P7S Medical Review: Spring 1997, Vol.4, No.1
The Changing Concepts in the Treatment of Brain AVMs: Thirty Years' Experience

I. Introduction

Arteriovenous malformations (AVMs) of the brain represent approximately 2,500 new patients in the United States yearly and are about one-tenth as common as intracranial aneurysms to which they are often associated^1,2 (Figure 1). Although rare, even for a specialist in the field of neurosurgery, brain AVMs represent an excellent model for the study and treatment of a particularly lethal, congenital lesion which can have a profound effect on the physiology and anatomy of the brain.^3-5 Our experience at the Neurological Institute of the Columbia-Presbyterian Medical Center is unique in managing these complex and rare lesions.⁶ We have operated on over 450 AVMs of the brain and see at least double that number in consultation. At our institution an AVM teamhas been organized comprised of the following disciplines: neurosurgery, neurology (Stroke Service), neuropsychology, interventional neuroradiology, neuroanesthesiology, and specialists in isotope imaging techniques. This application of many specialists, prolonged hospital time, and exotic procedures all drive up the cost of treating AVMs. Therefore, why treat these rare lesions? The answer is twofold: first, we have the technology to effect a cure in a relatively safe and effective fashion and second, perhaps more importantly, these conditions are most treacherous to young individuals during the most productive phase of their lives, often resulting in severe, permanent morbidity or death.

A. Congenital Lesions

Figure 1: Drawing of AVM located in the left frontoparietal motor-sensory strip. Large draining veins are marked by V, feeding arteries by A.Shunts occur between the large arteries and veins, resulting in a tangle of abnormal blood vessels carrying high flow blood which displace brain function to the margins.

Arteriovenous malformations are presumed to be congenital and not acquired. However, there is no strong evidence for hereditary or familial tendencies. It would also appear that an AVM grows with the growing brain and its developing circulation, so that the absolute size of the AVM in an infant compared with its future size is considerably smaller, whereas the ratio to brain size remains constant.

Evidence for a congenital origin is the nature of the lesion which represents an arrest in the development stage of the embryonic circulation in which direct connections between arteries and veins remain. The AVM is therefore a tangle of abnormal arteries, veins and cavernous channels. These are tightly coiled in a round or wedge shaped configuration which involves the parenchyma of the brain to varying degree. Some AVMs are in the 1 cm range while others have greater than a 5 cm diameter (Figures 2,3). In developing with the brain, the AVM appears to displace normal brain function so that the anatomical architecture of the brain accommodates the AVM. Within the confines of the AVM, there is no normal or functioning tissue. The implications of brain mapping in these patients is obvious. The lesions are not reproducible by experimental technique and in the adult brain there is no evidence that they grow in a neoplastic fashion.

B. Circulatory Physiology

Since most AVMs contain high flow artery-to-venous shunts, the opportunity for study of the brain circulation, normal and abnormal, is legion.^4,5,7 In spite of these large shunts, there is no good evidence that ischemia occurs in areas of the brain adjacent to the AVM. This is most likely the case, since the heart in the adult can compensate for the loss of circulation through these shunts and still manage to supply an adequate blood supply to the uninvolved brain. Studies are being done at Columbia University College of Physicians and Surgeons (P&S) to elucidate the dynamics of these A-V shunts and, through the use of brain mapping, to show that brain function is often displaced to regions around the margin of the AVM.^4,5 In the past, studies in P&S laboratories demonstrated that the large arteries supplying an AVM do not autoregulate.⁷

II. Presentation

With the sophistication and availability of brain imaging techniques, a significant percentage of AVMs which are asymptomatic have been uncovered. These are cases such as individuals with unrelated symptoms who have a CT scan and MRI and, lo and behold, there is an AVM that had not caused symptoms. The fact that the lesion is asymptomatic makes a judgment to institute treatment, which may carry significant risk, a difficult dilemma. On the other hand, most AVMs, at least 50% to 60%, present with a hemorrhage.^3,9-14 This hemorrhage will affect the brain surrounding the AVM since these are primarily parenchymal lesions and may produce devastating results, even death. Fortunately, in at least 75% of bleeds, the patient will survive with or without significant neurological deficit. This provides us the opportunity to treat the AVM by the technique that is best suited to prevent future bleeds. Second in incidence to hemorrhage is the onset of seizures related to the particular area of the brain involved by the AVM. In the case of the posterior fossa, seizures are not a problem and the incidence of hemorrhage of posterior fossa AVMs is proportionately higher.

