- Email: [email protected]
Pretreatment Predictors of Adverse Radiation Effects After Radiosurgery for Arteriovenous Malformation
Int. J. Radiation Oncology Biol. Phys., Vol. 82, No. 2, pp. 803–808, 2012 Copyright Ó 2012 Elsevier Inc. Printed in the USA. All rights reserved 0360-3016/$ - see front matter
doi:10.1016/j.ijrobp.2010.12.014
CLINICAL INVESTIGATION
Central Nervous System Tumor
PRETREATMENT PREDICTORS OF ADVERSE RADIATION EFFECTS AFTER RADIOSURGERY FOR ARTERIOVENOUS MALFORMATION CAROLINE HAYHURST, M.D.,* ERIC MONSALVES, B.SC.,* MONIQUE VAN PROOIJEN, PH.D.,y MICHAEL CUSIMANO, M.D.,z MAY TSAO, M.D.,x CYNTHIA MENARD, M.D.,k ABHAYA V. KULKARNI, M.D., PH.D.,{ MICHAEL SCHWARTZ, M.D.,** AND GELAREH ZADEH, M.D., PH.D.* *Gamma Knife Unit, Division of Neurosurgery, University Health Network, yPhysics Department, Princess Margaret Hospital, Division of Neurosurgery, St Michael’s Hospital, Toronto, Canada; xRadiation Oncology Program, Sunnybrook Hospital, kRadiation Oncology Program, Princess Margaret Hospital, {Division of Neurosurgery, Hospital for Sick Children, **Division of Neurosurgery, Sunnybrook Hospital, University of Toronto, Canada
z
Purpose: To identify vascular and dosimetric predictors of symptomatic T2 signal change and adverse radiation effects after radiosurgery for arteriovenous malformation, in order to define and validate preexisting risk models. Methods and Materials: A total of 125 patients with arteriovenous malformations (AVM) were treated at our institution between 2005 and 2009. Eighty-five patients have at least 12 months of clinical and radiological follow-up. Any new-onset headaches, new or worsening seizures, or neurological deficit were considered adverse events. Follow-up magnetic resonance images were assessed for new onset T2 signal change and the volume calculated. Pretreatment characteristics and dosimetric variables were analyzed to identify predictors of adverse radiation effects. Results: There were 19 children and 66 adults in the study cohort, with a mean age of 34 (range 6–74). Twenty-three (27%) patients suffered adverse radiation effects (ARE), 9 patients with permanent neurological deficit (10.6%). Of these, 5 developed fixed visual field deficits. Target volume and 12 Gy volume were the most significant predictors of adverse radiation effects on univariate analysis (p < 0.001). Location and cortical eloquence were not significantly associated with the development of adverse events (p = 0.12). No additional vascular parameters were identified as predictive of ARE. There was a significant target volume threshold of 4 cm3, above which the rate of ARE increased dramatically. Multivariate analysis target volume and the absence of prior hemorrhage are the only significant predictors of ARE. The volume of T2 signal change correlates to ARE, but only target volume is predictive of a higher volume of T2 signal change. Conclusions: Target volume and the absence of prior hemorrhage is the most accurate predictor of adverse radiation effects and complications after radiosurgery for AVMs. A high percentage of permanent visual field defects in this series suggest the optic radiation is a critical radiosensitive structure. Ó 2012 Elsevier Inc. Stereotactic radiosurgery, Arteriovenous malformation, Complications, Radiation injury.
