ELSEVIER

GAMMA KNIFE RADIOSURGERY FOR CEREBRAL ARTERIOVENOUS MALFORMATIONS: AN AUTOPSY REPORT FOCUSING ON IRRADIATION-INDUCED CHANGES OBSERVED IN NIDUS- UNRELATED ARTERIES Masaaki Yamamoto, M.D.; Minoru Jimbo, M.D.; Mitsunobu Ide, M.D.; Makio Kobayashi, M.D.; Chisato Toyoda, M.D.; Christer Lindquist, M.D., Ph.D.; and Bengt Karlson, M.D. Department of Neurosurgery (MY, MJ, MI), Tokyo Women’s Medical College Dai-ni Hospital, Tokyo, Japan; Department of Pathology (MK, CT), Tokyo Women’s Medical College, Tokyo, Japan; and Department of Neurosurgery (CL, BK), Karolinska Hospital, Stockholm, Sweden

Yamamoto M, Jimbo M, Ide M, Kobayashi M, Toyoda C, Lindquist C, Karlson B. Gamma knife radiosurgery for cerebral arterio- venous malformations: an autopsy report focusing on irradia- tion-induced changes observed in nidus-unrelated arteries. Surg Neurol 1995;44:421-7.

BACKGROUND In radiosurgical treatment for an arteriovenous malfor- mation (AVM), the effects of irradiation on the intranidal and perinidal angioarchitectures have seldom been ana- lyzed histologically. An autopsy case is reported, study- ing an AVM treated by gamma knife radiosurgery. Post- mortem studies following AVM-unrelated death were performed after a 2-year angiography had demonstrated complete nidus obliteration. Irradiation-induced changes were also observed in surrounding nidus-unrelated arter- ies and the choroid plexus, both of which were within the irradiation target. METHODS Microscopic studies were performed using a coronal sec- tion of the brain including the center of the AVM, on which the percent isodose volume gradient, corrected with a magnification rate, was superimposed. RESULTS

This study disclosed that intimal hypertrophy can occur in a normal, AVM-unrelated pial artery due to irradiation of 10 Gy or more and that more remarkable intimal hy- pertrophy with fragmentation of the elastic laminae, or even complete occlusion, can occur in these arteries with 25 Gy. Similarly, irradiation-induced degeneration was

Address reprint requests to: Masaaki Yamamoto, M.D., Department of Neurosurgery, Tokyo Women’s Medical College Dai-ni Hospital, Z-1-10 Nishiogu, Arakawa-ku, Tokyo, Japan.

Received February 27, 1995; accepted May 26, 1995.

0 1995 by Elsevier Science Inc. 655 Avenue of the Americas, New York, NY 10010

present in the choroid plexus, which had been exposed to doses varying from 10 Cy to 25 Gy. CONCLUSIONS

A normal surrounding blood vessel may also be affected by high-dose, single-fraction irradiation though the abnor- mal vessels have been reported to be more susceptible.

KEY WORDS

Arteriooenous malformation, autopsy, choroid plexus, gamma knife, pial artery, radiosurgery.

I n a previous paper [24], we presented a brief description of an autopsy case with an arterio-

venous malformation (AVM) treated by gamma knife radiosurgery, in which postmortem studies following an AVM-unrelated death were performed after a 2-year angiography demonstrated complete nidus obliteration. This report histologically dem- onstrated that, although many of the AVM vessels had obstructed lumina owing to intimal hypertro- phy, there were still patent vascular channels in the nidus (persistent and recanalizing vessels), which could not be observed angiographically. The key question since has been whether these vessels with patent lumina have the potential to rebleed. Herein, we present this case again in greater detail focusing on this issue, as well as irradiation-induced and dose-related changes observed in the arteries sur- rounding the nidus and within the choroid plexus. As Guo et al pointed out [9], irradiation by gamma

00903019/95/$15.00 SSDI 0090-3019(95)00260-B

422 Surg Neurol 1995;44:421-7

Yamamoto et al

q Percentage isodose vol- ume gradient superim-

posed upon a coronal section of the brain including the center of the nidus (elastica hematoxylin and eosin, original magnification x 1). Arrows and arrow heads indicate the vessels to be shown at higher magnification in Fig- ures 3 and 4.

knife affects not only the target vessels of the nidus but also the immediate surrounding or intervening brain tissue and its nutrient arteries. Several animal experiments have revealed radiosurgically induced changes observed in a normal artery within a target volume [4,15,18]. However, relatively high doses were used in these experiments: 150 Gy (100% iso- dose) or 125 Gy (84% isodose). Furthermore, these experimental animals were sacrificed for histologic study within 6-63 postradiosurgical weeks. There- fore, as yet, neither pathologic studies nor animal experiments have disclosed long-term changes in the normal arteries irradiated with a relatively low

dose, such as lo-25 Gy; these doses are often se- lected for the treatment of an AVM or a brain tumor. This case disclosed that intimal hypertrophy and occlusion can occur in a normal, AVM-unrelated pial artery due to an irradiation dose of 10 Gy or more.

