurol 9:15-18, 1978. 5. Buchsbaum HW, Colton RP: Anterior third ventricle cysts in infancy: Case report. J Neurosurg 26:264-266, 1967. 6. Bull JWD, Sutton D: The diagnosis of paraphysial cysts. Brain 72:487-516, 1949. 7. Busch E: A new approach for the removal of tumors of the third ventricle. Acta Psychiatr Neurol {Copenh) 19:57-60, 1944. 8. Chan R, Thompson G: Third ventricular colloid cysts presenting with acute neurological deterioration. Surg Neurol 19:358-362, 1983. 9. Ciric I, Zivin I: Neuroepithelial (colloid) cysts of the septum pellucidum. J Neurosurg 43:69-73, 1975. 10. Ciric I: Neuroepithelial cysts. J Neurosurg 44:134, 1976 (letter). 11. Cobb CA, Youmans JR: Brain tumors of disordered embryogenesis. In Youmans JR (ed): Neurological Surgery, ed 2. Philadelphia, WB Saunders, 1982, Vol5, ch91,p2899. 12. Cossu M, Lubinu F, Orunesu G, Pau A, Viale E, Sini M, Turtas S: Subchoroidal approach to the third ventricle: Microsurgical anatomy. Surg Neurol 21:325331, 1984. 13. Dandy WE: Benign Tumors in the Third Ventricle of the Brain: Diagnosis and Treatment. Springfield, IL, Charles С Thomas, 1933. 14. Deladsheer J, Guyot J, Jomin M, Scherpereel B, Laine E: Acces au troisieme ventricule par voie inter-thalamo-trigonale. Neurochirurgie 24:419-422, 1978.

15. Ehni G: Interhemispheric and percallosal (transcallosal) approach to the cingulate gyri, intraventricular shunt tubes, and certain deeply placed brain lesions. Neurosurgery 14:99-110, 1984. 16. Fulton JF, Bailey P: Tumors in the region of the third ventricle. J Nerv Meat Dis 69(1, 2, 3): 1929. 17. Geffen G, Walsh A, Simpson D, Jeeves M: Comparison of the effects of transcortical and transcallosal removal of intraventricular tumors. Brain 103:773-788, 1980. 18. Ghatak NR, Kasoff I, Alexander E: Further observation on the fine structure of a colloid cyst of the third ventricle. Acta Neuropathol [Berl) 39:101-107, 1977. 19. Greenwood J: Paraphysial cysts of the third ventricle: With report of eight cases. JNeurosurg 6:153-159, 1949. 20. Guner M, Shaw MDM, Turner JW, Steven JL: Computed tomography in the diagnosis of colloid cyst. Surg Neurol 6:345-348, 1976. 21. Gutierrez-Lara F, Patino R, Hakim S: Treatment of tumors of the third ventricle: A new and simple technique. Surg Neurol 3:323-325, 1975. 22. Harper RL, Ehni G: The anterior transcallosal approach to brain tumors. Adv Neurosurg 7:91-94, 1979. 23. Hines M: Studies in the growth and differentiation of the telencephalon in man: The fissura hippocampi. J Comp Neurol 34:73-171, 1922. 24. Hirano A, Ghatak NR: The fine structure of colloid cysts of the third ventricle. J Neuropathol Exp Neurol 33:333-341, 1974. 25. Hirsch J, Zouaoui A, Reiner D, Pierre-Kahn A: A new surgical approach to the third ventricle with interruption of the striothalamic vein. Ada Neurochir [Wien) 47:135-147, 1979. 26. Isaccson RL: The Limbic System, ed 2. New York, Plenum Press, 1982. 27. Jeeves M, Simpson D, Geffen G: Functional consequences of the transcallosal removal of intraventricular tumors. J Neurol Neurosurg Psychiatry 42:134142, 1979. 28. Kelly R: Colloid cysts of the third ventricle: Analysis of twenty-nine cases. Brain 74:23-65, 1951. 29. King T: Removal of intraventricular craniopharyngioma through the lamina terminalis. Ada Neurochir {Wien) 45:277-286, 1979. 30. Larson CP: Colloid (paraphysial) cyst of third ventricle with rupture into caudate nucleus and internal capsule. J Nerv Ment Dis 91:557-565, 1940. 31. Lavyne M, Patterson R: Subchoroidal transvelum interpositum approach to mid-third ventricular tumors. Neurosurgery 12:86-94, 1983. 32. Little JR, MacCarty CS: Colloid cysts of the third ventricle. J Neurosurg 40:230-235, 1974. 33. McKissock W: The surgical treatment of colloid cysts of the third ventricle: A report based upon twenty-one personal cases. Brain 74:1-9, 1951. 34. Oldberg E, Eisenhardt L: The neurological diagnosis of tumors of the third ventricle. Trans Am Neurol Assoc, 64:33-35, 1938. 35. O'Rahilly R, Gardner E: The developmental anatomy and histology of the human central nervous system. In Vinken P, Bruyn G (eds): Handbook of Clinical Neurology. Amsterdam, North Holland Publishing Co, 1977, vol 30, ch2, pp 15-40. 36. Ottoson D: The limbic system. In Physiology of the Nervous System. New York, Oxford University Press, 1983, ch 19, pp 286-288. 37. Palacios E, Azar-Kia B, Shannon M, Messina AV: Neuroepithelial (colloid) cysts: Pathogenesis and unusual features. AJR 126:56-62, 1976. 38. Pecker J, Guy G, Scarabin JM: Third ventricle tumors. In Yincken PJ, Bruyn GW (eds): Handbook of Clinical Neurology. Amsterdam, North Holland Pub lishing Co, 1974, vol 17, pp 440-489. 39. Poppen JL, Reyes V, Horrax G: Colloid cysts of the third ventricle: Report of seven cases. J Neurosurg 10:242-263, 1953. 40. Powell M, Torrens M, Thomson J, Horgan J: Isodense colloid cysts of the third ventricle: A diagnostic and therapeutic problem resolved by ventricu loscopy. Neurosurgery 13:234-237, 1983. 41. Riddoch G: Progressive dementia without headache or changes in the optic discs due to tumors of the third ventricle. Brain 59:222-233, 1936. 42. Sackett JF, Messina AV, Petito CK: Computed tomography and magnification vertebral angiography in the diagnosis of colloid cysts of the third ventricle. Radiology 116:95-100, 1975. 43. Sage MR, McAllister VL, Kendall BE, Bull JWD, Moseley IF: Radiology in the diagnosis of colloid cysts of the third ventricle. Br J Radiol 48:708-723, 1975.

44. Sharp JA: The junctional region of the cerebral hemisphere and the third ventricle in mammalian embryos. J Anal 93:159-168, 1959. 45. Shuangshoti S, Roberts MP, Netsky MG: Neuroepithelial (colloid) cysts. Arch Pathol {Chicago) 80:214-224, 1965. 46. Shuangshoti S, Netsky MG: Neuroepithelial (colloid) cysts of the nervous system. Neurology {NY) 16:887-903, 1966. 47. Shucart W, Stein B: Transcallosal approach to the anterior ventricular sys tem. Neurosurgery 3:339-343, 1978. 48. Stovall E: Ubereine Ependymeyste embryonalen charakters (Paraphyse) im dritten him Ventrikel mit Todlichem Ausgang. Beitr Pathol Anat 47:248269, 1909. 49. Stein B: Third ventricular tumors. Clin Neurosurg 27:315-331, 1980. 50. Stookey B: Intermittent obstruction of the foramen of Monro by neuroepithe lial cysts of the third ventricle: Symptoms, diagnosis and treatment. Bull Neurollnst NY 3:446-500, 1934. 51. Viale G, Turtas S: The subchoroid approach to the third ventricle. Surg Neurol 14:71-76, 1980. 52. Weinberger LM, Boshes B: Paraphyseal cysts of the third ventricle: Diagnosis and surgical management. Surgery 13:368-384, 1943. 53. Weisenburg TH: Tumors of the third ventricle. Brain 33:236, 1911. 54. Yenermen MH, Bowerman CI, Haymaker W: Colloid cyst of the third ventricle: A clinical study of 54 cases in the light of previous publications. Ada Neuroveg 17:211-277, 1958. Limbic System with Special Attention to Structures Involved in the Third Ventricular Approach 1. Apuzzo MLJ, Chikovani OK, Gott PS, et al: Transcallosal, interforniceal approaches for lesions affecting the third ventricle: Surgical considerations and consequences. Neurosurgery 10:547-554, 1982. 2. Benson DF, Gerschwind N: Shrinking retrograde amnesia. J Neurol Neuro surg Psychiatry 30:539-544, 1967. 3. Bogen JE: The callosal syndrome. In Heilman KM, Valenstein E (eds): Clinical Neuropsychology. Oxford, Oxford University Press, 1979. 4. Cairns H, Mosberg WH: Colloid cyst of the third ventricle. Surg Gynecol Obstet 92:545-570, 1951. 5. Carr AC: Memory deficit after fornix section. Neuropsychologia 20:95-98, 1982. 6. Choi D, Sudarsky L, Schachter S, et al: Medial thalamic hemorrhage with amnesia. Arch Neurol 40:611-613, 1983. 7. Cossu M, Lubinu F, Orunesu G, et al: Subchoroidal approach to the third ventricle. Surg Neurol 21:325-331, 1984. 8. DeJong RN: The hippocampus and its role in memory. J Neurol Sci 19:7383, 1973. 9. Delay J, Brion S, Derouesne C: Syndrome de Korsakoff et Etiologie Turmorale. Rev Neurol 111(2):97-133. 10. Dott NM: Surgical aspects of the hypothalamus. In Clarke WEL, Beattie J, Riddoch GG, et al (eds): The Hypothalamus: Morphological, Functional, Clinical, and Surgical Aspects. London, Oliver & Boyd, Ltd, 1938. 11. Ehni G: Interhemispheric and percallosal (transcallosal) approach to the cingulate gyri, intraventricular shunt tubes, and certain deeply placed brain lesions. Neurosurgery 14:99-110, 1984. 12. Gaff an D: Recognition impaired and association intact in the memory of monkeys after transection of the fornix. J Comp Physiol Psychol 86:11001109, 1974. 13. Garcia-Bengochea F, de la Torre O, Esquivel O, et al: The section of the fornix in the surgical treatment of certain epilepsies. Trans Am Neurol Assoc 79:176-178, 1954. 14. Hall E: The anatomy of the limbic system. In Mogenson GJ, Calaresu FR, Box В (eds): Neural Integration of Physiological Mechanisms and Behavior. Toronto, University of Toronto Press, 1981. 15. Hassler R, Riechert T: Uber einen Fall von Doppelseitiger Fornicotomie bie SogenannterTemporaler Epilepsie. Acta Neurochir {Wien) 5:330-340, 1957. 16. Heilman KM, Sypert GW: Korsakoff's syndrome resulting from bilateral fornix lesions. Neurology {NY) 27:490-493, 1977. 17. Henderson J, Green E: Behavioral effects of lesions of precommissural and postcommissural fornix. Brain Res Bull 2:123-129, 1977.

18. Horel JA: The neuroanatomy of amnesia. Brain 101:403-445, 1978. 19. Isaacson RL: The Limbic System, ed 2. New York, Plenum Press, 1982. 20. Kahn EA, Crosby EC: Korsakoff's syndrome associated with surgical lesions involving the mamillary bodies. Neurology (NY) 22:117-125, 1972. 21. Lavyne MH, Patterson RH: Subchoroidal transvelum interpositum approach to mid-third ventricular tumors. Neurosurgery 12:86-94, 1983. 22. McEntee WJ, Biber MP, Perl DP, Benson DF: Diencephalic amnesia: A reap praisal. J Neurol Neurosurg Psychiatry 39:436-441, 1976. 23. Markowitsch HJ: Thalamic mediodorsal nucleus and memory: A critical evaluation of studies in animals and man. Neurosci Biobehav Rev 6:351380, 1982. 24. Mathew NT, Meyer JS: Pathogenesis and natural history of transient global amnesia. Stroke 5:303-311, 1974. 25. Pay RG: Control of complex conation and emotion in the neocortex by the limbic entorhinal, subicular, and cingulate cortices and the hypothalamus, mamillary body, and thalamus. Intern J Neurosci 15:1-30, 1981. 26. Scoville WB, Milner B: Loss of recent memory after bilateral hippocampal lesions. J Neurol Neurosurg Psychiatry 20:11-21, 1957. 27. Speedie LJ, Heilman KM: Amnestic disturbance following infarction of the left dorsomedial nucleus of the thalamus. Neuropsychologia 20:597-604, 1982. 28. Speedie LJ, Heilman KM: Anterograde memory deficits for visuospatial ma terial after infarction of the right thalamus. Arch Neurol 40:183-186, 1983. 29. Spiegel EA, Wycis HT, Orchinik CW, Freed W: The thalamus and temporal orientation. Science 121:771-772, 1985. 30. Spiegel EA, Wycis HT: Multiple representation of various functions in the human subcortex. Confin Neurol 29:163-167, 1967. 31. Squire LR, Moore RY: Dorsal thalamic lesion in a noted case of human memory dysfunction. Ann Neurol 6:503-506, 1979. 32. Strub RL, Black FW: Organic Brain Syndromes. Philadelphia, FA Davis, 1981. 33. Swanson LW: The anatomical organization of septo hippocampal projections. In Ciba Foundation Symposium 58: Functions of the Septo Hippocampal System. Amsterdam, Elsevier, North Holland, 1978. 34. Sweet WH, Talland GA, Ervin FR: Loss of recent memory following section of the fornix. Trans Am Neurol Assoc 84:76-82, 1959. 35. Valenstein ES, Nauta WJH: A comparison of the distribution of the fornix system in the rat, guineapig, cat, and monkey. J Comp Neurol 113:337-362, 1959. 36. Victor M, Adams RD, Collins GH: The Wernicke-Korsakoff Syndrome. Phil adelphia, FA Davis Co, 1971. 37. Weiskrantz L: Experimental studies of amnesia. In Whitty CWM, Zangwill OL (eds): Amnesia. New York, Appelton, Century, Crofts, 1966. 38. Wood MW, Rowland JP: Bilateral anterior thalamotomy for the hyperactive child. South Med J 61:36-39, 1968. 39. Woolsey RM, Nelson JS: Asymptomatic destruction of the fornix in man. Arch Neurol 32:566-568, 1975. 40. Zaidel D, Sperry RW: Memory impairment after commissurotomy in man. Brain 97-.263-272, 1974.

14 Transcallosal Interforniceal Approach Michael L J. Apuzzo, M.D., and Steven L Giannotta, M.D.

Historical Perspective In 1944, Edward Busch published the description of a new approach for the removal of third ventricular tumors (6). Troubled by difficulties associated with efforts to remove colloid cysts and more particularly solid tumors through the foramen of Monro, he noted the following incidental surgical observation: "In removing a tumor of the septum pellucidum we had the opportunity of observing in detail the roof of the third ventricle from above and were impressed by its thinness and by the 'well-marked median raphe.' We thought that perhaps it might be possible to divide the roof of the third ventricle in the midline, thus gaining direct and satisfactory access to the ventricle itself." After substantiating the anatomical validity of such a deduction in several postmortem brains, he undertook, with some trepidation, the first procedure employing the interforniceal maneuver in a 48-year-old man with a colloid cyst. After a right frontal transcortical lateral ventricular exposure, "the septum pellucidum was split just above the roof of the third ventricle. . . . The roof of the third ventricle now bulged upwards and it was only necessary to start the division of the roof at the most prominent point by means of a blunt hook, after which the intraventricular pressure gradually completed the division, the tumor bulging upwards with every pulsation. The tumor was punctured but seemed solid; it was cautiously lifted up and seen to adhere to the Choroid plexus. A clip was placed anteriorly to the tumor and the stalk was cut through. The whole third ventricle was in full view." Busch had thus described division and definition of the midline forniceal raphe with subsequent perforation of the diencephalic roof along a natural plane between the two forniceal bodies. In a total of six operated patients in which this maneuver was used to gain third ventricular access, no untoward effects were observed. Little further commentary is available on the technique or the results of such a procedure. Based on the techniques of Dandy (8) and van Wagenen (23), during the 1960s a series of papers provided a substrate for the application of a strictly midline interhemispheric approach to the third ventricle that incorporated the technique for entry described by Busch. Prompted by the split brain animal studies of Sperry, Baldwin et al. (4) described a technique for medial cerebral incision in primates, which was later undertaken with greater refinement by Milhorat and Baldwin (20), with the addition of the technical aid of the operating microscope. The authors

Figure 14.1. Angled anatomical schematic diagram demonstrating major incisions in midline neural structures to the third ventricular chamber. Figure 14.2. Dorsal fornix view with anatomical elements of structure, line of incision, and silhouette of the underlying third ventricle.

stressed, "The approach affords a unique and extensive view of the third ventricular chamber. . . . Clinically, the transcallosal microdissection technique is of potential value in the exposure and removal of a number of midline tumors." In this chapter, we describe the technique of midline interhemispheric exposure via the corpus callosum and the forniceal component of the roof of the third ventricular chamber (Figs. 14.1 and 14.2). This technique affords excellent visualization primarily of the anterior and mid-third ventricular regions and may be safely used in the absence of lateral ventriculomegaly (1-3). Operative Corridor: Anatomy and Physiological Risks The major topographic elements of the midline corridor that require identification and consideration include the coronal suture, the sagittal sinus, the parasagittal veins, the falx cerebri, the cingulate gyrus, the pericallosal arteries, the corpus callosum, the fornix and its components, the tela choroidea, the medial posterior choroidal arteries, and the internal cerebral veins. Consideration of the consequences of injury to each neural or vascular component is essential as a stepwise progression evolves through the corridor of exposure. Parasagittal Veins Initial cortical visualization and manipulation of the parasagittal venous elements during the early stages of exposure and subsequent retraction may cause proximal or remote cortical injury. Other than direct manipulation, the consequences of potential venous infarction is an important consideration. Absolute preservation of the venous tributaries is an important goal within the surgical process. It has been our experience that appropriate placement of an economically sized bone flap may be suitably planned in relation to venous anatomy as defined by cerebral angiography rather than strictly by bony landmarks. Review of 100 angiograms (3) with particular attention to the distribution and confirmation of the parasagittal venous complex in the region of the coronal suture has disclosed that 42 of the studies showed evidence of significant venous tributaries from the mid- and posterior frontal lobe draining within 2 cm anterior or posterior to the coronal suture. The majority (70%) of these venous tributaries entered the sagittal sinus within the sector 2 cm posterior to the coronal suture, whereas 30% were evident in the anterior 2-cm region. Sacrifice of these major tributaries could cause either cortical deficit or seizures. Planning of bone flap placement in the pericoronal area and the extent and mode of brain retraction or deformation should take into account the parasagittal venous anatomy as an important guiding factor (Fig. 14.3). Corpus Callosum Major callosal incision and limited but strategically placed interruption of callosal pathways have been shown to result in disorders of interhemispheric transfer of information such as visual spatial transfer, tactile informational transfer, and bimanual motor learning (12, 13). The issue of limited incision through the superior genu and trunk has been addressed on a number of occasions (9, 11, 14, 18, 25). Jeeves et al. (18) reported three patients who underwent 30-mm incision in the mid- and anterior corpus callosum for the excision of third ventricular neoplasms. All were found to have deficits in transfer of tactile data, but not of information obtained visually. None of the patients was aware of this inability or inconvenienced by it. Winston et al. (25) studied four children ranging in age from 18 months to 10 years, who underwent 2-cm longi-

Figure 14.3. The bone flap is placed with consideration of the major parasagittal veins in the pericoronal area. Care is taken to avoid sacrifice of venous tributaries contributing to mid- and posterior frontal drainage. tudinal incision in the anterior body of the corpus callosum. None of the children demonstrated any of the signs of commissural disconnection. Tests of dexterity in intermanual and intramanual sensory transfer indicated successful interhemispheric transfer in spite of the young age of the patients tested and the location of their lesions. Gordon et al. (14) studied two patients with nearly complete section of the body, genu, and anterior commissure. No evidence of interhemispheric disconnection was evident. Apuzzo et al. (3) studied six patients with 2.5-cm incisions in the callosal trunk. All patients could readily name objects and numbers placed out of sight in the nondominant hand. Crossed and uncrossed localization of light touch was accurate from right to left hands. Interhemispheric transfer of tasks requiring tactile feedback in motor sequencing was successful and complete. These results indicated no measurable deficits in interhemispheric transfer of somesthetic information or complex perceptual motor learning tasks requiring continual sensory motor integration. Therefore, the majority of data support the concept that limited incision of the callosal trunk effects a minimal physiological alteration. * Fornix Once lateral ventricular entry has been effected, third ventricular lesions may be approached by a number of surgical maneuvers that involve an element of manipulation of the fornix (1). These maneuvers have been undertaken with trepidation because of the considered risk of amnesia during the postoperative period. Review of the literature related to the role of the fornix and memory processes provides contradictory opinions. The severe amnesia that has been attributed to damage of the hippocampus and mamillary bodies has logically led to the prediction that alteration of the forniceal structure would likewise produce amnesia. Review of available literature provides no consistently substantive evi-

dence of amnesia that has resulted from isolated forniceal injury. Dott (10) and Cairns and Mosberg (7) reported patients who had no memory deficits after section of the fornix. Bengochea et al. (5) sectioned the fornix bilaterally for epilepsy and noted no alterations in mentation in 12 patients. Woolsey and Nelson described neoplastic destruction of the fornix without memory loss (26). Hassler and Riechert (15) reported a patient who underwent bilateral fornicectomy for epilepsy. A lesion was placed stereotactically in the column at the level of the anterior commissure by electrical coagulation. The initial lesion was placed on the right and produced no amnesia. A second lesion in the left column 11 months later produced a severe memory deficit, however, autopsy revealed bilateral forniceal lesions with a tumor of the right medial frontal lobe and softening of the left hypothalamus. Sweet et al. (22) reported a patient with a colloid cyst of the third ventricle that was removed after bilateral sectioning of the column of the fornix, with subsequent manifestation of a severe amnestic syndrome. Memory for remote events was preserved, however; there was persistence of severe loss of recent events, which did not resolve during 2 years of follow-up. Horel (17) suggested that injury to the magnocellular medialis dorsalis during cyst removal could be considered the cause. Heilman and Sypert (16) reported a patient with amnesia and a glioma in the region of the hippocampal commissure. After subtotal excision of the lesion, the patient manifested a profound and persistent amnestic deficit. No structural verification for lesion site was reported. Memory loss has been reported after major extensive commissurotomy in which the hippocampal commissure was sectioned (27). In review, there are clear instances of destruction of the fornix with absence of detectable amnesia and also cases in which fornix damage has been implicated in amnesia. These reports may be subject to various interpretations based on the presence of attendant neural destruction or the lack of verification of structural alterations. Incision of the forniceal raphe and retraction of the body has not resulted in persistent amnestic syndromes in our hands, with mentation evaluated not only by standard bedside examination, but also by psychometric assessment (2, 3). The objective of the exposure is to take advantage of a natural plane of cleavage. This exposure often requires no retraction of the forniceal elements beyond that produced by the mass lesion within the third ventricular chamber. Structural Definition Preoperative radiographic assessment is directed toward the development of an absolute definition of the anatomical substrate peculiar to the given patient. The goal is to obtain clear perspectives of the individual's anatomy along the operative corridor and of all aspects of distortion and redisposition of anatomical elements by the lesion. In general, high resolution computerized tomography (CT) in axial, coronal, and particularly midline sagittal planes will accomplish adequate definition. Magnetic resonance imaging (MRI) in multiple planes will augment CT and at times provide important information related to lesion definition. Cerebral angiography provides important information related to cortical venous anatomy and occasionally augments imaging data related to the vascu-larity of the lesion (Fig. 14.4). Strict definition of surgical strategy and economy of tissue manipulation can be enhanced by careful consideration of the precise midpoint of bone flap placement, the angulation of retractor placement to the midpole of the lesion, and the distance of the corridor to the corpus callosum, fornix, and lesion. The size and extent of callosal and forniceal incisions

Figure 14.4. Cystic intraventricular craniopharyngioma. A. CT with stippled calcium. B. MRI demonstrating a sagittal view of the lesion with the anatomical corridor for entry. C. Venogram with the coronal suture landmark [arrow) as a guide for the bone flap plan. D. Postoperative (24 hours) CT. In this case the foraman of Monro (4 mm) was not distended by the mass, and some maneuver was clearly required for fornix manipulation. may be estimated preoperatively in relation to the apparent type, size, and disposition of the lesion within the ventricle. The relation of the fornix to the corpus callosum and the area of the septum pellucidum may be determined. There is considerable variability regarding not only the relationship of all normal anatomical elements, but also the structural substrate deformed by the pathological process; absolute definition of the entire complex will permit maximal economy of manipulation and thus provide for ultimate safety and satisfactory management of the pathological process. Operative Technique After the induction of general anesthesia with endotracheal intubation, the patient is positioned supine and the head is elevated 15 to 20° from the horizontal and supported by a Mayfield pin fixation headrest (Fig. 14.5). Consideration is given to the placement of a left frontal ventriculostomy. This is used especially in individuals with ventriculomegaly as an adjuvent for intraoperative intracranial pressure control and postoperative pressure monitoring. High potency glucocorticoid preparations are used in all, and prophylactic antibiotics are employed in patients with ventriculostomies. A paramedian right side 6- x 4- X 3-cm trapezoidal free bone flap is most commonly used for calvarial entry. Optimal placement for access to the foramen of Monro dictates that the bone flap be bisected by the

Figure 14.5. Patient is positioned in supine with head fixed in 20° of flexion from the natural longitudinal axis.

Figure 14.6. Visual access to the entire chamber is possible within variables of angles of entry at the cortical level. coronal suture. However, to reduce the possibility of cortical injury we plan the bone flap in relation to the specific cortical venous anatomy of the individual patient. Moving the bone flap posteriorly, although increasing the proximity to the paracentral hemispheric region, may produce a more optimal angle for corridor access to the forniceal body with raphe incision extending from the foramen of Monro posteriorly. This angle of access (Fig. 14.6) affords optimal visualization of all third ventricular

regions, but particularly the anterior third. Predominantly anterior placement of the bone flap in relation to the coronal suture promotes better visualization of the posterior third of the chamber. Depending on the thickness of the skull, the bone flap is developed on the midline or 1 cm to the left of the midline (Fig. 14.7). Generous and absolute midline exposure is required for ease of identification of the interhemispheric fissure, but also for minimization of retraction as the entry corridor is

Figure 14.7. Some options for developing the bone flap to assure midline exposure over 5 to 6 cm. With calvaria thicker than 1 cm, exposure is facilitated by incorporating 1- to 1.5-cm contralateral paramedian bone removal (D).

developed. The midline approach to the third ventricle, particularly in cases with small lesions (1 cm) and no hydrocephalus, requires access to an absolutely sagittal plane in the line of the falx. Therefore, if this can be initiated at a superficial level, retraction in the medial to lateral plane may be minimized. All retraction and deformation of parasagittal brain parenchyma may be localized completely to a 5-cm anterior to posterior plane. A number of scalp incisions (Fig. 14.8) are appropriate to gain adequate calvarial exposure and meet the individual requirements of cosmesis. However, we generally use a two-limbed curvilinear scalp flap. Once scalp incision is initiated, furosemide and/or mannitol is administered to reduce cerebral mass, thus facilitating interhemispheric exposure and minimizing the required force of retractor application. After the application of dural tenting sutures, the Budde self-retaining retractor system (Fig. 14.9) is brought into the field and appropriately installed in relation to the headrest for fixation. This system is versatile and flexible, offering the opportunity to introduce either retraction or instrumentation in any direction within the field and simultaneously offering a stabilization device for the surgeons' wrists and hands. The instrumentation offers an anchoring base for retraction sutures and, thus, the ring component of the system may be established adjacent to the operative field before the bone flap is fashioned. A trapezoidal dural incision (Fig. 14.10) is developed with the broad base of the form positioned at the line of the sagittal sinus. Traction sutures are applied so that the dural flap is reflected over the sagittal sinus with an element of compression of this structure. These sutures are secured to the left side of the retractor ring. Absolute exposure of the midline plane at the level of the superior falx is realized.

Figure 14.8. Various scalp incisions offer flexibility and economy of exposure at the midline in the pericoronal area and allow ventriculostomy placement on the left.

Figure 14.9. A. Ring retractor installed after scalp incision. Here it is pictured after the initial set-up with the bone flap turned and the dura mater reflected to the midline and secured to the left superior ring. B. Microscope in position with bilateral ribbon retraction in the midline and the operator's hands stabilized by the ring. A self-retaining retractor arm with a 19-mm blade is prepared to be applied from the right side of the field, and initial blunt retraction is undertaken at the falx-cortex interface with the broad end of a Penfield (No. 1) instrument. Indentification of the inferior falx is achieved using this technique, and Biocol is applied to the exposed medial and lateral cortical surfaces. This component of the exposure having been developed, the operating microscope with either a 275-mm or a 300-mm lens is brought into the field. The choice of objectives is related to the potential use of the micromanipulator laser adaptor and the individual depth of field. With the use of a combination of blunt and sharp microinstrumen-tation, a midline plane is then developed between the cingulate gyri. The retractor blade and Biocol are advanced to the corpus callosum. The

Figure 14.10. A. Dural incision takes advantage of the full extent of the bone flap with approximation of the midline and the sagittal sinus. B. Initial development of the midline plane at the sagittal sinus with identification of the falx. The cortex will be deformed no more than 5 cm in the longitudinal plane. C. Initial development of the exposure to the cingulate gyrus. Lateral retraction should not exceed 2 cm.

cingulate gyrus may occasionally be confused with the corpus callosum. However, the cingulate gyrus is the distinctive tan-gray of the cortical pial surface, whereas the callosum is strikingly white. With identification of the corpus callosum, the pericallosal arteries are readily identified (Fig. 14.11, A and B). At the callosum the exposure may be enhanced by the placement of cotton balls that are saturated with Ringer's lactate solution and cylindrically shaped to maintain the longitudinal exposure at the anterior and posterior poles of the midline slot. A 3.5- x 1-cm callosal exposure is the endpoint of this stage of the procedure (Fig. 14.13). The falx is maintained as the midline reference with the original position of the pericallosal arteries serving as a secondary landmark. Having established callosal exposure and identified the pericallosal arteries, the surgeon plans the callosal incision. This entry is 2.0 to 2.5 cm long and depends on variables related to the angle of entry and the size of the lesion. Five French (5 F) irrigating suctions and multiple microinstrumentations are utilized to mobilize the pericallosal arteries (Fig. 14.11С). In the majority of cases, the right pericallosal artery is displaced 2 to 3 mm to the right in the pericallosal cistern and incision of the callosum is initiated immediately to the right of the left pericallosal artery. Again the falx serves as a midline reference point during the initiation of the incision. Occasionally, with substantial third venrticular masses or ventriculomegaly there may be a domelike configuration of the corpus callosum; shift of the midline may occur with asymmetrical ventriculomegaly; rarely the falx is shifted; all of these variables will be apparent on the preoperative imaging and should be anticipated. Interhemispheric dissection may be complicated by midline shift or, as in cases of cysticercosis, adhesive arachnoiditis. The thickness of the corpus will vary according to the extent of hydrocephalus. As dissection with a 5 F suction commences, the surgeon should evaluate the point of callosal incision as it relates to underlying anatomy. Depending on the point of incision and its relation to the anterior to posterior and medial to lateral planes, the following entry points are possible: (a) right lateral ventricle, (b) left lateral ventricle, (c) septum pellucidum with or without cavum, and (d) forniceal body (Figs. 14.12 and 14.13). Study of preoperative sagittal imaging with particular attention to fornix anatomy and disposition in relation to the corpus callosum, as well as the area and configuration of the septum pellucidum, will orient the surgeon in relation to possible landmarks during callosal transit. As callosal transit is completed one identifies the ependymal surface of the lateral ventricles. This layer is readily appreciated because of its gray-black hue, and ependymal vessels are readily visualized under magnification (Fig. 14.14). At this time, the line of attachment of the septum is often identified, indicating that precise midline entry to the ventricular system has been achieved (Fig. 14.14). If the septum is not appreciated, either right or left lateral ventricular entry is most likely. The self-retaining retractor blade is then advanced to the inferior callosal margin and the Choroid plexus is identified. This is the most striking and reliable landmark in the mid- and anterior lateral ventricular system. It may be followed to the foramen of Monro with appropriate orientation being gained in relation to this important central landmark. In most cases the abnormality may be appreciated through the foramen of Monro, which is quite variable in area. Anatomical perspective will vary according to entry angle (Fig. 14.15). With identification of the foramen, the septal vein, septum pellucidum, forniceal structure, thalamostriate vein, thalamic tubercle, and head of the caudate are readily appreciated. Bipolar coagulating forceps are used to fenestrate the sep-

Figure 14.11. A. Initial exposure of the corpus callosum with a single retractor and cotton balls at the anterior and posterior poles creating a 3.5- x 1.5-cm area of visualization. B. Photomicrograph of the callosal exposure with the anterior cerebral arteries in the midline. C. Coronal view with the double retractor option. D. The anterior cerebral arteries are displaced toward the lateral callosal cistern. The falx gives a midline orientation for the incision.

Figure 14.12. А, В, С. А 5 F suction irrigation device and bipolar forceps are used to incise the corpus callosum over a 2.5-cm length. Note in В the options for ventricular entry.

Figure 14.13. Steps in retractor advancement in the sagittal plane with care to assure the appropriate entry angle to the septum and the foramen of Monro.

Figure 14.14. Photomicrograph showing initial callosal resection with ependyma of the left and right lateral ventricles with transverse veins and midline attachment of the septum pellucidum.

_

Figure 14.15. Possibilities for field of view at the periforaminal region depending on sagittal entry angles [А, В, С). The angle of visualization will affect access to the raphe and the predominant perspective of the third ventricular chamber; A, mid and posterior; B, middle third; C, mid and anterior. В is optimal for overall exposure.

turn, creating a single ventricular cavity in the event that a lateral ventricular shunt is required (Fig. 14.16). At times septal leafs or a cavum facilitates identification of the raphe or midline division between the two forniceal bodies. The septum serves as a landmark for guidance to the

Figure 14.16. A. The septum pellucidum is coagulated to the dorsal surface of the fornix. B. Visualization of the left foramen is gained by retraction. C. Photomicrograph with the septum being coagulated and soft glial tumor evident at the foramina of Monro bilaterally adjacent to forceps.

midline union of the forniceal columns and bodies superior and posterior to the anterior commissure and medial and superior to the foramen of Monro (Fig. 14.17). With midline entry, identification of the septum, and moderate ventriculomegaly, bilateral visualization of the foramen of Monro is readily achieved without a left-sided retractor (falx, cingulum, and corpus). The surgeon must remain cognizant of entry angles related to each level of the corridor as the perspective of the foramen of Monro and thus the fornix and finally the third ventricular chamber will vary according to minor angular sagittal variables. Minimal changes in entry angles may confuse the site of ventricular visualization and access. Having gained midline exposure with appropriate landmark orientation, the raphe is identified at the site of septum attachment on the dorsal fornix (Fig. 14.17); incision is commenced at the level of the foramen of Monro with a fine tip bipolar forceps and Sheehy canal knife (Fig. 14.18). The incision is carried posteriorly for 1.0 to 2.0 cm (Fig. 14.19). The size and shape of the fornix in the midline are quite variable and are influenced by individual variations and deformation by the pathological mass. What to expect during operative transit is indicated by preoperative imaging studies. With completion of the incision, the mass is readily identified (Fig. 14.20). The normal structures in the diencephalic roof including the tela choroidea, Choroid plexus, internal cerebral veins, and posterior choroidal arteries are often attenuated or displaced laterally by the presence of the mass. Secondary retraction at the level of the fornix is often not necessary; however, a 5-mm retractor blade may be utilized to maintain exposure either initially or once mass decompression has been partially achieved (Figs. 14.21 and 14.22). Access to the lesion may be possible through the midline raphe incision with views of anterior and posterior poles enhanced by minor alterations in retractor and microscope placements. Secondary manipulations of the lesion are possible through the foramina of Monro. Utilization of laser instrumentation and the introduction of miniature angular mirrors are accomplished readily.

Figure 14.17. Photomicrograph with the crest of the coagulated septum marking the raphe of the fornix adjacent to the foramen of Monro distended by craniopharyngioma.

Figure 14.18. A. The junction of the septum and the fornix provides the landmark for the forniceal raphe, which may be established and developed with a canal knife. В, С. The mass lesion is identified and the raphe is developed over a 2cm extent posteriorly from the foramen of Monro using bipolar forceps and a canal knife. D. A cystic lesion is incised and drained causing relaxation of the tense forniceal displacement and initiating the requirement for retraction on the forniceal body to maintain exposure.

Figure 14.19. Photomicrograph demonstrating initial dissection of the raphe during operation for a third ventricular ependymoma.

Figure 14.20. Photomicrograph demonstrates craniopharyngioma distending the raphe after the initial incision. The foramen of Monro is to the right of the field. In developing the interforniceal plane, one should not exceed 2 cm posterior to the foramen of Monro (Fig. 14.2). The surgeon must be cognizant of the hippocampal commissure in the posterior component of the forniceal structure. There is evidence in the literature suggesting that compromise of this area may lead to permanent memory impairment and, therefore, we have been reluctant to extend our incision posteriorly into this region. Likewise, we have not carried the incision beyond the region

Figure 14.21. Photomicrograph with a 5-mm retractor blade on the right fornix in a case of craniopharyngioma.

Figure 14.22. Use of the Budde ring retractor with multiple options exercised. The midline and posterior parasagittal retractors are 19 mm (cingulum, corpus callosum). The anterior parasagittal retractor is 5 mm (forniceal body). of the column-anterior commissure interface. The extent of the incision as described provides excellent visualization of all regions of the ventricle and in particular in the presence of a mass lesion, which enhances the overall exposure by its deformation of surrounding structures. At times the internal cerebral veins will be visualized. These have been retracted with self-retaining instruments without morbidity. Mass management strategy follows conventional standards of micro-surgical technique (Fig. 14.23). After initial exposure of the intraventricular lesion, needle aspiration may be undertaken or the mass may be incised and aspirated with a 5 F or 3 F suction. This may cause collapse

Figure 14.23. Lesion exposure and excision. A. Wall biopsy. B. Dissection of anterior pole (i), lateral recesses (2), and posterior pole (3). C. The internal cerebral veins may be identified laterally displaced by the mass expansion. of the forniceal bodies into the midline and necessitate further retraction; in this case a 5-mm retractor blade may be introduced from the right of the field. If the lesion is not cystic, after wall biopsy central decompression may be achieved with either a carbon dioxide laser or bipolar coagulation and microinstrumentation. Once internal decompression is achieved, anterior and lateral components of the mass may be dissected from the third ventricular wall with gentle traction on the lesion, developing the tumor-brain interface with a Sheehy canal knife. With lateral dissection the internal cerebral veins

may be identified and followed to the foramen of Monro anteriorly. The Choroid plexus is often intimately adherent to the junction with the thalamostriate vein. Fragments of tumor are removed, improving exposure. Visualization of the third venticular floor (if intact) or the sellar content is readily appreciated as the anterior and inferior poles are mobilized. Finally, the posterior component of the mass is approached. Depending on entry angles, this may require minor alterations of head position by reverse Trendelenburg maneuver, appropriate posterior angulation of the microscope, or the use of miniature mirrors (Figs. 14.24

Figure 14.24. A. Lesions may be manipulated simultaneously through the foramen of Monro and the interforniceal entry. В, С Visualization of the posterior chamber, if limited, may be gained with an angled dental mirror.

Figure 14.25. Completion of mass excision with an instrument in the foramen and the retractor removed. There is redundant fornix secondary to mass displacement. and 14.25). With mass excision extending from the diencephalic roof to the dorsum sellae, it is not unusual to visualize the clivus, basilar circulation, and prepontine cistern at the completion of lesion excision. Encounters with major venous and arterial structures are premeditated and defined by preoperative radiographic studies. Complications Although a number of complications may be attendant to dissection of lesions within the third ventricle (alteration of consiousness, gastrointestinal hemorrhage, increased endocrinopathy, visual loss, and other signs of diencephalic injury), the major considerations in establishing the transcallosal interforniceal corridor are the development of hemiparesis and memory loss (2, 3, 10, 17-19, 21, 24, 27). With care related to midline entry in relation to the cortical venous anatomy and minimization of midline retraction, the incidence of permanent paresis will approach zero and that of transient paresis will be less than 10%. Transient recent memory loss is the single most commonly encountered postoperative problem. This has been observed in 30% of our personal series, with complete resolution of this symptom generally observed within a 21-day period. This disorder of mentation is recognized as an amnesia for recent events that is most striking 24 to 72 hours postoperatively, with gradual resolution thereafter. Seventy-five percent of our cases manifested resolution of this problem within a 7-day period. All patients had returned to preoperative base line status or improved to normal mentation at 3 months postoperatively. The manifestation of this complaint was thought to be more related to the texture of the offending lesion than to the size of the lesion, with firm masses creating more need for local manipulation and the transmission of pressure to adjacent periventricular structures.

Utilization Advantages The transcallosal interforniceal corridor affords exposure of lesions primarily within the anterior and mid-third ventricle with minimal phys-

Figure 14.26. A. Multiple view CT of a partially cystic craniopharyngioma with a limited basilar and a significant third ventricular component. B. Early postoperative CT with a small asymptomatic frontal fluid collection. Adequate access and complete removal was accomplished by interforniceal exposure.

iological cost; in addition, exposure of the posterior component of the chamber may be effected. Lesions with basal components may be approached (Fig. 14.26). The technique allows for adequate exposure of the third ventricular chamber in the event of minimal ventricular enlargement or minimal mass expansion within the region. Additionally, it allows for simultaneous manipulation of and approach to the region through the foramen of Monro and the transforniceal corridor for added technical advantage. If the foramen of Monro is small or the lesion is not well visualized, the maneuver offers extensive exposure of the third ventricular chamber, aiding the comfort of the surgeon. Limitations Inherent limitations of the approach are stressed by lesions that are less than 1.5 cm in size and are located primarily posterior in the chamber and by lesions that are primarily basilar with lateral extension and present a restricted or narrow apex deforming the mid- and anterior fornix. For the latter, a combined basal (for lateral access) and transcallosal midline approach is often required, whereas for the former a direct posterior corridor should be considered. References 1. Apuzzo MLJ: Transcallosal Interforniceal exposure of lesions of the third ventricle. In Schmidek HH (ed): Operative Neurological Techniques: Indi cations and Methods. New York, Grune and Stratton, 1982. 2. Apuzzo MLJ: Inforniceal exposure of the third ventricular chamber. In Samii M (ed): Surgery in and around the Brain Stem and Third Ventricle. Berlin, Springer-Verlag, 1.987. 3. Apuzzo MJ, Chikovani OK, Gott PS, Teng EL, Zee CS, Giannotta SL, Weiss MH: Transcallosal, interfornicial approaches for lesions affecting the third ventricle: Surgical considerations and consequences. Neurosurgery 10:547554, 1982. 4. Baldwin M, Ommaya AD, Farrier R, MacDonald F: Mesial cerebral incision. J Neurosurg 20:679-686, 1963. 5. Bengochea FG, De La Torre O, Esquivel O, Vieta R, Fernandez C: The section of the fornix in the surgical treatment of certain epilipsies. Trans Am Neurol Assoc 79:176-178, 1954. 6. Busch E: A new approach for the removal of tumors of the third ventricle. Acta Psychiatr Scand 19:57-60, 1944. 7. Cairns H, Mosberg WH Jr: Colloid cyst of the third ventricle. Surg Gynecol Obstet 92:545-570, 1951. 8. Dandy WE: Diagnosis, localization and removal of tumors of the third ven tricle. Johns Hopkins Hosp Bull 33.188-189, 1922. 9. Dimond SJ, Scammell RE, Brouwers EYM, Weeks R: Functions of the centre section (trunk) of the corpus callosum in man. Brain 100:543-562, 1977. 10. Dott NM: Surgical aspects of the hypothalamus. In Clark WEL, Beattie J, Riddoch GG, Doh NM (eds): The Hypothalamus: Morphological, Functional, Clinical and Surgical Aspect. Edinburgh, Oliver and Boyd, 1938, pp 131185. 11. Gazzaniga MS, Risse GL, Springer SP, Clark E, Wilson DH: Psychologic and neurologic consequences of partial and complete cerebral commissurotomy. Neurology (NY) 25:10-15, 1975. 12. Geschwind N: Disconnexion syndromes in animals and man: Part I. Brain 88:237-294, 1965. 13. Geschwind N: Disconnexion syndromes in animals and man: Part II. Brain 88:585-644, 1965. 14. Gordon HW, Bogen JE, Sperry RW: Absence of disconnexion syndrome in two patients with partial section of the neocommissures. Brain 94:327-336, 1971. 15. Hassler R, Riechert T: Uber einen Fall von Doppelseitiger Fornicotomie bei sogenannter Teporaler Epilepsie. Acta Neurochir (Wien) 5:330-340, 1957. 16. Heilman KM, Sypert GW: Korsakoff's syndrome resulting from bilateral fornix lesions. Neurology (NY) 27:490-493, 1977. 17. Horel JA: The neuroanatomy of amnesia: A critique of the hippocampal memory hypothesis. Brain 101:403-445, 1978.

18. Jeeves MA, Simpson DA, Geffen G: Functional consequences of the trans callosal removal of intraventricular tumours. J Neurol Neurosurg Psychiatry 42:134-142, 1979. 19. Long DM, Chou SN: Transcallosal removal of Craniopharyngiomas within the third ventricle. J Neurosurg 39:563-567, 1973. 20. Milhorat TH, Baldwin M: A technique for surgical exposure of the cerebral midline: Experimental transcallosal microdissection. J Neurosurg 24:687691, 1966. 21. Shucart WA, Stein BM: Transcallosal approach to the anterior ventricular system. Neurosurgery 3:339-343, 1978. 22. Sweet WH, Talland GA, Ervin FR: Loss of recent memory following section of fornix. Trans Am Neurol Assoc 84:76-82, 1959. 23. van Wagenen WP: Surgery of the hypothalamic region. Res Publ Assoc Res Nerv Merit Dis 20:841-853, 1939. 24. Williams M, Pennybacker J: Memory disturbances in third ventricle tumours. J Neurol Neurosurg Psychiatry 17:115-123, 1954. 25. Winston KR, Cavazzuti V, Arkins T: Absence of neurological and behavioral abnormalities after anterior transcallosal operation for third ventricular lesions. Neurosurgery 4:386-393, 1979. 26. Woolsey RM, Nelson JS: Asymptomatic destruction of the fornix in man. Arch Neurol 32:566-568, 1975. 27. Zaidel D, Sperry KW: Memory impairment after commissurotomy in man. Brain 97:263-272, 1974.

15 Subchoroidal Trans-Velum Interpositum Approach Michael H. Lavyne, M.D., and RusselH. Patterson, Jr., M.D.

Excision of brain tumors lying within the anterior to middle portion of the third ventricle has been difficult because of the surgeon's inability to open the ventricle widely and safely. Although it is possible to remove small tumors from the anterior portion of the ventricle without opening the foramen of Monro, tumors within the midportion are relatively inaccessible unless the foramen is enlarged. Traditionally this has been done by opening the foramen anteriorly, sacrificing the ipsilateral column of the fornix. But this may result in severe short term memory loss, especially if either the tumor or the operation compromises the contralateral fornix. The second option, which we have chosen to utilize, has been to enlarge the foramen posteriorly (4). Dandy (1) described the removal of a mid-third ventricular tumor after opening the foramen posteriorly and ligating one of the "lesser veins" of Galen. However, most neurosurgeons have hesitated to sacrifice the vein, perhaps because McKissock (5) warned of the dangers of creating venous hypertension in the ipsilateral basal ganglia and internal capsule. This is not an important risk because there is excellent collateral circulation between the superficial cortical and medullary venous systems, as well as between the posterior medullary venous system and the galenic system. Hirsch et al. (3) and Deland-sheer et al. (2) were the first to describe the microsurgical subchoroidal approach to the third ventricle and emphasized that the ipsilateral thalamostriate vein could be coagulated and divided from the internal cerebral vein without risking venous hypertension in the ipsilateral basal ganglion. Viale and Turtas (6) modified this approach to the third ventricle by operating in the "figure eight" created between the thalamus and the ipsilateral internal cerebral vein posteriorly and the thalamostriate vein and foramen of Monro anteriorly. To gain access to the mid-third ventricle using this modification, one must heavily retract the body of the fornix, which could cause severe disturbance of short term memory. Therefore, we view Viale's modifications as a hindrance rather than an improvement upon the approach of Hirsch and Delandsheer. Exposure of the lateral ventricle is attained either through a cortical incision in the second frontal gyrus or through the corpus callosum in the midline. The transcortical route has the advantage that the key landmarks in the lateral ventricle that guide the surgeon into the third ventricle are in the direct line of sight and, therefore, less intraventricular retraction is required than is necessary after the transcallosal approach. On the other hand, the approach through the corpus callosum minimizes injury to the frontal lobe and may be less epileptogenic. In patients with massive ventriculomegaly due to associated hydrocephalus, the exposure is comparable with either approach.

Indications Tumors that are primarily extraaxial and indent the floor of the third ventricle should be removed via a subfrontal rather than a transventricular approach. These tumors arise from the sellar or parasellar region and it is sometimes difficult to determine the site of origin even at operation. Pituitary tumors are the most common type in this group, although epidermoid and dermoid tumors, meningiomas, atypical teratomas, ectopic pinealomas, and metastases, as well as nonneoplastic lesions, including arachnoid cysts and sarcoidosis, have been described. Tumors that are entirely confined within the ventricle and have not broken through the floor are best handled by the subchoroidal or interforniceal approaches. These tumors are frequently attached to the wall of the ventricles by pedicles through which they receive their blood supply. Low grade astrocytomas are relatively common in children and when they arise in the hypothalamus or the optic chiasm grow into the ventricular lumen, causing obstructive hydrocephalus as well as hypothalamic dysfunction. It is frequently difficult to differentiate these gliomas from lesions originating from within the thalamus that extend into the third ventricle. Other less frequently encountered tumors include ependymomas, papillomas of the Choroid plexus, and meningiomas of the tela choroidea. Entirely intraventricular Craniopharyngiomas, which originate from Rathe's pouch, have also been encountered. Vascular anomalies, including cavernous hemangioma, and nonneoplastic cysts secondary to parasitic or tuberculous granulomatous disease round out the list of exceptionally rare causes of obstructive hydrocephalus due to intraventricular abnormalities. If the patient has hydrocephalus, the surgical exposure will be much easier to attain. On the other hand, if the patient has a functioning ventricular shunt with normal or small-sized lateral ventricles, the transcortical incision, dissection, and retraction required to gain access to the mid-third ventricle make this approach less attractive than the transcallosal or transcallosal interforniceal approaches to the third ventricle. A specific example of this relative contraindication is seen in a patient who was shunted for "aqueductal stenosis" a decade before the development of a large third ventricular glioma (Fig. 15.1). In the face of small lateral ventricles, a transcortical subchoroidal approach to the glioma was performed for biopsy. A subop-timal view of this third ventricular tumor was attained because the small noncompliant lateral ventricle prevented adequate retraction of the Choroid plexus, fornix, and thalamus. In patients recently shunted the ventricular cavity is still relatively distensible, perhaps because of periventricular cerebrospinal fluid (CSF) accumulation. This enables the surgeon to retract the fornix and basal ganglia gently without fear of damaging them. In this situation we would prefer a stereotaxic biopsy to establish the diagnosis and, if that were not possible, the transcallosal interforniceal approach would be utilized to obtain tissue for pathological examination. If the ventricular shunt catheter had been placed through the right frontal approach, we would have performed a small cortisectomy around the ventricular catheter to gain access to the small lateral ventricle and the intraventricular landmarks needed to guide the surgeon to the third ventricle for tumor biopsy.

Figure 15.1. These two contrast-enhanced CT sections show small lateral ventricles in a patient who was shunted 10 years before the scan. The right hemi-cranium is smaller than the left. The shunt catheter passes through the thalamus into the right lateral ventricle, but caused no deficits. The contrast-enhancing tumor filling the velum interpositum emanates from the right thalamus. There is no evidence of periventricular low density due to transventricular uptake of water, which suggests to the surgeon that the shunt has been functioning properly for a long time.

Preoperative Evaluation Patients harboring third ventricular tumors usually present with signs and symptoms of hydrocephalus due to tumorous obstruction of CSF pathways or of pituitary or hypothalamic dysfunction due to direct invasion. Patients are usually referred to the neurosurgeon with a computerized tomographic (CT) scan that has defined the location of the lesion. Thin coronal sections, with and without intravenous contrast enhancement, should be studied to give the surgeon a clear knowledge of the location of the tumor with respect to the confines of the third ventricle and of the tumor's vascular pattern, which may help to determine the presumed preoperative diagnosis (Fig. 15.2). One must also study the peritumorous venous drainage with a high quality cerebral angiogram, preferably with magnification (Fig. 15.3). A digital arterial angiogram will not give the surgeon optimal information about third ventricular venous drainage and therefore is not an acceptable substitute for a standard angiogram. If the Venogram shows that the tumor has spread both internal cerebral veins laterally, we would remove it without dividing the thalomostriate vein from the internal cerebral vein. In this situation division of the septal vein from the internal cerebral vein may be necessary to enlarge the operative field from the foramen of Monro to the posterior portion of the third ventricle. Conversely, if the internal cerebral veins are together and displaced dorsally by the lesion, either or both the

Figure 15.2. A. This preoperative CT scan shows a well-circumscribed lesion filling the anterior and middle portions of the third ventricle, which is surrounded by edema. B. The speckled blood or calcium density within the lesion enhanced after intravenous contrast infusion. No hydrocephalus is noted. C. The sagittal reconstruction of the contrast-enhanced CT scan shows that the lesion is entirely confined within the third ventricle. This cavernous angioma was removed utilizing a subchoroidal exposure of the third ventricle. D. These operative photographs taken during the removal of this lesion show the view of the right lateral ventricle, including the right thalamostriate, septal, and internal cerebral veins, the fornix, and the thalamus, partially obscured by the small cottonoid. E. After both septal and thalamostriate veins were coagulated and divided, the lesion within the third ventricular cavity was identified, biopsied, and vaporized with a CO2 laser. Diagnosis: cavernous angioma.

Figure 15.3. A. The lateral Venogram shows the configuration of the veins draining the lateral ventricles into the internal cerebral vein. The venous angle made by the thalamostriate and internal cerebral vein is "open" (more obtuse than normal) because of the presence of a mid-third ventricular mass elevating the course of the internal cerebral vein (ICV) and partially obstructing the CSF pathways. The presence of direct venous communications between the ICV and the thalamus is not evident on this Venogram, but the posterior portion of the ICV is partially encased in tumor. The absence of many anteriorly draining parasagittal veins would make the transcallosal approach as attractive as the transcortical approach to the right lateral ventricle. B. Large thalamic perforating arteries from the top of the basilar artery and dilated medial posterior choroidal arteries suggest that this third ventricular mass is a primary thalamic tumor.

septal and ipsilateral thalamostriate veins are divided from the internal cerebral vein along the subchoroidal route to the third ventricle. Technique In patients who are obtunded due to tumorous hydrocephalus, we first place a ventricular shunt and later return the patient to the operating room for tumor biopsy or excision when the patient's level of consciousness improves. The placement of the scalp incision is critical when inserting a shunt. We prefer a frontal placement because the scalp flap can be utilized for definitive craniotomy (Fig. 15.4). The horseshoe right frontal scalp flap is hinged toward the coronal suture so that the shunt material will not lay beneath the incision. (If the transcallosal route is elected, the scalp flap crosses the midline. The patient is placed in the

Figure 15.4. A. The burr hole is placed in the midpupillary line anterior to the right coronal suture. A horseshoe scalp flap is fashioned so that it is hinged posteriorly. B. Another way of approximating the location of the single burr hole, which is utilized for the placement of the ventricular shunt catheter or as the centrum for the craniotomy, is to mark the skin 10 cm rostral from the nasion and 3 cm to the right of the midline. C. This conceptualization shows the sequence in which the saw cuts are made with the air-powered cranial saw after the single perforation has been drilled. The last cut is along the midline to assure that venous bleeding, if it should occur, can be controlled easily. Bleeding points are not coagulated, but rather are tamponaded with pledgets of thrombin-soaked gelatin sponge.

lateral position with the vertex elevated slightly above the equatorial plane so that the medial surface of the brain falls away from the falx, obviating the need for firm retraction.) The scalp flap is fashioned so that a single burr hole placed in the center of the flap 10 cm above the nasion and 3 cm to the right of the midline can be used for placement of the

Figure 15.5. A. The dural opening. Rather than hinging the dura mater toward the midline in a horseshoe fashion, we prefer to open the dura in a cruciate way to visualize as much of the cortex as possible. B. A small cortisectomy is made after coagulating the pial vessels with a bipolar cautery. С. А 16 gauge Cone ventricular needle is passed perpendicular to the cortex. Depending upon the thickness of the cortical mantle, the tip of the needle will enter the frontal horn of the right lateral ventricle just anterior to the foramen of Monro.

shunt catheter as well as for the small free bone plate needed for the craniotomy. The last saw cut is along the midline so that potentially troublesome parasagittal venous bleeding can be controlled promptly with a thin strip of thrombin-soaked gelatin sponge (Fig. 15.4.) The dura mater is opened in a cruciate fashion along the diagonal and is reflected along each saw cut (Fig. 15.5). The small superficial cortisectomy is made in the middle frontal gyrus and a 16 gauge Cone ventricular needle is passed into the right lateral ventricle (Fig. 15.5). A core of brain surrounding the needle track is removed so that the two 20-mm Silastic-coated Greenberg retractors can be placed within the ventricle (Figs. 15.6 to 15.8). The remainder of the operation is done with the help of the microscope. We prefer the 300-mm lens and the 12.5x oculars. The first key anatomical landmark to be encountered is the Choroid plexus. The Choroid plexus is attached along the superior and medial aspect of the thalamus at the choroidal fissure. After elevating the Choroid plexus, this ependymal lining becomes apparent. By following the Choroid plexus anteriorly to the foramen of Munro, the surgeon will also identify the subependymal course of the caudate and the thalamostriate veins as they run toward the foramen of Monro (Fig. 15.9). In patients with massive hydrocephalus, decompression of the ipsilateral ventricular CSF causes the septum pellucidum to bow out toward the collapsed ventricle, partly occluding the foramen of Monro (Fig. 15.10). The septum should be fenestrated, allowing CSF from the contralateral ventricle to escape. Care should be taken to puncture the midportion of the septum at least a few millimeters away from the column of the fornix. The foramen of Monro will become apparent as the septum returns to the middle. The Choroid plexus is elevated with the tip of the retractor blade revealing the dorsal aspect of the thalamus (Fig. 15.12A). Cutting the leptomeninges rather than stripping them from the superior aspect of the thalamus allows one to elevate

Figure 15.6. The artist has sketched an overview of the operative set-up. Note that the patient's head has been rotated and the vertex has been tilted a few degrees to the left of the midsagittal plane. The retractor blades are advanced to the edge of the cortisectomy so that the surgeon can easily view into the lateral ventricle. The cortisectomy should be lengthened so that 25 to 35 mm of intraventricular exposure is obtained.

Figure 15.7. The artist's drawings (A and B) and the operative photograph (C) delineate the intraventricular anatomy. The Choroid plexus, septal vein, thalamostriate vein, thalamus, and fornix, as well as the tumor filling the foramen of Monro, are seen.

Figure 15.8. These drawings show coronal views of the approach to the midportion of the third ventricular cavity. the choroid plexus and dislocate it toward the midline. If the Choroid plexus is excessively abundant, the surgeon may elect to coagulate and remove it. The subependymal veins that drain the anterolateral aspect of the ventricle, the septal and thalamostriate veins, shed their subependymal coat and join directly into the internal cerebral vein within the leaves of arachnoid called the velum interpositum. Anteromedial thalamic and posteromedial thalamic veins join the terminal segments of the internal cerebral vein posterior to this dissection. Visualization of these and occasional anomalous direct lateral veins allows one to divide both the septal and the thalamostriate veins, if it is necessary, to gain wider exposure of the third ventricle without fear of internal cerebral vein thrombosis (Fig. 15.11). After the thalamostriate or septal vein is divided from the internal cerebral vein (Fig. 15.12), the retractors are advanced along the medial wall of the thalamus and the body of the fornix, respectively, to gain access to the third ventricle. Only leptomeningeal adhesions coursing through the velum interpositum and the thin inferior pial layer underlying the ependymal membrane prevent access to the third ventricle. The medial posterior choroidal arteries travel through the velum interpositum and supply the Choroid plexus of the third and lateral ventricles as well as the posterior medial aspect of the thalamus. The Choroid plexus of the third ventricle lies in the roof of the cavity, but in the presence of a tumor it is usually atrophic, presumably as a result of chronically increased intracranial pressure. Once the tumor is encountered, biopsy is taken with a small cup forceps (Fig. 15.1 ЗА). The tumor

Figure 15.9. These two photographs (B and C) and the accompanying drawing (A) show the ventricular anatomy in a formalin-fixed brain. The right frontal cortex has been removed. The artist has enhanced the conceptualization of the microsurgeon's operative view by superimposing the subjacent third ventricle. B. Coagulation and then division of the thalamostriate vein from the internal cerebral vein as the former leaves its subependymal course over the thalamus. C. The remaining filmy strips of arachnoid that course through the velum interpositum. Division of these strands allows the surgeon free access to the midportion of the ventricle.

Figure 15.10. Left. The artist's conceptualization of the intraventricular anatomy clearly shows the Choroid plexus and subependymal course of the caudate and thalamostriate veins, but the view of the foramen of Monro has been obstructed by a ballooned septum pellucidum under pressure due to CSF in the isolated left lateral ventricle. The surgeon perforates the septum so that CSF under tension can drain into the operative field. Right. Aspiration of this CSF will allow the septum to drift out of the way, enabling the surgeon to identify the foramen of Monro and the confluence of the septal and thalamostriate veins. The retractors gently depress the thalamus and elevate the body of the fornix from right to left. Forceful retraction is not necessary. The tip of medial retractor is bent slightly so that the column of the fornix can be gently elevated as well as dislocated medially. Taking down the filmy strips of arachnoid in the velum interpositum may be necessary before one enters the third ventricular cavity. Sometimes the tumor spreads apart both internal cerebral veins, and taking the septal vein from the ipsilateral internal cerebral vein will allow one to retract the thalamus laterally a few millimeters, giving the surgeon an excellent view of the tumor cavity.

Figure 15.11. The communications between the deep thalamic veins and the internal cerebral vein. The presence of the middle and posterior venous communications enables the surgeon to coagulate the septal or thalamostriate vein without concern about the development of regional venous hypertension.

Figure 15.12. A tumor being removed. A and D. The thalamostriate vein is divided from the internal vein, and the leptomeninges in the velum interpositum are taken down. В and E. A flat dissector is passed between the thalamus and the internal cerebral vein, and the Silasticcoated retractors are advanced so that the medial retractor slightly dislocates the fornix and the lateral retractor depresses the thalamus. The tumor can then be attacked between the internal cerebral vein and the thalamus with standard microsurgical instruments, the laser, or the cavatron ultrasonic emulsifier. С and F. After the tumor is removed one should be able to see the contralateral foramen of Monro and both columns of the fornices.

Figure 15.13. The artist has shown the three usual tools for removing the tumor: A, microvascular biopsy forceps (for the avascular tumor); B, carbon dioxide laser (for the scirrhous tumor); and C, Cavitron aspirator (for the gelatinous tumor).

vascularity and consistency is assessed during this maneuver. Firm avascular tumors are vaporized with the carbon dioxide laser (Fig. 15.13B) and soft tumors are removed with suction (Fig. 15.13C) or the Cavitron ultrasonic emulsifier. Hemostasis is usually not a problem, and it is managed with bipolar cautery or cottonoid pledgets soaked in throm-bin. By gutting the centrum of the tumor, the surgeon identifies the plane between the tumor and the walls of the surrounding thalamus and hypothalamus. When possible, as in the case of intraventricular craniopharyngioma or low grade glioma, the lesion is completely removed. Virtually all of these patients are left with obstruction of the CSF pathways and, for this reason, a flanged ventricular catheter is left in the lateral ventricle when the craniotomy is closed. The tube is passed through the single burr hole made in the center of the free bone flap and is attached to an Ommaya reservoir (Fig. 15.14). We usually elect not to convert the Ommaya reservoir to a functioning shunt until the patient demonstrates clinical evidence of persistent obstructive hydrocephalus.

Figure 15.14. A. The relationship of the center-filling Ommaya reservoir to the scalp flap. At a later date the ventriculostomy can be converted to a shunt if communicating hydrocephalus ensues. B. The tip of a ventricular catheter in the frontal bow of the right lateral ventricle. The tip of the catheter should not rest near the coagulated deep cerebral veins.

When aspirating CSF through the catheter, it is possible to injure the subjacent deep cerebral veins stripped of their ependymal cover. To prevent this complication, it is important to leave just enough catheter to pierce the most superior aspect of the lateral ventricle. Complications Complications associated with this procedure have been few. Although the transcortical approach to the ventricle has been associated with postoperative seizures, the use of perioperative anticonvulsant medications has essentially eliminated this problem. An additional problem could be related to heavy retraction on the thalamus, caudate, or body of the fornix. However, this potential problem is essentially eliminated by coagulation and division of the thalamostriate vein from the internal cerebral vein, which gives the surgeon an excellent view in the third ventricle from the level of the foramen of Monro to its most posterior portion. This expanse will allow one to operate safely on anterior, middle, and posterior third ventricular lesions without injuring the columns of the fornix. Whether interruption of one or both columns of the fornix will result in permanent short term memory loss is controversial. We think that there has been enough unfavorable experience with unilateral anterior fornicotomy to urge avoidance of this maneuver, especially when the contralateral fornix could be compromised by the third ventricular tumor. References 1. Dandy WE: Benign Tumors in the Third Ventricle of the Brain: Diagnosis and Treatment. Springfield, IL, Charles С Thomas, 1933. 2. Delandsheer JM, Guyot JF, Jomin M, Sherpereel B, Laine E: Acces au triosieme ventricle par voie inter-thalamo-trigonale. Neurochirurgie 24:419421, 1978. 3. Hirsch JF, Zouaoui A, Renien D, Pierre-Kahn A: A new surgical approach to the third ventricle with interruption of the striothalamic vein. Acta Neurochir (Wien) 47:135-147, 1979. 4. Lavyne MH, Patterson RH, Jr: Subchoroidal trans-velum interpositum ap proach to mid-third ventricular tumors. Neurosurgery 12:86-93, 1983. 5. McKissock W: The surgical treatment of the colloid of the third ventricle: A report based upon twenty-one personal cases. Brain 74:1-9, 1951. 6. Viale GL, Turtas S: The subchoroidal approach to the third ventricle. Surg Neurol 14:71-76, 1980.

16 Subfrontal Transsphenoidal and TransLamina Terminalis Approaches RusselH. Patterson, Jr., M.D.

Many surgeons favor a pterional approach along the sphenoid ridge to tumors involving the anterior third ventricle, such as Craniopharyngiomas. The advantages of this approach are several. It is the shortest distance from the scalp to the sella, it provides a glimpse of the tumor both behind and in front of the carotid artery, and it avoids opening the frontal sinuses. However, the pterional approach has substantial disadvantages; the opposite optic nerve and carotid artery are poorly visualized, tumor removal is difficult because one must operate with the carotid arteries and optic nerves obstructing access, and the various arteries and nerves are positioned at a peculiar angle that is difficult to conceptualize without a substantial amount of experience. For these reasons, I prefer a straight frontal approach. A frontal approach has some drawbacks but, fortunately, these are minor or can be readily overcome. The surgeon may have to open the frontal and sphenoid sinuses, anosmia results in 30 to 50% of patients, and short optic nerves may block access to the tumor. On the other hand, exposure of the tumor is better than by any other approach with the consequence that the chances of damaging the optic nerves and major arteries are minimized. Surgical Technique The patient is positioned supine with two #20 spinal needles in the lumbar subarachnoid space for the withdrawal of spinal fluid. We have had constructed for the operating table a thick mattress with a notch that provides sufficient clearance for the lumbar needles (Fig. 16.1 A). At the appropriate time, spinal fluid is allowed to flow by gravity through plastic tubes into a sterile 50-ml syringe with the plunger removed and taped underneath the table. By means of a three-way stop cock on the syringe, spinal fluid in excess of 50 ml can be removed from this reservoir for injection back into the patient at the end of the operation. A frame that holds a Mayo tray clamps to the operating table. This is placed

Figure 16.1. A. Patient supine with table tilted head up about 20° and the lumbar spinal needle in place. The neck is extended so that the floor of the frontal fossa is tipped back about 45° from the vertical, which will reduce the need for frontal lobe retraction. B. The head is not rotated, which makes orientation around the sella easier (I, surgeon; 2, assistant; 3, nurse; 4, anesthesiologist; 5, microscope base).

Figure 16.2. A. The neck is extended so that the roof of the orbits is tipped back about 45° from the vertical. B. Position of Mayfield pin head holder. approximately 20 cm caudal to the patient's brow, and an extra outrigger provides an additional surface on which to place instruments. An ordinary back table as used in general surgery provides the reserve of instruments for the nurse, who stands on a platform near the first assistant. The nurse is close enough that she can provide an extra set of hands for irrigation or manipulation of the electrocautery, if necessary (Fig. 16.1B). The operating table is tipped with the head elevated and the feet down. The neck is extended so that the roof of the orbits, were they visible, would be tipped back approximately 45° from the vertical (Fig. 16.2A). This means that the frontal lobe will tend to fall back by gravity from the roof of the orbit, which minimizes the amount of retraction necessary to expose the region of the optic chiasm. The position probably will require the surgeon to use straight eyepieces on the operating microscope, which is most conveniently done by having the surgeon seated using a Mayo stand as an armrest (Fig. 16.1 A). It is also possible to place the head in a neutral position with the roof of the orbit and the undersurface of the frontal lobe positioned vertically, which is convenient for the surgeon because he can use angled eyepieces and operate with his line of vision straight down. However, in this position the frontal lobe tends to fall over the optic chiasm with the consequence that retraction pressure on the frontal lobe must be greater. Consequently, we prefer the approach with the head thrown back and neck extended. The head is held with the Mayfield pin head holder. This is not ideal for a coronal incision because invariably the pin position compromises the incision to a degree. We usually put the double pins in the left posterior frontal region and the single pin behind the right ear. It is then possible to drape for a coronal incision on the right side that extends from near the zygoma to a point sufficiently past the midline to allow a bifrontal incision if desired (Fig. 16.2B). After the patient is positioned, the hair is shaved for about 4 cm behind the hairline. If the tumor is small and the surgeon elects a unilateral

Figure 16.3. A and B. The coronal scalp incision is placed in the hair line, which varies from patient to patient. The bone plate reaches the midline and is flush with the orbital roof. C. For complex tumors, such as Craniopharyngiomas, a bifrontal bone plate is used. It can be carried to the zygoma insertion on either side, although usually 4 cm on either side of the midline is adequate. approach, the scalp incision is made from near the zygoma in front of the ear in the hairline across to the left side of the midline. The incision, particularly if it is as low as the zygoma, must be kept within 1 cm of the helix of the ear to avoid dividing the frontalis branch of the facial nerve. After making the incision, a soft tissue flap consisting of scalp and temporal muscle is turned down toward the brow. Because the bone opening must be made flush with the brow, it is important that the flap

Figure 16.4. В. Scalp and temporal muscle turned down toward the brow. В, С, and D. The dura mater is opened transversely, dividing the sagittal sinus and falx in the case of a bifrontal exposure. be turned down far enough to expose the supraorbital ridge and the insertion of the zygoma into the frontal bone at the keyhole. Small tumors can be handled readily through a right frontal exposure alone, but larger tumors are better exposed through a bifrontal craniotomy. Whether the exposure is unilateral or bilateral, we prefer to make the first hole a trephine opening centered on the midline just at the brow. The trephine makes for an easier passage through the frontal sinus than using a high speed burr (Fig. 16.3). Most often the frontal sinus will be entered in the process, but sometimes it is possible to preserve the mucosa by pushing it downward with an instrument before the trephine button

Figure 16.5. A. The retractor is put in obliquely along the roof of the orbit toward the optic chiasm. B. A more direct exposure can be gained by sliding the retractor medially to the falx and retracting the gyrus rectus. The danger is that the increased retraction pressure may damage the frontal lobe. С The tumor is exposed. The skull opening must be flush with the floor of the frontal fossa to minimize retraction. Usually the frontal sinus is opened widely. is removed. An air drill may be used to place the other holes in the skull. One of them should be at the keyhole on the right at the zygomatic process of the frontal bone, and others may be necessary in older patients in whom the dura mater is likely to be adherent to the intertable of the skull. In case of a bifrontal exposure, one burr hole is placed over the sagittal sinus approximately 4 cm posterior to the trephine opening and

Figure 16.6. A to C. Sometimes the brain is full and the chiasm is difficult to expose. In a unilateral approach, splitting the sylvian fissure enough to expose A1 and M1 will allow more retraction. D. In a bifrontal exposure, the olfactory tracts are dissected and preserved so that olfaction will not be lost. The frontal lobes are then separated and retracted. others are placed as needed in the skull to the left of the midline. The flap need not be large; 4 cm from the roof of the orbit is enough (Fig. 16.4A). It is possible to feel the roof of the orbit with an instrument from the medial trephine opening and also the lateral opening in the skull at the keyhole. In this way, it is possible to use the craniotome to make an opening flush with the roof of the orbit. The frontal sinus will be opened in most patients, but this is easy to repair at the end of the procedure. After the bone plate is removed, we request the anesthesiologist to open the spinal drainage system and begin the removal of spinal fluid. Perhaps

Figure 16.7. Preservation of the olfactory tracts during bifrontal craniotomy. The sense of smell is retained. 30 to 60 ml of spinal fluid are removed at first, and during the course of the procedure additional fluid is removed in quantities sufficient to keep spinal fluid from welling up around the optic chiasm. It is probably wise not to remove all of the spinal fluid so that blood spilled at the operative site will be less likely to fill the subarachnoid space. Brain relaxation is achieved by the position of the head, the withdrawal of spinal fluid, the administration of 50 g of urea at the beginning of the procedure, and adjustment of ventilation to keep the pCO2 in the range of 28 to 30 mm Hg. The dura mater is opened starting at the keyhole and curving across toward the trephine opening (Fig. 16.4C). The dural opening is started laterally to avoid any bridging veins draining from the cerebrum into the sagittal sinus. In the case of a bifrontal exposure, the sagittal sinus and the falx are divided just above the glabella. The microscope is then brought into the field and the surgeon sits using a draped Mayo stand as an arm support and also as a convenient place to put commonly used instruments (Fig. 16.1 A). A self-retaining brain retractor that attaches to the pin head holder has proved quite satisfactory (Fig. 16.5A). The older retractors that attached to the skull opening take up so much room that they are not practical with the small opening that we favor. For the unilateral approach (Fig. 16.5B), retraction of the brain is begun laterally over the roof of the orbit, and the olfactory tract is followed down to the region of the optic chiasm (Fig. 16.6C). The arachnoid around the optic chiasm and optic nerves is divided, and we strive to expose both optic nerves as well as the optic chiasm (Fig. 16.6D). If it is difficult to

Figure 16.8. If the optic nerves are long, it may be possible to extract the tumor from between the nerves. A layer of arachnoid usually separates a craniopharyngioma from the arteries, which makes separation relatively easy. However, the optic chiasm, hypothalamus, and pituitary stalk are often quite adherent to the tumor and difficult to separate. get adequate exposure of the chiasm, which will be the case if the optic nerves are long, it is often helpful to divide the arachnoid in the sylvian fissure sufficiently to expose the middle cerebral artery (Fig. 16.6C). The resulting separation of the frontal lobe and the temporal lobe allows better exposure of the chiasm. Gradually, adequate exposure of the optic nerves and chiasm can be obtained. Usually it is wise to coagulate and divide the olfactory nerve and try to have the retractor along or near the falx because of the advantages of a straighter approach to the optic chiasm. If the tumor is large, and often this can be anticipated from the preoperative roentgenographic studies, then a bifrontal exposure is ad-

Figure 16.9. Comparison of tumor removal with long or short optic nerves. vantageous. In this case a bone plate can be fashioned on the left side that is similar to the one on the right. However, it is often unnecessary to go as far as the keyhole on the left side, and the sawcut can be stopped perhaps 4 cm to the left of the midline. However, the sawcut still must be made flush with the roof of the left orbit to minimize retraction of the frontal lobe (Fig. 16.4B). After removal of the bone plate on the left, the dura mater is opened on the left side and the sagittal sinus and falx are divided at the glabella. It may be possible then to retract the frontal lobe and dissect the olfactory tracts from the undersurface of the frontal lobe, preserving them. This may be difficult, and often both olfactory tracts will be stretched in the process, leading to permanent anosmia. Brain retraction then is not only posteriorly, but the frontal lobes are separated. Callosal branches of the anterior cerebral artery are identified and followed to the region of the anterior communicating artery. This will allow exposure of the entire optic chiasm, even if the optic nerves are long, and it gives excellent exposure of the lamina terminalis (Fig. 16.6D). In a bifrontal craniotomy, ability to preserve the olfactory tracts is improved (Fig. 16.7). In cases in which the optic nerves are long, the craniopharyngioma usually will be identified between the nerves above the sella (Fig. 16.8A). The tumor is covered by a layer of arachnoid, and it is often possible to extract the tumor by a steady pull on the capsule and dissection of arachnoid off the sides. The cystic tumors provide the most difficulty because the cyst wall is quite delicate. It is easy to leave fragments of the cyst wall inadvertently behind, and because the cyst wall is composed of neoplastic tissue the pieces will continue to grow. Often the pituitary stalk, which can be identified by the vertical striations of the portal system, can be identified and preserved (Fig. 16.6D). In many patients, the optic nerves will be quite short and only a millimeter may exist between the chiasm and the tuberculum (Fig. 16.9B). This space is not sufficient for resection of the tumor (Fig. 16.9 A).

Figure 16.10. A. If the optic nerves are short and the tumor is large and hard to break into small pieces, the tuberculum sellae should be removed to protect the optic apparatus from injury. B. A flap of dura mater is peeled back from the tuberculum and the sphenoid sinus is entered with a chisel. Often the sinus mucosa can be preserved. С to E. The tuberculum and anterior wall of the sella are removed with bone instruments or a high speed burr. At this point, one can make a hole with a blunt instrument in the lamina terminalis and examine the tumor from above. The lamina is sometimes difficult to distinguish from the optic tracts and chiasm, but it has a faint reddish blush that identifies it. This is the place to make the hole. If the tumor is quite friable, it is sometimes possible to remove a good deal of tumor from between the optic tracts. However, it may be that the tumor under the chiasm is difficult to reach or that parts of the tumor will be so firm that manipulations to remove the tumor risk damage to the optic tract. If this is the case, we recommend that the tuberculum sellae be removed. This is easy to accomplish by cutting a flap of dura mater over the tuberculum and hinging it either down toward the chiasm or up

Figure 16.11. Removal of the tuberculum sellae (cadaver).

Figure 16.12. Craniopharyngioma with short optic nerves. toward the crista galli. A chisel then can be used to make a hole through the planum sphenoidale into the sphenoid sinus. The sphenoid sinus mucosa is quite thick and often can be preserved. Fine bone instruments are then used to remove the bone in front of the sella between the optic nerves (Fig. 16.10). Another option is to open the sphenoid sinus and remove the bone at the tuberculum with a high speed drill (Fig. 16.11). In either case, the whole process takes no more than 15 minutes. The dura mater enclosing the circular sinus then is coagulated and divided. A craniopharyngioma with short optic nerves is shown in Figure 16.12.

Figure 16.13. A. The dura mater is divided and pushed out of the way. В and C. Tumor is removed by pushing it down and away from the chiasm into the sphenoid sinus from which it can be extracted without brushing the optic apparatus. With the tumor exposed through the lamina terminalis and the tuberculum removed, it is possible to push chunks of tumor downward into the sphenoid sinus where they can be removed without fear of damaging the optic apparatus (Fig. 16.13B). Problems and Complications The main difficulty with the procedure occurs with large tumors that are adherent to the underside of the chiasm or to the hypothalamus. As a rule, steady traction and gentle dissection will free the tumor from both of these structures, but it is to these structures that the tumor is most adherent. Restraint must be exercised in separating adherent tumor from the optic chiasm. If the chiasm is stretched or vision is already impaired, small amounts of manipulation can cause severe visual loss. Occasionally, prudence will dictate leaving tumor behind in preference to causing blindness. Sometimes the tumor adheres to the carotid, posterior communicating, or basilar arteries, but this is exceptional. For the most part

Figure 16.14. Appearance after the lamina terminalis is opened, the tuberculum sellae is removed, and the craniopharyngioma is excised. the tumor is insulated from the arteries by a layer of arachnoid, and separation is easy. After tumor removal is completed, the hole in the sphenoid sinus is repaired by pushing the sphenoid sinus mucosa down and stuffing the cavity of the sinus with fat (Fig. 16.14). Fat obtained from a patient undergoing an operation in another room has been quite satisfactory (Fig. 16.15B). The fat is stuffed into the sinus cavity and held in place by a few sutures placed across the top of the fat between remaining fragments of dura mater. Previously withdrawn spinal fluid, if it is not blood-stained, is returned through the lumbar needles. This maneuver flushes out air and refills the brain. We try to close the dura watertight, if possible, and then use more fat to seal the open frontal sinuses. A flap of pericranium is cut from the underside of the scalp flap and sutured to the dura near the cut edge of the opening in the skull (Fig. 16.15C). If the dura mater is shredded, we do not patch it but place a sheet of gelatin sponge under the bone plate. The bone plate is repositioned and held with sutures passed through drill holes. The trephine button is replaced and wedged into position with some gelatin foam. Bone dust, silicone rubber glue, or acrylic cranioplasty material is used to fill the burr holes. The scalp is closed in two layers without drainage. The complications of the surgical procedure are few if retraction on the frontal lobe has been gentle. Anosmia occurs in about 30% of the patients for whom a unilateral approach has been used and at least 50% of the patients for whom a bifrontal approach has been used, even if the olfactory tracts are saved. Hypopituitarism is common, but often this is present before operation. Diabetes insipidus is a likely consequence, but this can be well managed with pitressin and desmopressin acetate and tends to self-correct. The worst endocrine complication is diabetes insipidus occurring in a patient with adypsia. Because the patient has inadequate thirst mechanisms, the fluids lost due to diabetes insipidus are not adequately replaced by oral fluids. This leads to dehydration with high serum osmolalities and hypernatremia. We have seen this primarily in patients who have had prior radiation therapy. It has been rare in patients undergoing operation for the first time or in patients whose prior therapy has been only operation.

Figure 16.15. A and B. The sphenoid sinus is closed by packing it with fat and tacking back the dural flap with a few sutures. С Fat is placed in the frontal sinus and, after the dura is closed, a flap of pericranium is placed over the fat and tacked to the dura mater. D. The bone plate is replaced and the holes and saw cut are filled with something like bone dust or silicone rubber.

17 Bifrontal Anterior Interhemispheric Approach Jiro Suzuki, M.D.

The first attempt to remove a tumor from the anterior part of the third ventricle was that of Lewis (9) in 1910, operating on a craniopharyngioma. After the total excision of a craniopharyngioma by Gordy et al. (4) in 1949, total removal of such lesions soon became commonplace throughout the world, although the operative mortality rate was high at 25 to 57% (4, 5, 10, 11). Subsequent to the advent of steroid replacement therapy in 1952 (7) and, more recently, microsurgery, the operative mortality has decreased, but several problems have remained. Moreover, because many tumors of the anterior part of the third ventricle are imbedded in or surround the hypothalamopituitary system, complete excision is not always the best therapy. In light of findings concerning the nature of the tumor, its location, the patient's age, etc., it is desirable to attempt to assure maximal prolongation of the patient's useful survival time and to take measures that will minimize the possibility of recurrence (1,2). For these purposes, therapy— including radiotherapy and chemotherapy—must be considered. When, however, operation is to be performed, operative findings often influence the decision of whether complete excision of the tumor is desirable. Shillito's caution (12) to the neurosurgeon should be kept in mind: "Mature judgment is needed to determine how far to go and when to stop." It is therefore essential to obtain an operative field of maximal size and to use an approach that minimizes damage to brain tissue. We have consequently devised a bifrontal interhemispheric trans-lamina terminalis approach (Fig. 17.1) for mass lesions, such as Craniopharyngiomas, that are known to raise the floor of the third ventricle and develop near the suprasellar region or to develop within the third ventricle itself. Using this approach, we have obtained favorable operative results (15). Thus far, at least six approaches to the anterior third ventricle have been attempted. They include (a) a subfrontal approach, (b) an interhemispheric trans-lamina terminalis approach, (c) a pterional approach, (d) a subtemporal approach, (e) a transcallosal approach, and (/) a transsphenoidal approach. For complete excision of a tumor, however, it is best to use an approach that minimizes damage to the brain tissue and

Figure 17.1. Interhemispheric trans-lamina terminalis approach (bifrontal craniotomy).

cranial nerves and allows sufficient treatment of arterial hemorrhage during the surgical operation. Using the bifrontal interhemispheric trans-lamina terminalis approach, both the anterior circle of Willis and the optic chiasm can be visualized and tumors within the third ventricle itself can be excised by first making a small incision in the lamina terminalis—which has often been widened and thinned by the tumor growth. Moreover, favorable surgical results can normally be obtained using this approach because the amount of damage to the brain and vessels is kept to a minimum. Needless to say, optimal results will be achieved using this approach when operation is undertaken with careful consideration of the anatomy, physiology, and pathology of the pituitary and hypothalamic regions (Figs. 17.2 and 17.3). Problems that arise during operations in this region are often due to the fact that, even if the tumor is benign, such as craniopharyngioma, it may invade the brain substance around the tumor itself. Moreover, a dense gliosis will normally surround the tumor. With regard to the question of the tissue surrounding a craniopharyngioma, Sweet (16) and Hoffman et al. (6) have argued that complete excision is possible due to the presence of a layer of gliosis, which provides a margin of safety between the tumor and the functioning neruons. In contrast, Ghatak et al. (3) and Shillito (13) have held that it is hard to perform complete excision of the tumor without damaging neural tissue when fingers of tumor invade the surroundings.

Figure 17.2. A sagittal section through the hypothalamus. A. Preoptic area. B. Lateral hypothalamic area. C. Paraventricular nucleus. D. Anterior hypothalamic nucleus. E. Supraoptic nucleus. F. Dorsomedial nucleus. G. Ventromedial nucleus. H. Arcuate nucleus. I. Posterior hypothalamic nucleus. J. Mamillary nucleus. In an autopsy study of six cases of craniopharyngioma, Kuboda et al. (8) found that the thickness of the reactive gliosis was 0 to 2 mm and the distance from the tumor to the hypothalamic nuclei was 0 to 3 mm (averaging 2 mm). Moreover, the vascular distribution around the tumor generally was abundantly concentrated within the gliosis, rather than within the tumor itself. Consequently, they concluded that dissection in cases of strong adhesion or in cases likely to hemorrhage is, in effect, inviting severe sequelae or a fatal outcome, an opinion with which we concur. In pediatric cases, even if damage to small arteries is incurred during excision, collateral circulation will develop and the plasticity of the brain is high. In contrast, in adult cases, there is little neovascularization and plasticity is low. It is thought that recovery from various sequelae is likely in pediatric cases, even after complete excision, whereas it is likely that among adults there will be some whose sequelae are severe and permanent. We therefore believe that the dissection and excision of tumors should be dealt with in completely different ways for pediatric and adult cases.

Figure 17.3. Drawings of coronal sections through portions of the human hypothalamus.

Figure 17.4. Preoperative CT scans. A. Plain CT scan demonstrates the cystic tumor containing a solid mass within the third ventricle. B. Enhanced CT scan, showing a cystic tumor with a strongly enhanced mass after the injection of contrast medium.

Figure 17.5. T1 image of MRI scan. The CSF and the cystic contents appear white (increased MRI signal). The solid part of the tumor is shown as regions of decreased image intensity. A. Coronal scan. B. Sagittal scan. Details of our trans-lamina terminalis approach are described within the context of reporting a typical case of a tumor of the third ventricle. The patient was a 24-year-old man. He had experienced fatigue and a tendency toward somnolence for about 2 years and reported having occasionally fallen asleep in the midst of important business meetings. He was admitted to the hospital with the complaint of headache. Both computerized tomographic (CT) (Fig. 17.4) and magnetic resonance imaging (MRI) (Fig. 17.5) scans showed a solid mass with a cyst extending from the suprasellar region to within the third ventricle. Angiograms did not reveal any notable abnormalities, such as elevation of the A1 segment. Operation was performed under a preoperative diagnosis of craniopharyngioma developing mainly within the third ventricle. Surgical Technique This section gives details of the surgical removal of the tumor in the anterior part of the third ventricle by means of a bifrontal craniotomy and an interhemispheric translamina terminalis approach. The patient is placed in the supine position, and the head is slightly extended with the nose pointing straight up (Fig. 17.6). The skin incision is made a few millimeters behind the frontal hairline and is extended inferiorly on both sides, without severing the temporal branch of the facial nerve (Fig. 17.7). The skin flap is then folded anteriorly over the periosteum and dissection is continued as far as the upper edge of the

Figure 17.6. Position of head.

Figure 17.7. Bifrontal scalp incision behind the hair line. orbits. Two burr holes are drilled directly over the midline—one at the middle of the orbital ridge and the other 5 cm superiorly. Two further burr holes are then made bilaterally at the junction of the orbital ridge, the zygomatic process of the frontal bone, and the linea temporalis (Fig. 17.8). Because increased adhesion of the bone to dura mater is sometimes found in elderly patients, two additional burr holes can be made on both sides of the upper midline burr hole, taking suitable precautions against dural adhesion. The craniotomy is begun once the dura mater has been sufficiently separated from the inner table of the skull. Because the frontal sinus is opened in approximately 80% of our cases (Fig. 17.9), the following procedure is also required. First, a nonstimulating antiseptic is

Figure 17.8. Sites of burr holes.

Figure 17.9. The frontal sinus is opened after bifrontal craniotomy. applied to all exposed regions, including the bone flap, and treatment of the frontal sinus is begun. While the mucous membrane within the sinus is cauterized, it is stripped from the bone and pressed toward the channels communicating with the nasal passage. Next, the internal bone lamina of the sinus is rongeured away, and the dead space in the sinus is reduced. Particularly the internal lamina of the lateral portions of the sinus must be thoroughly scraped. Because of inflammation of the mucous membrane, it may become thickened so that excess tissue will need to be excised. For the elimination of any dead space, bone chips or bone dust obtained during craniotomy can be packed into small recesses and bone wax can be applied to provide a level surface without any dead space (Fig. 17.10). Antiseptics are then applied to this surface and hands are washed. The development of a postoperative epidural hematoma is prevented by suturing the dura mater to the pericranial tissues at six sites anteriorly and four sites posteriorly. Opening of the dura mater is done as far anteriorly as possible along the edge of the orbita in a configuration that resembles a double-U or the number 3 (Fig. 17.11).

Figure 17.10. Closure of the opened frontal sinus. The cavity is filled with bone dust. A small amount of bone wax is smeared over the sinus opening.

Figure 17.11. Opening of the dura mater and division of the anterior part of the superior sagittal sinus and falx. A ventricular drainage tube is inserted into the frontal horn of the lateral ventricle. The further anteriorly that the superior sagittal sinus is divided the better, because this sinus becomes narrower anteriorly and damage to the bridging cortical veins just beneath the dura mater can then be avoided. This precaution allows one to preserve the bridging cortical veins as far as possible. In incision of the dura mater, it is important when approaching the midline region to inspect the subdural space carefully to identify the presence of bridging veins there. If this precuation is not taken, control of various points of hemorrhage from severed bridging veins may become difficult. When such hemorrhage is not easily controlled, the head can be elevated to decrease the venous pressure. Care should be taken, however, not to elevate the head excessively because the danger of air embolism is increased correspondingly. Bringing a halt to hemorrhage from the superior sagittal sinus is safely and easily done by means of cauterization. Only infrequently is the sinus

Figure 17.12. Dissection of the right olfactory bulb.

Figure 17.13. Dissection of the left olfactory bulb. so wide that it requires closure by suturing. Moreover, because there is a danger of the suture slipping off during or after the operation, thereby producing a massive subdural hemorrhage, cauterization is recommended. Furthermore, if the falx is also shortened by means of cauterization, the obstruction of the operative field is reduced. By rongeuring the ridge of the bone 3 cm from the midline at the right-sided craniotomy, a small incision in the dura mater beneath can be made and a ventricular catheter 3 m in diameter can be placed within the lateral ventricle after the frontal horn has been punctured with a Dandy cannula. The catheter is then immobilized in the subcutaneous tissue and led on top of the scalp and immobilized there. Cerebrospinal fluid (CSF) can then be drained

Figure 17.14. Dissection of the right olfactory tract.

Figure 17.15. Dissection of the left olfactory tract. through the tube, thereby reducing the total brain volume and producing a small gap between the dura mater and the brain itself. The surgical wound is kept moist with wet gauze and not allowed to dry out, and the entire brain surface is protected with thin cotton pledgets. The operation then proceeds toward the olfactory bulbs. The region of the bulb is easily approached by elevating the head a few degrees and viewing along the orbital surface of the brain. The arachnoid membrane and small arterioles and venules adhere to the olfactory bulbs, making it impossible to visualize the bulb directly. By cauterizing those small vessels, one can see the bulb itself. This process is carried out by alternating between the two sides until both bulbs are visible (Figs. 17.12 and 17.13).

Figure 17.16. Completion of dissection of both olfactory tracts.

Figure 17.17. Blunt dissection of the interhemispheric fissure between the frontal lobes. The head, which had been previously elevated, is now lowered and the dissection of the olfactory tracts, alternating between the left and the right, is greatly facilitated (Figs. 17.14 and 17.15). If a unilateral olfactory tract is completely dissected before the start of dissection of the other, there is the danger of avulsion of the contralateral olfactory tract. We therefore believe that it is important to dissect bluntly the olfactory tracts in parallel—proceeding by alternating between the left and the right sides. In the dissection of these tracts, pressure should not be applied inferiorly, but rather in a superior direction. Even when adhering by only a single fine fiber, inadvertent traction during the operation can lead to

Figure 17.18. The interhemispheric dissection should proceed in an "overhanging" manner.

