Neurosurgical Operative Atlas 8
Neurosurgical Operative Atlas 8
Neurosurgical Operative Atlas 8
Volume 8
AANS Publications Committee
Editors SETTI S. RENGACHARY ROBERT H. WILKINS
1999 For copies of this Cumulative Index contact: The American Association of Neurological Surgeons 22 South Washington St. Park Ridge, Illinois 60068-4287 email: aanspubs@valley.net voice: 847-692-9500 fax: 847-692-6770
Contents
Volume I
Optic Gliomas. Edgar M. Housepian / 1-13 Fibrous Dysplasia Involving the Craniofacial Skeleton. James T. Goodrich, Craig D. Hall / 14-22 Depressed Skull Fracture in Adults. Fred H. Geisler / 23-33 Cervical Hemilaminectomy for Excision of a Herniated Disc. Robert H. Wilkins, Sarah J. Gaskill / 34-38 Lateral Sphenoid Wing Meningioma. Joseph Ransohoff / 39-45 Selective Microsurgical Vestibular Nerve Section for Intractable Mnires Syndrome. Edward Tarlov / 46-53 Chiari Malformations and Syringohydromyelia in Children. W. Jerry Oakes / 54-60 Carotid Body Tumors. Fredric B. Meyer, Thoralf M. Sundt, Jr. / 61-69 Olfactory Groove Meningiomas. Joshua B. Bederson, Charles B. Wilson / 70-78 Cerebral Aneurysms at the Bifurcation of the Internal Carotid Artery. Eugene S. Flamm / 79-88 Treatment of Unilateral or Bilateral Coronal Synostosis. John A. Persing, John A. Jane / 89-98 Convexity Meningioma. Sarah J. Gaskill, Robert H. Wilkins / 99-105 Occipital Lobectomy. Milam E. Leavens / 106-112 Spinal Meningiomas. Michael N. Bucci, Julian T. Hoff / 113-116 Percutaneous Trigeminal Glycerol Rhizotomy. Ronald F. Young / 117-123 Lumbar Hemilaminectomy for Excision of a Herniated Disc. Patrick W. Hitchon, Vincent C. Traynelis / 124-129 Transoral Surgery for Craniovertebral Junction Anomalies. Arnold H. Menezes / 130-135 Anterolateral Cervical Approach to the Craniovertebral Junction. Dennis E. McDonnell / 136-153 Correction of Malposition of the Orbits. John A. Persing / 154-163 Removal of Cervical Ossified Posterior Longitudinal Ligament at Single and Multiple Levels. Ralph B. Cloward / 164-170 Technique of Ventriculostomy. Joseph H. Piatt, Jr., Kim J. Burchiel / 171-175 Cerebellar Medulloblastoma. Arthur E. Marlin, Sarah J. Gaskill / 176-183 Shunting of a Posttraumatic Syrinx. David J. Gower / 184-190 Direct Surgical Treatment of Vein of Galen Malformations. Harold J. Hoffman / 191-200 Spinal Nerve Schwannoma. Phyo Kim, Burton M. Onofrio / 201-206 Combined Craniofacial Resection for Anterior Skull Base Tumors. Ehud Arbit, Jatin Shah / 207-217 Diagnostic Open Brain and Meningeal Biopsy. Richard P. Anderson, Howard H. Kaufman, Sydney S. Schochet / 218-222 Ventriculoperitoneal Shunting. David C. McCullough / 223-230 Ventriculoatrial Shunting. Paul J. Camarata, Stephen J. Haines / 231-239 Excision of Acoustic Neuromas by the Middle Fossa Approach. Derald E. Brackmann / 240-248 Upper Thoracic Sympathectomy by a Posterior Midline Approach. Prem K. Pillay, Issam A. Awad, Donald F. Dohn / 249-255 Carotid Endarterectomy. Daniel L. Barrow, Christopher E. Clare / 256-266 Transsphenoidal Excision of Macroadenomas of the Pituitary Gland.
George T. Tindall, Eric J. Woodard, Daniel L. Barrow / 267-278 Computer-Directed Stereotactic Resection of Brain Tumors. Patrick J. Kelly / 279-293 Sagittal Synostosis. A. Leland Albright / 294-300 Glossopharyngeal Rhizotomy. Burton M. Onofrio / 301-304 Occipitocervical and High Cervical Stabilization. Volker K.H. Sonntag, Curtis A. Dickman / 305-315 Petroclival Meningiomas. Ossama Al-Mefty, Michael P. Schenk, Robert R. Smith / 316-326 Facial Reanimation without the Facial Nerve. Mark May, Steven M. Sobol / 327-336 Omental and Musculocutaneous Free Flaps for Coverage of Complicated Neurosurgical Wounds. Daniel L. Barrow, Foad Nahai / 337-348 Repair of Growing Skull Fracture. Tadanori Tomita / 349-354 Occipital Encephaloceles. William O. Bell / 355-362 Foramen Magnum Meningiomas and Schwannomas: Posterior Approach. Chad D. Abernathey, Burton M. Onofrio / 363-371 Penetrating Wounds of the Spine. Edward C. Benzel / 372-378 Percutaneous Radiofrequency Rhizolysis for Trigeminal Neuralgia. James Fick, John M. Tew, Jr. / 379-390 Extended Costotransversectomy. Eddy Garrido / 391-396 Surgical Resection of Posterior Fossa Epidermoid and Dermoid Cysts. Lee Kesterson / 397-406 Luque Rod Segmental Spinal Instrumentation. Edward C. Benzel, / 407-412 En Bloc Anterior Temporal Lobectomy for Temporolimbic Epilepsy. Michel F. Levesque / 413-422 Cingulotomy for Intractable Pain Using Stereotaxis Guided by Magnetic Resonance Imaging. Samuel J. Hassenbusch, Prem K. Pillay / 423-432 Cerebellar Astrocytomas. A. Leland Albright / 433-439 Extreme Lateral Lumbar Disc Herniation. Robert S. Hood / 440-444 Tentorial Meningiomas. Laligam N. Sekhar, Atul Goel / 445-455
Volume II
Surgical Repair of Trigonocephaly. Ken R. Winston, Michael J. Burke / 1-8 Dorsal Root Entry Zone (DREZ) Lesioning. Blaine S. Nashold, Jr., Amr O. Ei-Naggar / 9-24 Ophthalmic Segment Aneurysms. Arthur L. Day / 25-41 Chronic Subdural Hematoma. James E. Wilberger, Jr. / 42-48 Tailored Temporal Lobectomy Using Subdural Electrode Grids. Issam A. Awad, Joseph F. Hahn / 49-55 Gunshot Wounds of the Brain. Suzie C. Tindall, Ali Krisht / 56-59 Transtorcular Occlusion of Vein of Galen Malformations. J. Parker Mickle, Ronald G. Quisling, Keith Peters / 60-66 Detection of an Epileptic Focus and Cortical Mapping Using a Subdural Grid. Sumio Uematsu / 67-78 Anteromesial Temporal Lobectomy for Epilepsy. Issam A. Awad, Prem K. Pillay / 79-87 Anastomosis of the Facial Nerve After Resection of an Acoustic Neuroma. Charles M. Luetje / 88-90 An Extended Subfrontal Approach to the Skull Base. Chandranath Sen, Laligam N. Sekhar / 91-100
Pansynostosis: Surgical Management of Multiple Premature Suture Closure. James T. Goodrich, Craig D. Hall / 101-112 Distal Anterior Cerebral Artery Aneurysms. H. Hunt Batjer, Duke Samson / 113-126 Tethered Spinal Cord, Intramedullary Spinal Lipoma, and Lipomyelomeningocele. W. Jerry Oakes / 127-135 Interstitial Brachytherapy. Jeffrey D. McDonald, Philip H. Gutin / 136-144 Lateral Extracavitary Approach to the Thoracic and Lumbar Spine. Dennis J. Maiman, Sanford J. Larson / 145-153 An Extreme Lateral Transcondylar Approach to the Foramen Magnum and Cervical Spine. Chandranath Sen, Laligam N. Sekhar / 154-162 Retrolabyrinthine Presigmoid Approach for Sectioning of the Vestibular Nerve for Mnires Disease. Charles M. Luetje / 163-166 Stereotactic Surgical Ablation for Pain Relief. Ronald F. Young / 167-177 Anterior Screw Fixation of Odontoid Fractures. Ronald I. Apfelbaum / 178-188 Carpal Tunnel Syndrome. Setti S. Rengachary / 189-199 Transantral Ethmoidal Orbital Decompression For Graves Ophthalmopathy. Lawrence W. DeSanto / 200-206 Middle Fossa Approaches for Invasive Tumors Involving the Skull Base. Laligam N. Sekhar, Atul Goel, Chandranath Sen / 207-218 Transthoracic Excision of a Spinal Metastasis with Vertebral Body Reconstruction. Gregory J. Bennett / 219-228 Anterior Cervical Discectomy and Fusion-the Cloward Technique. Ralph B. Cloward / 229-240 Cubital Tunnel Syndrome. Setti S. Rengachary / 241-245 Caspar Plating of the Cervical Spine. H. Louis Harkey, Wolfhard Caspar, Yaghoub Tarassoli / 246-256 Surgical Management of Anterior Communicating Artery Aneurysms. Timothy C. Ryken, Chistopher M. Loftus / 257-265 Basilar Bifurcation Aneurysm: Pterional (Transsylvian) Approach. H. Hunt Batjer, Duke S. Samson / 266-281 Thalamotomy for Tremor. Roy A. E. Bakay, Jerrold L. Vitek, Mahlon R. Delong / 282-295 Endovascular Treatment of Carotid Cavernous Fistulas. Arvind Ahuja, Lee R. Guterman, Kimberly Livingston, Leo N. Hopkins / 296-304 Combined Transsylvian and Middle Fossa Approach to Interpeduncular Fossa Lesions. Chandranath Sen, Laligam N. Sekhar / 305-311 Aneurysms of the Ophthalmic Segment of the Internal Carotid Artery. Daniel L. Barrow / 312-322 Lumbar-Peritoneal Shunting. Setti S. Rengachary / 323-333 Surgery of the Cavernous Sinus. Harry van Loveren, Magdy El-Kalliny, Jeffrey Keller, John M. Tew, Jr. / 334-344 Encephaloceles of the Anterior Cranial Base. Alan R. Cohen / 345-353 Cotrel-Dubousset Instrumentation: Internal Fixation for Thoracolumbar Fractures and Tumors. Bruce E. van Dam / 354-358 Posterior-Lateral Lumbar Spinal Fusion. Edward S. Connolly / 359-366 Correction of Exorbitism. Constance M. Barone, Ravelo V. Argamaso, David F. Jimenez, James T. Goodrich / 367-372 Meralgia Paresthetica. Setti S. Rengachary / 373-379 Depressed Skull Fracture in Infants. Lyn C. Wright, Marion L. Walker / 380-383 Combined Presigmoid-Transtransversarium Intradural Approach to the Entire Clivus and
Anterior Craniospinal Region. Mario Ammirati, Melvin Cheatham / 384-395 Partial Median Corpectomy with Fibular Grafting for Cervical Spondylotic Myelopathy. Setti S. Rengachary / 396-409 Correction of Orbital Hypertelorism and Orbital Dystopia. Constance M. Barone, David F. Jimenez, Ravelo V. Argamaso, James T. Goodrich / 410-416 Percutaneous Radiofrequency Rhizotomy for the Treatment of Paraplegic Spasms. Sumio Uematsu / 417-427 Endocrine-Inactive Pituitary Adenomas. Charles B. Wilson / 428-437 Posterior Decompression and Fusion for Cervical Spondylotic Myelopathy. Paul Kurt Maurer, Charles Nussbaum / 438-447 Surgical Correction of Swan Neck Deformity. Peter M. Klara, Kevin T. Foley / 448-461
Volume III
Tuberculum Sellae Meningiomas. Ossama Al-Mefty / 1-11 Craniofacial Techniques Used in Resection of Anterior Skull Base Tumors. James T. Goodrich, Ravelo V. Argamaso / 12-20 Occipital Transtentorial Approach to Pineal Region Neoplasms. James I. Ausman, Balaji Sadasivan / 21-26 Meningioma of the Lateral Ventricle. Edward Tarlov / 27-30 Preauricular-Infratemporal Fossa Approach to Tumors that Involve the Lateral Cranial Base. Robert L. Grubb, Peter G. Smith / 31-37 Repair of the Myelomeningocele. David G. McLone / 38-44 Anterior Clinoidal Meningiomas. Franco DeMonte, Ossama Al-Mefty / 45-57 Dandy-Walker Malformation. Arthur E. Marlin, Sarah J. Gaskill / 58-65 Acoustic Neuromas: Surgical Anatomy of the Suboccipital Approach. Martin B. Camins, Jeffrey S. Oppenheim / 66-75 Exposure of the Skull Base via the Midface. James T. Goodrich, Sidney Eisig, George J. Cisneros, Allen B. Kantrowitz / 76-83 Exposure of the Skull Base by Transoral, Translabial, and Transmandibular Routes. James T. Goodrich, Sidney Eisig, Joseph G. Feghali, Allen B. Kantrowitz / 84-93 Surgical Management of Chiari I Malformations and Syringomyelia. Richard B. Morawetz / 94-102 Open-Door Maxillotomy Approach for Lesions of the Clivus. H. Louis Harkey, Vinod K. Anand, H. Alan Crockard, Michael P. Schenk / 103-112 Peripheral Nerve Repair. Allan J. Belzberg, James N. Campbell / 113-128 Surgical Management of Split Cord Malformations. Dachling Pang / 129-143 Tethered Cord Syndrome Secondary to Previous Repair of a Myelomeningocele. Timothy A. Strait / 144-150 Craniofacial Techniques for Managing Orbital Trauma. James T. Goodrich, Simeon A. Lauer, Ravelo V.Argamaso / 151-158 Transoral-Transclival Approach to Basilar Artery Aneurysms. R. A. de los Reyes, Paul W. Detwiler / 159-166 Frontal Lobectomy. Setti S. Rengachary / 167-175 Thoracic Outlet Syndrome: Supraclavicular First Rib Resection and Brachial Plexus Decompression. Susan E. Mackinnon, G. A. Patterson / 176-182 Transfacial Approaches to the Clivus and Upper Cervical Spine. Ivo P. Janecka / 183-192 Surgical Management of Prolactinomas. Andrew D. Parent / 193-202
Sectioning of the Filum Terminale. Frederick A. Boop, William M. Chadduck / 203-209 Repair of Diastematomyelia. Frederick A. Boop, William M. Chadduck / 210-214 Repair of a Lipomyelomeningocele. Frederick A. Boop, William M. Chadduck / 215-219 Untethering of the Spinal Cord After a Previous Myelomeningocele Repair. Frederick A. Boop, William M. Chadduck / 220-224 Secondary Carpal Tunnel Syndrome. Susan E. Mackinnon / 225-234 Spheno-Orbital Craniotomy for Meningioma. Joseph C. Maroon, John S. Kennerdell, Danko V. Vidovich / 235-243 Surgical Treatment of Anterior Sacral Meningocele. K. Stuart Lee / 244-251 Acrylic Cranioplasty. Setti S. Rengachary / 252-259 Preauricular Transzygomatic Infratemporal Craniotomy for Skull Base Tumors. Stephen L. Ondra, Michael G. Donovan / 260-269 Medial Sphenoid Ridge Meningiomas. Vallo Benjamin, Jules M. Nazzaro / 270-282 Surgical Treatment of Arteriovenous Malformations of the Cerebral Convexity. Wink S. Fisher III / 283-291 Lumbar Microdiscectomy. Peter M. Klara, Kevin T. Foley / 292-301 Microvascular Decompression of the Facial Nerve. Robert H. Wilkins / 302-311 Craniofacial Resection of Neoplasms of the Anterior Skull Base. Vincent C. Traynelis, Timothy M. McCulloch, Henry T. Hoffman / 312-323 Postlaminectomy Instability: Posterior Procedures. Seth M. Zeidman, Thomas B. Ducker / 324-336 Vertebral Artery and Posterior Inferior Cerebellar Artery Aneurysms: Surgical Management. Fernando G. Diaz, Richard D. Fessler / 337-343 Anterior Cervical Discectomy and Fusion: Smith-Robinson Technique. Philip R. Weinstein / 344-358 Management of Basilar and Posterior Cerebral Artery Aneurysms by Subtemporal Approaches. Robert M. Crowell, Christopher S. Ogilvy / 359-374 Subcutaneous Transposition of the Ulnar Nerve for Tardy Ulnar Palsy. Melvin L. Cheatham, Fredric L. Edelman, Martin Holland / 375-381 Image-Guided Neurosurgery: Frame-Based and Frameless Approaches. Lucia Zamorano, Lutz Nolte, Charlie Jiang, Majeed Kadi / 482-401 Anterior Stabilization of the Cervical Spine Using a Locking Plate System. Setti S. Rengachary / 402-413 Endoscopic Neurosurgery. Alan R. Cohen / 414-426 Surgical Management of Brain Abscess. Timothy C. Ryken, Christopher M. Loftus / 427-435 Submuscular Transposition of the Ulnar Nerve at the Elbow: Musculofascial Lengthening Technique. A. Lee Dellon / 436-443 Superficial Temporal Artery to Middle Cerebral Artery Bypass Grafting. Issam A. Awad / 444-456
Volume IV
Spinal Vascular Malformations. Edward H. Oldfield / 1-18 Posterior C1-2 Screw Fixation for Atlantoaxial Instability. Ronald I. Apfelbaum / 19-28 Supracerebellar Infratentorial Approaches to the Pineal Region. Michael L. Levy, Michael L. J. Apuzzo / 29-36 Third-Ventricle Exposure by the Interhemispheric Corridor. Peter Gruen, Michael L. J. Apuzzo / 37-42
Arteriovenous Malformations of the Basal Ganglia, Thalamus, and Adjacent Ventricles. Ghaus M. Malik, Fady T. Charbel / 43-58 Selective Denervation for Spasmodic Torticollis. Antonio A. F. DeSalles / 59-66 Unilateral Coronal Synostosis. James T. Goodrich, Ravelo Argamaso / 67-74 Neurosurgical Approaches to the Orbit. Part 1: Orbital Anatomy and Lateral Orbitotomy. Johnny B. Delashaw, Jr. / 75-84 Neurosurgical Aproaches to the Orbit. Part 2: Craniotomy for Surgical Exposure of the Orbit. Johnny B. Delashaw, Jr. / 85-94 Fourth Ventricular Ependymoma. J. Gordon McComb, John H. Schneider / 95-106 Sectioning of the Corpus Callosum for Epilepsy. Issam A. Awad / 107-116 Surgical Treatment of Intracranial Glomus Tumors. Vinod K. Anand, Michael P. Schenk, John P. Leonetti, Ossama Al-Mefty / 117-130 Technique of Temporal Lobectomy. Allen R. Wyler / 131-138 Treatment of Moyamoya Syndrome in Children with Pial Synangiosis. Richard G. Ellenbogen, R. Michael Scott / 139-146 Isthmic Spondylolysis and Spondylolisthesis: Treatment by Reduction, Interbody Fusion, and Lateral Stabilization. Timothy C. Wirt / 147-158 Translabyrinthine Removal of Acoustic Neuromas. John T. McElveen, Jr. / 159-164 Transsphenoidal Surgical Treatment of Cushings Disease. William F. Chandler / 165-172 Upper Thoracic Spinal Exposure Through a Lateral Parascapular Extrapleural Approach. Richard G. Fessler, Donald Dietze, David Peace / 173-182 Selective Dorsal Rhizotomy for the Spasticity of Cerebral Palsy. T. S. Park / 183-190 Surgical Treatment of the Subclavian Steal Syndrome. George E. Pierce / 191-198 Surgery for Tumors Affecting the Cavernous Sinus. Franco DeMonte, Vinod K. Anand, Ossama Al-Mefty / 199-208 Lambdoidal Synostosis. David F. Jimenez, Constance M. Barone, Ravelo V. Argamaso, James T. Goodrich / 209-214 Gamma Knife Radiosurgery of Intracranial Lesions. Robert J. Coffey / 215-224 Submuscular Transposition of the Ulnar Nerve at the Elbow. Susan E. Mackinnon / 225-234 Ulnar Nerve Entrapment at the Wrist. V. Leroy Young, Jill M. Young / 235-249
Volume V
Endoscopic Pituitary Surgery. Hae-Dong Jho, Ricardo L. Carrau, Yong Ko / 1-12 Torcular and Peritorcular Meningiomas. Griffith R. Harsh IV / 13-22 Surgical Resection of Lower Clivus-Anterior Foramen Magnum Meningioma. Vallo Benjamin, Ramesh P. Babu / 23-32 Basilar Bifurcation Aneurysms: Transsylvian Transclinoidal Transcavernous Approach. Murali Guthikonda, Fernando G. Diaz / 33-42 Surgical Management of Posterior Plagiocephaly. Richard G. Ellenbogen, Michael H. Mayer / 43-56 Acute Subdural Hematoma. Fred H. Geisler / 57-64 Intracranial Pressure Monitoring. Andrew D. Firlik, Donald W. Marion / 65-74 Temporal Lobectomy Under General Anesthesia. Diana L. Abson Kraemer, Dennis D. Spencer / 75-84 Far-Lateral Disc Herniation Treated by Microscopic Fragment Excision. Bruce V. Darden II, J. Robinson Hicks / 85-90 Stabilization of the Cervical Spine (C3-7) with Articular Mass (Lateral Mass) Plate and
Screws. T. Glenn Pait, Luis A. B. Borba / 91-100 Stabilization of the Cervical Spine with the Orion Anterior Cervical Plate System. Gary L. Lowery / 101-108 Texas Scottish Rite Hospital System for Internal Stabilization of Thoracolumbar Fractures. Bradford M. Mullin, Gary L. Rea / 109-120 Application of Frameless Stereotaxy in the Management of Intracranial Lesions. Dennis A. Turner, Paul B. Johnson / 121-128 A Modified Transfacial Approach to the Clivus. Brooke Swearingen, Michael P. Joseph, Matthew Cheney, Robert G. Ojemann / 129-134 Management of the Vertebral Artery During Excision of Extradural Tumors of the Cervical Spine. Chandranath Sen, Mark Eisenberg / 135-142 Posteroventral Pallidotomy for Patients with Parkinsons Disease. Robert P. Iacono, Shokei Yamada / 143-154 Functional Hemispherectomy. Joseph R. Smith, Mark R. Lee / 155-164 Microsurgical Decompresson of the Root Entry Zone for Trigeminal Neuralgia. Chandranath Sen / 165-170 The Anterior Cervical Approach to the Cervicothoracic Junction. Julian K. Wu / 171-176 Management of Extradural Non-Neoplastic Lesions of the Craniovertebral Junction via the Transcondylar Approach. Luis A. B. Borba, Ossama Al-Mefty, T. Glenn Pait, Ronald Tribell / 177-184 Far Lateral Lumbar Disc Herniation. Nancy E. Epstein, Joseph A. Epstein / 185-198 Repair and Reconstruction of Scalp and Calvarial Defects. Warren Schubert, Jeffrey Aldridge / 199-218 Sagittal Synostosis. Larry A. Sargent, Timothy A. Strait / 219-226 Microsurgical Lumbar Decompression Using Progressive Local Anesthesia. Stephen D. Kuslich / 227-232 Banked Fibula, the Locking Anterior Cervical Plate, and Allogeneic Bone Matrix in Anterior Cervical Fusions Following Cervical Discectomy. Scott Shapiro / 233-240 Endoscopic Third Ventriculostomy for Obstructive Hydrocephalus. Jonathan J. Baskin, Kim H. Manwaring / 241-246
Volume VI
Treatment of Carotid-Cavernous Sinus Fistulas Using a Superior Ophthalmic Vein Approach. Neil R. Miller, Lee H. Monsein, Rafael J. Tamargo / 1-4 The Separation of Craniopagus Twins. Harold J. Hoffman, James T. Rutka / 5-12 Posteroventral Pallidotomy for Parkinsons Disease Patients. Kim J. Burchiel, Jamal M. Taha, Jacques Favre / 13-26 Microelectrode-Guided Pallidotomy. Andres M. Lozano, William D. Hutchison / 27-34 Anterolateral Transforaminal Approach for a Large Dumbbell-Shaped Cervical Neurinoma. Isao Yamamoto / 35-42 Bridge Bypass Coaptation for Cervical Nerve Root Avulsion. Shokei Yamada, Russell R. Lonser, Robert P. Iacono / 43-50 Sinus Skeletonization Technique: A Treatment for Dural Arteriovenous Malformations at the Tentorial Apex. Evandro De Oliveira, Helder Tedeschi / 51-56 Microsurgical Carotid Endarterectomy. Julian E. Bailes, Patrick P. Flannagan / 57-64 Endoscopic Approaches to the Ventricular System. David F. Jimenez / 65-74 Surgical Management of Cranial Dural Arteriovenous Fistulas. Lokesh S. Tantuwaya, Julian E. Bailes / 75-84
Intraventricular Endoscopy: Diagnostic Ventriculoscopy, Tissue Biopsy, Cyst Fenestration, and Shunting. Jonathan J. Baskin, Kim H. Manwaring / 85-98 Endoscopic Carpal Tunnel Release Through a Monoportal Approach. Jay Menon / 99-108 Endoscopic Excision of Colloid Cysts. Jonathan J. Baskin, Kim H. Manwaring / 109-114 Surgical Anatomy of the Temporal Lobe. Steven N. Roper / 115-124 Multiple Subpial Transection. Walter W. Whisler / 125-130 Stereotactic Depth Electrode Implantation in the Evaluation of Candidates for Ablative Epilepsy Surgery. Joseph R. Smith, Mark R. Lee / 131-146 Trans-Sulcal Approach to Mesiotemporal Lesions. Isabelle M. Germano / 147-156 Anterior Cervical Spine Stabilization with the Codman Locking Plate System. R. John Hurlbert, Volker K. H. Sonntag / 157-166 Posterior Cervical Fusion with Tension Band Wiring. Thomas J. Lovely / 167-172 Primary Anterior Treatment of Thoracolumbar Burst Fractures. David W. Polly, Jr., Richard G. Ellenbogen / 173-182 Technique for Reduction of Spondylolisthesis Using Custom Texas Scottish Rite Hospital Forceps. Gary L. Lowery, David A. Fernandez, Atul L. Bhat, A. Eugene Pennisi / 183-192 Surgical Management of Infected Ventriculoperitoneal Shunt. Timothy M. George, Sohaib A. Kureshi / 193-200 Combined Fronto-Orbital and Occipital Advancement for Total Calvarial Reconstruction. Ian F. Pollack, H. Wolfgang Losken / 201-212 Repair of Meningoceles. Timothy M. George, Eric M. Gabriel / 213-220 Installation of a Dorsal Column Stimulator for Pain Relief. John P. Gorecki / 221-236 Implantation of Drug Infusion Pumps. John P. Gorecki / 237-250 Stereotactic Microsurgical Craniotomy for the Treatment of Third Ventricular Colloid Cysts. Kyle L. Cabbell, Donald A. Ross / 251-256 Hemispherectomy. Benjamin S. Carson, Aaron L. Zuckerberg / 257-264
Volume VII
Posterior Lumbar Interbody Fusion Augmented With the Ray Threaded Fusion Cage. Peter Klara, Berkley Rish, Charles D. Ray / 1-10 Total Sacrectomy. Ziya L. Gokaslan, Marvin M. Romsdahl, Stephen S. Kroll, Theresa A. Gillis, David W. Wildrick, Milam E. Leavens / 11-20 Treatment of Fractures at the Thoracolumbar Junction with Kaneda Anterior Spinal Instrumentation System. Seth M. Zeidman, Randy F. Davis / 21-28 Cannulated Screws for Odontoid and Atlantoaxial Transarticular Screw Fixation. Curtis A. Dickman, R. John Hurlbert / 29-42 Anterior Microforaminotomy for Cervical Radiculopathy: Disc Preservation Technique. Hae-Dong Jho / 43-52 Pedical Subtraction and Lumbar Extension Osteotomy for Iatrogenic Flatback. Gary L. Lowery, Atul L. Bhat, A. Eugene Pennisi / 53-58 The Surgical Treatment of Dolichoectactic and Fusiform Aneurysms. Michael T. Lawton, John A. Anson, Robert F. Spetzler / 59-68 Petrosal Approach for Resection of Petroclival Meningiomas. William T. Couldwell / 69-82 Surgical Resection of Esthesioneuroblastoma. Scott L. Henson, John A. Jane, Sr. / 83-92 Stereotactic Radiosurgery of the Trigeminal Nerve Root for Treatment of Trigeminal Neuralgia. Ronald F. Young / 93-98 Techniques of Peripheral Neurectomy for Control of Trigeminal Neuralgia. Raj Murali / 99-106
Percutaneous Balloon Compression for the Treatment of Trigeminal Neuralgia. Jeffrey A. Brown, Jan J. Gouda / 107-116 Microvascular Decompression for Hemifacial Spasm. Thomas J. Lovely / 117-124 Thalamic Deep Brain Stimulation for the Control of Tremor. Andres Lozano / 125-134 Magnetic Resonance Image-Guided Stereotactic Cingulotomy for Intractable Psychiatric Disease. Osama S. Abdelaziz, G. Rees Cosgrove / 135-140 Magnetic Resonance Image-Guided Pallidotomy. Antonio A.F. De Salles, Marwan Hariz / 141-148 Endoscopic Carpal Tunnel Release via a Biportal Approach. David F. Jimenez / 149-156 Thoracic Sympathectomy. J. Patrick Johnson, Samuel S. Ahn / 157-162 Blood Flow-Monitored Transthoracic Endoscopic Sympathectomy. Ricardo Segal, Peter M. Ferson, Edwin Nemoto, Sidney K. Wolfson Jr. / 163-172 Surgical Management of Craniopharyngiomas. Harold J. Hoffman / 173-182 Surgical Resection of Craniopharyngiomas. Ali F. Krisht, Ugur Tre / 183-190 Optic Nerve Sheath Fenestration in the Management of Pseudotumor Cerebri. Eric L. Berman, Jonathan D. Wirschafter / 191-200 Surgical Correction of Unilateral and Bilateral Coronal Synostosis. Ann Marie Flannery, Jack C. Yu / 201-210 Tethered Cord Syndrome: Management of Myelomeningocele, Diastematomyelia, and Hypertrophied Filum Terminale. Robert F. Keating, James Tait Goodrich / 211-218 Tethered Cord Syndrome: Management of Lipomyelomeningoceles. James Tait Goodrich / 219-226 Excision of Colloid Cyst via the Transcallosal Approach. Deepak Awasthi, John J. Kruse / 227-234 Laparoscopy Assisted Lumboperitoneal Shunt Placement in the Management of Pseudotumor Cerebri. Florence C. Barnett, Dennis E. McConnell / 235-240 The Transparaspinal Approach to Dumbbell-Shaped Spinal Tumors. Stephen T. Onesti, Ely Ashkenazi, W. Jost Michelsen / 241-248 Posterior Occipito-axial Fusion for Atlantoaxial Dislocation Associated with Occipitalized Atlas. Vijendra K. Jain, Sanjay Behari / 249-256 Evaluation and Management of Severe Facial Nerve Injury Resulting From Temporal Bone Trauma. Aijaz Alvi / 257-260
Volume VIII
Surgical Management of Paraclinoid Carotid Aneurysms. Murali Guthikonda, Fernando G. Diaz / 1-12 Surgical Management of Middle Cerebral Artery Aneurysms. Philip E. Stieg, Robert M. Friedlander / 13-22 Surgical Removal of Tentorial and Posterior Fossa Dural Arteriovenous Malformations. Adam I. Lewis, John M. Tew Jr. / 23-34 Surgical Resection of the Arteriovenous Malformations of the Posterior Fossa. Thomas Kopitnik, Duke Samson, Michael Horowitz / 35-46 Surgical Treatment of Arteriovenous Malformations of the Ventricular Trigone. Daniel L. Barrow, Roger H. Frankel / 47-56 Dural Arteriovenous Malformations of the Transverse and Sigmoid Sinuses. Todd A. Kuether, Gary M. Nesbit, Stanley L. Barnwell / 57-68
Operative Management of Anterior Fossa, Superior Sagittal Sinus, and Convexity Dural Arteriovenous Malformations. Aman B. Patel, Wesley A. King, Neil A. Martin / 69-78 Use of the Operating Arm System in Skull Base Surgery. Jeffrey J. Larson, Ronald E. Warwick, John M. Tew Jr. / 79-86 The Orbitocranial Zygomatic Approach to Aneurysms of the Upper Basilar Trunk. T. C. Origitano / 87-94 Extradural Approaches for Resection of Trigeminal Neurinomas. J. Diaz Day / 95-106 Surgical Management of Trigeminal Schwannomas. Madjid Samii, Ramesh Pitti Babu, Marcos Tatagiba / 107-120 Surgical Management of Cholesterol Granulomas of the Petrous Apex. Mark B. Eisenberg, Ossama Al-Mefty / 121-126 Surgical Management of Angiographically Occult Vascular Malformations of the Brainstem, Thalamus, and Basal Ganglia. Gary K. Steinberg, Steven D. Chang / 127-134 Management of Jugular Foramen Tumors. Jeffrey Bruce, Ian Storper / 135-142 Surgical Management of Esthesioblastomas. Ramesh Pitti Babu, Mark S. Persky / 143-152 Surgical Treatment of Brainstem Gliomas. Mark R. Lee, Michael Cowan / 153-160 Brainstem Gliomas. Harold J. Hoffman / 161-170 The Contralateral Transcallosal Approach to Lesions In or Adjacent to the Lateral Ventricle. Michael T. Lawton, Robert F. Spetzler / 171-178 Posterior Fossa Decompression Without Dural Opening for the Treatment of Chiari I Malformation. Jonathan Sherman, Jeffrey J. Larson, Kerry R. Crone / 179-184 Computed Tomography-Assisted Preformed Prosthesis for Repair of Cranial Defects. Manuel Dujovny, Celso Agner, Fady T. Charbel, Lewis L. Sadler, Raymond Evenhouse, Dierdre McConathy / 185-194 Chronic Subthalamic Nucleus Stimulation for Parkinsons Disease. Ali R. Rezai, William Hutchison, Andres M. Lozano / 195-208 Arthroscopic Microlumbar Discectomy. Kenneth F. Casey, Parviz Kambin, Marc Chang / 209-216 Excision of Herniated Thoracic Disc Via the Transthoracic Approach. Mary Louise Hlavin, Russell W. Hardy / 217-224 Surgical Management of Advanced Degenerative Disease of the Lumbar Spine with Multiplanar Deformity. Michael F. OBrien, Gary L. Lowery, A. Eugene Pennisi / 225-234 The Retropleural Approach to the Thoracic and Thoracolumbar Spine. Theodore H. Schwartz, Paul C. McCormick / 235-242 Surgical Treatment of Lateral Lumbar Herniated Discs. Giuseppe Lanzino, Christopher I. Shaffrey, John A. Jane, Sr. / 243-252 Trap Door Exposure of the Cervicothoracic Junction. Ziya L. Gokaslan, Garrett L. Walsh / 253-260 Peripheral Nerve Suture Techniques. Rajiv Midha, Margot Mackay / 261-269
A
Ablative epilepsy surgery, 6:131-146 Acoustic neuromas, 4:159-164 Acrylic cranioplasty, 5:214-215 Acute subdural hematoma, 5:57-63 Allogeneic bone matrix, 5:233-239 Aneurysms basilar bifurcation, 5:33-42 broad-based siphon, 8:3-4; 8:10-11 carotid cave, 8:3-4 carotid ophthalmic, 8:2-3; 8:10-11 carotid-superior hypophyseal, 8:3-4 dolichoectatic, 7:59-67 fusiform, 7:59-67 middle cerebral artery, 8:13-22 paraclinoid carotid artery, 8:1-12 superior hypophyseal, 8:1-2; 8:10-11 upper basilar trunk, 8:87-94 ventral paraclinoid, 8:3-4; 8:10-11 Angiographically occult vascular malformations, 8:127-133 Angioma, cavernous, 4:13-18 Anterior cervical spine discectomy, 5:233-239 implant systems, 5:101-108 stabilization, 6:157-166 Anterior foramen magnum meningioma, 5:23-32 Anterior fossa dural AVMs, 8:69-78 Anterior microforaminotomy, 7:43-52 Apert syndrome, 7:201 Arteriovenous fistulas (AVFs) cranial dural, 6:75-84 dural, 4:3-7; 6:51-56 intradural, 4:11-16 perimedullary, 4:11-13 Arteriovenous malformations (AVMs) anterior fossa, 8:69-78 basal ganglia, 4:43-58 cerebellar hemisphere, 8:36; 8:40-42 cerebellar tonsil, 8:36; 8:43-44 cerebellar vermis, 8:35-36, 8:38-40 deep parenchymal, 8:36 dural, 8:23-34; 8:69-79 glomus, 4:9-10 juvenile, 4:8-10 posterior fossa, 8:23-46 spinal cord, 4:7-10
superior sagittal sinus dural, 8:69-78 tentorial dural, 8:23-34 thalamic, 4:43-58 upper basilar trunk, 8:87 ventral paraclinoid, 8:3-4; 8:10-11 ventricular trigone, 8:47-56 Astrocytomas brainstem, 8:162; 8:164; 8:165-169 craniocervical, 8:169-170 Atlantoaxial dislocation with occipitalized atlas, 7:249-256 Atlantoaxial instability, C1-2 screw fixation, 4:19-28 Atlantoaxial transarticular screw fixation, 7:29-41
A B C D E F G H I J K L M N O P Q R S T U V W X Y Z
B
Banked fibula, 5:233-239 Basal ganglia AVMs, 4:43-58 Basilar bifurcation aneurysms, 5:33-42 Bilateral coronal synostosis, 7:201-210 Birth defects, 5:219-225 Bone graft harvesting fractures, 7:24-25 atlantoaxial dislocation, 7:251-254 calvarial defects, 5:199-217; 6:201-211 posterior lumbar interbody fusion, 7:6-10 Bone-wiring procedures, 5:91-100 Brainstem AVMs, 8:36; 8:43; 8:45-46 craniocervical astrocytomas, 8:169-170 dorsally exophytic gliomas, 8:161-163 diffuse intrinsic astrocytomas, 8:162; 8:164 focal intrinsic astrocytomas, 8:165-169 gliomas, 8:153-159 Bridge bypass coaptation, 6: 43-50 Broad-based siphon aneurysms, 8:3-4; 8:10-11 Burst fractures, 5:110; 6:173-182
C
Callostomy, 7:227-233 Calvarial defects, 5:199-217 Calvarial reconstruction, 6:201-211 Cannulated screws, 7:29-41
Carotid cave aneurysms, 8:3-4 Carotid endarterectomy, 6:57-64 Carotid-cavernous sinus fistulas, 6:1-4 Carotid ophthalmic aneurysms, 8:2-3; 8:10-11 Carotid-superior hypophyseal aneurysms, 8:3-4 Carpal tunnel syndrome, 6:99-108; 7:149-156 Cavernous angiomas, 4:13-18 Cavernous sinus tumors, 4:199-207 Cerebellar hemisphere AVMs, 8:36; 8:40-42 Cerebellar tonsil AVMs, 8:36; 8:43-44 Cerebellar vermis AVMs, 8:35-36; 8:38-40 Cerebral palsy, 4:183-190 Cervical fusion, 5:233-239; 6:167-171 Cervical nerve root avulsion, 6:43-50 Cervical neurinoma, 6:35-41 Cervical radiculopathy, 7:43-52 Cervical spine C1-2 screw fixation, 4:19-28 degenerative disc disease, 7:43-52 discectomy, 5:233-239 extradural tumors, 5:135-141 implant systems, 5:101-108 stabilization, 6:157-166 stabilization (articular mass), 5:91-100 stabilization (Orion system), 5:101-108 Chiari I malformation, 8:179-183 Children cerebral palsy, 4:183-190 moyamoya syndrome, 4:139-146 Cholesterol granulomas of petrous apex, 8:121-125 Chondrosarcoma, 5:129 Cingulotomy for psychiatric disease, 7:135-140 Clivus, 5:129-133 Codman locking plate system, 6:157-166 Colloid cyst, 6:109-114; 6:251-256; 7:227-233 Complex spinal schwannomas, 7:241-242 Convexity dural AVMs, 8:69-78 Coronal synostosis, 4:67-73; 6:201-211; 7:201-210 Corpus callosum sectioning, 4:38-39; 4:107-116 Cranial defects, 8:185-194 Cranial dural arteriovenous fistulas,
6:75-84 Craniocervical brainstem astrocytomas, 8:169-170 Craniopagus twins, 6:5-11 Craniopharyngiomas, 7:173-181; 7:183-190 Craniovertebral junction lesions, 5:177-184 Crouzons syndrome, 7:201 Cubital tunnel syndrome, 4:235-249 Cushings disease, 4:165-172 Cyst colloid, 6:109-114; 6:251-256; 7:227-233 fenestration, 6:85-98
D
Decompressive corpectomy, 5:101 Deep brain stimulation control of tremor, 7:125-134 subthalamic nucleus, 8:169-197; 8:200-201; 8:205-206 Deep parenchymal AVMs, 8:36 Degenerative disc disease, 7:43-52; 8:225-233 Denervation for spasmodic torticollis, 4:59-65 Diastematomyelia, 7:219-226 Direct end-to-end repair of peripheral nerves, 8:263-269 Disc herniation far lateral, 5:85-89 far lateral lumbar, 5:185-197 lateral, 8:243-251 thoracic, 8:217-224 Disc preservation, 7:43-52 Discectomy, cervical, 5:233-239 Dolichoectatic aneurysms, 7:59-67 Dorsal column stimulation, 6:221-235 Dorsal lipomyelomeningocele, 7:221-225 Dorsal rhizotomy, 4:183-190 Drug infusion pumps, 6:237-250 Dumbbell-shaped cervical neurinoma, 6:35-41 Dumbbell-shaped spinal tumor, 7:241-248 Dural AVFs, 4:3-7; 6:51-56; 6:75-84 Dural AVMs anterior fossa, 8:69-78 convexity, 8:69-78 inferior petrosal sinus, 8:29-32
petrous apex, 8:24-27 posterior fossa, 8:23-46 superior sagittal sinus, 8:69-78 tentorial, 8:23-34
E
Elbow, ulnar nerve transposition, 4:225-233 Electrode implantation, 6:131-146 Endarterectomy, carotid, 6:57-64 Endoscopy approaches to the ventricular system, 6:65-74 carpal tunnel release, 6:99-107; 7:149-156 colloid cysts, 6:109-114 fenestration of the third ventriculostomy, 5:241-246 intraventricular, 6:85-98 pituitary surgery, 5:1-12 thoracoscopic sympathectomy, 7:157-162; 7:163-171 Ependymoma, fourth ventricular, 4:95-106 Epilepsy ablative surgery, 6:131-146 corpus callosum sectioning, 4:107-116 medial temporal onset, 5:75-83 Esthesioblastomas, 7:83-91; 8:143-151 Exophytic gliomas, 8:161-163 Extradural non-neoplastic lesions, 5:177-184 Extradural cervical spine tumors, 5:135-141
Focal intrinsic brainstem astrocytomas, 8:165-169 Foramen magnum, 5:23-32 Fourth ventricular ependymoma, 4:95-106 Frameless stereotaxy, intracranial lesions, 5:121-128 Full facetectomy, 5:190-191 Full thickness calvarial bone graft, 5:215-217 Functional hemispherectomy, 5:155-164 Fusiform aneurysms, 7:59-67 Fusion tension band wiring, 6:167-171
G
Galen, vein of, 8:32-34 Gamma Knife radiosurgery, intracranial lesions, 4:215-224 Gliomas, brainstem, 8:153-159 Glomus AVM, 4:9-10 Glomus tumors, intracranial, 4:117-130 Grafts, 5:233-239; 8:267-269 bone, 5:199-217; 7:6-10; 7:24-25; 7:251-254 Granulomas, petrous apex cholesterol, 8:121-125
H
Hematoma, acute subdural, 5:57-63 Hemicorticectomy, 5:155 Hemifacial spasm, 7:117-124 Hemispherectomy, 5:155-164; 6:257-264 Herniation far lateral disc, 5:85-89 far lateral lumbar disc, 5:185-197 thoracic disc, 8:217-224 Horners syndrome and anterior microforaminotomy, 7:51 complication of thoracoscopic sympthectomy, 7:162 Hydrocephalus, 5:241-246; 6:65; 6:76; 6:98; 6:261-264 Hyperhidrosis, 7:158 Hypertrophied filum terminale, 7:219-226
F
Facial nerve injury, 7:257-260 Facial pain, 5:227-232 Far lateral disc herniation, 5:85-89; 5:185-197 Fascicular peripheral nerves repair, 8:267 Fields of Forel, 8:200 Fistulas carotid-cavernous sinus, 6:1-4 dural arteriovenous, 4:3-7; 6:51-56; 6:7584 intradural, 4:11-16 Flat-back syndrome, 7:53-58
I
Idiopathic intracranial hypertension,
7:191-200 Implantation of drug infusion pumps, 6:237-250 Infection of ventriculoperitoneal shunt, 6:193-200 Inferior dental neurectomy, 7:103-104 Inferior petrosal sinus dural AVMs, 8:29-32 Infraorbital neurectomy, 7:101-103 Interbody fusion, 4:147-157 Intercostal neuralgia, 7:162 Interhemispheric corridor and thirdventricle exposure, 4:37-42 Internal stabilization, 5:109-119; 5:233-239 Intervertebral disc damage, 7:51-52 Intracranial glomus tumors, 4:117-130 Intracranial hypertension, 7:191-200 Intracranial lesions, 4:75-83; 4:85-93; 4:215-224; 5:121-128 Intracranial pressure monitoring, 5:65-74 Intradural arteriovenous fistulas, 4:11-16 Intraventricular endoscopy, 6:85-98 Intraventricular shunt, 6:85-98 Isthmic spondylolysis/spondylolisthesis, 4:147-157
Locking plate system, 6:157-166 Low back pain, 5:227-232 Lower clivus-anterior foramen magnum meningioma, 5:23-32 Lumbar decompression, 5:227-232 Lumbar disc herniation, far-lateral, 5:185197 Lumbar extension osteotomy for flat-back syndrome, 7:53-58 Lumbar spine arthroscopic microlumbar, 8:209-216 degenerative disease, 8:225-233 far lateral disc herniation, 5:85-89 far lateral lumbar disc herniation, 5:185-197 foraminal stenosis, 8:227 thoracolumbar fractures, 5:109-119 Lumboperitoneal shunt placement for pseudotumor cerebri, 7:235-240
M
Meningioma anterior foramen magnum, 5:23-32 lower clivus, 5:23-32 petroclival, 7:69-81 torcular/peritorcular, 5:13-21 Meningoceles, 6:213-219 Mental neurectomy, 7:103-106 Mesiotemporal lesions, 6:147-156 Microelectrode-guided pallidotomy, 6:27-33 Microforaminotomy, anterior, 7:43-52 Microsurgery carotid endarterectomy, 6:57-64 craniotomy for colloid cysts, 6:251-256 lumbar decompression, 5:227-232 root entry zone decompression, 5:165-170 Microvascular decompression for hemifacial spasm, 7:117-124 Middle cerebral artery aneurysms, 8:13-22 Moyamoya syndrome, 4:139-146 MRI-guided pallidotomy, 7:141-148 MRI-guided stereotactic cingulotomy, 7:135-140 Multiple subpial transection, 6:125-129 Myelomeningocele, 7:219-226
J
Jugular foramen tumors, 8:135-142 Juvenile AVMs, 4:8-10
K
Kambin instrumentation for microlumbar discectomy, 8:211 Kaneda anterior spinal instrumentation system, 7:21-27
L
Labb, vein of, 8:58-60 Lambdoidal synostosis, 4:44-45; 4:209-214 Lateral disc herniation, 8:243-251 Lateral mass plate and screws, 5:91-100 Lateral orbitotomy, 4:81-83 Lateral ventricles, 5:67-69 Lipomyelomeningoceles, 7:219-226 Lobectomy, temporal, 4:131-137; 5:75-83 Locking anterior cervical plate, 5:233-239
N
Nerve root avulsion, 6: 43-50 Nerve root injury, 7:51 Neurectomy for trigeminal neuralgia, 7:99-106 Neurinoma, 6:35-41; 8:95-105 Neuroblastomas, olfactory, 7:83-91; 8: 143-151 Neuroma, acoustic, 4:159-164 Non-neoplastic lesions of the craniovertebral junction, 5:177-184
O
Obstructive hydrocephalus, 5:241-246 Occipitalized atlas, 7:249-254 Occipitoaxial fusion, 7:249-254 Odontoid transarticular screw fixation, 7:29-41 Olfactory neuroblastomas, 7:83-91; 8:143-151 Operating Arm System, 8:79-85; 8:133 Optic nerve injury, 8:12 Optic nerve sheath fenestration, 7:191-200 Orbit anatomy, 4:75-81 craniotomy, 4:85-93 lateral orbitotomy, 4:81-83 Orbitotomy, 4:81-83 Orion anterior cervical plate system, 5:101-108
P
Pain facial, 5:227-232 low back, 5:227-232 relief, 6:221-235 trigeminal neuralgia, 5:165-170 Pallidotomy microelectrode-guided, 6:27-33 MRI-guided, 7:141-148 posteroventral, 5:143-153; 6:13-26 subthalamic nucleus, 8:196-197; 8:200-201; 8:205-206 Paraclinoid carotid artery aneurysms, 8:1-12
Parkinsons disease deep brain stimulation for control of tremor, 7:125-134 MRI-guided pallidotomy, 7:141-148 posteroventral pallidotomy, 5:143-153; 6:13-26 subthalamic nucleus, 8:196-197; 8:200-201; 8:205-206 Pedical screw, 5:112-113; 5:116-117 Pedicle subtraction for flat-back syndrome, 7:53-58 Percutaneous balloon compression for trigeminal neuralgia, 7:107-116 Perimedullary AVFs, 4:11-13 Peripheral nerve suture techniques, 8:261-269 Peripheral neurectomy for trigeminal neuralgia, 7:99-106 Peritorcular meningiomas, 5:13-21 Petroclival meningiomas, 7:69-81 Petrous apex cholesterol granulomas, 8:121-125 dural AVMs, 8:24-27 Pfeiffers syndrome, 7:201 Pial synangiosis, 4: 139-146 Pineal region masses, 4:29-36 Pituitary Cushings disease, 4:165-172 surgery, 5:1-12 Plagiocephaly, posterior, 5: 43-55 Pneumothorax, postoperative, 7:162 Posterior C1-2 screw fixation, 4:19-28 Posterior cervical fusion with tension band wiring, 6:167-171 Posterior fossa dural AVMs, 8:23-46 Posterior lumbar interbody fusion, 7:1-10 Posterior occipitoaxial fusion for atlantoaxial dislocation, 7:249-254 Posterior plagiocephaly, 5: 43-55 Posterior stabilization, 5:91-100 Posterolateral tentorium dural AVMs, 8:25-29 Posteroventral pallidotomy, 5:143-153; 6:13-26 Pseudotumor cerebri lumboperitoneal shunt placement, 7:235-240
optic nerve sheath fenestration, 7:191-200 Psychiatric disease, surgery for, 7:135 Pulse generator for subthalamic nucleus stimulation, 8:205-206
R
Radiosurgery of intracranial lesions, 4:215-224 Radiosurgical dose planning, 7:94-96 Radiosurgical localization, 7:94-96 Ray Threaded Fusion Cage, 7:1-10 Raynauds syndrome, 7:158 Revascularization and dolichoectatic/fusiform aneurysms, 7:61-65 Rhizotomy dorsal, 4:183-190 spasmodic torticollis, 4:59-65 Rod placement and thoracolumbar junction fractures, 7:24-27 Root entry zone decompression, 5:165-170
S
Sacrectomy, 7:11-20 Sacrum tumors, 7:11-20 Sagittal synostosis, 5:219-225 Sathre-Chotzen syndrome, 7:201 Scalp reconstruction, 5:199-217 Schwannomas complex spinal, 7:241-242 trigeminal, 8:107-120 Screw fixation atlantoaxial instability, 4:19-28 atlantoaxial transarticular, 7:29-41 odontoid transarticular, 7:29-41 Seizures ablative epilepsy surgery, 6:131-146 corpus callosum sectioning, 4:38-39; 4:107-116 temporal lobectomy, 4:131-137 Shunt intraventricular, 6:85-98 ventriculoperitoneal, 6:193-200 Sinus fistulas, carotid-cavernous, 6:1-4 Sinus skeletonization technique, 6:51-56
Sinus, sagittal, 8:74-77 Sinus, transverse-sigmoid, 8:57-68 Spasmodic torticollis, 4:59-65 Spasticity, 4:183-190 Spina bifida, 7:219-226 Spinal cord AVMs, 4:7-10 Spinal exposure, upper thoracic, 4:173-182 Spinal instrumentation, 7:21-27 Spinal plate/screw placement, 7:23-25 Spinal stabilization cervical spine, 6:157-166 cervical spine with articular plates and screws, 5:91-100 cervical spine with the Orion system, 5:101-108 posterior, 5:91-100 thoracolumbar fractures, 5:109-119 Spinal tumor, dumbell-shaped, 7:241-248 Spinal vascular malformations, 4:1-18 Spondylolisthesis, 4:147-157; 6:183-191 Spondylolysis, 4:147-157 Stabilization cervical, 6:157-166 lateral, 4:147-157 posterior, 5:91-100 thoracolumbar fractures, 5:109-119 Stereolithography for cranial repair, 8:188 Stereotactic cingulotomy for psychiatric disease, 7:135-140 Stereotactic depth electrode implantation, 6:131-146 Stereotactic imaging and deep brain stimulation for control of tremor, 7:127-128; 7:141 Stereotactic microsurgical craniotomy, 6: 251-256 Stereotactic radiosurgery of trigeminal nerve root, 7:93-97 Stereotaxy, frameless, 5:121-128 Subclavian steal syndrome, 4:191-198 Subdural hematoma, 5:57-63 Substantia nigra pars reticulata/pars compacta, 8:201 Subthalamic nucleus, 8:196-197; 8:200-201; 8:205-206 Superior hypophyseal aneurysm, 8:1-2; 8:10-11 Superior sagittal sinus dural AVMs, 8:69-78 Supraorbital, supratrochlear neurectomy,
7:99-101 Sympathectomy, 7:157-162 Synostosis coronal, 4:67-73; 6:201-211; 7:201-210 lambdoidal, 4:44-45; 4:209-214 sagittal, 4:219-225
T
Temporal bone trauma, 7:257-260 Temporal lobe, 4:131-137; 5:75-83; 6:115-124 Tension band wiring, 6:167-171 Tentorial apex, 6:51-56 Tentorial dural AVMs, 8:23-34 Tethered cord syndrome, 7:219-226 Texas Scottish Rite Hospital forceps, 6:183-191 system, 5:109-119 Thalamic AVMs, 4:43-58 Thalamic mapping for control of tremor, 7:125-134 Third ventricular colloid cysts, 6:251-256 Third ventriculostomy for obstructive hydrocephalus, 5:241-246 Third-ventricle exposure, 4:37-42 Thoracic disc herniation, 8:217-224 Thoracic spine exposure, 4:173-182 Thoracolumbar spine burst fractures, 6:173-182 fractures, 5:109-119 junction fractures, 7:21-27 Thoracoscopic sympathectomy, 7:157-162 Thrombectomy, 7:61-62 Torcular/peritorcular meningiomas, 5:13-21 Transthoracic endoscopic sympathectomy, 7:163-171 Transverse-sigmoid sinus, 8:57-68 Tremor, 7:125-134 Trigeminal neuralgia percutaneous balloon compression, 7:107-116 peripheral neurectomy, 7:99-106 microvascular decompression of root entry zone, 5:165-170 stereotactic radiosurgery of the trigeminal nerve root, 7:93-97 Trigeminal schwannomas, 8:107-120
Tumors cavernous sinus, 4:199-207 dumbell-shaped spinal, 7:241-248 ependymomas, 4:95-106 extradural cervical spine, 5:135-141 intracranial glomus, 4:117-130 jugular foramen, 8:135-142 orbital region, 4:87-90 pineal region, 4:36 sacrum, 7:11-20 Twins, craniopagus, 6:5-11
U
Ulnar nerve entrapment, 4:235-249 submuscular transposition, 4:225-233 Unilateral coronal synostosis, 4:67-73; 7:201-210 Upper basilar trunk aneurysms, 8:87-94 AVMS, 8:87 Upper clivus dural AVMs, 8:24-27
V
Vascular malformations angiographically occult, 8:127-133 spinal, 4:1-18 Vein of Galen, 8:32-34 Vein of Labb, 8:58-60 Ventral intermediate thalamotomy, 7:125; 7:134 Ventral paraclinoid aneurysms, 8:3-4, 8:10-11 AVMs, 8:3-4; 8:10-11 Ventricular AVMs, 8:52-58 Ventricular system, 6:65-74 Ventricular trigone AVMs, 8:47-56 Ventriculoperitoneal shunt, 6:193-200 Vertebral artery, 5:135-141
W
Wrist, ulnar nerve entrapment, 4:235-249
Z
Zona incerta, 8:200
The paraclinoid segment of the internal carotid artery (ICA) extends from the proximal dural ring up to the origin of the posterior communicating artery. Aneurysms arising from this segment of the ICA account for 5% to 10% of all intracranial aneurysms. The surgical management of these aneurysms has technically challenged most neurosurgeons because of the complex anatomy of this region and, most importantly, the difficulty in obtaining proximal carotid control prior to aneurysm clipping. A clear understanding of the anatomy of this ICA segment and its adjacent osseous, vascular, and neural structures will optimize chances for successful surgical treatment of aneurysms of the paraclinoid segment.
INTRODUCTION
Of the many nomenclatures proposed to designate the segments of the ICA, we have used the modified Fischer classification proposed by the University of Cincinnati group as it describes the segments of the ICA in an anterograde sequence (Figure 1). The ICA traverses anteriorly through the cavernous sinus as the C4 segment and bends twice as it exits the sinus: medially and superiorly at first and subsequently posteriorly and superiorly, thus form-
ANATOMY
ing a loop and reversing its course by 180 degrees. This so-called anterior loop is oriented approximately 45 degrees to the base of the skull. As the oculomotor nerve crosses this anterior loop coursing toward the superior orbital fissure, its membranous epineurium blends with the adventitia of the ICA and extends across to fuse with the periosteum of the sphenoid bone. As it encircles the ICA, this layer is termed the proximal dural ring; the tissue between the oculomotor nerve and the ICA is called the carotid-oculomotor membrane. After the ICA completes the second bend, it emerges either from the roof or from the cavernous sinus, under the anterior clinoid process and continues intracranially as the C6 segment. The dura encircling the ICA as it exits the cavernous sinus forms the distal dural ring. The segment of the ICA between the proximal and distal dural rings is extracavernous, but not intradural, and is termed the C5 segment. The C5 and C6 segments of the ICA together constitute the paraclinoid segment. The proximal C6 segment of the ICA remains hidden under the anterior clinoid process beyond the distal dural ring. Two branches arise from this segment of the ICA: the ophthalmic artery and the superior hypophyseal artery. The ophthalmic artery arises distal to the distal dural ring on the superior surface of the ICA, bends forward, and travels through the optic foramen lying inferior and lateral to the optic nerve. At times, the ophthalmic artery is adherent to the dura of the optic canal floor and hence is insepa-
ACA
MCA
PCoA
Figure 1. Depiction of the segments of the ICA in an anterograde sequence. The paraclinoid segment consists of C5and C6segments (University of Cincinnati modification of the Fischer classification). ACA = anterior cerebral artery; PCoA = posterior communicating artery. (Reproduced from Bouthillier A, van Loveren HR, Keller JT: Segments of the internal carotid artery: a new classification. Neurosurgery 38:425-433, 1996, with permission)
rable. The superior hypophyseal artery, which can be single or multiple, arises from the medial surface of the ICA as it emerges from the distal dural ring and crosses over the diaphragma sellae.
Based on the site of origin and direction of projection as it relates to the C5 and C6 segments, the aneurysms in this location can be classified into four variants, types I through IV (Figure 2). Any of the four variants of the aneurysms described below can enlarge, making classification difficult and impractical. Types I(a) and I(b): Type I aneurysms are also called carotid-ophthalmic aneurysms. The type I(a)
CLASSIFICATION
aneurysm variant is the most common paraclinoid aneurysm; it arises from the dorsal surface of the C6 segment close to the ophthalmic artery. These aneurysms project superiorly into the subarachnoid space, displacing the optic nerve upward and medially. Presenting symptoms are either visual or due to rupture with subarachnoid hemorrhage (SAH). These aneurysms can at times erode the anterior clinoid process to a thin shell. The type I(b) aneurysm variant represents a small subset of aneurysms that arise from the superior surface of the C6 segment of the ICA, often a few millimeters from the ophthalmic artery origin. They are typically broad-based, sessile, and thin-walled.
OA
type IV
OA
type III(a) OA
SHA DS
type III(b)
Figure 2. Illustration showing the four variants of aneurysms. Type I(a) and I(b) variants arise from the dorsal surface of the C6 segment; the type I(a) aneurysm is closely related to the ophthalmic artery (OA) origin and the type I(b) aneurysm has no branch relation and is often sessile. The type II variant arises from the ventral surface of the C6 segment without branch relation. Insets A and B show variants in types III and IV. Type III(a) and III(b) aneurysms (supra- and infradiaphragmatic variants), closely related to the superior hypophyseal artery (SHA) origin, arise on the medial surface of C6 and C5 segments. The type IV variant is a large broad-based aneurysm extending from the distal C4 segment to the proximal C6 segment, widening the distal dural ring. DS = diaphragma sellae.
Type II: Type II aneurysms, also called ventral paraclinoid aneurysms, arise from the ventral surfaces of the C6 segment of the ICA. They have a broad base and the dome projects toward the r oof of the cavernous sinus. The dome may extend into the cavernous sinus in larger aneurysms. These aneurysms often present with third cranial nerve palsy and rarely with SAH. Types III(a) and III(b): Type III aneurysms (also called carotid-superior hypophyseal aneurysms) are closely related to the origin of the superior hypophyseal artery. Type III(a) aneurysms arise from the medial surface of the proximal C6 segment, project above the diaphragma sellae, and can present with SAH. This variant is the so-called carotid cave aneurysm. The type III(b) variant arises from the medial surface of the C5 segment and projects below the diaphragma sellae and can be mistaken for a sellar mass on computed tomography (CT). Both variants can enlarge to the extent of presenting both above and below the diaphragma sellae. Type IV: Type IV aneurysms (also called broadbased siphon aneurysms) are often large or giant, and span from distal C4 to proximal C6 segments. At least one half of the circumference of the vessel wall forms the broad base of the aneurysm. The base expands into the cavernous sinus and into the subarachnoid space. The dome often elevates the roof of the cavernous sinus and, at times, enlarges the proximal and distal dural rings, thus projecting outside the cavernous sinus under the anterior clinoid process.
All patients diagnosed with paraclinoid-carotid aneurysms undergo a detailed preoperative neuroophthalmological evaluation. Patients presenting without SAH undergo detailed visual field evaluation, and those presenting with SAH undergo bedside evaluation by confrontation methods.
Computed Tomography
Cerebral Angiography
Based on the type of paraclinoid-carotid aneurysm, presenting symptoms can include SAH, ocular symptoms of decreased visual acuity, visual field impairment, and diplopia; at times, retro-orbital pain is also a presenting symptom. Visual symptoms are usually ipsilateral but can be bilateral, with contralateral central scotoma because of the involvement of the knee fibers of von Willebrand.
PRESENTING SYMPTOMS
Selective internal carotid angiography is performed on all patients via transfemoral selective catheterization. The direction of projection allows classification of an aneurysm into one of the four types: types I(a), I(b), II, and IV are best visualized on the lateral projection; type III is best seen in anteroposterior and submento-vertical projections (Figure 3).
Patients presenting with symptoms of SAH are admitted to the intensive care unit. Their clinical condition is classified according to the Hunt and Hess grading system. Respiratory support is provided for patients with impaired sensorium. Indwelling catheters are used to monitor systemic arterial pressure, central venous pressure, and pulmonary arterial and wedge pressures. Cardiac output, cardiac index, and systemic vascular resistance are optimized for each patient. All patients receive intravenous phenytoin, and therapeutic levels are maintained. Nimodipine is administered in doses of 60 mg every 4 hours orally or via a nasal gastric tube.
INITIAL MANAGEMENT
In patients with large or giant aneurysms, especially a type IV variant or an aneurysm with extensive calcification, balloon test occlusion (BTO) with systemic hypotension and single photon emission CT (SPECT) are used to evaluate the patients tolerance for carotid occlusion as a definitive therapy and to assess the need for a bypass procedure. If a patient tolerates the test occlusion clinically and SPECT does not reveal perfusion defects, permanent balloon occlusion of the ICA and trapping of the aneurysm are offered as definitive therapy. A vascular bypass is carried out in patients who do not tolerate the BTO or who have significant perfusion defects on SPECT. If the superficial temporal artery (STA) is greater than 1.5 mm in diameter, an STA to M2 segment of the middle cerebral artery (MCA) bypass is carried out; if the STA is less than 1.5 mm, a vein bypass graft is used from the cervical ICA to the M2 segment of the MCA.
Figure 3. The four variants of aneurysms seen by angiography. Types I (A), II (B), and IV (D) are best visualized on lateral angiography and type III (C) on anteroposterior projection.
6
Magnetic Resonance Imaging and MR Angiography
The exact dimensions of an aneurysm filled with a thrombus are better determined by a magnetic resonance imaging (MRI) study. Coronal MRI of the sellar area may preoperatively identify types III(a) and III(b), the supra- and infradiaphragmatic variants of the paraclinoid aneurysm. This knowledge will help in planning the extent of diaphragmatic division needed to provide adequate exposure of the aneurysm. The infradiaphragmatic variant is visualized only after the diaphragma sellae is divided around the superior surface of the dome of the aneurysm. Small and asymptomatic infradiaphragmatic variants may not require surgical intervention if diagnosed preoperatively.
The neck is included in the sterile operative field. The angle of the mandible and the anterior margin of the sternocleidomastoid muscle are marked after prepping so that the carotid artery can either be compressed digitally or exposed for proximal control, trapping, or a vein bypass procedure.
SURGICAL TECHNIQUE
Timing of Surgery
Patients who are classified in Hunt and Hess Grades I-IV undergo surgery within 24 hours after admission. Surgery is delayed for patients classified in a Hunt and Hess Grade V or with multiple systemic problems (e.g., sepsis or aspiration pneumonia) until their clinical grade or general condition improves.
Intra-arterial and Swan-Ganz catheters are inserted to monitor blood pressure and pulmonary arterial and wedge pressures. The anesthetic agents used include propofol (2.5 mg/kg), sufentanil (1-2 mcg/kg), vecuronium (0.1 mg/kg), lidocaine (1.5 mg/kg), and 100% oxygen. Anesthesia is maintained with a percentage concentration of Forane at subminimal alveolar concentration (i.e., the amount required to prevent movement with skin incision), continuous infusion of propofol (25-75 mcg/kg/min), and sufentanil (0.5-1.0 mcg/kg/min). Mean arterial pressure is maintained at 70 to 80 mm Hg. Mannitol is administered as a 0.5-gm/kg bolus prior to craniotomy. If a temporary clip cannot be applied or if the aneurysm ruptures during dissection, the mean arterial pressure is lowered to 40 to 60 mm Hg by increasing the concentration of Forane, a nitroprusside infusion, or both.
Anesthetic Technique
The skin incision starts 1 cm anterior to the preauricular area just above the zygomatic arch, continues toward the midline, and curves forward toward the forehead, ending at the hairline in the opposite midpupillary line (Figure 4). If the anterior branch of the STA is larger than the posterior branch, the scalp incision starts posterior to the main trunk, thus keeping the main trunk with the scalp flap. If the posterior branch is larger, the incision is made anterior to the STA trunk. The temporalis muscle is incised along with the scalp and retracted forward toward the orbital ridge. The supraorbital ridge and the frontozygomatic suture are exposed. The supraorbital nerve and vessel are retracted with the scalp flap; an osteotome is used to isolate the bony margins around the neurovascular bundle, if enclosed in an osseous foramen, thus allowing the bundle to be mobilized along with the scalp. The orbital periosteum is separated superiorly and laterally from within the orbit. Precautions are taken to avoid perforation of the periorbital fascia, which could cause herniation of the periorbital fat. If the subperiosteal plane is maintained around the orbital ridge, the chance of disruption of the orbital periosteum is lessened. Figure 5 shows frontal and oblique views of the craniotomy and oblique osteotomy. Three burr holes are placed using a 7-mm Acra-Cut perforator. The first burr hole is made 1 cm above and lateral to the nasion. If the hole fails to penetrate both the inner and outer tables because of a large frontal sinus, the inner table is drilled separately to expose the dura. A second burr hole is made 5 cm posterior to the first one and 1 cm from the midline, midway between the coronal suture and orbital ridge. A third burr hole is made in the squamous temporal bone above the midzygomatic arch. The three burr holes are connected parallel to the midline and posteriorly. For left-sided aneurysms, the craniotomy is made as close to the midline as possible. A keyhole opening is made at the level of the orbital roof, posterior to the frontozygomatic suture, using an acorn-shaped dissecting burr; this opening exposes the floor of the anterior cranial fossa superiorly and the orbit inferiorly. Once the dura mater and orbital periosteum are separated from either side of the orbital roof, its lateral-most part is nibbled using a needle-nose Leksell rongeur.
OPERATIVE PROCEDURE
The patient is positioned supine with the head and thorax elevated by 15 degrees; the neck is neutral so that the projected plane of the orbital roof is perpendicular to the ground. The head is rotated 15 degrees to the contralateral direction. A motorized operative table permits most changes in positioning during surgery. A full-length silicone gel pillow is placed on the table.
Patient Positioning
facial nerve
STA
Figure 4. Scalp incision preserving the STA trunk and the frontalis innervation.
Figure 5. Frontal (A) and posterior oblique (B) views outlining the frontotemporal craniotomy with supraorbital ridge osteotomy, elevated as a single flap. Note the keyhole burr hole with orbit inferiorly and frontal dura superiorly, and that the pterion is drilled down.
An osteotomy extends across the orbital ridge from the supraorbital burr hole into the orbital roof using a C1 dissecting tool. A second osteotomy is performed across the frontozygomatic suture and the lateral wall of the orbit. The pterional area is drilled down and thinned. At this point, the cranial bone flap with the supraorbital ridge is lifted from the dura and rotated forward and down. As a result, the orbital roof will fracture, connecting the keyhole opening to the supraorbital osteotomy across the orbital roof. The bone flap is removed in one piece with the orbital rim. A self-retaining retractor system is attached to the Mayfield clamp. The dura is elevated from the anterior cranial fossa and the tip of the middle fossa. The pterion and greater wing of the sphenoid are drilled flat until the entire shiny inner cortical bone is seen and the lateral edge of the superior orbital fissure is visualized. The posterior part of the orbital roof and the lateral part of the lesser wing are removed using a fine rongeur. The frontal sinus is often entered and the entire mucosa and posterior wall are removed, thus cranializing the sinus. The frontonasal duct is obliterated by a small piece of temporalis muscle graft.
basal surface of the frontal lobe. The dura over the anterior clinoid process and the optic canal is coagulated, incised, and elevated medially as a flap in order to expose the roof of the optic canal and the anterior clinoid process (Figure 6A). Using a diamond-tipped dissecting tool, the entire optic canal is unroofed in a posteroanterior direction under continuous irrigation. Next, the anterior clinoid process is drilled from its tip to the base, keeping in mind that the contents of the superior orbital fissure are on its inferolateral aspect, the anterior loop of the ICA is directly underneath, and the optic nerve is medial (Figure 6B). Once the anterior clinoid process is resected, the optic strut is further isolated by the elevation of the dura from the floor of the optic canal superiorly and the second deflection of the anterior loop of the ICA downward. Using a fine diamond-tipped burr, the optic strut is drilled further. Curettes and rongeurs are not used, as they are not as precise and atraumatic as the drill.
The dura is opened in a T-shaped fashion. Its stem continues along the sylvian fissure: one limb extends medially across the orbital ridge toward the midline and the other limb extends laterally toward the temporal pole and the floor of the middle fossa, thus leaving the dural flaps to protect the brain surface. Under microscopic magnification, the sylvian fissure is opened widely, usually from its lateral to medial end. The entire MCA complex is exposed. The sylvian veins are preserved and are left attached to the temporal lobe. The frontal lobe is elevated from the optic nerve and optic chiasm; all arachnoid strands are sharply divided. The lamina terminalis is exposed and opened to allow drainage of cerebrospinal fluid from the ventricles. The cisterns around the optic nerve and the ICA are opened widely. In type I aneurysms that project superiorly, the dome of the aneurysm may be adherent to the frontal lobe. Elevation and traction of the frontal lobe must be done gently, dividing all arachnoid strands until the entire area is exposed.