CASEREPORT A 42-year-old woman was admitted to a local clinic on May 15, 1987, following the abrupt onset of se- vere headache. Neurologic examination on admis-

q The majority of remaining vessels within the obliter-

ated nidus were found to be thin-walled vessels containing elastic fibers (elastica van Gie- son, original magnification X 33).

Histology of Radiosurgery-Obliterated AVM Surg Neurol 423 1995;44:421-7

Pial arteries (arrows in Fig- ure 1) considered to have

been exposed to an irradiation dose of about 25 Gy (A). Signifi- cant intimal hypertrophy and marked fragmentation of the elastic laminae were observed in the larger-sized vessels. In the smaller-sized vessels, the lu- mina was completely ob- structed by intimal hypertrophy and no elastic laminae were dis- cernible. Recanalizing vessels were observed within the ob- structed vessel (B, arrow in Fig- ure 3 A) (elastica van Gieson, original magnification A, x 10; B, x 100).

sion was within normal limits. Computed tomogra- phy (CT) scan showed a hematoma in the left temporal lobe. Angiography demonstrated an AVM in the lower part of the left temporal lobe. Although surgical resection was strongly recommended, she refused operative intervention. On May 29, she was referred to our clinic to evaluate whether or not the AVM could be treated with a gamma knife. Repeat angiography disclosed that, although the AVM was fed by the vertebrobasilar and internal carotid ar- tery systems, the maximum nidus diameter was less than 3.0 cm. Thus, gamma knife radiosurgery was carried out at the Karolinska Hospital, Stockholm, on August 2, 1988. The nidus was covered with a

50% isodose volume using two target points with l&mm and 14-mm collimators. A central dose of 50 Gy was used to obtain a marginal dose of 25 Gy. Angiography showed nearly complete obliteration of the nidus 1 year after irradiation, and complete obliteration was confirmed by reexamination car- ried out on October 18, 1990.

On the day after angiography, the patient suffered angor pectoris of sudden onset when she began to walk after the 24 hours’ rest required following ex- amination. Cardiopulmonary arrest occurred within a few minutes. There was no response to resusci- tation procedures. Autopsy revealed the course of death to be pulmonary arterial embolism.

424 Surg Neurol 1995;44:421-7

Yamamoto et al

0 , Pial arteries (arrow heads in Figure 1) considered to

have been exposed to about 10 Gy of irradiation. Intimal hyper- trophy and fragmentation of the elastic laminae were observed. However, these changes were less marked than those ob- served in Figure 3 (elastica van Gieson, original magnification X 5).

HISTOLOGIC FINDINGS IN RELATION TO IRRADIATION DOSES Figure 1 shows the percentage isodose volume gra- dient superimposed upon a coronal section of the brain including the center of the AVM. The percent- age isodose volume gradient was corrected with a magnification rate determined by comparing a coro- nal magnetic resonance imaging section with the corresponding brain section. As we have already reported [24], many meandering blood vessels within the nidus had obstructed lumina due to in- timal hypertrophy. However, some of the vessels had patent lumina: remaining patent vessels and recanalizing vessels. Only these patent blood ves- sels had endothelial cells, which were shown immu- nohistochemically to be positive for factor VIII. The majority of remaining vessels were thin walled and contained elastic fibers, and were thus presumed to be veins (Figure 2).

Figure 3 shows pial arteries, which were consid- ered to have been exposed to an irradiation dose of about 25 Gy. Significant intimal hypertrophy and marked fragmentation of the elastic laminae were observed in the larger-sized vessels (Figure 3 A). In the smaller-sized vessels, the lumina were com- pletely obstructed by intimal hypertrophy, and no elastic laminae were discernible. Recanalizing ves- sels were observed within the obstructed vessel (Figure 3 B). Figure 4 also shows pial arteries, con- sidered to have been exposed to an irradiation dose of about 10 Gy. Intimal hypertrophy and fragmen- tation of the elastic laminae were observed. How- ever, these changes were less marked than those observed in Figure 3. Neither intimal hypertrophy

nor fragmentation of the elastic laminae was ob- served in vessels outside the 5 Gy isodose gradient, or in any vessels of the whole brain so far as we could investigate.

