Editorial. Radiosurgery in cavernous malformations: anatomy of a controversy

Transcription

1 See the corresponding article in this issue, pp J Neurosurg 113:16 22, 2010 Editorial Radiosurgery in cavernous malformations: anatomy of a controversy La d i s l a u St e i n e r, M.D., Ph.D., 1 Be n g t Ka r l s s o n, M.D., Ph.D., 2 Ch u n-po Ye n, M.D., 1 Ja m e s C. To r n e r, Ph.D., 3 Ch r i s t e r Li n d q u i s t, M.D., Ph.D., 4 a n d Da v i d Sc h l e s i n g e r, Ph.D. 1 1 Lars Leksell Center for Gamma Surgery, University of Virginia Health System, Charlottesville, Virginia; 2 Singapore Gamma Knife Centre, Republic of Singapore; 3 College of Public Health, University of Iowa, Iowa City, Iowa; and 4 Gamma Knife Centre, Bupa Cromwell Hospital, London, United Kingdom The invitation to write an editorial on the article by Lunsford et al. 19 Stereotactic radiosurgery for symptomatic solitary cerebral cavernous malformations considered high risk for resection published in this issue of the Journal of Neurosurgery provided the opportunity for rigorous scrutiny of both our and Lunsford s role in this controversy. We looked for possible flaws: biases and the quest to prevail, rather than the pursuit of truth, which might have tainted the approach to the issue. A short narrative of the start and the development of the dispute may make it easier to understand the controversy s resilience. Our successful obliteration of a series of cerebral arteriovenous malformations using the Gamma Knife in the 1970s 26,27 prompted a trial using the same technique for cavernous malformations (CMs). Between 1983 and 1987 Steiner treated 15 cases, and the unexpected unfavorable outcomes were first reported on April 30, 1990, in an oral presentation at the American Association of Neurological Surgeon s Meeting in Nashville, Tennessee. 31 Lunsford et al. 19 started using the Gamma Knife for CMs in 1988, and Coffey from Pittsburgh, commenting on our presentation, 31 explained our bad results based on the use of CT imaging instead of MR imaging (unavailable in Stockholm at that time) and too-high prescription doses. Steiner decided on a total moratorium on radiosurgery for CMs, which Karlsson, Kihlström, Rähn, and Noren implemented in Stockholm only after unsatisfactory outcomes in an additional 8 patients treated between 1990 and Lindquist treated another 4 brainstem and 1 thalamic CM at Bupa Cromwell Hospital in London. Despite low prescription doses in the latter 2 series (10 16 Gy) and the use of MR imaging for dose planning, the results were not better than for the first 15 cases. The follow-up for the cases treated in London was 6 27 months. Rebleeding after radiosurgery occurred in 2 cases, and radiation-induced complications were noted in another 2 cases. One case remained unchanged with moderate neurological deficits. Kondziolka et al. 15,16 published 2 articles: 1 prospective study assessing the natural course of CMs and 1 retrospective study suggesting that radiosurgery protects CMs from rupture. These authors later reported longterm results emphasizing an improvement in the natural history of CMs following radiosurgery and advocating radiosurgery in cases of difficult approaches for microsurgery. 6,14,15 An increasing number of neurosurgeons followed the policy established in Pittsburgh without any reservation. 7,10,17 Others did so but with some caveats. 1,20,22 Pollock and colleagues 22 stated that limitations in our knowledge of the natural history of untreated CMs make it impossible to conclude that radiosurgery protects against the future risk of bleeding. Radiosurgery of CMs does appear to entail a greater risk of radiation-related complications compared with that of [arteriovenous malformations]. Huang et al., 7 commenting on the impressive results of Hasegawa and associates, 6 emphasized selection bias as well as the difference between hemorrhage and outcomes described by him as compared with those previously reported. Huang and colleagues also criticized the use of the study group as its own control. Friedman s complimentary review of the same paper was tempered by the same critical mention of selection bias. 6 We described both an occasional limited decrease in the hemorrhage rate after radiosurgical treatment and partial obliteration of a CM excised by Steiner when 5 years after a Gamma Knife procedure no improvement in the frequency of seizure had been achieved (Fig. 1), 8,28,29 but we believed that the high incidence of radiation-induced complications did not justify the limited protection the treatment might afford. We suggested, a prospective randomized study is needed to establish the role of radiosurgery in the management of these lesions. 8 The most revealing part of the current report by Lunsford et al. 19 is the candid account of the study s weaknesses. In their Discussion, while emphasizing that they did not observe in their material any clustering of bleeding such as that reported by Barker et al., 2 they did admit that their data cannot refute the hypothesis that some CCMs may bleed repeatedly for some interval and then cease to bleed a possibility that could prove to be a confounding variable at odds with our results. Lunsford and colleagues also agree that the results of the retrospective method they used may have been partly skewed 16 J Neurosurg / Volume 113 / July 2010