Once the frontal lobe is elevated, the retractor blades are adjusted. The tips of the blade are inclined toward the tip of the anterior clinoid process to create a conical exposure (wide near the surface and narrow at the depth near the anterior clinoid process), thus minimizing the retraction pressure that is on the
To attain proximal control, the ICA can be exposed in three areas: Cervical carotid artery (C1). If the aneurysm is large and encroaches on the entire cavernous sinus, the cervical carotid artery is exposed for proximal control. Cavernous carotid segment (C4). The temporal lobe is retracted laterally to expose the oculomotor nerve as it enters the roof of the cavernous sinus. A sickle-shaped knife is used to incise the dural sleeve on the medial margin of the oculomotor nerve from the point of its entrance into the roof of the cavernous sinus to the superior orbital fissure. This allows the entire contents of the superior orbital fissure to be retracted away from the anterior loop of the ICA. The cavernous sinus is opened by incising its roof, extending from the distal dural ring to the posterior clinoid process. Bleeding encountered from the cavernous sinus, both medial and lateral to the cavernous carotid artery, is controlled by gentle packing with small pieces of Gelfoam. Dissection of the lateral aspect of the ICA must remain close to the artery to prevent injury of the sixth cranial nerve. The cavernous carotid artery is isolated (Figure 6C), thus enabling placement of a temporary clip for proximal control. C5 segment. The distal dural ring is divided circumferentially around the ICA across the floor of the optic canal (after the ophthalmic artery is dissected free) and across the roof of the cavernous sinus toward the diaphragma sellae. The C5 segment is dissected from the inferior surface of the optic strut, enabling temporary clip placement for proximal control.
optic canal
II
A
III
II
optic strut
C5
III C6
II
B
distal dural ring
ophthalmic artery
III
Gelfoam
Figure 6. A, dural incision outlined over the anterior clinoid process and the optic canal; B, the optic canal is unroofed and anterior clinoid process drilled away; note the oculomotor nerve seen through the membranous layer after clinoidectomy. C, carotidoculomotor membrane is incised medial to the third nerve and the nerve is retracted laterally. The cavernous sinus roof is opened and the C4 and C5 segments of the ICA are exposed for proximal control. II and III indicate cranial nerves.
The four variants of aneurysms are shown with ideal placement of clips in Figure 7.
In type I(a) aneurysms, the optic canal sheath is incised longitudinally to allow gentle retraction of the optic nerve. The arachnoid strands, which lie between the optic nerve and the ICA and the aneurysm, are sharply divided. The ophthalmic artery is identified. The carotid artery is displaced laterally. With large aneurysms, proximal control is achieved by exposing the C4 or C5 segments of the ICA. Most aneurysms projecting superiorly can be clipped by using a 45-degree angled clip, placing the blades parallel to the long axis of the carotid artery. At times, a carotid segment can be so ectatic that the aneurysm neck may be located medial to the optic nerve. In these instances, the tuberculum sella is drilled medial to the optic nerve. The optic nerve is retracted laterally and the clip is applied from its medial aspect.
Type I(b) aneurysms are difficult to obliterate because they are broad-based and sessile. Wide mobilization of the carotid artery with circumferential division of the distal dural ring followed by proximal and distal temporary clipping make the segment of the ICA become slack, thus allowing satisfactory clip placement with the blades parallel to the carotid artery. At times, a portion of the parent vessel may need to be included in the clip blades, as the base is very thin and fragile and may tear with approximation of the clip blades.
Preoperative studies do not always clarify whether a type III aneurysm is projecting above or below the diaphragma sellae. Definitive assessment of the location can only be made by exploration. Proximal control is achieved in large aneurysms by exposing the cavernous carotid artery (C4 segment). The optic nerve sheath is incised and the optic nerve is gently retracted medially to bring the supradiaphragmatic variant into view. After isolating the ophthalmic artery, the dura along the floor of the optic canal is incised and the C5 segment is mobilized laterally. The supradiaphragmatic variant aneurysm can be visualized and a 90-degree angled fenestrated clip is applied from a lateral direction. In the infradiaphragmatic variant, the aneurysm is hidden under the diaphragm and not visible when the C6 segment is displaced laterally. When faced with this variant, the distal dural ring is incised circumferentially around the carotid artery, and extending medially across the diaphragma sellae, leaving a cuff of the dura attached to the periphery of the aneurysm. The roof of the cavernous sinus is opened during this process, and bleeding is controlled by packing with Gelfoam. The pituitary gland is identified medial to the aneurysm. A 90-degree curved fenestrated clip (placed encircling the ICA) will obliterate the aneurysm.
The optic nerve sheath is incised longitudinally to facilitate gentle medial retraction of type II aneurysms. The distal dural ring along the floor of the optic canal is incised and the C5 segment is mobilized laterally for proximal control. A right-angled fenestrated clip is placed, with the blades on the undersurface of the ICA parallel to its long axis and the parent vessel passing through the fenestration. At times, a second fenestrated clip may need to be placed parallel to the first one so as to totally obliterate and occlude the neck of the aneurysm. If the aneurysm projects significantly into the cavernous sinus, the rigid dura of the roof of the cavernous sinus that encircles the aneurysm does not allow the clip blades to approximate. Dura forming the roof of the cavernous sinus can be incised circumferentially around the waist of the aneurysm, allowing the clip blades to approximate and occlude the neck of the aneurysm. The tips of the clip blades must be
When clipping a type IV aneurysm, the optic nerve sheath is incised and the nerve is gently retracted medially. The ophthalmic artery is identified. The dura is incised along the roof of the cavernous sinus medial to the oculomotor nerve, and the nerve is retracted laterally. The distal dural ring is often widened or made incompetent by the large aneurysm. This dural ring, along with the dura on the roof of the cavernous sinus, is incised around the aneurysm. The dura on the floor of the optic canal under the ophthalmic artery is incised, thus enabling the mobilization of the C5 and C6 segments. A temporary clip is placed on the C4 and C6 segments, and the aneurysm is trapped. A right-angled fenestrated clip is applied encircling the carotid artery; the clip blades are placed on the inner curvature of the artery along the neck of the aneurysm. Often, multiple serial clips are needed to obliterate the entire length of the aneurysm neck. If the clip slides toward the carotid artery and compromises the lumen, a second clip is placed tangentially; the aneurysm is collapsed with a 25-gauge needle and the first clip is removed. After restoration of the patency of the carotid artery and obliteration the aneurysm, the temporary clips are removed.
11
ophthalmic artery
MCA
type I
ACA
type II
type III
type IV
Figure 7. Illustrations depicting the four variants of aneurysms with ideal clip placement.
12
Pitfalls
Type IV aneurysms and, less frequently, variants of paraclinoid aneurysms may contain significant calcification in the vessel wall. Attempts to clip a heavily calcified aneurysm may be dangerous because of the risks of avulsion of the aneurysm, embolic phenomena with ischemic complications, or the inability to obliterate the aneurysm. Obliteration of a markedly calcified aneurysm is difficult in elderly patients. Preoperative CT scans may alert the surgeon to the possibility of calcification. If a calcified aneurysm is encountered and if the presenting symptoms are not due to SAH, the procedure may be terminated or a bypass pr ocedure performed, followed by endovascular balloon occlusion of the ICA.
medially. The ophthalmic artery must be identified before incising the dura on the floor of the optic canal and the distal dural ring to avoid inadvertent injury.
The frontal sinus entry must be identified and properly managed. The sphenoid sinus and the posterior ethmoid sinuses may have opened during clinoidectomy and be recognized and obliterated.
The anterior clinoid process and optic strut may be pneumatized and must be recognized while drilling. This sinus communication must be obliterated prior to closure. A small fat graft obtained from deep temporal fat or from the abdomen is secured in place with fibrin glue. Wound closure is done in the usual fashion to achieve a watertight dural closure. The frontal sinus is obliterated with fat or muscle and secured in place with cryoprecipitate. A small local periosteal flap is rotated from the scalp over the sinus and sutured to the adjacent dura. The fronto-orbital bone flap is replaced and secured with plates and screws. A subgaleal suction drain is left in place for 24 hours.
Closure
Patency of the carotid artery is easily evaluated using microDoppler ultrasound after aneurysm obliteration. In large and complex aneurysms, a radiolucent headholder is used in preparation for angiography and the patients neck is included in the sterile operative field. The ICA is punctured percutaneously or by an open method, and the angiography is completed.
Using the above classification, 99 (95%) of 115 patients who had aneurysms near the paraclinoid segment underwent successful clipping of the aneurysm. Four patients underwent a bypass procedure to the M2 segment of the MCA, using a saphenous vein bypass in two patients and an STA in the other two. Postoperative endovascular trapping of the aneurysm was completed following the confirmation of the patency of the bypass. The remaining patients underwent trapping without a bypass.
Surgical Experience
COMPLICATIONS
This is the most frequently encountered postoperative complication in the management of paraclinoidcarotid aneurysms. Although the optic nerve can be injured while unroofing the optic canal, the chances for injury can be decreased by following the principles of drilling: Hold tight and stroke gently. The dura covering the optic nerve must not be disrupted by the drill. The field is continuously irrigated with cool saline to decrease the chance of thermal injury to the nerve. The pial vessels of the optic nerve must not be disrupted while incising the dural sleeve. Retraction must be minimal and brief. Complete exposure of the anterior loop and extensive mobilization is preferred, thus allowing the artery to be retracted laterally rather than retracting the nerve
Patients with ICA aneurysms of the paraclinoid segment who are considered for surgical intervention should be evaluated thoroughly using selective angiography. In patients with complex and large aneurysms, CT and MRI studies can provide information regarding their size, relation to the adjacent structures, and presence of calcification. Identification of the aneurysms into one of four types facilitates operative planning, thus allowing successful clipping. As optic nerve dysfunction is the most frequent complication, very little retraction should be applied to the optic nerve, and the microvasculature of the optic nerve should not be disturbed. A clear understanding of the paraclinoid segment of the ICA anatomy is important to ensure successful treatment of aneurysms presenting in this area.
CONCLUSIONS
The middle cerebral artery (MCA) is the second most common site of aneurysm formation; in Yasargils 1984 series it represented 20% of all aneurysms. In addition, aneurysms in this region are the third most common site associated with subarachnoid hemorrhage (SAH). Typically, the aneurysms occur at the bifurcation of the M1 segment of the MCA; however, they may occur proximal or distal to this point. Aneurysms in this location are particularly challenging to the microvascular surgeon for several reasons, including the following: 1) they are frequently associated with hematomas that may be adherent to and difficult to dissect from the fundus of the aneurysm; 2) the anatomical variability of the M1 segment and the associated medial and lateral lenticulostriate vessels require meticulous technique; and 3) the overall anatomic complexity of the sylvian fissure demands patient microsurgical dissection. In addition, MCA aneurysms are frequently broad-based and include a portion of the distal M2 segment, thereby requiring unusual or multiple clip configurations. In 1944, Dandy provided the first report on surgical treatment of MCA aneurysms. Each case resulted in a fatality. Successful treatment of an aneurysm in this region was first reported by Dott. Currently,
INTRODUCTION
there are three surgical approaches to MCA aneurysms: 1) proximal to distal, exposing the internal carotid artery within the basal cistern and dissecting along the path of the sylvian fissure; 2) a transtemporal approach through the superior temporal gyrus; and 3) the transsylvian approach with direct splitting of the sylvian fissure. This chapter reviews the anatomy of the sylvian fissure as well as the preoperative assessment and management of aneurysms associated with the MCA. More specifically, the various surgical techniques for treatment of these lesions and the complications associated with their management are discussed.
The internal carotid artery bifurcates lateral to the optic chiasm, at the junction of the carotid and sylvian cisterns, into the MCA and the anterior cerebral artery (Figure 1). In approximately 70% of cases, the MCA is the larger branch. The first segment (M1) of the MCA courses laterally 1 cm posterior to the sphenoid wing and enters the deep sylvian fissure running horizontally in its sphenoidal portion. The mean length of the M1 segment is 14 to 17 mm. Several important and variable vessels arise from the M1 segment, which can be classified in the inferior medial (perforating) or superior lateral (temporal) groups. It is of paramount importance to understand the normal neurovascular course as well as the common variants of the MCA branches in order to adequately interpret angiographic anatomy and effectively deal with surgical lesions.
SURGICAL ANATOMY
13
14
temporal lobe
optic nerve
artery of Heubner
sylvian fissure
frontal lobe
Figure 1. Schematic view of the sylvian fissure after microsurgical opening. The origin of the MCA from the internal carotid artery and its proximal branches are demonstrated. M2: branches of the main MCA (M1) segment.
15
The inferior medial group (lenticulostriate perforators) originates on the inferior medial surface of the M1 segment, demonstrating significant variations in size and number (range two to 29). They may arise as individual branches directly from the M1 segment or as larger feeders that branch into multiple smaller arteries. Yasargil notes that in 40% of cases, a single large vessel originates from the M1 segment and then divides into smaller perforators. Two additional patterns are described, one (30%) where there are two main branches that thereafter divide and the other (30%) where the small lenticulostriate vessels directly arise from the M1 segment. Lenticulostriate vessels may also arise from the proximal portion of one of the two M2 segments. The lenticulostriate vessels course medially into the lateral aspect of the anterior perforated substance, supplying a portion of the nuclei of the basal ganglia, the superior half of the internal capsule, and the lateral segment of the anterior commissure. The superior lateral group of vessels includes the uncal artery, the polar temporal artery, and the anterior temporal artery. The anterior temporal branch courses anteriorly and turns above the superficial aspect of the sylvian fissure en route to the anterior temporal lobe surface. The uncal artery arises 30% of the time as a lateral M1 branch and 70% as a direct branch from the internal carotid artery. On occasion, the polar temporal artery is hypoplastic and the anterior temporal artery is reciprocally larger. In this case, the anterior temporal artery supplies the distribution of both vessels. The surgeon must also be careful in this situation not to confuse the anterior temporal artery take-off with the true M1 bifurcation. An alternate variant in which both the polar temporal artery and the anterior temporal artery are hypoplastic has been described. In this situation, a large branch originating from the inferior M2 segment usually supplies the respective territories. The MCA bifurcation is located just medial to the high point of the limen insulae at the junction of the sphenoidal and opercular-insular subdivisions of the sylvian cistern. Because the sylvian cistern is surrounded by brain, MCA bifurcation aneurysms are often embedded in the parenchyma. Therefore, ruptured lesions commonly present with intraparenchymal hemorrhage in addition to SAH. The junction of the temporal and frontal operculas can be used as superficial cortical landmarks to commence exploration of the bifurcation of M1 and M2. The M1-M2 junction is usually located 1.5-2 cm posterior to the anterior aspect of the insular portion of the sylvian fissure and 2-2.5 cm deep to the superior temporal sulcus. Two M2 branches arise from the bifurcation, the superior and inferior segments. The M2 segments then turn superior-posterior reaching the insular surface. The two segments separate initially at the
bifurcation and then come into close proximity in the opercular-insular portion of the sylvian fissure. Often distal to the true bifurcation, large M3 branches might arise, creating the impression of a trifurcation as well as causing confusion as to the location of the true bifurcation. On the other hand, a true trifurcation or greater number of branching vessels is found in up to 22% of cases. In true M1 bifurcations, the inferior trunk is larger in 41% of cases, the superior trunk in 36%, and they are of equal size in 23%. The territory supplied by the superior trunk includes the inferior frontal cortex, the frontal opercular cortex, and the parietal and central sulcus territories. The territory supplied by the inferior trunk includes the middle temporal cortex, the posterior temporal cortex, and the angular and posterior parietal regions.
It is important to be aware of the more common vascular abnormalities present in the MCA. McCormick noted in his landmark autopsy study of 1000 consecutive brains that the MCA was the least likely major intracranial vessel to harbor an anatomical abnormality. Albeit uncommon, aneurysms appear more frequently in patients with anatomical neurovascular abnormalities. Fenestrations are reported in the proximal portion of the M1 in 2% of cases and are often associated with aneurysms, which form secondary to intimal deficiencies at the proximal bifurcation. An accessory MCA may also be present and originate from the junction of A1 and A2, proximal A1, or the distal internal carotid artery.
We recommend that the patient presenting in good neurological status (Grades I-III on the Hunt-Hess scale) after an SAH be treated urgently as the risk for rehemorrhage is significant. However, each case must be individualized and surgical decisions should be based on age, grade of patient, aneurysm size and complexity, medical issues, and operating room personnel. The patient in a poor neurological status (Grades IV-V on the Hunt-Hess scale) may improve following a ventriculostomy and become a good candidate for surgery. Aneurysms in this region are frequently associated with large hematomas and mass effect causing a worsened neurological picture. In this case, emergent surgery may also be indicated based on mass effect. The management of incidental aneurysms in this region is controversial because of conflicting data on their natural history. Recent data suggest that size is the critical issue in deciding to operate. All aneurysms greater than 10 mm are treated unless there are medical contraindications. For lesions less than 10 mm in size, the surgeon must perform a r elative
PATIENT SELECTION
16
risk analysis between the surgical morbidity/mortality rate and the risk for aneurysm rupture which, after a rupture occurs, has a mortality rate of up to 60%. Again, surgical decision making is multifactorial and each case must be treated individually.
nous run-in. In addition, all bony margins are waxed to prevent air emboli. Subsequently, the dura is placed on stretch with sutures and covered with moist patties to facilitate dural closure upon completion of the case.
SURGICAL TECHNIQUE
The patient is placed supine on the operating table with a roll under the ipsilateral shoulder. The skull is immobilized with the Mayfield headholder, elevated above the heart, and rotated 45 degrees contralateral to the incision; the neck is flexed slightly to the contralateral shoulder to flatten the orbital roof and the vertex is lowered slightly. These maneuvers facilitate opening of the sylvian fissure with minimal traction on the frontal and temporal lobes. Care is taken to avoid venous outflow obstruction in the neck, either through positioning or methods used to secure the endotracheal tube. This position allows the surgeon to pursue any of the three surgical approaches. A standard frontotemporal or pterional incision, made completely behind the hairline, is routinely used (Figure 2). The incision is started no more than 1 cm anterior to the tragus of the ear at the root of the zygoma and extends behind the hairline up to the midline. Placing the incision in this location prevents injury to the frontalis branch of the seventh cranial nerve. The skin is dissected separately and reflected inferiorly over a rolled sponge with fishhook retractors. It is important to stay deep to the fat-pad in the keyhole region to avoid injury to the frontalis branch of the seventh nerve. The temporalis muscle is then incised along the superior temporal line and the posterior margin of the incision. The muscle is reflected inferiorly with the skin flap to provide adequate bony exposure. Leaving the superior tuft of muscle allows for a better cosmetic closure upon completion of the case. Three burr holes are placed individually in the pterion or keyhole region, the temporal squamous bone slightly above the root of the zygoma, and at the posterior margin of the incision at the height of the temporalis muscle. A standard pterional bone flap is elevated with cuts between the burr holes, as demonstrated in Figure 2. The temporal squamous bone is removed with rongeurs and the lateral roof of the orbit and lesser wing of the sphenoid bone are drilled down until the orbital meningeal artery is visualized. This bony removal provides better exposure of the proximal sylvian fissure and allows gentle traction on the temporal lobe. After Surgicel and narrow cottonoid patties are placed around the margins of the craniotomy, the dura is opened in a crescent moon-shaped fashion centered over the sylvian fissure. This minimizes ve-
SURGICAL APPROACHES
The transtemporal approach utilizes the skin incision and bony removal described above with slightly less removal of the lateral orbital roof and lesser wing of the sphenoid bone. A 3-4 cm cortisectomy is made in the superior temporal gyrus and centered approximately 2 cm posterior to the anterior-most aspect of the sylvian fissure. A subpial resection of the superior temporal gyrus is utilized to enter the horizontal portion of the sylvian fissure, enabling the surgeon to identify the M2 segments of the MCA. Dissection is then taken from distal to proximal within the sylvian fissure to identify the M1 segment, thereby providing proximal and distal control of the aneurysm. The lenticulostriate vessels, branching points of the M1 segment of the MCA, the aneurysm neck, and the fundus are also identified. This approach was initially advocated and popularized for patients who presented with large temporal lobe hematomas. A transtemporal approach is advantageous in that it requires a smaller craniotomy and provides a direct approach to the essential anatomy with good visualization of the inferior M2 trunk. In addition, there is less brain retraction and less manipulation of the M1 segment. The disadvantages of this approach include minimal release of cerebrospinal fluid (CSF) resulting in manipulation and retraction of a firm, stiff brain parenchyma. There is also more tissue resection than required by the other approaches. This may subject the patient to an increased risk for seizures. Finally, the aneurysm fundus is often exposed first with no proximal control.
Position
Transtemporal Approach
The subfrontal/pterional approach utilizes the positioning, craniotomy, and dural opening described earlier. Adequate bone removal of the orbital roof and lesser wing of the sphenoid down to the orbital meningeal artery is required to provide proximal exposure to the sylvian fissure. In addition, a temporal craniectomy is required to allow retraction of the temporal lobe. A retractor blade is advanced subfrontally and secured under microscopic visualization to provide exposure of the olfactory tract and optic and carotid cisterns. These cisterns are opened sharply with an arachnoid knife, thereby allowing the egression of spinal fluid and subsequent softening of the brain parenchyma. The exposure is then
Subfrontal/Pterional Approach
17
temporalis muscle
Figure 2. Schematic views of the head position, skin and muscle incisions, placement of burr holes, and location of the craniotomy. A, note the location of the skin and muscle incision within 1 cm of the tragus to avoid injury to the frontalis division of the facial nerve. The temporalis muscle is incised within its superior attachment to facilitate closure and provide better cosmesis. Note the position of temporal squamous bone, removed with rongeurs. This allows gentle traction on the temporal lobe. B, the skin flap is reflected individually, elevating the facial nerve with it. Subsequently the temporalis muscle is elevated, leaving the superior tuft of muscle; both the skin and muscle are isolated with fishhook retractors. The craniotomy is centered over the sylvian fissure.
craniotomy
18
taken distally along the internal carotid artery to its bifurcation and into the proximal sylvian fissure, again using sharp dissection. The entire M1 segment is exposed within the horizontal portion and into the opercular-insular region of the sylvian fissure, providing exposure of the bifurcation of the M1 segment into the M2 segments, as well as the neck of the aneurysm. With this exposure, the surgeon can also visualize the medial lenticulostriate vessels, which extend along the frontal surface of the anterior cerebral artery, and the M1 and M2 segments of the MCA. This dissection provides an excellent view of the anatomy of the sylvian fissure. Because of this wide opening, retraction is required of both the frontal and the temporal lobes. This approach offers several advantages, including early release of CSF and proximal control of the aneurysm. Concerns regarding a wide opening of the sylvian fissure include prolonged frontal lobe retraction and extensive manipulation of the entire M1 and M2 segments, which may induce spasm in these vessels.
sequentially providing exposure of the lateral lenticulostriate vessels, as well as the limen insula, the neck of the aneurysm, and finally the M1 segment. Arteries within the sylvian fissure either run frontally or temporally and can be reflected in their respective directions. Disadvantages of this approach include late proximal control in addition to frontal and temporal retraction to provide complete exposure of the aneurysm neck and branches of the MCA. Advantages of this technique include early CSF release and brain softening, minimal brain dissection, removal of subarachnoid blood, and wide exposure of the MCA bifurcation. Once dissection is complete, the objective is to occlude the aneurysm with a clip while preserving the normal anatomy without any stenosis, as there is minimal collateral flow in this distal region of the arterial tree.
The transsylvian approach is the one that we most commonly use and it has recently been described by many other authors. The positioning, incision, and craniotomy described above facilitate the application of sharp microdissection techniques to enter the opercular-insular segment of the sylvian fissure for the initial exposure. The atmosphere of the operating suite is calm and quiet with the surgeon in a seated position. The self-retaining retractor system, microscope, surgical assistant, surgical technician, controlled suction devices, and bipolar cautery are all ergonomically positioned to maintain economy of motion and ease of access. The sylvian fissure is opened 3 cm distal to the anterior lip of the inferior frontal surface. Initially, this is a superficial exposure and non-bayoneted, short, straight instruments are utilized. The venous drainage typically runs parallel to the sylvian fissure and is associated with the temporal surface. Small cortical veins within the sylvian fissure can be sacrificed without great concern; however, their preservation is always ideal. The superficial and thickened arachnoid over the sylvian fissure is opened from the point of entry anteriorly, providing wide exposure. Subsequently, an M3 segment of the MCA is identified and followed proximally, using sharp dissection. Blood is removed from the sylvian fissure during dissection and the thin wisps of arachnoid can be identified and cut sharply. Commonly, the fissure will open with gentle retraction on both the frontal and temporal lobes. This is not the case when there is a thick dense hematoma that r equires sharp dissection and suctioning. Dissection is taken along the M2 segment of the MCA,
Transsylvian Approach
Saccular narrow-necked aneurysms are the least common form found in this region (Figure 3). An aneurysm clip is selected after the anatomy is completely displayed. Ideally, we try to place a clip parallel to the M1 and M2 segments; however, on occasion a clip can be placed perpendicular to the proximal and distal vessels, as demonstrated. Lesions of this size rarely require temporary occlusion for placement of the permanent clip. After the aneurysm has been obliterated, it is punctured with a 26-gauge spinal needle to ensure occlusion. Broad-based aneurysms, which include in their neck either a portion of the proximal M1 or distal M2 vessel wall, are more commonly seen in this location (Figure 4). In addition, a trifurcation instead of bifurcation is a common anatomical variant. This should be visualized on the preoperative angiogram; however, vessel spasm may diminish the surgeons ability to visualize all vessels preoperatively. Thus, complete dissection of the neck and fundus of the aneurysm as well as the surrounding vessels in this region is essential to avoid postoperative complications. One must be prepared for temporary occlusion at all times but particularly under conditions of untimely aneurysm rupture or for complete dissection of the aneurysm. In many cases, a proximal temporary clip on the M1 segment may be all that is needed to soften the neck of the aneurysm and facilitate clip placement. However, as demonstrated in Figure 4, temporary trapping may be needed for sharp dissection of the aneurysm fundus from a distal branch. There are several guidelines to follow during temporary clip application. The proximal clip should be placed first and none of the clips should limit the surgeons view, the preservation of flow within the lenticulostriate vessels should be maximized, and finally the clips should not be placed in a
CLIP APPLICATION
STIEG AND FRIEDLANDER : SURGICAL MANAGEMENT OF MIDDLE CEREBRAL ARTERY ANEURYSMS M1 segment inferior trunk of the M2 segment
19
Figure 3. A, high-magnification view of the MCA bifurcation demonstrating the M1 and M2 segments with lenticulostriate vessels on the M1 segment. A narrow-necked aneurysm between the superior and inferior M2 trunks is demonstrated. B, a permanent clip has been placed across the neck of the aneurysm perpendicular to the axis of the parent vessel. In addition, the fundus of the aneurysm has been punctured and aspirated to ensure complete occlusion and removal of any mass effect. M1 segment
A B
Figure 4. high-magnification views of the MCA trifurcation. A, the association between the M1, three M2 trunks, and a broad-based aneurysm including a segment of the inferior M2 trunk in its neck is demonstrated. B, temporary clips have been placed across the M1 and M2 segments. The clips are placed to allow maximum exposure for the surgeon and maintain flow through the lenticulostriate vessels. Sharp dissection techniques are utilized to create a plane between the fundus of the aneurysm and the parent vessel. Preservation of all distal arteries is essential. C, clip application is parallel to the parent vessels and occludes the portion of the aneurysm neck that extends into the M2 trunk. All three distal M2 branches have been preserved and the aneurysm is punctured to ensure complete occlusion. Larger aneurysms with hematoma may be opened widely for evacuation of the mass. The fundus of the aneurysm is left intact. The distal temporary clips are removed first and the proximal M1 temporary clip is then removed.
20
region of atheroma which could result in an embolic complication. In addition to facilitating dissection between the aneurysm fundus and the parent vessels, temporary trapping allows the surgeon to perform an aneurysmorrhaphy. Indications for opening the aneurysm widely include a large thrombus within the aneurysm or calcium in the fundus, both of which limit closure of the clip. The thrombus can be removed with an ultrasonic aspirator, micropituitaries, or sharp dissection. Calcified plaque in the wall of the aneurysm can be either surgically removed or crushed with a hemostat, thereby enabling the clip to approximate the walls of the aneurysm. The surgeon must remember to open the aneurysm distal enough on the fundus to allow closure of the aneurysm with a clip. Moreover, dissecting the hematoma or atheroma too proximally may result in the creation of an intimal flap in the parent vessel. Should an intimal flap occur, dissection and occlusion may result once flow is restored in the parent vessels. Cerebroprotective maneuvers are used in order to minimize the effects of induced cerebral ischemia during temporary occlusion. The mechanisms of action for the protective measures are poorly understood but focus on decreasing cerebral oxygen requirements (general anesthesia, cooling to 33C and placing the patient in burst suppression with etomidate or barbiturates), increasing collateral flow (elevating systolic blood pressure 20% to 30%) and fr ee radical scavenging (0.5 gm mannitol/kg body weight). Intermittent 10-minute periods of temporary occlusion with reconstitution of flow between the periods can be used in specific circumstances when the aneurysm has not been opened. Five minutes of temporary occlusion is usually well tolerated. Beyond 15 minutes, the risk of ischemic injury rises sharply. Hypothermic circulatory arrest is considered only in the most extreme cases. Multiple clip configurations have been designed to enable the surgeon to apply them parallel to the parent vessels. The clips should also be placed to minimize torque on the M1 and M2 segments, which could result in vessel stenosis and ischemia. Fenestrated clips including one of the M2 branches have also been used. After the aneurysm has been secured, we recommend puncture with a 26-gauge spinal needle to rule out residual filling. Under conditions of temporary trapping, aneurysmorrhaphy, and clipping, the distal M2 temporary clips are removed first. At that time, the surgeon may note bleeding from the aneurysm which would require placement of a tandem clip or adjustment of the primary permanent clip. After removal of the distal temporary clips, the proximal temporary clip is removed. Fusiform aneurysms of the MCA bifurcation also present significant challenges (Figure 5). These aneurysms, which include a portion of the parent ves-
sel, cannot be excluded from the circulation by straightforward clip applications. The aneurysm may include the entire trifurcation, as demonstrated, or may involve one of the M2 branches. In either situation, reconstitution of flow in the distal M2 vessels is often needed and can be provided by bypass procedures. The surgeon must anticipate the need for bypass on the basis of the preoperative angiogram. Prior to making the skin incision, the superficial temporal artery (STA) is mapped out on the scalp using Doppler ultrasound. Infiltration of the scalp with lidocaine is avoided. The STA and its adventitia are isolated along its full length for use later in the case. As demonstrated, the aneurysm has been trapped by permanent clips on the M1 and M2 segments and subsequently opened for debulking of the hematoma within the aneurysm. In addition, an STA-to-MCA bypass in an end-to-end fashion is performed. One or two branches of the STA are isolated for use in the procedure. A proximal temporary clip is placed on this vessel and the distal ends are transected at the appropriate length. The adventitia is dissected away from the distal end. The vessel is irrigated with heparinized saline and anastomosed to the M2 branch with running 10-0 Prolene sutures. A suture is placed at each pole of the anastomoses with one run along the frontal surface and the other along the temporal surface of the anastomosis. A rubber dam is placed under the recipient vessels to facilitate visualization of the suture. Again, the distal (M2) temporary clips are removed first to assess the anastomosis; finally, the temporary clip on the STA is removed. Intraoperative angiography is used to evaluate the bypass before closure of the craniotomy. Care must be taken to closely oppose the galea during the closure to avoid leakage of CSF from the wound, as the dura cannot be closed tightly in this situation. Also, adequate bone removal will ensure patency of the STA as it enters the sylvian fissure. When only one M2 branch is involved in a fusiform aneurysm, an end-to-side anastomosis between the involved M2 segment and the normal M2 segment, respectively, is performed. Cerebral protection is used during performance of the bypass procedure.
We will discuss general and location-specific management issues as well as strategies to avoid complications in patients with MCA aneurysms.
Preoperative Management
For All Patients Adequate angiograms need to be obtained and carefully studied prior to surgery in order to thoroughly understand the variable MCA anatomy for the individual patient. A general medical evaluation needs to
21
M1 segment
A1 segment
STA
Figure 5. A, schematic of a fusiform aneurysm involving a distal portion of the M1 segment as well as proximal portions of both M2 trunks. Frequently, the lenticulostriate vessels are adherent to the surface of large aneurysms in the region and must be meticulously dissected free prior to placing the permanent clips as demonstrated. The clips are placed as close to the aneurysm as possible. B, demonstrations of permanent clip application across the M1 segment and two distal M2 segments, placed as close to the aneurysm as possible, can maintain flow to the lenticulostriates. In addition, an end-to-end STA-to-MCA anastomosis is demonstrated. The aneurysm is opened and the hematoma evacuated as needed. Branches of the STA are anastomosed to the M2 trunks in an end-to-end fashion to maintain distal flow.
22
be performed to optimize the preoperative condition of the patient. For Patients with Ruptured Aneurysm Patients with ruptured MCA aneurysms may present in extremis, secondary to a large intraparenchymal hematoma. Because of the urgency in such situations, conventional angiography may not be performed. Computed tomography (CT) angiography provides invaluable information prior to proceeding with aneurysm clipping following hematoma evacuation. Avoidance of aneurysmal rehemorrhage can be achieved by early and aggressive blood pressure control as well as proceeding with early surgery. Seizures can be avoided by using antiepileptic medications at the time of admission. When the patient presents with early symptomatic hydrocephalus, we prefer, if possible, to delay placement of a ventricular drain until surgery because of the associated increased risk of rehemorrhage. However, if the patient requires immediate CSF drainage, placement of a ventriculostomy with gradual lowering of the intracranial pressure is the goal.
formed. In order to avoid a stroke following temporary vessel occlusion, ischemic time must be kept at a minimum. Cerebroprotective aids employed to avoid ischemic complications after temporary occlusion include moderate hypothermia (33-34C), induced hypertension (160-170 mm Hg), mannitol (0.5 gm/kg body weight), and etomidate or barbiturateinduced burst suppression.
Postoperative Management
If subacute neurological deterioration (several hours postoperatively) occurs and is not related to seizures, hemorrhage, or persistent brain edema, one should consider an M2 occlusion or stenosis due to clip rotation. This has been described after closure or postoperative swelling of the sylvian fissure. An urgent CT scan will rule out a hemorrhagic complication and expeditious angiography will confirm vessel patency. Delayed hydrocephalus and vasospasm should be treated in the standard fashion.
Intraoperative rupture can be prevented by avoiding hypertension as well as careful microsurgical technique. A specific challenge of MCA bifurcation aneurysms is the inadvertent occlusion of an M2 or M3 branch during clipping of the aneurysm. This complication can be avoided by thorough preoperative evaluation of the angiogram and circumferential evaluation of the aneurysm neck either during temporary occlusion or following clipping. If one cannot confirm either complete obliteration of the aneurysm or patency of all the distal branches, intraoperative or early postoperative angiography should be per-
Intraoperative Management
MCA aneurysms are common and often very challenging for the microvascular neurosurgeon. Surgical results have improved immeasurably over the past several decades with the advancements in microsurgical technique, cerebral protection, and the aggressive management of vasospasm. The surgeon must be prepared for using any of the three described approaches for aneurysms in this region, as well as the need for possible bypass procedures. We feel most comfortable with the transsylvian approach; however, occasionally we open the proximal sylvian fissure. This latter technique is utilized when we have concern regarding the stability of the aneurysm fundus, projection of the aneurysm, or the complexity of the aneurysm neck. A methodical approach to these aneurysms with a thorough understanding of the anatomy, as well as patient, meticulous dissection techniques, should result in a good surgical outcome.
CONCLUSIONS
Our understanding of the anatomy and pathophysiology of dural arteriovenous malformations (DAVMs) has evolved rapidly with advancements in the field of interventional neuroradiology. DAVMs arise from abnormal arteriovenous shunts, yet venous hypertension from arterialized veins is responsible for their clinical presentation. The arterialized veins dilate, elongate, and rupture to cause subarachnoid or intraparenchymal hemorrhage. Ischemia may develop in the surrounding brain from stagnant venous drainage and poor perfusion. Communicating hydrocephalus occurs because venous hypertension impedes cerebrospinal fluid (CSF) absorption, and obstructive hydrocephalus occurs when large venous aneurysms obstruct CSF pathways. Many intracranial and spinal DAVMs can be cured by ligating the draining vein or occluding an associated dural venous sinus. Unlike parenchymal arteriovenous malformations (AVMs), the arterial supply does not have to be eliminated before the venous drainage because the leaves of the dura protect the nidus from bleeding. Venous occlusion is safe when the primary drainage is antegrade through a dural venous sinus or when a single draining vein
INTRODUCTION
can be identified as it exits the dura. When a DAVM is not associated with a venous sinus and cortical venous drainage is present, ligating multiple draining veins may be catastrophic. Ligation may worsen the venous hypertension, produce a venous infarction, or cause hemorrhage by diverting flow into nonarterialized veins. Tentorial and posterior fossa DAVMs are a unique subset of intracranial DAVMs that do not have an associated venous sinus and frequently have multiple draining veins that are tortuous and aneurysmal. The common presentation is subarachnoid hemorrhage (SAH) caused by a ruptured venous aneurysm. Therefore, treatment is directed to the arterial side, an approach that is similar to that for parenchymal AVMs. The therapeutic strategy for deep-seated DAVMs includes transarterial embolization followed by either stereotactic radiation or surgical excision. Transarterial embolization is performed to reduce the blood flow to the DAVM and decrease the venous hypertension. Because some dural arteries are too small to cannulate, endovascular techniques are rarely curative. Instead, embolization is followed by stereotactic radiation or surgical excision of the nidus. DAVMs with no cortical venous drainage can be partially treated; however, tentorial and posterior fossa DAVMs require complete obliteration to prevent recurrent hemorrhage. In the past, microsurgery for deep-seated DAVMs was associated with high morbidity from postopera-
23
24
tive hemorrhage if draining veins were occluded prematurely and residual DAVM was present. Limited exposure with conventional craniotomies and lar ge venous aneurysms accompanying the dural nidus also discouraged the use of microsurgery. Combining skull base approaches with conventional craniotomies has been an important addition in the management of patients with deep-seated DAVMs. The skull base approach expands the field of view, widens the angle of exposure, improves maneuverability that is needed to coagulate, clip, and excise the dural leaflet, and minimizes retraction of the brain. Four approaches are illustrated as the applicable technique for the surgical management of tentorial and posterior fossa DAVMs.
Patients with tentorial and posterior fossa DAVMs are commonly men between 50 and 70 years of age who have suffered an SAH. Less frequently, they suffer progressive neurological deterioration or symptoms related to hydrocephalus. Because DAVMs with cortical venous drainage have an aggressive natural history, the goal of treatment is complete angiographic obliteration. If patients improve neurologically after embolization and do not have a venous aneurysm, stereotactic radiation is a reasonable option. Direct surgical excision of the dural nidus is indicated if the ruptured venous aneurysm is not eliminated by embolization or if the feeding arteries are too small to embolize. Surgery is also recommended if the symptoms fail to resolve or new symptoms arise after embolization. Some authors recommend open craniotomy with direct transvenous embolization or ligation of the draining veins. This approach is safe in a small subset of patients who have had transarterial embolization and a single fistula (e.g., petrous apex DAVMs).
PATIENT SELECTION
ruptured saccular aneurysm. Selection of the external carotid arteries is important because a majority of the arterial supply may arise from branches of the middle meningeal artery, superficial temporal artery, and occipital artery. To identify venous aneurysms and to determine the extent of venous congestion and stagnant flow, the late venous phase should be imaged as well. Bone windows on the angiogram, which establish the location of the nidus in relation to the skull base, are particularly helpful to determine if a skull base exposure is required in addition to a conventional craniotomy. After the imaging studies are complete, treatment begins with transarterial embolization of the feeding arteries through the external carotid artery. Large branches from the meningohypophyseal trunk may also be embolized. Decreasing arterial flow reduces the vascularity to the DAVM and the effect of venous hypertension. Occluded vessels that have been embolized preoperatively also provide a good road map intraoperatively for understanding the surgical anatomy. Intraoperative angiography with imaging of the internal and external carotid arteries and the vertebrobasilar system documents obliteration of the DAVM. Postoperative angiography provides better resolution, selective arterial injections can be safely performed, and a variety of projections can be performed to expand the findings of intraoperative angiography.
The first imaging study obtained is usually computed tomography (CT), which frequently shows SAH or obstructive hydrocephalus from a large dilated venous aneurysm. Magnetic resonance (MR) imaging and MR angiography are useful for identifying the source of hemorrhage (usually a ruptured venous aneurysm), determining the degree of venous hypertension, establishing the ventricular size, and planning the operative approach based on the relationship of the nidus to the surrounding brain structures. MR imaging and MR angiography also serve as screening studies for DAVM thrombosis after stereotactic radiation. The findings of the initial cerebral angiogram may be negative if only the internal carotid arteries and vertebrobasilar circulation are injected in search of a
SURGICAL PLANNING
The best approach to deep-seated DAVMs includes a scalp incision, craniotomy, and dural opening that removes a significant portion of arterial supply from the external carotid artery. Removal of the petrous bone eliminates the transosseous supply and exposes dura around the entire DAVM nidus. The dura around the nidus is resected to eliminate the remaining dural supply. After removal of the dural component, the remaining arterial supply from the internal carotid artery is coagulated and incised. The following are descriptions of the surgical approaches that provide access to most tentorial and posterior fossa DAVMs, including those located at the petrous apex, the posterolateral tentorium, the falx-tentorial junction, the inferior petrosal sinus, the lower clivus, and the foramen magnum.
SURGICAL APPROACHES
At the petrous apex, the most common surgical finding is the tentorial artery from the meningohypophyseal trunk feeding a tentorial nidus with a single draining petrosal, mesencephalic, or pontine vein (Figure 1A). A subtemporal craniotomy with zygomatic osteotomy or suboccipital craniotomy is performed to provide access to the fistula. An anterior
LEWIS AND TEW : SURGICAL REMOVAL OF TENTORIAL AND POSTERIOR FOSSA DAVMS
25
petrosectomy in combination with a subtemporal approach is necessary when there is transosseous arterial supply through the petrous bone or when the nidus is located along the upper one half of the clivus (Figure 1B). An anterior petrosectomy requires removal of bone between the petrous portion of the internal carotid artery, the third division of the trigeminal nerve, and the internal auditory canal. This approach provides access to the upper half of the clivus and anterior aspect of the posterior fossa down to the inferior petrosal sinus. Exposure below the internal auditory canal requires a posterior petrosectomy and suboccipital approach. After induction of general endotracheal anesthesia, a lumbar catheter placed to drain CSF improves exposure and prevents postoperative CSF leakage. The patients head is placed in a radiolucent Mayfield headrest to permit intraoperative angiography. Positioning for a subtemporal approach is supine with a gelatin roll under the right shoulder, and the thorax is elevated 15 degrees. The head is rotated so that the sagittal sinus is parallel to the floor and then tilted downward 10 degrees so that the zygoma is the highest point in the surgical field (Figure 1C). This position brings the skull base into view and diminishes the need for retraction of the temporal lobe. A frontotemporal, or inverted question-mark, incision provides exposure from the cavernous sinus to the arcuate eminence of the petrous bone. The scalp is reflected inferiorly with the frontalis fat pad to avoid injury to the facial nerve. A zygomatic osteotomy may be performed to permit reflection of the temporalis muscle below the floor of the middle fossa and minimize temporal lobe retraction. The zygoma remains attached to the masseter muscle to improve healing after closure. The temporalis muscle and fascia are reflected inferiorly, leaving a superior cuff of fascia for closure. A high-speed drill is used to perform the subtemporal craniotomy and additional bone is removed with a burr to make the craniotomy flush with the floor of the middle fossa. When an anterior petrosectomy is required, the microscope is used to dissect the temporal fossa dura away from the petrous bone. Removal of approximately 30 ml of CSF aids in retraction of the temporal fossa dura. The middle meningeal artery is sacrificed at the foramen spinosum. Removal of bone around the foramen ovale allows mobilization of the trigeminal root and improved retraction of the temporal fossa dura medially. Selfretaining retractors are placed at the edge of the petrous ridge before drilling the petrous bone. The greater superficial petrosal nerve, identified with a facial nerve stimulator, is sectioned near the facial hiatus to prevent injury to the geniculate ganglion. The petrous carotid artery, which lies under the greater superficial petrosal nerve, is exposed
using a high-speed diamond-tipped drill. The posterior loop of the internal carotid artery is not exposed, thus avoiding injury to the geniculate ganglion and cochlea. The superior semicircular canal, which provides the posterior border of the anterior petrosectomy, is located beneath the hard white bone of the arcuate eminence at a 120-degree angle to the greater superficial petrosal nerve (Figure 1D). Drilling begins anterior to the arcuate eminence in the soft bone of the meatal plane to expose the dural sleeve of the internal auditory canal (Figure 1E). To avoid injury to hearing or the facial nerve, the geniculate ganglion, falciform crest, and cochlea are not exposed. The remaining bone between the third division of the trigeminal nerve, petrous carotid artery, and internal auditory canal (i.e., Kawases triangle) is removed inferiorly along the posterior fossa dura to the inferior petrosal sinus (Figure 1F). The dura is opened along the inferior temporal lobe and the tentorium is split from the superior petrosal sinus toward a point posterior to the trochlear nerve. Clips are placed on the superior petrosal sinus and the sinus is divided. Frequently, the sinus is thrombosed. The dural incision is then extended inferolateral to the inferior petrosal sinus to expose the upper posterior fossa (Figure 1G). After opening the dura, numerous feeding arteries in the tentorium are usually visualized and coagulated using bipolar electrocautery. The tentorium is incised with scissors and coagulated using bipolar electrocautery. The electrocautery causes the tentorial dura to shrink and retract, thus exposing the tentorial artery. The tentorial artery from the meningohypophyseal trunk can be seen communicating with arterialized vein(s). Occlusion of the feeding artery leads to collapse of the venous aneurysm. Similar to ligating spinal DAVMs, the veins and venous aneurysms of intracranial DAVMs need not be removed. Intraoperative angiography is performed to document complete obliteration of the petrous apex DAVM. To avoid postoperative CSF leakage, exposed mastoid air cells are closed with bone wax. Pericranium and fibrin glue (i.e., combination of thrombin and cryoprecipitate) are placed along the defect in the temporal lobe dura, and a pericranial fat graft is packed. A lumbar catheter is connected to a closed drainage system for 24 to 72 hours, with the level set at the external auditory canal. Drainage of CSF should not exceed 10 mL per hour to avoid pneumocephalus.
Posterolateral tentorial DAVMs are usually the largest and most complex deep-seated DAVMs (Figure 2A). The arterial supply is generally bilateral with contributions from the anterior circulation, posterior circulation, and external carotid system to form a plexi-
26
form nidus with multiple draining veins. A subtemporal craniotomy with removal of the posterior petrous bone and mastoid minimizes retraction of the temporal lobe, removes the transosseous arterial supply, enables the surgeon to excise a large area of the tentorium, and provides access to draining veins and venous aneurysms below the tentorium (Figure 2B). The addition of a posterior petrosectomy provides a presigmoid approach with sectioning of the superior petrosal sinus to mobilize the junction of the transverse-sigmoid sinus and protect the vein of Labb when the temporal lobe is retracted. This presigmoid approach gives access to the upper two thirds of the clivus, anterior cerebellum, and brainstem. The fifth through the 10th cranial nerves can be visualized. The addition of a suboccipital craniotomy provides a retrosigmoid approach with access to the lower clivus and foramen magnum. Sectioning of the transverse-sigmoid sinus and a labyrinthectomy is not r equired for most posterolateral DAVMs. Further exposure may be obtained with sacrifice of conduction hearing by total petrosectomy and sealing the semicircular canals. In our experience, occlusion of the dominant transverse-sigmoid sinus worsens venous hypertension and may cause a hemorrhagic venous infarction or postoperative hydrocephalus. A lumbar catheter inserted to drain CSF will improve exposure and prevent postoperative CSF leakage. The patients head is placed in a radiolucent
Mayfield headrest. The femoral arterial catheter used for intraoperative angiography is placed prior to positioning the patient in the right lateral decubitus position. An axillary roll is used and the thorax is elevated 15 degrees. The head is rotated until the sagittal sinus is parallel to the floor and is tilted 10 degrees downward so that the zygoma is the highest point in the surgical field (Figure 2C). To gain additional exposure of the petrous bone, mastoid tip, and deep venous structures, a J-shaped incision extends vertically from the root of the zygoma crosses the superior temporal line, and descends 3 cm posterior to the ear. The incision terminates at the insertion of the neck musculature to the skull. The temporal skin flap is reflected inferiorly and the temporalis muscle is reflected separately, leaving a superior fascial cuff. Burr holes are made at the anterior mastoid point and asterion, which are above and below the transverse-sigmoid junction, respectively. Bone overlying the sinus between the burr holes is removed with a drill. A pneumatic craniotome is used to perform the temporal craniotomy (Figure 2D). Under the microscope, the bone overlying the transverse sinus is removed with a high-speed drill. The mastoid air cells are removed. The middle fossa and meatal bone plates are dissected until the antrum is identified (Figure 2E). The bone of the sinodural angle is removed with a diamond-tipped drill to expose the
Figure 1. Thrombosis of the superior petrosal sinus is frequently seen in petrous apex dural arteriovenous malformations. A, schematic drawing of the petrous apex DAVM with arterial supply from the tentorial artery and drainage into a petrosal vein with an associated bilobed venous aneurysm. The vein drains into the contralateral superior petrosal sinus. B, hatched area represents boundaries of subtemporal craniotomy (1) and anterior petrosectomy (2). The area in red represents the location of the DAVM nidus. C, positioning for the subtemporal approach. An inverted question-mark incision (hatched line) is outlined over the temporal squama. Solid lines represent bony removal including the suboccipital craniotomy and zygomatic osteotomy. D, the temporal lobe dura is retracted to visualize the floor of the middle cranial fossa and petrous bone. Kawases triangle is bounded by the mandibular division of the trigeminal nerve, petrous internal carotid artery, and the seventh and eighth cranial nerve complex. Removal of the bone within this anatomical triangle represents the anterior petrosectomy. GSPN = greater superficial petrosal nerve. E, under the microscope, the bone overlying the petrous internal carotid artery (ICA) and internal auditory canal (IAC) is removed to define the borders of the anterior petrosectomy. The cortical bone of the superior semicircular canal forms a 50-degree angle with the IAC. F, after removal of the bone in the meatal plane, a line drawn from the tip of transverse crest to the intersection of the carotid artery with the trigeminal nerve (Millers line) approximates the basal turn of the cochlea. G, the dural reflection of the temporal lobe, posterior fossa, and tentorium forms the superior petrosal sinus. After the dura is opened, the contents of the upper posterior fossa are exposed. (Figure A reprinted with permission from the Mayfield Clinic; B reprinted from Tew JM Jr, van Loveren HR, Keller JT: Atlas of Opera tive Microneurosurgery, Vol II. Philadelphia: WB Saunders (in press); C-G reprinted from Tew JM Jr, van Loveren HR: Atlas of Operative Microneurosurgery, Vol I. Philadelphia: WB Saunders, 1994, pp 40-43)
LEWIS AND TEW : SURGICAL REMOVAL OF TENTORIAL AND POSTERIOR FOSSA DAVMS
27
A
GSPN
ICA
IAC
Kawases triangle
G
CN V CN VIII
CN IV
Figure 1.
28
anterior border of the sigmoid sinus and the posterior semicircular canal. The mastoid air cells of the mastoid tip are then removed to expose the posterior fossa dura and digastric ridge, which serves as a landmark to the stylomastoid foramen, the beginning of the fallopian canal, and the location of the facial nerve (Figure 2F). Thirty milliliters of CSF are removed and a dural incision is made from the posterior fossa (inferior to the superior petrosal sinus), to the middle fossa, and up to the petrous apex. The superior petrosal sinus is divided with clips. Under the microscope, the tentorium is divided toward the incisura directed posterior to the trochlear nerve (Figure 2G). The leaves of the tentorium are coagulated with bipolar electrocautery to obtain hemostasis and to shrink and retract the dura. Feeding arteries entering the tentorium are coagulated with bipolar electrocautery and cut with microscissors. Frameless stereotaxis may be useful to guide the approach and corroborate location of the DAVM in the tentorium. After excising the dura leaflet, the draining veins are coagulated and divided. In addition to documenting obliteration of the DAVM, intraoperative angiography reveals residual DAVMs, which cannot be visualized because they are concealed by cavernous and tentorial dura. Placing a metal clip at the operative site provides information on the intraoperative angiogram regarding the relative position of the remaining nidus to the skull base. A free abdominal fat graft used to close the defect in the mastoid and temporalis muscle is split to cover the mastoid defect. The bone flap and zygo-
matic bone are secured with titanium plates. The temporalis and cervical muscles are reattached to their respective fascial cuffs. The lumbar catheter is connected to a closed drainage system for 24 to 72 hours with the level set at the external auditory canal. CSF drainage should not exceed 10 mL per hour. A compression dressing is applied.
DAVMs of the inferior petrosal sinus are often supplied by the ascending pharyngeal artery, occipital artery, posterior meningeal artery, and muscular branches of the vertebral artery (Figure 3). The posterior auricular artery and meningohypophyseal trunk may contribute to the arterial supply. The arteriovenous connection commonly occurs at the end of the inferior petrosal sinus near the jugular bulb. Venous drainage is antegrade through the internal jugular vein or spinal medullary veins and may also occur retrograde through cortical veins into the transverse-sigmoid or cavernous sinuses (Figure 4A and B). Lower clivus and foramen magnum DAVMs gain supply from the posterior meningeal, meningojugular, occipital, and anterior spinal arteries, as well as the muscular branches of the vertebral artery. Drainage is into either the spinal medullary veins or the mesencephalic veins that drain toward the vein of Galen and straight sinus. A right far lateral suboccipital craniotomy is required to isolate the transverse-sigmoid sinus and jugular bulb, visualize the ventral brainstem
Figure 2. DAVM of the posterolateral tentorium. A, arterial supply from the intracavernous carotid artery branches and external carotid artery supply. Venous drainage flows into mesencephalic and cerebellar veins. B, hatched area represents boundaries of the subtemporal, suboccipital, and posterior petrosectomy. The dotted area represents sigmoid and transverse sinuses exposed but not sacrificed. The area in red represents the location of the DAVM nidus. C, positioning for the posterior petrosectomy. D, the temporalis muscle is reflected anteroinferiorly, leaving a superior fascial cuff. Burr holes placed on either side of the transverse sinus reduce the risk of injury to the sinus. The bone between the burr holes is removed with a burr. A pneumatic craniotome is used to per form the craniotomy. E, the spine of Henle is a landmark to the antrum. The antrum floor is the cortical bone of the lateral semicircular canal, which serves as a guide to deeper structures. The mastoid air cells are removed, and the middle fossa and meatal bone plates are dissected until the antrum is identified. F, the bone of the sinodural angle is removed with a diamond-tipped drill. The sigmoid sinus and posterior semicircular canal are exposed. The posterior semicircular canal is identified by following the lateral semicircular canal until it bisects the posterior semicircular canal. The mastoid air cells of the mastoid tip are then removed to expose the digastric ridge, which serves as a landmark to the stylomastoid foramen and the beginning of the fallopian canal. Removal of the sigmoid plate exposes the superior petrosal sinus. G, the superior petrosal sinus is divided and the tentorium is incised ventrally toward the incisura in a course toward the trochlear nerve. The feeding arteries and dural nidus are encountered along the path to the incisura. (Figure A reprinted with permission from the Mayfield Clinic; B reprinted from Tew JM Jr, van Loveren HR, Keller JT: Atlas of Operative Microneurosurgery, Vol II. Philadelphia: WB Saunders (in press); C-G reprinted from Tew JM Jr, van Loveren HR: Atlas of Operative Microneurosurgery, Vol I. Philadelphia: WB Saunders, 1994, pp 40, 45, 47-49, 51)
LEWIS AND TEW : SURGICAL REMOVAL OF TENTORIAL AND POSTERIOR FOSSA DAVMS
29
C A D
spine of Henle
transversesigmoid junction
Figure 2.
30
Figure 3. Thrombosis or stenosis of a major sinus may cause retrograde flow into cortical veins. A, right anterior posterior vertebral artery angiogram demonstrating a DAVM of the inferior petrosal sinus with stenosis of the right internal jugular vein and venous drainage through the contralateral transverse-sigmoid sinus into the left internal jugular vein. B, lateral vertebral angiogram demonstrating retrograde venous drainage into the superior sagittal sinus and vein of Galen via the straight sinus. (Reprinted with permission from Lewis AI, Rosenblatt SS, Tew JM Jr: Surgical management of deep-seated dural arteriovenous malformations. J Neurosurg 87:198-206, 1997)
Figure 4. DAVM of the inferior petrosal sinus. A, the inferior petrosal sinus DAVM is fed by both posterior meningeal arteries and muscular branches from both vertebral arteries. The occipital artery is causing the constriction of the right internal jugular vein. The jugular bulb is dilated and arrows within the lumen of the dural sinuses show that venous drainage is retrograde. B, hatched area represents boundaries of the far lateral suboccipital approach. The area in red represents the location of the DAVM nidus. C, a lateral hemisphere incision is outlined over the suboccipital bone and cervical spine. The hatched line represents the scalp incision. The solid line represents the bony removal including the suboccipital craniotomy, opening the foramen magnum, laminectomy of the posterior arch of C1, and partial resection of the posteromedial condyle. D, mastoid emissary veins and the epidural venous plexus may be arterialized. Hemostasis of these veins is best achieved with bipolar electrocautery or compression with Oxycel or other hemostatic agent. E, the posterolateral one third of the occipital condyle is removed with a high-speed diamond-tipped drill. F, after removal of the posterior arch of C1, the extradural horizontal segment of the vertebral artery is mobilized out of the vertebral groove on C1 to obtain proximal control of the primary arterial supply to the DAVM. G, the posterior fossa and spinal dura are reflected laterally with dural retention sutures. An incision begins in the spinal arachnoid and proceeds through the cisterna magna while cerebrospinal fluid is aspirated. (Figure A reprinted with permission from the Mayfield Clinic; B, D-F reprinted from Tew JM Jr, van Loveren HR, Keller JT: Atlas of Operative Microneurosurgery, Vol II. Philadelphia: WB Saunders (in press); C and G reprinted from Tew JM Jr, van Loveren HR: Atlas of Operative Microneurosurgery, Vol I. Philadelphia: WB Saunders, 1994, pp 69 and 72)
LEWIS AND TEW : SURGICAL REMOVAL OF TENTORIAL AND POSTERIOR FOSSA DAVMS
31
C D
occipital condyle
vertebral artery
G F
cisterna magna foramen transversarium vertebral artery mobilized
C2 nerve root
Figure 4.
32
and lower clivus, and resect the involved dural segment. The far lateral exposure isolates the vertebral artery, which is the primary arterial supply to most posterior fossa DAVMs. This exposure provides a caudal-to-rostral and lateral-to-medial field of view for DAVMs of the inferior petrosal sinus and lower clivus. After induction of general endotracheal anesthesia, the patients head is placed in a radiolucent Mayfield headrest for intraoperative angiography. A lumbar catheter is optional. The femoral arterial sheath for intraoperative angiography is inserted before the patient is placed in the lateral oblique position. An axillary roll is placed and the thorax is elevated 15 degrees. The head is rotated 45 degrees from parallel toward the floor and flexed until the posterior neck muscles are stretched but not tight. The mastoid process is the highest point. A midline suboccipital incision is performed from the spinous process of C2 toward the superior temporal line and inferior to the mastoid tip (Figure 4C). This incision allows dissection of the extracranial vertebral artery and exposes the suboccipital bone, mastoid region, and deep venous structures. The skin flap is reflected inferiorly, leaving a superior fascial cuff. A cuff of nuchal fascia and splenius capitis muscle is preserved to reapproximate the cervical musculature during closure. The suboccipital bone is removed in one piece with the fascial cuffs attached using a pneumatic drill. Additional bone is removed in the mastoid region with a burr. A subperiosteal dissection of the lamina of C1 and C2 is performed 2 cm lateral to the midline. Sharp dissection of the soft tissues between the foramen magnum and the arch of C1 identifies the vertebral artery as it pierces the dura. Dissection continues from medial to lateral until the foramen transversarium is identified. The surrounding venous plexus and vertebral muscular branches are coagulated and divided using bipolar electrocautery (Figure 4D). The arch of C1 is removed with a rongeur or drill. The lateral portion of the foramen magnum and the posteromedial third of the occipital condyle are removed with a diamondtipped drill (Figure 4E). An occipitocervical fusion is not necessary if only one third of the condyle has been removed. Bone overlying the foramen transversarium is removed with a diamond-tipped drill to mobilize the vertebral artery (Figure 4F). The dura is opened in the midline at C1 and extended laterally up to the transverse-sigmoid sinus (Figure 4G). The intracranial vertebral artery is identified and followed to the nidus, which is anteromedial to the jugular bulb. The feeding branches to the nidus are coagulated and the dural leaflet is excised. Bovine pericardium, fascia lata, or other dural substitutes are used to close the dural defect. The bone flap is replaced with plates and a standard closure
The arterial supply of midline DAVMs is often bilateral from the tentorial, middle meningeal, and superficial temporal arteries. Venous drainage is via cerebellar and vermian veins. The straight sinus is frequently thrombosed and a vein of Galen aneurysm is present. Venous aneurysms are very common and, in most cases, are responsible for hemorrhage or obstructive hydrocephalus (Figure 5). Because DAVMs of the vein of Galen and falx-tentorial junction are midline, an interhemispheric approach is used (Figure 6A and B). The trajectory is posterior to the splenium of the corpus callosum. There are four major steps in the operation. First, the scalp incision, craniotomy, and dural opening assist with the elimination of the arterial supply over the convexity. Second, the remaining dural arterial supply is eliminated by excising the falx and tentorium surrounding the nidus. Third, arterial branches from the meningohypophyseal trunk and posterior cerebral arteries are coagulated and cut as they enter the vein of Galen aneurysm. Division of the feeding arteries as they enter the DAVM is important to avoid occluding vessels of passage. Fourth, the thrombus within the venous aneurysm is removed to decompress the mesencephalon. The wall of the aneurysm remains in situ to avoid injury to the brainstem. A femoral arterial sheath is inserted before positioning the patient in a left lateral oblique position; for a right occipital approach, the head is placed in neutral position. The vertex becomes the highest point in the operative field (Figure 6C). The head is placed in a radiolucent Mayfield headrest to permit intraoperative angiography; the right hemisphere is dependent. A right posterior parasagittal craniotomy is performed for midline DAVMs (Figure 6D); bilateral approaches can be achieved by sectioning the falx and retracting the contralateral hemisphere. The skin flap is reflected toward the transverse sinus and the dura is opened in cruciate fashion (Figure 6E). The interhemispheric trajectory places the neurosurgeon at the falx-tentorial junction, behind the splenium of the corpus callosum and above the internal cerebral veins and the vein of Galen. Arterial supply from branches of the superficial temporal artery are coagulated during the opening and approach to the vein of Galen DAVM (Figure 6F). The tentorial and falx dura are coagulated and cut to eliminate the remaining meningeal arterial supply. The arachnoid is opened and the tentorial arteries from both meningohypophyseal trunks are coagulated and divided. The resected tentorium and falx dura overlying
LEWIS AND TEW : SURGICAL REMOVAL OF TENTORIAL AND POSTERIOR FOSSA DAVMS
33
Figure 5. MR imaging and MR angiography are performed to identify the source of hemorrhage, follow the ventricular size, identify thrombus within the venous aneurysm, and plan the operative approach based on the relationship of the nidus to the surrounding brain structures. A, T1-weighted axial MR image showing a partially thrombosed vein of Galen aneurysm with edema in the right subcortical nuclei due to venous hypertension. B, T1-weighted sagittal MR image showing the DAVM posterior and inferior to the splenium of the corpus callosum. An interhemispheric approach with a parieto-occipital trajectory avoided sectioning the corpus callosum. (Reprinted with permission from Lewis AI, Rosenblatt SS, Tew JM Jr: Surgical management of deep-seated dural arteriovenous malformations. J Neurosurg 87:198-206, 1997)
the vein of Galen aneurysm is removed with scissors. Thrombus within the aneurysm is removed with an ultrasonic aspirator to decompress the mesencephalon. The capsule is left in situ to prevent injury to the brain stem from excessive manipulation.
Surgical obliteration of a DAVM requires coagulation of the dura followed by excision of the dural leaflet to prevent the fistula from redirecting into another location or diverting flow into nonarterialized veins. Patients may develop transient, delayed neurological worsening. The presumed cause of the delayed postoperative deterioration is the exacerbation of preoperative venous hypertension. Patients may also require postoperative ventriculoperitoneal shunting for
communicating hydrocephalus after SAH or preoperative shunting for obstructive hydrocephalus caused by giant venous aneurysms. Ligating the draining veins prior to obliterating the arterial supply may lead to devastating hemorrhagic venous infarction. In addition, clipping or removing the venous aneurysm is unnecessary because it will collapse and thrombose when the DAVM nidus is obliterated. Thrombus within the venous aneurysm can be removed to decrease mass effect; however, the aneurysm wall should be left in situ to avoid injury to the surrounding brain (e.g., vein of Galen DAVMs). Intraoperative angiography is useful to identify a residual DAVM. However, postoperative angiography with selective injection of the external carotid artery and venous phase images should be performed to confirm complete DAVM obliteration.
34
B C
D
feeding arteries
Figure 6. Schematic drawing of the falx-tentorial DAVM with bilateral arterial supply from the meningohypophyseal trunk, superficial temporal artery, and superficial temporal artery. A, the venous drainage is into a partially thrombosed vein of Galen aneurysm and the right basal vein of Rosenthal. B, the area in red represents the location of the DAVM nidus along the tentorium and falx. C, the affected side is positioned down to allow the hemisphere to retract with the aid of gravity. This position is more comfortable than the sitting position because it allows the surgeon to use both hands side-by-side. Also, the risk of air embolism is significantly reduced compared with the sitting position. D, the site of the bone flap is shown. There are a paucity of bridging veins to the sagittal sinus in this location. E, a cruciate incision is performed to retract the dura toward the sagittal sinus. F, in the dependent position, minimal retraction of the occipital lobe is required to expose the falx-tentorial junction. Coagulation of the feeding arteries is performed as they enter the vein of Galen DAVM. (Figure A reprinted with permission from the Mayfield Clinic; Figures B-F reprinted with permission from Tew JM Jr, van Loveren HR, Keller JT: Atlas of Operative Microneurosurgery, Vol II. Philadelphia: WB Saunders (in press))
Approximately 20% of all arteriovenous malformations (AVMs) of the brain occur in the posterior fossa. They are more likely to become symptomatic than supratentorial AVMs because of the concentration of vital neurological structures within the limited confines of the posterior fossa. Infratentorial AVMs commonly present with hemorrhage, but may also present with progressive neurological deterioration or cranial neuropathy. Mixed intraparenchymal, subarachnoid, and intraventricular hemorrhage are the most common presenting findings of patients newly diagnosed with posterior fossa AVMs. The natural history of these lesions left untreated is poor because of the high incidence of hemorrhage and the neurological deterioration associated with AVMs of the posterior fossa. For this reason, most patients newly diagnosed with an AVM of the cerebellum or brainstem benefit from surgical treatment directed at eliminating the lesion and preventing future hemorrhage.
INTRODUCTION
Accurate classification of an AVM in the posterior fossa is necessary to guide preoperative planning and the optimal surgical treatment. Attempts at classifying cerebral hemispheric AVMs has met with limited acceptance, due in part to the variability of most
AVM CLASSIFICATION
supratentorial lesions and the complexities of the proposed classification systems. We classify AVMs of the posterior fossa into five groups relative to the involved brain tissue and blood supply. These five groups lend predictability to the anticipated vascular supply, venous drainage, and the ease or difficulty of surgical resection. We subdivide posterior fossa AVMs into lesions involving the following: 1) the cerebellar vermis, 2) the cerebellar hemisphere, 3) the cerebellar tonsil, 4) the pial surface of the brainstem, and 5) the deep parenchyma of the brainstem. This classification system accurately predicts the vascular supply to the lesion, aids in preoperative planning including the optimal surgical approach, and aids in determining the necessity of adjunctive preoperative embolization. Because 95% of posterior fossa AVMs are confined to the cerebellum and only 5% are true brainstem lesions, we will focus on AVMs involving the cerebellum. Despite the complexity of the posterior fossa, it is extremely rare for AVMs to simultaneously involve both the cerebellum and the brainstem. Large AVMs of the cerebellum may appear to involve the entire contents of the posterior fossa, but invariably the lesion is confined to the cerebellar tissue or the brainstem tissue, not both. AVMs of the cerebellar vermis are some of the most common vascular malformations found in the posterior fossa. The primary arterial supply to vermian AVMs is usually via the superior cerebellar arteries (SCAs) and the posterior inferior cerebellar
35
36
arteries (PICAs) bilaterally (see Figure 2C). The dominant arterial feeding supply of either the SCA or the PICA distribution depends upon the anatomical location of the vascular malformation within the vermis relative to the horizontal fissure. AVMs occurring superior to the horizontal fissure and involving the folium, declive, culmen, central lobule, and lingula usually have significant arterial input from the vermian branches of the SCA and lesser input from the PICAs. AVMs below the horizontal fissure within the tuber, pyramid, uvula, and nodulus typically derive the dominant arterial supply from the PICA bilaterally and the SCA to a much lesser extent. Although the SCA and PICA provide the majority of arterial input to vermian AVMs, the anterior inferior cerebellar artery (AICA) often provides deep lateral feeding to larger and deeper vermian AVMs through branches of the AICA. These deep branches of the AICA enter the foramen of Luschka and supply the lateral roof of the fourth ventricle and middle cerebellar peduncle. Vermian AVMs usually drain superiorly into the galenic system through the precentral cerebellar vein or through superior vermian veins bridging into the tentorium. The cerebellar hemisphere also represents a common location of cerebellar AVMs due to the relatively large anatomical region represented by the cerebellar hemisphere. The arterial supply to AVMs of the cerebellar hemisphere is unilateral and typically involves all three vascular territories (SCA, AICA, and PICA), with the dominant supply depending largely on the specific hemispheric location of the lesion (see Figure 5C). The closer the AVM is located to the cerebellopontine angle (CPA) cistern, the more prominent the supply from the AICA. Similarly, the more superior the malformation is located within the cerebellar hemisphere, the more arterial input from the SCA can be anticipated. If angiography demonstrates prominence and significant feeding from the deep branches of the AICA, the AVM usually involves the lateral aspect of the ventricular wall and middle cerebellar peduncle. AVMs within the cerebellar hemisphere usually have venous drainage superiorly and laterally into the petrosal system or superiorly into the galenic system. Cerebellar AVMs confined to the cerebellar tonsil occur less frequently and are limited in size by the nature of their location. The arterial feeding to tonsillar AVMs is usually via a unilaterally dilated PICA with occasionally some secondary feeding from distal inferior branches of the AICA if the lesion extends into the cerebellar hemisphere (see Figure 6B). Venous drainage can be lateral into the sigmoid sinus but is more common into the midline inferior vermian veins. Brainstem AVMs are rare lesions that pose unique challenges to treatment and should be classified as
separate and distinct entities from cerebellar AVMs. Although AVMs of the cerebellar hemisphere may partially involve the middle cerebellar peduncle, AVMs of the cerebellum are usually sharply demarcated and distinct from AVMs of the brainstem. Brainstem AVMs are usually small and should be further subgrouped into superficial lesions confined to the pia or the deep parenchyma. The superficial pial lesions are usually supplied by dilated branches of the SCA or AICA and drain into the prepontine or petrosal venous system. Pial AVMs of the brainstem do not usually present on the ventricular surface and have superficial representation of both arterial supply and venous drainage. Conversely, deep parenchymal AVMs of the brainstem are rarely seen ominous lesions with arterial supply usually through deep vertebrobasilar perforator vessels into the ventral aspect of the malformation. Ventricular involvement is common and venous drainage is via periependymal venous channels that ultimately connect into the galenic system.
Because of the natural history of untreated AVMs and the limited tolerance of the posterior fossa contents to hemorrhage, most patients with cerebellar AVMs should undergo treatment aimed at eliminating the vascular malformation. As potential treatment options are contemplated, the estimated risk of future hemorrhage, the patients age, life expectancy, general medical condition, and neurological status are considered together. Aneurysms commonly occur on the major feeding vessels associated with posterior fossa AVMs and are often the source of both intraparenchymal and subarachnoid hemorrhage. These aneurysms usually occur distally along the major feeding vessels to the AVM nidus. Small pre-aneurysmal dilatations on the intracranial vertebral or basilar artery are frequently found in association with large cerebellar AVMs and do not usually require treatment. On the other hand, proximal saccular aneurysms associated with an AVM should be treated prior to or in conjunction with treatment of the AVM. Failure to treat a proximal aneurysm may result in catastrophic subarachnoid hemorrhage after an AVM is removed distal to an aneurysm. Surgical resection of the AVM without treatment of a proximal saccular aneurysm markedly increases the risk of aneurysm rupture due to an abrupt increase in arterial pressure within the feeding vessel as surgical resection of the AVM is performed. Unlike ruptured cerebral aneurysms presenting with subarachnoid hemorrhage, AVMs of the cerebellum typically present with mixed intraparenchymal, subarachnoid, and intraventricular hemorrhage, with a majority of the hemorrhage usually within the
37
cerebellar parenchyma, and some degree of obstructive hydrocephalus. The volume of subarachnoid blood is usually underestimated on computed tomography (CT) due to the bony confines of the posterior fossa and the poor parenchymal resolution obtained with routine CT techniques. When an acute posterior fossa hemorrhage occurs from a suspected cerebellar AVM, one of the first considerations must be whether external ventricular drainage is immediately necessary. Patients with posterior fossa and intraventricular hemorrhage may rapidly deteriorate due to acute obstructive hydrocephalus and should either be treated with external ventricular drainage or undergo close neurological observation to detect early deterioration from acute hydrocephalus. If hydrocephalus has been adequately treated or is not present, the next consideration is whether an associated hematoma is producing significant mass effect within the posterior fossa and whether emergency clot evacuation is required. If the patient is neurologically stable, operative intervention is best delayed for 4-6 weeks until a portion of the intraparenchymal clot has liquefied and the peak time for parenchymal edema has passed. Patients often tolerate large hematomas of 5-6 cm without significant neurological sequelae if hydrocephalus is adequately treated. This delay in operative intervention allows for liquefication of the clot, potential improvement of the patients preoperative neurological status, and time for extensive radiological investigation of the vascular lesion. If the patients neurological condition has deteriorated to such an extent as to warrant emergency posterior fossa exploration and clot evacuation, several factors must be considered. If an arteriogram has been obtained in the presence of a large intraparenchymal hematoma, the hematoma may compress and partially obscure a significant portion of the AVM during angiography. The surgeon must therefore be prepared to potentially encounter an AVM significantly larger than the lesion visualized on an arteriogram obtained shortly after the initial hemorrhage. An adequate bone flap should be fashioned to accommodate cerebellar swelling and possible hemorrhaging from the malformation during resection. If surgery is required acutely following AVM hemorrhage, clot evacuation without AVM resection should be performed if possible and, if necessary, accompanied by large bony craniectomy and duraplasty. Occasionally, due to the location of the clot, complete hemostasis unfortunately cannot be obtained until the AVM has been resected. Anticipation of this possible eventuality should be factored into the decision-making process when emergency clot evacuation is considered. We prefer to defer definitive AVM resection for 4-6 weeks following a significant posterior fossa hemorrhage if the patients neurologi-
cal condition remains stable. Once a hematoma cavity begins to resolve and liquefy, the surgical resection is greatly facilitated. Repeat arteriography prior to the delayed surgical procedure often reveals portions of the AVM that were not apparent on the initial arteriogram. We reserve acute emergency surgical intervention only for patients with a potentially recoverable posterior fossa hemorrhage who are either moribund despite external ventricular drainage or deteriorate neurologically during a period of observation.