Figure 17.19. Exposure of the A2 segment interhemispherically. the complete freeing of the olfactory tract and loss of olfaction, so that no region of adherence should be overlooked. By further dissection the trigonal area of the olfactory tract will be reached and from there it is safe to proceed within the substance of the frontal lobe. Sufficient dissection of the surrounding arachnoid membrane should then be undertaken without applying pressure to the olfactory tract. By proceeding in this manner, alternating dissection of the left and right olfactory tracts, it is unlikely that either tract will be damaged or that they will be a hindrance during operation. The olfactory tract should not be allowed to become dry during the operation, and care should be taken to prevent the adherence of moistened cotton pledgets. Moreover, uneven pressure on the pledgets as they are placed can lead to severance of the olfactory tract and the nerves can

Figure 17.20. Complete dissection of both A2 segments.

Figure 17.21. Exposure of the AComA and branches of the A2 segments. be damaged due to heat conduction as a result of the coagulation of vessels in this region. At the completion of dissection of the olfactory tracts, a small portion of both optic nerves should be visible beneath the arachnoid (Fig. 17.16). Next, the position of the head is further lowered, and the left and right frontal lobes are bluntly dissected (Fig. 17.17). A key point at this time is that the interhemispheric dissection should be done by proceeding forward in an "overhanging" position (Fig. 17.18). In other words, the dissection is done such that the distal A2 segment of the anterior cerebral artery (АСА) will appear (Fig. 17.19). After both A2 segments beneath the

Figure 17.22. Complete dissection of the small perforating arteries from A1 and A2 segments or from the AComA.

Figure 17.23. If necessary for a wider operative field, the AComA and chiasm can be divided. genu of the corpus callosum have been exposed, dissection proceeds toward the anterior communicating artery (AComA)—thus allowing complete dissection of the frontal lobes with a minimum of damage to brain tissue (Fig. 17.20). Branches of the A2 segment can then be appreciated, but they must not be sacrificed (Fig. 17.21). Sometimes it is unclear which vessel is the A2 segment on the left and right and, in such a case, it is important to keep note of which and how many vessels on the left and on the right have been found. Hemostasis of venous hemorrhage can usually be achieved simply by means of applying a cotton pledget to the bleeding vessel. Often the A2 segment on both sides will adhere at several locations. The A2 segments should be completely separated to left and right and the course of the

Figure 17.24. The groove left in the chiasmatic region by compression can be seen after complete dissection of both A1 segments and the AComA from the chiasm and optic nerves. The tumor is visible through the lamina terminalis.

Figure 17.25. Dissection of the left internal carotid artery. perforating arteries leaving from the A1 and A2 segments or from the AComA will become evident (Fig. 17.22). Aspiration of the contents of any cysts will greatly facilitate the dissection of those arteries from the anterior wall of the third ventricle and separation of the perforating arteries to the left and right will allow visualization of the tumor beneath the lamina terminalis. In cases where a larger operative field is required and where the AComA is large enough that two vascular clips can be

Figure 17.26. Dissection of the right internal carotid artery. Bulging of the lamina terminalis has disappeared because the cystic contents have already been aspirated.

Figure 17.27. The hypophysial stalk can be identified posterior from the chiasm. applied, the AComA is divided and both the A1 and A2 groups of perforating arteries are separated to the left and right, respectively. This procedure will provide a wide operative field. However, the clips on the AComA might detach during operation, leading to various difficulties, so this procedure should not be used unless a wider operative field is required. When the color of the optic chiasm has been altered because of compression from below or tumor cell infiltration—and thus the postoperative recovery of chiasma functions is thought to be unlikely—

sectioning of the midline of the chiasm will provide the widest possible operative field (Fig. 17.23). In such cases, needless to say, bitemporal hemianopsia will remain postoperatively. When using the bifrontal interhemispheric trans-lamina terminalis approach, the worst possible circumstance is damage to arterial vessels during the operation. Regardless of how hemostasis is ultimately achieved, severe brain damage is likely to have been incurred. For this reason, the surrounding blood vessels must be identified and dissected from the chiasm and optic nerves until they hang completely free from one another. In Figure 17.24, a groove can be seen alongside the chiasm and bilateral optic nerves; these blood vessel grooves in the chiasmatic region can be seen after complete dissection of both A1 segments and the AComA. A tumor within the third ventricle can be seen to bulge the lamina terminalis forward. Dissection proceeds along both A1 segments as far as both internal carotid arteries (ICAs) (Figs. 17.25 and 17.26). Because the cyst has been aspirated, swelling of the lamina terminalis has disappeared (Fig. 17.26). In this particular case, there was no tumor within the sella and the pituitary stalk can be seen clearly posterior from the chiasm (Fig. 17.27). After the blood vessels in this region have been displaced posteriorly, a small incision in the lamina terminalis, just anterior of the AComA, is made (Fig. 17.28). In most cases of craniopharyngioma, the tumor elevates the floor of the third ventricle, reducing the ventricular space and allowing the flattened walls of the third ventricle to lie against one another. Consequently, the incision in the lamina terminalis will sever both the superior and the inferior wall. A large tumor located in the anteroinferior portion of the third ventricle will then be visible. This component of tumor will have been the enhanced portion on CT scans, and the cystic contents will already have been evacuated. Further dissection should proceed painstakingly. The number of thin cotton pledgets should be counted as they are inserted one by one between the internal wall of the third ventricle and the tumor. This procedure causes the tumor to rise gradually toward the lamina terminalis as a result of displacement by the cotton pieces (Fig. 17.29). At this time, simply because the tumor is visible it should not be decided to begin excision immediately. If piecemeal tumor excision were then to begin using forceps, the interface with normal brain tissue would not be discernible, damage to brain would be likely, and portions of the tumor would undoubtedly be left behind. Only after the tumor has been dissected and carefully delimited and it has been determined that damage to brain tissue will not be incurred should excision of the tumor itself be performed. Because of removal of tumor tissue, the ventricular space may be further increased so that the insertion of pledgets will let the tumor rise further into the visual field. By repeating this procedure, the tumor is made still smaller and, in the end, complete excision is achieved (Fig. 17.30). During the entire excision procedure, arterial bleeding must be avoided. If venous bleeding occurs, hemostasis can always be achieved by compression with a pledget if sufficient patience is shown. Once the tumor has been completely excised, the cerebral aqueduct can be seen at the deepest region of the third ventricle (Fig. 17.31); inferiorly, the basilar artery and posterior cerebral arteries, superior cerebellar arteries, and oculomotor nerves can also sometimes be exposed. Olfactory tracts that have previously been sufficiently dissected will be preserved using these surgical procedures (Fig. 17.32).

Figure 17.28. A small incision in the lamina terminalis is made strictly in the midline.

Figure 17.29. Removal of the tumor via the lamina terminalis. Thin cotton pledgets are inserted between the internal walls of the third ventricle and the tumor. In achieving complete hemostasis, one should rinse the operative field repeatedly with normal saline. Hemorrhage during the operation will flow into the recesses near portions of the ICA, the A1 and A2 segments, the AComA, and the M1 segment and will collect there. In such cases, the formation of a hematoma around these vessels is likely to cause vaso-spasm postoperatively between the 4th and the 14th days. There are cases in which the tumor itself has been successfully removed but the

Figure 17.30. The tumor is removed piecemeal and, in the end, complete excision is achieved.

Figure 17.31. After the removal of the tumor, the cerebral aqueduct is well visualized through the defect in the lamina terminalis. patient has been lost because of vasospasm, and therefore the importance of complete hemostasis must be kept in mind. Nonetheless, regardless of how much care is taken during operation, some blood will work its way down posteriorly toward the deep portions of the bilateral sylvian fissures or into the space produced by tumor excision or anteriorly toward the chiasm. To prevent that blood from remaining, the many cotton pledgets that have been inserted during tumor excision must now be removed—taking care that none remain

Figure 17.32. The basilar artery and both superior cerebellar arteries can be seen. The olfactory tracts are completely preserved.

Figure 17.33. Watertight closure of the dura mater. behind. Once all pledgets have been removed, the dural opening is closed with interrupted silk sutures, and aron alpha adhesive is smeared over them so that the suture line is watertight (Fig. 17.33). Two twist drill holes are then made in the skull flap that has been removed and stay sutures, attached to the dura mater, are brought through the drill holes. The epidural space is made as small as possible to protect against the postoperative development of hematoma there and to allow the skull flap to be fastened in place (Fig. 17.34). The frontal burr holes are filled with a cranioplasty material (methyl methacrylate) for cosmetic purposes. The tube for continuous ventricular drainage (CVD) is brought out through a counter incision and is maintained for postoperative control of intracranial pressure. After muscle, galea, and

Figure 17.34. The dura mater is hitched up to the middle of the bone flap with silk passed through twist drill holes.

Figure 17.35. Scalp closure. The CVD tube is immobilized on the skin. skin are sutured, the CVD tube is wound two full turns on the scalp and carefully sutured at several sites—thus completing the operation (Fig. 17.35). An incision in the flattened anterior wall of the third ventricle has been made and considerable trauma of the third ventricle has been incurred; therefore, steroids should be administered before operation and body weight, water, and electrolyte balance should be checked at hourly intervals. Moreover, detailed care must be provided to prevent gastrointestinal bleeding and various infections. Of course, intracranial pressure is con-

Figure 17.36. Principal operative approaches to a tumor of the third ventricle in each portion of the ventricle. trolled by means of CVD, but a 2-week limit is recommended because of the danger of infection, after which the CVD tube should be removed and a shunt operation should be performed if intracranial pressure remains high. CT scanning 1 week postoperatively confirmed complete removal of the tumor (Fig. 17.36). The postoperative status of this patient is satisfactory and he is now reappointed as a computer technologist.

Indications This approach is suitable for tumors of the anterior part of the third ventricle, especially for tumors that develop anteriorly from the line joining the anterior ridge of the foramen of Monro and the cerebral aqueduct. For tumors developing from the pineal region, complete excision by opening only the lamina terminalis is not possible (Fig. 17.37). Among tumors that grow mainly within the third ventricle, there are optic gliomas, pituitary adenomas, Craniopharyngiomas, meningiomas, cavernous angiomas, and arteriovenous malformations. In the case of suprasellar mass lesions that displace the third ventricle inferoposter-iorly, however, this approach produces a wide operative field and consequently good operative results without incurring any damage to the brain tissue itself. For example, this approach is indicated in cases of giant suprasellar meningioma, giant AComA aneurysm, and giant olfactory groove meningioma—for any of which opening of the lamina terminalis is not required. Moreover, we use a bifrontal interhemispheric approach in all cases of AComA aneurysm. The degrees of invasion and of adhesion to the tissue surrounding the third ventricle can be anticipated by means of MRI-CT scans, and total excision is never attempted when such adherence or invasion is severe. It is also important that rigorous endocrine testing has been done preoperatively, so that replacement of hormone can be begun before operation. Particularly glucocorticoid replacement should be begun on the day before the operation. The following are brief descriptions of three cases in which this approach was used for the excision of a mass lesion. Case 1 The patient was a 60-year-old man with a giant pituitary adenoma growing within the third ventricle (Fig. 17.38A). The chief complaint was bitemporal hemianopsia. Complete excision was done using this approach. Postoperatively transient diabetes insipidus and hypernatremia appeared, but he was discharged in excellent condition after 2 months. Postoperative CT scan showed no residual tumor (Fig. 17.38B).

Figure 17.37. Postoperative CT scan shows total removal of the tumor. A. Axial scan. B. Coronal scan.

Figure 17.38. Case 1. CT scans with contrast medium enhancement. A. Preoperative CT scan showing a giant tumor of the third ventricle. B. Postoperative CT scan showing that the tumor has been totally removed.

Figure 17.39. Case 2. CT scans with contrast medium enhancement. A. Preoperative CT scan showing a tumor in the anterior region of the third ventricle. B. Postoperative CT scan showing that the tumor shadow has disappeared.

Case 2 The patient was a 40-year-old man with a cavernous angioma in the seat ot the left wall of the third ventricle (Fig. 17.39Л). He suffered from polydipsia, polyuria, and rightsided homonymous hemianopsia. With the bifrontal trans-lamina terminalis approach, the lamina terminalis was opened, revealing a conglomerate vascular tumor directly below. The tumor was surrounded by a mudlike substance thought to be an old hematoma. These findings suggested the occurrence of several previous hemorrhages within the lesion. The tumor was totally excised and the patient was discharged in excellent condition (Fig. 17.39B). Case 3 The patient was a 26-year-old man with an arteriovenous malformation extending along the anterior wall of the third ventricle, along the medial walls of the frontal horns of the lateral ventricles, and along the entire length of the corpus callosum (Fig. 17.40A). With the same approach, the third ventricle was entered via the lamina terminalis to reveal an irregular, bumpy internal surface of the third ventricle caused by the abnormal vessels. The draining vein poured into the galenic vein, which courses on the superior surface of the third ventricle. A silver clip was placed to the draining vein and total excision of the arteriovenous malformation nidus of the lateral and third ventricles was performed. Next the head was extended extremely and the abnormal vessels leaving both A3 segments of the АСА along the corpus callosum were cauterized one by one (Fig. 17.40B). By making use of this head position when performing bifrontal craniotomy, it is possible to observe from both sides of the falx as far as the posterior portion of the corpus callosum.

Figure 17.40. Case 3. A. The angiogram before operation showing an arteriovenous malformation at the third ventricle and corpus callosum. B. The angiogram after operation showing disappearance of the arteriovenous malformation.

Summary Advantages The advantages of the bifrontal interhemispheric trans-lamina terminalis approach can be summarized as follows. 1. The operative field is extremely wide, allowing visualization of all of the important nerves and blood vessels at the base of the brain, including the anterior half of the circle of Willis, both A2, M1 and M2 segments and the perforating branches leaving from those segments, the olfactory nerves, the optic nerves, etc. With the conventional unilateral subfrontal approach these important structures are behind in a dead angle, and it is more difficult to remove a tumor from the contralateral half of the third ventricle. 2. No damage is done to the brain tissue, except the lamina terminalis itself which, depending upon the nature of the tumor, is often widened and thinned and shows reduced functional capacities. 3. When necessary, the AComA can be divided and, under some cir cumstances, the optic chiasm can also be cut, providing a still larger operative field.

4. This approach is suitable not only for tumors within the third ventricle, but also for many other lesions located on the midline of the anterior fossa. Disadvantages The disadvantages of this approach can be summarized as follows. 1. Treatment of the frontal sinus must be done with extreme caution to prevent infection. 2. Additional care is needed when the superior sagittal sinus has been divided. 3. Care must be taken to preserve the olfactory nerves (14). These three points are not thought to be insurmountable demerits of the trans-lamina terminalis approach and complications using this approach are in fact rare. Pitfalls Pitfalls that we have experienced during operations using this approach include the following. 1. In elderly patients, there is sometimes severe adhesion of the bone to the dura mater and damage to the dura mater may be incurred during craniotomy. 2. The use of a large amount of bone wax to reduce the dead space of the frontal sinus may result in an aseptic epidural abscess. 3. When the superior sagittal sinus has been cut and ligated with sutures, the ligation of the severed end may become loosened during operation or after closure of the dura mater and a large hemorrhage from the venous sinus may occur. 4. When two clips have been placed on both severed ends of the AComA, either clip may slip off during operation and control of the subsequent hemorrhage from the AComA may be difficult. 5. A perforating branch may become separated from its trunk artery, causing severe hemorrhage. Complications Finally, with regard to postoperative complications, the following points should be mentioned. 1. Diabetes insipidus should be considered a virtual inevitability. It is consequently essential to measure the urine output at 1-hour intervals during and after operation and to check the water and electrolyte balance, serum osmolality, etc. Before the appropriate period has been lost, des mopressin acetate should be administered, thereby maintaining an ap propriate water and electrolyte balance. In pediatric cases, the diabetes insipidus is always transitory, whereas symptoms in adults may be prolonged. 2. Gastrointestinal bleeding is rare when the trans-lamina terminalis approach is used, but administration of cimetidine to protect against that possibility is recommended. 3. Adrenal shock is also infrequently encountered, but steroids are normally given starting preoperatively and continuing for 2 or 3 weeks postoperatively. Care must be taken in the gradual reduction of the steroid dose. 4. Intracranial pressure increases due to acute hydrocephalus. The CVD instituted during operation should be continued postoperatively to control intracranial pressure. In those cases where the pressure remains high for 2 weeks postoperatively, control can then be obtained by means of a shunt operation.

5. Infection is a relatively frequent occurrence after operation of the third ventricular region. Because steroids are also used postoperatively, the administration of antibiotics is recommended as a preventive measure. 6. CSF leakage may occur. There is often slow healing of the wound after complete excision of a tumor in this region and CSF leakage from the site of the scalp closure can easily occur. Because such leakage can lead to infection, it is essential to make the dura closure completely watertight. These complications can normally be overcome in children, but may prove more difficult in some adult patients. References 1. Cabezudo JM, Vaquero J, Areito E, Martinez R, Sola RG, Bravo G: Cranio pharyngioma: A critical approach to treatment. J Neurosurg 55:371-375, 1981. 2. Danoff BF, Cowchoch FS, Kramer S: Childhood craniopharyngioma: Survival, local control, endocrine and neurologic function following radiotherapy. Int J Radiat Oncol Biol Phys 9:171-175, 1983. 3. Ghatak MR, Hirano A, Zimmerman HM: Ultrastructure of craniopharyn gioma. Cancer 27:1465-1475, 1971. 4. Gordy PD, Peet MM, Kahn BA: Surgery of craniopharyngioma. J Neurosurg 6:503-517, 1949. 5. Grant FC: Surgical experience with tumors of pituitary gland. JAMA 136:668-672, 1948. 6. Hoffman HJ< Hendrick EB, Humphreys RP, Buncic JR, Armstrong DL, Jenkin RDT: Management of craniopharyngioma in children. J Neurosurg 47:218227, 1977. 7. Ingraham FD, Matson DD, McLaurin RL: Cortisone and ACTH as an adjunct to the surgery of Craniopharyngiomas. N Engl J Med 246:568-571, 1952. 8. Kubota T, Yamamoto S, Kohno H, Ito H, Hayashi M: Operative procedures of craniopharyngioma estimated by autopsy findings. Neurol Med Chir 20:341354, 1980 (in Japanese). 9. Lewis DD: A contribution to the subject of tumors of the hypophysis. JAMA 55:1002-1008, 1910. 10. Love JG, Marshall TM: Craniopharyngioma (pituitary adamantinoma). Surg Gynecol Obstet 90:591-601, 1950. 11. Northfield DWC: Rathke-pouch tumors. Brain 80:293-312, 1957. 12. Shillito J: Paediatric Neurological Surgery. New York, Raven Press, 1978. 13. Shillito J: Craniopharyngiomas: The subfrontal approach, or none at all. Clin Neurosurg 27:188-205, 1980. 14. Suzuki J, Yoshimoto T, Mizoi K: Preservation of the olfactory tract in bifrontal craniotomy for anterior communicating artery aneurysms, and the functional prognosis. J Neurosurg 54:342-345, 1981. 15. Suzuki J, Katakura R, Mori T: Interhemispheric approach through the lamina terminalis to tumors of the anterior part of the third ventricle. Surg Neurol 22:157-163, 1984. 16. Sweet WH: Radical surgical treatment of craniopharyngioma. Clin Neurosurg 23:52-79, 1976.

18 Pterional Approach George T. Tindall, M.D., and Suzie C. Tindall, M.D.

For more than 20 years the pterional (frontotemporal) approach has been used for the surgical exposure and treatment of aneurysms located on the anterior portion of the circle of Willis. Only in recent years has this approach become popular for exposing other lesions situated in the suprasellar and parasellar regions. In 1972 Krayenbuhl et al. (4) reported a series of 250 patients with cerebral aneurysms operated upon utilizing microsurgical techniques. The good results achieved—83% of the entire series and 94% of patients who were Grades 1, 2, or 3 preoperatively— showed the importance of microsurgical techniques in the modern management of intracranial aneurysms. In their publication, which emphasized a number of important principles of aneurysm management, the authors described their relatively small osteoplastic frontotemporal craniotomy with the bone flap centered on the pterion. Although this bony opening seems relatively small compared to some of the bone flaps used in earlier years to expose aneurysms and other lesions in the vicinity of the chiasm, when used with the operating microscope and appropriate partial removal of the sphenoid bone it serves the purpose of providing excellent exposure of the chiasmal region. As Krayenbuhl et al. (4) commented, aneurysm exposure was achieved at the expense of the subarachnoid space and the bony floor of the anterior fossa rather than by undue retraction of the brain. Also, in this publication the authors described the use of a self-retaining retractor that ensured a constant but minimal amount of retraction, a distinct advantage over the inconsistent and sometimes dangerous use of hand-held brain retractors. In 1975, Yasargil and Fox (8) reported their impressive results in 505 patients operated upon for aneurysm. They described the pterional approach in considerably more detail. Although it is difficult to assign any specific individual credit for developing the pterional approach, it is certainly appropriate to give special mention to Yasargil for providing the details and popularizing this particular approach. The pterional approach described by Krayenbuhl and Yasargil resembles very closely the anterior craniotomy approach described by Dandy for aneurysms on the anterior portion of the circle of Willis. In his

excellent publications on the surgery of intracranial aneurysms (2, 3) he described a craniotomy that with minor modifications could be considered an early version of a pterional approach. In the procedure described by Dandy, the head was turned to the opposite side and a concealed, curved anterior frontal skin incision within the hairline was made. The extent of the bone flap removal was such that the sylvian fissure, anterior temporal lobe, and lateral inferior frontal lobe were exposed. Dandy's approach to the aneurysm was made by upward retraction of the inferior portion of the frontal lobe, essentially the same procedure used in the modern pterional approach. His bone flap was much larger than the one currently used and no attempt was made to drill the sphenoid flush so as to provide deep exposure with minimal retraction. The major difference between Dandy and modern neurosurgeons is, of course, the current use of microtechniques, which were not available to Dandy and other neurosurgeons of his era. In 1981, Rhoton et al. described a number of operative approaches to the third ventricle, among them a frontotemporal approach that is similar to the pterional approach for aneurysms except that it is extended further posteriorly into the temporal region (6). These authors recommended this latter approach only if a tumor involving the third ventricle was centered lateral to the sella or extended into the middle cranial fossa. Symon advocated the use of a midtemporal approach, which is essentially a pterional approach with temporal extension, for the operative removal of Craniopharyngiomas (7). Based on 100 cases, he thought that this approach, which provides access to lesions behind the optic nerves, makes it possible to achieve effective tumor removal. The pterional approach is used most frequently today for aneurysm surgery. It does, however, have a place in the surgical treatment of lesions in the suprasellar and inferior third ventricular regions. In general, the remaining comments in this chapter apply to nonaneurysmal lesions accessible by the pterional approach. Indications for the Pterional Approach The pterional approach has limited application for lesions in or about the third ventricle. From an anatomical standpoint, suprasellar and parasellar lesions that are located within the cross hatched area in Figures 18.1 and 18.2 can be adequately reached and surgically managed through a pterional approach. As shown in the coronal view (Fig. 18.1), suprasellar lesions that extend laterally into the anterior and middle cranial fossae are particularly suitable for the pterional approach. There is no limit to the lateral tumor extension that can be managed by the pterional approach if the lateral extension is ipsilateral to the surgical flap. The area depicted in the sagittal view (Fig. 18.2) extends from the planum sphenoidale anteriorly to the basilar artery caudally and vertically from within the sella to a point approximately 1.5 to 2.0 cm directly above the diaphragma sellae. From the standpoint of pathology, a variety of lesions can be expected in this anatomical area and all are suitable for attempted excision by this approach. In many instances an exact pathological diagnosis will not be apparent until frozen section examination of biopsy material at the time of operation. A list of possible lesions encountered in this area is shown in Table 18.1. In clinical practice, the most common neoplastic lesions are Craniopharyngiomas and pituitary adenomas and the most common vascular lesion is an aneurysm. However, the anatomical location of the lesion in question is the primary concern in the choice of the pterional

Figure 18.1. Coronal view showing the region of parasellar access via the left pterional approach.

Figure 18.2. Sagittal view showing the region of ideal access: from the planum sphenoidale anteriorly to the basilar artery posteriorly; from the sella to 2.0 cm superior to the diaphragm.

approach. Many of the lesions mentioned in the table are better handled by alternative exposures if they extend outside the designated anatomical area. Contraindications Contraindications to the operative procedure are both general and specific. General contraindications are those that mitigate against any major operative procedure and include poor health, an unstable medical condition, or a septic process anywhere in or on the body. Advanced age (i.e., >65 years) in itself is not a contraindication as long as the patient's general health and medical condition are both good. The only specific contraindication to the pterional approach would be in patients in whom the lesion is primarily out of the region cross hatched in Figures 18.1 and 18.2. This would include lesions that appear on computerized tomographic (CT) scan or other diagnostic studies to be predominantly within the third ventricle (although a lesion mainly in the third ventricle but with definite exophytic extension into the aforementioned cross hatched area can at least be biopsied and thus positively identified using the pterional approach). Other circumstances that may be encountered, particularly invasion of the cavernous sinus, do not, in our opinion, contraindicate the pterional approach. Although a gross total excision of the lesion may not be possible in this latter situation, nonetheless, a positive tissue diagnosis and a partial resection of a tumor can be accomplished via a pterional approach. Preoperative Evaluation Patients presenting with lesions in or immediately below the inferior third ventricle in an anatomical location suitable for the pterional approach will usually present with signs or symptoms referable to increased intracranial pressure or ophthalmological or neuroendocrine disorders (Table 18.2). It is important that the preoperative assessment of each patient be individualized depending on age, duration of symptoms, major complaints, etc., rather than applying a rigid protocol to each patient with a suspected lesion in the suprasellar region. Symptoms of increased intracranial pressure include nausea, vomiting, headache, apathy, and lethargy and are usually progressive. These symptoms usually follow the development of hydrocephalus due to obstruction of cerebrospinal fluid (CSF) flow through the third ventricle. However, increased intracranial pressure may also simply reflect the volume of the tumor or the amount of peritumoral edema. Headache may occur in isolation and may be due to dural involvement at the skull base or irritation of the fifth cranial nerve within the cavernous sinus.

Ophthalmological involvement is common with lesions in the suprasellar region. Visual acuity may be compromised in one or both eyes by either intrinsic (e.g., optic glioma) or extrinsic lesions. An afferent pupillary defect (Marcus-Gunn pupil) may accompany the diminished acuity. The type of visual field defect may give some hint of lesion localization within the suprasellar area. A defect confined to one eye suggests anterolateral localization compromising a single optic nerve; a junctional sco-toma suggests anteromedial localization with compromise of the optic nerve at the anterior chiasm; bitemporal superior quandrantanopsia or hemianopsia implicates chiasmal compression from below; and a congruous homonymous hemianopsia suggests more posterior involvement of the optic tract. Cavernous sinus invasion may be associated with involvement of the third through sixth cranial nerves. The large number of hypothalamic syndromes that can result from pathological involvement of this structure include disorders of consciousness and sleep, cognition, emotional behavior and affect, autonomic balance, caloric balance, and water-osmolar balance. As a detailed discussion of these syndromes is beyond the scope of this chapter the reader is referred to more definitive references, such as the publication of Plum and Van Uitert (5). Endocrinological involvement usually consists of varying degrees of hypopituitarism due in most instances to involvement of the region of the median eminence or pituitary stalk rather than to actual damage to the pituitary gland itself. This is understandable in view of the fact that the portal venous system that traverses the stalk is the major vascular supply of the adenohypophysis. Thus, interruption of the stalk results in extensive necrosis of the anterior pituitary. Also, varying degrees of diabetes insipidus (DI) can result from stalk impairment. Interestingly, in a patient with DI the subsequent development of hypopituitarism is associated with significant amelioration of the DI. The improvement is attributable to the low rate of glomerular filtration resulting from growth hormone and adrenal steroid deficiencies and increased tubular resorp-

tion of water due to adrenal insufficiency. The clinical findings associated with hypopituitarism are due to deficiencies in pituitary-target organ endocrine function and are outlined in Table 18.2. Endocrinological Testing Endocrinological evaluation is important in patients with parasellar and suprasellar lesions for the following reasons: (a) to determine whether there is deficient pituitary function; (b) to monitor the effects of therapy (e.g., either a loss or a return of function as a result of operation); and (c) to aid in following the status of a lesion in terms of progressive growth or recurrence. It is important to assess the status of the pituitaryadrenal and pituitary-thyroid axes before performing invasive diagnostic studies and surgical procedures. Any deficiency in these axes should be replaced and, in the case of the pituitary-adrenal axis, not only replaced but supplemented significantly before, during, and immediately after invasive diagnostic and surgical procedures. The stress of a major procedure in a patient with an undetected deficiency of endogenous adrenal steroids can provoke an Addisonian crisis during the postoperative period which can, in turn, be fatal unless recognized and treated quickly and appropriately. Minimal endocrine testing for patients with lesions in the suprasellar region includes an evaluation of the pituitary-target organ axes; tests recommended for this purpose are listed in Table 18.3. From these data, pituitary endocrine function can be evaluated. Additionally, in a patient in whom there is DI or a suspicion of DI, a water deprivation test to evaluate antidiuretic hormone or vasopressin reserve should be performed. Also, serum prolactin levels in patients with suprasellar lesions should be determined for two important reasons. First, the suprasellar lesion in question and for which operation is planned may be a pituitary tumor that secretes prolactin, a so-called prolactinoma. In these cases levels of serum prolactin are usually in excess of 150 ng/ml and may even exceed 1000 ng/ml. The second and a more likely cause for the elevated prolactin is related to the "stalk phenomenon." The "stalk phenomenon" is that situation which results when prolactin inhibitory factor (PIF), which is conveyed in the portal system, is prevented from reaching the normal adenohypophysis. Structural lesions that compress the pituitary stalk interfere with the delivery of PIF, a substance thought to be dopamine, thus allowing the normal adenohypophysis to secrete an ex-

cess of prolactin. In these instances, prolactin levels are usually in the range of 70 to 100 ng/ml and should not exceed 150 ng/ml. Structural lesions (other than a prolactinoma) that elevate serum prolactin by this mechanism are referred to as pseudoprolactinomas. Tumor Markers and CSF Cytology The presence of alpha-fetoprotein (AFP) or the beta subunit of human chorionic gonadotropin (HCG) in the lumbar CSF may help to predict the pathological diagnosis of a suprasellar mass. Elevated levels of CSF HCG are frequently seen in patients with suprasellar germinomas, and elevated CSF levels of AFP have been reported in patients with malignant teratomas that have yolk sac elements within them. If a germinoma is suspected CSF cytological examination may give positive identification. Neuroradiological Studies Neuroradiological studies have changed significantly over the past several years and, in general, have become more accurate from a diagnostic standpoint and safer and less painful for the patient. CT scanning has made a major impact on neurological diagnosis and has replaced procedures such as pneumoencephalography and polytomography. The latest diagnostic breakthrough, magnetic resonance imaging, is almost certain to play a major role in the treatment of these suprasellar lesions, but sufficient experience has not yet accumulated to define its role. Currently, recommended neurodiagnostic studies for patients with suprasellar lesions include anteroposterior and lateral skull films and high resolution CT scans in both axial and coronal views. A contrast-enhanced scan should be part of the routine CT examination. Although not a routine procedure, conventional cerebral angiography or high quality arterial digital angiography should be performed in patients with suprasellar lesions if there is any doubt as to the pathological nature of the lesion from a review of the skull x-ray films and CT scan and to determine the degree of vascularity of the lesion. Vascular lesions and the extent of vascularity of certain tumors can only be accurately defined by conventional angiography or high quality arterial digital angiography. In our opinion, digital intravenous angiography does not provide adequate quality imaging of these lesions. In some patients, special studies such as metrizamide cisternography (MC) and cavernous sinus venography (CSV) may be helpful in further defining the boundaries or invasiveness of a lesion. In our opinion, MC may be indicated in patients with a relatively small suprasellar tumor associated with a normal-sized sella turcica. In some instances, we have found that the metrizamide study provides a more accurate depiction of the boundaries of the lesion and its relationship to surrounding normal structures than can be obtained from a routine CT scan. MC involves a lumbar puncture using a #22 gauge needle. After satisfactory placement of the needle tip in the subarachnoid space, 5 ml of metrizamide (190 mg/ml) is injected slowly. The contrast medium is allowed to enter the basilar cisterns by gravity by placing the patient in a 60° Trendelenberg position for 2 minutes. CT scans are then obtained using serial sections 5 mm thick in both sagittal and coronal planes. CSV, which is a rarely used procedure, is accomplished by transfemoral venous catheterization of the inferior petrosal sinus and may help to determine whether the cavernous sinus is invaded by tumor. It is relatively uncomfortable for the patient and the amount of valuable diagnostic information provided by the test often leaves much to be desired.

Technical Aspects of the Pterional Approach to Inferior Third Ventricular and Parasellar Lesions Before planning the procedure the surgeon should have adequately evaluated several important factors about the patient. These include (a) the exact anatomical position of the lesion in question; (b) the degree of extension, if any, into the frontal fossa, the middle fossa, or the retroclival area and on which side; (c) the relationship of the lesion to the major surrounding anatomical structures such as the optic nerves and chiasm, the internal carotid artery and its major branches, the basilar artery, and the cavernous sinus; and (d) the status of the intracranial pressure (ICP); i.e., whether normal or elevated. If elevated, it is important that the cause of the elevation—hydrocephalus, mass effect, or peritumoral edema—be identified so that an appropriate treatment program can be considered or instituted before a definitive pterional approach to the lesion. Preoperative Considerations Once the exact anatomical position of the lesion has been ascertained and it has been determined that the pterional approach is the appropriate surgical procedure for the lesion in question, the surgeon must decide from which side the approach will be made. If the lesion has significant lateral extension into the frontal or middle fossa, it should be approached from the side of maximal lateral extension. If this is not a determining factor, the lesion should be approached from the side of maximal visual loss. If neither of these rules apply, we prefer to operate from the side of the nondominant hemisphere, generally the right side. If the patient has significant hydrocephalus and there is doubt that the parasellar lesion can be resected completely, we advise performing a ventriculoperitoneal shunting procedure at least several days before the definitive craniotomy. If the foramen of Monro is compromised by the lesion, biventricular shunting is carried out. Careful planning is necessary for placement of the shunt catheters and tubing so that the incisions and the course of the tubing will not interfere with the pterional approach. If a right pterional approach is planned, we position the left ventricular catheter transfrontally and the right ventricular catheter transparietally, situating Rickham reservoirs within the holes used for passage of the catheters and joining the distal drainage tubing of both systems through a "T" connector above a suitable valve situated beneath the mastoid fascia behind the right ear. This leaves the right frontal area free of tubing at the time of definitive craniotomy (Fig. 18.3). If there is significant peritumoral edema or tumor mass effect the patient is pretreated with relatively high doses of steroids (6 to 10 mg of dexamethasone QID or the equivalent) for several days before the definitive surgical procedure. We do not use prophylactic anticonvulsants preoperatively unless a cortical resection is planned or the lesion in question has invaded cortical structures. In view of current cost containment awareness, routine type and cross match of the blood of patients undergoing craniotomy by the pterional route are not performed unless the lesion has the potential for rapid blood loss (e.g., meningioma, arteriovenous malformation, aneurysm, etc.). The patient undergoes a thorough shampoo with Betadine or a comparable antiseptic soap on the night before the procedure, and the surgical area on the scalp is shaved immediately before transport of the patient to the operating room. Usually, only that area of the patient's scalp around the planned scalp incision is shaved. Perioperative antibiotic coverage is not used as a routine unless: (a) the

Figure 18.3. Representation of biventricular shunt placement designed to facilitate right frontal craniotomy. patient has been previously shunted or has other previously implanted foreign bodies (heart valves, vascular grafts, etc.); (b) it is anticipated that the operative procedure will last longer than 5 hours; (c) the patient has a large frontal sinus that will be entered during the procedure; or (d) the patient requires cardiac prophylaxis because of previous rheumatic fever. Patient Positioning In most instances the patient can be placed in the supine position with a soft roll under the ipsilateral shoulder. However, if the patient has a short, relatively nonflexible neck or if significant neck turning might compromise jugular venous return, the lateral decubitus position is used. If the lesion is large, there is significant peritumoral edema, or there is some other reason to believe that brain retraction will be difficult, a lumbar subarachnoid catheter or 18 gauge lumbar puncture needle may be inserted before positioning the patient. A drainage bag with stopcock for on/off control by the anesthesiologist is connected to the needle or indwelling catheter. During the dural opening and particularly during elevation of the frontal lobe, the catheter can be opened to allow CSF to

Figure 18.4. Fixation points of headrest before final positioning. drain slowly. This procedure considerably facilitates exposure to the region of the chiasm. Placement of the head in a three-point skull fixation clamp is necessary to prevent any unwanted head movement throughout the procedure. The device is applied with the single pin in or near the midline on the forehead and the remaining two pins straddling the occiput (Fig. 18.4). The head is then positioned appropriately for the approach. Usually, the sagittal plane of the head is parallel to the floor with the pterion uppermost. However, if the approach is to be directed more anteriorly into the frontal fossa or retrochiasmatic region the forehead may be depressed slightly. If there is to be significant dissection in the region of the tentorial hiatus and in subtemporal regions, the vertex may be depressed slightly; if an approach to the basilar artery or supraclival region is contemplated, the forehead may be elevated slightly. Operative Procedure Once the patient has been suitably positioned the scalp incision is planned before the application of drapes. The usual incision is begun 1 cm anterior to the tragus of the ear just at the zygomatic arch and is extended up just behind the hairline to the midsagittal plane 3 cm anterior to the coronal suture. As depicted in Figure 18.5, the scalp incision may be varied so as to permit more frontal exposure by bringing it across the center of the head in a curvilinear fashion (this is cosmetically preferable to curving it down over the forehead in front of the hairline) or to permit more temporal exposure by gently curving it farther back above the ear and then proceeding in a frontal direction. Once the planned incision is marked, the head is scrubbed with Betadine soap and reprepared with Betadine solution and sterile towels and draping are applied circumfer-entially. The use of a plastic shield such as Vi-drape isolates the exposed skin from the surgical field. In most instances we inject lidocaine with 1:100,000 epinephrine subcutaneously before making the scalp incision to aid in scalp hemostasis. Raney clips are used along the scalp edges. It is important to limit the amount of dissection in the subgaleal space between the galea and the temporalis fascia anteriorly so as to avoid injuring the frontalis branch of the facial nerve. With the electrocautery,

Figure 18.5. Variability of scalp incision to permit greater frontal or temporal exposure. the temporalis fascia and muscle are incised immediately beneath the scalp incision and all layers are turned forward as a unit (Fig. 18.6). The temporalis muscle is dissected off the skull far forward to the orbitofrontal angle and the superior orbital ridge and low over the temporal fossa. The muscle can be retracted to achieve maximal exposure with the use of rubber-banded sutures or fishhooks. A self-retaining retractor is used to spread the superior aspect of the incision. While turning the scalp and bone flaps, we administer 30 to 50 g of mannitol (20% solution) i.v. If there is a possibility that increased ICP might be a significant problem, 40 mg of furosemide is also given. The arterial pCO2 is kept in the range of 25 to 30 torr during the procedure and the patient is paralyzed and mechanically ventilated. The exact position of the bone flap will again depend upon the need for maximal exposure in certain areas. The standard flap, as depicted in Figure 18.7) is usually fashioned with four burr holes, one at the orbitofrontal angle, one over the middle of the coronal suture, one in the midtemporal region, and one low in the temporal fossa. The size of the flap is dictated somewhat by the surgeon's estimate of the amount of retraction necessary for appropriate exposure of the lesion. It is better to err on the side of the larger flap than to have to revise the flap during the procedure. For more exposure in the subfrontal or perichiasmatic region the flap can be expanded anteriorly by placing an additional hole over the supraorbital ridge. The presence of a large frontal sinus with more than the usual amount of lateral extension that would obviously require entry to achieve a pterional bone flap with an anterior extension is not a contraindication to the use of this approach. In these cases, we would proceed to enter the sinus, exenterate as much mucosa as feasible, and pack off the remainder of the sinus with either muscle or adipose tissue held in place with dural tack-up sutures. For more temporal exposure the single midtemporal hole can be replaced by two holes placed more pos-

Figure 18.6. A. Incision with respect to bony landmarks. B. Course of the frontalis branch of the facial nerve; dissection of the subgaleal space anteriorly should be limited. C. The temporalis fascia and muscle are reflected as a unit. teriorly. The options for bone flap placement are indicated in Figure 18.7. We prefer to connect the orbitofrontal and subtemporal holes by ron-guering a channel between them before completing the flap with the pneumatic craniotome. This permits better control of the sometimes vascular sphenoid ridge and avoids inadvertent dural tears from the use

Figure 18.7. Burr hole placement in reference to bony landmarks with options for increasing frontal or temporal exposure. of the craniotome in the irregular terrain of the sphenoid ridge. Once the flap is elevated, the dura mater is tacked up to small holes previously drilled obliquely in the bone edges. Attention is then directed to the sphenoid ridge. The lateral aspect of the sphenoid ridge is drilled away flush with the frontal fossa (Fig. 18.8). This maneuver is combined with a small subtemporal craniectomy. The importance of this aspect of the procedure cannot be overemphasized. If accomplished correctly it provides for much better visualization of the deep structures and less necessity for brain retraction later in the procedure. The usual dural opening is shown in Figure 18.9. The initial opening

Figure 18.8. A combination of temporal craniectomy (A) and excision of the sphenoid wing (B) develops a basal pterional exposure (C). is curvilinear, 1.5 to 2 cm back from the sphenoid ridge, and the small dural flap is tacked back over the ridge. Several cruciate incisions are made in the dura mater overlying the frontal and temporal lobes. These dural flaps can be left in place for a modest amount of brain protection during the remainder of the procedure. At this point we place moist gauze sponges and cottonoids strategically over the exposed flap to prevent excessive drying of the tissues and thrombin-soaked gelatin sponge with cottonoids along the bone flap circumference to prevent oozing into the wound during the remainder of the procedure.

Figure 18.9. Dural opening.

A self-retaining retractor system of the surgeon's choice is set up (Fig. 18.10). Two retractors, a frontal and a temporal, are generally used. The frontal is usually the wider of the two retractors. Strips of moist cottonoid can be used between the brain and the retractors. The frontal blade is carefully placed beneath the frontal lobe and gently advanced, exposing first the ipsilateral olfactory nerve and then the optic nerve at the anterior clinoid process (Fig. 18.11). The temporal blade is used for gentle retraction of the anterior-superior temporal lobe. As this retractor is advanced, one needs to visualize, coagulate, and incise any bridging veins from the surface of the temporal lobe. A bridging vein torn later in the procedure because this step was not accomplished adequately at the outset can be a source of frustration for the surgeon. We prefer to use magnifying loupes of 2 or 2.5X for all parts of the procedure described to this point. Once the retractors are set, the operating microscope is brought into the field and used throughout the rest of the operation until closure is begun. The microsurgical anatomy encountered during the pterional approach is depicted in Figure 18.12. It is convenient to think of this anatomy in terms of three triangles— frontal, middle, and posterior. The frontal triangle is that area bounded by the two anterior limbs of the optic chiasm and the tuberculum sellae. The middle triangle is bordered by the poster-olateral aspect of the ipsilateral optic nerve, the anterolateral aspect of

Figure 18.10. Self-retaining retractor blades are applied along frontal and temporal vectors. the ipsilateral internal carotid artery, and the A1 segment of the anterior cerebral artery. The posterior triangle is defined by the posterolateral aspect of the ipsilateral internal carotid artery, the tentorial edge, and the retracted temporal lobe. Working over the frontal retractor the surgeon first exposes the ipsilateral optic nerve and the infrachiasmatic cistern by gentle dissection of the surrounding arachnoid. Entry into the interchiasmatic region, or the frontal triangle, frequently results in a substantial egress of spinal fluid, which facilitates brain relaxation. Moving the frontal retractor posteriorly and after more arachnoidal dissection in the middle triangle the internal carotid artery comes into view lateral and posterior to the optic nerve. With the addition of temporal retraction, the tentorial edge, third cranial nerve, posterior communicating artery, and interpeduncular cistern can be explored in the posterior triangle. With slightly more retraction with the frontal retractor the anterior choroidal artery and the A1 segment of the anterior cerebral artery can be exposed. Working between and around these structures it is possible to visualize adequately the anteromedial portion of the contralateral optic nerve, the diaphragma sellae, the infundibulum, the contralateral internal carotid artery, the ipsilateral posterior clinoid process, the superior portion of the clivus, and the basilar artery. The anatomical appearance may be altered significantly depending upon the pathological condition in the area. In Figure 18.13 a cyst of a suprasellar craniopharyngioma presenting itself in the interchiasmatic region spreading the optic nerves apart is shown. Figure 18.14 shows the more solid part of this tumor as it insinuates itself between and around the structures of the middle and posterior triangles. Mass effects of such

Figure 18.11. A. The frontal blade is advanced to identify first the ipsilateral olfactory nerve. B. Next, the anterior clinoid process and optic nerve are appreciated. C. The temporal blade is advanced to enhance the lateral carotid exposure. lesions may significantly distort and dislocate normal anatomical structures. If the pathological character of the lesion is not immediately apparent upon initial inspection, the surgeon should consider needle aspiration using a 20 gauge spinal needle on a small syringe. If this maneuver is productive of arterial blood, the diagnosis of an aneurysm not visualized on the preoperative studies can be made and a potential catastrophe can be averted as the plan of management is altered.

Figure 18.12. A. The frontal triangle bounded by the two anterior limbs of the optic chiasm and tuberculum sellae. B. The middle triangle bounded by the posterolateral aspect of the ipsilateral optic nerve, the anterolateral aspect of the ipsilateral internal carotid artery, and the A1 segment of the anterior cerebral artery. C. The posterior triangle bounded by the posterolateral aspect of the ipsilateral internal carotid artery, the tentorial edge, and the retracted temporal lobe.

Figure 18.13. Suprasellar craniopharyngioma with distortion of the interchiasmatic region.

Figure 18.14. Distortion and displacement of the right optic nerve and carotid artery by a tumor mass developing a tumor presence in the middle and posterior triangles. The techniques used for resection will depend to some extent on the pathological nature of the lesion as determined by gross inspection and frozen section biopsy. With a cystic craniopharyngioma care should be taken to aspirate as much of the cystic contents as possible to prevent contamination of the spinal fluid with this material, which could lead to a postoperative chemical meningitis. Likewise, in dealing with a germi-noma, which has the propensity to seed through the subarachnoid space, the area around the tumor site should be protected by cottonoids and, as

much as possible, the tumor should be removed intact to prevent contamination of the spinal fluid. Sharp dissection of the arachnoidal and anatomical planes using bipolar coagulation, microscissors, and knives is preferable to a technique of pulling or tearing with dissectors, although the latter instruments can be used to advantage to define the planes initially. We believe that the limitations of exposure in this area make use of the Cavitron ultrasonic surgical aspirator difficult. However, the carbon dioxide laser is extremely helpful when dealing with tough, non-suckable tumor tissue such as may be encountered in meningioma or solid craniopharyngioma. From the pterional approach there will be certain areas that are blind to the surgeon such as the inferior surface of the ipsilateral optic nerve and the hypothalamic recesses in the floor of the third ventricle. A small 3-mm micromirror may help in visualizing such areas. The majority of the visible tumor should be resected initially with the expectation that the component extending up into the floor of the third ventricle may then collapse into the surgical field. Attempts to pull at tumor outside of the field of direct vision can have serious consequences. Tolerance of the normal anatomical structures encountered in the pterional approach to surgical dissection can only be learned by experience. An optic nerve or chiasm already stretched and distorted by adjacent tumor is extremely sensitive to any type of manipulation. The same can be said for the third cranial nerve. In general, the large vascular structures encountered in this approach will tolerate more manipulation than the adjacent neural structures. With completion of definitive removal of the lesion, hemostasis should be meticulously obtained using bipolar coagulation and gentle irrigation with saline. The anesthesiologist is asked to perform a sustained Valsalva maneuver on the patient to evaluate hemostasis before closure. The self-retaining retractors are then removed under direct vision. Closure of the dura mater is accomplished with 4-0 absorbable Vicryl (polygalactin; Ethicon, Inc., Somerville, New Jersey) suture and the aid of a pericranial graft if necessary. The bone flap is firmly resecured using 26 gauge wire or 2-0 silk. The authors use acrylic cranioplasty to fill all bony defects except the area of subtemporal decompression. Careful attention is paid to reconstituting the normal contour of the orbitofrontal angle and the burr hole sites in the superior aspect of the flap. Placing acrylic in the region of the removed subtemporal and inferior sphenoid ridge may result in an unacceptable bulge in this area, but using acrylic for the remaining bony defects results in an excellent cosmetic result and the patient appreciates not being able to feel the burr hole sites postoperatively. Closure of the temporalis muscle and fascia is with 2-0 absorbable Vicryl interrupted sutures. The galea is closed with interrupted 3-0 absorbable Vicryl sutures with buried knots. The skin is closed with a running, continuous 3-0 or 4-0 nylon suture, which is removed 7 to 10 days postoperatively. A small dressing is secured with paper-based tape. Pitfalls during the Intraoperative Period A number of pitfalls or technical mishaps can occur during the pterional approach. We do not include every possible problem that can arise, but we do describe the more common ones. In developing the skin, fascial, and muscle incision, one must keep in mind the course of the motor branch of the facial nerve that supplies the frontalis muscle. The nerve is difficult to visualize during the course of the operation and can be damaged by undue traction or dissection in the

subgaleal plane or divided during the skin and muscle incisions. Generally, one can avoid injury to this branch by opening the temporalis muscle and fascia immediately below the scalp incision and avoiding extensive dissection in the plane between the galea and temporalis fascia anteriorly. The superficial temporal artery is taking on more importance in the field of neurosurgery and we make an effort to preserve the largest branch of this artery whenever possible. A pitfall that one should strive to avoid is inappropriate placement of the bone flap; with careful planning this can be avoided. The bone flap should be tailored to the individual needs of the case, a point already emphasized. It is necessary to drill away the lateral sphenoid ridge flush or this bony protuberance will partially block the exposure of deeper structures. Adequate relaxation of the brain is necessary for safe and effective brain retraction and exposure of the inferior third ventricular and suprasellar areas. Occasionally, a "tight" brain is encountered before or at the time of dural incision. When this occurs, the operator should check to see that the usual measures for decreasing ICP are in use—i.e., hyper-ventilation to a PCO2 of 25 to 30 torr, mannitol, furosemide, steroids, and lumbar drainage. If, despite these measures, the brain continues to bulge, the operating table may be tilted to 10 to 15° of reverse Trendelenberg postiion, thus elevating the head slightly. A readjustment of the head position may also help to improve venous drainage and brain relaxation. If these measures fail, the frontal horn of the ipsilateral lateral ventricle may be tapped to drain enough ventricular fluid to relax the brain sufficiently for dural opening. Placement of self-retaining brain retractors can give problems. A common problem is for the retractor blade that is placed over the anterior temporal region to tear one of the bridging veins and thus cause troublesome bleeding. This can be avoided by cauterizing and dividing these veins as the retractor is carefully and slowly inserted. Another problem encountered particularly by the inexperienced operator is to place the frontal lobe retractor in too far and expose not the ipsilateral, but the contralateral olfactory bulb, thus causing confusion as the latter structure is followed posteriorly. Some difficulty may be experienced in exposing the retrochiasmatic area adequately so that a lesion in this area can be excised. Exposure in this case may be facilitated greatly by opening the arachnoid of the sylvian fissure before placing retractors. Excessive retractor pressure and frequent alteration in retractor position are to be avoided as they may lead to intracerebral swelling or hemorrhage. Complications The complications that potentially can follow the pterional approach are those that can be seen after any neurosurgical intracranial procedure. Infection, hemorrhage, and edema are the most common problems one is likely to encounter and each can be detected by close postoperative observation and the use of the appropriate diagnostic procedure, i.e., CT scanning. Complications peculiar or special to operation in the parasellar region include visual loss, diabetes insipidus, varying degrees of hypopituitarism, and rarely coma due to hypothalamic damage. Visual loss is usually a result of undue manipulation of the optic nerves or chiasm and is probably related to impairment in blood supply, particularly from damage to the arterial supply that reaches the chiasm inferiorly. Both diabetes insipidus and hypopituitarism can result from injury to the pituitary stalk. Both endocrine deficiency states can be confirmed by

appropriate laboratory determination. The syndrome of inappropriate antidiuretic hormone secretion can also follow operation in this area. References 1. Daughaday WH: The adenohypophysis. In Williams RH (ed): Textbook of Endocrinology, ed 5. Philadelphia, WB Saunders, 1974, pp 31-79. 2. Dandy WE: Intracranial Arterial Aneurysms. New York, Havner Publishing Co, 1969, pp 103-132. (Fascimile of the 1944 edition reprinted by arrange ment with Corness University Press.) 3. Dandy WE: Surgery of the Brain (A monograph from Lewis' Practice of Surgery, Volume XII). Hagerstown, MD, WF Prior Co, Inc, 1945, p 405. 4. Krayenbuhl HA, Yasargil MG, Flamm ES, Tew JM: Microsurgical treatment of intracranial saccular aneurysms. J Neurosurg 37:678-686, 1972. 5. Plum F, Van Uitert R: Nonendocrine Diseases and Disorders of the Hypo thalamus. In Reichlin S, Balderssarini RJ, Martin JB (eds): The Hypothala mus. New York, Raven Press, 1978, pp 415-473. 6. Rhoton AL, Yamamoto I, Peace DA: Microsurgery of the third ventricle: Part 2. Neurosurgery 8:357-373, 1981. 7. Symon L: Microsurgery of the hypothalamus with special reference to cran iopharyngioma. Neurosurg Rev 6:43-49, 1983. 8. Yasargil MG, Fox JL: The microsurgical approach to intracranial aneurysms. SurgNeurol 3:7-14, 1975.

19 Combined Approaches M. Gazi Yasargil, Prof. Dr. med., Peter J. Teddy, D. Phil., F.R.C.S., and Peter Roth

It has been known for many years that Craniopharyngiomas extending upward from the infundibulum of the third ventricle to the foramen of Monro (whether unilateral or bilateral, solid or cystic) cannot be completely removed using solely a frontobasal (pterional) approach and par-achiasmal dissection, even after opening the lamina terminalis. Although the basal and laterobasal portions of the tumor can be exposed by this route, the most superior and superoposterior portions, within the posterior half of the third ventricle, cannot be adequately dissected (Figs. 19.1 and 19.2). A subtemporal approach to Craniopharyngiomas has not been used in Zurich. It seems unlikely that complete removal of the tumors with large superior extension can be achieved in this manner for the same reason as for the pterional approach. Transcortical, transventricular operations are rejected for fear of producing porencephalic cysts or postoperative epilepsy and because, although the contralateral aspect of the third ventricle can usually be well demonstrated, it is difficult to control the ipsilateral dissection. The transcallosal approach has been used for arteriovenous malformations of the anterior third of the corpus callosum, colloid cysts of the third ventricle, and third ventricular gliomas. The pterional approach has been developed to facilitate the repair of aneurysms involving the circle of Willis. A combined paramedian-pterional approach was utilized quite successfully in three cases of multiple aneurysms (middle cerebral and/or basilar bifurcation with pericallosal aneurysm). The idea arose that Craniopharyngiomas could also be removed completely and safely by using this combined transcallosal and pterional operation. The transcallosal approach can be used to dissect the superior, superoposterior, and superolateral aspects of the mass, especially from the posterior half of the third ventricle and the origin of the aqueduct (Fig. 19.3). Removal of the parachiasmatic and suprachiasmatic segments of the tumor within the interpeduncular and ambient cisterns can be achieved from the frontobasal pterional opening. This would allow dissection of

Figure 19.1. Schematic representation of the multidimensional growing tendency of craniopharyngioma and its relation to the brain stem and the basal cisterns. A. Lateral. B. Anteroposterior view.

Figure 19.2. Direction of growth of parachiasmal tumors: A. Into the basal cisterns. B. Upward into the infundibulum without occlusion of foramina of Monro. С With occlusion of both foramina of Monro. D. Paramedian expansion and displacement of the third ventricle. the mass from around the internal carotid artery and its branches. The bilateral A1 segments and the anterior communicating artery, the basilar bifurcation with both P1/P2 segments, both superior cerebellar arteries, and the mamillary bodies. This frontobasal pterional approach is also effective for tumor dissection in the region of the optic nerves, chiasm, optic tracts, and third nerves. A particularly useful technique in removing

Figure 19.3. A combined approach (subfrontal-parachiasmal and transcallosaltransforamlnal) is required for proper dissection in a case of superoposterior expansion of the tumor into the third ventricle. the preinfundibular retrochiasmatic portion of these tumors is opening the lamina terminalis, delivering the tumor from beneath the chiasm to the chiasmatic and lamina terminalis cisterns, with subsequent removal from between the optic nerves (see Fig. 19.8). This combined approach was successfully used in 22 procedures for very large Craniopharyngiomas and in 2 cases of optic glioma arising from the optic tract with medial extension into the third ventricle causing obstruction of the foramen of Monro. We have also seen hypothalamic gliomas in three cases growing superiorly toward the foramen of Monro and extending into the suband prechiasmatic areas; these were removed using the combined approach. It is not always possible to predict the true extent of third ventricular tumors on preoperative computerized tomographic (CT) scans or angiograms. Often this must be established by a comprehensive exploration. Technique Of critical importance to this operation is the proper positioning of the patient. Particular care must be taken with regard to flexion of the neck (Fig. 19.4A) to bring the head into the correct degree of elevation. It is necessary to make one's approach through the corpus callosum and

Figure 19.4. A. Position of the patient for a combined approach to a third ventricular tumor. The arrows indicate the extent of the maneuverability available to the surgeon. B. One skin flap for two separate craniotomies (frontal-paramedian and pterional). The arrows indicate the surgical entrance for tumor dissection and removal.

foramina of Monro with the patient fully recumbent. This course would take a horizontal line to its lowermost limit. The patient must therefore be supine with the head somewhat flexed so that the surgeon is looking perpendicularly through the corpus callosum to the foramina and along this projected line to the infundibulum (Fig. 19.4B). The slight anterior tilting of the head is contrary to the normally adopted method of extending the neck to reduce frontal lobe retraction, as in the anterolateral approach where the head is also turned slightly to the left. As the head cannot easily be turned from right to left during the operation, a compromise must be accepted. However, as we need to reach midline structures there is little reason to require lateral tilt of the head. Furthermore, almost without exception there is obstruction of the foramina of Monro in these patients and nearly all have hydrocephalus. This helps the surgeon a good deal; after only a small opening is made in the corpus callosum, cerebrospinal fluid (CSF) is released and the frontal lobe falls back, thereby facilitating inspection of the parachiasmatic area with minimal retraction. One practical point is that the table position may need to be quite high and the arrangement of surgical drapes and instrument trays may not allow easy alteration. The surgeon may need a platform for the frontolateral part of the operation, but may sit for the transcallosal approach and dissection. The skin incision can be bifrontal or unilateral but must cross the midline (see Fig. 19.4B). The important factor is to recognize the position of the coronal suture, which is usually perpendicular to the foramen of Monro. Occasionally, gross chronic hydrocephalus may produce distortion of the calvarium so that the coronal suture is displaced (usually) backward. The general rule is, therefore, that the foramen of Monro is on the perpendicular from the coronal suture line or up to 1 cm anterior to it (see Fig. 19.4B). There is no need to turn a large bone flap, as under the same scalp flap two small craniotomies can be made. In cases of pronounced hydrocephalus the transcallosal approach is made first to facilitate release of CSF and minimize frontal lobe retraction during the second stage. In cases of only moderate hydrocephalus (especially Craniopharyngiomas) the frontolateral approach is made first. The sylvian fissure and then the carotid and lamina terminalis cisterns are opened such that the parachiasmatic extension of the tumor can be inspected and its relation to the optic nerve and tracts, the internal carotid artery, the third nerve, etc., can be studied. In cases of severe hydrocephalus and a tight brain, it may be necessary first to tap the lateral ventricle with a brain needle before starting any form of dissection or retraction. For the parasagittal craniotomy, three burr holes are fashioned—two in the midline (one 2 cm in front of the coronal suture and the other 1 cm behind) and the third 2 cm lateral to the midline so that a triangular free flap may be created and removed. The pterional craniotomy is carried out in the standard fashion previously described (1). After dissection of the temporal skin and muscle flaps four burr holes are made, the first behind the base of the zygomatic process of the frontal bone, the second over the glabella, the third 3 cm posterior to the first along the linea temporalis, and the fourth over the anterior part of the squama temporalis. A small quadrilateral free flap that is quite adequate for the frontobasal approach can thus be created. Midline Approach There are several points of surgical technique that are important when gaining access to midline structures using the transcallosal approach.

Figure 19.5. The corpus callosum is longitudinally divided at the level of the coronal suture. The dashed line indicates the site of the incision between the two pericallosal arteries. The craniotomy, although small (elongated anterosuperiorly), could be smaller yet if one could positively identify the location of the bridging veins to the sagittal sinus. Direct inspection determines the direction of the dissection, anteriorly or posteriorly, to these venous structures that are to be preserved. If a large leash of veins is encountered, dissection of their arachnoid sheath from the dura mater will permit mobilization. Wrapping these venous structures in small muscle fragments limits venous oozing. Occasionally, moderately sized veins must by necessity be sacrificed; excessive time must not be lost attempting to preserve these structures. Exploring the medial surface of the frontal lobe is generally straight-forward, especially when the subarachnoid cisterns are dilated. Small veins crossing the subarachnoid space are coagulated and divided as the frontal lobe is dissected free from the falx. Separating the falx and cingulate gyrus is also usually straightforward. Occasionally the falx is a fibrous network with numerous large defects. In this instance the variable depth of the falx suggests fusion of the cingulate gyri in the midline. Care must be taken not to confuse the opposed cingulate gyri for the corpus callosum. Under these circumstances the callosomarginal arteries within their respective sulci may be mistaken for the pericallosal vessels. The callosomarginal and pericallosal arteries occasionally provide small branches to the falx, but inadvertent avulsion is of no consequence. However, avulsion of the parent vessels will lead to troublesome hemorrhage. Upon reaching the body of the corpus callosum it is important to realize that there may be one to three A2 arteries (see Ref. 1). The arterial pattern in this region must be carefully studied on the preoperative angiograms and correlated with the surgical anatomy. One must always

Figure 19.6. A. The right foramen of Monro is obstructed by the expanding tumor. Sp, septum pellucidum bulging to the right as the left foramen of Monro is blocked. Pl, Choroid plexus. B. The septum is opened. Both foramina of Monro are visualized obstructed by the tumor. C. Schematic representation of the transcallosal-transseptal approach to both foramina of Monro.

inspect the lateral sulcus, beneath the cingulate gyrus, for accessory A2 vessels. Normally, our dissection proceeds by approaching the corpus callosum between the two pericallosal arteries; however, in some cases, anatomical variations will force us to remain lateral to the right pericallosal artery. To avoid potential mechanical vasospasm during dissection, cottonoids soaked with papaverine or phenytoin should be applied locally (Fig. 19.5). Using fine bipolar forceps, one makes a 10- to 15-mm opening into the corpus callosum. This allows free flow of CSF and decompression of the right ventricle. Unless there are natural openings in the septum pellucidum, the CSF of the left ventricle will remain unvented and the septum pellucidum will bulge to the right, obscuring the view of the surgeon. A 10-mm fenestrating incision in the septum pellucidum would decompress the left ventricle, giving the surgeon good access to both foramina of Monro (Fig. 19.6). Dissection of the Tumor at the Foramina of Monro and in the Third Ventricle The foraminal region should first be inspected to ascertain the course of the veins running toward the internal cerebral veins. In particular, one should identify the septal, frontal subependymal, thalamostriate, and smaller posteriorly to anteriorly directed veins, all of which run to the posterior inferior corner of the foramen of Monro and are usually covered by Choroid plexus. The foramen is usually grossly enlarged and it has not been found necessary to mobilize the Choroid plexus and tela choroidea (Fig. 19.7A).

Figure 19.7. A. Dissection of the tumor. B. Third ventricle visualized through the enlarged foramen of Monro after tumor removal.

Dissection of the tumor then proceeds bilaterally within the foramina from the posterior inferior corners and then further into the ventricles bilaterally. The tumor is punctured and a central decompression is carried out. This permits a deeper inspection and the massa intermedia may be seen (superiorly and posteriorly compressed and displaced), together with the origin of the aqueduct. At this stage it is most important to establish, if possible, a good plane between normal and tumoral tissue to prevent damage to the periaqueductal structures and the delicate lateral walls of the dilated third ventricle (Fig. 19.7B). The infundibulum and subchiasmatic region may now be inspected and the tumor dissected from these areas. This is generally quite safe in the midline; however, laterally care must be exercised to avoid damage to the optic tracts. Deeper dissection reveals the infundibulum, which may be membranous or actually breached by the tumor. Careful exploration is necessary here as one quickly approaches the basilar bifurcation and mamillary bodies; separation of the tumor from these structures is not difficult as there are few arachnoidal adhesions present (see Fig. 11 A). Within the interpeduncular cistern there is rarely serious bleeding from damage to small arteries during dissection. However, rupture of some of the venous branches may produce small hemorrhages that may be difficult to control by simple coagulation. In such instances, small pieces of muscle held in place for a few minutes with a moist sponge will usually be more effective in this sensitive area than excessive use of bipolar coagulation. The anterosuperior aspect of the third ventricle may now be inspected cautiously as this is where lie the A1 segments, anterior communicating arteries, hypothalamic vessels, and optic chiasm. If a large segment of tumor is still present and the anatomy is unclear, the midline approach is discontinued and the dissection is carried further from the frontobasal direction. In general, there is a relatively well-defined cleavage plane between the wall of the third ventricle and the tumor itself (at least in cases of craniopharyngioma). However, careful inspection and early orientation are mandatory to avoid damage to the ventricular wall. This may occasionally be complicated by the fact that the tumor may not lie entirely within the ventricle—a fact not easily established on CT scan. The mass may displace the ventricle to the right or to the left, indent it from below symmetrically or asymmetrically, or lie within the third ventricle (see Fig. 19.2). Pterional Approach This standard approach has been described in detail elsewhere (1) and only an outline is given here. After removal of the bone flap, the tension of the intracranial compartment can be assessed and if necessary, through a small dural rent, CSF may be released by puncturing the ventricle with a brain needle. One or two small dural rents are also cut along the frontal or temporal portions of the planned dural incision to provide further CSF release. Pterional craniotomy is designed to take advantage of the natural planes and spaces and to expose the base of the brain without significant brain retraction. One natural plane is provided by the sphenoid ridge as it separates the frontal and temporal lobes. Another is provided by the roof of the orbit projecting superiorly and indenting the basal surface of the frontal lobe. These planes project from the surface of the brain to the parasellar area and form the base of this pyramidal space, the apex of which is formed by the junction of the frontal and temporal lobes. The

Figure 19.8. A. Right-sided pterional approach showing an anteriorly fixed chiasm. The lamina terminalis is bulging and the tumor adheres to the anterior communicating and hypothalamic arteries. B. The lamina terminalis is opened and the superoanterior portion of the tumor has been dissected. base of this pyramid may be enlarged by removal of bone from the sphenoid ridge and by flattening of the orbital roof. The apex of the pyramid may be enlarged by opening the basal sylvian fissure, thus forming a further pyramid-shaped working space with the apex directed toward the limen insulae. The width of this pyramid is the shortest distance from the calvarium to the sella. The base of the pyramid is enlarged by using a high speed electric drill to smooth the convolutions of the posterior lateral orbital roof adjacent to the frontal sphenoid suture. The posterior ridge of the greater wing of the sphenoid is also flattened until a small ridge representing the most lateral aspect of the lesser wing is reached. Any bleeding from the bone is readily controlled by coagulation and the application of bone wax. If the frontal sinus extends far laterally, it may be inadvertently entered, in which case the mucosa is stripped from the bone and the sinus is plugged with a small piece of muscle and sealed with an acrylic adhesive before being covered with bone wax at the end of the procedure. The dura is opened in a semicircular fashion around the sylvian fissure, arching toward the sphenoid ridge, and is retracted. The direction and sequence of arachnoid dissection and Cisternal opening depend to some degree on the location of the lesion. Both the sylvian cistern between the basal frontal and temporal lobes and the lamina terminalis cistern are opened. The apex of the pyramid is enlarged using sharp dissection with a round arachnoid knife, and the sylvian cistern is entered at the level of the opercular frontal gyrus. The arachnoid of the sylvian cistern should be opened on the frontal side of the superficial middle cerebral veins.

Occasionally two or three frontoorbital venous tributaries crossing the sylvian fissue must be sacrificed to complete the dissection of the most medial aspect of the fissure. The sylvian cistern is then followed laterally around the orbital frontal gyrus, at which stage the frontal lobe will be retracted away from the temporal lobe. The right carotid, the lamina terminalis, and the interpeduncular cisterns are opened to release more CSF and to expose the base of the tumor. Although the approach to midline tumors is generally more frontomedially directed than for aneurysm surgery, the concept of "enlarging the base of the pyramid" by flattening the posterolateral orbital roof remains useful in minimizing retraction of the frontal lobe and in gaining maximal access to the parachiasmatic region (Fig. 19.8). Summary The combined approach provides a systematic method for attacking and excising these large Craniopharyngiomas. The initial step is lateral chiasmatic and subfrontal evaluation via the pterional dural opening. The anatomical relationships of the tumor to the optic nerve and tract, the internal carotid artery and its branches, the chiasm, and the anterior cerebral vessels are determined by direct inspection. Tumor removal via the subfrontal pterional approach is not undertaken until after the trans-

Figure 19.9. A to C. CT scan of an 11-year-old boy with craniopharyngioma occupying the third ventricle. D. Coronal CT showing extension of the tumor into the third ventricle and up to the foramina of Monro. Note bilateral hydrocephalus.

Figure 19.10. A. Anteriorly fixed chiasm visualized through the right pterional approach. B. Tumor visualized through the foramen of Monro using the transcallosal approach.

Figure 19.11. A. Basilar artery and its branches are now visualized through the opened infundibulum of the third ventricle after tumor removal. B. The anterosuperior portion of the tumor is removed through an opening of the lamina terminalis.

Figure 19.12. A to C. Postoperative axial CT scan showing radical removal of the tumor. D. Coronal CT scan showing normal size ventricles and patent foramina of Monro. callosal intraventricular approach to the tumor is completed. Frequently the basilar cisterns are blocked by the presence of these large tumors, compounding the hydrocephalus produced by obstruction of the foramen of Monro. Opening of the superficial cisterns, frontal and temporal, associated with the sylvian fissure provides escape for CSF otherwise blocked at the basal cisterns. The next step is to proceed with the paramedian dural exposure and transcallosal approach to the third ventricle. Upon opening into the lateral ventricles after the placement of a septal window to control hydrocephalus, one can proceed with a direct attack on the superior and superoposterior portions of the tumor, which is the major advantage of the combined approach. After the superior and superoposterior aspects of the tumor have been dissected and removed, the anatomical relationships of the foramen of Monro, origin of the aqueduct, and lateral and

basal aspects of the third ventricle are established, and hemostasis is completed, one reinspects the subfrontal area. During this step, one removes residual tumor from the retrochiasmatic area, interpeduncular cisterns, and optic tracts and nerves. Once one is satisfied with tumor removal and hemostasis is complete one may then proceed with the fourth and final step, the return to the transcallosal transventricular tumor bed to assure that no residual tumor is present and that hemostasis is absolute (Figs. 19.9A to C, and 19.10-19.12). Following these four steps of the combined transcallosal-pterional approach provides the most detailed evaluation of the subfrontal and intraventricular anatomy and simultaneously assures control of hydrocephalus. Tumors of this origin require precise anatomical evaluation, meticulous dissection, and absolute hemostasis, all provided by this combined approach. Reference 1. Yasargil MG: Microneurosurgery, vol I, Microsurgical Anatomy of the Basal Cisterns and Vessels of the Brain, Diagnostic Studies, General Operative Techniques and Pathological Considerations of the Intracranial Aneurysms. Stuttgart, Georg Thieme Verlag, 1984, p 371.

20 Transnasal Transsphenoidal Approach Martin H. Weiss, M.D.

Tumors of the floor of the cranial vault, particularly those arising from and extending along the midline, frequently impact upon the third ventricle as they extend superiorly into the intracranial compartment. These masses (whether they be neoplastic, inflammatory, or cystic) may derive from intradural or extradural sources as well as extracranial processes that extend in a superior direction. Efficient extraction of these lesions can frequently be accomplished most effectively by attacking them from the site of origin inferior to the third ventricle utilizing a transnasal procedure, which can be tailored to involve either intradural or extradural masses. Historical Perspective Schloffer, a rhinologist from Innsbruck, Austria, first recommended utilization of the transnasal approach to intracranial neoplasms and reported the successful removal of a pituitary tumor utilizing this approach in 1907 (5). Subsequent modifications by a number of international as well as American neurosurgeons enabled Cushing to report a 15-year experience of some 231 transnasal operations upon pituitary tumors with a mortality rate of 5.6%. This was significantly less than the early mortality figures for transcranial approaches to the area of the sella; the primary sources of morbidity lay with the difficulty in irradi-cating cerebrospinal fluid (CSF) fistulas developing during the course of the procedure and the occurrence of meningitis as a consequence of persistent CSF fistulas. As our results in the use of transcranial approaches to the sella and the parasellar area dramatically improved, interest in the transnasal approach to skull base lesions waned, particularly in the United States, even though our European colleagues effectively persisted in these pursuits. Advances in microsurgical techniques rekindled the North American interest in transnasal surgery led by the pioneering efforts of Jules Hardy in his promulgation of the technique of transsphenoidal hypophysectomy (3); our early successes with hypophysectomy of the normal gland in the treatment of disseminated breast and prostate cancer as well as diabetic retinopathy encouraged extension of

this technique to pituitary tumors as well as parasellar lesions with significant degrees of intracranial extension. The mortality rate in experienced hands with such approaches consistently is significantly less than 1 % with postoperative radiographic confirmation of radical resection of extensive lesions (2, 4). These considerations have led to permanent inclusion of this technique in the neurosurgical armamentarium with continued expansion of the indications for utilization of transnasal approaches to the base of the skull (1-3). General Guidelines At present, lesions that optimally lend themselves to transnasal resection are midline lesions of the skull base extending from the inferior aspect of the clivus (foramen magnum) to the level of the planum sphen-oidale whether they be extracranial, extradural, or intradural (Fig. 20.1). Varying degrees of lateral extension from the midline may lend themselves to inclusion in these operative approaches depending upon the experience of the surgeon and the broad base upon which such extension has arisen. It has become commonplace to utilize combined procedures to approach extensive lesions of the skull base that broadly involve the midline yet have considerable and inaccessible lateral extension. In general, our strategy has been to utilize a transnasal approach to these lesions initially to remove as much as possible of the tumor mass via what we think is a significantly more benign procedure; residual lateral tumor can then be approached via a transcranial intradural procedure in which tumor dissection will have been minimized by the previous transnasal procedure. Utilizing these criteria we have successfully approached lesions arising within the sphenoid sinus such as sphenoid sinus muco-celes, lesions of the clivus such as chordomas, lesions of the sella such as pituitary tumors and various intrasellar cysts, and lesions arising within the suprasellar cistern such as Craniopharyngiomas and dysger-minomas. There are relatively few contraindications to the implementation of this technique other than those already mentioned with specific respect to extensive lateral extension. The presence of a nonpneumatized sphenoid sinus complicates the surgical procedure and may require more extensive utilization of radiographic conformation of position intraoperatively, but the availability of appropriately sized high speed drills allows one to obviate this problem with relative ease. The position of the intracaver-nous carotid artery, in a similar fashion, adds a complicating dimension with respect to access to structures or lesions lying superior to the carotid arteries. Indeed, major ectatic carotid arteries may mitigate against a successful approach to a sellar or a suprasellar lesion although this is an extremely rare finding. The presence of acute sinusitis is an obvious temporary limiting factor that may delay but not prohibit the utilization of a transnasal approach. In general, if the bulk of the lesion lies within the midline with a modest or moderate degree of lateral extension, one would elect to approach these lesions transnasally. However, if the vast majority of the lesion lies lateral to the midline with a small degree of midline extension and invasion, one would utilize a standard transcranial approach because it would be more likely to allow for radical tumor resection. A previous rhinoplasty or submucosal resection for nasal septal deviation will also add to the complexity of the intranasal dissection because of postoperative changes in the relationship between the nasal mucosa and its adjacent structures, but appropriate dissection planes can invariably be established even under such circumstances.

Figure 20.1. Suprasellar extension arising from a broad-based mass may be easily approached via a transnasal transsphenoidal route. The lateral or anterior extension of a mass that has a small or narrow basal attachment would limit access to the contents of the tumor and is a relative contraindication to approaching via the transnasal route. Approaches to intradural lesions offer the greatest technical challenge. Frequently, such lesions extensively embrace the major vascular structures that cannot be readily appreciated during the course of transnasal surgery. It has been my experience that invasion of the cavernous sinus is best approached transnasally when such invasion has occurred from a midline lesion. The exposure of the entire sella via the transnasal route enables access to the extension of the tumor into the cavernous sinus well beyond that which can be generated via a transcranial approach. On the other hand, in those cases in which a tumor incorporates the carotid or basilar arteries within the substrate of the tumor mass, one may not visualize the relationship of the tumor and such vessels via a transnasal approach; one would therefore incline one's operative strategy toward a transcranial approach under such circumstances. Radiographic Considerations Preoperative assessment of these lesions requires meticulous radiographic definition of the extent of the lesion so that one can apply the selection criteria. In addition, however, one must recognize the frequent involvement of the hypothalamichypophysial axis by lesions so positioned, which impacts upon the risk factors. Any lesion proximal to the pituitary or hypothalamus requires adequate endocrine evaluation preoperatively to minimize the potential for an intraoperative or postoperative catastrophe because of inadequate pituitary reserve. The two most

important hormone axes that must be considered are those relating to cortisol and thyroid release from their respective organs. Because many of these procedures call for the concomitant utilization of exogenous glucocorticoids, the risk of intraoperative hypocortisolemia is generally not a major factor. However, preexistent hypothyroidism may become acutely manifest during the early postoperative period, making preoperative recognition of this potential an essential consideration. Adequate reestablishment of a euthyroid state requires approximately a week of treatment before surgical procedures can be electively pursued. It has been our policy to evaluate a complete thyroid profile as well as a baseline cortisol level preoperatively in anticipation of intraoperative manipulation of the hypothalamic-hypophysial axis. The existence of normal hormonal activity preoperatively obviously does not guarantee normalcy postoperatively, but a reasonable understanding of preoperative pituitary function can help to mitigate against particular perioperative complications. Radiographic evaluation of parasellar lesions has been radically affected by the advent of high resolution computerized tomographic (CT) scanning. Utilization of varying window widths enables definition of bony landmarks (whether or not these are pathologically involved) and of the extent of soft tissue extension and related vascular distortion. CT scanning with a high resolution scanner utilizing contrast enhancement along with coronal and sagittal reconstructions has enabled us to visualize the peritumoral vasculature to the extent that preoperative angiography is rarely required. In those cases in which questions of the existence of an aneurysm or vascular invasion by tumor are of concern, intravenous digital subtraction angiography has frequently proved to be an excellent alternative to selective angiography. Cystic lesions that involve the suprasellar cistern may defy adequate delineation by any of the aforementioned means. Positive contrast CT cisternography has been a helpful resource in outlining such lesions. Recently, magnetic resonance imaging (MRI) has allowed us to delineate the extent of such lesions. MRI may, in the reasonably near future, replace CT scanning as the primary or definitive imaging modality for such skull base lesions. The availability of these advanced imaging forms has essentially eliminated the need for conventional polytomographic studies in which visualization is limited to the definition of bony structures, obviously of limited use. Surgical Procedures Considerations of operative technique begin with the induction of general anesthesia. When a patient is defined as having compromised neurological function, most commonly visual compromise due to tumoral compression, or when preexistent hypopituitarism is defined, it is essential to administer adequate glucocorticoids before induction to minimize the potential for an undesirable preoperative hemodynamic crisis or intraoperative progression of neurological compromise. Patients with tumors of sufficient extent to compress neural structures are given sizable doses of high potency glucocorticoids (40 mg of methylpredniso-lone or 10 mg of dexamethasone) before anesthesia induction. The placement of the endotracheal tube with respect to the mouth itself is a significant consideration (Fig. 20.2A). The tube is best located slightly to the left of the patient's midline so as not to put untoward traction on the lips and therefore mitigate against satisfactory elevation intraoperatively. It is certainly reasonable to include an esophageal stethoscope along with the endotracheal tube to monitor pulmonary ventilation intraoperatively; it is essential to assess the status of the endotracheal balloon to evaluate

Figure 20.2. A. The endotracheal tube should be placed to the left of the midline so as to allow access easily from the patient's right side but not so far as to cause undue traction on the lips and therefore impede elevation of the lip during the course of the procedure. B. The patient is positioned in a supine position with the face parallel to the floor and the head elevated approximately 15° above the heart. C. In the final position, the surgeon (1) stands just to the right of the patient with the Mayo stand (2), the nurse (3), the microscope base (4), and the assistant (5) around the head of the patient. The anesthesiologist (6) stands to the left of the patient. The patient's right posterolateral thigh is approximately positioned to allow access should a fascia lata graft be required. the potential existence of any leaks that might allow intraoperative aspiration of blood, mucus, or irrigating fluid accumulating in the posterior oropharynx during the course of the procedure. The vast majority of our operations are done under balanced anesthesia utilizing nitrous oxide and narcotic, which allows us to awaken patients

Figure 20.3. Slight flexion of the head allows easiest visualization of the structures that lie caudal to the sella turcica whereas extension allows better visualization of those structures lying anterior to the sella turcica (particularly the planum sphenoidale and tuberculum sellae). For most situations, a neutral position (head parallel to the floor) is optimal for access to the sella and suprasellar structures. rapidly postoperatively to assess their visual and neurological status. Occasionally, inhalation anesthesia is required, particularly to control blood pressure when there is a tendency to hypertension. In normotensive patients, we prefer to keep the blood pressure at approximately a systolic of 100 mm Hg (a mean of 70 to 80 mm Hg) to reduce venous ooze from the operative site. Once anesthesia has been satisfactorily induced, we generally prefer to position the patient supine with the face parallel to the ceiling with the head elevated approximately 15° above the heart (Fig. 20.2, В and C). Flexion or extension of the head from the position just mentioned can help in approaching lesions either anterior or posterior to the sella turcica (Fig. 20.3). With the head parallel to the ceiling, those intra- and suprasellar tumors most commonly encountered may be most readily approached. However, if there is significant extension inferiorly along the clivus, visualization is enhanced by some flexion from this neutral position, whereas subfrontal extension of the tumor to a level of the tuberculum or planum may be best approached with extension of the head from this neutral position. One must also allow for exposure of the right superior posterolateral thigh or the anterior wall of the abdomen to gain access for a fascial or subcutaneous fat graft should that be necessary during the course of the procedure. Considerations of intracranial pressure control during this procedure generally relate to assurance that venous drainage will be adequate to

Figure 20.4. A gingival incision leaving approximately 3 mm of a gingival cuff for closure is made extending from the canine fossa to the contralateral canine fossa and is carried down to the level of the subperiosteal plane of the maxilla. The dissection is carried superiorly in a subperiosteal fashion to expose the upper aspect of the rostrum of the maxilla, including the nasal spine and the lateral recesses. If there should be a particularly prominent nasal spine or markedly heightened maxillary rostrum a resection of these will enable better dissection along the floor and along the mesial surface, avoiding any undue visual obstruction. Once the floor of the nose is readily identified, the mucosa of the nose is dissected off the floor along its mesial aspect, with the dissection then carried mesially up along the vomer and over the vomer to dissect the mucosa from the quadrangular cartilage. The dissection is carried superiorly along one side to free the mucosa from the quadrangular cartilage so that it can be visualized posteriorly to recognize the attachment of the quadrangular cartilage to both the vomer inferiorly and the perpendicular plate of the ethmoid posteriorly. avoid venous congestion of the operative site. This is generally assured by positioning the head above the heart to encourage venous drainage. Intracranial hypertension is generally not a major problem in these cases except with lesions that grow sufficiently superior to obstruct the fora-

Figure 20.5. Once the quadrangular cartilage is adequately visualized, it can be mobilized from its attachment to both the perpendicular plate of the ethmoid and the vomer so as to allow mobilization of the quadrangular cartilage to the side with the overlying attached nasal mucosa. This allows visualization of the perpendicular plate of the ethmoid and the superior aspect of the vomer for further dissection. Once the quadrangular cartilage has been mobilized from the vomer and the perpendicular plate of the ethmoid, a dissection is carried posteriorly along the perpendicular plate of the ethmoid and the posterior-superior aspect of the vomer to the rostrum of the sphenoid sinus. Identification of the sphenoid sinus ostia will indicate the superior aspect of the rostrum of the sphenoid sinus so as to limit the extent of the section and also allow access to the sphenoid sinus. mina of Monro, generating secondary hydrocephalus. Under such circumstances, reduction of intracranial pressure before the induction of anesthesia is generally desirable. We have utilized direct ventricular cannu-lation (either unilateral or bilateral depending upon interventricular communication) with removal of ventricular fluid before induction to control such intracranial hypertension. Utilization of dehydrating agents intraoperatively is rarely necessary or desirable, obviating the need for placement of a Foley catheter with its potential attendant morbidity. This operative procedure is performed through a "semisterile" field. As a consequence, it is imperative to irrigate the operative field vigorously during the course of operation to reduce the potential for contamination. A number of authors have advocated the utilization of preoperative antibiotics to reduce the risk of postoperative infection due to the semi-sterile nature of the operative field. We abandoned the use of such agents some 12 years ago with no apparent untoward effect. Our infection rate remains under 1% in our operative series. We do, however, irrigate vigorously with bacitracin-impregnated lactated Ringer's solution (50,000 units/500 ml). We think that this is the single most critical factor in reducing the potential for operative contamination from the paranasal sinuses.

Figure 20.6. One can gain access to the entire extent of the clivus by directing the dissection inferiorly along the vomer and the roof of the hard palate with dissection of the mucosa of the nasopharynx from the clivus to gain access to lesions as low as the foramen magnum. In this same manner a lesion lying over the tuberculum sellae and even the planum sphenoidale can be approached by carrying the dissection anterior to the sella turcica after the rostrum of the sphenoid has been resected. Once the patient is adequately positioned after the satisfactory induction of endotracheal anesthesia, a povidone-iodine preparation of the nasopharynx and sublabial gingival area of the maxilla is carried out to reduce potential contamination from these sources. The nasal mucosa and gingival mucosa are infiltrated with 0.5% lidocaine containing epinephrine 1/200,000 and the nose is packed with cottonoids impregnated with 5% cocaine to shrink the nasal turbinates. A gingival incision extending from canine fossa to canine fossa is then made and carried down to expose the rostrum of the maxilla (Fig. 20.4). The dissection should be carried in a subperiosteal fashion to minimize the amount of blood loss and superiorly to identify the entrance to the nose. A great variability has been encountered in the size of the nasal spine, which may or may not require resection superiorly to allow adequate visualization of the posterior aspect of the nasal pharynx. In addition, the superior aspect of the rostrum of the maxilla is variable in height from the floor of the nose and may require some resection inferiorly and laterally to increase the size of the nasal cavity and allow smooth and direct dissection of the nasal mucosa from the floor and mesial structures. Once this preliminary portion of dissection is completed, one must remember to remove the nasal cottonoids before proceeding with the intranasal dissection. The mucosa is first dissected from the floor of the nasal canal along its mesial aspect and then dissection is carried mesially and superiorly up along the vomer to the junction with the quadrangular cartilage. There are multiple adhesive bands extending from the basal mucosa to this junction, which must be meticulously dissected to avoid a tear in the nasal mucosa. The dissection is then carried superiorly up along the inferior aspect of the quadrangular cartilage, coming anteriorly to expose the columella at its inferior portion (Fig. 20.5). The nasal cartilage can then be dissected free from the superior aspect of the vomer along with the anterior aspect of the perpendicular plate of the ethmoid so that the cartilage can be observed as it is mobilized away from the midline with

Figure 20.7. In В one can appreciate resection of the rostrum of the sphenoid sinus utilizing the sphenoid ostia to gain access to the sphenoid sinus with the punch. If the sphenoid sinus is nonpneumatized or only partially pneumatized, one can use a high speed drill to drill through the bone to gain access to the sella turcica. Under such circumstances, radiographic or fluoroscopic control of the depth of the drilling dissection is probably desirable because it may be difficult, under these unusual circumstances, to appreciate the extent of drilling that has occurred. the nasal mucosa to which it remains attached on one side. This allows visualization of the vomer and the perpendicular plate of the ethmoid so that the dissection can then be carried posteriorly to identify the rostrum of the sphenoid sinus. The perpendicular plate of the ethmoid is an invaluable landmark in that it has never failed in guiding this surgeon to the rostrum of the sphenoid sinus. The dissection at this point will be dependent upon the area of particular involvement that one is addressing. In the unusual situation in which a chordoma of the clivus extending as far down as the foramen magnum is under surgical attack, one would want to dissect the mucosa of the nasopharynx from the inferior posterior wall of the clivus to gain access to this retrosphenoid structure. One would follow along the floor of the nose, utilizing the landmarks of the hard and soft palate to establish the plane between the mucosa and the clivus to gain access to this structure (Fig. 20.6). Ordinarily, one would utilize the rostrum of the sphenoid as a landmark to gain access to the sphenoid sinus once both of the sphenoid sinus ostia had been identified. In the usual course of events, where a suprasellar cistern tumor with an enlarged sella turcica is compressing the third ventricle, the sphenoid sinus ostia mark the anterior extent of the exposure with respect to the sphenoid sinus that will enable adequate access to the suprasellar cistern (Fig. 20.7). The rostrum of the sphenoid sinus can be readily removed, gaining access to its interior via the ostia. Should one be dealing with a sphenoid sinus mucocele, radical exenteration of the mucocele along with any remaining mucosa of the sphenoid sinus will allow one to complete

Figure 20.8. Once the rostrum of the sphenoid sinus has been resected to allow visualization of the full extent of the sphenoid, the sphenoid sinus mucosa should be exenterated, which will help control any bleeding from this mucosa as well as avoiding the potential for a postoperative sphenoid sinus mucocele. One must also note that there is almost always a bony septum within the sphenoid sinus that also must be removed to allow access to the entire extent of the floor of the sella turcica. Frequently, this septum is somewhat eccentric and may not appear to be present to the untrained eye. Dissection along one side of the septum only will give compromised exposure of the floor of the sella with resultant compromise to the extent of resection of the intra- or suprasellar contents. the procedure at this point. Should the abnormality lie superior to the sphenoid sinus, one should exenterate the sphenoid sinus mucosa to prevent the genesis of a postoperative mucocele along with any undue bleeding from residual sphenoid sinus mucosa (Fig. 20.8). If one is dealing with a chordoma of the clivus, it is imperative to recognize that these are frequently if not always surrounded by a layer of cortical bone, which must be drilled off before ready access to the intraclival chordoma can be secured. One can recognize the chordoma capsule after resection of

Figure 20.9. Clivus chordomas are generally found overlying a layer of cortical bone which must be drilled off to gain exposure to the capsule. These structures (remnants of the primitive notochord) reside within the confines of the clivus in their original phase. Once the capsule of the chordoma has been opened, one may readily remove the moderately vascular and gelatinous internal contents of the chordoma. Utilizing this technique, one can then progressively infold the capsule of the tumor so as to separate it from its surrounding structures to gain access to the extracapsular space to enable a radical resection. One must always remember that bone destruction is the hallmark of the chordoma; it may either penetrate or intimately attach to the dura, preventing potential complete resection utilizing these techniques. the clival cortical bone; radical resection is then most readily accomplished by removal of the central portion of the gelatinous chordoma and then folding in of the capsule (Fig. 20.9). These tumors are in general extradural tumors, which allow one to gauge the extent of resection by identification of the dural margins. Limiting oneself to the extradural compartment allows resection of the tumor while avoiding the potential for surgical damage to the basilar artery or disruption of the arachnoid, necessitating a grafting procedure to avoid a postoperative CSF fistula.

Figure 20.10. Once the floor of the sella has been removed so as to identify at each lateral boundary the cavernous sinus and the circular sinus superiorly (an opening of approximately 1.5 cm square), the dura is opened in a cruciate fashion so as to elevate four dural flaps that will give maximal surface area of exposure in the form of a square. In the unusual event that the dura mater is invaded, one may settle for an intracapsular removal of tumor, allowing the capsule to occlude the opening in the dura mater, or one may incise the dural margins surrounding the capsule longitudinally in an effort to extract the intradural portion of the capsule. When possible under these circumstances, preservation of the arachnoid will mitigate against the potential for a CSF fistula even though the dura mater has been breached. For those lesions arising within the sella turcica and extending superiorly it is necessary to resect the floor of the sella before gaining access to the primary lesion. If the floor has been markedly thinned by tumor expansion, one may simply chip off a significant segment of the floor using a long small bone curette; a thick or normal sellar floor can be easily opened using a high speed diamond drill while preserving the integrity of the overlying dura. When working within and beyond the sella turcica, it is easiest to utilize a cruciate incision when opening the dura so that, after elevating the dural flaps, one is left with a square opening as opposed to a diamond-shaped opening. A square shape maximizes the area of the dural opening (Fig. 20.10). Identification of the arachnoid, once again, is of critical importance in an effort to avoid a CSF fistula (Fig. 20.11). The arachnoid is actually a reasonably tough membrane which is, even in the face of major suprasellar extension, generally intact so as to allow resection of the tumor without disturbance of the overlying arachnoid. Lesions arising primarily within the suprasellar cistern, such as a craniopharyngioma arising from the pituitary stalk and growing superi-

Figure 20.11. Once the dural flaps have been elevated, access to the tumor both within the confines of the sella and in the suprasellar cistern can be readily obtained utilizing various curettes and scoops. The critical relationship between the tumor, diaphragma sellae, and suprasellar arachnoid must be recognized because it is imperative to preserve the integrity of the arachnoid to minimize or virtually eliminate the potential for a postoperative CSF fistula.