III. Treatment Decisions

In considering whether or not to treat an AVM which has bled or is asymptomatic, the paramount consideration is the patient's age. Age is important because of the subsequent normal life expectancy related to that age and thereby the exposure of the individual to the ravages of an untreated AVM.^15,16 Additionally, the younger patient, while having greater potential for recovery from a bleed, is also better equipped to handle the treatment. The natural history of AVMs, although suggested, is not known with statistical proof. Most studies have been done in retrospective fashion and the age of the patient, the location, size, anatomical and physiological characteristics of the AVM vary so greatly that it has been impossible for any one group of specialists to acquire the large numbers of patients required for statistical analysis. Furthermore, another variable to consider is the fact that the treatment through interventional, microsurgical and radiosurgical techniques has made so many advances in the last decade that it is impossible to compare the outcome results of these various forms of treatment.

Figure 2A: Side by side MRI showing a medial hemisphere AVM (arrows).	Figure 2B: This AVM is shown by angiography. Note how the angiographic pattern correlates with the MRI pattern. This patient with a medial hemisphere AVM presented with a seizure secondary to a small hemorrhage. The AVM was removed successfully.

Figure 3: Lateral carotid angiogram showing the grossly enlarged feeding arteries (A) to the AVM nidus (arrows) which is drained by an enlarged vein (V).

Further consideration relates to the size, the anatomical location in the brain and the presence or absence of certain physiological and anatomical features which we can now measure. The decision then becomes a somewhat philosophical one and involves a major commitment on behalf of a treating physician and the patient to analyze the situation. The physician informs the patient as to the risks of leaving it alone and the risks of treatment, whereas the patient informs the physician of his or her concern, either of leaving it alone or undergoing treatment, which carries a small but significantly severe morbidity.

Two patients are worthy of review: one was a 60-year-old cardiologist whose AVM in the right posterior nondominant temporal lobe was picked up incidentally when he went for evaluation of a 4th nerve palsy which turned out to be unrelated. The AVM was 3 cm in diameter and located in a noneloquent area of his brain. He had never had a hemorrhage, a seizure or any other symptom from the AVM. I suggested to him that he leave it alone. It seemed to me, without benefit of statistical analysis, that he had beaten the odds of the AVM causing him a problem. He initially agreed and then came back to me and said that he wished to have the AVM removed. I asked him "Why?" He told me that being a cardiologist, he realized that when men get older, they might have to go on anticoagulation therapy for cardiovascular disease and he could not live with the idea of having this potential "bomb" in his head and be anticoagulated. Fortunately, we were able to remove the AVM without any adverse sequelae and both - the surgeon and the patient - were quite pleased with the result (Figure 4).

Figure 4: Photograph of an AVM showing the surface features (outlined by arrows) of an underlying larger AVM. The draining veins (V) are well identified whereas the arteries remain obscure deep in the sulci.	Figure 5: Carotid angiogram (AP view) showing an AVM (outlined by arrows), deep to the dominant frontal operculum reached and resected by stereotactic approach. This 2 cm AVM had presented with a seizure.

Figure 6: Lateral posterior fossa vertebral angiogram showing an AVM of the cerebellum outlined by small arrows and a proximal arterial aneurysm on a feeding artery (long arrow). This patient presented with a cerebellar hemorrhage from the aneurysm.

Figure 7: Selective lateral vertebral angiogram identifying the posterior choroidal artery supply to an intraventricular AVM. The AVM (outlined by arrows) contains a venous aneurysm (V). This AVM had presented with an intraventricular hemorrhage and was identified by selective angiography. It was resected by an interhemispheric approach as shown in Figure 11.

At the opposite end of the spectrum is a 19-year-old person whose AVM was picked up following routine scan for head trauma. The 2 cm AVM in this case is located in the dominant hemisphere, motor-speech area and deep to the surface (Figure 5). Its anatomical features suggested that it would be a treacherous lesion in this patient's lifetime exposure, presumed to be approximately 65 years. Given the known natural history of AVMs, unlike the physician who had an excellent track record, this young man's record is unknown and we would strongly advocate an aggressive intervention. Discussions with the patient and the family were lengthy and numerous, so that the patient and the family fully understood the importance of the disease, either treated or untreated, and the various modes of treatment. I spend an inordinate amount of time with my patients regarding these matters. Since these are unusual lesions, most of the patients have little comprehension of the implications and risks.