2 Fig. 1. Photomicrograph showing a section of a CM that was treated with Gamma Knife surgery. Imaging did not reveal a change in lesion size, and the patient continued to have seizures. The CM was extirpated and histological analysis revealed that more than 75% of the CM had been obliterated. Masson trichrome, original magnification 100. by selection bias, and thus the pretreatment hemorrhage rate in their highly selected patient population does not necessarily represent the natural course of the disease. Natural Course of CMs Studies analyzing the natural course of CMs are either prospective or retrospective. Retrospective studies can be divided into 2 groups depending on the methodology used. With Method A, one simply divides the number of observed hemorrhages by the number of risk years. This method assumes that the patients are born with CMs and that the annual risk for hemorrhage is constant. In Method B, the number of observed hemorrhages is divided by the number of risk years as well, but both values are from the day after the first hemorrhage. Thus, the first hemorrhage as well as the time at risk before the first hemorrhage is excluded, making the method insensitive to the time of the development of the CM. Retrospective studies document the incidence of hemorrhage, which may or may not equal the prospective risk for hemorrhage. A selection bias can lead to a significant difference between the found incidence of and the actual annual risk for hemorrhage. The likelihood of drawing skewed conclusions is very high should the patient selection be based on prior ruptures. The impact of this selection bias is much weaker in prospective studies, making reasonable the assumption that the calculated incidence of hemorrhages equals the actual risk for hemorrhage. The annual risk of hemorrhage in 4 published studies analyzing the natural history of CMs is provided in Table J Neurosurg / Volume 113 / July ,16,23,24 The incidence of hemorrhage using prospective methods ranged between 0.8 and 3.8%. The incidence calculated from the same group of patients ranged between 0.3 and 2.3% if Method A was used. The minimal difference can be explained by the way patient-years at risk are calculated. In Method A one assumes that patients were born with CMs, which might not be true because the low incidence of hemorrhage in children and de novo CMs in familial CMs contradicts this assumption. 3,32 Because the natural course of CMs is still poorly understood it is tempting to use the pretreatment incidence of hemorrhage in the treated patient population as an indicator of the pretreatment risk instead of the % annual risk documented in papers analyzing the natural course of CMs. 5,12,16,21,23,24 The argument that CMs in the treated patient population represent a higher risk for hemorrhages than other CMs the proof being that they frequently bled before treatment sounds like circular reasoning to us and may lead to erroneous conclusions. In Table 2 we analyzed the annual risk of hemorrhage in radiosurgical series of patients with CMs, which selected only so-called high-risk patients. Using Method A, hemorrhage rates in these high-risk patients ranged between 3.9 and 6.5%, a little higher than the reported rates of % in the studies of natural history, but they became 17 36% when Method B was used. 1,6,11,18,22,30 Can this difference be explained by a much higher hemorrhage risk among patients selected for radiosurgery, or is it mainly caused by a methodological flaw? If we agree that the higher risk of subsequent hemorrhages for some period of time occurs in patients with prior symptomatic hemorrhages, the calculation of the hemorrhage rate using Method B will be problematic, because the increased number of hemorrhages at this specific time period and the prematurely terminated follow-up time by radiosurgery lead to a skewed, overly high hemorrhage rate. Increased Hemorrhage for Some CMs Available evidence suggests that CMs are not static lesions for which the risk of a clinically detectable hemorrhage is constant and independent of anatomical and clinical factors. It has been suggested that the risk for hemorrhage