Irradiation-induced changes in the choroid plexus are shown in Figure 5. The most lateral part was assumed to have been exposed to irradiation of 25 Gy or more while the most medial part was considered to have been exposed to irradiation of about 10 Gy. Loss of blood vessels in the swollen stroma as well as epithelial cell loss were observed, most prominently in the lateral part, and to a lesser degree, in the medial part.

DISCUSSION It is believed that, since Steiner et al reported the first patient with an AVM treated by gamma knife radiosurgery [21], more than 10,000 AVM patients worldwide have been radiosurgically treated using a gamma knife, a linear accelerator system, or a system incorporating Bragg peak particles [3,5-8, lo-14,16,19,20,22]. Overall, approximately 80% of patients show complete angiographic obliteration within a latency period of 2 to 3 years, at which point the risk of subsequent hemorrhage is elimi- nated. Because of its benign nature and bias toward the adolescent and young adult populations, post- mortem examination of the obliterated AVM has been performed in only a limited number of patients who died due to an AVM-unrelated cause [24], de- spite a relatively large number of patients having been treated. Pathologic studies of the treated ni-

Histology of Radiosurgery-Obliterated AVM Surg Neurol 425 1995;44:421-7

Ia Irradiation-induced changes in the choroid plexus. The

most lateral part (A) was con- sidered to have been exposed to 25 Gy or more and the most medial part (IS) was considered to have been exposed to about IO Gy. Note: Loss of blood ves- sels in the swollen stroma and epithelial cell loss were ob- served, most prominently in the lateral part and, to a lesser ex- tent, in the medial part (elastica van Gieson, original magnifica- tion x 16).

dus have occasionally been performed using solely a surgical specimen or performed in patients who died due to massive intracranial hemorrhage before the AVMs were confirmed obliterated by angiogra- phy [201.

Since we first reported an autopsy case with a radiosurgically obliterated AVM and demonstrated that there were a number of patent blood vessels within the treated nidus that could not be observed angiographically [24], debate has continued as to whether these patent vessels have the potential to rebleed. As reported herein, the majority of remain- ing vessels were veins. Almost all arteries that had composed the AVM were completely occluded. Al-

though there were many recanalizing vessels within the occluded arteries, these vessels were relatively small and had thick walls composed primarily of collagen fiber. Therefore, such vessels with patent lumina, recanalizing or persistent vessels, could not be considered to have the potential to rupture. Kawamoto concurs with this opinion based on the autopsy findings of a case with an essentially iden- tical condition (Kawamoto S, Department of Neuro- surgery, University of Tokyo: personal communica- tion, 1993).

Another important issue disclosed by this au- topsy study is that obliteration initially occurs in arteries, rather than veins. As shown in Figure 2, no

426 Surg Neurol 1995;44:421-7

irradiation-induced changes were observed within the walls of veins near a completely occluded ar- tery. It is favorable for AVM treatment that occlu- sion occurs initially in the arteries of the AVM, and thereby reduces the risk of rebleeding. This obser- vation is consistent with the fact that the incidence of postradiosurgical rebleeding during the latency period is not higher than that of rebleeding during the natural course of an untreated AVM [2,22,23]. Furthermore, radiosurgery is more advantageous than embolization, in which it is impossible to oc- clude the arteries selectively and in which, there- fore, the risk of rebleeding during, or immediately after, the procedure cannot be ignored. Rebleeding in a residual nidus after embolization has been re- ported in 7% to 11% of cases [1,17].

We studied irradiation-induced changes ob- served in AVM-unrelated pial arteries and the cho- roid plexus in relation to irradiation doses. This study cannot be regarded as highly scientific due to the absence of reference points within the brain section corresponding to the matrix center or the frame center given on the percentage isodose vol- ume gradient. Furthermore, the nidus itself had ap- parently shrunk over the 2 years following radio- surgery. Nevertheless, this study disclosed that intimal hypertrophy can occur in a normal, AVM- unrelated pial artery due to an irradiation dose of 10 Gy or more and that more marked intimal hyper- trophy with fragmentation of the elastic laminae, or even complete occlusion, can occur in these arter- ies following exposure to 25 Gy. Similarly, irradia- tion-induced degeneration was observed in areas of the choroid plexus exposed to doses varying from 10 Gy to 25 Gy. As recently reported by several authors, animal experiments have revealed that ra- diosurgery can also induce intimal hypertrophy and/or obliteration within a normal artery [4,15,18]. However, the doses used in these experiments were relatively high and all materials were used for his- tologic studies within a short postradiosurgical pe- riod (maximum, 63 weeks). Future animal experi- ments should be designed to analyze long-term changes induced by relatively low-dose irradiation.