unclear. Although radiation injury is known to be a function of dose and volume, the extent to which the vascular morphology of the nidus contributes to adverse radiation effects (ARE) and the interaction with radiation dosimetry characteristics is unclear. Previous multi-institutional studies led by the University of Pittsburgh and the Arteriovenous Malformation Radiosurgery Study Group have proposed a model of symptomatic postradiation injury based on the 12-Gy volume and nidus location proposing that the presence of clinical sequelae of T2 signal change is primarily related to location (8, 9). Flickinger et al. developed a risk model of AVM location, stratified into four tiers, with the frontal lobe representing the least risk
INTRODUCTION Radiosurgery provides an effective therapeutic option for selected brain arteriovenous mal-formations (AVM), achieving complete obliteration in 60–90% of cases with higher obliteration rates of >90% in smaller lesions (1–3). However, new areas of T2 signal change denoting radiation effects in normal surrounding brain are frequently seen. Such imaging changes occur much more frequently after radiosurgery for AVM than for other pathologies (4), occurring in 30–60% of cases (5, 6). Although white matter imaging changes are frequent, only a proportion will develop clinical complications after treatment (6, 7). The reasons for this discrepancy in response to radiosurgery in AVMs are Reprint requests to: Dr. Gelareh Zadeh, M.D., Ph.D., University of Toronto, Department of Neurosurgery, 399 Bathurst St., 4-439 West Wing, Toronto, Ontario, Canada M5G 2S8. Tel: (416) 6035679; Fax: (416) 603-5298; E-mail: [email protected]
Conflicts of interest: none. Received July 28, 2010, and in revised form Nov 24, 2010. Accepted for publication Dec 3, 2010. 803
I. J. Radiation Oncology d Biology d Physics
804
Volume 82, Number 2, 2012
Table 1. Significant postradiosurgery injury expression score (SPIE) (Flickinger et al., 2000)
Table 2. Patient characteristics and univariate analysis Patient characteristics
SPIE score
Age 1 2 3 4
Value
p value
Median 36 (range, 6–74) M 45 F 40 37 (43.5%) 22 (25.9%) 6 (7%) 13 (15.3%) 5 (5.9%) 1 17 (20%) 2 24 (28.2%) 3 37 (43.5%) 4 7 (8.2%) 50 40 (47.1%)
0.99
Location Frontal Cerebellum Temporal Parietal Occipital Basal ganglia Medulla Thalamus Intraventricular Pons/midbrain Corpus Callosum
and brainstem location the highest risk, denoted the significant postradiosurgery injury expression (SPIE) score (Table 1) (7). Although these models take into account radiation dose, they do not include the AVM vascular characteristics. Similarly, the radiosurgery-based AVM grading system was developed to predict the chance of complete obliteration without new deficit and is based on factors associated with a successful outcome (10). It incorporates location stratified into three tiers of frontal/temporal, parietal/occipital/corpus callosum, or basal ganglia and brainstem, with age and nidus volume. This score has been validated as a predictor of outcome, but includes obliteration and deficit from posttreatment hemorrhage as dependent variables (10, 11). In summary, the radiosurgery-based AVM score is a function of lesion volume and location and does not take into account additional AVM vascular characteristics or radiosurgical dosimetry variables. The aim of the current study was to identify vascular parameters inherent in the AVM, in addition to radiosurgery treatment characteristics that could predict ARE, to develop a clinically relevant tool for pretreatment patient counseling and clinical decision-making. Additionally, identifying predictive factors of ARE can provide important insights into the impact of radiation on the central nervous system. METHODS AND MATERIALS
Sex No previous hemorrhage Previous treatment: Surgery Embolization Radiosurgery (linac) Spetzler Martin grade
Eloquent location Venous drainage Deep Superficial Arterial supply Deep perforator Cortical Location (SPIE):
– 0.005 0.979
0.04
0.12 0.21
50 (58.8%) 35 (41.2%) 0.34
Maximum diameter (cm) Volume T2 signal change (cm3) Time to maximum T2 change
29 (34.1%) 56 (65.9%) 1 16 2 29 3 11 4 29 Median 2.42 cm (range, 0.1–5.13 cm) 11.65 (0.53–127.3)
0.12
– 0.001
12 months (5–36)
0.091
Analysis variables The clinical data and pretreatment imaging were reviewed for each patient to define pretreatment characteristics to be used for the prediction analysis. Table 2 demonstrates the variables included. Previous hemorrhage, previous treatment, and SpetzlerMartin grade were included. The presence of deep venous drainage and deep arterial perforator supply were recorded separately. The dosimetric variables studied include target volume, 12 Gy volume, maximum dose (Dmax), V100, and the Radiation Therapy Oncology Group Conformity Index (Table 3). All dosimetric values are calculated on the day of treatment and maintained prospectively in a locally developed web publishing database.
Patients
T2 signal change
In a retrospective review of patients with AVM treated between 2005 and 2009 at the University of Toronto with a Model 4C Gamma Knife facility (Elekta Instruments, Atlanta, GA), patients with a minimum of 12 months of complete clinical and radiological follow-up were included in this study of adverse radiation effects. Approval for this study was granted by the University Health Network research ethics board. All treatments were planned using stereotactic CT angiography, magnetic resonance angiography and two-dimensional digital subtraction angiogram. In our institution, a single dose of 25 Gy is prescribed to the edge of the nidus, but is reduced to 20 Gy for treatment volumes >4 cm3 and further reduced to 15 Gy for lesions in eloquent cortex. Clinical follow-up is undertaken at 3 months. Thereafter, clinical and magnetic resonance imaging follow-up are obtained 6 months, 12 months, and then yearly.