There are few other lesions in vascular neurosurgery where the surgeons conceptualization of the normal and pathological anatomy is as critical as in the surgical management of posterior fossa AVMs. Complete angiographic evaluation combined with high-quality multiplanar magnetic resonance (MR) imaging is crucial to preoperative planning. CT is useful for determining the size and configuration of an initial hemorrhage, if present, and for sequential follow-up study of hemorrhage resolution. Diagnostic cerebral angiography, which includes injection of both internal carotid and both vertebral arteries, is the definitive diagnostic test. For complete evaluation of posterior fossa AVMs, both vertebral arteries must be studied individually. Despite a lesion lateralized to one cerebellar hemisphere, angiographic opacification of the contralateral vertebral artery will yield valuable information regarding feeding to the lesion that is often obscured during ipsilateral vertebral angiography due to overlap and prominence of an individual feeding pedicle. Potential adjunctive preoperative embolization can be planned only after completion and careful review of the diagnostic arteriogram. Following cerebral angiography, MR imaging provides information to aid in determining the anatomical limits of the AVM, the subtle involvement of adjacent structures, and the limits of a planned resection margin. MR imaging will demonstrate the degree to which AVMs of the cerebellar vermis may involve the superior medullary velum or whether large hemispheric and tonsillar AVMs involve the middle and inferior cerebellar peduncles, respectively. At the present time, there is limited indication for CT angiography (spiral CT) or MR angiography in the management of a posterior fossa AVM. The information obtained from these computer renderings does not replace high-quality angiography or multiplanar MR imaging. Spiral CT shows great potential for future study of complex vascular lesions due to the relatively quick acquisition times and threedimensional spacial reconstruction capability. Preoperative embolization of selected feeding vessels can be of significant benefit prior to a surgical
38
floor Figure 1. Prone-concord position used for midline exposure of a posterior fossa AVM, lateral (A) and superior (B) views.
bolization of the SCA territory does not eliminate the necessity for dissection over the cerebellar tentorial surface, with preoperative embolization of the SCA territory, this dissection can be done late in the operative procedure when a portion of the arterial pressure has been eliminated from the venous structures and exposure of the superior aspect of the cerebellum has been maximized. Patients with brainstem AVMs are not ideal candidates for adjunctive embolization. We embolized nine brainstem AVMs prior to the attempted surgical resection of 23 such lesions and found an unacceptably high morbidity rate with preoperative embolization of these dangerous lesions. The vascular supply to these extremely rare lesions is best accessed surgically as close to the malformation as possible, to avoid iatrogenic ischemia to exceptionally vulnerable tissue such as the brainstem.
SURGICAL TECHNIQUE
AVMs of the cerebellar vermis are midline lesions and are best approached via a midline exposure. We prefer the prone-concord position for patients undergoing resection of vermian AVMs, regardless of the location of the lesion within the vermis. The patients are positioned prone on chest rolls with the neck flexed. The operating room table is placed in extreme reverse Trendelenburg position to bring the patients nuchal region horizontal with respect to the floor, and the patients lower extremities and knees are padded and flexed against table support to prevent caudal movement (Figure 1A). The patients head is then laterally canted toward the shoulder contralateral to the side of the operating table where the surgeon will stand. The table is rotated so the patients head is aligned to the surgeons preference with regard to operating room orientation (Figure 1B). Vermian AVMs can be resected via a large midline incision extending from above the inion to the C3-4 region. A large craniotomy bone flap is elevated such that the superior, inferior, and lateral margins of the AVM can be easily visualized, with no portion of the AVM concealed beneath bone. This axiom usually requires a customized removal of bone to include the foramen magnum for caudal lesions and exposure of the transverse sinus and torcula for superiorly located vermian lesions. A wide stellate durotomy is performed in a fashion to optimize posterior fossa exposure (Figure 2A). Prior to initiating resection of the AVM, the superficial margins of the malformation are inspected to ensure that the bony removal and dural opening have been adequate and that subsequent AVM resection can proceed unhindered. Obscuring a margin of the AVM by poor bony exposure can limit the ability to control hemorrhage during the procedure.
procedure to resect a posterior fossa AVM. Embolization is not without inherent risk and an estimated 5% morbidity to overall patient management risk is introduced with each embolization procedure. Embolization should be directed primarily at deep arterial feeding to the AVM, which is not easily accessible early in an operative resection of the lesion. Preoperative embolization of deep AICA feeding arteries to large vermian or cerebellar hemispheric AVMs can significantly lessen the difficulty with intraoperative bleeding from along the roof of the fourth ventricle. Feeders such as these are difficult to access early in the procedure and can be a significant source of hemorrhage deep in the resection bed late in the operative procedure. On the other hand, although the PICA may be easily accessed with an embolization catheter, it is also the most surgically accessible vessel within the posterior fossa and can almost always be surgically transected immediately adjacent to the AVM nidus, much closer to the nidus than can be routinely achieve with an embolization catheter. Aggressive embolization of SCA feeders distal to the dorsal aspect of the brainstem can be very helpful prior to a planned surgical resection of superiorly located cerebellar AVMs, which typically have significant SCA supply. Because large cerebellar AVMs and small AVMs superior to the horizontal fissure typically have venous drainage directed superiorly into the galenic system, preoperative SCA embolization aids the surgical resection and requires less initial dissection over the tentorial surface of the cerebellum early in the procedure. This avoids initial dissection around fragile arterialized venous structures prior to surgical transection of other more accessible feeding vessels early in the procedure. Although em-
39
Figure 2. AVM of the cerebellar vermis. A, exposure. B, schematic representation of the arterial supply and venous drainage. C, clip ligation of the PICA bilaterally adjacent to the nidus.
40
Figure 3. AVM of the cerebellar hemisphere. Vertex view of the patient for the lateral (A) and park-bench lateral (B) positions.
The initial microdissection of a vermian AVM is directed at accessing both PICAs within the subarachnoid space at the midline and dissecting the vessels to their entrance into the inferior aspect of the malformation (Figure 2C). The distal PICA arterial supply to the AVM is divided as close to the nidus as possible to minimize iatrogenic infarction of adjacent but uninvolved cerebellar tissue. After PICA feeding has been eliminated, circumferential dissection of the superficial pial margin of the AVM from an inferior to superior direction is performed on each lateral margin. This will eliminate possible superficial AICA feeding to the malformation, which is usually minimal in all but large vermian AVMs. Superficial dissection along the superior margin is then performed with caution so as not to jeopardize rostral venous outflow from the AVM, which is typically through the superior vermian venous system. After superficial feeding has been eliminated, continued feeding to the AVM is usually present through deep SCA and deep lateral AICA feeders within the superior medullary velum and lateral roof of the fourth ventricle. It is advisable to protect the floor of the fourth ventricle with a soft cotton strip so that if bleeding occurs during later, deeper stages of the resection from small but often numerous periependymal feeders, the lesion can be expediently removed without inadvertently injuring the floor of the fourth ventricle. Venous drainage of vermian AVMs is usually directed superiorly through the dilated superior vermian or precentral cerebellar venous system (Figure 2B). The venous outflow is transected only after
the AVM resection has been completed and the lesion is ready to be delivered from the operative field.
The optimal surgical approach and operative patient positioning for AVMs involving the cerebellar hemisphere is greatly dependent upon the size of the malformation, the location of the lesion within the hemisphere, the necessity for exposure of the CPA cistern, and surgeon preference. We prefer the lateral position for most operative procedures involving AVMs of the cerebellar hemisphere. The patient can usually be positioned supine with a blanket-roll under the ipsilateral shoulder and hip and the head positioned lateral and horizontal with respect to the floor (Figure 3A). If the patient is very large or has limited range of motion of the cervical spine, the patient is best placed in the park-bench lateral position (Figure 3B). The disadvantage of the park-bench lateral position is that the ipsilateral shoulder can decrease the working room of the operative field, especially in a large patient. To resect cerebellar hemispheric AVMs, we generally use a large C-shaped incision centered on the mastoid process and elevate a large subgaleal-subcutaneous skin flap (Figure 4A). The majority of the posterior cervical muscles are detached from the occiput and reflected posteriorly and inferiorly (Figure 4B). Most lateral cerebellar hemispheric AVMs have significant blood supply from the AICA, and access to the CPA cistern is imperative to surgically control this blood supply to the AVM. In order to
41
Figure 4. Surgical resection of AVMs of the lateral cerebellar hemisphere. A, skin incision. B, scalp and muscle reflected. C, bone removal (shaded area).
maximize exposure of the CPA cistern, we extend the lateral extent of the bony craniotomy to the level of the sigmoid sinus laterally. The superior, inferior, and posteromedial extents of the bony resection depend on the size and location of the AVM, but sufficient bone must be removed so that exposure is adequate and dissection unhindered (Figure 4C). After the dura has been widely opened and the malformation inspected to confirm the location of the major arterial feeders and the venous drainage, the primary initial task is to microdissect and transect the arterial feeders as close to the nidus as possible (Figure 5A). The PICA can be identified caudally in proximity to the cerebellar tonsil and followed to its point of entry into the malformation, where it is transected. Similarly, the AICA is identified in the CPA cistern and dissected to its entry into the AVM where it is ligated and divided. Often, the deep supply from the SCA to cerebellar hemispheric AVMs is partially obscured by the superiorly and laterally draining venous outflow of the malformation. By working
around the venous structures and undercutting the cerebellar tissue under the venous drainage, the major feeding from the hemispheric SCA branches can be transected close to the malformation without compromising the venous drainage (Figure 5C). Occasionally, large cerebellar hemispheric AVMs have deep arterial supply from deep branches of the AICA along the lateral aspect of the roof of the fourth ventricle and superior medullary velum (Figure 5B). On review of a preoperative MR image, if the AVM presents or is immediately adjacent to the fourth ventricle, or partially involves the middle cerebellar peduncle, there is a high likelihood of AICA feeding along the deep margin of the AVM. If troublesome bleeding occurs during resection from deep periependymal feeding vessels, expedient entry into the fourth ventricle allows the surgeon to access the periependymal feeding from the AICA supply proximally as the vessels enter the foramen of Luschka.
42
Figure 5. AVM of the cerebellar hemisphere. A, exposure. B, schematic representation of the arterial supply and venous drainage. C, transection of SCA feeding to an AVM close to superiorly located venous drainage.
43
Because of the somewhat small size of the cerebellar tonsil, AVMs confined to the tonsil are relatively rare but easy to resect. These patients can be positioned prone-concord or lateral, and adequate bony exposure should be per formed in the caudal direction to include wide opening of the foramen magnum. Occasionally, the posterior arch of C1 must be removed to completely expose the caudal extent of tonsillar AVMs (Figure 6A). These lesions usually derive their blood supply solely from the PICA (Figure 6C), which can be easily accessed in the subarachnoid space at its lateral and posterior medullary segments. After the PICA feeding has been transected close to the AVM, it is usually simple and expedient to perform a limited tonsillar resection in order to remove the AVM. The venous drainage of AVMs involving the cerebellar tonsil is either lateral into the sigmoid sinus or, more commonly, into the inferior vermian venous system (Figure 6B).
Brainstem AVMs include those lesions with parenchymal representation in the mesencephalon, tectal plate, pons, medulla, floor of the fourth ventricle, and CPA cistern. These malformations are extremely rare and can present with intraparenchymal or subarachnoid hemorrhage or a variety of fluctuating neurological symptoms. AVMs of the brainstem are best divided into two distinct groups: superficial or pial AVMs and deep parenchymal AVMs. Although AVMs from both groups may present with similar clinical signs and symptoms, the indications for surgical treatment and outcomes from the two groups are markedly different. Superficial AVMs of the brainstem primarily involve the pial-arachnoid layers, and minimally involve the underlying parenchyma. They often occur on the anterior or lateral surface of the brainstem in the CPA, and the arterial supply of these lesions is usually via dilated branches of the AICA with occasional supply from the SCA (Figure 7). Superficial brainstem AVMs that involve the anterior surface of the brainstem usually have no deep perforator feeding, with the venous drainage through lateral pontine veins into the petrosal system (Figure 8). We have found preoperative embolization of superficial brainstem AVMs to be extremely hazardous and of little value as an adjunct to their resection. Brainstem AVMs in the CPA are usually approached with the patient in a lateral position with an extreme lateral bony exposure in order to maximize access to the CPA cistern (Figure 9). Bone is removed laterally to the sigmoid sinus, superiorly to the tentorium, and inferiorly to include the foramen magnum. The cistern of the CPA is opened widely from the cisterna magna to the tentorium, and cerebrospinal fluid is allowed to egress to provide relaxation of the
cerebellum. Under high-power magnification, the feeding vessels to the malformation are identified and sequentially transected. The complexity of pial AVMs of the lateral brainstem in conjunction with the limited space available in the CPA usually precludes complete circumferential dissection during the initial phases of AVM resection. Following the initial dissection of the AVM, it is often necessary to gently elevate the nidus away from the brainstem within a shallow plane of pial dissection beginning posterolaterally and continuing anteromedially along the ventral surface of the brainstem (Figure 10). With sequential occlusion of the smaller feeding vessels entering the malformation, the malformation can be completely undercut from the pial surface, and the remaining feeding vessels and draining veins cauterized, divided, and delivered from the operative field. Deep parenchymal AVMs of the brainstem pose an entirely different surgical risk than superficially located lesions. The feeding vessels to deep brainstem AVMs are usually vertebrobasilar perforator arteries originating ventrally and coursing through normal brainstem tissue prior to penetrating the deep margins of the malformation (Figure 11). Deep parenchymal brainstem AVMs often involve the floor of the fourth ventricle, and venous drainage is usually through dilated periependymal veins into the galenic venous system. These malformations are intimately associated with normal brainstem parenchyma, and the arterial supply to these malformations cannot be accessed beyond the margins of the AVM without significant risk of iatrogenic ischemia and infarction of adjacent normal brainstem parenchyma. The small size of the feeding vessels along with their irrigation of adjacent normal tissue makes adjunctive embolization difficult and minimally successful. In a similar fashion, surgical resection of deep parenchymal AVMs of the brainstem carries extremely high morbidity and mortality rates. Occasionally, when the lesion is small and associated with an intraparenchymal hematoma, surgical resection can be accomplished by working within the resolving hematoma cavity without significant disruption of normal brainstem tissue. The approach to parenchymal AVMs of the brainstem should traverse the minimum amount of normal tissue required to access the lesion. The margins of the malformation are identified and the lesion is circumferentially dissected until the small feeding vessels at the depths of the resection bed are reached. The surgical resection of these lesions is usually associated with considerable and persistent deep arterial bleeding that can only be managed by cautery and careful, persistent microdissection. In general, patients with deep parenchymal brainstem AVMs are poor candidates for microsurgical resection because of the multitude of problems mentioned above.
44
Figure 6. AVM of the cerebellar tonsil. A, exposure. B, schematic representation of the arterial supply and venous drainage. C, electrocautery of PICA feeding vessel.
45
B
Figure 7. MR image (A) and anteroposterior angiogram (B) of a superficial AVM of the brainstem.
AVMs of the posterior fossa frequently present with clinically significant hemorrhage within the substance of the cerebellum or brainstem, within the subarachnoid space, or both. It is common to find aneurysms on proximal feeding arteries of posterior fossa AVMs that are the source of subarachnoid or intraparenchymal hemorrhage. AVMs of the posterior fossa are best subdivided by location into cerebellar vermian AVMs, cerebellar hemispheric AVMs, cerebellar tonsillar AVMs, superficial brainstem pial surface AVMs, and deep parenchymal brainstem AVMs. Classification into one of the above locations helps anticipate the suspected arterial supply and likely venous drainage. AVMs of the posterior fossa require full radiological evaluation prior to treatment, including four-ves-
SUMMARY
sel angiography and high-quality MR imaging. Embolization as an adjunct to surgical resection is useful to decrease the blood flow to cerebellar AVMs primarily through embolization of the deep feeding vessels, which are not easily accessed initially during an operative resection. Feeding arteries that are easily accessed microsurgically should not undergo unnecessary embolization due to the risk of iatrogenic tissue infarction. Embolization of brainstem AVMs carries a high risk and is of little benefit as an adjunct to surgical resection. The surgical resection of posterior fossa AVMs is best delayed until partial resolution of any hematoma, dependent upon the patients neurological condition. The natural history of posterior fossa AVMs suggests that surgical treatment to eliminate these aggressive lesions is indicated in the majority of circumstances.
46
Figure 8. Illustration depicting an AVM of the superficial CPA surface of the pons.
Figure 9. Illustration depicting the initial lateral exposure of a superficial pontine AVM presenting to the CPA cistern.
Figure 10. Illustration depicting ligation of feeding arteries and elevation of a superficial pontine AVM.
Figure 11. Illustration depicting a deep parenchymal AVM of the brainstem with deep perforators feeding from the basilar artery.
Arteriovenous malformations (AVMs) in the region of the ventricular trigone present unique challenges to the neurosurgeon. These lesions are invariably located in eloquent or functionally important brain and have a complex arterial supply and deep venous drainage. Involvement of the choroid plexus with the AVM may add to the overall size of the lesion. To expose and resect these complex lesions, a number of tangential approaches have been utilized rather than the more conventional perpendicular approaches used for AVMs on the cerebral convexity.
INTRODUCTION
As with other AVMs, lesions in the region of the ventricular trigone most commonly come to clinical attention during the third or fourth decades of life. These deep-seated lesions most frequently produce symptoms after an intraventricular and/or intracerebral hemorrhage. Infrequently, these malformations present with seizures or a visual field deficit in the absence of hemorrhage. Since the incidence of AVM rebleeding is very low immediately after the initial hemorrhage and cerebral vasospasm is exceed-
PATIENT SELECTION
ingly rare, immediate surgery is usually not indicated for the patient presenting with a hemorrhage. Surgical intervention is typically delayed for 2 or 3 weeks to allow for recovery from the hemorrhage. Occasionally, a patient presenting with a large hematoma and associated mass effect may require urgent evacuation of the hematoma. In this situation, the emergency operation is usually performed to remove the hematoma, with the actual resection of the AVM reserved for a later date. Once the patient has stabilized, the intricate neurovascular operation can be carried out more electively. After the AVM has been localized and defined by appropriate imaging studies, the various therapeutic options are reviewed with the patient and family. Treatment recommendations are then tailored to the individual patient and lesion. In addition to no treatment, options include embolization, surgical resection, and stereotactic radiosurgery. The decision is made after careful consideration of a variety of factors related to both the patient and the malformation. The patients age, health, neurological condition, occupation, and psychological reaction to the knowledge that he or she harbors a potentially dangerous lesion must all be taken into consideration. Factors related to the AVM that are of importance in decision-making include the precise location, size, arterial supply, venous drainage, and relationship of the malformation to associated hematomas. Because of the poor natural history of untreated AVMs, most of these
47
48
lesions require some form of therapy. Embolization alone rarely cures an AVM in the region of the ventricular trigone and is therefore not often of benefit as a sole treatment. It is, however, a very useful adjunct to surgery. Stereotactic radiosurgery is a reasonable option for treating small AVMs in this region, but the significant time interval required for the radiation to have its beneficial effects as well as the risk of radiation injury represent significant detractors to this therapeutic modality. In the majority of trigonal AVMs, the risk of surgical removal with an immediate cure is significantly less than the risk of the natural history of the untreated malformation.
of the deep arterial supply by embolization reduces the critical blood supply and diminishes surgical difficulty and risk. Furthermore, preoperative embolization allows for a more gradual redistribution of arterial blood flow to the surrounding normal brain and theoretically reduces the risk of postoperative reperfusion complications.
Patients presenting with an intracerebral and/or intraventricular hemorrhage are managed aggressively to control intracranial hypertension and may require a ventriculostomy for control of hydrocephalus, if present. Blood pressure control using oral or parenteral medications is also commonly necessary in the acute period. Since all operative approaches to AVMs in the region of the ventricular trigone require violation of the cortex, patients are placed on prophylactic anticonvulsants. After initial stabilization of the patient, diagnostic imaging studies are carried out to define the anatomical features of the lesion. Noncontrasted computed tomography is the initial diagnostic study performed acutely in the patient presenting with a hemorrhage. This allows determination of the extent and location of the hemorrhage and the diagnosis of hydrocephalus. Invasive cerebral angiography is essential to define the angioarchitecture of the AVM. The posterior cerebral artery usually feeds these lesions, with the posterior choroidal and posterior temporal arteries being predominant. Other arteries that occasionally contribute include the anterior choroidal, middle cerebral, pericallosal, and thalamoperforating arteries. Venous drainage is typically deep into the galenic system. Occasionally, a small amount of superficial cortical venous drainage is there as well. In rare instances, the vein of Labb is involved or the lesion has purely superficial venous drainage. Magnetic resonance imaging (MRI) provides anatomical information that complements the angiogram and is extremely useful in planning the most appropriate operative approach. MRI demonstrates the precise relationship of the vascular malformation and associated hematoma to the trigone, indicates the proximity to adjacent anatomical structures, and reveals the cortical surface closest to the AVM. We have found preoperative embolization to be an important adjunct to the surgical resection of AVMs in this region. The surgical approaches to these malformations often expose the deep arterial supply at a later stage of the procedure. Reduction or elimination
PREOPERATIVE PREPARATION
The advantages of general anesthesia to achieve a relaxed brain and to control blood pressure provide a more optimal environment for successful operations on these lesions. Prior to the induction of general anesthesia, patients are given anticonvulsants, systemic corticosteroids, and prophylactic antibiotics. An invasive arterial line is placed to provide access for arterial blood gas determinations and to allow for continuous monitoring of the arterial pressure. It is our preference to operate on AVMs under normal blood pressure and to prevent hypotension. Since autoregulation is often impaired in the brain adjacent to an AVM, profound hypotension or excessive retraction to the adjacent brain may result in ischemic injury. If the patients head is elevated above the heart to any significant degree, a precordial Doppler ultrasound and an atrial catheter are placed preoperatively to optimally monitor for and manage potential air emboli. A Foley catheter is used for urinary drainage and measurement.
ANESTHETIC TECHNIQUE
We have utilized three basic approaches in the surgical management of AVMs in the region of the ventricular trigone. They include: 1) a lateral temporal approach through the inferior or middle temporal gyrus; 2) a parieto-occipital transcortical approach; and 3) an occipital interhemispheric approach. The latter can be ipsilateral and parafalcine or contralateral and transfalcine. Selection of the approach should be based upon the presence of preoperative neurological deficits, the presence of an intracerebral hemorrhage, and the relationship of the AVM to the trigone based on angiographic and MRI localization (Figure 1). The transtemporal approach is used for AVMs located along the lateral, anterior, or inferior wall of the trigone (Figure 1A). This approach is particularly useful for AVMs associated with a hematoma extending into the temporal lobe. The interhemispheric approach is reserved for selected AVMs situated along the medial aspect of the trigone that may also involve the splenium of the corpus callosum (Figure 1B). The parieto-occipital approach, as described by Heros, is used for AVMs involving the superior and medial walls of the trigone (Figure 1C).
SURGICAL TREATMENT
49
Figure 1. Various relationships of trigonal AVMs to the ventricular trigone. A = AVM along the lateral, anterior, or inferior wall of the trigone. B = AVM on the medial aspect of the trigone. C = AVM involving the superior wall of the trigone.
After induction of anesthesia, the patient is placed in a supine position with the ipsilateral shoulder elevated and the head parallel to the floor with the vertex angled slightly downward. The head is held in position by a radiolucent headholder. A horseshoeshaped incision is made around the auricle of the ear from the zygomatic arch to the retromastoid region. A temporal craniotomy is extended to the floor of the middle cranial fossa (Figure 2). Dural tack-up sutures are placed and the dura is opened. The AVM is localized along with any associated hematoma utilizing intraoperative ultrasound. The vein of Labb is identified and a 1- to 2-cm incision is created in the inferior temporal gyrus. The middle temporal gyrus may be traversed in the nondominant hemisphere. The cortisectomy should be planned in order to provide the most direct access to the AVM while protecting the vein of Labb. A small horizontal cortical incision parallel to the optic radiations will minimize the
Temporal Approach
risk of visual field deficits (Figure 3 upper). The operating microscope is used to assist in guiding the dissection slightly posteriorly and superiorly toward the trigone with suction and bipolar cautery (Figure 4). Self-retaining retractors are used to maintain exposure (Figure 3 lower). To minimize injury to the eloquent cortex, minimal force should be used in placing the retractors. If a hematoma is present, it is evacuated to provide more relaxation and exposure. Once the trigone is entered, the brain relaxes significantly due to the drainage of cerebrospinal fluid and minimal retraction is necessary to maintain adequate exposure. Utilizing this approach, the surgeon usually encounters the nidus of the AVM prior to the feeding arteries. Dissection must be carried circumferentially around the nidus to the feeders that are sequentially coagulated and/or clipped and divided (Figure 5). The deep venous drainage is usually on the medial aspect of the AVM and is preserved until the AVM is completely devascularized.
50
Figure 2. Skin incision (solid line) and temporal craniotomy (hatched line) for the temporal approach to AVMs of the ventricular trigone.
Figure 4. Temporal approach. Illustration of the direction of the operative approach through the inferior or middle temporal gyrus to the ventricular trigone.
51
Figure 3. Temporal approach. Illustrations demonstrating the horizontal cortical incision parallel to the optic radiations (upper) and use of self-retaining retractor to maintain exposure (lower).
Figure 5. Temporal approach. Illustration of coronal section of the brain demonstrating the relationship of the AVM to the ventricular trigone and the operative approach through the inferior temporal gyrus.
52
Parieto-Occipital Approach
The patient is placed in a semi-sitting slouch or lounging position, with the head slightly flexed (Figure 6). The external landmark for the cortical incision is a point approximately 9 cm cephalad to the inion and 1.5 to 2.0 cm lateral to the midline (Figure 7A). A horseshoe-shaped skin incision is centered over this point, extending to the midline and based inferiorly. Burr holes are placed and a parieto-occipital craniotomy is created that is centered on the above landmark (Figure 7A). Once the dura is exposed, it can be opened with the flap based on the sagittal sinus. After opening the dura, the exact location of the cortical incision should be made by directly measuring 7.0 cm above the occipital pole and 2.0 cm lateral to the midline (Figure 7A). A laterally oriented cortical incision is then created, approximately 2.0 cm in length, and the approach to the trigone is aimed toward the ipsilateral pupil (Fig-
ure 7B). Again, intraoperative ultrasound or an image-guided stereotactic navigational system may be used to increase the precision of the approach. As with the transtemporal approach, the surgeon usually initially encounters the nidus of the AVM; circumferential dissection around the superior, medial, or lateral surface of the nidus exposes the feeding arteries (Figure 8). The surgeon has a direct view to the trigone from a superior angle using this approach. This is opposed to the oblique tangential view given to the superior portion of the trigone obtained from a parasagittal approach. In addition, there is less retraction required than for the parasagittal approach in order to reach this superior region of the trigone, and prolonged retraction of parasagittal veins is avoided. Since placement of the cortical incision is between the parietal sensory association fibers and the occipital visual association fibers, the risk of neurological injury is minimized.
Figure 6. Parieto-occipital transcortical approach. Patient is in a semi-sitting slouch or lounging position with the head slightly flexed. Arrow = direction of approach to the ventricular trigone.
53
Figure 7. Parieto-occipital transcortical approach. Posterior (A) and lateral (B) views of external and internal landmarks for approach to the ventricular trigone. Solid line = skin incision; hatched line = craniotomy.
Figure 8. Parieto-occipital transcortical approach. Illustration of laterally oriented cortical incision with exposure maintained by self-retaining retractors. The AVM is encountered on the superior surface of the ventricular trigone and extends into the ventricular system.
54
The patient is positioned in the lateral position with the ipsilateral shoulder on a roll and the head parallel to the floor with the AVM side down, so that gravity assists in retraction of the occipital lobe (Figure 9). A horseshoe-shaped incision is made in the occipital region based laterally. An occipital craniotomy is created with the medial burr holes placed over the superior sagittal sinus (Figure 10). The dural opening is based medially on the superior sagittal sinus. The occipital lobe can then be retracted from the falx since there are usually no veins draining from this area of the cortex to the superior sagittal sinus. Retraction is minimized by the positioning of the patient, which allows the side with the lesion to fall away from the falx. The calcarine or parieto-occipital sulci are opened to expose the most medial aspect of the AVM (Figure 11). One disadvantage of this approach is that the surgeon often exposes the venous
drainage before the arterial supply. The former must be protected until the AVM is devascularized. Feeding vessels are then coagulated and divided with larger arteries requiring clips. Venous drainage is then coagulated and/or clip ligated and divided. A variation of this approach involves parafalcine dissection contralateral to the side of the AVM in order to approach the AVM from a more perpendicular trajectory. The patient can be positioned in a semisitting slouch position. The horseshoe-shaped incision is made across the midline with the base of the flap on the side contralateral to the lesion. The craniotomy crosses the midline instead of basing itself on the superior sagittal sinus and is mainly contralateral to the AVM. The dura is opened contralateral to the AVM and the flap is based on the superior sagittal sinus. The falx is split, allowing access to the hemisphere ipsilateral to the lesion, and is approached transcortically as described above.
Figure 9. Parasagittal interhemispheric approach. Patient is positioned with the side of the craniotomy dependent to allow gravity to assist with occipital lobe retraction. Solid line = skin incision; hatched line = craniotomy.
55
Figure 10. Parasagittal interhemispheric approach. Illustration of an AVM on the medial surface of the ventricular trigone. The craniotomy is centered over the occipital lobe extending to the occipital tip, and the AVM is exposed via an interhemispheric approach. Arrow points to location of AVM and relationship to the ventricular trigone on insert.
Figure 11. Parasagittal interhemispheric approach. The calcarine fissure has been opened and the AVM is approached beneath the surface of the occipital lobe. Circumferential dissection around the AVM is utilized to devascularize the lesion.
56
Once the AVM has been surgically removed, we routinely per form intraoperative angiography to document complete obliteration of the malformation. If discovered postoperatively, incompletely resected AVMs are associated with a significant risk of hemorrhage and require reoperation. If any portion of the AVM is identified on intraoperative angiography, the resection bed is re-explored and residual malformation resected.
INTRAOPERATIVE ANGIOGRAPHY
Once complete removal of the malformation is documented by intraoperative angiography, the resection bed is carefully inspected under the operating microscope to be certain that there is absolute hemostasis. Under constant irrigation, the surgeon examines the resection bed with a piece of cotton or cottonoid patty and suction. Any bleeding points are carefully coagulated with bipolar cautery because one should not rely on hemostatic agents to arrest bleeding in the resection bed. During this stage of inspection, the patients blood pressure is maintained at a normal or slightly elevated level. Once excellent hemostasis is obtained, the resection bed is lined with a single layer of Surgicel. The Surgicel layer should turn brown when adequate hemostasis has been achieved. Any areas of red indicate inadequate hemostasis and require removal of the Surgicel and complete control of the bleeding site. Gelfoam or Avitene packing may obscure underlying bleeding and are avoided. Once proper hemostasis is achieved, the anesthesiologist is asked to perform a Valsalva maneuver on the patient to ensure that there are no venous bleeders. The resection cavity is filled with saline, and the dura is closed in a watertight fashion. The bone flap is replaced with sutures or miniplates, and the wound is closed in layers.
The surgical management of intracranial AVMs presents some of the most complex decision-making and technical challenges in neurosurgery. Complications and misadventures may occur at any point in the diagnostic or therapeutic process in the management of trigonal AVMs.
COMPLICATIONS
Preoperatively, the patient must undergo appropriate risk stratification with regard to other medical conditions that may affect the intraoperative and postoperative hospital course. Multimodal imaging obtained during the preoperative work-up is important for providing the surgeon with spatial conceptualization of the lesion with respect to the regional anatomy. All risks that apply to craniotomy in general and AVM surgery specifically exist in the surgical management of trigonal AVMs. Some potential complications are unique to these lesions due to their anatomical relationships to the optic radiations, language cortex, and thalamus. Intraoperatively, most complications can be avoided by using meticulous technique. Since these AVMs are in eloquent or functionally important cortex, the margin of resection must not be too wide. Dissection must closely follow the nidus as there may or may not be a clear gliotic plane to follow. Choosing the correct approach to expose the AVM will avoid retraction injury to the surrounding normal cortex and minimize the risk of damage to the optic radiations. Arteries en passage that supply the AVM but extend beyond to perfuse normal brain must be identified. Only pedicles devoted to the AVM are coagulated or clipped. As with other AVMs, care must be taken to protect the venous drainage until the malformation has been devascularized to avoid congestion and bleeding from the nidus. Postoperative hemorrhage may occur due to rebleeding from the retained AVM, poor intraoperative hemostasis, or normal perfusion pressure breakthrough. AVMs in the region of the ventricular trigone are not common. Their surgical treatment is confounded by the lack of cortical representation, involvement of the choroid plexus, deep arterial feeders and venous drainage, and location in functionally important cortex. These technical challenges have resulted in the development of the surgical approaches described above. Selective angiography and MRI assist in determining which approach is optimal for the individual AVM. The judicious use of preoperative embolization to devascularize the lesion and intraoperative angiography to document achievement of the surgical goals are additional adjuncts that enhance the safety and efficacy of surgical treatment of these lesions.
Dural arteriovenous malformations of the transverse and sigmoid sinuses are abnormal direct shunts between the dural arterial supply and the dural venous system. They represent 10% of all intracranial vascular malformations. The lesions are acquired in life and are more appropriately referred to as fistulas rather than malformations. The management of dural arteriovenous fistulas (DAVFs) r equires a clear understanding of the angioarchitecture of the lesion. This chapter describes the treatment approach, which uses the combined neurosurgical and neurointerventional services with specific emphasis on surgical treatment. A case is also presented to illustrate the concepts presented in this chapter. These lesions can be extremely complex and require an understanding of the arterial supply, nidus location, venous drainage, and location of the vein of Labb. The arterial supply to these lesions can originate from any of the arteries that supply the dura. These arteries include transcranial perforating vessels from the occipital, posterior auricular, middle meningeal, and ascending pharyngeal branches of the external carotid artery. Supply can also come from the dural
INTRODUCTION
branches that arise from the anterior, middle, and posterior cerebral arteries, in addition to petrous and cavernous branches of the internal carotid artery and the posterior meningeal branch of the vertebral artery. Infrequently, there can also be branches from the subclavian artery. This vast number of potential supplying arteries is the basis for treatment failure when only supplying arteries are embolized or surgically ligated. Occluding any portion of the arterial supply often leads to hypertrophy of the remaining supplying vessels. For this reason, the arterial supply to DAVFs does not strongly guide the therapy. The nidus of the DAVF is located in the wall of the sinus, is usually small and can be spread over several centimeters of the sinus. The nidus can generally be defined on the early phase of the angiogram, and therapy is directed at this portion of the sinus. Hypertrophied arteries and veins supplying the fistula can give the appearance that the bone, brain, or convexity dura is involved in the lesion, although the true nidus is located within the dura and around the sinus. The crucial aspect of these lesions is the venous drainage. Venous drainage is the most important factor in assessing the risk of stroke as well as directing treatment of the lesion. A distinction must be made between the normal venous drainage pattern of the brain into cortical veins and dural sinuses and the drainage of the DAVF, which may also fill dural sinuses and cortical veins in a retrograde direction.
57
58 PATIENT SELECTION
Patients with DAVFs can present with a number of different clinical symptoms. These symptoms depend on the location, flow, and venous drainage of the DAVF. Most commonly, the patient will develop a pulse-synchronous bruit. This bruit may be barely audible to the patient or, in rare instances, so loud it can be heard by others in the room. The character of the bruit can change as flow dynamics within the fistula change. Many patients present with a new headache pattern that may relate to pressure changes within the sinus. These headaches are often more prominent on the side of the fistula and can change in character with the fistula. Venous hypertension without hemorrhage can also result in neurological deficits as a result of cerebral edema and infarction. Memory or visual impairment may be noted by the patient. Symptoms related to raised intracranial pr essure may also occur as a result of impaired cerebrospinal fluid absorption. DAVFs with cortical drainage can also result in seizures. DAVFs may present with epidural, subdural, subarachnoid, or intracerebral hemorrhage. Patients with spontaneous brain hemorrhages must be fully evaluated for a possible DAVF.
1B). The sagittal and straight sinuses drain via the torcula and contralateral transverse sinus and jugular vein. Any stenosis in the ipsilateral sigmoid sinus or jugular vein may increase the retrograde flow. In the presence of jugular venous outflow restriction, even small DAVFs may have some retrograde flow.
Some fistulas appear to drain entirely toward the torcula, contralateral transverse sinus, and jugular vein. This pattern occurs when the ipsilateral sigmoid sinus, jugular bulb, or jugular vein is occluded (Figure 1C). The flow may also appear angiographically to be entirely retrograde in cases of severe stenosis. Only when a microcatheter is advanced through the stenosis can a determination of patency be made.
The angioarchitecture of DAVFs involving the transverse and sigmoid sinuses can be divided into five categories based on the patterns of venous drainage of the fistulas. These categories are not rigid and some lesions may not fit exactly within a single category. It is equally important to understand and characterize the venous drainage of the brain with specific attention to the vein of Labb. This chapter emphasizes these aspects of the DAVF. These categories are graphically illustrated in Figure 1. DAVFs with normal antegrade flow drain into the involved sinus and down the ipsilateral jugular vein (Figure 1A). The arterial supply to the fistula is usually small and the sinus pressure is not sufficiently elevated to cause retrograde flow toward the torcula. The direction of flow is not only dependent on the flow rate but also the presence of any stenosis in the ipsilateral sigmoid sinus or jugular vein. The sagittal and straight sinuses may drain to the right and/or left transverse sinus.
ANGIOARCHITECTURE
This category applies to fistulas that, in addition to drainage through the involved dural sinuses, have drainage into cortical veins (Figure 1D). The vein of Labb is commonly involved with cortical venous drainage. This group would also include those fistulas that drain retrograde into the straight sinus, rather than across the torcula to the contralateral transverse sinus. The critical finding in this group is cortical venous drainage. This pattern has a much higher risk for hemorrhage. Cortical venous drainage may result from large fistulas with high flow rates and/or venous outflow restriction in the dural sinuses or jugular vein. Both states result in high pressures in the transverse and sigmoid sinuses. The result is reflux of blood into the cortical veins. A variety of fistulas may drain entirely into cortical veins and not the adjacent dural sinuses (Figure 1E). This may occur angiographically if the associated sinus is thrombosed. Rarely, the fistula may drain entirely into cortical veins even if the underlying sinus is completely normal. The connection between the fistula and the sinus has either thrombosed or was never present. These fistulas have the highest risk of hemorrhage because of the high venous pressure. They are similar to the DAVFs of the anterior cranial fossa/ethmoidal groove that virtually always drain to cortical veins and present with hemorrhage. The vein of Labb provides venous drainage from the parietal and posterior temporal lobes. Usually, it is not a single vein but several veins that enter the sinus at the junction of the transverse and sigmoid sinuses. This vein may be very small and inconsequential or large and provide a significant and cru-
Antegrade and Retrograde Flow Through the Transverse-Sigmoid Sinus with Cortical Venous Drainage
DAVFs with antegrade and retrograde flow are usually larger and the sinus pressure is sufficient to cause antegrade flow to the ipsilateral jugular vein as well as retrograde flow toward the torcula (Figure
59
Figure 1. The angioarchitecture of a right DAVF of the transverse-sigmoid sinus supplied by the right occipital artery. A, normal antegrade flow down the right jugular vein. B, antegrade and retrograde flow toward the torcula with right sigmoid sinus stenosis. C, retrograde flow to the torcula with right sigmoid sinus occlusion. D, antegrade and retrograde flow with cortical venous drainage into the right vein of Labb. E, flow into cortical veins only. Flow in the sinus is normal with no connection to the fistula located in the wall of the sinus. All drainage is through the sinus wall into the right vein of Labb.
60
cial pathway of venous drainage from the brain. The vein is seen on late phases of an internal car otid angiogram on lateral and frontal projections. DAVFs of the transverse and sigmoid sinus may alter drainage of the vein of Labb. The pressure in a normal dural sinus is less than 10 mm Hg. Usually, the vein of Labb drains the posterior temporal lobe into the transverse-sigmoid sinus junction. Direct arterial to sinus shunts increase the pressure in the sinus. Depending on the size of the shunt and the presence of venous outflow restriction in the sigmoid sinus or jugular vein, pressure rises within the sinus. As the pressure in the sinus rises, flow in the vein of Labb ceases; if the pressure is sufficiently high, flow will reverse direction. This is what is referred to as cortical venous drainage. The vein of Labb may not be seen on the delayed phase of the internal carotid angiogram if there is r etrograde flow that is dif ficult to distinguish from a small or absent vein.
more broad. Low-risk fistulas may not require surgical treatment if the patient can tolerate the symptoms. If the fistula is treated, palliative therapy may be more appropriate for the initial treatment. If the fistula recurs, the fistula may eventually need more definitive treatment for cure. The following is a review of the different treatment options. These therapies are primarily based on fistula angioarchitecture.
The spontaneous remission of DAVFs has been reported in rare instances. Although the incidence of spontaneous closure is not known, it needs to be considered when discussing the natural history of these lesions. Low- or high-risk DAVFs may close without treatment.
CONSERVATIVE TREATMENT
The decision to surgically treat an individual must be based on knowledge of the patients symptoms, angiographic anatomy, risk factors for hemorrhage or neurological deficits, and an understanding of the risks of surgical treatment. A common indication for treatment is patient intolerance of the bruit. In many cases, the patient is able to cope very well with a loud bruit and no treatment is indicated. However, appropriate treatment is offered to those who cannot tolerate the bruit. Risk factors for hemorrhage and stroke include cortical venous drainage and retrograde flow into the torcula and straight sinus. These conditions lead to venous hypertension, which can also result in cerebral edema and associated neurological deficits. Nearly all DAVFs associated with hemorrhage, raised intracranial pressure, or neurological deficits have one of these risk factors. Every effort should be made to cure these high-risk lesions. Subtotal occlusion may not protect the patient from further episodes of bleeding; therefore, the goal of treatment is complete occlusion. Once the indications for treatment and the angioarchitecture of the fistula have been determined, treatment plans can proceed. The risks and benefits of each treatment must be weighed against the risk of stroke from the fistula. Patients with hemorrhage, infarction, or cerebral edema with associated neurological deficits are considered at high risk and should have definitive treatment that results in complete obliteration of the fistula. Subtotal treatment may not provide patients with protection from further hemorrhage, since the fistula may return with hypertrophy of residual arterial feeders. However, in patients who have low-risk fistulas, the options for treatment are
In many instances, DAVFs of the transverse and sigmoid sinuses have arterial supply from the occipital artery. Before being evaluated in a clinic, many patients have already noticed that they can stop their bruit by compressing this artery. Compression therapy is an appropriate option for the compliant patient with a low-risk fistula. Compression of the carotid artery may also lead to thrombosis of the fistula. The arteries are compressed in an increasing fashion over several days to a maximum of 30 seconds 3 times per hour. This treatment is continued for 3 weeks before follow-up. However, patients with high-risk fistulas, such as those with cortical venous drainage, hemorrhage, or infarction, should not perform compressive therapy and if possible should undergo more definitive treatment. Compressive therapy has a relatively low chance of resulting in fistula thrombosis, with most patients eventually undergoing additional treatment.
COMPRESSION THERAPY
TREATMENT
More definitive treatment of DAVFs of the transverse and sigmoid sinuses requires one or a combination of the following: transarterial embolization, transvenous embolization, or intraoperative transvenous embolization with surgical resection. Transarterial embolization may be used as either definitive treatment of a DAVF or as a preoperative adjunct. Microcatheters are navigated into as many arteries supplying the fistula as possible which are then occluded with an embolic agent. General classes of embolic agents include particles (Gelfoam or polyvinyl alcohol (PVA)), platinum microcoils, liquid adhesives, or sclerosing agents (glucose, alcohol, or Sotradecol). A thorough understanding of possible collateral connections between branches is imperative to reduce the risk of neurological deficit resulting from
EMBOLIZATION
Transarterial Embolization
61
arterial embolization. Virtually all external carotid artery branches relevant to a DAVF of the transverse and sigmoid sinuses also have anastomoses with either the internal carotid artery or the vertebral artery. Knowledge of the arterial supply to the cranial nerves is also important to avoid complications. Provocative testing of an artery can easily be performed by neurological testing during selective intraarterial lidocaine injection. If neurological deficits result, it is probably not safe to embolize that artery with small particles or liquid adhesives. Although arterial embolization can be performed with little risk if particles or coils are used, the main concern is a low cure rate with isolated arterial embolization. Even if the arterial supply to the DAVF is almost completely closed, a high rate of recurrence remains. This recurrence is due to the large number of arterial feeders that supply a fistula. It is usually not possible to embolize every arterial pedicle, and the remaining arteries will hypertrophy and re-establish the fistula. However, transarterial embolization is usually effective in eliminating or reducing the bruit for some time. This may alleviate the patients symptoms to a degree that can be tolerated. The fistula may also progress to complete thrombosis. Tranvenous embolization can be performed either percutaneously, via the femoral vein, or intraoperatively with direct exposure of the sinus. This approach is safe but carries increased risk in the presence of cortical venous drainage. In those cases embolization may divert flow into the cortical veins, increasing the risk of hemorrhage. Without cortical venous drainage, embolization is safe due to the low chance of hemorrhage from dural sinuses. Embolization of cortical veins is generally not safe secondary to their thin walls, which increase the risk of perforation. In the presence of cortical venous drainage, a DAVF may still be embolized safely if the sinus is surgically exposed. The sinus is exposed via a process described later in this chapter. After exposing the cortex around the sinus, the arterialized veins draining retrograde from the sinus are divided. Any normal draining vein is preserved. The sinus can then be punctured and coils placed in the sinus along the entire length involved in the DAVF. This operative exposure allows for occlusion of these fragile retrograde draining veins prior to sinus embolization, preventing a catastrophic hemorrhage. The placement of coils transvenously must be precise, whether performed percutaneously or directly after operative exposure. The coils are placed across the entire length of involved sinus. If the entire involved portion is not occluded, any residual will continue to allow flow through the fistula and will enlarge with time. Care must be taken to avoid sinus
occlusion where normal veins, especially the vein of Labb, drain antegrade into the sinus. There is usually enough separation between a normal vein of Labb and the fistula to allow placement of coils across the fistula while preserving the vein. In cases of close proximity, operative exposure and intraoperative coil placement allow for more precise placement of the coils. There are relatively few instances in which it is necessary to resect the involved sinus. When the sinus is thrombosed and venous drainage of the DAVF is only into cortical veins, surgery is the best approach to occlude the veins draining the fistula. Resection of the sinus is not needed. An example of when surgical resection may be required is the case of a fistula with underlying transverse and sigmoid sinus thrombosis that drains into the superior petrosal sinus. This sinus may be more difficult to embolize due to its small size and deep location. Surgical resection of the involved sinus is technically superior and more likely to result in cure of the DAVF. Surgical resection of lesions affecting the transverse and sigmoid sinuses has been well described previously but is reviewed in this chapter.
Surgical Resection
Transvenous Embolization
The approach to treating DAVFs of the transverse and sigmoid sinuses depends largely on the angioarchitecture of the fistula. As mentioned previously, this is based on the five patterns of venous drainage from the fistula, which include the following. Normal antegrade flow through the transversesigmoid sinus to the jugular vein Antegrade and retrograde flow through the transverse-sigmoid sinus Retrograde flow to the torcula Antegrade and/or retrograde flow through the sinus with cortical venous drainage Flow into cortical veins only
CLINICAL APPROACH TO DURAL ARTERIOVENOUS FISTULAS OF THE TRANSVERSE AND SIGMOID SINUSES
In this type of malformation, the clinical problem relates to a bruit or to headaches. The risk of hemorrhage or stroke is low since the fistula flows antegrade through the sinuses. Treatment is offered only when there is intolerance of these symptoms. Many patients are surprisingly tolerant of the bruit, especially when reassured that the risk of stroke is small. However, other patients experience serious interruptions of sleep, normal daily activities, and work. Arterial embolization is an effective approach to eliminating or reducing the intensity of the bruit, although the effect may be short-lived. Catheter
Normal Antegrade Flow Through the Transverse-Sigmoid Sinus to the Jugular Vein
62
improvements have allowed for more complete arterial occlusion, but it is important to understand the limitations of this approach and to avoid multiple lengthy arterial embolization procedures in those patients with recurrent DAVF after a similar procedure. A curative approach is taken in those patients with symptomatic recurrence after one or two arterial embolization procedures if the contralateral transverse and sigmoid sinuses can provide venous drainage from the brain. Prior to occluding the sinus transvenously, it must be confirmed that flow in the superior sagittal sinus can enter the opposite transverse sinus; flow in the vein of Labb must also be preserved during the embolization. If flow in the vein of Labb cannot be preserved, or the opposite transverse sinus is small, transvenous embolization should not be performed.
The presence of retrograde flow in the transverse sinus, directed toward the torcula, indicates that the involved sinus can be safely occluded percutaneously without causing a venous occlusive stroke. The contralateral transverse and sigmoid sinuses will provide sufficient venous outflow from the brain. The venous phase of an angiogram of the ipsilateral internal carotid artery generally shows venous drainage to the opposite transverse sinus or anteriorly to the cavernous sinus. Once again, the important factor is the vein of Labb. If transvenous embolization is performed, the vein must not be occluded. Generally, the fistula is located closer to the sigmoid sinus, allowing for the coils in the sinus to occlude the fistula without compromising flow in the vein of Labb. After treatment, the vein drains into the transverse sinus that flows retrograde to the torcula and contralateral transverse sinus. If the vein of Labb is located at the same portion of sinus that contains the fistula, the fistula cannot be safely occluded. In this instance, transarterial embolization may be all that can be offered since surgical resection is likely to also result in occlusion of the vein of Labb. The first step in treating these lesions is transarterial embolization to reduce the flow into the DAVF. Transvenous embolization is then performed during the same procedure to occlude venous drainage from the fistula. Coils are placed in the involved section of sinus, maintaining flow in the vein of Labb. Often the coils are packed down to the jugular bulb.
eral sigmoid sinus. Although the sigmoid sinus may appear occluded, a microcatheter can sometimes be navigated through the apparent occlusion. The vein of Labb usually does not fill because it is either occluded or blood from the posterior temporal lobe drains anteriorly toward the cavernous sinus. The high pressure in the sinus prevents normal drainage from the vein of Labb into the transverse sinus. A more extreme situation arises when the pressure is elevated enough to result in retrograde flow in the vein of Labb from the fistula and cortical venous drainage. The treatment of these lesions is similar to that described above. First, transarterial embolization is performed to reduce flow through the fistula. Transvenous embolization is required to cure this lesion. Coils are placed in the involved segment of sinus draining the fistula. To place the coils, the catheter must either be navigated through the occluded sinus or through the opposite transverse sinus and across the torcula. If the vein of Labb does not fill, it can be sacrificed.
The presence of cortical venous drainage defines these lesions as high risk for causing stroke, so therapy is directed at curing these lesions. Incomplete therapy may provide a risk reduction if the cortical venous drainage is occluded. Transarterial embolization is first performed to reduce flow through the DAVF. Occlusion of the venous drainage is then performed, which requires surgical resection or surgical exposure with intraoperative embolization.
Antegrade and/or Retrograde Flow Through the Sinus with Cortical Venous Drainage
A DAVF that drains entirely toward the torcula is usually a high-flow lesion associated with severe venous outflow restriction or occlusion of the ipsilat-
The fistula present when the involved sinus is thrombosed and the venous drainage of the DAVF is entirely through the cortical veins is perhaps the fistula with the highest risk. This case may represent a progression of the above lesions as the sinus thrombosis progresses to occlusion. Surgery is the definitive therapy for these lesions. Preoperative transarterial embolization may be useful to reduce flow through the fistula and reduce surgical blood loss. However, it is the surgical interruption of the draining veins that provides the cure. Intraoperative angiography is particularly helpful to ensure that all draining veins are occluded. A portable digital subtraction angiographic machine can be used in conjunction with a radiolucent headholder. After the patient is placed under general anesthesia, a sheath can be placed with a continuous infusion of heparin. This sheath is then kept sterile; when an intraoperative angiogram is required, the selective angiogram can be performed rapidly through this sheath. The surgical exposure is determined by whether the veins drain supra- or in-
63
C B
Figure 2. Head position for surgical exposure of a DAVF of the right transversesigmoid sinus. Skin incision (A), craniotomy (B), and dural incisions (C) are shown. The patient is positioned three quarters prone on the operating table.
fratentorially, or both. It is not necessary to remove the sinus since it is already occluded. These lesions cannot be treated percutaneously since the draining veins are too tortuous and thin to safely catheterize and embolize. The authors have encountered a patient who spontaneously thrombosed before surgery. Rarely, a fistula will be identified that drains entirely into cortical veins only, with a normal patent transverse sinus. These DAVFs are treated the same way as above with the exception that the underlying sinus must not be damaged since it is carrying normal venous blood. The vein of Labb may drain in a normal fashion into the sinus and must be preserved when this is encountered.
drops and air embolism may occur. The sitting position increases the risk of air embolism.
The patient can be positioned in a park bench threequarter prone position or in a supine position with the shoulder elevated and the head turned sharply in a headholder. The region that must be accessible extends from the mastoid to the inion. Either of these positions allows for intraoperative angiography if a sheath has been placed in the femoral artery prior to positioning.
As with any surgery around the dural sinuses, the appropriate anesthetic precautions must be taken to treat possible air embolism. Initially, the high pressure in the sinus prevents air from entering the sinus. As the fistula is closed, pressure in the sinus
The incision is generally shaped like a hockey stick, which allows for scalp retraction and access from the mastoid to the inion (Figure 2). Bleeding from the scalp edges can be significant secondary to hypertrophy of either the occipital, superficial temporal, or posterior auricular artery. These arteries should always be embolized preoperatively if they supply the
Scalp Flap
64
fistula. Care is taken to avoid any incision that may result in an ischemic flap. Raney clips are applied to the scalp edges, and the flap is reflected to its base, exposing the mastoid tip. Some of the deep cervical fascia and musculature are separated from the occipital bone with a Bovie. Occasionally, large feeding or draining vessels are encountered that can be coagulated and divided. Venous bone bleeding is easily controlled with bone wax. Fishhooks are then used to retract the flap for adequate exposure.
The craniotomy must be carefully planned to include the necessary portion of the sinus as well as to avoid any catastrophic bleeding. If the patient has had transarterial embolization, flow through the fistula should be significantly lower. Once the extent of exposure has been determined, two sets of burr holes are drilled; one burr hole set is located above the sinus and one set below the sinus (Figure 2). These sets of burr holes are placed at the medial and lateral extents of the craniotomy. A #3 Penfield dissector can then be used to separate the dura and sinus from the overlying bone. The drill can also be used to remove the bone directly over the sinus until it can be visualized through a very thin layer of bone. Once the sinus has been safely crossed, a craniotome can be used to complete the removal of the bone flap. It cannot be emphasized enough that great care must be taken when using the craniotome to avoid laceration of the sinus. The dura should be bluntly dissected off the bone before cutting the bone with the craniotome. A burred bit can be used to remove the bone overlying any area of concern. Pieces of Gelfoam are prepared to be placed over any bleeding site in the dura or laceration in the sinus. The bone flap is then carefully elevated while separated from the underlying dura. Once the flap is elevated, Gelfoam is quickly placed on the dura and sinus and pressure is manually applied. Beginning at the edges, bleeding is controlled as quickly as possible with bipolar coagulation. Gelfoam is then systematically elevated while coagulating using bipolar cautery. This is continued until hemostasis is obtained. During this process, attention must be directed to the patients monitors for signs of venous air embolism. Pressure in the sinus is generally elevated enough as a result of the fistula in addition to the recumbent position to prevent air embolism.
Craniotomy
The dura 1-2 cm above and below the transverse sinus is opened parallel to the transverse sinus. This exposes the occipital lobe and cerebellum. Cortical veins draining the fistula are coagulated using bipolar cautery, clipped with Weck clips, and cut. Usually the vein of Labb is one of the veins draining retrograde and can be safely occluded, decreasing the venous hypertension in the brain. The sinus is then punctured with a catheter, and coils are placed in the sinus over the involved segment. Intraoperative angiography is necessary to ensure that the coils are deposited in the proper area. A large Weck clip, or ligature, is then placed across the transverse sinus between the portion of the sinus involved by the fistula and the torcula. This ensures that there is no retrograde flow from the fistula to the torcula. Preoperatively, it is important to verify that there is no drainage through the superior petrosal sinus because, if left open, it may allow the fistula to continue to drain toward the cavernous sinus. Operative exposure prior to embolization is safest because of the risk of diverting flow to the cortical veins if a purely percutaneous approach is used. As the coils are placed in the sinus during a percutaneous procedure, there is a risk that the outflow of the fistula through the sinus may then be diverted to the cortical veins. These cortical veins are too thin to catheterize and embolize. This excessive flow in the veins could cause catastrophic hemorrhage.
Preoperative transarterial embolization is very effective at reducing blood loss during sinus exposure and may prevent massive hemorrhage. Reports have warned that blood transfusions should be started prior to elevating the bone flap. Blood loss could approach 300 ml/min in some cases. With routine preoperative embolization, the need for transfusion has been significantly reduced and is usually not required. Preoperative embolization is usually performed within 48 hours prior to surgery. The positioning, incision, and craniotomy should be performed in a manner similar to that described for surgical exposure with intraoperative embolization.
SURGICAL RESECTION
After hemostasis has been achieved, the margins of the dura are tacked up to the bone edges with multiple interrupted sutures. These sutures are placed 2 cm apart and serve to provide hemostasis of epidural bleeding and prevent a postoperative epidural hematoma.
After the dura has been opened parallel to the transverse sinus both above and below the sinus, any retrograde-filling cortical veins that are identified can be coagulated, clipped, and divided. Hemostats are then placed across the sinus and the sinus is cut between the hemostats, leaving enough cuff of sinus to close with a running suture. The cuff is closed with a running suture and the hemostat removed.
Sinus Ligation
65
Figure 3. Angiogram of the left common carotid artery, lateral projection, demonstrating a markedly enlarged left occipital artery (arrow) and multiple external carotid and small cavernous carotid artery branches (arrowhead) supplying the left transverse-sigmoid sinus DAVF.
As mentioned previously, when initially opening the dura, the arterialized veins draining the sinus can be sacrificed since the brain is not dependent on venous drainage from these sites. However, normalappearing veins, particularly the vein of Labb if flow is antegrade, must be preserved as well as that portion of the sinus to avoid venous infarction. If the vein of Labb appears to be draining antegrade into the sinus, the sinus must be subtotally resected allowing the vein of Labb to drain through the superior petrosal sinus or the distal sigmoid sinus. The dural vascular supply is interrupted, preserving flow through the sinus. Prior to closure, an intraoperative angiogram can be easily performed to verify complete closure of the fistula. If an intraoperative angiogram is planned, a femoral artery introducer can be placed prior to positioning the patient.
The dura is closed using an allograft dural substitute in a watertight fashion. The bone flap is replaced with titanium microplates and the scalp is closed in a layered fashion.
Closure
Once the sinus has been divided, the next step is to begin dividing the tentorium toward the petrous bone. The hemostat may be used to begin elevating the lateral portion of the sinus from the wound, exposing the tentorium. The tentorium is then slowly incised, controlling any dural bleeding with cautery or Weck clips. As the fistula is encountered, the surgeon must be prepared to control brisk hemorrhage. The fistula is most commonly located near the sigmoid-transverse sinus junction. This is the point at which preoperative transarterial embolization can play a very beneficial role. As resection is carried further anteriorly, the petrous bone and superior petrosal sinus are ultimately encountered and may contain many vascular channels feeding the fistula. Bleeding from these channels can be controlled with cautery, bone wax, or Surgicel, or by drilling the bone with a diamond-tipped burr. The superior petrosal sinus is cut and the dura is reflected off the floor of the middle fossa laterally to the transverse-sigmoid sinus junction. When the fistula has been resected, the residual sigmoid sinus can simply be packed with Surgicel and sutured, taking care to avoid any air embolism. It is usually not necessary to resect the sigmoid sinus since the fistula most commonly involves the sinus at the junction with the transverse sinus.
Fistula Resection
DAVFs involving the torcula are mentioned since they are almost always associated with cortical venous drainage and hemorrhage. Surgery is the best approach and may be preceded by preoperative transarterial embolization to reduce blood flow in the fistula. First, cortical venous drainage from the fistula is occluded as described above. If the fistula drains entirely to the cortical veins and not into the torcula, it must be preserved. If the fistula drains into the torcula, the torcula can be sacrificed only if all venous drainage from the brain is directed away from the torcula toward the cavernous sinus.
Patients are monitored closely in a neurointensive care unit following surgery. Their neurological status is closely observed for any signs that may indicate cerebral edema, hemorrhage, or venous infarction. Depending on the treatment, anticoagulation of any type is avoided to improve the likelihood of fistula thrombosis. A postoperative angiogram can be performed either prior to discharge or on an outpatient basis. Repeat angiography is performed to document the status of the fistula if there is any change in the characteristics of the patients symptoms.
POSTOPERATIVE CARE
ILLUSTRATIVE CASE
As an example of the above treatment discussion, we present the case of a 75-year-old woman who presented with a loud bruit, headaches, ataxia, and cog-
66
Figure 4. Angiograms of the left common carotid artery following arterial embolization. A, lateral projection. Reduction of flow to the fistula with residual arterial supply from the left occipital artery, ascending pharyngeal artery, and intracavernous branches of the internal carotid artery. B, anteroposterior projection. Left transversesigmoid DAVF seen with retrograde venous drainage toward the torcula and contralateral sigmoid sinus (arrow).
nitive changes. An angiogram was performed (Figure 3) which demonstrated a large high-flow left transverse-sigmoid sinus DAVF with supply from the left vertebral artery and both right and left internal and external carotid arteries. The sigmoid sinus appeared thrombosed, with the predominant venous drainage retrograde toward the torcula. There also was some reflux into cortical veins. No vein of Labb was identified. Transarterial embolization was performed to reduce flow through the fistula. The left occipital, accessory meningeal, middle meningeal, internal maxillary, and posterior auricular arteries were selectively catheterized, studied angiographically, tested with lidocaine, and embolized with PVA particles (500-1000 ). Post-embolization angiography (Figure 4) demonstrated marked devascularization of the DAVF, although supply remained from the internal carotid artery and a branch of the occipital artery. As discussed above, transarterial embolization may result in improvement of symptoms but will not cure these lesions due to hypertrophy of remaining arterial feeders. The patient r eturned 2 months later for transve-
nous embolization. No cortical venous drainage was identified at this time. An attempt was made to catheterize the left sigmoid sinus, but it could not be accessed either through the left internal jugular vein, due to occluding thrombus, or across the torcula from the right transverse sinus, due to multiple torcular septations. Also noted during this procedure was some reflux into the straight sinus, which may be an indication that this represented a lesion at higher risk for stroke or hemorrhage. Definitive treatment of this lesion with occlusion of the venous drainage was necessary. This would require surgical exposure of the sinus. The patient underwent surgical exposure of the left transverse-sigmoid sinus as previously described. The left sigmoid sinus was embolized by placing an angiocatheter into the left transverse sinus through which a microcatheter was positioned at the fistula site near the sigmoid sinus (Figure 5). The sinus was then occluded with multiple fiber platinum coils (Figure 6). An intraoperative angiogram was obtained that confirmed sinus occlusion and no residual fistula.
67
retention suture
MEDIAL
A
left occipital lobe LATERAL retention sutures
Figure 5. A, intraoperative photograph showing exposure of the transverse and sigmoid sinuses and angiocatheter positioned in the transverse sinus directed toward the sigmoid sinus. B, line diagram of photograph depicting anatomic structures of the transverse sinus, the left occipital cortex, and the cerebellum.
Figure 6. Intraoperative fluoroscopy image with an angiocatheter (arrow) being used to guide a microcatheter for coil embolization of the left transverse-sigmoid sinus (arrowhead).
68
Figure 7. Postoperative angiogram, lateral projection, of the left common carotid artery injection with no evidence of residual DAVF.
The patient returned 6 months later with complete resolution of her symptoms, and a follow-up angiogram (Figure 7) demonstrated no residual fistula.
Dural arteriovenous fistulas of the transverse-sigmoid sinus can be very complex to treat and require a clear understanding of the arterial supply, nidus location, venous drainage, and location of the vein of Labb. These lesions can present with a wide variety of symptoms and may pose a significant risk to the patient. The venous drainage is the most important factor in assessing the risk of stroke and hemor-
CONCLUSIONS
rhage. Depending on the angioarchitecture of the D A V F , a treatment regimen can be developed which may include conservative therapy, transarterial embolization, transvenous embolization, surgical resection, or a combination of the above. These lesions can only be cured by occlusion of the venous drainage. Based on the fistula characteristics, this can be successfully achieved by transvenous embolization, either percutaneously or after operative exposure, or surgical resection. This approach to DAVFs requires a combined effort from both the neurosurgical and neurointerventional services to provide each patient with the best treatment possible.
OPERATIVE MANAGEMENT OF ANTERIOR FOSSA, SUPERIOR SAGITTAL SINUS, AND CONVEXITY DURAL ARTERIOVENOUS MALFORMATIONS
AMAN B. PATEL, M.D. WESLEY A. KING, M.D. NEIL A. MARTIN, M.D.
Dural arteriovenous malformations (DAVMs) of the anterior fossa are a rare subgroup of malformations that have been only sporadically reported. They account for only 10% of all DAVMs and are distinguished by their high incidence of hemorrhage and unusual anatomy. Lesions associated with the dural convexity and superior sagittal sinus are also unusual, comprising approximately 5% to 10% of DAVMs. Nevertheless, these DAVMs are quite interesting in that they are distinguished by their high incidence of hemorrhage, which is in contrast to the indolent symptomatology associated with the more common lesions involving the cavernous, transverse, and sigmoid sinuses. The pertinent microsurgical anatomy and operative treatment of anterior fossa and superior sagittal sinus DAVMs are reviewed in this chapter.
INTRODUCTION
The anterior fossa dura receives its blood supply from both the internal and the external carotid arteries. In most cases, the internal carotid artery supplies the anterior fossa dura through branches of the ophthalmic artery. The ophthalmic artery usually arises from the supraclinoid portion of the internal carotid
NORMAL ANATOMY
siphon and passes through the optic canal inferior to the optic nerve. After entering the orbit, it gives off a small branch, the recurrent meningeal artery, which runs back into the middle fossa through the lateral part of the superior orbital fissure and may anastomose with branches from the middle meningeal artery. The intraorbital ophthalmic artery also gives off anterior and posterior ethmoidal arteries in the region of the ethmoidal air cells. The posterior ethmoidal artery, the smaller of the two branches, supplies the basal dura in the region of the planum sphenoidale, while the anterior ethmoidal artery serves as the principal branch to the dura of the anteromedial anterior fossa floor. The anterior ethmoidal artery also gives rise to the anterior falx branch that supplies the falx cerebri. After the anterior falx branch, the anterior ethmoidal artery perforates the cribriform plate, and its branches course along the anterior cranial fossa floor and ascend in the falx cerebri. In some cases, the anterior meningeal branches of the anterior ethmoidal artery supply a large portion of the dura of the frontal convexity as well (Figure 1). The principal external carotid arterial supply to the anterior fossa dura comes from the middle meningeal branches, which is also the predominant supply to the frontal convexity dura. These branches may then anastomose with meningeal branches of the ophthalmic artery. Four sites of anastomosis exist between the ophthalmic artery and meningeal branches of the external carotid artery: 1) through the recurrent meningeal branch; 2) through meningeal branches of the posterior ethmoidal artery; 3)
69
70
mid. meningeal branches ant. falx branch ant. ethmoidal a. ant. meningeal branches ant. ethmoidal a.
3 2 1
maxillary a.
Figure 1. Normal anatomy of the blood supply to the anterior fossa dura. The four sites of anastomosis between the ophthalmic artery and meningeal branches of the external carotid are depicted. 1 = Middle meningeal artery to recurrent meningeal artery; 2 = middle meningeal artery to meningeal branch of the posterior ethmoidal artery; 3 = middle meningeal artery to anterior falx branch of the anterior ethmoidal artery; 4 = middle meningeal artery to meningeal branch of anterior ethmoidal artery.
through the anterior falx branch of the anterior ethmoidal artery; and 4) through meningeal branches of the anterior ethmoidal artery. These collateral routes can become prominent in pathological conditions such as meningiomas, DAVMs, and internal car otid artery occlusion. In addition, the external carotid artery can contribute blood supply through transcalvarial branches of the superficial temporal artery and through branches of the internal maxillary artery (Figure 1). Arterial supply to the dura of the superior sagittal sinus and convexity is primarily from branches of the external carotid artery. The most prominent branches that supply this site are the middle meningeal, superficial temporal, internal maxillary, and occipital arteries. The latter three supply the dura via transosseous perforations. Less commonly, there is supply from the anterior falcine branch of the ophthalmic artery or the posterior meningeal branch of the vertebral artery.
In most cases, an intracranial hemorrhage can be seen on computed tomography (CT). Bleeding is either intraparenchymal, subarachnoid, or a combination of the two. In this way, the behavior of a DAVM mimics that of a pure pial AVM. Most frequently, intraparenchymal bleeding is massive and located in the anteromedial aspect of the frontal lobe. Subarachnoid hemorrhage is most often seen in the interhemispheric fissure. Enlarged veins or varices may be demonstrated on contrast-enhanced CT. In addition to demonstrating acute or subacute intraparenchymal or subarachnoid hemorrhage, magnetic resonance imaging (MRI) without contrast enhancement may reveal the draining vein at the floor of the anterior cranial fossa as a flow void, especially in the coronal projection. The presence of venous ectasia or varices may be confirmed. Anterior fossa DAVMs almost universally involve the dura in the region of the cribriform plate and the
Radiographic Features
71
ophthalmic a.
Figure 2. Anterior fossa DAVM is demonstrated in the region of the cribriform plate and the anterior falx. The lesion shows a hypertrophied anterior ethmoidal artery with a fistulous connection to leptomeningeal veins. This connection is dilated into a varix, which is usually the source of hemorrhage. The goal of operative treatment is to divide this fistulous connection.
anterior falx. Usually, a hypertrophied anterior ethmoidal artery supplies the lesion (Figure 2). The anterior ethmoidal supply is bilateral in about one half of cases. Additional blood supply can be obtained from the posterior ethmoidal artery or the anterior falx artery. The external carotid artery may contribute additional feeders, supplying the DAVM through frontal branches of the superficial temporal artery (transosseous) and the anterior branches of the middle meningeal artery. Additionally, there may be supply from the internal maxillary artery. Anterior fossa DAVMs usually drain exclusively into pial veins of the anterior frontal lobe, from which they empty into the superior sagittal sinus or, in rare cases, subfrontally into the cavernous sinus. In patients with hemorrhage from anterior cranial fossa DAVMs, cerebral angiography consistently demonstrates focal aneurysmal dilatation of the venous channel (venous aneurysm or varix) near the site of the dural-to-pial anastomosis. Such a varix
has been described in approximately 90% of cases of anterior fossa DAVM and appears to be the source of hemorrhage. When an aneurysm is not present, the lesion more frequently causes insidious nonhemorrhagic symptoms or is an incidental finding. DAVMs of the superior sagittal sinus are most often located in the middle and posterior thirds of the sinus and are supplied by hypertrophied branches of the external carotid artery. Enlarged middle meningeal arteries and veins may lead to prominent dural grooves on plain skull radiography. In patients with a long history of hyperemia, thickening and sclerosis of the overlying bone may be found. CT may also show associated bony thickening, sclerosis, and enlarged dural grooves. However, CT is most useful for detecting hemorrhagic complications (e.g., intraparenchymal, intraventricular, or subarachnoid) and associated hydrocephalus from prior hemorrhage or increased intracranial pressure (ICP). Contrast enhancement is useful to demonstrate the enlarged
72
draining vein and varices. In cases of thrombosis of the superior sagittal sinus, contrast enhancement may demonstrate the so-called empty delta sign, in which a triangular area of non-enhancement is seen within the sinus, corresponding to clot. MRI will demonstrate acute and subacute intraparenchymal, subarachnoid, and subdural hemorrhage, as well as the flow voids of the enlarged draining veins. Associated superior sagittal sinus venous thrombosis may also be detected. Contrast enhancement may demonstrate the nidus within the leaves of the sinus, confirm sinus thrombosis, and further enhance the pial venous drainage. Cerebral angiography reveals that the primary pedicles supplying superior sagittal sinus and convexity DAVMs are the middle meningeal arteries. Additionally, other branches of the external carotid artery, namely the superficial temporal and occipital arteries, may supply these lesions through transosseous perforations. It is important to carefully evaluate both external carotid arteries, since DAVMs in this location not uncommonly have bilateral symmetric blood supply. Less commonly, there is supply from the anterior falcine branch of the ophthalmic artery or the posterior meningeal branch of the vertebral artery. The most posterior DAVMs may derive some blood from the meningohypophyseal trunk through the arteries of Bernasconi and Casaneri. Pial cortical contributions from branches of the anterior cerebral, middle cerebral, or posterior cerebral arteries may also supply the DAVM. DAVMs may drain directly into the superior sagittal sinus, in which case hemorrhage is rare. Pial veins that normally connect with a dural sinus involved by a DAVM may also be recruited for venous drainage. In these, due to shunting or to sinus occlusion, the high dural sinus pressure causes retrograde flow in the pial veins that normally drain into the sinus. Dementia may be present if pressure within the sinus is elevated in such a way as to prevent normal pial cerebral venous drainage. It is not uncommon to find DAVMs a distance away from the superior sagittal sinus because of drainage into cortical veins which may travel a variable path before reconnecting with the superior sagittal sinus. Again, with DAVMs in this location, venous ectasias (or aneurysms) may be seen; these are notoriously associated with intracerebral hemorrhage.
more urgent. Intracranial hemorrhage is serious and quite possibly is an indication for emergency treatment. Other indications for surgical treatment include severe papilledema that threatens visual function and local cortical dysfunction that may increase without treatment. For example, a large accumulation of cortical veins over a silent area such as the frontal lobe might in time cause serious functional impairment without giving adequate warning. Headache and bruit are symptoms that require individual judgment regarding the circumstances of the particular case. A serious headache may indicate an elevated ICP, or the headache itself may be intolerable (DAVMs cause more prominent headaches than cortical malformations). In addition, a continuous bruit can be quite intolerable and may be an indication for surgical treatment.