Through the dedicated studies of our AVM team at P&S, we have uncovered certain important anatomical and physiological criteria which may assist in projecting the risk of any given AVM.^4,5 At the time of interventional procedures, with selective arteriography, we are able to do brain mapping and also to develop an anatomical footprint of these AVMs, being able to visualize deep arteries and veins, as well as other important anatomical features; especially the presence of an adjacent aneurysm (Figures 6, 7). Pressures are recorded in the various vessels and cerebral blood flow studies accomplished. Preliminary evaluation of these pioneering studies at CPMC suggest that we now have additional information about an AVM to help predict its future course. The features that appear to increase the natural risk are the presence of deep draining veins, small size, increased pressure in the feeding arteries and the location, either deep, periventricular or in an eloquent area of the brain, such as the dominant motor-speech area. Brain mapping allows us to talk with confidence to the patient and family about what functions are at the border of the AVM so that we can discuss preventing injury to these areas.

IV. Workup

The history and physical are of lesser importance in considering various forms of treatment. It is quite obvious, from what we have said, that the asymptomatic patient, without neurological deficit and with a treacherous AVM is an ideal one to treat and cure, since this is prophylactic treatment, preventing what is most likely to be a devastating future. CT scans, with and without contrast, are the most common tests that have been widely utilized to screen patients. However, to have a detailed evaluation of an AVM, the CT scan is usually the least important. An MRI and MRA (Figure 2) give a broad brush indication as to many of the features of the AVM, plus its location and relation to other areas of the brain, whereas the angiogram is still the gold standard in determining many of the anatomic details and is absolutely necessary for decision making (Figures 2, 3).

Angiography has been extended to include interventional techniques which also embrace selective arteriography where fine catheters can be manipulated into the branches of the major intracranial arteries to selectively identify arteries related to the AVM. Dr. Pile-Spellman, as part of the AVM team, has been carrying out flow studies, psychological evaluation, brain mapping and pressure studies of the involved blood vessels. With the microcatheter techniques, selective studies of the surrounding brain can be done through injection of amytal or xylocaine to "numb" that part of the brain and study its function by psychological testing at the time of the procedure. Additionally, SPECT scanning, normal or activated, has been utilized to further analyze blood flow and its relation to brain activity. This has been of enormous value in selecting a subsequent curative treatment plan.

Figure 8: Chart listing the names of physicians in various centers. The only surgical series is "Stein". The other authors represent radiosurgery centers, using various radiosurgical techniques. Note that the extent of residual AVM following treatment has been much higher in the radiosurgery group, as opposed to microsurgery. Also note, the complication rate or neurological deficit is generally higher with radiosurgery than with microsurgery.

Figure 9: Chart comparing the onset of neurological deficits, the preoperative status of deficits with the postoperative status after convalescence in a series of 67 surgically resected small AVMs.

V. Treatment

Generically there are only two forms of treatment for an AVM. These are: (1) Definitive or curative and (2) Adjunctive or modifying. In some instances, no treatment whatsoever may be the best course.

Only surgery or radiosurgery are definitive treatments for an AVM and that can lead to a cure. Adjunctive therapies are primarily embolic therapies carried out by an interventional neuroradiologist. These are rarely curative. There have been no other treatments found of value, such as long-term hypotensive therapy, hemoconcentration, and carotid compression or ligation. Seizures can be managed by anticonvulsants, however, this does nothing to modify or treat the AVM.

Both surgery and radiosurgery depend on the size of the AVM, so that the smaller AVMs, under 3 cm, show the best results from microsurgery as well as radiosurgery. Unfortunately, the radiosurgical cure rate, even with the small lesions, falls in the range of 80% whereas microsurgical cure rate falls in the range of 96%, with cure defined as total obliteration (Figure 8). The difference is in the risk of the procedure. The risk, although small, is immediate for surgery (Figure 9). For radiosurgery, there is no immediate risk. However, years later, radiation necrosis, and/or obliteration of major vessels supplying normal brain in the vicinity of the AVM, may ensue. These changes may occur so gradually and so slowly that the brain can accommodate. In other patients, there may be progressive neurological deficits as a result of radiation damage.