Figure 20.12. A pure suprasellar tumor may be approached by carrying the bony resection anteriorly over the tuberculum sellae with exposure of the dura mater lying anterior to the circular sinus. A transverse incision is then made in the dura rostrum and caudal to the circular sinus with coagulation of the circular sinus to allow transection of the circular sinus and exposure of the suprasellar cistern without disturbing the pituitary. orly, present a particularly interesting problem. In these cases, the sella turcica is usually of normal size and one strives to preserve a normal pituitary gland while resecting the suprasellar lesion. Under such circumstances, one optimally must resect not only the floor of the sella turcica but also extend the bony resection superiorly to remove the tuberculum sellae, which can be visualized through the sphenoid sinus opening (Fig. 20.12). This will generally require crossing the circular sinus at the anterior lip of the sella turcica, which generally can be readily accomplished. The sinus is an intradural structure contained between the leaves of the dura so that a pure extradural resection of bone should not compromise sinus integrity. Once the bone of the tuberculum sellae has

Figure 20.13. If a CSF fistula is recognized at operation, one closes this fistulous component with a fascia lata graft. The fascia is placed against the diaphragma sellae within the intrasellar compartment. This is then held in position with a section of quadrangular cartilage if that is readily available or, more recently, a small piece of Marlex mesh. Fat taken from the subcutaneous tissue at the time of the fascial resection is utilized to fill the sphenoid and hold the graft material in position. been removed one would open the dura anterior to the circular sinus as well as inferior to the circular sinus (along the floor of the sella) to isolate the circular sinus. One would then coagulate and transect this sinus to gain direct visualization of the suprasellar cistern, preserving the pituitary in its position. Under these circumstances, it is frequently if not always necessary to open the arachnoid intentionally to drain out CSF so that one may adequately visualize the suprasellar compartment. Once the capsule of the craniopharyngioma is recognized, a needle may be inserted to drain any cystic fluid. The contents can then be resected piecemeal with progressive infolding of the capsule of the craniopharyngioma or other cystic lesions to effect a radical removal of tumor. This

Figure 20.14. Once the retractor has been removed, the nasal mucosa is approximated and tamponaded by placing a posterior and anterior nasal pack of Vaseline gauze. The posterior pack is placed up against the rostrum of the sphenoid to hold the fat graft in position when this is required. The gingival mucosa, which heals rapidly and readily, is approximated with three inverted knot sutures. same technique obviously allows for resection of those suprasellar tumors that extend anteriorly over the tuberculum sellae and defy extraction via a resection solely of the sellar floor. Once the mass has been resected, attention must be paid to the need for obliteration of the CSF fistula if one has been created (Fig. 20.13). Under such circumstances, we prefer to use a fascial graft constructed from a small segment of fascia lata. The graft is placed on the intradural side of the opening and then is held in position by a section of cartilage wedged into the bony opening to hold the dura mater against the graft. The sphenoid opening or the opening behind this graft is then filled with fat taken from the site of the graft to buttress the graft into position. After removal of the self-retraining retractor, the nasal mucosa is approximated and tamponaded by utilization of bilateral posterior and anterior nasal packs (Fig. 20.14). The posterior nasal pack is placed up against the sphenoid opening to prevent migration of the fat that may have been placed in the sphenoid. If a graft is not necessary, strips of Surgicel (absorbable hemostat; Johnson & Johnson, New Brunswick, New

Jersey) are placed down along the tumor bed, and the sphenoid sinus is not filled with a fat graft. Approximation of the gingival mucosa before insertion of the nasal packs is accomplished with three plain catgut sutures that create relative approximation of the mucosal margins without an attempt at hemostasis, which is accomplished by the insertion of the nasal packs. Pitfalls As with any operative procedure, anatomical variations or intraoperative misadventures may occur. With the exception of the aforementioned variability in the size of the nasal spine and height of the vomer and rostrum of the maxilla, most anatomical variations can be anticipated on the basis of the preoperative roentgenograms. There are a number of points during the course of the dissection that may allow an individual to go astray. The first of these most commonly occurs during the dissection of the nasal mucosa from the vomer, particularly in its transition to the quadrangular cartilage. Because of dense adhesions at this level, it is easy to tear through the nasal mucosa, resulting in entrance into the nasal cavity. A small tear, particularly one that is unilateral, can be well tolerated and not prevent one from continuing with a meticulous dissection to accomplish one's objectives. However, bilateral tears, particularly with resection of the quadrangular cartilage when that becomes necessary, can leave the individual with a large nasal perforation that can be quite symptomatic with respect to respiratory noises and chronic nasal drainage. Beyond the mucosal dissection, one must be concerned about the relationship of the parasellar vascular structures to any sellar and suprasellar lesion. Invasion of the cavernous sinus is a commonplace event under such circumstances, and one may frequently enter the cavernous sinus to resect such invasive tumor without any undue or excessive morbidity. Deviations in the course of the carotid arteries, however, can be of grave consequence; one must exert extreme caution when opening the dura mater to make sure that an ectatic carotid artery does not underlie the sellar dura mater. I have seen both carotid arteries touching within the midline, emerging from the cavernous sinus as they traverse that structure. A graft that is not adequately secured to obliterate a fistulous opening in the subarachnoid space will not inhibit the genesis of a postoperative CSF leak. A good rule of thumb is to perform a Valsalva maneuver after insertion of the graft to assess the integrity of the graft. There should be no CSF transgressing the graft. Finally, inadequate insertion of nasal packs does not allow for adequate tamponade of the nasal mucosa to inhibit postoperative bleeding from this site. Such bleeding can occur in a retrograde fashion, resulting in the genesis of a suprasellar postoperative clot, or may emerge through the pack or gingival opening, causing persistent bleeding through the gingiva. Under the latter circumstances, reinforcement of the nasal pack will usually result in sufficient tamponade. When retrograde hemorrhage occurs, however, it is necessary to reexplore the operative site, extract any accumulated clot, and repack the nasal mucosa. If, during the course of the operative procedure, there is excessive bleeding from within the cavernous sinus, this can be readily controlled by the application of a Surgicel pack against the bleeding site. The pack is held in position for a good 10 to 15 minutes to tamponade the bleeding site. Inadvertent entrance into the carotid artery at this level presents a much more serious problem. This would call for the insertion of a permanent pack to tamponade the bleeding site sufficiently while at the

same time avoiding excess compression of the artery, which could result in carotid stenosis or occlusion. Such a pack can be temporarily fashioned from Surgicel while a fascial or muscle graft is obtained from the preoperatively prepared leg. This graft would be inserted directly against the bleeding site and then be held in position by a fat graft that would come out through and fill the sphenoid sinus. Complications The above pitfalls and potential complications, although distinctly unusual, are relatively specific to this technical approach. There are, however, additional potential complications that relate to the neural structures involved with tumor progression. These potential complications are, of course, inherent in any surgical procedure that undertakes decompression of neural structures from such masses. If one should extend beyond the confines of the arachnoid in extracting one of these masses via the transnasal route, there is an obvious potential for entanglement of optic nerves, suprasellar carotid artery, hypothalamus, and pituitary stalk. Interestingly, a number of sizable series in addition to ours have demonstrated that the likelihood of such complicating factors seems significantly less with the transnasal route rather than the transcranial route. The incidence of visual recovery, for instance, seems to be at least as good if not better with transnasal resection of suprasellar tumors as compared to the transcranial approach. When it is necessary to breach the arachnoid utilizing the transnasal approach, extreme caution must be exercised in assessing the extent of available tumor for resection utilizing this technique. Any undue stress on adjacent structures and on tumor capsule is to be avoided during the course of resection. If one remains confined to the extraarachnoidal compartment, the potential for such complications is virtually nonexistent. It has become evident over the past 2 decades that compressive lesions of the third ventricle arising from the suprasellar cistern or structures inferior to this may be resected effectively and safely via a transnasal approach. The duration of operation is generally significantly less than for transcranial operation in experienced hands, and the reported mortality and morbidity is such as to make this procedure the optimal choice for resection of such lesions. References 1. Ezrin C, Horvath E, Kaufman B, Kovacs K, Weiss MH: Pituitary Surgery. Boca Raton, FL, CRC Press, 1980, p 180. 2. Guiot G: Transsphenoidal approach in surgical treatment of pituitary ade nomas: General principles and indications in non-functioning adenomas. Excerpta Med Int Cong Ser 303:158, 1973. 3. Hardy J: L'exeresa des adenomas hypophysaimes par voie trans-sphenolpale. Union Med Can 91:933, 1962. 4. Henderson WR: The pituitary adenomata: A followup study of the surgical results in 338 cases (Dr. Harvey Cushing's series). Br J Surg 26:811, 1939. 5. Schloffer H: Erfulgreiche Operation Eiwes Hypophysewtunions auf Nasallam. Weg WienKlin Wochamschr 20:621, 1907.

21 Anterior and Mid-Third Ventricular Lesions: A Surgical Overview Michael L J. Apuzzo, M.D., Chi-Shing Zee, M.D., and Robert E. Breeze, M.D.

Surgery of the anterior and mid-third ventricular region remains a challenging issue. In spite of improvements in radiological imaging, diagnosis is often in doubt; however, structural definition is realized to the extent that preoperative planning of surgical strategies may be performed in greater detail than ever before. The advent of imagingdirected ster-eotaxis (12, 73) now allows for histological assay, safe endoscopic visualization, and occasionally excision of lesions in the region and must be considered a vital element in the composite of strategical alternatives. The development of microsurgical techniques and their utilization with a number of operative corridors (144) and intraoperative maneuvers provide a variety of approaches that may be undertaken according to the structural and histological presentation of an individual lesion (145, 157, 168). This chapter reviews structural presentations of the diverse pathological spectrum affecting the anterior and mid-third ventricle, as well as the advantages and risks of available operative corridors, in an effort to develop guidelines for strategies in the management of these lesions. Pathology: Patterns of Structural Presentation In spite of a diverse spectrum of histological processes (48), the structural presentation of masses affecting the anterior and mid-third ventricular region may be divided into three major groups with secondary divisions (Figs. 21.1 and 21.2). Development of this classification is pertinent to the selection of the appropriate surgical corridor for excision and alternative management. Extraaxial Intraventricular These lesions are histologically benign masses with minimal areas of origin and adherence to elements of the ventricular walls. Microscopic surgical planes are, in general, well defined. This category consists of lesions of developmental, neoplastic, vascular, and infectious etiologies and includes colloid cysts, Craniopharyngiomas, epidermoids, dermoids,

teratomas, papillomas, cysticercosis cysts, ependymal cysts, granulomas, and arteriovenous malformations (5, 24, 27, 38, 54, 63, 64, 65, 80, 83, 109, 114, 136, 137, 138, 149, 150, 152, 153, 156, 175, 178). Intraaxial with Ventricular Component These lesions originate within the neural or supportive elements adjacent to the ventricle with concomitant mass producing ventricular compression or an exophytic component growing primarily into the ventricular cavity. Within this group are lesions that have extended to the region by local metastasis or by remote metastasis. Examples of the group

include the spectrum of glial tumors that may affect the thalamus, hypothalamus, or optic structures, medulloblastomas, and germinomas (19,57, 155). Basal These lesions have origin from structures within the skull or brain bases and extend superiorly to involve the third ventricular cavity. These

are most commonly histologically benign and potentially excisable lesions with various extents of origin and involvement of the basal structures. Such lesions are, in general, developmental, neoplastic, vascular, or infectious in origin and include pituitary adenomas, Craniopharyngiomas, chordomas, epidermoids, meningiomas, germinomas, arachnoid cysts, cysticercosis cysts, and aneurysms (3, 16, 17, 57, 62, 90, 125, 126, 176). It is important to consider that these processes may be secondarily subdivided into lesions that (a) elevate the third ventricular floor or (b) disrupt the third ventricular floor during their evolution. Those lesions that disrupt the third ventricular floor are clearly visualized through the foramen of Monro. Radiology: Role of Computerized Tomography and Magnetic Resonance Imaging Computerized tomography (CT) is the primary diagnostic imaging procedure in the evaluation of patients with anterior and mid-third ventricular lesions. With the availability of CT, the role of skull roentgenography, skull angiography, and radionuclide brain scanning has diminished markedly while pneumoencephalography has been virtually eliminated. CT has certain advantages compared to these other imaging modalities, including safety, economy of time, and information provided. These features have made CT an integral part of the evaluation of patients with anterior and mid-third ventricular lesions. Not infrequently, instillation of metrizamide into the ventricular system is required to evaluate such lesions, and magnetic resonance imaging (MRI) will undoubtedly be a helpful modality in the evaluation of patients with lesions in this location. MRI has the advantage of increased soft tissue contrast and rapidly provides multiplanar imaging. As experience accumulates, MRI may well replace CT. Mass lesions in the anterior and middle portions of the third ventricle are not common; they can be divided into three major groups: Extraaxial intraventricular These lesions are important because of their potential resectability. Histologically, they are benign masses with minimal areas of adherence to the ventricular wall. Certain intraventricular lesions can only be identified radiographically on metrizamide ventriculograms or metrizamide CT ventriculograms. To confirm that one is dealing with an intraventricular lesion, one must show that the lesion can be delineated by metrizamide on all sides except for a small area of attachment. Metrizamide studies will also demonstrate whether the lesion has smooth margins, a characteristic feature in many of the intraventricular lesions. MRI is also potentially useful in the evaluation of these lesions, and the availability of coronal and sagittal images is an asset. By selecting optimal imaging sequences, demonstration of the intraventricular location of these lesions surrounded by cerebrospinal fluid may be possible. Colloid cysts are the most common anterior third ventricular lesions in many series. They are located in the roof of the third ventricle, arising from the most anterior extremity of the tela choroidea. The findings of CT are highly characteristic in most cases, a well-circumscribed, high attenuation mass being seen on the noncontrast scan (18). A slight increase in attenuation is generally seen after intravenous contrast infusion. Occasionally, the cysts may be isodense or only slightly hyperdense (128). Widening of the septum pellucidum may be associated with colloid cysts (59). Large colloid cysts commonly bulge into the foramen of Monro and can result in intermittent hydrocephalus.

Ependymomas are relatively rare tumors that are seldom found within the third ventricle. On CT, ependymomas exhibit patchy contrast enhancement with cystic areas. Ependymomas have a relatively high incidence of internal calcification (172, 189). Distinguishing ependymomas from deeply located gliomas may be difficult. Both tumors may have a lobulated appearance with contrast enhancement. Intraventricular Craniopharyngiomas are extremely rare lesions. On CT, they may present as enhancing lesions with and without calcification, not different from suprasellar Craniopharyngiomas. Cystic areas within the lesion may also be seen. Cysticercosis cysts can be freely mobile or adherent to the ventricular wall. Intraventricular cysticercosis cysts generally require the instillation of metrizamide into the ventricular system for their demonstration because their density tends to be similar to ventricular fluid and they tend to be thin-walled (8, 190). Choroid plexus papillomas of the third ventricle are very rare. They are usually attached to the roof of the third ventricle, as are colloid cysts. However, in contrast to the smooth margin of colloid cysts, a finely lobulated, irregular margin may be appreciated with Choroid plexus neoplasms. In addition, calcification may be seen. Meningiomas of the anterior third ventricle are rare. CT usually shows a high density lesion before contrast administration, with homogeneous enhancement after contrast infusion. Calcification is frequent; this is a feature not seen with colloid cysts (101). Intraaxial Lesions with a Ventricular Component These lesions include the spectrum of glial tumors that may affect the optic chiasm, hypothalamus, and thalamus as well as germinomas and medulloblastomas. On CT, gliomas may show variable density, whereas germinomas and medulloblastomas generally have a slight increase in density before contrast infusion; with contrast they are usually enhancing mass lesions (122). An intraventricular component of these lesions is also seen as an enhancing mass. Differentiation of these lesions from ependymomas may be difficult. Frequently, metrizamide ventriculograms or metrizamide CT ventriculograms are necessary to determine the extent of intraventricular involvement. MRI can play a significant role in defining the lesion and demonstrating its intraventricular involvement in axial, coronal, and sagittal planes. A major advantage of MRI is that the optic chiasm and pituitary stalk can be readily identified on sagittal and coronal images. The relationship between the lesion and the optic chiasm can be delineated readily due to the difference in signal intensity between the lesion and the optic chiasm, which can be achieved by selecting appropriate imaging sequences. Critical information regarding whether the chiasm is prefixed or superiorly displaced are readily available on coronal and sagittal images. CT can demonstrate the location of a normal optic chiasm. However, in the presence of an abnormal mass lesion, separation of the optic chiasm from a mass lesion in the suprasellar cistern may not be possible because of a lack of significant density difference. A further advantage of MRI is that both internal carotid arteries can be identified in the parasellar region due to the lack of signal from rapidly flowing blood. On the coronal images, the chiasmal carotid window can be readily defined. Encasement or displacement of the carotid arteries by tumor mass can be readily identified. Basal These lesions originate from structures at the base of skull or brain

and extend superiorly to involve the third ventricle. Lesions originating from the skull base may produce bony changes on skull films and CT scans. Lesions originating from a parasellar location may extend occasionally into the third ventricle causing foramen of Monro obstruction. Those lesions with a significant intraventricular component may obliterate the anterior third ventricle on metrizamide ventriculography. Intraventricular lesions with only a small area of attachment to the ventricular wall may be shown to have metrizamide surrounding the lesion on metrizamide ventriculography in contrast to suprasellar lesions that extend to the third ventricle and obliterate that structure. In these cases, no metrizamide will be visualized within the anterior third ventricle because it has been compressed. MRI can provide information regarding changes involving the bone marrow in the clivus. The relationship between the lesion and the brain stem or suprasellar cistern can be well shown on sagittal images. Again, involvement or displacement of the carotid arteries and optic chiasm by the lesion can be clearly demonstrated. A disadvantage of MRI is its poor sensitivity in the detection of calcification in lesions such as Craniopharyngiomas, chordomas, or os-teochondromas. Osteochondromas and chordomas are examples of lesions arising from the skull base. Their origin can be traced easily to the osseous structures of the skull base (16, 90). Pituitary adenomas typically cause enlargement of the sella turcica with the bulk of the lesion seen within the sella. There may be varying degrees of suprasellar extension. With sufficient suprasellar extension, involvement of the anterior third ventricle can also be seen. CT will demonstrate an enhancing mass lesion with or without cystic changes. Calcification in adenomas is uncommon (147). Involvement of the anterior third ventricle is common in Craniopharyngiomas. On CT, they often present as enhancing lesions that are frequently calcified and not infrequently contain cystic areas (56). Suprasellar arachnoid cysts or cysticercosis cysts extending into the third ventricle may be quite difficult to differentiate from cysticercosis cysts or ependymal cysts within the third ventricle on CT. Metrizamide ventriculography or metrizamide CT ventriculography is generally required to define these lesions. Suprasellar meningiomas can be identified as high attenuation lesions with homogeneous, intense contrast enhancement on CT scans. Suprasellar epidermoidomas are low density lesions that occasionally have fatty values and that may show calcification and contrast enhancement. An aneurysm at the tip of the basilar artery may occasionally mimic a lesion in the third ventricle and even cause obstruction at the foramina of Monro. Caution should be exercised to avoid confusing an aneurysm of the basilar artery with a colloid cyst. Careful CT analysis should permit diagnosis of an aneurysm, which should be contiguous with the basilar artery. Of course, angiography provides definitive information in such cases. General Technical Principles of Transcranial Approaches to the Third Ventricle The overlying objectives in surgery of this region relate to obtaining maximal exposure and intraoperative flexibility for lesion access and manipulation (6, 144). These objectives are to be obtained with minimal manipulation, injury, or sacrifice of normal neural and vascular structures. Consideration of a number of principles is essential for a satisfactory outcome.

Corridor Selection Multiple alternatives are available for access to the third ventricular chamber. However, for an individual lesion usually one corridor offers optimal exposure and maximal intraoperative flexibility. Selection of the corridor is dependent upon adequate structural definition on imaging studies as well as possibly angiographic or ventriculographic assessments. Proper selection of the operative corridor will minimize the "bottom line" neural manipulation, injury, or sacrifice. Craniotomy Flap Position The craniotomy flap position should approximate the skull base or midline to minimize brain retraction and maximize exposure corridors. Ventricular or spinal drainage as well as osmotic agents should be used to reduce the mass of retracted tissues and the need for excessive retractor pressure. Only self-retaining retractor systems should be used. Incision of Neural Structures Because the third ventricular chamber can be reached only with neural transit, neural incision is frequently required. The incision required in the lamina terminalis, corpus callosum, or forniceal raphe should be minimized within limits of safe or adequate exposure. Venous Preservation The number of veins sacrificed during transit and at the site of the lesion should be minimized. This principle applies to both parasagittal veins and veins of the galenic and Paraventricular systems. Techniques of alternate routes of access as well as displacement should be used before sacrifice. Cortical injury and deep venous thrombosis are potential causes of major neurological deficit (20, 37, 79, 127, 164, 167). Arterial Preservation All arterial structures in the region of a mass displacement should be preserved. Vessels should be displaced from the tumor capsule. Numerous important arteries are encountered when removing tumors of the third ventricle. The posterior circle of Willis and basilar apex appear below the floor. Perforating branches of the anterior circle are intimately associated with the anterior wall. The posterior cerebral, pericallosal, superior cerebellar, and choroidal arteries are adjacent to the posterior wall. Both anterior and posterior cerebral arteries supply the roof. The internal carotid, anterior choroidal, anterior communicating, and posterior communicating arteries supply perforating branches to the walls. Disorders of memory, personality, and consciousness are the cost of otherwise technically adequate procedures involving minimal vascular injury or sacrifice. Lesion Management Every effort should be made to minimize trauma and the transmission of pressure to normal structures in the region of the abnormality. This may be achieved not only by obtaining adequate and comfortable expo-sure, but also by utilizing careful principles of lesion excision. Mass presence should be minimized initially by internal decompression. Initially, aspiration should be undertaken. Next the capsule should be opened, a biopsy obtained, and intracapsular removal completed at high power magnification by microsurgical technique with or without microsurgical adjuvents. Only after this is completed should an effort be made to separate the capsule from the neural and vascular structures. With decompression, a combination of brain pulsation, microirrigation, and microdissection most often will produce satisfactory dissection planes.

Operative Approaches After the recognition and radiological definition of the pathological process in this region, the goals of the surgical undertaking should include (a) definition of histology, (b) maximal excision of the lesion, (c) relief of the alteration in cerebrospinal fluid dynamics, and (d) relief of local signs and symptoms attendant to the regional mass. A number of operative techniques (Table 21.1) are available and of safe and proven value. Each must be performed appropriately from the technical standpoint and applied properly to the structural and histological substrate if optimization of result is to be realized. Major operative categories include: (a) craniotomy by basal, superior, or combined approaches; (b) imaging guidance stereotaxy; and (c) cerebrospinal fluid diversion.

Craniotomy Basal Masses with basal components and origin are appropriately exposed by a number of alternative routes (Figs. 21.3 and 21.4). All are essentially extraaxial corridors.

Transsphenoidal This approach offers excellent and rapid access of the sella turcica (184). In the event that this structure is enlarged by the mass, even greater suprasellar access is available in vectors that are in the line of sellar distension. Visual system and third ventricular region masses are readily decompressed, particularly if lesions are cystic or soft. Figure 21.3. Sagittal (A) and coronal (B) representations of corridors of access to the anterior and mid-third ventricular region, as well as basal components of lesions affecting the chamber.

Lesion excision and surgical expectations are limited in the event that angular frontal or temporal masses exist or when masses are solid in texture. In addition, excision of dense Capsular components that may be adherent to vascular and neural structures is rarely complete and is undertaken with risk (4, 41, 70, 100). Therefore, tumor excision is generally limited to the removal of tissue within the tumor capsule, permitting the capsule to retract from neural and vascular elements. A number of potential complications must be considered with a transsphenoidal exposure (72, 144, 184): (a) stretch injury of the infraorbital nerve, secondary to sublabial manipulation and speculum insertion; (b) olfactory injury with misdirection of superior dissection; (c) optic nerve injury with optic foramen fracture secondary to forceful retractor opening; (d) multiple cranial nerve (optic, oculomotor, trochlear, trigeminal, abducent) and carotid artery injury with cavernous sinus trauma; (e) carotid, trigeminal, or optic nerve injury with forced retractor opening and forceful manipulation in the sphenoid sinus; (f) optic nerve, chiasmal, circle of Willis, and hypothalamic injury with excessive manipulation during tumor excision, and (g) cerebrospinal fluid fistula with arachnoid disruption. Transcranial (Fig. 21.5) Subfrontal. Basal midline tumors with suprasellar extension may be readily approached by the subfrontal route either unilaterally or bilaterally (134, 135). This approach may be undertaken along midline or oblique frontal corridors. With the midline corridor, angulation for visualization of the optic nerves and carotid arteries is perhaps best in terms of general anatomical comprehension (Fig. 21.6). The surgeon can work between the optic nerves and easily identify the pituitary stalk and carotid arteries. The arachnoid of the anterior chiasmatic cistern is opened. Retraction and manipulation of the optic nerves and chiasm is minimized. After internal decompression of the mass by initial techniques of needle aspiration and microsurgical reduction of the interior of the mass, microdissection of the capsule is undertaken. In general, posterolateral exposure along the ipsilateral carotid artery and the visual system is limited. This may be optimized by lateral extension of the basal component of the bone flap to the pterion. With the utilization of a midline corridor, the transsphenoidal (133) entry via the planum sphenoidale may be used if the chiasm seems to be prefixed, and the lamina terminalis (often distended by the mass) may be incised to offer tumor access or visualization. These maneuvers offer a secondary route for tumor manipulation and delivery (91, 169, 170). The lamina terminalis entry is of value with masses located above the sella turcica, but below the foramen of Monro, particularly if the chiasm seems to be prefixed and subchiasmatic access is not available. In attaining this exposure, care should be taken to preserve perforating branches from the anterior cerebral and anterior communicating arteries. Midline incision of the lamina is optimal. Lack of superior displacement of the A1 (horizontal) portion of the anterior cerebral artery with a suprasellar mass often will be anatomically attended by a "prefixed" chiasm (2). Such an angiographic observation usually indicates a primary anteroinferior vector of chiasmal displacement as the evolving mass affects the normal anatomy of the region. The transsphenoidal approach is indicated if the chiasm is prefixed, the sinus is aerated, opticocarotid access is limited, and the lamina terminalis is not distended by tumor. Pterional. The pterional approach offers the shortest transit from

scalp to sella turcica, as well as good anterolateral visualization of the ipsilateral carotidneurovisual pathway (Fig. 21.7). With opening of the medial sylvian fissure and use of the opticocarotid and lateral carotid corridors, visualization of the retrosellar space is realized (177). Such exposures are effective for suprasellar tumors where the chiasm is prefixed. The opticocarotid corridor is utilized if asymmetrical superolateral extension of the mass distends the interval between the optic nerve and the carotid artery, particularly if the chiasm appears prefixed. Care must be taken to preserve perforating branches from the internal carotid artery to the visual system. When using this exposure, one must realize that the opposite carotid

artery and optic nerve are not fully visualized initially. Tumor removal is often difficult because of the obstructing disposition of major neural and vascular structures, with various nerves and arteries positioned at peculiar angles that are difficult to conceptualize without considerable experience. Subtemporal. The subtemporal corridor offers optimal exposure for lesions in the posterior, parasellar, dorsum sella, and posterior perforated space regions (Fig. 21.8). If sufficient mass is present, exposure of such a lesion may be gained via the pterional opticocarotid approach. However, direct access for posterior suprasellar lesions is optimized subtemporally. The major consideration with this approach is the posterior communicating artery with its perforating branches to the region. In addition, medial angulation of the tentorium may limit exposure. This approach requires extensive temporal retraction and provides poor visualization of structures in the prepontine cistern. The surgeon may visualize the ipsilateral posterior cerebral and superior cerebellar arteries, but the contralateral structures are not fully visualized. In addition, the ipsilateral third and fourth nerves are in the line of the approach. All of the basal approaches, although providing access to basal components of an offending mass, provide limited superior access or visualization.

Superior These approaches offer exposure of the third ventricular via superior entry through the trunk of the corpus callosum or the frontal cortex. These corridors are used for exposure and excision of intraventricular lesions or intraventricular components of basal lesions that are not accessible from a basilar approach (2, 16). Transcortical Originally described by Walter Dandy, this approach classically is undertaken through the right middle frontal gyrus and is optimally used

in the presence of ventriculomegaly, which enhances exposure without the need for excessive brain excision, retraction, or manipulation. It offers excellent visualization of the ipsilateral foramen of Monro with a satisfactory visual alignment for lesions of the middle and midanterior sections of the third ventricular chamber. It provides the optimal angulation for use of the subchoroidal exposure, but a less satisfactory visual alignment for the interforniceal maneuver or visualization of the contralateral

foramen. In addition to the sacrifice of neural tissue required, it is considered to have a higher incidence of seizure complications than the transcallosal exposure. Guidance during the pial-ependymal transit may be enhanced by using real time ultrasonography. Transcallosal The transcallosal corridor, gained by interhemispheric exposure of the body of the corpus callosum in the pericoronal region followed by a 2- to 3-cm incision, offers the major advantages of constant anatomy, shorter transit to the diencephalic roof, and flexibility of exploration within the entire third ventricular cavity through the optimal access provided by interforniceal exposure simultaneously with visualization of both foramina of Monro. There is no disruption of hemispheric tissue and, importantly, ventricular size is irrelevant (10, 110, 111, 162, 163, 166). Section of the callosal trunk has been extensively evaluated and does not carry a physiological cost currently considered to be measurable or recognizable (10, 21, 82). However, the interhemispheric exposure carries the risk of contralateral hemiparesis (10, 162). The incidence of such a complication may be reduced by the utilization of preoperative angiography as an adjunct to operative planning and refinement of operative technique (10). Mutism, a rare complication, may be related to bilateral cingulate retraction. Both superior approaches offer exposure of the foramen of Monro and diencephalic roof with optimal maneuvers for third ventricular entry including (a) transforaminal, (b) transforaminal with ipsilateral forniceal column section, (c) subchoroidal, and (d) interforniceal visualization (Fig. 21.9). The goal of surgery is total excision with minimal trauma to adjacent neural tissues. The amount of exposure required to accomplish this end will vary with the nature of the pathological process and the skill and

experience of the operator. Requirements for exposure are variable, but knowledge and familarity with each method of exposure of the third ventricular chamber is imperative to expand the spectrum of options and thus the safety of the operative endeavor. Each of these methods and options carries certain advantages and disadvantages (53). Transforaminal Visualization. Transforaminal exposure may be op-

timized by the presence of a large lesion that distends the foramen, although this is not a dependable feature of third ventricular lesions (Fig. 21.10). In such a case, if the angulation of exposure is satisfactory, excision may be accomplished with minimal midline manipulation. This is dependant in all cases on the texture of the individual lesion and its resistance to methods of excision initiated by the operator. Often, especially in the presence of a firm lesion, inadequate access to the posterior and anterosuperior component necessitates further exposure maneuvers. Transforaminal Exposure with Unilateral Section of the Ipsilateral Column of the Fornix. This method of enlargement of the area of exposure has been advocated in the past. However, it is not recommended as other maneuvers afford greater exposure without the sacrifice of neural elements. Subchoroidal Visualization. This technique, which takes advantage of the natural plane at the region of the lamina affixa, allows mobilization of both forniceal bodies from ipsilateral to medial (Fig. 21.11). This approach provides access to the central portion of the ventricle by lateral displacement of the fornix via the velum interpositum (43, 49, 75, 98, 99, 179). It is suited for lesions in the superior half of the third ventricle adjacent to the roof and posterior to the foramen. This exposure is expanded to include directly the foramen of Monro by sacrifice of the thalamostriate vein at the foramen—a maneuver that has not been recognized to initiate complicating events. This tolerance confirms the capacity of collateral connections to shunt blood from the deep medullary to the superficial subependymal venous systems. These collaterals are apparent in certain pathological processes. In patients with a direct lateral tributary to the internal cerebral vein, both the septal and thalamostriate veins may be safely sacrificed. The direct lateral vein is similar to the thalamostriate vein; both receive transverse caudate tributaries, although the direct lateral vein enters the internal cerebral vein 1 to 2 cm posterior to the thalamostriate vein. With the presence of a mass and inadequate foramen exposure, this maneuver will often provide satisfactory exposure without active retraction of the fornix. Lines of visualization for this maneuver are optimized by the transcortical approach. In our experience the transcallosal approach may be used as an initial stage before this maneuver, but it is somewhat less satisfactory. A drawback of this technique is that, if a lesion is small or moderately sized, retraction of the thalamus may be required to gain adequate working visualization with the transcortical corridor. Interforniceal Visualization. This maneuver (10, 11) takes advantage of a natural plane of division between the columns and bodies of the fornices that opens into the diencephalic roof (Fig. 21.11). This division is occasionally apparent with biventricular exposure in the presence of a mass and is clearly evident in the presence of a cavum septum pellucidum or septal leafs. It is readily defined by the passage of a microinstrument at the line of the septal insertion in the presence of a mass. When combined with the transcallosal approach, this maneuver affords complete exposure of the third ventricular chamber and midline basal structures. In the presence of a significant mass, active retraction is not required; the internal cerebral veins are displaced by the mass or may be retracted for inferior or posterior visualization as required. This maneuver is undertaken only if transforaminal exposure is inadequate or if manipulation seems excessive. This exposure combined with bilateral foraminal exposure as afforded by the transcallosal technique provides the most extensive exposure of the third ventricular chamber.

Although a number of problems related to the hypothalamic region and its elegant functional components are possible (34, 106, 110, 111, 130), the major complication experienced with micromanipulation of the midline diencephalic structures is transient memory loss. This problem is observed with each of the previously discussed techniques and is generally transient. It is considered to be related to direct and transmitted trauma to the deep midline and paramedian structures.

Combined A single corridor occasionally provides inadequate exposure for satisfactory and safe excision of basal masses with simultaneous third ventricular involvement. In such cases consideration must be given to an approach that combines features of both basal and superior exposures.

Yasargil et al. (187) described simultaneous pericoronal and pterional bone flaps to effect transcallosal and pterional exposures. Ehni and Ehni (53) described a large frontotemporoparietal bone flap for multiple corridor exposure. We have developed a miter bone flap (Fig. 21.12) that affords exposure of the cerebral midline in the pericoronal region for transcallosal exposure and the entire frontal floor medial to the pterion for subfrontal and lateral optic access. This flap may be developed after a bicoronal incision and requires incision of the temporalis muscle in a quadrant cresent at its posterior base. This flap has proved to be economical in terms of effort versus extent of exposure and has proven to be ideal for basal masses with large mid- and anterior sellar attachments. It is not recommended for masses with temporal components, which will require more extensive temporal exposure. Transcortical exposure of the foramen is easily accomplished. Subfrontal exploration is undertaken during the initial stage of the procedure; superior corridors are established if required as a secondary step after the extraaxial operation.

Imaging Guidance Stereotaxy The wedding of computerized tomographic scanning or magnetic resonance scanning and stereotactic techniques has added a new dimension to the management of intracranial masses and most particularly lesions in the third ventricular region (7-9, 12, 13, 23, 85, 87, 88, 105, 129). Low risk biopsy may be achieved, providing guidance for operative or nonoperative decision making (1, 44, 81, 108, 142). In addition, by using ventriculoscopic (78, 180) techniques with imaging guidance stereotaxy, cystic lesions may be aspirated (under direct vision) or excised with microinstrumentation. We have used such methods satisfactorily with craniopharyngioma cysts, colloid cysts, and cysticercosis cysts, totally excising cysticercosis cysts. Intracavitary brachytherapy of neoplastic cystic lesions seems to provide either an alternative to craniotomy or a postoperative adjuvant of substantive importance (68, 124). Cerebrospinal Fluid Diversion Biventricular shunting offers a method of management for lateral ventriculomegaly and its attendant symptoms secondary to bilateral foramen of Monro obstruction. At times third ventricular masses permit communication between the lateral ventricles via the formina-anterior third ventricular complex. This may be indicated on imaging studies or be established by ventriculostomy followed by metrizamide ventriculography. Internalized ventriculoperitoneal or jugular shunting is occasionally required after direct management of third ventricular masses. At craniotomy, third ventriculostomy (78) is accomplished with trans-lamina terminalis exposures with lesion excision or perforation of the region of the tuber cinereum (Fig. 21.13). Silastic (Dow Corning, Midland, Michigan) conduits may be established from the prepontine cistern through the ventricle and midline corridor to a subgaleally placed Rickham reservoir. Internal measures of fluid diversion can be judged competent only by close observation of the patient with intracranial pressure monitoring over a 48- to 72-hour period, followed by frequent clinical and radiographic assessment. Fenestration of the septum pellucidum is advisable with all direct third ventricular exploration to preclude the need for bilateral shunting procedures.

Memory In the microsurgical era, operation of the third ventricle with associated manipulation of midline basal cerebral structures is often attended by postoperative manifestations of an amnestic syndrome (34, 10-12). Fortunately this complication is usually transient (days to several weeks

duration). This striking complication is observed in what is, in most cases, an otherwise functionally intact patient, generating considerable concern and providing a major obstruction to optimal patient performance. The fornix, the major interconnecting limbic pathway and a major component of the diencephalic roof, is by "tradition" considered to be the primary focus of neural injury in such cases. However, the substance of this belief may be challenged. There is no agreement in the literature whether fornix integrity is required for normal memory. Although this structure represents a major limbic pathway, significant comparable fiber bundles would remain intact after isolated forniceal injury. Importantly, nearly all hippocampal connections with the associated cortices would be structurally patent. More logically, current data seem to indicate

that diffuse multifocal midline injuries to a number of areas concerned with the memory process are collectively required for the amnestic syndrome (45, 61). These areas probably include the basal forebrain nuclei, the thalamic nuclei, and the inferior thalamic peduncle. The basal nucleus of Meynert lies predominantly in the substantia innominata. This structure extends inferiorly to the basal ganglia and pallidum from the midline to the temporal lobe. Its major component lies adjacent to the midline millimeters from the third ventricle (Fig. 21.14). This region provides major cholinergie input to the cortical region. Fornix injury may alter cholinergie input to the hippocampal formation, which, when combined with pathological basal nuclei, could impact upon all cortical cholinergie innervation. Important thalamic nuclei include the nucleus reuniens and other associated midline nuclei such as the Paraventricular nuclei. These groups are situated between the anterior pole of the thalamus and the dorsal medial thalamic nucleus directly over the third ventricle. These nuclei provide major input to the entorhinal cortex. The nucleus reuniens projects to the hippocampus, with logically afferent brain stem influence. With associated forniceal injury, the hippocampal formation would be partially restricted from afferent and efferent perspectives. The inferior thalamic peduncle carries a major component of amygdaloid output and provides a major connection with the dorsal medial thalamic nucleus. Both are considered to be important anatomic elements of memory. This structure is limited in anteroposterior extent, but is immediately adjacent to the third ventricle as it passes inferior and medial to the globus pallidus. Craniopharyngioma Surgical Pathology Although comprising between 1 and 4% of all primary intracranial neoplasms, this histological entity represents one of the more common surgical problems in the third ventricular region. Great variability in size is evident, with components of the lesions often involving the sellar, suprasellar, and third ventricular compartments with combinations of cystic and solid components (2, 92). As Craniopharyngiomas are classically considered to arise from rests of buccal epitheleum that accompany migration of Rathke's pouch from the premature stomadeum upward to join the infundibulum and form the pituitary gland and its stalk, the spectrum of compartmental and structural variability is broad (48). Lesions may be totally intrasellar, intrasellar and extrasellar, suprasellar only, or intraventricular only. The most common structural manifestation is a suprasellar mass with displacement and/or disruption of the floor of the third ventricular chamber. These lesions may be totally extraarachnoidal, but may also be intrapial with finger-like projections invading neural tissue. The presence of this intrapial component may be attended by intense glial reaction in adjacent brain. This reaction acts as an impediment to establishing dissection planes in some cases and, particularly with cystic lesions with fine membranes, may further complicate efforts at excision. This response is variable in intensity and extent and, in certain cases, may provide a "buffer area" that allows dissection exclusive of functioning neural tissue (159, 173, 174). A major technical issue relates to involvement of the components of the circle of Willis with elements of the neoplasm. Such intimate adherence often provides the major limitation in excision (31, 32, 35, 36, 40, 52, 60, 67, 77, 84, 86, 94, 115, 116, 119, 123, 141, 151, 158, 171).

Operative Techniques (Suprasellar) For the purposes of this discussion the operative considerations for a suprasellar craniopharyngioma without a temporal or major frontal component are described. This may be considered, with some modification, a "traditional" basal frontal exposure. If hydrocephalus is present a left frontal ventriculostomy is established. Bilateral ventriculostomies may be required in the event that complete obstruction of the right foramen is present. Components of a right basal frontotemporal exposure or a complete frontopterionotemporal craniotomy may be performed. We have preferred a right frontal craniotomy that incorporates the entire frontal floor from the midline to the sphenoid wing. This craniotomy is accomplished by incising the temporalis muscle parallel to the frontal floor to the region of the pterion and then extending the incision at right angles to the base along the line of the coronal suture. A free bone flap is turned with the crescent of temporal muscle at the posteroinferior base. This exposure offers options of pterional and lateral wing excision and opening of the sylvian fissure for added basilar and lateral exposure. The approach to the lesion is extraaxial via a number of optional angular corridors. The medial subfrontal approach offers optimal visualization of both optic nerves, but poor angulation for appreciation of the lateral component of the optic nerve, carotid artery, and optic tract anatomy. The laterofrontal corridor afforded by the posterior (lateral) exposure affords such exposure and adds another intraoperative option. During initial exposure the orbital surface of the frontal lobe is protected by Bicol (Trimedyne, Santa Ana, СA) as identification of the optic nerve and tumor surface is made under low power magnification. Self-retaining retractor systems are essential. After exposure of the parasellar region, a number of options are available, including the subchiasmal, opticocarotid, lateral carotid, transsphenoidal, and translaminar exposures. Selection of each of these options is related to anatomic presentation. Therefore, an effort should be made to accomplish complete appreciation of all elements of anatomic presentation and distortion of normal anatomic relationships. The maximal exposure of tumor surface is the goal of this stage of the procedure. After initial tumor exposure the arachnoid layer is identified and incised, with care being taken to preserve the margin of the tumor-arachnoid envelope. Internal decompression by aspiration is the next stage of the excision, as even minimal aspiration may enhance dissection planes at the tumor margins. Further internal decompression is afforded by methods of aspiration, bipolar cauterization, and sharp excision or laser vaporization (51). Care should be taken in the presence of calcium to avoid transfer of excessive heat generated by laser energy. It is preferable to crush calcium with microinstrumentation and remove it piecemeal. Use of a 5 French suction-irrigation system with sharp dissection within the tumor capsule-arachnoid envelope facilitates dissection. The major feeding arteries are most often derived from the anterior circulation components of the circle of Willis (anterior cerebral, anterior communicating, carotid, and posterior communicating arteries). These small branches may be coagulated, with care being taken to preserve feeding branches to the inferior chiasm and tracts. Capsular components are removed in sections, with care taken to avoid "blind" or excessive traction. Anatomic maintenance of the pituitary stalk is a major objective in any microsurgical dissection in the sellar region and is a requirement for normal pituitary function. With high-power magnification the stalk is

readily recognized by a distinctive striate pattern (portal system) that is directed vertically from the inferior hypothalamic surface to the diaphragma region. This structure is most often identified in the posterior or posterolateral tumor margins. At completion of the excision, angular dental mirrors or endoscopes aid in regional visualization. Although considerable variability of opinion exists regarding the resectability of such lesions, available data imply that initial direct exploration is the most appropriate management, especially in lesions with solid components (36, 76, 77, 173, 174, 159). Such action is required to assess properly the operability of these lesions. Major objectives in the procedure include decompression of the visual system and reduction of local mass effect at the foramen of Monro. With attainment of these goals the procedure should probably be terminated if microdissection planes are obscure. Radiation therapy (76, 95, 96, 146) should be initiated during the postoperative period if operative knowledge or postoperative imaging indicate the presence of tumor or residual calcium. Early recurrence (months) is most commonly cystic and should be managed by imaging stereotactic methods with conduit/reservoir placement for drainage and consideration of colloid-based radionuclide installation (14, 15, 93, 103, 104). Reoperation is not precluded by previous operations and radiation treatment. However, surgical expectations are less optimistic and risks increased in relation to the initial procedure when dissection planes were optimal (77, 173). In the event that reoperation is required, it is worthwhile to consider an alternate corridor. Guidelines for exploration of the variations in structural presentations are detailed later in this chapter. Colloid (Neuroepithelial) Cyst Developmental and Surgical Pathology Colloid cysts of the third ventricle are relatively rare lesions accounting for 0.5 to 1.0% of primary intracranial, neurally derived tumors. Many terms have been applied to these lesions, including neuroepithelial cysts, neuroepithelial tumors, paraphyseal cysts, ependymal cysts, epithelial cysts, Choroid plexus cysts, and cysts of the foramen of Monro (47, 66, 89, 131, 132, 183, 192). Although the first documented case a of colloid cyst of the third ventricle was reported by Wallmann in 1858 (181), it was not until 1933 that Dandy (46) described the radiological definition and surgical management of these lesions (185). There has been considerable debate and controversy over the origin of the cyst. In 1910, Sjovall suggested that the cyst was a remnant of the paraphysis. The paraphysis is a constant feature of human embryos in the 17- to 100-mm stage and lies at the rostral end of the diencephalic roof. This widely accepted concept carried with it the belief that such cysts occurred only at the anterior part of the third ventricle. In 1955, Kappers suggested that, although occasional examples are paraphyseal in origin, most arise from diencephalic ependymal pouches (160). The Choroid plexus has been proposed as an alternate site of origin of such cysts. The term neuroepithelial cyst was proposed by Shuangshoti and Netsky in 1966 (161). They proposed that these lesions may occur anywhere along the course of the central nervous system from the Choroid plexus or ependyma, both being derived from a common neuroepithelium. Most commonly, colloid cysts arise in the anterior superior part of the third ventricle immediately posterior to the foramen of Monro (4, 28, 33,

59). They generally project inferiorly into the third ventricle and vary in extent superiorly and rostrally. Attachments of various dimensions are present with the tela choroidea of the third ventricular roof. Occasionally, the forniceal column superior to the anterior commissure will be separated by the mass (41). This separation may involve the leaves of the septum pellucidum. Generally, the cyst is well circumscribed, smooth, and spherical with dimensions varying from 0.3 to 9 cm. The wall of the cyst is composed of a layer of epithelial cells, either low columnar or cuboidal, and surrounded by a connective tissue capsule (74). The cyst is filled with homogeneous viscous material containing cellular debris. This material is variable in viscosity. Various numbers of desquamated epithelial cells, leukocytes, red cells, gitter cells, and cholesterol pigment have been described in the colloid material. It is important to recognize that these lesions may occur posterior to the foramen of Monro and that they may be attended by various degrees of aqueductal stenosis. Symptomatology Numerous perspectives relating to symptomatology are available in the literature. Symptom complexes include the following. Acute and Intermittent Increases in Intracranial Pressure This causes the "classical" history of paroxysmal headache with changes in head position. These episodes may be attended by vertigo, vomiting, alterations in conscience level, alterations in mentation, seizure activity, vital sign aberrances, or cerebrospinal fluid rhinorrhea. The predisposition to acute demise has been repeatedly stressed (30, 39, 55, 58, 102, 113, 117, 186, 188). Chronically Increased Intracranial Pressure In this complex chronic dementia alone or in combination with gait disturbances and urinary incontinence are encountered (107). Local Pressure Effects These effects related to the mass include sensorimotor, extrapyramidal, and autonomic disturbances and memory disorders (97). Identification of the presence of these lesions by the "classical presentation" is unusual. Long-standing, intermittent, nonspecific headache without attendant postural exacerbations is the most frequently encountered complaint (39, 118). Radiology As has been noted, imaging studies provide the primary indicator toward suspicion of this lesion, with masses generally presenting in the anterior third ventricle with attendant lateral ventriculomegaly (154, 191). The spheroid masses are usually slightly hyperdense, but may be isodense (50, 140). The lesion generally shows some elements of uniform enhancement with contrast medium, but no ring enhancement has been described (29). Attendant structural alterations include enlargement of the septum pellucidum and collapse of the posterior third ventricle. Fullness of the posterior third ventricle may be present if aqueductal stenosis is present. Occasionally, such a lesion may be disclosed as a truly "incidental" finding with no evidence of lateral ventriculomegaly. Focal enhancement may indicate the presence of a xanthogranulomatous component.

Management In spite of their benign histological character these lesions are life-threatening (39). Although the absolute risk of demise has not been accurately determined, there is no doubt that, because of their location and repeated documentations of rapid and fatal neurological deterioration, definitive management to reduce the cyst mass or maintain normal cerebrospinal fluid dynamics is most appropriate. An element of controversy exists regarding the best method for treatment. Arguments exist for the following management options: (a) biventricular shunting, (b) stereotactic aspiration, and (c) direct excision. Because of the potential incidence of aqueductal stenosis (30 to 40%), some method of cerebrospinal fluid diversion may be required if option b or с is exercised initially. Stereotactic aspiration with imaging-guided techniques may be accomplished at small risk and offers an apparently plausible management option (7, 12, 13, 22, 148). In our experience lesions less than 1 cm in diameter may be difficult to puncture and aspirate; however, published experience indicates that cyst aspiration is possible in 80% of lesions greater than 1 cm in diameter if a cannula larger than 1.5 mm may be introduced. Use of this technique should be attended by a meticulous and vigilant monitoring of ventricular size or intraventricular pressure with a consideration of ventriculography postaspiration to assess patency of the aqueduct of Sylvius. This technique requires expertise in the utilization of such instrumentation and requires further evaluation and experience before it can be uniformly endorsed for all cases. It is optimally undertaken with endoscopic control. In the event that craniotomy is undertaken, both the transcortical and transcallosal corridors afford exposure of the foramen of Monro. Occasionally, the forniceal raphe may be opened by the mass presence. With appreciation of the mass at the foramen, aspiration should be undertaken with a #20 needle or #3 F suction after incision of the capsule. Miniature curettes may be introduced within the cyst cavity to assist with reduction of the content volume. At all times, care is taken to avoid traction or manipulation of the cyst attachment at the tela choroidea adjacent to the foramen, which may cause venous or arterial injury. After evacuation of the cyst contents, the wall may be coagulated and gently delivered in fragments through the foramen. If the foramen is small or adequate access for manipulation is not available, added exposure may be gained via subchoroidal or interforniceal corridors. After cyst excision, ventricular exploration may be afforded by fiber-optic endoscopy or direct visualization if additional corridors have been established. The development of a third ventriculostomy in the region of the tuber cinereum may be considered at that time with or without a Silastic conduit being led through the main operative corridor to a Rickham reservoir. Subchoroidal or interforniceal corridors may be developed if the cyst is not apparent at the foramen to identify the abnormality and provide a route for manipulation and excision. In our experience "complete" third ventricular exploration is optimally accomplished via the transcallosal-interforniceal approach. After cyst excision, monitoring of intracranial pressure is required for 48 to 72 hours. Biventricular shunting as a definitive treatment may be considered. However, 70% of the patients will be unnecessarily shunt-dependent when the capabilities of other operative methodologies are considered.

Selection of a Surgical Procedure The selection of an appropriate surgical procedure may be governed by the presenting structural substrate as defined on imaging studies. Many factors bear on such a selection including those related to the individual patient, surgeon, and institution; however, our approach to the utilization of management alternatives based on the presenting structural substrate is detailed in Table 21.2.

Extraaxial Intraventricular Mass (6) Masses contained within the third ventricular chamber are approached primarily via a transcallosal corridor, which offers maximal options for midline exploration irrespective of ventricular size (Table 21.2). If foraminal exposure is inadequate, interforniceal or subchoroidal exposure is easily initiated. Stereotactic excision, evacuation, and/or biopsy of cystic lesions is a primary consideration. Cerebrospinal fluid diversion as the isolated management of such a lesion, although occasionally indicated because of patient factors such as age or medical fragility, is not our primary choice. Selected cases are shown in Figures 21.15 to 21.21. Intraaxial Ventricular Mass (6) Such lesions with intraaxial components and ventricular involvement are often either radiosensitive or managable by indirect means after stereotactic histological assay (Table 21.2). Management is based on information derived from stereotactic biopsy and includes cerebrospinal fluid diversion with radiochemotherapy for malignant lesions and transcallosal third ventricular exploration when a defined and separate intraventricular component is present with histologically benign lesions. Cerebrospinal fluid diversion without tissue assay is not considered a prudent alternative. Selected cases are shown in Figures 21.22 to 21.27. Basal Masses Basal masses may be considered in three groups (Table 21.2): intrasel-lar-suprasellar, suprasellar, and superior extension (third ventricular). Intrasellar-Suprasellar Masses Intrasellar lesions (Table 21.2) with enlargement of the sella and mid-

line suprasellar extension are optimally managed by the transsphenoidal corridor, particularly if the lesion is cystic. In such cases with pneumatized sphenoid bones, anterior sellar entry allows permanent fenestration of the cystic cavity. In the event that frontal or temporal extensions are present, a craniotomy allowing access to the appropriate fossa should be undertaken. If the sellar contour and volume are normal a subfrontal approach is generally the most effective corridor. Stereotactic drainage of cystic lesions with the establishment of permanent reservoirs should be given consideration in the treatment of cystic lesions primarily in elderly or medically fragile patients. Suprasellar Masses Suprasellar masses (Table 21.2) with midline subfrontal superior extension are managed by a subfrontal exposure that allows options initiated on the basis of chiasmal position. If radiographic assessments do not define the strict chiasmal and perichiasmal anatomy, basilar exposure from the frontal midline to the pterion affords the optimal exposure for the exercise of multiple options at the sellar region, including the initiation of subchiasmatic, transsphenoidal, translaminal, opticocarotid, and lateral carotid exposures. Midline retrosellar masses are approached optimally through a pter-

ional corridor with opticocarotid and lateral carotid exposures. Consideration may be given to subtemporal exposures particularly if a temporal component of tumor mass is present. Major subfrontal and temporal masses require exposure of their appropriate fossae. Predominantly cystic masses in the suprasellar region are effectively managed by stereotactic techniques in experienced hands; however, dependent on histology, an effort at wall excision by an appropriate exposure should be considered. Superior Extension Masses When superior extension (Table 21.2) is apparent above the foramen of Monro and the midline base is greater than 2.5 cm, an exposure that allows combined basal and superior approaches is most appropriate. If the basal component is less than 2.5 cm (midline), a transcallosal interforniceal exposure will afford access that will allow complete excision of the lesion. In cases of this type, the hypothalamic floor is not only elevated, but is in fact parted by the lesion so that visualization of the tumor at the foramen of Monro is possible. Stereotactic endoscopy has been recommended as an initial step before undertaking craniotomy for midline exposure and lesion excision. If visualization of the lesion at the foramen is realized, initial transcallosal exposure is appropriate; otherwise, initial subfrontal exposure is indicated. Selected cases are shown in Figures 21.28 to 21.31. References 1. Adams JH, Graham DI, Doyle D: Brain Biopsy: The Smear Technique for Neurosurgical Biopsies. Philadelphia, JB Lippincott Co, 1981. 2. Al-Mefty O, Hassounah M, Weaver P, Sakati N, Jinkins JR, Fox JL: Micro surgery for giant Craniopharyngiomas in children. Neurosurgery 17:585595, 1985. 3. Al-Mefty O, Holoubi A, Rifai A, Fox JL: Microsurgical removal of suprasellar meningiomas. Neurosurgery 16:364-372, 1985. 4. Antunes JL, Kenneth ML, Ganti SR: Colloid cysts of the third ventricle. Neurosurgery 7:450-455, 1980. 5. Antunes JL, Kvam D, Ganti SR, Louis KM, Good J: Mixed colloid cysts— Xanthogranulomas of the third ventricle. Surg Neurol 16:256-261, 1981. 6. Antunes JL, Muraszko K, Quest DO, Carmel PW: Surgical strategies in the management of tumours of the anterior third ventricle. In Borck M (ed): Modern Neurosurgery. Heidelberg, Springer-Verlag, 1982, pp 215-224. 7. Apuzzo MLJ, Chandrasoma PT, Zelman V, Giannotta SL, Weiss MH: Com puted tomographic guidance stereotaxis in the management of lesions of the third ventricular region. Neurosurgery 15:502-508, 1984. 8. Apuzzo MLJ, Dobkin WR, Zee CS, Chan JC, Giannotta SL, Weiss MH: Surgical considerations in treatment of intraventricular cysticercosis: An analysis of 45 cases. J Neurosurg 60:400-407, 1984. 9. Apuzzo MLJ, Sabshin JK: Computed tomographic guidance stereotaxis in the management of intracranial mass lesions. Neurosurgery 12:277-285, 1983. 1 10. Apuzzo MLJ, Chikovani OK, Gott PS, Teng EL, Zee C-S, Giannotta SL, Weiss MH: Transcallosal, interfornicial approaches for lesions affecting the third ventricle: Surgical considerations and consequences. Neurosurgery 10:547-554, 1982. 11. Apuzzo MLJ, Giannotta SL: Transcallosal interforniceal approach. In Apuzzo MLJ (ed): Surgery of the Third Ventricle. Baltimore, Williams & Wilkins, 1987, p. 354. 12. Apuzzo MLJ: CT guidance stereotaxis in the management of 94 lesions of the third ventricular region. In Samii M (ed): Surgery in and around the Brain Stem and the Third Ventricle. Berlin, Springer-Verlag, 1987. 13. Apuzzo MLJ, Chandrasoma P, Zelman V, von Hanwehr R: Applications of computerized tomographic guidance stereotaxis. In Apuzzo MLJ (ed): Sur gery of the Third Ventricle. Baltimore, Williams & Wilkins, 1987.

14. Backlund E-O: Studies on Craniopharyngiomas: III. Stereotaxic treatment with intracystic yttrium-90. Acta Chir Scand 139:237-247, 1973. 15. Backlund E-O: Studies on Craniopharyngiomas: Stereotaxic treatment with radiosurgery. Acta Chir Scand 139:344-351, 1973. 16. Bahr AL, Gayler BW: Cranial chondrosarcomas. Radiology 124:151-156, 1977. 17. Bartlett JR: Craniopharyngiomas—a summary of 85 cases. J Neurol Neurosurg Psychiatry 34:37-41, 1971. 18. Batnitzky S, Sarwar M, Leeds NE, Schechter MM, Azar-Kia B: Colloid cysts of the third ventricle. Radiology 112:327-341, 1974. 19. Beal MF, Kleinman GM, Ojemann RG, Hochberg FH: Gangliocytoma of third ventricle: Hyperphagia, somnolence, and dementia. Neurology [NY] 31:1224-1228, 1981. 20. Bedford THB: The nervous system of the velum interpositum of the rhesus monkey and the effect of experimental occlusion of the great vein of Galen. Brain 57:255-265, 1934. 21. Bogen JE: Physiological consequences of complete or partial commissural section. In Apuzzo MLJ (ed): Surgery of the Third Ventricle. Baltimore, Williams & Wilkins, 1987, p. 175. 22. Borsch DA, Rahn T, Backlund EO: Treatment of colloid cysts of the third ventricle by stereotactic aspiration. Surg Neurol 9:15-18, 1978. 23. Bosch DA: Indications for stereotactic biopsy in brain tumours. Acta Neurochir (Wien) 54:167-179, 1980. 24. Britt RH, Silverberg GD, Enzmann DR, Hanbery JW: Third ventricular Choroid plexus arteriovenous malformation simulating a colloid cyst. J Neurosurg 52:246-250, 1980. 25. Brodeur GM, Howarth CB, Pratt CB, Caces J, Hustu HO: Malignant germ cell tumors in 57 children and adolescents. Cancer 48:1890-1898, 1981. 26. Brun A, Nils E: The pathogenesis of cerebral symptoms in colloid cysts of the third ventricle: A clinical and pathoanatomical study. Acta Neurol Scand 49:525-535, 1973. 27. Buchsbaum HW, Colton RP: Anterior third ventricular cysts in infancy. J Neurosurg 26:264-266, 1967. 28. Bull JWD, Sutton D: The diagnosis of paraphysial cysts. Brain 72:487516, 1949. 29. Bullard DE, Osborne D, Cook WA: Colloid cyst of the third ventricle pre senting as a ring-enhancing lesion on computed tomography. Neurosurgery 11:790-791, 1982. 30. Burkle FM, Lipowski ZJ: Colloid cyst of the third ventricle presenting as psychiatric disorder. Am J Psychiatry 135:373-374, 1978. 31. Cabezudo J, Vaquero J, Areito E, Martinez R, Garcia-de-Sola R, Bravo G: Craniopharyngiomas: A critical approach to treatment. J Neurosurg 55:371-375, 1981. 32. Cabezudo JM, Vaquero J, Garcia-de-Sola R, Leunda G, Nombela L, Bravo G: Computed tomography with Craniopharyngiomas: A review. Surg Neurol 15:422-427, 1981. 33. Cairns H, Mosberg WH Jr: Colloid cyst of the third ventricle. Surg Gynecol Obstet 92:545-570, 1951. 34. Carmel PW: Surgical syndromes of the hypothalamus. Clin Neurosurg 27:133-159, 1980. 35. Carmel PW, Antunes JL, Chang CH: Craniopharyngiomas in children. Neu rosurgery 11:382-389, 1982. 36. Carmel PW: Craniopharyngiomas. In Wilkins RH, Rengacharyss (eds): Neu rosurgery. New York, McGraw-Hill, 1985, pp 905-916. 37. Caron JP, Nick J, Contamin F, Singer B: Tolerance de la ligature et de la thrombose aseptique des veines cerebrales profondes chez l'homme. Ann Med Interne 128:899-906, 1977. 38. Cashion EL, Young JM: Intraventricula craniopharyngioma. J Neurosurg 34:84-87, 1971. 39. Chan RC, Thompson GB: Third ventricular colloid cysts presenting with acute neurological deterioration. Surg Neurol 19:358-362, 1983. 40. Choux M, Lena G: Bases of surgical management of craniopharyngioma in children. Acta Neurochir (Wien) [Suppl] 28:348, 1979. 41. Ciric I, Zivin I: Neuroepithelial (colloid) cysts of the septum pellucidum. J Neurosurg 43:69-73, 1975. 42. Ciric IS, Cozzens W: Craniopharyngiomas: Transsphenoidal method of ap proach—for the virtuoso only? Clin Neurosurg 27:169-187, 1980.

43. Cossu M, Lubinu F, Orunesu G, Pau A, Sehrbundt Viale E, Sini MG, Turtas S: Subchoroidal approach to the third ventricle. Surg Neurol 21:325-331, 1984. 44. Crain BJ, Bigner SH, Johnston WW: Fine Needle Aspiration Biopsy of Deep Cerebrum. A Comparison of Normal and Neoplastic Morphology. Ada Cytol (Baltimore) 26:772-778, 1982. 45. Damasio AR, Van Hoesen GW: Pathological correlates of amnesia and the anatomical basis of memory. In Apuzzo MLJ (ed): Surgery of the Third Ventricle. Baltimore, Williams & Wilkins, 1987, p. 195. 46. Dandy WE: Benign Tumors in the Third Ventricle of the Brain: Diagnosis and Treatment. Springfield, Thomas, 1933, p 171. 47. Danziger J, Bloch S:; Paraphysial cysts of the third ventricle. S Afr Med J 48:2277-2280, 1974. 48. Davis RL: Pathological lesions of the third ventricle and adjacent structures. In Apuzzo MLJ (ed): Surgery of the Third Ventricle. Baltimore, Williams & Wilkins, 1987, p. 235. 49. Delandsheer JM, Guyot JF, Jomin M, Sherpereel B, Laine E: Acces autroisieme ventricule par voie interthalamo-trigonale. Neurochirurgie 24:419422, 1978. 50. Donaldson JO, Simon RH: Radiodense ions within a third ventricular colloid cyst. Arch Neurol 37:246, 1980. 51. Edwards MSB, Boggan JE, Fuller ТА: The laser in neurological surgery. J Neurosurg 59:555-566, 1983. 52. Ehni G: Craniopharyngioma surgery in children. J Neurosurg 52:439, 1980 (letter). 53. Ehni G, Ehni B: Considerations in transforaminal entry. In Apuzzo MLJ (ed): Surgery of the Third Ventricle. Baltimore, Williams & Wilkins, 1987, p. 326. 54. Esteban H, Marquez V, Perez J, Bueno J, Gomez P, Fernandez M: Primary intraventricular choriocarcinoma. Surg Neurol 11:21-23, 1979. 55. Findler G, Cotev S: Neurogenic pulmonary edema associated with a colloid cyst in the third ventricle. J Neurosurg 52:395-398, 1980. 56. Fitz CR, Wortzman G, Harwood-Nash DC, et al: Computed tomography in Craniopharyngiomas. Radiology 127:687-691, 1978. 57. Fox JL, Al-Mefty O: Suprasellar arachnoid cysts: An extension of the membrane of liliequist. Neurosurgery 7:615-618, 1980. 58. Fulton JF, Bailey P: Tumors in the region of the third ventricle: Their diagnosis and relation to pathological sleep. J Nerv Ment Dis 69:1-25, 1929. 59. Ganti SR, Antunes JL, Louis KM, Hilal SK: Computed tomography in the diagnosis of colloid cysts of the third ventricle. Radiology 138:385-391, 1981. 60. Garcia-Uria J: Surgical experience with craniopharyngioma in adults. Surg Neurol 9:11-14, 1978. 61. Garretson H: Memory in man: A neurosurgeon's perspective. In Apuzzo MLJ (ed): Surgery of the Third Ventricle. Baltimore, Williams & Wilkins, 1987, p. 209. 62. Ghatak NR, Hirano A, Zimmerman HM: Intrasellar germinomas: A form of ectopic pinealoma. J Neurosurg 31:670-675, 1969. 63. Giannotta SL, Pauli F, Farhat SM: Epidermoid cyst of the third ventricle. Surg Neurol 5:164-166, 1976. 64. Goldstein SJ, Wilson DD, Young AB, Guidry GJ: Craniopharyngioma intrin sic to the third ventricle. Surg Neurol 20:249-253, 1983. 65. Gradin WC, Taylon C, Fruin AH: Choroid plexus papilloma of the third ventricle: Case report and review of the literature. Neurosurgery 12:217220,1983. 66. Greenwood J Jr: Paraphysial cysts of the third ventricle. J Neurosurg 6:153-159, 1949. 67. Guidetti B, Fraioli B: Craniopharyngiomas: Results of surgical treatment. Acta Neurochir [Wien] [Suppl] 28:349-351, 1979. 68. Gutin PH, Phillips TL, Wara WM, Leibel SA, Hosobuchi Y, Leven VA, Weaver KA, Lamb S: Brachytherapy of recurrent malignant brain tumors with removable high-activity iodine-125 sources. J Neurosurg 60:61-68, 1984. 69. Gyldensted C: Measurements of the normal ventricular system and hemi spheric sulci in 100 adults with computed tomography. Neuroradiology 14:183-192, 1977. 70. Hamer J: Removal of craniopharyngioma by subnasal-transsphenoidal op eration. Neuropaediatrie 9:312-319, 1978.

71. Hammock MK, Milhorat TH, Earle К, DiChiro G: Vein of Galen ligation in the primate. J Neurosurg 34:77-83, 1971. 72. Hardy J: Trans-sphenoidal approach to the pituitary gland. In Wilkins RH, RengacharySS(eds): Neurosurgery. New York, McGraw-Hill, 1985, pp 889898. 73. Heilbrun MP, Roberts TS, Apuzzo MLJ, Wells TH, Sabshin JK: Preliminary experience with Brown-Roberts-Wells computerized tomographic stereo taxic guidance system. J Neurosurg 59:217-222, 1983. 74. Hirano A, Ghatak NR: The fine structure of colloid cysts of the third ventricle. J Neuropathol Exp Neurol 33:333-341, 1974. 75. Hirsch JF, Zouaoui A, Renier D, Pierre-Kahn A: A new surgical approach to the third ventricle with interruption of the striothalamic vein. Acta Neurochir (Wien) 47:135-147, 1979. 76. Hoff JT, Patterson RH Jr: Craniopharyngiomas in children and adults. J Neurosurg 36:299-302, 1972. 77. Hoffman HJ, Hendrick EB, Humphreys RP, Buncic JR, Armstrong DL, Jenkin RD: Management of craniopharyngioma in children. J Neurosurg 47:218-227, 1977. 78. Hoffman HJ, Harwood-Nash D, Gilday DL: Percutaneous third ventriculos tomy in the management of noncommunicating hydrocephalus. Neurosur gery 7:313-321, 1980. 79. Huang YP, Okudera T, Ohta T, Robbins A: Anatomic variations of the dural venous sinuses. In Kapp JP, Schmidek HH (eds): The Cerebral Venous System and Its Disorders. New York, Grune and Stratton, 1984. 80. Ito J, Kadekaru T, Hayano M, Kurita I, Okada K, Yoshida Y: Meningioma in the tela choroidea of the third ventricle: CT and angiographic correlations. Neuroradiology 21:207-211, 1981. 81. Jane JA, Bertrand G: A cytological method for the diagnosis of tumors affecting the central nervous system. J Neuropathol Exp Neurol 21:400409, 1962. 82. Jeeves MA, Simpson DA, Geffen G: Functional consequences of the trans callosal removal of intraventricular tumours. J Neurol Neurosurg Psychia try 42:134-142, 1979. 83. Jensen H-P, Pendl G, Goerke W: Head bobbing in a patient with a cyst of the third ventricle. Childs Brain 4:235-241, 1978. 84. Kahn EA, Gosch HH, Seeger JF, Hicks SP: Forty-five years of experience with the Craniopharyngiomas. Surg Neurol 1:5-12, 1973. 85. Kandel El, Peresedov VV: Stereotaxic evacuation of spontaneous intracer ebral hematomas. J Neurosurg 62:206-213, 1985. 86. Katz EL: Late results of radical excision of Craniopharyngiomas in children. J Neurosurg 42:86-90, 1975. 87. Kelly PJ, Alker GJ, Goerss S: Computer-assisted stereotactic laser micro surgery for the treatment of intracranial neoplasms. Neurosurgery 10:324331, 1982. 88. Kelly PJ: Computer-assisted stereotactic laser microsurgery. In Apuzzo MLJ (ed): Surgery of the Third Ventricle. Baltimore, Williams & Wilkins, 1987. 89. Kelly R: Colloid cysts of the third ventricle. Brain 74:23-65, 1951. 90. Kendall BE, Lee BCP: Cranial chordomas. Br J Radiol 50:687-698, 1977. 91. King TT: Removal of intraventricular Craniopharyngiomas through the lamina terminalis. Acta Neurochir (Wien) 45:277-286, 1979. 92. Kitano I, Yoneda K, Yamakawa Y, Fukui M, Kinoshita K: Huge cystic craniopharyngioma with unusual extensions. Neuroradiology 22:39-42, 1981. 93. Kobayashi T, Kageyama N, Ohara K: Internal irradiation for cystic cranio pharyngioma. J Neurosurg 55:896-903, 1981. 94. Konovalov AN: Microsurgery of tumours of diencephalic region. Neurosurg Rev 6:37-41, 1983. 95. Kramer S, McKissock W, Concannon JP: Craniopharyngiomas: Treatment by combined surgery and radiation therapy. J Neurosurg 18:217-226, 1961. 96. Kramer S, Southard M, Mansfield CM: Radiotherapy in the management of Craniopharyngiomas: Further experiences and late results. AJR 103:4452, 1968. 97. Larson CP: Colloid (paraphysial) cyst of third ventricle with rupture into caudate nucleus and internal capsule. J JVeru Ment Dis 91:557-565, 1940. 98. Lavyne MH, Patterson RH Jr: Subchoroidal trans-velum interpositum ap proach to mid-third ventricular tumors. Neurosurgery 12:86-94, 1983.

99. Lavyne MH, Patterson RH Jr: Subchoroidal trans-velum interpositum approach. In Apuzzo MLJ (ed): Surgery of the Third Ventricle. Baltimore, Williams & Wilkins, 1987, p. 381. 100. Laws ER Jr: Transsphenoidal microsurgery in the management of cranio pharyngioma. J Neurosurg 52:661-666, 1980. 101. Lee YY, Lin SR Horner FA: Third ventricle meningioma mimicking a colloid cyst in a child. AJR 132:669-671, 1979. 102. Leestma JE, Konakci Y: Sudden unexpected death caused by neuroepithelial (colloid) cyst of the third ventricle. J Forensic Sci 26:486-491, 1981. 103. Leksell L: The stereotaxic method and radiosurgery of the brain. Acta Chir Scand 102:316-319, 1951. 104. Leksell L, Backlund E-O, Johansson L: Treatment of Craniopharyngiomas. Acta Chir Scand 133:345-350, 1967. 105. Leksell L, Leksell D, Schwebel J: Stereotaxis and nuclear magnetic reso nance. J Neurol Neurosurg Psychiatry 48:14-18, 1985. 106. Lipton JM, Rosenstein J, Sklar FH: Thermoregulatory disorders after re moval of a craniopharyngioma from the third cerebral ventricle. Brain Res Bull 7:369-373, 1981. 107. Little JR, MacCarty CS: Colloid cysts of the third ventricle. J Neurosurg 39:230-235, 1974. 108. Liwnicz BH, Henderson KS, Masukawa T, Smith RD: Needle aspiration cytology of intracranial lesions: A review of 84 cases. Acta Cytol (Baltimore) 26:779-786, 1982. 109. Loftus CM, Marquardt MD, Stein BM: Hemangioblastoma of the third ven tricle. Neurosurgery 15:67-72, 1984. 110. Long DM, Leibrock L: The transcallosal approach to the anterior ventricular system and its application in the therapy of craniopharyngioma. Clin Neu rosurg 27:160-168, 1980. 111. Long DM, Chou SN: Transcallosal removal of Craniopharyngiomas within the third ventricle. J Neurosurg 39:563-567, 1973. 112. Mair WGP, Warrington EK, Weiskrantz L: Memory disorder in Korsakoff's psychosis: A neuropathological and neuropsychological investigation of two cases. Brain 102:749-783, 1979. 113. Malik GM, Horoupian DS, Boulos RS: Hemorrhagic (colloid) cyst of the third ventricle and episodic neurologic deficits. Surg Neurol 13:73-77, 1980. 114. Markwalder TM, Markwalder RV, Markwalder HM: Meningioma of the anterior part of the third ventricle. J Neurosurg 50:233-235, 1979. 115. Matson DD, Crigler JR Jr: Management of craniopharyngioma in childhood. J Neurosurg 30:377-390, 1969. 116. McLone DG, Raimondi AJ, Naidich TP: Craniopharyngiomas. Childs Brain 9:188-200, 1982. 117. McDonald JA: Colloid cyst of the third ventricle and sudden death. Ann EmergMed 11:365-367, 1982. 118. McKissock W: The surgical treatment of colloid cyst of the third ventricle. Brain 74:1-9, 1951. 119. Metzger J, Nachanakian A, Gardeur D, Sichez JP: CT follow-up studies in postoperative Craniopharyngiomas. Acta Neurochir (Wien) [Suppl] 28:357361, 1979. 120. Michaels LG, Rutz D: Colloid cysts of the third ventricle. Arch Neurol 39:640-643, 1982. 121. Miller CF II, White RJ, Roski RA: Spontaneous ventriculocisternostomy. Surg Neurol 11:63-66, 1979. 122. Miller JH, Pena AM, Segall HD: Radiological investigation of sellar region masses in children. Radiology 134:81-87, 1980. 123. Mori K, Handa H, Murata T, Takechi J, Miwa S, Osaka K: Results of treatment of craniopharyngioma. Childs Brain 6:303-312, 1980. 124. Mudinger F: Implantation of radioisotopes (curi-therapy). In Schaltenbrand G, Walker AE, (eds): Stereotaxy of the Human Brain. New York, ThiemeStratton, Inc, 1982. 125. Murali R, Epstein F: Diagnosis and treatment of suprasellar arachnoid cyst. J Neurosurg 50:515-518, 1979. 126. Northfield DWC: Rathke-pouch tumours. Brain 80:293-312, 1957. 127. Ono M, Rhoton AL Jr, Peace D, Rodriguez RJ: Microsurgical anatomy of the deep venous system of the brain. Neurosurgery 15:621-657, 1984. 128. Osborn AG, Wing SD: Thin-section computed tomography in the evaluation of third ventricular colloid cysts. Radiology 124:257-258, 1977. 129. Ostertag CB, Mennel HD, Kiessling M: Stereotactic biopsy of brain tumors. Surg Neurol 14:275-283, 1980.

130. Page RB: Hypothalamic control of anterior pituitary function: Surgical Implications. In Wilkins RH, Rengachary SS (eds): Neurosurgery. New York, McGraw-Hill, 1985, pp 791-805. 131. Palacios E, Azar-Kia B, Shannon M, Messina AV: Neuroepithelial (colloid) cysts. AJR 126:56-62, 1976. 132. Paterson JE, Leslie M: Colloid cyst of the third ventricle of the brain. Br. Med J 1:920-922, May 1935. 133. Patterson RH, Danylevich A: Surgical removal of Craniopharyngiomas by a transcranial approach through the lamina terminalis and sphenoid sinus. Neurosurgery 7:111-117, 1980. 134. Patterson RH Jr: Subfrontal approach to the pituitary gland. In Wilkins RH, Rengachary SS (eds): Neurosurgery. New York, McGraw-Hill, 1985, pp 902-904. 135. Patterson RH Jr: Subfrontal transsphenoidal and trans-lamina terminalis approaches. In Apuzzo MLJ (ed): Surgery of the Third Ventricle. Baltimore, Williams & Wilkins, 1987, p. 398. 136. Pecker J, Guy G, Scarabin J-M: Third ventricle tumours including tumours of the septum pellucidum, colloid cysts and subependymal glomerate astro cytoma. ClinNeurol 17:440-489, 1974. 137. Pheline C, Engel P, Leguyader B, Tamic P: Meningioma of the third ventricle. J Neurol Neurosurg Psychiatry 40:1024, 1977. 138. Pheline C, Engle P, Leguyader B, Tamic P: Meningioma of the third ventricle. J Neurosurg 51:266, 1979. 139. Posner JB, Block GA: Anatomy and physiology of consciousness: syndromes of altered consciousness related to third ventricular Surgery: In Apuzzo MLJ (ed): Surgery of the Third Ventricle. Baltimore, Williams & Wilkins, 1986. 140. Powell MP, Torrens MJ, Thomson JLG, Horgan JG: Isodense colloid cysts of the third ventricle: A diagnostic and therapeutic problem resolved by ventriculoscopy. Neurosurgery 13:234-237, 1983. 141. Ray BS: Surgery of recurrent intracranial tumors. Clin Neurosurg 10:1 -30, 1964. 142. Rekate HL, Ruch T, Nulsen FE, et al: Needle biopsy of tumors in the region of the third ventricle. J Neurosurg 54:338-341, 1981. 143. Rhoton AL Jr, Yamamoto I, Peace DA: Microsurgery of the third ventricle: Part 2. Operative approaches. Neurosurgery 8:357-373, 1981. 144. Rhoton AL Jr, Yamamoto I: Operative approaches to the third ventricle. In Wilkins RH, Rengachary SS (eds): Neurosurgery. New York, McGraw-Hill, 1985, pp 938-951. 145. Rhoton AL Jr: Microsurgical anatomy of the third ventricular region. In Apuzzo MLJ (ed): Surgery of the Third Ventricle. Baltimore, Williams & Wilkins, 1987, p. 92. 146. Richmond IL, Wara WM, Wilson CB: Role of radiation therapy in the management of Craniopharyngiomas in children. Neurosurgery 6:513-517, 1980. 147. Rilliet B, Mohr G, Robert F, Hardy J: Calcifications in pituitary adenomas. Surg Neurol. 15:249-255, 1981. 148. Rivas JJ, Lobato RD: CT-assisted stereotaxic aspiration of colloid cysts of the third ventricle. J Neurosurg 62:238-242, 1985. 149. Roberts В Jr, Horner FA, McDonald JV: Anterior third ventricle meningioma in a child. Childs Brain 8:145-148, 1981. 150. Rodrigues EA, Lawson AAH: Meningioma of the third ventricle. Postgrad Med J 58:498-500, 1982. 151. Rougerie J: What can be expected from the surgical treatment of Cranio pharyngiomas in children: Report of 92 cases. Childs Brain 5:433-449, 1979. 152. Rozario R, Adelman L, Prager RJ, Stein BM: Meningiomas of the pineal region and third ventricle. Neurosurgery 5:489-495, 1979. 153. Rush JL, Kusske JA, Porter RW, Pribram HW: Xanthogranulomas of the third ventricle. Neurosurgery 4:329-333, 1979. 154. Sage MR, McAllister VL, Kendall BE, Bull JWD, Moseley IF: Radiology in the diagnosis of colloid cysts of the third ventricle. Br J Radiol 48:708723, 1975. 155. Sato M, Ushio Y, Arita N, Mogami H: Hypothalamic hamartoma: Report of two cases. Neurosurgery 16:198-206, 1985. 156. Servo A, Porras M, Jaaskinen J: Diagnosis of ependymal intraventricular cysts of the third ventricle by computed tomography. Neuroradiology 24:155-157, 1983.

157. SeegerW(ed): Microsurgery of the Brain. New York, Springer Verlag, 1980, vol 1 and 2. 158. Shapiro K, Till K, Grant DN: Craniopharyngiomas in childhood: A rational approach to treatment. J Neurosurg 50:617-623, 1979. 159. Shillito J Jr.- Craniopharyngiomas: The subfronal approach, or none at all? CUn Neurosurg 27:188-205, 1980. 160. Shuangshoti S, Roberts MP, Netsky MG: Neuroepithelial (colloid) cysts. Arch Pathol 80:214-224, 1965. 161. Shuangshoti S, Netsky MG: Neuroepithelial (colloid) cysts of the nervous system. Neurology {NY) 16:887-903, 1966. 162. Shucart W: Anterior transcallosal and trancortical approaches. In Apuzzo MLJ (ed): Surgery of the Third Ventricle. Baltimore, Williams & Wilkins, 1987, p. 303. 163. Shucart WA, Stein BM: Transcallosal approach to the anterior ventricular system. Neurosurgery 3:339-343, 1978. 164. Smith RR, Sanford RA: Disorders of the deep cerebral veins. In Kapp JP, Schmidek HH (eds): The Cerbral Venous System and Its Disorders. New York, Grune and Stratton, 1984. 165. Stein BM: Supracerebellar-infratentorial approach to pineal tumors. Surg Neurol 11:331-337, 1979. 166. Stein BM: Third ventricular tumors. Clin Neurosurg 27:315-331, 1980. 167. Sugita K, Kobayashi S, Yokoo A: Preservation of large bridging veins during brain retraction. J Neurosurg 57:856-858, 1982. 168. Sugita К (ed): Micro-neurosurgical Atlas. Heidelberg, Springer-Verlag, 1985. 169. Suzuki J, Katakura R, Mori T: Interhemispheric approach through the lamina terminalis to tumors of the anterior part of the third ventricle. Surg Neurol 22:157-163, 1984. 170. Suzuki J: Bifrontal anterior interhemispheric approach. In Apuzzo MLJ (ed): Surgery of the Third Ventricle. Baltimore, Williams & Wilkins, 1987, p. 413. 171. Svien HJ: Surgical experiences with Craniopharyngiomas. J Neurosurg 23:148-155, 1965. 172. Swartz JD, Zimmerman RA, Bilaniuk LT: Computed tomography of intra cranial ependymomas. Radiology 143:97-101, 1982. 173. Sweet WH: Recurrent Craniopharyngiomas: Therapeutic alternatives. Clin Neurosurg 27:206-229, 1980. 174. Sweet WH: Radical surgical treatment of craniopharyngioma. Clin Neuro surg 23:52-79, 1975. 175. Szper I, Oi S, Leestma J, Kimm KS, Wetzel NE: Xanthogranuloma of the third ventricle. J Neurosurg 51:565-568, 1979. 176. Takeuchi J, Handa H, Nagata I: Suprasellar germinoma. J Neurosurg 49:41-48, 1978. 177. Tindall GT, Tindall SC: Pterional approach. In Apuzzo MLJ (ed): Surgery of the Third Ventricle. Baltimore, Williams & Wilkins, 1987, p. 440. 178. Tomasello F, Albanese V, Bernini FP, Picozzi P: Choroid plexus papilloma of the third ventricle. Surg Neurol 16:69-71, 1981. 179. Viale GL, Turtas S: The subchoroid approach to the third ventricle. Surg Neurol 14:71-76, 1980. 180. Vries JK: An endoscopic technique for third ventriculostomy. Surg Neurol 9:165-168, 1978. 181. Wallmann H: Eine Colloid cyste im dritten Hirnventrikel und ein Lipom im Plexus Choroides. Virchows Arch {Pathol Anat) 11:383-388, 1858. 182. Weber FP, Worster-Dought C, Dickson WEC: Cholesterol tumour (cranio pharyngioma) of the pituitary body. J Neurol Psychopathol 15:39-45, 1934. 183. Weinberger LM, Boshes B: Paraphysical cysts of the third ventricle. Surgery 13:368-384, 1943. 184. Weiss MH: Transnasal transsphenoidal approach. In Apuzzo MLJ (ed): Surgery of the Third Ventricle. Baltimore, Williams & Wilkins, 1987, p. 476. 185. Wilkins RH: History of surgery of the third ventricular region. In Apuzzo MLJ (ed): Surgery of the Third Ventricle. Baltimore, Williams & Wilkins, 1986. 186. Williams PC, English GM, Edgar RE: Colloid cyst causing rhinorrhea. Surg Neurol 14:185-187, 1980. 187. Yasargil MG, Teddy PJ, Roth P: Combined approaches. In Apuzzo MLJ (ed):

Surgery of the Third Ventricle. Baltimore, Williams & Wilkins, 1987, p. 462. 188. Yenermen MH, Bowerman CI, Haymaker W: Colloid cyst of the third ventri cle; a clinica study of 54 cases in the light of previous publications. Acta Neuroveg 17:211-277, 1958. 189. Zee CS, Segall HD, Ahmadi J, et al: Computed tomography of posterior fossa ependymoma in childhood. Surg Neurol 20:221-226, 1983. 190. Zee CS, Segall HD, Miller C, et al: Unusual neuroradiological features of intracranial cysticercosis. Radiology 137:397-407, 1980. 191. Zilkha A: Computed tomography of colloid cysts of the third ventricle. Clin Radiol 32:397-401, 1981. 192. Zimmerman HM, German WJ: Colloid tumors of the third ventricle. Arch Neurol Psychiatry 30:309-325, 1933.

Commentary В Technique and Strategies of Direct Surgical Management of Craniopharyngioma Alexander N. Konovalov, M.D. At the Moscow N. N. Burdenko Institute of Neurosurgery the problem of surgical treatment of craniopharyngioma has always been given special attention. About 1200 patients with these tumors have been observed and treated. My personal experience in dealing with that neurosurgical problem includes more than 300 cases in which microsurgical removal of Craniopharyngiomas was performed. In 1974, we started microsurgical radical removal of these tumors. Since that time our surgical technique has evolved and our results have improved. However, this problem is far from resolution and needs further serious investigation. Anatomical Variants of Craniopharyngiomas From the surgical point of view it is of great importance to clarify the extension and exact localization of the tumor, especially its relation to the third ventricle before operation. In our practice we use the following classification of Craniopharyngiomas according to the point of initial growth and the relation to the third ventricle and to the sella: endosuprasellar-extraven-tricular, and intraventricular tumors (Fig. B.1). Endosuprasellar Craniopharyngiomas Small, purely intrasellar tumors are rare and can be removed with the transsphenoidal approach. An endosuprasellar craniopharyngioma starts to grow in the sella and displaces upward the diaphragm, chiasm, optic nerves, and third

ventricular floor. The pituitary stalk is usually seriously damaged and included into the tumor capsule. In some cases the diaphragm can be destroyed by the tumor as it grows in various directions to reach a giant size. Suprasellar-Extraventricular Craniopharyngiomas The point of initial growth of these tumors is on the level of the pituitary stalk. The tumor destroys the stalk partially or completely, spreads in various directions (retro-, para-, an-tesellar), and displaces the third ventricular floor. Intraventricular Craniopharyngiomas If the tumor starts to grow at the level of the infundibulum it destroys the floor of the ventricle and penetrates into its cavity. There are two possible variants: (a) the main part of the tumor occupies the third ventricle (intraventricular type of craniopharyngioma) and (b) the tumor spreads into the third ventricle as well as into the extraventricular space. All of these variants of Craniopharyngiomas are summarized in Figure B. 1. This classification is just an approximation. In practice it is not always easy to classify the tumor according to these groupings, even at operation. There may be mixed forms as the tumor can grow from several points along the hypophyseal axis. In addition, we have identified a group of giant Craniopharyngiomas that spread far beyond the chiasma-sellar region.

The precise diagnosis of the type of craniopharyngioma and its relation to the third ventricle, optic pathways, sella, and other structures is of great importance. Computerized tomography (CT) and magnetic resonance imaging are the best methods, but even they cannot in all cases determine whether the floor of the third ventricle is penetrated or just displaced by the tumor. Sometimes a preoperative ventriculoscopy may help in defining the intraventricular propagation of the craniopharyngioma. Surgical Technique Third Ventricular Craniopharyngiomas In comparison with Craniopharyngiomas of other groups third ventricular Craniopharyngiomas present the greatest difficulty for radical removal. That is why many surgical approaches have been suggested for their management (trans-lamina terminalis, transcortical, transcallosal, and others). My surgical technique differs for variants of these tumors and has changed significantly during the last 10 years. At first, trans-lamina terminalis access was regarded as preferable. It was used in about two-thirds of my 150 patients with intraventricular Craniopharyngiomas. More recently, my attitude toward this method of crani-

opharyngioma treatment has changed. I now prefer to remove third ventricular tumors through the transcallosal route alone or in combination with the subfrontal approach. If there is exact information about the intra-and extraventricular extent of the lesion, a wide right-sided trepanation is used; its posterior edge reaches the coronal suture and crosses the midline (Fig. B.2). It permits transcallosal removal of the intraventricular part of the tumor as the first stage and after that, if a basal portion of the tumor is still inaccessible through the transcallosal route, the subfrontal approach is used to extirpate the extraventricular part of craniopharyngioma through the lamina terminalis incision, under the chiasm, or through the opticocarotid triangle. Transcallosal Approach The corpus callosum is divided between the anterior cerebral arteries by a 2-cm incision. Then the tumor may be removed through one and if necessary both foramina of Monro (in both cases in incision is made in the septum pellucidum). If the foramen of Monro is small it may be enlarged by the anterior incision or the tumor may be approached by splitting the fornix strictly in the midline. By this method the upper part of the tumor can be widely exposed. This permits

its radical removal (Figs. B.3 to B.6). I now use all of these methods and the choice of the operative corridor depends on the actual anatomical situation. In comparison with the trans-lamina terminalis exposure, the transcallosal approach has some important advantages: the tumor is widely exposed, all parts of the third ventricle are accessible, and all stages of tumor removal are under visual control. With the transcallosal approach

important arterial branches supplying the diencephalon can be preserved more effectively than with the basal approach. The first stage of tumor resection is opening of the capsule and evacuation of the cystic fluid. The solid part of the tumor can be removed with the help of an ultrasound aspirator, although the size of this instrument limits its usefulness in a narrow operative channel. The basal suprasellar part of the tumor is usually hard and contains

large calcifications. The removal of these calcified masses is difficult and necessitates breaking the tumor into fragments. This can be done by means of bipolar coagulation. In some cases with large and very hard calcified components we have successfully used a special small ultra-sound knife. When the solid part is removed the tumor capsule must be separated from the ventricular walls and then removed in pieces. After tumor ablation, there usually is an opening in the ventricular floor through which one can see the pedunculi and the basilar artery with its branches (Fig. B.6). The basal part of large tumors inaccessible by the transcallosal route must be extirpated by a subfrontal exposure during the same operation. Trans-Lamina Terminalis Approach In all cases when the tumor was situated intraventricularly a similar picture was revealed during operation. The optic chiasm was very stretched by the tumor and adjacent to the tuberculum sellae. The lamina terminalis was expanded and thinned and the tumor was usually

visible through it. After tumor puncture through the lamina terminalis and evacuation of cystic fluid, dislocation of the chiasm and optic nerves diminished our observation of the sellar regions and the remnants of the pituitary stalk. The incision in the lamina terminalis (on the average 1 to 1.5 cm long) is made strictly in the midline. At first, the solid, calcified part of the tumor is removed; then the capsule is gently separated from the ventricular cavity and resected (Figs. B.7 and B.8). Using this approach, it is usually impossible to visualize all details of the third ventricle, which makes radical tumor removal difficult and dangerous. Therefore, we prefer the transcallosal route and use the lamina terminalis exposure rarely. The extraventricular part of the tumor may be approached and excised through the opticocarotid triangle. Endosuprasellar Craniopharyngiomas Tumors of this group can vary markedly in size and suprasellar extension. The pituitary gland as well as the pituitary stalk may be partially or completely destroyed. In some cases the tumor

can penetrate into the third ventricle. In an overwhelming majority of cases endosuprasellar Craniopharyngiomas can be radically removed with the subfrontal approach (usually right-sided). After tumor exposure the capsule is widely opened between the optic nerves, and the tumor contents—cystic liquid, cholesterol masses, and calcification—are removed as completely as possible; with capsule collapse, the vessels passing over it can be more easily detached; they include the posterior communicating artery, anterior choroidal arteries, and small branches of the internal carotid arteries to the diencephalic region. Under the microscope, it is possible to distinguish between vessels that merely pass across the capsule and those that contribute to its blood supply. The latter are coagulated and divided. Adhesion of the anterior cerebral arteries to the tumor capsule occurs in cases of pronounced antesellar and suprasellar growth. As it is very important to preserve the vascular

connections of the infundibular part of the third ventricle and pituitary gland, I try at the beginning to divide the tumor from the remnants of the sella diaphragm, which constitute the external layer of the tumor capsule. In some cases it is possible to identify a place between these two layers and preserve the posterior part of diaphragm with its vascular net. If these attempts are unsuccessful the suprasellar part of the capsule is thoroughly separated from the chiasm, the third ventricular floor, and remnants of the pituitary stalk and is excised just above the sellar entrance. The next stage of the operation is removal of the intrasellar part of the tumor with the help of a blunt dissector. The tumor capsule is separated from the dura mater of the sella. Manipulation must be very careful to preserve the remnants of the pituitary gland, a fine strip of light pink tissue lining the floor and posterior wall of the sella turcica. Complete removal of the endosuprasellar craniopharyngioma usually results in a wide expo-

sure of the interpeduncular cistern, the brain stem, and the upper part of the basilar artery with its branches (Figs. B.9 and B.10). Suprasellar-Extraventricular Craniopharyngiomas This type of craniopharyngioma occurs more often in adults than in children. The tumors are frequently situated asymmetrically, displacing the hypophyseal stalk and the third ventricle to one side. In the majority of cases unilateral subfrontal access permits radical tumor removal, but if most of the tumor growth is in the middle or posterior fossa other approaches (such as subtemporal or transtentorial) may be used. Depending on the location of the tumor, its dissection has been carried out between the optic nerves and also through the opticocarotid triangle. The tumor capsule is frequently thin and contains fluid, cholesterol masses, yellow tumor tissue of a lobular structure, and collections of calcium. Sometimes the tumor appears as dense homogeneous tissue that includes the large vessels and optic and oculomotor nerves. After evac-

uation of the cystic fluid and the solid part of the tumor that can be removed more easily with the help of an ultrasound aspirator, the capsule must be meticulously separated from the optic pathways, the third ventricle floor, the sella diaphragm, and the carotid arteries and their branches and then removed. Giant Craniopharyngiomas In this group we include tumors, usually cystic, that spread far beyond the chiasmal-sellar region. These tumors may occupy the anterior, middle, and even posterior fossae or the ventricular system. Quite often the walls of the large cystic cavities are so thin that they could hardly be completely separated from the cerebral tissue and particularly the arachnoid membrane. Besides the cysts there are also solid portions of the tumor in which the inmportant vessels and cranial nerves may be included. About half of our cases of giant craniopharyngioma were in children. Depending on the predominant location of the tumor, various approaches can be used (frontal,

temporal, transventricular, and others). Separation of the important arteries and their branches, the cranial nerves, and the diencephalic structures is the most critical part of the operation (Fig. B.1 1). Special attention should be given to careful hemostasis because of the large operative field. The removal of large cysts may result in brain collapse and rupture of the cortical veins. In some cases, the cyst should be punctured repeatedly to evacuate fluid or an Ommaja system should be used before radical extirpation. Fluid

evacuation diminishes cyst size, facilitating extirpation. Clinical Considerations At the Burdenko Institute, Craniopharyngiomas are approximately as common in adults as in children. There is some difference in the presentations of different types of craniopharyngioma in children and adults. Endosuprasellar, third ventricular, and giant craniopharynyn-giomas are more frequent in children. Suprasel-lar-extraventricular (pituitary stalk) craniopha-

ryngiomas occur more often in adults (about one-half of all cases). We operate as radically as possible upon children with Craniopharyngiomas to prevent a tumor recurrence. With adults, especially aged patients, we are more conservative. Reoperations are more difficult and dangerous because of adhesions. Nevertheless, the radical removal of

Craniopharyngiomas in subsequent operations is possible. Concerning palliative surgery on the ventricular system, our policy is first to try to remove the tumor radically by the direct approach. This is easier to do when the ventricles are large than when they collapse after diversionary shunt operations.