is higher in female patients 21,24 and in centrally located CMs. 16,23 Kondziolka et al. 16 have also observed that the risk for a clinically detectable CM hemorrhage increases after an earlier bleed; however, it is unknown for how long this elevated risk persists. It is reasonable to believe that the increase in risk is temporary; if not, a larger number of CMs would have a higher risk for hemorrhage because the majority of these lesions show imaging evidence of either acute or subacute hemorrhage. It should also be noted that while blood degradation products suggest prior hemorrhages, they might also occur from extravasations of a small amount of erythrocytes. To our knowledge, the duration of the increased risk has not been analyzed in any scientific publications; therefore, we cannot exclude the possibility that the decrease in the bleeding rate after the treatment of recently hemorrhaged CMs is to some extent caused by a reset to the prehemorrhage bleeding risk rather than by treatment. 17

3 TABLE 1: Incidence of annual hemorrhage in studies analyzing the natural history of CMs* Authors & Year Available for FU Prospective Retrospective Method A Kim et al., Porter et al., Kondziolka et al., Robinson et al., * FU = follow-up. Hemorrhage rates are calculated by us from the numbers of hemorrhages and risk years reported by the authors in the publications. Checking the Validity of the Methods and Conclusions To analyze the consequences of different methodologies and selection biases, we used a method previously reported by Karlsson and Söderman 9 in which a virtual patient population and a random number generator simulated outcome. However, instead of a 1 and 2% annual hemorrhage risk, we ascribed a 3% risk to 2000 virtual CM patients and followed them up for 8 years. We could then use Methods A and B to analyze the results and see if they were different depending on the analytical method used. Both methods resulted in a 3.1% annual incidence of hemorrhage, a nonsignificant difference compared with the ascribed 3% risk (p < 0.001). Things changed when we introduced the bias of selecting patients with single or multiple hemorrhages only, thus trying to mimic the selection criteria for patients treated with radiosurgery. If we only included patients with at least 2 hemorrhages and treated them soon after the last hemorrhage, the results using Method B changed dramatically. The calculated pretreatment risk for hemorrhage is now 26%, which should be compared with the true risk of 3% an increase with a factor of almost 8. Comparing Published Results With the Virtual Model Let us set the parameters in our virtual patient population so that the summarized pre- and posttreatment hemorrhage risks in 5 papers will be matched. 1,11,18,22,30 Let us assume that the 2000 virtual patients will be followed up for 8 years before the virtual treatment. Let us also assume a 3% annual risk of hemorrhage, increasing it to 8% the first 2 years after hemorrhage, which resets to the prehemorrhage level 2 years later. Assume that the patient is treated 1 year after the last hemorrhage and that we treat all patients with a hemorrhage and follow them up 6 years after the virtual treatment (Table 3). Neither the pre- nor the posttreatment risks differ significantly between the virtual and the actual or published patient population. Thus, the published results are compatible with a 3% annual risk and an additional 8% for 2 years following a hemorrhage. The fact that the risks for hemorrhages before and after radiosurgical treatment are compatible with the 3 and 8% risks ascribed to our virtual patients cannot be taken as evidence that these risk numbers apply to the CMs in actual patients treated with radiosurgery. It does, however, show that the results following CM irradiation cannot be taken as evidence that radiosurgery affects the risk of hemorrhage, as the changes seen can be caused by a reset to a prehemorrhage bleeding risk or other factors. Comparing the findings in the virtual patient population with the data in the current paper by Lunsford et al., 19 we noted that the pretreatment incidence of hemorrhage was significantly higher (32.5%) compared with that in the virtual population and that the risk of hemorrhage more than 2 years after treatment was significantly lower (1.1%). When Lunsford and colleagues results were com- TABLE 2: Incidence of annual hemorrhage from radiosurgical series using retrospective Methods A and B* Authors & Year Method A Method B Amin-Hanjani et al., Hasegawa et al., Kim et al., Pollock et al., Liu et al., NA NA Tsien et al., * NA = not available in the published article. 18 J Neurosurg / Volume 113 / July 2010