A. Surgery

The surgery for the removal of an arteriovenous malformation is a carefully planned procedure.^3,15-17 The surgeon must be aware of all of the anatomical details of the AVM and is abetted in this task by the interventionalist who has reduced major shunts by embolization. The operation microscope must be utilized in these procedures as the detail of the supplying vessels and relationship to the AVM may be semi-microscopic. The patient is prepared in the operating room, the head held rigidly, and spinal drainage instituted to relax the brain so that "working room" may be obtained around the margins of the AVM. The craniotomy is then effected and, if all plans have gone according to everyone's concept of the localization of the AVM, the craniotomy will result in the AVM being in the center of the field surrounded by normal brain so that anatomical features, especially draining veins, may be seen for landmarks (Figure 4). In many cases, the AVM may be subsurface or only a tip of it may show, much like an iceberg. Using anatomical references from the angiograms, in some cases a stereoscopic pair in the operating room, the surgeon may then make a cortical incision around the AVM, retracting the normal brain (Figure 10). Care must be taken not to damage the normal surrounding brain, therefore relaxation is critical in terms of working under close tolerances. The draining veins are left intact, while the feeding arteries are interrupted to reduce the pressure within the AVM. The most troublesome area is the deep portion where small thin-walled arteries which do not autoregulate traverse the white matter carrying high blood flow. This is where hypotension provided by the neuroanesthesiologist is of great value. Since these vessels do not appear to autoregulate, hypotension has a direct effect on their blood flow and will produce a marked decrease in the bleeding from these blood vessels. Many of the AVMs go into the ventricle and are supplied by the arteries of the choroid plexus (Figure 7). All these must be secured and the surgeon satisfied that the entire AVM is removed. A postoperative angiogram to confirm this fact is usually done.

Figure 10A: Drawing showing the surface exposure of an AVM with meticulous attention to interrupting the arterial supply by cautery and scissors prior to resecting the AVM. Figure 10B: Drawing of a coronal section of the brain showing the interruption of deep, high flow thin wall arteries at the margin of the ventricle. This is the most difficult portion of an AVM resection. Figure 10C: Only at the final moment may the major draining vein be ligated in the total removal of the AVM.

Most AVMs, especially the smaller ones, are accessible to the surgeon as long as they present to the surface of the brain - either the convexity or medial surface, the ventricular or the brain stem or fourth ventricle surfaces^{6,15,16,18-20} (Figures 11,12). Those small AVMs which are located deep to the surface of the brain and which do not front on any of the aforementioned surfaces can be reached through small transcortical incisions guided by a stereotactic probe aimed at the AVM as seen on a scan or arteriogram²¹ (Figure 5). Stereotaxis provides us a three-dimensional localizer for the deep AVM and the cortical incision is kept to the minimum, thereby avoiding, even in eloquent areas, significant neurological deficits.

It is the large AVM that presents the utmost difficulty for the surgeon and is generally inappropriate for radiosurgery, even though it may be reduced in a patchy fashion by embolization. Embolization, on the other hand, with surgery, has been a significant adjuvant, since it decreases the flow and makes the management of these AVMs easier during the operation.^15,16 Notwithstanding, even large AVMs located adjacent to eloquent areas of the brain can be removed with reasonable safety.

No operation is guaranteed and devastation may occur from postoperative hemorrhage, intraoperative hemorrhage, or the interruption of circulation to viable areas of the brain. Once an operation on an AVM is started, it rarely is abandoned part way through. The surgeon is committed to a final resection of the AVM.

Figure 11: Drawing showing six alternative routes to AVMs located on the medial aspect of the cerebral hemisphere; most involving portions of the limbic region. These approaches have been utilized to totally resect these deep seated small AVMs.	Figure 12: Drawing showing the various approaches to AVMs located in the cerebral hemispheres, diencephalic and brain stem regions. These approaches require maximal relaxation in order to retract large portions of the cerebral hemisphere or cerebellum. The approach "B" is subtentorial over the cerebellum.

B. Radiosurgery

Radiosurgery involves the focus of a number of high-energy radiation beams onto the area of the AVM, creating endarteritis obliterans and eventual blockage of the AVM and sometimes cure. There are various methods to provide focused radiation to brain lesions. The AVM has been the most popular lesion to be treated by the radiotherapists with this new technique. The systems that are well known are the Proton Beam, the Gamma Knife® and the Linear Accelerator, modified to provide multi-beam focus on the target.^6,22,23 At CPMC, radiosurgery is provided by Dr. Isaacson of the Radiation Oncology Department and Dr. Michael Sisti of the Neurosurgery Department. In most instances, the AVMs which they treat are deemed to be inoperable. This means inaccessible, buried in the brain, not to be reached by stereotactic or other techniques.