Patients presenting with intraparenchymal hemorrhage should be given an anticonvulsant agent. We use Dilantin as the first drug of choice, with an intravenous loading dose of 15 mg/kg, followed by a maintenance dose of 300 mg at bedtime. All patients are also given 10 mg Decadron on arrival to the operating room, which is continued every 6 hours during the course of the operation. Intravenous antibiotics (1 gm Ancef and 80 mg gentamicin) are given on arrival and throughout the pr ocedure and are continued for 24 hours postoperatively. After the induction of anesthesia and insertion of a catheter into the bladder, 100 gm of mannitol is given intravenously; 10 to 20 mg furosemide can be given if additional brain relaxation is required.
PREOPERATIVE PREPARATION
One of the most serious and urgent indications for treatment is the presence of a single cortical draining vein that could rupture. This is especially true if it drains a sinus that has a major obstruction, particularly if it is a sinus cut off from its normal drainage channel. If there is dilatation of the draining vein (venous aneurysm or varix), intervention becomes even
In most patients, there is a decrease in intravascular volume immediately prior to intracranial surgery. Therefore, the replacement of intravascular volume by the intravenous administration of crystalloid or colloid solution is indicated before the induction of anesthesia. Minimal preoperative medication with a benzodiazepine eliminates the risk of narcotic-induced obtundation, hypoventilation, and resultant CO2 retention. Patients with altered levels of consciousness should not receive premedication. Monitoring includes direct arterial pressure, central venous pressure, cardiac rate and rhythm by electrocardiography, temperature, end-tidal CO2 and blood oxygen saturation, urinary output, and intermittent measurement of arterial blood gases, hematocrit, serum electrolytes, and osmolality. A peripheral nerve stimulator is used to measure the degree of muscle relaxation. Venous air embolism is detected by a change in the precordial ultrasonic Doppler signal and the presence of increased end-tidal nitrogen or decreased end-tidal CO2.
Anesthetic Technique
73
Anesthesia is induced with the intravenous administration of propofol or thiopental. Fentanyl (3 to 6 g/kg) or sufentanil (0.5 to 1 g/kg) and vecuronium (0.1 mg/kg) or atracurium (0.5 mg/kg) are added after controlled hyperventilation with 100% oxygen has been instituted. The larynx is intubated after intravenous lidocaine (1.5 mg/kg) and additional thiopental (2 mg/kg) have been administered to prevent systemic and intracranial hypertension during intubation. Anesthesia is maintained with the continuous administration of oxygen, nitrous oxide or air, low concentrations of isoflurane, and intermittent doses or infusion of narcotic, muscle relaxant, and barbiturate or propofol. Ventilation is controlled to keep the arterial pCO2 between 35 and 40 mm Hg before the skull is opened and between 25 and 30 mm Hg after.
General Considerations
Figure 3. The skin incision, craniotomy, and dural opening for the operative approach to anterior fossa DAVMs. This is a unilateral low frontal craniotomy approach. The patient is placed in the supine position with the head maintained in a vertical position and slightly extended. The skin incision is just behind the hairline and extends past the midline to allow exposure farther anteriorly. The craniotomy is performed unilaterally to the superior sagittal sinus.
A unilateral low frontal craniotomy approach allows exposure to the region of dural pathology and is usually undertaken from the side of maximal leptomeningeal venous drainage or from the nondominant hemisphere. A unilateral exposure allows transfal-
Operative Technique
cine access for interruption of ipsilateral and contralateral leptomeningeal venous drainage. This is preferred over a bifrontal approach because it avoids unroofing of a potentially arterialized superior sagittal sinus. In addition, when approached via a low frontal craniotomy, minimal brain retraction is required. The patient is placed in the supine position with careful attention to padding of all pressure points. The head is elevated, slightly extended, and held in the vertical position with the skeletal-fixation headrest. The extension of the head will facilitate retraction of the frontal lobe. A radiolucent headrest may be used if there may be a need for intraoperative angiography. A skin incision is made just above the zygomatic process behind the hairline and extending medially, ending at or just beyond the midline at the hairline (Figure 3). If the incision is too far forward or extends below the zygomatic process, the frontal branch of the facial nerve may be injured. The skin, underlying
74
temporalis muscle, and the pericranial tissue are turned together, exposing the anterior and lateral inferior frontal and anterior temporal bone. A burr hole is placed just below the anterior end of the superior temporal line, which allows exposure of the floor of the anterior fossa. Two additional burr holes may be placed, one in the posterior frontal region and the other in the temporal region. A free bone flap is then turned. The bone flap should extend just off the midline. The lateral portion of the sphenoid wing is removed using the drill. After drill holes are made around the craniotomy opening, dural sutures are placed to control epidural bleeding. The dura is then opened over the inferior frontal and anterior temporal regions. The frontal lobe is protected with Instat and is carefully elevated over the orbital roof with self-retaining retractors. If a large intracerebral hematoma is present, it should be evacuated initially to allow safe retraction of the frontal pole. In most situations, the vascular malformation consists of simple fistulous connections between the dura and the pial veins without a significant dural nidus (Figure 2). The most important step in surgery is occlusion of the vascular connection between the dura of the cribriform plate area and the pial vessels. If the falx or the anterior fossa dura is extensively involved by a nidus, which is rare, it should be excised since residual DAVM can enlarge, recruit pial veins, and thus increase the chance of future hemorrhage. It is probably not necessary to excise the venous aneurysm from the cerebral parenchyma because once the varix is disconnected from the dura it will thrombose and involute.
tions must be carefully considered and planned. In addition to the anatomical characteristics of the DAVM, the clinical status of the patient must be taken into account. In cases where the DAVM has ruptured, operative treatment with or without preoperative embolization must be considered. However, when the DAVM is discovered incidentally or secondary to symptoms not related to hemorrhage, a strong case can be made for palliative treatment to prevent the possibility of hemorrhage. Palliative treatment can also be considered for medically unstable patients or to prevent further bleeding in patients who have hemorrhaged and who are not able to tolerate an open surgical pr ocedure. Embolization of the DAVM is generally the best option for palliation.
The major potential complications include worsening of mental function due to frontal lobe retraction or anterior cerebral artery injury, cerebrospinal fluid (CSF) leak, anosmia, infection, and postoperative seizure. Unilateral or bilateral anosmia is an expected complication of this operation. If the frontal sinus was entered on turning the bone flap it must be cranialized, the mucosa removed, packed with muscle or Gelfoam, and sealed with a pericranial flap. If this is not done, the patient will be at risk for the development of an intracranial infection. Any potential site of CSF leakage, especially in the region of the cribriform plate, must be sealed. We use pericranium and muscle in addition to fibrin glue to seal potential sites of CSF leakage.
Complications
Superior sagittal sinus DAVMs may be treated by a combination of embolization and surgery for complete extirpation. Preparatory embolization may minimize the risk of scalp and dural bleeding during operative treatment of these DAVMs. The surgical treatment of superior sagittal sinus DAVMs is based upon the venous flow pattern determined on preoperative or intraoperative angiography. To minimize venous congestion and edema, the head is positioned so that the DAVM is uppermost in the operative field and is held in skeletal fixation. With DAVMs of the convexity that can be visualized on MRI or CT, the use of a frameless stereotactic system may aid in the positioning, minimal skin incision, and subsequent craniotomy. The planned skin incision must allow for full exposure of the DAVM, adequate removal of the dura surrounding the AVM, and adequate access to any potential bleeding. A wide area of the head should be prepped and draped so that the opening can be enlarged if necessary. Blood supply to the scalp flap must be adequate, and a wide enough base left to provide sufficient vascularization. The cosmetic result of the scar and bone flap should also be considered. For superior sagittal sinus DAVMs located anterior to the coronal suture (anterior third of the superior sagittal sinus), the patient is placed supine with the head slightly extended (Figure 4). A bicoronal skin incision is used. The incision should be placed well behind the posterior limit of the DAVM. The temporalis muscle can be left attached to the superior temporal line on both sides of the head. The pericranium should be left intact to serve as a dural substitute at the end of the operation. For lesions in the middle third of the sagittal sinus, the patient is placed in a supine or semilateral position with the head elevated so that the scalp over the center of the DAVM is uppermost (Figure 5). A horseshoe-shaped incision, with the base broader than the apex, is made. The apex of the flap must extend sufficiently far across
Operative Technique
75
AVM incision
Figure 4. Patient positioning, skin incision, and craniotomy for DAVMs of the anterior third of the superior sagittal sinus. The patient is placed supine with the head slightly extended. A bicoronal skin incision is used.
Figure 5. Patient positioning, skin incision, and craniotomy for DAVMs of the middle third of the superior sagittal sinus. The patient is placed supine with the head elevated so that the scalp over the center of the DAVM is uppermost. A horseshoe-shaped incision is used.
76
osteotomy
incision
Figure 6. Patient positioning, skin incision, and craniotomy for DAVMs of the posterior third of the superior sagittal sinus. The patient is placed in the lateral position and the head is well elevated and turned 45 degrees toward the floor to bring the center of the DAVM to the highest point. The skin incision is horseshoe shaped.
AVM Figure 7. Superior sagittal sinus DAVM following the craniotomy. This shows a malformation at the middle third of the superior sagittal sinus, with a nidus located over the sinus. The DAVM depicted here receives blood supply from bilateral middle meningeal arteries.
the midline to allow adequate bone removal on both sides of the sagittal sinus. For DAVMs located in the posterior third of the sagittal sinus, the patient is placed in the lateral position (Figure 6). The head is elevated and turned at least 45 degrees toward the floor to bring the center of the DAVM to the highest point. The skin incision is horseshoe shaped, extends about 2 cm across the midline, and is based in the
posterior temporal-inferior occipital region. A free bone flap is raised and a sufficient number of burr holes on either side of the sinus are made to allow easy stripping of the dura. A craniotome is used after the dura has been stripped. The dural stripping and bone cuts over the sinus are made last, after all other burr holes have been connected (Figure 7). If the dura is densely adherent over the
77
Figure 8. Drawing showing the arterial supply to the superior sagittal sinus DAVM interrupted following coagulation. This opening of the dura and coagulation of vessels is done lateral to the sinus. In the figure depicted here, this would need to be accomplished bilaterally secondary to the bilateral arterial supply. This is usually sufficient treatment for DAVMs that drain exclusively into the sinus, since the high-pressure arterial flow into the superior sagittal sinus will be interrupted. Leptomeningeal venous drainage, if present, can be interrupted when the dura is opened.
sinus, a free flap should be raised unilateral to the sinus, the dura should be stripped over the sinus under direct vision, and a second free bone flap should be raised across the midline. As the bone flap is elevated, Gelfoam and cottonoids are placed directly over the sagittal sinus. Bleeding from the bone is controlled with bone wax, and dural bleeding is controlled with cautery, Gelfoam, and Surgicel. The dura is then tacked up in the periphery by holes drilled into the bone. Paramedian lesions, which are occasionally associated with an occluded dural sinus and variceal or aneurysmal leptomeningeal venous structures, can be treated by coagulation of the pathological dural leaflet and interruption of leptomeningeal arterialized venous drainage (Figure 8). If a fistula is located in the wall of the sinus and drains exclusively through cortical veins, one should divide the attachment of the fistulous vein to the sinus. It is possible to do this with impunity since cortical blood flow has already found collateral pathways. The wall of the sinus can thus be coagulated without entering the sinus lumen. If the nidus drains directly into the sinus, it is sufficient to interrupt the arterial supply
to the DAVM by coagulating and interrupting the feeding vessels. This will decrease the venous hypertension within the superior sagittal sinus, which is the main source of symptoms. If the sinus is not used as a primary outlet for the brain venous drainage, it may be possible to excise the sinus and not cause significant parenchymal venous hypertension. This is most frequently encountered in the anterior third of the sinus.
Postoperative complications include venous air embolus, hemorrhage, increased neurological deficit, and wound infection. If an air embolus is detected by a reduction in end-tidal CO2 or if air is heard on the precordial Doppler, the patients head can be lowered. Once the opening is covered and the vital signs are stable, the patient can be returned to the original position. CT visualizes the presence of a hematoma or increased cerebral edema. Increased cerebral edema can be treated with an increase in the steroid dose and a prolonged steroid taper. As in anterior fossa DAVMs, openings in the frontal sinus need to
Complications
78
The dura can be closed primarily or by taking a piece of pericranium and sewing it into the dural defect. For anterior fossa DAVMs, closure of the dura in the region of the cribriform plate may be difficult. This area can be packed with muscle or pericranium and subsequently sealed with fibrin glue. Central tack-up sites are drilled into the bone flap and tack-up sutures are placed partial thickness through the dura. The bone flap is replaced and held in place using miniplates, and mesh or burr-hole covers can be used to cover large defects. The temporalis muscle is reapproximated using 3-0 Vicryl sutures in a simple interrupted fashion. The galea is then closed using 3-0 Vicryl sutures in an inverted interrupted fashion. The skin is closed using skin staples or 3-0 nylon in a running interlocking pattern. The wound is then covered with a sterile dressing, and a pressure head wrap is placed.
As the patient awakens from anesthesia, intravenous agents are used to control blood pressure tightly. A portable monitor is used to follow blood pressure and heart rate as the patient is transported to the recovery room or intensive care unit (ICU). Close observation in the ICU setting is indicated for at least the first postoperative night. Steroid doses are usually tapered over 5 to 10 days, depending on the patients neurological status and the extent of cerebral edema. Antibiotics are continued for 24 hours. The length of time for anticonvulsant medication has not been established. If there is no history of seizures, anticonvulsant medication can likely be stopped within 2 to 3 months. If a seizure disorder has been present, anticonvulsants are continued for 6 months to a year, and further treatment is planned depending on electroencephalographic findings. Prior to discharge from the hospital, the patient should undergo cerebral angiography to confirm the absence of residual DAVM.
The surgical treatment of skull base lesions is frequently limited because of extension of the tumor into the bony, neural, and vascular structures around the skull base as well as by the presence of critical neurovascular structures surrounding or invested by the lesion. These limitations have in part been overcome by the use of frameless stereotactic systems, such as the Operating Arm System (OAS) (Radionics, Inc., Burlington, MA). The OAS aids in the preoperative planning of surgical approaches by allowing the surgeon to visualize the operative trajectory to a skull base lesion on the computer graphic workstation. During surgery, the OAS allows the neurosurgeon to visualize the anatomy along the trajectory to the lesion and delineate lesion boundaries, achieving a more complete resection, with potentially decreased morbidity.
INTRODUCTION
room. Frameless stereotactic systems such as the OAS are powerful new tools for preoperative surgical planning and for interactive, frameless stereotactic navigation. The OAS includes an ergonomic digitized arm that senses the position of its probe tip and an integrated computer graphic workstation that stores and manipulates imaging data. The system facilitates planning of the optimal surgical approach and, in doing so, plans the craniotomy trajectory. During surgery, the OAS aids in navigating around critical neural and vascular structures. This improved surgical technique increases the safety, accuracy, and efficiency of skull base surgery, reduces operating time, and may reduce complications from the operation.
The OAS is an advanced neurosurgical instrument used in the operating room for interactive, imageguided surgery. It provides precise, real-time localization of patient anatomy by displaying a point in the surgical field on a preoperative image set, which is viewed on a computer workstation in the operating
BACKGROUND
The OAS is effective for cortical and subcortical lesions, skull base lesions, and deep-seated vascular lesions. The accuracy of the OAS ranges from 1 to 4 mm, which suffices for most neurosurgical procedures. However, target shifts resulting from intraoperative tumor swelling, lesion debulking, retraction, or cerebrospinal fluid (CSF) drainage can increase the error, limiting the usefulness of the OAS in defining the margins of some intra-axial lesions. The OAS is well suited for skull base lesions because target shifts are usually not encountered when a lesion is affixed to the skull base.
PATIENT SELECTION
79
80
Figure 1. A and B, frontal (A) and lateral (B) views of the placement of four reference fiducials on rigid areas of the forehead and one fiducial in the mastoid region respectively. C, the correct attachment of the Operating Arm System to the Mayfield headholder and the optimum placement of the arm so as not to interfere with access to the surgical target. The articulating operating arm should be tested for proper positioning before calibration to ensure proper reach and probe trajectory. Dashed line indicates skin incision. (Reproduced with permission from the Mayfield Clinic)
81
Computed tomography (CT) or magnetic resonance imaging (MRI) can be used for frameless stereotactic surgery. MRI is preferred for skull base lesions when superior resolution of soft tissue is required. CT bone windows are useful for visualizing skull base approaches that require extensive drilling in the temporal bone, occipital condyles, or clivus, such as during anterior or posterior petrosal approaches, far lateral transcondylar approaches, and transoral, transmaxillary, or transsphenoidal approaches.
IMAGE SELECTION
Stereotactic registration is obtained following fixation of the operating arm and skull. Imaging data and registration are performed in the operating room and take approximately 15 minutes. Following registration, the computer workstation calculates and displays the registration error. Error of less than 3 mm is generally acceptable for skull base lesions. The fiducials can be removed after registration so that they will not interfere with the sterile preparation or skin incision.
Stereotactic Registration
Fiducial markers (Radionics, Inc.) are placed on the patients head prior to imaging. For optimal performance, four reference fiducials are applied to rigid areas of the head (i.e., forehead and mastoids) where scalp elasticity is minimal (Figure 1A and B). This application scheme forms a plane passing through the target structure. It is not necessary to tailor the fiducial placement to account for location of the skin incision since stereotactic registration of the OAS is performed before surgery. The fiducials can be outlined with a permanent marking pen to ensure accurate replacement in the event that a marker is inadvertently removed, such as when the stereotactic imaging study is performed on an outpatient basis prior to hospital admission.
The OAS facilitates planning the surgical approach to a skull base lesion by allowing the surgeon to visualize trajectories to the boundaries of the lesion and directing the optimal surgical pathway. The hand-held depth probe is a retractable calibrated probe that allows the surgeon to scan into the depths of a patients anatomy before making an incision and compare various approaches to a lesion (Figure 2).
PRESURGICAL PLANNING
The OAS can be used at all stages of surgery, including skin incision, craniotomy, intraoperative navigation and localization, and lesion resection.
OPERATIVE PROCEDURE
Frameless stereotaxy with the OAS requires threepoint skull fixation, as well as fixation of the operating arm to the head holder. Care is taken during skull fixation to minimize scalp and fiducial movement that would increase error during stereotactic registration. The articulating operating arm should be tested for reach and probe trajectory before registration. The articulating arm must have freedom of movement to allow the probe tip to reach all fiducials and also provide the desired trajectory to the lesion (Figure 1C).
Skin incisions are minimized with the OAS because the incision is placed precisely over the desired craniotomy site. Smaller skin incisions are cosmetically more appealing and, when the scalp and muscle dissection is minimized, can result in less postoperative pain.
Skin Incision
Using the OAS, the craniotomy is customized to provide a corridor to the skull base lesion. The OAS depth probe is indispensable in showing the lesion in relation to a proposed craniotomy. Additionally,
Craniotomy
B
Figure 2. The Operating Arm System depth probe shown fully expanded (A) and fully contracted (B).
82
sigmoid sinus
asterion
transverse sinus
Figure 3. Illustration of the depth probe identifying the transverse and sigmoid sinuses before craniotomy (dashed line). (Reproduced with permission from the Mayfield Clinic)
underlying critical structures and landmarks are identified with the depth probe (rather than with topographical landmarks, which can be inconsistent) during placement of the burr holes and craniotomy. For example, when basing a craniotomy on the sigmoid sinus-transverse sinus junction for a suboccipital or posterior petrosal exposure, the depth probe identifies the exact position of the sinuses, allowing for the safe and accurate positioning of the burr holes and craniotomy and eliminating the search for accurate topographical landmarks (Figure 3). During anterior approaches to the skull base, the OAS shows the corridor to a clival or upper cervical lesion. This allows for maximum exposure through a transoral, transmaxillary, or transsphenoidal approach. The lesion is surveyed with the depth probe before soft-tissue dissection. This interactive presurgical planning helps determine the extent of the
opening required (i.e., inclusion of a transmandibular or transpalatal approach) and maximizes exposure to this region (Figure 4). During transsphenoidal surgery, the OAS helps identify the midline of the sella and the medial extent of the internal carotid artery in the cavernous sinus (Figure 5). The sella opening is configured to the specific anatomy of the lesion.
The boundaries of a skull base lesion are easily detected using the OAS. This is especially helpful when anatomical landmarks are not clearly visible or when the lesion is invested in critical structures. For example, tumor extensions are difficult to identify intraoperatively because they frequently involve the foramina, fissures, and dural folds of the skull base
83
Figure 4. Operating room setup for using the Le Fort approach to a clival chordoma. A, depth probe passing through the presumed transoral opening. B, computer screen of the Operating Arm System showing the location of the tip of the depth probe (green circle) in relation to the target.
Figure 5. Computer screen of the Operating Arm System used to identify the location of the internal carotid artery in a patient with a clival giant cell tumor. Green circle indicates location of the probe tip.
84
Figure 6. Patient with a tentorial meningioma with dural tail. A, preoperative MRI. B, postoperative MRI demonstrating complete resection. C, intraoperative view of the computer screen for the Operating Arm System showing the dural tail (green circle).
(Figure 6A). Using the operating arm probe, the neurosurgeon can more completely resect a tumor by identifying extensions of skull base tumor intraoperatively (Figure 6B and C). Brainstem lesions are visualized with the OAS before opening the surface of the brainstem (Figure 7). This technique minimizes morbidity to cranial nerve nuclei often related to resecting brainstem lesions that do not extend to the surface of the brainstem. The OAS can be used to stereotactically guide surgical instruments, such as endoscopes, into ventricles or cysts, making the OAS an important navigational tool in the operating room for real-time interactive surgery.
As with any frameless stereotactic system, errors can occur while using the OAS during image acquisition, stereotactic registration, and intraoperative navigation. The accuracy of the operating arm is 1 to 4 mm. Errors during image acquisition are due to patient movement. Thin-slice, fast-acquisition images minimize the error associated with patient movement and
CAUTIONS
improve the resolution. Errors during stereotactic registration occur when different parts of a fiducial are selected on the computer workstation and the scalp. Fiducials that have recognizable parts on both the fiducial itself and the image of the fiducial on the computer workstation decrease selection error. Scalp elasticity increases error by displacing the fiducials. Care must be taken to avoid scalp movement when placing the head in three-point fixation. Error during intraoperative navigation results from target shifts during surgery. Target shifts are reduced by avoiding hyperventilation, osmotic diuresis, CSF, and brain retraction. Target shifts are less of a problem for skull base lesions because the lesions are often fixed to the skull base.
The OAS is a useful adjunct to skull base surgery. Although frameless stereotactic technology has not obviated the need for high-level surgical skills and comprehension of the complex anatomy of the skull base, it does provide useful information that can be incorporated when performing a skull base procedure.
CONCLUSION
85
Figure 6C.
nucleus VI
motor nucleus V
Figure 7. Illustration indicating placement of the depth probe to identify the brainstem lesion and location of the cranial nerve nuclei before opening the floor of the fourth ventricle. (Reproduced with permission from the Mayfield Clinic)
The orbitocranial zygomatic approach represents a modification of the classical pterional approach to the pre-pontine cistern that allows maximal exposure of the region of interest. This approach takes advantage of removal and/or relocation of soft tissue and bone to maximize the viewing and working area while minimizing the overall working distance. While the overall dimensions of the pre-pontine space (upper basilar trunk) are fixed by the inherent spatial anatomy, the overall surface area of exposure is enhanced utilizing this cranial base approach (Figure 1).
INTRODUCTION
PATIENT POSITIONING
All surgical approaches begin with patient fixation and positioning. Lesions of the pre-pontine cisterns, being deep and constrained, require three-point pin fixation of the head for maximum stability. Threepoint pin fixation is carried out in such a fashion as to place all pins behind the ears. This allows for maximum soft-tissue mobilization and minimal
87
88
interference of pin(s) to the operators hands; in addition, the transition of instruments between surgeon and nurse and, when necessary, assistants aid is facilitated. Treatment with a pentobarbital bolus prior to pin placement will diminish the patients physiological response to pin placement. Head positioning r equires three degrees of freedom of movement: rotation, approximately 30 degrees from the neutral position, and extension 15 to 20 degrees with neutral head tilt. The ability to rotate the table sideto-side during the operative procedure greatly enhances the surgeons operative flexibility. Prior to pin fixation, a lumbar spinal drain is placed to be utilized if necessary. Care should be taken not to substitute spinal drainage for microdissection to gain operating space. Spinal drainage should remain closed during the extradural drilling so that the cerebrospinal fluid (CSF) can act as a buffer for the brain against inadvertent dural contact. Also, overdrainage of CSF collapses the cisterns, making their dissection more difficult.
4 to 6 cm behind the posterior flap and cutting from the temporal line to the temporal line with a unipolar cautery. The flap is now elevated with further cutting along both temporal lines. Care should be taken not to tear the tissue, especially where it is most adherent at the coronal suture. As the flap is elevated to the level of the supraorbital rims, care is taken to find and preserve the supraorbital neurovascular bundle. It may be in a notch or a true foramen. A foramen can be converted to a notch by utilizing a 1to 2-mm Kerrison to open the foramen or by drilling circumferentially around the foramen. The neurovascular bundle is then brought forward with the pericranial sheet graft. Further forward mobilization of all planes of the soft tissue anterior flap occurs by releasing the periorbitum from the superior and lateral orbital walls, utilizing a combination of #1 and #4 Penfield dissectors. At this point, the anterior flap is well away and below the bone. These maneuvers flatten the exposure, thereby beginning the process of bringing the surgeon closer to the target (Figure 2).
The skin incision is marked out behind the hairline, starting in front of the tragus of the ear, running along the hairline to the contralateral temporal line. First, lidocaine with epinephrine is injected. The initial incision is made down to, but not including, the temporalis fascia or pericranium. Scissors are utilized to protect the underlying soft tissue. Skin clips are applied for additional hemostasis. Attempts should be made to preserve, in continuity, the frontal branch of the superficial temporal artery. Anterior and posterior skin flaps are raised. The anterior flap is raised to the level of the keyhole. At this point, a curvilinear cut in the temporalis fascia is made through both the superficial and the deep temporalis fascia. A cut is started just behind the keyhole down to the muscle. Scissors are placed over the muscle, and the fascia is cut to the root of the zygomatic process of the temporal bone. Two veins are encountered during the transaction, which should be anticipated and controlled with bipolar cautery. The temporalis fascia, both superficial and deep with the fat pad in between, is elevated off the muscle to the level of the zygoma. At the zygoma the fascia splits, with the superficial fascia and fat pad going superficial to the bone and the remaining fascia deep to it. Splitting the fascia is accomplished by gentle dissection with a periosteal elevator starting at the root(s) of the zygomatic process of both the temporal bone and the frontal bone and moving to the zygoma. Once accomplished, the frontal branch of the facial nerve is safely protected and the resulting maneuver exposes the bone from the root of the zygomatic process of the frontal bone to the zygomatic process of the temporal bone. The pericranial flap is now harvested by reaching
The zygomatic osteotomies are now made. These cuts are oriented to maximize the area for temporal muscle displacement. The first osteotomy is made parallel to the zygomatic process of the frontal bone using the posterior bony rim as a cutting guide. A brain spatula should be passed under the zygoma to protect the muscle. The second osteotomy at the temporal root must be made diagonally to reflect the architecture of the bony attachment. Again, the muscle below is protected by a brain spatula. The zygoma itself can be left attached to the muscle on its underside. At this point, the temporalis muscle is taken down. A variety of techniques can be used depending on how reattachment along the temporal line is planned. Generally, a cut is made 1 cm below and parallel to the temporal line to the coronal suture. This will leave a cuff of tissue for reapproximating during reconstruction. A second cut is made perpendicular to the temporal line from behind the pterion to the base of the zygoma. The anterior border of the temporalis muscle is not attached to the bone. Mobilization of the inferior border of the temporalis muscle now occurs. A large diploic vein is encountered in the region of the keyhole. This must be controlled with bone wax packing. The temporal muscle is folded inferiorly over the cut zygoma. Bleeding on the back wall of the muscle should be controlled with precise bipolar cautery. Dissection is taken down to just below the original level of the zygoma. Great care must be taken to obtain and maintain hemostasis to avoid postoperative discoloration down into the neck. Remember that the only remaining blood supply to the temporalis muscle at this point is from its deep arterial supply. Reckless use of cautery can result in
89
Figure 1. Increased view and working space afforded by orbitocranial zygomatic approach. The surface area viewed is enhanced and the working distance shortened with multiple operative corridors (purple = classical pterional approach; blue = orbitocranial zygomatic approach).
Figure 2. Initial soft tissue transfers and bone work begin the process of widening and flattening the operative approach. Care is taken to preserve the neurovascular bundle of the pericranial flap and mobilization of the temporalis fascia to preserve the facial nerve.
90
a dead temporalis muscle. Mobilization of the temporalis muscle in this fashion again leads to a flatter operative site. Bony exposure places the surgeon in front of and below the temporal lobe.
A pterional craniotomy is now per formed (Figure 3). Burr holes are placed: 1) at the base of the zygomatic process of the temporal bone, providing access to the floor of the middle temporal fossa; 2) behind the pterion; 3) at the keyhole (giving access to the anterior cranial fossa and orbit); and 4) just behind the orbital ridge medial to the temporal line. The dura is stripped off the bone with a #3 Penfield dissector. The burr holes are connected with a craniotome, raising a free bone flap. Dural bleeding is controlled with bipolar cautery. The frontal and temporal lobe dura are now elevated off their respective bones. A rongeur is utilized to remove the bone from around the temporal tip, exposing the dura over the temporal lobe superolaterally from temporal pole to root of the zygoma. Elevation of the temporal dura anteriorly would allow exposure of the V2 and V3 roots. Gentle dissection of the periorbitum from the superior and lateral orbital walls is now undertaken. Utilizing a fine cutting blade or reciprocating saw, the orbital osteotomies are performed, protecting the dura and periorbitum with soft, malleable retractors under direct visualization. The extent of the supraorbital osteotomy is tailored to the size of the lesion. Often only a superolateral, supraorbital osteotomy is necessary, sparing entry into the frontal sinus. Likewise, the extent of the lateral orbital cut can be modified. With removal of the supra- and lateral orbital ridges, the periorbitum is further dissected free of the superior and lateral orbital walls, which can then be removed as a single piece with a drill or piecemeal with microdissection back to the base of the anterior clinoid process. Occasionally, a robust anterior clinoid process can project into and limit the oculocarotid triangle. With prepontine lesions, the clinoid process can be safely removed extradurally. A second benefit to removing the anterior clinoid process and opening the optic canal is the ability to intradurally mobilize both the ipsilateral optic nerve and the carotid artery, increasing their mobility and thus opening both the opticocarotid and the oculocarotid corridors to the basilar tip. At this point, great care must be taken to ensure hemostasis of all dural and bony surfaces to avoid continued bleeding during the intradural dissection, which can be a nuisance.
Craniotomy
microsurgical dissection. The dural opening starts low across the frontal lobe, approximately 1 cm above the floor of the anterior fossa, and proceeds down into the sylvian fissure. This maneuver allows the frontal lobe dura to be pleated back on itself and tacked to the pericranium, compressing the orbit. Tacking the dura flat increases the flatness of the approach while decreasing the necessity for retraction and CSF diversion. The dura is then cut around the front and side of the temporal lobe, again tenting over the adjacent soft tissue covered by Oxycel, further flattening the approach and improving hemostasis. A final cut up the sylvian fissure completes the dural opening. At this point, the entire temporal lobe from tip to the root of the zygoma is exposed, with visualization of the superior, middle, and inferior temporal gyri (Figure 4A).
The dura is now opened. To protect the brain during bone work, the spinal drain has not been opened to this point. The drain remains closed to keep the cisterns and sylvian fissure expanded during the initial
The microscope is now brought onto the field; under microscopic magnification and illumination, the sylvian fissure is opened. Again, it is important to not open the CSF drain at this point. This keeps the fissure and cisterns expanded with CSF and easier to dissect. Following Yasargils technique, the fissure is dissected by opening the arachnoid and dissecting down to the artery; using the gentle opening pressure of the bipolar cautery to dissect, the arachnoid is opened deep to superficial. Sharp dissection with scissors is used for the most superficial arachnoid, if necessary. Furthermore, it is suggested that no retractors be placed at this time. While use of retractors may be tempting, they tend to immobilize the brain and limit the freedom and visualization of dissection. There is also great potential risk of over-retraction leading to brain injury in a still tethered brain. Microdissection is carried down to the carotid cistern, which is opened. The optic nerve is freed from the frontal lobe, and the A1 segment is untethered. Dissection is now directed toward releasing the uncus and opening the space lateral to the carotid artery. The posterior communicating artery is freed and the third nerve is identified and freed as it enters the tentorial edge. At this point, the temporal lobe has been released circumferentially. If necessary for maximal working space, the veins entering the sphenoparietal sinus are bipolar coagulated and cut. The temporal lobe can now be mobilized up and out of the middle cranial fossa and held by a small amount of traction via a brain blade. The difference between traction and retraction is not a matter of mere semantics. I define retraction as the application of a pulling force against the brain. Traction, on the other hand, simply contours an object in place without significant pulling force. With complete microdissection of the sylvian fissure, the brain blade can be placed to position the entire temporal lobe in space. The blade is brought in
91
Figure 3. Bone work. A, pterional craniotomy, tailored orbital osteotomy, and tailored zygomatic osteotomy. B, removal of the superior and lateral orbital walls permitting an unobstructed view to the tentorial edge.
A
Figure 4. A, with soft tissue transfer and bony removal, a wide view of the temporal lobe and frontal/subfrontal region is possible. The dura is opened as demonstrated by the dotted line and tented back against the soft tissue with some compression of the periorbitum. B, the view down the orbitocranial zygomatic approach with its fullest exposure. The temporal lobe being freed from the middle cranial fossa is positioned up and out. The third nerve is the pathway to the brainstem and the basilar artery. Mobilization of the carotid artery and optic nerve can enhance the operative space.
92
Figure 5. Operative view through the orbitocranial zygomatic approach to the upper basilar trunk and bifurcation.
anteriorly to the temporal tip and slightly medially. The untethered temporal lobe will assume a position allowing a corridor for visualization and work of approximately 4 to 5 cm (Figure 4B). The head and microscope are now positioned to bring the third nerve into the middle of the field. Utilizing microsurgical dissection, the arachnoid along the nerve is cut and the nerve is followed back to the brainstem. The third nerve acts as an approach vector to the basilar artery. The basilar, superior cerebellar, and posterior cerebral arteries can all be identified. The posterior communicating artery is followed along its entire course with great care to preserve all of its perforators. The posterior clinoid process is identified just medial to the third nerve. At this point, one has proximal control of the basilar artery. Dissection is carried along the posterior cerebral artery to the basilar junction (and aneurysm). By rotating the table, a more subtemporal view is obtained and dissection of the back wall and perforators is performed. Rotation of the scope allows a view through the opticocarotid and oculocarotid triangles for visualization
of the contralateral posterior cerebral artery. The approach thus maximizes the viewing angles and operative corridors, minimizing the limitations of the traditional approaches (Figure 5).
A number of adjuncts can be applied to cope with variations in aneurysm size, geometry, and position along the clivus. At times, the perforators off the posterior communicating artery, in combination with a foreshortened length, can hinder advancement of the clip. Injury of a posterior communicating artery perforator should not be taken lightly as it can lead to hemiparesis. Transection of the posterior communicating artery at its most proximal perforator (as related to the carotid artery) will allow the entire complex to be rotated back out of harms way. This permits further mobilization of a now laterally untethered carotid artery, expanding the oculocarotid triangle. At times, with a low-riding basilar artery, the pos-
93
terior clinoid artery can become an obstacle, especially with regard to clip trajectory. With the middle cranial fossa empty, drilling of the posterior clinoid becomes less onerous as it allows for the use of shorter drills as the hand is brought closer to the target. The dura over the posterosuperior aspect is bipolarly cut and gently dissected forward, exposing several millimeters of the bone, which can then be drilled. Overzealous takedown of the bone or dura can lead to cavernous sinus bleeding, which can be robust. Bleeding is generally controlled by packing with a hemostatic agent, which can lead to a greater obstruction than the primary anatomy. Generally, no more than 3 to 5 mm of bone needs to be exposed and drilled. Great care must be taken not to injure the third nerve during the drilling process. For low-lying aneurysms or large aneurysms that require lower basilar artery control, the third nerve can be mobilized and the tentorium split. The third nerve is visualized entering the tentorium. With the aid of a scythe blade or a #11 bladed knife, the dura over the course of the third nerve is opened for approximately 1 cm. Care is taken to stay directly over the nerve while transecting the dura. Again, entry into the cavernous sinus can occur and should be controlled with minimal packing with a hemostatic agent and pressure with a cottonoid patty. The third nerve may now be mobilized forward or back. The tentorium is then split behind the third nerve, taking care not to injure the fourth nerve which is situated behind it. To avoid injury, the fourth nerve must be identified prior to this maneuver and its trajectory projected. The tentorial edges can then be tented back with 6-0 retention sutures to give a V-shaped corridor to the basilar artery. A more sophisticated adjunct to expose the basilar trunk is the anterior petrosectomy, or the Kawase approach. This extradural approach relies on the identification of key temporal bone anatomy and requires an intimate knowledge of the temporal bone anatomy; it is not for the uninitiated. Complications from this approach can include carotid artery injury, deafness, and facial paralysis. The reader is referred to the works of Kawase for a detailed description. A full description is beyond the scope of this chapter and is mentioned here for completeness. Radiological adjuncts to basilar artery surgery include intraoperative balloon proximal control and angiography and preoperative computed tomography (CT) angiography. Preoperative CT angiography is performed utilizing spiral CT with venous injection. Data collection takes approximately 60 seconds with reformat times of 5 to 10 minutes. Studies can be reformatted in: 1) simultaneous axial, coronal, and sagittal planes with point-to-point tracking; 2) multiplanar volume rendering; and 3) three-dimensional
rendering. The vessels of interest can be followed with a cursor along their route and cross-referenced in all three planes. The precise anatomical relationship of the aneurysm geometry and spatial orientation to its osseous environment are well visualized. Calcification of the neck can be readily identified. Most importantly, aneurysm neck geometry can be rendered from multiple views including inside of the aneurysm. Intraoperative angiography may be critical for large and giant basilar apex aneurysms to assure major vessel patency (particularly contralateral posterior cerebral and superior cerebellar arteries) and aneurysm obliteration. Remember that perforator vessel integrity is beyond angiographic resolution and must be assured by careful postclipping inspection. To assure intraoperative angiography when needed, we place the catheter in the parent artery of interest preoperatively with a slow heparin drip through the catheter. In our series of 100 intraoperative angiograms, constituting 425 injections, we have experienced one embolic event utilizing this protocol. Key to assuring obliteration is to take multiple views, reproducing those views that best demonstrate the lesion preoperatively. Intraoperative balloon proximal control can also be of help in dealing with large and giant basilar apex lesions. This maneuver avoids placement of a proximal clip, which can obscure visualization of critical structures. Retrograde suction can collapse or at least soften the aneurysm, allowing for final dissection and clip application. Even with proximal endovascular control, the surgeon is encouraged, if possible, to prepare for proximal clip control should the balloon fail. Retrograde suction can lead to large volumes of blood loss if the collateral anterior circulation is robust. Utilization of a cell-saver to reconstitute the blood volume should be considered. This approach and adjuncts can be utilized to reach aneurysms of the basilar apex, P1 and P2 segments of the posterior cerebral artery, and the superior cerebellar artery. Aneurysms below the upper one third (midbasilar) of the basilar artery are better approached petrosally. Occasionally, a very low-lying short basilar artery exists with an apex aneurysm whose dome is at or below the posterior clinoid artery. These lie anatomically in the midbasilar region and should be approached petrosally. Manipulation of the third nerve will lead to palsy, with the patient generally recovering over 4 to 6 weeks. The surgeon should be prepared for the presence of a unilateral dilated pupil in the recovery room. Generally, this can be determined clinically. CT can confirm if there is any doubt. Intraoperative angiography usually relieves the surgeon of a vascular catastrophe in these cases.
94
Wound closure begins with opening the lamina terminalis. This may aid in reducing the incidence of postoperative hydrocephalus in a patient with subarachnoid hemorrhage. The brain blade is removed. Generally, the view without the blade is adequate (Figure 5). It should be stressed that the blade need only be placed for the moments before and during clipping when a static field is required. Generally, I perform a duraplasty with a generous piece of pericranium to provide nonconstricting dural closure. Just prior to closure, the intradural space is filled with warm saline to check for a watertight closure and to evacuate air. If the frontal sinus was entered, the mucosa is cauterized and the sinus packed with an antibiotic-soaked hemostatic agent. A small piece of temporalis muscle is then packed into the opening into the sinus. Epidural tenting sutures are placed circumferentially over the hemostatic agent along with a central tenting suture. The pericranial flap is then pulled over the sinus opening packed with muscle and sutured to the dura. A sheet of bacteriostatic hemostatic agent is cut and placed over the superior and lateral orbital bone defect prior to tacking down the pericranium. If the sinus is not entered, the pericranium is laid back down in its anatomical position. The supraorbital bone is now reapproximated to the pterional free flap by a microplate and screws on the underside. This construct is reapproximated to the cranium with microplates and screws. It is possible to plate with all plates hidden by the temporalis muscle. If further cosmesis is required, hydroxyapatite bone cement can be used to fill the bone cuts, burr holes, and temporal craniectomy site. If used, one must have the strictest hemostasis, leave the area well drained, and allow the material to set before continuing the closure (20 to 40 minutes). The temporalis muscle is now brought back into its anatomical location. Care should be taken to secure it high above the keyhole to avoid a cosmetic defect. A wire-passer drill hole at the junction of the zygomatic process of the frontal bone and supraorbital ridge can facilitate this maneuver. If possible, the temporalis fascia should be reapproximated anatomically. The zygoma is now reattached with a microplate and screws. With
reattachment, care should be taken not to constrict the base of the temporalis muscle; this now represents the primary blood supply of the muscle. Postoperative swelling against a constricting zygoma can lead to loss of the temporalis muscle. Critical inspection of the skin flaps is now undertaken with a goal of maximum hemostasis. A drain is placed under the posterior skin flap, using grenade suction, not as a substitute for hemostasis but to encourage obliteration of the large subgaleal dead space. A galeal closure of inverted suture is placed, followed by skin staples. Blood loss for this exposure (excluding retrograde suction or aneurysm rupture) is estimated to be approximately 100 to 150 cc. The spinal drain is removed prior to extubation, and the subgaleal drain is removed in 24 to 48 hours. If there is no facial nerve function, return is generally seen in 1 to 6 weeks; however, it may be hard to determine immediately postoperatively. Any forehead and scalp numbness noted by the patient should resolve and is a result of manipulating the supraorbital neurovascular bundle. Eye swelling and ecchymosis are rare if the periorbitum has not been violated. It either occurs, an ice bag to the eye appears to control and diminish the consequences. Because of the lower, wider, flatter operative field, the surgeon will find that long-bayoneted instruments are cumbersome and no longer necessary. A mixture of short-bayoneted and straight bipolars of variable lengths can now accommodate the dissection. With experience, the additional osteotomies will add approximately 20 minutes to the approach and closure times.
It is always asked, Is all of this necessary? Yasargil never did this. Drake never did this. Sugita never did this. Which one of us is a Yasargil, Drake, or Sugita? The concepts and techniques in this chapter are not my own but ones that have been learned from the master surgeons willing to share with their students. I humbly acknowledge and am grateful to Drs. Gazi Yasargil, Ossama Al-Mefty, and Evandro DeOliveira for sharing their experience and wisdom.
CONCLUSION
Neurinomas arising from the trigeminal nerve are rare tumors that occupy the cavernous sinus and parasellar regions. These tumors typically cause a variable degree of trigeminal dysfunction and sometimes diplopia. This condition necessitates surgical intervention. No benefit has been demonstrated by other therapies, including radiosurgery. These tumors are divided into four types and classification is by their location. The first two types are not addressed in this chapter. The first type includes tumors arising from the trigeminal nerve root, contained within the posterior fossa. These are best approached via a retrosigmoid approach. A second type of tumor, the dumbbell type, spans the posterior and middle fossas. These masses involve both the nerve root and the ganglion, and, when very large, may extend peripherally. This tumor requires a combined petrosal type of strategy and is not addressed here. This chapter is concerned with the surgical treatment of the ganglion and peripheral types of neurinomas of the trigeminal nerve. The large majority of ganglion-type tumors may be approached via an
INTRODUCTION
entirely extradural temporopolar approach. Ganglion tumors located laterally may also be approached via an extradural subtemporal transpetrosal approach. Peripheral neurinomas are handled by one of these extradural strategies or by a variation. The key concept is that these two types of neurinomas are located extradurally and that a contemporary cranial base strategy can result in total resection with low morbidity. An important consideration in the surgical approach to these tumors is determining an appropriate surgical strategy for extradural bone removal. The preoperative imaging studies must be carefully reviewed to determine the exact location of the mass. The necessary degree and locations of extradural bone removal are then determined for the individual patient. In the case of a ganglion tumor, for example, a lateral location will not necessitate removal of the anterior clinoid process and decompression of the optic canal (Figure 1A-C). Also, laterally located tumors approached via an extradural subtemporal transpetrosal approach may not require full petrous apex removal for adequate exposure (Figure 1D-H). This is mainly determined by the size of the tumor and the presence of any inferior extension of tumor eroding the petrous bone or involving the carotid canal. Anterior and medially located tumors may not require unroofing of the foramen ovale or rotundum for adequate exposure of the tumor (Figure 1I). These tumors may be approached adequately by unroofing
95
96
Figure 1. A, illustration of an anterolaterally located ganglion tumor. B, axial MRI of a small ganglion tumor that would not require optic canal unroofing and anterior clinoid resection. C, coronal view of the same neurinoma. D, illustration of a laterally positioned neurinoma approached via an extradural subtemporal approach. E, axial MRI of a small lateral trigeminal neurinoma.
97
J
Figure 1. F, intraoperative photograph of an exposed pigmented neurinoma located between V2 and V3. (Photo courtesy of T. Fukushima, M.D.) G, exposure following removal of a well-encapsulated tumor. H, pigmented trigeminal neurinoma. I, anterior and medially located tumor not requiring unroofing of foramen rotundum or ovale. J, extensive tumor requiring global exposure of the cavernous sinus.
98
the optic canal, the superior orbital fissure, and the anteromedial triangle via removal of the anterior clinoid process. Large tumors involving the entire cavernous sinus region will require a global strategy in order to optimize extradural bone removal (Figure 1J). This process of strategy development is important so as to limit potential morbidity to the pericavernous structures.
The patients medical history is obtained and a detailed neurological examination completed. In particular, trigeminal nerve function is carefully tested and documented. All sensory modalities are tested and the intensity of sensation as compared to a normal area is graded using a 10-point scale. This is important in terms of determining postoperative improvement in sensory function. The corneal reflexes are tested with a cotton wisp. Patients with any sign of keratitis are referred to an ophthalmologist for consultation. The majority of patients will be referred with a diagnostic imaging study, typically magnetic resonance imaging (MRI). Patients in whom the anterior transpetrosal approach is indicated will undergo imaging work-up supplemented by a fine-cut bone window computed tomography scan. Particular attention is directed toward the relationship between the cochlea, the internal auditory canal, and the carotid canal in interpreting this study. Measurements may be taken from the images to provide the surgeon with an accurate concept of the operative window that will be created through petrous apex removal. This study is also important in cases where frameless stereotaxis will be used as an adjunct. Patients undergo the usual preoperative laboratory blood work, electrocardiogram, and chest x-ray. Medical clearance is obtained when warranted by any chronic medical conditions such as diabetes or hypertension. Patients are administered anticonvulsant medication in the operating room, typically diphenylhydantoin at a dose of 10 mg/kg. For a short period, the patient receives a maintenance dose of 3-5 mg/kg/day. In the majority of cases involving an entirely extradural surgery, anticonvulsant prophylaxis is stopped on the seventh to 10th postoperative day if the patient has not had any seizures. All patients are given antibiotics perioperatively. A broad-spectrum cephalosporin is used in the majority of cases. Vancomycin is given to patients who have exhibited an allergy to penicillin and its derivatives.
PREOPERATIVE PREPARATION
to two large-bore intravenous lines, some patients require central venous access. Preferably, this is placed in the subclavian vein. Several maneuvers are important in terms of managing the intracranial pressure during the procedure. Selected patients undergoing an anterior transpetrosal approach have a lumbar drain inserted for drainage of cerebrospinal fluid during retraction of the temporal lobe. This is especially important in the older patient who may not tolerate any degree of temporal lobe retraction. In the younger patient in whom there is usually an adequate degree of cerebral compliance, I do not insert a drain. The end-tidal CO2 is monitored and maintained between 25 and 30 mm Hg. Furosemide, 10-20 mg, and mannitol, 25 gm/kg, are also administered intravenously at the time of cutting the cranial bone flap to initiate a diuresis. These maneuvers are directed at providing a compliant brain, although it will be retracted only sparingly and under the protection of the overlying dura. When performing the anterior transpetrosal approach, facial nerve monitoring is a useful adjunct. The nerve monitor is utilized to assist with identifying the geniculate ganglion when located deep to the bone. Otherwise, no specific neurophysiological monitoring modality is of particular benefit in these cases.
The patient is placed supine on the operating table and the table is couched, elevating the back and the knees. All bony prominences are well padded with foam. Sequential compression boots are placed on the patients legs and the device is activated. The head is placed in the Mayfield three-pin headrest with the two-pin side on the side of the approach, which is the dependent side. The single pin arm is placed in the area of the midpupillary line on the contralateral side, within the hairline. The head is turned approximately 45 degrees toward the opposite shoulder. The vertex is placed in the neutral position. Dropping the vertex of the head in these procedures is not suggested, as this is an operation where the surgeon is looking down onto the middle fossa and cavernous sinus areas. The head is draped in the usual fashion for a frontotemporal craniotomy.
Operative Positioning
Patients are administered general anesthesia via inhalation agents. All patients have a radial arterial line placed to monitor blood pressure. In addition
ANESTHETIC TECHNIQUE
The scalp incision is made beginning just anterior to the tragus of the ear, over the root of the zygomatic process. The incision continues superiorly, within the hairline, gently curving to end at the midline (Figure 2). Scalp clips are applied to the margins of the scalp, incorporating galea, for hemostasis. The galeocutaneous flap is then elevated, utilizing an interfascial
Procedure
99
Figure 2. The extradural frontotemporal temporopolar approach. The patient is positioned with the head turned approximately 45 degrees (A). The scalp incision and craniotomy are similar to that of a routine pterional-type approach (B).
dissection over the temporalis muscle. This will result in rotating the temporalis muscle and fascia inferiorly and posteriorly, clearing out the frontozygomatic recess and providing a wide, flat access to the middle fossa. The galea is separated from the pericranium medial to the superior temporal line, taking care to preserve the supraorbital nerves with the galeal layer. At the superior temporal line, the fascia innominata (loose connective tissue between the temporalis fascia and the galea) is incised from its connection to the medial pericranium. This fascia is then elevated with the galeal layer, leaving bare temporalis fascia. This maneuver places the surgeon in the correct plane to separate the superficial and deep components of the temporal fat pad. The superficial fat pad layer contains the frontalis branches of the facial nerve, which are protected when this maneuver is properly performed. Scalp elevation proceeds until the supraorbital and lateral orbital rims are exposed, covered by the periosteum. Inferiorly, the periosteum over the zygomatic arch is exposed. The scalp is held with large blunt scalp hooks and rubber bands. A vascularized pericranial flap is preserved, incising along the margin of the scalp incision and the superior temporal line. The flap is then raised via subperiosteal dissection and wrapped in moist gauze. The pericranial flap is held anterior with the scalp flap using large blunt scalp hooks. This may be used later for dural repair or coverage of an opened air sinus. The temporalis fascia is separated at the superior temporal line using a sharp instrument. Periosteum
over the lateral orbital rim and zygomatic arch is incised next, such as to leave a cuff of tissue for later reattachment. The periosteum is then elevated toward the temporalis muscle, freeing the fascia of its anterior and inferior attachment. The temporalis muscle is then elevated from the frontozygomatic recess and the temporal squama via subperiosteal dissection. This is done cold in order to avoid damage to the neural or vascular supply to the muscle; the result will be superior cosmetic outcome without loss of muscle bulk and tone. The muscle is held inferiorly and posteriorly using large blunt hooks. The frontotemporal craniotomy is then cut using a high-speed drill. Burr holes are placed over the pterion and in the low temporal region. The craniotomy is roughly centered one third above and two thirds below the superior temporal line. The bone flap typically measures roughly 6 4 cm. After removal of the flap, hemostasis is secured and the posterior dural margins are tacked up to the bone margin with fine suture. The dura is elevated away at the anterior temporal margin with a dissector, and any remaining bone overhanging the middle fossa is rongeured away. Extradural removal of bone at the anterior and temporal base is the next step in the procedure, and is performed with a high-speed drill. The dura is elevated away from the sphenoid ridge, and self-retaining retractor blades are placed to hold it away (Figure 3). The sphenoid ridge is reduced until a smooth contour of the superior and lateral orbit is created. The
100
Figure 3. Elevation of the dura exposes the neural foramina at the middle fossa base and the superior orbital fissure.
Figure 4. The defining maneuver of the extradural approach for neurinomas is separation of the lateral wall of the cavernous sinus.
Figure 5. Tumor is exposed and resected utilizing the various entry corridors to the cavernous sinus.
101
bone of the middle fossa is shaved down toward the periosteum forming the roof of the infratemporal fossa, leaving the outer cortical table intact where possible. The dura is further elevated toward the temporal base to expose the superior orbital fissure, foramen rotundum, and foramen ovale. From this point of the procedure, the structures exposed via bone removal are tailored specifically to what is required by the anatomy of the individual lesion being treated (as outlined above in the Introduction). At a minimum, bone must be removed over the superior orbital fissure, foramen rotundum, and foramen ovale. In selected cases, the anterior clinoid process will be removed and the optic canal unroofed. Beginning at the superior orbital fissure, the cleavage plane between the temporal dura propria and periorbital fascia is separated using sharp dissection technique. A retractor blade is placed over the temporal pole and pressure is directed in the posterior direction. This cleavage plane is developed and opened over V2, V3, and the ganglion. At this point of dissection, the tumor becomes readily evident. Usually, trigeminal nerve fibers are spread and flattened over the capsule of the tumor. Separation of the dura propria from the connective tissue covering of the nerve (i.e., the outer cavernous membrane) continues toward the incisural edge. This is the defining maneuver of the extradural approach to these tumors (Figure 4). The tumor and the cavernous sinus are exposed without opening the dura and exposing the underlying temporal lobe (Figure 5). Additionally, the temporal tip bridging veins are left undisturbed. Resection of the tumor now proceeds utilizing the various triangular entry corridors to the cavernous sinus. The routes accessible by this approach are the anteromedial, anterolateral, far lateral, posterolateral, lateral, superior, and medial. Typically, the anterolateral, far lateral, and lateral triangles will be the most frequently employed for ganglion tumors. Standard principles of tumor resection are used in these procedures. The tumor capsule is identified and opened, usually exposed via the anterolateral or far lateral triangle. Sometimes, fibers of the trigeminal nerve are separated to expose the capsule. After opening the capsule, generous internal debulking is performed. This can be adequately performed with sharp dissection techniques and the use of microring curettes. After internal debulking, the plane between the trigeminal nerve and tumor capsule is developed with sharp microdissectors, first working within one triangular corridor. This same maneuver is then performed in the various triangular entry corridors used, pushing tumor and capsule toward the primary entry corridor with soft cottonoid patties. Any cavernous bleeding is controlled with Surgicel packing. Usually, cavernous sinus bleeding is not particularly problematic in these cases.
After tumor resection, the wound is closed. The primary goal of closure in this approach is r econstruction of the skull base. Any openings made in the dura must be reconstructed in watertight fashion. Usually, adipose and fascial grafts are not necessary. However, procedures in which an air sinus has been opened through the course of extradural bone removal will require adipose and fascial grafting for adequate closure. Any opened air sinus must be exenterated of its mucosa and packed with fat or muscle to prevent infection and mucocele formation. The bone flap is then reapproximated with stainless steel wire or titanium microplates for best cosmesis.
Wound Closure
Operative Positioning
Procedure
102
B
Figure 6. The extradural subtemporal transpetrosal approach. The patients head is positioned laterally (A). The incision and craniotomy are shown (B).
scalp incision. The muscle is cleared away from the root of the zygoma and pulled anteriorly. The temporalis muscle is held anterior and posterior with blunt scalp hooks. When a larger craniotomy is necessary, a question mark-shaped incision is made (Figure 6). The incision extends superiorly to the level of the superior temporal line. After applying scalp clips to the galeocutaneous margins, the scalp is elevated from the temporalis fascia. This requires an interfascial dissection, splitting the superficial and deep components of the temporal fat pad in order to protect the frontalis branches of the facial nerve. The key to this dissection is identification of the loose areolar layer of connective tissue, called the fascia innominata, between the galea and the temporalis fascia. The fascia innominata is elevated with the galea, opening the appropriate dissection plane to split the components of the fat pad. The temporalis fascia and muscle are then incised in the line of the scalp incision with unipolar cautery. The muscle is then elevated via subperiosteal dissection from the temporal squama and elevated anteriorly, held by blunt scalp hooks. Again, it is important to free the muscle from the zygomatic root and pull it anteriorly to maximize exposure. A temporal craniotomy is then cut, centered two thirds anterior and one third posterior to the external auditory canal. The craniotomy needs to measure no more than approximately 4 4 cm. A highspeed drill is then used to reduce any remaining
bony overhang of the middle fossa floor. A flat viewing angle across the floor is necessary in terms of limiting temporal lobe retraction. The dura is elevated from the middle fossa floor with a fine dissector, beginning over the petrous ridge (Figure 7). Initiating dural elevation posteriorly is important in terms of avoiding traction on the GSPN, which may lead to postoperative facial nerve dysfunction. The dura is elevated medially until the lateral margin of the trigeminal impression is exposed. Elevation in the anterior direction separates the dura propria from the periosteum covering the GSPN, lying in the major petrosal groove. Elevation continues anteriorly until the middle meningeal artery is exposed at the foramen spinosum. The artery is coagulated and divided. The dura is further elevated to expose the foramen ovale and V3. Two tapered self-retaining retractor blades are then placed on the dura over the temporal lobe. The cleavage plane between the temporal dura propria and the connective tissue sheath over V3 is next identified and developed sharply. The dura propria is separated from V3 and the lateral portion of the gasserian ganglion, similar to the temporopolar approach (Figure 8). This maneuver exposes the trigeminal fibers stretched over the tumor mass. In some cases, removal of bone at the petrous apex may be necessary for full exposure and removal of the tumor. To safely remove the bone of the petrous apex via the extradural route, sequential dissection of the
103
Figure 7. Elevation of the middle fossa dura exposes key landmarks of the middle fossa floor.
Figure 8. Separation of the dura propria from the outer cavernous membrane provides exposure of the posterolateral cavernous sinus. The exposure is now adequate for petrous apex removal, if indicated.
Figure 9. Petrous apex removal increases exposure into the posterior cavernous sinus, down to the clivus and entrance of the sixth cranial nerve into Dorellos canal.
104
middle fossa floor is critical (Figure 9). It is helpful to consider the key landmarks of the middle fossa floor as a guide to bone removal. Four landmarks outline the rhomboid-shaped volume of bone which may then be removed and which is devoid of neural or vascular structures. These landmarks are: 1) the GSPN-V3 junction, 2) the porus trigeminus, 3) the arcuate eminence-petrous ridge junction, and 4) the junction of the axes of the GSPN and the arcuate eminence. Obliquely projecting this rhomboid-shaped complex through the bone toward the clivus delimits the volume of bone removed to the level of the inferior petrosal sinus. Radical petrous apex removal is seldom required in these patients, and the approach must be tailored for the individual patient. To begin, the intrapetrous carotid artery is exposed by unroofing the carotid canal in Glasscocks triangle. This triangle is defined by the foramen ovale, the cochlea, and the V3 origin. It is easiest to begin drilling near V3, just anterior to the GSPN. The artery is followed laterally until the crossing point of the tensor tympani muscle is identified. This marks the genu of the petrous carotid artery. Next, the internal auditory canal is unroofed. This is done by first drilling near the petrous ridge over the bisection axis of the angle between the GSPN and the arcuate eminence. This axis is roughly 60 degrees from either the GSPN or the superior semicircular canal. With the internal auditory canal and the petrous carotid exposed, the bone between these two structures may then be quickly removed with the drill. The majority of this bone is vascular cancellous bone that can be easily drilled away. The bone under the gasserian ganglion may also be removed by thinning with the drill followed by removal using a dissector. The tumor is then resected by utilizing the lateral triangular entry corridors to the cavernous sinus. Most typical via this approach are the far lateral, posterolateral, and posteromedial triangles. The same principles apply as for the temporopolar approach, outlined above. This approach is typically performed without any dural disruption. In some cases, the posterior fossa subarachnoid space will come into communication with the extradural space as a consequence of tumor resection near the trigeminal root in Meckels cave. This circumstance requires exclusion of the subarachnoid space from communication with the middle fossa. Two methods for closure are possible. My preferred method is placement of a pedicled temporalis muscle flap into the defect. This flap is taken from the posterior to the middle section of the temporalis muscle. This tends to heal very well due to the vascularized nature of the flap. An alternative method is to place a free adipose graft in the defect. Typically, this is harvested from the abdomen. The bone flap is then replaced and the wound closed.
Certain complications are common after resection of trigeminal neurinomas. Certainly, deficits of the fifth cranial nerve may be expected in the early postoperative period. Most patients present with some dysfunction of the trigeminal nerve, whether it be hypesthesia or pain. Many patients will have some resolution or a diminution of their hypesthesia. Patients with pain as a presenting complaint are typically relieved. A large number of patients suffer trigeminal motor dysfunction after resection. This is less common in patients with small, medially situated ganglion tumors and those with peripheral lesions of V1 and V2. Because of the high risk of motor dysfunction, the rare patient with bilateral tumors is treated only on the side producing the bulk of symptoms. The opposite side must be managed expectantly and may only be treatable by radiation therapy. Probably the most troublesome complication from removal of these tumors is keratitis suffered as a consequence of a diminished corneal reflex. However, most patients will present with a diminished corneal reflex. Therefore, eye care is extremely important beginning at the time of presentation. Diplopia is also a frequent presenting complaint that may or may not be helped by surgery. Certainly, any operation in or around the cavernous sinus has the potential to injure the ocular motor nerves. Diplopia may be a consequence of surgical resection, although this is uncommon. Careful dissection of the tumor capsule while limiting manipulation of these nerves is the key to reducing the chances of this complication. When utilizing the extradural subtemporal transpetrosal approach, the seventh and eighth cranial nerves are especially at risk. The seventh nerve in particular may be damaged via traction on the GSPN, which translates to the geniculate ganglion, damaging the nerve. Elevating the dura posteriorly to anteriorly will reduce the chance of this complication. Also, the seventh and eighth nerves are at risk during unroofing of the internal auditory canal. The cochlea may also be damaged during this drilling. Avoidance of this structure is achieved by observing the anatomical relationship of the cochlea to the landmark structures. The cochlea resides in the volume of bone located under the geniculate ganglion, between the internal auditory canal and the genu of the petrous carotid artery. This may also be described by the premeatal triangle, defined by the carotid genu, the geniculate ganglion, and the medial lip of the porus acusticus. Of course, practice in the cadaver laboratory and clinical experience are key factors in avoiding the cochlea during drilling. Other structures at risk from the high-speed drill are the optic nerve, the carotid artery, the oculomotor nerve, and the trigeminal peripheral branches.
COMPLICATIONS
105
All structures are at risk both from direct damage by the burr and from the heat generated by using a diamond-tipped burr. Such complex bone removal must be done carefully after sufficient training in the cadaver laboratory to reduce complications. Also, all drilling is performed with strict attention to maintaining a cool local environment with continuous irrigation.
CONCLUSIONS
marily with surgical resection. In the vast majority of cases, tumors classified as ganglion or peripheral types may be treated through an entirely extradural approach. The two main extradural approaches for these tumors have been presented. It should be kept in mind that each approach is tailored according to the individual anatomy of the tumor in terms of the necessary extradural bone removal. When successfully employed, these approaches result in a high rate of success with regard to tumor resection and resolution of symptoms.
Schwannomas arising from the trigeminal nerve are quite uncommon and represent only 0.2% of all intracranial tumors. Although 50% of intracranial trigeminal schwannomas arise from the trigeminal ganglion, schwannomas can arise from any part of the trigeminal nerve, including its origin in the posterior fossa, and extend to the extracranial branches on the face. Approximately 20% of trigeminal schwannomas arise from the fifth cranial nerve in the posterior fossa and present as a mass in the cerebellopontine angle, causing dysfunction of the fifth, seventh, and eighth cranial nerves with or without cerebellar dysfunction. It is estimated that 25% of trigeminal schwannomas have an hourglass type of extension above and below the tentorium. Symptoms may consist of paresthesias, numbness, pain, and/or difficulty in chewing on the side where the tumor is located. Although some patients experience facial pain, decreased sensation is more commonly reported than pain. Fifteen percent of patients with a middle fossa schwannoma may be asymptomatic. However, as the tumor increases in size, it can grow in a variety of ways depending on which branch is involved; thus, signs and symptoms depend upon the direction of tumor growth. These tumors may appear as masses
INTRODUCTION
in the cavernous sinus with third, fourth, and sixth cranial nerve dysfunction, or as para-pharyngeal masses, nasopharyngeal masses, or masses in the pterygopalatine and infratemporal fossae. Among the extracranial branches, the ophthalmic division is the most commonly involved, followed by the maxillary and mandibular divisions. Schwannomas of the ophthalmic division present with proptosis. However, lesions arising from the maxillary or mandibular divisions can be quiescent for a long time, progressively eroding the skull base; they are often accidentally discovered during work-up for sinus conditions or facial injury.
Patients with intracranial lesions presenting with increased intracranial pressure (ICP) and neurological deficits are candidates for surgery. However, patients presenting with small, incidentally discovered, asymptomatic, extracranial lesions at the skull base without erosion of the skull base should be periodically followed. Although rare, an increase in size or a change in the character of the lesion should alert the surgeon to a possible malignant transformation requiring surgery. Patients with extracranial lesions and erosion of the skull base, or the enlargement of exit foramen (such as the foramen rotundum or ovale) may be candidates for surgery since these tumors have a tendency to violate the dura and become intracranial. Resection of these tumors may not be a technical challenge, but significant morbidity can result from cerebrospinal fluid (CSF) leakage
PATIENT SELECTION
107
108
if the floor is not repaired well. In elderly patients presenting with intracranial lesions, for whom long hours of surgery cannot be undertaken for medical reasons, intratumoral debulking may be ideal to reduce the mass. All patients should be advised of the potential risks of deficits of the fifth cranial nerve following surgery.
The availability of computed tomography (CT) and magnetic resonance (MR) imaging has made plain x-rays obsolete. CT with bone windows is an excellent diagnostic tool by which to visualize the bony erosions at the skull base. On CT, schwannomas appear isodense and enhance with contrast medium. On MR imaging, these lesions are hypo-, iso-, or hyperintense on T1-weighted images and significantly enhance after intravenous contrast. MR angiography may be performed to visualize displacement of the major vessels, particularly in large tumors. In addition, conventional cerebral angiography may be undertaken to better delineate the carotid artery and to study the venous anatomy with attention to the vein of Labb and the sigmoid sinus, particularly in tumors with infra- and supratentorial extensions (hourglass type). The balloon occlusion test may be necessary in cases where the carotid artery needs to be sacrificed and a vein bypass graft is contemplated, or where temporary occlusion of the carotid artery is required during tumor dissection.
RADIOLOGICAL EVALUATION
increase ICP. The hypertensive response to pin fixation of the head may be minimized or eliminated by prior administration of an intravenous anesthetic. Anesthesia is generally maintained with a narcotic by infusion or intermittent dosage of nitrous oxide and isoflurane. The PaCO2 is maintained in the range of 25 to 30 mm Hg. Induced hypotension may be required during surgery to reduce blood loss and the need for transfusion. A mean arterial pressure of 50-60 mm Hg is acceptable in a healthy individual, but may not be tolerated by a patient with cardiovascular disease or hypertension. In highly vascular tumors, deliberate hypotension may be induced by increasing the level of isoflurane or by a direct-acting vasodilator, such as sodium nitroprusside or nitroglycerine. After intubation, the bladder is catheterized and 20% mannitol is administered intravenously (1 gm/kg body weight). Evoked potential monitoring necessitates some modification of the anesthetic technique. If just the sensory modalities are monitored, the level of isoflurane is maintained at <0.4% and anesthesia is sustained primarily with a narcotic infusion and nitrous oxide. If motor evoked potentials and electromyography are also employed, halogenated anesthetics cannot be used. In this case, propofol and opioid (fentanyl or sufentanil) infusions are substituted and only nitrous oxide is used as an inhalant.
Starting at midnight prior to surgery, the patient is given intravenous Solu-Medrol, 250 mg every 6 hours. The day prior to surgery, a loading dose of 1 gm Dilantin is administered, which is continued postoperatively at 100 mg three times a day. When premedication is used, it may be one of the shorteracting benzodiazepines such as diazepam or midazolam, on call to the operating room. Narcotics should be avoided because they tend to produce respiratory depression and/or nausea and vomiting, which can result in increased ICP. Routine monitoring in the operating room consists of pulse oximetry, noninvasive blood pressure measurement, electrocardiography, and capnography. Capnography assesses the ICP level and is titrated to obtain optimal ICP control. An arterial catheter and one or two largebore intravenous lines are inserted. A short-acting opioid and ultrashort-acting intravenous anesthetic (generally thiopental or propofol) are used for induction, followed by a nondepolarizing muscle relaxant for intubation. Intravenous lidocaine (1.0-1.5 mg/kg body weight) may be useful prior to laryngoscopy. A combination of these agents permits a smooth induction of anesthesia, thus avoiding hypertension, hypoxia, hypercarbia, or coughing, all of which may
The surgical approach to a trigeminal schwannoma depends upon the location of the tumor. These tumors are classified into four categories: Type A tumors are located entirely within the middle cranial fossa, commonly arising from the trigeminal ganglion. Type B tumors are situated in the posterior fossa and arise from the root of the fifth cranial nerve, proximal to the ganglion. Type C tumors occupy the supra- and infratentorial compartment in an hourglass manner and grow across the tentorial hiatus. Type D tumors arise from the extracranial branches of the fifth cranial nerve.
OPERATIVE TECHNIQUE
Type A tumors are entirely confined to the middle cranial fossa (Figure 1). They arise either from the trigeminal ganglion or from one of the branches of the fifth cranial nerve prior to exiting the intracranial compartment. These tumors may invade the cavernous sinus or extend into the orbit via the superior orbital fissure, such as in tumors arising from the ophthalmic division. The frontotemporal approach is best suited for type A tumors situated in the middle cranial fossa (Figure 2). Patients are placed in the supine position with the head fixed in a three-pin Mayfield head-
109
B
Figure 1. MR imaging, axial (A) and coronal (B) views, of the brain with gadolinium demonstrating tumor of the middle cranial fossa (type A tumor).
Figure 2. Supine position used for resection of tumors via the frontotemporal approach. Note that the patient is well padded and secured and the head is maintained above the level of the heart.
110
Figure 3. Skin incision. Note that the lower part of the skin incision is within 1 cm of the ear to avoid injury to the facial nerve. Figure 4. Skin flap has been retracted. After interfacial dissection at the zygomatic process, the temporalis muscle and fascia are incised as a second layer.