4 TABLE 3: Postradiosurgical incidence of hemorrhage within and after 2 years compared with the virtual patient group and patients in the Lunsford et al. report Authors & Year Before Radiosurgery Incidence of Hemorrhage <2 After Radiosurgery >2 Amin-Hanjani et al., Kim et al., Liu et al., Pollock et al., Tsien et al., total virtual patients p value Lunsford patients p value < pared with the cumulative findings of the 5 referred articles, 1,11,18,22,30 the result was 4 ruptures in 378 years versus 24 ruptures in 380 years (p < 002). Comparing our results 8 to the findings of Lunsford et al., we observed 4 hemorrhages in 39 risk-years within 2 years after treatment as compared with 22 hemorrhages in 204 years, a nonsignificant difference (p = 0.99). There was, however, a significant difference in the hemorrhage rate 2 years or more after treatment. We registered 5 hemorrhages in 76 risk-years as compared with 4 hemorrhages in the 378 years reported by Lunsford et al., a significant difference (p < 0.01). These contradictory findings compel us to conclude that the impact of radiosurgery on the risk for CM hemorrhage is still not established. Postradiosurgery Hemorrhage Rate and Incidence of Complications If all patients in the 15 studies reporting results following CM radiosurgery are considered, we find that a complication developed in approximately 132 patients (19%) 1,4, 6 8,10,11,13,17,18,20,22,25,30,33 and around one-half of the complications were permanent. Amin-Hajani et al. 1 reported 3 lethal complications. Thus, the incidence as well as the severity of complications seems not to be negligible after radiosurgery for CMs. -Benefit Analysis Based on the presented facts, we believe that one cannot use the pretreatment incidence of hemorrhage to determine if radiosurgery has or has not affected the natural course of the disease. Instead, one can use a fundamental principle in radiobiology the relation between the radiation dose and the risk of complications; that is, the higher the dose given, the higher the risk of complications. If CMs respond to radiosurgery, there must be a relation between the dose and the response; the higher the dose given, the lower the posttreatment incidence of hemorrhages after 2 years. Thus, by extracting the complications and posttreatment hemorrhage rates from different clinical studies, we J Neurosurg / Volume 113 / July 2010 can get an idea of how steep the dose-response curve is for CMs in relation to the dose-risk curve. For this analysis, we used 10 studies for which the results could be extracted from the information published. 1,4,6,8,13,18,20,22,30,33 As seen in Fig. 2, there was a big difference in outcome at the different centers the most common, a 2 4% risk of hemorrhage and a 10 30% risk of symptomatic complications. The trend in the meta-analysis was that the incidence of complications increased with an increasing incidence of hemorrhages 2 years or more after radiosurgery. Given the small number of patients in the studies used, the conclusion was tentative. However, the data suggested that the dose-response relation is weak, which can be interpreted as a very limited response to radiation, or that the maximum radiation response is already reached when a low radiation dose is given. Radiosurgery for CMs Considered High for Resection The controversy on the use of radiosurgery for CMs boils down to the validity of the methods used to calculate and compare the pre- versus the postradiosurgery hemorrhage rates. In this context the 20-year-old debate remains unsolved. The current report by Lunsford et al. includes only CMs considered taboo for surgery. In this existential scenario the choices available to the patient are limited to doing nothing or undergoing radiosurgery. Actually, the controversy discussed in this editorial addresses the validity of methods used to define the natural history of CMs, the incidence and annual risk of hemorrhage before and after radiosurgery, the role of selection biases, and a methodology that can generate flawed conclusions. These problems apply to both operable and inoperable CMs. Notwithstanding, in dealing with patients harboring inoperable CMs in whom none of the management alternatives meet their expectations, neurosurgeons with serious arguments against the rationales for using radiosurgery will face the agonizing task of answering questions posed by a desperate patient or family in an existential situation. Delicately balanced information based on knowledge, 19