The authors would like to thank Bierta E. Barfod, M.D., University of Washington School of Medicine, for her assis- tance in the preparation of this manuscript.

This study was supported by scholarships fmm the Royal Swedish Academy of Sciences, the Japanese Society for the Promotion of Science, the Japan Private School Promotion Foundation, and the Dr. hoe Okamoto International Ex- change Fund.

Yamamoto et al

REFERENCES 1.

2.

3.

4.

5.

6.

7.

Andrew BT, Wilson CB. Staged treatment of AVM’s of the brain. Neurosurgery 1987;21:314-23. Barrow DL, Reisner A. Natural history of intracranial aneurysms and vascular malformations. Clin Neuro- surg 1993;40:3-39. Betti 00, Munari C, Rosler R. Stereotactic radiosur- gery with the linear accelerator: treatment of arterio- venous malformations. Neurosurgery 1989;24:311-21. Blatt DR, Friedman WA, Bova FJ, Theele DP, Mickle JP. Temporal characteristics of radiosurgical lesions in an animal model. J Neurosurg 1994;80:1046-55. Colombo F, Benedetti A, Pozza F, Marchetti C, Chierego G. Linear accelerator radiosurgery of cere- bral arteriovenous malformations. Neurosurgery 1989;24:833-40. Fabrikant JI, Frankel KA, Phillips MH, Levy RP. Ster- eotactic heavy charged-particle Bragg peak radiosur- gery for intracranial arteriovenous malformations. In: Edwards MSB, Hoffman HJ, eds. Cerebral vascular diseases in children and adolescents. Baltimore: Wil- liams & Wilkins, 1989:389-409. Fabrikant JI, Lyman JT, Hosobuchi Y. Stereotactic heavy-ion Bragg peak radiosurgery for intracranial vascular disorders: method for treatment of deep arteriovenous malformations. Br J Radio1 1984;57: 479-90.

8.

9.

10.

11.

12.

13.

14.

15.

Friedman WA, Bova FJ. Linear accelerator radiosur- gery for arteriovenous malformations. J Neurosurg 1992;77:832-41. Guo WY. Radiological aspects of gamma knife radio- surgery for arteriovenous malformations and other non-tumoural disorders of the brain. Stockholm: Karolinska Institute (Thesis), 1993:734. Jimbo M, Kitamura K, Steiner L. Radiotherapy to ce rebral arteriovenous malformation. Neurosurgeons (Tokyo) 1987;6:254-7 (in Japanese, English abstract). Kawamoto S. Radiosurgery of arteriovenous malfor- mations. No Shinkei Geka (Neural Surg, Tokyo) 1994; 22:403-20 (in Japanese, English abstract). Kjellberg RN, Hanamura T, Davis KR, Lyons SL, Adams RD. Bragg-peak proton-beam therapy for arte- riovenous malformations of the brain. N Engl J Med 1983;309:269-74. Lindquist C, Steiner L. Stereotactic radiosurgical treatment of arteriovenous malformations. In: Lunsford LD, ed. Modern stereotactic neurosurgery. Boston: Nijhoff, 1988:491-505. Loefiler JS, Rossitch E Jr, Siddon R, Moore MR, Rock- off MA, Alexander E III. Role of stereotactic radiosur- gery with a linear accelerator in treatment of intra- cranial arteriovenous malformations and tumors in children. Pediatrics 1990;85:774-84. Lunsford LD, Altschuler EM, Flickinger JC, Wu A, Mar- tinez AJ. In vivo biological effects of stereotactic radiosurgery: a primate model. Neurosurg 1990;27: 373-82.

16.

17.

Lunsford LD, Kondziolka D, Flickinger JC, Bissonette DJ, Jungreis CA, Maitz AH, Horton JA, Coffey RJ. Ste- reotactic radiosurgery for arteriovenous malforma- tions of the brain. J Neurosurg 1991;75:512-24. Pevsner PH, George ED, Doppman JL. Interventional radiology polymer update: acrylic. Neurosurgery 1982;10:314-6.