All follow-up imaging was reviewed for the presence of T2 signal change after treatment. Where new T2 signal change occurred, Table 3. Dosimetry variables and univariate analysis Dosimetry variables
Value (median/range) 3
Target volume (cm ) 12 Gy volume Prescription dose Isocenter number D max V100 CI (Radiation Therapy Oncology Group)
3
2.210 cm (0.1–14.6) 6.79 cm3 (0.38–41.01) 20 Gy (15–25 Gy) 10 (1–27) 40 (30–71) 98.59 (88.31–100) 1.54 (1.20–4.56)
p value <0.001 <0.001 0.062 0.04 0.20 0.187 0.41
Predictors of Radiation Affects on Arteriovenous Malformation d C. HAYHURST et al.
the time to onset in months was recorded and the volume of T2 signal change calculated on DICOM FLAIR images using ITKSNAP software (Fig. 1) (12). The time to maximal volume of T2 signal change was recorded.
Adverse radiation effects Clinical data were recorded for all patients. Clinical adverse radiation effects were defined as severe headaches requiring steroid administration, new or worsening seizures, and new focal neurological deficit. No patients experienced new or repeat hemorrhage after radiosurgery in our series. Adverse radiation effects were classified as transient or permanent when still present at last follow-up.
Statistical analysis Univariate logistic regression analysis was used and pretreatment variables with p < 0.10 were then included in a multivariate analysis, after excluding highly collinear variables. Threshold analysis was performed by calculating the area under a receiver operating characteristic curve (AUC) for outcome. All analyses were performed with SPSS version 17 (SPSS Inc., Chicago, IL).
RESULTS Patient characteristics Eighty-eight patients with AVM were treated at our institution between 2005 and 2009. Eight-five patients were included in this study. Three patients were excluded for lack of complete radiological follow-up. There were 66 adults, and 19 children age <18. Table 2 demonstrates the patient’s pretreatment characteristics. Forty-eight (56.5%) had presented with hemorrhage. 22 (25.9%) patients had undergone previous treatment, including surgery, embolization, or linac radiosurgery. The median duration of follow-up is 26 months (range, 12–49 months). No patients experienced posttreatment hemorrhage in the follow-up period. Twenty-three (27%) patients developed adverse radiation effects. Fourteen (16.5%) developed transient complications including headaches, new seizures, and new focal neurological deficit. Nine (10.6%) patients had permanent neurological deficit after treatment: 4 patients had mild hemiparesis
805
and 5 (5.8%) patients had fixed visual field defects. Of those with visual field defects, only one AVM was located within the calcarine cortex. The median time to onset of ARE was 6 months (0.5–36 months). T2 signal change Forty-two (49.4%) patients developed areas of new T2 signal change after treatment. The median time to onset of T2 signal change was 12 months (range, 5–36 months). The median volume of T2 signal change was 11.66 cm3 (0.53–127.3). Increasing volume of T2 signal change is associated with ARE. Those who developed ARE had a median T2 signal change volume of 34.4 cm3 vs. 5.68 cm3 in those without ARE (p = 0.001, OR 1.05). The only significant predictor of the volume of T2 signal change is target volume (p = 0.002). The presence of deep venous drainage, deep perforator arterial supply, and location were not significant. There was no significant association between ARE and early (<12 months) or late (>12 months) onset T2 signal change. Predictors of adverse effects Both transient and permanent ARE were included as the dependent variable for the analysis. On univariate analysis, no prior hemorrhage, target volume, 12 Gy volume, number of isocenters, and Spetzler Martin grade were significant (Tables 2 and 3). However, the significance of the Spetzler Martin grade is related to nidus size because when analyzed separately, deep venous drainage, and eloquent location are not significant. Age at treatment and previous treatment, interestingly including previous radiosurgery, were not significant in this study. The presence of deep perforator arterial supply was also not significant. No additional dosimetric variables contributed to ARE. Location as categorized by the SPIE score or as independent variables were not a significant predictor of ARE in this series (p > 0.12). The category with the highest risk for ARE was SPIE 3, corresponding to occipital and basal ganglia (OR 4.17 [0.91–19.18]).