Not reported in detail by the advocates of radiosurgery are complications which we have witnessed.²³ We have seen a young man of 25 years who had a dominant hemisphere, 3 cm AVM involving the middle cerebral artery (Figure 13). A Gamma Knife®was used in the treatment of this lesion. After a period of one year, an angiogram showed that the lesion was reduced but was not obliterated. After a period of 3 years, the lesion was obliterated and the middle cerebral artery and its small branches were intact. However, MRI showed diffuse white matter changes 3 inches away from the target of the radiation therapy and the patient was developing right-sided weakness and difficulty of speech. He was recently seen at 7 years follow up. White matter changes are less, there is significant radiation necrosis with cavitation in the region of the AVM, and now the middle cerebral artery is not visualized, nor are its branches (Figure 14). A SPECT scan shows marked decrease in flow to the entire middle cerebral-supplied left hemisphere. Amazingly, probably because the process has been slow, the patient is improved, in terms of right-sided weakness and his speech problem, but he still has occasional temporal lobe seizures. Anticipating that he will have problems as he grows older, we are considering an augmentation procedure for his cerebral circulation. This demonstrates the effectiveness and low risk of radiosurgery during the initial phase, but the horrendous complications of radiation that can occur later.

Figure 13A: AP left carotid angiogram showing a 3 cm AVM (arrows) intimate to the middle cerebral artery (MCA).	Figure 13B: AP left carotid angiogram, three years following Gamma Knife® radiosurgery, shows total obliteration of the AVM with preservation of the middle cerebral, lenticulostriate and choroidal arteries.
Figure 14A: MRI showing significant white and gray matter radionecrosis (arrows) seven years following Gamma Knife® radiosurgery with successful obliteration of AVM shown in Figure 13B.	Figure 14B: MRA showing previously patent middle cerebral artery and its branches (Figure 13B), now completely occluded (at arrow) with a circulatory deficit, which is relatively asymptomatic, to the left hemisphere.

Figure 15: Drawing showing how important arteries "en passage" are gradually occluded by embolic material (dark spheres) during staged embolization while collateralization (c) to the distal watershed arterial perfusion territory is encouraged and eventually takes over the function of the occluded major branches of the middle cerebral. This is most likely a situation similar to what has occurred as a result of Gamma Knife® therapy demonstrated by the patient in Figures 13 and 14.

It goes without saying that an AVM partially treated by whatever means, embolization, microsurgery, radiosurgery is not a cured AVM. In fact, it may be more dangerous than the original, considering the anatomical features which we have addressed early in the paper, namely that the small AVMs seem to be more treacherous than the large ones.

C. Embolization

Interventional neuroradiologists using techniques of embolization have been of great value in modifying AVMs prior to definitive treatment.^4,5,24 At CPMC Dr. Pile-Spellman uses an acrylate glue which polymerizes at a known rate to plug some of the major arteries supplying an AVM, simplifying the surgery or other subsequent treatment. The intravascular blockage of arteries can also be used to block arteries which are important to surrounding cortex and pass right through the AVM (Figure 15). We know at surgery that it is impossible to dissect these arteries out and, therefore, if one is occluded by emboli and the patient made artificially hypertensive, it is quite possible to preserve and augment the surrounding circulation so that by the time surgery is carried out, this artery has become useless and may be resected with the AVM. Additionally, deep feeders which contain aneurysms at some distance from the AVM may also be blocked prior to operative intervention, facilitating the operation.

VI. Conclusion

Although AVMs of the brain are relatively uncommon in the spectrum of neurological disorders, they do present an interesting substrate with the opportunity to study cerebrovascular physiology, normal and abnormal, as well as brain function as related to the abnormal circulation of an AVM. We have gathered a team at P&S that addresses virtually all of the clinical and research aspects as related to AVMs and have provided additional data to the field. The treatment techniques have developed to a high degree of sophistication and there is nothing to suggest that in the future there will not be further amazing advances in both the evaluation and therapeutic techniques that are applied to these lesions. As a result, it has been possible to treat many more patients with AVMs in an intelligent manner and protect them from the devastating effects of the natural history during the most productive period of their life. Still, further studies are needed to determine if large AVMs are better treated or left alone; or will there be newer techniques developed so that they may be treated with the same degree of efficiency, cure rate and minimal morbidity and mortality that so far has been present for the small AVMs? In our series treating large and small AVMs, the mortality has been 1.5%. For the small AVMs the cure rate has been in the range of 96%.

We must not lose sight of the implications of the disease as they pertain to a particular patient. These are horrifying problems for the patient. The patients don't understand, especially those who present in an asymptomatic fashion. Those that have had a hemorrhage can attest to the devastation of these lesions. However, even in these circumstances, there is reluctance to undergo treatment that may recreate devastation. The patient and the family must by guided through the decision. It is very difficult to put one's own interest in the background and substitute the patient's interest. However, this is a path that is assigned to the specialist dealing with these disorders and it must not be abdicated.