18. Spiegelmann R, Friedman WA, Bova FJ, Theele DP,

Histology of Radiosurgery-Obliterated AVM

19.

20.

21.

22.

23.

24.

Mickle JP. LINAC radiosurgery: an animal model. J Neurosurg 1993;78:638-44. Steinberg GK, Fabrikant JI, Marks MP, Levy RP, Frankel KA, Phillips MH, Sher LM, Silverberg GD. Stereotactic heavycharged particle Bragg peak radi- ation for intracranial arteriovenous malformations. N Engl J Med 1990;323:96-101. Steiner L. Treatment of arteriovenous malformation by radiosurgery. In: Wilson CB, Stein BM, eds. Intra- cranial arteriovenous malformations. Baltimore: Wil- liams & Wilkins, 1984:295-313. Steiner L, Leksell L, Greitz T, Forster DM, Backlund EO. Stereotaxic radiosurgery for cerebral arterio- venous malformations: report of a case. Acta Chir Stand 1972;138:459-64. Steiner L, Lindquist C, Adler JR, Torner JC, Alves W, Steiner M. Clinical outcome of radiosurgery for cere- bral arteriovenous malformations. J Neurosurg 1992; 77:1-8. Toya S, Kawase T. Natural history of cerebral arterio- venous malformations: summarized data of 10 pre- sentations. Surgery for Cerebral Stroke (Tokyo) 1990; l&477-81 (in Japanese, English abstract). Yamamoto M, Jimbo M, Kobayashi M, Toyoda C, Ide M, Tanaka N, Lindquist C, Steiner L. Long-term results of radiosurgery for arteriovenous malformation: neu- rodiagnostic imaging and histological studies of an- giographically-confirmed nidus obliteration. Surg Neurol 1992;37:219-30.

COMMENTARY Yamamoto et al have amplified their original obser- vations in a patient who had successful angio- graphic obliteration of an arteriovenous malforma- tion (AVM) of the temporal lobe. The effects of radiosurgery on the intranidal angioarchitecture were amplified in the first report of this case, and additional studies have been performed of the sur- rounding brain cerebrovasculature. These studies indicate that while the abnormal vessels of an AVM are perhaps more susceptible to the effects of a high-dose single-fraction ionizing radiation, normal surrounding blood vessels may also be affected as well. The original case report stirred considerable anxiety among both friends and foes of radiosurgi- cal management of AVMs. Both were concerned that an angiographically obliterated AVM may still have patent blood vessels if analyzed by his- topathologic criteria. The fact that no patient with an angiographically obliterated AVM (when con- firmed by suitable high-resolution magnification subtraction angiography) has yet to sustain a brain

Surg Neurol 427 1995;44:42 l-7

hemorrhage provides considerable reassurance. Yamamoto and his colleagues have added this in- formation to a surprisingly scant literature that de- fines the radiologic response of radiosurgery. In vascular malformations, this response is character- ized by the development of obliterative changes within the blood vessel related to endothelial pro- liferation and vessel wall damage. These changes eventually lead to luminal obliteration.

L. Dade Lunsford, M.D. Pittsburgh, Pennsylvania

We welcome this single-case report, which is one of the few dealing with histologic follow-up following gamma knife irradiation of the brain. This occurred fortuitously (not for the patient) when sudden death occurred due to unrelated causes. The au- thors do an excellent job in analyzing various por- tions of the radiated brain and involved blood ves- sels at varying distances and doses from the epicenter of the targeted lesion. The changes are not totally unexpected; however, unexpected is the change seen within the isodose area of 25 Gy. Here, a recanalizing vessel is observed and it brings up the issue of recanalization of vessels previously considered obliterated by radiosurgery and the usual follow-up angiogram. This might suggest that the promoters of radiosurgery as a cure-all should be encouraged to repeat angiograms at varying long-term intervals following their apparent recog- nition of successful therapy of AVMs on the first posttreatment angiogram. This study also demon- strates the potential risks of angiography in the evaluation and management of AVMs. As cerebral angiography remains the gold standard diagnostic test for the follow-up of this procedure, patients shoulcl be informed of the risks when undergoing radiosurgery or microsurgery for AVMs [ 11.

Bennett M. Stein, M.D. Michael B. Sisti, M.D.

New York, New York

REFERENCE 1. Kondziolka D, Lunsford LD, Kanal E, Talagala L. Ster-

eotactic magnetic resonance angiography for targeting in arteriovenous malformation radiosurgery. Neuro- surgery 1994;35:585-91.