5 admissible in any field of science.... For the belief in a single truth and in being the possessor thereof is the root cause of all evil in the world. Disclosure Drs. Karlsson and Lindquist are both consultants for, and Dr. Lindquist owns stock in, Elekta AB. Acknowledgment The authors thank the skillful and dedicated secretarial assistance of Mrs. Brenda Melan from the Lars Leksell Center for Gamma Surgery, Department of Neurological Surgery, University of Virginia Health System, Charlottesville, Virginia. Fig. 2. Graph depicting the relation between the incidence of complications (x axis) and the annual incidence of hemorrhages 2 years after radiosurgical treatment (y axis). Line represents the linear regression after the highest and lowest annual hemorrhage rates have been excluded. conscience, psychological insight, compassion, and empathy may provide the facts necessary for an informed decision and simultaneously keep alive the hope of the patient. If a neurosurgeon decides to decline to perform radiosurgery, he or she should encourage the patient to obtain a second opinion from peers who disagree with his or her conclusion. Concerning the feasibility of surgical extirpation with reasonable risks, the opinions of a neurosurgeon with substantial experience in the management of this kind of lesion should be sought as well as a decision by the local multidisciplinary stereotactic radiosurgery conference. Criteria for the inoperability of CMs are in a process of revision. More frequent use of neurophysiological testing when selecting the appropriate approach to the lesion and changing the technique dissecting fibers of the brainstem instead of cutting them may permit surgery for CMs located a distance from the pia. Conclusions In summary, the paper by Lunsford et al. is suggestive but not conclusive regarding the possible reduction of hemorrhage in CMs, and the cause and effect of radiosurgery cannot be established. Person-years analysis (Methods A and B) has some flaws, such as the questions of constancy or randomness of risk. Survival modeling or a pre-post design of individual rates of hemorrhage (for example, a comparison of the individual hemorrhage rate per period of time pre- and postradiosurgery) may serve as an alternative. However, even these methods lead only to estimates that should be confirmed in prospective comparative studies. To end the discussion of this elusive topic, it seems appropriate to quote Max Born s thoughtful phrase, ideas such as absolute certitude, absolute exactness, final truth, etc. are figments of the imagination which should not be References 1. Amin-Hanjani S, Ogilvy CS, Candia GJ, Lyons S, Chapman PH: Stereotactic radiosurgery for cavernous malformations: Kjellberg s experience with proton beam therapy in 98 cases at the Harvard Cyclotron. Neurosurgery 42: , Barker FG II, Amin-Hanjani S, Butler WE, Lyons S, Ojemann RG, Chapman PH, et al: Temporal clustering of hemorrhages from untreated cavernous malformations of the central nervous system. Neurosurgery 49:15 25, Brunereau L, Levy C, Laberge S, Houtteville J, Labauge P: De novo lesions in familial form of cerebral cavernous malformations: clinical and MR features in 29 non-hispanic families. Surg Neurol 53: , Chang SD, Levy RP, Adler JR Jr, Martin DP, Krakovitz PR, Steinberg GK: Stereotactic radiosurgery of angiographically occult vascular malformations: 14-year experience. Neurosurgery 43: , Del Curling O Jr, Kelly DL Jr, Elster AD, Craven TE: An analysis of the natural history of cavernous angiomas. J Neurosurg 75: , Hasegawa T, McInerney J, Kondziolka D, Lee JY, Flickinger JC, Lunsford LD: Long-term results after stereotactic radiosurgery for patients with cavernous malformations. Neurosurgery 50: , Huang YC, Tseng CK, Chang CN, Wei KC, Liao CC, Hsu PW: LINAC radiosurgery for intracranial cavernous malformation: 10-year experience. Clin Neurol Neurosurg 108: , Karlsson B, Kihlström L, Lindquist C, Ericson K, Steiner L: Radiosurgery for cavernous malformations. J Neurosurg 88: , Karlsson B, Söderman M: Stereotactic radiosurgery for cavernomas, in Lunsford LD, Sheehan JP (eds): Intracranial Stereotactic Radiosurgery. New York: Thieme, 2009, pp Kida Y, Kobayashi T, Tanaka T: Treatment of symptomatic AOVMs with radiosurgery. Acta Neurochir Suppl 63:68 72, Kim DG, Choe WJ, Paek SH, Chung HT, Kim IH, Han DH: Radiosurgery of intracranial cavernous malformations. Acta Neurochir (Wien) 144: , Kim DS, Park YG, Choi JU, Chung SS, Lee KC: An analysis of the natural history of cavernous malformations. Surg Neurol 48:9 18, Kim MS, Pyo SY, Jeong YG, Lee SI, Jung YT, Sim JH: Gamma knife surgery for intracranial cavernous hemangioma. J Neurosurg 102 Suppl: , Kondziolka D, Lunsford LD, Coffey RJ, Bissonette DJ, Flickinger JC: Stereotactic radiosurgery of angiographically occult vascular malformations: indications and preliminary experience. Neurosurgery 27: , Kondziolka D, Lunsford LD, Flickinger JC, Kestle JR: Reduction of hemorrhage risk after stereotactic radiosurgery for cavernous malformations. J Neurosurg 83: , J Neurosurg / Volume 113 / July 2010