Commentary С Diencephalic Structures at Risk in Third Ventricular Surgery Robert B. Page, M.D. The median and paired third ventricle lies between the two hemispheres of the brain and is filled with fluid. Operating within it to remove a tumor or a parasitic cyst should be relatively safe as long as its boundaries are not violated. The problem presented to the surgeon when faced with a tumor in the third ventricle is rather like the problem presented to Willie Sutton when faced with a payroll in a bank vault. It is to determine by what route the vault (ventricle) can be entered and by what techniques the payroll (tumor) can be removed so that as little disturbance as possible will be created and the bank (patient) will continue to function as before. To answer that question for each individual job, we would do well to emulate Mr. Sutton and "case the joint." Anatomy of the Third Ventricle The front wall of the third ventricle is formed by the optic chiasm, the lamina terminalis, and the anterior commissure. The optic chiasm and the anterior commissure are fiber tracts through which information is passed between neural structures of the two hemispheres. The lamina terminalis, although not serving that function in adulthood, is the embryological site of development of the commissural plate, which differentiates into the anterior, callosal, and hippocampal commissures (3). The Supraoptic nuclei and the columns of the fornix lie in the anterior wall of the hypothalamus, just dorsal to the optic chiasm and just lateral to the lamina terminalis.

The organum vasculosum of the lamina terminalis (OVLT), a structure implicated in body fluid homeostasis and reproduction, lies beneath the anterior commissure in the midline in the lamina terminalis. More dorsally, at the level of the foramen of Monro lies the subforniceal organ (SFO), which is also implicated in the maintenance of fluid homeostasis (2). The ceiling and floor of the third ventricle are also attenuated. Its ceiling is vascular and is made up of the Choroid plexus of the third ventricle, which is attached to the thalamus along the tinea thalamica, and of the paired internal cerebral veins. It is termed the velum interpositum. Overlying that ceiling, as a roof over the attic, are the hippocampal commissure and the crus of the fornices. The floor of the third ventricle is formed by the mamillary bodies (the ultimate destination of many fibers in the fornix, which lies in its roof) and the tuber cinereum, the site of attachment of the hypophyseal median eminence to the hypothalamus. The posterior surface of the third ventricle is formed by the midbrain, the posterior commissure, the habenula, and the pineal gland. The aqueduct of Sylvius serves as a conduit carrying cerebrospinal fluid out of the third ventricle and lies just beneath the posterior commissure. The posterior wall of the third ventricle is attenuated only at the posterior and habenular commissure and the suprapineal recess. The lateral walls of the third ventricle are stout and are made up of the nuclear masses of the hypothalamus.

Approaches to the Third Ventricle As a consequence of this architecture, the surgeon is forced to enter the third ventricle by one of three routes. The first is through the lamina terminalis (the anterior wall). The second is through the forniceal raphe, hippocampal commissure, and velum interpositum or just lateral to the fornices through the tinea thalamica (the roof). The third is through the posterior commissure or the suprapineal recess (the second story of the back wall). Entering from below is barred by the attachment of the pituitary gland of the brain. Entering from the side is barred by the nuclear masses of the hypothalamus. Entering from the posterior ventral aspect is barred by the connection of the midbrain with the hypothalamus. The surgeon will thus enter and leave the third ventricle by the same routes as many fiber tracts that enter and leave the hypothalamus, and the surgeon must perforce be cognizant of them. While working in the third ventricle, the surgeon will work next to nuclear masses and fiber tracts of the hypothalamus and must respect them lest alarms go off that cannot be safely silenced. Hypothalamus: General Anatomical Features If the brain is sectioned in the midsagittal plane, the medial wall of the third ventricle can be visualized. A horizontal groove, the hypothalamic sulcus, demarcates the thalamus above from the hypothalamus below. The anterior margin of the hypothalamus does not strictly correspond to the anterior wall of the third ventricle (the lamina terminalis). Instead it is defined by a line drawn from the rostral margin of the optic chiasm to the anterior commissure (5). In midsagittal section the hypothalamus is frequently divided into three zones. From anterior to posterior they are the Supraoptic, the tuberal, and the mamillary regions. In coronal section the hypothalamus is divided into a periventricular (median), a medial, and a lateral zone. Finally, a horizontal line drawn parallel to the anterior-posterior commissural line and midway between the floor and the hypothalamic sulcus separates the hypothalamus into a dorsal and a ventral zone. This schoolbook parcellation of the hypothalamus into a threedimensional grid (5) has several important consequences for surgeons. First, ascending and descending input from the brain stem and from the medial forebrain areas, respectively, for the most part enters the medial forebrain bundle—a tract coursing through the lateral zone of the hypothalamus and far from

the sites of surgical appraoch. Second, neurosecretory groups involved with regulating the individual's ability both to survive and to reproduce lie in the periventricular zone and are at risk during surgical exploration of the third ventricle. Third, chemosensory sites with the ability to sense sodium concentration and glucose concentration lie in the periventricular zone of the hy-pothalamus. Specific Nuclei The periventricular zone contains all of the neurosecretory cell nuclei except the Supraoptic nuclei (SON). These nuclear groups include the Paraventricular nuclei (PVN) and the arcuate nuclei (AN). The suprachiasmatic nucleus also lies in the periventricular zone and its cells synthesize a neurosecretory peptide—vasopressin. However, these cells synapse with other neurons and do not make neurohemal contact. In these suprachiasmatic cells, vasopressin serves as a neurotransmitter or a neuromodular and not as a neurosecretion. The Supraoptic nuclei are em-bryologically derived from the periventricular zone but migrate from it during the course of brain maturation to take up a new position over the optic tract (5). Similarly, gonadotropin-re-leasing hormone (GnRH)-containing cells are thought to migrate from the periventricular zone into the preoptic area of the adult. The PVN and AN are of particular importance; if recent findings in laboratory animals are applicable to humans, these nuclear groups are the primary sites in which peptides and amines regulating the pituitary gland's response to changes in the internal milieu and the external environment are synthesized. Corticotropin-releasing factor (CRF), arginine vasopressin (AVP), oxyto-cin (ОТ), thyrotropinreleasing hormone (TRH), enkephalin, and somatostatin (SRIF) are also synthesized in cells residing in the PVN. Axons of many of these neurons terminate in the median eminence near the fenestrated vessels of its primary capillary plexus. Axons containing CRF, AVP, ОТ, TRH, or SRIF leave the PVN laterally and course toward the medial forebrain bundle. Upon entering the lateral zone they turn to course downward toward the ventral surface of the brain and then pass as a discrete fiber bundle medially through the lateral retrochiasmatic area to gain the midline and to enter the median eminence from anteriorly. They terminate in its medial third (7). The Paraventricular nucleus may well serve as a site for the integration of the autonomic neural and the endocrine systems (10). For example, with hypotension, neural traffic is increased over

the glossopharyngeal nerves and there is increased input into the nucleus and tractus soli-tarius. Connections between the nucleus solitar-ius and the dorsal motor nucleus of the vagus and the lateral reticular nucleus have been demonstrated. Ascending noradrenergic systems originating in these nuclei project to the parvo-cellular division of the PVN. Some noradrenergic neurons terminate on AVPcontaining neurons in the PVN. Some of these neurons, along with neurons containing CRF, project to the median eminence. AVP and CRF released into median eminence capillaries and carried by portal routes to the pars distalis synergistically stimulate the release of adrenocorpticotropic hormone (ACTH) from corticotrophs and of beta-endorphin from melanotrophs. ACTH can then enter the systemic circulation to stimulate the release of corticosteroids from the adrenal gland. The lateral reticular nucleus also projects to the SON (1). Stimulation of the SON results in release of AVP from axons terminating in the neural lobe. AVP can enter the systemic circulation to (a) conserve water loss to the kidney and to (b) elevate blood pressure by stimulating vasoconstriction and by sensitizing the gain and the feedback response of the carotid bodies and hence altering the neural traffic over the glossopharyngeal nerves. AVP released in the neural lobe can also cross (by short portal and capillary routes) into the neighboring pars distalis to release beta-endorphin from melanotrophs or ACTH from corticotrophs (6). Some AVP-synthesizing cells in the PVN project not to the median eminence but to the medulla where they synapse with ascending noradrenergic cells projecting back to the PVN. Some oxytocincontaining cells in the PVN project to the intermediolateral cell column of the spinal cord and presumably alter sympathetic outflow. CRF projections from cell bodies lying in the Paraventricular nucleus can also alter sympathetic tone. The details of this system are far from clear but it seems evident that the PVN is a "site for the integration of neuroendocrine and autonomic mechanisms" (10). Less well localized are sites for the integration of temperature control in the walls of both the anterior and posterior aspects of the third ventricle and sites for sensing the concentrations of sodium and glucose in the circulating blood. It is evident that cell systems subserving these functions vital for homeostasis lie close to the walls of the third ventricle and that the integrity of the ventricular surface should be respected by surgeons entering the third ventricle. The arcuate nucleus lies near the floor of the

third ventricle. Many of its cell bodies contain dopamine (11). Others are the site of the synthesis of proopiomelanocortin (POMC) and of its derivatives (ACTH, the melanotropins, beta-lipotro-pin, and betaand gamma-endorphin) (4). With the exception of growth hormone-releasing hormone (8), hypothalamic releasing and inhibiting hormones do not seem to be synthesized in the arcuate nuclei. Dopamine, synthesized in cell bodies in the arcuate nucleus and released predominantly in the lateral thirds of the median eminence, inhibits prolactin (PRO) secretion by pituitary lactotrophs. The role of the processed products of POMC in the median eminence is unknown at present but endorphins may modify gonadotropin, PRO, AVP, and ОТ secretion. The significance of projections of ACTH-containing fibers from the arcuate nuclei to the Paraventricular nuclei where they contact cells containing CRF (4) remains unknown. The Supraoptic nuclei lie in the medial zone of the Supraoptic region. They contain magno-cellular neurons that synthesize either AVP or ОТ with their associated neurophysins. In addition, some cells form part of a third opioid system and contain dynorphins. Dynorphins may coexist in the same neurons with the classical neurosecretory peptides (12). Axons of the Supraoptic nucleus project posteriorly to form the supraopticohypophyseal tract, which enters the median eminence of the pituitary in the midline. The supraopticohypophyseal tract passes through the internal zone of the median eminence. Its axons terminate on fenestrated capillaries in the neural lobe. The preoptic area contains GnRH-containing cell bodies that project directly to the median eminence and terminate in its lateral thirds. These fiber pathways join the fibers coursing medially from the lateral retrochiasmatic area to form a funnel where neuropeptide-containing pathways enter the median eminence from the front (7). These fiber tracts pass through the arcuate nuclei anteriorly as they descend into the median eminence. Frontal Approach When approaching the third ventricle from the front through the lamina terminalis, the surgeon will use a subfrontal approach and gain access to the lamina terminalis and the optic chiasm. In doing so, the surgeon will expose the region of the anterior perforated space and the preoptic area. The structures at risk then include descending pathways from the medial frontal lobe, which will enter the medial forebrain bundle in

the lateral wall of the hypothalamus. Such structures pass close to the surface of the brain in the region of the olfactory tubercle and anterior perforated space and include projections from the septal region destined for the mesencephalic tegmentum (5). Damage in this region through excess retraction or damage to perforating vessels originating from the anterior cerebral artery may result in significant alterations of consciousness. Damage to higher cortical function can be expected if the nucleus basalis of Meynert (the origin of the major cholinergie projection to the cortex) is injured. Changes in gonadotropic function are possible as GnRHcontaining cells lie in the preoptic area. The lamina terminalsi itself is avascular and contains no neural tracts. The approach into the third ventricle is safe as long as it is kept strictly in the midline. The organum vasculosum of the lamina terminalis and the subforniceal organ lie along the course of the approach. Both of these structures have been implicated in salt and water homeostasis (9). In addition, the SON lie just above the optic chiasm and just lateral to the midline. Hence, disturbances in water balance and conservation may be secondary to damage to the SON or to the OVLT and SFO. In addition, memory deficits may occur if the pillars of the fornices or if the SON are damaged during the approach. Memory dysfunction has been reported both with forniceal damage and with the loss or destruction of AVPsecreting neurons. Such a deficit may recover as diabetes insipidus improves. Superior Approach In the approach to the third ventricle from above, the fornices are at risk as they lie within the operative field. However, no other major input into the hypothalamus is present in this region. Posterior Approach The posterior approach through the suprapineal recess or through a stretched and thinned posterior commissure also spares the ascending and descending tracts leaving the hypothalamus posteriorly. These tracts include the mamillary peduncle and the dorsal longitudinal, fasciculus and lie well beneath the tectum of the midbrain and out of the field of the surgeon. Ventricular Floor The floor of the third ventricle is at risk when the surgeon is working within the third ventricle or beneath it as in the transsphenoidal approach carried above the diaphragma sellae. Damage to

the floor of the third ventricle will include the arcuate nuclei and the neurosecretory tracts passing through it as well as the median eminence itself as these systems are tightly packed within a small space. Panhypopituitarism is to be expected. The surgeon attempts to enter the third ventricle through windows placed anteriorly, superiorly, and posteriorly. These windows contain either no fiber tracts or fiber tracts such as the hippocampal or posterior commissure, which can be divided without apparent neurological deficit. Through these windows the surgeon gains access to the third ventricle, but once within the third ventricle one must take care not to violate its walls. References 1. Day ТА, Renaud LP: Electrophysiological evidence that noradrenergic afferents selectively facilitate the activity of Supraoptic vasopressin neurons. Brain Res 303:233-240, 1984. 2. Joseph SA, Knigge KM: The endocrine hypothal amus: Recent anatomical studies. Res Publ Assoc Res Nerv Merit Dis 56:15-47, 1978. 3. Loeser JD, Alvord EC Jr: Agenesis of the corpus callosum. Brain 91:553-570, 1968. 4. Mezey E, Kiss JZ, Mueller GP, Eskay R, O'Donahue TL, Palkovits M: Distribution of the pro-opiomelanocortin derived peptides, adrenocorticotrope hormone, a-melanocyte-stimulating hor mone and B-endorphin (ACTH, a-MSH, 0-End) in the rat hypothalamus. Brain Res 328:341-347, 1985. 5. Nauta WJH, Haymaker W: Hypothalamic nuclei and fiber connections. In Haymaker W, Anderson E, Nauta WJH (eds): Hypothalamus. Springfield, IL, Charles С Thomas, 1969, pp 136-209. 6. Page RB: Directional pituitary blood flow: A microcinephotographic study. Endocrinology 112: 157-165, 1983. 7. Palkovits M: Neuropeptides in the median emi nence: Their sources and destinations. Peptides 3:299-303, 1982. 8. Sawchenko PE, Swanson LW, Rivier J, Vale WW: The distribution of growth-hormone-releasing factor (GRF) immunoreactivity in the central nervous system of the rat: An immunohistochem ical study using antisera directed against rat hy pothalamic GRF. J Comp Neurol 237:100-115, 1985. 9. Simpson JB: The circumventricular organs and the central actions of angiotensin. Neuroendocrinology 32:248-256, 1981. 10. Swanson LW, Sawchenko PE: Paraventricular nucleus: A site for the integration of neuroendo crine and autonomic mechanisms. Neuroendocrinology 31:410-417, 1980. 11. Szentagothai J, Flerko B, Mess B, Halasz B: Hy pothalamic Control of the Anterior Pituitary. Bu dapest, Akedemiai, 1968. 12. Whitnall MH, Gainer H, Cox BM, Molineaux CJ: Dynorphin-A-(l-8) is contained within vasopres sin neurosecretory vesicles in rat pituitary. Sci ence 222; 1137-1139, 1983.

22 Posterior Transcortical Approach Kenichiro Sugita, M.D., and Kazuhiro Hongo, M.D.

Indications, Applications, and Contraindications The posterior transcortical approach through the superior parietal lobe to the third ventricle is not indicated for a lesion entirely within the third ventricle. It is indicated for such lesions as posterior thalamic and posterior parathalamic tumors, which are related to the posterior Paraventricular portion, and arteriovenous malformations around the posterior half of the ventricle (around the trigone), which are located in the lateral wall of the third ventricle. The approach can be made through a cortical incision in the parietal lobe, even in the dominant hemisphere. The posterior transcortical approach is contraindicated for lesions of the anterior portions of the thalamus and ventricle, for which the anterior transcortical approach through the frontal lobe is used. A lesion located solely in the third ventricle is best approached interhemispherically. The more lateral to the midline is a lesion, the more useful is the transcortical approach (1-10, 13, 14). In our own practice we have not used an approach through the middle temporal gyrus to lesions of the posterior ventricular portion and the posterior thalamus because a temporal approach to such lesions is likely to produce postoperative neurological deficits such as hemianopsia and hemiparesis, although the distance involved in this approach is shorter than that through the medial parietal incision. The temporal approach is indicated only for lesions located lateral to the temporal horn and in the temporal lobe. An approach through the lateral parietal cortical incision is inadvisable because the areas involved are related to more important neurological functions, such as speech and vision, than are involved in the medial approach. Preoperative Evaluation and Radiographic Assessment Exact radiological study is indispensable for the posterior transcortical approach. Three-dimensional reconstruction of the preoperative computerized tomographic (CT) scan is very helpful because it is often difficult to obtain topographical orientation during the procedures for deep-seated lesions through a small cortical incision under an operating microscope.

In addition to sagittal reconstruction of the CT scan, the new imaging by magnetic resonance would also be very helpful where available. Angiog-raphy provides us with much important information: How vascular is the lesion? Where do the main feeding arteries come from? For a highly vascular tumor we should plan to attack during the initial stage the side where the feeders are mainly located. How much have the important veins been shifted? When the preoperative CT scan shows a diffusely infiltrating lesion without a clear border or with an extensive invasion into the bilateral hemispheres, radical operation using this approach should not be attempted. There are also objections to direct surgical attack on malignant tumors located in the periventricular region because of the possibility of ventricular seeding during the transventricular approach. When a lesion is small and seated deep from the ventricular floor, a two-stage operation of stereotaxic exploration and direct attack is recommended. At the time of the stereotaxic exploration a metal marker such as a piece of a hemoclip should be left in the lesion in addition to taking a tumor specimen for histological study. The metal marker is useful for the second operation. Considerations of Position and Intracranial Pressure Control The patient's position is prone with the head elevated higher than the heart level, to achieve which the upper portion of the operating table is elevated about 35°. The patient's head is positioned slightly chin-up so that the operating microscope can be used perpendicularly in a sagittal plane. During the initial procedure from the skin incision to exposure of the trigone the patient's head is kept perpendicular to the floor because in this way the correct direction from the cortical incision to the ventricle is easily obtained. After the ventricle is entered, the patient's head is rotated to the side of the lesion so that we can use the microscope perpendicularly in a coronal plane during the procedure around the lesion; the angle of rotation is usually about 20°. The usual methods for controlling intracranial pressure, such as elevation of the patient's head, hyperventilation, and the administration of mannitol, are indispensable during the initial stage. Lumbar drainage of cerebrospinal fluid (CSF) is usually not done because the ventricle is opened during approach and lumbar drainage might promote tentorial herniation if the mass is large with perifocal edema. In some cases with advanced noncommunicating hydrocephalus ventricular drainage is instituted in the contralateral anterior horn before the craniotomy; the tap is helpful for the initial approach until the ventricle is exposed and CSF is suctioned from the whole ventricular system. Pictorial Illustration of the Technique Position of the Patient's Body and Head This is as previously described (Fig. 22.1A). Sugita Multipurpose Head Frame This assembly (12) is an essential tool in this approach; it holds four tapered spatulas that facilitate the approach through a small cortical incision with minimal damage to normal brain tissue (Fig. 22.1B). Skin Incision with Craniotomy A U-shaped skin incision is made in the medial parietal area, 6x6 cm in size. The top of the skin flap is made along the midline. The craniotomy

is 5 х 5 cm in size so as to correspond to the skin flap close to the sagittal sinus, which is not exposed. The center of the craniotomy is about 4 to 5 cm posterior to the central sulcus and 3 cm lateral to the midline (Fig. 22.1С). Dural Opening, Ventricular Tap, and Cortical Incision The dura mater is opened with an X-shaped incision. A ventricular tap needle is inserted at the cortical surface slightly posterior to the center of the craniotomy. The target of the tap is the trigone. The CSF is not drained from the tap except in a case where the brain bulges with advanced hydrocephalus because a direct approach to a collapsed ventricle is difficult. A cortical incision less than 3 cm long is made coronally on the superior parietal cortex; the approach to the trigone is made by coronal sectioning of the brain. Coronal sectioning of the cortex and brain helps to minimize damage to the connecting fibers of the corpus callosum. In a patient with advanced hydrocephalus the incision should be made slightly posterior to the center of the craniotomy because the brain sinks to the frontal side after the egress of CSF. Brain Retraction with Four Tapered Spatulas The operating microscope is used throughout the procedure after a successful ventricular tap. The route from the cortical incision to the trigone is made by sectioning the brain along the track of the ventricular tap. The approach is helped by the application of four tapered spatulas that are held with four self-retaining retractors. These retractors are set on the Sugita multipurpose head frame; in this way spatulas can be applied from any point around the operating field. There are two particularly important points regarding brain retraction. The first concerns the application of tapered spatulas. Three kinds of tapered spatula are available, differing in the width of the tip: 2, 4, and 6 mm. The application of tapered spatulas enables us to obtain an ample operating field with less brain retraction, and retraction with tapered spatulas results in less brain damage than retraction with the usual spatulas. Another important point is that the brain is not retracted with all four spatulas at a time but only with a pair; the other pair is kept loose. The pair in use is changed whenever the side of the operative procedure is changed. Retraction with a pair of spatulas facilitates visualization of a wider operating field than retraction with four spatulas, even through a 3-cm cortical incision. The self-retaining retractor must be light; our newly designed one made of titanium weighs 180 g and is tapered. As it is so light it is possible to feel the resistance of the brain and know how much the brain is retracted. Cotton patties must be placed under the spatulas; they must be soft and thin with little tendency to adhere to tissue (Fig. 22.2A). Techniques of Dissection of Tumors Usually it is difficult to discriminate thalamic and parathalamic tumors from the normal structure of the ventricular floor even with high magnification of the operating microscope because a thick normal ependymal layer covers the floor. Only the topographical information obtained by preoperative CT scanning helps us to decide which area of the ventricular floor should be sacrificed to advance into lesions in the thalamus. With a small and deep-seated lesion a direct approach to the lesion with minimal damage to the normal structure is more difficult than with a large lesion. Here intraoperative radiological study is helpful. Although the best investigation is intraoperative CT scanning, intraoperative plain skull films are also very helpful after a small hemoclip has been posi-

tioned in the bottom of the operating field. In addition to the two-stage operation mentioned previously, open operation combined with the stereotaxic method is recommended for dealing with small and deep-seated lesions. We isolate the tumor from the surrounding critical tissues with a small cotton patty held with bipolar coagulating forceps when the consistency

of the tumor is different from that of the surrounding tissues. The patties used for the purpose are conventional cotton pieces sized 1 X 2 or 1 x 3 cm. The patty should be changed frequently, usually every minute, because its surface becomes too smooth with adhesion and absorption of the surrounding brain and tumor tissues. Ideal dissection along the tumor border can be performed with the rough surface of a new patty. We can feel the different consistencies between the tumor and normal tissues through the small ball-shaped patty, something like the old-fashioned method of finger enucleation. When a tumor has no clear border, radical removal must not be attempted; only partial removal is performed inside the safety zone that is determined from the preoperative CT scan (Fig. 22.2B). When the normal brain is completely isolated from the tumor the normal areas should be protected with different kinds of cotton patty from the bemsheets used for dissection: the bemsheets are hard whereas the patties are soft. Using two different kinds of patty helps us to discriminate completely finished areas from ongoing ones. When we check the area completely isolated from the tumor a little later, its surface will often have developed a pathological appearance due to clots and microbleeding. Topographical Anatomy around the Third Ventricle The topographical anatomy along the approach from the parietal cortex through the trigone to the thalamus must be thoroughly understood. When we are operating in the depth of the brain via a small transcortical route, the anatomical orientation is sometimes lost, especially when we operate under a microscope. In this approach the most important landmark is the shape of the trigone and the three directions of the anterior, inferior, and posterior horns. If the ventricle is missing the approach cannot be successful. We must master the topographical atlas for stereotaxic surgery so that the size and location of the important structures around the ventricle can be known in millimeters. The actual location of a critical structure, however, will be different from normal because of the tumor. During operation we must therefore keep in mind not only the normal anatomy but also the pathological location of critical structures in each case, as obtained from the preoperative CT scan. The three topographical sections of Figures 22.2C and 22.3 are reconstructed from the atlas of Schaltenbrand and Bailey (11); Section A is the topography along the anterior route from the posterior parietal cortex to the thalamus, Section В is along the central route, and Section С is along the posterior route. The distance from the cortex to the ventricle is about 5 cm. The width of the thalamus is 15 to 25 mm. The dorsomedial portion of the thalamus about 10 mm lateral and 15 mm ventral from the dorsomedial corner of the thalamus (i.e., the pulvinar thalami and nucleus dorsomedialis) can be directly attacked: this portion is, in our experience, relatively silent. Pitfalls during the Intraoperative Period The craniotomy must be made in the correct area to make a correct cortical incision in the medial parietal lobe. When advanced hydrocephalus is present the cortical incision should be made slightly to the occipital side in the window of the small craniotomy because the slackened brain sinks to the frontal side after the egress of CSF. Although the incision is less than 3 cm long the arachnoid membrane around it should not be damaged because the arachnoid protects against excessive elongation of

the incision by retraction. At the end of operation, however, the length will often have increased to about 4 cm. Unless the patient's head is perpendicular the approach to the trigone is occasionally difficult. The orientation from the cortex to the trigone can be easily realized when the head is perpendicular. A ventricular tap before transcortical incision is recommended: tracing the tap facilitates entering the ventricle. After the ventricle is opened the patient's head is rotated to the affected side. The use of four tapered saptulas connected to self-retaining retractors and the Sugita frame is one of the most essential techniques in this approach. Although four spatulas are set up around the cortical incision, the brain around the route of approach should be retracted with only two spatulas at a time. Retraction of only one side of the brain helps to maintain a normal blood circulation in the other side. A larger operating field than the cortical incision is obtained in this manner by changing the direction of retraction. Retraction with four spatulas simultaneously does not enlarge the operating field much and damages the brain tissue more than does the use of only two spatulas. Another important point is that the observation angle of the microscope must be frequently changed: tilting and sloping of the microscope reduces the need for brain retraction. The communicating channels of the exposed ventricle located in the bottom of the operating field must be obstructed during the procedure with large pieces of cotton patty to prevent spread of blood to the entire ventricular system, especially when dealing with a hemorrhagic tumor. We occasionally become confused about the anatomical orientation during the procedure around a deep-seated lesion through an operating microscope that provides us a limited visual field, especially when we approach through a small cortical incision. To keep our sense of direction, the most important landmark is the ventricle. When the third ventricle has been exposed, the foramen of Monro and occasionally the entrance of the aqueduct give us further information about the anatomical relationships. The more the ventricle is exposed the more easily can we obtain anatomical orientation. During removal of thalamic and parathalamic tumors that are located close to critical structures such as the internal capsule and the optic radiation we must keep in mind the three-dimensional topographic mapping reconstructed from the preoperative CT scan. We should pay attention to the fact that the anatomical relationships are often far different from normal in the case of large thalamic tumor; for example, we encountered a case where the internal capsule was shifted more than 3 cm laterally from its normal position. The most difficult decision concerns how radically tumors in this area can be removed. We have not encountered postoperative additional neurological deficits that were permanent when only the tumor mass was removed, even in an area where the thalamus and internal capsule would have been located in the normal brain. A relatively safe zone to attack is the posterior dorsomedial area of the thalamus; this is the anterolateral floor of the trigone. We should refrain from advancing in a deeply anterolateral direction from the trigone, especially when the tumor border is unclear. Complications Associated with the Technique It is said that an approach through the parietal lobe of the dominant hemisphere is likely to cause the Gerstmann syndrome as a postoperative complication. However, we have never experienced this in 43 cases operated through the parietal transcortical approach, 20 of which were

on the left side. To minimize such postoperative complications the cortical incision should be made as small as possible and retraction of the brain should be as little as possible, with the aid of the four tapered spatulas connected to self-retaining retractors. If the cortical incision is made in the far lateral area of the parietal lobe, there is a higher possibility of such postoperative complications as homonymous hemianopsia or sensory aphasia. The complication one must be most careful to avoid is injury of the normal internal capsule and a large area of the normal thalamus. To avoid additional neurological deficits the normal tissues should never be removed in the deep anterolateral area around a tumor located in the thalamus. In this area one must leave a small portion of the tumor when the tumor border is not clear. Case Reports Case 1 This 60-year-old man had had right hemiparesis for 3 months. The CT scan showed a mass in the ventroposterior area of the left thalamus (Fig. 22.4). First, radiation therapy was tried without effect. A stereotaxic exploration was performed 1 month before the main operation, and a small piece of a silver clip was left in the lesion after the removal of two pieces of tumor tissue, which were histologically determined to be benign astrocytoma. With the patient in the prone position a direct approach to the tumor was performed through a left medial parietal corticotomy. After the trigone had been exposed an incision was made in the portion anterolateral to the trigone, which was a roof of the nucleus dorsomedialis. The tumor was found intramedullarly at a depth of 6 mm from the bottom of the ventricle (Fig. 22.5). The small piece of clip that had been left was very useful in getting the correct orientation to the tumor. The tumor had a hard capsule and its center was cystic, containing cheeselike necrotic tissue. After removal of the mass with the capsule, a thin

arachnoid membrane and many small arteries were visible in the bottom of the dead space; this was the third ventricle. After operation the right hemiparesis was slightly worse for 2 weeks and then returned to the preoperative state. No postoperative additional deficits were observed. The patient was able to walk. Case 2 This 46-year-old man had only a brief (2-week) history of severe headache. A transcortical approach was made through a 3-cm cortical incision in the right medial parietal lobe. At a depth of about 6 cm the trigone of the right lateral ventricle was encountered. After coagulating and resecting the choroidal plexus the ependymal layer of the ventricular floor and the thalamic mass was entered. Decompression was readily accomplished because many clots were found inside the tumor and its border was clear. Blood loss was less than 100 ml. The histology diagnosis was Grade III astrocytoma. The postoperative course was uneventful except for a transient left hemiparesis. Irradiation was done (Fig. 22.6).

Case3 This 57-year-old man had complained of nausea, vomiting, headache, and anorexia for more than 10 months. Slight right hemiparesis appeared a few weeks before operation. A transcortical approach was made through the left medial parietal lobe. The trigone of the left lateral ventricle was entered at a depth of 3 cm. Immediately a tumorous bulging was found in the anterolateral wall of the trigone. The tumor was covered with a thin ependymal layer, which was carefully peeled; the posterior border of the tumor was delineated. The tumor was rather vascular; we had a difficult time coagulating the bleeders. The medial wall of the trigone was free of tumor; communication to the contralateral ventricle was made through the medial wall and CSF was drained. The brain became slack. We began to remove the tumor by suction dissection. Most of it was removable by suction; its color and consistency were different from those of the normal brain. The two-suction method, with a large suction in the dominant hand and a small one in the other, was useful at times. As we proceeded with suction dissection we found that the anterior border was rather well demarcated. The lateral ventral extent of the tumor was not followed too far because we knew from the preoperative CT

scan (Fig. 22.7) that the tumor had extended down to the midbrain. At the end of the procedure, the space created in the depth measured 3 x 3.5 x 3.5 cm. Subtotal removal with substantial decompression was believed to have been accomplished. The histological finding was Grade II or III astrocytoma. Postoperatively right hemiplegia appeared for several days and then mild hemiparesis continued. The patient was transferred to another hospital for irradiation. References 1. Castaigne P, Ron dot P, Ribadeau-Dumas JL, et al: Ataxie optique localisee au cote gauche dans les deux hemichamps visuels homonymes gauches. Rev Neurol (Paris) 131:23-28, 1975. 2. Castaigne P, Pertuiset B, Rondot P, et al: Ataxie optiquedans les deux hemi champs visuels homonymes gauches apres exerese chirurgicale d'un anevrysme arteriel de la paroi du ventricule lateral. Rev Neurol (Paris) 124:262268, 1971. 3. Cramer F: The intraventricular meningiomas: A note on the neurologic determinants governing the surgical approach. Arch Neurol 3:98, 1960. 4. Dandy WE: Diagnosis, localization and removal of tumors of the third ven tricle. Johns Hopkins Hosp Bull 33-188, 1922. 5. Gardner WJ, Turner CA: Primary fibroblastic tumours of the Choroid plexus of the lateral ventricles. Surg Arch 66:804-809, 1938. 6. Hirsch JF, Zouaoui A, Renier D, et al: A new surgical approach to the third ventricle with interruption of the striothalamic vein. Ada Neurochir (Wien) 47:135-147, 1979. 7. Kempe LG: Operative Neurosurgery. Berlin, Springer, 1968, vol 1, pp 196202. 8. Lapras C, Deruty R, Bret PH: Tumors of the lateral ventricles. Adv Tech niques Neurosurg 11:103-167, 1984.

9. Maurizio F, Mario S, Giulio M, et al: Meningiomas of the lateral ventricles. J Neurosurg 54:64-75, 1981. 10. Okudera H, Kobayashi S, Sugita K: A simple three-dimensional display model based on recorded CT scan films for surgical reference. Acta Neurochir (Wien) 71:47-53, 1984. 11. Schaltenbrand G, Bailey P: Einfuhrung in die stereotaktischen Operationen, mit einem Atlas des menschlichen Gehirns. Stuttgart, Thieme, 1959, vol 3. 12. Sugita K, Hirota T, Mizuno T: A newly designed multipurpose microneurosurgical head frame. J Neurosurg 48:656-657, 1978. 13. Tonnis W: Behsndlungsergebnis bei Geschwulsten des Seitenventrikels. Langenbecks Arch Klin Chir 284:446-450, 1956. 14. Torre E, Alexander E Jr, Davis CH Jr, et al: Tumors of the lateral ventricles of the brain: Report of eight cases, with suggestions for clinical management. J Neurosurg 20:461-470, 1963.

23 Infratentorial Supracerebellar Approach Bennett M. Stein, M.D.

The material in this chapter is based on over 90 operations for tumors of the pineal region. Not accepting the premise that tumors of the pineal region are generally malignant and therefore should be treated by a more conservative approach (1, 2, 6) (a shunt operation to relieve hydrocephalus followed by indiscriminate radiation), I have approached the problem in a more radical fashion that requires definition of the tumor before the institution of therapy (24-27). With an increased awareness of the pathology of these tumors from autopsy series and histological confirmation at operation, it has become apparent that there are a wide variety of tumor types existing in the pineal region (5, 10, 12, 19, 22, 30). These tumors take origin from various structures, not only the pineal body but the tela choroidea, thalamus, and midbrain. The tumors run a gamut from benign encapsulated tumors to malignant invasive tumors that are prone to seed. The categorization that we have adopted is based on the scheme of Rubinstein (21), which places the tumors into three broad categories: (a) Germ cell tumors including the benign dermoids and epidermoids ranging through the intermediate tumors of the mature teratoma type (Fig. 23.1) to the extremely malignant, invasive, and prone to seed germinoma, embryonal carcinoma, and choriocarcinoma, (b) Pineal cell tumors including the pineoblastoma and the pineocytoma—the latter benign or malignant, (c) Supporting cell tumors including the astrocytoma (Grade 1 to Grade 4), meningioma, ependymoma, oligodendroglioma, and other rare tumors arising from cells in association with the pineal body but not actually part of it. The exact identification of these tumors allows a more enlightened form of therapy. Some of the principles in the treatment of these tumors are old and well established, such as the treatment of hydrocephalus by a shunt. Others are new and innovative, such as the operative microsurgical resection of benign encapsulated tumors and the identification of malignant tumors followed by radiation, chemotherapy, or combinations thereof. The armamentarium of the neurooncologist has been broadened through the use of multiple chemotherapeutic agents. The

cornerstone of our method of treatment rests on the development of a safe and effective operative procedure. Accordingly, we have found that the only certain way to identify and thereby effectively treat tumors of the pineal region is to obtain abundant tissue for histological examination including light microscopy, histochemical studies, and electron micros-copy. The use of computerized tomographic (CT) scanning with and without contrast administration, angiography, biological markers (including the beta subunit of human chorionic gonadotropin alpha-fetoprotein) (13,14) and more recently the magnetic resonance imaging (MRI) scanner has not allowed us to identify the histological nature of these tumors with a high degree of accuracy (Figs. 23.2 to 23.4). Historical Perspective The operative experience with pineal tumors is intriguing to review. Dandy (4) was the first neurosurgeon to tackle tumors of the pineal region aggressively. At that time operation on these deep-seated tumors was frowned upon by his colleagues (3). Dandy was a student of the pineal body, studying the functions of the gland as well as developing operative approaches to reach this region in the experimental animal and humans. Unfortunately, microsurgical technique was not available to him at that time and anesthesia for these difficult neurosurgical cases was in its infancy. With these two strikes against him it was not unexpected that his operative mortality and morbidity were high. Perhaps as a result of this and other fruitless attempts to remove or otherwise surgically treat tumors of this region, the direct operative approach to pineal tumors fell into a state of disrepute. However, other neurosurgeons developed and promoted various operative techniques to explore the region. Whereas Dandy (4) utilized a parafalx, parietal approach with resection of a portion of the posterior corpus callosum and approach to these tumors through the deep venous system, Van Wagenen (29) used a transventricular approach through the dilated trigone of the lateral ventricle (a rather convoluted and involved approach dependent entirely upon the presence of hydrocephalus). Poppen (15) fostered an approach above the tentorium either medial or inferior to the occipital lobe, thereby approaching the tumor low and adjacent to the deep venous system. None of the results from these various operative procedures, however, were what one could call spectacularly successful. Krause (11) in the 1920s with a primitive neurosurgical technique utilized the posterior fossa approach to the pineal region and was able to explore the region successfully in three cases, removing some of the tumors. We have exclusively used the operation as developed by Krause (18, 27). Advantages and Contraindications The Infratentorial Supracerebellar approach is essentially a posterior fossa approach to the pineal region and in our estimation has certain advantages, (a) The approach is central to tumors that are basically midline, (b) The exposure is excellent when carried out with the patient in the sitting position and provides space comparable to that obtained with supratentorial approaches, (c) The tumors are approached ventral to the deep venous system, (d) There is no morbidity related to the parietal or occipital lobes as is experienced with supratentorial exposures. It is for these reasons that I prefer this approach and have used it in virtually all of the cases of pineal region tumor and of some vascular malformations. This approach, however, should not be used when the tumor extends dorsally, in which case it may envelop or grow above the deep venous system. Neither should the approach be used when the

tumor is large and extends laterally into the region of the trigone of the lateral ventricle (Fig. 23.3). Occasionally, additional lateral exposure may be gained by cutting the tentorium from the posterior fossa exposure. Fortunately, most of the tumors are of modest size and do not extend to a major degree in the aforementioned directions. Operative Exposure and Technique As with any innovative operative approach to the central nervous system, utmost attention to detail is absolutely essential in achieving the desired result. This statement holds for the simplest operation such as a shunt to complicated ones such as the approach to a basilar aneurysm or acoustic tumor. Obviously, the same principle applies when one is reaching the central area of the brain with a limited degree of exposure (27). The operation microscope is absolutely essential to these operations. It would be impossible to perform them with headlight and loupes and this should not be attempted. Also the use of long instruments is preferred because the distances are of significant magnitude. Some of the conventional microsurgical instruments are not applicable to operation in this region. Similarly, innovative instruments such as the laser and the Cavitron are used with difficulty in this region because of the small exposure and the angulation required during the operative approach. Nevertheless, these two instruments have added to the armamentarium of the surgeon approaching tumors of this region. Perhaps the most important aspect of the operative procedure is the positioning of the patient so that the trajectory utilized is comfortable and effective in exposing the pineal region. This aspect of the operation as well as other fine points are stressed in the following discussion. Utilizing illustrations and operative photographs, a step-by-step approach to this region is presented. Patient Preparation The patients are prepared for operation by the performance of CT scanning with and without contrast administration. Especially helpful are coronal and sagittal reconstructed views. More recently we have been using the MRI scanner to identify the location of the tumor. Rarely, we perform arteriography. Some of the tumors are quite vascular but that fact will be apparent at the time of operation and knowing it beforehand does not reassure the surgeon or the patient or make the operation significantly easier. The vascularity of these tumors derives from branches of the posterior medial and lateral choroidal arteries. These can be sacrificed with impunity. Certainly, one does not want to wander into a vascular malformation such as a vein of Galen annomaly or a true arteriovenous malformation in this region. However, the presence of this type of lesion is usually suspected through the use of the CT scanning and in these instances arteriography is mandatory. Twenty-four to 48 hours before the operation, the patient is given high dose steroids, which are continued. If the patient has untreated hydrocephalus, this must be relieved at the time of the operation by ventricular drainage or the performance of extrancranial shunting (ventriculoperitoneal shunt) at some time before the definitive operative procedure. Interestingly, many of the operations that we have carried out are referrals and in many of these patients shunts have been performed previously; unfortunately in some of the cases radiation has also been given. Nevertheless, it is absolutely critical to the operative procedure that the ventricular system be thoroughly decompressed before the operation or during the operative procedure. If more relaxation is required, we do not hesitate to use

mannitol. Often when the cisterna magna is exposed, the release of cerebrospinal fluid (CSF)— even if a block exists at the level of the incisura or the aqueduct — will relieve tension within the posterior fossa. The choice of anesthesia is left to the anesthesiologist. Ours prefer the use of inhalation anesthesia, which will generally raise intracranial pressure. This elevation, however, can be defeated with hyperventilation and the appropriate use of ventricular decompression and dehydrating agents. Spinal drainage is not utilized in these cases. All of the operations are done with the patient in the sitting position with the usual preparation for this position, such as Doppler monitoring and elevation of the legs. Depending upon the preference of the anesthesiologist, a central venous catheter may be utilized. Because of the difficulty experienced in placing a central venous catheter either the evening before or at the time of the operation, we have tended to abandon this procedure. However, Doppler monitoring is essential to detect small amounts of air entering the venous system. When this phenomenon is detected early, it is possible to take measures to prevent further ingress of air to the venous system. Positioning The patient is positioned in a "sitting-slouch" position (Fig. 23.5A). The patient's body is placed well down on the operating table, the shoulders are supported with a firm pillow or sandbag, and the body is placed in a С configuration. A small pillow is required under the lumbar lordosis to support the back. The head is then grasped by the three-point vise type of head holder and positioned with a moderate degree of flexion so that two fingers can be placed between the patient's chin and the sternum. Care must be taken not to overflex the head. In discussing the positioning, an anecdotal tragic experience should be recounted. We treated a patient with a large tumor of the pineal region that turned out to be an infiltrating astrocytoma. One month before operation on the pineal tumor, the patient was involved in an automobile accident; he was not rendered unconscious but suffered from headaches afterward. The pineal tumor was picked up on a routine CT scan done for headaches. There had been no complaints referable to the neck. When positioning him in the characteristic sitting-slouch position, the anesthesiologist noted that there was some difficulty in controlling hypotension. The operation went well, and a modest portion of the tumor was resected. In the recovery room, it was noted that the patient was quadriplegic, but alert. Cervical spine films and a myelogram that evening showed no abnormalities. One week later, flexion-extension views were taken of the cervical spine. No abnormalities were noted as the quadriplegia persisted. At that point, on a hunch that there might have been another disease process within the cervical spinal cord, a CT scan with and without contrast infusion was carried out. To our surprise, fractures dating back to the auto accident were noted in the laminae and pedicles of C-5 and C-6. At the time of the CT scan these were nondisplaced but obviously, with the degree of distortion of the cervical and thoracic spine during the positioning of the patient, these fractures permitted a guillotine effect on the cervical cord, resulting in quadriplegia. Therefore, if there is any hint of abnormality in the cervical spine either acute or long-standing, of necessity the position should be modified or the operative procedure should be abandoned. Alternatively, we have not been comfortable with the prone position nor have we tried the three-quarter prone position. A self-retaining retractor of the Greenberg type is positioned with the holder affixed to the operating table and the retractor post arranged in a U-shaped fashion, open inferiorly around the operative procedure (Fig.

23.5В). These bars will be the recipients of the two retractor arms (position to be detailed later), a remote irrigating system, and the tray for small cottonoids and Telfa pads. The operation microscope is arranged so that the inclined eyepiece is rotated upward, allowing the surgeon to look down rather than straight ahead or upward. A 275-mm objective is utilized. With this objective, it is possible to place long instruments comfortably into the wound and to use the Cavitron aspirator with the long curved tip. A television camera and the binocular observer arm are located in the usual position on the microscope and the whole affair is covered by a transparent drape. Exposure A forearm rest is arranged at the head of the table. This can be attached to the seat upon which the surgeon sits or directly to the top of the table after the hinged head piece has been removed. A long midline incision is performed. This should extend down to approximately C-4 and up to the

region of the lambdoid suture, well above the torcular region (Fig. 23.6A). The extra length of the incision is necessary to gain separation of the muscles and fascia from the suboccipital region. If the patient's ventricular system requires decompression, a separate incision is made to the right side in the region of the lambdoid suture to drain the trigone of the lateral ventricle. The amount of CSF removed can be regulated during the operation to obtain maximal relaxation of the posterior fossa contents, while not overdraining the system. The midline incision is brought directly down through the nuchal ligament to the bone of the suboccipital region. It is not necessary to denude the first or second cervical spines of

their muscle attachments although the incision must go down to the posterior aspect of these two structures. The craniectomy not involving the foramen magnum does not require the removal of the oblique or recti muscles from C-l, C-2. A wide exposure of the suboccipital bone, however, is necessary (Fig. 23.6B). A self-retaining retractor of low profile is utilized to hold back the muscles and fascia (Fig. 23.6C). It is apparent from Figure 23.9, which is a sagittal view, that the center of the route to the pineal gland is just below the tentorium or in this case the torcular region. The remainder of the craniectomy is only to provide access for the various instruments and the beam of the operation microscope. Using the air drill, numerous burr holes are placed because the bone is the thickest over the torcular and the sinus regions. A number of passes with the drill in this area are necessary to at least thin out the bone so that rongeuring is done in an effortless fashion. The craniectomy should extend just above the transverse sinuses and include all of the torcular region so that the view is not obscured by overhanging bone and the tentorium can be elevated ever so slightly by a self-retaining retractor. Laterally, the craniectomy is carried out almost to the mastoid groove and inferiorly 1 cm short of the foramen magnum region. This provides a wide oblong-shaped craniectomy (Fig. 23.6B). Again, the most essential portion of the craniectomy is that uncovering the sinsuses and the torcular region. At this point the surgeon may determine the pressure within the posterior fossa by palpating the dura mater. If the dura is tense and nonpulsatile, then the ventricle must be further decompressed or dehydrating agents must be instituted. It is folly to open the dura of the posterior fossa with a tense cerebellum. There is little choice of the fashion in which the dura is opened (Fig. 23.7). It is opened to expose maximally the superior surface of the cerebellum. This requires a V-shaped incision with the limbs relatively close to the midline so that the

"cathedral effect" of the tentorium does not obscure the central or midline view and so that the dura in the central area may be retracted upward to the greatest extent (Fig. 23.7B). Secondary incisions are then carried out laterally toward the lateral sinuses so that, at the completion of the incision, there are three flaps, a central one and two lateral ones, which reflect upward and expose the superior surface of the cerebellum (Figs. 23.7С, 23.8). These flaps may be held upward by stay sutures attached to rubber bands placed around the Greenberg retractor. Inferiorly, the dura mater is left as a sling to support the cerebellar hemispheres during retraction. As noted previously, if additional space is required, this may be obtained by making a small opening in the arachnoid of the cisterna magna and releasing CSF from this cistern. The Greenberg retractor is then arranged with the various posts and bars as shown in Figure 23.7 A. This provides two selfretaining retractors, a cottonoid tray, and a fixed

irrigating system, which is composed of a 16 gauge needle attached to intravenous tubing and then to a syringe. The tip of the needle may be bent and then positioned with the retractor so that the irrigation goes directly into the area of the pineal tumor when this portion of the operation is attained. Before placement of the retractor blades, the veins bridging between the cerebellum and the overlying dura mater must all be coagulated and divided (Figs. 23.9 and 23.10). By gravity, with the patient in the sitting position, the cerebellum will drop under most circumstances and immediately provide 1 to 1.5 cm of space between its surface and the tentorium (Fig. 23.8). At this point, protective Telfa is placed over the cerebellum and a self-retaining retractor is placed into the region of the incisura hard against the torcular and straight sinus, lifting these structures upward. Another retractor usually 0.5-in. wide is

looped low to come over the cerebellum in the region of the vermis. This is bent in an Sshaped fashion and depresses the cerebellum (Fig. 23.10A and B). The tip of this retractor should be placed directly at the anterior aspect of the cerebellum. At this point with the naked eye or with loupes it is possible to see the pineal region and in almost all cases the arachnoid will be thickened and opalescent. The operation microscope is now brought into the field and is used during the remaining portion of work on the tumor. The thickened arachnoid in the region of the pineal tumor and the incisura is then opened with a small arachnoid knife, long bayonet scissors, and the assistance of a long bipolar cautery under

irrigation. The precentral cerebellar vein will be noted coming directly from the region of the anterior vermis toward the great vein of Galen, which is just visible through the thickened arachnoid (Figs. 23.11 and 23.12). The precentral vein may be coagulated with impunity to release the anterior vermis, permitting further exposure of the pineal region. A few small branches of the choroidal and superior cerebellar arteries may

be noted in the thickened arachnoid overlying the tumor, and these can be indiscriminantly coagulated and divided. Care must be taken not to injure the large veins laterally (the veins of Rosenthal running along the medial aspect of the temporal lobe). At the time of this initial exposure the trajectory of the operation microscope is generally in line with the vein of Galen and slightly dorsal to the main bulk of the pineal tumor. Care must be taken not to continue to pursue this direction, for this will bring the surgeon into conflict with this important deep venous structure (16). As soon as the vein of Galen is identified, the trajectory of the microscope is angled downward 10 to 20°. At the same time the inferior retractor is made to retract the cerebellum further, exposing a broad area

of the posterior aspect of the tumor (Fig. 23.13). It may be necessary to cut the thickened arachnoid well out laterally just above the margin of the cerebellum to release the cerebellum from this tethering. In Figure 23.11A the sequence of these incisions is shown. First, the incision along the anterior border of the cerebellum (I and 2), then division of the precentral cerebellar vein (3), and then cuts extending upward to the region of the vein of Galen and the confluence of venous structures (4 and 5). Tumor Management The posterior face of most tumors will now be exposed adequately to permit progress to further stages of the operation, including tumor removal. Small branches of the choroidal arteries especially in vascular tumors will course over this posterior surface of the tumor and can be coagulated with impunity (Fig. 23.11B). The tumor is then opened with a fine long-handled knife and bayonet scissors. Coagulation is used where necessary depending upon the vascularity of the tumor capsule (Fig. 23.14). Portions of the tumor are then removed by tumor forceps for frozen tissue diagnosis. The accuracy on these frozen tissues is about 50%, perhaps no more than the law of chance. This is explained by the difficulty in diagnosing pineal region tumors on frozen tissue even by an experienced pathologist and is one reason that we prefer open biopsy with abundant tissue for permanent sections. The interior of the tumor then may be decompressed as necessary (Figs. 23.14B and 23.15). In those tumors that are benign and encapsulated, it is possible to dissect laterally, superiorly, and inferiorly around the capsule of the tumors after gutting the interior. Many tumors are attached superiorly in the region of the velum interpositum and sometimes laterally to the medial aspect of the pulvinar or the walls of the third ventricle. Attachments inferiorly to the region of the quadrigeminal portion of the midbrain vary according to the nature of the tumor but are generally modest in those encapsulated benign tumors because of the small choroidal arteries bridging this gap. In tumors that are exophytic or that fungate out of the midbrain, the area

of attachment will be broad and ill-defined and should not be violated (Fig. 23.16). Various instruments may be used to core out the interior of the tumor. Some of the tumors are extremely soft and, although varying in vascularity, may be suctioned by a 7 F suction tip. Other firmer tumors such as meningiomas, some astrocytomas, and teratomas are variegated and much of them are firm, even calcified; this tissue requires instrumentation such as the laser or Cavitron aspirator. With the long curved tip of the Cavitron, it is now possible with some difficulty to place it into the operative area to remove these central cores of firm, sometimes calcified

tissue (Fig. 23.14B). Some of the teratomas have even contained structures resembling malformed teeth, and these have been the most difficult to remove. Unfortunately, the heavily calcified areas are often the most vascular. A long bipolar cautery is of assistance in these maneuvers and the use of long tumor forceps, curettes, and scissors can be combined to core out the interior of the tumors. Once a large central portion of the tumor is removed, it will become apparent whether the tumor is infiltrating or encapsulated. With infiltra-tive tumors, it is wise not to attempt radical resections; these are tumors that best respond to radiation, chemotherapy, or combinations thereof. Pursuing malignant infiltrating tumors into surrounding structures will often lead to severe neurological deficits that may be permanent. The most difficult portion of encapsulated tumors to remove is that portion that hollows out the midbrain or extends laterally into the trigone of the lateral ventricles. Because of the sitting position the portion intimate to the midbrain is most difficult to remove because it must be elevated and then microdissection techniques must be carried out between it and the critical midbrain structures. This can be accomplished, however, and it may be possible for a skilled assistant to hold the tumor up with forceps while the surgeon works in this interface, coagulating and dividing the vessels. The prepositioned irrigator is of immense help in this maneuver. As the tumor is gradually delivered or the large decompression is made through the center of the tumor, openings into the posterior third ventricle frequently occur. This gives the surgeon an excellent perception of the relationship of his operative exposure and the third ventricle as related to the tumor. Tumors with minimal attachment will often roll out at this point and large pieces may be removed as a whole (Figs. 23.15 and 23.17). Again, particular attention is paid to the attachment, especially to the velum interpositum, because a hole in the internal cerebral veins or the vein of Galen will lead to profuse bleeding that is difficult to control. Other than packing, it is difficult to place hemostatic agents on these structures and have them stay in place. Similarly, an attempt to cauterize these distended veins may lead to larger openings. With experience, it is possible to determine which tumors are resectable and which should be left short of total resection and also the tumors in which only a biopsy should be done. Surprisingly, we have encountered our only severe difficulties — three mortalities — in patients with malignant tumors that were highly vascular and difficult to manage at operation. Only small portions of the tumors in these patients were removed and yet hemostasis was inadequate, leading to massive postoperative hemorrhage and death. One of these cases was a glioblastoma in which hemorrhage occurred 1 week after the operation, presumably as a vessel was eroded. The other two hemorrhages occurred immediately after operation on malignant pineocytomas where bleeding was difficult to control at the time of operation. We have used all of the standard hemostatic agents and prefer none over others. When total removal is possible, the tumor is usually not highly vascular and hemostasis after such radical resections is usually quite easy. CSF Diversion and Closure In some cases where a block to the aqueduct remains after the central portion of the tumor is cored out, it is possible to leave a small ventricular catheter through the tumor into the third ventricle, subsequently leading it down and attaching it to the dura mater while placing the distal end into the cisterna magna. This permits an internal shunting of CSF much like a Torkildsen procedure and may be used in lieu of an extracranial

shunt. Once hemostasis is secured, we have not left metallic markers in the area, foregoing this because of the possible use of MRI scanning. All protective mechanisms are removed from the cerebellum and the dura is closed in a sling like fashion to resupport the cerebellum and replace it to its normal position. All of the patients had craniectomies and there has been no replacement of the bone. The muscles and fascia are closed in appropriate layers, and we have used wound drainage for 12 hours. The patient is kept in a sitting position, extubated, and taken to the recovery room where ventricular drainage is continued as necessary. A high dose of dexamethasone is also continued. Malignant Tumor Management In the comprehensive treatment of these patients when malignant tumors are encountered, our protocol mandates the performance of three CSF evaluations during the postoperative period and myelography studying the lumbar and thoracic regions looking for evidence of seeding. Wherever possible we have been removing CSF via either the shunt system or the cisterna magna to evaluate it from a cytological viewpoint (8). We have developed a protocol for the treatment of these patients which is founded on a detailed identification of the tumor. This information plus the evaluation of CSF and myelography has led us to the following treatment regimen: Surgery In those benign and encapsulated tumors that are totally removed no further treatment is necessary (27). This group comprises approximately 30% of the tumors encountered and consists of the low grade cystic or solid astrocytoma, dermoid, epidermoid, and meningioma (20) and the

mature encapsulated three-germ layer teratoma. Some of the pineocytomas have fallen into this cateogry, as has the rare ependymoma. Radiation Those tumors that have the features of a germinoma, including the small patches of lymphocyte-like cells in the interstices and the larger nucleated cells, have a tendency to spread in the region of the third ventricle or by seeding throughout the CSF spaces. Our primary treatment has been radiation therapy. When seeding is not evident, radiation is directed at the tumor and the region of the third ventricle. When seeding is apparent from a myelogram or evaluation of the CSF, radiation is given to the entire neuraxis, 5500 rads to the tumor, 3500 rads to the entire brain, and 3500 rads to the spine (17, 28). Chemotherapy In tumors of the primitive embryonal or choriocarcinoma type, chemotherapy has been the first line of defense. This consists of the Einhorn triple chemotherapy treatment (7). In tumors that have received radiation treatment and recur, chemotherapy may follow (9, 13, 23). Results Using this operative approach, we have had 3 mortalities out of 90 operative experiences. These 3 mortalities occurred in patients with malignant tumors in which hemostasis was a problem. Morbidity was generally temporary and consisted of somnolence and disturbance of extraocular movements such as Parinaud's syndrome (Fig. 23.18).

References 1. Camins MB, Schlestnger EB: Treatment of tumors of the posterior part of the third ventricle and the pineal region: A long term followup. Ada Neurochir (Men) 40:131-143, 1978.

2. Cummins FM, Taveras JM, Schlesinger EB: Treatment of gliomas of the third ventricle and pinealomas: With special reference to the value of radiotherapy. Neurology (Minneap) 10:1031-1036, 1960. 3. Cushing H: Intracranial tumors: Notes upon a series of two thousand verified cases with surgical mortality pertaining thereto. Springfield, IL, Charles С Thomas, 1933. 4. Dandy WE: Operative experience in cases of pineal tumor. Arch Surg (Chi cago) 33:19-46, 1936. 5. DeGirolami U, Schmidek H: Clinicopathological study of 53 tumors of the pineal region. J Neurosurg 39:455-462, 1973. 6. Donat JF, Okazaki H, Gomez MR, Reagan RJ, Baker HL Jr, Laws ER Jr: Pineal tumors: A 53-year experience. Arch Neurol 35:736-740, 1978. 7. Einhorn LH, Donohue J: Cis-diamine dichloroplatinum, vinblastine, and bleomycin combination chemotherapy in disseminated testicular cancer. Ann Intern Мей 87:293-298, 1977. 8. Gindhart TD, Tsukahara YC: Cytologic diagnosis of pineal germinoma in cerebrospinal fluid and sputum. Acta Cytol 23:341-346, 1979. 9. Ginsberg S, Kirshner J, Reich S, et al: Systemic chemotherapy for a primary germ cell tumor of the brain: A pharmaco-kinetic study. Cancer Treat Rep 65:477-483, 1981. 10. Herrick MK, Rubinstein LJ: The cytological differentiating potential of pineal parenchymal neoplasms (the pinealomas): A clinicopathological study of 28 tumors. Brain 102:321-332, 1979. 11. Krause F: Operative frielegung der vierhugen, nebst beobachtungen uber hirndruck und dekompression. Zentbl Chir 53:2812-2819, 1926. 12. Neuwelt EA, Ginsberg M, Frenkel et al: Malignant pineal region tumors: A clinico-pathological study. J Neurosurg 57:597-607, 1979. 13. Neuwelt EA, Frenkel EP, Smith RG: Suprasellar germinomas (ectopic pineal omas): Aspects of immunological characterization and successful chemo therapeutic responses in recurrent disease. Neurosurgery 7:352-358, 1980. 14. Ono N, Takeda F, Uki J, et al: A suprasellar embryonal carcinoma producing alphafetoprotein and human chorionic gonadotropin; treated with combined chemotherapy followed by radiotherapy. Surg Neurol 18:435-443, 1983. 15. Poppen JL: The right occipital approach to a pinealoma. J Neurosurg 25:706710, 1966. 16. guest DO, Kleriga E: Microsurgical anatomy of the pineal region. Neurosur gery 6:385-390, 1979. 17. Rao YTR, Medini E, Haselow RE, et al: Pineal and ectopic pineal tumors: The role of radiation therapy. Cancer 48:708-713, 1981. 18. Reid WS, Clark K: Comparison of the Infratentorial and transtentorial ap proaches to the pineal region. Neurosurgery 3:1-8, 1978. 19. Ringertz N, Nordenstam H, Flyger G: Tumors of the pineal region. J Neuropathol Exp Neurol 13:540-561, 1954. 20. Rozario R, Adelman L, Prager RJ, Stein BM: Meningiomas of the pineal region and third ventricle. Neurosurgery 5:489-495, 1979. 21. Rubinstein LF: Cytogenesis and differentiation of pineal neoplasms. Hum Pathol 12:441-448, 1981. 22. Schmidek HH: Pineal Tumours. New York, Masson, 1977. 23. Seigal T, Pfeffer R, Catane R, et al: Successful chemotherapy of recurrent intracranial germinoma with spinal metastases. Neurology (Cleve) 33:631633, 1983. 24. Stein BM: The Infratentorial Supracerebellar approach to pineal lesions. J Neurosurg 35:197-202, 1971. 25. Stein BM: Supracerebellar-infratentorial approach to pineal tumors. Surg Neurol 11:331-337, 1979. 26. Stein BM: Surgical treatment of pineal tumors. Clin Neurosurg 26:490-510, 1979, ch 19. 27. Stein BM: Supracerebellar approach for pineal region neoplasms. In Schmi dek H, Sweet W (eds): Operative Neurosurgical Techniques. New York, Grune & Stratton, 1982, vol 1, pp 599-607. 28. Sung D, Harisiadis L, Chang CH: Midline pineal tumors and suprasellar germinomas: Highly curable by irradiation. Radiology 128: 745-751, 1978. 29. Van Wagenen WP: A surgical approach for the removal of certain pineal tumors: Report of a case. Surg Gynecol Obstet 53:216-220, 1931. 30. Whittle IR, Allsop JL, Besser M: Tuberculoma mimicking a pinealoma. J Neurosurg 59:875-878, 1983.

24 Occipital Transtentorial Approach W.Kemp Clark, M.D.

Historical Perspective Historically, tumors in the region of the pineal gland were treated by cerebrospinal fluid diversion and radiation therapy (5). The chief reasons for this surgical nihilism were the high operative mortality and morbidity associated with direct surgical approaches to tumors in this area (2, 3, 9, 13, 15). The older approaches to this region usually involved either an occipital lobectomy with its concomitant permanent homonymous hemianopsia or division of some part of the corpus callosum (4, 8, 14). Usually the splenium was divided. This results in a partial cerebral disconnection syndrome with the patient having difficulty transferring visual information from one hemisphere to the other. These older approaches to this region often resulted in division of one or both internal cerebral veins or even the vein of Galen itself (2). The literature is confusing as to the exact results of this action. The collateral venous drainage from the deep galenic system is through septal and choroidal veins. This may be adequate to carry the venous circulation but, on the other hand, it may not. However, mutism, a flat affect, even coma and death have been reported (12). Seizures and hemiplegia have also resulted from these older approaches (13). Other approaches to this region have involved the Supracerebellar infratentorial approach (7, 11). To reach the pineal region, one must coagulate the veins draining the cerebellar hemisphere. This may result in cerebellar venous infarction (7). The advantages of the occipital transtentorial approach over the Supracerebellar infratentorial approach have been described by surgeons using both approaches (10). Advantages and Indications The principle advantages of this approach over the supracellular are a greater ability to mobilize the tumor and to visualize more of the third ventricle and better orientation of the surgeon. An anatomical fact makes this approach feasible. There are no veins that cross from the occipital lobe into the superior sagittal sinus. Therefore, no risk of infarction

occurs by this approach unless the surgeon makes the mistake of ligating the inferior cerebral vein. As this vein drains the occipital lobe into a transverse sinus, its interruption produces edema in the occipital lobe. The vein lies quite lateral and should be outside the usual operative exposure. The occipital transtentorial approach was initially described by Jamie-son (5). He operated without the use of the operating microscope. The addition of this instrument has made the procedure both more effective and safer. The improved illumination and magnification make it possible to gain access to the pineal region with substantial ease (6). Today, the indications for operation of pineal region tumors can best be outlined as follows. In many conditions operation is the only curative procedure available. This is true for aneurysms of the vein of Galen and for teratomas of the pineal gland itself. During operation, one may obtain tissue for pathological diagnosis. This in itself is a worthwhile goal, as it makes possible a better approach to therapy and prognosis. Because chemotherapy is a distinct possibility in cases of germinoma of the pineal gland, a tissue diagnosis of tumor type is a worthwhile reason to operate on these patients. The mere act of operating on the patient regardless of the type of tumor results in reducing the size of the tumor, reducing the burden to the patient. This will improve the chances of radiation controlling the tumor. Finally, operative procedures in this region should result in opening of the third ventricle into the subarachnoid space. This avoids the need for shunt devices. Control of intracranial pressure without a shunt is a worthwhile goal and can only be achieved by direct surgical approach to the tumor. Turning specifically to the advantages of the occipital transtentorial approach, these are many. This approach allows a great deal of exposure without sacrifice of any neural tissue. It is remarkably free of complications or surgical difficulties. Access is gained to the entire third ventricle, the superior cerebellar vermis, the upper fourth ventricle, and the posterior corpus callosum. Excellent visualization of the normal anatomy and the abnormality is obtained. All types of pathology existing in this location can be handled. Positioning of Patient The occipital transtentorial approach traditionally was done with the patient in the sitting position with the attendant risks of air embolus (5). More importantly, in my personal series, all visual field cuts postoperatively occurred in patients operated on in the sitting position. Since changing to the semiprone position, we have eliminated this complication. If the patient is to be operated on in the sitting position, an indwelling catheter in the right heart must be placed for aspiration of air. A Doppler probe should be placed on the precordium. The most likely time of air embolus is when the bone flap is being turned or when the dura mater is being opened. In the sitting position, care should be taken to be sure that the patient is placed far enough forward that the operating microscope can be brought into position without producing hyperextension of the surgeon's neck. An alternative is the semiprone position (Fig. 24.1). This is the position preferred for operations on aneurysms of the vein of Galen. Small children do not tolerate the sitting position very well and it is easier to place them prone with the surgeon at the patient's head. The disadvantages is that the anatomy is reversed and the surgeon must be reminded of that fact. Of the three possible positions the best is undoubtedly the park bench

(semilateral) (Fig. 24.2). It allows easy access to the area and does not produce inversion of the anatomy. It has the additional advantage that the occipital lobe tends to fall away spontaneously and thus retraction of it is avoided. Patient Preparation In preparation for the procedure, certain adjuncts have proven helpful. Corticosteroids, usually dexamethasone in doses of 4 to 6 mg every 6 hours for 48 hours preoperatively, is helpful. A ventriculostomy or a preoperative ventriculoperitoneal shunt for control of intracranial pressure before direct approach to the tumor is also beneficial. Instrumentation The single most important instrument for the performance of this procedure is the operating microscope. Our preference is for the Zeiss operating microscope with the Contraves stand. A binocular observer tube, 35-mm camera, and TV camera are adjuncts to the use of the microscope. It is our practice to bring in the microscope at the time of opening of the tentorium or just after it has been opened. The microscope can be removed from the operative field after the work in the pineal region has been completed and the dural closure has begun. Microsurgical instruments are of the usual types, the preference being for bayonetted forceps, both sharp and blunt. Bipolar cautery is essential. The bipolar cautery forceps must be long enough to reach into the depths of the wound. Usually a 7.5- or 8-in. forceps will serve. Bayonetted scissors, usually curved slightly upward, are used to open the arachnoid and do the dissection around the veins. An arachnoidal knife is sometimes very useful in doing this part of the dissection.

Operative Technique The details of the opening of the operative incision are critical (1, 6). For a righthanded surgeon, a right occipital craniotomy is usually easier. For the left-handed surgeon, it is usually the opposite side. In either case, the crux of the exposure lies in ensuring that the bony opening is placed to be both across the midline to the opposite side and at the level of the transverse sinus. What follows is a description for a righthanded surgeon and is therefore a right occipital craniotomy. The skin incision is made to the left side of the midline, beginning at

the level of the external occipital protuberance (Fig. 24.3). It is carried up approximately 7 to 8 cm and then turned across the midline laterally. This is important to make the craniotomy on the down side of the skull, which ensures the use of gravity to help retract the occipital lobe. The incision is again turned downward to end along the mastoid groove. The scalp is then reflected downward. It is easiest to turn the scalp in one layer including the periosteum. Hemostasis from the scalp is obtained in the usual manner. It is our preference to use Rainey clips for this purpose. However, hemostats or Michelle clips may be used. After reflection of the scalp, the bony opening is made (Fig. 24.4). Burr holes are placed on the right side along the midline just lateral to the superior sagittal sinus. Usually two holes are sufficient as the length of the craniotomy flap is usually about 7 cm. Six burr holes are usually required, four on the right side and two on the left side of the superior

sagittal sinus. Burr holes are then connected with the use of the Gigli saw or a craniotome. The bone is removed by rongeurs across the superior sagittal sinus. The bone flap is then removed as a free flap and kept sterile on the scrub nurse's table. Next, removal of the occipital bone over the transverse sinus is carried out by rongeur. It is not necessary to carry this completely into the posterior fossa, but the transverse sinus must be visualized throughout its extent in the operative field. Bleeding from either of the dural sinuses is easily controlled by using thrombin-soaked gelatin sponge and gentle pressure. The dura mater is opened as an L-shaped or T-shaped incision. In the first cases in our personal series, the dura was opened in a conventional

horseshoe fashion, basing the horseshoe on the superior sagittal sinus. We have abandoned this procedure because the roll of dura along the interhemispheric fissure interfered with access to the lesion. If an L-shaped incision is used, it should begin superiorly as close to the superior longitudinal sinus as possible. The incision is carried inferiorly along the sinus to the level of the torcular. It is then turned laterally, close to the transverse sinus. We leave a cuff of dura of sufficient size to enable a watertight dural closure. Alternatively, we use an incision opening the dura along the sagittal sinus, then along the transverse sinus, and finally obliquely up toward the superior lateral burr hole. This incision has been used in most of our recent cases (Fig. 24.5A). It allows easier access for retraction of the occipital lobe. In any event, the dura is then reflected upward and outward, exposing the occipital lobe. Then, a self-retaining retractor may be gently applied to the occipital lobe, retracting it laterally and superiorly. The crux here is to avoid placing the retractor directly on the calcarine cortex, keeping it more to the occipital pole than deep into the interhemispheric fissure (Fig. 24.5B). If the park bench position is used, it is usually not necessary to retract the occipital lobe as it will fall away spontaneously (Fig. 24.6). Next, the tentorium is opened (Fig. 24.7A). This is done by making an incision parallel to the straight sinus, approximately 1 to 1.5 cm from it. It is easiest to open the tentorium going anteriorly. The surgeon is cutting

away from himself. The beginning incision is made at the junction of the straight sinus and the torcular and is carried anteriorly toward the incisura. The tentorium is vascular and the surgeon should be prepared to control the bleeding by bipolar coagulation or, occasionally, by the use of clips. Retention sutures are placed to hold the tentorium open, enabling visualization of the superior cerebellar vermis and the arachnoid over the veins of the galenic system. Much of the difficulty encountered during earlier operative procedures in this region was due to the tearing of this arachnoid, resulting in damage to the underlying veins. Once, it was thought that the preoperative use of radiation produced the thickening of the arachnoid. This is probably not the case as the arachnoid in this area is always thick. It is necessary to free these veins from their arachnoidal sheath to gain

access to the tumor (Fig. 24.7B). Using microsurgical technique, one opens sharply the dense arachnoid over the deep veins. The first vein visualized is usually the vein of Galen; next is usually the right basilar vein of Rosenthal, which enters the field laterally, coming in to join with the internal cerebral vein. The internal cerebral vein and the precentral vein of the cerebellum are usually the next veins visualized. The arachnoid is dissected from the vein of Galen, the internal cerebral vein, and the basal vein of Rosenthal. In most circumstances, it is necessary to

dissect the arachnoid off the right internal cerebral vein and the right basal vein of Rosenthal only (Fig. 24.7B). However, it is possible to dissect the left with facility should it be necessary to mobilize the veins to gain access to the lesion. Once the arachnoid has been dissected free from the veins, the surgeon can deal with any lesion of the superior vermis, quadrigeminal plate, posterior third ventricle, splenium, or pineal gland (Fig. 24.8). Lesion Management In a typical pineal region tumor, the tumor mass lies deep to these venous structures (Fig. 24.9). A more difficult situation arises when the tumor rises from the free edge of the tentorium or the junction of the falx and the tentorium (Fig. 24.10). In this circumstance, the veins will be carried anteriorly, producing the problem of identifying them through the tumor. The surgeon must reduce the volume of the tumor using the ultrasonic aspirator. Working cautiously along the lateral inferior margin of the tumor, the surgeon can begin to identify the arachnoid and the veins. It is usually easier to deal with meningiomas arising in this location by approaching their removal from lateral to medial and from inferior to superior. However, no hard and fast rule can be made as each case must be approached individually, and the most promising plane of dissection should be pursued by the surgeon. Intrinsic tumors, such as gliomas of the quadrigeminal plate or gliomas of the splenium of the corpus callosum, will again take the deep venous structures posteriorly toward the surgeon, and they may be displaced upward or downward depending on the site of origin of the tumor. It is usually simple to identify an intrinsic tumor of either the brain stem or the splenium by the lack of encapsulation of the mass. However, if a primary pineal tumor is very large this may be rather difficult. Occasionally, the identification of the capsule may prove difficult. The glioma in this region should be cautiously excised and the third ventricle should be entered through its posterior aspect in the pineal recess. This should be a large opening so that cerebrospinal fluid (CSF) flow will be facilitated. If the lesion dealt with is a vein of Galen aneurysm, the major feeders to the malformation should have been identified by preoperative angiog-raphy. Their location can then be confirmed by direct inspection by rotating the mass of the aneurysm until they can be seen. These can be dividied between aneurysm clips. Thus, a thorough centimeter by centimeter inspection of the wall of the vein of Galen is carried out. Arterial feeders entering into it are either coagulated or clipped and divided. As this dissection proceeds, the operator will note that the vein of Galen becomes progressively bluer and less turgid, making it easier to complete the dissection. When the surgeon feels that all arterial connections have been divided, a 25 gauge needle is introduced into the vein of Galen and a blood sample is drawn. Simultaneously blood is drawn from a peripheral vein. Determination of blood gases on the two samples is done and the values should be the same. This indicates that the blood from the vein of Galen is venous and that all arterial anastomoses have been interrupted. If one is dealing with the typical pineal region tumor, the tumor should be entered by sharp dissection and a biopsy obtained (Fig. 24.11). A frozen section will help with the decision of how to proceed. If, as occasionally happens, the tumor proves to be a glioma arising either from the quadrigeminal p?ate or from the corpus callosum, cytoreduction of the tumor should be done cautiously and the CSF pathway should be opened.

If the tumor proves to be of embryonal origin such as a teratoma or dermoid, piecemeal or en bloc resection should be done. The surgeon should make every attempt to remove the tumor completely. If the tumor is a germinoma of the pineal region, thorough cytoreduction of the tumor with removal of as much as possible should be carried out (Fig. 24.12). In any case, the end stage is when one can visualize the interior of the third ventricle, seeing all the way to the foramen of Monro. This ensures that there is no block of the CSF pathway.

Pineal region tumors may extend inferiorly down under the vermis of the cerebellum (Fig. 24.13A). If this is the case, that structure should be divided so that the lower pole of the tumor may be identified, mobilized, and removed. Similarly, the tumor may extend superiorly into the splenium of the corpus callosum (Fig. 24.13B). Usually it is possible by reducing the tumor volume internally and delivering the capsule of the tumor downward into the operative field to avoid having to divide the splenium. The surgeon should be aware that opening the superior vermis of the cerebellum produces little in the way of neurological deficit but too vigorous a division of the splenium will produce a cerebral disconnection syndrome. Through this approach, it is possible to operate on tumors of the

superior cerebellar vermis. Upon opening the tentorium, the surgeon will see the culmen of the cerebellum (Fig. 24.1ЗА). Either the tumor presents on the surface or it may be found subcortically by ultrasound or by needle. It is then approached through a transcortical incision. The great advantage of approaching tumors in the cerebellar vermis by this approach is the avoidance of manipulation of the cerebellum. The tumor may be removed in toto or in part depending on its histological characteristics without producing neurological deficit by retraction or dissection within the cerebellum. Likewise, tumors lying in the superior part of the fourth ventricle may be approached through the superior cerebellar vermis by this approach, again with the advantage of approaching the tumor through the shortest possible route. Intraoperative Decisions Clearly, the advantage of this procedure is the possibility of total excision of the lesion. This is the nub of the whole operative procedure. There are a number of points that enter into the decision to excise totally or simply to perform an internal decompression. Similarly, there is a problem relating to when to stop the internal decompression of a tumor. The factors that enter into the decision to attempt total resection include the nature of the lesion, the location of the major deep venous structures,

the ease of the procedure up to the decision-making point, the age of the patient, and the availability of adjunctive therapy. Obviously the type of abnormality with which the surgeon is dealing is a critical determinant of the possibility of total excision. It is rarely possible to excise a germ cell tumor totally unless it is a teratoma. An occasional germinoma can be excised. For other tumors that occur in this region, such as meningiomas, arachnoidal cysts, lipomas, and gliomas, total resection becomes increasingly difficult. One is usually unable to remove a meningioma if it is invading the dural sinuses. Total resection is also limited by the location of the great veins, which typically are displaced anteriorly to meningioma when one is working through tumor toward these vascular structures. Because most meningiomas in this area are truly benign in terms of their growth characteristics, it is usually not necessary to risk the patient's life and function in attempting a total removal. Tumors that are intrinsic to the brain stem such as lipomas and gliomas cannot be surgically resected except at the price of a ghastly neurological deficit. Here internal decompression is the trick. How well the procedure has gone to the point of decisionmaking for a total excision is also critical. If it has been a struggle with a number of problems, such as blood loss or anesthetic difficulty, it is usually the better part of valor to abandon the procedure. One can return another day for a second attempt. If things have gone well and there are no problems with the patient in general, it is safe to consider total resection if all other criteria are met. The age of the patient is also a consideration. The surgeon should be more aggressive in the younger age group. The availability of adjunctive therapy must be considered because some of the germ cell tumors have been successfully managed with chemotherapy and radiation. The critical dimension for total excision is the ability of the surgeon to define a clean plane between the tumor and the surrounding normal structures. As long as this plane can be easily identified, it is possible to proceed toward total resection. This is a matter of experience and judgment. The question of when the tumor has been adequately reduced is again a matter of surgical judgment. One must open the posterior third ventricle adequately to allow for the normal flow of CSF. It is critical that the ventricular system be unblocked before resection is abandoned; otherwise, the patient will require some type of shunt. One of the purposes of the operation is to avoid the need for a shunt, with its concomitant problems of intermittent obstruction or infection (Fig. 24.12). Once the ventricular system has been decompressed, one must consider the amount of tumor burden that remains to be treated by radiation, chemotherapy, or both. The best estimate of this is how much normal tissue can be identified around the edges of the mass. The direction in which the remaining tumor is going is also critical. For instance, in a quadrigeminal plate glioma the remaining tumor is in the brain stem. It becomes critical to stop excising the tumor before entering the brain stem proper. A very difficult problem lies in the surgical treatment of lipomas in this region and here cytoreduction should stop as soon as the ventricular system is decompressed. The margins between lipomatous tumor and brain are extremely difficult to define, and the surgeon may injure critical areas of the brain. Closure After removal of the tumor and local hemostatis, the retractor, if any, should be removed from the wound, the wound thoroughly irrigated, and

the dura replaced and sutured (Fig. 24.14A). Although, it is not necessary to close the tentorium watertight, a few sutures to prevent the occipital lobe from herniating downward are worthwhile. Closure of the dura mater can be done either with interrupted or running suture. The dura mater should be closed as watertight as possible (Fig. 24.14B). The bone flap is replaced and wired in position, and the small, narrow craniectomy along the transverse sinus is repaired using stainless steel mesh as an onlay cranioplasty (Fig. 24.14C). The scalp flap is then replaced and closed either in one layer with mattress sutures or in the two-layer closure of the galea and the skin.

Drainage using a suction-type drain such as the Jackson Pratt may be used. It should be placed in the epidural space as an optional measure. Summary and Complications In summary, the occipital transtentorial approach to the pineal region, posterior third ventricle, superior cerebellar vermis, quadrigeminal plate, and deep galenic venous system has proven to be one of the most successful surgical procedures introduced during the past 2 decades. The older approaches had unacceptable morbidity and mortality. This has proven not to be the case with this approach. The complications of this approach are the usual ones of any surgical intervention in the head. Infection, osteomyelitis, and intracranial hemorrhage of various types are possible. In my experience none have occurred. This may be a function of the relatively small size of my series of these tumors. Specific complications using this approach include the production of a homonymous hemianopsia. This is the direct result of improper placement of the retractor with two vigorous a retraction of the occipital lobe. This may be avoided by the use of hypertonic solutions during operation and the preexistent use of ventricular drainage to reduce ventricular size. Finally, the use of the park bench position produces falling away of the occipital lobe without retraction. Other complications involve injury to the splenium of the corpus callosum. If this is inadvertently or deliberately divided, the patient will be left with a cerebral disconnection syndrome. This will produce difficulty in reading, as visual information cannot be transferred from one hemisphere to the other. Damage to the venous structures are the final specific problems related to this approach. The easiest one to avoid is damage to the inferior cerebral vein, which drains the occipital lobe. This should lie too far laterally to be of any concern to the operating surgeon. It rarely is in the operative field but can be visualized occasionally lying just lateral to the bony opening. Damage to the internal cerebral veins may result in mutism. Damage to the vein of Galen may produce a similar effect. The secrete in avoiding damage to the deep venous system is to open the arachnoid sharply under direct vision with magnification. In the case of intrinsic tumors approached through this exposure, it is possible to enter into the quadrigeminal plate of the upper brain stem with disastrous results. However, the surgeon should recognize that the goal in such a case is to open the third ventricle, not to effect a great deal of tumor resection. This approach to this anatomically deep region in the center of the brain has proven to be effective, safe, and relatively simple to perform technically. It is a significant advance in our surgical capacity to deal with tumors of the human nervous system. References 1. Clark K: The occipital transtentorial approach to the pineal region, in Schmidek HH, Sweet WH (eds): Operative Neurosurgery and Techniques. New York, Grune and Stratton, 1983, pp 595-598. 2. Dandy WE: An operation for the removal of pineal tumors. Surg Gunecol Obstet 33:113-119, 1921. 3. Glasauer FE: An operative approach to pineal tumors. Acta Neurochir (Wien) 22:177-180, 1970. 4. Horrax G: Treatment of tumors of the pineal body. Arch Neurol Psychiatry 64:227-242, 1950. 5. Jamieson KG: Excision of pineal tumors. J Neurosurg 35:550-553, 1971. 6. Lazar ML, Clark WK: Direct surgical management of masses in the region of the vein of Galen. Surg Neurol 2:17-21, 1974.

7. Page LK: The infratentorial Supracerebellar exposure of tumors in the pineal area. Neurosurgery 1:36-40, 1977. 8. Poppen JL: The right occipital approach to a pinealoma. JNeurosurg 25:706710, 1966. 9. Poppen JL, Marino R Jr: Pinealomas and tumors of the posterior portion of the third ventricle. J Neurosurg 28:357-364, 1968. 10. Reid WS, Clark WK: Comparison of the infratentorial and transtentorial aproaches to the pineal region. Neurosurgery 3:1-8, 1978. 11. Stein BM. The infratentorial Supracerebellar approach to pineal lesions. J Neurosurg 35:197-202, 1971. 12. Stern WE, Batzdorf U, Rich JR: Challenges of surgical excision of tumors in the pineal region. Bull LA Neurol Soc 36:106-118, 1971. 13. Suzuki J, Iwabuchi T: Surgical removal of pineal tumors (pinealomas and teratomas). J Neurosurg 23:565-571, 1965. 14. Van Wagenen WP: A surgical approach for removal of certain pineal tumors: Report of a case. Surg Gynecol Obstet 53:216-220, 1931. L5. Ward A, Spurling RG: The conservative treatment of third venticle tumors. J Neurosurg 5:124-130, 1948.

25 Posterior Intrahemispheric Retrocallosal and Transcallosal Approaches J. Gordon McComb, M.D., and Michael L J. Apuzzo, M.D.

Historical Perspective The posterior interhemispheric transcallosal approach to the pineal region was originally described by Walter Dandy in 1915 using a canine model (5, 22, 35). Based on his experience with 12 animals he suggested the feasibility of such an approach in humans, and in 1921 he published his initial experience using this method with 3 cases of pineal region masses (7). With the head in a lateral position with the right side up, an extensive right midline parietooccipital bone flap was developed, the dura was reflected over the sagittal sinus, and bridging veins were sacrificed. After midline exposure, the posterior 3 to 4 cm of the corpus callosum was incised, providing visualization of the vein of Galen, its tributaries, and the posterior third ventricle and quadrigeminal regions. In the first instance a silent cerebellar tumor had secondarily involved the region of the pineal body and the corpora quadrigemina; after exposure of the tumor no attempt was made to remove it, because of its infiltrating character. This case, however, showed that a good exposure of this region is possible. On two subsequent occasions tumors of the pineal body have been completely removed. In one case an encapsulated tubercle of the pineal body... it measured 5 cm by 4 cm. The results of this case demonstrated not only the feasibility of the removal of tumors of the pineal body, but also the absence (in this case at least) of any injurious mental or physical effects due to the operation. In this chapter we describe modifications of this essential concept (7, 28) that, when coupled with contemporary microsurgical instrumentation, allows safe exposure of the pineal gland, posterior third ventricle, quadrigeminal cistern, and quadrigeminal complex (Fig. 25.1) via a posterior interhemispheric corridor, which allows regional entry and exposure via maneuvers of (a) tentorial incision, (b) falcine incision, (c) splenial retraction, (d) incision of the posterior body of the corpus callosum, and (e) splenial incision.

Operative Corridor: Anatomy and Physiological Risks The major topographic elements of the posterior midline corridor that require identification and consideration include the inion, the lamdoidal and sagittal sutures, the sagittal sinus, the parasagittal cortical veins, the falx cerebri, the tentorium, the incisura, the straight sinus, the vein of Galen and its tributaries, and the posterior trunk and splenium of the corpus callosum (20, 21, 23, 31). Consideration of the consequences of injury to each neural or vascular component is essential as a stepwise progression evolves through the corridor of exposure. Dural Venous Sinuses During bone flap development and dural reflection, attention must be directed toward maintaining the integrity of the sagittal sinus, which is exposed over its posterior third, the lateral sinus, and the torcular Hero-phili (14, 17). Particularly, if the sitting position is used aspiration of air must be avoided. With the retrocallosal approach, it may be necessary to gain more exposure by incising the falx cerebri, which necessitates dividing the inferior sagittal sinus or the tentorium parallel to the straight sinus (Fig. 25.2). Parasagittal Veins Although primary consideration must be given (9) to parasagittal venous tributaries during anterior callosal approaches (1, 34), this is not an important issue in posterior callosal approaches that involve the posterior third of the nondominant cerebral hemisphere. The posterior one-third of the sagittal sinus receives few significant tributaries, and the posterior, medial superior nondominant hemisphere is generally without a detectable functional component (Fig. 25.3). Deep Cerebral Veins (Galenic System) The venous structures comprising the galenic system provide an apparent potential threat to satisfactory outcome in surgical endeavors in this region (16, 21, 30, 32). Most reports addressing the surgical approaches to the pineal region caution that these vessels should not be compromised as either hydrocephalus or venous infarction could result. The concept that hydrocephalus resulted from occlusion of the vein of Galen originated with the work of Dandy in 1919 (6). However, subsequent studies by Bedford (1934) (2), Schlesinger (1940) (28), and Hammock et al. (1971) (13) failed to confirm this observation. The issue of venous infarction remains clouded. Clinical experience with thrombosis of the vein of Galen has not added relevant information as the circulatory and pathological changes that follow occlusion thrombosis are not selective to the vein of Galen, but involve many of the veins comprising the deep venous system and are often associated with other pathological processes such as inflammation or mass lesions. Schlesinger (29) reported in 1939 that no valves are present in the deep or the superficial venous systems of the brain and that there is free bidirectional communication between the two drainage pathways. Some of the vascular changes observed after ligation of the vein of Galen in earlier studies potentially relate to operative techniques rather than actual ligation of the structure. Any objections to technique seem to have been avoided by Hammock et al. (13), who selectively occluded the vein of Galen with clips, effecting minimal trauma to adjacent tissues. Pre- and postoperative angiograms confirmed pre- and postocclusion vascular anatomy in the rhesus monkey. The only circulatory change noted on the postclipping

angiogram was dilation of the adjacent venous pathways; there was no evidence of thrombosis. Examination of the brains at various intervals after vein of Galen occlusion showed no evidence of vascular infarction or encephalomalacia, but only dilation of the diencephalic and choroidal vessels. Clinically these animals showed no untoward effects. No study has detailed selective occlusion of the internal cerebral veins, basal veins of Rosenthal, or precentral cerebellar veins, which in combination or total might deprive access to collateral drainage pathways. Based upon available information, it seems that isolated partial occlusion of the deep venous system should have no untoward consequences. However, more detailed laboratory and clinical information is required before an absolute statement may be made in this regard. By necessity, the authors have occluded one or two of these tributaries without evidence of clinical consequences. Dandy noted a similar clinical observation in 1936 (8). Caron et al. (4) described two cases in which both internal cerebral veins were sacrificed without harmful effects.

Spectrum of Approach The interhemispheric retrocallosal or transcallosal approaches allow a number of corridors and exposures depending on individual patient anatomy and size and location of the lesion (Fig. 25.4). Possible exposures of the quadrigeminal and posterior and mid-ventricular regions are gained by a number of intraoperative maneuvers in subsequent corridors: (a)

exposure of the posterior incisural margin, (b) incision of the tentorium with exposure of the quadrigeminal cistern and posterior third ventricle, (c) incision of the falx (inferior sagittal sinus) with enhanced contralateral visualization, (d) retraction of the splenium of the corpus callosum with posterior third ventricular and superior quadrigeminal exposure, (e) section of the posterior trunk of the corpus callosum with increased exposure of the posterior and middle third ventricle, and (f) section of the splenium (if unavoidable) to increase third ventricular exposure. The individual anatomy and lesion (Fig. 25.5) will dictate the particular need for exposure (25); however, a number of possibilities are evident dependent on the location of the mass (Fig. 25.6): (a) Masses in the superior quadrigeminal cistern encroaching on the posterior third ventricle may be approached with a retrocallosal and transincisural corridor with or without sectioning of the tentorium and/or falx by either an 8-cm bone flap extending from the torcular or a 6-cm bone flap 2 cm superior to the torcular, (b) Lesions predominantly in the quadrigeminal cistern with minimal third ventricular involvement may be managed by a 6-cm midline bone flap 2 cm superior to the inion. Tentorial incision is generally required. (c) Lesions involving the posterior and mid-third ventricular chamber with or without superior quadrigeminal involvement generally require exposure that allows maximal flexibility of the corridor. In this case a 10-cm midline bone flap extends from 2 cm superior to the torcular anteriorly. This allows access to the posterior trunk of the corpus callosum as well as the posterior incisural margin.