Note: All drawings are the product of Robert J. Demarest, former head of Audiovisual Department, P&S, except Figure 1, which is the work of Robert Ullrich.

References:

1. McCormick WF, Rosenfield DB: Massive brain hemorrhage: A review of 144 cases and an examination of their causes. Stroke 1973;4:946-954.

2. Perret G, Nishioka H: Report on the cooperative study of intracranial aneurysms and subarachnoid hemorrhage. Section VI. Arteriovenous malformations. J Neurosurg 1966;25:467-490.

3. Wilson C, Stein BM. Intracranial Arteriovenous Malformations in Current Neurosurgical Practice. Baltimore:Williams & Wilkins, 1984:1-342.

4. Young W, Kader A, Ornstein E, et al.: The Columbia University AVM Project: Cerebral hyperemia after AVM resection is related to "breakthrough" complications but not to feeding artery pressure. Neurosurgery 1996;38:1085-1095.

5. Young WL, Prohovnik I, Ornstein E, et al.: The effect of arteriovenous malformation resection on cerebrovascular

reactivity to carbon dioxide. Neurosurgery 1990;27:257-267.

6. Sisti M, Kader A, Stein B: Microsurgery for 67 intracranial arteriovenous malformations less than 3 cm in diameter. J Neurosurg 1993;79:653-660.

7. Spetzler RF, Wilson CB, Weinstein P, et al.: Normal perfusion pressure breakthrough theory. Clin Neurosurg 1978;25:651-672.

8. Muraszko K, Wang HH, Pelton G, Stein BM: A study of the reactivity of feeding vessels to arteriovenous malformations: correlation with clinical outcome. Neurosurgery 1990;26:190-200.

9. Fults D, Kelly DL: The natural history of arteriovenous malformations of the brain: A clinical study. Neurosurgery 1984;15:658-662.

10. Graf CJ, Perret G, Torner JC: Bleeding from cerebral arteriovenous malformations as part of their natural history. J Neurosurg 1983;58:331-337.

11. Ondra SL, Troupp H, George E, Schwav K: The natural history of symptomatic arteriovenous malformations of

the brain: a 24-year follow-up assessment. J Neurosurg 1990;73:387-391.

12. Pool JL, Potts DG. Aneurysms and Arteriovenous Anomalies of the Brain. Diagnosis and Treatment. New York:Harper and Row, 1965: 463.

13. Troupp H, Marttila I, Halonen V: Arteriovenous malformations of the brain: Prognosis without operation. Acta Neurochir 1970;22:125-128.

14. Wilkins RH: Natural history of intracranial vascular malformations: A review. Neurosurgery 1985;16:421-430.

15. Stein BM, Wolpert SM: Arteriovenous malformations of the brain I: Current concepts and treatment. Arch Neurol 1980;37:1-5.

16. Stein BM, Wolpert SM: Arteriovenous malformations of the brain II: Current concepts and treatment. Arch Neurol 1980;37:69-75.

17. Drake CG: Cerebral arteriovenous malformations: Considerations for an experience with surgical treatment in 166

cases. Clin Neurosurg 1979;26:145-208.

18. Solomon RA, Stein BM: Interhemispheric approach for the surgical removal of thalamocaudate arteriovenous malformations. J Neurosurg 1987;66:345-351.

19. Solomon RA, Stein BM: Management of arteriovenous malformations of the brain stem. J Neurosurg 1986;64:857-864.

20. Solomon RA, Stein BM: Surgical treatment of arteriovenous malformations that follow the tentorial ring. Neurosurgery 1986;18:708-715.

21. Sisti MB, Solomon RA, Stein BM: Stereotactic craniotomy in the resection of small arteriovenous malformations.

J Neurosurg 1991;75:40-44.

22. Heros RC, Korosue K: Radiation treatment of cerebral arteriovenous malformations. N Engl J Med 1990;323:127-129.

23. Lunsford LD, Kondziolka D, Flickinger JC, et al.: Stereotactic radiosurgery for arteriovenous malformations of the brain. J Neurosurg 1991;75:512-524.

24. Vinuela FV, Debrun GM, Fox AJ, Kan S: Detachable calibrated-leak balloon for superselective angiography and embolization of dural arteriovenous malformations. J Neurosurg 1983;58:817-823.