Figure 6. Tumor has been exposed after widely splitting the sylvian fissure. Note the relationship of optic nerve and cerebral vessels to the tumor. fissure
111
holder. A shoulder roll is placed under the ipsilateral shoulder, elevating and turning the upper torso to the opposite side, minimizing rotation of the head. The head is tilted to the other side by 30 to 35 degrees and extended by 150 to 120 degrees, making the zygomatic eminence the highest point in the operating field. The head is elevated above the level of the heart, facilitating the venous drainage. All pressure points are well padded. The skin incision is marked starting within 1 cm in front of the external auditory meatus, beginning at the zygoma and gently curving around the temporal line onto the frontal area, all within the hairline (Figure 3). A single myocutaneous flap is elevated and retracted using fishhooks, exposing the frontal and temporal bones and centered on the pterion (Figure 4). A cuff of fascia along the superior temporal line is left for later closure. A frontal keyhole and another burr hole at the squamous portion of the temporal bone, just above the zygomatic root, are placed before the craniotome is used. In elderly patients, since the dura is stuck to the bone, it is better to place two or three burr holes and gently separate the dura from the bone before the craniotomy flap is lifted, to expose the dura covering the frontal and temporal lobes and the sylvian fissure. Extradural drilling of the lesser wing of the sphenoid bone is accomplished, making the base very flat. The dura is then opened in a semicircular manner based on the floor; from it, a second T-shaped cut is made vertically corresponding to the sylvian fissure (Figure 5). Beginning at this stage, microscopic magnification and illumination are used. The sylvian fissure is split from medial to lateral, exposing the tumor, carotid artery, and the sylvian vessels (Figure 6). Retraction of the frontal and temporal lobes is accomplished by the Greenberg retractor system. Initially, the brain is covered with a Penrose drain and cottonoids are placed over the brain; later, a malleable brain retractor is used on the surface of the cottonoids. These tumors are extracerebral and subdural. They have a well-defined capsule and often are not vascular. Initially, the surface of the tumor is coagulated and cut, allowing the tumors interior to be exposed. An ultrasonic aspirator is generously used to reduce the size of the tumor and thoroughly decompress it so that the surrounding structures, such as the carotid artery bifurcation and the third and fourth nerves on the medial aspect of the tumor near the tentorial margin, are well visualized. It also helps to identify the arachnoid membrane around the tumor, avoiding injury to the blood vessels, which are inside the arachnoid toward the pia around the brain. Progressive internal decompression helps identify the branches of the fifth cranial nerve as well. On most occasions, the tumor can be removed without causing undue damage to the fifth nerve.
Once the tumor is removed, the gasserian ganglion and the posterior cavernous sinus are carefully examined and exposed if necessary, since in massive tumors, remnants of the tumor may lurk that can cause recurrence if not removed. Once the tumor is removed from the cavernous sinus, bleeding that originates from the sinus is readily stopped by packing it with surgical cotton or Avitene. At the conclusion of surgical resection, the dura is closed in a watertight fashion and the operation is completed in a routine manner. We use a Jackson Pratt drain in the submuscular-galeal plane, which is discontinued after 48 hours.
Tumors situated in the posterior fossa (Type B) are approached via conventional retromastoid craniectomy (Figure 7). These tumors are approached with the patient in the semi-sitting or lounging position (Figure 8). Alternatively, the lateral decubitus position can be used. If the semi-sitting position is chosen, a precordial Doppler ultrasound is used to detect air embolism and a central venous line with its tip at the junction of the superior vena cava and the right atrium is placed after intubation to aspirate the embolized air. Otherwise, no special anesthetic considerations are necessary. Initially, the patients head is placed in a three-pin Mayfield headholder and is slowly positioned in the sitting position. The head is slightly turned to the ipsilateral side of the tumor. A retromastoid skin incision is made from the superior nuchal line up to the base of the occiput (Figure 9). The muscles are cut using a knife or Bovie coagulator; however, incision and muscle dissection at the skull base is carefully done using a knife, lest injury to the vertebral artery occur. Cerebellar retractors are later applied to retract the muscles that are supporting the bone. A single burr hole is placed just below the Frankfurt horizontal plane and 4.5-5 cm behind the external auditory canal and is converted into a 4-5 cm diameter craniectomy exposing the edge of the transverse and sigmoid sinus junction, including the posterior fossa dura (Figure 10). During bone removal, the mastoid air cells may be opened; these are packed with muscle at the end of surgery. Alternatively, in younger patients, a craniotomy flap exposing the posterior fossa with additional required drilling toward the sinuses can be accomplished. Dural opening is carried out several millimeters away from and along the sinuses and then retracted with tack-up sutures. The cerebellum is then covered with cottonoids and retracted using a malleable brain retractor, exposing the cerebellopontine angle. Initially, brain relaxation is achieved by draining CSF through an opening of the cistern in the angle (Figure 11). Prior to tumor resection, it is usually not necessary to cut the superior petrosal
112
B
Figure 7. MR imaging, sagittal (A) and axial (B) views, of the brain with gadolinium demonstrating tumor in the posterior fossa (type B tumor).
Figure 8. Semi-sitting position commonly employed for resection of type B tumors. Note the elevation of the legs and ipsilateral turning, with mild flexion, of the head, without causing jugular vein compromise.
113
Figure 10. Craniectomy adequate to expose the edge of the transverse and sigmoid sinuses. The dura is opened a few millimeters away from the sinuses.
Figure 11. Dura is tightly tacked up to the nearby soft tissue, maximizing the exposure. Once the cerebellum is retracted, the tumor is exposed in the cerebellopontine angle. Note the relation of the superior cerebellar artery and the seventh and eighth nerve complex to the tumor.
114
vein, but sacrificing the vein is not known to cause any deficits. Once the cerebellopontine angle exposed, the fifth cranial nerve at the brainstem, as well as the tumor are observed. Since these tumors arise from the sensory component, the motor component is separately identified, pushed to the caudal aspect of the tumor, and preserved. In large tumors, the entire nerve is often splayed on the surface of the tumor; with progressive debulking using the ultrasonic aspirator, the nerve is visualized after the displaced nerve fibers become compact, and can be preserved completely. To achieve total tumor resection, it may be necessary to sacrifice a few nerve fibers without any major neurological deficit. The extraarachnoidal dissection and debulking are performed by teasing with the help of two tumor-holding forceps and an ultrasonic aspirator. Enthusiastic separation from the brainstem using cottonoids is to be avoided. Constant suction and irrigation help to clear the operative field. The seventh and eighth cranial nerve complex (which has been pushed down) is identified, both anatomically as well as physiologically, using electrophysiological monitoring. Once the tumor is completely removed, jugular compression is performed to test hemostasis, and the dura is closed in a watertight fashion. The mastoid air cells are packed with fibrin glue-soaked muscle to prevent CSF leak. Routine closure of muscle and fascia is later achieved using 0-0 Vicryl sutures. Subcutaneous closure with interrupted 3-0 Vicryl sutures, and skin closure with either staples or monofilament nylon, is performed.
Figure 12. MR imaging, axial view, of the brain with gadolinium demonstrating tumor in both the posterior and middle fossa (ype C tumor).
Patients with type C tumors are operated on using a combined pre-sigmoid supra- and infratentorial approach (Figures 13-18). This approach is done with the patient either in the semisitting position, in the supine position with the head turned, or in a strict lateral position. We believe that the sitting position is ideal for operating on these patients. As far as the sitting position is concerned, similar precautions and maneuvers are undertaken as mentioned in the previous section. The skin incision is marked as shown in Figure 13, in a 7 shape around the ear. The area of skin incision is infiltrated with 1:100,000 lidocaine with epinephrine. The skin is incised and reflected toward the ear, and the temporal and suboccipital bones are exposed after raising a single musculoperiosteal flap. Initially, by using a high-speed drill, the bone over the sigmoid sinus and the junction of the transverse sinus is drilled. Part of the transverse sinus closer to the midline is also drilled. Later, two additional burr holes are placed in the temporal area. A helmet-
shaped bone flap is then removed using a craniotome, thus exposing the subtemporal and suboccipital dura as well as the transverse and sigmoid sinuses. At this stage, a total mastoidectomy extending to the semicircular canals is accomplished. The entire dura from the sigmoid sinus to the semicircular canals is then exposed. The pre-sigmoid dura is cut vertically and the subtemporal dura is incised horizontally along the base of the temporal fossa. Before the tentorium is cut, the superior petrosal sinus is doubly heated using 2-0 silk ties. It is advised not to simply coagulate and leave the sinus since it can open during the surgery and cause troublesome bleeding. It is advisable to cut the tentorium in its posterior portion to prevent injury to the trochlear nerve. The last cut on the tentorium is made after identifying and isolating the fourth cranial nerve in the subarachnoid space with the assistance of a fine nerve hook. Once the tentorium is cut, the tumor is exposed after placing two retractors, one on the cerebellum and another on the temporal lobe. The fourth cranial nerve and the posterior cerebral artery at the tentorial hiatus are protected while the retractors are applied. Progressive intratumoral decompression with an ultrasonic aspirator is carried out, maintaining the integrity of the peritumoral arachnoid. In the posterior fossa, the tumor lies above the seventh and eighth nerve complex, push-
115
Figure 13. Retromastoid skin incision is continued as a low temporal incision around the ear is used.
Figure 14. The mastoidectomy, suboccipital craniectomy, and temporal craniotomy are completed, exposing the dura in the retrosigmoid, pre-sigmoid, and temporal areas.
Tumor resection via the combined pre-sigmoid supraand infratentorial approach (type C tumor)
Figure 15. Dural incision as marked by the dotted line. Note the relationship of the superior petrosal sinus relative to the dotted line. The superior petrosal sinus needs to be ligated prior to the tentorial incision.
116
Figure 16. Tumor is exposed after the tentorium is cut. Gentle retraction of the temporal lobe and the cerebellum in the presigmoid area exposes the tumor, blood vessels, and cranial nerves.
ing it downward. Intratumoral decompression without disturbing the arachnoid will shrink the tumor and separate it from the seventh and eighth nerve complex. Residual tumor situated near the brainstem is managed after resecting the tumor from the posterior cerebral and superior cerebellar arteries, the fourth cranial nerve, and the seventh and eighth nerve complex. Large tumors in this area significantly compress the brainstem. Careless peeling of the tumor from the brainstem before adequate decompression of the tumor should be avoided. With progressive decompression, the tumor delivers itself and becomes separated from the brainstem, since there is always a layer of arachnoid at the interface. Peritumoral arachnoid separates the tumor from the important surrounding neurovascular structures, and if the surgeon stays within the tumor and within the arachnoid, inadvertent injury to these structures is avoided. At the end of the procedure, the dura is approximated with an interrupted 4-0 Nurolon suture. If watertight closure cannot be achieved, lyophilized dura or temporalis fascia is used for duraplasty and fibrin glue is used to reinforce the suture line. The myocutaneous flap and skin are closed in layers in the usual fashion. In the case of duraplasty, lumbar spinal drainage is used for a period of 3-5 days until the dura seals.
Type D tumors are primarily extracranial tumors with or without intracranial extension. If the tumor arises in the maxillary division, the tumor lies in the pterygopalatine fossa; if it arises in the mandibular division, the tumor occupies the subtemporal and infratemporal fossa. These tumors cause erosion of the foramen ovale or rotundum and extend intracranially. Once they extend intracranially, they may compress the temporal lobe, often embedded into the dura yet never violating it. In the case of giant-sized tumors, the petrous bone is eroded and the petrous carotid artery becomes incorporated into the tumor, or at least lies very densely adherent to the tumor. A thorough understanding of this information acquired by preoperative neuroimaging is very important for proper surgical planning. Type D tumors are resected using the subtemporal-infratemporal approach in collaboration with a faciomaxillary surgeon (Figure 19). Patients are operated on while in the supine position (Figure 20). The skin incision begins in front of the ear, just below the level of the zygoma, and extends to the temporal area, to the top of the superior temporal line, gently curving forward (Figure 21). A standard interfacial dissection is carried out at the zygoma, thus protecting the facial nerve branch traveling to the forehead (Figure 22). The zygomatic osteotomy, including part
117
Figure 17. Intraoperative photograph showing the extent of craniectomy and mastoidectomy.
Figure 18. Intraoperative photograph axial view, showing the tumor exposed.
118
Figure 19. MR imaging, coronal view, with gadolinium demonstrating extracranial tumor (type D tumor).
Figure 20. Supine position used for resection. Note the generous padding of pressure points, as well as the adequate taping to secure the patient and position the legs above the level of the heart.
of the frontal process, is performed later (Figure 23). There is no need to perform an orbital osteotomy for tumors located in the pterygopalatine and infratemporal fossa area. In large tumors, the facial nerve may need to be identified, at the parotid fascia, to avoid injury during tumor dissection. Even if the tumor is not compressing the temporal lobe, it is advisable to per form a craniectomy and expose the dura to avoid inadvertent dural and temporal lobe injury. A small low temporal craniotomy and craniectomy, including the margins of the foramen ovale and spinosum, is carried out, identifying the mandibular division and the middle meningeal artery.
After bipolar cauterizing, the middle meningeal artery is carefully sectioned. The tumor borders are identified and its relation to the temporal dura defined. If there is radiographic evidence of compression on the temporal lobe, it is advisable to retract the temporal dura before the tumor is resected. If retraction is difficult, an elective dural-arachnoidal incision is made to allow drainage of some CSF. Initially, the mandibular division is followed from its extracranial course toward the middle fossa. Once the tumor in this area is identified, it is progressively decompressed and resected. If the tumor capsule is densely adherent to the dura and is difficult to sepa-
119
Figure 22. After interfacial dissection to preserve the frontal branch of the facial nerve, the temporalis muscle has been separated from the zygomatic arch.
Figure 23. Zygomatic osteotomy and temporal craniotomy have been performed and the tumor is exposed. Note the relation of the tumor to the dura and the parotid gland.
120
rate, the involved dura is resected and duraplasty is later performed. In tumors with a tongue-like intracranial extension after extracranial tumor resection, the dura is opened, the temporal lobe retracted, and the tumor completely excised. Once the tumor is resected, osteotomy and craniotomy are replaced and the wound is closed in layers in a routine fashion. If the approach described above does not give adequate exposure, particularly in very large tumors eroded into the pterygopalatine fossa, extra space is created by dislocation of the mandible of the temporomandibular joint. Sometimes, mobilization of the facial nerve (by detaching the pterygoid muscles to the coronoid process of the mandible and section of the ramus of the mandible) gives extra room. In tumors arising from the maxillary division, in rare cases, a transfacial-transmaxillary approach is necessary. This is more useful in very large and malignant tumors and often is not necessary in routine schwannomas. In tumors arising from the ophthalmic division or tumors with intraorbital extension, a frontotemporal craniotomy and orbitozygomatic osteotomy are performed with removal of the roof of the orbit and the superior orbital fissure. The dura in the superior orbital fissure area is cut and continued to the periorbita. Intraorbital tumors are very well localized and can be removed without causing injury to the ocular muscles or nerves. The optic nerve, which lies in the area, needs to be identified and preserved with great attention. Tumors that extend into the cavernous sinus are located primarily in the leaves of the lateral wall and can be completely removed after opening the lateral wall of the cavernous sinus. After removing the tumor, the periorbita is closed with 0-0 Vicryl, and the craniotomy and osteotomy are closed with 0-0 Tevdek stitches. The remainder of the closure is carried out as usual.
Other than routine complications of craniotomy, such as infection, hemorrhage, CSF leak, and other related problems, fifth nerve dysfunction is the most common complication associated with these operations. Dysfunction of the ophthalmic division can cause exposure keratitis and corneal ulceration, which need to be recognized and treated promptly. Gold weights in the upper eyelid and temporary tarsorrhaphy are two of the best methods for protecting the eye. To prevent life-threatening infection, it is essential to recognize the entry into the paranasal sinuses during surgery, and prompt attention must be paid to repair it appropriately at that time. The skull base defects are repaired using fat, muscle, and fibrin glue to prevent potential CSF leak; occasionally, postoperative lumbar drainage is necessary for 3-5 days.
COMPLICATIONS
Postoperatively, patients undergoing surgery for trigeminal schwannomas are given intravenous antibiotics for a period of 72 hours, tapering doses of intravenous steroids and anticonvulsants. Except in those with a spinal drain, patients are mobilized out of bed as soon as possible, to prevent potential complications such as deep venous thrombosis. Pneumatic compression boots are used for patients while in bed. After 3 months, follow-up MR imaging with and without gadolinium enhancement is performed. If there is no residual tumor, MR imaging every year for 2 years and then every 5 years is suggested. Patients with small residual tumors are closely followed with MR imaging since these tumors grow very slowly. Reoperation is not indicated unless there is significant compromise of the neural structures from the residual tumor.
Cholesterol granulomas are benign expansile lesions most commonly found in either the middle ear or the mastoid region in association with some type of inflammatory ear disease. Petrous apex cholesterol granulomas (PACGs), however, have been described less often and in many cases have been erroneously grouped with epidermoid tumors, which are etiologically and histologically distinct from PACGs. Petrous apex cholesterol granulomas are the end result of chronic obstruction of normal aeration to the petrous air cells. This obstruction leads to the development of negative pressures within the air cells, causing mucosal engorgement. The mucosal hemorrhage that ensues sets the stage for a cycle of hemorrhage and inflammation ending in the formation and growth of a cholesterol granuloma. Because the progression from hemorrhage to granuloma formation is an ongoing process, these lesions have varied clinical and radiographic characteristics depending upon their stage of formation. In the past, PACGs have usually been drained by otolaryngologists via either a transtemporal or a transsphenoidal route; placement of a small Silastic tube was then inserted to re-establish aeration. Unfortunately, the recurrence rates following the procedure using these approaches have been reported in the literature to be as high as 60%, with some pa-
INTRODUCTION
tients requiring multiple revisions. An alternative approach is complete excision of the lesion, including the mucosal lining, followed by obliteration of the cavity with vascularized tissue.
The presenting signs and symptoms of PACG are largely dependent upon the size and extent of the lesion. Smaller asymptomatic lesions may be discovered only during the evaluation of an unrelated complaint (e.g., headache or dizziness). As it grows, the PACG has a compressive effect upon the surrounding neurovascular structures. The trigeminal and abducens nerves are particularly vulnerable because of their proximity to the petrous apex. As such, many patients present either because of diplopia or facial numbness or pain. With larger lesions, the patient may present with hearing loss and facial nerve dysfunction. While there is no pathognomonic symptom of a PACG, the time course of symptoms often gives a clue as to the diagnosis. The cyclical pattern of hemorrhage and inflammation often leads to periodic worsening of symptoms rather than a steady progression, as in a neoplastic growth.
CLINICAL PRESENTATION
Small asymptomatic or minimally symptomatic lesions in a compliant patient may be safely followed with serial magnetic resonance imaging (MRI) studies. In a small subset of these patients, the progressive cycle will be arrested and no further treatment will be required. Patients with progressive symptoms or enlarging lesions obviously require treatment. Patients who present with either significant symp-
PATIENT SELECTION
121
122
toms (r egardless of lesion size) or large lesions (regardless of symptoms) will also require surgical intervention. The traditional mode of therapy involves a drainage procedure via either a transtemporal or transsphenoidal route. Because the r ecurrence rate for these types of procedures is relatively high, we have approached these lesions via an extended middle fossa approach, which allows complete removal of the granuloma and pseudocapsule. This is followed by obliteration of the cavity with vascularized tissue in the form of a pedicled strip of temporalis muscle.
Because of frequent involvement of the sixth and eighth cranial nerves, all patients should undergo a formal neuro-ophthalmological examination as well as an audiogram and monitoring of brainstem auditory evoked responses. The radiographic evaluation consists of an MRI with and without contrast enhancement in order to confirm the diagnosis. Petrous apex cholesterol granulomas often display a highsignal characteristic on both T1- and T2-weighted images with only meager peripheral enhancement (Figure 1). Depending on the age of the lesion, however, this classic appearance may not be present. MRI also provides information regarding the extent of the lesion and whether intradural extension is present. A computed tomography (CT) scan with bone window settings should also be obtained in order to evaluate the extent of erosion of the petrous apex and the temporal bone. Any involvement of the middle ear will also be readily seen on CT.
PREOPERATIVE EVALUATION
The extended middle fossa approach is an extradural, subtemporal approach to the petrous apex. Although the dissection is entirely extradural, there is a small risk of postoperative seizures because of elevation of the temporal lobe. Patients are therefore started on an anticonvulsant medication the night prior to surgery. This is continued for 3 to 6 months following surgery. No other special preparations or medications are required apart from the routine preoperative evaluation and preparation for surgery.
PREOPERATIVE PREPARATION
Following the induction of general endotracheal anesthesia, a lumbar drain is placed to allow cerebrospinal fluid (CSF) drainage during the extradural subtemporal dissection. The patient is also maintained in a hypovolemic state and is hyperventilated. These maneuvers are important to maximize brain relaxation. Electrodes are then placed to allow for monitoring of somatosensory evoked potentials and facial nerve electromyography (EMG). In addition, an earphone is placed in each external auditory canal
and scalp electrodes are inserted for monitoring of brainstem auditory evoked potentials. During the exposure, excessive traction or manipulation on the greater superficial petrosal nerve, which runs along the floor of the middle fossa, may result in traction injury to the facial nerve via the geniculate ganglion. This complication has been minimized by continuously monitoring the facial nerve EMG. Therefore, the patient must not receive muscle relaxants following induction and intubation. The patient is positioned supine and the head is fixed in a three-pin head clamp. The head of the bed should be elevated to bring the patients head above the level of the heart, thus maximizing venous drainage and brain relaxation. The head is turned to bring the zygoma nearly horizontal, and a roll is placed under the ipsilateral shoulder to maintain the neck in a neutral position (Figure 2). Alternatively, the patient may be positioned in the true lateral position with a subaxillary roll in place. The operating microscope is then balanced and brought over to the field to check that it will be in a comfortable operating position prior to draping. The ipsilateral frontotemporal region is then shaved, prepped, and draped in a standard manner. For cases in which intradural extension is suspected, the lateral thigh is also exposed and prepared, should the need for a fascial or fat graft arise. A preauricular curvilinear incision is made (Figure 2) and the scalp flap is elevated along the subgaleal plane. Care is taken to preserve the superficial temporal artery. At the anterior third of the temporalis muscle, the superficial layer of the deep temporal fascia is incised down to the muscle and reflected anteriorly with the scalp flap, thus sparing the frontalis branches of the facial nerve. The zygomatic arch may then be safely dissected in a subperiosteal manner and sectioned at its most anterior and posterior ends. This allows downward displacement of the arch along with the temporalis muscle. This maneuver provides access along the middle fossa floor with <1 cm of temporal lobe elevation. A limited temporal craniotomy abutting the floor of the middle fossa is performed. It is important to drill the edge of the craniotomy flush with the middle fossa floor in order to avoid unnecessary temporal lobe r etraction (Figure 3). At this point, the operating microscope is brought into the field and the remainder of the surgery is performed under the microscope. The lumbar drain is opened and 30 to 50 cc of CSF is allowed to drain. The temporal lobe is supported extradurally with a malleable retractor, and extradural dissection is carried medially until the middle meningeal artery is identified at the foramen spinosum. Small dural vessels traversing the middle fossa floor are easily controlled with either direct bipolar coagulation or bone
123
Figure 1. Axial T1- (A) and T2-weighted (B) MRIs. Note the high signal of the right petrous apex cholesterol granuloma.
Figure 2. Patient positioning and scalp incision for an extended middle fossa approach to the petrous apex.
Figure 3. Temporal craniotomy flush with the middle fossa floor to avoid retraction of the temporal lobe. Note osteotomy of zygoma (asterisks).
124
Figure 4. Surgeons view following extradural elevation of the temporal lobe and ligation of the middle meningeal artery (1). 2 = petrous portion of carotid artery; 3 = V3; 4 = greater and lesser petrosal nerves; 5 = cholesterol granuloma.
Figure 5. The emptied petrous apex cavity obliterated using a pedicled strip of temporalis muscle.
125
wax. The middle meningeal artery is coagulated and sectioned sharply, and the foramen is packed with bone wax or oxidized cellulose. The foramen ovale anteriorly and the greater superficial petrosal nerve (GSPN) medially are then identified. The lesser superficial petrosal nerve may be seen lateral to the GSPN. The GSPN is sharply dissected from its dural attachment and preserved; in some cases, the GSPN must be divided to avoid traction injury to the facial nerve. Further medially, the petrous portion of the internal carotid artery can be seen deep to the posterolateral border of the third division of the trigeminal nerve, and, depending upon the size of the PACG, may be deviated superolaterally (Figure 4). Care must be taken in exposing this region because the bone over the carotid canal is often thin or even dehiscent. Once the PACG is identified and entered, the fluid is drained and the soft tissues, including the pseudocapsule, are completely excised. This is most often done with a combination of microdissectors, bipolar forceps, and smooth ring curettes. When using ring curettes, it is important to be sure that they are smooth in order to avoid inadvertent injury to the carotid artery. Often, a portion of the remaining petrous apex must be removed with a high-speed drill in order to gain access to all areas of the cavity. If this is necessary, a diamond bit is used in order to minimize the chance of injuring any surrounding structures (e.g., the dura or the internal carotid artery). Obliterating the resultant cavity with fat fragments makes follow-up MRI arduous because of the difficulty in distinguishing the signal intensities of fat versus postoperative fluid collections. We therefore fashion a pedicled strip of temporalis muscle that can be used to obliterate the PACG cavity (Figure 5). This permits easy postoperative MRI follow-up, with the muscle flap giving a consistently hypointense signal on T1-weighted images. The bone flap is returned and secured with titanium miniplates and screws, and the remainder of the flap is closed in standard fashion. No subgaleal drain is utilized. The lumbar drain is removed prior to the patient being reversed from anesthesia unless a dural repair was necessary, in which case the drain is usually left in place for 24 to 48 hours. Patients are reversed from anesthesia in the operating room and observed overnight in either the neurosurgical intensive care unit or the recovery room. They are maintained on dexamethasone for 1-2 days
followed by tapering doses, and antibiotics are maintained for 48 hours. All patients receive antiseizure medication for 3 to 6 months, which is then tapered off if the patient remains seizure-free.
Because this is an extradural approach, primary neurological complications are rare. Maximal brain relaxation using a combination of hypovolemia, hyperventilation, and CSF drainage has nearly eliminated temporal lobe-related problems. Despite this, however, there is a risk of postoperative seizures; patients are therefore maintained on antiseizure medication for 3 to 6 months. If a dural repair was necessary because of intradural extension of the granuloma, then close postoperative surveillance for a CSF leak is crucial. The use of either a fat or a fascial graft, combined with the vascularized temporalis flap and postoperative lumbar drainage, helps minimize the risk of a CSF leak in patients requiring a repair. If it should occur, efforts to stop the leak must be taken, including re-exploration. As mentioned above, postoperative facial weakness is related to traction injury via the superficial petrosal nerve. This is prevented with careful intraoperative monitoring of facial nerve EMG.
POSTOPERATIVE COMPLICATIONS
CONCLUSIONS
1. Cholesterol granulomas of the petrous apex are distinct lesions resulting from blockage of the normal aeration of the petrous apex air cells, which in turn leads to a repeating cycle of mucosal engorgement, hemorrhage, and granuloma formation. 2. The extended middle fossa approach provides a safe, shallow, extradural exposure of the petr ous apex. It enables the surgeon to radically remove the granuloma and obliterate the cavity with vascularized tissue. 3. Complications related to temporal lobe r etraction are minimized by combining a zygomatic osteotomy and drilling the temporal squama flush with the middle fossa floor; adequate brain relaxation is obtained using hyperventilation, hypovolemia, and CSF drainage. 4. Ipsilateral facial nerve function is protected by attending to the greater superficial petrosal nerve and intraoperative EMG monitoring.
The authors thank Ron Tribell for his expert creation of the illustrations in this article.
SURGICAL MANAGEMENT OF ANGIOGRAPHICALLY OCCULT VASCULAR MALFORMATIONS OF THE BRAINSTEM, THALAMUS, AND BASAL GANGLIA
GARY K. STEINBERG, M.D., P H.D. STEVEN D. CHANG, M.D.
Resection of angiographically occult vascular malformations (AOVMs) of the brainstem, thalamus, and basal ganglia has only recently been possible; this has come about largely as a result of improvements in micr osurgery and radiographic imaging. Prior to the advent of magnetic resonance (MR) imaging, it was extremely difficult to diagnose the presence of AOVMs in these regions. Many patients suffering episodic clinical deterioration secondary to hemorrhage from an AOVM were diagnosed as having multiple sclerosis or cerebral infarctions. The use of microinstrumentation and the neuromicroscope along with the aid of stereotaxis and electrophysiological monitoring have enabled neurosurgeons to safely resect many of these lesions, most of which were believed to be unresectable. The natural history of untreated AOVMs in the brainstem, thalamus, and basal ganglia is variable and incompletely described, with bleed rates ranging from 0.25% to 43% per year. Data suggest that brainstem AOVMs presenting with clinical hemorrhage may rebleed at a higher rate of 4.5% to 21.5% per year. The neurological morbidity and mortality associated with clinical bleeding from AOVMs in these regions can be significant, due to the critical nature of structures affected by a single or recurrent hemorrhage.
INTRODUCTION
Angiographically occult vascular malformations typically become symptomatic following a hemorrhage. The volume of the hemorrhage, as well as the particular location, determine the type and severity of the neurological symptoms. Given the critical functions of the brainstem, thalamus, and basal ganglia, most hemorrhages in these regions are symptomatic. Acute hemorrhages usually yield new deficits, while other neurological deficits may result from edema or gliosis surrounding the AOVM. Seizures (supratentorial AOVMs) and cranial nerve deficits (brainstem AOVMs) as well as generalized symptoms such as headache and nausea may be noted.
CLINICAL PRESENTATION
The radiographic appearance of AOVMs is quite characteristic. These lesions usually appear as well-demarcated areas of heterogenous signal intensity on both T1- and T2-weighted images, surrounded by a peripheral rim of hypointensity that is more easily seen on the T2-weighted images. The high-attenuation areas on T1- and T2-weighted images represent subacute blood (extracellular methemoglobin), while the peripheral rim of hypointensity represents hemosiderin and ferritin (chronic hemorrhage). Small AOVMs may appear as a punctate region of low attenuation. Occasionally, a large subacute hemorrhage will dominate the MR appearance, obscuring the characteristic mixed attenuation appearance.
RADIOGRAPHIC APPEARANCE
PATIENT SELECTION
1999 The American Association of Neurological Surgeons
Current indications for surgery include multiple, clinically symptomatic hemorrhages or a single large
127
128
hemorrhage associated with significant neurological deficit. It is also important to consider the location of the AOVM in relation to a ventricular or pial surface, since operative exposure through intact brainstem, thalamic, or basal ganglia structures usually results in unacceptable neurological consequences. An exception to this is the presence of deep lesions in the thalamus that can be stereotactically approached through the pulvinar region and microsurgically resected with good clinical results. The T1-weighted MR image is best for assessing proximity to such an ependymal or pial surface. If the AOVM or its hemosiderin ring is immediately adjacent to such a surface, the AOVM can be approached safely. T2-weighted images, including first echo or second echo, cause some blooming in the hemosiderin ring and a false sense of proximity to the pial or ventricular surface. Patients in the following categories are poor anesthetic risks and are excluded from surgical consideration: elderly patients, those severely and chronically debilitated by neurological deficits, or those with associated severe medical problems. Patients with AOVMs located within the posterior limb of the internal capsule are also not good surgical candidates, because of fear of producing permanent contralateral hemiplegia. However, large AOVMs located in the basal ganglia or thalamus and adjacent to the posterior limb of the internal capsule can be successfully removed with good clinical outcomes.
control of blood pressure. Patients undergo surgery while under normotensive conditions. Mild brain hypothermia is used, dropping the core body and brain temperature to 33-34C by applying a cooling blanket. This degree of mild brain hypothermia has been shown to provide excellent protection against experimental ischemic and traumatic cerebral injury. Under operative conditions of AOVM resection, this hypothermic technique is safe, feasible, and economical with excellent overall clinical outcomes. Spinal cerebrospinal fluid (CSF) drainage is not routinely utilized. Adequate relaxation is achieved thorough the use of ventricular or subarachnoid CSF drainage. Upon completion of the procedure and during emergence from anesthesia, the patients blood pressure should be controlled to avoid any hypertensive episodes, particularly during extubation.
The timing of surgery following the last clinical hemorrhage is important. A 4-week delay from the most recent significant clinical hemorrhage allows for stabilization and possible improvement of the clinical condition. This also provides time for acute hemorrhagic components to soften or liquefy, facilitating easier dissection. Another advantage of operating in the subacute period is that the hematoma has already performed some of the dissection and, after drainage or removal of this blood, one can work through the clot cavity without entering normal parenchyma. Removal of subacute hematoma may also immediately improve the patients neurological condition by relieving pressure on normal neural pathways. Over several months time, the hematoma will organize, fibrose, and become surrounded by glial scarring, obscuring the well-demarcated dissection planes and making surgical removal more difficult. Thus, resection more than several months after the last hemorrhage may be more difficult than procedures per formed in the subacute phase. Prior to surgery, hypertension is controlled with oral antihypertensive medications.
PREOPERATIVE PREPARATION
Anesthetic Technique
The surgical approach is chosen according to the site where the AOVM presents at the pial or ependymal surface. For AOVMs in the midline medulla or pons that present to the fourth ventricular surface, a midline suboccipital approach is utilized, with the patient positioned prone (Figure 1A). If the medullary or pontine AOVM is located laterally or anteriorly and presents to one of these pial surfaces, but not to the fourth ventricular surface, the far lateral suboccipital approach is employed. The patient is positioned laterally or supine with the head turned (Figure 1B). Midbrain or pontomesencephalic AOVMs are exposed via a subtemporal or a subtemporal, transpetrosal approach if they present to the lateral or anterior surface of these structures. The patient is positioned supine with a shoulder roll and the head is turned or operated in the lateral position. Some of the midline pontomesencephalic AOVMs can also be approached via a midline suboccipital exposure if they present to the fourth ventricle or the cerebral aqueduct. If mesencephalic AOVMs are located in the tectal plate region, a supracerebellar, infratentorial exposure is utilized. We prefer the patient to be in a sitting position to allow easier positioning of the microscope (Figure 1C). Others have used the supracerebellar, infratentorial approach with the patient in the Concorde position. AOVMs of the thalamus are exposed via a midline posterior parietal interhemispheric approach, usually transcallosal to the lateral ventricle, but sometimes inferior to the splenium of the corpus callosum. The patient is positioned laterally with the involved side down to allow gravity relaxation of the ipsilateral parietal-occipital mesial cortex (Figure 1B). AOVMs of the basal ganglia are exposed, either transcallosal transventricular as the thalamic lesions, or transsylvian, depending on their precise size and location.
STEINBERG AND CHANG : AOVMS OF THE BRAINSTEM, THALAMUS, AND BASAL GANGLIA
129
Figure 1. A, for AOVMs in the midline medulla or pons that present to the fourth ventricular surface, the patient is placed in the prone position. B, for lateral medullary or pontine AOVMs, the patient is placed in the lateral position with head turned. C, for tectal plate or posterior mesencephalic AOVMs, the patient is positioned in the sitting position.
After general anesthesia is induced and the patient positioned, a sterile surgical scrub is performed and the incision site draped. The skin incision is made using a No. 10 blade, and bipolar coagulating forceps are used to cauterize any bleeding scalp vessels. Dissection is carried through the subcutaneous tissue, galea, and/or muscle using a No. 15 blade or electrocautery, and the bone is exposed. A perforator and craniotome are used to perform the craniotomy. Bone wax is used to control bleeding along the bone edges, and dural tack-up sutures are placed. The dura is opened using dural forceps and a No. 15 blade. For AOVMs in the midline medulla or pons that present to the fourth ventricular surface, a sub-
OPERATIVE PROCEDURE
occipital craniotomy (removing the foramen magnum but not the posterior arch of C1) is used. The cerebellar tonsils are retracted and the inferior midline vermis is sectioned to facilitate exposure to the floor of the fourth ventricle (Figure 2). A region of blue discoloration (hemorrhage) or hemosiderin indicates the location of the AOVM (Figure 3). Microsurgical resection is done under high-power magnification with very fine bipolar irrigating coagulation forceps and using low coagulation power. It is critical to perform the dissection on the edge of the AOVM, leaving the hemosiderin-stained parenchyma intact (Figure 4). Hemostasis is achieved with small pieces of Surgicel, Gelfoam, or cotton. Dissection of the AOVM is done using extreme caution to avoid
130
A
Figure 2. The prone position as viewed through microscope with surgeon at the top of the patients head. A, the floor of the fourth ventricle is exposed for medial medullary or pontine AOVMs, and a midline suboccipital approach is utilized. B, once the AOVM is localized, a gently placed small retractor blade may aid in exposure.
Figure 3. Intraoperative low-power photograph showing hemorrhagic discoloration where the AOVM presents to the ependymal surface of the fourth ventricle (arrow).
Figure 4. Intraoperative high-power photograph showing circumferential dissection around the AOVM, taking care to stay on the AOVM margin, thereby avoiding injury to surrounding brainstem parenchyma.
STEINBERG AND CHANG : AOVMS OF THE BRAINSTEM, THALAMUS, AND BASAL GANGLIA
131
excessive coagulation with spread of current to surrounding brain tissue. Occasionally, the smallest sized tapered retractor blade applied gently to the edge of the brainstem is used for exposure (Figure 2B). It is important to employ piecemeal removal of the AOVM if the parenchymal opening is smaller than the AOVM in order to avoid excessive retraction of normal parenchyma; however, it is essential to resect the entire AOVM, since any residual has a high chance of recurrence with rebleeding and enlar gement. It is also critically important to preserve any developmental venous anomaly (venous angioma) that is associated with the AOVM to avoid a venous infarction with subsequent neurological injury. For AOVMs located in the lateral or anterior medulla or pons, a far-lateral suboccipital approach is used. Meticulous drilling of bone up to the transverse and sigmoid sinuses maximizes exposure. Once the craniotomy is complete, the dura is opened while carefully preserving the transverse and sigmoid sinuses. Retractor blades are used to gently retract the cerebellum to expose the brainstem. Discoloration of the normal tissue indicates the AOVM location, and microsurgical resection proceeds as above. Cranial nerves and vascular branches of the basilar and vertebral artery are carefully avoided (Figure 5). AOVMs in the tectal plate or posterior mesencephalon are approached via a suboccipital craniotomy, and a supracerebellar, infratentorial exposure is most frequently used (Figure 6). This approach requires dissection (using bipolar electrocautery and small bore suction) of the superior cerebellar vermis and superior medullary velum. The precentral cerebellar vein may be sacrificed to improve exposure. Exposure is maximized using retractor blades to retract the cerebellar hemispheres. Care must be taken to avoid injury to the vein of Galen, the two draining veins of Rosenthal, and the straight sinus. Hemostasis is maintained at all times to reduce blood clots within the fourth ventricle in order to minimize postoperative hydrocephalus. Once the AOVM location is identified, microsurgical resection proceeds as described previously. Following the craniotomy, AOVMs of the thalamus are approached via a midline posterior interhemispheric transcallosal approach if the malformation is medial (Figure 7A). Retractor blades placed on the ipsilateral cortical hemisphere improve exposure (the dura maintains retraction of the contralateral hemisphere if the involved side faces down). The corpus callosum is identified, and a callosotomy performed using bipolar electrocautery forceps and small bore suction to achieve exposure to the third ventricle (Figure 7B). Hemostasis must be maintained at all times to ensure visibility while working in the ventricle. The AOVM is identified on the surface of the pos-
terior thalamus or, if not directly visible, located with the aid of stereotaxis. Microsurgical resection can be safely per formed within the pulvinar nucleus without any significant neurological deficit. Occasionally, the smallest sized tapered retractor blade applied gently to the edge of the thalamus is used to increase exposure. Care must be taken to avoid significant amounts of intraventricular blood to minimize the risk of postoperative hydrocephalus. Posteroinferior thalamic AOVMs may be difficult to reach via the above exposure, and are optimally approached through a transcallosal interhemispheric approach inferior to the splenium. Basal ganglia AOVMs are exposed via a transcallosal transventricular approach if they are medially located. For these lesions, the case proceeds as described above for the thalamic AOVMs. Lateral basal ganglia AOVMs are generally exposed via a transsylvian approach (Figure 8). Following completion of a temporal craniotomy, the dura is reflected and the sylvian fissure opened under the microscope to expose the insular cortex. Stereotaxis is useful in the approach to these lateral lesions, which generally do not come to a pial surface. Once the AOVM location is identified, bipolar electrocautery is used to dissect through a small portion of the insular surface directly overlying the AOVM. Small retractor blades may be gently placed to maximize exposure. Microdissection is performed similarly to thalamic lesions, using bipolar electrocautery forceps and gentle small bore suction. Perforating arteries from the middle cerebral artery traverse the insula and the basal ganglia, and should be avoided if they do not directly feed the AOVM.
Meticulous hemostasis is obtained in the resection bed of the AOVM, and a final inspection is performed to ensure that the entire AOVM has been completely removed. Transient induced hypertension can be used to test hemostasis. The resection bed is then lined with Surgicel. The dura is closed in the standard fashion using 4-0 nylon suture, and the bone replaced to complete the craniotomy. A subgaleal drain is attached to light suction for 12-24 hours. The scalp incision is closed in two layers, with approximation of galeal layer using 3-0 Dexon sutures and the skin with staples. A sterile head dressing is applied.
CLOSURE TECHNIQUES
Electrophysiological monitoring and mapping of brainstem motor nuclei are extremely valuable aids for the resection of AOVMs of the brainstem, thalamus, and basal ganglia, and will improve clinical results. Monitoring routinely consists of bilateral somatosensory evoked potentials, bilateral brainstem
MONITORING
132
Figure 5. The far-lateral suboccipital approach to the anterolateral medulla or pons. Cranial nerves V (short arrow), VII/ VIII (long arrow), and IX, X, XI complex (arrowhead) are well visualized.
Figure 6. The infratentorial supracerebellar approach to tectal plate or posterior mesencephalic AOVMs. Gentle retraction on the cerebellum can sometimes facilitate exposure.
Figure 7. Thalamic and mesial basal ganglia AOVMs are approached via a midline interhemispheric approach (A) and a callosotomy is performed providing access to the lateral ventricle and thalamus or basal ganglia (B). Arrow indicates corpus callosum.
Figure 8. Lateral basal ganglia AOVMs are exposed via a transsylvian approach with gentle retraction of the frontal (A) and temporal (B) lobes.
STEINBERG AND CHANG : AOVMS OF THE BRAINSTEM, THALAMUS, AND BASAL GANGLIA
133
auditory evoked potentials, and, if appropriate, bilateral fifth, seventh, 11th, and 12th nerve motor function for brainstem surgery. Bilateral upper and lower somatosensory evoked potentials are used during resection of thalamic and basal ganglia AOVMs. Continuous monitoring of these sensory pathways and motor nuclei has allowed early detection of excessive retraction or manipulation of critical structures. Electrophysiological mapping of the brainstem fifth, seventh, and 12th nerve motor nuclei function, using tiny stimulating electrodes to localize these functions, has allowed more precise planning of safe trajectories through the floor of the fourth ventricle to approach AOVMs. A midline approach to the floor of the fourth ventricle should be avoided, if possible, since this risks causing bilateral internuclear ophthalmoplegia. Special precautions should also be taken to avoid the sixth/seventh cranial nerve complex.
while the subgaleal drain is in place. Postoperative anticonvulsants are utilized on a routine basis only if the patient presented with preoperative seizures.
Intraoperative stereotaxis helps localize portions of the deep AOVMs not presenting directly to the ventricular or pial surface. The CRW (Cosman-RobertsWells) frame was previously used, but was found to be somewhat cumbersome. The Radionics frameless Operating Arm System (OAS) was easier to use, although we now routinely employ the new Radionics Optical Tracking System (OTS) or the Elekta Viewscope, which obviate the need for any articulating arm system and maintain a precision of a few millimeters. This stereotactic approach has allowed better planning for the initial incision into the brainstem, thalamus, or basal ganglia, and also has helped confirm complete resection of all components of the larger AOVMs that may have irregular configurations.
STEREOTAXIS
The patient is monitored for 24 hours in an intensive care setting. Careful control of blood pressure through the judicious use of antihypertensive medication minimizes hemorrhage at the AOVM resection site. The subgaleal drain is removed on the first postoperative day. Antibiotics are routinely used only
POSTOPERATIVE CARE
The most devastating complication is perioperative recurrent hemorrhage which, fortunately, is extremely rare. This can be due either to lack of adequate hemostasis or failure to completely resect the AOVM and emphasizes the importance of removing the entire malformation, including any irregular extensions. MR scanning 3-6 months following surgery is generally reliable in confirming complete resection of the malformation, although hemosiderin staining will persist forever in the resection bed. There should be no evidence of persistent subacute hemorrhage in the form of methemoglobin. Other areas of postoperative morbidity can be related to manipulation, retraction, and edema of critical brain parenchyma surrounding these deep AOVMs. Transient internuclear ophthalmoplegia is common after a midline fourth ventricular approach, but usually resolves. Transient worsening of hemiparesis is associated with resection of some thalamic and basal ganglia AOVMs. Worsening of facial nerve function, sixth nerve paresis, appendicular ataxia, or gait is sometimes noted following resection of certain brainstem AOVMs. Ondines curse is a risk in patients with large medullary AOVMs. The far-lateral approach can sometimes be accompanied by transient lower cranial nerve palsies. Postoperative hydrocephalus is a known complication in patients with significant intraventricular blood, but this can be avoided with meticulous hemostasis throughout the procedure. A watertight dural closure minimizes the risk of CSF leaks and pseudomeningoceles. As with other craniotomies, postoperative meningitis and wound infections are rare, but well-described, events. Better clinical results are achieved in patients who are in good preoperative condition; however, patients in poor preoperative condition have been found to make remarkable recoveries.
COMPLICATIONS
The anatomical complexity and functional importance of jugular foramen structures present a formidable challenge to the surgical removal of tumors in that region. Optimal therapeutic outcomes are dependent upon the cooperation of a multidisciplinary team consisting of neurosurgeons, otologists, headneck surgeons, and neuroradiologists. A thorough understanding of the jugular foramen and its adjacent anatomy is essential for a successful surgical outcome. The jugular foramen contains the jugular bulb, representing the continuation of the sigmoid sinus and progressing to the internal jugular vein. Additional venous drainage enters the jugular bulb via the inferior petrosal sinus. The lower cranial nerves (the glossopharyngeal, the vagus, and the spinal accessory nerves) also traverse the foramen. Additional adjacent anatomical structures encountered during jugular foramen surgery include the facial nerve, carotid artery, hypoglossal nerve, semicircular canals, middle ear, and the temporal components for hearing function.
INTRODUCTION
The most common tumors of the jugular foramen include neuromas of the lower cranial nerves and glomus jugulare tumors. Glomus jugulare tumors require a much greater operative exposure than neuromas due to their highly vascular and locally invasive characteristics. Neuromas are less of a surgical challenge since they are noninvasive and are generally localized to one nerve fascicle, making them easily separable from the surrounding cranial nerves. Less commonly presenting tumors include chordomas, epidermoids, and a variety of carcinomas. A detailed radiological workup is necessary for establishing the diagnosis and determining the anatomic extent of the tumor. Magnetic resonance imaging (MRI) with and without gadolinium enhancement is best for demonstrating the tumor and its relationship to adjacent structures (Figures 1 and 2). Highresolution computed tomography (CT), with and without bone windows, delineates the destruction and erosion of the temporal bone and its relationship to the carotid artery, hypoglossal canal, temporal bone structures, and arteries (Figure 3). Angiography is useful for establishing the diagnosis of a glomus tumor; in addition, it is necessary for determining the degree of vascularity, the location of feeding vessels, and routes of venous drainage (Figure 4A). The diagnosis of a glomus tumor should be anticipated prior to angiography so that simultaneous embolization may be performed if indicated (Figure 4B). When
PATIENT SELECTION
135
136
Figure 1. Contrast-enhanced axial MRI demonstrating enhancing mass in the jugular foramen.
Figure 2. Contrast-enhanced coronal MRI demonstrating an enhancing mass in the jugular foramen and extending into the jugular vein.
appropriate, a carotid occlusion test may also be performed if the tumor involves the carotid artery and potential damage to the artery during surgery exists. In addition to the radiographic workup, an endocrine evaluation including a 24-hour urine collection to measure for catecholamines should be performed. Approximately 4% of glomus tumors secrete catecholamines; if not recognized, the presence of catecholamines can lead to postoperative complications. Complete hearing and swallowing evaluations and otoscopic examinations are also indicated. Nearly all symptomatic patients with jugular fora-
men tumors are candidates for surgical resection. Since many tumors grow slowly, however, consideration must be given to the patients age and any underlying medical conditions that might contraindicate surgery. There is no simple formula for determining the risk-benefit ratios of surgery other than on an individual basis. These operations are rarely emergencies and adequate time for complete diagnostic workups and thoughtful preparations can be made. More controversial are asymptomatic tumors, which must also be evaluated within the context of the patients age and medical condition. Conserva-
137
Figure 4. A, angiogram showing prominent vascularity of the glomus tumor. B, angiogram following embolization demonstrating marked reduction in vascularity.
tive management is often desirable under these circumstances. The selection of alternatives to surgery is also controversial. Embolization alone is thought to reduce the rate of growth of these tumors; however, it is generally used in conjunction with surgery rather than as a primary, single therapeutic option. The effectiveness of radiation therapy on a long-term basis is disputed. Most of the literature addressing alternatives to surgery is distorted by the lack of scientific methodology. Similarly, even less is known about the long-term effects of radiosurgery. The decision to radiate has important implications. If radiation fails and surgery is ultimately necessary, there is significantly greater difficulty and morbidity associated with surgery. The patient must be well prepared for the extensive surgery that is required. Significant risks to hearing, swallowing, facial functions, and speech must be anticipated. In certain instances where there is extensive tumor, functions such as hearing may need to be deliberately sacrificed as part of the resection. Patients must be prepared for the possibility of either a temporary or permanent tracheostomy. In most instances, cranial nerve impairment is temporary. Patients must also be aware that because of
the labor-intensive surgery involving the jugular foramen, it may be necessary to stage the operation into two parts.
All patients with glomus tumors should undergo preoperative embolization (Figure 4). The timing of this is controversial but generally should be done within 24 to 72 hours of surgery. Extensive tumors may require two separate embolizations. Tracheostomy, performed prophylactically, may be preferable to risking postoperative aspiration pneumonia. The tracheostomy can be removed if the swallowing function is intact.
PREOPERATIVE PREPARATION
There are no special general anesthesia considerations. Electrophysiological monitoring is recommended, consisting of brain stem auditory evoked responses as well as monitoring of the lower cranial nerves and facial nerve. Routine broad-spectrum antibiotics are used. Since a spinal drain is usually necessary to reduce postoperative cerebrospinal fluid (CSF) leakage, it is best placed preoperatively, unless intradural exploration is not likely to be per-
Anesthetic Technique
138
formed.
For optimal positioning, the patient is placed supine with the head turned 90 degrees in the contralateral direction (Figure 5). A sandbag is placed underneath the ipsilateral shoulder, facilitating head turning. If an individual is particularly large and of limited mobility, a lateral position may be needed. A threepoint head fixation device may be used to maintain the head position, although a simple cerebellar headrest is often preferable so that the head position may be moved as desired.
Operative Positioning
A wide operative exposure of the petrous bone and neck is necessary for safe removal of jugular foramen tumors. It is important to have control of the vessels proximally and distally as well as to identify the cranial nerves proximally and distally where the anatomy has not been disturbed by tumor (Figure 6). Inadequate exposure adds to the difficulty of the operation and may result in failure to completely remove the tumor. Smaller glomus tumors and noninvasive tumors such as neuromas generally require a less extensive exposure. As the surgeons experience increases, the size of the operative field may be reduced; however, this is not of major concern. A C-shaped posterior auricular incision is made extending anteriorly into the temporalis muscle and down along the anterior border of the sternocleidomastoid muscle. This will provide the exposure necessary to the lower cranial nerves, the carotid artery, and the jugular vein in the neck, as well as to be able to mobilize the temporomandibular joint and perform temporal bone dissection.
Skin Incision
The scalp flap is retracted anteriorly, leaving the temporalis muscle and fascia intact. The ear canal is transected at the bony-cartilaginous junction and is then oversewn. Inferiorly, a flap is raised off the periosteum of the skull continuous with the sternocleidomastoid muscle. Subplatysmal flaps are elevated in the neck until the parotid gland is visualized, with the flap remaining continuous with the fascia of the parotid gland. The sternocleidomastoid muscle is retracted laterally and the internal jugular vein is identified and dissected from an inferior to superior direction. The vagus nerve is identified and followed along its path between the carotid artery and the internal jugular vein. The jugular vein and common carotid arteries are both tagged with a vessel loop. The internal carotid artery is dissected up toward the base of the skull after elevating the submandibular gland. The spinal accessory and hypo-
OPERATIVE PROCEDURE
glossal nerves are identified and dissected above the bifurcation of the carotid artery. The spinal accessory nerve is followed through the sternocleidomastoid muscle and up to the jugular foramen. The parotid gland is sharply dissected off the tympanic and mastoid bones to the level of the posterior belly of the digastric muscle. The facial nerve is carefully dissected adjacent to the styloid process and peeled off the parotid gland up to where it branches. The facial nerve is dissected from an inferior to superior direction until it is completely freed from the parotid gland. The parotid gland can then be removed in its entirety. At this point, the neck dissection is completed, with exposure and control of the internal jugular vein, the carotid arteries, and the ninth through 12th cranial nerves up to the jugular foramen. The facial nerve is also dissected free from the parotid gland. The next phase is temporal bone dissection (Figure 7). This can be performed using either loupes and a strong headlight or the operating microscope. A high-speed drill with copious irrigation is used to begin the mastoidectomy. The bone over the middle cranial fossa (tegmen) and the sigmoid sinus is exposed. The posterior external auditory canal wall is then ground to Koerners septum. The septum is removed and the incus can be seen. As dissection proceeds anteriorly into the zygomatic route, the descending segment of the facial nerve is visualized. This is followed down to the stylomastoid foramen. The facial nerve is dissected and drilled away from its bony canal up to the level of the external genu (Figure 8). The bony external auditory canal is drilled away circumferentially to the area of the facial recess. The incudostapedial joint and incus are removed from the middle ear. Dissection proceeds until the external auditory canal itself is removed. The tympanic membrane, the tympanic malleus, and the annulus are removed. The temporal bone posterior to the facial nerve is drilled away down to the level of the jugular bulb. The tumor can often be seen by this point. Dissection continues until the facial nerve is completely mobilized distal to the lateral genu. At this point, a decision can be made as to whether or not the petrous carotid artery needs to be mobilized. To accomplish this, the temporomandibular joint must be drilled away on all sides, taking care to avoid the carotid artery lying just medial to it. The zygoma is then divided, enabling the mandible to be mobilized anteriorly. Drilling continues into the posterior cranial fossa to the retrosigmoid region. Enough bone is removed to be able to mobilize the sigmoid sinus. At any point, the sigmoid sinus can be opened and packed with Surgicel or other suitable hemostatic agent. The entire sigmoid sinus is drilled away down to the jugular foramen as well as down to the occipital
139
Figure 5. Photograph showing patient positioned supine with the head turned approximately 90 degrees. A C-shaped posterior auricular incision provides the exposure necessary to expose the temporal bone and lower cranial nerves in the neck.
Figure 6. Drawing demonstrating a postauricular incision. Smaller tumors can be removed with a smaller incision and less exposure. (Drawing courtesy of George Card)
140
sigmoid sinus
Figure 7. Drawing of the tumor in the jugular foramen after temporal bone drilling. (Drawing courtesy of George Card)
facial nerve
jugular bulb
sigmoid sinus Figure 8. Operative photograph demonstrating facial nerve exposure after temporal bone drilling.
141
Figure 9. Drawing showing the jugular foramen anatomy following tumor removal. (Drawing courtesy of George Card)
Figure 10. Operative photograph demonstrating closure with free fat graft overlying vascularized temporalis muscle flap that has been rotated into temporal bone defect.
142
condyle and hypoglossal canal. Drilling continues until the endolymphatic sac and semicircular canals are drilled away. A complete labyrinthectomy is often needed. The facial nerve can be followed up to the geniculate ganglion and to the cerebellopontine angle. Throughout drilling, it is important to leave the capsule of the tumor intact to reduce bleeding in the operative field. Once the cochlea is drilled away, the middle fossa dura is easily freed up. Dissection is continued anteriorly into the clivus and the bone overlying the Eustachian tube is removed. The petrous carotid artery can be identified and safely dissected away, but enough exposure must be maintained to have control of the artery should a tear occur. The middle meningeal artery can also be ligated at this point. Following the extensive exposure, removal of the tumor begins. If there is a significant intradural component, the dura may be opened adjacent to the internal auditory canal. En bloc removal of the tumor is preferred; however, piecemeal removal may be necessary depending on the size of the tumor. The outer capsule is generally vascular, even with embolization. Cautery may be used to reduce the capsule; because of the adjacent cranial nerves, this must be done with great care. Generally, it is preferable to allow minor bleeding to continue rather than to risk cranial nerve damage from cautery. The tumor can be internally debulked, and hemostatic material such as Avitene placed into the tumor will control most bleeding. The cranial nerves can be identified proximally as they go into the jugular foramen, although is often difficult to follow the nerves as they enter the canal. It is generally best to mobilize the sigmoid sinus and resect it down to the jugular foramen (Figure 9). The internal jugular vein will already have been ligated distally and following the jugular bulb through the foramen and down to the divided jugular vein is straightforward. Venous back bleeding from the inferior petrosal sinus can be controlled with Avitene. Tumor attachments to the carotid artery as well as the surrounding nerves may require sharp dissection. This puts the carotid artery at some risk and, therefore, it is important to have complete control of the vessel. Once the tumor has been completely removed, complete hemostasis is
easier to obtain.
Following tumor removal, the defect in the dura must be closed. Watertight closure is accomplished by mobilizing a vascularized portion of the temporalis muscle and fascia and sewing it to the dural defect with a running suture. Fascia lata may be used if necessary; however, vascularized tissue is preferred. Non-autologous materials should be avoided as they provide less capacity for watertight closure and carry a higher risk of infection. Fibrin glue may be used to reinforce this closure, although its benefit is questionable. Fat harvested from the thigh can be used to fill any open space from the temporal bone dissection (Figure 10). The eustachian tube is sealed using a piece of temporalis muscle. The temporomandibular joint is reapproximated with microplates. The sternocleidomastoid muscle is reapproximated at its site of origin with interrupted sutures. A subfascial drain is often helpful for at least the first 48 hours. The scalp flap is reapproximated with subgaleal sutures, followed by staples for the skin. A spinal drain is left in place for approximately 5 days. Continuous drainage is best avoided in favor of intermittent drainage of 12 to 15 cc of CSF every 2 hours. The drain is clamped in between draining to avoid overdrainage.
CLOSURE
Major complications include facial nerve and lower cranial nerve palsies, most of which are transient. If there is obvious damage to the facial nerve during surgery, an interposition graft with the greater auricular nerve or a direct re-anastomosis may be performed. If there is significant vagus nerve palsy, a medialization procedure or Teflon injection may be necessary. Following a labyrinthectomy, the patient no longer has the capacity for hearing. CSF leakage can be a problem and is best prevented by the meticulous use of layered closure and a spinal drain. Because of the length of the procedure and the potential for open space around the temporal bone, infection is common. Although unusual, there is always risk of catastrophic injury to the carotid artery which could result in a stroke.
COMPLICATIONS
Esthesioneuroblastomas, or olfactory neuroblastomas, are uncommon tumors constituting 3% of all intracranial neoplasms. They are derived from the neuroepithelial sensory cells of the olfactory mucosa. They usually arise in the nasal vault, presenting as nasal masses. These tumors are locally aggressive and invade the paranasal sinuses, nasopharynx, palate, orbit, or the basicranium. In 20% to 25% of cases, they produce neurological symptoms due to intracranial invasion. When present in the nasal cavity, the common findings are nasal obstruction and epistaxis. However, the intracranial masses present with anosmia and signs and symptoms of raised intracranial pressure (ICP) or proptosis and visual defects once they invade the orbit. The disease may also begin with the signs of a brain tumor or meningitis. Esthesioneuroblastomas are often slow-growing tumors and have a tendency to metastasize to regional lymph nodes and the lung. Regional cervical metastases are known to occur in 10%-40% of cases, even after a silent period lasting years; distant metastases in the lung, bone, liver, mediastinum, and spleen occur in 8% of cases. No definite correlation between the histological grading of these tumors and the biological behavior
INTRODUCTION
is known; rather it is the location and extent of the tumor at the time of the diagnosis that appear to bear a direct relation to the prognosis. Management of these tumors is primarily surgical. Although they have a certain degree of radiosensitivity, esthesioneuroblastomas tend to recur after radiotherapy. The role of chemotherapy in the management, either as primary treatment or as an adjuvant, remains to be evaluated.
Computed tomography (CT) and magnetic resonance imaging (MRI) are the two most valuable diagnostic tools in evaluating this disease. CT with bone algorithms shows the extent of bone destruction (Figure 1), and MRI, with and without gadolinium enhancement, serves to detail the location and extent of submucosal spread of the tumor and define dural involvement (Figure 2). In large tumors, CT of the paranasal sinuses demonstrates the entire extent of the disease process, the obstruction of paranasal sinus outlets, and the destruction to the bone (Figure 3). A three-dimensional CT with bone windows would show the defect in the skull base; the study helps plan reconstruction of the skull base. In MRI, these tumors are hypointense on T1-weighted images and enhance with contrast. The differential diagnoses include hemangiopericytoma and olfactory groove meningioma. Tumors with intracranial extension associated with cysts in the peritumoral area are often diagnostic of esthesioblastomas. In large tumors, magnetic resonance angiography and/or
RADIOLOGICAL EVALUATION
143
144
Figure 1. Coronal section of CT with bone algorithm demonstrating the extent of bone destruction caused by the esthesioneuroblastoma.
B
Figure 2. MRI with gadolinium enhancement in sagittal (A) and coronal (B) views verifies the location and submucosal spread of the tumor.
conventional cerebral angiography are necessary to determine the vascularity of the tumor and the displacement of the major intracranial vasculature. In selective cases, the work-up should include a complete skeletal survey complemented with a bone scan to detect the metastatic spread to the skeleton.
Esthesioneuroblastoma most frequently occurs in patients 40-60 years of age. However, a second peak is noted in persons aged 10-20 years. Caucasians are more commonly affected than Blacks. These tumors
PREOPERATIVE CONSIDERATIONS
145
combined craniofacial surgery and postoperative radiation. Currently, the approach is multidisciplinary and multimodality based. For optimal treatment planning, the tumor should be staged based on the classification proposed by Kaddish and coworkers (Figure 4): Stage A: Tumor confined to the nasal cavity. Stage B: Tumor in the nasal cavity with extension to one or more paranasal sinuses. Stage C: Tumor extending beyond the nasal cavity into the orbit and/or the intracranial cavity, cervical lymph nodes, or distant structures. In Stages A and B craniofacial resection with postoperative radiation has been found to give the best tumor-free survival time; however, it is a matter of debate whether radiotherapy prior to surgery or postoperatively (even in the absence of residual tumor) is useful. Some oncologists favor using radiotherapy as the sole treatment. However, for Stage C tumor, the consensus is to perform a craniofacial resection with the addition of radiotherapy and chemotherapy (cyclophosphamide and vincristine) for distant metastases, although all tumors respond to radiation. Since local recurrence is as high as 60% in patients treated with radiotherapy followed by extracranial surgery, we feel that in all stages (A, B, and C) craniofacial resection is necessary, as there is only a 12% recurrence. The term cure in these patients is to be used with caution because 50% of the patients who are survivors eventually succumb to recurrent or metastatic disease. Radical tumor dissection or any of its modifications followed by postoperative radiation is warranted for clinically evident neck metastases.
Figure 3. Coronal section of CT with contrast enhancement showing the full extent of the disease process as well as obstruction of the paranasal sinus outlets.
are slow growing and remain asymptomatic for a long time, except for a few episodes of epistaxis and nasal obstruction, as long as they are confined to the nasal cavity. Although between 19% and 47% of patients have neurological complications associated with the disease, only 7% have primary symptomatology referable to the nervous system; 70% have rhinological complaints, hence, the initial evaluation is often done by an otolaryngologist. Radiological studies are warranted in patients with the above symptoms and clinical evidence of nasal polypoid masses (especially unilateral). At times, either a direct endoscopic or open biopsy is necessary for the diagnosis prior to definitive surgery. A thorough examination of the neck to rule out cervical lymphadenopathy or a work-up for distant metastases, possibly in the lungs, is undertaken along with a routine general medical preoperative evaluation.
There is an ongoing debate as to the optimal method of treatment. When large series are critically evaluated, surgery and radiation seem to offer a similar benefit, particularly in small intranasal tumors. In larger tumors, radiation alone or radiation prior to surgery appears to yield less favorable results than
Treatment Choice
A correct preoperative diagnosis is crucial in lesions confined to the nasal cavity with either minimal or no intracranial extension. Hemangiopericytoma, lymphoma, plasmacytoma, undifferentiated squamous cell carcinoma, oat cell carcinoma, melanoma, embryonal rhabdomyosarcoma, and aggressive inverting papilloma are some of the tumors considered in the differential diagnosis. Transnasal biopsy with the patient under local anesthesia is easily performed. The results of the biopsy help in planning for definitive treatment. Although routine hematoxylin-eosin staining is helpful in diagnosis, further clarification is obtained by electron microscopy and immunohistochemical studies.
OPERATIVE TECHNIQUE
Esthesioneuroblastomas presenting at the anterior skull base pose a surgical challenge; they are managed via a combined craniofacial or rhino-neurosurgical approach. Some surgeons advocate a form of
146
Figure 4. The Kaddish tumor staging process, frontal (left) and lateral (right) views, assessing the extent of tumor. Appropriate treatment planning follows staging. A = tumor confined to the nasal cavity; B = tumor in the nasal cavity extending to one or more paranasal sinuses; C = tumor extension beyond the nasal cavity into the intracranial compartment.
transfacial exposures such as lateral rhinotomy, ethmoidectomy, and medial maxillectomy for tumors that do not extend intracranially or do not involve the cribriform plate. However, a combined transfacial and transcranial approach is necessary because there is microscopic infiltration of the cribriform plate and olfactory bulbs, even in small tumors.
Starting at midnight prior to surgery, the patients are given 250 mg intravenous Solu-Medrol every 6 hours. The day prior to surgery, patients are given a loading dose of 1 gm Dilantin, which is continued after surgery at 100 mg three times a day. If sedation is desired, one of the shorter-acting benzodiazepines such as diazepam or midazolam is administered on call to the operating room. Narcotics should be avoided because of their tendency to produce respiratory depression and/or nausea and vomiting, which in turn can result in increased ICP. Routine monitoring in the operating room consists of pulse oximetry, con-
tinuous blood pressure measurement, electrocardiography, and capnography. Capnography assesses the level of hyperventilation and is titrated to obtain optimal ICP control. An arterial catheter and one or two large-bore intravenous lines are inserted. A short-acting opioid and ultrashort-acting intravenous anesthetic (generally thiopental or propofol) are used for induction, followed by a nondepolarizing muscle relaxant for intubation. Intravenous lidocaine, 1.0-1.5 mg/kg body weight, may be useful prior to laryngoscopy. A combination of these agents permits a smooth induction of anesthesia, thus avoiding hypertension, hypoxia, hypercarbia, and coughing, all of which may increase ICP. The hypertensive response to pin fixation of the head may be minimized or eliminated by the prior administration of an intravenous anesthetic. Anesthesia is generally maintained with a narcotic by infusion or by nitrous oxide and isoflurane. The PaCO 2 is maintained in the range of 25 to 30 mm Hg. Controlled hypotension may be desired during surgery to reduce blood loss and the need for
147
to the upper lip is thoroughly prepped with Betadine scrub and solution and draped. Prior to the Betadine prep, a temporary tarsorrhaphy using 6-0 nylon is undertaken. The right quadrant of the abdomen is also prepped and draped for harvesting the fat graft at a later stage.
Figure 5. The skin incisions (dotted lines) for resection of esthesioneuroblastomas, including a high bicoronal incision behind the hairline (which is at least 13 cm from the root of the nose) as well as a separate lateral rhinotomy incision.
A high bicoronal skin incision within the hairline at least 13 cm from the root of the nose and a separate lateral rhinotomy incision are marked as shown in Figure 5. This long skin flap is necessary to obtain an adequate pericranial graft. In the case of a large tumor, it is safer to deal with the intracranial portion initially, defining the tumor and protecting the brain; otherwise, any brain herniation through the skull base might cause injury to the brain parenchyma during the rhinological portion of surgery. The proposed line of incision is infiltrated with 1% lidocaine with epinephrine. The skin is incised and the bleeding skin edges are controlled with Raney clips. The posterior edge of the cut skin is lifted with hooks, and subcutaneous dissection is carried out 2 to 3 inches behind the line of the incision. Using a #11 blade, the pericranium is cut right to the bone. Thus obtained (at least 2 to 3 inches longer than the skin flap), the pericranium is later used to cover the anterior cranial fossa during the repair. The pericranium, along with the thin layer of fascia on the temporalis muscle (which is located superficial to the temporalis fascia), is stripped and reflected along with the skin up to the nasion, exposing both of the supraorbital ridges and held retracted until the end of surgery.
Skin Incision
transfusion. A mean arterial pressure of 50 to 60 mm Hg is acceptable in a healthy individual, but may not be tolerated by a patient with cardiovascular disease or hypertension. In highly vascular tumors, deliberate hypotension may be induced by increasing the level of isoflurane or by a direct-acting vasodilator, such as sodium nitroprusside or nitroglycerine. After intubation, the bladder is catheterized, the table padded, and 20% mannitol is given intravenously (1 gm/kg body weight).
Prior to positioning the patient, a tunneled lumbar drain is placed; cerebrospinal fluid (CSF) is not drained unless necessary for brain retraction during surgery. Venodyne pneumatic compression boots are applied with their use continued into the postoperative period. The patient is then placed in the supine position. All pressure points are well padded. The head is slightly extended and fixed to the bed after securing it in a three-pin Mayfield headholder, without any side-to-side tilt. The entire head and face up
Patient Positioning
Initially, a low frontal craniotomy is performed using the Midas Rex drill. It is necessary to raise a low frontal bone flap but not a large one. Two burr holes are placed just above the supraorbital ridges on the lateral aspect (keyholes) and another burr hole is placed exactly in the midline just anterior to the coronal suture. Special attention is paid while the burr hole is drilled near the frontal sinus. When the burr hole is made, the surgeon may decide to stop after penetrating the outer wall of the frontal sinus and use the Midas Rex drill to cut the bone. This technique could cause a tear of the dura and/or the superior sagittal sinus. Instead, it is better to drill the inner wall of the frontal sinus as well, separate the dura and sagittal sinus from the overlying bone, and use the side-cutting Midas Rex drill to complete the craniotomy. The frontal sinuses entered in these exposures are obliterated and repaired at the end of the surgery to prevent potential CSF leak or the formation of mucoceles. One can also raise the bone flap in two pieces. The first piece comes up almost to the midline, then the superior sagittal sinus is sepa-
148
skin reflected
supraorbital osteotomy
craniotomy
Figure 6. After the craniotomy, a supraorbital osteotomy is performed. This is particularly necessary with large tumors.
rated from the bone; later the second piece is removed from the other side. This would avoid injury to the superior sagittal sinus and provide excellent cosmetic results. After the frontal bone is removed, the addition of a supraorbital osteotomy, at least on one side, will minimize the need for brain retraction, especially in large tumors (Figure 6). In giant tumors, which occupy the anterior cranial fossa about equally from the midline, bilateral supraorbital osteotomies are necessary. Prior to the osteotomy, the supraorbital nerve is released from the notch using a Cottle elevator. If the notch is converted into a foramen, the rim of the foramen on either side is broken with a small osteotome and the nerve is released
toward the orbit along with the periorbita which, in turn, is sharply separated from the inner surface of the orbit.
Two horizontal incisions are made in the dura along the floor of the anterior cranial fossa up to the midline, and two rubber dam cottonoid patties are applied on the surface of the brain next to the midline and retracted posteriorly. This exposes the falx and undersurface of the superior sagittal sinus. The sinus is divided between two 2-0 silk ligatures. The frontal lobes are retracted back, exposing the cribriform plate and the olfactory bulbs (Figure 7).
Resection of Tumor
149
frontal lobe retracted cranial nerve II Figure 7. The cribriform plate and the olfactory bulbs are exposed, while the frontal lobes are retracted back, just prior to tumor resection.
In extracranial tumors or tumors with minor intracranial extension, the olfactory nerve filaments should be microdissected and excised at the time of removal of the intranasal component of tumor. This should be done even in the absence of CT evidence of intracranial involvement, to prevent recurrence of the tumor, because of its origin from olfactory epithelium. After resection, a routine closure is undertaken. In medium-sized tumors, resection is carried out piecemeal. Pr ogressive intratumoral decompression and shrinkage of the tumor with bipolar coagulation under constant irrigation would separate the tumor from surrounding compressed brain and open the subarachnoid space. Once partial decompression is accomplished, the base of the tumor at the cribriform plate is managed. The ethmoidal feeding arteries are generally found in this area. They are cauterized and cut, and the intranasal part of the tumor is removed. In giant tumors, the dura is attached to the surface of the tumor; in order to obtain a satisfactory result, the dura (along with the tumor) is resected. Giant tumors could extend posteriorly and press upon the optic nerves and anterior cerebral artery complex. Progressive extra-arachnoidal debulking of the tumor would create a peritumoral space, avoid-
ing injury to the arachnoid toward the brain side. After tumor resection, it is important to resect the olfactory nerve filaments and bulbs.