Structural Definitions High resolution computerized tomography (CT) with and without contrast administration and additional reconstructions in the coronal and sagittal planes usually provide excellent detail of the region of the posterior third ventricle (Fig. 25.7). Magnetic resonance imaging (MRI) with multiplanar views augments CT data and provides further definition of vascular anatomy (Figs. 25.8 and 25.9). At present, only the water content with various relaxation times can be measured on MRI, but it is anticipated that in future various doping compounds will become available to delineate further the nature of the lesion. Small lesions in the posterior third ventricle can even be further delineated by the addition of metri-

zamide (Fig. 25.10) to the ventricular cerebrospinal fluid (CSF) by way of a ventriculostomy, which have often been placed previously to control raised intracranial pressure (ICP). Unless there is question as to the vascularity of the lesion, there is little need for angiography when the retrocallosal approach to the posterior third ventricle is planned. For the more anterior transcallosal approach, it is suggested that angiography be done if for nothing more than to delineate the location of the parasagittal draining veins (1). This will allow better placement of the bone flap to minimize the number of these veins that must be occluded and divided to gain access to the midline. Operative Techniques General Measures The majority of patients with tumors in the region of the posterior third ventricle have hydrocephalus. For control of the associated raised ICP,

many patients have received a ventriculostomy before coming to the operating room; if not, one can be placed after the induction of general anesthesia. Antibiotics are given to cover the perioperative period (at the beginning of the operation and for 24 to 48 hours thereafter). As three-fourths of ventriculostomy infections are from Staphylococcus species (approximately one-half of the total infections are Staphylococcus epi-dermidis and one-fourth are Staphylococcus aureus), the antibiotics need to cover this group of organisms. We are presently using vancomycin as 30 to 40% of the S. epidermidis cultured at our institution are resistent to methicillin. Glucocorticoids have generally been started 1 or more days before operation. If a ventriculostomy is placed before operation, the ventricles are deliberately kept moderately dilated by only draining CSF if the ICP exceeds 15 to 20 torr. This allows decompression of the ventricles by further drainage of CSF intraoperatively when the medial aspect of the right hemisphere is gently retracted from the falx cerebri, allowing the corpus callosum or retrocallosal region to be adequately exposed. Rather than inserting a CSF-diverting shunt preoperatively, the authors prefer an attempt to reestablish CSF circulation by tumor removal, thereby obviating the need for shunting. During the postoperative period, the presence of a ventriculostomy will allow monitoring of ICP as well as determining whether CSF circulation has been adequately reestablished. If the exposure is not adequate with CSF drainage, osmotic and nonosmotic dehydrating agents may be used. Intraoperative evoked response monitoring is a helpful and recommended adjunct (12).

Position Although a number of options in positioning are possible to obtain access to the posterior hemispheric midline (15, 18, 19, 24, 26), the authors prefer placing the patient in a right lateral decubitis (Figs. 25.11 to 25.13) or three-quarters prone position, with the head fixed in a dependent oblique posture (Figs. 25.14 and 25.15). This head position is developed by placing the superior sagittal sinus parallel to the floor and then elevating the vertex 30° to 40° from the horizontal. This allows the dependent hemisphere (usually the right) to fall away from the midline while the falx cerebri supports the superior (left) hemisphere. Removal of CSF from the dilated lateral ventricles, as most frequently is done, provides excellent exposure with minimal retraction of the hemisphere. Compared to the sitting position (Figs. 25.16 and 25.17), the surgeon enjoys a closer working distance and experiences less fatigue as his hands and arms are at waist or chest level and not extended. The use of the hands are parallel rather than one over the other as necessitated by sitting position. Another advantage is that the position of the microscope to the patient can be varied over a greater distance, gaining further visibility even though the corridor of access is narrow. Raising and lowering the patient will add another degree of flexibility for the operative exposure and reduces the manipulation necessary to reach the lesion. Of most importance is the fact that the risk of air embolism is considerably reduced. Because of the clinical and technical superiority of the dependent oblique head position we reserve the semi-sitting position for excessively obese individuals in whom utilization of the dependent oblique position would be unduly difficult or potentially threatening to pulmonary status during a lengthy procedure (Figs. 25.18 and 25.19).

Retrocallosal Approach Before draping, so as to observe readily the pertinent anatomical landmarks, one marks the midline, the torcular, and the transverse sinus on the scalp. Either a linear, 2-limbed mitre or an S-shaped incision may be used. Exposure should include the sagittal and lambdoidal sutures to act as bone flap landmarks. The scalp is then injected with 0.25% lidocaine, 1:400,000 epinephrine solution to reduce bleeding. Either three or four burr holes are placed to develop the paramedian bone flap. The two medial burr holes are placed directly on the midline 6 to 8 cm apart with the posterior burr hole at or just above the torcular. If two lateral burr holes are placed, the most posterior one is at or just above the transverse sinus. The lateral one or two burr holes need be no more than 4 to 6 cm off the midline. Absolute midline exposure is imperative. In individuals with a thick calvarium it is desirable to extend the bone flap over the midline. The dura mater is separated from the overlying calvarium and a free bone flap is cut. The dura mater is opened in a trapezoidal shape with the lateral extent of the opening at the point where the hemisphere needs maximal retraction. The intact portion of the dura mater supports the hemisphere. The infrequent bridging veins are coagulated and divided as necessary, allowing the posterior parietal and occipital lobes to be easily retracted. Occluding and dividing a bridging vein within the operative field should not significantly compromise venous drainage in this portion of the hemisphere. Stay sutures are placed in the dura mater just lateral to the superior sagittal sinus and retraction is obtained by securing the ties to the Budde retractor ring. If a ventriculostomy is in place, it is

opened and the pressure is reduced to atmospheric while gentle pressure is applied to the medial surface of the parietooccipital lobes to aid in expelling CSF from the ventricles. The Budde self-retaining retractor system and the operating microscope are brought into position. A 19-mm retractor blade is used and exposure of the incisural margin is rapid as the falx is complete. Compression of the medial hemisphere is minimal and a 2-cm slot is easily realized (Fig. 25.20). With the retrocallosal approach, it is not necessary to separate the cingulate gyri or the pericallosal arteries. If the lesion is posteriorly directed, it can be seen distending the dense arachnoid in the notch between the falx cerebri and the

tentorium. If necessary, either the falx cerebri, the tentorium, or both may be divided to gain additional exposure. The dura mater of the falx cerebri (Fig. 25.21) is cauterized with a combination of straight or angu-lated bipolar forceps and is sharply incised, usually with a #11 blade, after the inferior sagittal sinus has been occluded. Stay sutures can be placed in the cut edges of the falx for further exposure or to apply traction to stop bleeding from the inferior sagittal sinus if it continues. The tentorium (Fig. 25.22) is divided in a similar fashion but lateral to the straight sinus. Stay sutures can also be placed in the cut edges of the tentorium to gain excellent exposure to the posterior fossa. A second 5/8-or 1/4-in. blade may be applied medially on the falx and a third V4-in. retractor may be placed on the splenium to enhance exposure.

It is generally not necessary to expose the splenium of the corpus callosum although one needs to be cognizant of its location (Fig. 25.23). Occasionally retraction or partial excision of a distended and thinned commissure may be necessary. The arachnoid is divided as needed. The internal cerebral veins, basilar veins of Rosenthal, and precentral cerebellar veins may or may not be encountered and are frequently displaced by the tumor mass. Usually one can work around them but sometimes it is necesary to coagulate and divide one or two of these vessels. Tumor removal can be accomplished with a variety of techniques including aspiration, bipolar coagulation, ultrasonic aspiration, and laser vaporization, depending upon the consistency of the mass (Fig. 25.24).

Depending upon the location of the lesion and the degree of resection, the third ventricle may or may not have been entered. Feeding arterial vessels are usually inferior and lateral to the mass. If possible, these should be identified early in the dissection. One frequently has some idea as to whether CSF circulation has been reestablished. The ventriculostomy is closed and opened when the patient has returned to the intensive care unit. No attempt is made to reinflate the ventricles by injecting fluid via the ventriculostomy. The self-retaining retractor is removed. The dura mater is closed using a running 4-0 Vicryl suture. Dural tenting sutures are placed if desired. Gelfoam is placed over the exposed dura mater. The bone flap is replaced and fixed with #28 gauge wire. The scalp is closed as a single layer using either interrupted 3-0 or 4-0 Vicryl sutures in the younger patient or staple sutures in the adult. The skin edges are approximated with Steri-strips if scalp bleeding is not a problem or a running subcuticular suture if it is.

Transcallosal Approach If the lesion is totally confined to the posterior portion of the third ventricle without extension beneath and posterior to the corpus callosum or there is significant middle third ventricular extension it may be necessary to use a transcallosal approach to augment exposure of the lesion. If possible, it is best to maintain the integrity of the splenium of the corpus callosum. For this reason, the exposure needs to be such that a 2to 2.5-cm incision can be made in the body of the corpus callosum 2 to 3 cm anterior to the tip of the splenium, thus leaving it intact (Fig. 25.25). As the approach is more anterior, the calvarial opening needs to be moved accordingly. This necessitates that the patient undergo angiography to determine the location of the major draining cerebral veins so one can

position the bony opening to avoid dividing many of the major draining veins in the posterior parietal region. The depth of the falx cerebri varies considerably, as does the extent of the adhesions between the two hemispheres that must be divided before the corpus callosum is reached. It is necessary to identify both pericallosal arteries to make certain that the dissection to the corpus callosum is between them and not on their medial aspect. The corpus callosum is easily identified as a glistening white structure that is totally different from the cortex. After estimating the distance from the tip of the splenium of the corpus callosum, one makes the posterior extent of the incision at least 2 to 3 cm anteriorly with a microsucter. A narrow (19-mm)-bladed self-retaining retractor can be placed through the incision in the body of the corpus callosum. The dilatation of the third ventricle either by tumor mass or hydrocephalus thins the corpus callosum and posterior fornix, simplifying the dissection. The connective tissue of the tela choroidea is next encountered. Lesions in this region tend to displace the internal cerebral veins laterally, although on some occasions they can be found separated from the midline by only a millimeter or two of connective tissue. Usually it is possible to coagulate and divide the intervening connective tissue and displace the cerebral veins laterally but, if necessary, one may be occluded and divided. With large masses or significant hydrocephalus the dissection required to enter the third ventricle can be surprisingly little; however, with small lesions and little hydrocephalus, the depth of the operative corridor before reaching the third ventricle is considerable. Sagittal MRI or CT will guide depth measurements. Entry into the third ventricle is readily appreciated by noting the smooth ependymal surface and the presence of CSF. Tumor removal can be accomplished with a variety of techniques. After tumor excision the aqueduct of Sylvius can be observed. Closure is the same as for the retrocallosal approach. Utilization The posterior interhemispheric retrocallosal and transcallosal approaches offer rapid, direct, and safe exposure of the quadrigeminal and pineal regions and posterior twothirds of the third ventricle with minimal physiological consequence. By concurrent utilization of the dependent oblique head position, the procedural exposure and ease of intraoperative manipulation may be facilitated. The corridor may be used for access to any mass or vascular lesion within these regions, irrespective to its size. Considering the histological spectrum of tumors, diagnostic capability of imaging devices, availability of biological markers, current imaging-guided stereotactic methods, and therapeutic modalities, it seems that direct operation should be reserved for (a) benign potentially excisable lesions and (b) malignant and mixed germinal tumors where multimodal-ity therapy offers the best opportunity for palliation. Such lesions may be considered largely radioresistant. Advantages Using the dependent oblique head position, gravity and CSF drainage are frequently all that is required to obtain excellent exposure of the pineal region of the corpus callosum. The lateral decubitus or 3/4 prone body position are well tolerated by the patient and reduce the risk of air emboli. The retrocallosal approach is the preferred corridor as it is not necessary to sacrifice any important intracranial structures. By opening the falx cerebri, one has access to the opposite side and, by opening the tentorium, the superior portions of the quadrigeminal region are exposed.

The dependent oblique position provides the surgeon with a comfortable working position as the distance allows the hands to be placed by the side of the waist or at chest level and the arms are not extended. In addition, the operating microscope can be used over a wide arc, which is further increased by raising or lowering the operating room table. The assistant surgeon and scrub nurse are conveniently located, which aids a smooth flow of the operative procedure. The technique offers visualization of all elements of the midline and posterior third ventricle and the pineal and quadrigeminal regions with minimal physiological risks and minimizes the utilization of active retraction. The corridor provides the shortest and most direct access to the pineal region (26, 27, 31, 33). Disadvantages This corridor is advocated only for those lesions that are primarily in the pineal or posterior third ventricular locations. It is not an appropriate approach for lesions of the anterior third ventricle or those of the posterior fossa. In general, the galenic system and its tributaries are encountered before the pathological structure. Optimal positioning is difficult in large patients. References 1. Apuzzo MLJ, Chikovani OK, Gott PS, Teng BL, Zee CS, Giannotta SL, Weiss MH: Transcallosal, interfornicial approaches for lesions affecting the third ventricle: Surgical considerations and consequences. Neurosurgery 10:547554, 1982. 2. Bedford THB: The nervous system of the velum interpositum of the rhesus monkey and the effect of experimental occlusion of the great vein of Galen. Brain 57:255-265, 1934. 3. Bogen JE: Physiological consequences of complete or partial commissural section. In Apuzzo MLJ (ed): Surgery of the Third Ventricle. Baltimore, Williams & Wilkins, 1987. 4. Caron JP, Nick J, Contamin F, Singer B, Comoy J, Keravel Y: Tolerance de la ligature et de la thrombose aseptique des veines cerebraler profondes chez l'homme. Ann Med Interne (Paris) 12:899-906, 1977. 5. Dandy WE: Extirpation of the pineal body. J Exp Med 22:237-247, 1915. 6. Dandy WE: Experimental hydrocephalus. Ann Surg 70:129-142,1919. 7. Dandy WE: An operation for the removal of pineal tumors. Surg Gynecol Obstet 33:113-119, 1921. 8. Dandy WE: Operative experience in cases of pineal tumor. Arch Surg 33:1946, 1936. 9. Dandy WE: Benign Tumors in the Third Ventricle of the Brain: Diagnosis and Treatment. Springfield, Thomas, 1933, p 171. 10. Foerster O: Das operative Vorgehen bei Tumor der Vierhugelgegend. Men Klin Wochenschr 41:986-990, 1928. 11. Geschwind N: Disconnexion syndromes in animals and man. Brain 88:237294,585-644, 1965. 12. Greenberg RP, Ducker DP: Evoked potentials in the clinical neurosciences. J Neurosurg 56:1-18, 1982. 13. Hammock MK, Milhorat TH, Earle K, DiChiro G: Vein of Galen ligation in the primate: Angiographic, gross, and light microscopic evaluation. J Neurosurg 34:77-83, 1971. 14. Huang YP, Okudera T, Ohta T, Robbins A: Anatomic variations of the dural venous sinuses. In Kapp JP, Schmidek HH (eds): The Cerebral Venous System and its Disorders. New York, Grune and Stratton, 1984, pp 109167. 15. Jamieson KG: Excision of pineal tumors. J Neurosurg 35:550-553, 1971. 16. Kaplan HA: Results of obliteration of specific cerebral veins and dural venous sinuses: Animal and human studies. In Kapp JP, Schmidek HH (eds): The Cerebral Venous System and its Disorders. New York, Grune and Stratton, 1984, pp 275-282. 17. Kapp JP, Schmidek HH: Surgery of the cerebral venous system. In Kapp JP, Schmidek HH (eds): The Cerebral Venous System and Its Disorders. New York, Grune and Stratton, 1984, pp 597-623. 18. Kobayashi S, Sugita K, Tanaka Y, Kyoshima K: Infratentorial approach to

19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30.

31. 32. 33. 34. 35.

the pineal region in the prone position: Concorde position. J Neurosurg 58:141-143, 1983. McComb JG, Barky K: Lateral decubitus position for posterior fossa surgery in children. Concepts Pediatr Neurosurg 5:207-213, 1985. Ono M, Rhoton AL Jr, Barry M: Microsurgical anatomy of the region of the tentorial incisura. J Neurosurg 60:365-399, 1984. Ono M, Rhoton AL Jr, Peace D, Rodriguez RJ: Microsurgical anatomy of the deep venous system of the brain. Neurosurgery 15:621-657, 1984. Pendyl G: The surgery of pineal lesions—historical perspective. In Neuwelt EA (ed): Diagnosis and Treatment of Pineal Region Tumors. Baltimore, Williams & Wilkins, 1984, pp 139-154. Pendyl G: Microsurgical anatomy of the pineal region. In Neuwelt EA (ed): Diagnosis and Treatment of Pineal Region Tumors. Baltimore, Williams & Wilkins, 1984, pp 155-207. Poppen JL: The right occipital approach to a pinealoma. J Neurosurg 25:706710, 1966. Raimondi AJ, Tomita T: Pineal tumors in childhood. Child Brain 9:239-266, 1982. Reid WS, Clark WK: Comparison of the infratentorial and transtentorial approaches to the pineal region. Neurosurgery 3:1-8,1978. Rhoton A Jr, Yamamoto I, Peace DA: Microsurgery of the third ventricle: Part 2. Operative approaches. Neurosurgery 8:357-373, 1981. Schlesinger B: The tolerance of the blocked galenic system against artificially increased intravenous pressure. Brain 63:178-183, 1940. Schlesinger B: The venous drainage of the brain, with special reference to the galenic system. Brain 62:274-291, 1939. Schmidek HH, Kapp JP: Traumatic and neoplastic involvement of the cere bral venous system. In Kapp JP, Schmidek HH (eds): The Cerebral Venous System and Its Disorders. New York, Grune and Stratton, 1984, pp 581595. Seeger W (ed): Microsurgery of the Brain. New York, Springer-Verlag, 1980, vol 1 and 2. Smith RR, Sanford RA: Disorders and the deep cerebral veins. In Kapp JP, Schmidek HH (eds): The Cerebral Venous System and Its Disorders. New York, Grune and Stratton, 1984, pp 547-555. Stein BM: The infratentorial Supracerebellar approach to pineal lesions. J Neurosurg 35:197-202, 1971. Sugita K, Kobayashi S, Yokoo A: Preservation of large bridging veins during brain retraction. J Neurosurg 57:856-858, 1982. Wilkins RH: History of surgery of the third ventricular region. In Apuzzo MLJ (ed): Surgery of the Third Ventricle. Baltimore, Williams & Wilkins, 1987, pp3-31.

Commentary D Operative Management of Malformations of the Vein of Galen J. Gordon McComb, M.D., and Michael L J. Apuzzo, M.D.

It is necessary to differentiate between primary and secondary vein of Galen aneurysms. In the secondary form, the galenic system, including the inferior sagittal and straight sinuses as well as the vein of Galen, is dilated in response to an increased venous outflow from either an adjacent or a remote angioma. The treatment for the sec-

ondary forms is the same as for angiomata in general. In the primary vein of Galen malformation the arterial feeders connect directly into the sack of the aneurysm, which is thick and tough, making its rupture very unlikely. In some cases there can be a small adjacent angioma in association with what is predominantly a primary

vein of Galen aneurysm. Whereas hemorrhage is unusual in the primary form, it is much more frequent when the primary form is associated with an angioma (Figs. D.1 and D.2). The comments that follow are directed to what is predominantly the primary form of vein of Galen aneurysm. Clinical Features and Selection of Management Modes The management of vein of Galen aneurysms is best considered in the context of the patients'

ages at the time of presentation. This consists of four groupings: neonatal, infantile, childhood, and adult, as characterized by Amacher and Shil-lito(2)(TableD.l). The neonate presents at birth with massive high output cardiac failure that responds little to maximal medical therapy. The diagnosis is usually made by cardiac angiography for a suspected heart anomaly and is subsequently confirmed by computerized tomographic (CT) scanning, although intracranial ultrasound can now be considered an initial imaging technique. Surgical

intervention in an infant with intractable cardiac failure has been nearly universally fatal; this demise is usually on a cardiovascular basis because of poor myocardical perfusion. More recently, embolization techniques are proving to be of value. Using a transfemoral route Berenstein and associates (7) have been able to diminish the flow through the fistula significantly by one or more embolization procedures (Fig. D.3). This technique is not without risk as it is possible for the embolized material to pass through the malformation and lodge in the lungs. A more direct

approach has recently been described by Mickle (5). Coiled copper wires are introduced into the aneurysmal sac through the dura mater over the torcular after a burr hole has been made at this location. This technique seems to have greater success with fewer risks. By reducing the shunting through the fistula one can bring the cardiac failure under control. Rarely does the embolization procedure result in complete obliteration of the feeding vessels. It is uncertain whether the benefits of subsequent direct operative intervention to remove the remaining feeders are worth

the risk. Such a decision is also dependent upon the presence or absence of an associated adjacent angioma, making the possibility of hemorrhage a consideration as well. In spite of appropriate and successful medical and surgical treatment of this malformation, the prognosis might still be very poor. Norman and Becker (6) noted extensive parenchymal changes in neonates with this lesion dying shortly after birth without attempted surgical intervention. This indicates that, in some neonates, widespread irreversible parenchymal damage has occurred by the time of delivery. The ability to predict those neonates who will not respond to any form of therapy remains to be established. Hydrocephalus is usually the presenting factor in the infantile group. These infants frequently had mild neonatal cardiac failure that responded to therapy or resolved spontaneously. In these patients it has been the authors' approach to shunt the hydrocephalus if progressive but not to attempt to treat the vein of Galen aneurysm surgically unless there is evidence of hemorrhage, increasing compression of surrounding neural structures, or additive parenchymal damage from the "steal phenomenon" as judged by progressive neurological deterioration and changes on CT or MR imaging. Some of these patients would be suitable for embolization but not those in whom there is evidence of hemorrhage or increasing mass effect. The childhood group may present with progressive hydrocephalus or subarachnoid hemorrhage; the diagnosis is then made by CT scanning and is further defined by angiography. As with the infantile group, the approach would be to intervene surgically in the presence of hemorrhage or mass effect. The hydrocephalus, if progressive, would require shunting. Patients in the adult group often present with

subarachnoid hemorrhage or symptoms and signs of a pineal region mass. In these situations direct surgical intervention is warranted. If hydrocephalus is present, shunting should be undertaken before occlusion of the malformation, as subependymal veins are often more distended after occlusion of the malformation and pose a risk with subependymal passage of instrumentation (8). Operative Approach The surgical approach has been described by Amacher (1), Yasargil et al. (8), and Hoffman et al. (3). It is advocated that a vein of Galen malformation be technically considered like any other pineal lesion and approached via a posterior interhemispheric route, as described in Chapter 25. Vascular Anatomy and Positioning The most common arterial supply to these malformations is from the anterior cerebral arteries, the posterior cerebral and superior cerebellar arteries, and the posterior thalamic perforators arising from the peduncular segment of the posterior cerebral artery. This supply is frequently bilateral (Fig. D.4). Therefore, the operative position and corridor must be determined with these facts taken into consideration. A Supracerebellar subtentorial attack is not appropriate as this approach will not allow access to the more anterior arterial feeders. Although the patient could be placed in either a sitting or a prone position, either a lateral decubitus (pediatric) or 3/4 prone dependent oblique position (adult) is favored (4). Hypotensive anesthetic technique may be required during dissection and collapse of the aneurysm dome.

Operative Technique Basic elements of the approach have been defined in detail in the previous chapter (4). A bone flap that extends from the torcular 8 to 10 cm anteriorly (childhood and adult cases) affords the optimal exposure required for isolation of all vascular input (Fig. D.5). In these cases the exposed cortex appears more vascular, and greater care than usual is required as the interhemispheric

corridor is developed (cuneus region). The Budde ring retractor system is indispensable as a retraction device and hand stabilization platform. In the deep midline the thickened arachnoidal membrane of the callosal, dorsal ambient, and quadrigeminal cisterns is encountered and opened. With the right side approach soft cottonoid paddies are used to isolate the plane between the malformation and the brain (Fig. D.6).

The ambient cistern is opened and the posterior cerebral and superior cerebellar arteries are identified. Distal segments of the anterior cerebral arteries are exposed over the splenium and traced to the dome of the malformation. In each case terminal feeding branches are divided at the malformation, which may be manipulated without significant risk of rupture because of the generally thickened wall. Left lateral feeders may be identified by a transfalcine exposure with elements of rotation and collapse of the dome. A constant vigilance regarding completeness of oc-

clusion requires observations of the color changes in the vein of Galen. When its color changes from red to blue, all feeders have been divided. If no change is observed after occlusion of the distal pericallosal, posterior cerebral, and superior cerebellar artery contributions, it is necessary to explore the region of the large perforating posterior thalamic vessels from the peduncular segment of the posterior cerebral artery. This is accomplished by a retrosplenial approach with depression of the malformation with a cottonoid and retraction of the splenium with a 1/8-

or 1/4-in. self-retaining retractor blade. If no alternative is comfortable, the splenium may be split to expose the cistern of the velum interpositum. At the base of the cistern the tela choroidea of the third ventricle is identified and opened and the striae medullares, massa intermedia, posterior commissure, and habenular trigone are visualized. The posterior thalamic perforators may be identified passing through the habenular trigone to the galenic malformtion at its anterior base. If the superior exposure fails to provide adequate visualization, the inferior surface may be explored with deformation of the inferior and posterior wall of the malformation. After these maneuvers the malformation should be blue. Further checks include Doppler auscultation and evaluation of the O2 and CO2 content of aspirated blood. If the sac is flaccid at the end of the procedure it need not be excised, especially if doing so might damage the hypothalamus or thalamus. Galenic venous drainage need not be of concern when deciding whether to remove the aneurysmal sac as venous drainage by necessity has developed along other channels because of the increased pressure. The younger the patient, the more concern for myocardial function and hemodynamics, both intra- and postoperatively. Evidence of increasing heart failure must be aggressively treated, including pharmacological support, reduction of peripheral vascular resistance, and the withdrawal of blood. Removal of the malformation in several stages, especially in the younger patient, may be advisable. Postoperative angiography is

indicated to assure that all fistulous connections have been disrupted. The patient must be followed indefinitely for signs of hydrocephalus and shunted if necessary. References 1. Amacher AL: Vein of Galen aneurysms. In Wilkins RH, Rengachary SS (eds): Neurosurgery. New York, McGraw-Hill, 1985, pp 1459-1465. 2. Amacher AL, Shillito J Jr: The syndromes and surgical treatment of aneurysms of the great vein of Galen. J Neurosurg 39:89-98, 1973. 3. Hoffman JH, Chaung S, Hendrick EB, Humphreys RP: Aneurysms of the vein of Galen. J Neurosurg 57:316-322, 1982. 4. McComb JG, Apuzzo MLJ: Posterior Intrahemi spheric retrocallosal and transcallosal ap proaches. In Apuzzo MLJ (ed): Surgery oj the Third Ventricle. Baltimore, Williams & Wilkins, 1987, pp 611-640. 5. Mickle JP: The transtorcular approach to vein of Galen aneurysms—a preliminary report. Pre sented at the 13th Annual Meeting of the Section of Pediatric Neurological Surgeons of the Ameri can Association of Neurological Surgeons, Salt Lake City, Utah, December 11-13, 1984. 6. Norman MG, Becker LE: Cerebral damage in neo nates resulting from arteriovenous malformation of the vein of Galen. J Neurol Neurosurg Psy chiatry 37:252-258, 1974. 7. Wisoff JH, Berenstein A, Epstein FJ: Vein of Galen aneurysm: Combined treatment of emboli zation and surgery. Presented at the 13th Annual Meeting of the Section of Pediatric Neurological Surgeons of the American Association of Neuro logical Surgeons, Salt Lake City, Utah, December 11-13, 1984. 8. Yasargil MG, Antic J, Laciga R, Jaim KK, Boone SC: Arteriovenous malformations of vein of Galen: Microsurgical treatment. Surg Neurol 6:195-200, 1976.

26 Controversies, Techniques, and Strategies for Pineal Tumor Surgery Claude Lapras, M.D., and J. D. Patet, M.D.

Different approaches for operation in the pineal region have been proposed. After various experiences we finally chose the occipital supra-and transtentorial approach as the most convenient in cases of pineal tumors. We have now used it for 85 patients. Controversies about the Approaches Transcallosal Approach The transcallosal approach as proposed by Dandy (1) was used in our first eight cases, but problems were more frequent than advantages. We were unable to perform a total removal in any case. The approach is performed along the falx at the level of the parietooccipital junction. It is posteriorly free from veins to the superior sagittal sinus but anteriorly the parietal bridging veins are often encountered and could be damaged by excessive retraction; dividing these veins always initiates a mild hemiparesis or a contralateral astereognosia. When the corpus callosum is reached, there are no landmarks, the approach can be oblique, the pericallosal arteries are variable, and the venous anatomy is not uniform. As the exposure is narrow, finding exactly where the splenium is located is difficult. After dividing the corpus callosum, the internal cerebral veins are dissected. Normally tumor removal is performed between the internal cerebral veins or by retracting them laterally. Keeping veins displaced anteriorly is easy because they are mobile, but posteriorly they join together to form the great vein of Galen and are fixed on the midline; operation between or under them is dangerous and blind. The transcallosal approach was proposed especially for superiorly developed tumors. In these cases (Fig. 26.1 A), starting dissection by locating the internal cerebral veins is difficult because the anatomy is modified. The veins are sometimes displaced laterally, one vein may remain in the midline when the other is distorted laterally, or both may be included in or hidden by the exophytic superior growth of the tumor. The main arguments against using the transcallosal approach are

related to vascular pedicles and tumor extension. Vessels of pineal tumors are almost exclusively situated posterior to the tumor. Arteries come from the medial branches of the posterior choroidal arteries. With a superior approach they are seen only at the end of the procedure. In contrast, they are initially and easily dissected by approaching posteriorly. Small additional arteries run anteriorly and laterally from the distal branches of the pericallosal arteries. They are never large enough to justify a superior approach. Two or three short pineal veins tether the tumor under the vein of Galen. These short tumoral veins are more safely divided after a posterior approach allowing one to dissect below the vein of Galen then after a superior approach. Approaching through the corpus callosum, anterior tumor extensions, exophytic as well as limited, are dissected first. At this time and at this level it is impossible to know precisely whether the tumor invades brain structures. The answer usually lies

posteriorly on the midline near the quadrigeminal plate or posteriorly and laterally near the thalamopulvinar mass. Posterior approaches provide better conditions for the judgment of tumor features. Poppen (3) proposed performing the approach more posteriorly, through an occipital bone flap, dividing only the splenium of the corpus callosum. But division of the corpus callosum itself and even its visualization are not necessary. Operating along a lower route across a sagittal opening of the tentorium is easier. This modified occipital approach was described by Jamieson (2); we have used it since 1970. Transcortical-Transventricular Approach The transcortical-transventricular approach described by Van Wagenen (5) has the main disadvantage of being too lateral. Operation is performed after entering the posterior part of the lateral ventricle through a cortical incision located in the parietal region. This approach is limited to the nondominant hemisphere. In the lateral ventricle, the only landmark is the Choroid plexus and its hilum. The tumor, bulging behind the thalamus beneath the posterior medial wall, is seen after Choroid plexus removal. The route is direct and medial behind the hilum of the Choroid plexus and tangential to the posterior thalamic mass. This approach is deep; the thalamic mass is restricting, and it is difficult to work on the midline and more difficult beyond it. Removal is performed through a narrow venous triangle made by the internal cerebral vein and the basilar vein of Rosenthal. This approach was suitable to only two young patients and one recurrent case. They presented with large pineal tumors developed almost completely laterally toward a dilated lateral ventricle on the nondominant hemisphere. Anterior Subchoroidal Approach If the lateral ventricle is large, entering the frontal horn and posterior opening of the foramen of Monro by section of the thalamostriate vein gives good access to the middle and eventually the posterior part of the third ventricle. In the case of a pineal tumor, this gives access only to the anterior tumor extension. This approach has the same disadvantages as the transcallosal; the vessels are behind the tumor and the critical region to check tumor infiltration is posteriorly and remote from the skull opening. We have never used this approach for pineal operation. Occipital Infratentorial Approach Our experience with this approach (4) is limited. Its main advantage is that with retraction of the upper cerebellum the surgical corridor is free of delicate nervous structures. Theoretically, cerebellar retraction is easy. The sitting position is advantageous. Sacrificing the bridging veins between cerebellum and tentorium is without consequence but the addition of venous sacrifice to cerebellar retraction can explain some persistent postoperative ataxia. In the case of a preoperative tonsillar herniation, retraction could aggravate the situation. This danger was encountered in one of our patients who manifested bradycardia after retractor application. The main disadvantage of the infratentorial approach is the presence of the tentorium. It is restricting, giving two lateral blind corners. The slope of the tentorium produces restricted visualization on each side; therefore, it should be used for small tumors without lateral extension. We found it difficult to perform hemostasis in front of a tumor extending into the cavity of the third ventricle and to dissect the roof of the third ventricle. Therefore, it seems better to use another approach for ante-

riorly developed tumors. Pineal tumors may also extend caudally into two directions. Most commonly, the tumor extends posteriorly above the quadrigeminal plate, filling the quadrigeminal cistern (Fig. 26.1B), displacing the superior cerebellum, and finally being encased behind the brain stem and cerebellum. With this approach it is difficult to remove the posterior extension under direct view. The other posterior extension is into the dilated upper part of the aqueduct of Sylvius (Fig. 26.1С) where a tumoral nodule can prolapse and distend the brain stem. The bulging appearance of the quadrigeminal plate must be carefully evaluated and differentiated from the tumor invasion. The infratentorial approach is not optimal for this visualization or for removing the nodule. However, the infratentorial approach seems appropriate for small pineal lesions that are well cirumscribed, plainly situated in the midline, and associated with small lateral ventricles. Retraction of the upper cerebellum is not modified by ventricular size, whereas retraction of the occipital lobe needed for the occipital supra- and transtentorial approach is more difficult when ventricles are small spontaneously or after a lengthy delay between shunting and direct operation. The Occipital Supratentorial and Transtentorial Approach Position The approach is performed with the patient in the sitting position with the hips completely flexed. The knees are semiflexed and the legs are elevated to the level of the heart. The position is more fetal than sitting. A large belt is applied to the abdomen to reduce peripheral venous volume and diminish blood pressure variations associated with the sitting position. For the same purpose, elastic stockings are used. The head is completely flexed anteriorly and turned slightly downward toward the right side (Fig. 26.2). Head flexion and inclination of the operating table tend to place the right occipital pole at the highest point of the operating field so that during the exposure the occipital lobe is retracted and not simultaneously forcibly elevated. At the completion of operation after retractors are removed, when the flattened hemisphere resumes its normal contour and position, head flexion reduces venous traction. Before dural closure, the cranial cavity can be completely filled with fluid without leakage, preventing postoperative pneumocephalus and decreasing the tension on bridging veins. After finalization of position with flexion of the head the tracheal tube can be incidentally displaced forward to the carina or even into the main stem bronchus. Thus, after head flexion, it is necessary to check the free airway. Most frequently, pineal operation is performed on young patients without cervical spondylosis but, if the patient is elderly, forced head flexion should be avoided. Positive end expiratory pressure is used, FiCO2 is continuously analyzed by a capnograph, and arterial pressure and central venous pressure are monitored. A free access at the patient's neck is preserved to be able to compress the jugular veins at each step of the operation, checking the quality of hemostasis and enlarging venous sinuses and veins when they need to be accurately recognized. Opening The skin flap around the right occipital pole is large enough to identify the sagittal suture, the lambdoid suture, and the upper insertions of the neck muscles. When a shunt is required before direct operation, it is better to insert it on the left side to avoid the catheter crossing the skin flap. Operation is performed on the right side to avoid the development of an alexia without agraphia.

The bone flap is lozenge-shaped. The upper lateral burr hole is situated as far laterally and as high as possible to give room for occipital pole retraction and to prevent brain compression against the craniotomy opening (Fig. 26.2). The upper medial burr hole is made just tangent to the midline marked by the sagittal suture. The lower medial burr hole at the torcular is enlarged by bone removal until the confluence is well exposed and an area of occipital dura mater free of venous sinuses is available. The inferolateral burr hole is less lateral than the upper one to be far from the temporal posterior veins. It is located higher than the level of the lateral sinus located when making the medial burr hole 1 cm above the neck muscle insertions. A free bone flap is turned. Before dividing the dura mater it is useful to ask the anesthesiologist to compress the jugular veins to see precisely where the dural sinuses are located. The intradural walls of the venous sinuses are also checked. The dural incision is blocked by a cross stitch at its end to prevent the danger of the incision spontaneously enlarging after dural retraction and lacerating the venous sinuses. The dural flap is turned along the midline and the dura mater is suspended. Bridging veins between the convexity and the dural flap are never found.

Retraction Minimal retraction of the medial occipital cortex from the falx allows division of posterior arachnoid granulations. At times, especially when the patient is elderly, the dura mater sheds bloody tears at the granulation site. Laying on a piece of Surgicel is preferable to extensive coagulation, which retracts the dura. The parietal posterior bridging veins are not within the area of exposure. The occipital lobe is retracted. Usually, it only has one or two small veins between it and the lateral sinus or directly to the tentorium. They are carefully divided. The retractor is advanced until it reaches the free edge of the tentorium. It is not necessary to retract deeper to see, for example, the vein of Galen or even the splenium of the corpus callosum. It is adequate to appreciate the free border of the tentorium. Some cases complicated by postoperative hemianopsia are probably related to inappropriate use of the retractor. This happens when the retractor is used not only to retract, but simultaneously to elevate the occipital lobe if head position is not appropriate, or an effort is made to see the splenium. Self-retaining retractors are used after the occipital lobe is protected with a collagen sponge. Two retractors are usually used: (a) a larger one upon the inferior surface of the occipital lobe, directed in a superior lateral plane, and (b) a medium-sized instrument vertical and lateral supporting the first retractor as it advances to the tentorium. Dissection The operating microscope is now required. The tentorium is opened along a line parallel and 1.5 cm lateral to the straight sinus. This opening is a key step of the procedure. It is at least half the sagittal length of the tentorium. It is made by opening posteriorly and advancing anteriorly toward the incisural margin. Sometimes, an aberrant venous sinus is seen in the middle of the tentorium (Fig. 26.3B). This venous sinus is divided after opening the dura anterior and posterior to it and closing it by a perforating stitch on each side. The venous sinus of the free edge of the tentorium is small and needs only to be coagulated. It is necessary to take care not to divide the tentorium obliquely and to remain at least 1 cm lateral to the vein of Galen. The lateral flap of the tentorium is retracted by coagulation. The medial flap is suspended by two sutures. An incision made medially at the posterior end of the sagittal incision further allows optimal retraction of the medial flap of the tentorium. To avoid traction upon the vein of Galen one does not forcibly retract the anterior suture. The posterior part of the ambient cistern or pineal cistern is examined (Fig. 26.3C). At times with large pineal tumors that have developed posteriorly, the cistern is filled and the tumor appears immediately. But in the majority of cases, the culmen of the cerebellum is close to the vein of Galen and covers the pineal region. Approaching pineal tumors requires separation of the upper cerebellum from the vein of Galen by dissecting arachnoid and dividing veins crossing the cistern. Dissection of arachnoid is performed near the top of the cerebellum, eventually slightly encroaching upon it, but never at the highest point of the cistern close to the vein of Galen. The arachnoid sheet of the vein, firmly adherent to its walls, is carefully preserved. Arachnoid dissection is extended far laterally around the brain stem. Lateral dissection is easier as arachnoid is less adherent to vessels and brain structures. Two or three veins, the precentral veins, are sagittally arranged in the midline, crossing the cistern and joining the cerebellum and quadrigeminal plate to the vein of Galen. All are sacrificed. Sacrificing these veins is devoid of neurological consequences. (In our experience of 58 patients operated

on according to this technique and anlayzed for late follow-up, 32 have no cerebellar sequellae.) Dissecting the arachnoid and dividing the veins allows the cerebellum to slip slightly into the posterior fossa, enlarging the corridor, displacing the cerebellum away from the vein of Galen (Fig. 26.4A), and facilitating dissection of the quadrigeminal plate and pineal region. Pineal tumors appear under a venous arch made by the vein of Galen (Fig. 26.4B) at the midline and the basilar veins of Rosenthal on each side. This venous arch is well defined and must be preserved. It is the limiting margin under which operation is performed. Pineal tumors encroach upon the rostral quadrigeminal plate where it is impossible to recognize the structure of the posterior commissure. Removal Biopsies for frozen section are taken. Dissection is first performed to ascertain whether the tumor is infiltrating and if removal is possible. Beginning at the midline, between tumor and quadrigeminal plate, where few vessels are observed, eventual extension to the brain stem is recognized. If dissection is possible, the floor of the third ventricle is reached and operation can proceed. Laterally there is a groove between tumor and thalamopulvinar mass where the posterior choroidal artery is visible (Fig.

26.5A). At this level the artery is ascending in the subarachnoid space and is never included in the brain parenchyma or in the tumor mass (Fig. 26.5B). Tumor branches originating along the vertical portion of the artery are also initially situated in the subarachnoid space, where they are easily divided before disappearing into the tumor mass (Fig. 26.5C). Even when pineal tumors fill the cistern, the posterior choroidal artery is not displaced (Fig. 26.5D). After dividing the arteries, it is possible to perform lateral dissection between the tumor and the lateral wall of the third ventricle and again to determine whether the tumor is infiltrating. If tumor is infiltrating toward the brain stem or the thalamopulvinar mass, total removal is impossible and operation is limited to biopsy or partial removal. If the tumor does not seem to be infiltrating at those levels, tumor removal is attempted. Excision begins inferiorly and laterally with care taken not be retract the tumor forcibly. One progressively works forward until the third ventricle, anterior to the tumor, is appreciated. Generally, in approaching from the right side, the left lateral wall of the third ventricle is better visualized than the right. If a nodule extends laterally farther than the right wall of the ventricle, it is possible to remove it under view after venous dissection. Displacing the retractor laterally the upper part of the basilar vein of Rosenthal is exposed. Dissection on the external side of

the basilar vein is possible as there are no venous afferents at this level. Care is required at the lower part of the basilar vein near the level of the tentorium where it receives small, but functionally important, internal occipital veins. External dissection of the basilar vein of Rosenthal gives good access to the ipsilateral wall of the third ventricle. When the tumor is large, it is at times necessary to perform an internal decompression to determine the extent of anterior and superior involvement. Retrograde removal of the upper part of the tumor is then performed, separating its anterior extension from the internal cerebral veins. Anterior dissection is relatively safe because venous afferents are rare at this level. Dissection of the last fragments of tumor remaining fixed under the vein of Galen is painstaking because they are united by a dense, adhering, and thick arachnoid. Sometimes achieving a total removal is impossible, and the fragments below the vein of Galen are coagulated in place. When this is required, we use anticoagulant therapy during the postoperative period. At the completion of the removal, the cavity of the third ventricle is visualized with the massa intermedia, the foramen of Monro, the posterior part of the floor, and eventually the origin of the aqueduct. The space between the quadrigeminal plate and the upper cerebellum is evaluated. The occipital supratentorial and transtentorial approach probably gives the most extensive view of the entire pineal region, allowing total removal

even for large tumors in 66% of cases. We consider that total removal of pineal tumors is the most important factor for late results independent of histological findings. Strategies Some clinical or laboratory data may influence surgical indication. Clinical Symptoms Clinical symptoms are not necessarily reliable guides for outcome. The only noticeable fact is the poor prognosis for patients with impaired consciousness not improved by shunting and for patients presenting a complete Parinaud's syndrome (paralyzed, dilated pupils and convergence palsy associated with the classical upward and downward gaze palsy). These symptoms have been observed in patients with large, malignant, invading tumors. Preoperative diabetes insipidus does not necessarily indicate that the tumor is a germinoma, is invading the floor of the third ventricle, or has even seeded the optochiasmatic region. This symptom has been reported in all types of pineal tumors and can disappear after simple surgical removal. In our experience of five patients with preoperative diabetes insipidus, three have been cured after operation and two still require treatment, but diabetes insipidus is sometimes observed in cases of bipolar germinomas, simultaneously developed in the pineal and optochiasmatic regions. Operation is not indicated in these cases. From the literature, it is difficult to have an idea of the value of pubertas precox for prognosis and surgical indication. We have observed two male patients with pubertas precox. They had malignant pineal tumors, large and diffusely invading in one case and rather small and only superficially encroaching upon the quadrigeminal plate in the other case. Removal was partial in the first case and apparently total in the second case. However, late results were poor in the two patients. Etiology Surgical indications are not influenced by the age of the patient. The same surgical approach can be used in children and adults. The sitting position is no more difficult in children than older patients. However, during infancy, pineal tumors are often associated with extensive hydrocephalus, making operation in the sitting position dangerous when the flattened hemisphere stretches the bridging veins. It is better to delay direct operation until the ventricles become smaller after shunting or to operate with the patient in an alternate position. We have no experience of approaching the pineal region with the patient prone, but it probably is more difficult than with the patient sitting because there is no spontaneous opening of the pineal cistern after dissection of the arachnoid and division of the precentral veins. In infancy, when a pineal tumor is associated with large ventricles, the best approach is probably through the cortex, with the patient lying laterally. The sex of the patient gives some indication of tumor type; pineal germ cell tumors are extremely rare in female patients (17 male/2 female in our statistics; 1 female among 10 germinoma cases). The small probability of a germinoma is an argument for direct operation in a female patient with a pineal tumor. Computerized Tomographic (CT) Appearance It is difficult to know on CT studies whether a tumor is infiltrating (Fig. 26.6). Volume is not a good guide. Large tumors completely filling the

posterior cistern are amenable to total removal. On the other hand, small pineal tumors theoretically suited for total removal could infiltrate the quadrigeminal plate. The limits of tumor on CT scans are better guides than volume (Fig. 26.7). If a tumor shape is clearly delineated, whatever its volume may be, total resection is probably possible. The only difficult point concerns tumors that are richly vascularized. On enhanced CT studies it is sometimes difficult to differentiate between tumor extension toward the corpus callosum and the simple injection of numerous vessels in the roof of the third ventricle or between tumor extension toward the brain stem and a rich vascular network behind the brain stem extending into the tentorium (Fig. 26.8). Laboratory Data Normal cytology of cerebrospinal fluid (CSF) has no diagnostic value. If cytological examination shows tumor cells, any type of malignant pineal tumor is possible, germ cells (malignant teratoma or germinoma) as well as pineal cells (pineoblastomas) or even others (glioblastoma, metastasis, etc.). Therefore, as only a diagnosis of malignancy is made, surgical indications and techniques cannot be directed by this assay. The prognosis of pineal tumors does not change if malignant cells are found in CSF before operation. Tumors associated with malignant cells are limited and resectable as frequently as others. But postoperative radiation therapy and chemotherapy are indicated if tumor cells were observed before operation. The presence of persistent malignant cells in spite of these treatment initiatives carries a bad prognosis. When cytology shows two

types of abnormal cells (large clear germ cells associated with small lymphocytes), the most likely diagnosis is germinoma. Evaluation of blood levels of alpha-fetoprotein (AFP) and beta-chorionic gonatropin hormone (HCG) is actually routinely performed. Elevated levels of HCG alone are observed for all types of malignant germ cell tumors (germinomas, malignant teratomas, choriocarcinomas) whera where elevated levels of AFP would be more specific for malignant teratomas. In our experience, elevated levels of HCG and AFP have also found in patients with pineoblastomas. Thus, tumor markers seem more indicative of a malignancy than a particular histological type. Tumor markers are useful during followup to indicate total removal when they disappear and recurrence when they are again elevated.

Radiotherapy Test Analysis of the preceding data indicates that the accurate clinical diagnosis of a germinoma is rarely possible. However, this diagnosis is the critical point for surgical indications because germinomas are very radiosensitive and could be cured without direct operation. Other malignant pineal tumors (malignant teratomas, pineoblastomas) are also radiosensitive, but the possibility of cure without operation is much less. If a stereotaxic biopsy is performed, the diagnosis of germinoma is possible and radiotherapy first is a good strategy. When stereotaxy is not available or if the tumor seems too small or to have a rich vascular bed, thus making stereotaxy dangerous, the diagnosis of germinoma may be suspected in relation to certain clinical variables. It is established on variables of sex (male), clinical signs (eventually diabetes insipidus), CT appearance (medium size tumor, nonhomogeneous, developed on the midline, spontaneously dense, becoming irregularly hyperdense after enhancement, without calcifications), angiographic appearance (vascular bed composed of many thin capillary vessels, with a regular blush, without large spots injected by contrast agent), cytology of CSF (eventually two cell types), and blood levels of tumor markers (eventually elevated HCG, no AFP). If germinoma is suspected clinically, others have proposed abbreviated radiotherapy or a full radiotherapy cycle. If the tumor size distinctly diminishes after 2000 rads and if the tumor disappears after radiotherapy is completed, the diagnosis of germinoma is proven. If the tumor size is not modified by radiation or diminishes only slightly or even increases, the diagnosis of germinoma is questionable. Operation is then undertaken, but time has been consumed and operative conditions are worse than before radiation. Dissection is more difficult because the irradiated tumor is fibrous and strongly adherent to adjacent vessels and nervous structures. The arachnoid is thickened and

dense. Direct operation was necessary for three patients after the failure of shunting and radiotherapy, the so-called "conservative treatment." We could achieve a total removal in only one case (benign teratoma); the patient is still alive. Total removal was impossible for two others (meningioma and malignant teratoma). Both patients died later of local progression, after 7 years with meningioma and after 3 months with teratoma. We do not recommend the use of a radiotherapy test. To treat malignant tumors it is preferable to resect the tumoral mass before radiotherapy, even performing a total removal in some cases. This strategy gives the best chance for prolonged survival for the patient. In the event that stereotaxis is not available, we also recommend the direct approach. The diagnosis is made by frozen section and the occipital transtentorial approach allows for a large surgical removal without particular risks. Radiotherapy is used after surgical resection. We have had no operative mortality and have achieved a 90% late survival for germinoma patients treated according to this strategy. References 1. Dandy WE: Benign Tumors in the Third Ventricle of the Brain: Diagnosis and Treatment. Springfield, Charles С Thomas, 1933. 2. Jamieson KG: Excision of pineal tumors. J Neurosurg 35:550-553, 1971. 3. Poppen JL: The right occipital approach to a pinealoma. J Neurosurg 25:706710, 1966. 4. Stein BM: Supracerebellar-infratentorial approach to pineal tumors. Surg Neurol 11:331-337, 1979. 5. Van Wagenen WP: A surgical approach for the removal of certain pineal tumors. Surg Gyncecol Obstet 53:216-220, 1931.

27 Pineal Region and Posterior Third Ventricular Tumors: A Surgical Overview Keiji Sano, M.D., D.M.Sc, FAC.S.(hon.)

So-called Germ Cell Tumors and Pineal Tumors in the Pineal-Posterior Third Ventricular Region Rene Descartes (1596-1650), the greatest philosopher of 17th century France, thought that the pineal body might act as a valve or sphincter to regulate the passage of spirits between the ventricular chambers, and this tiny organ thus gained supreme importance as the seat of the soul. This concept is, of course, wrong. The pineal body or gland, however, is certainly the seat of various kinds of tumors. Tumors in the pineal region occupy 4% of all intracranial primary neoplasms in Japan (583 of 14,672 cases) (9, 47) (Table 27.1). This is definitely a higher incidence than in any other country (2, 45-48, 56). This high incidence is mostly comprised of the socalled germinoma, or pinealoma, of the two-cell pattern type (PTC), which forms 2.6% of all intracranial primary neoplasms according to the most recent Japanese statistics (9, 47). In these same statistics, pineocytoma occupies 0.2%, pineoblastoma 0.1%, and teratoma including malignant teratoma 0.7%. In Cushing's series (10), pinealoma is found in 0.7%, and in the Chinese statistics (8) it appears in 0.9%. Most European and American statistics show a lower incidence of this tumor than these figures (8). Table 27.2 shows 125 cases of tumors in the pineal region and the posterior third ventricle (excluding falcotentorial meningiomas) treated in my clinic and its affiliated hospitals. The pineal tumor of the two-cell pattern type, which has been called pinealoma (17, 18, 24) or germinoma (12, 13, 42-44), is described in this table as PTC. Pineoblastoma corresponds to medulloblastoma of the pineal of Zulch (63). Mixed germ cell tumors are composed of two or three types of tumors such as germinoma (seminema), teratoma, teratoma with malignant transformation, embryonal carcinoma, choriocarcinoma, and endodermal sinus tumor (Table 7). According to the studies of Zulch and my group (31, 40, 47), PTC can be divided into germinoma (seminoma) of germ cell origin and pinealoma of pineal parenchymal origin. If we admit, although there is no convincing

evidence, the germ cell theory that teratoma, teratoma with malignant transformation, germinoma (seminoma), embryonal carcinoma, endodermal sinus tumor, choriocarcinoma, and mixed germ cell tumors all arise from the imaginary germ cell (12, 13, 55), they can be grouped under the term "germ cell tumors." The tumors in the pineal region and the posterior third ventricle are classified as follows: (a) germ cell tumors — including germinoma (seminoma), teratoma (and epidermoid or dermoid), teratoma with malignant transformation, embryonal carcinoma, endodermal sinus tumor, choriocarcinoma, and mixed germ cell tumors; (b) tumors of pineal parenchymal origin — including pinealoma, pineocytoma, and pineoblastoma; (c) gliomas; and (d) tumors of the velum interpositum — including meningioma, hemangiopericytoma, angiomas, and cysts. In this chapter, I will deal mainly with Groups a and b. PTC, as its name implies, is composed of two distinct cell types. Groups of large, polygonal cells with frequently clear cytoplasm are separated by fibrous strands with perivascular lymphocytic infiltration, displaying a "mosaic pattern." Because of its morphological similarity to seminoma or dysgerminoma, the term germinoma, which was proposed by Friedman (12, 13) and later adopted by Russell and Rubinstein (42-44), or atypical teratoma (Russell and Rubinstein [44]), has now received wide acceptance. The term "pinealoma" was introduced by Krabbe in 1923 (24) in a very

casual way as "an adenoma, or more exactly, 'pinealoma'," was adopted by Horrax and Bailey (17, 18), and has since been generally used. It is held to signify a tumor of the pineal parenchymal cells. The tumor is likewise composed of masses of large spherical epithelial cells separated by a reticular connective tissue containing lymphoid cells. Both Bailey (3, 17, 18) and Zulch (63) regarded these epithelial cells as originating from the parenchymal cells of the pineal body. There are, less frequently, tumors composed of a homogeneous mass of smaller epithelial cells without lymphoid cells. Bailey (3, 17, 18) called tumors of this type "pinealoma of spongioblastic type" or "pineoblastoma." Russel and Rubinstein regarded this type of tumor as the true pinealoma of pineal parenchymal origin and divided it into "pineocytoma" and malignant "pineoblastoma" (44). Zulch named these two "isomorphic pinealoma" and "medulloblastoma of the pineal," respectively, and called the two-cell pattern type tumor "anisomorphic pinealoma" (63). These rather chaotic nomenclatures are listed in Table 27.3. PTC is found not only in the pineal region but also in other areas of the brain. Table 27.4 lists PTCs in various sites that have been treated in my clinic and histologically verified. The results seem to favor the classification of PTC as a germinoma (seminoma). For reevaluation of the histogenesis of PTC, 48 surgical specimens, a number sufficient for detailed study, were selected and carefully reviewed by both light and electron microscopic examinations, as was reported elsewhere (47). Of the total 48 cases of PTC, 28 cases were located in the pineal body, 11 in the suprasellar-hypothalamic area, and the remaining 9 in other sites (Table 27.5). They consisted of 41 males and 7 females, indicating a marked preponderance of males over females. Thirty-three patients were younger than 20 years of age and 15 patients were 20 or older.

The pineal organ of lower vertebrates functions as a photoreceptor that transduces an exteroceptive input (light) into neural signals. In birds, the pineal is not a true photoreceptor but in response to light it changes the rate of synthesis of hormonal products like melatonin (photoendocrine transducer). In mammals the pineal gland is an endocrine organ with some of the properties of a "neuroendocrine transducer": one of its major functions is to convert the input of neural signals to a hormonal output, i.e., melatonin and other methoxyindoles and probably Polypeptides (7). In the normal pineal gland, lobules consisting of pineal parenchymal cells are incompletely separated by irregular trabeculae and fibrous septae carrying blood vessels along with them. Like all other endocrine organs, the pineal body has a very rich capillary blood supply (sinusoids). The parenchymal cells attach closely to the sinusoidal sheaths. The intimate proximity of the cells to the sinusoids is regarded as a common feature of the endocrine glands. Thus, it might be expected that true tumors originating from the pineal parenchymal cells would show a close relationship between the tumor cells and the sinusoids as the tumor stroma. The seminiferous tubules constitute the exocrine portion of the testes. There is no direct interrelationship between the tubules and blood vessels from the tunica vasculosa testis. Typical microscopic pictures of seminoma reveal regular, fibrous supporting stromata containing capillaries and lymphocytes that divide the tumor into lobules. There is no evidence of tumor cells attaching to the capillaries. Based on this new criterion that in a true tumor of pineal origin tumor cells should be, as in any endocrine neoplasm, in close proximity to the capillaries (sinusoids) without intervening mesenchymal tissues, histological reevaluation of 48 specimens was performed. Besides this fundamental finding of close proximity to the capillaries, the pattern of lymphocytic infiltration is different in tumors of pineal origin and semino-mas. In the former, lymphocytes are mainly seen along the margins of tumor nests or clusters with occasional sparse scattering in tumor cell clustlers. In the latter, lymphocytes are seen mainly in fibrous septa with occasional granulomatous reaction, something never noticed in tumors of pineal origin. In addition, tumor cells in seminoma are somewhat larger than those in tumors of pineal origin. According to this criterion, PTC can be divided into two groups: one corresponds to a true tumor of pineal origin, which may be called pinealoma, and the other to the seminoma, which, according to Friedman (12), may be called germinoma. In some parts of pinealoma, transition from the main component of

large, ovoid cells with central, vesiculated nuclei to rather smaller cells with eccentric, dark-staining nuclei and eosinophilic cytoplasm is frequently observed (47). The cytological features of these smaller cells are exactly compatible with those of pineocytoma postulated by Russell and Rubinstein (16, 42, 44). Therefore, the available evidence strongly favors that Russell's pineocytoma be regarded as the mature type of pinealoma here described. Ultrastructural observations of this pinealoma (31, 47) also support the concept of its endocrine properties. The cytoplasmic processes of the tumor cells extend into the perivascular interstices, and the tumor cells remain closely adjacent to the fenestrated pores of the endothelium. The tumor cells extensively show smooth-surfaced endoplasmic reticula in the tubular and vesicular forms of cisterns, which may represent their secretive phase. Also, microtubules, as one of the characteristics of pineal cells, are noticed in the cytoplasm of the tumor cells. According to the formulated criterion, 24 of the 48 cases could be assigned to the category of pinealoma (Table 27.5). Of these 24 cases of pinealoma, 19 were found in the pineal region, 2 in the suprasellar-hypothalamic area, and 3 in other sites. The sites of the other 24 cases with germinoma included the pineal region in 9 cases, the suprasellarhypothalamic area in 9, and the other sites in 6. Of tumors involving the pineal region, pinealoma accounted for two-thirds of the 28 cases (19 cases); in contrast, 80% of suprasellar-hypothalamic neoplasms were germinoma (Table 27.5). In both pinealoma and germinoma, males were predominant (83 to 88%). However, 2 pinealoma patients in the suprasellar region and 3 of 9 germinoma patients in the suprasellar region were female. Therefore, in pinealoma and germinoma in the suprasellar region the male predominance was not marked. The results found in earlier ultrastructural studies (54) supporting the close resemblance between pineal tumors with a mosaic pattern and testicular seminomas may have been because those specimens had been mostly obtained from suprasellarhypothalamic tumors. The age of predilection of patients with pinealoma involving the pineal region ranged from 12 to 20 years, with an average of 16.4 ± 4.8. Suprasellarhypothalamic germinomas mostly appeared between the ages of 8 and 12, with a mean age of 10.4 ± 2.5—slightly younger than that for pinealoma. However, germinomas of the pineal region did not show such a predilection for younger ages. The average age of pineoblastoma cases was 13.8 ± 10.9 (16), i.e., younger than that of pinealoma cases. According to Herrick and Rubinstein (16), the average age of pineocytoma cases was 42.6 ± 20.6. In terms of age, therefore, pinealoma is situated between immature pineoblastoma and mature pineocytoma. Neoplasms of endocrine organs usually appear after endocrine functions have developed to some extent, and so the average age of 16 years for pinealoma seems compatible. In terms of sex distribution, pinealoma, pineoblastoma, and probably pineocytoma, as well as germinoma showed a male predominance (Tables 27.2, 27.4, and 27.5). Table 27.6 shows cases of the so-called intracranial germ cell tumors experienced in my clinic. All of these tumors were located in the area of the third ventricle including the pineal and suprasellar regions. Germinoma (9 cases) is included in this list; pinealoma, however, is excluded for the reason mentioned previously. All mature teratoma cases (16 cases), 4 cases of teratoma with malignant transformation, 15 of 30 mixed germ cell tumor cases, and 1 case of unclassified germ cell tumor were located in the pineal region (see Table 27.2). Table 27.7 shows the histological components of the mixed germ cell tumors described. As seen in Table 27.7, two or three components can be noted.

Among 84 patients with germ cell tumors, 66 (78.6%) were under the age of 20 and 18 (21.4%) were at least 20 years. There were 70 (83.3%) males and 14 (16.7%) females (Table 27.6). This age and sex distribution for intracranial germ cell tumors is essentially similar to that for germ cell tumors of other organs. Therapeutic Principles and Preoperative Evaluation Diagnosis of a medium-sized or large tumor arising in the pineal region and the posterior third ventricle is not difficult because of the presence of increased intracranial pressure, paralysis of the conjugate upward gaze (Parinaud's sign), pseudo-Argyll Robertson pupil, etc. Cerebral an-

giography is useful to detect the extent and vascularization of the tumor (38). Calcification in plain craniograms may sometimes be pathogno-monic, especially in patients under 10 years of age. The most powerful and noninvasive diagnostic tool, however, is computerized tomography (CT). Magnetic resonance imaging may likewise be useful; I, however, have had no experience with its use. Before the advent of CT, ventriculography, especially positive-contrast ventriculography, had widely been used, but it is now used only in exceptional cases. Examination of levels of alpha-fetoprotein (AFP), human chorionic gonadotropin (HCG), and carcinoembryonic antigen in the serum or in the cerebrospinal fluid should be done in all cases. The first two tumor markers (AFP and HCG) are especially informative as to the nature of tumors. If HCG is positive, the tumor must be choriocarcinoma or a mixed germ cell tumor with choriocarcinomatous elements. If AFP is positive, the tumor must be an endodermal sinus tumor or a mixed germ cell tumor with endodermal sinus elements. In both cases, prognosis is rather poor even after surgical removal of the tumor and postoperative radiotherapy of the whole neuraxis. Recently, chemotherapeutic agents such as actinomycin D, cis-platinum, vinblastine, bleomycin, etc., or their combinations have been reported to be worthy of use. If AFP and HCG are negative, the tumor may be pinealoma, germinoma, embryonal carcinoma, teratoma with malignant transformation, or mature teratoma. If 2000 rads of local radiation effectively abolishes the tumor on CT, it may be pinealoma or germinoma, and radiotherapy should be continued with or without a shunting procedure. However, if the tumor is more than 2 cm or sometimes more than 1.5 cm in diameter, direct operation and removal of the tumor followed by radiation should be recommended. (For pinealoma and germinoma, local radiation is usually

sufficient.) Mature teratoma can be cured only by removal of the tumor, and its prognosis is good. Embryonal carcinoma and teratoma with malignant transformation usually carry a poor prognosis even after surgical removal and postoperative radiotherapy of the whole neuraxis. The previously mentioned chemotherapeutic agents are also recommended as being worthwhile. These principles are schematically illustrated in Figure 27.1. Stereotaxic biopsy of tumors of this region has recently been gaining in popularity. I, however, am not particularly enthusiastic about this procedure because different parts of the same tumor of this region may show different histological findings so that biopsy of a small piece of the tumor may be misleading as to the true nature of the tumor. Therefore, I prefer exploration and partial resection (unless total removal is possible) of the tumor. Cytological examination of the cerebrospinal fluid is important for diagnosis. If malignant neoplastic cells are identified at cytology, the case may develop dissemination metastases in the cerebrospinal fluid space. For diagnostic purposes, I recommend Millipore filter-cell culture (49) of the cerebrospinal fluid. This method is more sensitive than conventional cytological studies. Therefore, positive culture does not necessarily mean that the probability of dissemination metastases is high. After treatment is finished, during follow-up, repeated examinations of the tumor markers are useful to detect recurrence of the tumor producing these markers at the earliest possible stage. Historical Perspectives of Various Approaches Horsley (19) was probably the first to try to remove a pineal tumor. He used the infratentorial Supracerebellar approach, but with poor results. He therefore recommended supratentorial approaches. In 1913, Krause successfully removed a huge tumor in the region of the corpora quadrigemina from a 10-year-old boy by the infratentorial Supracerebellar approach, which he reported with Oppenheim, who diagnosed the case (34). The tumor was reported to be a fibrosarcoma or an encapsulated mixed cell sarcoma, but seems, however, to have been a teratoma or meningioma according to modern pathological designation. In 1926 (25), Krause added two more cases operated upon by the same approach; in these two, however, he could not remove the tumors. Practically the same approach was reported by Zapletal in 1956 (62). He reported four cases: a malignant astrocytoma in the quadrigeminal region, a medulloblastoma of the upper vermis, an epidermoid, and a pinealoma. The last was successfully removed. In the era of microsurgery, Stein (51), in 1971, revived and elaborated this infratentorial Supracerebellar approach, which has been widely used since (35, 47, 48). The occipital or parietooccipital approach with or without splitting the splenium of the corpus callosum or the tentorium along the straight sinus has been proposed by Brunner (5, 41), Pratt (37), Heppner (15), Poppen (36), Kempe (23), Glasauer (14), Jamieson (20), Lazar and Clark (29), Sano (45-48), Lapras (27), and others. Special credit should be given to Jamieson, who established the occipital transtentorial approach. The parietal transcallosal approach was proposed by Dandy (11), then by Kunicki (26), Suzuki and Iwabuchi (53), and others. In 1931, Van Wagenen (57) proposed the posterior transventricular approach. This has, however, been used only occasionally. For huge tumors in the pineal region or in the posterior third ventricle, Sano proposed the anterior or frontal transcallosal approach (47).

Other approaches such as the subchoroidal approach (28, 58) or the transcallosal interforniceal approach (1), which are primarily for lesions in the middle or anterior third ventricle, may be used for tumors in the pineal-posterior third ventricular region. These various approaches and the pertinent anatomy were summarized by Rhoton and his group (39, 61). Of these approaches, I prefer (a) the occipital transtentorial approach, (b) the infratentorial Supracerebellar approach, and for huge tumors (c) the frontal transcallosal approach. Indications for Utilization of the Various Approaches For the occipital transtentorial approach to a pineal tumor, I prefer the incision (usually on the nondominant side) illustrated in Figure 27.2. The midline portion of the incision can be elongated to the suboccipital region if it is necessary to open the posterior fossa. Craniotomy is close to the superior sagittal sinus and the transverse sinus. The patient is either in the prone position or the lounging position, as seen in Figure 27.2. The occipital lobe is punctured through the dura

mater, and a silicone rubber tube is inserted into the posterior portion of the lateral ventricle and secured to the dura (Fig. 27.10). The tube drains the cerebrospinal fluid during the operation and makes lateral retraction of the occipital lobe easier. The tentorium is split close to the straight sinus, and the superior surface of the cerebellum is exposed. The operating microscope is used from this stage. A pineal tumor is often visible rostral to the vermis, underneath the vein of Galen. If the tumor is located further rostrally, the splenium of the corpus callosum is split by suction to expose the tela choroidea of the third ventricle. If the tumor is large, it is often already breaking through the tela choroidea; if not, the tela is incised along the midline or close to the nondominant occipital lobe after cauterization with a bipolar coagulator. Therefore, this approach enables the operator to use the low (parie-to-)occipital approach (as proposed by Poppen (36) and Jamieson (20)) and, if necessary, the high parietooccipital approach (as proposed by Dandy (11) and others) as well. If the tumor is a pinealoma or a germinoma (germinoma is usually slightly tougher), the tumor is removed piecemeal. If the tumor is a teratoma, removal en bloc will often be feasible. After removal of the

tumor, the other end of the silicone rubber tube (which has been inserted into the lateral ventricle) is brought into the lateral cistern or the pontine cistern to secure the cerebrospinal fluid pathway. This is done because the rostral portion of the aqueduct is often compressed by the tumor, and even after removal of the tumor the effect of the compression may remain for a certain period.

The infratentorial Supracerebellar approach is utilized when the pineal tumor is not too large. The superior surface of the cerebellum is pressed down after electrocauterizing and severing the bridging veins between the cerebellum and the tentorium (Fig. 27.3). The tumor can be seen beneath the vein of Galen and between the basal veins of Rosenthal. The precentral cerebellar vein and the superior vermian vein are identified and severed. The tumor is usually removed piecemeal. Figure 27.4 shows CT of a 7-year-old boy with a medium-sized tumor in the pineal region that was removed by this approach. Figure 27.5 shows the tumor being removed piecemeal with the aid of the Cavitron ultrasonic aspirator. In Figure 27.6, no tumor is visible; the normal wall of the third ventricle can now be seen. Figure 27.7 is a postoperative CT (1 week after the operation). The tumor is totally removed. This tumor turned out to be a

teratoma with malignant transformation. The patient is well and attending school a little more than 3 years postoperatively. I previously performed the infratentorial Supracerebellar approach with the patient in a sitting position as advocated by Stein (51). However, because of the fear of air embolism I am now using this approach with the patient in an oblique prone position as shown in Figure 27.8. The operator stands to the left of the patient and looks down on the pineal region through the operating microscope.

Another point that should be mentioned and has hitherto never been described is that in rare cases the vein of the cerebellomesencephalic fissure (30) or the superior and inferior quadrigeminal veins are so well developed that the tumor cannot be reached directly from behind. In that case the tumor should be approached slightly obliquely between these veins and the basal vein of Rosenthal. Another alternative for such cases is the occipital transtentorial approach. As shown in Figure 27.9, the tumor is attacked from a more superior and lateral direction without touching the quadrigeminal veins. I have experienced no operative mortality either by the occipital or the infratentorial approach. Steroid is administered before, during, and after the operation. Postoperative irradiation for pinealoma or germinaoma was done with Co-60 or LINAC, the total dose being 5000 to 6000 rads (the daily dose, 100 to 200 rads). The field of irradiation was 6 X 6 to 8 X 8 cm, centering on the pineal region. In cases of malignant tumors, whole brain irradiation and spinal cord irradiation were added, but even in these cases the total dose did not exceed 5000 to 6000 rads. Several years ago, I treated two cases (4- and 7-year-old boys) of huge malignant pineal teratoma with an endodermal sinus tumor element that showed increased serum AFP. I thought that these were both too large to attack either by the occipital transtentorial approach or by the infratentorial Supracerebellar approach (Fig. 27.10 and 27.11). I decided to remove the tumors en bloc by the frontal transcallosal approach. Frontal craniotomy was fashioned, as shown in Figure 27.12A. The corpus callosum was longitudinally split, about 4 cm in length, at its junction with the nondominant cingulate gyrus (Fig. 27.12B). The tumor was easily identified and gently detached from the surrounding tissues under the micro-scope and then was removed en bloc (Fig. 27.13 and 27.14). The tumors were 4.5 cm and 5 cm in diameter. After the operation, a marked decrease of AFP was noted. The postoperative course was uneventful and the patients returned to school. The 4-year-old boy is still healthy more than 3 years postoperatively. The 7-year-old boy, however, later developed diffuse metastases in his body and died 1.5 years later.

I favor the occipital transtentorial approach or the infratentorial Supracerebellar approach if the pineal tumor is located posterior to the adhesio interthalamica. If the tumor is too large and reaches anterior to the adhesio iterthalamica, the transcallosal, especially the frontal transcallosal approach may be recommended. Radiotherapy or Direct Operation Both pinealoma and germinoma are radiosensitive (4, 6, 21, 33, 50, 52, 59, 60). Therefore, radiation alone or radiation plus ventriculoperitoneal shunting prevails as the established treatment. The greatest criticism against direct operation is the anticipated significant operative mortality and the increased probability of dissemination metastases in the cerebrospinal fluid space.

However, there was no operative mortality in my series. I believe that meticulous surgical maneuvers, especially microsurgical ones, do not carry a significant operative risk. I also believe that reduction of the bulk of a neoplasm by direct operation, even if that neoplasm is radiosensitive, is important for the effectiveness of radiotherapy. This may be clearly demonstrated if one compares the results of treatments of pineal pinealoma-germinoma with those of suprasellar pineal-oma-germinoma, where reduction of the bulk of the tumor is easier and more extensive (Table 27.8). The 10-year survival rate (calculated by the Kaplan-Meier method (22)) of suprasellar pinealoma-germinoma cases was 100%, whereas that of pineal pinealoma-germinoma cases was 74.9% in my series. In the literature, the incidence of dissemination metastases of pinealoma or germinoma in the cerebrospinal fluid space is 8 to 20% (48); in my series it was 3.7%. Therefore, one cannot say that direct operation of the tumor will increase the probability of dissemination metastases. In Japan, the general trend is first to irradiate the tumor with 2000 rads. If the tumor disappears on CT, radiation therapy is continued; if not, direct operation is performed because in that case the tumor is most probably a neoplasm other than pinealoma or germinoma. However, when a pinealoma or a germinoma is of considerable size, it may recur even if the tumor disappears on CT after radiotherapy. Therefore, I recommend direct operation and postoperative radiotherapy for

pinealoma-germinoma cases (except for cases with small or multiple tumors). Patients harboring mixed germ cell tumors usually have a poor prognosis even if their tumors are totally removed and radiotherapy follows. Cases of mature teratoma, on the other hand, show good outcome in the follow-up. Table 27.9 shows survival rates of patients with tumors other than pinealoma and germinoma. Summary and Conclusion There are various kinds of tumors in the pineal-posterior third ventricular region. Among them, the so-called two-cell pattern tumor is most frequent. These tumors can be divided into tumors of pineal parenchymal origin (pinealoma) and tumors of germ cell origin (germinoma or seminoma). These two-cell pattern tumors and the so-called germ cell tumors are also found in other areas of the brain, most frequently, however, in the area of the third ventricle. The germ cell tumors are grossly divided into two groups by the tumor markers (AFP and HCG), with those being either positive or negative. However, exact diagnosis can be made only by operation and histological examination of the surgical specimens. In treatment, mature teratoma should be surgically removed because it is not radiosensitive. Pinealoma and germinoma (seminoma) are radiosensitive; if they are of a considerable size, however, surgical removal and postoperative radiotherapy should be recommended (for these tumors, local radiation may be sufficient because dissemination metastases are not frequent). Embryonal carcinoma, teratoma with malignant transformation, choriocarcinoma, endodermal sinus tumor, and their combinations (mixed germ cell tumors) should be submitted to operation and postoperative radiation (whole neuraxis) combined with chemotherapy. Pinealoma, germinoma, and mature teratoma carry good prognoses, but other tumors carry rather poor prognoses. Among the various surgical approaches to tumors in the pineal-posterior third ventricular region, the infratentorial Supracerebellar approach and the occipital transtentorial approach may be the methods of choice if the tumor is confined posterior to the adhesio interthalamica. If the tumor is large and reaches anterior to the adhesio interthalamica, the frontal transcallosal approach may be preferable. References 1. Apuzzo MLJ, Chlkovani OK, Gott PS, Teng EL, Zee C-S, Giannotta SL, Weiss MH: Transcallosal interforniceal approaches for lesions affecting the third ventricle: Surgical considerations and consequences. Neurosurgery 10:547554, 1982. 2. Araki C, Matsumoto S: Statistical re-evaluation of pinealoma and related tumors in Japan. J Neurosurg 30:146-149, 1969. 3. Bailey P: Intracranial Tumors. Springfield, IL, Charles С Thomas, 1948. 4. Bradfield JS, Perez CA: Pineal tumors and ectopic pinealomas: Analysis of treatment and failures. Radiology 103:399-406, 1972. 5. Brunner C: Cited by Rorschach H (Ref. 41). 6. Camins MB, Schlesinger EB: Treatment of tumours of the posterior part of the third ventricle and the pineal region: A long term follow-up. Acta Neurochir{Wien) 40:131-143, 1978. 7. Cardinali DP: Melatonin: A mammalian pineal hormone. Endocrine Rev 2:327-346, 1981. 8. Cheng M-K: Brain tumors in the People's Republic of China: A statistical review. Neurosurgery 10:16-21, 1982. 9. Committee of Brain Tumor Registry in Japan: Brain Tumor Registry in Japan, 1982, vol 4. 10. Cushing H: Intracranial Tumors: Note upon a Series of 2000 Verified Cases

with Surgical Mortality Pertaining Thereto. Springfield, IL, Charles С Thomas, 1932, pp 62-65. 11. Dandy W: An operation for the removal of pineal tumors. Surg. Gynecol Obstet 33:113-119, 1921. 12. Friedman NB: Germinoma of the pineal: Its identity with germinoma ("semi noma") of the testis. Cancer Res. 7:363-368, 1947. 13. Friedman NB: The comparative morphogenesis of extragenital and gonadal teratoid tumors. Cancer 4:265-276, 1951. 14. Glasauer FE: An operative approach to pineal tumors. Acta Neurochir (Wien) 22:177-180, 1970. 15. Heppner F: Zur Operationstechnik bei Pinealomen. Zbl Neurochir 19:219224, 1959. 16. Herrick MK, Rubinstein LJ: The cytological differentiating potential of pineal parenchymal neoplasms (true pinealomas): Clinicopathological study of 28 tumours. Brain 102:289-320, 1979. 17. Horrax G, Bailey P: Tumors of the pineal body. Arch Neurol Psychiatry 13:423-470, 1925. 18. Horrax G, Bailey P: Pineal pathology: Further studies. Arch Neurol Psychia try 19:394-414, 1928. 119. Horsley V: Discussion. Proc Roy Soc Med 3(Pt 2):77, 1910. 20. Jamieson KG: Excision of pineal tumors. J Neurosurg 35:550-553, 1971. 21. Jenkin RDT, Simpson WJK, Keen CW: Pineal and suprasellar germinomas: Results of radiation treatment. J Neurosurg 48:99-107, 1978. 22. Kaplan EL, Meier P: Non-parametric estimation for incomplete observations. J Am StatAssoc 53:457-481, 1958. 23. Kempe LG: Operative Neurosurgery. New York, Springer-Verlag, 1968, vol 1. 24. Krabbe KH: The pineal gland, especially in relation to the problem of its supposed significance in sexual development. Endocrinology 7:379-414, 1923. 25. Krause F: Operative Freilegung der Vierhugel nebst Beobachtungen iiber Hirndruck und Dekompression. Zbl Chir 53:2812-2819, 1926. 26. Kunicki A: Operative experience in 8 cases of pineal tumor. J Neurosurg 17:815-823, 1960. 27. Lapras C: Experience with the direct surgical approach in 52 tumors of the pineal region. In Paoletti P, Walker MD, Butti G, Knerich R (eds): Multidisciplinary Aspects of Brain Tumor Therapy. Amsterdam, Elsevier/North Hol land Biomedical Press, 1979, pp 173-176. 28. Layne MH, Patterson RH: Subchoroidal trans-velum interpositum approach to mid-third ventricular tumors. Neurosurgery 12:86-94, 1983. 29. Lazar ML, Clark K: Direct surgical management of masses in the region of the vein of Galen. Surg Neurol 2:17-21, 1974. 30. Matsushima T, Rhoton AL Jr, de Oliveira E, Peace D: Microsurgical anatomy of the veins of the posterior fossa. J Neurosurg 59:63-105, 1983. 31. Matsutani M: Ultrastructural study of pinealoma in view of its histogenesis. Brain Nerve {Tokyo) 28:41-56, 1976 (in Japanese). 32. Matsutani M, Sano K, Takakura K, Seto T: Intracranial germ cell tumors. Neurosurgeons [Tokyo) 3:83-100, 1984. 33. Mincer F, Meltzer J, Botstein C: Pinealoma. Cancer 37:2713-2718, 1976. . 34. Oppenheim H, Krause F: Operative Erfolge bei Geschwulsten der Sehhugel-und Vierhugelgegend. Berl Klin Wchschr 50:2316-2322, 1913. 35. Pendl G: Infratentorial approach to mesencephalic tumors. In Koos WT, Bock FW (eds): Clinical Microneurosurgery. Stuttgart, Georg Thieme, 1976, pp 143-150. 36. Poppen JL, Marino R Jr: Pinealomas and tumors of the posterior portion of the third ventricle. J Neurosurg 28:357-364, 1968. 37. Pratt OW, Brooks EF: Successful excision of a tumour of the pineal gland. Can Med Assoc J 39:240-243, 1938. 38. Raimondi AJ, Tomita T: Pineal tumors in childhood: Epidemiology, pathophysiology, and surgical approaches. Child Brain 9:239-266, 1982. 39. Rhoton AL Jr, Yamamoto I, Peace DA: Microsurgery of the third ventricle: Part 2. Operative approaches. Neurosurgery 8:357-373, 1981. 40. Riverson EA, Zulch KJ: Pineal parenchymal tumours and germinomas (the problem of the so-called pinealomas). Neurosurg Rev 1:3-l 1, 1979. 41. Rorschach H: Zur Pathologie der Tumoren der Zirbeldruse. Beitr Klin Chir 83:451, 1913.

42. Rubinstein LJ: Tumors of the Central Nervous System. Atlas of Tumor Pathology, series 2, fascicle 6. Washington, DC, AFIP, 1972, pp 269-284. 43. Russell DS: The pinealoma; its relationships to teratoma. J Pathol Bacteriol 56:145-150, 1944. 44. Russell DS, Rubinstein LJ: Pathology of Tumours of the Nervous System, ed4. London, Arnold, 1977. 45. Sano K: Pinealoma in children. Child Brain 2:67-72, 1976. 46. Sano K: Diagnosis and treatment of tumours in the pineal region. Acta Neurochir (Wien) 34:153-157, 1976. 47. Sano K: Pineal region tumors: Problems in pathology and treatment. Clin Neurosurg 30:59-91, 1983. 48. Sano K, Matsutani M: Pinealoma (germinoma) treated by direct surgery and postoperative irradiation: A long-term follow-up. Child Brain 8:81-97, 1981. 49. Sano K, Nagal M, Tsuchida T, Hoshino T: New diagnostic method of brain tumors by cell culture of the cerebrospinal fluid—Millipore filter-cell culture method. Neurol Med Chir 8:17-27, 1966. 50. Smith NJ, EL-Mahdi AM, Constable WC: Results of irradiation of tumors in the region of the pineal body. Acta Radiol Ther Phys Biol 15:17-22, 1976. 51. Stein BM: The infratentorial Supracerebellar approach to pineal lesions. J Neurosurg 35:197-202, 1971. 52. Sung D II, Harisiadis L, Chang CH: Midline pineal tumors and suprasellar germinomas; highly curable by irradiation. Radiology 128:745-751, 1978. 53. Suzuki J, Iwabuchi T: Surgical removal of pineal tumors (pinealomas and teratomas): Experience in series of 19 cases. J Neurosurg 23:565-571, 1965. 54. Tabuchi K, Yamada O, Nishimoto A: The ultrastructure of pinealomas. Acta Neuropathol (Berl.) 24:117-127, 1973. 55. Teilum G: Classification of endodermal sinus tumor (mesoblastoma vitellinum) and so-called embryonal carcinoma of the ovary. Acta Pathol Microbiol Scand 64:407-429, 1965. 56. Ueki K, Tanaka R: Treatments and prognosis of pineal tumors—experience of 110 cases. Neurol Med Chir [Tokyo) 20:1-26, 1980. 57. van Wagenen WP: A surgical approach for the removal of certain pineal tumors: Report of a case. Surg Gynecol Obstet 53:216-220, 1931. 58. Viale GL, Turtas S: The subchoroid approach to the third ventricle. Surg Neurol 14:71-76, 1980. 59. Wara WM, Fellows CF, Sheline GE, Wilson CB, Townsend JJ: Radiation therapy for pineal tumors and suprasellar germinomas. Radiology 124:221223, 1977. 60. Wara WM, Jenkin RDT, Evans A, Ertel I, Hittle R, Ortega J, Wilson CB, Hammond D: Tumors of the pineal and suprasellar region: Childrens Cancer Study Group treatment results 1960-1975. Cancer 43:698-701, 1979. 61. Yamamoto I, Rhoton AL Jr, Peace DA: Microsurgery of the third ventricle: Part 1. Microsurgical anatomy. Neurosurgery 8:334-356, 1981. 62. Zapletal B: Ein neuer operativer Zugang zum Gebiet der Incisura Tentorji. Zbl Neurochir 16:64-69, 1956. 63. Zulch KJ: Brain Tumors: Their Biology and Pathology. New York, Springer, 1957.

28 Technical Aspects of Excision of Giant Basal Tumors with Third Ventricular Involvement Madjid Samii, Prof. Dr. med.

The third ventricle with its central location in the intracranial cavity can be involved by the different pathological processes arising from structures at the skull base. Giant tumors of the sellar and parasellar region reach the third ventricle anteriorly. They push or invade its wall and further propagate, filling its cavity. Processes along the tentorial margin can grow and involve the third ventricle from a caudolateral aspect. The posterior part of the ventricle can be affected by those pathological processes originating in the pineal region. The planning of our operative strategy for excision of these lesions will depend on their primary site of origin and their propagation. Now, with our increased knowledge of the microanatomy of the third ventricle and its surrounding delicate structures as well as the tremendous advancement of microsurgical techniques, we are able to deal with these lesions with the least percentage of morbidity and nearly negligible mortality. In this book all different approaches to the third ventricle are described in other chapters. Our only concern is directed to the surgical management of giant pathological processes that secondarily involve the third ventricle. Giant Space-occupying Lesions Involving the Anterior Portion of the Third Ventricle Tumors of the sellar and parasellar region may secondarily involve the anterior wall of the third ventricle, later filling its cavity. There they may compress its wall or obstruct the cerebrospinal fluid (CSF) flow by closure of the foramen of Monro, producing different neurological deficits. The most important pathological processes in this group are Craniopharyngiomas and pituitary adenomas. In the past, surgical excision of these giant tumors was usually partial; total removal was accompanied by a high percentage of both morbidity and mortality. This is due to the involvement of many important structures like the optic nerves, chiasm, hypothalamus, and perforating vessels of the circle of Willis. Delicate

and meticulous microsurgical techniques are required to avoid damaging these structures. During surgical excision the strategy for total removal will depend on whether the tumor is compressing the surrounding neural tissues or, as in rare cases, infiltrating these tissues, e.g., craniopharyngioma (28). Another fact that influences the ease or difficulty of operation is how the nerves and vessels are involved in the neoplasm. In general, we should divide all tumors into two categories. One group of tumors grows and enlarges symmetrically, compressing and displacing the structures in all possible directions. In such a tumor all important structures are localized around the tumour capsule. Gradual resection and central reduction of the tumor will make the capsule collapse, giving more and more space. The tumor capsule can then be easily dissected from the important structures. In such cases the huge size of the tumor does not add to the gravity or risks of total removal. In contrast to this group we are sometimes confronted by a second group of tumors that grow and then engulf surrounding vessels and nerves, like the carotid, anterior cerebral, anterior choroidal, posterior communicating, and perforating arteries, as well as the cranial nerves. The removal of these tumors can be hazardous with a high rate of morbidity and sometimes mortality. Therefore, the strategy of microsurgical approach and tumor excision requires a firm comprehension of the microanatomy that is altered to a great extent by the growing neoplasm. In handling these tumors one must start at definite anatomical landmarks, beginning where the surrounded vital structures are not involved and obscured by the pathological process. The tumor can be removed piecemeal and the structures dissected stepwise and freed. A direct attack into the center of these tumors can lead to traumatization of the structures engulfed and increases the risk of morbidity or even mortality. Fortunately these types of suprasellar tumors are only a minority. For the removal of giant sellar and parasellar tumors with extensive involvement of the third ventricle, many approaches have been described (3, 17, 18, 29, 30). As the tumor arises primarily basely and secondarily extends into the third ventricle, we prefer subfrontal access (6, 27) compared to a transcortical (4) or transcallosal approach (10). This approach is more direct and gives much more space than the subtemporal (5) or transtemporal technique (31). A giant basal tumor elevates the suprasellar structures and enlarges the suprasellar space. This may lead to a retrofixed chiasm with opening of the angle between the optic nerves and the chiasm (Fig. 28.1). In certain cases the tumor can separate the supracavernous part of the internal carotid artery (ICA) from the optic nerve, giving sufficient distance for a lateral approach to the tumor. In these situations a subchiasmatic or a laterochiasmatic approach can be performed. If the distance between the tuberculum sellae and the optic chiasm is not large enough for total removal, a retrochiasmatic approach through the lamina terminalis should be performed. The choice of the frontal craniotomy, whether right, left, or bifrontal, depends on the additional extension of the tumor to the parasellar and retrosellar region. In the case of midline tumors with no lateral extension a right frontal craniotomy is performed. With extreme parasellar extension to one side we prefer a craniotomy on the opposite side. This is important as it gives a better subchiasmatic visual access to the structures at the tentorial margin. In cases of bilateral extension of the tumor, which is common with giant basal tumors, we prefer a small bifrontal craniotomy. The crani-

otomy should be quite basal, extending to the roof of the orbit without respect to the frontal sinus. The superior sagittal sinus will be ligated and the attachment of the falx to the crista galli will be transected. With microsurgical dissection both olfactory nerves can be preserved (19, 26). Case 1 A 9-year-old girl presented in a somnolent state with a history of headache and gradual deterioration of visual acuity. Ophthalmological examination revealed central tubular vision in the left eye, perception of light in the right eye, and bilateral primary optic atrophy. Computerized axial tomography (CT) (Fig. 28.2) showed a large suprasellar cystic craniopharyngioma with a small solid sellar part. The tumor filled the third ventricle, occluding the foramen of Monro and producing obstructive hydrocephalus. A ventriculoperitoneal shunt was done to improve the level of consciousness by relieving the obstructive hydrocephalus. Two weeks later a bifrontal osteoclas-tic craniotomy was carried out and the anterior part of the tumor was exposed.

The tumor was punctured and the cyst was aspirated. Gradual resection of the tumor capsule from the left and right optic nerves and carotid arteries was done. The majority of the capsule and the tumor could be removed subchiasmatically, but there was still a tumor remnant left behind the chiasm and in the third ventricle. The third ventricle was opened through the lamina terminalis and the tumor was totally excised. After total removal we had a good view of the pons, interpeduncular fossa, and basilar artery and its tributaries (Fig. 28.3). During the early postoperative period the child suffered from a transient diabetes insipidus and was for 3 days somnolent. The patient was discharged 5 weeks after operation in a very good general condition. Her visual deficit was the same. A postoperative control CT scan showed complete excision of the tumor (Fig. 28.4). Case 2 A 56-year-old woman presented with a giant recurrence of a pituitary adenoma. She had undergone operation 6 years before and postoperatively had suffered from mental disturbances, bilateral dyskinesia, panhypopituitarism, and bitemporal hemianopsia that did not improve. CT (Fig. 28.5) showed a giant parasellar tumor extending suprasellarly into both lateral ventricles and compressing the third ventricle. Angiography showed definitive displacement and engulfment of the right carotid artery and its bifurcation. The tumor was approached through a bifrontal craniotomy. The parasellar part was compressing the cavernous sinus and right optic nerve. The supracavernous part of the ICA and its bifurcation were completely surrounded by the tumor. Superiorly, the tumor extended through the frontal lobe to both lateral ventricles. Posteriorly, it indented the third ventricle. To avoid damaging the surrounding structures we used the first few millimeters of the right ICA as a landmark and followed the tumor around the vessels. After exposure of the carotid bifurcation further dissection and removal of the tumor around the anterior and middle cerebral arteries was carried out, sparing the important perforating vessels. In the next step the tumor in front of the lamina terminalis was removed, exposing the cavity of the third ventricle. At the last stage the part of the tumor extending to the lateral ventricle was removed. Postoperatively the patient made an excellent recovery and was discharged in the same clinical condition as preoperatively. (Fig. 28.6 shows a postoperative CT scan.)