Fig. 1. T2 signal change: (a) T2 weighted magnetic resonance image acquired on day of stereotactic radiosurgery, (b) new onset T2 signal change at 6 months.
806
I. J. Radiation Oncology d Biology d Physics
Volume 82, Number 2, 2012
ARE. No children had permanent ARE. There is no significant difference in the rate of T2 signal change between adults and children or the time to onset of T2 signal change. The median volume of T2 signal change in children is 14.04 cm3, compared with 13.281 cm3 in adults (p = 0.911). The median target volume in children was 1.67 cm3, which is not significantly different to that in adults (p = 0.212). DISCUSSION
Fig. 2. Incidence of adverse radiation effects according to target volume for stereotactic radiosurgery.
The median target volume in those with ARE is 6.49 cm3 and 2.76 cm3 in those without ARE. Using the AUC to determine significant target volume thresholds, there is a significant threshold of 4 cm3 (AUC 0.77, Fig. 2). 55 AVMs had a target volume of less than 4 cm3 and 6 (10.9%) developed ARE, 30 AVMs were >4 cm3 in volume, and 17 (56.7%) developed ARE. On multivariate analysis of pretreatment variables (excluding 12 Gy volume because it can only be calculated at time of treatment and is correlated to treatment volume), only a treatment volume greater than 4 cm3 and the absence of prior hemorrhage are significant (p < 0.001 OR 9.14 [2.90–28.79] and p = 0.036 OR 3.43 [1.09–10.85], respectively; Fig. 3).
Pediatric AVM Nineteen children were included in this study, representing 22% of the study cohort. Seventeen children (89.4%) presented with hemorrhage. Seven children (36.8%) developed new T2 signal change and 4 developed transient
This study is the first to demonstrate that a target volume threshold exists as a predictor of ARE after radiosurgery for AVMs, independent of location, in patients without a history of hemorrhage. The significant volume threshold of 4 cm3 was seen in this series. There were no vascular parameters, such as venous or arterial anatomy, or nidus morphology and no dosimetry parameters that contributed significantly to ARE. Radiosurgical treatment planning for AVM balances the radiation dose required to achieve nidus obliteration with a dose that risks radiation-induced injury to surrounding normal brain. Knowledge of additional predictors of complications allows for more accurate patient counselling regarding management options and potentially alteration of the radiosurgical plan to minimize complications. In an initial singleinstitution analysis of imaging changes after AVM radiosurgery, Flickinger et al. (5) elegantly demonstrated an actuarial risk of developing imaging changes of 31% at 2 years, with the risk of symptomatic imaging changes 14% at 2 years. Symptomatic imaging changes were more frequent in brainstem locations compared with cortical location. In multivariate analysis only treatment volume was significantly associated with symptomatic imaging changes. However, the study concludes that AVM location determines whether imaging changes are symptomatic or not. Subsequent studies by the University of Pittsburgh Group defined the 12-Gy volume as a significant predictor of symptomatic imaging changes (9). A 12-Gy volume has become the standard for complication
Fig. 3. Multivariate analysis: (a) boxplot depicting target volume for patients with and without adverse radiation effects. (b) Incidence of adverse radiation effects in patients with and without prior hemorrhage.
Predictors of Radiation Affects on Arteriovenous Malformation d C. HAYHURST et al.
807
Fig. 4. Postradiosurgery visual field defect: (a) treatment planning magnetic resonance image showing occipital white matter arteriovenous malformations, (b) extent of occipital white matter T2 signal change 6 months after radiosurgery.