If the tumor has extensively penetrated the skull base, then en bloc resection of the cribriform plate, the ethmoid sinuses, and a portion of the planum sphenoidale including both the olfactory bulbs with their filaments is attempted. En bloc resection can be achieved either by drilling at the tumor margins or through saw cuts. If saw cuts are preferred, cut at the glabella parallel to the anterior cranial fossa. A second cut just behind the cribriform plate approximately 2.5 cm from the root of the nose and two lateral cuts on either side of the cribriform plate will surround the tumor and avoid injury to the optic nerves. Occasionally, in patients with extensive tumors presenting with poor vision, unroofing the optic nerves from the chiasm to the orbital apex, either unilaterally or bilaterally with the resection of the entire planum between the nerves, is undertaken. The cut margins are examined to ensure that they are tumor free. Orbital exenteration is necessary only if the tumor extends beyond the periorbita into the orbit. We advised earlier to complete the intracranial
150
portion of resection before the rhinosurgical portion; however, in massive tumors, it is easier and safer to have the otolaryngologist and the neurosurgeon work in concert so that the neurosurgeon, after protecting the frontal lobes with rubber dam cottonoid patties and placing retractor blades, should be able to push the tumor through the cribriform plate so that the otolaryngologist can remove it from below through the nose. Because these tumors are notorious for their recurrence, total gross surgical resection without causing undue complications should be the goal.
An incision is made from the inferior medial third of the eyebrow along the lateral nasal area and around the base of the ala nasi (Figure 5). The soft tissues are elevated in a subperiosteal plane to expose the nasal bone, the frontal process of the maxilla, and the maxillary process of the frontal bone. If wide exposure of the anterior superior nasal cavity is necessary, a lateral nasal osteotomy extending into the nasal cavity is performed. The nasal bone is replaced using miniplates at the end of the procedure. The medial orbit is then dissected in a subperiorbital fashion to the orbital apex. The medial canthal ligament is mobilized without disrupting it. The nasolacrimal duct is sharply transected and the anterior and posterior ethmoid arteries are bipolar cauterized and transected. Identification of the fronto-ethmoidal suture indicates the border between the anterior cranial fossa above and the ethmoid cavity below. A complete anterior and posterior ethmoidectomy and sphenoidectomy are done; the lacrimal bone and lamina papyracea are resected. Adequate tumor exposure is now obtained. Large tumors may require a medial maxillectomy. The medial third of the orbital floor can also be resected without need for subsequent reconstruction. The orbital roof is resected as necessary; this is safely per formed since the frontal lobe has been retracted via an intracranial approach.
Lateral Rhinotomy
second layer is the fat graft, and the third layer is the vascularized pericranial flap (Figure 8). The bony defect in the anterior cranial fossa could be left alone, but we advocate repairing the floor with a split-thickness bone graft obtained from the craniotomy flap. This would act as a bridge, preventing herniation of the frontal lobes and sagging of the soft tissue layers placed on the floor. However, some surgeons do repair it with a methylmethacrylate mold. A primarily watertight dural closure is not possible; hence, closure of the dura using either cadaver fascia lata or autologous temporalis fascia should be undertaken. To further achieve a tight seal, a fat graft obtained from the abdominal wall is placed on the floor over the bone graft. At this stage, the vascularized pericranial flap is separated from the skin and laid down on the floor over the fat graft. This flap is also tacked down with 4-0 Nurolon to the posterior end of the dura of the anterior cranial fossa. The suture lines and the tissue grafts are routinely sealed by fibrin glue. In reoperations or in the absence of a strong pericranium, the floor is covered with vascularized temporalis muscle; the free rectus abdominus flap hooked to the superficial temporal artery (with the help of a plastic surgeon) may be used. Postoperatively, a lumbar drain is routinely used for 5-7 days. Repair of the cribriform plate from the nose is accomplished using fat and full-thickness grafts of the skin and subcutaneous tissue placed on a gelatin sponge; keeping the raw surface upward toward the intracranial cavity and packing with iodoform gauze impregnated with bacitracin ointment gives the best results of repair. The nose is then sewn back in position.
After en bloc resection, the floor of the anterior cranial fossa is repaired. Three layers of grafting are used. The bottom-most layer is the bone graft, the
All patients are maintained on antibiotics, preferably 500 mg vancomycin every 6 hours and 80 mg tobramycin or gentamycin administered intravenously every 8 hours, until the drains and packing are removed. Alternatively, Unasyn can be used. If no drains are placed, antibiotics are given for a period of 48 hours. Patients are also continued on 100 mg Dilantin three times a day. Bed rest is required until the lumbar drain is removed, and the patient should be carefully monitored for any signs or symptoms of CSF leak through the nose. This potential problem is hopefully prevented via a lumbar drain for 5-7 days, after all layers of repair are stuck down and well healed. CSF, 50 to 60 cc every 8 hours, is drained. In rare instances, it is necessary to seal the leak via a transfacial, transsphenoidal approach by packing the sphenoid sinus. After the spinal drain is removed, the patient is mobilized. While on bed rest, the patient is given 5000 U heparin subcutaneously administered every 8 hours, beginning 72 hours after surgery. Nasal packing is removed on the 7th postoperative day.
POSTOPERATIVE CARE
151
pericranial flap
fat graft
bone graft
Figure 8. Illustration demonstrating complete repair of the anterior cranial fossa, which is carried out after the en bloc resection. The bottom-most layer consists of the bone graft, the middle layer consists of the fat graft, and the top layer is the vascularized pericranial flap.
Operative complications include injury to the anterior cerebral artery complex and the optic nerves, especially when the tumors are large. Care is exercised during dissection of the posterior aspect of the tumor to avoid these structures. During tumor resection, one should not violate tumor-arachnoid interface. Bilateral frontal lobe damage, either due to excessive retraction or to inadvertent injury during tumor removal, may result in serious cognitive problems. Supraorbital osteotomy and good anesthetic brain relaxing technique would help to avoid these known complications. Other complications include meningitis secondary to CSF leak and the development of meningoencephalocele or mucoceles. As
COMPLICATIONS
mentioned earlier, all of these complications can be avoided by strict adherence to the repair, reconstruction principle, and postoperative care.
The 5-year survival rate ranges from 18% to 71%. Survival depends on the size of the tumor, intracranial extension, and lymph node involvement. In patients with Stage C disease, the 5-year survival is reported in the range of 60%, while with Stage A it is almost 100%. Postoperatively, these patients are carefully followed by CT or MRI about every 6 months. In suspicious cases, endonasal endoscopy is performed and any recurrence is appropriately managed by either surgery or adjuvant therapy.
Brainstem gliomas represent approximately 10% of pediatric brain tumors, but are rare in adults. As a group, these tumors have a poor prognosis and are associated with a 30% 5-year survival rate. However, it has become clear that brainstem gliomas are a heterogeneous group of tumors in regard to biological behavior and surgical accessibility. Certainly, some categories of brainstem gliomas possess a lowgrade pathology and have a better prognosis. In addition, with modern imaging and technical advances, many of these tumors represent surgical challenges that can be overcome. A very important tool in the surgical management of brainstem gliomas is magnetic resonance (MR) imaging. This makes it possible to identify the precise anatomical location of the neoplasm. In conjunction with the clinical picture, MR imaging allows one to know the pathology of the tumor with some degree of certainty. Thus, therapeutic decisions, including those regarding the surgical management of brainstem gliomas, can be made based on characteristic MR images.
INTRODUCTION
Patient selection is vital to the successful surgical treatment of brainstem gliomas. The most rewarding surgical outcomes will be in patients who harbor
PATIENT SELECTION
low-grade tumors that displace rather than infiltrate normal brainstem structures (Figure 1). In addition, patients who are most able to withstand the surgical manipulation of brainstem structures are those who have not yet been injured by their tumor; thus, surgery early in the disease course is preferable to surgery late in the disease course. There is a subgroup of patients with diffuse brainstem tumors in whom surgical intervention plays no role. These patients typically present with a very short history (a duration of weeks) of progressive cranial nerve palsies and gait ataxia. MR imaging reveals an expanded brainstem throughout, with diffuse infiltration by neoplastic cells as evidenced by a marked increase in signal on T2-weighted images of the pons and the medulla. Because these tumors diffusely infiltrate normal brainstem structures, resection of any portion of the tumor also involves removal of functional brainstem tissue. These tumors behave quite aggressively and the patients prognosis is dismal. The histology typically is anaplastic astrocytoma or glioblastoma multiforme. Focal brainstem tumors, on the other hand, are amenable to surgical resection, especially those with an exophytic component. Because these tumors are very slow-growing, patients typically present with a long history (years) of a mild cranial nerve palsy and occasionally with long-tract signs. MR imaging reveals a focal tumor that frequently enhances with gadolinium contrast. As noted above, the tumor often has a large exophytic component. The growth pattern of these tumors suggests that they respect anatomical boundaries and they remain restricted to the compartment of the brainstem from which they
153
154
Figure 1. An illustration of a partially exophytic focal medullary tumor. The tumor displaces the lower cranial nerve motor nuclei anterior to the tumor rather than infiltrating normal brainstem structures.
originated (e.g., in the medulla). These tumors behave in a fairly indolent manner until they reach a critical mass, at which time the clinical course may accelerate. The histology typically is low-grade astrocytoma, juvenile pilocytic astrocytoma, or more rarely ganglioglioma. Focal or exophytic brainstem gliomas typically displace and compress normal brainstem structures. This characteristic, coupled with their low-grade pathology and slow growth, can make these lesions amenable to complete surgical resection and possible cure.
All patients should undergo preoperative MR scanning with contrast enhancement. Not only does this provide for anatomic localization of the tumor and its relationship with normal structures, it also provides for a baseline study in which to judge the efficacy of the operation. It is unusual for patients with brainstem gliomas to suffer from hydrocephalus, and therefore cerebrospinal fluid (CSF) diversion is rarely an issue.
PREOPERATIVE PREPARATION
Routine laboratory studies including a complete blood count with differential, prothrombin time, and partial thromboplastin time are obtained. The patient is typed and cross-matched for two units of packed red blood cells. The patients hair is shampooed with antiseptic soap the night before surgery. Patients receive 10 to 20 mg of dexamethasone intravenously every 6 hours starting at midnight before surgery. In addition, 10 to 20 mg of Pepcid (famotidine) is given intravenously at midnight. Prophylactic antibiotics (usually a first-generation cephalosporin) are given the morning of surgery, early enough to obtain a therapeutic tissue level at the time of incision.
OPERATIVE TECHNIQUE
The patient is induced with intravenous Pentothal and intubated in the supine position. Although not essential, it may be worth a discussion with the anesthesiologists as to the merits of using an armored
Anesthesia
155
endotracheal tube. We frequently use one to prevent the collapse of the endotracheal tube that occasionally occurs with aggressive flexion of the patients neck to obtain optimal suboccipital exposure during positioning. Patients are generally maintained with inhalation anesthesia consisting of Forane and nitrous oxide. Supplemental intravenous fentanyl is also administered. An intra-arterial line is placed to monitor blood pressure and blood gas levels. There is no need for hyperventilation since early in the course of the procedure, the cisterna magna is opened with release of CSF and resulting relaxation of the posterior fossa. Patients undergoing brainstem surgery require strict hemodynamic monitoring. It is not unusual to note wide fluctuations in blood pressure and heart rate during tumor dissection and removal. Typically, the patient will become either hypertensive or bradycardic, and it is essential that the anesthesiologist be prepared for pharmacological intervention if necessary. If hemodynamic changes are encountered, tumor dissection and manipulation should be halted temporarily. The hemodynamic instability usually self-corrects, allowing for the procedure to continue.
The patient is placed in a pinned headholder (either Mayfield or Sugita) and is turned to the prone position onto two gel chest rolls. The arms are tucked on the sides under the sheets. The headholder is attached to the operating table with the neck moderately flexed and the chin pulled posteriorly. This gives excellent exposure of the suboccipital region (Figure 2A). The operating table is placed in reverse Trendelenburg position until the posterior neck and inion are parallel to the ground. In children under the age of 3 years, a horseshoe-shaped headholder is used instead of a pinned headholder. In this case, care is taken to protect the eyes. Unfortunately, optimal exposure of the suboccipital region is difficult without a pinned headholder. An alternative method of positioning patients for this operation is the sitting position. However, not only is there a risk for air embolism, but there are also major concerns regarding adequate cerebral perfusion during times of hemodynamic instability brought on during the tumor surgery. Therefore, we do not advise the sitting position for brainstem glioma surgery.
Patient Positioning
We typically do not shave the patients hair, but plan for the incision in a midline part. If a patients hair is too short to comb, we will shave a small strip. The incision line is infiltrated with 0.5% lidocaine with 1:200,000 epinephrine. After the usual prep and drape, an incision is made from several centimeters
above the inion to the palpated spinous process of C2 or C3, and taken down to the pericranium of the occiput and the fascia of the nuchal muscles (Figure 2B). The fascia is cut on the midline with Bovie current and taken down to the cranium, the arch of C1, and the spinous process of C2. Care must be taken when cutting down on the midline of the arch of C1 since this can be cartilaginous, especially in younger children. The nuchal muscles are dissected in a subperiosteal manner utilizing Bovie current on the suboccipital cranium and a combination of blunt and sharp dissection on the foramen magnum and the arch of C1. There is no need to perform further muscle dissection on C2 unless a laminectomy is required for resection of the cervical portion of some brainstem gliomas. An oval-shaped burr hole is then created on the midline just below the inion. Typically, an air-powered drill with an acorn-shaped bit is utilized. It should be remembered that a keel of bone very commonly exists on the midline in the posterior fossa. Therefore, when drilling one must be aware that the bone thickness will vary widely over the small area of the burr hole, and violation of the dura could occur if care is not taken. After exposure of the dura, the burr hole is enlarged with a curette and the dura is widely stripped with a #3 Penfield. Attention is then turned to the foramen magnum where the periosteum, which melds into the outer leaf of dura of the posterior fossa, is cut widely utilizing a curette. A small amount of bleeding may be encountered at this point from the marginal sinus, which can be easily controlled with Gelfoam soaked in thrombin. The dura is again widely stripped at the foramen magnum with a #3 Penfield. A bone flap is then created utilizing a craniotome. Bone cuts are made from the burr hole, bringing them laterally and then inferiorly and diagonally toward the foramen magnum to make a heart-shaped craniotomy (Figure 3). The bone flap is elevated with a #3 Penfield. There is no need to expose the transverse sinus. The lateral edges of the posterior rim of the foramen magnum are removed using rongeurs. The posterior arch of C1 is not removed except in cases where the brainstem tumor extends into the cervical spinal cord. After epidural hemostasis is obtained, attention is turned to the dural opening. A small dural opening is made with a #11 blade at one of the upper corners of the craniotomy. Using Metzenbaum scissors, the dural opening is continued diagonally to the midline at the craniocervical junction. Care is taken not to cross the midline occipital sinus or the usually present marginal sinus. This is repeated on the opposite side. Two Ligaclips are placed across the occipital sinus, and the sinus is divided at the base of the two diagonals. If the marginal sinus is significant, two Ligaclips are placed across it on the midline. The
156
Figure 2. A, patient positioning. The patient is placed in the prone position on two gel chest rolls. The head is fixed with a pinned head clamp and attached to the operating table with the neck moderately flexed; the chin is pulled posteriorly to give good exposure of the suboccipital region. B, a skin incision is made from several centimeters above the inion to the palpated spinous process of C2 or C3. The hair is not shown for illustrative purposes.
Figure 3. A suboccipital craniotomy. A high-speed drill is utilized to create an oval-shaped burr hole over the midline. One must be aware of the varying bone thickness because of the midline keel. After stripping the dura at the foramen magnum, two bone cuts are made, creating a heart-shaped craniotomy (dotted lines). Any remaining bone of the posterior foramen magnum is removed with rongeurs (also shown with dotted lines).
Figure 4. Dural opening. A Y-shaped dural opening is created. When cutting across the occipital and marginal sinus, Ligaclips may need to be placed to control bleeding. After cutting the dura, the superior and lateral flaps are sutured out of the field.
157
dural opening is then completed, bringing a final limb inferiorly on the midline in the upper cervical region. The dural opening appears as a Y (Figure 4). The three corners are sutured and tacked laterally and superiorly. If not already violated during the dural opening, the cisterna magna should be opened at this time to allow for release of CSF and relaxation of the posterior fossa. The degree and method of magnification during the remainder of the operation depend on the individual surgeon. However, we elect to bring in the operating microscope at this time and perform the remainder of the operation under microscopic illumination.
To provide adequate exposure of the fourth ventricle, the inferior half of the vermis is split. Initially, the retractors from the system of choice (we use either the Sugita or Greenberg systems) are placed to split the tonsils and provide good exposure of the inferior vermis. The bipolar cautery is then utilized to coagulate the pia and veins on the surface of the vermis. The coagulated surface is then cut with microscissors. The corticectomy is created using bipolar coagulation and suction. With completely intrinsic brainstem tumors, the floor of the fourth ventricle is frequently elevated with partial or complete obliteration of the fourth ventricle. Therefore, care must be taken when removing tissue of the inferior vermis since the floor of the fourth ventricle can be inadvertently violated. To help guard against this, we typically split the vermis from a caudal to a cranial direction after retracting the cerebellar tonsils. This allows direct visualization of the dorsal surface of the upper cervical spinal cord, which can be followed to the obex, and the floor of the fourth ventricle as the vermis is split. With dorsally exophytic brainstem tumors, the tumor is usually encountered first, and the floor of the fourth ventricle will not be identified until the superior pole of the tumor has been breached.
Tumor Exposure
Tumor Resection
Focal intrinsic brainstem tumors are completely intra-axial; however, a portion of the tumor will usually have an abnormal-appearing layer of ependyma over it where it broaches the floor of the fourth ventricle. This can be readily identified. This area can be coagulated utilizing bipolarity and entered using suction, after which tissue is removed for pathological examination. An ultrasonic aspirator is then used for the majority of the tumor removal (Figure 5). These tumors are typically quite soft and suckable; therefore, it is advisable that the aspirator settings of low amplitude and suction be used. Typically, these lesions will displace brainstem nuclei either superiorly, laterally, or inferiorly (Figure 6). Although displaced,
the important nuclei will be intimately associated with the tumor margins, and, therefore, it is vitally important that tumor resection proceeds from the inside out (internal resection). Tumor margins are recognized by a change in the way the tumor tissue responds to suction. The area where there is little residual tumor becomes floppy, as if there remains a fragile capsule. The difference in tissue consistency between tumor and brainstem causes tumor tissue to pull away from brainstem tissue under light suction when there is little tumor bulk left at a specific margin. When this is encountered, the remainder of tumor removal should be accomplished with light suction and bipolar coagulation. Once white matter is encountered, tumor removal is complete at this margin (Figure 7). Significant bleeding is uncommon and is usually easily controlled. One should be loath to use bipolar coagulation in the tumor bed unless a specific bleeding vessel can be identified, since this can easily injure normal brainstem tissue. Usually, hemostasis is obtained with gentle, warm saline irrigation, and if necessary with application of microfibrillary collagen and thrombin. With dorsally exophytic brainstem tumors, the tumor bulk is outside the brainstem. In fact, when first reviewing imaging studies, it is common to believe that one is dealing with a cerebellar tumor until invasion of the brainstem via the floor of the fourth ventricle or the inferior and middle cerebellar peduncles is recognized. After opening the posterior fossa as described above, we usually perform a corticectomy of the inferior vermis or over the dome of the mass of exophytic tumor coming closest to the cerebellar cortex as defined by intraoperative ultrasound. Upon encountering the tumor, we take several specimens for pathological examination and then proceed with surgically defining the tumor. Lateral, inferior, and superior margins of the tumor must be identified and dissected from the surrounding cerebellar tissue prior to tumor debulking with the aspirator. Continuously alternating between defining tumor borders and removing tumor tissue allows one to safely proceed with tumor resection without risking inadvertent entry into the brainstem. Although these tumors can be very large, with adequate debulking, the superior pole of the tumor can be identified and the floor of the fourth ventricle visualized. At this point, it is prudent to place a cottonoid patty over the floor of the upper fourth ventricle to guard against injury to the facial colliculi and to block any blood from entering the aqueduct. Working over the dome of the debulked and more mobile tumor, the tumor can begin to be amputated at the level of the floor of the fourth ventricle, bringing this dissection inferiorly until the tumor has been completely mobilized and removed. This allows removal of the tumor tissue while maintaining the
158
Figure 5. Tumor exposure and resection. A corticectomy is made with bipolar coagulation and suction, and the lower half of the vermis is split. Identifying the floor of the fourth ventricle is critical. Intrinsic tumors usually can be readily identified broaching the ventricular floor. After tumor samples are taken for pathology, an internal resection is performed, primarily utilizing the ultrasonic aspirator.
Figure 7. View of the floor of the fourth ventricle after tumor resection and hemostasis have been achieved.
Figure 6. An illustration of an intrinsic focal pontine tumor displacing the facial colliculi (abducens nuclei) inferiorly and laterally.
159
orientation of the normal brainstem tissue. It is extremely important that the small amount of residual tumor tissue which may invade the brainstem is not chased below the level of the floor of the fourth ventricle, as this can lead to permanent injury of the motor nuclei of the lower cranial nerves.
At completion of the tumor resection, the tumor bed is irrigated with warm saline, and hemostasis is assured. The dura is closed with running 4-0 silk sutures, and the bone flap is secured with 2-0 silk sutures. A few 2-0 Vicryl sutures are placed in the muscle to decrease the amount of dead space, followed with interrupted 3-0 Vicryl sutures for the fascia and the galea or subcutaneous tissue. The skin is closed with a running 3-0 nylon suture. No drains are placed.
Closure
Patients with lesions restricted to the pons are extubated in the operating room. They are usually managed in the intensive care unit for about 24 hours. This is primarily for close observation and monitoring. At this point, they are most often fully responsive and able to assist in their own care; they are
POSTOPERATIVE MANAGEMENT
then transferred to the regular ward. Physical and occupational therapy is often begun on the third postoperative day to assist in their recovery. Dexamethasone is usually tapered over 7-10 days. Patients with lesions involving the medulla remain intubated for 72 hours. Sedation may be necessary to keep them comfortable. Continued intubation and ventilation is a precaution against the loss of respiratory drive suf fered by some patients 24 to 48 hours postoperatively. This is probably related to edema of the medulla and leads to hypoventilation, hypoxia, and hypercarbia. Patients who do well are transferred to the regular ward usually 24 hours after extubation. Physical and occupational therapy is begun at this time. If they have an adequate gag reflex, they are fed orally. If not, a nasal feeding tube is placed temporarily until it is no longer needed. Most patients recover their ability to swallow quickly and do not require a percutaneous gastrostomy tube. Patients are maintained on high-dose steroids (6 to 10 mg dexamethasone every 6 hours depending on their size) for 3 days, and then tapered over 7-10 days. Skin sutures are usually removed 7-10 days after surgery, and most patients are discharged at that time. Some patients may require inpatient rehabilitation and are usually transferred to an appropriate facility at that time.
BRAINSTEM GLIOMAS
HAROLD J. HOFFMAN, M.D., F.R.C.S.(C)
Brainstem gliomas have long been regarded as inoperable tumors. In 1939, Bailey et al described the treatment of brainstem gliomas as a pessimistic chapter in the history of neurosurgery. Traditionally, all tumors involving the brainstem were considered infiltrative with diffuse gliomatous proliferation. Before the advent of modern neuroimaging, brainstem tumors were diagnosed using air and contrast x-ray techniques. Although these studies gave a very poor view of the brainstem, they allowed neurosurgeons to recognize tumors in the brainstem and to operate on them. In 1967, Olivecrona described 26 patients with tumors of the medulla; of these, seven survived for 10 to 25 years after an operation that comprised either partial removal or biopsy and decompression. No mention was made on whether these patients received radiotherapy. In 1968, Poole reported three cases of confirmed astrocytoma in the brainstem, in which the lesions were intrinsically solid and cystic; these patients survived for long periods. Poole ascribed this long-term survival to treatment including radiotherapy. In 1971, Lassiter et al described five patients with brainstem masses that were largely cystic and contained mural nodules. These patients were treated by uncapping the cyst, biopsy of the nodule, and postoperative radiotherapy. Again, there were long-term survivors among these patients. Vil-
INTRODUCTION
lani, in writing about brainstem gliomas in 1975, stated that during the previous 30 years, no progress had been achieved in this field. Brainstem gliomas are a relatively common tumor in children. Russell and Rubinstein stated that 77% of brainstem gliomas occur in patients less than 20 years of age. Matson stated that 18.7% of posterior fossa tumors were in the brainstem. Koos and Miller found that 13% of the posterior fossa tumors they encountered were in the brainstem, and in my own institution 29% of posterior fossa tumors were located in the brainstem. A review of brainstem gliomas has indicated that 20% are dorsally exophytic into the fourth ventricle, 50% are diffuse intrinsic tumors that start in the pons and spread throughout the brainstem, 20% are focal in the brainstem, and 10% are located in the craniocervical junction. These four types of brainstem gliomas are discussed below. Magnetic resonance images (MRIs) and operative photographs are used in describing these tumors.
In 1980 at The Hospital For Sick Children, Toronto, we encountered a small but distinct group of eight brainstem tumors that behaved in a significantly different fashion from the typical infiltrating brainstem glioma. In these patients, symptoms frequently began early in childhood and sometimes during infancy. In infants, the history tends to be one of intractable vomiting and failure to thrive, whereas in older children, signs of raised intracranial pressure, including papilledema, are frequently present. A long history of symptoms can usually be elicited.
161
162
Dorsally exophytic brainstem tumors protrude into and largely fill the fourth ventricle (Figures 1 and 2). They mushroom down over the dorsal surface of the cervical spinal cord and typically enhance brightly with the administration of a contrast agent. These tumors are typically low-grade astrocytomas or, more rarely, gangliogliomas. Furthermore, the vast majority of these tumors are solid and not cystic. Subtotal surgical excision has been extremely successful (Figure 3).
Intrinsic astrocytomas of the brainstem start in the pons and spread up into the midbrain and down into the medulla; they typically present with cranial nerve palsies and long-tract signs. Initially, the tumor shows no enhancement and enlarges the brainstem. If the patient is treated with radiotherapy, the brain-
stem typically returns to a normal size. For a brief period the child responds positively, until MRI reveals the tumor to be enhancing, with nodular excrescences extending from the brainstem (Figure 4). Patients with these diffuse intrinsic tumors rarely survive for more than 2 years, despite aggressive therapy. At autopsy, these tumors are revealed to be either malignant astrocytomas or glioblastoma multiforme (Figure 5). Furthermore, at autopsy, extensive leptomeningeal metastases are frequently present (Figure 6). No treatment has been shown to be successful in enhancing the length of survival in children with this type of tumor. In numerous instances, surgeons have operated on these tumors with no effective long-term results. Adjunctive therapy, including radiotherapy and chemotherapy, provides short-term relief for these patients, but eventually the tumor returns and survival for more than 2 years is rare.
163
Figure 3. A, operative view of a dorsally exophytic astrocytoma. B, operative view following resection of tumor.
164
Figure 4. Sagittal T1-weighted MRI showing a diffuse tumor of the pons extending up into the midbrain and down into the medulla with nodular excrescences.
165
Focal intrinsic gliomas make up about 20% of brainstem gliomas. They are most commonly located in the midbrain, although they have been known to occur in the medulla; in rare instances, they are located in the pons itself. Depending on the location, they can give rise to cranial nerve palsies and to long-tract signs. They can be solid or cystic. After resection, no additional therapy is necessary in 60% of patients, whereas 40% require adjunctive therapy because of recurrence. Focal tumors can occur in any location within the brainstem. They usually involve a portion of the tegmentum of the midbrain (Figures 7 to 10). In my experience, I have found that one can elevate the temporal lobe, section the tentorium, and then expose the side of the midbrain; a horizontal incision is then made in the midbrain exposing the tumor, which can then be debulked using an ultrasonic aspirator or a contact laser. Small tectal tumors
can be very indolent and typically close off the aqueduct, producing hydrocephalus. Hydrocephalus can be treated with a third ventriculostomy or a ventriculoperitoneal shunt. There is no need to resect these small indolent tectal gliomas. However, one can resect a very large tectal glioma that mimics a pineal tumor. These tumors must be explored and debulked (Figure 11). Frameless stereotaxis is essential in removing focal tumors of the brainstem (Figure 12). Focal tumors in the pons are rare; they are approached usually via a midline incision in the floor of the fourth ventricle (Figure 13). Focal tumors in the medulla are approached via an incision in the midline; if they bulge out laterally, they can be approached from the external surface of the medulla (Figure 14). In patients in whom resection is undertaken in the medulla, it is important for the patient to maintain postoperative ventilation in the intensive care unit until it can be ascertained that breathing is effective and there is no evidence of CO2 retention.
B
Figure 7. A, computed tomography scans in an 8-month-old infant showing an enhancing tumor of the midbrain extending into the thalamus. B, T1weighted MRIs of the patient at 8 years old. Six years later, hemiparesis has disappeared and there is very little tumor.
166
C
Figure 8. A, axial T-1 weighted MRIs showing large enhancing tumor of the midbrain. B, 3 months following resection of the tumor, some enhancing tumor is still seen; C, 2.5 years later, no residual tumor is seen.
167
Figure 9. A, MRIs showing focal tumor filling the left midbrain. B, operative view of the tumor. C, following resection, little tumor remains.
Figure 10. A and B, MRIs showing focal tumor in the midbrain with a cystic component in the fourth ventricle. C, frameless stereotaxy is used in removing the tumor.
168
B
Figure 11. A, T1-weighted MRIs showing a large tectal tumor. B, operative view showing the left colliculus filled with tumor. C, operative view after resection of the tumor.
169
Figure 13. Axial T1-weighted MRIs showing a focal tumor in the pons.
Figure 14. T1-weighted enhanced MRIs showing a focal tumor in the medulla with a cyst.
Craniocervical brainstem gliomas arise in the upper cervical cord and extend into the medulla (Figures 15 to 17). These patients typically present with long-tract signs and may have lower cranial nerve signs as well. Laminotomy is a useful procedure if the tumor extends down into the spinal cord. A midline laminotomy is made, and the tumor is debulked using an ultrasonic aspirator. Resection is carried out rostrally to the level of the medulla, where resection becomes more conservative. The use of motor and sensory evoked potentials is essential.
Diffuse intrinsic brainstem tumors are malignant and patients do not benefit from surgical intervention. Dorsally exophytic tumors, focal intrinsic tumors, and craniocervical tumors are usually benign and surgical debulking is beneficial. Some of these tumors undergo involution, and no adjuvant therapy is needed. In patients with small tectal gliomas, only control of the hydrocephalus and observation are needed. In patients with large tectal gliomas that can mimic a pineal tumor, the glioma should be explored and resected.
CONCLUSION
Figure 15. MRIs showing a large enhancing tumor extending from the cervical cord and into the medulla.
170
Figure 17. A, MRI of a craniocervical tumorthe patient underwent biopsy and radiation at another institution. B, following resection and fusion of the spine, the child responded to treatment. C, operative view of tumor extending into fourth ventricle.
Lesions located laterally in or adjacent to the lateral ventricle can be difficult to surgically expose. T raditional approaches to such lesions are transcortical, through the superior or middle temporal gyrus, the superior parietal lobule, or the frontal lobe. The transcallosal approach is usually performed via an ipsilateral route with the patient supine and the midline of the head vertically oriented. Dissection proceeds through the ipsilateral interhemispheric fissure and the ipsilateral ventricle to reach the lesion. The authors prefer a contralateral transcallosal approach. In this approach, the patient is supine, the midline of the head is oriented horizontally with the lesion on the upside, and dissection proceeds through the contralateral interhemispheric fissure and ipsilateral ventricle to reach the lesion. The contralateral transcallosal approach has many advantages compared to other routes. The use of this approach avoids a transcortical incision; in addition, gravity retracts the hemisphere to open the interhemispheric fissure. A more lateral angle of approach increases the surgical exposure and minimizes retraction of the medial hemisphere. Finally, lesions located in the dominant hemisphere can be resected without an interhemispheric dissection on that side.
INTRODUCTION
We performed our first contralateral transcallosal approach in 1983 and recently published our results with 32 patients. The technique is safe and effective and warrants consideration for appropriate lesions.
The contralateral transcallosal approach can be used to expose several different types of lesions located laterally in or adjacent to the lateral ventricle, including cavernous malformations, arteriovenous malformations (AVMs), and tumors. Cavernous malformations located in the caudate nucleus, anteromedial thalamus, lateral ventricular wall, and posterior thalamus are well suited to this approach. These lesions reach an ependymal surface laterally in the ventricular system. AVMs located in the corpus callosum, choroid plexus/lateral ventricle, and thalamus are also well suited to this approach. Typically, such AVMs are fed by the pericallosal and choroidal arteries, and the nidus extends laterally. Likewise, tumors in similar locations, including gliomas, central neurocytomas, intraventricular meningiomas, choroid plexus papillomas, and some craniopharyngiomas, can be exposed via the contralateral transcallosal approach. Patient age can be a contraindication to this approach because the craniotomy flap crosses the superior sagittal sinus. In older patients with more adherent dura, the risks of sinus injury, significant blood loss, sinus occlusion, and air embolism are increased.
PATIENT SELECTION
171
172
Figure 1. Lateral (A) and superior (B) views of the patient positioned for a contralateral transcallosal approach to a left intraventricular lesion. The head is turned to the right, the left shoulder is bolster ed, and the skin flap is based on the right.
The patient is placed in the supine position with bolsters under the shoulder ipsilateral to the lesion (Figure 1). The head is placed in a Mayfield headholder and turned so that the hemisphere contralateral to the lesion is down, the lesion is up, and the sagittal midline is parallel to the floor (Figure 2). The head is angled 45 upward (lateral neck flexion) to optimize the angle of view into the ventricle (Figure 3). A Ushaped skin incision in the frontoparietal region creates a scalp flap that is based laterally on the contralateral side and crosses the midline to the other side. The anterior limb of the incision parallels the hairline, and care is exerted not to incise the skin on the forehead. A fishhook retraction system, with the hooks pulling inferiorly on this limb of the incision, helps to increase the anteroinferior exposure of the bone flap. The coronal suture is identified, and a craniotomy is made with two thirds of the bone flap anterior and one third posterior to the coronal suture, crossing the superior sagittal sinus twice to expose the other side. Great care must be exercised in crossing the sagittal sinus with the craniotome. Its epidural position is established visually before the sinus is crossed by aiming a light into the cut of bone and irrigating enough to see intact dura. If a dural breach is detected, the flap is taken in two pieces, with the first piece entirely over the down hemisphere and the second piece crossing the sinus to complete the craniotomy. With two pieces, the dura and sagittal sinus can be carefully dissected off the inner table of skull after the first piece is removed under excellent visualization. This technique is particularly important in elderly patients with adherent or thin dura.
SURGICAL TECHNIQUE
The dura is opened contralateral to the lesion with a C-shaped flap based along the superior sagittal sinus. Important bridging veins must be protected (Figure 4A). The dural flap is tented superiorly to open the interhemispheric fissure, which is then opened further by removing adhesions between the brain and falx, working between the veins bridging to the superior sagittal sinus. Every effort is made to preserve these bridging veins. The arachnoid membrane deep to the free edge of the falx is opened to expose the pericallosal arteries and corpus callosum. Gravity allows the contralateral (down) hemisphere to separate from the falx, eliminating the need for retraction. A retractor is placed along the inferior free edge of the falx to expose the pericallosal arteries and corpus callosum. The superior falx should not be resected aggressively, to avoid stenosis or occlusion of the superior sagittal sinus which would thereby promote sinus thrombosis and venous infarction. The pericallosal arteries are separated, and the corpus callosum is incised 2 cm to permit entry to the lateral ventricle ipsilateral to the lesion. The incision site can be identified using an intraoperative navigational system or by retracting the ipsilateral pericallosal artery laterally and entering the ventricle beneath the arterys normal location (Figure 4B). Once inside the ventricle, normal ventricular anatomy is identified (including the choroid plexus, thalamostriate vein, and septal vein) and the appropriate ventricle is confirmed (Figures 5 and 6). The lesion is then identified and treated. Further exposure can be obtained by deepening the retractor so that its tip is in the ventricle retracting the corpus callosum. At the end of the procedure, the ventricles are extensively irrigated and filled with saline. If not in-
LAWTON AND SPETZLER : THE CONTRALATERAL TRANSCALLOSAL APPROACH TO LESIONS IN OR ADJACENT TO THE LATERAL VENTRICLE
173
Figure 2. A, gravity retracts the contralateral hemisphere, opens the interhemispheric fissure, and pulls the lesion into the surgeons view. B, the ipsilateral transcallosal approach (line A) requires retraction of the medial hemisphere to reach the lateral aspect of the lesion, whereas the contralateral transcallosal approach (line B) requires no additional retraction.
Figure 3. Patient position and lesion location from the surgeons perspective.
174
planned callosotomy
dura falx
B
falx midline corpus callosum section of corpus callosum
callosomarginal artery
cingulate gyrus
Figure 4. A, the interhemispheric dissection has been completed, and the corpus callosum is in view at the depth of the exposure. The fishhooks pull the scalp forward to increase exposure for the bone flap while keeping the incision off the forehead. The bone flap crosses the sagittal sinus twice to expose it completely. This bone work enables the dural flap to be tacked superiorly, permitting a view directly down the falx and the ability to open the interhemispheric fissure widely. B, the callosotomy is made beneath the pericallosal artery ipsilateral to the lesion in order to enter the appropriate ventricle.
LAWTON AND SPETZLER : THE CONTRALATERAL TRANSCALLOSAL APPROACH TO LESIONS IN OR ADJACENT TO THE LATERAL VENTRICLE
175
pericallosal artery
thalamus
foramen of Monro
septal vein
septum pellucidum choroid plexus of the lateral ventricle fornix Figure 5. The relationship between the lateral thalamostriate vein and the medial choroid plexus confirms entry into the appropriate ventricle, and the lesion is visualized. Note how gravity pulls the lesion down into the corridor of operative exposure. Deepening the tip of the retractor into the ventricle helps visualize the pathology better.
Figure 6. Sagittal view illustrating that all regions of the lateral ventricle are accessible via this exposure, depending upon the angle of approach.
176
serted preoperatively, a ventriculostomy catheter is left in the ventricle at the end of the operation to clear residual blood that could obstruct the flow of cerebrospinal fluid. The contralateral transcallosal approach is also used for lesions located laterally in the corpus callosum and the cingulate gyrus. The corpus callosum may not be traversed, but the advantages of the contralateral approach apply as well to these lesions. With the more superficial lesions, the falx often is cut inferiorly after the inferior sagittal sinus has been coagulated to enable a sharper angle across to the opposite side.
The contralateral transcallosal approach was used in lesions laterally in or adjacent to the lateral ventricle in 32 patients. Three patients had right-sided lesions, and they were positioned with the left side down; the remaining 29 patients had left-sided lesions and were positioned with the right side down. In four patients, the falx was cut to increase exposure of the contralateral medial hemisphere. In two patients with left hypothalamic lesions, the foramen of Monro was enlarged by opening the choroidal fissure posteriorly to gain access to the third ventricle. In one patient with a thalamic AVM, the posterior contralateral transcallosal approach was used; in the remainder, the standard anterior contralateral transcallosal approach was used. In six patients, cavernous malformations were resected completely in a single surgical stage. In five patients, AVMs were resected in a single stage, and in two, a second-stage surgery was required. Choroidal and pericallosal AVMs can be resected completely via this approach; however, thalamic AVMs may require treatment with stereotactic radiosurgery to the deep residual nidus. In four patients, benign tumors (two meningiomas, one choroid plexus papilloma, and one prolactinoma) were resected completely via the contralateral transcallosal approach. More malignant tumors (e.g., astrocytomas, glioblastomas multiforme, and central neurocytomas) can also be accessed through this approach, but their resection is typically subtotal. Ventriculostomy catheters were often indicated preoperatively due to hydrocephalus from intraventricular hemorrhage, trapped ventricular horns, or obstruction of the foramen of Monro. When not already present, ventricular catheters were inserted during the operation and left in place for several days. Only four patients required ventriculoperitoneal shunting. There were no surgical deaths in this series. Two (6%) of the 32 patients experienced transient neurological deficits. After resection of a thalamic AVM, one patient developed a small thalamic infarction
RESULTS
with right-arm weakness, which resolved completely. After resection of a left atrial meningioma, a second patient developed left moderate hemiparesis, which also resolved completely. Two patients (6%) experienced permanent neurological morbidity. One patient with a central neurocytoma experienced postoperative intraventricular hemorrhage that required multiple ventriculostomies. He subsequently developed a Serratia ventriculitis. The other patient developed a venous infarction after resection of a left glioblastoma multiforme, with resultant right hemiparesis and expressive aphasia. At last examination, patient outcome in terms of the Glasgow Outcome Scale score (GOS) was determined over a mean follow-up of 1.6 years. Twenty-six patients had a good recovery (GOS 1), four had moderate disabilities (GOS 2), and two had severe disabilities (GOS 3).
The contralateral transcallosal approach is an excellent technique for dealing with laterally placed lesions in the lateral ventricles, which can be particularly difficult to resect. With this approach, the craniotomy and interhemispheric approach are contralateral to the lesion and the transventricular approach is ipsilateral to the lesion. We are not the first to report the contralateral transcallosal approach. Other surgeons, including Long and Chou, Shucart and Stein, Machado de Almeida, and Ehni, also have mentioned it. It seems, however, that most neurosurgeons use this approach rarely, if at all. The outcomes in this series demonstrate the utility of the contralateral transcallosal approach for treating a wide variety of lesions located laterally in or adjacent to the ventricle or in the interhemispheric fissure, including cavernous malformations, AVMs, and tumors of all types. The modifications described here make this a safe and effective approach, which may warrant more than occasional use. Two other approaches for exposing ventricular lesions are the transcortical-transventricular approach and the transcallosal-transventricular approach. The transcortical approach is more direct with a short working distance to the lesion, but it has distinct disadvantages. First, transcortical approaches traverse normal brain tissue, increasing the risk of neurological deficits. Various trajectories through cortex have been described (e.g., through the superior and middle temporal gyri and through the inferior (Van Wagenen) and superior parietal lobules), each associated with neurological sequelae. Second, transcortical approaches are more difficult to perform than the transcallosal approach. Fewer anatomical landmarks are present to guide the surgeon transcortically to the
DISCUSSION
LAWTON AND SPETZLER : THE CONTRALATERAL TRANSCALLOSAL APPROACH TO LESIONS IN OR ADJACENT TO THE LATERAL VENTRICLE
177
ventricle. This problem, however, has been mitigated by the advent of intraoperative navigational systems. The approach is particularly difficult when the ventricles are not dilated. The exposure tends to be limited because of the inclination to minimize the extent of cortical incision. Finally, the transcortical approach can cause postoperative seizures. The limitations of the transcortical approach led Dandy, Milhorat and Baldwin, Shucart and Stein, and others to develop the transcallosal approach. The transcallosal approach does not traverse cortex but goes through the interhemispheric fissure and corpus callosum instead. This route has identifiable anatomical landmarks to guide the surgeon to the ventricle and enlarged ventricles are not a necessity. The third ventricle can be exposed simply by enlarging the foramen of Monro, a maneuver not easily accomplished through a more oblique transcortical approach. Most importantly, many studies have demonstrated that the transcallosal approach is well tolerated and produces no functional neurological deficits or postoperative seizures. Most neurosurgeons position their patients supine or sitting with the head and neck in the neutral position and the sagittal midline oriented vertically. We prefer to rotate the neck laterally to orient the midline horizontally. This horizontal position is easier on the surgeons hands, allowing the hands to work in the same plane rather than on top of each other. The surgeon can use an instrument in each hand without obscuring the operative field. This position also permits gravity to retract the downside hemisphere and open the interhemispheric fissure
while the falx retracts the upside hemisphere. Increasingly, the advantages of this position are being recognized. The horizontal position necessitates a decision regarding laterality. In most cases, the head is positioned with the lesion side down (i.e., right-sided lesions are approached with the right side down and left-sided lesions with the left side down). This position works well for lesions located near the midline. However, exposure of laterally placed lesions in or adjacent to the lateral ventricle requires significant retraction of the hemisphere and risks injury to the parafalcine and cingulate gyri. The contralateral transcallosal approach offers a better angle of approach that increases lateral exposure of the lesion and minimizes the retraction required on the medial hemisphere. In addition to opening the interhemispheric fissure to facilitate the approach, gravity pulls the upside lesion medially into the surgeons view. Without this angle of approach, the lateral margin of the lesion can be reached only with excessive retraction. The contralateral transcallosal approach is ideal for left-sided lesions. Although added lateral exposure is gained regardless of the side of approach, the approach to a left-sided lesion from the right spares the dominant hemisphere from retraction injury and the potential sacrifice of bridging veins. Because of this, it is believed to be safer than the approach to a right-sided lesion from the left. Thus, the contralateral transcallosal approach from the left can be used judiciously when the need for more lateral exposure offsets the additional risks of a left interhemispheric dissection.
POSTERIOR FOSSA DECOMPRESSION WITHOUT DURAL OPENING FOR THE TREATMENT OF CHIARI I MALFORMATION
JONATHAN SHERMAN, M.D. JEFFREY J. LARSON, M.D. KERRY R. CRONE, M.D.
The treatment of patients with Chiari I malformation traditionally consists of posterior fossa decompression, with or without cervical laminectomy, combined with duraplasty. Posterior fossa decompression has been developed to relieve impaction created by tonsillar herniation and to cause regression of symptoms. Modifications of this treatment include fourth-ventricle-to-subarachnoid shunting to maintain free egress of cerebrospinal fluid (CSF), as well as plugging of the obex to prevent syrinx expansion. However, it is unclear whether the syrinx is in direct communication with the subarachnoid space. We have adopted the use of intraoperative ultrasound to evaluate decompression of the foramen magnum and the need for duraplasty in children with Chiari I malformation. Our treatment for patients with Chiari I malformation is based on the theory that posterior fossa decompression without duraplasty is adequate if CSF flow can be established.
INTRODUCTION
Children with Chiari I malformation are candidates for posterior fossa decompression. Contraindications to this treatment include the presence of any intradural pathology that requires dural opening, such as arachnoid scarring or sequestrations. Ultrasound may be performed in such patients; however, intradural exploration is still required. In patients with medical conditions that preclude surgery (e.g., bleeding disorders or cardiac anomalies), this procedure would be contraindicated until treatment is initiated.
PATIENT SELECTION
Magnetic resonance imaging (MRI) of the craniocervical junction is obtained preoperatively to determine the extent of the Chiari I malformation. Sagittal and axial T2-weighted images are important to visualize the CSF spaces ventral and dorsal to the cervicomedullary junction. MRI of the spine is helpful to demonstrate the presence and extent of a syrinx, especially in patients with signs and symptoms suggestive of a syrinx. Preoperative evaluation should include complete blood cell count, electrolyte profile, coagulation studies, and blood type and crossmatch. Prior to being placed on the operating table, the patient is intubated, an inhalation anesthetic is ad-
179
180
video monitor
instrument table
surgeon anesthesiologist
Figure 1. Superior view of the operating room layout during posterior fossa decompression. The surgeon stands at the patients left side, while the patient is prone. The assistant and the scrub nurse are on the patients right side. The ultrasound equipment is behind the assistant, allowing the surgeon a clear view of the image while ultrasonography is performed.
ministered, and intravenous antibiotics are completely infused. Prior to patient positioning, cardiac monitor leads, intravenous access, pulse oximetry, and an arterial line or a blood pressure cuff are placed.
quacy of decompression. A 7-MHz transducer (actual head size 1.2 2.2 cm) is introduced into the surgical bed, which has been irrigated with 0.9% NaCl.
SPECIAL EQUIPMENT
In addition to the standard surgical instrumentation needed for posterior fossa decompression, intraoperative ultrasound equipment is necessary. Sagittal and axial imaging of the cervicomedullary junction is obtained intraoperatively to evaluate the ade-
In the surgical suite (Figure 1) the patient is positioned prone, with the neck slightly flexed. The head is placed on a Mayfield headrest or in three-point fixation. The shoulders are pulled down to expose the suboccipital region and the posterior cervical spine from C1 to C6.
SURGICAL SET-UP
SHERMAN ET AL : POSTERIOR FOSSA DECOMPRESSION WITHOUT DURAL OPENING FOR THE TREATMENT OF CHIARI I MALFORMATION
181
A midline skin incision is made from the inion to the level of C3. The paraspinous fascia is dissected at the avascular linea albae and is retracted laterally to expose the occiput and the upper cervical laminae. Using a high-speed drill, the surgeon thins the suboccipital bone. Bone rongeurs are used to expose the posterior fossa and spinal dura. Bony decompression, which is carried laterally to the occipital condyles, provides lateral decompression of the cervicomedullary junction. An upper cervical laminectomy is performed to accommodate the extent of the descended cerebellar tonsils. Laminectomy is carried to the lateral aspect of the canal (Figure 2A). After completing the bony decompression of the suboccipital and cervical canal, the surgeon identifies a transverse dural band at the level of the cervicomedullary junction. This dural band, which consists of thickened dura and periosteum of the bony foramen magnum, must be incised to achieve decompression (Figure 2B). The band is divided with a #15-blade scalpel, being careful that the underlying dura is not incised. The thickened band is then peeled away from the dura using forceps (Figure 2C). Ultrasound images are obtained to reveal the level of tonsillar herniation. Sufficient decompression has been achieved if CSF can be identified dorsal to the tonsils and cervicomedullary junction by ultrasound on sagittal and axial views and if tonsillar pistoning (i.e., pulsatile rostral-caudal movement of the tonsils with little anteriorposterior pulsation (Figure 3A)) is absent. Ultrasonography is used to determine whether there is adequate CSF dorsal to the cervicomedullary junction and tonsils. Movement of the tonsils is also assessed. If the CSF spaces are small (i.e., less than 2 mm between the tonsils and dura (Figure 3B)) or if there is pistoning, then duraplasty is performed using a graft of periosteum or dural substitute (Figure 3C). An incision is made in the dura from the top of the bony decompression to the bottom. The dural opening is made lateral to the midline on one side, above the level of the foramen magnum to avoid opening the occipital sinus. To allow sufficient space inside the dura for adequate decompression, a graft is sewn in place using 5-0 Prolene sutures, running a simple suture to obtain a watertight closure. Intradural exploration and coagulation of the cerebellar tonsils elevate the tonsils above the level of the foramen magnum. Release of an arachnoid membrane, which is frequently found overlying the fourth ventricle and foramen of Magendie, is also performed. After adequate decompression is obtained, a Valsalva maneuver is performed to ensure, via ultrasonography, that foramen magnum obstruction by further tonsillar descent does not occur and to verify
OPERATIVE PROCEDURE
that no holes or defects were created in the dura. The paraspinous musculature is reapproximated using widely spaced, simple, interrupted 3-0 Vicryl sutures. The overlying cervical fascia is tightly closed, using simple, interrupted 3-0 Vicryl sutures. The skin is closed in the usual fashion.
The patient remains in a high-acuity care setting for 24 hours following surgery. Activity is increased progressively. The patient is discharged when dietary intake is adequate and he or she is ambulating well.
POSTOPERATIVE CARE
Approximately 6 weeks after the operation, MRI is performed to evaluate the CSF space around the tonsils and cervicomedullary junction. We have used cine-MRI to assess CSF flow. If the patient has an associated syrinx, repeat MRI at the appropriate spinal level is per formed to monitor the size. In our experience, a decrease in the size of the syrinx can be observed by 4 to 6 months postoperatively.
Follow-up Care
Most postoperative complications in posterior fossa decompression are a result of duraplasty. Frequently encountered complications such as CSF leakage, meningitis, hemorrhage, and arachnoiditis can be prevented when duraplasty is avoided. In one patient, a portion of the cerebellar tonsil herniated through a small hole that occurred in the dura during dural thinning. This caused infarction of the herniated portion of tonsil. After presenting with recurrent, intractable posterior headaches, the patient underwent duraplasty. Her symptoms resolved following this second procedure. We have subsequently discontinued the practice of dural thinning and release the fibrous band only at the level of the foramen magnum, as described.
Complications
We have found that posterior fossa decompression without duraplasty can adequately treat patients with Chiari I malformation. Intraoperative ultrasound is a useful adjunct to evaluate adequate decompression by assessing the amount of CSF around the descended tonsils in the obstructed foramen magnum. Avoiding duraplasty has decreased the incidence of postoperative complications. The use of this procedure without duraplasty has been limited to the pediatric population, but may also prove effective in adults. Further studies are needed to assess the amount of CSF necessary for adequate decompression, as well as to determine the long-term outcomes in patients in whom decompression is performed with and without dural opening.
CONCLUSIONS
182 A
transverse sinus
occipital sinus
B
vertebral artery
C1 laminectomy
bulging tonsil
marginal sinus
Figure 2. A, the posterior fossa and spinal dura are exposed following a suboccipital craniectomy and C1 laminectomy. Note the thickened transverse fibrous band at the level of the foramen magnum, still compressing the dura and underlying tonsils and brainstem. B, a right-angle dissector is inserted between the fibrous band and the dura. The fibrous band is then incised to prevent cutting the dura or marginal sinus. C, the fibrous band is peeled from the dura after incision, thus providing further decompression of the foramen magnum.
SHERMAN ET AL : POSTERIOR FOSSA DECOMPRESSION WITHOUT DURAL OPENING FOR THE TREATMENT OF CHIARI I MALFORMATION 2 mm CSF space
183
confluence
systole
C2
fourth ventricle
Figure 3. A, illustration of tonsillar pistoning (arrow) or the rostral-caudal movement of the tonsils that occurs in Chiari malformation during systolediastole pulsations. Loss of this pistoning movement, combined with adequate CSF space between the tonsils and the dura, indicates sufficient decompression of the foramen magnum. B, ultrasonic image demonstrating adequate decompression of the foramen magnum. Note the 2 mm of CSF space between the cerebellar tonsils and the dura posteriorly. C, in cases in which adequate decompression cannot be obtained without dural opening, a midline dural incision, placement of a dural graft, and duraplasty are used to widen the foramen magnum.
tonsil
Cranioplasty is among the oldest neurosurgical procedures. Trauma, infections, tumors, and compression caused by brain edema are some of the reasons for removal of cranial bone (Figure 1). Table 1 lists common causes of bone defects. Many authors suggest that cranioplasty may have a therapeutic effect that broadens previous indications for the procedure.
INTRODUCTION
Table 1 Causes of Cranial Defects Trauma Motor vehicle accidents Occupational accidents Urban violence Sports injuries War Neoplasia Meningioma involving the bone Eosinophilic granuloma Posterior fossa tumors Bone metastasis Infection Osteomyelitis Fungi Tuberculosis Miscellaneous Aneurysms Subdural hematoma Cranial decompression Cranial nerve neuralgias
DIAGNOSTIC STUDIES
1. skull x-ray 2. diagnostic and three-dimensional (3D) reconstruction computed tomography (CT) 3. diagnostic and phase-contrast magnetic resonance imaging (MRI) 4. transcranial cerebral oximetry (TCCO) with orthostatic challenge (Figure 2A and B) 5. electroencephalography (EEG) 6. xenon-enhanced CT cerebral blood flow (CBF) (Figure 2C and D) 7. neuropsychiatric evaluation (exit interview and Cognistat)
185
186
Figure 1. Causes of cranial defects. A, precranioplasty 3D reconstruction of the skull in a patient with a large cranial defect secondary to trauma. B, postcranioplasty 3D reconstruction of the skull in the same patient. C, precranioplasty 3D reconstruction of the skull in a patient with a large convexity meningioma. D, postcranioplasty 3D CT reconstruction in the same patient. E, precranioplasty plain CT in a patient with frontal sinusitis. F, precranioplasty 3D CT reconstruction of the same patient.
187
D
Figure 2. Diagnostic tests in cranioplasty. A, precranioplasty TCCO with orthostatic challenge. B, postcranioplasty TCCO with orthostatic challenge, showing return to precranioplasty values in the same patient. C, precranioplasty anatomical and quantitative Xenon-enhanced CT CBF study of a patient with a large cranial defect. D, postcranioplasty anatomical and quantitative Xe-CT CBF of the same patient, showing generalized improvement of CBF after the surgical procedure.
188
Table 2 Characteristics of a Good Cranioplasty structural integrity infection-resistant properties stability over time ease of fabrication precise fit minimal inflammation and graft-host reaction radioneutral for visualization material properties
Table 3 Characteristics of a Good Implant Material biocompatible high-impact strength Food and Drug Administration-approved and readily available nonbiodegradable non-exothermic during surgery no toxic fumes and/or residues MRI compatible
The fabrication and surgical fitting of cranial implants by traditional methods is dependent on subjective skills and procedures, producing in most cases a close but imperfect fit that necessitates revisions during the surgical placement of the implant. Results from our experience using computer-aided design and computer-aided manufacturing (CAD/ CAM) technology indicate that presurgical design and the fabrication of cranial implants using CT data from the patient results in a near perfect fit. Tables 2 and 3 list characteristics of a good cranioplasty and of a good implant material.
SURGICAL RECONSTRUCTION
Rapid prototyping is a term for several manufacturing processes that permit the creation of solid, physical objects directly from a digital file or computer data. The specific rapid prototyping technology used in our procedure is stereolithography technology from 3D Systems (Valencia, CA). Stereolithography was first applied commercially in the automobile industry by Chrysler Motors in 1989. The first primitive demonstration of the technology was in 1984,
Stereolithography
but its full development was not realized until a few years later. Since its development, stereolithography has been applied to a wide range of objects, from archaeology and environmental science to physical and paleoanthropology. In the medical field, it was first applied for maxillofacial prosthetics in the late 1980s and was later used to design cranial implants in the middle 1990s. Table 4 lists the characteristics of implant materials used in stereolithography vs. those used in intraoperative molding. Several authors have claimed to be the first to adopt the technology for cranioplasty. A procedure for generating CAD/CAM files to manufacture cranial implants to repair intermediate and large-sized skull defects was developed in 1996 by the Department of Neurosurgery and researchers in the Biomedical Visualization Laboratory, School of Biomedical and Health Information Sciences at the University of Illinois at Chicago in collaboration with commercial design and rapid prototyping products. The process involves the following steps: data acquisition (image acquisition, data processing, and model building); implant fabrication; and implantation.
Table 4 Comparison Between Intraoperative Molding and Stereolithography of Methylmethacrylate Intraoperative Molding Exothermic reaction Fume production Tissue reaction Graft-host reaction Operating time Hospital stay severe severe severe severe long long Stereolithography none none mild mild short short
189
laser beam is repeated, according to this sequence, slice-by-slice. In this way, a solid plastic form is gradually created in the photocurable polymer. The finished model is removed from the vat, drained of excess polymer, and submitted to a final curing stage using ultraviolet radiation. The final model is an exact physical representation of the computer model of the patients skull and defect (Figure 3E). To save time and the cost of additional material and buildtime, only the portion of the skull immediately surrounding the defect is manufactured.
The physical model generated through stereolithography (Figure 4A) serves as a template for the manufacture of the final implant. Initial Steps Modeling clay is applied to the plastic model of the patients defect. It is smoothed and contoured to represent the inner surface of the prosthesis/neurotissue boundary. A silicone rubber mold is made of this aggregated piece (Figure 4B). Dental stone is poured into the Silastic mold to produce a solid base upon which the prosthesis is designed. Design of the Implant A dental grade wax is applied to the negative space representing the skull defect. Care is exercised in maintaining appropriate contour and thickness to assure the best cosmetic appearance following implantation of the custom prosthesis. The Mold When the wax model is completed, the second half of the mold can be made. Keys are placed in the back piece of the mold so that the final two pieces can be perfectly realigned. A separator is applied to all exposed surfaces so that the two halves of the final mold will not bond together. Dental stone is poured onto the surface to an appropriate thickness and allowed to set until thoroughly hardened (Figure 4C). Casting the Implant The mold halves are separated and the wax is removed, leaving a negative space to receive the final implant material. A medical-grade methylmethacrylate is mixed to form a soft, pliable lump the consistency of warm butter. The material is placed into the mold and clamped tightly (Figure 4D). The clamped mold is placed in a hot-water bath to slow cure. Following a prescribed cure time, the clamps are removed and the solid methylmethacrylate implant is taken from the mold (Figure 4E). Additional curing in an oven releases free monomers trapped in the implant. After trimming and polishing, the finished implant is sent to the operating suite for ethylene oxide gas sterilization.
Implant Fabrication
190
Figure 3. Steps for using stereolithography in cranioplasty. A, plain skull CT slice. B, precranioplasty 3D soft tissue reconstruction. C, precranioplasty 3D bone reconstruction. D, computer-generated 3D model of the bone defect. E, final cranial prosthesis.
191
Figure 4. Steps in the prosthesis fabrication. A, stereolithographic model of the cranial defect. B, positive clay model. C, wax pattern or stone positive. D, mold in clamp.
192
Implantation
Patient Selection Any patient with a large cranial defect (9 sq cm) or who has a cranial defect with a complex shape is a candidate for implantation (Figure 5A and B). Presurgical tests are important for establishing the intracranial hemodynamic environment. The decision to perform a cranioplasty may be influenced by tests for seizure control, as well as knowledge of the neuropsychiatric status of the patient. Additionally, patients presenting with sinking scalp syndrome or the syndrome of the trephined may benefit from cranioplasty. Contraindications Contraindictions of cranioplasty include infection of the skin or bone, thin skin with compromised blood supply after radiotherapy, missing skin, or patients in poor clinical condition who may not withstand anesthesia and surgery well. Cranioplasty as a secondary procedure is per formed only after r esolution of the basic pathological processes that led to its avoidance as a primary procedure. Infection of the surgical site, allergic reaction to the implant material, graft rejection, hematoma formation, and carcinogenic potential are some of the general risks of the procedure. Preoperative Preparation In cases where there is concern about an active infection of the bone, gallium single photon emission CT (SPECT) is performed and an appropriate antimicrobial chemotherapy initiated. Seizure control, as evidenced by EEG, plasma drug levels, and no clinical evidence of seizures, and normal preoperative psychiatric status need to be established in order to avoid major complications postoperatively. Anesthesia and Monitoring Induction is performed with mesethlin and thiopental, and the patient is paralyzed with pancuronium. General anesthesia is administered with isoflurane after intubation with an endotracheal tube, controlled mechanical ventilation, arterial and venous
lines, and Foley catheter placement. Ventilatory control is established. Electrocardiography, arterial blood gases, pulse oximetry, body temperature, and mean arterial pressure are generally monitored intraoperatively. If necessary, a spinal drain may be introduced to control increased postoperative intracranial pressure. Patient Postioning and Draping Positioning is predicated on the location of the bone defect. Accordingly, the patient is positioned in the appropriate supine, prone, or lateral position (Figure 5C). After shaving and cleansing the surgical site, the patient is draped in the standard fashion dictated by the cranial defect. Skin Incision To avoid postoperative healing that might compromise the flap, the skin incision replicates that used in the previous surgery. Operative Procedure The incision site is infiltrated with 1% lidocaine and epinephrine, and the incision made. Raney skin clips (Codman, Inc., Randolph, MA), if needed, are used for hemostasis. The flap is undermined and the dural attachments are released (Figure 5D and E). Dissection continues until the edges of the skull defect are visualized. The surgical site is irrigated with bacitracin. The implant is removed from the bacitracin bath in which it was preserved following the ethylene oxide sterilization (Figure 5F). Holes are drilled in the surrounding bone and implant with a Midas Rex high-power drill (Midas Rex, Fort Worth, TX). The implant is fixed in place and screwed with a minimum of three to four titanium plates and screws to effect immobilization of the prosthesis and restitution of the calvarium (Figure 5G). After careful hemostasis of the periosteum and musculocutaneous flap, the incision is closed in two layers. The subcutaneous/galea incision is closed with 0-0 interrupted inverted Dexon, and the skin is closed using staples (Figure 5H and I).
Figure 5. Implantation of the prosthesis. A, posterior view of the cranial defect. B, lateral view of the cranial defect. C, patient in prone position. D, opening of the skin and dissection of the musculocutaneous flap. E, liberation of adhesions. F, implant in bacitracin solution. G, detail of the screwed titanium plates. H, postoperative posterior view. I, postoperative lateral view.
193
Figure 5.
Hematoma, infection, seizures, cerebrospinal fluid leak, and skin dehiscence may occur. With the intention of early detection of any acute postoperative complication, a 3D reconstruction of the skull is made from a 1-mm CT slice, within the first 24 hours, to assess the fit of the implant. Patients are generally sent home 1 day after surgery and return to the outpatient clinic 1 week later. At that time, xenon-enhanced CT, transcranial cerebral oximetry with orthostatic challenge, neuropsychiatric analysis, and plasma anti-epileptic drug levels are measured.
Computer-aided cranial implant design and fabrication procedures appear to provide a better quality product, a near perfect custom fit to the individual patient, thereby decreasing patient risk in the form of reduced operating room time and better implant fit and function. Computer-aided preoperative implant manufacture methods yield cost savings as a result of reduced operating room time and shortened hospital stays for patients requiring implant surgery.
CONCLUSION
In the last several years, there has been a renaissance in the neurosurgical management of patients with movement disorders. Advances in anatomical and functional imaging, improvements in physiological localization techniques, the wide availability of stereotactic systems, and the realization of the limitations of medical therapy have all been instrumental in the resurgence of surgery for the treatment of movement disorders. Currently, stereotactic thalamotomy and posteroventral pallidotomy are the most commonly employed lesioning procedures for movement disorders. Despite being effective, lesioning has implicit limitations. Lesioning offers a one-time benefit and is not adaptable to clinical changes or disease progression. Additionally, the irreversible nature of lesioning can result in significant morbidity, particularly with bilateral procedures. Complications associated with thalamotomy include speech disturbances such as dysarthria, cerebellar signs, gait difficulties, and cognitive dysfunction. Complication rates as high as
INTRODUCTION
60% have been reported with bilateral thalamotomy. In patients undergoing pallidotomy, cognitive, speech, motor, and visual field deficits have been reported. An increasingly utilized alternative mode of therapy is chronic electrical stimulation of specific intracranial structures, also called deep brain stimulation (DBS). DBS involves the placement of multicontact electrodes in the thalamus, basal ganglia, or white matter tracts. The exact mechanism of DBS action is unknown. However, prevailing theories implicate a depolarizing blockade or jamming of neurons and/ or axonal tracts resulting from high-frequency stimulation. This chronic stimulation effectively mimics a lesion, but is reversible. The major advantage of DBS is its adjustable and reversible feature, allowing for maximal efficacy while minimizing complications. Currently the most common application of DBS is thalamic (ventralis intermedius (VIM)) stimulation for refractory tremor (essential tr emor, Parkinsons disease, cerebellar) and pallidal stimulation for rigidity from Parkinsons disease. The improved patient safety and the striking benefits of DBS have expanded the possibilities of intervention into new targets. One such target area of increasing attention is the subthalamic nucleus (STN).
195
196
The STN plays a crucial modulatory role in influencing the basal ganglia projections to the motor thalamus and the brainstem. Although a logical target for intervention, the association of the STN with hemiballismus has been worrisome. Recently, however, reports from several centers have shown safety and efficacy of bilateral STN stimulation for Parkinsons disease. These studies have demonstrated that chronic high-frequency bilateral STN stimulation results in improvements in all of the cardinal motor manifestations of Parkinsons disease (akinesia, bradykinesia, rigidity, and tremor), including the most resistant and disabling axial components of gait, posture, and balance. The overall results have been very encouraging and, at times, dramatic.
symptomatology of Parkinsons disease causing abnormalities of the motor thalamus and its upstream projections (supplementary motor cortex, premotor cortex, and primary motor cortex) as well as in the brainstem regions involved with the control of posture and gait. Indeed, in MPTP-treated primates, dopamine agonists, radiofrequency lesioning, or chronic high-frequency electrical stimulation of the STN result in the resolution of motor dysfunction.
The rationale for targeting the STN lies in its distinct anatomical connections to the basal ganglia and the brainstem (Figure 1). The STN comprises approximately 540,000 neurons and measures 10 mm (medial-lateral) 8 mm (anterior-posterior) 6 mm (dorsal-ventral). It lies inferior to the thalamus, superior to the substantia nigra, and posteromedial to the pallidum and the internal capsule. The STN receives input from the motor cortex as well as inhibitory gamma-aminobutyric acid (GABA) input from the globus pallidus externus (GPe) via the indirect pathway. The STN sends excitatory (glutamate) efferents to the output nuclei of the basal ganglia (i.e., the globus pallidus internus (GPi) and the substantia nigra pars reticulata (SNr)). Additionally, the STN sends projections to the GPe and striatum as well as to brainstem nuclei such as the pedunculopontine nuclei, a nucleus of importance for axial control and locomotion. The currently proposed models of basal ganglia function are based on work in the 1-methyl-4phenyl-1,2,3,6-tetrahydropyridine (MPTP) primate model of parkinsonism and studies in patients with Parkinsons disease. These models have shown that the degeneration of the substantia nigra pars compacta (SNc) and the resultant dopamine deficient state lead to an overactivity of the STN and GPi/SNr. The increased STN activity is believed to be secondary to the loss of inhibition from the GPe via the indirect pathway. The overactive STN through its excitatory projections to the GPi/SNr causes hyperactivity in the basal ganglia output, which has also lost its striatal inhibition via the direct pathway. The overall effect in patients with Parkinsons disease is overactive GPi/SNr nuclei with corresponding excessive inhibitory (GABA) efferents to the motor thalamus and the brain stem. This hyperactive basal ganglia output is implicated in the generation of the
The indications for treating patients using STN stimulation are expanding and evolving. Patients considered for STN DBS must be examined by a movement disorder specialist and meet the clinical criteria for the diagnosis of idiopathic Parkinsons disease. These include tremor, rigidity, akinesia/bradykinesia, postural instability, and gait disturbance. It is not known if patients with secondary parkinsonism and the Parkinson plus syndrome are suitable candidates for STN stimulation. At our center, STN DBS candidates typically will have bilateral, axial manifestations of Parkinsons disease and are advanced in their disease, with a significant disability despite optimization of their medication regimen. Many also have dyskinesias and severe on-off fluctuations. Advanced age is not necessarily a contraindication for the procedure, as patients in their 80s have successfully undergone DBS. It should be noted that the selection process should be case specific. A patient with predominantly unilateral drug-induced dyskinesias may be a better candidate for pallidotomy than STN stimulation. Surgical candidates should not have a bleeding disorder or contraindications to general anesthesia. Patients must be motivated and have sufficient cognitive functions and reserves to undergo an awake surgery requiring their active participation. Additionally, patients and family must be able to effectively use the stimulator and be willing to participate in frequent follow-up visits for detailed evaluation and adjustments of the stimulator. Other important considerations include living in close proximity to or having ready access to a center with expertise in DBS and having resources to deal with long-term hardware/system maintenance and associated complications.
PATIENT SELECTION
Preoperative Investigations
All patients must be evaluated preoperatively in the off phase (approximately 12 hours without medication) as well as in the on phase (1 hour after the usual morning dose of medication). This baseline evaluation should consist of objective standardized parkinsonian scales incorporated into the Core As-
197
Figure 1. Schematic diagram demonstrating the inhibitory (GABA, blue arrows) and excitatory (glutamate, red arrows) projections between the striatum, the subthalamic nucleus (STN), basal ganglia, motor thalamus, brainstem locomotion/gait centers, and cortical motor areas. Note the excitatory projection of the STN to the output nuclei of the basal ganglia: globus pallidus internus (GPi) and substantia nigra pars reticulata (SNr). These output nuclei, in turn, have inhibitory input to the motor thalamus as well as descending inhibitory input to the pedunculopontine (PPN) and the mesencephalic area (MEA) of the brainstem (believed to play an important role in axial control of posture and locomotion). The STN receives inhibitory input from the globus pallidus externus (GPe) via the indirect pathway, whereas the GPi receives inhibitory projections from the striatum via the direct pathway. Not all projections (including the cholinergic pathways) are demonstrated in this diagram. The PPN/MEA also have reciprocal ascending input to the Gpe, GPi/SNr, STN, thalamus, and striatum. The substantia nigra pars compacta (SNc) with its dopaminergic projections to the striatum is also not shown. In Parkinsons disease, degeneration of the SNc results in the loss of dopaminergic input to the striatum. This causes a hyperactivity of the GPi/SNr secondary to loss of inhibition via the direct pathway and hyperactivity of the STN via loss of inhibition through the indirect pathway.
sessment Program for Intracerebral Transplantation (CAPIT), such as the Unified Parkinsons Disease Rating Scale (UPDRS), the Hoehn and Yahr staging scale, and the Schwab and England disability scale. Additionally, all patients should undergo a formal neuropsychological evaluation.
held overnight. This accentuates the pathophysiological findings seen with microelectrode recording and allows for optimal assessment of the intraoperative clinical improvements in patients with stimulation.
Whenever possible, the patient should undergo the procedure in the off state, with all medications with-
OPERATIVE TECHNIQUE
Any one of several commercially available stereotactic frames can be applied under local anesthesia (1:1 mixture of 0.25% bupivacaine and 2% lidocaine). This can be performed in the operating room or in a
198
procedure room adjacent to the imaging suite. At our institution, we use the Leksell G frame (Elekta, Inc., Atlanta, GA) which can be attached to the skull with sharp pins or with carbon fiber pins inserted into holes drilled into the outer table of the skull. The surgeon also has the option of applying ear bars to assist in frame alignment. Most patients, however, find the ear bars to be uncomfortable. The frame should be applied such that it is parallel to the anterior commissure (AC)-posterior commissure (PC) line. As a guideline, an imaginary line from the external auditory meatus to the lateral canthus can approximate this AC-PC line. After frame application, the magnetic resonance imaging (MRI) or computed tomography (CT) localizer (fiducial box) is attached to the frame and the patient undergoes either CT or MRI for anatomical localization. Ventriculography is not used in most centers as the currently available MRIs with high-resolution volumetric acquisitions obviate the need for this invasive technique. MRI has a higher anatomical resolution than CT, but is more prone to distortions in spatial accuracy. Any of the currently available image-correction algorithms or MRI-CT image fusion software can minimize these distortions. At the Toronto Hospital, we use a Signa 1.5-tesla MRI magnet (General Electric, Milwaukee, WI). After the initial image acquisition, a midline sagittal slice is chosen to best identify the AC and the PC. A subsequent volumetric scan is performed along the ACPC line providing coronal, sagittal, and axial images. The sections are 1 mm thick and non-overlapping. Images are acquired using a gradient echo sequence with a relaxation time of 43 msec and an excitation time of 13 msec (flip angle 450, FOV 24 24, matrix 256 256).