Giant Space-occupying Lesions Involving the Lateral Portion of the Third Ventricle Processes arising along the tentorial margin can reach the third ventricle in a caudolateral aspect. This region is of particular importance because of its close relationship to functionally important cranial nerves and vessels as well as brain stem structures (12). Pathological processes of the tentorial notch that reach a huge size and may compress the third ventricle are usually meningiomas, chordomas, and neurinomas. There are many operative approaches (11, 14, 16) to deal with lesions in the region of the tentorial notch. We prefer the pterional approach modified by Yasargil (33) for lesions in the region of the anterior and middle third of the tentorium. A curved frontotemporal incision is made inside the hairline and running 1 cm anterior to the tragus. In this way one secures an adequate approach to the lesser wing of the sphenoid and the floors of the anterior and middle fossae. After raising a small frontotemporal flap and removing the lateral part of the lesser wing, one continues intradurally along the sphenoid wing in the direction of the anterior clinoid process by gradually opening the cisterns and removing CSF. This procedure must not be rushed; by constant aspiration of CSF and gentle retraction the frontal and temporal lips of the sylvian fissure can be gradually opened without any undue pressure on the brain. After exposure of the optic nerve and further opening of the cisterns along the edge of the tentorium, the internal carotid, posterior communicating, and anterior choroidal arteries and the oculomotor nerve are exposed in the anterior third of the tentorial notch. By further dissection along the free margin of the middle third of the tentorium in a posterior direction, the trochlear nerve and the posterior cerebral artery and its origin from the basilar artery become visible. After incising the margin of the tentorium and stitching it back in its middle third one has a very good view into the depth of the infratentorial space. Our further surgical strategy depends on the anatomical variations of the tentorial notch (9). In some individuals the tentorial notch is narrow. In such a situation, exposure of the tentorial margin does not permit any view to the infratentorial region. The previously mentioned incision of the tentorium can be helpful in certain circumstances, but would still not

be sufficient for those giant tumors that extend infratentorially in the cerebellopontine angle and prepontine space. To avoid a second stage operation through a lateral suboccipital approach, we coagulate and transect the superior petrosal sinus and expose the petrous apex. This can be removed carefully up to the internal auditory meatus with a high speed diamond drill (Fig. 28.7). The seventh and eighth cranial nerves are now visible and can be dissected free from the lesion. The manner in which we deal with the fifth nerve depends on the clinical findings and morphological behavior of the tumor. If there is no functional deficit and the nerve is only compressed by the tumor, all efforts are undertaken to preserve the nerve. In cases of functional deficit and infiltration of the nerve fascicles, e.g., invasive meningioma, one must sacrifice the nerve for more radical removal of the tumor. The most important aspects of such giant tumors at the tentorial margin are their relation to vessels supplying the brain stem and the severe adhesion of the tumor to this very delicate area. According to my personal experience one can decompress the brain stem by gradual resection of the tumor from lateral to medial in a piece-meal manner until the tumor is completely removed. During dissection some small arteries can be sacrificed without compromising the blood supply to the mesencephalon. As the circle of Willis demonstrates a tremendous degree of variation we are not able to determine in each individual case the exact blood supply and collateral circulation of the brain stem (13). There are no fixed criteria to be taken into consideration for this special problem. The

outcome of our operated cases has shown that many of those vessels that supply the tumor and also the brain stem can be sacrificed without any postoperative consequences. It seems that it is more dangerous to apply traction to brain stem vessels by pulling on the tumor. Traction of any vessel at the brain stem can lead to interstitial rupture and bleeding, which causes severe brain stem lesions and marked neurological deficits. It seems wiser to coagulate and cut such vessels when necessary. Case 1 A 55-year-old Italian man complained of headache, diplopia, right hemiparesis, and regurgitation that had been progressive over 1 year. On examination the patient suffered from bilateral papilledema, right sixth and seventh nerve palsy, diminished pharyngeal reflexes, bilateral pyramidal syndrome (more on the right side), and a positive rombergism sign. Plain x-ray tomography showed widening of the sella with destruction of the dorsum sellae and clivus and parasellar calcification. CT scanning and magnetic resonance imaging (MRI) showed a huge

space-occupying lesion in the region of the clivus, displacing the brain stem upward and dorsally, compressing the third ventricle and mesencephalon with secondary obstructive hydrocephalus, and extending bilaterally (more to the right side) down to the foramen magnum. Anteriorly it extended supra- and parasel-larly, invading the right cavernous sinus (Fig. 28.8A). Angiography showed dorsal displacement of the basilar artery (Fig. 28.8B) and opening of the carotid bifurcation. The right pterional approach was chosen as it allows handling of the tumor in the suprasellar, parasellar, and petroclival region. Through a right frontotemporal osteoplastic flap, the right sylvian fissure was opened. The suprasellar part was removed using a laterochiasmatic approach. The supracavernous part of the ICA was dissected from the tumor up to the carotid bifurcation. Incision of the tentorium was carried out and the superior surface of the tumor in the right middle fossa was exposed. The oculomotor nerve was gradually exposed by piecemeal removal of the tumor. The tumor was followed to the cavernous sinus, which was opened, excising the intracavernous portion of the tumor and freeing the third, fourth, and sixth nerves. The last step was meticulous dissection and piecemeal removal of the part of the tumor adherent to the mesencephalon and brain stem, freeing the basilar artery and its branches. At the end of this procedure complete excision of the tumor was accomplished to the contralateral side of the brain stem. One could then have a good view of the structures on both sides of the tentorial edge. The patient made an excellent recovery with improvement of his paresis. The histopathological report was chordoma. Follow-up CT scan showed complete excision of the tumor (Fig. 28.9). Four weeks later the patient left the hospital, walking unassisted, and flew back home. Case 2 An Indian man, 56 years old, complained of gradually progressing left hemiparesis of 6 months' duration. This was accompanied by emotional instability and a complete fifth cranial nerve lesion. His CT scan showed a petroclival mass destroying the clivus and the petrous apex and compressing the third ventricle, producing secondary obstructive hydrocephalus (Fig. 28.10). A ventricular shunt was performed in India. After verification of the benign nature of the tumor (meningioma) through a stereotoaxic biopsy, the patient was referred to our clinic. The modified pterional approach was carried out for total removal of the tumor. After removal of the supratentorial part, the tentorium was incised, the superior petrosal sinus was coagulated and transected, and the petrous apex was drilled out to the internal auditory meatus. Because of the previous clinical loss of

trigeminal nerve function and its infiltration by the tumor, the trigeminal as well as the trochlear nerves were sacrificed to allow complete removal of the tumor. The patient had an uneventful postoperative course. He left the hospital 4 weeks later completely recovered from his hemiparesis. Giant Space-occupying Lesions Involving the Posterior Part of the Third Ventricle The pineal region may show a wide variety of pathological processes, ranging from the most malignant to the most benign (21). These tumors indent the posterior aspect of the third ventricle or infiltrate its wall, finally filling the ventricular cavity. The most common tumors in this region are the so-called germinomas or pinealomas of the twocell pattern type (PTC) (23). Other lesions that may sometimes reach huge sizes are falcotentorial meningiomas, teratomas, and pineoblastomas. The strategy for management of these pathological processes is still a matter of controversy. Germinomas or the two-cell pattern type of pinealom is radiosensitive. Therefore some advocate radiation alone or radiation plus ventricular shunting as the established treatments. The greatest criticism against direct operation is the anticipated significant operative mortality and the increased probability of dissemination metastases in the CSF. With the recent advances in the technology of diagnosis, improved pre- and post-

operative management, and the perfection of operative techniques, no significant operative risk is encountered when dealing with these tumors. It seems also that reduction of the bulk of the neoplasm by direct operation, even if that neoplasm is radiosensitive, is important for the effectiveness of radiotherapy. Sano (20, 22) also reported no increased incidence of dissemination metastases in his series of operated patients with such tumors. By using the conservative line of management we can miss patients harboring benign lesions that could be cured completely by a direct attack. In our clinic we manage these cases by direct surgical attack. The following treatment (radiotherapy, chemotherapy) will depend on the pathological character of the tumor. Usually a preliminary idea of the pathological classification of the tumor is obtained before operation from radiography (CT scan, angiography, MRI) and hormonal assay (alpha-fetoprotein (AFP), human chorionic gonadotropin (HCG)) in the serum and CSF. Several approaches have been proposed for the excision of tumors in

the pineal region. These include the parietal transcallosal (2, 25), the occipital or parietooccipital transtentorial (1, 15), the infratentorial Supracerebellar (8, 24), the posterior transventricular (32), and the frontal or anterior transcallosal approach (20). The infratentorial Supracerebellar approach is preferred for most tumors invading the pineal region because the access is not obstructed by the deep venous system that caps the dorsal and lateral aspects of pineal tumors. The occipital transtentorial approach is used only for tumors above the tentorial edge if there is no major extension to the posterior fossa or opposite side and for those located above the vein of Galen. In the infratentorial Supracerebellar approach the patient is in the semisitting position. Some surgeons prefer the concorde position (7). The head and neck is flexed to bring the tentorium near the horizontal plane. A vertical midline incision is used. The suboccipital craniectomy extends above the lower edge of the torcular and both transverse sinuses. The dural incision extends up to the inferior margin of the transverse sinus. Gravity assists the cerebellum in falling down away from the tentorium cerebelli. The straight sinus and tentorium may be elevated with a retractor and the vermis is retracted gently inferiorly. Bridging veins over the superior surface of the cerebellum may be coagulated and transected without risk. Incision of the arachnoid over the quadrigeminal cistern brings the tumor and deep venous system in view. The vein of Galen and the internal cerebral veins are above the tumor. The precentral vein is posterior, the thalamus, the medial posterior choroidal and posterior cerebral arteries, and the basal veins are lateral. The superior cerebellar arteries and veins are below. Case 1 A 15-year-old boy presented in a stuporous state. His symptoms had begun 2 years previously with headache and diabetes insipidus followed by bilateral hearing disturbance. On examination the patient had bilateral papilledema and a positive Parinaud's sign. A CT scan showed a hyperdense space-occupying lesion in the pineal region with hydrocephalus (Fig. 28.11, A and B). A ventricular shunt was inserted. After his general condition improved, excision of the tumor was accomplished through a Supracerebellar subtentorial approach. Postoperatively the patient showed an excellent recovery; his follow-up CT scan demonstrated complete excision of the tumor (Fig. 28.11С). The tumor excised was histopathologically verified as a teratoma with malignant changes. The patient received postoperatively deep x-ray therapy as a supplementary treatment. References 1. Ausman J: A new operative approach to the pineal region. In Samii M (ed): Surgery in and around the Brain Stem and Third Ventricle. Berlin, Sprin ger- Verlag (in press). 2. Dandy W: An operation for the removal of pineal tumours. Surg Gynecol Obstet 33:113-119, 1921. 3. Hardy J, Vezina JL: Transphenoidal neurosurgery of intracranial neoplasm. AdvNeurol 15:261-274, 1976. 4. Hirsch JF, Zouaoui A, Renier D, Pierre K: A new surgical approach to the third ventricle with interruption of the striothalamic vein. Acta Neurochir {Wien) 47:135-147, 1979. 5. Kempe LG: Operative Neurosurgery. Berlin, Springer-Verlag, 1968, vol 1. 6. King TT: Removal of intraventricular craniopharyngioma through the lamina terminalis. Acta Neurochir {Wien) 45:277-286, 1979. 7. Kobayashi S, Sugita K, Tanaka Y, Kyoshima K: Infratentorial approach to the pineal region in the prone position: Concorde position. J Neurosurg 58:141-143, 1983. 8. Krause F: Operative Freilegung der Vierhugel nebst Beobachtungen uber Hirndruck und Dekompression. Zbl Chir 35:2812-2819, 1926. '9. Lang J: Anatomy of the tentorial margin. Adv Neurosurg 13:173-182, 1985. 10. Long DM, Chou SN: Transcallosal removal of craniopharyngioma within the third ventricle. J Neurosurg 39:563-567, 1973.

11. Malis LI: Surgical resection of tumours of the skull base. In Wilkins RH, Rengachary SS (eds): Neurosurgery, 1985, pp 1011-1021. 12. Michio O, Makiko O, Rhoton Jr AL: Microsurgical anatomy of the region of the tentorial incisura. J Neurosurg 60:356-399, 1984. 13. Mitterwallner FV: Variationsstatistische Untersuchungen an den basalen HintergefaBen. Acta Anat [Basel) 24:51, 1955. 14. Perneczky A, Horaczek A: Approaches to the tentorial edge, demonstrated by references to 30 cases of tentorial meningiomas. Adv Neurosurg 13:186190, 1985. 15. Poppen JL, Marino R Jr: Pinealomas and tumours of the posterior portion of the third ventricle. J Neurosurg 28:357-364, 1968. 16. Ramina R, Samii M, Baumann H, Hermans B: Surgical management of skull base meningioma. Presented at the Third International Course of Neurosur gery, Toronto, Canada, 1985. 17. Rand R: Transfrontal, transphenoidal craniotomy in pituitary and related tumours. In Microneurosurgery, ed 3, 1985, pp 135-145. 18. Rhoton AL Jr, Yamamoto I, Peace DA: Microsurgery of the third ventricle: Part 2. Operative approaches. Neurosurgery 8:357-373, 1981. 19. Samii M: Olfactory nerve. In Samii M, Jannetta PJ (eds): The Cranial Nerves. Berlin, Springer-Verlag, 1981. 20. Sano K: Microsurgery of tumours in the pineal region. In Microneurosurgery, ed 3, 1985. 21. Sano K: Pineal region tumours: Problems in pathology and treatment. Clln Neurosurg 30:39-91, 1983. 22. Sano K, Matsutani M: Pinealoma (germinoma) treated by direct surgery and postoperative irradiation: A long-term followup. Child's Brain 8:81-97, 1981. 23. Sano K, et al: Brain Tumour Registry in Japan, 1981, vol 3. 24. Stein BM: The infratentorial Supracerebellar approach to pineal lesions. J Neurosurg 35:197-202, 1977. 25. Suzuki J, Iwabuchi T: Surgical removal of pineal tumours: Experience in a series of 19 cases. J Neurosurg 23:565-571, 1965. 26. Suzuki J, Yoshimoto T, Mizoi K: Preservation of the olfactory tract in bifrontal craniotomy for anterior communicating artery aneurysms and functional prognosis. J Neurosurg 54:342-345, 1981. 27. Suzuki J, Katakura R, Mori T: Interhemispheric approach through the lamina terminalis to tumours of the anterior part of the third ventricle. Surg Neurol 22:157-163, 1984. 28. Sweet WH: Recurrent craniopharyngioma. In Edwards JMR (ed): Topical Reviews In Neurosurgery. Bristol, Wright, 1982, vol I, pp 134-160. 29. Symon L: The temporal approach for resection of craniopharyngioma. In Symon L (ed): Operative Surgery: Neurosurgery. London, Butterworth, 1979, pp 185-186. 30. Symon L, Jakubowski J, Kendall B: Surgical treatment of giant pituitary adenomas. J Neurol Neurosurg Psychiatry 42:973-982, 1979. 31. Symon L, Sprich W: Radical excision of craniopharyngioma: Results in 20 patients. J Neurosurg 62:174-181, 1985. 32. Van Wagenen WP: A surgical approach for the removal of certain pineal tumours: Report of a case. Surg Gynecol Obstet 53:216-220, 1931. 33. Yasargil GM, Fox JL,Ray MW: The operative approach to aneurysms of the anterior communicating artery. Adv Technical Standards Neurosurgery 2:113-170.

29 Cerebrospinal Fluid Diversion J. Gordon McComb, M.D., and F. Miles Little, M.D.

Cerebrospinal Fluid Dynamics The multiple functions of cerebrospinal fluid (CSF) coupled with its fairly rapid turnover indeed make CSF "the third circulation" as first referred to by Cushing (2). The circulation of CSF throughout the central nervous system is maintained primarily by the hydrostatic difference between newly formed CSF within the ventricles and parenchyma and that at its sites of absorption. Less important factors influencing CSF circulation are associated with pulsations of the brain and choroid plexus from the arterial tree, respiratory variations, changes in bodily position, and ciliary action of the ependyma and choroid plexus epithelium. CSF is produced in all four ventricles although the majority originates from the lateral ventricles, consistent with the larger amount of choroid plexus tissue present in the lateral as compared with the midline ventricles. The lateral ventricles are joined together at the anterior superior aspect of the third ventricle by the foramina of Monro, each of which measures less than a centimeter at its greatest diameter. Obstruction to CSF flow can occur at one or both of these foramina by a fairly small mass. The third ventricle, normally only 3 to 5 mm in width, drains into the aqueduct of Sylvius, the diameter of which usually measures only 2 mm. The relatively long and narrow aqueduct of Sylvius is the most frequent site of obstruction within the ventricular system. Tumors in and around the third ventricle can either obstruct the CSF before it enters the third ventricle from the lateral ventricles or prevent its egress from the posterior portion of the third ventricle into the aqueduct of Sylvius. The first step in the formation of CSF is the passage of an ultrafiltrate of plasma through the non-tight junctioned choroidal capillary endothelium by hydrostatic pressure into the surrounding connective tissue stroma beneath the epithelium of the villus. The ultrafiltrate is subsequently transformed into a secretion (namely, CSF) by an active metabolic process within the choroidal epithelium via a mechanism that is largely speculative. Approximately 80 to 90% of CSF formation occurs at the choroid plexus with the remainder being produced in the parenchyma, most likely at the capillary-glial complex. The CSF formation rate is about

20 ml per hour or 500 ml per day. Because the total amount of CSF in the ventricles and subarachnoid space (SAS) in the adult averages approximately 150 ml, a threefold turnover of CSF occurs daily. Production of CSF is only slightly pressure-responsive and does not diminish noticeably until cerebral prefusion pressure is markedly reduced» interfering with the first step in CSF production, reducing the quantity of ultrafiltrate from the choroidal capillary. Absorption of CSF and its constituents depends upon bulk flow in addition to passive or facilitated diffusion and active transport of specific solutes. With the single rare exception of CSF overproduction by a choroid plexus papilloma, hydrocephalus results from impaired CSF absorption. Production of CSF in hydrocephalus is either normal or nearly normal. In compensated hydrocephalus, the rate of absorption must equal the rate of formation, approximately 500 ml per day, and in the noncompensated state only a small fraction of the total amount secreted is retained; thus the overwhelming majority of CSF output is still absorbed. As CSF formation is relatively constant, the change in resistance to absorption determines CSF pressure and whether the hydrocephalus is progressive. Impairment of CSF absorption in communicating hydrocephalus could occur at some or all of the following sites: the arachnoid villus, the lymphatic channels associated with cranial and spinal nerves, the lymphatic channels in the adventitia of the cerebral vessels, or the arachnoid membrane. If CSF outflow from the ventricles is blocked (i.e., noncommunicating hydrocephalus), fluid flow could still take place via the PierreRobin spaces surrounding the blood vessels and the extracellular space of the cortical mantle to reach the SAS on the brain's surface. Assuming a complete blockage within the ventricular system, which is not always the case, additional ways for CSF to exit the ventricles would be through a fistulous opening created by rupture of the ventricular system into the SAS at a location such as the lamina terminalis or suprapineal recess. CSF drainage could then occur as if the ventricular obstruction did not exist (15). The question of whether CSF can be absorbed by the brain has been debated for some time. Penetration of substances into the periventricular region of the hydrocephalic animal has been well documented. With the advent of computerized tomographic (CT) scanning, periventricular hypodensity may be noted in the presence of hydrocephalus and has been shown to be the result of CSF migrating into the area surrounding the ventricles in the face of increased intraventricular pressure (ICP). CSF in the parenchyma, indicative of migration, does not necessarily equate with absorption. The extracellular space in the brain, which amounts to 15%, readily allows fluid flow in the parenchyma under normal physiological conditions and its velocity and direction is responsive to changes in hydrostatic and osmotic pressure gradients. Macromolecules injected into the CSF of the ventricles or SAS have been observed to penetrate readily the extracellular space of the parenchyma and vice versa, there being no barrier to free movement across either the ependyma or the pial membranes. Evidence thus supports the contention that the brain, rather than absorbing CSF, is acting as a conduit for fluid to move from the ventricles through the parenchyma to the SAS. With very high ICP, it is possible that absorption of CSF may also occur via the choroid plexus. This would involve the flow of CSF (or extracellular fluid within the parenchyma) into the stroma of the choroid plexus from the subependymal region. Absorption of CSF into the blood could take place via the choroidal capillaries as these capillaries are of the fenestrated type and thus do not have tight junctions. There is no evidence to support CSF being absorbed by the parenchymal capillary epithelium. This does

not preclude, however, net water changes between the parenchymal extracellular fluid and blood when dysequilibrium in the blood-brain osmotic gradient occurs as there is no barrier in this regard. Ventriculostomy

General Considerations The initial management of hydrocephalus associated with tumors in or about the third ventricle often requires CSF diversion. This may be accomplished either by placing a ventriculostomy or by inserting an extracranial shunting system before a direct surgical procedure of the third ventricle. We prefer a ventriculostomy to a shunt followed by an attempt to reestablish CSF circulation at the time of tumor removal, thereby obviating the need for a shunt if this is successful. Also, placing a ventriculostomy and keeping the ventricles somewhat dilated by draining CSF only if the ICP exceeds 15 to 20 torr will permit further decompression of the ventricles interoperatively, improving exposure to the third ventricular region. With a permanent shunting system in place, the ventricles may come down to normal size by the time of the definitive operative procedure and this opportunity will have been lost. In the immediate postoperative period the presence of a ventriculostomy will allow monitoring of ICP as well as determination of whether CSF circulation has been adequately reestablished. Operative Corridor: Anatomy and Physiological Risks The ventriculostomy insertion site is usually the frontal or parietal region. The right side is usually chosen as it is rarely the dominant hemisphere. If one lateral ventricle is significantly more dilated than the other, ventriculostomy placement should be into the larger ventricle. Our personal preference for site of ventriculostomy insertion is the frontal region under most circumstances because in such a location (a) the ventriculostomy can be placed more easily in an intensive care unit (ICU) setting; (b) the external landmarks are more readily apparent; (c) smaller ventricles may be more easily cannulated, decreasing the risk for damage to other structures; (d) the frontal region is easier to bandage; (e) there is less problem with the patient lying on the bandage or tubing; and (f) if it is necessary to insert a permanent CSF-diverting shunt, the ventriculostomy in the frontal region would not be in the operative field at the posterior parietal region, our preferred site for the insertion of a permanent diverting system. Physiological risks are approximately the same when comparing the frontal and parietal approaches, with the one exception of an increased risk of hemiparesis if the catheter is passed through the internal capsule. A possible risk with frontal insertion is that of damage to the hypothalamic structures if the tubing is passed through the lateral ventricles and into the floor of the third ventricle. CT scans have shown on more than one occasion that the tip of the catheter has obviously gone into the hypothalamic region without any clinical evidence of dysfunction. Structural Definitions Initial diagnostic studies are either CT scanning with or without contrast agent and magnetic resonance imaging (MRI). Angiography is not necessary unless the mass might be vascular. If a question exists as to whether adequate communication is present between the lateral ventricles via the foramina of Monro, metrizamide can be placed into the ventricular system once the ventriculostomy is in place. Small obstructing

lesions in the posterior portion of the third ventricle can be more accurately outlined with metrizamide. Operative Technique To diminish the incidence of infection, which is the primary ventriculostomy complication, strict attention to aseptic technique is very important whether the ventriculostomy is placed in an ICU setting or in the operating room. Hair is removed for a considerable distance around the planned site of ventriculostomy insertion; in fact, in most instances it is probably just as well to remove all of the hair as it will be necessary to do so later at the time of the definitive procedure. The skin is scrubbed with alcohol and a degreasing agent (such as Freon) and then well prepared with povidone-iodine (Betadine) solution. At the site of planned incision, the skin is injected with lidocaine containing epinephrine. In the child or adult, a linear incision 1 cm long, is made 1 to 2 cm anterior to the coronal suture in the midpupillary line (Fig. 29.1 A and B). The coronal suture is used as a site of entry in the infant. A hand or twist drill may be used to make the opening through the skull. A hand drill is faster and easier but is less safe than the twist drill. In the infant or young child in whom the skull is not thick, a smallerdiameter drill can be used. In the older child or adult, a larger hole is necessary. It is easier to make the initial hole with a smaller diameter bit followed by a larger one rather than using the larger diameter drill at the outset. Keeping the head in a midline position with the face exposed will provide visualization of the external landmarks and thereby promote a more accurate placement of the ventriculostomy (Fig. 29.1С). A #11 blade is used to make certain that the inner table of bone has been adequately removed and to incise the dura mater. This makes it less likely that the dura mater will be stripped from the ventriculostomy entrance site and decreases the chance of an epidural hematoma. For the past several years we have been using a ventriculostomy tube modified by Holter-Hausner (Fig. 29.1С). This tubing has several advantages over the others presently available. The holes at the distal end of the catheter are covered with another layer of tubing containing longitudinal slits. The overcovering slits make it less likely that the holes will be plugged with brain on catheter entry. In addition, if the ventricles should collapse around the tubing, the holes are protected and have a greater chance of remaining patent. Markers are placed on the tubing at 5-, 10-, and 15-cm intervals, allowing one to know the depth of penetration without having to resort to the use of a ruler. Tabs are bonded to the tubing and easily pass with it when making a subcutaneous tunnel. The tabs are sewn to the scalp to prevent dislodgment. The tubing is passed through the twist drill hole perpendicular to a tangent at the plane of the site of insertion with the tip directed toward the middle of the nose. (Fig. 29.1С). Unless the ventricle is markedly dilated, it is usually entered 5 to 7 cm beneath the skin surface, which is often confirmed by CSF emerging from the end of the tubing. The stylette is removed from the tubing and a curved, flanged trocar is attached to the end of the ventriculostomy tubing and secured with a suture (Fig. 29.2A). Using the catheter entry site, one pushes the trocar under the skin so as to exit 5 or more cm from the point of insertion, thus creating a subcutaneous tunnel through which the tubing is pulled (Fig. 29.2B). Care is taken not to disturb the placement of the tube tip within the ventricular system. The trocar is removed and patency is checked again to make certain that CSF is freely flowing from the catheter. A Luer-lok tip is placed on the tubing along with a 1 -ml tuberculine syringe to prevent

any further loss of CSF (Fig. 29.2С). The Luer-lok tip is secured to the tubing with a suture. At the initial site of entry, the skin is closed with a single horizontal mattress suture. As the tube exists from the subcutaneous tunnel it is secured to the skin in a manner similar to that of a chest tube by wrapping a suture affixed to the skin several times around

the tubing and tying it. The tubing is then brought anteriorly in a gentle loop and the two tabs are sutured to the skin to provide additional security against dislodgment (Fig. 29.2C). The wounds are covered and taped. A tape mesentery is made for the tubing to secure the catheter to the head (Fig. 29.2D). Using pressure (arterial type) tubing the ventriculostomy catheter is connected to a transducer to monitor ICP and/or drain CSF. With the ventriculostomy placed in such a location the patient has good mobility of the head and the head does not rest upon the tubing or dressing. Complications The principle complication is infection. In a pediatric series of over 250 ventriculostomies, our infection rate was between 2 and 3%. The duration of monitoring averaged 7 days. The infection rate did not seem to be related to the duration of placement, leakage of CSF at the catheter site, or whether the ventriculostomy tubing was tunneled, but rather to factors that reduced the patient's resistance to infection. Tunneling has decreased the incidence of CSF leakage, however. There have been no wound infections. Approximately three-quarters of the infections were from staphylococcal organisms (half Staphylococcus epidermidis and the other quarter Staphylococcus aureus). Prophylactic antibiotics have not been used routinely. Another major complication would be a hematoma (epidural, subdural, or intraparenchymal) at the site of the ventriculostomy. This is a rare complication and most often has been associated with a known or unsuspected coagulation disorder. Although possible, we have not encountered either hemiparesis or hypothalamic dysfunction associated with ventriculostomy placement. Seizures originating at the site of insertion are another risk. As these patients often have one or more reasons to experience seizures, it is usually not possible to ascribe seizures to the ventriculostomy insertion site unless electroencephalogram (EEG) tracings show epileptiform activity localized to the appropriate site and not elsewhere. Utilization It is thought that a ventriculostomy can be used in the situation where a mass in or adjacent to the third ventricle has obstructed CSF flow resulting in ventricular enlargement and raised ICP. We prefer the placement of a ventriculostomy to a definitive shunting procedure because a permanent shunt may not be necessary if the obstruction to normal CSF flow can be surgically removed. In addition, it is helpful to have dilated ventricles at the time of the definitive operative procedure as deflating them will allow further access to deep midline structures. A ventriculostomy will also allow monitoring of ICP postoperatively. The ventriculostomy can be placed in the ICU setting without having to bring the patient to the operating room as would be necessary for the placement of a permanent CSF-diverting shunt. The major complication to the placement of a ventriculostomy is infection, the incidence rate of which should be 5% or under (this is the same as for a permanent shunt). Most infections are fairly easily cleared with appropriate antibiotic therapy, especially if the ventriculostomy can be removed, and are usually not associated with any permanent sequelae. Intracranial CSF Diversions Historical and General Considerations Dandy (3) performed the first operative procedure for the treatment of hydrocephalus by fenestrating the anterior third ventricle. He had pre-

viously abandoned fulguration of the choroid plexus and cannulation of the aqueduct of Sylvius because of their high mortality and failure rates. He, as well as others, had also found that sectioning of the corpus callosum or incision of the cortex to allow CSF to escape from an obstructed ventricular system was rarely successful. Dandy did not have much enthusiasm for Torkildsen's procedure of using a tube to divert CSF from the ventricles into the cisterna magna as he thought that the rate of obstruction was higher than with third ventriculostomy. As White and Michelsen (34) pointed out (in the pre-CT and MRI scanning era) the Torkildsen procedure permitted the surgeon to explore the posterior fossa before placing the shunt. An anterior third ventriculostomy could relieve the symptoms associated with hydrocephalus and raised ICP, thereby masking the presence of a surgically resectable mass in the posterior fossa. In 1932 Dandy (4) significantly modified his operative procedure for third ventriculostomy and advocated a temporal rather than subfrontal approach, as the former avoided sacrificing an optic nerve and diminished the chance of CSF collecting on the surface of the brain to produce an "external hydrocephalus." It was not until 1945 that Dandy (5) reported his full series of 92 cases of hydrocephalus treated by an open operative anterior third ventriculostomy. The mortality was 12% with 50% good results; most of the failures were in infants less than 1 year of age. It became evident that success was dependent upon the patency of the subarachnoid drainage pathways outside the ventricular system rather than on the technique of third ventricular fenestration. Dye circulation studies sometimes were useful in determining such patency, but often were not. It also became evident that the chance of reestablishing adequate CSF flow in the subarachnoid pathways in congenital hydrocephalus was considerably less likely than in an acquired form. Scarff (25-27, 29) for decades reminded us of the benefits of third ventriculostomy while an ever-increasing assortment of shunting devices carried the day as the standard treatment for hydrocephalus. Mixter (19) was the first to report an endoscopic means of fenestrating the third ventricle into the surrounding cisterns. This technique has more recently been updated by Vires (32). In the last decade most third ventriculostomies have been accomplished by stereotaxis (9, 24) with virtually no attention being given to an open operative procedure not associated with an attempt to biopsy or remove a mass lesion in or about the third ventricle. Percutaneous stereotaxic methods, which have been even further refined with CT-guided assistance, have markedly reduced the mortality and morbidity associated with third ventriculostomy, but this technique will not increase the rate of success for it is tied to the patency of the CSF drainage pathways beyond the ventricular system, the adequacy of which is difficult to ascertain beforehand. Utilization Third ventriculostomy is very appealing as it poses the chance to reestablish physiological CSF circulation without having to insert foreign bodies and eliminates all of the problems associated with these devices. For a third ventriculostomy to succeed it is essential that the resistance to CSF drainage beyond the ventricular system be within the normal range. As the majority of cases with obstruction to CSF flow in and around the third ventricle are acquired rather than congenital it seems that there would be a much greater likelihood that the subarachnoid pathways for CSF drainage would be adequate and thus a third ventriculostomy would be successful. Unfortunately, there is still a significant

failure rate even when an attempt is made to preselect carefully the patients for this procedure. We are not confident that the clearance studies using radiopharmaceuticals are all that reliable in determining the adequacy of the CSF pathways beyond the ventricular system. We have no experience with either endoscopic or stereotaxic procedures

for third ventriculostomy because it is our practice to insert an extracranial CSF-diverting shunt if it has not been possible to reestablish normal CSF circulation at the time of the third ventricular operative procedure. If removing the obstructing mass has not resulted in the reestablishment of normal CSF circulation, especially if the third ventricle has been fenestrated at its floor, the lamina terminalis, or the suprapineal recess in the process of or in addition to tumor removal, the likelihood of accomplishing normal CSF circulation by a subsequent endoscopic or stereotaxic procedure is remote (Fig. 29.3). In the current era of fairly reliable extracranial CSF shunting systems, it seems that most attempts at third ventriculostomy are made in frustration, after repeated shunt failures in patients with nonmalignant disease. The occasional and relatively isolated enthusiasm for third ventriculostomy is neither catching nor sustained. Even if third ventriculostomy is successful in restoring normal CSF circulation it is necessary to follow these patients routinely with CT or MRI scanning for some will subsequently develop progressive hydrocephalus and require extracranial CSF diversion.

Extracranial CSF Diversion Historical and General Considerations Torkildsen in 1939 (31) produced the first clinically practical method to divert CSF past an obstruction in the ventricular system by inserting a rubber tube between the ventricles and the cisterna magna. This type of shunt did not require a valve but did necessitate that the hydrocephalus be noncommunicating. Irrespective of whether or not the hydrocephalus is noncommunicating, it is not always possible to tell beforehand whether diverting CSF into the SAS from the ventricles will successfully control the hydrocephalus. Although a lateral decubitus position can be used, in the past most patients were placed prone to insert a ventriculocisternal or ventriculocervical (VC) shunt, requiring more set-up time and incurring more anesthetic risk than if the patient were supine. In addition it is necessary to do an occipital craniectomy or a cervical laminectomy to insert the distal end of the shunt into the SAS. Revision of the distal end of the shunt is also cumbersome. For these reasons and with the development of pressure-regulated one-way flow valves, the VC shunt is only occasionally used today. The modern extracranial shunting era for noncommunicating hydrocephalus began in 1949 when Matson introduced diversion of CSF into the ureter (13). The subsequent development of new synthetic materials and pressure-regulated one-way flow valves in the 1950s (20, 22, 23) subsequently made this shunting procedure unnecessary and avoided the sacrifice of a kidney. In addition the ureteral shunts were associated with such severe complications as gram-negative ventriculitis and life-threatening dehydration in infants (14). Steady improvement in shunting hardware associated with a progressive decline in shunt-related complications have made extracranial shunting the procedure of choice when direct surgical removal of an obstruction is not feasible. As noted previously, we prefer the use of a ventriculostomy to initial shunting if it seems that CSF circulation may be reestablished by surgical removal of the obstruction in or about the third ventricle. The ideal shunt would be free of all complications and would drain the right amount of CSF on a minute to minute basis so as to maintain the appropriate normal physiological intraventricular pressure at all times. Such a goal is obviously not obtainable with any implanted mechanical device.

Extracranial CSF-diverting shunts commonly in use today include the ventriculoperitoneal (VP), ventriculopleural (VP1), ventriculoatrial (VA), and lumboperitoneal (LP) varieties. Until 10 to 15 years ago the VA shunt was the type most frequently inserted as it had the highest degree of success. The VA shunts have been largely supplanted by VP shunts as the latter are technically easier to insert or distally revise; have less severe complications, even though the incidence of such for the two types is similar; and, although not a factor in the adult, lengthening of the shunt can be avoided as enough tubing can be placed into the infant's abdominal cavity to allow for growth into adulthood. VA shunts inserted in infants and children necessitate elective lengthening of the distal end on one or more occasions so that the tip is properly positioned in the atrium, a consideration not relevant in the adult patient. The development of newer silicone elastomers and the insertion of enough tubing to allow the abdominal end to migrate freely about the peritoneal cavity have largely eliminated the problem of distal obstruction and allowed the VP shunt to supplant the VA variety. On our service, a pediatric one, less than 1 % of shunt insertions or revisions are of the VA type. The shunt that we use next most frequently after the VP variety is the VP1 type. The technical ease of inserting the distal end into the pleural space is very nearly equivalent to that of insertion into the peritoneal cavity (in fact probably even easier in the obese patient), and the incidence and severity of complications is about the same. This form of shunt, however, cannot be used in infants and children as the absorptive capacity of the pleural space is often not adequate and hydrothorax develops. In the adolescent or adult patient this is rarely a problem. An advantage to using a VP1 shunt is that the distance required to enter the pleural cavity is less than that when the shunt is directed into the abdominal cavity. As obstruction to CSF flow in and about the third ventricle produces a noncommunicating form of hydrocephalus, a LP shunt cannot be used so there need be no further discussion of this type of shunt. Whereas lesions in the posterior third ventricle rarely produce hydrocephalus that obstructs CSF flow from one lateral ventricle to another, this happens with some frequency with lesions situated in the anterior portion of the third ventricle where one or both of the foramina of Monro can be blocked. Removal of an anteriorly located third ventricular mass can often reestablish the communication between the two lateral ventricles, but if questionable it is suggested that the septum pellucidum be fenestrated. If it is necessary to place catheters in both lateral ventricles at the outset to control progressive hydrocephalus, the two systems can be connected, requiring only one distal tube. What frequently happens, however, is that a catheter placed in one lateral ventricle initially drains both lateral ventricles satisfactorily. Then with subsequent tumor growth or scarring, the two ventricular systems become isolated. The patient may be symptomatic with evidence of raised ICP or the problem may be detected only on routine follow-up CT or MRI studies. At this point it becomes necessary to place a second ventricular catheter in the nondraining lateral ventricle. The choice in this situation would be either to connect the second ventricular catheter to the existing shunt or to insert an entirely separate new shunt. The technical ease of putting in a completely new second shunt is probably equivalent to that of connecting it into the existing shunt in the infant or child, but most likely not for the adult. The advantage of two completely separate shunts is that they facilitate the diagnosis and treatment of shunt malfunction and sometimes infection.

If the proper cephalic skin incision site is not outlined before draping, orientation can be lost and the entry site can be placed over the superior sagittal sinus. The superior sagittal sinus often can be just to the right of midline in the posterior parietal area and may account for the tendency to place the catheter entry site too far laterally. With growth of the skull in infants it has been observed that the entry site will migrate laterally, resulting in the ventricular catheter having a definite curve whereas initially it was straight. It is important that the skin incisions be placed so that they are adjacent to and not directly over the shunt hardware, particularly in infants where the scalp is thin, as this will lessen the incidence of wound breakdown and resultant shunt infection. Thought must be given to the location of the incisions and various shunt components in regard to possible future shunt revisions. Operative Corridor: Anatomy and Physiological Risks Whereas we prefer a frontal location to insert a catheter for a ventriculostomy we advocate a posterior parietal location for the placement of a permanent shunt. Disadvantages of the frontal approach include more extensive hair shaving and scalp preparation and an additional skin incision needed in the posterior parietal region as it is very difficult in the infant and almost impossible in the adolescent or adult to tunnel the catheter from the frontal area to the distal entry site. Minimizing the number of skin incisions should lead to a lower shunt infection rate. If the tubing is tunneled too close to the ear the wearing of glasses becomes a problem. The rationale for placing the tip of the ventricular catheter in the frontal horn, anterior to the foramen of Monro, is that the absence of choroid plexus tissue in this region reduces the chance that the ventricular catheter will be obstructed. The advantage of the frontal approach is a shorter distance into the ventricle and less chance that the catheter could be passed through the hemisphere without reaching the ventricle, a problem when the ventricles are not significantly enlarged. We have not found this to be a particular problem with the posterior parietal approach if the projection of the frontal horn has been properly marked to provide good intraoperative orientation showing where to pass the ventricular catheter. The expanding use of intraoperative ultrasound will also be of benefit in canulating the lateral ventricle at the desired site of entry. The lateral ventricle is most easily entered posteriorly at the site of the occipital horn projected on to the skin surface. Many published drawings show the ventricular catheter being inserted anterior to and often inferior to this location. As noted in the ventriculostomy section, a risk to a frontal approach includes passing the catheter through the lateral ventricles into the basal ganglia or through the floor of the third ventricle into the hypothalamic structures. The chance of producing neurological sequelae with inadvertent misdirection of the ventricular catheter from the frontal approach seems to be roughly equivalent when the entry site is from the occipital horn. A subxiphoid midline abdominal incision is preferred to a subcostal incision as it is easier to reach the peritoneum without having to traverse muscle layers. If the midline is not available a subcostal incision is used. If it becomes necessary to reposition the catheter later in the abdominal cavity it is easier to use a new site of entry rather than trying to trace the established tract into the peritoneal cavity. A new entry site will also reduce the risk of injury to a viscus. Determining the site of entry into the pleural cavity must take into

consideration the location of the neurovascular bundle, which lies under the rib. Staying on the superior surface of the lower rib of the intercostal space chosen avoids this possible problem. It is also necessary to appreciate the location of the internal mammary artery and vein running a centimeter or two lateral to the border of the sternum so as to avoid these vessels. The only other concern is not to enter the chest so low that the diaphragm is encountered. For cosmesis, an incision just below the breast is often used. If cosmesis is of no concern or the patient is obese, the second-third interspace in the midclavicular line is technically easier. The quantity of air entering the pleural cavity has not been a problem. It is not necessary to place a chest tube. For an atrial shunt, a neck incision is made midway between the mandible and the clavicle. This allows ready access to the carotid sheath and the common facial vein. The incision is made parallel to the creases in the neck to minimize scarring. Structures at risk in the carotid sheath are the common carotid artery, internal jugular vein, and vagus nerve. If the common facial vein is available, it is simpler to introduce the atrial catheter into this vein than directly into the internal jugular vein. With atrial shunt placement it is necessary to ensure that no significant air embolus occurs. Use of a soft pliable catheter will avoid the possibility of perforating the vascular system and subsequent complications that this might cause. Structural definitions As noted in the ventriculostomy section the diagnostic studies are either CT scanning with and without contrast or MRI. If a question exists as to whether adequate communication is present between the two lateral ventricles metrizamide can be placed into the ventricular system. Operative Technique Ventriculoperitoneal Shunt The patient is placed in the upper right corner of the table with the head turned to the left. This allows the anesthesiologist complete access to the airway (Fig. 29.4A). In infants, an esophageal stethoscope is used to avoid placing a stethoscope over the chest in the region of the operative field. A doughnut made of any soft material is placed under the head. Padding is also placed beneath the neck to make the angle between the head and chest as flat as possible, thereby making it easier to pass the tubing subcutaneously from the head to the abdomen. It is helpful to place a rubber button, taken from a medicine bottle, on the patient's forehead in a position representing the anterior surface projection of the right frontal horn (Fig. 29.6B). The patient's skin is shaved and prepared with soap, alcohol, and a degreasing agent (such as Freon) and then is well prepared with povidone-iodine (Betadine) solution. A thorough skin cleansing should markedly reduce the skin bacterial count and presumably lower the incidence of shunt infections as it is thought that skin contamination accounts for a significant percentage of shunt infections. Excess povidone-iodine is removed with more alcohol to ensure better adherence of a barrier drape to the skin. The skin is sprayed with an adhesive (such as Vidrape) to ensure better adherence of the barrier drape. At this step in the procedure, the incisions are outlined with a skin marker to ensure proper orientation before draping. Cloth towels are placed at some distance from the incision sites to prevent the barrier drape from sticking to the endotracheal tube, i.v.s, etc. A barrier drape impregnated with povidone-iodine (Ioban) is placed over the exposed skin.

Care is taken to ensure that the barrier drape adheres well to the skin, particularly at the incision sites. Paper drapes, impenetrable to liquid, are placed closer to the proposed incision sites. A second layer of paper drapes are then placed around the patient and operating table. The skin is injected with 0.25% lidocaine and 1:400,000 epinephrine solution to decrease bleeding. A small curvilinear incision is made in the right posterior parietal area in the region of the projection of the occipital horn to the skin surface (Fig. 29.4). The scalp is reflected. A cruciate incision is made in the periosteum, which is elevated enough to allow a perforator to make a hole through the underlying bone (Fig. 29.5A). The wound is next covered with a sponge moistened with either Ringer's lactate containing bacitracin or povidone-iodine and is covered with a towel to prevent contamination. Attention is then directed to the abdomen where a subxiphoid midline incision is made through the abdominal wall until the peritoneum is encountered and two 4-0 absorbable sutures are placed into the peritoneum without opening this membrane. This eliminates hunting for the opening subsequently. A shunt tube passer is then tunneled from the cephalic to the abdominal incision without making any additional incisions along the pathway and the tubing is implanted subcutaneously (Fig. 29.5B and C). The peritoneum is opened and the remainder of the tubing is placed into the abdominal cavity (Fig. 29.5D). We are using abdominal tubing 120 cm long even in newborn infants to eliminate the subsequent need to lengthen the shunt. The tubing is flushed with Ringer's lactate containing bacitracin. The injection of this solution into the tubing lumen establishes that the distal end is patent and that there is no resistance to flow. If a valve is to be used, it is inserted onto the proximal end of the tubing and affixed with nonabsorbable 3-0 sutures (Fig. 29.5E). Pinpoint cautery is used to make a hole through the dura mater exposed by the perforator and the underlying arachnoid and pia (Fig. 29.6A). If the cortex is thick the dural hole size is not critical; it should be big enough to allow the ventricular catheter to be introduced easily without resistance. If the cortex is greatly thinned care is taken to make certain that the dural hole is no bigger than the diameter of the ventricular catheter to minimize the possibility of CSF leaking around the entry site resulting in a subcutaneous collection of fluid. Cortical bleeding is occasionally encountered but it can be readily controlled with bipolar cautery or direct pressure. From the CT or MRI scan and by measuring the skull the appropriate length of ventricular tubing needed to reach the frontal horn can be estimated (Fig. 29.6B). In the older child and the adult, a 12cm-long ventricular catheter is usually appropriate. After the ventricle is entered, CSF is obtained and sent for analysis for cells, sugar, and protein and for culture (Fig. 29.6C). This establishes the nature of the CSF for future reference. If indicated, CSF can also be obtained for cytology or biological marker examination. Care is taken not to drain an excessive amount of CSF, especially if the ventricles are large and the cortical mantle is thin, as doing so increases the risk of subdural hematoma formation. If desired, an antibiotic may be injected into the ventricles. The side arm of the ventricular catheter reservoir is then attached to the proximal end of the valve and secured with a 3-0 nonabsorbable suture (Fig. 29.7). Attaching the sidearm of the reservoir to the valve or peritoneal tubing immediately after placing the ventricular catheter diminishes the CSF loss. If a separate reservoir is used rather than a one-piece ventricular catheter reservoir unit, it is better to attach the reservoir to the proximal end of the valve or peritoneal tubing before inserting the

ventricular catheter. Once again, this minimizes the loss of CSF from the ventricles. The wounds are scrubbed with povidone-iodine solution followed by irrigation with Ringer's lactate containing bacitracin. The incisions are then closed using either interrupted 3-0 or 4-0 absorbable sutures. The skin edges are approximated with tape such as Steri-strips. The technique described can be used in the infant, child, or adult. Skin stitches are not used; the wounds heal better and there is no need to remove the sutures, particularly a factor in the uncooperative patient. If there is any question about the integrity of the closure in regard to possible CSF leakage, a second layer of sutures placed in the skin is advisable. This would also be true if difficulty exists in controlling scalp bleeding at the termination of the procedure. Ventriculopleural Shunt As with a VP shunt, the cephalic incision is made first and the burr hole is placed. The wound is covered and attention is directed to the chest wall. For cosmesis, an incision approximately 3 cm long is made just below the breast and the midclavicular line (Fig. 29.4A). If the patient is obese or the presence of an incision higher on the chest wall is not objectionable the site may be moved up to the second-third intercostal space. The subcutaneous tissue, deep fascia, and pectoralis muscles are divided (Fig. 29.8A). It helps to wear a headlight to see more easily into the depths of the incision. The external and internal intercostal muscles are divided at the superior aspect of the lower of the two ribs of the intercostal space chosen. A self-retaining retractor placed between the two ribs opens the intercostal space even further (Fig. 29.6B). The parietal pleura is next visualized with the lung beneath moving with respiration. The pleura is not opened at this point. The tubing is passed from the cephalic to the chest incision, with no additional incisions being necessary between. Once the tubing is in its subcutaneous location the pleura is opened just enough to admit the tubing (Fig. 29.8C). Twenty to 40 cm of tubing is inserted into the pleural cavity to provide redundancy and to ensure that the tubing will continually migrate in the pleural space. If the pleural opening is small it need not be sutured, but if larger it can be closed about the tubing with a 4-0 absorbable suture. Before the pleura is closed the anesthesiologist is asked to expand the lung to expel as much air as possible. This step can be repeated at closure of the first muscle layer. The remainder of this incision is closed like the abdominal wall. There is no need to place a chest tube. A postoperative chest x-ray film is routinely obtained. Ventriculoatrial Shunt The cranial portion is handled in the same manner as with a VP or VP1 shunt. An incision following a skin crease is made midway between the mandible and the clavicle (Fig. 29.4A). The platysma muscle is divided and the anterior border of the sternocleidomastoid muscle is identified, dissected, and retracted posteriorly until the carotid sheath is located (Fig. 29.9A). The internal jugular vein is then separated from the common carotid artery and the vagus nerve. With further exposure the common facial vein is isolated for at least 1 cm before its entry into the internal jugular vein (Fig. 29.9B). If the common facial vein is not suitable as an entry site, the tubing can be placed directly into the internal jugular vein after placing a purse string suture in the vessel wall. The common facial vein is then ligated and put on tension so as to facilitate entry of the tubing. For control, temporary vascular clamps or silicone elastomer

vascular loops are placed about the internal jugular vein proximal and distal to the entry site of the common facial vein (Fig. 29.9C). A stick tie through the wall of the common facial vein is made but not tied proximal to the vein's entry into the internal jugular vein. The length of tubing needed to place the tip of the catheter into the right atrium is estimated and a suture or mark is placed on the tubing for future reference (Fig. 29.9D). The tip of the tubing is cut on a slight angle to facilitate its entry. The tubing is filled with normal saline and clamped so as not to allow air

into the atrium. The common facial vein is opened enough to admit the tubing, which is advanced to what is thought to be the desired location. As the necessary length of tubing is only an estimate and it is possible that the tubing can go into other vascular channels than the one desired, it is necessary to confirm the location of the catheter tip. A monometer can be placed on the tubing and the pressure measured. If the pressure is low then the tip is most likely in the right atrium. If the pressure is over 10 cm H2O and pulsatile then the end most likely is in the right ventricle, the tip having gone through the tricuspid valve. By pulling the tubing back, a pressure drop should be noted, indicating that the end now resides in the right atrium. In addition there is usually a change in the P-wave of the electrocardiogram when the tubing tip resides in the right atrium. To be absolutely certain that the tip is in the position desired, it is suggested that intraoperative fluoroscopy be used to visualize the tubing by the injection of a contrast agent. In an adult the midportion of the atrium is a satisfactory location for the end of the tubing; in an infant or child it is best to position the end as low as possible within the atrium to allow for future growth of the child. The mark or suture on the tubing is noted in relation to its entry into the common facial vein so that final proper positioning is assured and to indicate inadvertent movement of the tubing during subsequent steps in the operative procedure (Fig. 29.9E). The valve may be attached before or after the tubing is tunneled to the cephalic incision (Fig. 29.9F). This would not be necessary if a distal slit valve type of catheter is used. The stick tie through the wall of the common facial vein before its entry into the internal jugular vein is now securely tied. Previously it need only have been tied loosely if bleeding had occurred. The remainder of the operative procedure is as described for a VP shunt. A concern with a VA shunt is to prevent a significant amount of air from entering the system. This is rarely a problem. Biventricular Shunt Biventricular shunts can be inserted in either frontal or parietal locations (Figs. 29.10 and 29.11). If the frontal region is chosen the patient is positioned as described previously. If a parietal approach is used the main problem is one of positioning. The patient is placed with the head face down to assure good access to both posterior parietal regions. The torso is rotated so as to expose the right chest wall or the right flank. In this position it is easy to make an incision in either the lateral chest wall for a VP1 shunt or the right lower quadrant for a VP shunt. It is not possible to insert a VA shunt in this position. Scalp incisions are made as before as are chest or abdominal incisions. There are three options at this point. The two shunts can be treated independently with two tubes tunneled to the pleural or abdominal cavity, the ventricular catheter can be connected by a Y or T connector proximal to a valve (if a proximal valve is used), or a valve can be placed just after the reservoir on each side and the two shunts can be joined at a more distal location. In the infant or child it is probably just as easy to make two entirely separate shunting systems as it will simplify future revisions. In the adult a single distal tube would often be technically easier. It is also necessary to make an incision in the neck as it is not possible to tunnel subcutaneously between the cephalic and distal incisions. When designing the system and placing the incisions it is always best to consider that subsequent revision of the shunt might be necessary. Complications As the best way to avoid shunt complications is not to insert a shunt, it is necessary to establish beyond a reasonable doubt that the shunt is

needed. Shunt complications can be divided into those that are common to all types of shunts and those that are unique to a particular shunt type. All shunts are subject to obstruction, disconnection, and infection. The frequency and location of obstruction or disconnection depend to a fair degree on the type of shunt hardware utilized. In general, the most frequent malfunction is occlusion somewhere within the shunt lumen. In the pediatric age group, particularly those under 1 year of age, the most frequent site of obstruction is in the ventricular catheter by choroid plexus or ependymal or glial tissue, which can grow into the lumen. Another source of malfunction is disconnection, which can occur at any point in the system but most commonly where the various components are joined. Pressureregulated valves can block, drain CSF at higher or lower than the intended pressure, and rarely allow retrograde flow, a concern only in a vascular shunt. Although considerable apprehension has been expressed regarding the protein content of CSF and the likelihood of valve obstruction, no definite relationship has been established (17). It is our experience that, in the pediatric age group, the increased protein of the CSF is usually associated with more frequent ventricular catheter obstructions. The use of radiopaque materials for the entire length of the shunt is advocated, making it possible to determine the continuity of the shunting system on plain x-ray films. It is also recommended that a reservoir be placed in the system to allow access to the CSF. By tapping the reservoir, pressure measurements can be made, function of the system can be ascertained, and CSF can be obtained for examination as needed. If, after tapping the reservoir, the adequacy of function of the shunt is still in doubt, imaging techniques that use contrast agents or radiopharmaceuticals can determine patency. Each type of shunting system has its own peculiarities that determine how best to establish the adequacy of its function. Most shunting systems contain a pumping mechanism in either the valve or the reservoir that allegedly tests whether the shunt is functioning properly. The correlation between response to pumping and proper functioning of the shunt may be at variance, with some shunts that do not pump normally functioning satisfactorily while others that pump normally are malfunctioning. If a shunt malfunction is suspected, it is strongly advocated that response to pumping not be the sole criterion used to make any definite conclusions in this regard. The use of implanted foreign materials always presents the risk of infection. Infection prevention obviously is preferable to treatment. The vast majority of shunt infections occur at the time of insertion, revision, or improper tapping. The use of meticulous aseptic surgical technique and possibly prophylactic antibiotics has steadily reduced the risk of infection so that a per case infection rate of 5% or less should be the standard, with some centers reporting infection rates in the 1 to 2% range (30, 33). A shunt infection may manifest itself by swelling and redness over a portion or all of the shunt tract and such generalized symptoms as peritonitis (with a VP shunt) or septicemia (with a VA shunt). Skin breakdown over the hardware and resultant shunt infection are problems that can be seen in the infant but are rarely a consideration in the adult.

Proper incision and hardware placement and the use of low profile and pliable materials should virtually eliminate this form of complication. Infection can be solely confined to the CSF in the ventricles and shunt, giving no external evidence other than a shunt malfunction. Initial infection rates for both VP and VA shunts are similar; however, the presence of a catheter within the atrium allows subsequent colonization during episodes of bacteremia (11). A VA shunt infection may lead to septicemia with subsequent renal damage, lung damage, endocarditis, and septic emboli, whereas a VP shunt infection most commonly leads only to obstruction of the distal end of the tubing. A complication of a chronic low grade shunt infection that is limited to VA shunts is that of nephritis,

which develops from deposition of antibody-antigen and complement immune complexes in the glomeruli. The most reliable way to confirm a shunt infection is to obtain multiple CSF samples from the shunting system (28). Approximately one-half of the infecting organisms are S. epidermidis and one-quarter are S. aureus, with the remaining onequarter being a wide variety of pathogens (7, 8, 28). The management of shunt infections is still subject to considerable debate. All advocate the use of appropriate intraventricular and systemic antibiotics. Some authorities attempt to clear the infection without replacement of the shunting system, others favor immediate replacement, whereas others advocate delayed replacement (7, 10, 16, 21). Our preference is complete removal of the infected system with delayed replacement as this has the highest chance of success with the lowest morbidity and the shortest hospital stay (10). Delayed replacement usually necessitates drainage of CSF intermittently via a reservoir or continuously by the establishment of an external ventricular drainage system. Complications unique to VP shunts are ascites, pseudocyst formation, perforation of a viscus or the abdominal wall, intestinal obstruction, spread of infection or neoplasm from the ventricles to the abdominal cavity, and a higher incidence of inguinal and umbilical hernias in infants (6). The use of a coiled spring inside the peritoneal catheter to prevent its kinking is responsible for the majority of abdominal complications because the stiffness of the catheter tip results in its penetration of surrounding structures. The length of tubing in the peritoneal cavity does not seem to be a factor in subsequent complications. We have used an extended length of peritoneal tubing for over 10 years even in the most premature neonates without any complications due to the length. Use of extended length tubing eliminates the need to lengthen the distal end of the shunt to accommodate for growth. Complications unique to VP1 shunts are primarily those of fluid accumulation within the pleural cavity. This is particularly a problem in the infant and child and is considerably less a problem in the adolescent or adult. Hydrothorax is treated by moving the distal catheter to a site outside of the thoracic cavity. Complications unique to vascular shunts involve the heart, lungs, and vasculature and include vena cava obstruction, mural thrombosis, bacterial endocarditis, cardiac arrhythmias, cardiac tamponade (secondary to perforation of the heart wall), embolization of the distal catheter into a pulmonary artery, and chronic pulmonary thromboembolization that could result in pulmonary hypertension or even cor pulmonale (17). The development of a significant subdural fluid collection after ventricular shunting relates directly to the age of the patient and the size of the ventricles; the older the patient, the bigger the ventricles, the more likely this complication, not only at the time of shunt insertion but ever thereafter. Even a seemingly minor head injury may produce a significant subdural fluid accumulation. The use of a higher pressure valve helps to reduce this problem. If the subdural collection is asymptomatic, treatment is not necessarily indicated, but if a progressive increase in volume and associated clinical symptoms occur the subdural fluid must be drained, usually by the placement of a subdural peritoneal shunt without a valve to provide a pressure gradient between the two shunting systems. The incidence of seizures in patients with shunted hydrocephalus is higher than can be attributed to hydrocephalus alone (1, 17). This is substantiated by the finding of EEG abnormalities local to the site of shunt insertion (12). Whether prophylactic anticonvulsant treatment is needed and for how long is not clear and is probably best decided on a case by case basis. We routinely do not advocate such for our patients.

Utilization A CSF-diverting shunt must be inserted when there is evidence of progressive hydrocephalus after failure to reestablish CSF circulation by removal of a third ventricular obstruction. A second attempt at intracranial CSF diversion is rarely warranted.

References 1. Copeland GP, Foy PM, Shaw MD: The incidence of epilepsy after ventricular shunting operations. Surg Neurol 17:279-281, 1982. 2. Cushing H: The Third Circulation. London, Oxford University Press, 1926. 3. Dandy WE: An operative procedure for hydrocephalus. Johns Hopkins Hosp Bull 33:189-190, 1922. 4. Dandy WE: Surgery of the brain. In Lewis D (ed): Practice of Surgery. Hagerstown, MD, WF Prior Co, Inc, 1932. 5. Dandy WE: Diagnosis and treatment of strictures of aqueduct of Sylvius (causing hydrocephalus). Arch Surg 51:1-14, 1945. 6. Davidson RI: Peritoneal bypass in the treatment of hydrocephalus: Historical review of abdominal complications. J Neurol Neurosurg Psychiatry 39:640646, 1976. 7. Forward KR, Fewer HD, Stiver HG: Cerebrospinal fluid shunt infections: A review of 35 infections in 32 patients. J Neurosurg 59:389-394, 1983. 8. George R, Leibrock L, Epstein M: Long-term analysis of Cerebrospinal fluid shunt infections. J Neurosurg 51:804-811, 1979. 9. Hoffman HJ, Harwood-Nash D, Gilday DL: Percutaneous third ventriculos tomy in the management of noncommunicating hydrocephalus. Neurosurgery 7:313-321, 1980. 10. James HE, Walsh JW, Wilson HD, et al: Prospective randomized study of therapy in Cerebrospinal fluid shunt infection. Neurosurgery 7:459-463, 1980. 11. Keucher TR, Mealey J Jr: Long-term results after ventriculo-atrial and ventriculo-peritoneal shunting for infantile hydrocephalus. J Neurosurg 50:197-186, 1979. 12. Laws ER Jr, Niedermeyer E: EEG findings in hydrocephalic patients with shunt procedures. Electroencephalogr Clin Neurophysiol 29:325, 1970. 13. Matson DD: A new operation for the treatment of communicating hydroceph alus: Report of a case secondary to generalized meningitis. J Neurosurg 6:238-247, 1949. 14. Matson DD: Neurosurgery of Infancy and Childhood, ed 2. Springfield, Charles С Thomas, 1969. 15. McComb JG, Hyman S, Weiss MH: Cerebrospinal fluid drainage following acute obstruction of the fourth ventricle in the rabbit. Concepts Pediatr Neurosurg 4:90-101, 1983. 16. McLaurin RL: Treatment of infected ventricular shunts. Child Brain 1:306310, 1975. 17. McLaurin RL: Shunt complications. In Section of Pediatric Neurosurgery of the American Association of Neurological Surgeons (eds): Pediatric Neuro surgery: Surgery of the Developing Nervous System. New York, Grune and Stratton, 1982, pp 243-253. 18. Miller CF II, White RJ, Roski RA: Spontaneous ventriculo-cisternostomy. Surg Neurol 11:63-66, 1979. 19. Mixter WJ: Ventriculoscopy and puncture of the floor of the third ventricle. Boston Med Surg J 188:277-278, 1923. 20. Nulsen FE, Spitz EB: Treatment of hydrocephalus by direct shunt from ventricle to jugular vein. Surg Forum 2:399-403, 1952. 21. O'Brien M, Parent A, Davis B: Management of ventricular shunt infections. Childs Brain 5:304-309, 1979. 22. Pudenz RH, Russell FE, Hurd AH, et al: Ventriculo-auriculostomy: A tech nique for shunting Cerebrospinal fluid into the right auricle. Preliminary report. J Neurosurg 14:171-179, 1957. 23. Pudenz RH: The surgical treatment of hydrocephalus—an historical review. Surg Neurol 15:15-26, 1981. 24. Sayers MP, Kosnick EJ: Percutaneous third ventriculostomy: Experience and technique. Childs Brain 2:24-30, 1976. 25. Scarff JE: Treatment of obstructive hydrocephalus by puncture of the lamina terminalis and floor of the third ventricle. J Neurosurg 8:204-213, 1951.

26. Scarff JE: Treatment of hydrocephalus: An historical and critical review of methods and results. J Neurol Neurosurg Psychiatry 26:1-26, 1963. 27. Scarff JE: Third ventriculostomy by puncture of the lamina terminalis and the floor of third ventricle. J Neurosurg 24:935-943, 1966. 28. Schoenbaum SC, Gardner P, Shilito J Jr: Infections of Cerebrospinal fluid shunts: Epidemiology, clinical manifestations and therapy. J Infect Dis 131:543-552, 1975. 29. Stookey B, Scarff J: Occlusion of the aqueduct of Sylvius by neoplastic and non-neoplastic processes with a rational surgical treatment for relief of the resultant obstructive hydrocephalus. Bull Neurol Inst NY 5:348-377, 1936. 30. Tabars Z, Forrest D: Colonisation of CSF shunts: Preventive measures. Kinderchir 37:156-157, 1982. 31. Torkildsen A: A new palliative operation in cases of inoperable occlusion of sylvian aqueduct. Acta Chir Scand 82:117-124, 1939. 32. Vries JK: An endoscopic technique for third ventriculostomy. Surg Neurol 9:165-168, 1978. 33. Welch K: Residual shunt infection in a program aimed at its prevention. Kinderchir 28:374-377, 1979. 34. White JC, Michelsen JJ: Treatment of obstructive hydrocephalus in adults. Surg Gynecol Obstet 74:99-109, 1942.

30 Considerations and Techniques in the Pediatric Age Group Harold J. Hoffman, M.D., F.R.C.S.(C)

Tumors in and around the third ventricle are common in childhood. During the years 1950 to 1984 a total of 1367 children with brain tumors were seen at the Hospital for Sick Children. Three hundred ninety-three of these children had third ventricular tumors (28.7%); 72 had optic nerve gliomas, 122 had suprasellar tumors, 66 had pineal region tumors, 17 had tumors within the confines of the third ventricle, 55 had hypothalamic tumors, and 61 had thalamic tumors. The majority of these tumors were benign, 110 being low grade astrocytomas and 95 being craniopharyngiomas.

Optic Nerve Gliomas Optic pathway gliomas are relatively common in children. They constituted 3.6% of all brain tumors in Matson's series (11) and made up 6% of our series of brain tumors at the Hospital for Sick Children (4). Davies initially described the close relationship between optic nerve tumors and von Recklinghausen's disease (3). Thirty percent of patients with optic nerve gliomas at the Hospital for Sick Children have evidence of von Recklinghausen's disease. Optic gliomas have a highly unpredictable course; in one patient the tumor will spread and quickly become fatal, but in another it can remain quiescent for many years (Fig. 30.1). As a result, opinion concerning the management and treatment of this tumor has been radically divided. Some clinicians advocate surgical intervention, others rely on radiotherapy, and still others refuse to undertake any active form of treatment, believing that these tumors are not true neoplasms, but hamartomas. The tumors typically begin within the optic nerve in the optic canal and from this point usually extend forward into the orbit as well as backward toward the chiasm. They can infiltrate adjacent brain. Of the 72 optic pathway tumors seen during the past 25 years at the Hospital for Sick Children, 24 lay anterior to the optic chiasm and 48 involved chiasm and optic tracts. The tumors are usually Grade 1 to 2 astrocyto-

mas, but they may occasionally show some signs of aggressivity and have been graded as high as Grade 3. The vast majority of patients with optic nerve gliomas present before the age of 6. The most common presenting symptom is visual loss, often in conjunction with optic atrophy. Proptosis is common when the tumor extends forward into the orbit. At first the globe protrudes forward and then it is displaced laterally (Fig. 30.2). Visual field defects indicate involvement of the optic tract and optic chiasm. Papilledema is usually produced by tumors that encroach posteriorly on the third ventricle and produce hydrocephalus (Fig. 30.3). As the tumor extends beyond the confines of the optic system, it can involve the internal capsule and produce hemiparesis (Fig. 30.4A).

The diencephalic syndrome consisting of emaciation in association with an alert hyperkinetic infant can be seen in patients with an optic nerve glioma. Optic nerve gliomas that involve the optic tract often produce extensive distension of the basal cisterns and Cerebrospinal fluid (CSF) subarachnoid pathways (Figs. 30.4B and 30.5). The resulting hydrocephalus is frequently extraventricular and a computerized tomographic (CT) scan may show only minimal ventricular enlargement. Rarely, the distended

CSF pathways over the surface of the brain can rupture, producing massive subdural effusions (Fig. ЗОЛА). Ascites after diversionary ventriculoperitoneal shunting is a rare complication seen in patients with optic nerve gliomas. Kronlein described the operative technique for resection of the intraorbial portion of the optic nerve glioma in 1888 (9). However, because these tumors involve the intracranial portion of the optic nerve, a two-stage procedure emerged: the neurosurgeon would resect the intracranial portion of the optic nerve tumor and disengage it from the optic canal through a craniotomy approach and the ophthalmologist would remove the orbital tumor through a Kronlein procedure. An entirely transcranial route developed by Jackson (8) and Housepian (7) allowed access to intracranial as well as orbital portions of the tumor in one stage. Optic nerve gliomas are almost always low grade astrocytomas. There is no chance of tumor recurrence if they can be totally resected. Unfortunately, total resection is only possible when the tumor is present in one optic nerve and is well in front of the optic chiasm. When the tumor involves the optic chiasm, optic tracts, optic radiation, and geniculate bodies, total resection is not possible. However, a partial or sometimes subtotal removal of such posterior optic pathway tumors can be carried out. When vision is compromised by an optic nerve glioma that can be only partially removed, radiotherapy is indicated. Low grade tumors can respond to radiation and may even disappear completely with such a course of radiotherapy (Fig. 30.4). If vision is not significantly compromised, radiotherapy can be delayed and the patient can be followed clinically and on CT scan to see whether the tumor is aggressive and will require radiation. In patients in whom an anteriorly placed tumor is located in one optic nerve, there is frequently intraorbital as well as intracranial tumor. An attempt should be made to remove the tumor en bloc. We position the patient face up with neck extended and head in the pin fixation headrest. Brain relaxation is achieved by the intravenous instillation of mannitol in a dose of 2 g/kg and hyperventilation of the patient. A bicoronal incision is used and a front craniotomy is carried out on

the side of the optic nerve involved. A small frontal flap is turned. This extends from the midline medially to the outer edge of the supraorbital margin laterally and from supraorbital margin to midway between nasion and coronal suture posteriorly. The dura mater is opened just above the supraorbital margin and the frontal lobe is retracted, exposing optic nerve and chiasm. Frontopolar veins in the way can be coagulated and divided. The olfactory tract is coagulated and divided to prevent avulsion of the olfactory bulb. The posterior extent of the tumor is visualized. If the tumor does not involve the chiasm, one should then prepare for total excision of the tumor. If the tumor grossly appears to involve chiasm and vision is severely compromised or if significant proptosis is present, resection of the optic nerve is still indicated. The dura mater in the floor of the anterior fossa is divided back to the level of the optic foramen. With a retractor between the orbital fascia and the roof of the orbit, a high speed drill is used to open the orbital roof, which is then removed with rongeurs and punches back to the optic canal. The optic canal is then unroofed using high speed burrs and 1mm punches. This done, the dura of the optic nerve is opened and the optic nerve is followed into the orbit. The anulus of Zinn is divided, and traction sutures are left to hold the two cut edges in place for eventual reconstitution. The ocular muscles are retracted and the nerve is followed directly to the globe. The nerve is then amputated behind the globe, with care being taken not to penetrate the globe. Such penetration can lead to irreversible globe damage. The entire optic nerve is then lifted out of the orbit and the optic canal and amputated just in front of the chiasm. Care should be taken to label its proximal and distal ends for histological examination to confirm that there has been no histological spread of tumor beyond the posterior line of resection. The orbital roof is repaired with tantalum mesh. In patients with posteriorly placed tumors that do not go forward into

the orbit but extend back along the optic tract and involve the chiasm, it is possible to remove large portions of tumor while preserving vision. The ultrasonic aspirator is an ideal tool for removing such tumors. The tumor can be resected partially under microscopic vision. After such an operation, the patient can be followed clinically and with CT scanning (Fig. 30.6). If there is no further evidence of tumor growth and vision is good, a watchful wait is indicated. However, if the tumor does show evidence of growth or if vision is bad or deteriorating, radiotherapy is indicated (Fig. 30.4). Craniopharyngiomas Craniopharyngiomas are benign suprasellar tumors that are seen predominantly in childhood. Although there is great controversy on the definitive management of these tumors, the overwhelming majority of craniopharyngiomas can be totally resected to cure the patient (5, 1214). Craniopharyngiomas can present as small tumors that enlarge the sella and produce endocrine deficiency and headache (Fig. 30.7). These tumors do not displace vessels on angiography. Although such tumors can be resected transsphenoidally, I prefer a subfrontal approach. The patient's head is positioned in anatomical position, with the nose directly upward and the head extended (Fig. 30.8). The pin fixation headrest should be used in patients over 2 years of age. It is essential that the brain be relaxed when the frontal lobe is elevated. Mannitol in a dose of 2 g/kg is given at the start of the operation and the patient is hyperventilated. I use a small right frontal bone flap, cut low just above the supraorbital margin and extending to the midline, with burr holes placed over the sagittal sinus and the medial cut made just to the left of the midline. The lateral burr hole is made at the pterion. The dura mater is opened just above the supraorbital margin and the dural opening is extended trans-

versely to the sagittal sinus medially and to the pterion laterally. The frontal lobe is elevated. Any frontopolar veins in the way are divided, and the olfactory nerve is divided just behind the olfactory bulb. Self-retaining brain retractors are used to elevate the frontal lobe, the arachnoid is incised over the right optic nerve, and the chiasmatic cistern is opened and emptied of CSF. Using the operating microscope, one exposes the craniopharyngioma and incises its capsule. Whether the contents are liquid or solid, they are then emptied with the ultrasonic aspirator. The collapsed tumor can then be separated from the optic nerves and internal carotid arteries and pulled down from the hypothalamus. Frequently, the pituitary stalk ends in the tumor and must be divided at this point. The tumor can then be shelled out of the sella. If the diaphragma sellae is deficient, the tumor invades the pituitary gland and, as one pulls on the neoplasm, fingers of tumor will come out along with some adherent pituitary gland. The larger craniopharyngiomas extend either forward between the two optic nerves (prechiasmatic) (Fig. 30.9) or backward into the third ventricle (retrochiasmatic) (Fig. 30.10). The prechiasmatic tumors typically compress one optic nerve more than the other and the patients usually present with severe visual loss in one eye. Endocrine problems are commonly present in this group of patients. Angiography will show elevation of the A1 segment of the anterior cerebral arteries with no displacement of the basilar artery. In much the same way as for sellar tumors, prechiasmatic tumors are approached through a small right frontal bone flap that brings one down on the optic nerves, the chiasm, and the tumor situated between the two optic nerves. The tumor capsule is coagulated and incised and the tumor contents are emptied with the ultrasonic aspirator. This procedure collapses the capsule and allows it to be pulled forward and dissected free from the optic nerves, internal carotid arteries, posterior communicating arteries, third cranial nerves, and hypothalamus. The tumor is situated above the diaphragma sellae unless the diaphragma is deficient in which

case the tumor invades the pituitary gland. Occasionally the pituitary stalk can be saved but it usually comes out with the tumor when the latter is pulled free from surrounding structures. The retrochiasmatic tumors protrude posteriorly and extend into the third ventricle and commonly present with ventricular dilatation. If hydrocephalus is present the patients will have papilledema but rarely any other visual problem. Endocrine problems are relatively uncommon at the time of presentation with this type of tumor. On angiography, these tumors will show no elevation of the A1 segment of the anterior cerebral artery, but they will show posterior displacement of the basilar artery

and stretching of the posterior communicating arteries. Retrochiasmatic tumors frequently thin out the chiasm to a sheet that is pushed up against the tuberculum sellae, providing a cover for the dome of the tumor (Fig. 30.10C). Despite this severely distorted appearance of the visual apparatus, these patients usually have normal acuity and often have no visual field defect. The retrochiasmatic tumors are reached through a combination of routes always involving an approach between the optic nerve and internal carotid artery and frequently involving opening of the lamina terminalis. To gain access to the route between the optic nerve and the internal carotid artery, the right frontal flap used for both sellar and prechiasmatic tumors is enlarged by removal of the pterion along with some temporal bone and the outer half of the sphenoid wing. This allows one to retract the temporal lobe posteriorly and the frontal lobe upward, thus bringing the optic nerve, optic tract, and internal carotid artery into clear view.

Under the microscope, with self-retaining brain retractors on the frontal and temporal lobes, one can retract the optic tract medially and the internal carotid artery laterally to expose the tumor. If only solid calcified tumor is encountered here, the frontal lobe is elevated off the chiasm and the lamina terminalis is exposed. This structure is greyish and can be easily distinguished from chiasm. The lamina is incised and the third ventricle is entered. The floor of the third ventricle is then opened to expose the tumor cyst, which can be entered and emptied. One can then go back to the route between the internal carotid artery and the optic tract and shift calcified tumor forward to the space that opens up between the two optic nerves as the collapsed tumor allows the chiasm to move back into normal position. Once the tumor has been decompressed, it becomes possible to remove it between the optic nerve and internal carotid artery as well as between the two optic nerves. Occasionally, it must be extracted through the opening in the lamina terminalis.

The membrane of Lilliequist is always intact so there should be no fear of damage to the basilar artery or brain stem. These structures come into clear view once the tumor is removed (Fig. 30.11). Claude Lapras recommends inserting a cotton patty into the third ventricle when the lamina is opened to prevent the migration of tumor contents up into the third ventricle during tumor removal (10). This can certainly happen if no measures are taken to prevent it; one does not become aware of such migration until the postoperative CT scan is seen.

Pineal Region Tumors In recent years the ability to determine serum and CSF markers in the case of pineal region tumors has allowed identification of particular germ cell tumors (1). Choriocarcinomas lead to a marked increase in human chorionic gonadotropin (HCG) in both serum and CSF. Endodermal sinus tumors (yolk sac tumors) are associated with increased levels of alphafetoprotein (AFP) in both serum and CSF. Embryonal carcinomas produce increased levels of both HCG and AFP in serum and CSF. The common germ cell tumor, the germinoma, has no effect on these markers. Among 46 verified pineal tumors treated in our institution, there were only 4 that could be identified with markers (6). With small tumors in this region a CT-guided stereotaxic biopsy is of value. The procedure can be done with little in the way of morbidity and can provide diagnostic specimens of tumor in a safe fashion. In the case of large tumors there is a significant risk of mixed histology, and a small biopsy can therefore be misleading. Six of the 46 verified tumors seen at our institution had mixed histology (6). Radical resection of a malignant tumor has the advantage of leaving less of a tumor burden to be dealt with by radiotherapy. The approach to the pineal region, located in the center of the brain, has always been a challenge to neurosurgeons. There are three main approaches in current use: the transcallosal, the supracerebellar-infratentorial, and the occipital-transtentorial. I use the

posterior transcallosal approach for tumors that extend anteriorly into the third ventricle as well as for those that extend upward toward the corpus callosum (Fig. 30.12). The supracerebellar-infratentorial approach is used for tumors that extend inferiorly into the posterior fossa (Fig. 30.13). The occipital-transtentorial approach is used for tumors that extend posteriorly above the level of the tentorium (Fig. 30.14). A sagittal cut of a magnetic resonance imaging (MRI) scan or a sagittal reconstruction of the CT scan will show the direction of tumor growth (Fig. 30.15). Pineal neoplasms commonly occlude the aqueduct and the posterior

third ventricle; consequently, hydrocephalus is usually manifest when these patients present. Because many of the tumors in the pineal region can disseminate systemically through a diversionary shunt, we utilize a filtered ventriculoperitoneal shunt to treat the attendent hydrocephalus as an initial step in management (15). The diversionary shunt relieves the patient of many of his symptoms and also helps to slacken the brain at the time of surgical exposure. Although it has been assumed best to leave the ventricles dilated and to institute drainage at the time of operation, we have found that diversionary shunting will create a slack brain for several weeks after shunting, allowing one to operate on a well patient with a relaxed brain. In addition to the use of a diversionary shunt, patients are given mannitol in a dose of 2 g/kg during opening of the bone flap and are hyperventilated. For the posterior transcallosal approach the patient is positioned supine with the body and head flexed and the head in the pin fixation headrest in anatomical position (Fig. 30.16). A right parietal bone flap extending from the midline to 4 cm laterally is turned. The length of the bone flap is determined by the preoperative angiogram, which shows the draining parasagittal veins. The bone flap never reaches as far as the motor strip anteriorly and never goes as far back as the lambdoid suture. The medial cut is made just to the left of the midline, thus exposing the sagittal sinus in the operative field (Fig. 30.17). The dural reflection starts out laterally and extends anteriorly and posteriorly depending on the course of the draining parasagittal veins. Every effort is made not to compromise any of these veins during the opening of the dura mater. The dural incision is carried up to the sagittal sinus anteriorly and posteriorly, allowing the dura to be reflected to the left. These vertical limbs of the dural incision are made to preserve the draining veins in safety. The parietal lobe is then gently retracted away from the falx utilizing self-retaining retractors, and the corpus callosum is exposed. Care must be taken not to damage the anterior cerebral arteries. With the aid of the

microscope a 1 -cm-long incision is made in the corpus callosum just in front of the splenium. With most pineal tumors the internal cerebral veins will be situated over the dome of the tumor (Fig. 30.18). These veins are venae communicantes and can be retracted to one side of the dome of the tumor. Once this is done the tumor can be dissected from surrounding structures. In the case of large tumors, we have found it useful to reduce the tumor with the ultrasonic aspirator, which allows for histological examination of all removed tumor tissue. This is particularly important in the pineal region, where so many tumors have mixed histology. In the case of benign tumors such as teratoma or dermoid, it should be possible to remove the tumor totally. With infiltrating tumors such as geminomas, pineoblastomas, and astrocytomas, a subtotal resection is carried out. In those cases in which the tumor surrounds the internal cerebral veins, care must be taken not to injure these veins. Germinomas may occasionally surround the internal veins so that one comes directly down on tumor as the corpus callosum is divided. This is more characteristic of glial tumors and is never seen with large benign teratomas and dermoids, which always elevate the internal cerebral veins in this region. In the callosal approach, the collicular plate and brain stem are not well visualized until the tumor has been removed. Therefore, brain stem function must be carefully monitored as the tumor is removed as there can be serious disturbances in vital signs with traction on tumors that infiltrate the collicular plate. The supracerebellar-infratentorial approach is ideal for tumors that extent inferiorly into the posterior fossa. In these cases, the tentorial notch is large, allowing for downward protrusion of the tumor. The patient is positioned prone, with the head flexed and fixed in anatomical position in the pin fixation headrest.

A midline incision is used to strip muscles off occipital bone. The bone is removed to just above the level of the lateral sinuses and torcular. The foramen magnum is removed. A V-shaped dural incision is made with the two sides of the V coming together in the region of the foramen magnum. In children there is a prominent suboccipital sinus and there may be a falx overlying the vermis. Removing the foramen magnum and bringing the two dural incisions together at the level of the foramen magnum allows one to open dura mater below the level of the midline sinus and falx. The operating microscope is then used and the superior surface of the cerebellum is retracted with self-retaining brain retractors. The draining veins coming off the superior surface of the cerebellum are coagulated and divided, allowing for inferior retraction on the cerebellum. The arachnoid in the pineal regions is extremely thick, obscuring veins and tumor in the region. This arachnoid is opened by sharp dissection. The precentral cerebellar vein is coagulated and divided, allowing further downward retraction on the superior surface of the cerebellum. Care must be taken not to damage the basal veins of Rosenthal, which lie out laterally. Once the tumor is exposed, it can be gutted with the ultrasonic aspirator. If it is not an infiltrating tumor, it can then be easily separated from surrounding structures and removed. The occipital transtentorial approach is utilized for tumors that extend posteriorly above the level of the tentorium. We position the patient supine, with head flexed. A right occipital bone flap is turned. This is hinged inferiorly and exposes the sagittal sinus medially and the torcular and transverse sinus inferiorly. The occipital lobe is retracted superiorly and laterally, exposing the tentorium, the falx, and the straight sinus. Draining veins going into the tentorium and falx are divided. The tentorium is then incised 1 cm from the torcular, with the incision extending laterally and anteriorly to the tentorial notch. The edges of the divided tentorium are retracted with traction sutures and the arachnoid overlying the deep venous system is exposed. Under the microscope, the thinned arachnoid is opened sharply, thus bringing the quadrigeminal plate, the superior vermis, the splenium of the corpus callosum, and the major veins draining into the straight sinus (including the internal cerebral veins, the vein of Galen, the basal veins of Rosenthal, and the precentral cerebellar vein) into view. The veins are dissected from the tumor capsule and the tumor is dissected free from surrounding structures. In the case of a noninfiltrating tumor, the tumor is gutted with the ultrasonic aspirator allowing the capsule to be totally removed once it has been freed from surrounding structures. Infiltrating tumors are resected subtotally. Thalamic Tumors Thalamic tumors are common in childhood. About half of these tumors are benign and almost all are glial in origin (2). Obstruction of CSF pathways is common, and diversionary shunts will frequently be required before the treatment of the tumor. In the case of small thalamic tumors presenting with the classical signs of contralateral hemiplegia, we favor a CT-guided sterotaxic biopsy. In our unit the Brown-Robert-Wells frame is used; we think that we can safely biopsy such small tumors and obtain representative tissue samples to guide us in subsequent tumor management. However, in the case of large thalamic tumors, particularly those that protrude into the lateral ventricle, we recommend an attempt at subtotal excision. This allows the tumor to be made smaller and provides

a more reliable sample for histological diagnosis. With such a reduction procedure, further therapy is frequently unnecessary for low grade astrocytomas in the thalamus. Thalamic tumors that protrude into the lateral ventricle do so anteriorly near the foramen of Monro, and frequently occlude the foramina of Monro and produce hydrocephalus (Fig. 30.19). With such patients, we use an anterior transcallosal approach. We turn a small right frontal bone flap that is centered over the coronal suture and extends anteriorly to the region of the hairline and posteriorly to the pre-Rolandic region. The medial extent of the bone flap is just to the left of midline, and the bone flap extends 4 cm to the right of midline. The patient frequently has a diversionary shunt, is given mannitol in a dose of 2 g/kg, and is hyperventilated. The brain is consequently slack. Every effort is made to preserve draining parasagittal veins. The dural opening is made so as to achieve this end. The frontal lobe is then retracted away from the falx and the corpus callosum is exposed. The corpus callosum is then incised with the ultrasonic aspirator over the lateral ventricle into which the thalamic tumor protrudes. The ultrasonic aspirator is used because it will incise the corpus callosum very gently and bring one down to the ependymal surface of the lateral ventricle. There may be some

veins in the roof of the lateral ventricle that must be coagulated at this point. Self-retaining brain retractors are used in the lateral ventricle and, with the microscope in place, the tumor, which has frequently breached the ependymal surface of the lateral ventricle, can be visualized and partially resected (Fig. 30.20). If the ependymal surface is intact, a thin layer of thalamus must be incised to expose the tumor. Occasionally the cingulate gyri of the two frontal lobes are fused anteriorly and it therefore becomes impossible to expose the corpus callosum. In such cases an incision is made in the cingulate gyrus to allow access to the corpus callosum. In all cases of thalamic tumor in which this approach is used it is relatively simple to make a large opening in the septum pellucidum so that one ventricular shunt will drain both lateral ventricles effectively. Thalamic tumors that expand laterally can be approached transcortically using a posterior parietal bone flap. The parietal cortex is incised, thus allowing both access to thalamus and cytoreduction of the tumor. Operative ultrasound is invaluable in guiding one to the tumor. Posterior thalamic tumors can be resected through the occipital transtentorial approach. In the case of large benign tumors that have been satisfactorily resected, radiotherapy is not used (Fig. 30.19). Radiotherapy is reserved for histologically benign thalamic tumors that show evidence of further growth on CT scan after resection and for histologically malignant tumors. Hypothalamic Tumors Hypothalamic tumors that fill the third ventricle and produce hydrocephalus are approached through an anterior transcallosal route. The third ventricle is entered through the foramen of Monro (Fig. 30.21). If the tumor is large the third ventricle can be entered by dividing the thalamostriate vein and splitting the choroidal fissure, thus gaining access to the entire third ventricle.

In the case of hypothalamic gliomas that present as suprasellar masses, a combined subfrontal and transpterional route allows adequate access to the tumor. As long as one works in tumor, one can satisfactorily remove tumor without disturbing normal function. In the case of hypothalamic astrocytomas that fill the third ventricle and are removed by the transcallosal route, endocrine dysfunction is uncommon, particularly if the patient is not given radiotherapy.

When a low grade astrocytoma is resected, radiation therapy is frequently unnecessary. Many such patients have now been followed for over 10 years with no change in their residual tumor without radiotherapy (Fig. 30.22). On the other hand, if the tumor does show evidence of further growth after resection a course of radiotherapy is indicated; some of these low grade tumors will disappear with radiotherapy. The hypothalamic hamartomas present with precocious puberty. The transpterional route allows access to the tumor between the optic nerve and internal carotid artery. The tumor frequently hangs down from the hypothalamus in the suprasellar cistern like a grape and can be readily resected (Fig. 30.23).

Suprasellar Germinomas These tumors typically occur in young girls who present with visual and endocrine problems. The tumor is in the suprasellar area and a subfront and transpterional route is used to expose the tumor in an attempt to resect it subtotally (Fig. 30.24). These suprasellar germinomas will frequently infiltrate optic nerves and enlarge the chiasm and nerves, thus mimicking the appearance of an optic nerve glioma (Fig. 30.25). If the tumor extends into the third ventricle, filling it and producing hydrocephalus, I prefer to use an anterior transcallosal approach and resect the tumor from within the third ventricle.

Choroid Plexus Papillomas These tumors are also approached through an anterior transcallosal route. The third ventricle is entered by dividing the thalamostriate vein and opening the choroidal fissure. This allows access to the entire third ventricle. The tumor's blood supply can then be controlled and the tumor resected.

Subependymal Mixed Gliomas of Tuberous Sclerosis These tumors can reach enormous size (Fig. 30.26). They frequently occlude the foramen of Monro and produce hydrocephalus. They too can be approached through an anterior transcallosal route and frequently require posterior enlargement of the foramen of Monro to allow resection of the tumor. The Cavitron aspirator has proved invaluable in resecting these tumors.

Conclusion The advances in neuroradiological imaging of lesions in and around the third ventricle during the past decade have allowed the neurosurgeon to plan his approach to tumors in this region in a rational fashion. The technological advances during the same period have allowed safe execution of a variety of routes to tumors in the region of the third ventricle and expeditious removal of these tumors. Many of the intraaxial tumors in the region of the third ventricle are low grade astrocytomas. In the past these low grade tumors were frequently only biopsied and were then treated with radiotherapy. Because of the follow-up that CT scanning provides we now know that many of these low grade tumors do not respond to radiotherapy. With radical resection there frequently seems to be no further growth of the residual tumor over a follow-up of many years. However, there is little doubt that radiotherapy is of use for some of these low grade tumors. Therefore, we can now reserve radiotherapy for low grade tumors that cannot be removed subtotally and for low grade tumors that behave in an aggressive fashion and continue showing evidence of growth after subtotal removal. The extraaxial tumors in and around the third ventricle can be totally and safely removed with the technological advances that we now have available to us, thus avoiding the risks inherent in courses of radiotherapy to the developing brain of the child.

References

1. Allen JC, Nisselbaum J, Epstein F, Rosen G, Schawartz MK: Alphafetoprotein and human chorionic gonadotrophin determination in Cerebrospinal fluid: An aid to the diagnosis and management of intracranial germ-cell tumors. J Neurosurg 51:368-374, 1979. 2. Bernstein M, Hoffman HJ, Halliday W, Hendrick EB, Humphreys RP: Thala mic tumors in children: Longterm follow-up and treatment guide lines. J Neurosurg 61:649-656, 1984. 3. Davies FA: Primary tumors of the optic nerve (a phenomenon of von Recklinghausen's disease); a clinical and pathological study with a report of five

4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15.

cases and a review of the literature. Arch Ophthalmol 23:735-821, 9571022, 1940. Hoffman HJ: Supratentorial tumors in children. In Youmans JR (ed): Neu rological Surgery. Philadelphia, 1982, pp 2702-2732. Hoffman HJ: Craniopharyngioma: The continuing controversy on manage ment. Concepts Pediatr Neurosurg 2:14-28, 1982. Hoffman HJ, Yoshida M, Hendrick EB, Humphreys RP: Pineal tumors in childhood. Concepts Pediatr Neurosurg 4:360-386, 1983. Housepian EM: Surgical treatment of unilateral optic nerve gliomas. J Neu rosurg 31:604-607, 1969. Jackson H: Orbital tumors. J Neurosurg 18:317-439, 1912. Kronlein RV: Zur Pathologie und Operativen Behandlung der Dermoid Cysten der Orbita. Beitr Z Klin Chir 4:149-163, 1988. Lapras C, Patet JD, Mottolese C, Lapras CA Jr: Results after surgery for craniopharyngioma in 42 children. Presented at XIII Annual Meeting of the International Society for Pediatric Neurosurgery, Mexico City, July 18, 1985. Matson DD: Neurosurgery of Infancy and Childhood, ed 2. Springfield, IL, Charles С Thomas, 1969, pp 523-536 Mori K, Handa H, Murata T, Takeuchi J, Nura S, Osaka K: Results of treatment of craniopharyngioma. Childs Brain 6:303-312, 1980. Rougerie J: What can be expected from the surgical treatment of craniophar yngiomas in children: Report of 92 cases. Childs Brain 5:433-449, 1979. Sweet WH: Radical surgical treatment of craniopharyngioma. Clin Neurosurg 23:52-79, 1976. Wilson ER, Takei Y, Bikhoff WT, O'Brien M, Tindall GT: Abdominal metastases of primary intracranial yolk sac tumors through ventriculo-peritoneal shunts: Report of three cases. Neurosurgery 5:356-364, 1979.

31 Applications of Computerized Tomographic Guidance Stereotaxy Michael L J. Apuzzo, M.D., Parakrama T. Chandrasoma, M.D., Vladimir Zelman, M.D., Roger I. von Hanwehr, M.D., and Craig A. Fredericks, M.D.

Refinements in the technology of radiographic imaging have revolutionized the surgeon's appreciation of intracranial mass processes. From the standpoints of localization and extent, lesions may be defined to comprehensive limits that have previously been unappreciated. The wedding of computerized tomographic (CT) radiological imaging and techniques of stereotaxy allows the formulation of totally novel and improved methods of management strategies for various intracranial disorders (9, 37) and mass processes (4, 6, 31, 41, 42, 50, 51, 53, 65). Perhaps in no area of the brain is this more apparent than in the third ventricular region, where the spectrum of pathology and the controversial nature of many management strategies argue for low risk biological assays of offending lesions and management techniques that are minimally traumatic in view of the physiologically fragile nature of the region and the potential risks versus the benefits of major operative endeavors (4, 68). Such has been the experience at the University of Southern California Medical Center, where 500 consecutive CT stereotaxic guidance operations have been analyzed. Over 140 were studied for the definition of management strategies of third ventricular region masses. These operations were undertaken for purposes of biopsy, culture, aspiration, permanent conduit placement, radionuclide implantation (4, 5, 28, 29, 38, 56, 57, 82), endoscopic visualization with biopsy (4, 6), excision, or aspiration. Although a number of stereotaxic instruments applicable to the techniques of computerized guidance stereotaxy are available, our experience and familiarity with the Brown-Roberts-Wells (BRW) system is presented as the basis for point access, the guidance for point instrumentation as defined in this chapter.

Instrumentation and Techniques Stereotaxic Instrument The BRW guidance system (Radionics, Inc., Burlington, MA) consists of four major components: a head ring, a localizer unit, an arc guidance system, and a phantom base simulator. Head ring This nickel-plated aluminum ring is fixed to the cranium at four points by vertical graphite epoxy posts with plastic and steel pins (Fig. 31.1). This component of the system acts as a stationary platform for the localizer and arc guidance systems, which are applied during scanning and the stereotaxic invasive procedure. Localizing Unit This component is comprised of six vertical and three diagonal graphite rods and is attached to the head ring during the imaging process (Fig. 31.2). The rods act as reference markers on each scan slice and may be mathematically related to an intracranial target point. Arc Guidance System This component consists of a base ring, a rotatable ring, and a perpendicular arc (Fig. 31.3). The base ring is accurately and securely affixed to the patient head ring by three mounting balls. The rotatable ring surrounds the base ring and allows 360° (alpha) manipulation. Attached to the rotatable ring is a perpendicular arc. This arc may be pivoted through the base ring to 30° (beta). A radial slide is attached to the arc. This slide has 180° (gamma) of movement along the arc. In addition, the slide contains a sleeve that pivots 90° (delta) about the axis pin of the arc. This sleeve accepts various bushings that act as carriers and direct the trajectory of surgical instrumentation to the target point. The combination of angle settings (alpha, beta, gamma, delta) are computed in an Epson HX-20 programmable calculator and allow the

operator to design entry points to use with intracranial target points anywhere within the sphere of the guidance arc. Phantom Base This device consists of a base ring (head ring equivalent) and a movable pointed tip designated a "phantom target" (Fig. 31.4). This point may be fixed to any x, y, and z setting. These settings are established when

calculation output coordinates for target point or entry point coordinates are chosen with the arc system. After entry of scan coordinates and entry point data derived from the phantom, the Epson HX-20 portable computer provides the arc frame settings and the distance to the target. The target coordinates are set in the phantom. Trajectory coordinates are placed on the arc and checked on the phantom target. The arc system on the phantom ring bears the same relationship to the phantom target as it will to the intracranial target when it will be affixed to the patient base ring. This provides an extracranial check of the target settings, arc coordinates, and distances for the instrument placement in intracranial point access.