6 16. Kondziolka D, Lunsford LD, Kestle JR: The natural history of cerebral cavernous malformations. J Neurosurg 83: , Liscák R, Vladyka V, Simonová G, Vymazal J, Novotny J Jr: Gamma knife surgery of brain cavernous hemangiomas. J Neurosurg 102 Suppl: , Liu KD, Chung WY, Wu HM, Shiau CY, Wang LW, Guo WY, et al: Gamma knife surgery for cavernous hemangiomas: an analysis of 125 patients. J Neurosurg 102 Suppl:81 86, Lunsford LD, Khan AA, Niranjan A, Kano H, Flickinger JC, Kondziolka D: Stereotactic radiosurgery for symptomatic solitary cerebral cavernous malformations considered high risk for resection. Clinical article. J Neurosurg [epub ahead of print February 19, DOI: / JNS081626] 20. Mitchell P, Hodgson TJ, Seaman S, Kemeny AA, Forster DM: Stereotactic radiosurgery and the risk of haemorrhage from cavernous malformations. Br J Neurosurg 14:96 100, Moriarity JL, Wetzel M, Clatterbuck RE, Javedan S, Sheppard JM, Hoenig-Rigamonti K, et al: The natural history of cavernous malformations: a prospective study of 68 patients. Neurosurgery 44: , Pollock BE, Garces YI, Stafford SL, Foote RL, Schomberg PJ, Link MJ: Stereotactic radiosurgery for cavernous malformations. J Neurosurg 93: , Porter PJ, Willinsky RA, Harper W, Wallace MC: Cerebral cavernous malformations: natural history and prognosis after clinical deterioration with or without hemorrhage. J Neurosurg 87: , Robinson JR, Awad IA, Little JR: Natural history of the cavernous angioma. J Neurosurg 75: , Stea RA, Schicker L, King GA, Winfield JA: Stereotactic linear radiosurgery for cavernous angiomas. Stereotact Funct Neurosurg 63: , Steiner L: Radiosurgery in cerebral arteriovenous malformations, in Fein JM, Flamm ES (eds): Cerebrovascular Surgery. New York: Springer-Verlag, 1985, Vol 4, pp Steiner L, Leksell L, Greitz T, Forster DM, Backlund EO: Stereotaxic radiosurgery for cerebral arteriovenous malformations. Report of a case. Acta Chir Scand 138: , Steiner L, Lindquist C, Steiner M: Radiosurgery. Adv Tech Stand Neurosurg 19:19 102, Steiner L, Sheehan J, Lindquist C, Stroila M, Prasad D, Steiner M: Gamma knife surgery for cerebral vascular malformations, tumors and functional disorders, in Schmidek HH, Roberts DW (eds): Schmidek and Sweet s Operative Neurosurgical Techniques: Indications, Methods, and Results, ed 5. Philadelphia: WB Saunders, 2005, Vol 1, pp Tsien C, Souhami L, Sadikot A, Olivier A, del Carpio-O Donovan R, Corns R, et al: Stereotactic radiosurgery in the management of angiographically occult vascular malformations. Int J Radiat Oncol Biol Phys 50: , Weil S, Tew JM Jr, Steiner L: Comparison of radiosurgery and microsurgery for treatment of cavernous malformations of the brain stem. J Neurosurg 72:336A, 1990 (Abstract) 32. Zabramski JM, Wascher TM, Spetzler RF, Johnson B, Golfinos J, Drayer BP, et al: The natural history of familial cavernous malformations: results of an ongoing study. J Neurosurg 80: , Zhang N, Pan L, Wang BJ, Wang EM, Dai JZ, Cai PW: Gamma knife radiosurgery for cavernous hemangiomas. J Neurosurg 93 (Suppl 3):74 77, 2000 Response Douglas Kondziolka, M.D., and L. Dade Lunsford, M.D. University of Pittsburgh, Pittsburgh, Pennsylvania J Neurosurg / Volume 113 / July 2010 We thank Dr. Steiner and colleagues for their thoughtful commentary and analysis of the epidemiological issues related to deep-brain CMs and the potential role of radiosurgery in this setting. They discuss many of the issues on this topic, arriving at the conclusion that truth may be difficult to obtain and that our recent report is suggestive but not conclusive about the reduction of a hemorrhage risk for CMs. Cavernous malformations are common. Essentially, they are a diagnosis of the era of MR imaging. Although rarely identified before that period, they are now frequently seen in our hospitals and clinics. Indeed, most patients are asymptomatic and are followed up with serial imaging studies and periodic clinical assessments. Resection is used most frequently for hemorrhagic malformations or those causing symptoms when the malformation is accessible with acceptable risks. We advocate Gamma Knife surgery in the setting of symptomatic, hemorrhagic CMs for which resection is believed to be an unacceptable choice. In such cases when patients choose radiosurgery over observation, they do so because they know well the neurological impact of the CM. As part of our comprehensive assessment of CM radiosurgery, it is important to note that most patients treated in our practice are simply monitored. Indeed, this patient population was the genesis of the first prospective natural history study of CMs, which the authors cite. Importantly, we found that in patients with 2 symptomatic hemorrhages, a 32% annual risk for