PC intercommissural line onto the digitized map at the sagittal laterality of interest. We usually start with the laterality of 12 mm from the midline that corresponds to the plane with the maximum diameter of the STN in an anteroposterior direction. To correct for individual variations in the AC-PC distance, this corresponding 12-mm sagittal map is shrunk or stretched and reformatted to the length of the intercommissural line for each specific patient. The subsequently generated brain map is overlaid onto a millimeter grid ruled in stereotactic coordinates in the anteroposterior and dorsoventral scales with a corresponding diagram of the brain nuclei and tracts depicted in the chosen laterality (Figure 2). Our final anatomical target is chosen at the center of the STN on this reformatted brain map. Typically, structures encountered in our trajectories include the anterior thalamic nuclei such as the nucleus reticularis, ventralis oralis anterior, ventralis oralis posterior nuclei, zona incerta, H2 fields of Forel, STN, and SNr/SNc in the midbrain.
Anatomical localization of the STN can be achieved in two ways: 1) direct targeting; or 2) indirectly using a standardized anatomical atlas as a function of the AC and PC. For direct targeting, STN can be visualized on coronal T2-weighted MRI or various other sequences such as an inversion recovery sequence. On axial MRI, the STN is located anterolateral to the red nucleus and posterior to the cerebral peduncle. At our institution, we utilize the anatomical brain atlas method for targeting. On MRI and/or axial CT slices, the stereotactic x, y, and z coordinates of the AC and PC are calculated using the MRI/CT console software program. The calculated AC and PC stereotactic coordinates are fed into a computer with a commercially available program containing digitized diagrams of sagittal brain sections from a standardized brain atlas (G Schaltenbrand, W Wahren: Atlas for Stereotaxy of the Human Brain. 2nd ed, 1977). The program transcribes the patients calculated AC-
Subsequent to stereotactic CT/MRI image acquisition and anatomical target localization, the patient is taken to the operating room and placed in a supine position, with the head elevated approximately 30 degrees. Intravenous prophylactic antibiotics are administered. The frontal region is shaved and prepped in the standard fashion. The procedure is performed with the patient under local anesthesia with 2% Xylocaine. A 5-cm linear or curvilinear incision is made 2.5 cm lateral to the midline and centered anterior to the coronal suture. Raney clips are placed for hemostasis, and a self-retaining retractor is applied. Using a 14-mm diameter drill bit on a power drill, a burr hole is made anterior to the coronal suture. The dura is coagulated and incised, in a cruciate fashion. The pial surface is also coagulated and incised, avoiding surface vessels. Fibrin glue is applied to minimize cerebrospinal fluid leaks and entry of air into the cranial cavity. When performing simultaneous bilateral STN procedures, an identical sequence of steps is carried out on the opposite side. The stereotactic arc is then applied, and the x, y, and z coordinates for the anatomical STN target are set. A guide tube cannula with a blunt tip stylet is then introduced into the brain parenchyma to a point 10 mm proximal to the chosen target. Physiological localization begins at this time, with the ultimate aim of correlating the anatomical and physiological findings.
Surgical Procedure
In our experience, physiological localization is mandatory for definitive target determination. Sole reliance on anatomical localization can be problematic because of the frequent discrepancy between the
199
Figure 2. A 12-mm sagittal map of the Schaltenbrand and Wahren Atlas reformatted to an actual AC-PC distance of a patient and ruled in stereotactic coordinates for the Leksell frame. A typical trajectory of the microelectrode through various nuclei is depicted. Each gradation is 1 mm. The + in the center of the subthalamic nucleus (STN) represents the chosen anatomical target. This point is annotated as 0 on the trajectory. H2 = fields of Forel, Hpth = hypothalamus, IC = internal capsule, mcp = midcommissural point, Rt = nucleus reticularis of the thalamus, SNr = substantia nigra pars reticulata, Vim = ventralis intermedius, Voa = ventralis oralis anterior, Vop = ventralis oralis posterior, ZI = zona incerta.
200
expected location and actual position of the stereotactic targets. Physiological corroboration can be achieved with microelectrode recording and stimulation and/or macrostimulation. Macroelectrode stimulation is rapid and requires minimal equipment. However, it has low spatial resolution and is not able to record neurons and to discriminate between axons and neurons. Micr oelectrodes, however, provide exquisite physiological identification of receptive fields and neuronal firing patterns via direct measures of individual single-unit neuronal activity and are able to distinguish somatodendritic from axonal activity. We rely exclusively on microelectrode recording and stimulation for STN localization. Microelectrodes provide precise mapping information, which guides the surgeon in determining the most optimal location for final electrode implantation. The microelectrodes are made of Parylene-C insulated tungsten electrodes plated with gold and platinum. Tip lengths range from 15 to 40 m with impedance of 0.2 to 0.5 MOhm. The microelectrode is housed in its own guide tube, which is the same length as the larger guide tube described above. A manual hydraulic microdrive is used to advance the microelectrode in submillimeter steps beyond the most distal aspect of the cannula into the brain parenchyma (typically 10 mm superficial to the chosen anatomical STN target (Figure 2)). The electrical signal derived from single- and multi-unit neuronal recordings is amplified, filtered, displayed on an oscilloscope, and fed to an audio monitor. A window discriminator is used to determine the discharge frequency of the recorded units. Thus the relative size and shape of the action potential is visualized, audio monitoring of neuronal activity is achieved, and the firing frequency of individual neurons determined. As an adjunct to microelectrode recording, microelectrode stimulation is performed by passing electrical current of 0.1 to 100 A (1-second train, 300 Hz, and 100-sec pulse width) through the tips of the electrode. The recording is typically performed continuously, whereas stimulation is done every 1-2 mm. The clinical stimulation responses are correlated with electrical thresholds to further determine the proximity to a given location. Usually, the lower the threshold of stimulation, the closer the proximity to the area of interest. In a typical STN case, three or four trajectories (anterior, posterior, medial, and lateral) are made for physiological localization. With simultaneous bilateral pr ocedures, the opposite side usually requires one or two trajectories because of the brain symmetry. Trajectories are made by removing the guide tube assembly and moving the arc in the desired direction in increments of 2-3 mm. The ultimate goal of physiological localization is to identify the STN. In order to accomplish this, struc-
tures anterior (internal capsule), dorsal (zona incerta), superior (zona incerta, nucleus reticularis, and the ventralis oralis anterior and posterior), and ventral (SNr) must be localized (Figure 2). The characteristic frequencies of cellular activity encountered in each of these structures are shown in Figure 3.
There are two characteristic types of neurons observed in the anterior thalamic nuclei. One type is called a bursting cell, which has a slow firing rate interposed by characteristic rapid bursts of activity. These cells usually have a firing rate of 15 to 25 spikes/second but the bursts can reach firing rates up to 300 or 400 spikes/second. These characteristic bursting cells have also been described in other thalamic regions as well as in r ecordings from laboratory animals. The other major cell type is a nonbursting but irregular firing cell. These have firing rates ranging from 15 to 25 spikes/second and may be cells of the nucleus reticularis, the ventralis oralis anterior, or the ventralis oralis posterior.
As the microelectrode passes ventrally to the anterior thalamic nuclei, a relatively quiet region is encountered. This is the thalamic fasciculus (H1 fields of Forel), which contains pallidofugal fibers with very few cells. The zona incerta is a thin strip of gray matter below this fiber tract with a variable recording pattern. Occasionally, we have recorded large-amplitude regular-firing or, less frequently, burst-firing neurons from this region in the range of 25 to 45 spikes/second. Alternatively, the zona incerta can be a relatively quiet region. Ventral to the zona incerta is the lenticular fasciculus (H2 fields of Forel) containing pallidofugal fibers, which is also a relatively quiet region.
As the microelectrode enters the STN, an increase in background noise is detected reflective of a region of increased cellular density. In Parkinsons disease patients in the off state, STN neurons have an irregular firing pattern with a mean rate of 37 Hz (range 25 to 45 Hz). The cellular activity in the dorsal portion of the STN can be modulated by active and/or passive movements of limbs. These movement-related responses are clear, reproducible, and should be actively sought during the recording. The ventral aspect of the STN may also have movement-related neurons, but in our experience these are more prevalent in the dorsal regions.
Subthalamic Nucleus
201
Figure 3. Graph showing the mean neuronal firing frequencies (spikes/second) of neurons encountered in trajectories traversing the anterior thalamus, STN, and SNr. The anterior thalamic nuclei consist of the nucleus reticularis, ventralis oralis anterior, and ventralis oralis posterior. AnTh-B = anterior thalamic cells with bursting characteristics; AnTh-NB = anterior thalamic cells with no bursting activity.
In the patient with clinically apparent tremor, cells with spontaneous discharge rates synchronous with the tremor (i.e., tremor cells) can be observed. Microstimulation can result in tremor reduction or tremor arrest. In addition to tremor cells, occasional bursting cells can be encountered in the STN, but with less prevalence than in the anterior thalamic nuclei. Microstimulation can occasionally result in the patient experiencing paresthesias, presumably from current spread to the medial lemniscus. In more anterior and lateral trajectories, axons of the internal capsule are encountered and, with microstimulation, contralateral muscle contraction can be observed.
The substantia nigra pars reticulata and pars compacta lie ventral and posterior to the STN and are characterized by cells with a regular pattern of firing with higher rates than the STN (mean 71 Hz, range 60 to 80 Hz). Occasionally, cells can be found that respond to saccades. Microstimulation (particularly in the medial regions of the SNr) can result in ocular
deviation due to current spread to the oculomotor nerve fascicles. Occasionally, paresthesias can be reported, presumably from current spread to the posteriorly located medial lemniscus. The characteristic physiological findings of microelectrode recording and stimulation are summarized in Table 1. The physiological information derived from the corresponding microelectrode recording (receptive fields) and microelectrode stimulation (projected field) along each tract/trajectory is annotated on the corresponding stereotactic brain maps, allowing simultaneous visualization and correlation of the physiological and anatomical findings (Figure 4). This assists in the selection of the optimum location for implantation of the DBS electrode.
General principles guiding the final implantation of the electrode involve placement of the electrode at a region allowing for maximal clinical response while minimizing the undesired side effects. The ideal location for the placement of the electrode is in the trajectory, which is traversing the midportion of the
202
Table 1 Characteristic Physiological Findings Using Microelectrode Recording and Stimulation Structure Thalamic nuclei: reticularis, ventral oralis anterior and posterior Zona incerta Subthalamic nucleus Microelectrode Recording Bursting cells Microelectrode Stimulation No clinical effect
Sparse cells (bursting) Irregular firing pattern Mean rate (37 Hz) Range (25-45 Hz) Movement responsive Neurons Tremor cells
Regular firing pattern Ocular deviation Mean rate (71 Hz) Paresthesias Range (60-80 Hz) Saccade and movement responsive Neurons
Paresthesias are most likely due to current spread to the medial lemniscus. *Contralateral muscle contraction secondary to current spread to the internal capsule. Ocular deviation secondary to stimulation of third cranial nerve.
STN with the most distal lead being placed in the most ventral portion of the STN. This arrangement allows for the contacts/poles to be within the STN, thus providing maximal programming possibilities. The DBS electrode we currently use is a quadripolar electrode (Figure 5) supplied by Medtronic, Inc. (Minneapolis, MN). Each pole/contact is made of cylindrical platinum/iridium alloy and is 1.5 mm in length, separated from the other pole by an insulated distance of 1.5 mm or 0.5 mm, depending on the model (Figure 5). The diameter of the electrode is 1.27 mm, and the entire electrode length can be 28 or 40 cm, depending on the model. To confirm the position and trajectory of the actual DBS electrode, the use of intraoperative x-ray or fluoroscopy is mandatory. This is accomplished via a beam orthogonal to the stereotactic frame. At this time, the x, y, and z coordinates of the determined target are set on the frame. A lead holder assembly (Medtronic) is attached to the frame. This assembly device stabilizes and secures the DBS electrode and the cannula. A stylet-coupled cannula (1.6 mm in diameter) is inserted to the target under fluoroscopic guidance and verification. Subsequently, the DBS electrode is inserted under fluoroscopic
guidance to the target and the cannula is removed (Figure 6). The mere introduction of the cannula/ DBS lead causes the patient to exhibit dyskinesias and choreiform movements, which is a good prognostic sign that clinically verifies the target. At this time, intraoperative test stimulation is performed by attaching the proximal portion of the DBS lead to a hand-held pulse generator device (Screener). The patient is tested at the most proximal and distal pole combinations for therapeutic and adverse effects. If adverse effects (motor contraction, paresthesias, or ocular deviation) are persistent at low thresholds (<2 V, 120 Hz, 60 msec), the electrode may be repositioned to a more favorable site according to the anatomical/physiological map. Finally, the stylet is removed and the DBS electrode is anchored to the burr-hole ring (a plastic ring designed to fit the 14-mm burr hole) by wedging the electrode into one of its two grooves. An additional plastic burr-hole cap further locks the DBS lead into the burr hole ring. This maneuver is designed to prevent lead migration, which was a significant problem in the past. At this time, final fluoroscopic imaging is performed for verification. For simultaneous bilateral STN electrode implantation, the same procedure is
203
Figure 4. Microelectrode recording findings of three trajectories superimposed on the 12-mm lateral sagittal brain map are used for initial anatomical targeting as shown in Figure 2. Each gradation is 1 mm. Black circles represent single unit cellular recordings. The corresponding frequencies are shown. Solid bars = quiet regions with sparse neurons; red dots = neurons responsive to passive and active movements of areas depicted by red circles; hfd = high frequency discharge >80 Hz; BC = cells with characteristic bursting activity; red P = paresthesias experienced with microstimulation in the distal lower extremity, most likely due to current spread to the medial lemniscus; hpth = hypothalamus.
204
Figure 5. The quadripolar DBS electrode. Each pole/contact is made of cylindrical platinum/iridium alloy and is 1.5 mm in length and 1.27 mm in diameter, separated from the other pole by an insulated distance of 1.5 mm (Model 3387, left) or 0.5 mm (Model 3389, right). (Courtesy of Medtronic)
Figure 6. Intraoperative stereotactic radiograph of the DBS electrode at the target. The electrode is introduced from a precoronal burr hole and directed posteriorly and ventrally. The position of the electrode is monitored via fluoroscopy. The black circle represents the target, which is also the center of the stereotactic ring. Note the appearance of the four contacts/poles and the cannula, which has been withdrawn proximally.
205
Figure 7. The implantable pulse generator (Itrel II) unit is powered by a lithium battery and fully programmable via telemetry. The quadripolar DBS electrode is also shown. (Courtesy of Medtronic)
repeated on the opposite side. Alternatively, the opposite side can be implanted at a later time. We routinely implant bilateral STN stimulators simultaneously. After DBS implantation, intraoperative test stimulation, and securing of the DBS electrode, one can proceed directly to the implantation of the pulse generator. Alternatively, the DBS lead can be connected to a percutaneous extension for prolonged testing and obtaining imaging studies prior to permanent internalization with the pulse generator.
This stage of the procedure should be carried out with the patient under general anesthesia as it requires a tunneling procedure that can cause significant discomfort and anxiety. Intravenous prophylactic antibiotics are administered. The patient is positioned in a similar fashion as for a ventriculoperitoneal shunt, placed supine with the head turned toward the opposite side, with a shoulder roll on the ipsilateral side. A 5-cm horizontal incision is made below the clavicle and a subcutaneous pocket is made for placing the generator. The implantable pulse generator (IPG) device (Itrel II, Medtronic) is powered by a lithium battery and is fully programmable via telemetry (Figure 7). The scalp opening is connected to the infraclavicular opening using a tunnel-
ing device similar to a shunt passer (provided by Medtronic). An electrode extension wire is passed retrograde from the infraclavicular opening to the scalp. The distal end of this extension is connected to the IPG device. The pulse generator is sutured to the pectoralis fascia to prevent migration. Similarly, at the cranial opening, the proximal portion of the extension wire is connected to the proximal portion of the DBS lead by tightening the four screws. A plastic cover/ sheath is place over the connector and sutured at both ends to keep fluids out and maintain contact point isolation. Finally, a two-layered standard wound closure is performed (Figure 8). Figure 9 demonstrates the final position of the DBS electrodes with postoperative MRI.
The goal of DBS programming is to provide the best clinical response with minimal or no side effects. DBS programming is achieved via a telemetry device capable of adjusting parameters such as the choice of poles, mode of stimulation, frequency, pulse width, amplitude, and continuous or cycling stimulation. The mode of stimulation can be unipolar, bipolar, or multipolar. Each of the four poles/contacts (designated as 0, 1, 2, or 3) can be positive, negative, or off. The direction of current is from cathode () to anode
206
Figure 8. Composite photograph showing the overall configuration/positioning of the implanted bilateral DBS electrodes, connectors, and pulse generators.
(+) in bipolar stimulation, and the case of the IPG device serves as the ground for unipolar stimulation. The amplitude of stimulation can range from 0.1 to 10 V (usually 2 to 3.5 V for STN). The frequency (number of pulses/second) can go as high as 185 Hz with the currently available models. Like thalamic (VIM) stimulation for tremor, effective STN stimulation is best achieved using a high frequency (120 Hz). The pulse width (duration of pulses) is commonly 60-90 sec. Choosing the optimal stimulation parameters can be a complex task and can require hours of programming. Most stimulation parameter adjustments are made in the initial 1-3 months after implantation. After this period, stimulation parameters usually remain stable. The patient also has the capability to turn the stimulator off and on with a hand-held magnet. The life expectancy of the pulse generator varies depending on the parameter and duration of stimulation. In general, a pulse generator can be expected to last 3 or 4 years. The multiple combination of stimulation settings is particularly advantageous as it allows the best clinical response to be achieved with the fewest side effects. The determination of the optimal selection parameters, however, is individualized. Each time a setting is changed, the patient must be assessed objectively and subjectively in a systematic and consistent fashion. This process can consume much
B
Figure 9. Postoperative T1-weighted MRI, axial (A) and sagittal (B) views, demonstrating the final position of the DBS electrodes.
207
time and resources, particularly with bilateral STN stimulation. Patients with advanced disease and axial symptomatology require more frequent adjustments and careful clinical evaluation, while those with tremor-dominant Parkinsons disease require less programming. An experienced team with available resources is required to achieve these goals most efficiently and effectively.
The reversible and adjustable features of DBS technology have permitted its application to the STN in the patient with advanced, bilateral, and medically refractory Parkinsons disease. DBS not only prevents undesired side effects, but also can be adaptable as the symptoms change or the disease progresses. There have been several recent reports of follow-up with chronic high-frequency bilateral STN stimulation. These studies have documented striking improvements in parkinsonian symptoms. Improvements in overall off state UPDRS motor scores have been 60% or greater with the most striking improvement in tremor, followed by akinesia/bradykinesia and rigidity, as well as significant improvements in gait. There have also been significant impr ovements in dyskinesias and on/off motor fluctuations, as well as reduction in the use of drugs by 40%-50%. In our own series of patients who have undergone chronic high-frequency bilateral STN stimulation, we have observed total UPDRS motor score improvements in the medication off state of 63%, including tremor (82%), akinesia (62%), rigidity (60%), and gait and postural instability (51%). In the on state, the overall improvement of total UPDRS motor scores was 47% and L-dopa-induced dyskinesias were reduced by 76% with total drug dosage being decreased by 40%. The complications associated with STN DBS can be categorized as those related to the surgical procedure, stimulation, and technical/equipment. The complications of the surgical procedure are similar to those seen following other stereotactic DBS or lesioning procedures. The most serious complication inherent to any intracranial stereotactic procedure is that of intracranial hemorrhage, which occurs on the order of 1%-3%. The additional small risk of pneumocephalus is minimized by sealing the dural opening with fibrin glue. Transient complications such as intraoperative and postoperative confusion and agitation have also been seen (up to 2 weeks), but most commonly in older individuals. Other complications
have involved neuropsychiatric changes with subtle memory deficits, word finding and concentration difficulties, and personality changes. The stimulation-induced side effects include apraxia of eyelid opening, diplopia, paresthesias, and involuntary movements such as chorea, all of which can be alleviated by adjusting stimulation settings. Transient hemiballism has also been observed with stimulation. The potential hardware-related complications are on the order of 1%-3% and include electrode migration, breakage, disconnection, skin erosion, and equipment failure. Additionally, there is approximately a 5% chance of infection that is most often superficial.
Important issues to take into consideration in the initial decision-making process include the cost of a bilateral system and the long-term system maintenance (including equipment-related complications and battery replacement). Another consideration is the proximity of the patient or ready access to a center with expertise in the management of DBS. Additionally, programming can require significant time and resource expenditures. A center dedicated to this endeavor must be carefully planned and have adequate resources for the long term. Finally, patients must be willing and have the ability to cooperate with the extensive evaluation and prolonged programming that may be necessary to reach the optimal settings.
PRACTICAL CONSIDERATIONS
The reversible and adjustable features of DBS technology have allowed its application to the STN. The STN is an ideal target in patients with Parkinsons disease, because of its overactivity and its projections to the basal ganglia output nuclei (GPi/SNr). Single cell microelectrode r ecording technology can precisely and reproducibly localize the STN. STN neurons have an irregular firing pattern with a firing rate in the range of 25-45 Hz. They can also exhibit movement-related activity and/or activity in synchrony with the tremor. Microelectrode stimulation is less useful for STN localization. Although complications related to the procedure, hardware, and stimulation can occur, the overall marked benefits obtained by these disabled patients outweigh the adverse effects. In patients with advanced Parkinsons disease, bilateral STN DBS results in a definitive improvement in all of the cardinal motor manifestations of Parkinsons disease.
CONCLUSION
Minimally incisional spine surgery is undergoing a revolution at the current time. Over the last decade, we have seen advances in anesthetic techniques, coupled with technical advances in the size, diversity, and scope of the instruments available to the surgeon. With those advances have come radically different attitudes toward the possibilities of minimally incisional surgery. Orthopedic and neurological surgeons have had at their disposal a variety of surgical techniques to approach the lumbar disc since it was first approached in 1934. The evolution of these techniques has revolved around the need to remove the offending material and reduce the patients pain, while preserving neurological function. Although fraught with complications, the open pr ocedure is one of the most common techniques for alleviating pain in this area and remains a procedure commonly performed by orthopedic and neurological surgeons. The percutaneous approach utilizing the endoscope extends the knowledge that we have acquired over the past 50 years. This approach dates to 1951, when Hope described an anterior retroperitoneal approach to the disc space. In 1975, Hijikata expanded this approach when he described the percutaneous technique. Kambin has pioneered and extensively devel-
INTRODUCTION
oped instruments for targeted posterior fragmentectomy, herein called arthroscopic microlumbar discectomy. In 1982, Kambin described the technical aspects of the operation as we shall describe it in this chapter. Since that time, the further development of instrumentation has made this technique one that, with the appropriate training, can be a part of the armamentarium of all spine surgeons.
Patients are selected for arthroscopic microlumbar discectomy as they would be for any discectomy approach. Patients with a history of lumbar root symptomatology of either myotomal loss, dermatomal pain, or/and sensory loss are candidates. Patients usually have not responded to conservative therapy. Conservative therapy persists for a period of not less than 6 weeks nor longer than 18 weeks, and consists of land- or water-based exercises for flexion/extension of the spine. Patients who present with motor loss or other acute neurological syndromes can certainly be evaluated and treated more promptly as the situation dictates. At the conclusion of the conservative therapy phase, patients should undergo imaging studies. At the current time, magnetic resonance imaging (MRI) of the spine affords the most comprehensive view as to the origin, extent, and contributing factors of compressive nerve root pathology. Patients with a clear-cut single-level root symptom ideally present with MRI evidence for focal disc pathology. In some instances, disc pathology may be combined with a minimum of anterior bony disease causing a mini-
PATIENT SELECTION
209
210
mum of lateral recess stenosis. Patients whose imaging studies reveal extensive lateral recess stenosis, focal bilateral stenosis, or additional other bony pathology would not be candidates for this lateral extracanalicular approach. In addition, local instability such as antero- or retrolisthesis is a relative contraindication to the procedure. Also not feasible for removal due to limitations with the equipment is disc material that has migrated the equivalent of one disc space level above or below the index. At the time of counseling for endoscopic surgery, patients are advised of the possibility of conversion to an open procedure should the pathology so dictate. Patients with tumors in the foramen, within the canal space, and a combination of bony and epidural pathology should be counseled as to the limited nature of biopsy in resection possible through the scope. However, that is not the subject of this review. The risks and complications of arthroscopic microlumbar discectomy include those experienced in the open procedure: 1) infection (seen in 1% to 3% of patients in the current series); 2) failure to relieve symptomatology (seen in 13% of our patients; symptomatology was relieved in 87%); and 3) transient worsening of symptomatology (seen in 2% to 4% of patients). This worsening takes the form of dysesthesia occurring frequently along the anterior aspect of the calf or thigh. Dysesthesia is described as burning in nature, occurs most often on Day 5 following the procedure, and has been self-limited in most patients. There have been no neurovascular complications to date. The preoperative selection of the patients is neither age nor sex limited. However, patients who cannot comfortably be positioned in the prone position due to orthopedic or soft-tissue deformity of the lower extremities are not candidates. In addition, patients whose anatomical size would preclude the instrument use are also not candidates. This can be determined by a simple measurement from the lateral aspect of the disc space to the skin entry point described in this article. This is best accomplished using computed tomography (CT) at the symptomatic level.
become obvious candidates for general anesthesia when their level of anxiety and/or their previous experience with conscious sedation so dictates to the anesthesiologist. The anesthesia of choice, however, is intravenous conscious sedation. Occasionally, while relying predominantly on midazolam or propofol, we have found intravenous ketamine to be useful, especially if the patient is large and the apparent volume of distribution makes the administration of conscious sedation more challenging. Despite adequate intravenous sedation for the initial steps of the procedure described here, the occasional patient complains of pain at the time of manipulation of the annular fibers. We have found that the topical use of fentanyl applied via cottonoid patty to the annular fibers has been helpful. In some cases, 50 g fentanyl injected into the annular fibers will afford and extend relief. Antibiotics are also administered as part of the preoperative preparation, including 1 gm Ancef administered intravenously within 1 hour of incision; some patients presenting with penicillin allergies are given 120 mg gentamycin intravenously when renal function is appropriate. Single preoperative doses are employed.
The anesthetic technique used is determined by the surgical team. Intravenous conscious sedation utilizing fentanyl and midazolam and/or propofol has enjoyed large acceptance in the institutions in which we currently practice this surgery. In some patients, however, the use of spinal anesthetics or general anesthesia is necessary for the successful completion of the procedure. During preoperative counseling, some patients
Patient positioning commences with the patient being requested to turn prone on a radiolucent operating frame (Kambin Frame, Northstar Medical, Philadelphia, PA), which allows for the straightening of lumbar lordosis and gentle positioning and cushioning of the chest and hips (Figure 1). The arms are brought up alongside the patients head and the legs are flexed over two pillows. To avoid the complication of strap pressure, adequate padding is required over the lower extremities. The fibular head can inadvertently be trapped under the operating table belt. Bladder catheterization is not generally employed. Preoperatively, we utilize the biplanar fluoroscopic x-ray device (OEC Medical Systems, Salt Lake City, UT) to ensure proper positioning and readily readable x-ray images. The surgeon has the option of standing on the side of the patient ipsilateral or contralateral to the pathology. The fluoroscopic arm is best positioned opposite the side that the surgeon stands to allow for ease of use during the operation with a minimum of movement in the area of the field. The camera tower (Stryker Systems, Stryker Instruments, Kalamazoo, MI) and additional arthroscopic materials including the printing device are best positioned at the foot of the patient in clear view of the surgeon. We usually allow the camera device to be positioned at the foot of the patient, the x-ray fluoroscopic television immediately to the surgeons left of the camera tower, and the C-arm device to the left, completing a 90-degree arc from the foot of the patient to the point opposite the surgeon.
PATIENT POSITIONING
211
Skin preparation is accomplished utilizing surface agents (e.g., Betadine or alcohol) supplemented with an Ioban drape (3M Corp., Minneapolis, MN) to cover the entire operative field, three disc levels above and below the target level. Laterally, the drape should extend at least 12 cm off the midline. The preparation is augmented by a shower curtain-type drape (Ioban 2 patient isolation system 6617, 3M Corp.). This facilitates manipulation of the biplanar fluoroscope during the procedure. The electrocautery device and suction are positioned at the head or foot of the patient to facilitate C-arm movement and to avoid the lines crossing the patient during x-ray evaluation. The C-arm is ideally positioned in the anteroposterior (AP) plane at the start of the procedure, and the angle of the C-arm is adjusted so that the
OPERATIVE TECHNIQUE
disc space is seen straight ahead after the appropriate disc level has been selected. A line is drawn on the skin utilizing a marking pen to indicate the level selected and identified, for example, L4-5. Laterally, the skin entry point 9 to 11 cm from each midline is marked and the entry point is selected (usually 9 cm) and infiltrated at the skin level using 0.5% lidocaine without epinephrine. The muscles are infiltrated as well with lidocaine. The instruments used are the Smith and Nephew arthroscopic surgical system and Kambin surgical set (Memphis, TN) (Figure 2). The camera equipment is a Dyonics Surgical System (Andover, MA). We use a Sony color video printer to record the pr ocedure. The pictures are useful for review and teaching purposes. A 22-gauge spinal needle measuring 18 cm is introduced at the skin entry point at an angle of 35-45
212
degrees. The preoperative imaging study, whether CT or MRI, may be utilized to determine the exact angle of entry as measured from the triangular working zone to the point of entry on the skin. The triangular working zone (Figure 3) is the annular entry port through a posterolateral approach that is formed anteriorly by the descending spinal nerve from the level above, inferiorly by the proximal plate of the inferior vertebrae, and posteriorly by the proximal articular process of the lower lumbar segment. As the needle is introduced and penetrates the sacrospinalis, quadratus lumborum, and psoas major muscles, it is gently rotated to allow the bevel to carry the needle through a straight line toward the target level. The biplanar fluoroscopic arm can be utilized to check the progress of the needle. The needle has entered the triangular working zone when the tip of the needle seen in the AP projection lies at the lateral margin of the bipedicular line formed by the ipsilateral outer margin of the pedical above and below the disc space selected. In the lateral view, the needle will lie just at the inferior margin of the disc space at the level selected (Figure 4). Needle geometry is important so as to avoid too steep an entry angle which would project the surgeon past the disc space and into the retroperitoneal structures or too shallow an angle which would project posteriorly behind the lamina and other spinal structures. Needle geometry is also important so as to avoid too lateral an entry, which carries the potential to traverse the peritoneal space prior to entering the retroperitoneal muscles. The biplanar fluoroscopic C-arm images are most helpful in avoiding incorrect needle positioning when seen in true orthogonal relationship. A technical note regarding the L5-S1 disc space: This disc space is best approached by imaging the sharp angle of the endplates in the lateral view and projecting that point to the skin. It will become apparent that to remain parallel with the endplates, the needle entry point may come to lie at the level of the L4-5 disc space as seen in the AP projection. Preoperative evaluation of the imaging studies will also identify the position of the iliac crest relative to the disc space. A high crest as seen on CT or MRI will force a more cephalad entry point for successful cannulation of the disc space. Occasionally, the height of the crest will obstruct the approach. The goal of the lateral projection is to maintain the plane of the needle parallel with one of the endplates in the disc space. This will allow the safe docking and entry of the instruments into the disc space. This emphasis distinguishes the angle normally employed during a biopsy when simple triangulation of the lateral edge of the disc space is appropriate. Following successful needle placement, the needle stylet is removed and a small metal wire is placed in the center barrel of the needle into the disc space,
confirmed by biplanar fluoroscopy. At this point, a #15 scalpel blade is used to incise the skin and subcutaneous fascia. The cannulated obturator is introduced over the stylet through the muscle layers until within 1-2 cm of the annulus. The stylet is removed while approaching the annulus to avoid inadvertently pinning or trapping the nerve root in the triangular working zone. The biplanar C-arm is utilized to demonstrate placement of the cannulated obturator at the lateral aspect of the bipedicular line and at the posterior margin of the disc space. The universal access cannula is now introduced over the end of the obturated cannula. At this point, the edge of the annulus can often be assessed by tapping against the obturated cannula and feeling a sense of steady pressure, which then springs back against the thumb. The operation then proceeds by attaching the suction cap to the access cannula. The zero-degree forward-looking endoscope can be introduced through the suction cap. Irrigation is provided via a 1- or 3-L bag of saline positioned at the edge of the operative field. Control of flow is accomplished in the field with the usual stopcocks. The scope allows for inspection of the annulus. The pattern of annular fibers and fat can be visualized to distinguish the annulus from the periosteum of the bone above and below (Figure 5). The biplanar C-arm can be utilized to demonstrate the level of visualization. The annulus can then be cleaned utilizing a cottonoid patty introduced into the universal access cannula and moved around the end of the cannula with either the back of the obturator twirled between the fingertips or a small grasping forceps to place the patty against the annulus and wipe off some of the fat. At this point, the cottonoid patty can be used to introduce the fentanyl at the level of the annulus. We soak the 1-cm cottonoid patty in 50 g fentanyl and apply it directly to the annulus, leaving it in place for 1 minute. After inspection of the annulus, it is helpful to use a spinal needle to check each of the four quadrants in the universal access cannula, with the 12 oclock position posteriorly or dorsally on the patient, the 3 oclock caudally, the 9 oclock rostrally, and the 6 oclock ventrally as viewed from the end of the access cannula. This confirms the position of the cannula over the disc space. Prior to the application of the local anesthetic, We find it helpful to move the scope to visualize the exiting root, thereby avoiding inadvertent injury. The annulus can be infiltrated during these maneuvers as well with 25 g fentanyl or 1-2 ml 1% lidocaine without epinephrine. Straight trephines 3 and 5 mm in diameter are introduced through the universal access cannula and twisted until they penetrate the annulus. Once the 5-mm trephine has been introduced, the access cannula can be gently advanced 0.5 cm to allow docking or firmly seating it in the annular fibers. This allows for
213
Figure 3. Drawing of the triangular working zone, viewed axially (upper) and sagittally (lower).
Figure 4. Radiographs in anteroposterior (left) and lateral (right) projections showing the needle position.
214
stability at the annular fibers as well as at the operators hand at the skin level. Alternatively, a small locking cap can be placed over the cannula. We find hand stabilization to be superior, especially for the initial cases. A zero-degree endoscope can be introduced for visualization of the intradiscal material (Figure 6). When introducing the scope, it is useful to briefly irrigate to clear material and occasional blood products in the access cannula; then reduce or turn off the suction to allow a column of water for visualization and magnification. Via the arthroscope photographic towers, photos can be taken through the endoscope and the steps of the operation documented as well as providing a teaching file. Once the presence of disc material has been confirmed through a zero-degree endoscope, it can be removed utilizing the various forceps in the instrument set. Commonly, one starts with the straight forceps to evacuate material directly in the path of the approach. The deflecting forceps are used to move more posteriorly in the disc space. Enough disc material is removed to enable room to introduce the shaver. The shaver has a forward and reverse foot pedal and is provided with constant irrigation via a universal access cannula. Suction is attached to the shaver head itself. A suction trap is used to collect disc material as it is removed. Caution should be taken to engage the drive motor for the shaver only when it is fully engaged in the disc space. Traversing the edge of the cannula with the shaver engaged can cause neural elements to be drawn toward the shaver. The shaver is utilized in a 270-degree arc which excludes the posterior 90 degrees running roughly between 10 oclock and 2 oclock positions in the clock face previously described (i.e., avoiding the most dorsal portion of the disc space at the level of the root). Zero- and 30-degree scopes can be reintroduced as needed during the shaving process to inspect the posterior-most elements. When a biportal exposure is utilized, similar steps are followed to this point with the introduction of the needle, stylet, obturated cannula, access cannula, scope, trephines, and disc space evacuation. During this phase, it is possible to introduce a zero- or 30degree endoscope through one portal and the shaver through another portal and observe the function and motion of the shaver and other grasping instruments (Figure 7). Alternatively, through a unilateral approach, a working scope with a central channel can be introduced to allow simultaneous visualization of the removal of disc material. Another alternative technique utilizes the oval cannula with an external diameter of 9 mm, which allows for simultaneous as well as metachronous visualization. Disc removal continues posteriorly with the angle-edged instruments until elements of the posterior longitudinal
ligament are identified. These fibers are oriented in parallel, as opposed to the amorphous nuclear material. On many occasions, disc material that has penetrated the posterior longitudinal ligament and/or annular fibers can be retrieved, allowing the dura to be inspected (Figure 8). Once disc removal is judged to be adequate, the universal access cannula can be disengaged and moved somewhat more posteriorly along the edge of the annulus, allowing for visualization of the lateral end of the foraminal contents. This maneuver enables inspection of the posterior margin of the triangular working zone and, on occasion, removal of small bony elements from the vertebral body that may impinge on the lateral aspect of the foramen. During routine disc removal, this also allows for inspection of and removal of fragments that are in a paraforaminal location. Fragments that are totally extraforaminal, the so-called lateral disc, are particularly advantageous to remove as the extracanalicular approach of the scope brings the surgeon down directly on the offending pathology. After completing the operation, inspection of the annular surface allows for confirmation of the extent of removal of the protruding disc material and subsequent decompression of the nerve root visualized immediately rostral to the disc space. Bleeding occasionally occurs in the venous elements lateral to the annulus and can be controlled with gentle irrigation, the application of Surgicel through the scope apparatus, or the utilization of a monopolar coagulation wire. The monopolar coagulation wire can be applied with the working channel scope and performed under direct vision (Ellman Wire, Surgitron Corp.). We have found that the initial inspection of the disc space and the subsequent inspection at the end are best accomplished with the zero-degree scope when moving along the annulus exterior to the disc space. However, within the disc space, the 30-degree and 70-degree endoscopes allow for visualization especially along the posterior margin of the disc space at the level of the expected pathology. The extent of fragment removal and subsequent disc removal is obviously determined by the operating surgeon. We have found a useful classification regimen that identifies disc material in the axial and sagittal planes as visualized on MRI. In the axial plane, disc material can occur in a medial, paramedial, foraminal, or extraforaminal location moving from medial to lateral in any given disc space. In the sagittal view, the material can be interannular, subligamentous, or extraligamentous and, in the latter category, the extraligamentous material is at the level of the disc space or either above or below the disc space. Disc fragments that have migrated beyond the confines of the posterior ligament and/or migrated within the epidural space are going to present the greatest challenge in the early phases of
215
Figure 8. The posterior longitudinal ligament and the dura are visible at left.
216
acquiring the skill of arthroscopic microdiscectomy. Removal of these disc fragments should not be attempted until a surgeons learning curve justifies that approach. The instruments have limitations as to the degree of posterior reach that they enjoy. In addition, exploration through the annulus and posterior ligaments into that epidural space increases the possibility of inadvertent dural injury and subsequent cerebrospinal fluid egress. Irrigation is continued throughout the procedure, often using 3 L of normal saline. We have not seen any systemic changes or local nerve root problems using this irrigation regimen. On occasion, when there is evidence of mild venous oozing near the root at the end of the procedure, a small Hemovac drain can be placed directly through the universal access cannula and removed in 6-8 hours. In most patients, the access cannula is removed, a single nonabsorbable suture is placed at the skin line, and a single SteriStrip bandage is placed on the skin; the patient is then taken to the recovery room.
Postoperatively, clinical signs and symptoms dictate the degree of success. Occasionally, however, patients present with possible residual symptomatology
POSTOPERATIVE FOLLOW-UP
or symptomatology that can be attributed to other structural abnormalities. We evaluated the usefulness of immediate postoperative imaging in this series of patients. We found that MRI and CT examinations are both quite sensitive to the removal of offending disc material in the canal space. When the preoperative studies are compared to the studies obtained in the first 12-24 hours after surgery, residual air in the previous area of disc protrusion, resumption of a more normal canal configuration along the posterior margin of the vertebral body, and removal of the laterally placed disc herniation can be clearly visualized. We have looked at the percent of canal clearance in these patients and demonstrated a high degree of correlation with successful canal clearance and subsequent clinical outcome. In the current series, a successful outcome has been seen in approximately 87% of patients. As with many procedures in this region, other factors, including the presence of workers compensation, litigation, and recurrent disc disease have been associated with reduced incidence of successful outcomes. Although the procedure has many similarities to open disc surgery, advantages include a smaller incision, no canalicular dissection, and minimal disc space collapse as indicated by long-term follow-up.
The use of thoracotomy to treat spinal disease originated with the treatment of spinal tuberculosis or Potts disease. The abysmal results following laminectomy for herniated thoracic disks have led surgeons to seek novel approaches for this problem, and in 1969, Perot and Munro, and later Ransohoff and colleagues described the transthoracic removal of herniated discs. This is now a standard and very useful procedure for treating thoracic intervertebral disc disease, particularly lesions that are large or midline in location. Symptomatic thoracic disc herniations are unusual. It is estimated that 0.25%-0.75 % of clinically apparent discs will occur in this region and most frequently present between T8 and T11. The typical presentation involves radicular pain, which can either be acute or chronic. Given the risks of this type of surgery, nonsurgical options such as use of anti-inflammatory drugs or nerve blocks should be considered in patients with pain alone and no neurological deficit. Involvement of the cord can produce paraparesis or sensory disturbances that are generally subacute or chronic in nature, although sudden paraplegia can be seen. Myelopathy is an indication for surgery.
INTRODUCTION
Precise preoperative imaging is key in the surgical planning for herniated thoracic disc. Magnetic resonance imaging (MRI) (preferably) or computed tomography myelography must clearly define the level of the offending disc. Thoracic spine x-rays are useful to provide identifying landmarks for proper intraoperative localization. Although a seemingly formidable surgery, with careful patient selection morbidity and mortality rates using transthoracic discectomy are not unreasonably high. We have performed the procedure in patients up to 70 years old with underlying coronary disease. However, certain patients, including the elderly, those with chronic obstructive pulmonary disease, and persons who smoke do represent higher risk groups. Generally, however, surgery can be planned on an elective basis to allow thorough evaluation of the preoperative pulmonary and cardiac status. However, urgent or emergent surgery is mandatory when there is rapid loss of lower extremity function. Pulmonary function should be assessed with an arterial blood gas and spirometry. A satisfactory arterial blood gas value includes a pO2 >90 mm Hg, pCO2 <42 mm Hg and O2 saturation >92%. Value ranges of pO2 60-80 mm Hg, pCO2 42-45 mm Hg, or O 2 saturation 90%92% would be considered borderline acceptable. Values out of these ranges place the patient at a much higher surgical risk. On spirometry, the following results are associated with significantly higher pul-
217
218
Figure 1. T1-weighted non-contrasted MRI demonstrating a large right-sided, laterally herniated disc at the T10-11 level which resulted in sudden paraplegia. A right-sided thoracotomy was used to removed the fragment.
monary complications: forced vital capacity (FVC) <1.5 liters, FEV1 (forced expiratory volume at 1 second) less than 1.0 liter, FEV1/FVC <35% or forced expiratory flow of 25%-75% (FEF25%-75%) <1 liter/ second. Premorbid cardiac disease, including a history of angina, congestive heart failure, or even diabetes, carries a much greater risk for perioperative cardiovascular complications. Cardiac evaluation should be directed by the patients internist or cardiologist and may include a treadmill test in asymptomatic individuals between ages 45 and 60 years or a thallium exercise scan in patients aged over 60 or with a history of myocardial infarction, angina, or heart failure. (An intravenous dipyridamole-thallium scan may be obtained in a patient unable to exercise.) All patients should refrain from smoking for at least 2 weeks before surgery and those with underlying lung disease should have their pulmonary status optimized with bronchodilators or treatment of any chronic bronchitis or bronchial edema.
brae are easily accessible, with excellent visualization of the anterior and lateral portions of the disc space as well as the spinal canal. To reach the extreme levels of the thoracic spine, specialized techniques must be utilized. The uppermost thoracic vertebrae can be approached via either a posterior transthoracic technique elevating and retracting the scapulae or by an anterior route, possibly requiring a median sternotomy. The thoracolumbar junction is obscured by the diaphragm; therefore, techniques to expose the T11, T12, and L1 vertebrae must include managing the diaphragm.
Using the posterolateral thoracotomy approach described here, the fourth to the 10th thoracic verte-
OPERATIVE TECHNIQUE
Selecting the side of the surgical approach for a thoracotomy is largely a matter of preference. Some surgeons prefer using a left-sided thoracotomy to provide access because of greater ease in dealing with the aorta than the venous structures on the right, although at the superior aspect, the arch and great vessels may be of concern. Other surgeons prefer to use a right-sided approach because of concern of injuring the artery of Adamkiewicz, which usually enters on the left side in the region of the thoracolum-
Surgical Approach
HLAVIN AND HARDY : EXCISION OF HERNIATED THORACIC DISC VIA THE TRANSTHORACIC APPROACH
219
B
Figure 2. Patient positioning and incision. The patient is secured in a lateral position with the chest centered over the break in the table to facilitate opening the interspace during surgery. A beanbag device or padded sand bags (not shown) can support the patient. Care must be taken to protect the axillary and peroneal areas as well. The course of a typical incision is shown (dotted lines). A, surgeons view from above. B, side view, surgeon facing the patient.
bar junction. On the other hand, when working at the lower portions of the thoracic spine, the liver is more difficult to retract, making a left-sided approach more desirable. Should instrumentation be required, injury to the aorta is of concern, and again a right-sided approach is preferred by some. Clearly, a lateralizing disc, as shown in Figure 1 (or other pathology), can dictate the route of choice. Other considerations include coexistent lung pathology, which might dictate opening a particular side of the chest. In most cases, however, approach from either side can be used.
bronchial blocker. A urinary catheter is placed. Spinal cord monitoring with somatosensory evoked potentials may be helpful, although the infor mation provided is clearly limited and injury to the anterior spinal cord can occur without significant changes in the signal. As with other procedures, perioperative antibiotics are given. Perioperative steroids may be used in cases with significant cord compression.
Surgery is carried out with a general anesthetic technique utilizing at least two large-bore intravenous access (16 gauge) catheters and arterial blood gas monitoring. Some patients may require more invasive central venous pressure or Swan-Ganz cardiac monitoring. The use of a double lumen-cuffed endotracheal tube (e.g., Carlens tube) permits deflating the exposed lung out of the way should additional room for retraction be necessary. Other methods can be used to separate lung ventilation if necessary, such as a
Anesthesia
As shown in Figure 2, the patient is placed in a lateral position for surgery, rotated slightly forward so that the chest contents will tend to fall away from the spine with gravity. A beanbag device or sand bags padded in towels can be used to maintain this position, as long as the support is radiolucent to allow intraoperative x-ray localization. The chest should be positioned over the break in the operating table so that it may be flexed during surgery to open the interspace and thus improve working room. At completion, the break can be returned to neutral to facilitate closure. The axilla is supported and protected with a roll while the legs are cushioned with pillows. The lower leg is bent while the upper leg is kept
Patient Postioning
220
extended. The upper arm is supported with pillows and bent at about a 90-degree angle to rotate it out of the field. The patient is secured well to the operating table with 3-inch tape to enable safe rotation during the procedure. The patient may be tilted slightly toward the surgeon to allow for a better view.
For a standard thoracotomy, the incision is typically centered over the rib, which potentially may be removed. The nipple is a marker for the fourth interspace. In young patients, intercostal ligaments may be reasonably supple such that use of a rib spreader will provide adequate retraction and exposure. Alternatively, a portion of the rib numbered by the vertebral body superior to the offending intervertebral disc can be removed. In the thoracic spine, it is important to remember that any particular rib articulates with the posterior/superior margin of the vertebral body with which it shares a number. For example, as shown in Figure 3, the head of the T9 rib articulates with the body of T8 and T9 at the posterolateral aspect of the offending disc space. One accurate way to count ribs is to place ones hand inside the pleural cavity and count down from the top and then up from the bottom. The counts should agree before a resection is performed. Alternatively, a plain spine x-ray should be obtained. To obtain optimal exposure and working room, it is best to remove the rib numbered by the vertebrae superior to the disc space. (Thus, as shown in Figure 3, the T8 rib is removed laterally to gain access to the T8-9 space.) This compensates for the angulation of the ribs inferiorly across the anterior aspect of the vertebral body and facilitates working inferiorly rather than superiorly. As depicted in Figure 2, the incision should follow the rib and curve slightly cephalad posteriorly, where it extends to approximately halfway between the spinous process and the medial portion of the scapula. Anteriorly, the incision curves toward the sternum. The skin incision is made initially with a scalpel. Thereafter, Bovie cautery is used for hemostasis and division of the musculofascial layers, which includes the trapezius and rhomboid muscles medially and the latissimus dorsi and serratus anterior muscles deeper and extending anteriorly. Large vessels should be ligated and divided. The periosteum of the rib to be resected is incised with the Bovie cautery, then stripped from the superficial surface using a Cobb or other periosteal elevator. After clearing off the undersurface of the rib with a curette, a pigtail rib stripper can be used to dissect off the remaining periosteum and elevate the rib out of its bed (Figure 4A). Care should be taken to identify and save the neurovascular bundles that travel along the inferior border of each rib. (Some surgeons prefer ligating the neurovascular bundle with 2-0 or
Operative Procedure
3-0 silk ties because it is believed to eliminate stretch injury to the nerve that has been implicated in postoperative pain syndrome.) At this point, the rib to be resected may be cut out with a rib cutter, dividing as posteriorly as possible, close to the costovertebral angle. Should an attempt be made to preserve the rib, the intercostal muscles may be split additionally to enable further separation of the intercostal space. The pleura is grasped with smooth forceps and opened using scissors (Figure 4B). After protecting the ribs with cotton tape, a self-retaining (Finochietto) retractor is placed in the pleural cavity and opened. The lung may be covered with a moist lap pad and retracted out of the way. A wide ribbon retractor or sweetheart retractor may be used to accomplish this. Judicious placement of the ribbon under the Finochietto retractor can eliminate the need for manual retraction. At this point, the vertebral bodies and other anatomic structures should be visible through the parietal pleura (Figure 5). The intervertebral discs can be both visualized and palpated as the prominent bulging areas of the vertebral column while the intervening concavities represent the vertebral bodies. The bottom of the concavity approximates the mid portion of the body. Radiographic confirmation of the proper disc space before proceeding is crucial at this point. An open-sided rectangular incision is then made in the parietal pleura, extending from the mid portion of the vertebral bodies above and below the affected disc space and extending laterally at either end (Figure 6). The pleura is then dissected carefully from the vertebral body and reflected laterally, thereby exposing the intervertebral vessels as well as the sympathetic chain. The segmental vessels should be identified and ligated using 3-0 silk ties in order to provide adequate bony exposure. Running with the segmental artery and vein are the intercostal nerves. The intercostal nerve can be traced to the neural foramina of interest, which then enables identification of the pedicle, an important landmark. It is important to remember, however, that the head of the rib overlies the pedicle and occasionally needs to be removed to obtain adequate exposure of the thecal sac (Figure 7). In some instances, although not routinely, it may be necessary to drill off a portion of the pedicle as well. At this point, the operative microscope is brought into place. The annulus of the offending disc is incised anteriorly and the disc space is grossly evacuated using a combination of pituitary ronguers and curettes. This initial debulking is crucial prior to any attempts to remove the extruded fragment compressing the neural elements, as this provides exposure and working room. It is important to remember to work away from the spinal cord. Flexion of the operating table at this juncture may be used as a maneu-
HLAVIN AND HARDY : EXCISION OF HERNIATED THORACIC DISC VIA THE TRANSTHORACIC APPROACH
221
1 2 3 4 5 6
Figure 3. Anatomic diagram of the rib cage, posterior view. This picture depicts the articulation of the ribs with their respective bodies and interspaces and demonstrates the bony resection that would be used for a T8-9 herniated disc through a left-sided approach. The head of the T9 rib articulates with the bodies of T8 and T9 at the posterolateral aspect of the offending disc space. The head has been removed to gain access to the T8-9 disc space. In addition, the T8 rib has been removed laterally to provide working room.
1 2 3 4 5 6 7 8
7 8
9 9 10 11 12 12 10 11
Figure 4. Rib resection and pleural opening. A, the rib to be resected, still lying in its bed, has been cleared off with periosteal elevators and the underside dissected free to allow insertion of a pigtail rib stripper. B, after bony resection, the now-exposed rib bed and pleura are grasped with forceps and divided.
222
CEPHALAD
sympathetic trunk
aorta CAUDAL
lung
CEPHALAD
diaphragm
esophagus
B
intercostal arteries, veins, nerves Figure 5. Intrathoracic anatomy. A, right-sided view. B, left-sided view.
Figure 6. Pleural incision. The dashed line indicates the window opening through the pleura (left-sided approach).
HLAVIN AND HARDY : EXCISION OF HERNIATED THORACIC DISC VIA THE TRANSTHORACIC APPROACH
223
Figure 7. Removal of the rib head. The parietal pleura has been reflected, segmental vessels ligated, and the head of the rib drilled off. The dashed line indicates where the rib was removed.
Figure 8. Bony resection and disc removal. A, the annulus has been incised and the disc space preliminarily emptied using long pituitary ronguers and curettes. Additional bone is being drilled off the superior and inferior endplates to widen working room. B, large osteophytes or calcified ligament can require more extensive bony removal of the bodies and pedicle to visualize the thecal sac and ensure adequate decompression. C, anatomic diagram demonstrating extent of final bony removal.
224
ver to open the disc space. Alternatively, a disc space spreader may be placed into the interspace. Drilling off a small portion of both the superior and inferior endplates as well as the vertebral body enlarges the exposure and provides access to the herniated fragments (Figure 8A). Additional exposure of the dura may be obtained by resecting the head of the rib abutting the intervertebral disc space and r emoving the superior portion of the pedicle (Figure 8B). Any calcified posterior longitudinal ligament or fibrotic annulus should be resected either with Kerrison ronguers or by forcing the material into the interspace and removing it. At completion, palpation of the dura with a No. 4 Penfield should be performed to confirm adequate decompression. A typical vertebral bony resection is shown in transverse section in Figure 8C.
chest tube is attached to an underwater suction drainage. It is subsequently removed when drainage has subsided and no air leaks are present. An immediate postoperative chest film as well as daily chest xrays should be obtained to confirm resolution of any pneumothorax.
Return of the operating table to a flat position facilitates wound closure. The parietal pleura is reapproximated using interrupted absorbable sutures of either chromic catgut or 3-0 Vicryl sutures. An apical dependent chest tube is externalized through a stab wound and the chest is then closed serially in layers. A rib approximator is used to initially place 0-0 Nurolon stitches around the ribs. A running absorbable stitch is used for the individual muscular layers. The skin can be stapled or closed with an absorbable subcuticular stitch such as 4-0 Vicryl sutures. The
WOUND CLOSURE
Risks of surgery include death, paraplegia, hemorrhage, infection, cerebrospinal fluid (CSF)-pleural fistula, post-thoracotomy pain, or kyphotic deformity. Careful patient selection and preoperative cardiopulmonary assessment with judicious perioperative monitoring will help minimize morbidity and mortality from this procedure. Clearly, manipulation around the spinal cord poses a threat for neurological injury. An ample bony decompression to provide working room without retracting the cord as well as always working away from the cord will help decrease risk of damage. An extensive bony removal must be weighed against the risk for postoperative kyphotic deformity from collapse. Should a significant bony defect be present, placement of a bone graft can reduce the risk of long-term angulation. Care must be taken to watch for a CSF leak at the end of the procedure, with inspection under a Valsalva maneuver. Should there be suspicion of a rent, this should be sealed with a muscle or fascia graft and fibrin glue and a lumbar drain placed for several days postoperatively to avoid a CSF fistula.
SURGICAL MANAGEMENT OF ADVANCED DEGENERATIVE DISEASE OF THE LUMBAR SPINE WITH MULTIPLANAR DEFORMITY
MICHAEL F. OBRIEN, M.D. GARY L. LOWERY, M.D., PH.D. A. EUGENE PENNISI, M.A.
It is not possible to predict the natural history of untreated adult spinal deformity. The deformity may present in many forms, either idiopathic, degenerative, or the result of a post-surgical imbalance with induced deformity either above or below the index fusion area. In the vast majority of adults with severe spinal deformity, degenerative disc disease is the cause. The onset of adult degenerative scoliosis is rarely seen in patients younger than age 40. Once believed to be stable in comparison to adolescent idiopathic disease, long-term studies have shown that these deformities progress anywhere from at least 1 to 3 per year. Followed over 5 to 10 years, these deformities can result in a crippling disability. As with all spinal disease, the first course of treatment is conservative. For patients with minimal symptoms, observation may be sufficient. In those in whom the symptoms are back pain alone, a course of nonsteroidal anti-inflammatory medications, bracing, and aerobic conditioning programs may be of use. Unlike adolescent idiopathic scoliosis, bracing probably has less of a role in the elderly. In patients who present with symptoms of spinal claudication, an aerobic conditioning program may be of benefit. This, along with epidural steroid injections and nerve root blocks, may temporize through a period of symptom exacerbation or, in some fortunate patients, give sufficient relief to obviate the need to con-
INTRODUCTION
sider surgery. When all conservative means have failed and the patients symptoms suggest that surgery is the most appropriate option, patient selection and a clear understanding of the structural pathology are the two most important factors in the successful surgical treatment of this disease. The surgical treatments are really the sequential application of standard anterior and posterior spinal techniques. These include anterior lumbar intradiscal fusion, posterior lumbar intradiscal fusion, pedicle screw placement, anterior instrumentation techniques, corpectomies, and osteotomies. Other techniques include derotation for correcting coronal deformities and restoring lordosis. The appropriate application of these techniques in a well thought-out manner is necessary for the successful surgical treatment of these patients. An understanding of the peculiarities of each unique curve, as well as the underlying pathology, is essential. Without a clear understanding of the unique segmental abnormalities in each deformity, it is impossible to correctly determine the surgical technique most appropriate (whether anterior, posterior, or combined) and the area over which the surgery needs to be performed to stabilize the spine in the corrected coronal and sagittal planes. Failure to achieve a balanced correction, to address each abnormality at each level, and to anticipate changes in the remaining discs above and below will lead to long-term failure. Ultimately, success will result from the appropriate realignment of the structural elements in both the coronal and sagittal projections and through decompression of stenotic levels to eliminate the effects of nerve root traction and spinal stenosis.
225
226
The most important portions of the radiological workup are the plain x-rays. These should include long-standing anteroposterior and lateral scoliosis films. In addition, long-cassette right and left lateral side-bending films in the supine position and pushprone and extension films over a bolster are necessary to assess the structural components of the curve, including curve segment flexibility and the ability of the spine in the potentially uninstrumented segments to fall within the stable zone after operative connection. Push-prone films are done with the patient in the prone position, with the pelvis stabilized by an assistant and the surgeon pressing on the deformity to attempt to create a radiographic picture of the maximum possible correction. Not only does this help assess the flexibility of the main curve, it also approximates the response of the compensatory curves to this correction. Bolster films are particularly useful when considering the flexibility of a kyphotic segment and its ability to be corrected. The patient is placed in the supine position with a bolster over the apex of the kyphosis, allowing the torso to fall backward under the effect of gravity. This will usually give a good indication as to the ability of the kyphosis to be corrected with instrumentation alone. When a patient presents with neurological deficits secondary to spinal stenosis and nerve root compression, the level and the extent of the compressive pathology need to be delineated. In mild curves, magnetic resonance imaging (MRI) is useful to ascertain the intracanal pathology. In addition, MRI gives valuable supplemental information about the lower lumbar discs and the possibility of avoiding fusion to the sacrum. It is, however, unusual in a population of aging patients to have MRI normal discs. Computed tomography (CT) myelography may be a more useful study for investigation of intracanal pathology in the severely stenotic and/or severely deformed spine. Because the myelographic dye clearly and accurately outlines the neural space, a precise and detailed representation of the extent of neural element compression can be obtained. In patients who have undergone previous surgery and implantation of stainless steel instrumentation, this is often a useful study. Discography is useful for delineating the optimal operative levels. It is important that every attempt be made to spare lower lumbar segments in particular. This helps maintain some degree of flexibility and provides a stress relief point for the spinal instrumentation. CT discography is an excellent way to investigate both the morphology and the clinical condition of the disc. A chief concern in this population of patients is general nutrition. In patients with reduced serum
PREOPERATIVE PREPARATIONS
protein, some form of preoperative enteral alimentation or hyperalimentation may be necessary to optimize the patient for the surgical challenge. Preoperative laboratory investigations should include total albumin, total lymphocyte count, transferrin, and pre-albumin. If there is a possibility of anterior transthoracic surgery, a full pulmonary and cardiac workup should be performed. Cosmetic issues are also important. Patients with significant upper lumbar and lower thoracic abnormalities may have unsightly rotational prominences that are an integral part of the patients dissatisfaction. If the patient has a significant and unacceptable lumbar prominence, consideration should to be given to an anterior multilevel discectomy and intradiscal fusion to help decrease this rotational deformity. In a similar manner, if the thoracic prominence is a cosmetic issue, consideration may be given to a thoracoplasty in combination with the deformity surgery. Pulmonary function is decreased for several months following thoracoplasty, a factor that should be considered preoperatively in patients with pulmonary dysfunction.
Because of lengthy surgical time, the possibility of poor vascular access, and large fluid shifts, these patients should be monitored with intra-arterial lines and central venous access lines. Although hypotensive anesthesia is a possibility, serious complications can be associated with this. In the elderly population, severe hypotensive anesthesia can result in myocardial infarction, stroke, and multisystem failure including the renal and hepatic systems. In addition, hypotensive anesthesia, particularly during large corrective deformity surgeries, can result in blindness if direct ocular pressure is not applied. Normotensive anesthesia may be more appropriate in these extensive procedures. Blood loss can be offset by preoperatively donated autologous blood and the use of a cell saver.
ANESTHETIC TECHNIQUE
For the most part, both an anterior and a posterior approach will be needed for the correction of these deformities. Because of the severe nature of the deformity and the poor fusion rates expected in this population, structural grafting and fixation both anteriorly and posteriorly are warranted to ensure adequate fusion. However, in some instances the deformity may be addressed exclusively anteriorly or posteriorly. Many adequate positioning techniques have been described. Specific positioning schemes may depend to a large degree on anticipated intraoperative repositioning (i.e., closing or opening osteotomy sites) and maintenance of lumbar lordosis. Whichever technique is used, adequate decompres-
OPERATIVE POSITIONING
227
Neill BioMedical Art Co., 1999 Figure 1. Foraminal stenosis due to hypertrophic facets.
The primary dysfunction in patients with this pathology is the asymmetric collapse of the degenerating lumbar intervertebral disc. The goal of surgery is to decompress where appropriate, restore spinal balance, reconstruct the deficient disc(s), and stabilize the spine until fusion occurs. The hallmark of the disease is lateral listheses or rotatory subluxations. Early in the course of the disease, this is often identified as a slight asymmetric collapse without obvious
SURGICAL GOALS
rotation at the involved disc space. This asymmetric collapse increases until there is near complete loss of unilateral disc height in the coronal plane. This is followed by a lateral shift of the superior vertebral body on the inferior vertebral body and evidence of rotation as identified by using Nash/Moe observations for pedicle position. At this point, the patient frequently presents with radicular symptoms. These symptoms typically result from either standard foraminal or central stenosis due to collapse of the disc space anteriorly and overgrowth of the facet joints with redundant ligamentum flavum posteriorly (Figure 1). Characteristically, there is compression on the cauda
228
equina secondary to the loss of overall central canal diameter associated with rotatory subluxation (Figure 2). Another unique mechanism for the onset of radicular symptoms in this pathology is nerve root traction secondary to the rotatory translation of the pedicles, causing nerve root stretching as the roots exit the neural foramen (Figure 3). The curves range from subtle, tight radius deformities confined to the upper and middle lumbar spine to fairly large curves involving the lower thoracic and upper lumbar spine as the primary curve and fractional or compensatory curve in the lower lumbar spine which returns the spine to the pelvis (Figure 4). In the vast majority of cases, the apex of the degenerative scoliosis is at the L3-4 disc level, which typically corresponds to at least one rotatory subluxation. It is not uncommon at the level of the rotatory subluxation to also have a sagittal plane spondylolisthesis in addition to a loss of lordosis (see Figure 3). While it is most common to have only one such obvious level of instability, both the rotatory subluxations and the spondylolistheses can involve multiple levels. Contraindications for surgical treatment in these patients often center more around the medical conditions of the elderly and presence of osteoporosis than around specific difficulties with the procedure, although correcting large, stiff curves can be a formidable undertaking.
part, concerns associated with the posterior approach have to do with improperly or poorly placed instrumentation. Insufficient or suboptimal biomechanical placement of instrumentation can result in failure. In the osteoporotic spine, even rigid constructs may be inadequate due to the poor osseous purchase of the metallic implants.
Posterior osteotomies, fusion, and instrumentation frequently require supplementary intradiscal grafting in order to fully address discal pathology and secure a solid fusion. This may be performed via an anterior or posterior approach. Intradiscal fusion of L5-S1 is often needed to restore spinal balance and to reconstruct a deficient disc that is placed under increased shear and torsional loads (Figure 9). This may be performed retroperitoneally or via Pfannenstiels incision (Figure 10). Additionally, if the deformity is rigid, an anterior release of the remaining structures most responsible for resisting rotation may be required in order to achieve correction. Once released, the appropriate posterior osteotomies, operative positioning, and efficient placement of biomechanically correct posterior instrumentation can be accomplished. Instrumentation used both anteriorly and posteriorly is common to less complicated surgeries and is used in multiple combinations to achieve the desired correction in complicated deformities.
OPERATIVE PROCEDURE
The determination as to whether to proceed anteriorly, posteriorly, or both needs to be assessed on a case-by-case basis. When the deformity is supple or unstable, consideration should first be given to instrumentation. This usually implies a posterior approach with realignment and instrumentation first. The standard posterior approach for spinal cases is used (Figure 5). Realignment in such cases may often be achieved via a posterior osteotomy in which the lamina and facet are removed. In this technique, the inferior facets of the superior vertebra and the lamina and the superior facets of the inferior vertebra are removed (Figure 6). The osteotomy is then closed and stabilized under compression instrumentation (Figure 7). Finally, posterolateral fusion, including the facets, is performed to heal the osteotomy site. Pedicle screws are an excellent means of segmental fixation in the posterior lumbar spine (Figure 8). Their safe placement is certainly within the realm of possibilities for the accomplished spine surgeon, but the risk of serious neurological injury should not be neglected. Special care must be taken not to injure the dura and the neural elements secondary to impertinent penetration into the canal. For the most
Posterior Approach
Anterior Approach
In certain instances, anterior instrumentation may be sufficient for reconstruction. Anterior instrumentation is particularly helpful in substantially decreasing coronal plane deformities and in stabilizing the spine after corpectomy. The type of anterior approach used depends on the individual patients anatomy that needs to be accessed and the type of instrumentation to be used during surgery. Anterior instrumentation at multiple levels typically requires a lateral approach and may be best suited to a patient positioned in the right or left lateral decubitus position and an extensile flank incision (Figure 11). This allows the development of the retroperitoneal space and provides for a straight lateral approach to the spine. The major considerations during an anterior approach are predominantly anatomic and center on the safe dissection of the vascular structures, especially at L4 and below. At this level, adequate exposure of the anterior spine requires immobilization of the great vessels at the aortic bifurcation and safe identification and ligation of the often prominent and numerous recurrent lumbar veins. This safeguards against uncontrolled intraoperative bleeding and avulsion of these vessels from the iliac vein. The segmental vessels are more easily located and controlled;
229
Figure 3. Lateral view showing spondylolisthesis and loss of lordosis. Note the nerve root stretching around the pedicle at L3.
230
they also need to be considered when performing corpectomies and during placement of transvertebral body screws with bicortical purchase. Failure to consider these vessels during these procedures can result in significant bleeding that is difficult to control since the vessels in question are on the opposite side of the vertebral body. Other anatomic structures that need to be carefully identified and considered include the hypogastric plexus, the ureter, the ilioinguinal and genitofemoral nerves, the lumbar plexus, and of course the spinal cord and cauda equina. Care must be taken to mobilize all structures anterior to the periosteum over the sacral promontory (L5-S1 disc space) using blunt dissection. Cautery should not be used on this anterior soft tissue, and sharp dissection should be limited to intradiscal dissection. The anterior and posterior reconstruction techniques used in these complicated procedures are not significantly different than the techniques used in simpler degenerative procedures. They merely represent the sequential application of various techniques to address the numerous problems in these complicated deformities. It is obvious that multiple procedures performed in the same patient increase the risk of infection and other complications. Significant preoperative planning is necessary to be sure that each technique is applied in a sound and planned
order so as to achieve the goals of surgery with minimum blood loss and minimum operative time.
As with all types of deformity surgery, particularly when large changes are anticipated in the overall coronal and sagittal alignment, electrophysiological monitoring is necessary. It is optimal to have both somatosensory evoked potential and motor evoked potential monitoring available. To ensure the greatest safety for the neural elements, the use of monitored pedicle probes and pedicle screw stimulation should be considered during the placement of pedicle screws. For electrophysiological monitoring to be truly effective, there needs to be cooperation between the electrophysiologist and the anesthesiologist. In addition, the electrophysiologist needs to be familiar with sophisticated techniques that involve monitoring both nerve roots via EMG and monitoring of motor and sensory tracts. Also affecting the usefulness of the electrophysiological monitoring is the willingness of the surgeon to cooperate with the electrophysiologist. There must be adequate time for monitoring during dissection techniques using electrocautery and when probing pedicles and placing pedicle screws. In the event that electrophysiological monitoring is equivocal or suggests an intraoperative concern, a wake-up test should be considered as the
MONITORING
231
Figure 7. Stabilization of the lumbar and lumbosacral spine using Texas Scottish Rite Hospital instrumentation.
Figure 8. Pedicle screws placed for fixation in the posterior lumbar spine.
232
Figure 9. Total discectomy in preparation for intradiscal fusion in a combined anterior/posterior approach.
233
Neill BioMedical Art Co., 1999 Figure 11. Extensile flank incision for an anterior approach.
gold standard for intact neurological functioning, especially for the motor tracts.
High-speed burrs, which provide both high torque and high speed, are extremely helpful when performing complicated osteotomies and decompressions around the neural elements. Although a high-speed burr near the dural sac may appear somewhat dangerous, it is probably much less dangerous than dull osteotomes, Kerrisons, and rongeurs which have a tendency to slip off hard bone and may not be able to be applied to the osseous structures in a convenient manner. In spite of this, great care is necessary while using these tools near the neural elements. These challenging cases are certainly not the forum to practice using this instrumentation.
SPECIALIZED INSTRUMENTATION
actually be manipulated in such a way as to show the screw in a bulls eye position in the pedicle. No intraoperative imaging techniques, however, can substitute for careful preoperative planning.
INTRAOPERATIVE RADIOGRAPHY
Intraoperative imaging is most important in cases involving severe deformities in which large changes in patient positioning in either the frontal or the sagittal projection are anticipated. Intraoperative fluoroscopy also helps to ascertain the coronal alignment of pedicle screws since the head of the image machine can
As described above, anatomic or instrumentation problems can be avoided by careful preoperative planning and meticulous technique. Frequently, complications in this group of patients concern the metabolic consequences of surgery. Large transfusion volumes may result in disseminated intravascular coagulation or adult respiratory distress syndrome. Overzealous manipulation of vascular structures may result in intimal injuries leading to either local thrombosis or the propagation of thrombi with resulting stroke, myocardial infarction, or pulmonary embolism. In addition, large-volume fluid shifts may result in multisystem failures. To prevent these complications, careful preoperative planning is critical. Intraoperative decisions can be anticipated and alternative solutions developed, resulting in more safely performed complex operations. The optimal surgical goal cannot be achieved in all cases. Other solutions and a worse-case scenario option must be detailed.
COMPLICATIONS
The retropleural approach provides the shortest direct route to the ventral thoracic and thoracolumbar spine and spinal canal, avoiding entry of the pleural cavity. One advantage of a ventral appr oach to the spinal cord is the ability to visualize significant ventral pathology that extends past the midline of the involved vertebral body. Dorsolateral approaches such as the costotransversectomy and the lateral extracavitary provide exposure that barely extends to the midline; further ventral visualization may require extensive muscle mobilization and bloody transforaminal dissection. These approaches also require sacrifice of the intercostal nerve and risk possible occlusion of the radiculomedullary artery. The retropleural approach offers several advantages over the alternative transthoracic ventral approach. Since a shorter segment of the chest cavity must be traversed to reach the spine through the opening, significant retraction of the unprotected lung is avoided. In the retropleural approach, the parietal and visceral pleura shield the lung from the forces of retraction. Postoperative pulmonary complications are minimized and placement of a chest tube is usually not required. While some authors report a higher incidence of pleural effusion in the retro-
INTRODUCTION
pleural approach, this has not been the experience of the senior author (P.C.M.). In the case of a cerebrospinal fluid (CSF) leak, the transthoracic approach requires placement of a spinal drain as well as a chest tube on water seal. For the retropleural approach, no treatment is indicated. The retropleural approach also allows a smaller incision and minimizes soft tissue dissection, which may lessen postoperative pain and shorten the patients hospital stay. In one large series, post-thoracotomy pain syndrome lasting at least 6 months was reported in 9.2% of patients undergoing the transthoracic approach. A shorter, more secure operative field reduces the risk of untoward intraoperative events and facilitates achievement of the surgical objective. The retropleural approach also uses a less oblique angle, which affords improved visualization of the spinal canal unobstructed by the rib head, aorta, or vena cava. During the transthoracic approach, the curved dorsal cortical vertebral body margin may disorient the surgeon, especially if inadequate bony dissection has been performed. Finally, bleeding from the extensive ventral epidural thoracic venous plexus is easier to control from a less oblique angle. The approach also provides both lateral canal exposure through the resected pedicle bed and adjacent foramen as well as a direct line of vision that is anterior to the ventral canal floor. Thus, initial canal identification and complete canal decompression is more easily and confidently achieved. Some authors find that pleural adhesions encountered in adult patients make retropleural dissection difficult. In the
235
236
Figure 1. A, computed tomography scan demonstrating L2 flexion/compression fracture with retropulsion of bone fragment into the spinal canal. B, anteroposterior radiograph following r etropleural corpectomy and reconstruction demonstrating lateral titanium plate and autologous rib/femoral head allograft bone graft. C, lateral radiograph following retropleural corpectomy and reconstruction. (Reproduced from McCormick PC: Retropleural approach to the thoracic and thoracolumbar spine. Neurosurgery 37:908-914, 1995, with permission)
authors opinion, however, minor lacerations of the pleura are easily r epaired. If the pleurae are significantly compromised, the approach can easily be modified into a transthoracic approach.