Software The software for the system uses data from x and у coordinates of the central CT pixel of the nine localizer rods and CT target. Applying various mathematical formulae, the two-dimensional formulae (x and y) are transformed to three-dimensional coordinates (x, y, and z) relating the position of the localizer rods and the targets to the plane with its vertical height reference to the patient's base ring. With the x, y, and z coordinates of the target derived, the design of the arc system allows a course and distance to be plotted between any two points in space. These points are the entry point (scalp or dura mater) and the target. Software allows for selection of multiple targets, entry points, and parallel transits per scan plane. Steps in Point Access Point access is achieved readily, rapidly, and safely with local anesthesia with standby utilized in all but selected pediatric cases. Application of the Base Ring (10 Minutes) After preparation of the scalp, selection of appropriate post and pin position is made with the patient in the sitting position (Fig. 31.5). This is most easily achieved on an operating room table in an anesthesia

preparation room. In most cases for lesions in the third ventricular region the base ring is positioned in a plane defined by the tip of the nose and the inion. Not only is the ring secured during placement fixation by a Velcro strap, but observors at the foot of the operating table and lateral to the patient visually monitor ring position during the placement. The assistant lateral to the patient assesses ring position and manually maintains the sagittal suture in position parallel to the floor. The operator is free to direct and infiltrate the scalp at the four points of fixation with 1% xylocaine with epinephrine. He advances the nylon drive pins to the scalp and, adjusting the tracks, advances the four carbon fiber posts to set the pins securely to the pericranium. This evolution requires 5 to 10 minutes to complete. Generally, intravenous contrast medium is being infused during the early stages of this procedure. The patient is transported to the scanner with the anesthesiologist in attendance. Scan Target (10 Minutes) The localizer unit is affixed to the base ring in the scanner (Fig. 31.6). Because detailed imaging studies have been previously attained, an abbreviated study defining desired aspects of the target region may be obtained. Depending on the number of slices, this is generally accomplished in 10 minutes and the patient is returned to the operating room,

where he is prepared for the formal surgical aspect of the procedure. The surgeons then study the scan slices, select a slice for targeting, and derive and record pixel coordinates for the nine localizer rods and the targets. These steps require 5 to 10 minutes. Entry Point Selection (5 Minutes) Entry points are selected according to the target position and may be variable according to special elements of lesion, disposition, and extent. Once a scalp point is selected the arc system is affixed to the base ring and the entry point is marked in position by a probe placed to the point and rigidly fixed in the arc (Fig. 31.7). The arc is then moved to the phantom base where the x, y, and z coordinates are derived from three appropriate scales and are recorded. Optimal entry point selection may be attained within the scanning unit. If scanner software is appropriate, an entry site may be marked during the imaging and planes of transit may be reconstructed to the target point. This technique used concurrently with rapid bolus contrast infusion to define vascular structures

will increase the safety and control of the entry to target transit. This requires 5 minutes. Data Processing (10 Minutes) All localization values for reference, the target values, and, finally, the entry point coordinates are entered on the Epson HX-20 programmable computer and appropriate alpha, beta, gamma, and delta settings are derived for the arc system along with data describing the depth of the target from the arc slide bushing. In addition, coordinates for localization of the phantom are rendered (Fig. 31.8A). Arc Settings and Phantom (10 Minutes) The settings for the arc are entered and initially checked on the phantom base against the entry point (phantom target) (Fig. 31.8, В to D). The target coordinates in vertical, lateral, and anterior planes (x, y, z) are then placed on the phantom as the target and the arc coordinates and depth of transit to the phantom target point are checked, thus providing the extracranial assessment of arc setting and transit trajectories and

distances. In addition, instrumentation to be introduced to the true target point may be precisely calibrated to the phantom target. In general, these preparations are usually completed in 45 minutes, with additional time for transportation to and from the scanner. Instrumentation at the Target Point Biopsy For purposes of biopsy a #13 gauge cannula with a blunt stylette is introduced to the target site as a conduit (Fig. 31.9). In preparation for the introduction of this instrument to the target point, a flexible bronchoscopy cup biopsy forceps is advanced through the cannula and the distance to the emergence of the cups is carefully marked with a Steristrip. The cup closure and length of introduction is determined on the phantom base target point and marked with an adjustable sleeve in reference to the rigid bushing and guide tube fixation of the arc system (Fig. 31.10). The scalp entry point is infiltrated with 1 % xylocaine with epinephrine, and a 7- to 10-mm incision is made. The arc system is now affixed to the

base ring, and a guide tube appropriate for fixation of a long 4.5-mm twist drill is passed and secured within the arc slide bushing. The twist drill is then used to penetrate to the skull; the inner table is recognized by a catch of the drill and minimal discomfort is appreciated by the patient as the dura mater is encountered. A sharp probe is used to penetrate the dura after the drill guide tube is exchanged for a 2.7-mm guide tube in the slide bushing. The 13 gauge cannula is then advanced to the target site (along the trajectory of the drill hole) in rigid fixation. The blunt stylette is removed. The biopsy forceps is advanced to the Steristrip marker and opened, and biopsies are obtained at various vectors and, if appropriate, at various depths. An assistant opens and closes the cups of the instrument while the operator palpates the barrel of the flexible forceps, appreciating resistance and tissue texture changes. Aspiration for drainage and fluid analysis may be undertaken. A number of devices for tissue retrieval have been used on our service (Fig. 31.9 and 31.11). The flexible forceps are considered the best universal instrument for biopsy, while the slotted and coiled instruments are dependent on minimal tissue resistance for removal of satisfactory samples. The forceps have afforded satisfactory tissue samples with minimal morbidity. Permanent Conduit (Fig. 31.12) For the placement of a permanent drainage conduit we have used a 22cm x 2.5-mm-diameter Silastic tubing with a rigid stylette (Radionics, Inc., Burlington, MA). The device is available with either a single 0.5-cm slot 0.5 cm from the tip or two 0.5-cm slots at 180° orientation 0.5 and 1 cm from the catheter tip. The catheter is fixed to a large Rickham reservoir which is "countersunk" in the calvarium. The technique for conduit placement is similar to that of biopsy. A 3-cm curvilinear scalp flap is made. The arc-directed 4.5-mm twist drill penetration is then followed by the use of a Cushing perforator to form a cone-shaped

depression in the calvarium, which will accept a Rickham device. The 13 gauge cannula is then introduced to the target point, aspiration is attempted to assure entry to the cystic cavity, and the catheter is then advanced to the target point, aspirated, trimmed, and secured to the Rickham device, which seats in the calvarial depression. Radionuclide Instillation (Fig. 31.12) Colloid Beta Emitter. The use of colloid-based beta-emitting radionuclides offers a plausible management option for cystic lesions in the peri-third ventricular area. In particular, cystic craniopharyngiomas may be approached by this technique. To avoid leaking of the colloid, we wait 7 to 14 days after the placement of a catheter-reservoir system and perform a cystogram with metrizamide to determine cyst volume and the integrity of the system (44). Cyst volume has been an important parameter for our radiotherapists in calculating the quantity and distribution kinetics of the colloid. The radionuclide is then infused in a single bolus through the Rickham device. Point Source Gamma Emitter. To assure homogeneous dose delivery in solid tumors with strict control of the rate and dose distribution, multiple sources of iridium-192 have been used with multiple catheter arrays (5, 6). The rationale for this approach and elements of the technique have been previously described. It involves the transcutaneous placement of Silastic catheters in parallel array to CT-derived target points. The catheters are delivered through 4.5-mm twist drill calvarial openings and are fixed to the scalp by adjustable Silastic cuffs, which are initially secured to the catheters by Aron alpha acrylic. Later, following afterloading with iridium ribbons, a large vascular clip is used to consolidate the catheter ribbon and cuff complex for the 4- to 6-day period of brachytherapy. After dose delivery, the scalp sutures are cut, and the catheters with sources are removed at the bedside. The patients are maintained on full dose high potency glucocorticoids and prophylactic antibiotics intravenously by the heparin lock system during the period of treatment. Glucocorticoids are then tapered related to patient tolerance.

Endoscopy A 6.2-mm-diameter endoscope with a 20-cm-long barrel has been used for both cerebroscopy and ventriculoscopy (Fig. 31.13). The instrument, produced by Karl Storz Endoscopy (Tuttlingen, West Germany), provides capabilities of visualization and irrigation and a port for the introduction of point instrumentation for biopsy with either flexible or rigid cupped forceps. In addition, specialized cannulae for cyst wall puncture and aspiration are readily accepted (Fig. 31.14). Submersible quartz fibers for transmission of laser energy may be likewise introduced for applications of hemostasis or lesion coagulation. These procedures are undertaken with local anesthesia with a neuro-

anesthesiologist in attendance. For visualization of the foramen of Monro, a scan slice through the structure is obtained and the target point is selected at the orifice. A 18-mm burr hole is then made 1 cm anterior to the coronal suture in the right pupillary line, and the entry point coordinate is selected from the center of the exposed dura (Fig. 31.15). After checking of arc coordinate settings on the phantom base the dura is opened and a 5-mm corticopial window is fashioned with bipolar coagulating forceps. The endoscopy sheath with a blunt stylette in place is then introduced through the rigid bushing (with a Teflon sleeve for ease of passage) to the target point. The stylette is then removed and the fiberoptic visualization package is introduced, sealing the system to the target point. Irrigating fluid (Ringer's lactate) at body temperature is then introduced intermittently through adjustable side ports. Visualization may be enhanced in both clarity and scope by minor alterations of depth and angulation within the arc system as well as employment of irrigating fluid and instrumentation through the central port. Cyst puncture and aspiration may be undertaken under direct visualization at the foramen of Monro. A 13 gauge cannula with a blunt stylette is introduced to approximate the cyst wall. Next, a sharp-tipped stylette

replaces the blunt one and cyst wall puncture is undertaken; the cannula is then advanced into the cystic cavity and aspiration is undertaken. Calvarial Entry Point Selection Selection of an appropriate entry point is critical for the safety of transit to individual target points. Such a selection is made based on considerations of (a) the lesion's location, size, and suspected composition; (b) intervening neural structures; (c) intervening vascular structures; and (d) the target point and objective of the procedure. The safety of transit is increased by using computer software that allows real time reconstruction of transit planes or individual axial slice presentation with cursor marking of transit points. This data management capability, when coupled with rapid bolus infusion of contrast agent to define neighboring and other vascular structures, enhances the safety of transit and allows alteration of the entry point. Approximations of entry point selection are presented in Table 31.1. Entry point and transit trajectory are not standardized, but are selected individually related to the four variables noted in the preceding paragraph. Pathological Considerations Although the technique of imaging directed stereotaxic brain biopsy has been developed recently, the methods used for pathological interpre-

tation have been established for many years. The use of smears in the diagnosis of nervous system lesions was first reported by Eisenhart and Cushing and was perfected in later years (25, 40, 55, 70, 83). Many extensive reports are available on the appearances of brain tumors on smear preparations (2, 52, 83). Routine frozen sections and other histological techniques used are standard techniques. The specimen obtained at stereotaxic biopsy differs from other brain biopsy specimens in two ways, one a disadvantage and the other an advantage. The major disadvantage with the specimen is its size. The usual specimen is 1 to 2 mm in greatest diameter. Two or three such pieces are obtained initially, with more similarly sized pieces provided if the pathologist cannot make a diagnosis. Although it seems to be restrictive initially, as experience with the technique increases this represents an adequate sample in the majority of cases. A more serious limitation is the potential sampling error resulting from the small specimen. When a neoplasm is a composite of various populations, errors may result. Differences in the degree of malignancy in astrocytomas and variations in the appearance of pineal neoplasms in different regions are obvious examples of such sampling problems. The major advantage with the specimen is that it is obtained from a predetermined CT scan target point in the lesion. With little exception, this provides a more representative specimen than an open brain biopsy at the edge of the lesion, where the greatest difficulty exists in establishing a diagnosis. The objective of a stereotaxic biopsy must be fully understood. It should be used to provide a guide in terms of tissue diagnosis so that appropriate treatment can be planned in a manner that is more rational and scientific than without such information. When so utilized, stereotaxic diagnosis is very successful because it provides excellent differentiation between broad groups of lesions, e.g., inflammatory versus neoplastic. The relatively minor variations between stereotaxic and final histological diagnosis rarely affect treatment. Interpretation of the Stereotaxic Biopsy Three factors are essential for the interpretation of the stereotaxic biopsy. These are the pathologist, the neurosurgeon who performs the biopsy, and the CT scan. In our institution, the neurosurgeon brings the specimen and CT scan to the pathology frozen section area. For transport from the operating room, the specimen is placed on a saline-soaked piece of Gelfoam and is carried in a Petri dish. In rare cases where there is difficulty obtaining a solid specimen, the pathologist comes to the operating room with fixative, stain, and microscope. The pathologist who interprets the stereotaxic biopsy may be either a neuropathologist with experience and interest in cytology and handling small specimens or a general surgical pathologist with experience and interest in surgical neuropathology and aspiration cytology. In our department the stereotaxic biopsies are interpreted by a surgical pathologist. The establishment of a team of pathologist and neurosurgeon who understand the procedure and work well together is imperative. The neurosurgeon who performs the biopsy provides invaluable information. The consistency of the specimen is important for two reasons. First, it may give a clue as to the possible nature of the lesion. Second, this information influences the way in which the specimen is processed. The CT scan and its interpretation are critical, and a stereotaxic biopsy should not be interpreted without this information. Discussion of the CT

scan always includes a clinical history and a differential diagnosis. If a pathological diagnosis does not conform to the CT scan appearance, the case is reevaluated. For example, a biopsy showing only low grade astrocytoma may be deemed insufficient and a request may be made for additional sampling if the CT scan or clinical history suggests a more histologically malignant neoplasm. Processing the Stereotaxic Biopsy We almost invariably prepare a smear as the first step in processing the specimen. The only exception to this is if the neurosurgeon reports an unusually firm consistency of the lesion at the time of biopsy. Very firm lesions tend to smear poorly, and we proceed to frozen section directly in this circumstance. This happens very rarely. To make the smear, we cut out with a sharp scalpel blade a minute piece from each biopsy specimen, lay these out in a horizontal row on a glass slide, and smear them using a second glass slide. The best method of smearing is to first crush the tissue between the flat surfaces of the two slides and then draw the slides apart. With experience the amount of pressure needed to make the optimal smear can be assessed by the way in which the specimen crushes between the slides. All pieces are sampled, and the smear from each piece can be separately identified by its position on the slide. When the biopsies are less than 0.5 mm, the entire specimen is used to make the smear. The amount of pressure needed for making smears varies. Normal brain smears very easily. Neoplasms vary in the way in which they smear even within a given class of neoplasm. Low grade astrocytomas sometimes smear very easily but at other times do not smear out at all. When a specimen does not smear out, we use maximal pressure to produce a crush preparation. While this is not as satisfactory as a smear, it frequently provides diagnostic information. Both slides used to make the smear are stained. The smear or crush preparation is immediately fixed in 100% methanol and stained by a rapid hematoxylin and eosin (H & E) technique. This takes 1 minute to perform, provides a permanent smear that can be stored indefinitely, and gives excellent cytological detail. We initially used air-drying and Romanowsky stain, as suggested in much of the literature, but discarded it. Romanowsky-stained smears are associated with cytological distortion due to drying and are much less satisfactory for permanent storage. The latter is a major problem as the smear represents the diagnostic material in many cases and the only diagnostic material in a few cases. H & E staining is also much closer to what the pathologist sees in routine sections and permits better correlation between smears and sections. We also tried Papanicolaou stain, but discarded it as it stained the background fibrillary material less effectively than eosin. Currently, we use Romanowsky stains in addition to H & E only if a diagnosis of malignant lymphoma is suspected. Microscopic examination of the smear determines the further processing of the specimen. If the initial smear is diagnostic, the remainder of the material is placed in 10% formalin for permanent paraffin-embedded sections. If the smear is not diagnostic, the rest of the material is processed, either by repeat smear or frozen section. If a diagnosis is still not reached, additional biopsy material from a slightly different target point is requested. We have never requested more than one additional series of biopsies and have been successful in providing accurate diagnostic information in over 90% of cases (4, 6). Where necessary, specimens may be taken for electron microscopy or snap frozen for immunoperoxidase

studies and additional specimens may be sent for culture if an infectious cause seems possible. The Normal Brain Smear It is part of training for the pathologist to examine normal brain smears. We used autopsy brain to provide normal controls for different parts of the brain. We found that such preparations provide more than adequate material with surprisingly good preservation of cytological detail. Recognizing normal cellularity of different areas of the brain and the appearance of the normal fibrillary background of the deep cerebral white matter is critical. The brain around the third ventricle resembles deep white matter in other areas of the cerebral hemispheres and does not cause confusion. Stereotaxic biopsies of intraventricular lesions frequently show cells from the choroid plexus. These appear as groups of columnar epithelial cells, arranged in clusters and singly. Calcification is frequently present in the choroid plexus. These features can cause confusion when diagnoses of craniopharyngioma and colloid cyst are considerations. Smears from the white matter of the cerebral hemispheres show scattered glial cells in a background of fine fibers. The fibers are mainly the long tracts passing through the area and do not originate in the glial cells. Most of the glial cells are small, very uniform, and round to oval and show few processes in routine smears. It is very difficult to identify specific types of glial cells in normal brain. Such identification is possible with special techniques (18). The thalamus and basal ganglia show the presence of neurons in addition to glial cells. Neurons appear as large cells with abundant cytoplasm and a central nucleus that contains a prominent nucleolus. Reactive Gliosis versus Astrocytoma The differentiation of low grade astrocytoma from reactive gliosis presents fewer problems on stereotaxic biopsy than open biopsy. Delineation of low grade astrocytoma at operation is difficult, and open biopsy specimens are frequently from the periphery, where normal brain is infiltrated by astrocytoma. We have great difficulty on such open biopsy specimens where infiltration of brain by neoplastic astrocytes is very difficult to differentiate from reactive gliosis. The stereotaxic biopsy specimen is from the center of the much more accurately delineated CT lesion, and the smear shows only neoplastic astrocytes and their processes and presents an appearance that is usually diagnostic. Astrocytomas include the fibrillary astrocytoma and the juvenile pilocytic astrocytoma. The latter occurs more commonly in the region of the hypothalamus. These are difficult to distinguish on smear, and we use the term astrocytoma to cover both lesions. The majority of astrocytomas smear poorly, the cells tending to remain in fibrillar masses (Fig. 31.16). The masses show a network of coarse fibrillary processes in which are uniformly scattered fibrillary astrocytes. The processes arise from and are traceable to the astrocytes, unlike in normal brain. The cells are slightly larger than normal glial cells and have short, spindle-shaped, often angulated nuclei with blunted ends. More rounded cells may also be present. There may be considerable cytological distortion. Although they smear with difficulty, the diagnosis is easily made on smear, mainly by the greatly abnormal, coarse fibrillar background. On frozen section, the cellularity is only slightly increased. A minority of astrocytomas smear easily (Fig. 31.17). The increased cellularity and the presence of coarse fibrillar processes arising from the

spindle or stellate astrocytic cells distinguishes these from normal brain. Astrocytomas tend to be more uniform than reactive gliosis, where the cell population seems to consist of several different cell types. Astrocytomas with microcystic change appear more cellular on the smear than on sections. Because of this, when a neoplasm with clinical or cytological

features of a low grade astrocytoma appears more cellular than usual, we proceed to a frozen section. The absence of significant cytological atypia, pleomorphism, endothelial proliferation, mitotic activity, or necrosis is essential for a diagnosis of low grade astrocytoma. The presence of any of these features, even in a small area of the smear, suggests the probability of a more malignant astrocytoma (26). Reactive gliosis usually presents a more heterogeneous appearance, with fibrillary, protoplasmic, and gemistocytic astrocytes and microglial cells. These occur in a background that may have the fine granularity of normal brain or be composed of coarse fibers. Inflammatory cells, including neutrophils, lymphocytes, and plasma cells, foamy histiocytes, and hemosiderin pigment may be present. We commonly encounter gliosis when a stereotaxic biopsy is taken from the edge of a cystic or necrotic lesion. In most of these cases, the CT scan appearance makes low grade astrocytoma unlikely, so that differentiation of astrocytoma and gliosis is not even a consideration. Malignant Astrocytoma, Glioblastoma Multiforme The diagnosis of malignant (or anaplastic) astrocytoma and glioblastoma multiforme is usually not difficult. Both present highly cellular smears in the characteristic fibrillary background of astrocytic neoplasms (Figs. 31.18 and 31.19). There is cytological pleomorphism, nuclear enlargement, and abnormal chromatin distribution in the nuclei (Fig. 31.20). These features are very prominent in glioblastoma multiforme. The presence of numerous gemistocytic astrocytes is also a common feature in high grade astrocytomas. Gemistocytic astrocytes are recognizable as large cells with homogeneous eosinophilic cytoplasm and a small, often eccentric nucleus. Differentiation between malignant (anaplastic) astrocytoma and glio-

blastoma multiforme is possible when there is evidence of necrosis on the smear. Necrosis is recognizable as either coagulated pink-purple debris or better, as individual cells that have undergone coagulative necrosis. We made a diagnosis of glioblastoma multiforme only when necrosis is present (26). Endothelial proliferation is recognizable as collections of cohesive spindle cells arranged in the linear fashion of a blood

vessel. A cellular astrocytoma that has endothelial proliferation, but no necrosis, is reported as a malignant (anaplastic) astrocytoma. Sampling is an important factor in the differentiation between these two astrocytomas. A report of malignant astrocytoma is always accompanied by a comment that glioblastoma multiforme cannot be excluded. The differentiation of glioblastoma multiforme from metastatic carcinoma can be made, usually without much difficulty, by the lack of cohesiveness of the malignant astrocytes, the presence of numerous gemistocytes, and, most importantly, the fibrillary background of astrocytic neoplasms (Figs. 31.10, 31.19, and 31.20). The presence of extensive necrosis presents problems at stereotaxic biopsy. A specimen composed of necrotic tissue is inadequate and necessitates further sampling. Careful selection of the target point in the lesion can avoid this problem. Subependymal Giant Cell Astrocytoma This neoplasm is composed of large cells with abundant pink cytoplasm, numerous fibrillary processes, and eccentric nuclei that resemble gemistocytic astrocytes very closely. The clinical history of tuberous sclerosis, the occurrence in a younger age group, and the ventricular location of the subependymal giant cell astrocytoma should permit its differentiation from gemistocytic astrocytoma. Metastatic Neoplasms Stereotaxic biopsy finds a major role in establishing a histological diagnosis of a brain lesion in a patient with a known primary malignancy elsewhere. In these cases, the diagnosis of metastatic carcinoma presents few problems. Two types of smears may be seen. The first is a highly cellular smear composed of cohesive malignant cells without a fibrillary background from an area where the neoplasm has replaced normal brain (Fig. 31.21). The cells vary in their differentiation. Melanin pigment of

malignant melanoma is more clearly seen in smears than frozen or permanent sections. The second type of smear shows a small number of cohesive cell groups in a background of necrosis or normal brain. In this type of specimen, the smear represents a much more sensitive method of establishing a diagnosis than a frozen section. When a patient is not known to have a primary lesion elsewhere, the diagnosis of metastatic tumor must be made on the features seen on the smear alone. Carcinomas show cohesive groups of round or oval malignant cells without fibrillary processes. The diagnosis of specific sites is possible for a few distinctive tumors like oat cell carcinoma of lung and renal adenocarcinoma. In most cases, however, the carcinoma is too poorly differentiated to indicate the primary site. Malignant neoplasms composed of noncohesive spindle cells are very difficult to diagnose specifically. Metastatic sarcoma, sarcomas of the brain, including meningeal sarcoma, and gliosarcoma are all possibilities. Cellular, Nonastrocytic, Primary Neoplasms Meningioma Meningioma rarely occurs as an intraventricular neoplasm. It occurs more commonly as a basal tumor that may extend up into the floor of the third ventricle. The neurosurgeon frequently recognizes the firm consistency of the neoplasm at the time of biopsy. On smear, meningiomas are highly cellular and composed of meningothelial cells (Fig. 31.22). These are easily recognizable as plump spindle cells with central nuclei and abundant pink cytoplasm. The cells present as cohesive groups with numerous single cells. Although they usually present a very regular cytological appearance, considerable cytological atypia is not uncommon. Enlargement of cells, pleomorphism, nuclear chromatin abnormalities, and rare mitotic figures may be seen. Tight whorls are seen in most cases. Psammoma bodies are seen in a few cases. Foamy histiocytes are

common and correlate well with xanthomatous degeneration. The background material is usually scant. In some cases, a collagenous background is easily distinguishable from the fibrillary background of an astrocytic neoplasm. It should be stressed that the critical diagnostic feature is the recognition of meningothelial cells on smear. Although we evaluate cytological atypia in meningiomas and include these in our report, we do not attempt to predict biological behavior on the basis of smears. In a few cases, extreme atypia on smear has correlated with malignancy. We also do not attempt to classify meningiomas according to histological subtypes. Oligodendroglioma Oligodendroglioma infrequently presents as an intraventricular neoplasm. The smear is highly cellular (Fig. 31.23). The cells are small and dyscohesive and have very uniform round nuclei with a delicate chromatin distribution. The cytoplasm and cell membranes cannot usually be seen on smears and a halo-like appearance is seen only rarely around cells. The background is nonfibrillary and usually scant. Calcospherites are commonly found, even when calcification is not appreciated radiologically, and are helpful in diagnosis. Another feature that is helpful is the presence of numerous small blood vessels in the smear. These are recognizable as linear structures containing erythrocytes and lined by regular endothelial cells. Ependymoma Although uncommon, ependymoma frequently enters into the differential diagnosis of third ventricular neoplasms. On smear, they present a fairly uniform population of small round to polygonal cells presenting as lightly cohesive groupings and single cells. The cells have somewhat hyperchromatic nuclei and scanty cytoplasm. The smear is highly cellular and the background lacks astrocytic processes. Mild cellular atypia is

commonly present. The tendency to form rosettes varies. In one of our cases, the smear was composed almost entirely of rosette-like cell clusters. These had a central mass of fibrillar material surrounded by ependymal cells (Fig. 31.24). In other cases, rosettes could not be identified on smears, except for a tendency for the cells to group around vascular structures. Choroid Plexus Neoplasms Normal choroid plexus is very distinctive, appearing as small cuboidal to columnar cells lining a fibrovascular stroma that is frequently thrown into papillary structures. We anticipate that cytological differentiation of Papillomas from normal choroid plexus will probably be a matter of quantity and accurate CT localization. Pineal Neoplasms Pineal neoplasms represent a very important group of lesions in the region of the third ventricle. They are among the most troublesome at stereotaxic biopsy because their frequent lack of uniformity make them very prone to sampling errors. We have encountered a number of germinomas presenting as intraaxial para-third ventricle lesions. In one of these, the diagnosis was made with relative ease because the smear showed the typical dual population of germ cells and lymphocytes (Fig. 31.25). The germ cells appear as highly active-looking, large, round cells with central nuclei. Cytological atypia and mitotic activity are present. Although not appearing cohesive, the cells tend not to separate extensively. Interspersed among these large cells are numerous small lymphocytes and collagenous bands of varying thickness. Another case presented the greatest difficulty in our experience with stereotaxic biopsies. Two separate stereotaxic biopsies showed chronic inflammatory cells on smear and epithelioid granulomas on sections (Fig. 31.26). Although culture was negative, the patient had an

unsuccessful trial of antituberculous therapy before the diagnosis was made on open biopsy (Fig. 31.27). The germinoma had areas of extensive granulomatous inflammation leading to a sampling error at stereotaxic biopsy. Pineal germ cell tumors are frequently mixed, with teratoma, embryonal carcinoma, and yolk sac carcinoma present in varying amounts (12). This presents obvious sampling problems. However, biochemical markers in serum and Cerebrospinal fluid may suggest such a composite if it is overlooked because of sampling error.

Germinomas and other germ cell tumors can occur in the floor of the third ventricle ("ectopic pinealoma"). Pineal parenchymal neoplasms also present problems. Pineocytomas are characterized by a uniform population of round, dyscohesive cells with scant cytoplasm (Fig. 31.28). These resemble Oligodendroglioma on smear. Pineoblastomas are composed of very primitive, intermediatesized cells that tend to form cohesive groups. The cells have hyperchro-

matic nuclei and scanty cytoplasm and tend to show nuclear molding. Extensive necrosis and a high mitotic rate are common. In one of our cases, two separate stereotaxic biopsies showed both of these types of neoplasm (Fig. 31.29), representing a transitional form (33). Pituitary Adenoma Pituitary adenoma appears as a third ventricular lesion only when it is large and invasive. Such tumors are usually obvious on CT scan, and stereotaxic biopsy is rarely undertaken. On smears, cells from pituitary adenomas appear as small, dyscohesive round cells with delicate nuclei and variable amounts of cytoplasm. Those lesions that are invasive tend to have greater degrees of cytological atypia. Craniopharyngioma Craniopharyngioma is a frequent diagnostic consideration in this region. The cytological identification of a craniopharyngioma depends on the presence of squamous epithelial elements. Tissue from solid areas shows large cohesive sheets of benign squamous epithelium (Fig. 31.30). Considerable cytological atypia may be present. Smears from cystic areas show squames. Examination of fresh smears made from the oily fluid in polarized light for the presence of cholesterol crystals is a useful diagnostic maneuver. Calcification is very common, both in smears and in sections, and is helpful in making a diagnosis in a cystic lesion of the third ventricle. Correlation of the CT scan and clinical features precludes confusion with well-differentiated metastatic squamous carcinoma, which may be suggested when cytological atypia is present. Distinction from an epidermoid or dermoid cyst is impossible on the smear, although the CT appearance of a solid component to the mass would suggest craniopharyngioma (Fig. 31.31).

Colloid (Neuroepithelial) Cyst Colloid cysts are characterized by their epithelial lining, which is usually cuboidal or columnar and may be ciliated, and its gelatinous contents (Fig. 31.32). Stereotaxic biopsies from outside the wall show gliotic brain and choroid plexus. Smears from the cyst wall that we have made from

excised specimens show columnar epithelium that resemble choroid plexus epithelium. In one of our cases, there was marked cytological atypia and increased mitotic activity (Fig. 31.33). Histological examination of these cases showed the epithelial lining to be three to four layers thick without any invasive tendency. Smears of cyst contents do not have any diagnostic characteristics.

Other Cysts Epidermoid and dermoid cysts rarely occur in this region. Smears from the cyst interior and wall show anucleate squames and squamous epithelial cells. Distinction from craniopharyngioma needs clinical correlation. Arachnoid cysts rarely enter into consideration in the diagnosis of a cystic lesion in this region. Its content of Cerebrospinal fluid distinguishes it from other cysts. The arachnoidal cell lining of these cysts is difficult to identify on small biopsy specimens.

Malignant Lymphoma Primary lymphomas of the brain may occur in the region of the third ventricle. They are more frequently encountered since the onset of the current epidemic of acquired immune deficiency syndrome (AIDS). In these patients, the neurological mass lesion is frequently the first manifestation of AIDS. Smears from lymphomas show a monomorphous population of transformed lymphocytes. The most common subtype is immunoblastic sarcoma, the immunoblasts appearing as round, large cells with large nuclei showing prominent nucleoli and abundant cytoplasm. More rarely, the lymphoma is composed of small lymphocytes, frequently showing plasmacytoid features. These cases are unusual and difficult to differentiate from a reactive lymphocytic infiltrate associated with an inflammatory process. Smears to establish monoclonal staining for immunoglobulin markers are essential for diagnosis in these cases. Airdried smears are optimal for such marker studies. Inflammatory Lesions

Granulomas Granulomas present several problems in diagnosis. First, the smear frequently consists entirely of the central necrotic material which is not diagnostic. Second, the epithelioid cells are difficult to recognize as such in smears. When an admixture of lymphocytes is present, as is frequent, the recognition of the mass as inflammatory is easier. Third, the identification of a granuloma does not provide a specific diagnosis. We have identified Coccidiodes immitis spherules in the center of one granuloma and acid-fast bacilli in two others. Culture, however, is essential for diagnosis. In one case, the granulomatous inflammation was part of a pineal germinoma.

Toxoplasmosis Infection with Toxoplasma gondii has become common in patients with AIDS. The lesions are difficult to distinguish from malignant lymphoma clinically. Stereotaxic biopsy is the method of choice in most cases to establish a tissue diagnosis. Toxoplasmosis is usually characterized by a necrotizing encephalitis in which Toxoplasma pseudocysts can be identified. We have successfully identified pseudocysts on both frozen sections and smears. Pseudocysts of Toxoplasma appear as large, rounded structures containing numerous trophozoites. The diagnosis is suggested on the initial smear by the presence of necrosis, a polymorphous inflammatory infiltrate containing lymphocytes, plasma cells, and numerous foamy histiocytes (Fig. 31.34). If no trophozoites are identified on the smear, frozen sections, often multiple, should be performed (Fig. 31.35). Immunoperoxidase techniques for toxoplasma pseudocysts and trophozoites are useful.

Acute Encephalitis Herpes simplex encephalitis rarely enters the differential diagnosis of a mass lesion in the region of the third ventricle. It does, however, come

into consideration when the smear shows a mixed inflammatory reaction and no infectious agent is identified. In herpes simplex encephalitis, there are numerous transformed lymphocytes, cells with large eosinophilic intranuclear inclusions, and bi- and multinucleated giant cells. The binucleate cells resemble Reed-Sternberg cells. Immunoperoxidase techniques and electron microscopy are extremely useful in confirming the diagnosis.

Cysticercosis Cysticercal cysts commonly occur in the third ventricle and the suprasellar region. In the Los Angeles County Hospital, which serves a large Mexican population, this is a common lesion. Stereotaxic biopsies are commonly from the gliotic inflamed brain adjacent to the cyst. These show increased cells with astrocytes and inflammatory cells. The presence of large numbers of eosinophils in this area is highly suggestive of cysticercosis. Fragments of choroid plexus in the biopsy may cause confusion with epithelium-lined cysts. The cysticercal cyst is lined by a homogeneous structureless eosinophilic cuticle that is thrown into scallops. Immediately below this are numerous small basophilic bodies that resemble daryorrhectic nuclei. Calcification is common. Aspiration of cysticercal cyst contents produces a watery fluid in which small fragments of cuticle may be seen. This, however, is difficult to distinguish from artifact, and a diagnosis is rarely made on smears of cyst fluid.

Anesthetic Considerations The successful outcome of stereotaxic neurosurgical procedures is, to a large extent, dependent on the smooth relationship between analgesic/ amnestic neuroleptic local standby anesthesia and the effective maintenance of a precise and reproducible stereotaxic routine. The demands placed on the neuroanesthesiologist center around amnestic sedation, neuroleptic analgesia, and the continuous assessment of neurological function. Issues governing the attainment of these objectives are: (a) the effective preoperative priming of patient expectations and comprehension relating to the procedure, (b) development of isolated voice recognition as a channel of communication between anesthetist and patient, (c) identification of functional neurological implications posed by the specific abnormality, (d) selection of adequate but judicious premedication, (e) the use of a controlled sedation-titration technique throughout the procedure, (f) adequate airway access for rapid intubation should the extremely infrequent need for induction of general anesthesia arise, and (g) monitoring of neurological assessment including speech, cognition, sensorimotor lateralization, and level of consciousness. Appropriately managed local standby anesthesia permits a maximum of operative control as well as patient comfort and safety. In over 500 cases managed by stereotaxis from 1981 to 1986 at LAC-USC Medical Center, discrete anesthetic complications have not been encountered. Clinically manifest seizure activity occurred in 2 patients with mass lesions. The induction of general neuroanesthesia was electively undertaken in 10 preoperatively obtunded patients. There are a number of key stages to local anesthesia during a stereotaxic procedure that relate to the various operative, radiographic, and transport steps described in the previous section. Before the procedure, patient evaluation, preparation, and premedication must be established. Once in the operating room area where the base ring is to be applied, reliable venous access and stable physiological monitoring are attained. Conray contrast infusion is generally instituted in all CT-guided procedures at this point. Depending on the nature of the lesion, prior delayeddose contrast infusion may be utilized. Contrast dye hypersensitivity reactions must of course be anticipated, and one must be prepared to treat these appropriately. Institution of appropriate neuroleptic sedationtitration technique along with the surgeon's delivery of generous local

cutaneous analgesia is required before application of the stereotaxic reference base ring. Transport of the patient to and from the CT scanner and the operating room suite must be well planned from an anesthetic point of view to avoid delays in the flow of the stereotaxic routine. Continuation of analgesic/ amnestic sedation as well as attention to maintaining proper physiological monitoring are important during these stages. Once in the operating theatre, appropriately titrated anesthesia is continued. Before the actual operative phase when the skin incision, twist drill hole, and biopsy are to be performed, the surgeon again applies local cutaneous analgesia. At the conclusion of the procedure the anesthetist begins reversal of anesthesia about when the base ring is being removed. Patient monitoring is continued during transport to the recovery room area or the CT scanner for the standard postoperative CT scan. The relevant neurological aspects underlying a patient's particular diagnosis must be adequately ascertained preoperatively. One must consider the presence of any significant intracranial mass effect, intracranial hypertension, or hydrocephalus, as well as any underlying systemic problems (cardiopulmonary and metabolic stability). The presence of stability in the patient's psychological status as it may be related to modification of neuroleptic anesthesia at certain points during the procedure is of considerable importance. Patient education is particularly relevant in relation to this last point. With repetitive and reassuring explanations of the details of the stereotaxic technique one will develop an operantly trained participant. Such a patient will take an active and interested but calmly cooperative and hypokinetic role. Pharmacological agents useful for neuroleptic sedation-titration technique during stereotaxic procedures include benzodiazepams, narcotics, hypnotics, butyrophenones, and related agents, of which the most commonly employed exampes are: diazepam (Valium), meperidine (Demerol), fentanyl (Sublimaze), droperidol (Inapsine), alphaprodine (Nisentil), and hydroxyzine (Vistaril). Glucocorticoids and, at times, anticonvulsants or diuretics are used in the presence of mass lesions, cerebral edema, or elevated intracranial pressure. Antimicrobial agents may be appropriate during endoscopy or protracted procedures entailing multiple stereotaxic corridors or the implantation of foreign bodies (interstitial stereotaxic radioisotope implantation for tumor brachytherapy is one example). If one adheres to the basic practical strategies outlined here for effective neuroanesthetic management of the various CT-guided stereotaxic procedures, successful and efficient realization of neurosurgical objectives may be more reliably attained. In undertaking such a procedure where the patient plays such an active role and multiple detailed steps must be carried out with reproducible meticulous precision, the role of the anesthetist is vital. Patient cooperation and responses during the transcerebral probe transit and target point surgical manipulations enhance the safety of the method. Experience and Rationale for Utilization of the Technique With the use of these techniques in over 140 operations in the third ventricular region, procedural objectives as described in the previous section were realized in 98% of cases (histological and/or microbiological verification in 94%) with a mortality rate of less than 1 % and a significant complication rate of less than 2% (Figs. 31.36 to 31.41). Based on the broad spectrum of abnormalities in both the anterior and the posterior third ventricular regions as well as the generally restricted ability of

imaging and indirect diagnostic procedures to provide absolute histological or microbiological diagnosis, the technique of imaging-guided stereotaxis should be considered as an alternative initial procedure to craniotomy in cases where midline lesions seem to be of intrinsic neoplastic or infectious origin or potentially radiosensitive or when the diagnosis is in doubt (15, 17, 62, 63). This stereotaxic procedure provides diagnosis at low risk and offers a rational guide for proceeding to further therapy, which may include craniotomy, radiotherapy (teletherapy and/or brachytherapy), or chemotherapy (antimicrobial or antineoplastic). In the anterior and mid-third ventricular regions, cystic lesions such as colloid cysts or cysticercosis may be aspirated or excised with or without direct visualization by using a variety of cerebroscopic techniques (4, 6, 7, 14, 19, 39, 48, 54, 66, 69, 76, 79). In high risk patients or those with recurrent lesions, craniopharyngioma cysts may be aspirated and treated with colloid-suspended radionuclides (10, 11, 44). The posterior third ventricular and pineal region provides a site for a diverse spectrum of pathological processes and diverse neoplastic involvement (21-24, 32, 35, 36, 45-47, 60, 67, 72, 74, 75).

In spite of their complexity, the diverse spectrum of pineal region tumors may be simply classified into four major groups: germ cell tumors (germinomas and teratomas), pineal parenchymal cell tumors (Pineocytomas and pineoblastomas), tumors of supporting (glial stroma) or adjacent tissues and others, and nonneoplastic cysts and vascular lesions. Of the true pineal tumors, approximately 80% are of germ cell origin. Of these, 70% are germinomas and 30% are teratomas (60% benign, 40% malignant). Pineal cell tumors (Pineocytomas and pineoblastomas) account for approximately one-fifth of the lesions. Long term survival has been reported after the excision of benign and circumscribed lesions (30% of the entire spectrum) as well as after radiotherapy of germinomas (20, 73, 80, 81). Histologically malignant lesions in this region have a uniformly poor prognosis in spite of multimodality aggressive therapy. These

lesions include embryonal carcinoma, endodermal sinus tumor, and choriocarcinoma in pure form or as a component of mixed lesions including other germinal components. Approximately 35% of germ cell tumors are comprised of mixed elements. Computerized radiographic imaging defines the presence and general extent of this spectrum of pathological lesions in the majority of cases. Most commonly, masses present as high density lesions with contrast enhancement. Although radiographic imaging is suggestive of a diagnosis in many cases (especially teratomas and cysts), strict histological appraisal is not uniformly obtainable. The presence of biological tumor markers in the serum or Cerebrospinal fluid may indicate a functioning cellular component of these tumors (3, 7, 8, 30, 34). In general, the functioning tumors have a poor prognosis. Beta chain human chorionic gonadatropic hormone is considered indicative of the presence of a choriocarcinoma or a mixed germ cell tumor with choriocarcinomatous elements. Alpha-fetoprotein in spinal fluid or

serum is considered indicative of an endodermal sinus tumor or a mixed germ cell tumor with endodermal sinus elements. Cerebrospinal fluid cytology assessment may indicate the presence of malignant cells, but is of limited overall practical value because of the lack of consistency of response (less than 20% in potentially disseminating neoplasms). With the complexity of histological types of lesions that may evolve there is little doubt that adequate histological diagnosis should be the

initial objective in patient management. Stereotaxic biopsy provides such data which, when combined with imaging studies and biological marker assays, may be used to direct further therapy, which may include operation, whole neural axis radiation, or polychemotherapy (1, 13, 16, 20, 27, 43, 49, 58, 59, 61, 64, 71, 77, 78, 80-82). Considering the histological spectrum of tumors, diagnostic capability of imaging devices, availability of biological markers, current imagingguided stereotaxic methods, and effectiveness of therapeutic modalities, it seems that direct surgical approaches should be reserved for benign potentially excisable lesions and malignant and mixed germinal tumors, where multimodality therapy offers the best opportunity for palliation. With techniques of imaging-directed stereotaxis it is possible to select patients who may derive the major operative benefit before undertaking a direct surgical approach.

References 1. Abay EO, Laws ER Jr, Grado GL: Pineal tumors in children and adolescents. JNeurosurg 55:889-895, 1981. 2. Adams JH, Graham DI, Doyle D: Brain Biopsy: The Smear Technique for Neurosurgical Biopsies. Philadelphia, JB Lippincott Co, 1981. 3. Allen JC, Nisselbaum J, Epstein F,: Alpha-fetoprotein and human chorionic gonadotropin determination in Cerebrospinal fluid. J Neurosurg 51:368374, 1979. 4. Apuzzo MLJ, Chandrasoma PT, Zelman V, Giannotta SL, Weiss MH: Com puted tomographic guidance stereotaxis in the management of lesions of the third ventricular region. Neurosurgery 15:502-508, 1984. 5. Apuzzo MLJ, Jepson JH, Luxton G, Little FM: Ionizing and nonionizing

6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32.

radiation treatment of malignant cerebral gliomas: Specialized approaches. Clin Neurosurg 31:470-496, 1984. Apuzzo MLJ, Sabshin JK: Computed tomographic guidance stereotaxis in the management of intracranial mass lesions. Neurosurgery 12:277-285, 1983. Apuzzo MLJ, Heifetz MD, Weiss MH, Kurze T: Neurosurgical endoscopy using the side-viewing telescope: Technical note. J Neurosurg 46:398-400, 1977. Arita N, Bitoh S, Ushio Y: Primary pineal endodermal sinus tumor with elevated serum and CSF alphafetoprotein levels. J Neurosurg 53:244-248, 1980. Backlund EO, Granberg PO, Hamberger B: Transplantation of adrenal med ullary tissue to striatum in parkinsonism. J Neurosurg 62:169-173, 1985. Backlund E-O: Studies on craniopharyngiomas: III. Stereotaxic treatment with intracystic yttrium-90. Acta Chir Scand 139:237-247, 1973. Backlund E-O: Studies on Craniopharyngiomas: Stereotaxic treatment with radiosurgery. Acta Chir Scand 139:344-351, 1973. Bjornsson J, Schiethauer BW, Okazak H, Leech RW: Intracranial germ cell tumors: Pathobiologic and immunohistochemical aspects of seventy cases. J Neuropathol Exp Neurol (in press). Bloom HJG: Primary intracranial germ cell tumors. Clin Oncol 2:233-257, 1983. Borit A: Embryonal carcinoma of the pineal region. J Pathol 97:165-168, 1969. Bosch DA: Indications for stereotactic biopsy in brain tumours. Acta Neurochir(Wien) 54:167-179, 1980. Chapman PH, Linggood RM: The management of pineal area tumors: A recent reappraisal. Cancer 46:1253-1257, 1980. Conway LW: Stereotaxic diagnosis and treatment of intracranial tumors including an initial experience with cryosurgery for pinealomas. J Neurosurg 38:453-460, 1973. Crain BJ, Bigner SH, Johnston WW: Fine needle aspiration biopsy of deep cerebrum: A comparison of normal and neoplastic morphology. Acta Cytol (Baltimore) 26:772-778, 1982. Crue BL, Lasby V, Kenton B, Felsoory A: Needle scope attached to stereotactic frame inspection of cisterna magna during percutaneous radiofrequency trigeminal tractomy. Appl Neurophysiol 39:58-64, 1976/77. Danoff B, Sheline GE: Radiotherapy of pineal tumors. In Neuwelt EA (ed): Diagnosis and Treatment of Pineal Region Tumors. Baltimore, Williams & Wilkins, 1984, pp 300-308. Dayan AD, Marshall AHE, Miller AA: Atypical teratomas of the pineal and hypothalamus. J Pathol Bacteriol 92:1-28, 1966. DeGirolami U, Schmidek H: Clinicopathological study of 53 tumors of the pineal region. J Neurosurg 39:455-462, 1973. Derek R, Jenkin T, Simpson JK, Keen CW: Pineal and suprasellar germino mas. J Neurosurg 48:99-107, 1978. Donat JF, Okazaki H, Gomez MR: Pineal tumors: A 53 year experience. Arch Neurol 35:736-740, 1978. Eisenhardt I, Cushing H: Diagnosis of intracranial tumors by supravital technique. Am J Pathol 6:541-552, 1930. Fulling KH, Nelson JS: Cerebral astrocytic neoplasms in the adult: Contri bution of histologic examination to the assessment of prognosis. Semin Diagnostic Pathol 1:152-163, 1984. Griffin BR, Griffin TW, Tong DYK: Pineal region tumors: Results of radiation therapy and indications for elective spinal irradiation. Int J Radiat Oncol Biol Phys 7:605-608, 1981. Gutin PH, Philips TL, Hosobuchi Y: Permanent and removable implants for the brachytherapy of brain tumors. Int J Radiat Oncol Biol Phys 7:13711381, 1981. Gutin PH, Phillips TL, Wara WM, Leibel SA, Hosobuchi Y, Leven VA, Weaver KA, Lamb S: Brachytherapy of recurrent malignant brain tumors with re movable high-activity iodine-125 sources. J Neurosurg 60:61-68, 1984. Hasse J, Neilsen K: Value of tumor markers in the treatment of endodermal sinus tumors and Choriocarcinomas in the pineal region. Neurosurgery 5:485-488, 1979. Heilbrun MP, Roberts TS, Apuzzo MLJ, Wells TH, Sabshin JK: Preliminary experience with Brown-Roberts-Wells computerized tomographic stereotaxic guidance system. J Neurosurg 59:217-222, 1983. Herrick MK: Pathology of pineal tumors. In Neuwelt EA (ed): Diagnosis and

33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. 46. 47. 48. 49.

50. 51. 52. 53. 54. 55. 56. 57. 58. 59.

Treatment of Pineal Region Tumors. Baltimore, Williams & Wilkins, 1984, pp 31-60. Herrick MK, Rubinstein LJ: The cytological differentiating potential of pineal parenchymal neoplasms (true pinealomas). Brain 102:289-320, 1979. Hill S, Martin E, Ellison EC, Hunt WE: Carcinoembryonic antigen in Cerebro spinal fluid of adult brain-tumor patients. J Neurosurg 53:627-632, 1980. Hitchon PW, Abu-Yousef MM, Graf CJ: Management and outcome of pineal region tumors. Neurosurgery 13:248-253, 1983. Ho K-L, Rassekh ZS: Endodermal sinus tumor of the pineal region: Case report and review of the literature. Cancer 44:1081-1086, 1979. Hoffman HJ, Harwood-Nash D, Gilday DL: Percutaneous third ventriculos tomy in the management of noncommunicating hydrocephalus. Neurosur gery 7:313-321, 1980. Hosobuchi Y, Philips TL, Stupar ТА, Gutin PH: Interstitial brachytherapy of primary brain tumors: Preliminary report. J Neurosurg 53:613-617, 1980. Iizuka J: Development of a stereotaxic endoscopy of the ventricular system. ConfinNeurol 37:141-149, 1975. Jane JA, Bertrand G: A cytological method for the diagnosis of tumors affecting the central nervous system. J Neuropathol Exp Neurol 21:400409, 1962. Kandel El, Peresedov W: Stereotaxic evacuation of spontaneous intracerebral hematomas. J Neurosurg 62:206-213, 1985. Kelly PJ, Kail BA, Goerss S: Transposition of volumetric information derived from computed tomography scanning into stereotactic space. Surg Neurol 21:465-471, 1984. Kirshner JJ, Ginsberg SJ, Fitzpatrick AV, Comis RL: Treatment of a primary intracranial germ cell tumor with a systemic chemotherapy. Med Ped Oncol 9:361-365, 1981. Kobayashi T, Kageyama N, Ohara K: Internal irradiation for cystic cranio pharyngioma. J Neurosurg 55:896-903, 1981. Kobayashi T, Kageyama N, Kida Y: Unilateral germinomas involving the basal ganglia and thalamus. J Neurosurg 55:55-62, 1981. Koide O, Watanabe Y, Sato K: A pathological survey of intracranial germi noma and pinealoma in Japan. Cancer 45:2119-2130, 1980. Komrower GM: Precocious puberty in association with pineal seminoma. Arch Dis Child 49:822, 1974. Kramer S, Soughard M, Mansfield CM: Radiotherapy in the management of craniopharyngiomas: Further experiences and late results. AJR 103:44-52, 1968. Laws ER, Abay EO, Forbes GS, Grado GL: Conservative management of pineal tumors—Mayo Clinic experience. In Neuwelt EA (ed): Diagnosis and Treatment of Pineal Region Tumors. Baltimore, Williams & Wilkins, 1984, pp 323-331. Leksell L: The stereotaxic method and radiosurgery of the brain. Acta Chir Scand 102:316-319,1951. Leksell L, Leksell D, Schwebel J: Stereotaxis and nuclear magnetic reso nance. J Neurol Neurosurg Psychiatry 48:14-18, 1985. Liwnicz BH, Henderson KS, Masukawa T, Smith RD: Needle aspiration cytol ogy of intracranial lesions: A review of 84 cases. Acta Cytol (Baltimore) 26: 779-786, 1982. Lunsford LD, Latchaw RE, Vries JK: Stereotactic implantation of deep brain electrodes using computed tomography. Neurosurgery 13:280-286, 1983. Mixter WJ: Ventriculoscopy with puncture of the floor of the third ventricle. Med Surg J 188:277-278, 1923. Morris AA: The use of the smear technique in the rapid histological diagnosis of tumors of the central nervous system: Description of a new staining method. J Neurosurg 4:497-504, 1947. Mundinger F, Hoefer T: Protracted long-term irradiation of inoperable midbrain tumors by stereotactic curie-therapy using iridium-192. Acta Neurochir(Wien) 21:93-100, 1974. Mundinger F: Implantation of radioisotopes (curie-therapy). In Schaltenbrand G, Walker AE (eds): Stereotaxy of the Human Brain. New York, ThiemeStratton, Inc., 1982. Neuwelt EA, Frenkel EP, Smith RG: Suprasellar germinomas (ectopic pineal omas): Aspects of immunological characterization and successful chemotherapeutic responses in recurrent disease. Neurosurgery 7:352-358, 1980. Neuwelt EA, Frenkel EP: Germinomas and other pineal tumors: Chemother-

60. 61. 62. 63. 64. 65. 66. 67. 68. 69. 70. 71. 72. 73. 74. 75. 76. 77. 78. 79. 80. 81. 82. 83.

apeutic responses. In Neuwelt EA (ed): Diagnosis and Treatment of Pineal Region Tumors. Baltimore, Williams & Wilkins, 1984, pp 332-343. Obrador S, Soto M, Gurierrez-Diaz JA: Surgical management of tumours of the pineal region. Ada Neurochir (Wien) 34:159-171, 1976. Onoyama Y, Ono K, Nakajima T, Hiraoka M, Abe M: Radiation therapy of pineal tumors; Radiology 130:757-760, 1979. Ostertag CB, Mennel HD, Kiessling M: Stereotactic biopsy of brain tumors. SurgNeurol 14:275-283, 1980. Pecker J, Scarabin J-M, Vallee B: Treatment in tumours of the pineal region: Value of stereotaxic biopsy. Surg Neurol 12:341-348, 1979. Pertuiset B, Visot A, Metzger J: Diagnosis of pinealoblastomas by positive response to cobalt therapy. Acta Neurochir (Wien) 34:151-152, 1976. Picard C, Oliviewr A, Bertrand G: The first human stereotaxic apparatus: The contribution of Aubrey Mussen to the field of stereotaxis. J Neurosurg 59:673-676, 1983. Powell MP, Torrens MJ, Thomas JLG: Isodense colloid cysts of the third ventricle: A diagnostic and therapeutic problem resolved by ventriculoscopy. Neurosurgery 13:234-237, 1983. Preissig SH, Smith MT, Huntington HW: Rhabdomyosarcoma arising in a pineal teratoma. Cancer 44:281-284, 1979. Rekate HL, Ruch T, Nulsen FE: Needle biopsy of tumors in the region of the third ventricle. J Neurosurg 54:338-341, 1981. Rivas JJ, Lobato RD: CT-assisted stereotaxic aspiration of colloid cysts of the third ventricle. J Neurosurg 62:238-242, 1985. Russell DS, Krayenbuhl H, Cairns H: The wet film technique in the histolog ical diagnosis of intracranial tumors: A rapid method. J Pathol 45:501-505, 1937. Salazar OM, Castro-Vita H, Bakos RS, Feldstein ML: Radiation of tumors of the pineal region. Int J Radiat Oncol Biol Phys 5:491-499, 1979. Sambasivan M, Nayar A: Epidermoid cyst of the pineal region. J Neurol Neurosurg Psychiatry 37:1333-1335, 1974. Sano K, Matsutani M: Pinealoma (germinoma) treated by direct surgery and postoperative irradiation. Childs Brain 8:81-97, 1981. Sheithauer BW: Neuropathology of pineal region tumors. Clin Neurosurg 32:351-383, 1985. Solarski A, Panke ES, Panke TW: Craniopharyngioma in the pineal gland. Arch Pathol Lab Med 102:490-491, 1978. Sugita K, Mutsuga N, Takoaka Y, Hirota T, Shibuya M, Doi T: Stereotaxic exploration of para-third ventricle tumors. ConfinNeurol 37:156-162, 1975. Takakura K: Nonsurgical pineal tumor therapy—the Japanese experience. In Neuwelt EA (ed): Diagnosis and Treatment of Pineal Region Tumors. Baltimore, Williams & Wilkins, 1984, pp 309-322. Ueki K, Tanaka R: Treatments and prognoses of pineal tumors—experience of 110 cases. Neurol Med Chir 20:1-26, 1980. Vries JK: An endoscopic technique for third ventriculostomy. Surg Neurol 9: 165-168. Wara WM, Jenkin RDT, Evans A, Ertel I: Tumors of the pineal and supra sellar region: Childrens Cancer Study Group treatment results 1960-1975. Cancer 43:698-701, 1979. Wara WM, Fellows CF, Sheline GE: Radiation therapy for pineal tumors and suprasellar germinomas. Radiology 124:221-223, 1977. Weigel K, Ostertag CB, Mundinger F: Interstitial long-term irradiation of tumors in the pineal region: Stereotactic cerebral irradiation. Inserm Symp No. 12:283-292, 1979. Willems J-S: Aspiration biopsy cytology of tumors of the central nervous system and base of the skull. In Linsk JA, Franzen S (eds): Clinical Aspira tion Cytology. Philadelphia, Lippincott, 1983.

Commentary E Contemporary European Contributions to Neurosurgical Stereotaxy Roger I. von Hanwehr, M.D.

Significant contributions within the field of stereotaxic neurosurgery have emerged from the European sector, especially since the advent of contemporary computerized tomography (CT), beginning in the early 1970s. The actual and potential application of stereotaxic technology in the diagnosis and therapeutic management of mass lesions either impacting on deep midline/ paramedian neural axis structures within the cranial vault or directly situated throughout the diencephalic region has been readily apparent in original work emanating from Sweden, France, the United Kingdom, Spain, Italy, the Federal Republic of Germany, and the Benelux countries. Diverse areas of progress within the field of neurosurgical stereotaxy have included: (a) development of various CT-guided stereotaxic systems; (b) numerous diagnostic biopsy techniques; (c) localization of deep-seated cystic, neoplastic, or vascular lesions followed by often innovative surgical management effected via stereotactically defined access corridors; (d) interstitial radioisotope implant brachytherapy; (e) high energy lesion radiosurgery; (f) miscellaneous but potentially applicable adaptions of stereotaxy ranging from cryosurgery to assessment of deoxyribonucleic acid (DNA) distribution patterns within gliomas; and (g) more recently the utilization of precise implantation methods intended for attempts at microtransplantation within the central nervous system (CNS). The evolution of CT-guided stereotaxic technology has involved the utilization and adaption of numerous systems with individual variations between differing national settings. The current British contribution has centered on both the Bennett stereotaxic frame with transfer of CTdefined target coordinates via the Leksell method and the Brown-Robert-Wells stereotaxic apparatus, which permits direct CT-guided stereotaxic intervention (76, 78). In contrast, stereotaxic systems used in Belgium and France (Suetens and Talairach methods, respectively) rely on indirect transfer of CT-reconstructed or cerebral angiographic data to provide multidimensional spatial definition of target lesions, including the

potential availability of stereoscopic imaging and simulation of biopsy probe/electrode positioning (72). An Italian team (Colombo/Dettori) has developed reliable techniques for CT control of diencephalic lesions in functional stereotaxis as well as a novel adaption of the Reichert-Mundinger apparatus, the Siemens Siretom 2000 CT scanner, and a rigidly attachable calvarial coordinate system, thus permitting combined or separate use of CT, angiographic, and encephalographically derived targeting data (22, 26). Various German groups have contributed toward the innovation of more elegant stereotaxic intervention, primarily relying on the Reichert-Mundinger device coupled to CT imaging technology during stereotaxic biopsy and interstitial brachytherapy procedures (9, 10, 47, 71). The development of various base ring, phantom coordinate grid, and stereotaxic coordinate assemblies has been extensively detailed in conjunction with computer-assisted direct axial tomography reference point and target point guidance (9, 10, 47, 50, 71). Digital radiographic systems have also been developed in conjunction with CT-guided stereotaxic systems (31), whereas new targeting devices within existing stereotaxic assemblies allow constant CT guidance during a given procedure in conjunction with universal reference modes that facilitate multiple entry and target points as well as varied transit trajectory settings (32). Although the Leksell stereotaxis unit has remained the cornerstone of the leading Swedish impact on this subdiscipline of neurosurgery (38), more contemporary stereotaxic instrument systems such as that of Boethius allow direct data interface between the surgical stereotaxic assembly and various diagnostic neuroradiological (CT, positron emission tomography (PET), magnetic resonance imaging (MRI), etc.) or therapeutic (linear accelerators, stereotaxic "gamma knife" radiosurgery units, cryosurgery devices, etc.) technology (12, 14). Other Swedish groups have provided valuable ancillary insight into the refinement of CT data acquisition for the purposes of ensuring accurate stereotaxic coordinate determination during surgical procedures (8, 75).

The major impact of increasingly accurate CTguided biopsy techniques is apparent in the Swedish as well as the overall European stereotaxic experience. Willems et al. demonstrated highly reliable (87 to 88%) cytodiagnostic accuracy in 112 stereotaxic aspiration biopsies of benign and malignant neoplasms in part located in sellar, suprasellar, diencephalic, and pineal locales (80). Edner similarly reported 94 stereotaxic puncture biopsies in cases of sellar, parasellar, and suprasellar lesions, of which 28 were CT-guided (29). Various therapeutic options including cyst aspiration (in the case of craniopharyngiomas), intracystic radioisotope instillation, etc., were included in the stereotaxic series, with the initial diagnostic impression gained from a stereotaxic intervention guiding eventual therapeutic action. Utilizing the Leksell stereotaxic apparatus, the same author detailed 345 biopsy procedures including 84 directed at mostly deep-seated sellar, diencephalic (third ventricular/foramen of Monro), pineal, basal ganglia, and hypothalamic cystic lesions, as identified with enhanced CT imaging in roughly half of these cases (30). Twenty-two solid lesions in the identical anatomical distributions were also encountered. Using enhanced CT data, pathological diagnostic accuracy approached 90% with a mortality of less than 1 % and a morbidity of 2.3%. Cystic lesions encountered in the suprasellar/diencephalic domain included craniopharyngiomas, cystic gliomas, and arachnoidal as well as colloidal cysts. Solid lesions of the third ventricle and pineal regions included astrocytomas, pineoblastomas, germinomas, and ependymomas as well as a rare dermoid and teratomatous lesion. Other Swedish teams have developed related issues relating to contrast enhancement techniques/patterns and CT criteria that aid in the effective lesion localization, definition, and target evaluation essential in effective stereotaxic diagnosis of deep midline glial neoplasms (44, 45). In Belgium (Waltregny) and in the Netherlands (Bosch), further experience with stereotaxic biopsy has resulted in clarification of specific indications (a) for serial stereotaxic biopsies where extensive pathological and structural/anatomical assessment of tumor constituents and volume is deemed appropriate and (b) for stereotaxic biopsies designed to establish histopathological diagnosis permitting nonsurgical management of extremely radiosensitive deep midline lesions (dysgerminomas) or conversely stereotaxic evacuation of cystic diencephalic lesions (third ventricular colloid cyst) (15, 79).

The reliability of stereotaxic biopsies from a diagnostic standpoint has been the focal impetus of the Italian contribution to this application of neurosurgical stereotaxy. Scerrati et al. found a near perfect reliability of diagnostic information in the stereotaxic biopsy of intracranial spaceoccupying lesions, whereas de Divitis cited an accuracy of 81 to 92% in the evaluation of 64 cases including 27 neuroglial tumors representing the lower margin of diagnostic accuracy (24, 65). Broggi specifically reported a small series of 17 pediatric deep-seated neoplastic lesions (including third ventricular, suprasellar, and pineal masses) where a CT-guided stereotaxic Reichert apparatus afforded a highly accurate and varied compendium of tissue diagnoses in part obtained through imprint cytological techniques (19). The same author separately cited a diagnostic accuracy of 85% in an overall series of 200 biopsies utilizing varied transit trajectory options, tissue impedance, and depth electroencephalogram recording techniques all aimed at maximizing stereotaxic localization and three-dimensional topographic definition of the mass target lesion involved (18). Lobato, from Spain, provided additional data on 28 diencephalic and deep-seated midline lesions which, in some cases, underwent CT-guided stereotaxic biopsy with an overall diagnostic accuracy of 89% (10- X 1.2-mm specimens harvested utilizing a Backlund spiral needle) and a complication rate of 8%, including 7% with transiently increased neurological deficits and a 1 % incidence of postbiopsy intracerebral hematoma (46). Although the French contribution to CT guidance in stereotaxic biopsy procedures remains limited because of continued reliance on stereotaxic angiography (Talairach system), a large series reported by Sedan (318 biopsies) detailed criteria for limiting potential complications, including less than 4 biopsy trajectory passes per procedure, decreasing biopsy sampling frequency, and judicious selection of patients as well as stereotaxic entry point to target point transit route (66). A fairly broad experience with stereotaxic biopsy techniques was also described by Daumas-Duport, with nearly exclusive emphasis on correlation between CT findings and histological configuration, which is presented as an important factor in the choice of successful trajectory routes during stereotaxic biopsies (27, 28). With reference to stereotaxic biopsies directed at deep midline and diencephalic target mass lesions, the German experience of the Freiburg group (Mundinger, Birg, Ostertag, Kiessling, and

Kleihues) remains the dominant contribution (36, 48, 60). Direct CT stereotaxis utilizing the Mundinger/Birg computer-compatible version of the Reichert/Mundinger stereotaxic apparatus has been refined to the point where, given certain reference relationships with the stereotaxic device and patient on the one hand and the CT scan gantry on the other hand, the CT screen and stereotaxic apparatus coordinates are identical, thus obviating any need for a reference frame or conversion system (48). This permits biopsies or implantation corridors to be effected in a given patient at exact and reproducible targets and trajectories. In over 600 patients, biopsies by the Freiburg stereotaxic team yielded accurate tumor diagnoses and approximate grading (combined cytological/histological evaluation) in 82% of cases (36). Immunohistochemical tumor marker and cytoskeleton protein identification may further enhance this rate of diagnostic accuracy. In stereotaxic biopsies of 302 deep-seated brain tumors by the same group, 30% of lesions were situated in the basal ganglia, 36% in the deep cerebral hemispheres, 13% in the parapineal region, and 21% in the diencephalic/suprasellar regions (60). A wide and comprehensive scope of pathological diagnoses cover the full spectrum of the patient population reported, but most of the lesions were gliomas (71%). Operative mortality was 2.3% with a 3% incidence of transient neurological deterioration within the overall series (60). In nearly all cases of complications, an intracerebral hematoma was the offending postbiopsy development. Biopsy samples with an average volume of 1 mm3 were routinely obtained with a special stereotaxic microbiopsy forceps (diameter 0.8 mm), and one or more samples were harvested at intervals of 5 to 10 mm within the transit pathway (60). Local anesthesia was routinely utilized. Stereotaxic biopsy led to therapeutic management decisions involving interstitial radioisotope implant brachytherapy in 74% of cases and further surgical intervention (stereotaxic or conventional) or external irradiation in 26% of cases (60). Innovations in operative stereotaxis other than biopsy-directed procedures are extensive throughout the European stereotaxic experience. Operative stereotaxic CT-guided management is usually based on valuable diagnostic information gained during the biopsy procedure. In addition, the stereotaxic approach offers an accurate means to localize the target lesion via a safe, selected transit corridor, while stereotaxic data also provide further information defining the stereoscopic three-dimensional topography of a

given lesion if the appropriate CT-guided imaging and stereotaxic system is utilized (64). Stereotaxic treatment of deep-seated diencephalic/ midline lesions may involve interstitial radioisotope implantation (brachytherapy), intracavitary Curie therapy utilizing radiocolloid agents, direct drainage of cystic structures (arachnoid cysts, colloid cysts, neoplastic cysts, abscesses, etc.), stereotaxic ventriculostomy or ventriculocisternostomy (cystic craniopharyngioma, arachnoid cyst), direct instillation of immunotherapeutic agents (interstitial interferon), and combined open-stereotaxic procedures for a direct surgical approach to small deep-seated neoplasms or vascular malformations (53). One particularly germane operative application of stereotaxy relates to diencephalic access with a minimum of risk in the case of colloid cyst aspiration techniques that have been reported from Sweden (Bosch, Rahn, Backlund) and Spain (Rivas, Lobato) (17, 62). The former group has reported four cases of stereotaxic third ventricular colloid cyst aspiration utilizing air ventriculographic or CT guidance and employing cannula probes 0.6 to 1.5 mm in inside diameter (ID) for cyst puncture. The volume yield of the viscous aspirates averaged 0.3 to 1 ml and this type of specimen allowed accurate cytological diagnosis. Two of four cases demonstrated no cyst recurrence on follow-up CT evaluation, whereas the remaining two cases demonstrated cyst remnants and a decreased cyst size, respectively (17). The report by Rivas et al. again cited three cases with stereotaxic colloid cyst aspiration via a Leksell-Jernberg CTadapted stereotaxic frame (62). In two of these three cases, aspirates of 1.8 and 40 ml were obtained utilizing 1.2- to 1.8-ID probe cannulas with an internal spiral biopsy device designed to enter through the cyst capsule. Cytological diagnosis was confirmatory with periodic acidSchiff-positive amorphous eosinophilic colloid cyst contents being present. Concomitant hydrocephalus was reduced or resolved along with the complete disappearance of both cysts. The third patient required repeat stereotaxic aspiration with a 1.8-mm cannula yielding 2 ml of viscous material, and again the postoperative course was eventually characterized by resolution of the cyst (62). Thus, stereotaxic approaches to third ventricular cysts of this type may represent a viable and potentially less risky alternative to conventional open surgical management techniques. Utilization of stereotaxic methods for the evacuation of deep-seated intracerebral hematomas has similarly been reported by numerous authors including Kandel (U.S.S.R.) and Broseta (Spain),

in both instances employing some modification of the Backlund/von Hoist Archimedes spiral cannula screw originally designed for stereotaxic hematoma extrication (20, 34). Hematoma volume and localization is determined from the CTderived stereotaxic coordinate data, and multiple evacuation access corridors are stereotaxically planned (34). Catheters with endpoint Silastic balloons have been used for cortical hemostasis after hematoma removal. These are inserted down the stereotaxic transit access and inflation is maintained at the pressure level of CSF in the contralateral ventricle (34). Kandel reported this procedure with documentation for 32 patients with deep-seated spontaneous hemorrhages including 30 hypertensive cases with an overall mortality of 20% and a 16% rebleeding rate (34). With appropriate selection of patients and enhanced knowledge regarding optimal time for hematoma evacuation after the acute ictus, stereotaxic hematoma evacuation may emerge as a useful surgical adjunct to open approaches in those cases deemed appropriate for intervention. A major issue that remains unresolved relates to the impact of complete versus incomplete hematoma removal on the risk of rebleeding after a stereotaxic hematoma evacuation (34). Broseta's experience with 16 spontaneous intracerebral hematomas presented similar data and raised identical questions (17). Of interest is the role of acute cortical edema after hematoma decompression and removal, as well as the pattern of intracranial pressure dynamics, which in the case of Brosetta's series remained decreased compared with the uniform preoperative intracranial hypertension in these patients (17). Although numbers of patients studied do not permit an accurate statistical assessment, Kandel reported the majority of patients representative of postoperative mortality (especially those related to rebleeding) underwent early operation within 3 days of the initial insult (34). Thus, delayed aspiration of hematomas deserves consideration although complications in this category related to pulmonary embolism were also observed, thus clouding resolution of this issue. This stereotaxic methodology has been adapted for the evacuation of intracerebral clots stemming from rupture of aneurysmal lesions or vascular malformations. Furthermore, Kandel has detailed experimental stereotaxic technology designed for angiographically guided clipping of arterial and arteriovenous aneurysmal lesions (33). Of more immediate practical value is the development of combined stereotaxic localization and access corridor placement followed by ster-

eotaxically directed microsurgical approaches to and excision of deep-seated midline arteriovenous malformations that present difficult surgical challenges by virtue of small dimensions or anatomical location in the paraventricular diencephalic region. De Sola detailed such a case (25) successfully managed with the Talairach angiography-stereotaxy apparatus, and a substantial potential for the utilization of combined CT and angiographically guided stereotaxic surgery of treacherous vascular malformations has progressively emerged. A comprehensive survey of stereotaxic radioisotope implantation brachytherapy is beyond the scope of this review. Extensive and internationally recognized innovation in this field has emerged from Germany, as well as from France and Sweden. Various series detailing a multitude of clinical experiences with brachytherapy of intracranial deep-seated neoplasms have emerged in recent years, including reports of stereotaxic intracavitary yttrium-90 and rhenium-186 radiocolloid irradiation of cystic craniopharyngiomas (Netzeband, 1984) and of active glioma cysts (Szikla, 1984), as well as combined stereotaxic interstitial and external radiation therapy advances directed at therapy of deep-seated glial neoplastic lesions (Rougier, 1984; Szikla, 1984; Mundinger and Weigel, 1984) (57, 58, 63, 73, 74). By far the most voluminous accumulation of experience in the subdiscipline of stereotaxic glioma brachytherapy rests with the noted Friburg group of Mundinger et al. Experimental data on the early and late sequential morphological effects of iridium-192, yttrium-90, and iodine125 gamma and beta interstitial radionuclide source implants in the brain are fairly unique in the international literature (35, 59, 61). The substantial clinical experience with treatment of intracranial glial neoplasms has been equally well detailed (21, 49, 51-56). In one series of patients compiled between 1965 and 1983, 329 deepseated midline/diencephalic region (mesencephalon excluded) lesions were treated with various stereotaxic interventions, and 168 patients remained alive at the conclusion of the retrospective review period. Of these, 185 patients underwent stereotaxic iridium-192 (91 patients) or iodine-125 (94 patients) curiebrachytherapy of their neoplastic lesions, with 34 and 54 patients respectively, remaining alive at the conclusion of the review period (57). A related study provided a spectrum of 28 diencephalic neoplasms subjected to iridium-192 or iodine-125 interstitial radiotherapy from among a total population of 251 intracranial tumors. Three- and 5-year sur-

vival rates with this form of therapeutic management for diencephalic lesions are 52% and 37%, respectively, whereas the overall 5-year survival rate for malignant astrocytomas (irrespective of intracranial anatomical site) is 47% in patients receiving interstitial implant brachytherapy (53). Various forms of high dose short range and long term graduated dose implant techniques are utilized in the treatment of hypothalamic and diencephalic gliomas, and these include procedures with stereotaxic permanent iridium-192 implantation under bioptic control (51). The integration of CT imaging with stereotaxy has permitted topographic definition of tumor mass, resulting in enhanced data useful for planning radioisotope implantation (53). The trend of Mundinger et al. toward the use of iodine-125 reflects the preference for a radioisotope source with soft (28 to 35keV) radiation (desirable from the standpoint of radiation protection for medical personnel) output, a steep dose curve fall off permitting tightly controlled isqdose curve configurations, a favorable half-life time factor of 60 days, and a sufficiently high initial dose delivery to the tumor, thus guaranteeing a sufficient cumulative focal dose (51, 53, 56). These factors, in addition to low intensity, long term permanent implantation techniques, allow a dosing sequence extending over 7 months in some cases. Under these circumstances the slow dose accumulation tends to spare viable neuraxis tissue, while resulting in a steady increase in radiation sensitivity on the part of neoplastic components, where cellular fatality is more probable in abnormal cells exhibiting a premitotic pause than in normal cellular elements (51). In contrast to the theoretical approach to tumor radiotherapy offered by stereotaxic radioisotope implant brachycurietherapy, stereotaxic radiosurgery delivers a singularly decisive radiotherapeutic/destructive impact to the neoplastic target lesion. The latter method is equally applicable to deep-seated tumors. Ever since the first pioneering ventures into stereotaxic neuroradiosurgery by Professor Lars Leksell, Sweden has remained synonymous with state of the art stereotaxic technology (38, 41). The development and scope of stereotaxic neuroradiosurgery in its evolution from the original utilization of stereotaxic methodology for open surgical procedures involving radiofrequency probes (38), to synchrocyclotron and linear accelerator heavy particle/proton beam closed stereotaxic radiosurgery (37, 42), and finally to the presently operational second and third generation gamma knife units (41, 40) remains beyond

the scope of this discussion. Nevertheless, with regard to deeply situated neoplastic mass lesions such as craniopharyngiomas, as well as small vascular malformations, the gamma knife has assumed an increasingly significant therapeutic role as an adjunct to more conventional modes of surgical treatment (3, 69, 70). As of 1983, over 760 patients had undergone stereotaxic radiosurgical treatment for mass lesions and other indications, in all cases utilizing the cobalt-60 gamma knife unit along with various methods of stereotaxic radiological localization (41). Twenty-two craniopharyngiomas and 204 vascular malformations were treated from 1968 to 1982 and are included in this series. In addition, between 1966 and 1980, a larger series of craniopharyngiomas exhibiting a predominantly cystic component (Backlund et al., 1980) was treated with stereotaxic injections into the tumor cyst, utilizing yttrium-90 radiocolloid in all 54 cases (2, 41). Of these 54 patients, 34 had not previously undergone treatment, surgically or otherwise, and no recurrences were reported. The overall Swedish experience with radiocolloid treatment of cystic craniopharyngiomas has generally yielded impressive results, with gradual shrinkage and eventual collapse of the cyst wall (2, 4, 5, 41, 43). However, in at least 31 other cases where a primarily solid or combined multicystic/solid craniopharyngioma was involved, treatment with external closed stereotaxic gamma knife radiosurgery (preceded by stereotaxic yttrium-90 instillation only where indicated by the presence of a sizable partial cystic component) was the preferred modality. The exclusive use of the gamma knife was usually (but not exclusively) employed for those cases where small (less than 2.5-cm diameter), relatively circumscribed, deep-seated neoplastic lesions permitted maximal utilization of the gamma unit's sharply demarcated multiportal intersecting beams providing selective, precise high intensity irradiation of the designated target field (41). Radionecrosis of craniopharyngiomas can occur at single dose magnitudes of only 2 to 3 Gy (1), however, and thus more sizable lesions have also been effectively treated with gamma knife radiosurgery (3, 6). Among the 31 craniopharyngiomas cited, 9 such lesions reported by Backlund ranged up to 5 cm in diameter and received target doses varying between 20 and 50 Gy (3). Doage delivery to the tumor was dependent on tumor size (larger lesions receive attenuated dosage levels at their periphery and thus require high target dosages), the relative contributions of cystic and solid components to the tumor volume

(solid tumors require higher delivered dosages), and the proximity of vital neural structures (the single dose tolerance of optic pathways is limited to 10 Gy, thus constraining dosage levels to the periphery of certain lesions extending into the region of the optic chiasm) (3). In maintaining the principle that the steepest gradient of the radiation field (50% isodose level) should coincide with the peripheral margin of the tumor, vital structures surrounding tumor are spared significant radiation injury in most cases. Conversely, the delivered dose (even in larger tumors receiving only 5 Gy to some areas of their periphery) in the cases reported still resulted in dramatic tumor shrinkage and striking clinical improvement (3). The relative paucity of acute and delayed radiation effects observed with high intensity gamma knife radiosurgery relates in part to the relatively focused area of radionecrosis and the resultant decreased risk of generalized edema involving the surrounding neural tissue (39), provided that the basic principles governing high energy beam irradiation, as noted, are followed. The contribution of Swedish neurosurgery with regard to stereotaxic radiosurgery of arteriovenous malformations has been equally impressive. Leksell (1983) cited over 204 vascular malformations treated with the stereotaxic gamma knife unit. In 67 of these patients, follow-up angiography 2 years after treatment demonstrated an 83% obliteration rate, most notably in those cases where intersecting stereotaxically focused high energy radiation beams covered most of an entire malformation (41). Mild to moderate residual or delayed onset hemiparesis in 5 cases and a latency of 6 to 18 months between treatment and angiographically verified obliteration of these lesions represent another drawback in terms of delayed bleeding risks. Best suited for this procedure are deep-seated, small (less than 2.5 to 3 cm in diameter) lesions presenting an otherwise high surgical risk with conventional management. As presented in some initial pioneering reports, stereotaxic angiography permits accurate localization through three-dimensional coordinates that define the lesion target field when transferred to the gamma knife unit (40, 67, 69, 70). Multiple, narrow, intersecting high energy beams (from up to 179 cobalt-60 sources) are focused on the lesion target field using a collimator arrangement with the appropriately selected cross sectional beam diameter, thus resulting in one or more high intensity radiation isodose fields stereotaxically centered within the target lesion (68). Preplanned dosimetry configurations placing the

50% isodose curve at the lesion margin are routinely utilized. Single dose treatment has varied between 3 and 12.5 krads over a period of 20 to 40 minutes, utilizing collimator diameters of 8 to 14 mm which correspond to target volume coverage of 0.5 to 3 cm3 (within at least the limits of the 50% isodose curve), respectively (68, 69). Serial follow-up angiography in 42 of 68 patients first reported by Steiner (1979) and then expanded to a series of 85 patients with 1-year and 67 patients with 2-year angiographic reevaluation by Leksell (1983) revealed that 83 to 85% of vascular malformations are completely obliterated at 1 year, with 6% of the lesions unchanged and the remaining 10.5% partially obliterated also at 2 years posttherapy (41, 68). A single dose of 5 krads (and possibly as little as 3 to 4 krads) seems to be a sufficient critical threshold dose for the obliteration of most arteriovenous malformations under 3 cm in diameter. Attendant damage of surrounding brain tissue emerging as a delayed consequence of treatment was noted in 5 of these cases, and this late radiation effect appearing 3 to 9 months after stereotaxic radiosurgery is mostly restricted to cases where the 5-krad limit was exceeded, especially in conjunction with the use of a 14-mm collimator. The associated delayed neurological deficit in such cases (hemiparesis, etc.) may be related to changes affecting regional venous drainage surrounding the target zone (68). Any summary or review of regional contributions to a field as rapidly changing and dynamic as neurosurgical stereotaxy requires a preview of impending or expected developments. The contemporary evolution of stereotaxic radiosurgical technology continues, as evidenced in the development of a radiosurgery apparatus merging stereotaxic technique with a high energy isocentric 4-MV linear accelerator radiotherapy unit (23). Colombo et al. from Italy have begun to investigate an original adaption along these lines and have reported the capacity for steeper isodose curve fall-off, a wider spherical dose coverage sector, and a significantly wider range and maneuverability of collimator setting parameters compared with the more conventional gamma knife unit previously described (23). By utilizing three-dimensional CT data and modifications in the specially constructed linear accelerator stereotaxic apparatus, the investigators may eventually succeed in obtaining molded radiation emission isodose curves, thus allowing extremely enhanced accuracy in the radiotherapeutic coverage of a target lesion's topographic variance. The fusion of stereotaxy and neurooncology is

already apparent with reports emanating from various European centers describing the utilization of stereotaxic techniques for the interstitial,intraneoplasticadministration/instillation of chemotherapeutic agents (bleomycin) and immunomodulating / immunotherapeutic substances (interferon) (16, 53). Therapeutic antineoplastic agents could thus be chosen and utilized irrespective of limitations imposed by issues impacting on vascular delivery and neurovascular blood-brain barrier dynamics. Stereotaxic cryosurgery as advocated by Boethius and Greitz in Stockholm may entail a potential for novel management of small, deepseated tumors by permitting the stereotaxic neurosurgeon to create a stereotaxically defined destructive lesion within a given neoplastic mass that is sufficiently large to cover the full extent of the target mass volume (13). The diameter of the cryoprobe determines the lesion dimensions, and the probe is left in place 15 to 20 minutes after creation of the lesion, primarily for considerations of hemostatic tamponade during the post-cold injury phase of transient capillary bleeding (13). Necrotized tissue may be reduced in volume through an outer suction cannula within the stereotaxic apparatus; however, care must be taken to avoid the ventricular space because this may provide a natural egress portal for any residual capillary oozing. No data are available on postsurgical evaluation, patient performance, and effect of the stereotaxic cryosurgical lesion on arresting progression of the offending neoplastic mass. Adaption of new cell culture techniques to the size constraints of stereotaxic microbiopsies will allow more extensive in vitro pathological scrutiny of tissue specimens within culture systems and as an adjunct to more conventional histopathological analysis (77). Related harvesting methods can provide biopsy tissue for a variety of sophisticated analytical methodologies, including rapid flow cytofluorometry of tumor cell suspensions with attempts at analysis of DNA distribution patterns among a series of glial tumors, or on a comparative basis between different biopsy target sites within a single lesion. One such preliminary study has demonstrated the feasibility and potential efficacy of such investigation, although no definite correlation of histopathological grading and DNA distribution patterns within tumor cell populations from different tumor biopsy sites and different patients was identified (11). Such techniques may eventually provide objective criteria for assessing a given tumor's biological activity and degree of malig-

nancy. If histopathological correlation is eventually determined with respect to assays of nucleic acid distribution patterns in CNS tumors, prognostic factors and therapeutic considerations may emerge as relevant correlates of DNA distribution analysis in tumors. The frontier of neurosurgical stereotaxy most likely points toward future involvement with microtransplantation techniques within the CNS. A recent pilot clinical trial reported by Backlund et al. from Sweden detailed the first documented attempt at autologous adrenal to striatal cross transplantation utilizing stereotaxic implantation in patients with severe parkinsonism refractory to pharmacological modulation (7). The theoretical aim of providing striatal tissue with a new cellular template source of dopaminergic catecholamines utilizing adrenal medullary tissue simultaneously harvested from the patient just before stereotaxic implantation is well founded on experimental animal data on reinnervation of nigrostriatal 6-hydroxydopaminelesioned striatum with fetal substantia nigra grafting. Initial results with the two selected patients receiving the autologous adrenal medullary grafts are sufficiently intriguing to merit further clinical trials on highly selected patients. Undoubtedly, autograft adrenal medullary cell populations will need to be cloned or selected out for maximal dopaminergic/catecholamine secretory capacity if more biochemically sophisticated attempts at microtransplantation within the CNS are entertained. This initial clinical trial of Backlund et al. performed in a neurosurgical stereotaxic setting with a deep-seated striatal target point represents the first attempt at microtransplantation within the human brain. References 1. Backlund E-O: Disorders of the skull base region. In Hamberger and Wersall (eds): Nobel Sympo sium # 10, 1969, pp 237-244. 2. Backlund E-O: Stereotactic treatment of cranio pharyngiomas: A 15 year material. Presented at the 32nd Annual Meeting of the Scandinavian Neurosurgical Society, Linkoping, 1979. 3. Backlund E-O: Solid craniopharyngiomas treated by stereotaxic radiosurgery. In: Szikla G (ed): Ster eotactic Cerebral Irradiation, INSERM Sympo sium * 12. Amsterdam, Elsevier North Holland, Biomedical Press, 1979, pp 271-281. 4. Backlund E-O: Studies on craniopharyngiomas: Stereotaxic treatment with intracystic yttrium90. Acta Chir Scand 138:749-759, 1972. 5. Backlund E-O: Studies on craniopharyngiomas: Stereotaxic treatment with intracystic yttrium90. Acta Chir Scand 139:237-247, 1973. 6. Backlund E-O: Studies on craniopharyngiomas:

7. 8. 9.

10. 11.

12. 13.

14.

15. 16.

17. 18.

19.

20. 21.

22.

23.

24.

Stereotaxic treatment with radiosurgery. Acta ChirScand 139:344-351, 1973. Backlund E-O: Transplantation of adrenal med ullary tissue to striatum in parkinsonism. J Neurosurg 62:169-173, 1985. Bergstrom M, Greitz T: Stereotaxic computed to mography. AJR 127:167-170, 1976. Birg W, Mundinger F: Direct target point deter mination for stereotactic brain operations from CT data and the calculation of setting parameters for polar-coordinate stereotactic devices. Appl Neurophyslol 45:387-395, 1982. Birg W, Mundinger F, Klar M: Computer assist ance for stereotactic brain operations. Adv Neuro Surg 4: 287-291, 1977. Boethius J, Bergstand G, Collins VP, Edner G, Tribukait B: DNA distribution in various parts of malignant gliomas assayed on stereotactic biop sies. Appl Neurophysiol 43:216-221, 1980. Boethius J, Bergstrom M, Greitz T, Ribbe T: CT localization in stereotactic surgery. Appl Neu.rophysiol 43:164-169, 1980. Boethius J, Greitz T, Kuylenstierna R, Lagerkranser M, Lundquest PG, Ribbe T, Wiksell H: Stereo: tactic cryosurgery in a CT scanner. Acta Neurochir[Wien) (Suppl) 33:553-557, 1984. Boethius J, Ribbe T: A new stereotatic instrument which can be used in conjunction with open sur gery. ActaNeurochlr[Wien) (Suppl) 33:559-565, 1984. Bosch DA: Indications for stereotactic biopsy in brain tumours. Acta Neurochir (Wien) 54:167179, 1980. Bosch DA, Hindmarsch T, Larsson S, Backlund EO: Intraneoplastic administration of bleomycin in intracerebral gliomas: A pilot study. Acta Neu rochir (Wien) (Suppl) 30:441-444, 1980. Bosch DA, Rahn T, Backlund E-O: Treatment of colloid cysts of the third ventricle by stereotactic aspiration. Surg Neurol 1:15-18, 1978. Broggi G, Franzini A, Giorgi C, Allegranza A: Di agnostic accuracy and multimodal approach in stereotactic biopsies of deep brain tumours. Acta Neurochir (Wien) (Suppl) 33:211-212, 1984. Broggi G, Franzini Migliavacca F, Allegranza A: Stereotactic biopsy of deep brain tumors in in fancy and childhood. Childs Brain 10:92-98, 1983. Broseta J, Gonzalez-Darder J, Barcia-Salorio JL: Stereotactic evacuation of intracerebral hematomas. Appl Neurophysiol 45:443-448, 1982. Busch M, Makoski B, Schulz U, Sauerwein K: Das Essener Nachlade-Verfahren fur die intrakavitare Strahlentherapie. Strahlentherapie 153:581588, 1977. Columbo F, Angrilli F, Zanardo A, Pinna V, Alexandr A, Benedetti A: A universal method to employ CT scanner spatial information in stereotactic surgery. Appl Neurophysiol 45:352-364, 1982. Colombo F, Benedetti A, Pozza F, Avanzo RC, Marchetti C, Chierego G, Zanardo A: External ster eotactic irradiation by linear accelerator. Neuro surgery 16:154-160, 1985. de Divitiis E, Spaziante R, Cappabianca P, Caputi F, Pettinato G, De Саго М: Reliability of stereotac tic biopsy: A model to test the value of diagnoses obtained from small fragments of nervous system tumors. Appl Neurophysiol 46:295-303, 1983.

25. de Sola GR, Cabezudo J, Areitio E, Bravo G: Com bined approach (stereotactic-microsurgical) to a para ventricular arteriovenous malformation. Acta Neurochir (Wien) (Suppl) 30:413-416, 1980. 26. Dettori P, Colombo F, Pinna V, Benedetti A: CT control of stereotactic surgery in the diencephalon. Neuroradiology 23:91-94, 1982. 27. Daumas-Duport C, Monsaingeon V, N'Guyen JP, Missir O, Szikla G: Some correlations between histological and CT aspects of cerebral gliomas contributing to the choice of significant trajecto ries for stereotactic biopsies. Acta Neurochir (Wien) (Suppl) 33:185-194, 1984. 28. Dauman-Duport C, Monsaingeon V, Szenthe L, Szikla GS: Serial stereotactic biopsies: A double histological code of gliomas according to malig nancy and 3-D configuration, as an aid to thera peutic decision and assessment of results. Appl Neurophysiol 45:431-437, 1982. 29. Edner G: Stereotactic biopsy in tumours of the sellar region other than adenomas. Acta Neuro chir (Wien) (Suppl) 28:326-328, 1979. 30. Edner G: Stereotactic biopsy of intracranial space occupying lesions. Acta Neurochir (Wien) 57:213-234, 1981. 31. Froder M, Seitzer D, Buren G, Dieckmann G: Dig ital radiography for target point evaluation in ster eotactic neurosurgery. Appl Neurophysiol 46:206-210, 1983. 32. Huk W, Baer U: A new targeting device for ster eotaxic procedure within the CT scanner. Neuro radiology 19:13-17, 1980. 33. Kandel El, Peresedov VV: Stereotaxic clipping of arterial and arteriovenous aneurysms of the brain. Acta Neurochir (Suppl) 30:405-412, 1980. 34. Kandel El, Peresedov VV: Stereotaxic evacuation of spontaneous intracerebral hematomas. J Neurosurg 62:206-213, 1985. 35. Kiessling M, Kleihues P, Gessaga E, Mundinger F, Ostertag CB, Weigel K: Morphology of intracranial tumours and adjacent brain structures following interstitial iodine-125 radiotherapy. Acta Neuro chir (Wien) (Suppl) 33:281-289, 1984. 36. Kleihues P, Volk B, Anagnostopoulos Kiessling M: Morphologic evaluation of stereotactic brain tum our biopsies. Acta Neurochir (Wien) (Suppl) 33:171-181, 1984. 37. Larssou B, Leksell L, Rexed B, Souvander P, Mair W, Andersson B: The high energy proton beam as a neurosurgical tool. Nature 182:1222-1223, 1958. 38. Leksell L: A stereotaxic apparatus for intracerebral surgery. Acta Clin Scand 99:229-233, 1949. 39. Leksell L, Larsson B, Andersson B, Rexed B, Sourander P, Mair W: Lesions in the depth of the brain produced by a beam of high energy protons. Acta Radiol (Stockh) 54:251-264, 1960. 40. Leksell L: Stereotaxis and radiosurgery: An oper ative system. Springfield, IL, Charles С Thomas Publisher, 1971. 41. Leksell L: Stereotactic radiosurgery. J Neurol Neurosurg Psychiatry 46:797-803, 1983. 42. Leksell L, Larsson B, Rexed B: The use of highenergy pistons for cerebral surgery in man. Acta ChirScand 125:1-7, 1963. 43. Leksell L, Uden C: A therapeutic trial with radio-

44. 45.

46.

47. 48. 49.

50.

51.

52.

53. 54.

55.

56.

57. 58.