reporting in the radiosurgical literature and, although it is significant in our study, it is of less relevance when considering pretreatment evaluation and risk stratification. Flickinger et al. subsequently devised and validated in a large multiinstitutional series, a model of 12-Gy volume and location for predicting ARE (7, 8, 13). In our series, we were unable to correlate location to the development of ARE, although we have included both transient and permanent ARE in our analysis. In this series, six brainstem AVMs were treated with target volumes between 0.6 cm3 and 2.6 cm3 and none developed complications. Eloquent location was not found to be associated with ARE in our series, although 47.1% of the AVMs treated involved eloquent cortex. A critical finding is that of the nine occipital AVMs treated, five (55.5%) developed visual field defects which are permanent at last follow-up. Only one of these involved eloquent cortex and the remainder are located in deep white matter, suggesting that the optic radiation is highly radiosensitive. Pollock et al. (14) reported 34 patients with AVM within the optic radiation or striate cortex, 27 of whom had normal vision before radiosurgery. Two (6%) developed documented visual field defects after treatment. Consequently, there is increasing interest to integrate diffusion tensor tractography into radiosurgical planning. Maruyama et al. (15) retrospectively integrated optic radiation tractography to the treatment plan of 10 patients treated for AVM located in the postgeniculate visual pathways and demonstrated a maximum dose of less than 12 Gy to the optic radiation did not cause any new visual symptoms. In a study of integrated tractography in cyberknife radiosurgery planning, Pantelis et al. (16) incorporated optic radiation tractography into the optimization process, demonstrating a dose reduction from 2,000 cGy to less than 1,200 cGy to the optic radiation when included in the treatment plan. Despite our initial hypothesis that additional dosimetry or vascular parameters would provide a more accurate prediction
model for ARE, we were unable to demonstrate in a multivariate analysis any additional significant variables in radiosurgical treatment planning. Flickinger et al. did not demonstrate any dosimetry characteristics predictive of ARE, including Dmax, isodose, and dose rate (5, 7–9, 13). Interestingly, prior hemorrhage was not associated with a reduced risk of complications in these studies, in contrast to previous reports and our data (17). An analysis of ARE by Friedman et al. (18), demonstrated a reduced rate of transient complications with improved conformality. The addition of vascular anatomical parameters to our analysis did not show any association with the development of ARE. Isolated case reports have suggested early occlusion of dominant draining veins contributes to perinidal edema and complications (19, 20). Van den Berg et al. (21) reported a higher volume of T2 signal change in those with AVMs drained by a single vein; however, there is no comparison to those without new T2 signal change or association with clinical ARE. It is clear that AVM nidus volume is a determinant of ARE after treatment. In our study, a target volume threshold of 4 cm3 means a large number of AVMs would be considered high risk. However, dose reduction to achieve lower complication rates is associated with significantly lower obliteration rates. Han et al. (22) reported a 12.5% obliteration rate for AVM >14 cm3 treated with a lowest dose of 10 Gy, with a rate of 9.5% postradiosurgery imaging changes. An alternative to balance an effective dose for obliteration with an acceptable risk of complications is staged radiosurgical treatment. Sirin et al. (23), published results of 28 patients with very large volume AVM (>15 mL) treated with 2 or 3 stage radiosurgery. Fourteen patients had more than 36 months’ follow-up with a 50% complete obliteration rate; however, 14% suffered hemorrhage in the follow-up period, appearing to confirm the theoretical risk of increased hemorrhage rates with partial obliteration. Further reports on the outcome of staged
808
I. J. Radiation Oncology d Biology d Physics
radiosurgery for large AVM are required to elucidate the efficacy of this treatment modality. Limitations of the study Our data represent a single-institution retrospective analysis and require validation in a prospective series. However, we would have expected to validate the University of Pittsburgh model regarding location in our patients. We included both transient and permanent neurological sequelae in our study, but it is possible that the location effects are different when permanent deficits alone are studied. Our study numbers remain relatively small, so our analyses might have failed to demonstrate significance for certain predictor variables simply because of lack of statistical power.
Volume 82, Number 2, 2012
CONCLUSION Increasing AVM nidus volume, without a history of prior hemorrhage, is a significant predictor of ARE after radiosurgery, with a volume threshold of 4 cm3, above which complications are more likely to occur. This volume effect appears to be independent of eloquent location, but deep white matter tracts, particularly the optic radiation, are likely to be important in the development of ARE. The optic radiation is most likely more sensitive to radiation injury and a better understanding of the radiation response in the white matter tracts is needed. Further studies integrating tractography into radiosurgical treatment planning will further clarify the role of location in the development of ARE after radiosurgery for AVM.