The retropleural approach is most appropriate for localized (two or fewer vertebral segments) pathology of the ventral thoracic and thoracolumbar spine between T3 and L1. Traumatic fractures and posttraumatic kyphoses occur frequently in the thoracolumbar spine and are often amenable to surgical correction via the retropleural approach. The traumatic lesion most suitable to the retropleural approach is the burst fracture, in which there is failure of the anterior and middle columns due to compression loading (Figure 1). Neurologically intact patients with minimal canal compromise and an intact dorsal column can usually be managed conservatively with bracing, bed rest, and physiotherapy. Likewise, pa-
INDICATIONS
tients with complete loss of neurological function below the level of injury for more than 48 hours will usually not improve following operative decompression but may require surgical stabilization for rehabilitation. Fractures with a greater than 40% loss of body height, greater than 50% canal compr omise, major kyphotic deformity, or significant neurological deficits are generally considered operative lesions. A ventral approach is also useful in posttraumatic kyphosis in the setting of increasing deformity, kyphosis greater than 30 degrees, intractable pain, or persistent or progressive neurological deficits since dorsal fusion alone is usually not sufficient. If there is dorsal stability, neurological improvement and significant pain relief are achieved in a majority of these patients approached ventrally. Adequate reduction and fusion, however, usually require ventral instrumentation or prolonged postoperative recumbency. If there is a question of compromised posterior stability, then dorsal spinal fusion may be required in
SCHWARTZ AND McCORMICK : RETROPLEURAL APPROACH TO THE THORACIC AND THORACOLUMBAR SPINE
237
Figure 2. Artists drawing depicting patient positioning. Skin incisions for upper thoracic (A), midthoracic (B), and thoracolumbar (C) levels are shown. (Reproduced from McCormick PC: Retropleural approach to the thoracic and thoracolumbar spine. Neurosurgery 37:908-914, 1995, with permission)
addition to ventral fusion, and a dorsolateral or separate dorsal and ventral approaches may be more appropriate. Disc herniations in the thoracic region are also easily decompressed via the retropleural approach. The increased exposure is particularly useful for calcified discs, which cannot be adequately removed via the transpedicular or costotransversectomy approach. Significant ventral compression from primary and metastatic tumors to the vertebral body or infections of the spine can also be appropriately managed via the retropleural approach. Laminectomy alone is often contraindicated in these situations and carries a risk of neurological deterioration. Posterolateral approaches are more appropriate for lesions involving three-column dysfunction or greater than three segments of involvement. Indications for the ventral approach include progressive neurological deficit, lack of tissue diagnosis, instability and kyphosis during or following radiation therapy, and patients with a known radio-resistant tumor such as melanoma or hypernephroma or Potts disease. Anterior decompression and stabilization also have the long-term benefit of minimizing progressive deformity.
racic levels. Routine induction and intubation are then accomplished. A double lumen endotracheal tube is requested for lesions above T6. A Foley catheter is placed and preoperative antibiotics are routinely administered. Air-compression boots are used to prevent deep vein thrombosis.
The side of the approach is determined primarily by the level and location of the pathology. The patient is turned to a lateral position on a beanbag with a small roll placed under the dependent axilla (Figure 2). The upper arm is rested on a pillow. Thoracolumbar lesions should be centered over the kidney break or flexion joint of the operating table to improve the operative exposure. The lower leg is flexed at the hip and knee to stabilize the patient. Suction aspiration of the beanbag secures the position. All bony prominences and subcutaneously coursing peripheral nerves are well padded.
Operative Positioning
Following appropriate venous and arterial line access, a lumbar epidural catheter can be placed for perioperative morphine installation, which markedly diminishes postoperative pain, even at upper tho-
For lesions between T5 and T10, a 12-cm skin incision extending from the posterior axillary line to a point 4 cm lateral to the posterior midline is made over the rib at the level of the lesion (Figure 2). For a T8-9 disc herniation, for example, a skin incision is made in line with the T9 rib, which will be removed. The skin incision must be altered for upper thoracic (T3-4) lesions. While forward flexion of the upper arm and shoulder rotates the scapula superolaterally to expose the T5 and T6 ribs, the T3-4 ribs remain under its cover. For these lesions, a hockey-stick
Skin Incision
238
incision, which parallels the medial and inferior scapular borders, is per formed (Figure 2). The incision is carried down through the scapular muscle attachments (i.e., the trapezius or rhomboid muscles) to the ribs. The detached scapula is rotated superiorly to expose the appropriate rib. The approach to the thoracolumbar junction is discussed separately.
After confirmation of the correct vertebral level, a subperiosteal detachment of the intercostal muscles over an 8-10 cm rib segment is performed. This rib segment is removed (Figure 3A). Note that the most proximal 4 cm of rib, which is attached to the transverse process and vertebral body, remains in place. A well-defined tissue layer is identified in the bed of the resected rib segment. This is the endothoracic fascia, which lines the entire thoracic cavity. The parietal pleura maintains its attachment to the inner chest wall through this layer. The potential space between the parietal pleura and the endothoracic fascia may contain a small amount of loose areolar tissue. The endothoracic fascia is analogous to the transversalis fascia which lines the abdominal cavity. Both line their respective visceral cavities and are reflected over the diaphragm as its fascia. The endothoracic fascia is continuous with the inner periosteum of the rib and the thoracic vertebral bodies. Occasionally, muscle fibers of the inconstant subcostal muscle may be identified within this fascia in the rib bed. The thoracic sympathetic chain, intercostal vessels and nerves, thoracic duct, and azygous veins are contained against the thoracic wall and vertebral bodies within this fascial layer. The endothoracic fascia is sharply incised in line with the rib bed (Figure 3B). The parietal pleura is dissected widely off the undersurface of the endothoracic fascia with Kitner (peanut) clamps, much the same way in which the peritoneum is freed from the transversalis fascia of the abdominal wall during a retroperitoneal exposure (Figure 4). The pleura is dissected proximally to expose the vertebral column. Small pleural tears are repaired with suture. A tablemounted malleable blade maintains lung retraction. Deflation of the lung with a double-lumen tube is not critical but does facilitate lung retraction at the upper thoracic levels (above T6). A self-retaining crank retractor on the adjacent ribs widens the exposure. The endothoracic fascia is opened over the remaining proximal rib segment with cautery. The softtissue attachments, costotransverse, and stellate ligaments are sharply divided and the rib head is disarticulated from the vertebral body (Figure 4b). For thoracic disc removal, the endothoracic fascia over the disc space is incised. This divides the sympathetic chain which descends on the vertebral column just anterior to the rib head insertion (Figure 5A). The fascia and the vertebral body periosteum
OPERATIVE PROCEDURE
are elevated in either direction away from the disc space. The intercostal vessels that run transversely at the midvertebral body level are preserved within this reflected tissue. The margins of the pedicle are defined sharply with curettes and nerve hooks. The disc space is incised and evacuated with curettes and rongeurs (Figure 5B). A high-speed drill removes the adjacent endplates and extends the dissection into the adjacent vertebral bodies. The pedicle is also removed with the drill and Kerrison rongeurs (Figure 6A). This provides access to the lateral spinal canal, which is probed with a dissector. The bony vertebral body dissection can now be completed posteriorly with knowledge of the canals location. Generous bony removal facilitates and assures adequate canal decompression. The depth of decompression should be 3 to 3.5 cm from the lateral vertebral body margin. The corpectomy should extend about 1.5 cm on either side of the disc space. This opening should be maintained back to the dorsal cortical margin. Bone bleeding may be intermittently controlled with smeared wax at the end of a Kitner clamp. Once the dorsal cortical margin has been sufficiently thinned, a reverse-angled curette sharply divides the posterior longitudinal ligament. The adherent endplates, annulus, and cortex are pushed down into the corpectomy defect. This invariably results in epidural venous bleeding so it must be completed quickly. Bipolar cautery forceps may be passed on either side of the vessels from the lateral canal entrance for effective coagulation of the ventral epidural vessels. The thin dorsal layer of the posterior longitudinal ligament usually remains. In the authors experience, the calcified thoracic disc is often suspended within this layer and is not as commonly intradural as the literature suggests. The ligament is gently probed with a nerve hook for identification and delivery of these fragments away from the spinal cord. Once adequate decompression has been achieved, troughs are drilled into the adjacent vertebral bodies and the harvested rib is placed as an interbody strut graft (Figure 6B). Vertebral body corpectomy requires removal of an additional rib head for adequate exposure. For a T7 corpectomy, which has been exposed through the T7 rib bed, for example, resection of the T8 rib head provides exposure of the T7-8 disc space and adjacent T8 vertebral body and pedicle (Figure 7). The intercostal artery and vein are divided between sutures, close to the midline. Proximal ligation of the intercostal artery can be safely performed because abundant muscular and osseous anastomoses will distally reconstitute the occluded vessel and maintain spinal cord blood flow. Following disc exposure and intercostal vessel ligation, the periosteum and endothoracic fascia are incised and elevated off the vertebral body. The discs are incised and curetted as previously described. The corpectomy is completed
SCHWARTZ AND McCORMICK : RETROPLEURAL APPROACH TO THE THORACIC AND THORACOLUMBAR SPINE
239
Figure 3. A, an 8-10 cm rib segment is removed following subperiosteal dissection. B, the endothoracic fascia is opened in line with the resected rib bed. The pleura has been bluntly freed from its inner surface and remains intact. (Reproduced from McCormick PC: Retropleural approach to the thoracic and thoracolumbar spine. Neurosurgery 37:908-914, 1995, with permission)
Figure 4. The pleura is bluntly freed and retracted (a) from the undersurface of the endothoracic fascia to widely expose the posterior thoracic wall and vertebral bodies. The neurovascular bundle (b) has been partially mobilized from its underlying fascia to improve visualization. The endothoracic fascia remains densely adherent to the posterior thoracic wall and vertebral column. Incision and subperiosteal elevation of the endothoracic fascia then continue proximally over the rib head (c), the disc space, and adjacent vertebral bodies. (Reproduced from McCormick PC: Retropleural approach to the thoracic and thoracolumbar spine. Neurosurgery 37:908-914, 1995, with permission)
240
B
Figure 5. A, the sympathetic chain is divided and the disc space sharply incised. B, the disc space is evacuated using curettes and rongeurs.
Figure 6. A, a high-speed drill is utilized for removal of adjacent endplates and pedicle. B, the harvested rib is placed as an interbody strut graft. (Figure 6A reproduced from McCormick PC: Retropleural approach to the thoracic and thoracolumbar spine. Neurosurgery 37:908-914, 1995, with permission)
SCHWARTZ AND McCORMICK : RETROPLEURAL APPROACH TO THE THORACIC AND THORACOLUMBAR SPINE
241
Figure 7. A, adequate exposure for corpectomy requires the removal of two proximal rib segments, incision and removal of discs both superior and inferior to the involved body, and proximal ligation of the intercostal vessels (a) at the affected level. B, reconstruction is performed following adequate spinal canal decompression. (Figure 7B reproduced from McCormick PC: Retropleural approach to the thoracic and thoracolumbar spine. Neurosurgery 37:908-914, 1995, with permission)
using rongeurs, curettes, and a high-speed drill. Appropriate reconstruction is then performed. The retropleural approach must be modified at the thoracolumbar junction (T11-L1) because of the diaphragm and the greater caudal angulation of the ribs. In these cases, the skin incision and rib resection are performed two levels rostral to the pathological segment. In the case of a T12 vertebral body exposure, for example, a 12-14 cm skin incision is made over the T10 rib from the posterior axillary line to 4 cm off the midline (Figure 2). The incision is carried down through the subcutaneous tissue and muscles to the T10 rib with cautery. A 10-cm segment of exposed rib is cleaned circumferentially and removed. The pleural surface of the diaphragm is identified. The endothoracic fascia is reflected over the diaphragm and tightly applied to its surface. The costophrenic pleural reflection may also be seen. The initial exposure is tight because of the diaphragm attachment to the ribs. If the lateral endothoracic fascia is present in the rib bed, it is opened. Caudally, the pleural surface of the diaphragm is depressed and detached from the inner surface of the T11 and T12 ribs with sharp periosteal dissectors. This immediately unites the retroperitoneum with the retropleural space. The detachment is easily continued medially to elevate
the arcuate ligaments off the quadratus and psoas muscles. The subcostal nerve can be seen running laterally beneath the lateral arcuate ligament on the quadratus lumborum surface. Division of the ipsilateral crus completes the diaphragm mobilization. Table-mounted malleable blades provide retraction. The proximal 4 cm of the T12 rib, including the rib head, is removed. Part of the psoas muscle, which attaches to the T12 vertebral body, may need to be elevated. This completes the exposure of the T12 vertebral body, adjacent disc, and T12 pedicle. Disc incision or corpectomy is performed utilizing the principles previously outlined.
Following decompression and stabilization, the arcuate ligaments are reattached to the psoas and quadratus muscles with suture. The pleura is then carefully inspected. A chest tube should be placed if any pleural tears are identified. The remainder of the wound is closed in layers. Approximation of the adjacent ribs with suture reduces the chest wall deformity associated with rib resection. Care is taken not to ensnare the intercostal nerves with suture during closure. This may substantially increase postoperative pain.
Closure
242 COMPLICATIONS
The following are commonly seen complications and suggestions for avoiding them. 1) Pneumothorax can be avoided by repairing any pleural tears with sutures or staples under positive pressure ventilation. In the face of large tears, or if air remains in the intrapleural space at the conclusion of the procedure, a chest tube should be placed. 2) CSF leak is prevented by visualizing the dura early in the operation. This can be achieved by identifying and removing the pedicle to expose the lateral canal. 3) Horners syndrome can be avoided by preserving the sympathetic chain in the upper thoracic region if possible. Do not use this approach above T3.
4) Spinal cord infarction can be avoided by not using cautery in the foramen. 5) Intercostal neuralgia is minimized by not suturing the intercostal nerve during rib approximation. This may be prevented by passing suture through holes drilled in the caudal rib. Also, avoid crushing the neurovascular bundle against rib with crank retractors. 6) Carefully ligate the intercostal neurovascular bundle if performing corpectomy. Preserve the intercostal neurovascular bundle if only removing disc. 7) Pleural effusion can be avoided with selective placement of a chest tube or even a Hemovac drain in the retropleural space in cases where hemostasis is not totally secure due to persistent oozing from epidural veins or bone to prevent significant retropleural fluid collection.
Lateral disc herniation has been denoted by a variety of terms, including herniation of the foramen, far lateral disc, posterior lateral disc, extreme lateral disc, and extraforaminal disc. Lateral disc herniation is defined here as herniated disc material predominantly located lateral to the midportion of the intervertebral foramen. This is compared to an intracanalicular disc herniation when the disc lies in the midline (central) or out to the midportion of the intervertebral foramen (paracentral). A further differentiation can be made of a lateral disc herniation between a foraminal lateral disc herniation (when the disc material is primarily located in the intervertebral foramen) and a far lateral herniation (when the herniated disc is mainly situated extraforaminally and the nerve is compressed in its paraspinal course). The surgical approach is significantly influenced by the location of the disc herniation relative to the intervertebral foramen (foraminal vs. far lateral) and whether any associated intracanalicular pathology, such as lateral recess stenosis, is present. Before the widespread use of computed tomography (CT) and magnetic resonance imaging (MRI) permitted an accurate diagnosis, lateral disc herniations were difficult to visualize and were often missed on routine myelography. In surgical series, the incidence
of lateral disc herniation ranges from 4% to 12% of all herniated lumbar discs. Lateral disc herniations generally occur in older patients and at higher interspace levels than routine (intracanalicular) herniations. Central and paracentral disc herniations involve the L4-5 or L5-S1 levels in 95% of reported cases. Lateral herniation occurs more frequently at higher lumbar spine levels; the L4-5 disc is affected in 40% of cases and the L3-4 level in 37%. A superolateral migration of disc material occurs in most cases of lateral lumbar disc herniation, which results in compression of the exiting nerve root at the level of the foramen (i.e., an L4-5 disc herniation compresses the L4 root). In contrast, with most intracanalicular disc herniations, the traversing nerve root is much more commonly affected (i.e., an L4-5 disc herniation compresses the L5 root). Because of the anatomical relationships of the neural foramen, the nerve root and ganglion are displaced cranially and/ or dorsally and are compressed against the lower edge of the upper pedicle. Downward migration of a fragment is prevented by the lower pedicle. This superior migration has important surgical implications because disc fragment removal is frequently possible without compromising the facet joint, which often overlies the intervertebral disc space. Since the L3 and L4 nerve roots are commonly involved, patients with a herniated lateral disc often complain of radiating pain and sensory changes in the medial or anterior thigh. Quadriceps weakness and reduced or absent patellar reflexes are common findings on neurological examination. The straightleg raising test is usually positive if the L5 nerve root
243
244
is involved, is sometimes present if the compressed nerve root is L4, and is usually absent when the disc involves the L3 root. The femoral stretch test is usually positive when there is involvement of the L3 and L4 nerve roots. Optimally, high-resolution CT and MRI will display lateral disc herniation when present. Lateral herniation can occur as an isolated condition or in association with other forms of spinal degenerative disease. To avoid overlooking this source of neural compromise, careful analysis of the intervertebral foramen on the axial and sagittal images is warranted despite either the normal appearance of the central and paracentral disc or cases with associated pathology. Signs of lateral disc herniation on MRI and/or CT studies include the presence of disc density or disc signal material at the level of or lateral to the neural foramen, displacement or obliteration of perineural fat, and nerve root or ganglion compression/displacement. Myelography alone has little value in the identification of far lateral disc herniation. In fact, this diagnostic test is often negative since the nerve compression usually occurs outside the nerve root subarachnoid space. We frequently use CT myelography as a complementary study for patients with lateral disc herniation; the addition of high-resolution, thincut axial CT improves the diagnostic accuracy by showing whether the disc fragment is in continuity with the intervertebral disc and whether it encroaches on the nerve root just inside or lateral to the intervertebral foramen. We believe that CT myelography continues to have a significant role in the diagnosis and management of lateral herniation because it permits a better evaluation of associated lateral recess and spinal stenosis, especially in older patients in whom lateral disc herniation often occurs. In addition, CT myelography defines the bony anatomy better than MRI and improves preoperative planning by enhancing visualization of the orientation and location of the facet joints, and demonstrating the relationship of the disc herniation to the isthmus laminae and transverse process.
Figure 1. Illustration showing the anatomical relationships of the neural foramen and the far lateral space at different levels of the lumbar spine.
Unless a severe neurological deficit is present, conservative management is recommended as a first-line treatment for the patient with a lateral disc herniation. The early phase of conservative therapy includes pharmacological pain relief (generally with nonsteroidal anti-inflammatory agents) and a reduction of activities such as heavy lifting, prolonged sitting, or strenuous exercise for several weeks. Prolonged bed rest is not indicated and is usually counterproductive. Frequently, symptoms resolve with time and no additional therapy is needed. Other treatment modalities such as physical therapy and selective perineu-
PATIENT SELECTION
ral infiltration with steroid solution may be considered in selected patients. Surgery is performed only when the patients pain is persistent and disabling following a complete 6- to 8-week course of conservative treatment. It has been suggested that patients with a lateral herniated disc may experience lasting, disabling pain more often than patients with an intracanalicular herniation; this may be caused by the involvement of the very sensitive posterior root ganglion.
Familiarity with the relevant anatomy is of utmost importance to fully understanding the surgical approach to lumbar lateral disc herniation. The intervertebral foramina are three-dimensional canals bordered anteriorly by the more cephalad vertebral body and the intervertebral disc and posteriorly by the pars interarticularis of the vertebral body above (Figures 1 and 2). The cranial vertebral body may constitute more of the anterior border of the foramen than the intervertebral disc. The superior margin is delimited by the inferior aspect of the pedicle of the upper
ANATOMICAL CONSIDERATIONS
245
partial facetectomy
L4 superior facet
Inset
Figure 2. Representation of the normal anatomy. Inset, the ligamentum flavum is opened and a small hemilaminectomy is performed on the lamina of L3. The lateral and most inferior portions of the inferior facet of the superior lamina are removed for an extension outlined by the dotted line. (Reproduced from Chenelle AG, Shaffrey CI, Lanzino G, et al: Treatment of far lateral discs by an approach lateral to the facet joint. Techn Neurosurg 2:249-253, 1996, with permission)
ventral nerve root Figure 3. Transverse section illustrating the neurovascular anatomy and the direction of the surgical exposure obtained via an intertransverse route. (Reproduced from OBrien MF, Peterson D, Crockard HA: A posterolateral microsurgical approach to extreme-lateral lumbar disc herniation. J Neurosurg 83:636-640, 1995, with permission)
246
vertebra, while the inferior margin is circumscribed by the superior aspect of the pedicle of the lower vertebral body. The entrance and exit to the foramen are the medial and lateral margins of adjacent pedicles. Lateral to the foramen is the far lateral space. This space is bordered medially by the foramen, laterally by the intertransverse ligament, anteriorly by the cranial vertebral body and disc covered by the annulus fibrosus, and posteriorly by the lateral fibers of the ligamentum flavum. The nerve root enters the foramen and far lateral space medially and superiorly, adjacent to the pedicle of the superior vertebra, then crosses the disc space extraforaminally, laterally and slightly anterior to the posterolateral margin of the disc. The dorsal root ganglion is found within the intervertebral foramen just medial to the lateral border of the pars interarticularis. The nerve divides into the posterior primary ramus and the ventral nerve root immediately distal to the dorsal root ganglion. The ventral root crosses the disc space obliquely and passes in close proximity to the lateral aspect of the caudal pedicle. The posterior primary ramus travels in a caudal and dorsal direction, passes between the pars interarticularis and the intertransverse muscle and ligaments, and then through a gap between the medial free edge of the intertransverse ligament and the lateral free edge of the ligamentum flavum (Figures 3 and 8). The lumbar segmental artery travels through the space defined medially by the emerging nerve root and laterally by fibers of the intertransverse ligament and divides into five recognizable branches. These branches include: 1) a branch to the undersurface of the transverse process; 2) a dorsally directed branch in the waist of the pars interarticularis; 3) a laterally directed branch that enters the space between the multifidus and longissimus muscles; 4) a branch into the midlateral aspect of the erector spinae muscles; and 5) a branch accompanying the lateral branch of the posterior primary ramus (Figures 3 and 8). From a surgical point of view, the operative approach to the far lateral space is limited dorsally by the isthmus laminae (defined as the lateral margin of the interarticular portion of the lamina); superiorly by the inferior margin of the cranial pedicle, the transverse process, and, when present, by a prominent accessory process; and inferiorly by the apophyseal joint of the articular facet. The relationships among these structures vary significantly at different levels and the operative exposure to the lateral space becomes progressively smaller in the lower lumbar spine, especially at L5-S1 (Figure 1). The lumbar spinal anatomy is a major consideration in determining the appropriate approach to a lateral disc herniation. Significant compromise of the facet joint should be avoided when possible. These
anatomical considerations strongly affect the microsurgical approach to an extraforaminal disc herniation. From L1-2 to L3-4, a midline approach, with lateral retraction of the paraspinal muscles past the facet joints to expose the isthmus of the lamina and the proximal transverse processes, allows for enough exposure of the lateral portion of the intervertebral foramen. In the upper lumbar spine (L1-2 to L3-4), the pedicles arise vertically from the dorsal aspect of the vertebral bodies; the distance between the inferior margin of the upper transverse process and the superior boundary of the facet joint averages 10 mm. The width of the isthmus lamina is small so that the lateral disc herniation at these levels frequently lies lateral to the isthmus. Because of the usual superior migration of the lateral disc herniation, a few millimeters of the isthmus need to be removed to identify the nerve root, the ganglion, and the disc fragment (following removal of the lateral extension of the ligamentum flavum, intertransverse muscle, and ligament). At these levels the facet joint partially or completely covers the intervertebral disc, but rarely is more than a small resection of the superior aspect of the facet needed. Often at L4-5 and always at L5-S1, an exposure through a paramedian incision permits a more direct approach and reduces the need for significant bone resection. However, in cases with significant spinal or lateral recess stenosis, or when severe facet tropism is present, partial or complete facetectomy is occasionally required to adequately decompress the nerve root. Conversely, at the L4-5 and L5-S1 levels the operative exposure decreases in size because the pedicles originate more obliquely, the isthmus laminae are substantially wider, a prominent accessory process may be present, and the facet joints substantially overlie the intervertebral disc space. There is also a progressive decrease in the cephalocaudal dimension of the operative window because of diminished distance between the inferior border of the transverse process of L4, the superior and lateral aspects of the facet joint and the transverse process at the L4-5 level, and the superior and lateral aspects of the facet joint and of the sacrum at L5-S1. The L4-5 and L5-S1 facet joints may overlap the intervertebral disc space significantly. Frequently, resection of the lateral aspect of the isthmus, the inferior aspect of the superior transverse process and accessory process, the lower one fourth of the superior pedicle, and the superior one fourth of the facet joint is needed to achieve satisfactory exposure, especially at the L5-S1 level.
SURGICAL CONSIDERATIONS
Lateral disc herniation presents a particular problem surgically because the disc fragment and disc space
247
are often hidden either under or lateral to the facet joint, especially at lower lumbar levels. Several surgical approaches to this area have been described: 1) a unilateral facetectomy through a midline incision; 2) a muscle-splitting paraspinal operation with lateral dissection between the transverse pr ocesses; 3) a partial hemilaminectomy with partial or complete facetectomy; 4) a retroperitoneal approach; and 5) a hemilaminectomy with foraminotomy (more recently, a midline approach with a partial resection of the isthmus laminae). This multiplicity of surgical approaches underscores the difficulty in achieving adequate exposure without potentially destabilizing the spine in the course of treating these herniations. Anatomical variations for each patient should be analyzed on both plain radiographs and imaging studies to help determine the optimal approach. Specific attention should be given to the bony structures in the intertransverse process region. Coexisting pathology within the canal should be defined and the operative approach tailored to address all significant pathology. Significant coexisting pathology such as severe facet joint arthritis, significant lateral recess or foraminal stenosis, or synovial cyst formation may require incomplete or total facetectomy. In cases where the facet joint has been significantly compromised, simultaneous fusion may be considered. We have gradually modified our surgical approach, placing greater emphasis on maintaining the integrity of the facet joint. The different approaches described can be divided in two broad categories: midline hemilaminectomy, and paramedian splitting of the paraspinal muscles. The paramedian muscle-splitting approach allows a lateral exposure either through the longissimus-iliocostalis interval or via a muscle-splitting approach through the iliocostalis muscle. This lateral approach provides more direct visualization of the disc fragment and permits reduced bone excision and muscle retraction, allowing greater exposure than the midline approaches. This approach can be effectively performed in cases of extraforaminal lateral disc herniation at any level and is frequently the procedure of choice for cases of foraminal or extraforaminal disc herniation at the L4-5 or L5-S1 levels. The anatomy involved, however, is somewhat unfamiliar to most neurosurgeons and it is not possible to directly address any intracanalicular pathology. The midline approach with several modifications is favored by many surgeons because of the ease of dissection, familiarity with the operative anatomy, and the ability to address pathology within the spinal canal. We believe that most disc herniations within the intervertebral foramen at the L2-3 and L34 levels can be reached via a hemilaminectomy approach (J.A.J.) or by an approach lateral to the isthmus lamina with resection of a portion of the isthmus (C.I.S.). Nevertheless, isolated disc hernia-
tions truly lateral to the facet joint from L1-2 to L3-4 can be very effectively undertaken via a midline approach lateral to the isthmus lamina without hemilaminectomy. For far lateral disc herniations, only a very small portion of the isthmus usually needs to be resected. This technique spares the majority of the facet joint and reaches the disc through the interval between the inferior aspect of the superior transverse process and the superior border of the inferior facet joint. If more than 50% of the facet joint is removed, the risk of spinal instability ranges between 1.6% and 2.4% and the incidence of back pain is substantially higher. Removing either the entire medial or lateral half of the facet joint results in total facet joint compromise. By preserving the facet joint, the risk of instability and postoperative pain is substantially reduced. If required, exploration of the intraspinal compartment can be performed through a partial hemilaminectomy for removal of any additional disc fragments or treatment of associated lumbar or lateral recess stenosis (present in 20% to 72% of the cases). If a simultaneous partial hemilaminectomy is planned, care must be taken not to make the isthmus overly thin after resection of the lateral aspect of the isthmus lamina. Consequences include predisposition to fracture as the patient regains full activity, back pain, and spinal instability.
After the administration of perioperative antibiotics and steroids and infiltration of the skin and paraspinal muscles with 0.25% bupivacaine with 1: 200,000 epinephrine, a midline skin incision of 6 to 10 cm is made slightly superior to the disc space of interest (Figure 4A). Dissection is carried down to the level of the thoracolumbar fascia, which is opened in a curvilinear fashion away from the insertion on the spinous process. This flap is retracted past the midline. Electrocautery and a periosteal elevator are used to subperiosteally dissect the paraspinal muscles away from the spinous processes and lamina. The paraspinal muscles are retracted just past the lateral border of the facet joint and a self-retaining retractor is inserted to maintain retraction (Figure 5). Care is taken to preserve the facet joint capsules during the exposure. The proximal aspect of the transverse processes, the isthmus lamina, and the facet joint capsules are identified. An intraoperative lateral radiograph is then taken to confirm the level of the dissection; a blunt marker placed lateral to the isthmus gives a good indication of position and facilitates orientation. We routinely use an operating microscope for bone resection and disc fragment removal to allow adequate magnification and proper illumination.
248
Figure 4. A straight incision is made along the midline for the medial hemilaminectomy approach (A) or 10 cm from the midline at the level of interest for the intertransverse process approach (B).
root ganglion. At the middle and lower lumbar levels, it can be difficult to enter the intervertebral disc space via this approach. The amount of aggressive disc removal from within the disc space has been reduced as we have observed that removal of the free fragment is all that is required for symptomatic relief in most cases. If exposure within the spinal canal is anticipated, the ligamentum flavum is opened more medially and a small hemilaminectomy is performed laterally on the superior lamina (Figure 2 inset). A small portion of the lateral section of the superior facet of the inferior vertebral body can be removed without compromising the facet joint (Figure 2 inset). Care is taken to maintain the integrity of the isthmus lamina. If there is associated intraspinal herniation and/or stenosis, the laminectomy can be enlarged to complete the intraspinal portion of the procedure (Figures 6 and 7A). To ensure complete decompression of the nerve root once the discectomy is completed, the foramen can be explored from inside and outside the canal using a blunt nerve hook (Figure 7B). At the end of the procedure, fat grafts are placed over the extraforaminal and intraspinal defects.
A microdrill is used to remove a small portion of the isthmus and, if required, the inferior aspect of the superior transverse process and accessory process, the lower one fourth of the superior pedicle, and the superior one fourth of the facet joint. The amount of bone resection required is frequently less with upper lumbar levels than with lower levels. This method usually provides adequate access to remove the disc herniation. The intertransverse fascia and the lateral extension of the ligamentum flavum are removed only when drilling and bony resection are completed because the dorsal root ganglion is vulnerable to thermal or mechanical trauma. The intertransversarii muscles and the intertransverse ligament can now be sharply removed from the transverse process of the superior vertebral body. This exposes the nerve root of interest, and ruptured disc material is usually seen inferior to the nerve root. The herniated lateral disc can then be removed with little nerve root retraction. Brisk bleeding can occur from the lumbar segmental artery and its branches if they are inadvertently compromised. Bipolar coagulation of bleeding vessels should be performed with good visualization of the vessel to avoid thermal trauma to the dorsal
The patient is placed in the prone position on a spinal frame. A spinal needle is inserted in the midline in the interspace of interest and an intraoperative x-ray is obtained to confirm the correct level. The skin and the paraspinal muscles are anesthetized with 0.25% bupivacaine with 1:200,000 epinephrine. A vertical incision, usually 8 to 10 cm in length, is made 10 cm from the midline (Figure 4B) through skin, fat, and then investing deep fascia. The transverse process is then palpated and access is gained in a 30-degree line to the horizontal by longitudinally splitting fibers of the iliocostalis muscle (Figure 3). The muscle is separated by blunt dissection, and handheld retractors are placed. A neurovascular band can be felt obliquely crossing the vertical muscle fibers and is followed ventromedially to the pedicle-transverse process junction (Figure 8A). A selfretaining retractor is inserted to maintain exposure. The operative microscope is then brought into position and the pedicle-transverse process junction is identified. When these steps have been followed carefully, this approach allows for a relatively avascular exposure. Adjacent to the intervertebral foramen is often a plug of fat which, when removed, exposes the origin of the posterior primary ramus, the ganglion, and the ventral root. Careful dissection of the tissue planes will separate the neurovascular structures from the disc fragment, which can be removed with pituitary microrongeurs without bone resection (Figure 8B). Most lateral disc herniations are sequestered frag-
Intertransverse Approach
249
Figure 5. After incising the lumbosacral fascia in a curvilinear fashion, the paraspinal muscles are dissected using a periosteal elevator and retracted laterally. (Reproduced from Chenelle AG, Shaf frey CI, Lanzino G, et al: Treatment of far lateral discs by an approach lateral to the facet joint. Techn Neurosurg 2:249-253, 1996, with permission)
Figure 6. A, the intertransverse muscle is sharply removed and the lateral herniated disc overlying the nerve root is visualized. B, the disc is removed. (Reproduced from Chenelle AG, Shaffrey CI, Lanzino G, et al: Treatment of far lateral discs by an approach lateral to the facet joint. Techn Neurosurg 2:249-253, 1996, with permission)
250
intertransverse muscle
herniated disc
Figure 7. A, if the herniated disc extends medially to the neural foramen, the hemilaminectomy is extended and the disc removed working on both sides of the foramen. B, a blunt nerve hook is inserted in the neural foramen to verify that no disc fragments are left behind. (Reproduced from Chenelle AG, Shaffrey CI, Lanzino G, et al: Treatment of far lateral discs by an approach lateral to the facet joint. Techn Neuro surg 2:249-253, 1996, with permission)
extreme lateral disc herniation lateral branch of the posterior primary ramus
A
transverse process
terminal branch of the segmental artery ventral nerve root dorsal root ganglion
Figure 8. A, the microscopic view obtained by the intertransverse route. B, after microdissection, the disc can be removed, decompressing the nerve root. (Reproduced from OBrien MF, Peterson D, Crockard HA: A posterolateral microsurgical approach to extreme-lateral lumbar disc herniation. J Neurosurg 83:636-640, 1995, with permission)
251
ments. If a diffuse disc bulge is present, the annulus can be incised and fine rongeurs used to remove the degenerated disc material. When a more medial exposure is required, the lateral margin of the isthmus lamina or facet joint can be removed without jeopardizing facet integrity. After discectomy, the split muscle comes together and sutures are confined to the deep fascia and superficial layers. No drains are necessary and the patient is mobilized a few hours after the surgical procedure.
With the availability and routine use of CT and MRI, intraforaminal and far lateral disc herniations are diagnosed more frequently. Lateral herniated discs pose particular surgical problems because the disc is either behind or lateral to the facet joint and it is difficult to treat using a standard lumbar hemilaminectomy without significantly compromising the facet joint. In cases of L1-2 or L3-4 lateral disc herniations,
CONCLUSIONS
we recommend a midline approach with resection of a small portion of the isthmus lamina that spares the majority of the facet joint. For some L4-5 and most L5-S1 far lateral disc herniations, the anatomical relationships often make an intertransverse approach more convenient. Each technique has its advantages and disadvantages. The midline approach allows for easy exploration of the intraspinal compartment through a hemilaminectomy when medial extension of the lateral disc fragment, foraminal stenosis, lumbar stenosis, or intracanalicular pathology is present at other levels. This technique is easy to learn and the anatomy involved is familiar to neurosurgeons. The amount of resection of the isthmus lamina and facet joint decreases as surgeons familiarity with this approach increases. The paramedian intertransverse technique is less familiar to many neurosurgeons and requires a new surgical orientation. This technique, however, allows removal of the disc with minimal bone resection and damage to the soft tissues and offers a more direct avenue to far lateral disc herniations, especially at the lower lumbar levels.
Primary and metastatic tumors may both involve the upper thoracic spine, and occasionally trauma affects this region of the spinal column. The standard anterolateral cervical approach combined with a median sternotomy provides satisfactory exposure of the cervicothoracic junction from C7 through the T2-3 junction, whereas a high posterolateral thoracotomy allows access to the thoracic spine up to the T3-4 disc space. The T3 through T4 region, on the other hand, is mostly hidden behind the great vessels, and neither of the approaches described above would be optimal for exposure of this region, which is located at the apex of the chest pyramid and offers little room for maneuvering instruments. The trap door exposure described by Nazzaro and coworkers in 1994 is a combination of the anterolateral cervical approach, median sternotomy, and anterolateral thoracotomy and is the most suitable intervention for gaining access not only to the T3 and T4 vertebral bodies anteriorly, but also to the entire ventral cervical spinal column, as well as to the mid and upper thoracic spine. Unlike previously described approaches to the cervicothoracic junction, it also spares the sternoclavicular junction and does not require transection of the clavicle.
INTRODUCTION
The main indication for this surgical procedure is an anteriorly located pathology resulting in spinal cord compression due to retropulsed bone or disc fragments and/or tumor mass or kyphotic angulation, primarily affecting the T3 and T4 levels. Although the anterior weight-bearing spinal column can be effectively reconstructed, either by using polymethylmethacrylate or a bone graft after the removal of a vertebra, the placement of supplemental anterior instrumentation is difficult to achieve due to the lack of appropriately designed upper thoracic spinal fixation devices. Whereas anterior cervical plate and screw constructs can be used for the fixation of this region in the absence of severe kyphotic angulation, additional posterior instrumentation is usually recommended for those who have significant spinal deformity. In patients who have significant spinal cord compromise at the time of presentation, surgical intervention is usually per formed within 12 to 24 hours of diagnosis. In patients with traumatic injuries, however, one may elect to postpone surgery until other concomitant injuries are dealt with and the patient is stabilized medically. Whereas magnetic resonance imaging (MRI) is preferred for evaluating the cervicothoracic junction and upper thoracic spine in most patients, computed tomography (CT) can provide important additional details of bony anatomy and is very helpful in deter-
PATIENT SELECTION
253
254
mining the nature of compressive pathology (i.e., the presence of retropulsed bone fragment(s) vs. tumor mass effect). This is particularly important if one is contemplating the use of radiation therapy as the primary mode of treatment. Isolated lesions involving the spine but not the lung can be addressed in patients with marginal pulmonary function. In addition, because the right recurrent laryngeal nerve is at considerable risk during the exposure, an appropriate preoperative laryngoscopic examination is usually needed to rule out vocal cord paralysis, especially in those in whom a left-sided recurrent laryngeal nerve dysfunction is already suspected.
In addition to the routine monitoring needed for general anesthesia, the patient should undergo placement of a large-bore intravenous catheter, because these surgical procedures may require a significant amount of blood volume replacement during the vertebrectomy pr ocedure. This catheter is placed in a femoral or left subclavian vein, thereby avoiding the right internal jugular and right subclavian veins, which are in the surgical field. The intravascular volume status is monitored with either a central line or a Swan-Ganz catheter, depending upon the patients cardiopulmonary reserve. Second-generation cephalosporins are routinely administered preoperatively for prophylaxis.
PREOPERATIVE PREPARATION
ANESTHETIC TECHNIQUE
The use of inhalation anesthetics should be minimized in order allow somatosensory evoked potential (SSEP) monitoring for assessment of spinal cord function during surgery.
OPERATIVE POSITIONING
OPERATIVE PROCEDURE
The skin incisions path and the location of critical cuts through the chest wall that are needed to expose
the important anatomical structures are illustrated in Figure 1 (with inset). This exposure is carried out through the right side. The incision extends along the anterior border of the sternocleidomastoid muscle down to the sternal notch, as shown in the intraoperative photograph (Figure 2A). From there the incision is carried down the midline and over the sternum to the fourth intercostal space, where it curves laterally and follows the fourth interspace. The neck platysma is then incised, and a plane is established between the carotid sheath (laterally) and the tracheoesophageal viscera (medially) using blunt dissection. During this stage of exposure, both the omohyoid muscle and middle thyroid vessels may need to be ligated and transected to gain access to the lower cervical spine. Subsequently, dissection is carried further down to the prevertebral fasciae, where the longus colli muscles are visualized bilaterally, running longitudinally over the anterior cervical spinal column off the midline. A plane is established at this point between the manubrium and the underlying vascular structures. The anesthesiologist is then instructed to selectively deflate the right lung, so that the chest cavity can be entered without injuring the lung underneath. Next, a sternal saw is used to transect the sternum in the midline down to the fourth intercostal space (Figure 2B), where this cut is connected laterally to the previously made thoracotomy incision. A chest spreader is placed to open the sternum in the midline and to retract the anterior chest wall superolaterally (Figure 2C). At this stage, the mammary artery is identified and may need to be ligated and transected to permit further retraction of the chest wall. Attention is then directed toward the carotid sheath. The sheath is opened and the vagus nerve identified (Figure 2D). The innominate artery is then located under the sternal notch and is dissected free laterally up to the take-off of the right carotid artery. Just distal to that, the right subclavian artery is identified and the vagus nerve is traced down to the takeoff of the right recurrent laryngeal nerve, which loops
Figure 1. Artists depiction of the trap door exposure. Anterolateral chest wall is retracted superolaterally. Retractors in the upper cervical region and over the sternal angle allow visualization of all important anatomical structures in the midline. The right carotid artery and right vagus nerve are pulled laterally, whereas the right recurrent laryngeal nerve (shown exaggerated to demonstrate its entire course and relationship to the subclavian artery and tracheoesophageal structures) is retracted ventrally and medially. The anterior cervical spinal column is visualized through the space between the carotid artery and the tracheoesophageal viscera. In the chest cavity, the aorta, the left common carotid artery (origin), the innominate artery, the right subclavian artery, and the take-off of the right common carotid artery are shown. Also visible are the transected azygos vein, the superior vena cava, and the left brachiocephalic vein. Rib heads have been removed, and T3 and T4 vertebrectomies have been completed. Through the vertebrectomy defect, the dura and the right T3 nerve root are seen.
255
esophagus right vagus nerve right recurrent laryngeal nerve cervical spine trachea innominate artery left brachiocephalic vein right common carotid artery sternum
T5
aorta
right ventricle
Figure 1.
256
innominate artery
A
aorta
D
right brachiocephalic vein right lung heart
257
under this vessel. This nerve is then traced more medially to where it enters the tracheoesophageal groove, and a vessel loop is placed around it for identification. Similarly, vessel loops are placed around the carotid, subclavian, and innominate arteries, as well as the brachiocephalic vein. Further dissection at this stage also allows visualization of the thymus. After the recurrent laryngeal nerve has been identified and properly protected, blunt dissection is used to obtain further exposure of the cervicothoracic junction. This provides exposure of the upper thoracic spine, usually at the T1-T2/3 junction. T3 and T4 are mostly hidden from view behind the great vessels. The junction of the mediastinal, apical, and posterior chest wall pleurae covers the anterior and anterolateral surface of the T3 and T4 vertebral bodies. The pleura is incised over the spinal column and is dissected off the spine (Figure 2E). The azygos vein usually crosses the T4-5 disc space from a lateral to a medial direction in joining the superior vena cava. This vein is dissected free of the surrounding areolar tissue and is ligated, transfixed, and transected. At this point it is possible to visualize the segmental vessels at the T3 or T4 levels and, if needed, at the T5 level. These vessels are doubly ligated with 20 silk ties, clipped, and transected, allowing them to be dissected free of the vertebral bodies and to be retracted more laterally. After the thoracic level has been confirmed, either with an intraoperative x-ray or by visual inspection, the vertebrectomy is carried out using the standard landmarks. Usually, the disc space between the vertebrae above and below is
identified, and then the annulus is incised using a #15 blade. Subsequently, the disc material is removed using curettes of various sizes and Leksell rongeurs. A high-speed diamond-tipped burr is typically used to remove the vertebral body down to the level of the posterior longitudinal ligament. In the case of tumor resection, an ultrasonic aspirator is often useful for breaking down and aspirating the tumor in a piecemeal fashion. Bipolar electrocautery is used for hemostasis. After the posterior longitudinal ligament is exposed, it is opened and removed, usually with a small upward-angled curette. Either transaxial illumination with headlights and loupe magnification or an operating microscope can be used during this part of the procedure.
After completion of the vertebrectomy, r econstruction of the spinal column is performed. In patients with tumors, polymethylmethacrylate is used. However, in trauma patients, autologous bone from the iliac crest is preferred. A fibular strut graft or a tibial allograft can also be used for reconstruction of the anterior weight-bearing spinal column. After completion of the reconstruction, the placement of anterior instrumentation can usually be carried out using a cervical plate and screw construct (Figure 2F). The low cervical exposure present at this point allows insertion of the screws into the vertebral bodies anteriorly at the T2 level, whereas the lower aspect of the plate can be carried down to T5 or even further, depending upon the number of levels of ver-
Reconstruction
Figure 2. Intraoperative photographs of the patient presented as an illustrative case. A, the position of the patient and planned skin incision prior to draping. The patients head is turned to the left and his neck is slightly extended by placing a roll between the scapulae. The arms are tucked in at the sides and padded. The incision follows the anterior border of the sternocleidomastoid muscle down to the sternal notch, from which it is carried to the fourth intercostal space over the sternum in the midline. The incision is then curved laterally, following the space between the fourth and fifth ribs. Also visible is a double-lumen endotracheal tube that allows selective ventilation of the lungs during surgery. B, the use of a sternal saw following cervical dissection and completion of the anterolateral thoracotomy. C, close-up view, showing the transition from the lower cervical region to the chest cavity. The anterolateral parietal pleura, along with the chest wall, is retracted superolaterally. Also visualized are the right lung and the right brachiocephalic vein, as well as the pericardium (more medially). D, all the anatomical structures of importance. The lower cervical dissection has been completed, and the right lung has been selectively deflated. Visualized structures include the right common carotid artery, right vagus nerve, right innominate artery, right subclavian arteries (origin), right brachiocephalic vein, pericardium, and aorta. Also seen is the right recurrent laryngeal nerve, which loops around the right subclavian artery and enters the tracheoesophageal groove, crossing the C7T1 vertebral region. The surgeons hand is retracting the tracheoesophageal viscera medially. E, close-up view of the posterior chest wall and the upper thoracic spine. The surgeons hand is retracting the right lung to show the thoracic spinal column, which is covered by parietal pleura extending from the mediastinum. Parietal pleura has been resected over the spine inferiorly, down to the T5-6 disc space. Also seen are the DeBakey forceps, which have been passed from the lower cervical region into the chest cavity through an opening in the apical pleura. The azygos vein, which crosses the field at the T4-5 level, has been ligated and transected, allowing further medial retraction of mediastinal pleura. F, close-up view showing the final construct. The T2 through T4 vertebrectomy has been completed, and the anterior spinal column has been reconstructed with polymethylmethacrylate, which is hidden from view under the plate. Lateral to the plate, decompressed dura can be observed. The cervical plate extends from T1 to T5 and has vertebral body screws located at the T1 and T5 levels.
258
tebrectomy performed during the procedure. At this point, exposure of the lower cervical spine is achieved in a standard fashion by mobilizing the longus colli muscles bilaterally. At no time during this procedure should metal retractors be used, as they could potentially compress the recurrent laryngeal nerve.
After completion of the instrumentation, the right lung is re-expanded. Following irrigation of the chest cavity, two chest tubes are usually placed, one located posteriorly and the other more anteriorly. Both tubes are brought out through the anterolateral chest wall along the anterior axillary line at the T8-9 level. The ribs are brought together with a rib approximator, and the rib cage is closed using 2-0 Vicryl pericostal interrupted sutures. The sternum is closed using interrupted 5-gauge stainless steel wires. The peristernal fascia and the subcutaneous layer are closed with 1-0 and 2-0 Vicryl sutures, respectively. The neck incision is closed at the platysmal layer using 3-0 Vicryl running sutures. The subcutaneous layer in the cervical region is approximated using 3-0 Vicryl inverted interrupted sutures, and the skin is closed using a subcuticular 4-0 Vicryl running suture. Both the sternal incision and the anterolateral thoracotomy incision can be closed using either staples or a subcuticular running suture. If needed, an additional drain can be placed in the cervical incision and brought out of the operative field through a separate stab wound incision.
Closure Techniques
Throughout the procedure, the important vascular and neural structures are protected and are in view at all times. A diamond-tipped burr is used exclusively for the removal of the vertebral body because it does not snag either the dura (during the exposure of the thecal sac) or the other soft tissue structur es around the exposed area. In addition, an ultrasonic aspirator is very helpful for removing tumor tissue.
Specialized Instrumentation
During the procedure, SSEP monitoring is r outinely used to assess the spinal cord function. Fluid volume replacement is determined by the blood loss observed during the procedure, as well as by intravenous pressure or Swan-Ganz catheter readings.
Monitoring
COMPLICATIONS
Complications of the trap-door exposure of the cervicothoracic junction procedure are those typical of anterolateral cervical exposure, as well as complications related to the thoracotomy procedure. The neural structure at highest risk is the recurrent laryngeal nerve, and great attention should be paid to preserve
this structure completely, as well as to avoid placing it in traction, especially with metal retractors, during the exposure. Other structures at risk include the carotid artery, the subclavian and vertebral arteries (as they are dissected out), the vagus nerve, and, more medially, the esophagus and trachea. Blunt dissection allows protection of the tracheoesophageal structures medially and the carotid artery more laterally. In the chest cavity, the great vessels, the innominate artery, the subclavian artery, and the brachiocephalic vein can potentially be injured during the dissection. However, injuries to these structures can be dealt with more easily if the thoracic surgical exposure has already been completed. Should a dural tear occur during the vertebrectomy procedure, its primary repair is extremely difficult. In this situation, one can usually place a piece of muscle or fascia lata over the dura along with Gelfoam and fibrin glue. If cer ebrospinal fluid (CSF) is visualized during surgery, placement of a postoperative lumbar drain for CSF drainage for 4 to 5 days following the operation is almost always required, and the patient is kept at bed rest. If the surgeon believes the potential exists for a continuous CSF leak, a vascularized intercostal muscle flap can be used to plug the dural opening. Again, this would require postoperative lumbar CSF drainage. Unfortunately, the chest tubes that are routinely placed to prevent postoperative hemo- or pneumothorax would also contribute to a CSF leak by creating a negative pressure. Therefore, in the presence of an intraoperative CSF leak, one should remove chest tubes as quickly as possible (i.e., as soon as the thoracic output has fallen below 150 cc in 24 hours or when there is no significant pneumothorax on the follow-up chest x-ray). A CSF leak occurring postoperatively in spite of these precautions may require reoperation and repair of the dural tear, usually with a vascularized muscle graft (as described above) and postoperative CSF drainage. Esophageal laceration can also occur, either as a result of sharp dissection or from the sharp teeth of the cervical retractors. If the laceration of the esophagus is recognized intraoperatively, it should undergo primary repair. The wound should be drained and the patient placed on nasogastric drainage for at least 7 to 10 days. Subsequently, a barium swallow study should be obtained to confirm that the perforation has sealed. However, in the majority of the cases, injury to the esophagus is not recognized at the time of surgery and presents later as a local infection, fistula, generalized sepsis, or mediastinitis. The presence of crepitus in the neck or of mediastinal air on a chest x-ray should strongly suggest the possibility of an esophageal perforation. The diagnosis can be confirmed with an esophagogram, although this test is not always positive. Esophagoscopy or CT scanning
259
following an esophagogram may also demonstrate perforation. The treatment of a delayed perforation consists of nasogastric drainage, antibiotics, and reexploration of the incision. Once found, a defect can be repaired. In order to avoid injury to the esophagus, one should minimize the use of sharp dissection below the level of the superficial cervical fascia. In addition, the longus colli muscles should be freed enough both rostrally and caudally that the sharp teeth of the self-retaining retractors can be safely placed under them without being likely to dislodge during the procedure. Also, attention should be paid to make sure that the esophagus or the other soft tissue structures are not protruding into the field between the retractors during the use of the power drill, because the esophagus can sometimes get caught in the drill burr, resulting in esophageal per-
foration. Occasionally, a displaced graft can perforate the esophagus, thereby necessitating re-exploration. Either replacement or removal of the graft may be indicated, depending upon the need for the graft in maintaining spinal stability. The esophageal perforation should be repaired, if possible, and the patient treated with antibiotics and nasogastric drainage, as described previously.
A 27-year-old white man with a history of renal cell carcinoma presented with a 3-month history of progressively worsening interscapular pain. His neurological examination was unremarkable. MRI revealed a T3 vertebral body metastasis involving primarily the left pedicle as well as radiographic spinal cord
ILLUSTRATIVE CASE
Figure 3. Preoperative MRI of the patient in Figure 2. The patient had already undergone a transpedicular T3 vertebrectomy for metastatic renal cell cancer, followed by insertion of posterior instrumentation with titanium rods, hooks, and Wisconsin wires. Six months later, he presented with recurrent tumor around the methylmethacrylate construct that was placed during the initial procedure, resulting in myelopathy. Epidural tumor extended rostrally and caudally behind the T2 and T4 vertebral bodies.
Figure 4. Postoperative MRI of the patient in Figure 2 showing the T2 through T4 vertebrectomy, the methylmethacrylate construct extending from T1 to T5, and the cervical locking plate/screw fixation. Artifacts visible posterior to the thecal sac are the result of previously placed cervicothoracic instrumentation consisting of hooks, rods, and Wisconsin wires.
260
and nerve root compression. The patient underwent a T3-4 laminectomy, a transpedicular T3 vertebrectomy, reconstruction with methylmethacrylate, and had posterior cervicothoracic instrumentation and fusion (C4-T8). Titanium rods, hooks, and Wisconsin wires were used initially for posterior fixation. The patient did well following surgery and was placed on chemotherapy. He returned 4 months later with left T3 radiculopathy. MRI (Figure 3) showed recurrent tumor posterior to the methylmethacrylate con-
struct, extending behind the T2 and T4 vertebral bodies. A neurological examination revealed myelopathy. At this point, the decision was made to proceed with a trap door surgical exposure and to per form a T2-4 vertebrectomy and an anterior reconstruction with instrumentation, as described in this chapter and figures. Postoperatively, the patients myelopathy resolved (Figures 4 and 5). In this patient, surgery was followed by local radiation therapy. At 1year follow-up review, the patient was fully ambulatory with no symptoms.
Figure 5. Postoperative anteroposterior (left) and lateral (right) x-rays of the patient in Figure 2 showing that anteriorly, the construct consists of methylmethacrylate (which replaces the T2-4 vertebral bodies) and a cervical locking plate (with screws) extending from T1 through T5. Also seen is the initially placed posterior cervicothoracic fixation including rods, hooks and Wisconsin wires.
To appreciate the suture repair of a peripheral nerve, an understanding of the connective tissue layers as well as the fascicular anatomy of a nerve is important. As seen in Figure 1, which contains a diagram of the peripheral nerve architecture and its components, an external epineurial sheath surrounds each peripheral nerve and is composed of connective tissue and longitudinal blood vessels. There is an external epineurium and an internal epineurium. The internal epineurium demarcates fascicles and groups of fascicles within the nerve. Each individual fascicle is surrounded by perineurium. The axons are contained within fascicles, in close association with Schwann cells and the basement membrane that surrounds Schwann cells, the endoneural basal lamina. The proportion of connective tissue within the nerve varies considerably, from 25% to 85%, throughout the cross section of the nerve. In general, there is more connective tissue in the nerve where it crosses the joint. The connective tissue, particularly the perineurium, is the source of the main tensile strength to the nerve. It is also the layer that can take and hold a suture. From a practical viewpoint, the smallest component of nerve that can be repaired is the fascicle.
INTRODUCTION
Nerves may be generally divided into four basic patterns of intraneural architecture, based on their fascicular structure (Figure 1). Nerves containing one large fascicle are termed monofascicular, while those containing a few or a discrete number of fascicles are oligofascicular. Most nerves contain many fascicles of varying sizes and are termed polyfascicular. The polyfascicular pattern exists either with fascicles grouped together or with a more diffuse (ungrouped) arrangement throughout the cross section of the nerve. The fascicular nature of a nerve changes as it extends from proximal to distal in the extremity. For example, the ulnar nerve is polyfascicular as it comes off the brachial plexus and then generally becomes organized into four fascicles at the level of the elbow. These fascicles are further segregated into motor and sensory groupings at the level of the wrist and, finally, the motor branch in the hand is monofascicular. Following a nerve injury resulting in axotomy, the distal axon undergoes Wallerian degeneration. Attempted regeneration from the proximal axon may or may not be successful, depending primarily on local and certain systemic factors. In the absence of effective spontaneous regeneration, repair of the peripheral nerve is in order. Methods of peripheral nerve repair fall under two basic categories: direct repair (neurorrhaphy), and bridge procedures in which autologous nerve grafts are most commonly utilized. The direct or suture repair may be performed using an epineurial, a grouped fascicular, a fascicular, or various combinations of these methods.
261
262
Figure 1. Schematic diagram of peripheral nerve architecture and composition. The patterns of fascicular structure of peripheral nerve are illustrated: monofascicular, oligofascicular, and polyfascicular (grouped and ungrouped). The anatomical composition, with emphasis on the connective tissue layers of the nerve (i.e., epineurium (external and internal), perineurium, and endoneurium) are illustrated.
It must be stressed at the outset that a peripheral nerve repair is not a type of cellular repair, but is actually a repair done at the level of the connective tissue to coapt a healthy proximal nerve to a healthy distal nerve stump. The appropriate anatomical environment is then provided, so that axons from the proximal stump can regenerate into endoneural tubes within the distal nerve stump and, hence, be led to end organs to restore function. Note that a nerve graft functions as a conduit whose axons are destined to undergo degeneration as soon as it is
removed from its harvest site. Thus, the graft provides an endoneural tube network available to be exploited by regenerating axons from the proximal host nerve stump. It also provides viable Schwann cells, as long as the caliber of the nerve graft is not too large. For this reason, small-caliber cutaneous nerves are most commonly used as graft material. The small-caliber nerves, when sutured in a series of parallel segments, are in close proximity to tissue fluid and are therefore nourished. They also undergo rapid revascularization, and thus remain viable.
263
A detailed knowledge of the gross anatomy of the extremities and peripheral nerves is imperative before undertaking nerve exploration and repair. The surgeon must be prepared to expose the nerve well proximal and distal to the area of injury. Appropriate positioning of the limb, padding of pressure points, and wide draping are essential. Special attention to draping of the limb or a different limb for procuring nerve grafts is also required. Since the nerve may need to be stimulated intraoperatively to evoke muscle contractions, only a short-acting paralyzing agent, given at the induction of anesthesia, should be used. The author prefers the use of general anesthesia and does not use tourniquets for these procedures. The isolation of the nerve itself should be performed using sharp dissection. The surgeon identifies normal nerve proximal and distal to the zone of injury, and then works toward the area of injury. In clean lacerating injuries, the area of exposure may be relatively small. However, most injuries leave the nerve in continuity; since these injuries are also explored weeks to months following trauma, there is considerable scar formation and distortion of tissue, necessitating a wide exposure. Using sharp dissection techniques, the area of injured nerve is circumferentially exposed. With the aid of an operating microscope, the gross anatomical details of the injury are identified. As advocated by Kline, intraoperative nerve action potential studies dictate an external neurolysis only if evidence of spontaneous regeneration (the presence of a nerve action potential studies) is found (Figure 2). On the other hand, the lack of evidence of spontaneous regeneration (the absence of a nerve action potential studies) dictates resection of the neuroma and appropriate repair. The placement of lateral stay sutures using 6-0 monofilament (as illustrated in Figure 2A and B) helps maintain the topographical alignment of the nerve. Under the operating microscope, the surgeon then cuts across the center of the neuroma. Small segments of the nerve are sliced in perfect cross section, using a fresh blade, until a healthy fascicular pattern is identified both at the proximal and at the distal stump (Figure 2C and D). This step is critical, as attempting to oppose or graft scarred proximal and distal stumps is a major cause of nerve repair failure. Healthy fascicular tissue is recognized when the epineurium retracts slightly and the endoneurium appears to pout or mushroom out of the fascicles (because of positive endoneural pressure). This type of adequate debridement invariably leaves some degree of gap between the proximal and the distal stump. If the gap is short and the two ends can be brought together without undue tension, a direct repair is appropriate. One good way to determine the
degree of tension present at the suture line is to bring the ends together using the stay epineurial sutures. If this can be performed without suture distraction, a direct repair is appropriate. However, if the ends are under considerable tension and the suture line appears to tear out, a graft repair must be performed. Several techniques are available to bring the proximal and distal stumps closer together and allow a direct repair to be performed. In all situations, proximal and distal mobilization of the nerve for considerable distances should be performed. Thus, tethering forces to adjacent surrounding fascial and subcutaneous tissue are removed, allowing short gap lengths to be overcome. In certain situations, such as the ulnar nerve at the elbow, the nerve may be transposed, allowing a considerable length to be obtained (Figure 3). Nerve repairs performed with a joint in slight flexion (e.g., the elbow joint for upper extremity and the knee joint for lower extremity) may be appropriate. In these situations, the extremity will need to be splinted in this position for 2-3 weeks postoperatively and then gradually mobilized with physiotherapy. Excessive flexion of joints, however, should be avoided. Indeed, most nerve repairs should be performed with the extremity in a state of near-maximal extension, so that postoperative mobilization occurs without risk of distraction at the suture line. One can therefore avoid prolonged limb immobilization. Under most circumstances for extremity nerve repairs, I place bulky dressings (but not splints) around the incision area and leave in place for 2 weeks as a reminder to the patient to avoid excessive motion. Brachial plexus repairs are protected with a shoulder immobilizer or sling for 3 weeks. Thereafter, the patient is allowed gradual and progressive range of motion supervised by an appropriately instructed physiotherapist.
Direct end-to-end repair is possible in the majority of clean lacerating injuries and in cases of delayed repair when the two ends can be brought together without tension. The author uses magnification with the operating microscope (others may prefer loupes) for all repairs. Indispensable tools include microinstruments with fine tips (such as jewelers forceps) and fine-tipped microsuture needle drivers. Commonly, 8-0, 9-0, or 10-0 monofilament nylon microsutures are utilized, determined by the caliber of the nerve undergoing repair. In general, 8-0 and more rarely 9-0 sutures are utilized for proximal repairs such as brachial plexus elements and 10-0 sutures are utilized for more distal repairs and fascicular coaptations. To control bleeding from the nerve ends, minor oozing is often halted by simple
DIRECT REPAIR
264
Figure 2.
265
Figure 3. Drawing demonstrating techniques available to gain length to allow direct repair of nerve ends without tension. An injury gap of the ulnar nerve just distal to the elbow is used as an example. In addition to slight flexion of the elbow joint, the ulnar nerve undergoes anterior transposition. These two maneuvers relax the nerve, overcoming the gap, and permit tension-free direct repair.
Figure 2. A and B: surgical management of a neuroma in continuity. A, operative photomicrograph illustrating injury to the peroneal nerve just proximal to the head of the fibula. The superficial peroneal nerve (in the upper part of photograph, encircled with the Penrose drain) conducted a nerve action potential and underwent neurolysis. There was no demonstrable evidence of regeneration across the deep peroneal branch (no return of clinical function and absence of a nerve action potential across the neuroma). The nerve appears attenuated at the site of the neuroma, with 6-0 stay sutures placed proximal and distal to the neuroma. B, drawing of a neuroma prepared for resection and debridement. Stay or lateral stay sutures are positioned to orient and topographically align the nerve. The site of the initial nerve section is across the middle of the neuroma (solid line in center). C and D: sectioning technique used to debride the neuroma in continuity until all scarring has been removed and a fresh fascicular pattern is observed at the nerve stump. C, operative photomicrograph demonstrating sequential sectioning of the neuroma under microscopic magnification to obtain a fascicular pattern. D, schematic diagram demonstrating progressive sectioning until a grossly normal fascicular pattern is observed. (Figure 2C reproduced courtesy of Dr. Alan Van Beek)
266
Figure 4. Epineurial suture repair technique. Lateral stay sutures and vascular landmarks are used to maintain topographic alignment. Sutures are placed with the needle passed through the internal and external epineurium from both stumps to approximate the nerve ends, until the final repair is achieved (inset).
pressure with cottonoids. However, more substantial bleeding from small epineurial vessels should be controlled using small amounts of bipolar current delivered via fine-tipped jewelers forceps. The use of a microirrigator for saline flushes enhances visibility and further aids the performance of the nerve repair. Direct repair techniques include epineurial, grouped fascicular, and fascicular repairs. The indications and use of each of these techniques are described here.
Epineurial suture repair has been a traditional method of nerve coaptation. These repairs are most appropriate for monofascicular (e.g., digital) nerves and diffusely grouped polyfascicular (most proximal limb and plexus element) nerves. Simplistically, this method achieves continuity of the connective tissue from the proximal to the distal stumps, without tension and with appropriate rotational alignment of both stumps. The goal is to obtain good coaptation of proximal and distal fascicular anatomy. Freshening of the two nerve ends to debride the nerve and re-
Epineurial Repair
move scar tissue is therefore critical. Achieving appropriate nerve alignment can be aided by inspecting for longitudinal blood vessels in the epineurium as well as attending to fascicular alignment. The use of lateral stay sutures (Figure 2) also aids this process. Neurorrhaphy is performed using 8-0 to 10-0 nonabsorbable nylon sutures. A small bite of the internal and external epineurium (being careful to avoid perineurium) is taken from both stumps and the suture is tied using only mild to moderate tension (Figure 4). It is critical to avoid tying the knot under too great a tension, as this will cause overriding or an accordion effect on the fascicles or, in fact, pouting out of a fascicle from the epineurial repair site, thus defeating the purpose of suturing. The two initial sutures are placed 180 degrees apart from each other (Figure 4). This distance is then divided in half and two further sutures are positioned. The number of epineurial sutures required varies; in the majority of cases, four to eight sutures suffice for approximating the proximal and distal stumps in a tension-free manner. Excess sutures may result in additional scarring and are to be avoided.
267
Figure 5. Grouped fascicular repair technique. Microsutures are placed through the interfascicular epineurial tissue, avoiding perineurium and allowing coaptation of fascicular groups from the proximal to distal stumps. The sutures are tied using mild to moderate tension, preventing overriding of fascicles.
A grouped fascicular repair technique is a potentially more accurate method than epineurial repair. Theoretically, a disadvantage of epineurial repair is the inability to precisely match the appropriate proximal and distal fascicles. However, experimental and clinical studies have not shown a clear advantage of one technique over the other. For practical purposes, the grouped fascicular technique is especially indicated in situations where an easily identifiable part of the cross section of the nerve supplies sensory function, while another portion of the nerve supplies motor function. More distal extremity nerves, such as the elbow-to-wrist segments of the ulnar and median nerves, are examples of nerves that merit this type of repair. Another indication is the nerve injury requiring a split repair. In this situation, a portion of the nerve that is clearly regenerating (by clinical and electrical criteria) is preserved in continuity using external and internal neurolysis techniques, while the groups of fascicles that are clearly neur otemetic
undergo repair. As in the epineurial repair method, the nerve ends are matched by resecting damaged tissue. Debridement is followed by careful analysis, under the operating microscope, of the anatomical cross sectional appearance of the nerve stump. Using the cross sectional appearance, the longitudinal blood supply, and other spatial landmarks (e.g., branching of nerve just proximal and distal to the injury site), the proximal and distal stumps can be matched. Interfascicular dissection is then performed within the internal epineurium to draw out groups of fascicles (Figure 5). Groups of fascicles may vary from two to several, each surrounded by a variable amount of internal epineurium, with the external epineurium dissected away (Figure 5). Once groups of fascicles are adequately matched, 8-0 or 9-0 microsutures are placed through the interfascicular epineurial tissue, avoiding perineurium and allowing coaptation of fascicular groups from the proximal to the distal stump (Figure 5).
268
Figure 6. Fascicular repair. A, an acutely lacerated nerve has its fascicular configuration dissected out, with topographical relationships used to align proximal and distal fascicles. B, fine 10-0 nylon microsutures, placed in the perineurium, are used to coapt individual or groups of two or three fascicles to each other. Note that each fascicular repair requires two or, at most, three sutures. (Figure 6B reproduced courtesy of Dr. Jack Tupper)
Figure 7. Graft repair of the peripheral nerve. The interfascicular technique is used to repair individual graft segments to groups of fascicles from the host proximal and distal stumps. A, schematic drawing of interpositional grafts being sutured to the host nerve. Note that in the graft undergoing suturing, the needle bites of the graft and host stumps allow widening (or fishmouthing) of the cross sectional area of the graft to allow complete coverage of the host stump. B, operative photograph showing the completed repair of posterior tibial nerve injury with four interposed sural nerve grafts. (Figure 7B reproduced courtesy of Dr. David Kline and from Dubuisson A, Kline D: Neurol Clin North Am 10:935-951, 1992 with permission)
269
In rare situations, a specific fascicular repair may be indicated. An example of where this repair would be appropriate is a clean lacerating injury of the median or ulnar nerve at the wrist, repaired primarily where individual motor and sensory fascicles can be defined. Another example is a partial injury across a portion of the cross section of the nerve. Here, individual injured fascicles from the proximal and distal stumps may be specifically repaired to each other. This repair requires the use of high magnification and fine 10-0 nylon microsutures swaged on a 75-m needle. The suture is placed through the perineurium of the proximal and distal stumps with careful avoidance of the internal endoneural contents (Figure 6). Again, the purpose is to attain tension-free coaptation of the proximal and the distal stumps. Tight suture ties should be avoided, as they will result in lateral bulging and herniation of endoneural contents. Also, only two or three sutures should be used in this situation, placed circumferentially around the nerve at approximately equal distances. The fascicular repair can then be strengthened by repairing loose epineurial sutures of the surrounding nerve and grouped fascicle repairs of the remaining nerve elements. In this sense, many direct nerve repair techniques are a combination of epineurial and group fascicular repairs. In the minority of situations, the fascicular repair of a specific nerve component may be required. Thus, the surgeon should be adept in all of the techniques described here. The use of a given technique or combinations will depend on the specific nerve injury encountered and the underlying specific fascicular pattern.
Fascicular Repair
The surgical technique of nerve grafting is similar in many ways to that of direct nerve repair. From a simplistic viewpoint, when direct repair cannot be performed without undue tension, nerve grafting must be undertaken. A corollary to this is that when nerve grafting is performed, there must be no tension at the proximal and distal repair sites to prevent postoperative distraction. Therefore, nerve grafting is generally performed with the extremity in full extension. In general, a nerve graft should be about 10% longer than the existing nerve gap. The majority of autologous nerve grafts are cutaneous nerves of the lower (e.g., sural) and upper extremities. (Donor nerves that are available for nerve grafting and methods for procuring these are discussed in the next volume of the atlas.) It is imperative to harvest the maximum amount of nerve graft material available. When the caliber of the host nerve stump undergoing
GRAFT REPAIR
grafting is the same dimension as the nerve graft, a slightly larger segment than the gap is needed. Frequently, the cross sectional area of the host nerve stump will be several folds larger than the diameter of the nerve graft. Several segments of nerve graft will therefore be needed to allow coverage of the entire cross sectional surface of the host nerve stumps. This must be taken into consideration when harvesting the donor nerve. The graft repair technique is similar to the various repair methods already outlined, with the exception that there are two suture lines. When preparing the proximal and distal stumps, the interfascicular and external epineurial tissue must be dissected away to allow groups of fascicles (or fingers) to be created at each stump (Figure 7). The diameter of the group of fascicles should approximate that of the cutaneous nerve graft obtained. The nerve graft is then sewn in place to the proximal group of fascicles, using epineurial and interfascicular epineurial techniques (Figures 5 and 7). Sutures, approximately 180 degrees apart, are positioned and passed through the epineurium of the nerve graft and then through the interfascicular epineurium of the host stump, spreading the cross section of the graft in a fish-mouth configuration (Figure 7). The end of the graft segment then covers as much of the fascicular area of the host nerve stump as possible, allowing capture of the maximum number of regenerating axons. Typically, two sutures, and occasionally three or four sutures, are required for each nerve graft at each coaptation site. The nerve graft segments are sutured to the distal group of fascicles in a similar manner. It is important to re-inspect all proximal and distal coaptation sites at the completion of the repair to ensure that no graft ends became disconnected while others were being sutured in place. As much as possible, the grafts are aligned in orientation so that groups of proximal fascicles will be directed to appropriate groups of distal fascicles. This allows for the most precise repair possible. For each nerve treated, an understanding of the gross anatomy and the branching pattern of the nerve, as well as knowledge of interfascicular anatomy, is helpful. As Sunderland has shown, the plexiform nature of the more proximal nerve elements, where groups of fibers are exchanged among fascicles in a longitudinal manner, defeats exact matching. In general, when dealing with proximal nerves and nerve elements such as the brachial plexus, spatial alignment of fascicles allows the best possible matching. In the most distal nerves (e.g., the median and the ulnar nerve at the wrist), more precise anatomical matching with regard to group fascicular and, occasionally, specific fascicular anatomy is possible.