another symptomatic event could be expected in contrast to a 0.6% annual risk in those who had not had a prior symptomatic hemorrhage. We acknowledge the fact that hemorrhage predictions are only as good as the data from which they are drawn. To date, nobody has followed CMs with serial imaging studies for extended periods of time, and no one can state with any strong degree of confidence what the natural history might be over a 10- to 20-year period. On the other hand, the authors use of Method A, simply dividing the number of observed hemorrhages by the number of risk years (assuming the patient was born with a CM), is not sound and can simply be discarded. It is well known to neurosurgeons that CMs can become evident on imaging over time, a finding well documented in both the neurosurgical and neuroradiological literature. Thus, their Method B, in which the number of observed hemorrhages is divided by the number of risk years (after the first hemorrhage), makes more sense. Indeed, it is a real-world patient scenario. The patient has a hemorrhage and learns that he or she has a CM. After that, they want to know what to do. It is at this point that the neurosurgeon offers advice. We have used this approach in our clinical studies of both observation and radiosurgery. The authors discuss their own results as well as a limited number of other series that have utilized different techniques. Why their own results were unsatisfactory remains unclear. Both higher and lower doses were used. What also remains unclear is how they defined the target. Even in our own experience, we have found that targeting within the hemosiderin rim rather than at its external edge is important. The target volume is reduced, and the radiation fall-off is steeper. is related to volume and location. Most CMs are located in the brainstem or di- 21

7 encephalon. We have not seen the degree of unsatisfactory outcomes that they noted. Perhaps this is our own perspective. When they discuss risks, they quote 1 report that documented 3 deaths. It is important to note that that study was a proton-beam therapy series and likely without the stereotactic accuracy of the Gamma Knife technique. The authors discuss at length a virtual patient model used to simulate the outcomes of 2000 virtual patients assuming variable hemorrhage risks. Previously, we conducted a decision analysis of CM radiosurgery by using different CM outcomes and found it to be so complex as to be truly unsatisfactory for publication. Rather than performing a virtual analysis, we prefer to publish our own data. Our conclusions are based on patients that we selected for radiosurgery. To that end, we acknowledge a selection bias. were chosen because they typically had multiple symptomatic hemorrhages in high-risk locations and chose to undergo the procedure. The authors then call for a prospective randomized trial. Perhaps they are not aware that such a trial was actually begun under our leadership. It included 8 large neurovascular academic programs in the US and Canada. A CM study group met over a 2-year period and designed a randomized trial comparing resection and radiosurgery for high-risk malformations. Unfortunately, after 1 year had passed, not 1 patient had been entered into the study. Truly, this is another example of neurosurgeons addressing a fascinating topic, but unfortunately not a common one. In summary, CM radiosurgery has been controversial, but after years of experience at our center, we do not think it must remain so. with multiple hemorrhagic and symptomatic CMs in high-risk locations seek treatment. The data presented in our reports consistently show a significant reduction in the hemorrhage rate, particularly after 2 years. The procedure meets the goals of the patient. The morbidity profile has been limited. Figure 1 in their commentary shows that the greater part of a CM has been obliterated. As in all surgeries, neurosurgeons and their colleagues must use the information at hand to best counsel the patient a reminder that recalls the words of Michel de Montaigne ( ) who noted that whenever a new scientific discovery is presented, it is first said that it cannot be true. When its truth has been demonstrated beyond question, it is said, Yes, it may be true, but it is not important. Finally, when time has elapsed to fully evidence its importance, it is said, Yes, surely it is important, but it is no longer new. Cavernous malformation radiosurgery is no longer new. There is no controversy in that. Disclosure Drs. Kondziolka and Lunsford are both consultants for, and Dr. Lunsford owns stock in, Elekta AB. Please include this information when citing this paper: published online February 19, 2010; DOI: / JNS J Neurosurg / Volume 113 / July 2010