REFERENCES 1. Karlsson B, Lindquist C, Steiner L. Prediction of obliteration after gamma knife surgery for cerebral arteriovenous malformations. Neurosurgery 1997;40:425–430. discussion 430–421. 2. Flickinger JC, Pollock BE, Kondziolka D, et al. A dose-response analysis of arteriovenous malformation obliteration after radiosurgery. Int J Radiat Oncol Biol Phys 1996;36:873–879. 3. Sun DQ, Carson KA, Raza SM, et al. The radiosurgical treatment of arteriovenous malformations: Obliteration, morbidities, and performance status. Int J Radiat Oncol Biol Phys 2010 Apr 16. [Epub ahead of print]. 4. Flickinger JC, Kondziolka D, Kalend AM, et al. Radiosurgeryrelated imaging changes in surrounding brain: Multivariate analysis and model evaluation. Radiosurgery 1996;1:229–236. 5. Flickinger JC, Lunsford LD, Kondziolka D, et al. Radiosurgery and brain tolerance: An analysis of neurodiagnostic imaging changes after gamma knife radiosurgery for arteriovenous malformations. Int J Radiat Oncol Biol Phys 1992;23:19–26. 6. Ganz JC, Reda WA, Abdelkarim K. Adverse radiation effects after Gamma Knife Surgery in relation to dose and volume. Acta Neurochir (Wien) 2009;151:9–19. 7. Flickinger JC, Kondziolka D, Maitz AH, et al. Analysis of neurological sequelae from radiosurgery of arteriovenous malformations: How location affects outcome. Int J Radiat Oncol Biol Phys 1998;40:273–278. 8. Flickinger JC, Kondziolka D, Lunsford LD, et al. Development of a model to predict permanent symptomatic postradiosurgery injury for arteriovenous malformation patients. Arteriovenous Malformation Radiosurgery Study Group. Int J Radiat Oncol Biol Phys 2000;46:1143–1148. 9. Flickinger JC, Kondziolka D, Pollock BE, et al. Complications from arteriovenous malformation radiosurgery: Multivariate analysis and risk modeling. Int J Radiat Oncol Biol Phys 1997;38:485–490. 10. Pollock BE, Flickinger JC. A proposed radiosurgery-based grading system for arteriovenous malformations. J Neurosurg 2002;96:79–85. 11. Andrade-Souza YM, Zadeh G, Ramani M, et al. Testing the radiosurgery-based arteriovenous malformation score and the modified Spetzler-Martin grading system to predict radiosurgical outcome. J Neurosurg 2005;103:642–648. 12. Yushkevich PA, Piven J, Hazlett HC, et al. User-guided 3D active contour segmentation of anatomical structures: Significantly improved efficiency and reliability. Neuroimage 2006; 31:1116–1128.
13. Flickinger JC, Kondziolka D, Lunsford LD, et al. A multiinstitutional analysis of complication outcomes after arteriovenous malformation radiosurgery. Int J Radiat Oncol Biol Phys 1999;44:67–74. 14. Pollock BE, Lunsford LD, Kondziolka D, et al. Stereotactic radiosurgery for postgeniculate visual pathway arteriovenous malformations. J Neurosurg 1996;84:437–441. 15. Maruyama K, Kamada K, Shin M, et al. Optic radiation tractography integrated into simulated treatment planning for Gamma Knife surgery. J Neurosurg 2007;107:721– 726. 16. Pantelis E, Papadakis N, Verigos K, et al. Integration of functional MRI and white matter tractography in stereotactic radiosurgery clinical practice. Int J Radiat Oncol Biol Phys 2010; 78:257-67. 17. Karlsson B, Lax I, Soderman M. Factors influencing the risk for complications following Gamma Knife radiosurgery of cerebral arteriovenous malformations. Radiother Oncol 1997;43: 275–280. 18. Friedman WA, Bova FJ, Bollampally S, et al. Analysis of factors predictive of success or complications in arteriovenous malformation radiosurgery. Neurosurgery 2003;52:296–307. discussion 307–298. 19. Pollock BE. Occlusive hyperemia: A radiosurgical phenomenon? Neurosurgery 2000;47:1178–1182. discussion 1182–1174. 20. Chapman PH, Ogilvy CS, Loeffler JS. The relationship between occlusive hyperemia and complications associated with the radiosurgical treatment of arteriovenous malformations: Report of two cases. Neurosurgery 2004;55:228–233. discussion 233–224. 21. van den Berg R, Buis DR, Lagerwaard FJ, et al. Extensive white matter changes after stereotactic radiosurgery for brain arteriovenous malformations: A prognostic sign for obliteration? Neurosurgery 2008;63:1064–1069. discussion 1069– 1070. 22. Han JH, Kim DG, Chung HT, et al. Clinical and neuroimaging outcome of cerebral arteriovenous malformations after Gamma Knife surgery: Analysis of the radiation injury rate depending on the arteriovenous malformation volume. J Neurosurg 2008; 109:191–198. 23. Sirin S, Kondziolka D, Niranjan A, et al. Prospective staged volume radiosurgery for large arteriovenous malformations: Indications and outcomes in otherwise untreatable patients. Neurosurgery 2006;58:17–27. discussion 17–27.