59.

active isotopes in cystic brain tumor. In Radioisotope Techniques. Oxford, H. M. Stationary Office, 1953, vol 1. Lewander R: Contrast enhancement with time in gliomas. Acta RadiolDiagn 20:689-702, 1979. Lewander R, Bergstrom M, Boethius J, Collins VP, Edner G, Greitz T, Willems J: Stereotactic com puter tomography for biopsy of gliomas. Acta Ra diolDiagn 19:867-888, 1978. Lobato RD, Rivas JJ, Cabello A, Roger R: Stereotactic biopsy of brain lesions visualized with com puted tomography. Appl Neurophysiol 45:426430, 1982. Mundinger F, Birg W: Stereotactic brain surgery with the aid of computed tomography (CT-stereotaxy). Adv Neurosurg 10:17, 1982. Mundinger F, Birg W: Stereotactic biopsy of intra cranial processes. Acta Neurochir {Wien) {Suppl) 33:219-224, 1984. Mundinger F, Hoefer T: Protracted long-term ir radiation of inoperable midbrain tumours by ster eotactic curie-therapy using iridium-192. Acta Neurochir {Wien) {Suppl) 21:93-100, 1974. Mundinger F, Birg W, Klar M: Computer-assisted stereotactic brain operations by means including computerized axial tomography. Appl Neuro physiol 41:169-182, 1978. Mundinger F, Metzel E: Interstitial radioisotope therapy of intractable diencephalic tumors by the stereotaxic permanent implantation of iridium192, including bioptic control. Confin Neurol 32:195202, 1970. Mundinger F, Noetzel H, Riechert T: Erfahrungen mit der lokalisierten Bestrahlung von malignen Hirngeschwulsten mit Radio-Isotopen. Acta Neu rochir {Suppl) 6:171-182, 1959. Mundinger F, Ostertag CB, Birg W, Weigel K: Ster eotactic treatment of brain lesions. Appl Neuro physiol 43:198-204, 1980. Mundinger F, Riechert T, Schulz A, Zysno E: Die Applikation von Radio-Isotopen zur Strahlenbehandlung Intracranieller Tumoren. Acta Neuro chir {Wien) {Suppl) 6:183-186, 1959. Mundinger F, Vogt P, Jobski C, Fischer HU, Os tertag C: Klinische und experimentelle Ergebnisse der interstitillen Brachy-Curietherapie in Kombination mit Radiosensibilisatoren bei infiltrierenden Hirntumoren. Strahlen Therapie 193:318323, 1972. Mundinger F, Weigel K: Indication and results of stereotactic curietherapy with iridium-192 and iodine-125 for non-resectable tumours of the hy pothalamic region. Acta Neurochir (Wien) {Suppl) 33:323-330, 1984. Mundinger F, Weigel K: Long-term results of ster eotactic interstitial curietherapy. Acta Neurochir {Wien) {Suppl) 33:367-373, 1984. Netzeband G, Sturm V, Georgi P, Sinn H, Schnabel K, Schlegel W, Schabbert S, Marin-Grez M, Gahbauer H: Results of stereotactic intracavitary ir radiation of cystic craniopharyngiomas: Compar ison of the effects of yttrium-90 and rhenium186. Acta Neurochir (Wien) {Suppl) 33:341-344, 1984. Ostertag CB, Groothuis D, Kleihues P: I. Effects on tumour and brain: Experimental data on early and late morphologic effects of permanently im planted gamma and beta sources (iridium-192,

60. 61.

62. 63. 64.

65.

•

66.

67.

68.

69.

70.

71.

72.

73.

74.

75.

iodine-125 and yttrium-90) in the brain. Acta Neurochir {Wien) {Suppl) 33:271-280, 1984. Ostertag CB, Mennel HD, Kiessling M: Stereotactic biopsy of brain tumors. Surg Neurol 14:275-283, 1980. Ostertag CB, Weigel K, Warnke P, Lombeck G, Kleihues P: Sequential morphological changes in the dog brain after interstitial iodine-125 irradia tion. Neurosurgery 13:523-528, 1983. Rivas JJ, Lobato RD: CT-assisted stereotaxic as piration of colloid cysts of the third ventricle. J Neurosurg 62:238-242, 1985. Rougier A, Pigneux J, Cohadon F: Combined in terstitial and external irradiation of gliomas. Acta Neurochir {Wien) {Suppl) 33:345-353, 1984. Scarabin JM, Pecker J, Brucher JM, Vallee B, Guegan Y, Faivre J, Simon J: Stereotactic explo ration in 200 supratentorial brain tumors. Neuroradiology 16:591-593, 1978. Scerrati M, Rossi GF: The reliability of stereotactic biopsy. Acta Neurochir {Wien) {Suppl) 33:201205, 1984. Sedan R, Peragut JC, Farnarier P, Hassoun J, Sethian M: Intraencephalic stereotactic biopsies (309 patients/318 biopsies). Acta Neurochir {Suppl) 33:207-210, 1984. Steiner L, Backlund EO, Grite T, Leksell L, Noren G, Rahn T: Radiosurgery in intracranial arteriovenous malformations. Excerpta Med Int Congr Series 433, 1978. Steiner L, Greitz T, Backlund EO, Leksell L, Noren G, Rahn T: Radiosurgery in arteriovenous malfor mation of the brain: Undue effects. In Szikla G (ed): Stereotactic Cerebral Irradiation, INSERM Symposium * 12. Amsterdam, Elsevier North Hol land, Biomedical Press, 1979, pp 257-269. Steiner L, Leksell L, Forster DMC, Greitz T, Back lund E-O: Stereotactic radiosurgery in intracra nial arterio-venous malformations. Acta Neuro chir {Wien) {Suppl) 21:195-209, 1974. Steiner L, Leksell L, Greitz T, Forster DMC, Back lund E-O: Stereotaxic radiosurgery for cerebral arterio-venous malformations: Report of a case. Acta Chir Scand 138:459-464, 1972. Sturm V, Pastyr O, Schlegel W, Scharfenberg H, Zabel H-J, Netzeband G, Schabbert S, Berberich W: Stereotactic computer tomography with a mod ified Riechert-Mundinger device as the basis for integrated stereotactic neuroradiological investi gations. Acta Neurochir {Wien) 68:11-17, 1983. Suetens P, Gybels J, Jansen P, Oosterlinck A, Haegemans A, Kierckx P: A global 3-D image of the blood vessels, tumor and simulated electrode. Acta Neurochir {Wien) {Suppl) 33:225-232, 1984. Szikla G, Musolino A, Miyahara S, Schaub C, Askienazy S: Colloid rhenium-186 in endocavitary beta irradiation of cystic craniopharyngiomas and active glioma cysts: Long term results, side effects and clinical dosimetry. Acta Neurochir {Wien) {Suppl) 33:331-339, 1984. Szikla G, Schlienger M, Blond S, Daumas-Duport C, Missir O, Miyahara S, Musolino A, Schaub C: Interstitial and combined interstitial and external irradiation of supratentorial gliomas: Results in 61 cases treated 1973-1981. Acta Neurochir {Wien) {Suppl) 33:355-362, 1984. Tampieri D, Bergstrand G: Postural displacement

of the brain—on the feasibility of using CT for determination of stereotactic coordinates. Neuroradtology 24:167-168, 1983. 76. Thomas DGT, Anderson RE, du Boulay GH: CTguided stereotactic neurosurgery: Experience in 24 cases with a new stereotactic system. J Neurol Neurosurg Psychiatry 47:9-16, 1984. 77. Thomas DGT, Darling JL, Watkins BA, Hine MC: A simple method for the growth of cell cultures from small biopsies of brain tumours taken during CT-directed stereotactic procedures. Acta Neurochir[Wien) (Suppl) 33:243-245, 1984.

78. Thomson GSM, Kingsley PE, Afshar F, Wylie IG: Stereotactic brain biopsy using a narrow aperture computed tomography scanner. Clin Radiol 35:209-214, 1984. 79. Waltregny A, Petrov V, Brotchi J: Serial stereo taxic biopsies. Acta Neurochir (Wien) (Suppi) 21:221-226, 1974. 80. Willems J, Alva-Willems J: Accuracy of cytologic diagnosis of central nervous system neoplasms in stereotactic biopsies. Int Acad Cytol 28:243-249, 1984.

Commentary F Role of Stereotaxis in the Management of Midline Cerebral Lesions Erik-Olof Backlund, M.D., Ph.D.

The integration of stereotaxic operative techniques into traditional, "nonfunctional" neurosurgery has been of great importance. For many years some clinical research groups have tried to follow the basic principle that surgical problems in the deeper regions of the intracranial compartment should be solved, if feasible, with the use of stereotaxic techniques, whereas superfi-

cial lesions should be attacked by the best "visual" techniques available, i.e., microsurgery (Fig. F.1). In my experience, consistent attention to the basic principle implied radically changed conditions of general policy, risk, and outcome for patients with lesions inside or near the third ventricle. The importance of this new, stereotaxic approach has been proven in controversial

issues such as the management of craniopharyngiomas. Since the integration of stereotaxy with various imaging techniques, the crucial value of a stereotaxic procedure as a first choice alternative instead of a conventional operation has been clearly demonstrated in many more patients than was first assumed possible. As a commentary to more than one chapter in this book, I will give some examples from my personal experience to illustrate these considerations.

Biopsy Inaccessible tumors in the central core of the brain have been treated with radiotherapy without a previous tissue diagnosis. Well-established stereotaxic biopsy techniques are now available, and such a policy can no longer be justified. Any intracranial tumor can now be classified by fine needle biopsy with reasonable safety (4, 6, 11). In spite of continuous improvement of the diagnostic imaging techniques, a tissue diagnosis by biopsy (histological sectioning or smears for cytological examination) increases the reliability of the diagnosis (8). Stereotaxic biopsy also allows new possibilities for the characterization of a tumor by other means, such as biological markers, immunohistochemistry, and DNA measurements. The advantages of the stereotaxic approach for biopsy are demonstrated in particular when lesions in the pineal region are encountered. Here there is a wide diagnostic spectrum, making radiological diagnosis less accurate and thus tissue classification more important for further man-

agement. New therapeutic modalities for pineal region lesions have appeared and new policies have been designed (2, 10). This makes tissue diagnosis a virtual prerequisite for a proper choice of treatment. The possible occurrence of a benign cyst in the pineal region simulating a tumor deserves special attention in this context. The computerized tomographic (CT) attenuation level is not always conclusive for a cyst and may even indicate a cystic

lesion to be solid. Stereotaxic biopsy always reveals the true condition. To perform a cystocisternal or cystoventricular shunting procedure during the same operation as the biopsy is often feasible and the treatment of choice (Fig. F.2). Tumors in the brain stem proper at the mesencephalic level or in the posterior fossa have been considered inaccessible for stereotaxic biopsy because of inherent risks related to this region. However, with the transcerebellar approach, any conceivable region of the intracranial compartment can be reached for fine needle biopsy with reasonable risk (Fig. F.3). Colloid Cysts of the Third Ventricle Because of the innocent character of a colloid cyst in the third ventricle, a conventional surgical intervention for its removal may be far too mutilating and risky. Any experienced brain surgeon may recall frustrating episodes during such operations, such as difficulty keeping a hydrocephalic hemisphere distended or troublesome bleeding from central blood vessels at the depth of a keyhole operating field. The introduction of an aspiration technique for these lesions opened new perspectives (5). My first such case, a colleague, has now been followed for 16 years without recurrent symptoms. The patient continued his academic career after the operation and is now working as a clinical professor. Major operation for these intraventricular cysts seems no longer necessary. The technique

is internationally recognized, with rewarding results reported in a significant number of patients (8, 12). Craniopharyngiomas For patients with huge craniopharyngiomas extending deeply into the central regions of the brain, the introduction of a standardized stereotaxic treatment program has been of great importance (1). My personal experiences prove the definite value of an exclusive stereotaxic approach in seemingly desperate cases with cystic tumor parts encroaching upon the third ventricular area far back to the aqueductal region. Long term results from a recent review of 15 years of material are shown in Table F. 1. Intracystic instillation of radioactive material (in a liquid state, usually colloidal yttrium-90, a pure beta-emitter) initiates a gradual collapse of the cystic part, with no dramatic peroperative consequences. For many patients, the results are dramatic (Fig. F.4). In the majority of craniopharyngioma cases the clinical problems are due to expanding cysts. Any treatment program should begin with a stereotaxic cyst puncture in preparation for radioisotope treatment. This less dramatic management is in bright contrast to the often sadly dramatic courses after attempts at tumor removal. Rewarding results after stereotaxic craniopharyngioma treatment reported by others have given this approach international recognition (7, 9, 13, 14). The range of beta-radiation, albeit very short

normal and that the subarachnoid space was patent. In spite of fulfilled selection criteria and a perfect "morphological" result from the operation, some patients in the first series obviously had a component of impaired CSF resorption capacity outside the ventricular system. The acquired patency of the aqueduct did not assure a decrease of ventricular size. If better selection methods can be designed, stereotaxic aqueduct reconstruction will deserve a place as a logical treatment in selected cases of obstructive hydrocephalus.

References

(50% of the dose at 1.1-mm depth) is of course a critical factor when risks and possible complications are considered. In less than 10% of my patients was there suspicion of late adverse effects from the radiation, most often sudden impairment of the visual fields after a long symptom-free interval. The side effect was disabling in only one patient. By improvement of the technique, probably using another radiocolloid, the complication rate may well be lower in the next series.

Occlusion of the Aqueduct The procedure most often used for the treatment of various hydrocephalic states, ventriculovenous or ventriculoperitoneal shunting, is not a logical approach when impaired Cerebrospinal fluid (CSF) flow through the aqueduct is the cause of the ventricular dilatation. By stereotaxic technique it is possible to restore the anatomical communication between the third and the fourth ventricles (3). The course of the aqueduct strictly in the midsagittal plane, a seeming morphological obstacle for the instrumental approach, is no problem with the use of a specially designed cannula. The technique allows shorter tubes or catheters to be placed safely through the aqueduct after mechanical recanalization guided by preoperative selective stereotaxic ventriculography. When my first material was analyzed, it was found in retrospect that the selection criteria (lumbar infusion/pressure test and isotope cisternography) had not been accurate. When formulating these criteria I assumed that normal findings would indicate that CSF resorption was

1. Backhand EO, Johansson L, Sarby B: Studies on craniopharyngiomas: II. Treatment by stereotaxis and radiosurgery. Acta ChirScand 138:749-759, 1972. 2. Backlund EO, Rahn T, Sarby B: Treatment of pinealomas by stereotaxic radiation surgery. Acta Radiol Ther Phys Biol 13:368-376, 1974. 3. Backlund EO, Grepe A, Lunsford D: Stereotaxic reconstruction of the aqueduct of Sylvius. J Neurosurg 55:800-810, 1981. 4. Bosch DA: Indications for stereotaxic biopsy in brain tumours. Acta Neurochir (Wien) 54:167179, 1980. 5. Bosch DA, Rahn T, Backlund EO: Treatment of colloid cysts of the third ventricle by stereotaxic aspiration. Surg Neurol 9:15-18, 1978. 6. Edner G: Stereotaxic biopsy of intracranial space occupying lesions. Acta Neurochir (Wien) 57:213-234, 1981. 7. Kobayashi T, Kageyama N, Ohara K: Internal ir radiation for cystic craniopharyngioma. J Neurosurg 55:896-903, 1981. 8. Lunsford LD, Martinez AJ: Stereotaxic explora tion of the brain in the era of computed tomogra phy. Surg Neurol 22:222-330, 1984. 9. Lunsford LD, Gumerman L, Levine G: Stereotaxic intracavitary irradiation of cystic neoplasms of the brain. Presented at the IX Meeting of the World Society for Stereotactic and Functional Neurosur gery, Toronto, 1985. 10. Neuwelt EA, Gumerlock MK: The challenge of pineal region tumors. In Neuwelt EA (ed): Diag nosis and Treatment of Pineal Region Tumors. Baltimore, Williams & Wilkins, 1984, pp 1-30. 11. Ostertag CB, Mennel HD, Kiessling M: Stereotaxic biopsy of brain tumors. Surg Neurol 14:275-283, 1980. 12. Rivas JJ, Lobato RD: CT-assisted stereotaxic as piration of colloid cysts of the third ventricle. J Neurosurg 62:238-242, 1985. 13. Schaub C, Bluet-Pajot MT, Videau-Lornet C, et al: Endocavitary beta irradiation of glioma cysts with colloidal 186rhenium. In Szilka G (ed): Stereotactic Cerebral Irradiation. Amsterdam, North-Hol land, 1979, pp 293-302. 14. Strauss L, Sturm V, Georgi P, et al: Radioisotope therapy of cystic craniopharyngiomas. Int J Radiat Oncol Biol Phys 8:1581-1585, 1982.

Commentary G Magnetic Resonance Stereotaxy Eric R. Cosman, Ph.D., M. Peter Heilbrun, M.D., and Trent H. Wells, Jr.

We describe here the development of a new localizer frame and head ring assembly that enables the calculation of stereotaxic targets from virtually any magnetic resonance imaging (MRI) scan plane, including those that are parallel to the axial, sagittal, or coronal planes and those that are skewed to these basic slice orientations. The assembly is compatible with the Brown-Roberts-Wells (BRW) computerized tomographic (CT) stereotaxic system and can be used in any standard MRI head coil without image aberration or a special means of fixation to the MRI machine. Just as the combination of CT scanning and stereotaxic surgery evolved into a powerful neurosurgical technique (1,3), MRI stereotaxy offers yet added potential for observing and approaching intracranial targets. MRI is complementary and sometimes superior to CT in diagnosing certain brain tumors and neurological disorders. MRI also has the significant advantage over CT of yielding easily coronal and sagittal images and, in the near future, possibly chemical shift and spectroscopic images. There are, however, several difficult technical problems in building a

universal stereotaxic MRI localizer that makes full use of the advantages of MRI and yet does not degrade the MRI images themselves. We summarize these problems here and describe a new localizer design that has overcome them and worked well in early clinical trials. The ideal MRI localizer system for stereotaxic surgery should enable target coordinates—anteroposterior (AP), lateral, and vertical positions relative to a head ring fixed to the head — to be calculated from any slice reconstruction, whether it be parallel to or skewed to the axial, sagittal, or coronal plane. This would in many cases simplify the planning of approaches to volumetric or functional targets in the brain. The standard CT localizer for the BRW system allows target calculations from any nearly axial plane to be made by means of a set of three N-type structures (2), each with two rods and one diagonal made of carbon fiber, arranged around an axial circumference of the head as shown in Figure G.1A. A nearly axial slice will intersect the six rods and three diagonal elements to give nine fiducial points on the CT image (Fig. G.1B)

from which the CT plane and any target within it can be determined. The MRI localizer extends this concept to include any arbitrary slice plane, not just those that are quasi-axial. It consists of a cubical array of N-type structures with rod and diagonal elements in three orthogonal planes, as shown in Figure G.2A. This assures that for any plane oriented approximately axially, sagittally, or coronally a sufficient number of rod and diagonal elements will be cut to determine fully that plane and any target within it. Furthermore, by making the pattern of fiducial points different for each orthogonal cut, as shown in Figure G.2B, the plane and parity of the cut can never be misinterpreted. Such a universal MRI localizer is scanner-independent in the sense that it does not have to be fixed in any particular alignment relative to the scanner gantry. It may be placed in the gantry with its axis skewed to the scanner axis, yet as

long as the appropriate rods and diagonals are cut by the scan plane, the coordinates relative to the head ring of any target seen in the scan image can be calculated. This is a significant feature because clamping a head ring or localizer to an MRI machine is made difficult by the small size of standard head coils and the variable shape of machine tables. The localizer frame and head ring must not cause distortion of the precision magnetic and radiofrequency (rf) fields of the MRI scanner that are required for proper imaging. Thus no ferromagnetic materials may be used in its construction and major eddy current pathways must be eliminated for metallic components. Figure G.3 shows the present design, which works well for MRI magnetic fields of 0.1 to 1.5 tesla. The head ring, head posts, and skull screws are made of anodized aluminum, which has no effect on the scanner's magnetic fields. The skull screws are

pointed and hardened to penetrate the scalp directly and are self-anchored to the skull without dulling. This simplifies applying the head ring to the patient and eliminates making scalp incisions and skull drill holes. The head ring is made in two aluminum semicircular segments that are connected by insulating Micarta spacers 180° apart. This is essential to prevent a macroscopic eddy current loop running circumferentially around the head ring, induced by the MRI scan-

ner's rf excitation field. Such a current loop would cause severe attenuation of the MRI signal. The design has yielded MRI images of negligible distortion and attenuation. Figure G.4 shows the MRI localizer attached to the head ring. The localizer is made of polycarbonate and contains tubular channels on the front, back, sides, and top. The channels may be filled with a variety of paramagnetic substances so that they will be detectable as fiducial spots

on an intersecting scan slice; however, petroleum jelly works well for a wide range of time and spin-echo MRI pulse sequences and eliminates the problems of replenishment. Figure G.5 is an axial MRI scan illustrating the fiducial spots. The x and у coordinate data for each fiducial spot, as well as any chosen target point or points on the MRI image, may be extracted from the MRI machine console software and are en-

tered into an off-line computer-calculator. The calculator computes the actual scan plane and the AP, lateral, and vertical coordinates of the target relative to the head ring. This is done very efficiently by means of a vector projection formalism. We note that the x and у data need only be linear (for example, pixel number or dimensions of arbitrary scale factor) and do not depend on their origin position for the calculator to make

this computation. The resulting head ring coordinates are subsequently used for the stereotaxically calculated approach to the target. To fit inside standard MRI machine manufacturers' head coils and also to mate with the BRW stereotaxic surgical system, the MRI head ring and localizer were made with their outer diameter less than 260 mm. The outer rim of the MRI head ring is formed to accept the standard BRW CT head ring (which is not MRI-compatible) and the two head rings are clamped together in a precise orientation during the surgical phase by fixation points, as shown in Figure G.6. After MRI scanning with the MRI head ring and localizer connected (Fig. G.4), the localizer is removed and the CT head ring is attached to the MRI head ring (Fig. G.6). From that point on, the BRW system components, such as the arc system, phantom base, etc., may be used in the usual way, all being referenced to the BRW CT head ring. Figure G.7 shows the BRW arc system attached to the combined BRW and MRI head ring as it would be for stereotaxic surgery. The MRI localizer system is modular so that individual components may be changed with minimal effort. For example, an alternate local-

izer frame with integrated surface coils that can also couple to the MRI head ring for high resolution and isotope imaging studies is being designed. The present system itself may also be used for CT scan imaging as well as MRI, making possible comparative scanning without change of the head ring.

Acknowledgments We are indebted to Dr. M. Apuzzo and Dr. E. Ganz for facilitating several MRI tests during the course of this development. E.R.C. was supported in part by the Whitaker Health Sciences Fund, a M.I.T. Faculty Research Grant, and the William W. Harris Children Charity Trust.

References 1. Apuzzo ML, Sabshin JK: Computed tomographic guidance stereotaxis in the management of intra cranial mass lesions. Neurosurgery 12:277-285, 1983. 2. Brown RA: A computerized tomography-computer graphics approach to stereotaxic localization. J Neurosurg 50:715-720, 1979. 3. Heilbrun MP, Roberts TS, Apuzzo ML, Wells TH, Sabshin JK: Preliminary experience with BrownRoberts-Wells (BRW) computerized stereotaxic guidance system. J Neurosurg 59:217-222, 1983.

32 Computer-assisted Stereotaxic Laser Microsurgery Patrick J. Kelly, M.D.

Tumors in and near the third ventricle may be biopsied accurately and safely utilizing CT-based stereotaxic techniques (1, 7). Computerized tomography (CT) provides a precise three-dimensional data base and since 1978 various systems have been developed that allow translation of points on CT scans into stereotaxic coordinates (2, 5, 10, 15). However, CT scanning is also precise volumetric data which, if gathered under stereotaxic conditions, can be utilized to reconstitute defined tumor volumes into stereotaxic space (8, 9, 12). These volumes may then be treated by interstitial radionuclides or resected utilizing computer-interactive stereotaxic control (8, 9, 11). We have applied the concept of volume stereotaxis to the treatment of subcortical tumors since 1979 to resect aggressively tumors located in the thalamus, basal ganglia, and subcortical white matter with acceptable postoperative results (11). Our experience has indicated that the method may be applied to all subcortical intraaxial lesions including tumors in and adjacent to the third ventricle, which are well suited to the instruments and technique. Computer-assisted stereotaxic procedures are performed in three stages: data acquisition, surgical planning, and the interactive surgical procedure.

Data Base Acquisition Stereotaxic Headholder A CT-compatible stereotaxic head holder is applied to the patient under local anesthesia. The head holder fixes to the skull by four carbon fiber pins inserted into drill holes through the outer table of the patient's skull into the diploe. Micrometer attachments to the vertical supports of the stereotaxic head holder provide a precise means for replacing the head holder if data acquisition and operation are performed on separate days. Stereotaxic CT Scanning A CT table adaptation plate secures the CT-compatible stereotaxic head holder to the CT table. A CT localization system fixes to the head holder

during CT scanning. This consists of nine reference bars in three sets of three rods each, which are arranged in the shape of the letter N and are located on either side of the head and anteriorly (Fig. 32.1). The localization system produces a set of nine reference marks on each CT slice (Fig.

32.2). The height and inclination of the plane of the CT slice with reference to the stereotaxic instrument can be calculated from the measured distance of the middle reference mark corresponding to the oblique reference bar to the end reference marks, which correspond to the vertical bars. Stereotaxic coordinates for any point on a CT slice can easily be calculated using mathematical transformations described previously (12). Stereotaxic Positioning Device The stereotaxic instrument is based on the arc-quadrant principle. The patient's head is moved with 3° of freedom to place an intracranial target into the focal point of a fixed sphere defined by arc and quadrant (Fig. 32.3). The frame consists of a three-dimensional slide system, a small internal arc-quadrant for directing probes and retractors intracranially, and a 400-mm horizontal arc-quadrant that holds the microscope and laser manipulation system. The stereotaxic head holder fixes to a threedimensional slide system that is activated by servomotors. Stereotaxic positions are output by optical encoders on each axis and are displayed on digital readouts. With this system, multiple areas within the target tumor volume may be accessed quickly and conveniently. Surgical Instrumentation Certain specialized surgical instruments are also used during stereotaxic laser craniotomies. These consist of stereotaxic retractors of 2- and 3-cm diameters that mount on and are stereotaxically directed by an internal arc-quadrant. Dilators for each retractor are used to spread the laser incision for advancement of the retractor during the procedure. Tear drop and fenestrated suckers and an extra-long stereotaxic bipolar forceps are essential. Long dissectors and alligator scissors must be custom made at present.

Surgical Planning Stereotaxic Volume Interpolation The archived data tape from the stereotaxic CT scan is read into the operating room computer system (independent physician's diagnostic console for the General Electric 8800 and 9800 CT scanning units). The CT slices that demonstrate the lesion are displayed. The surgeon digitizes all slices that demonstrate the tumor as follows: First, the computer recognizes the nine reference artifacts on the CT slice utilizing an intensity detection algorithm. The surgeon then inputs a series of points on the boundary of the tumor utilizing the cursor and trackball subsystem (Fig. 32.4). The computer places this series of points into a three-dimensional image matrix in relationship to their stereotaxic position, which has been calculated from the nine localizing artifacts, and then connects these points into a closed contour. All subsequent CT slices demonstrating the tumor are similarly digitized. The computer suspends all of these tumor outlines within a three-dimensional image matrix in relationship to their stereotaxic position. An interpolation program then creates intermediate slices at 1-mm intervals, and each of the digitized and interpolated slices is filled with 1-mm3 voxels. This creates a three-dimensional volume within the computer image matrix having the same configuration, spatial relationships, and location in stereotaxic space as the tumor within the patient's brain in reference to the stereotaxic localization system. If the surgeon specifies approach angles from horizontal and vertical planes (quadrant and arc angles, respectively), this is interpreted as a viewline by the computer. A series of two-dimensional slices of the tumor volume cut perpendicular to the surgical viewline may be displayed on an operating room video monitor at the precise depths encountered during operation. Viewline Selection The surgeon selects a surgical trajectory (viewline) after consideration of several factors. First, the most direct route to the tumor is optional but

the location of important cortical and subcortical areas and the direction of major white matter pathways that overlay the tumor must be considered. A trajectory that avoids major vascular structures may also be selected. In addition, a surgical approach that traverses the lesion along its major axis may be optimal as regards the efficiency with which the procedure may be performed. The surgical trajectory is expressed in arc and quadrant approach angles that will be set on the stereotaxic instrument. The mechanical control provided by the stereotaxic frame and the computer display of the lesion with reference to the frame will keep the surgeon oriented to the tumor volume at all times during the procedure. Surgical Stereotaxic Laser The carbon dioxide laser is a convenient method for removing tissue from the bottom of a deep cavity and is especially useful in the removal of deep-seated intracranial tumors. In addition, its thermal effects afford some degree of hemostasis. At high power settings its thermal damage is limited to the spot size of the laser and a thin zone of thermal necrosis usually no more than 200 to 300 mm. In our system the laser beam is reflected by mirrors on the stereotaxic frame that are controlled with precision by the computer. A laser manipulator system (microslad) attaches to the operating microscope on the stereotaxic frame. The laser beam is delivered to the manipulator system by an articulated arm and focusing lens. Within the microslad, x and у mirrors deflect the laser beam toward the focal point of the stereotaxic frame. The pitch of these mirrors is controlled by galvanometers whose voltage input is controlled by the digital to analog output of the computer system. The computer plots the position of the laser beam on the operating room video monitor as a cursor that is overlayed on reformatted tumor outlines sliced perpendicular to the surgical viewline at the level at which the microscope and surgical laser are focused (Fig. 32.5).

Surgical Procedure The surgical approach to lesions of the third ventricular region depends on whether the tumor is located within the third ventricle entirely or is located primarily within the thalamus or hypothalamus and displaces the third ventricle. The approach to anteriorly situated intraventricular lesions is through the lateral ventricle anteriorly. On the other hand, the surgical approach to thalamic tumors is through the frontal white matter and anterior limb of the internal capsule. Tumors of the posterior thalamus are best approached through the temporooccipital junction. Intraventricular Lesions After the data acquisition and surgical planning phases have been completed the patient is placed under general anesthesia and the CTcompatible stereotaxic head frame is replaced on the patient's head utilizing the same pin holes in the skull and micrometer settings that had been utilized during the stereotaxic CT scan. A central point within the tumor, selected by the surgeon, is positioned into the focal point of the stereotaxic frame. After preparing and draping the scalp, the surgeon makes an incision at the hairline approximately 3 in. long (Fig. 32.6). The skull is opened by a 1.5-in. cranial trephine and the dura mater is opened in a cruciate manner. The frontal cortex is coagulated with bipolar forceps and a vertical incision through the pia is made with scissors. The incision is deepened into the frontal white matter using the carbon dioxide laser as the laser manipulator system and microscope are advanced toward the focal point of the stereotaxic frame. The stereotaxic retractor is inserted into the cortical opening, and further deepening of the incision is done, directing the laser beam through the retractor as follows: The incision is extended 1.5 cm deep to the end of the retractor with the laser. The surgeon then inserts the dilator and the retractor is advanced over the dilator. The dilator is withdrawn, and the incision is extended deeper using the laser. Upon entering the lateral ventricle, the surgeon drains ventricular fluid with the sucker and advances the retractor to the floor of the lateral ventricle (Fig. 32.7, A and B). The fenestrated suction is used against a small cotton ball to keep ventricular fluid away from the surgical field. With the laser at 10 watts with a focused spot, an incision is made in the floor of the lateral ventricle lateral to the foramen of Monro and lateral to the septal and thalamostriate veins (Fig. 32.7C). This incision is deepened to the tumor. Gentle retraction is applied to the incision with the fenestrated suction as the laser is used to dissect ependyma of the superior portion of the third ventricle from the superficial aspect of the tumor (Fig. 32.8). Then the dilator is inserted through the retractor into this opening over the tumor and rotated 90° as the surgeon slides the retractor deeper (Fig. 32.9, A to D). This creates a shaft from the surface to the outer border of the tumor (Fig. 32.9E). The surgeon can measure the depth of the retractor from position calibrations in reference to the fixed radius of the stereotaxic arc-quadrant. At a depth corresponding to the superficial boundary of the tumor, which has been calculated along the viewline by the computer, the first slice of the tumor is displayed on the operating room video monitor (Fig. 32.10). The surgeon vaporizes the interior of the tumor utilizing a defocused laser beam with a 1 - to 2-mm spot size and 50 to 80 watts of power. The position of the laser within the tumor is displayed on the video monitor in reference to tumor outlines reconstructed from the CT-defined tumor volume (Fig. 32.11). This continues as the surgeon progresses from the most superficial aspect of the tumor to the deepest. A thin capsule of

tumor is left attached to the third ventricular ependymal wall. This cytoreduction process is easily accomplished by keeping the retractor within the outlines of the tumor using the computer display as a reference (Fig. 32.12). If the tumor is larger than the retractor, the computer image may be translated on the screen in reference to the retractor and new stereotaxic frame settings are calculated. In practice, a technician activates the servomotors controlling the three-dimensional slide to move the patient's head within the stereotaxic arc-quadrant and place another portion of the tumor under the retractor. The retractor will always be directed at the focal point of the stereotaxic arc-quadrant. After creating a large cavity within the tumor, the stereotaxic retractor is withdrawn to the superficial portion of the tumor and the capsule is

drawn under the retractor. This is accomplished by grasping it with a biopsy forceps or suction tip. The capsule may be gradually dissected away from the ventricular ependyma using short bursts of low power (10 to 15 watts) defocused laser energy, which results in contraction of the capsule as it separates from the ependyma (Fig. 32.13). It is advised that the procedure not extend too deeply on one side of the tumor before the other sides are freed. After portions of the capsule are freed, the plane of dissection is preserved by placing a cotton ball between ependyma and tumor capsule. This will result in the tumor capsule being directed into the center of the stereotaxic retractor. The capsule of the tumor may then be contracted by low power defocused laser energy and then vaporized utilizing higher power settings on the laser. The retractor is advanced to facilitate deeper capsule dissection as necessary. Ultimately, the entire capsule may be removed in this manner. Throughout the procedure, hemostasis should be meticulously maintained. An extra-long bipolar

forceps and defocused low power laser energy may be used for this purpose. After removal or, in some cases, internal resection of the tumor the retractor is slowly withdrawn as hemostasis is secured along the tract. Pre- and postoperative scans of patients who have undergone this procedure are presented in Figure 32.14. Paraventricular Thalamic Lesions We have found that in stereotaxic laser resections of tumors of the thalamus bordering on the third ventricle patients do better if the ventricular system is avoided and the dorsal thalamus is left intact. A surgical approach to thalamic lesions through the anterior limb of the internal capsule parallel to white matter fibers is advised. In this way, only anterior thalamus and anterior limb of the internal capsule risk injury, which should be well tolerated in the unilateral situation. Lesions located in the posterior regions of the thalamus can be approached from a temporooccipital incision that is directed horizontally and traverses the posterior temporal horn of the lateral ventricle. This may result in a contralateral superior temporal quadrantanopsia postoperatively. With either approach, the computer-assisted stereotaxic technique will maintain orientation within the lesion as tumor borders delineated by stereotaxic CT scanning and reformatted in three-dimensional space are portrayed to the surgeon interactively during the procedure. A surgical trajectory is planned to access the thalamic neoplasm from a trephine that will be performed at the frontal hairline in the case of

anterior thalamic tumors or at the temporooccipital junction if the approach to a posterior thalamic neoplasm is to be through the temporooccipital white matter. It is desirable to have the surgical approach along the long axis of the tumor to minimize the amount of retraction necessary to visualize the entirety of the lesion. To obviate the possibility of intraoperative shifts of position of the tumor in stereotaxic space, one should perform the following step. This step is especially important in very large or cystic lesions or in cases where there is a possibility of entering the ventricular system. A series of stainless steel reference balls (0.5 mm in diameter) are deposited in the lesion utilizing a biopsy cannula stereotaxically directed through a 1/8-in. twist drill hole in the skull. Anteroposterior and lateral roentgenograms are then obtained before craniotomy and dural opening. Spatial shifts of the tumor detected by movements of the stainless steel balls on subsequent roentgenograms and resulting from decompression of the brain during tumor removal, cyst drainage, or

escape of ventricular fluid can be corrected within the three-dimensional computer matrix. The scalp, bone, and dural openings and cortical and subcortical white matter incisions are made and deepened as described in the previous section. At the outer border of the tumor, the incision is undercut by reflecting the laser with a stainless steel instrument. Using the dilator, the retractor may be advanced to create a shaft through the brain. Deep to the retractor will be found the superficial aspect of the tumor. At this point the first slice of the tumor volume reconstructed from the stereotaxic CT is displayed on the operating room computer system. First, the surgeon uses the laser to cut around the tumor to separate it from

surrounding brain tissue. The computer has displayed the position of the laser beam in reference to the outline of the tumor at the level to which the stereotaxic retractor has been advanced and at the level at which the microscope and laser are focused. As the tumor is progressively separated from brain tissue, the retractor is advanced and new slices of the tumor are displayed on the screen. After the tumor has been isolated from surrounding brain tissue or in the event that it fills the retractor to obscure visualization, it may be resected down to the deepest extent of the retractor using biopsy forceps, defocused high power laser (85 watts, 2- to 5-mm spot size), or suction. Then the process of separating deeper levels from surrounding brain tissue continues. If the tumor is larger than the retractor, the patient's head can simply be moved with the servomotor-controlled stereotaxic instrument to place another portion of the tumor directly under the retractor as described previously. The depth of the procedure is controlled by noting the depth of the retractor in relationship to the stereotaxic arc-quadrant and by stereotaxic teleradiographs (Fig. 32.15). In addition, the stereotaxically inserted reference balls will give an indication of the depth of the procedure as they are encountered. Ultimately, the final slices of the tumor are vaporized from superficial to the deepest layer as the surgeon monitors not only the operative field,

but also the position of the laser beam represented by the cursor on the computer display monitor in relationship to the outlines of the tumor. Bleeding is controlled utilizing defocused low power laser energy with a power density of less than 10 watts/cm2 or by an extra-long (150-mm) bipolar forceps (Radionics Incorporated, Burlington, Massachusetts). Hemostasis is secured as the stereotaxic retractor is slowly withdrawn. A tight dural closure and double-layer scalp closure, after the trephine bone plug is secured, is utilized. Figures 32.16 and 32.17 demonstrate preand postoperative scans in two patients with thalamic pilocytic astrocytomas resected by this procedure. An 8-year-old boy (Fig. 32.16) had presented with apraxia of the right upper extremity of 3 years' duration. The 3-cm3 contrast-enhancing lesion in the left thalamus was approached anteriorly. Postoperatively no worsening of his right-sided apraxia was noted and speech was normal. He was discharged from the hospital 6 days after operation. CT scanning 18 months after operation demonstrated no evidence of residual or recurrent tumor. He continues to do well. Postoperative radiation therapy was not administered. The second patient is a 5-year-old girl who presented with increased intracranial pressure and mild right hemiparesis with dyskinesia. Her

tumor in the posterior left thalamus was resected from a temporooccipital approach (Fig. 32.17). A ventriculoperitoneal shunt had been inserted at another institution. Postoperatively she was neurologically normal and we were unable to document a visual field deficit. No residual tumor was noted on postoperative CT scanning.

Discussion Traditionally neurosurgeons have relied on knowledge of neuroanatomy and hand-eye coordination to guide their surgical procedures. However, a surgeon's three-dimensional orientation decreases the deeper a procedure extends below the cortical surface. Surgeons must therefore depend on the identification of normal anatomy that may be distorted by the pathological condition to avoid getting lost in their attempts to treat deep subcortical lesions. Within the lateral and third ventricles surgeons attempt to maintain their spatial orientation by structural landmarks such as the foramen of Monro, thalamostriate vein, choroid plexus, and fornix (19, 21). This may not be particularly difficult when the ventricles are dilated and visualization is optimal. However, a surgeon may have difficulty maintaining orientation within the lateral ventricle and indeed may have difficulty even finding the lateral ventricle and the structures within it in patients with normal-sized or small ventricles (3, 13, 14, 16, 18). It has therefore usually been recommended that the surgical procedure should traverse the corpus callosum to gain access to the third ventricle in patients with small or normal-sized lateral ventricles (1, 20). The structures of the corpus callosum and branches of the anterior cerebral artery can then spatially orient the surgeon. A callossal incision made perpendicular to its fibers is usually well tolerated (1) although it may be associated with some memory impairment and personality changes (4, 7). However, because normal anatomical structures must be identified in both the transventricular and transcallosal approaches, the craniotomies, cortical incisions, and openings into the ventricular system or through the corpus callosum may be more extensive than really necessary to deal only with the specific lesion. This problem is more serious for tumors located in the wall of the third ventricle and in the thalamus. Conventionally, the transventricular approach to identify normal ventricular system landmarks has been recommended before the tumor can be approached through the inferior wall of the lateral ventricle and through dorsal thalamic nuclei (6). Unfortunately, if the tumor is found it may not be clear where tumor ends and normal brain begins. The results of spatial miscalculations in this regard are severe neurological complications. Therefore, most procedures listed initially as "tumor excisions" are, in fact, usually nervous biopsies that could have been accomplished utilizing stereotaxic techniques with less hazard to the patient (2, 5, 10, 17). However, stereotaxic methodology provides a means of effectively biopsying lesions of and around the third ventricle and also may be used to guide a controlled craniotomy for tumor excision. The advantage of the stereotaxic approach described is that it maintains three-dimensional surgical orientation with respect to a computerinterpolated CT-defined tumor volume. The surgical approach that will minimize exposure of and injury to essential brain structures can be selected. Aggressive resections of circumscribed tumors from neurologically important subcortical areas can be accomplished with a high degree of accuracy. Recurrent or residual low grade lesions of the third ventricle or thalamus may be treated by a second stereotaxic procedure utilizing the same bone flap and surgical pathway as the first. It is probably best

to deal with especially large lesions in two separate operations performed several months apart. The instrumentation described in this report may seem complicated to many and unrealistically expensive to some. Nevertheless, commercially available stereotaxic instruments could be adapted to the procedures described in this chapter. In addition, computer systems are becoming more powerful and less expensive. Stock software could be adapted to stereotaxic volumetric surgery by individuals and instructions willing to make a commitment to the task. References 1. Apuzzo MLJ, Chikovani OK, Gott PS, Teng EL, Zee CS, Gianotta SC, Weiss MH: Transcallosal, interforniceal approaches for lesions affecting the third ventricle: Surgical considerations and consequences. Neurosurgery 10:547554, 1982. 2. Apuzzo MLJ, Chandrasoma PT, Zelman V, Giannotta SL, Weiss MH: Com puted tomographic-based stereotaxis in the management of lesions of the third ventricular region. Neurosurgery 15:502-508, 1984. 3. Busch E: A new approach for the removal of tumors of the third ventricle. Acta Psychiatr Scand 19:57-60, 1944. 4. Geffen G, Walsh A, Simpson D, Jeeves M: Comparison of the effects of transcortical and transcallosal removal of intraventricular tumors. Brain 113:773-788, 1980. 5. Heilbrun MP, Roberts TS, Apuzzo MLJ, Well TH Jr, Sabshin JK: Preliminary experience with Brown-Roberts-Wells (BRW) computerized tomography ster eotaxic guidance system. J Neurosurg 59:217-222, 1983. 6. Hirsch JF, Zouaoui A, Reinger D, Pierre-Kahn A: A new surgical approach to the third ventricle with interruption of the striothalamic vein. Acta Neurochir (Wien) 47:135-147, 1979. 7. Jeeves MA, Simpson DA, Geffen G: Functional consequences of the trans callosal removal of intraventricular tumors. J Neurol Neurosurg Psychiatry 42:134-142, 1979. 8. Kelly PJ, Alker GJ, Goerss S: Computer-assisted stereotaxic laser microsur gery for the treatment of intracranial neoplasms. Neurosurgery 10:324-331, 1982. 9. Kelly PJ, Kail BA, Goerss S: Computer simulation for the stereotactic place ment of interstitial radionuclide sources into computed tomography-defined tumor volumes. Neurosurgery 14:442-448, 1984. 10. Kelly PJ, Alker GJ, Kail BA, Goerss S: Method of computed tomographybased stereotactic biopsy with arteriographic control. Neurosurgery 14:172177, 1984. 11. Kelly PJ, Kail B, Goerss S, Alker GJ: Precision resection of intra-axial CNS lesions by CT-based stereotactic craniotomy and computer monitored CO2 laser. Acta Neurochir (Wien) 68:1-9, 1983. 12. Kelly PJ, Kail BA, Goerss S: Transposition of volumetric information derived from computed tomography scanning into stereotactic space. Surg Neurol 21:465-471, 1984. 13. Kempe LG: Operative Neurosurgery. New York, Springer-Verlag, 1968, vol 1. 14. Little JR, MacCarty CS: Colloid cysts of the third ventricle. J Neurosurg 40:230-235, 1974. 15. Lunsford LD, Martinez AJ: Stereotactic exploration of the brain in the era of computed tomography. Surg Neurol 22:222-230, 1984. 16. McKissock W: The surgical treatment of colloid cyst of the third ventricle: A report based on twenty-one personal cases. Brain 74:1-9, 1951. 17. Menezes AH, Bell WE, Perrett GE: Hypothalamic tumors in children: Their diagnosis and management. Childs Brain 3:265-280, 1977. 18. Poppen JL, Reyes V, Horrax G: Colloid cysts of the third ventricle. J Neuro surg 10:242-263, 1953. 19. Rhoton AL Jr, Yamamoto I, Peace DA: Microsurgery of the third ventricle: Part 2. Operative approaches. Neurosurgery 8:357-373, 1981. 20. Shucart WA, Stein BM: Transcallosal approach to the anterior ventricular system. Neurosurgery 3:339-343, 1978. 21. Viale GL, Turtas S: The subchoroid approach to the third ventricle. Surg Neurol 14:71-76, 1980.

33 Radiotherapy of Pineal and Suprasellar Tumors William M. Wara, M.D., and Philip H. Gutin, M.D.

Radiotherapy is commonly used in the treatment of pineal and suprasellar tumors. Because of the differences in histological classification of tumors occurring in this region (i.e., germ cell tumors, pineal parenchymal tumors, glial tumors, and benign tumors) and their differing biological behavior, selection of the appropriate treatment regimen requires knowledge of the tumors and of the value and limitations of the various therapeutic options. Physical Factors Pertinent to Radiotherapy Megavoltage irradiation (x-rays and gamma rays with energies of 1 to 40 MeV) is clinically used in the treatment of central nervous system (CNS) tumors. The biological effects of these types of radiation are due to their ability to produce charged ions and free radicals that secondarily cause permanent deoxyribonucleic acid damage. The distribution of energy deposited in tissues varies with the energy of the radiation, the size of the radiation source (60Co), the size of the beam treatment, the use of absorbing filters, the source to target distance, and the collimation system. Computer dosimetry that incorporates shape-shielding blocks, wedge filters, and individually designed tissue-compensating filters are used to localize the high dose volume and to minimize irradiation to sensitive normal tissues. In selected situations, interstitial or intracavitary radiation sources may be used.

Biological Factors The biological basis of radiation therapy is complex. The success of this treatment depends on including careful treatment planning to maximize the high dose region, the differential sensitivity of normal and tumor cells, the total volume of tumor, the repopulation of tumor cells in the irradiation volume, the rearrangement of cells within the cell cycle between individual radiation exposures, the differential capability of tumor and normal cells to repair sublethal radiation damage, and the amount of oxygen within the usually hypoxic tumor cells. Experimentally, each of these factors has been shown to be significant but in the complex clinical situation it is difficult to evaluate them simultaneously. In practice, treatment is selected to provide optimal physical distribution of radiation with minimal risk to surrounding critical normal structures. Radiation Tolerance of the CNS Radiation doses utilized for the treatment of most brain tumors carry a definite but poorly quantitated possibility of damage. The larger the daily dose, the total dose, and/or the volume irradiated, the greater the risk. If damage occurs, the functional deficit varies with anatomical location and extent of injury. Clinical experience has demonstrated that there is minimal risk if the total dose does not exceed 5000 to 5500 rads and individual fractions are 180 to 200 rads (33). Higher doses may be utilized for very malignant

tumors where without radical treatment death is assured. In children less than 18 months of age the dose is usually reduced by 10 to 20%. Adverse effects of brain radiation are generally divided into three groups depending upon time of appearance. Acute reactions during the course of therapy are probably secondary to cerebral edema. With conventionally fractionated radiotherapy, i.e., daily fractions of 180 to 200 rads, these reactions are uncommon and usually are reversible with corticosteroid treatment. Onset of the acute delayed reaction varies from a few weeks to 4 months after completion of irradiation. This reaction is thought to be due to a temporary demyelination. It tends to mimic the original tumor and new symptoms and occurs in approximately 25% of patients. The reaction is usually self-limited and no therapy is required. It is important to recognize this reaction so that there will be no premature alteration or institution of revised treatment for this situation. The delayed severe effects occur 6 months to 10 years posttreatment. The pathological mechanism is thought to be late damage of the vascular supply of the irradiated area. The onset is usually insidious and may progress to scarring or necrosis. The complications are infrequent and rare with doses of radiation below 5000 rads. In children who have received low dose irradiation, computerized tomographic (CT) scans have shown ventricular dilatation, wide arachnoid spaces, atrophy, and intracerebral calcifications (26). Psychosocial studies of irradiated children have shown impairment in quantitative tests and performance with abstract material after irradiation (10). The incidence of dysfunction and its relationship to various combinations of chemotherapy and radiation are currently being prospectively evaluated. Radiation to the pituitary and hypothalamic areas can produce growth hormone deficiency. This complication can occur in the treatment of tumors that do not involve the pituitary or hypothalamic area. Shalet and coworkers have reported decreased response of plasma growth hormone to insulin hyperglycemia in 11 of 16 children treated for intracranial tumors (32). Using insulin, arginine, and L-dopa challenge our group has documented similar growth hormone deficiencies in 8 children after CNS radiation (28). The incidence of this pituitary dysfunction is unknown as the studies reported have been on groups of patients with pituitary dysfunction. All children who have been treated to the pituitary and suprasellar areas should be followed with serial growth curves. If a decreased growth rate

is documented a complete endocrine evaluation should be undertaken. Conventional Radiotherapy Conventional radiotherapy is utilized in the treatment of the majority of pineal and suprasellar tumors. Most patients are treated with high energy (4 to 18 MeV) linear accelerators after simulation and computer dosimetry. In all types of tumors the radiation is focal to the tumor bed with a limited margin designed to spare as much normal brain as possible. The exact treatment recommendation is dependent on the pathological diagnosis. The dose of irradiation to the primary tumor has generally ranged from 4500 to 5500 rads given at 180 to 200 rads per day. Although germinomas might be more sensitive, no trial using a lower dose has been performed. Adjuvant radiotherapy to uninvolved areas of the brain and spine has used a dose of 2500 to 3500 rads. The need to treat the spinal area is controversial. The overall incidence of spinal seeding from pineal tumors and suprasellar germinomas is approximately 10% (38). Therefore, we do not recommend routine craniospinal axis irradiation (Table 33.1). However, there may be a subpopulation of patients who might benefit, so we screen all patients with Cerebrospinal fluid cytology and a myelogram. A positive myelogram would necessitate whole axis irradiation. Germ Cell Tumors Because of the operative morbidity associated with surgery for these locations, historically many of these tumors were often irradiated without a biopsy. Now such patients may receive a therapeutic trial of 2000 rads and if, on rescanning, the tumor has markedly shrunk, an empirical diagnosis of germinoma is made and the irradiation is completed to 5000 rads. If a biopsy is obtained we recommend a dose of 6000 rads for all germ cell tumors other than germinomas. Reported survival rates for patients with these types of tumors are good. The 2-year survival rate for various types is approximately 60 to 70% (6, 17, 24, 27, 30, 37, 38). Germinomas seem to

have higher survival rates, but the lack of biopsy in many cases makes it difficult to separate other tumor types (Table 33.2). Approximately 50% of patients experience improvement in symptoms after radiotherapy. With germinomas, CT scans return to normal but the more sensitive magnetic resonance scan shows a stable residual lesion. Treatment failures in our experience are usually recurrences at the primary tumor site. Pineal Parenchymal Tumors Primary pineal tissue tumors can be low grade Pineocytomas or malignant pineoblastomas. The pineocytoma is given local irradiation to a dose of 5000 rads. The malignant pineoblastoma is treated like a medulloblastoma and given 5500 rads to the primary and 2500 to 3500 rads to the remainder of the craniospinal axis. These patients do less well, with a survival rate of approximately 30% (38). Gliomas Gliomas in the anatomical region are treated with a regimen similar to that recommended for a more common supratentorial location. Low grade astrocytoma found in the posterior optic region, the hypothalamic area, and the suprasellar region is treated with focal irradiation, limiting the field to the minimal normal tissue possible, to a total dose of 5040 rads at 180 rads per day. A dose reduction of 20% is made for children less than 2 years of age at time of diagnosis. The rare patient's results in this group follow the more common supratentorial astrocytoma. The experience at UCSF was summarized by Leibel et al., who reviewed the results for all astrocytomas treated between 1942 and 1967. Patients were divided into three groups. In 14 patients with total tumor resection, none received irradiation and no tumor recurred. In 37 patients with incomplete resection but no postoperative irradiation the 5-year recurrence-free survival was 19%. The third group was composed of 71 patients who received postoperative irradiation after subtotal removal; they had a 46% recur-

rence-free survival. Based on this experience all subtotally removed astrocytomas are irradiated with the expectation of a 50% long term survival (19). The most difficult brain tumors to treat are the malignant gliomas. Fortunately, in this region they are quite rare. Data accumulated over the past decade has indicated a 16% survival for anaplastic astrocytoma and a 0% survival for glioblastoma multiforme. An increase in radiation dose from 5000 to 6000 rads can increase median survival from 28 to 42 weeks in these patients (39). Because of the lack of results in this area new approaches are now being investigated including hyperfractionation radiotherapy and the addition of multidrug chemotherapy and radiosensitizers. Craniopharyngioma Although these tumors are histologically benign, their proximity to critical structures causes major morbidity and difficulty for complete surgical removal. Repeated attempts at complete removal usually result in a poor quality of life. We have reviewed 8 patients with total removal, 12 patients with subtotal removal and irradiation, and 8 patients with biopsy and irradiation treated at UCSF from 1956 to 1974. The majority of the patients received focal irradiation to doses between 5000 and 5500 rads. Five-year survival rates were similar in all patients. Seven of 8 patients (88%) with total removal were alive, 11 of 12 patients with surgery plus subtotal irradiation (88%) were alive, and 8 of 8 patients (100%) with biopsy plus irradiation were alive (Table 33.3). Quality of life and complications were similar in both groups (29). Radical resection of craniopharyngiomas should be considered but only if it can be accomplished safely. Based on our data we currently recommend conservative resection followed by focal irradiation if tumor is left behind. Our current length and quality of survival is similar to that in previous reports (16).

Localized Radiotherapy: Intracystic Isotope Therapy Background For cystic craniopharyngiomas and thalamic gliomas teletherapy is the principal radiation modality used in the United States. However, at a number of European centers stereotaxic puncture and instillation of radioactive isotopes is the primary treatment for these tumors, particularly craniopharyngiomas, replacing microsurgical approaches and external irradiation. Stereotaxic instillation of a radioisotope was first performed in a patient with a cystic craniopharyngioma recurrence by Leksell at the Karolinksa Hospital (20). This patient was treated with colloidal phosphorus-32, and since that time a large number of patients have been treated with this isotope as well as colloidal suspensions of gold-198, ytrrium-90, and rhenium186(1, 2, 23, 31, 36, 34). Colloidal Isotopes For maximally localized treatment and ease of handling, pure beta emitters, like 32P and 90Y, are preferred against cystic craniopharyngioma (2). Although the lesser penetration of 32P might be an advantage in thin-walled cysts, the shorter half-life of 90Y makes evacuation of the cyst possible after a shorter interval because the prescribed dose is administered more quickly (2). A review of the recent literature seems to favor 90Y as the superior isotope against cystic craniopharyngioma (2, 23, 34). Unfortunately, only 32P is available in the United States.

Treatment In careful hands a therapeutic radioisotope is delivered to tumor cysts through a clean stereotaxic puncture (preferably computerized tomography-directed). The instillation is best monitored by use of intraoperative gamma-camera control (23) to assess isotope leakage from the cyst, a potentially disastrous complication. Beta radiation doses to the cyst wall have ranged from 5 to 100 krads (2). However, 20 krads is the dose used in most recent series. Szikla has favored colloidal 186Re for intracystic therapy of craniopharyngiomas and glioma cysts. This isotope has a combined beta and gamma emission, the beta energy (358 kev) of 186 Re being lower than that of 90Y (2.27 MeV), and therefore relatively limiting irradiation of surrounding normal tissue (31, 36). However, this lower penetration may have led to cyst recurrence in four of seven patients treated with 186 Re by Sturm and his coworkers (23). Szikla

maintains, however, that 186Re instillation has led to regression of cyst fluid formation in all craniopharyngiomas and low grade gliomas treated (31, 36). The experience with the treatment of patients with intracystic 90Y has been very favorable at Karolinska ( 1 , 2 ) and at Heidelberg (23). Radiographic regression of the craniopharyngioma cyst and concomitant clinical improvement can be expected. Radiation damage to the optic nerves is seen in about 5% of cases (23, 34). In the United States a small experience with instillation of the only available colloidal radioisotope, 32P, into recurrent, previously irradiated craniopharyngioma cysts has been accrued. Several centers around the country are now using this modality as an experimental primary therapy-

Stereotaxic Radiosurgery Background In 1959, Leksell and his coworkers introduced the prospect of stereotaxically directing highly collimated beams of radiation to destroy predictable and limited volumes of brain (21). By this technique, radiation is delivered as a single necrotizing dose from, as it is currently formulated, the gamma unit, an array of 179 cobalt-60 sources. The gamma irradiation beam from each source is directed toward the center of a collimator helmet at which the target has been stereotaxically positioned. The targets treated at Karolinska Hospital, where, until recently, the only gamma units resided, have included thalamic nuclei, arteriovenous malformations, acoustic tumors, pineal region tumors, pituitary tumors, and craniopharyngiomas (3). Other methods of stereotaxic radiosurgery depend upon the Bragg peak phenomenon for radiation dose localization and have included the proton beam at the Harvard University cyclotron (18, 35) and the heavy particle beams at the Lawrence Berkeley Laboratory (7). It is thought that, with careful and creative treatment planning, conventional linear accelerators can also be adapted for obliterative stereotaxic radiosurgery, and small series of patients are now being accrued (8, 14, 15). This technique may be very useful for tumors in and around the third ventricle. Craniopharyngioma Backlund has used the gamma unit to treat a number of solid craniopharyngiomas or the solid remnants of cystic craniopharyngiomas, demonstrating reduction in tumor size with regres-

sion of symptoms and without complication (3, 4). The cystic tumors had been treated previously with 90Y instillation. This is, in fact, the approach that Backlund now advocates for craniopharyngioma. A stereotaxic puncture with radiocolloid (90Y) instillation is done and followed by gamma unit radiosurgery if there are solid portions to the tumor (4). There is a very limited experience with proton irradiation of craniopharyngioma (18) and a diminishing small experience with stereotaxic linear accelerator treatment of craniopharyngioma (14). Pineal Region Tumors A small number of pineal region ependymomas, astrocytomas, and Pineocytomas have been treated on the gamma unit to doses of 2000 to 7500 rads after stereotaxic biopsy. Results have been variable (3). Tumors chosen for radiosurgery have been limited by size and geometry. Two patients with pineal region tumors have been treated with stereotaxic linear accelerator techniques with good early results (less than 18 months follow-up) (8). Interstitial Brachytherapy Background The history and scientific basis for the interstitial irradiation of brain tumors have been reviewed (5). The possibility of delivering from implanted sources a localized dose of radiation to tumors in and around the third ventricle, while sparing surrounding normal brain, is attractive, particularly in children, where such tumors commonly occur. However, the specter of even the

slightest brain injury in this region blunts one's enthusiasm because the normal structures here are so critical to neurological function. Thalamic and optic gliomas, pineal region tumors, and, in lesser numbers, several other tumors in this area have been treated with stereotaxically implanted iodine-125, and iridium-192 sources (11, 12, 22, 25, 40). Mundinger has treated a large number of patients with low grade or anaplastic gliomas in the region of the third ventricle by stereotaxic biopsy and interstitial irradiation from permanently implanted 125I and 192Ir sources, 125I being preferred for the better-demonstrated lesions (22). In some cases external beam treatment was also given. Mundinger's group has also treated many patients with pineal region tumors of all varieties with permanent implants (40). As could be predicted, tumor seeding along the Cerebrospinal fluid pathways was a problem in some cases of germinoma. Pecker et al. have reported treating two patients with germinomas similarly (25). As can be appreciated from the foregoing discussion of external beam radiotherapy, this modality is relatively effective against tumors in the region of the third ventricle. For this reason, in the United States external beam irradiation is used as the primary treatment for these lesions. Interstitial brachytherapy is reserved for reirradiating selected recurrences with the appropriate geometry or for giving a "boost" dose of radiation to the tumor bed after external radiotherapy. Reirradiation of tumors in the region of the third ventricle must be done with the full realization that catastrophic brain injury might result. Patients receiving interstitial implants for brain tumor recurrences develop, in many in-

stances, focal radiation necrosis that might exacerbate neurological deficit and require operation for resection of the necrotic material (12). For tumors located, for example, in the thalamus, the deficit might be devastating and the operation might be perilous. Our group has treated several children and adults with recurrent gliomas of the optic chiasm or thalamus with permanent or removable 125I implants (11-13). Because the interstitial radiation dose has been reduced in these patients, response has been, in general, favorable and injury has been limited. Only a single patient, a 5-year-old boy with a recurrent anaplastic thalamic glioma, has required reoperation for necrosis; and the outcome has been excellent. He is surviving without recurrence and only a mild exacerbation of his hemiparesis nearly 3 years after implantation. Figure 33.1 shows the clinical course of a 17year-old boy with a cystic recurrence of a thalamic glioma who received 5000 rads to the tumor periphery from centrally implanted removable 125 I sources. The radionecrotic reaction resolved spontaneously and left him with no residual deficit. He is off steroids and functioning normally in college 2 years after implantation. Clearly, for selected patients with tumor recurrences in the third ventricular region interstitial brachytherapy can be of value. Cases must be selected with rigorous attention to tumor geography and geometry and the techniques must be applied with the utmost care to avoid devastating brain injury. References 1. Backlund E-O, Johansson L, Sarby B: Studies on craniopharyngiomas: II. Treatment by stereotaxis and radiosurgery. Acta Chir Scand 138:749-759, 1972. 2. Backlund E-O: Studies on craniopharyngiomas: III. Stereotaxic treatment with intracystic yttrium90. Acta Chir Scand 139:237-247, 1973. 3. Backlund E-O: Stereotactic radiosurgery in intra cranial tumors and vascular malformations. In Krayenbuhl H (ed): Advances and Technical Standards in Neurosurgery. Vienna, SpringerVerlag, 1979, vol 6, pp 3-27. 4. Backlund E-O: Solid craniopharyngiomas treated by stereotactic radiosurgery. In Szikla G (ed): Ster eotactic Cerebral Irradiation. Amsterdam, Elsevier/North Holland Biomedical Press, 1979, pp 271-281. 5. Bernstein M, Gutin PH: Interstitial irradiation of brain tumors: A review. Neurosurgery 9:741750, 1981. 6. Camins MB, Schlesinger EB: Treatment of tumors of the posterior part of the third ventricle and the pineal region: A long-term follow-up. Acta Neurochir(Wien) 40:131-143, 1978. 7. Castro J, Saunders WM, Tobias CA, et al: Treat ment of cancer with heavy charged particles. Int

J Radiat Oncol Biol Phys 8:2191-2198, 1982. 8. Colombo F, Benedetti A, Pozza F, Avanzo RC, Marchetti C, Chierego G, Zanardo A: External ster eotaxic irradiation on a linear accelerator. Neu rosurgery 16:154-160, 1985. 9. Danoff B, Sheline GE: Radiotherapy of pineal tu mors. In Heuwelt EA (ed): Diagnosis and Treat ment of Pineal Region Tumors. Baltimore, Wil liams & Williams, 1984, ch 16, pp 300-308. 10. Eiser C: Intellectual abilities among survivors of childhood leukemia as a function of CNS irradia tion. Arch Dis Child 53:391-395, 1978. 11. Gutin PH, Phillips TL, Hosobuchi Y, et al: Perma nent and removable implants for the brachyther apy of brain tumors. Int J Radiat Oncol Biol Phys 7:1371-1381, 1981. 12. Gutin PH, Phillips TL, Wara WM, et al: Brachy therapy of recurrent malignant brain tumors with removable high-activity iodine-125 sources. J Neurosurg 60:61-68, 1983. 13. Gutin PH, Edwards MSB, Wara WM, et al: Prelim inary experience with 125I brachytherapy of pedi atric brain tumors. Concepts Pediatr Neurosurg (in press). 14. Hartmann GH, Schlegal W, Sturm V, Kober B, Pastyr O, Lorenz WJ: Cerebral radiation surgery using moving field irradiation at a linear acceler ator facility. Int J Radiat Oncol Biol Phys 11:1185-1192, 1985. 15. Heifetz MD, Wexler M, Thompson R: Single-beam radiotherapy knife: A practical theoretical ideas. J Neurosurg 60:814-818, 1984. 16. Hoffman HJ, Hendrick EB, Humphreys RP, et al: Management of craniopharyngioma in children. J Neurosurg 47:218-227, 1977. 17. Jenkin RDT, Simpson WJK, Keen CW: Pineal and suprasellar germinomas. J Neurosurg 48:99107, 1978. 18. Kjellberg RN: Stereotactic Bragg peak proton ra diosurgery results. In Szikla G (ed): Stereotactic Cerebral Irradiation. Amsterdam, Elsevier, 1979, pp 233-244. 19. Leibel SA, Sheline GE, Wara WM, Boldrey E, Nielson S: The role of radiation therapy in the treat ment of astrocytomas. Cancer 35:1551-1557, 1975. 20. Leksell L, Liden K: A therapeutic trial with radio active isotopes in cystic brain tumor. In Radioi sotope Techniques: I. Medical and Physiological Applications, 1952, pp 1-4. 21. Leksell L: Stereotaxis and Radiosurgery. Spring field IL, Charles С Thomas, 1971. 22. Mundinger F, Weigel K: Indication and results of stereotactic curietherapy with iridium-192 and iodine-125 for non-resectable tumours of the hy pothalamic region. Acta Neurochir (Wien) Suppl 33:323-330, 1984. 23. Netzeband G, Sturm V, Georgi P, et al: Results of stereotactic intracavitary irradiation of cystic craniopharyngiomas: Comparison of the effects of yttrium-90 and rhenium-186. Acta Neurochir (Wien) Suppl 33:341-344, 1984. 24. Onoyama Y, Ono K, Nakajima T, et al: Radiation therapy of pineal tumors. Radiology 130:757760, 1979. 25. Pecker J, Scarabin J-M, Vallee B, et al: Treatment in tumors of the pineal region: Value of stereotaxic biopsy. Surg Neurol 12:341-348, 1979.

26. Peylan-Ramu N, Poplack DG, Plzzo PA, et al: Ab normal CT scans of the brain in asymptomatic children with acute lymphocytic leukemia after prophylactic treatment of the central nervous sys tem with radiation and intrathecal chemotherapy. NEnglJMed 298:815-818, 1978. 27. Rao YTR, Medini E, Haselow RE, et al: Pineal and ectopic pineal tumors: The role of radiation ther apy. Cancer 48:708-713, 1981. 28. Richards GE, Wara WM, Grumbach M, et al: De layed onset of hypopituitarism: Sequelae of ther apeutic irradiation of central nervous system, eye and middle ear tumor. J Pediatr 89:553-559, 1976. 29. Richmond IL, Wara WM, Wilson CB: Role of radia tion therapy in the management of craniophar yngiomas in children. Neurosurgery 6:513-517, 1980. 30. Salazar OM, Castro-Vita H, Bakos RS, et al: Ra diation therapy for tumors of the pineal region. Int J Radiat Oncol Blol Phys 5:491-499, 1979. 31. Schaub C, Bluet-Pajot MT, Videau-Lornet C, et al: Endocavitary beta irradiation of glioma cysts with colloidal 186rhenium. In Szikla G (ed): Stereotactic Cerebral Irradiation. Amsterdam, Elsevier/North Holland Biomedical Press, 1979, pp 293-302. 32. Shalet SM, Beardwell CG, Morris-Jones PH, Pear son D: Pituitary function after treatment of intra cranial tumors in children. Lancet 2:104-107, 1975.

33. Sheline GE, Wara WM, Smith V: Therapeutic ir radiation and brain injury. Int J Radiat Oncol Biol Phys 6:1215-1228, 1980. 34. Strauss L, Sturm V, Georgi P, et al: Radioisotope therapy of cystic craniopharyngiomas. Int J Ra diat Oncol Biol Phys 8:1581-1585, 1982. 35. Suit H, Griffin T, Almond P, et al: Particle radia tion therapy. Cancer Treat Symp 1:147-160, 1984. 36. Szikla G, Musolino A, Miyahara S, et al: Colloidal rhenium-186 in endocavitary beta irradiation of cystic craniopharyngiomas and active glioma cysts: Long term results, side effects and clinical dosimetry. Acta Neurochir (Wien) Suppl 33:332339, 1984. 37. Waga S, Handa H, Yamashita J: Intracranial ger minomas: Treatment and results. Surg Neurol 11:167-172, 1979. 38. Wara WM, Jenkin DT, Evans A, et al: Tumor of the pineal and suprasellar region: Children's Can cer Study Group treatment results 1960-1975. A report from Children's Cancer Study Group. Can cer 43:288-291, 39. Wara WM: Radiation therapy for brain tumors. Cancer (Suppl) 55:2291-2295, 1985. 40. Weigel K, Ostertag C, Mundinger F: Interstitial long term irradiation of tumors in the pineal re gion. In Szikla G (ed): Stereotactic Cerebral Irra diation. Amsterdam, Elsevier/North-Holland Bio medical Press, 1979, pp 283-292.

34 Chemotherapy of Tumors of the Third Ventricular Region MichaelS. B. Edwards, M.D., and Victor A. Levin, M.D.

The majority of tumors developing in the region of the anterior third ventricle are glial in origin (24). In general, initial treatment includes open or stereotaxic biopsy (1) followed by local irradiation and, on occasion, adjuvant chemotherapy. At the time of tumor progression, chemotherapy has been used to control tumor growth. In the chapter that follows, we discuss our experience at the University of California and the relevant literature on treating patients with third ventricular tumors with chemotherapy. Materials and Methods Patient Population Using our computerized patient data base, we found 39 of 1103 patients who were treated with chemotherapy in an adjuvant or recurrent fashion for glial tumors in the third ventricular region (hypothalamus/chiasm/thalamus). The primary location and histology of these 39 patients are listed in Table 34.1. Thirty-six of the 39 tumors were glial in origin. The age range for hypothalamic/chiasm tumors was 3 to 60 years (5 patients were less than 17 years of age or younger) and for thalamic tumors was 1 to 64 years (6 were 17 years of age or younger). There were 21 males and 18 females. A literature search revealed 10 cases of successful (response) chemotherapeutic treatment of pineal region germ cell tumors (21). In addition, we have successfully treated 3 other patients with pineal germ cell tumors and 2 pa-

tients harboring pineoblastomas. These patients and their treatments are listed in Table 34.2. Chemotherapy—Glial Tumors Two patients with hypothalamic astrocytomas were treated with a two-drug chemotherapy regimen (actinomycin D and vincristine) at the time of initial diagnosis. This decision was based on a personal communication from Dr. Roger Packer (Childrens Hospital of Philadelphia) who had used this regimen and had observed objective computerized tomographic (CT) scan responses. Both patients continued to demonstrate clinical deterioration (loss of vision) and worsening CT scans. Treatment with local radiation therapy produced CT and clinical tumor regression and

neither had required further chemotherapy for up to 4 years. The seven other patients with hypothalamic astrocytomas were treated with a nitrosourea-based chemotherapy at recurrence. Although this is a small population upon which to base conclusions, the Kaplan-Meier representation (Fig. 34.1) indicates a median time to tumor progression (MTP) of 35 weeks. There is not enough information to assess the effectiveness of adjuvant chemotherapy. Thirteen patients with thalamic tumors were treated with a nitrosourea-based chemotherapy in an adjuvant fashion. The median age of this group was 43 years (range, 22 to 64 years). Their MTP was 32 weeks (Fig. 34.2). Seventeen patients with thalamic tumor were treated for tumor recurrence (nitrosourea-based chemother-

apy) and exhibited an MTP of 51 weeks. The median age of this group was 29 years (range, 1 to 60 years). No toxic deaths were observed in these series.

Case Report D.B. was 29 years old when she underwent a subtotal resection of a gemistocytic astrocytoma of the thalamus (July 6, 1981). She received 6500 rads of radiation, which was completed on September 25, 1981. The tumor progressed in December 1982 and she was started on poly-drug chemotherapy (BCNU, 5fluorouracil, hydroxyurea, 6-mercaptopurine) but failed to respond. In February 1983, therapy was changed to procarbazine (scans at start of procarbazine shown in Fig. 34.3). She returned at the completion of one cycle (April 1983) with clinical and CT scan evidence of improvement despite a stable dexamethasone dosage (Fig. 34.4). She has continued on procar-

bazine and remains clinically stable with continued but slight improvement on subsequent CT scans.

All responded for variable periods up to 6 months except one patient with metastatic germinoma who has been free of disease for greater than 6 years (18). However, three of these patients developed bleomycin pulmonary toxicity, two succumbing to this complication. We have treated two patients harboring pineal germ cell tumors with combination chemotherapy adjuvant to radiation therapy with good responses. The two children with recurrent pineoblastoma initially responded to combination chemotherapy but progressed within 1 year of the initial response. One patient with a pineal endodermal sinus tumor was treated with radiation therapy in combination with vincristine, adriamycin, and cytoxan and has remained free of disease for more than 10 years (21).

Chemotherapy—Pineal Tumors

Discussion

Six patients with a presumed or recurrent germinoma were treated with vincristine, bleomycin, and cisplatinum or adriamycin (12, 18, 24).

The majority of gliomas arising in the hypothalamic/optic nerve region occur in children with peak incidence at less than 5 years of age.

The majority of these tumors are slow-growing, although their biological aggressiveness is not easily predicted (10). As a rule, when transformation to a more aggressive tumor does occur (2), it is most commonly in older patients (11). Our patients demonstrated this same tendency at recurrence: the median age of our patients in the hypothalamic/chiasmal group at recurrence was 15 years (range, 3 to 60 years). The natural history of hypothalamic/chiasmal gliomas and the relative benefits of surgery and radiation therapy are well appreciated. Those lesions confined to one optic nerve without chiasmal involvement can be resected if proptosis and blindness develop. If the glioma involves the chiasm or hypothalamus, surgical treatment consists of biopsy to confirm the diagnosis. On rare occasions, a large exophytic portion compressing the third ventricle and producing hydrocephalus can be removed without morbidity. The biological behavior of these tumors is unpredictable, but frequently progressive visual loss or increased size on CT scans requires further treatment. Local radiotherapy can produce improvement in visual acuity, decrease tumor size (2, 8, 16), and improve survival (17, 22). The place of chemotherapy, instead of or before irradiation, is presumptive. The use of chemotherapy alone for lower grade astrocytomas has been only recently considered (9). Our initial attempt to treat two children with chemotherapy (actinomycin D and vincristine) based on information from Packer (19) failed. Both children showed signs of progressive visual loss and required radiation therapy. Visual loss returned within 3 months of the completion of radiation therapy. The use of chemotherapy adjuvant to radiation therapy is based on our prior experience with supratentorial tumors in adults (5, 7, 13-15) and children (20). As a result of reviewing our data for this chapter, we found that multiple agent chemotherapy based on nitrosoureas produced documented tumor response in five of our seven patients treated at recurrence. In this limited experience, tumor histology did not influence response rate. Although CT-documented regression was observed, median time to tumor progression was only 35 weeks. This is similar to the results obtained with nitrosourea-based chemotherapy for malignant cerebral astrocytomas (5, 7, 13-15). Like hypothalamic/chiasmal tumors, thalamic tumors are most frequently glial in origin (23). Surgical sampling is essential to establish a histological diagnosis and plan therapy (1). Approximately 50% recur within 5 years of initial treatment despite radiotherapy (6, 20). The use of

adjuvant chemotherapy for malignant thalamic astrocytomas has produced a median time to tumor progression (32 weeks) that is less than that observed for hemispheric astrocytic tumors. Median times to tumor progression in the latter groups vary by age at first treatment and histological classification: approximately 31 to 42 weeks for glioblastoma multiforme and 77 to 123 weeks for the less anaplastic astrocytomas (1315) in adults to 130 weeks in those under 18 years of age (20). Tumor recurrence was generally treated with chemotherapy. If a nitrosourea was used previously as an adjuvant to radiation therapy, other agents were used at recurrence because of the high likelihood of nitrosourea-resistant cells surviving in the tumor. In this case the use of noncross resistant chemotherapy, such as procarbazine, would seem more rational. Otherwise, nitrosourea-based therapy was our first choice. Recurrent tumors treated with nitrosoureabased poly drug chemotherapy resulted in improvement or stabilization in 7 of 17 patients. The MTP for the entire group was 51 weeks (Fig. 34.1), which was quite good in comparison to that observed in malignant hemispheric tumors treated at recurrence (5, 7, 13-15) although the median age of this group, 29 years, favors a somewhat better response. Treatment of recurrent or malignant germ cell pineal region tumors is fraught with failure. Despite their occasional sensitivity to chemotherapy (18), reported responses have been of short duration or drug toxicity produced unacceptable complications (i.e., pulmonary toxicity). Although the literature is replete with reports of attempts to treat these tumors with chemotherapy, the number of documented responses are few (3, 12, 18, 24) and the long term survivors can be counted on one hand. It is clear that no particular drug regimen is superior to another, although the polydrug regimens of vincristine, bleomycin, and cisplatinum and nitrosoureabased multidrug regimens have been partially effective. Single agent therapy has been of no benefit for these tumors. In summary, chemotherapy for hypothalamic/ optic nerve tumors, before or instead of irradiation, is of unproven benefit, although recent theoretical considerations (9) suggest that it should be pursued in some cases. In this regard, we have been evaluating the use of cell cycle-specific therapy consisting of 5-fluorouracil, hydroxyurea, and 6-thioguanine in these patients with non-contrast enhancing moderately anaplastic astrocytomas. If patients progress on this polydrug regimen, standard radiation therapy is ad-

ministered. The use of adjuvant chemotherapy for this group of patients awaits further confirmation. Since thalamic gliomas seem to behave in a similar manner to supratentorial gliomas, we believe that there is a good rationale to use chemotherapy adjuvant to radiation therapy for these patients. Unfortunately, our failure to use a consistent chemotherapy treatment for these patients leaves us in the awkward position of being unable to quantify precisely the benefit of chemotherapy. Nitrosourea-based chemotherapy for recurrent malignant hypothalamic, optic nerve, and thalamic tumors seems to have definite short term effectiveness and in select patients, particularly with thalamic gliomas, may produce prolonged tumor stability.

Acknowledgment These studies were supported in part by HEW Grants CA-13525. We thank Pamela Silver for data collection and Irene Asturias for manuscript preparation. We also thank the attending staff and previous Neun-Oncology Fellows of the Neuro-Oncology Service for their help in caring for these patients.

9. 10. 11. 12.

13.

14.

15.

References 1. Bernstein M, Hoffman HJ, Halliday WC, Hendrick EB, Humphreys RP: Thalamic tumors in children: Long-term follow-up and treatment guidelines. J Neurosurg 61:649-656, 1984. 2. Danoff BF, Kraner S, Thompson N: The radiotherapeutic management of optic nerve gliomas in children. Int J Radiat Oncol Biol Phys 6:4550, 1980. 3. deTribolet N, Barrelet L: Successful chemother apy of pinealoma. Lancet 2:1228-1229, 1977. 4. Doworetz DE, Blitzer PH, Wang CC, Linggood RM: Management of glioma of the optic nerve and/or chiasm. Cancer 45:1467-1471, 1980. 5. Edwards MB, Levin VA, Wilson CB: Brain tumor chemotherapy: An evaluation of agents in current use for Phase II-III studies. Cancer Treat Rep 64:1179-1205, 1980. 6. Eisenberg HM: Supratentorial Astrocytoma in Pediatric Neurosurgery: Surgery of the Devel oping Nervous System. Section of Pediatric Neu rosurgery of the American Association of Neuro logical Surgeons. New York, Grune & Stratton, 1982, pp 429-432. 7. Gutin PH, Levin VA: Surgery, radiation, and chemotherapy in the treatment of malignant brain tumors. In Thompson RA, Green JR (eds): Controversies in Neurology. New York, Raven Press, 1983, pp 67-86. 8. Harter DJ, Caderao JB, Leavens ME, Young SE:

16. 17. 18. 19. 20.

21. 22. 23. 24.

Radiotherapy in the management of primary gliomas involving the intra-cranial optic nerves and chiasm. Int J Radiat Oncol Biol Phys 4:681 686, 1978. Hoshino T: A commentary on the biology and growth kinetics of "low grade" and "high grade" gliomas. J Neurosurg 61:895-900, 1984. Hoyt WF, Baghdassarian SA: Optic glioma of childhood. Br J Ophthalmol 53:793-798, 1969. Hoyt WF, Mishel LG, Ussell S, Schatz NJ, Suckling RD: Malignant optic gliomas of adulthood. Brain 96:121-132, 1973. Kirshner JJ, Ginsberg SJ, Fitzpatrick AV, Comes RL: Treatment of a primary intracranial germ cell tumor with systemic chemotherapy. Med Pediatr Oncol 9:361-365, 1981. Levin VA, Wilson CB, Davis R, Wara W, Pischer TL, Irwin L: A Phase III comparison of BCNU, hydroxyurea, and radiation to BCNU and radiation therapy for treatment of primary malignant gliomas. J Neurosurg 51:526-532, 1979. Levin VA, Gutin PH, Wilson CB: Brain tumors. In Greenspan EM (ed): Clinical Interpretation and Practice of Cancer Chemotherapy. New York, Raven Press, 1982, pp 393-408. Levin VA, Wara WM, Davis RL, Vestneys P, Resser KJ, Yatsko K, Nutik S, Gutin PH, Wilson CB: Phase III comparison of BCNU and the combina tion of procarbazine, CCNU, and vincristine ad ministered after radiation therapy with hydrox yurea to patients with malignant gliomas. J Neurosurg 63:218-223, 1985. MacCarthy CS, Boyd AS, Childs DS: Tumors of the optic nerve and optic chiasm. J Neurosurg 33:439-444, 1970. Montgomery AB, Griffin T, Parker RB, Gerdes AJ: Optic nerve glioma: The role of radiation therapy. Cancer 40:2079-2080, 1977. Neuwelt EA (ed): Diagnosis and Treatment of Pineal Region Tumors. Baltimore, Williams & Wilkins, 1984. Packer R: Personal communication, 1984. Phuphanich S, Edwards MB, Levin VA, Vestnys PS, Wara WM, Davis RL, Wilson CB: Supratento rial malignant gliomas of childhood: Results of treatment with radiation and chemotherapy. J Neurosurg 60:495-499, 1984. Prioleau G, Wilson CB: Endodermal sinus tumor of the pineal region. Cancer 38:2489-2493,1976. Roberson C, Tiel K: Hypothalamic gliomas in chil dren. J Neurol Neurosurg Psychiatry 37:10471052, 1974. Russsell DS, Rubinstein LJ: Pathology of Tumors of the Nervous System. Baltimore, Williams & Wilkins, 1971. Siegal T, Pfeffer MR, Catane R, Sulkes A, Gomori MJ, Fuks: Successful chemotherapy of recurrent intracranial germinoma with spinal metastases. Neurology (NY) 33:631-633, 1983.

35 Therapeutic Modality Selection in Management of Germ Cell Tumors Kintomo Takakura, M.D., and Masao Matsutani, M.D.

Intracranial germ cell tumors are divided into three major histological types: germinoma and mature and immature teratoma. Immature teratomas include embryonal carcinoma, choriocarcinoma, endodermal sinus tumor (yolk sac tumor), and other teratocarcinomas. Mixed tumors of these histological types are often encountered. Because Choriocarcinomas produce human chorionic gonadotropin (hCG) and endodermal sinus tumors and some embryonal carcinomas produce alpha-fetoprotein (AFP), the immature teratomas containing components of those tumors can be properly diagnosed without histological verification when the levels of those tumor markers elevate in the serum or in the Cerebrospinal fluid of patients. Recent studies revealed that most immature teratomas produced either hCG, AFP, or both marker proteins. From therapeutic points of view, germinoma is quite sensitive to radiation and is properly treated by radiotherapy alone. Mature teratomas can be surgically removed completely in most cases. Immature teratomas are, however, resistant to radiation, grow rapidly, and metastasize frequently inside ventricles, to subarachnoid spaces, and to other organs. They require, therefore, multimodality treatment with surgical removal, radiotherapy, and chemotherapy. Because histological findings of intracranial germ cell tumors are heterogeneous in most cases, a biopsied specimen of the tumor hardly clarifies whole features of the tumor. There is always controversy about the therapeutic modalities of intracranial germ cell tu-

mors. Some neurosurgeons insist that histological verification is necessary before any treatment. On the other hand, others prefer so-called diagnostic radiotherapy before surgical removal whenever it is required, mainly for reasons of safer and better management of the patients. The volume of germinoma can be decreased by small doses of radiation (i.e., 20 Gy). Furthermore, many mixed-type teratomas contain germinoma components and respond remarkably to radiotherapy. Presurgical radiotherapy is, therefore, rational for reducing the size of the tumor before surgical removal and also for diagnostic aid. This short report describes our therapeutic approach for intracranial germ cell tumors.

Histological Types of Intracranial Germ Cell Tumors The histological types of surgically removed intracranial germ cell tumors are summarized in Table 35.1. Of a total of 84 tumors completely analyzed, 54 tumors showed a homogeneous histological pattern in each tumor tissue (pure type) and 30 other tumors contained mixed histological patterns. Pure type is, however, determined by the histological feature that more than 98% of tumor tissue shows the same pattern of histology; this does not mean real homogeneity of the tumor. Pure types of those tumors included: 24 germinomas and 22 mature and 8 immature teratomas. Mixed types of the tumor contained various components of germinoma and mature or immature teratoma in the same tumor tissue. The precise histological pattern of each tumor is

described in Table 35.1. Of 84 tumors, 47 cases (56%) contained germinomatous component and 25 cases (30%) contained some immature teratomatous components. Forty-five tumors (54%) were located in the pineal region, 29 tumors (35%) were in the suprasellar region, and 10 tumors (12%) were in other sites. Although the sex ratio of all germ cell tumors was 5 (male):l (female), it largely differs with age and the histological type of tumor. The most significant sex difference was noted in cases of immature teratoma. Male patients with immature teratomas were 10 times more frequent than females at ages between 10 to 20 years old, and no female patients at ages between 20 to 40 years old have ever been encountered (6).

Therapeutic Modality and Results The outcome of treatment for intracranial germ cell tumors depends on the histological type and location of the tumor. For tumors located in the pineal region, radiotherapy with a small dose of 20 Gy is locally given as the initial treatment (diagnostic radiation). Ventriculoperitoneal shunting for hydrocephalus is performed only when the increased intracranial pressure is suspected to be hazardous. When a patient is in good condition without papilledema, headache, or nausea, radiotherapy is generally given with glucocorticoid administration without VP shunting. If a tumor is germinoma, it shrinks rapidly and almost disappears on CT scan during this radiotherapy. In such a case, 30 to 40 Gy of radiation is further given to whole brain for preventing the recurrence and intraventricular dissemination. When a tumor does not respond fully to the initial radiotherapy, surgical removal of the tumor is indicated. The approach to the tumor is deter-

mined by the location and shape of the tumor in the third ventricle as appreciated on computerized tomographic (CT) scan and magnetic resonance imaging (MRI). In cases of mature teratoma, complete removal of the tumor is a goal of operation. No postsurgical irradiation is needed. In cases of immature teratoma, the total removal of the tumor is hardly possible because the tumor generally invades to the adjacent brain tissue. All immature teratomas are treated with postsurgical radiotherapy (30 to 40 Gy to whole brain and 25 Gy to spinal cord) concomitantly with chemotherapy using ACNU (100 mg/m2 i.v. on Day 1) and vincristine (1 mg/m2 i.v. on Days 2 and 3) for synchronization of the tumor cell cycle (5). The second course of chemotherapy is added during radiotherapy whenever possible. After finishing the whole course of radiotherapy, chemotherapy with cisplatin, vinblastine, and bleomycin (PVB therapy) is given for preventing the recurrence. Cisplatin (20 mg/m2 i.v.) is given 5 times on Days 1 to 5. Vinblastine (4 to 6 mg/ m2 i.v.) is given on Days 1 and 7. Bleomycin (10 to 15 mg/m2 i.v.) is given on Days 1, 8, and 15. These courses of chemotherapy (21-day course) are repeated 3 times as the initial maintenance chemotherapy. For further maintenance chemotherapy, the above one course of chemotherapy is repeated every 3 to 4 months. All patients are checked with regular examinations of CT scan or MRI and serum tumor marker assays for detection of recurrence. When the diagnosis of a pineal region lesion is suspected to be a tumor other than germ cell such as meningioma, epidermoid, or gliomas by imaging examinations before any treatment, surgical removal and the establishment of a histological diagnosis are generally the first steps of therapy.

In cases of suprasellar germ cell tumors, surgical removal and histological verification are the initial steps of treatment because operation in this location is safely performed and various tumors other than germ cell tumors such as craniopharyngioma or optic nerve gliomas are quite often encountered. The therapeutic modality after operation is similar to that for tumors located in the pineal region. Germ cell tumors in other locations require surgical verification before radiotherapy or chemotherapy because the diagnosis of germ cell tumor cannot be established without histological examination except in functioning tumors producing hCG or AFP. Germinoma is sensitive and responds well to radiotherapy. The survival rates of patients with germinoma (n = 79) in all sites at 5 and 10 years after the initial treatment were 74 and 72%, respectively (Fig. 35.1). The survival rates for suprasellar germinoma are much better than for pineal region germinoma mainly because of surgical morbidity. The survival rates for suprasellar germinoma (n = 22) at 5 and 10 years after operation were 91 and 84%, respectively, whereas those for pineal region germinoma (n = 49) were both 65% (Fig. 35.2). These survival rates included all cases treated during the past 20 years. The mortality and morbidity of operation for pineal region tumors have, however, much improved recently. The surgical mortality in the most recent 5 years was 0%. Therefore, the long term survival rates of patients with germinoma seem to be much better than the above data indicate. The outcome of treatment for mature teratoma depends on the size of the tumor and the clinical status of the patient at the time of admission. The survival rates at 5 and 10 years after operation were 64 and 48%, respectively. The survival rate of immature teratomas was the worst

compared to other tumors. The 5-year survival rate was 26% and only few patients have been able to survive more than 10 years. The response to multimodal treatment is related to the markerproducing function of the tumor. A serum hCG level over 1000 mlU/ml or an AFP level over 1000 ng/ml is a sign of poor prognosis. All long term survivors showed lower levels of markers in their serum. Our Phase II group study (2) revealed that PVB therapy was effective to suppress the growth of the immature teratoma and has extended the survival time. Thirty cases of immature teratoma were treated with PVB therapy and 47 historical controls were accumulated in this study. The median survival time of controls was 18.0 months after operation. Only 14.0% of the patients survived more than 5 years. Of 30 patients treated with PVB therapy, 20 primary cases received PVB therapy at the time of the initial treatment. The effectiveness rate for reducing tumor size appearing on CT scan was 71%. Nine

of 20 (45%) patients demonstrated complete response (disappearance of the tumor on CT scan). A reduction of serum tumor marker levels was noted in 90% of the patients. In the recurrent cases, the effectiveness rate for reduction of tumor size on CT scan was 50% and the tumor marker reduction rate was 69%. The survival rates at 2 years after operation were 46.5% for the patients who received radiotherapy alone and 67.7% for the patients who received PVB therapy (Fig. 35.3). The major side effects of PVB therapy were nausea and vomiting (52%) and myelosuppression (45%) and pulmonary fibrosis (1 case). Pneumonia (2 cases), temporary decrease of auditory function (2 cases), hyponatremia (1 case), and pigmentation of skin were noted.

Discussion In this chapter, we have described the therapeutic modality and results for intracranial germ cell tumors. Other tumors in the pineal region such as pineocytoma, pineoblastoma, various gliomas, and others were excluded. Because almost 80% of all pineal region tumors in Japan are germ cell tumors (6), the differential diagnosis can be quite satisfactorily established by CT scan, MRI, angiography, and serum tumor marker study. Although diagnostic radiotherapy for pineal region tumors is opposed by some neurosurgeons simply for lack of histological verification, it is beneficial for almost all patients from clinical points of view. Almost all tumors respond to the radiotherapy to some extent and the surgical management for the remaining tumor has never been hindered by this presurgical radiotherapy. Many patients with germinoma have been cured by radiotherapy alone without operation. Because the prognosis for immature teratoma is still very poor, the therapeutic modality for those tumors in the pineal region remains unsolved. PVB therapy has been reported to be effective for intracranial germ cell tumors by several investigators (1, 3, 4); the side effects on renal function, auditory disturbance, and pul-

monary fibrosis are limiting factors for continuing this chemotherapeutic schedule. Combined chemotherapy with cisplatin and VP-16 was reported to have the same potential therapeutic response as PVB therapy with less toxicity (7). Inasmuch as immature teratomas demonstrate a definite male predominance, endocrine factors controlling tumor should be investigated to establish the hormonal treatment of these tumors and better chemotherapies. Further studies are required to improve the therapeutic modality for these immature teratomas.

Acknowledgment This work was supported by Cancer Research Grant 59-22 from the Ministry of Health and Welfare and a Special Research Grant from the Japan Brain Foundation. Aid in preparing this manuscript by Miss Haruko Ichimura is also acknowledged.

References 1. Kirshner JJ, Ginsberg SJ, Fitzpatrick AV, et al: Treatment of a primary intracranial germ cell tumor with systemic chemotherapy. Med Pediatr Oncol 9:361-365, 1981. 2. Matsutani M: Cisplatin, vinblastine, bleomycin (PVB) combination chemotherapy in the treat ment of intracranial malignant germ cell tumor: Phase II study. Presented at the 8th International Congress of Neurological Surgery, Toronto, Can ada, 1985, Abstract 127, pp 81-82. 3. Neuwelt EA, Frankel EP, Smith RG: Suprasellar germinomas (ectopic pinealomas): Aspects of immunologic characterization and successful chem otherapeutic responses in recurrent disease. Neu rosurgery 7:352-358, 1985. 4. Siegel T, Pfeffer MR, Catane R, et al: Successful chemotherapy of recurrent intracranial germi noma with spinal metastases. Neurology [NY] 33:631-633, 1983. 5. Takakura K: Synchronized chemo-radiotherapy for brain tumors. Prog Nerv Res (Jpn) 26:105112, 1982. 6. Takakura K: Nonsurgical pineal tumor therapy— the Japanese experience. In Neuwelt EA (ed): Di agnosis and Treatment of Pineal Region Tu mors. Baltimore, Williams & Wilkins, 1984, pp 309-322. 7. Williams SD, Turner S, Loehrer PJ, et al: Testicular cancer: Results of reinduction therapy. Proc Am Soc Clin Oncol 2:137, 1983.