doi:10.1016/j.ijrobp.2010.12.014
CLINICAL INVESTIGATION
Central Nervous System Tumor
PRETREATMENT PREDICTORS OF ADVERSE RADIATION EFFECTS AFTER RADIOSURGERY FOR ARTERIOVENOUS MALFORMATION CAROLINE HAYHURST, M.D.,* ERIC MONSALVES, B.SC.,* MONIQUE VAN PROOIJEN, PH.D.,y MICHAEL CUSIMANO, M.D.,z MAY TSAO, M.D.,x CYNTHIA MENARD, M.D.,k ABHAYA V. KULKARNI, M.D., PH.D.,{ MICHAEL SCHWARTZ, M.D.,** AND GELAREH ZADEH, M.D., PH.D.* *Gamma Knife Unit, Division of Neurosurgery, University Health Network, yPhysics Department, Princess Margaret Hospital, Division of Neurosurgery, St Michael’s Hospital, Toronto, Canada; xRadiation Oncology Program, Sunnybrook Hospital, kRadiation Oncology Program, Princess Margaret Hospital, {Division of Neurosurgery, Hospital for Sick Children, **Division of Neurosurgery, Sunnybrook Hospital, University of Toronto, Canada
z
Purpose: To identify vascular and dosimetric predictors of symptomatic T2 signal change and adverse radiation effects after radiosurgery for arteriovenous malformation, in order to define and validate preexisting risk models. Methods and Materials: A total of 125 patients with arteriovenous malformations (AVM) were treated at our institution between 2005 and 2009. Eighty-five patients have at least 12 months of clinical and radiological follow-up. Any new-onset headaches, new or worsening seizures, or neurological deficit were considered adverse events. Follow-up magnetic resonance images were assessed for new onset T2 signal change and the volume calculated. Pretreatment characteristics and dosimetric variables were analyzed to identify predictors of adverse radiation effects. Results: There were 19 children and 66 adults in the study cohort, with a mean age of 34 (range 6–74). Twenty-three (27%) patients suffered adverse radiation effects (ARE), 9 patients with permanent neurological deficit (10.6%). Of these, 5 developed fixed visual field deficits. Target volume and 12 Gy volume were the most significant predictors of adverse radiation effects on univariate analysis (p < 0.001). Location and cortical eloquence were not significantly associated with the development of adverse events (p = 0.12). No additional vascular parameters were identified as predictive of ARE. There was a significant target volume threshold of 4 cm3, above which the rate of ARE increased dramatically. Multivariate analysis target volume and the absence of prior hemorrhage are the only significant predictors of ARE. The volume of T2 signal change correlates to ARE, but only target volume is predictive of a higher volume of T2 signal change. Conclusions: Target volume and the absence of prior hemorrhage is the most accurate predictor of adverse radiation effects and complications after radiosurgery for AVMs. A high percentage of permanent visual field defects in this series suggest the optic radiation is a critical radiosensitive structure. Ó 2012 Elsevier Inc. Stereotactic radiosurgery, Arteriovenous malformation, Complications, Radiation injury.
unclear. Although radiation injury is known to be a function of dose and volume, the extent to which the vascular morphology of the nidus contributes to adverse radiation effects (ARE) and the interaction with radiation dosimetry characteristics is unclear. Previous multi-institutional studies led by the University of Pittsburgh and the Arteriovenous Malformation Radiosurgery Study Group have proposed a model of symptomatic postradiation injury based on the 12-Gy volume and nidus location proposing that the presence of clinical sequelae of T2 signal change is primarily related to location (8, 9). Flickinger et al. developed a risk model of AVM location, stratified into four tiers, with the frontal lobe representing the least risk
INTRODUCTION Radiosurgery provides an effective therapeutic option for selected brain arteriovenous mal-formations (AVM), achieving complete obliteration in 60–90% of cases with higher obliteration rates of >90% in smaller lesions (1–3). However, new areas of T2 signal change denoting radiation effects in normal surrounding brain are frequently seen. Such imaging changes occur much more frequently after radiosurgery for AVM than for other pathologies (4), occurring in 30–60% of cases (5, 6). Although white matter imaging changes are frequent, only a proportion will develop clinical complications after treatment (6, 7). The reasons for this discrepancy in response to radiosurgery in AVMs are Reprint requests to: Dr. Gelareh Zadeh, M.D., Ph.D., University of Toronto, Department of Neurosurgery, 399 Bathurst St., 4-439 West Wing, Toronto, Ontario, Canada M5G 2S8. Tel: (416) 6035679; Fax: (416) 603-5298; E-mail: [email protected]
Conflicts of interest: none. Received July 28, 2010, and in revised form Nov 24, 2010. Accepted for publication Dec 3, 2010. 803
I. J. Radiation Oncology d Biology d Physics
804
Volume 82, Number 2, 2012
Table 1. Significant postradiosurgery injury expression score (SPIE) (Flickinger et al., 2000)
Table 2. Patient characteristics and univariate analysis Patient characteristics
SPIE score
Age 1 2 3 4
Value
p value
Median 36 (range, 6–74) M 45 F 40 37 (43.5%) 22 (25.9%) 6 (7%) 13 (15.3%) 5 (5.9%) 1 17 (20%) 2 24 (28.2%) 3 37 (43.5%) 4 7 (8.2%) 50 40 (47.1%)
0.99
Location Frontal Cerebellum Temporal Parietal Occipital Basal ganglia Medulla Thalamus Intraventricular Pons/midbrain Corpus Callosum
and brainstem location the highest risk, denoted the significant postradiosurgery injury expression (SPIE) score (Table 1) (7). Although these models take into account radiation dose, they do not include the AVM vascular characteristics. Similarly, the radiosurgery-based AVM grading system was developed to predict the chance of complete obliteration without new deficit and is based on factors associated with a successful outcome (10). It incorporates location stratified into three tiers of frontal/temporal, parietal/occipital/corpus callosum, or basal ganglia and brainstem, with age and nidus volume. This score has been validated as a predictor of outcome, but includes obliteration and deficit from posttreatment hemorrhage as dependent variables (10, 11). In summary, the radiosurgery-based AVM score is a function of lesion volume and location and does not take into account additional AVM vascular characteristics or radiosurgical dosimetry variables. The aim of the current study was to identify vascular parameters inherent in the AVM, in addition to radiosurgery treatment characteristics that could predict ARE, to develop a clinically relevant tool for pretreatment patient counseling and clinical decision-making. Additionally, identifying predictive factors of ARE can provide important insights into the impact of radiation on the central nervous system. METHODS AND MATERIALS
Sex No previous hemorrhage Previous treatment: Surgery Embolization Radiosurgery (linac) Spetzler Martin grade
Eloquent location Venous drainage Deep Superficial Arterial supply Deep perforator Cortical Location (SPIE):
– 0.005 0.979
0.04
0.12 0.21
50 (58.8%) 35 (41.2%) 0.34
Maximum diameter (cm) Volume T2 signal change (cm3) Time to maximum T2 change
29 (34.1%) 56 (65.9%) 1 16 2 29 3 11 4 29 Median 2.42 cm (range, 0.1–5.13 cm) 11.65 (0.53–127.3)
0.12
– 0.001
12 months (5–36)
0.091
Analysis variables The clinical data and pretreatment imaging were reviewed for each patient to define pretreatment characteristics to be used for the prediction analysis. Table 2 demonstrates the variables included. Previous hemorrhage, previous treatment, and SpetzlerMartin grade were included. The presence of deep venous drainage and deep arterial perforator supply were recorded separately. The dosimetric variables studied include target volume, 12 Gy volume, maximum dose (Dmax), V100, and the Radiation Therapy Oncology Group Conformity Index (Table 3). All dosimetric values are calculated on the day of treatment and maintained prospectively in a locally developed web publishing database.
Patients
T2 signal change
In a retrospective review of patients with AVM treated between 2005 and 2009 at the University of Toronto with a Model 4C Gamma Knife facility (Elekta Instruments, Atlanta, GA), patients with a minimum of 12 months of complete clinical and radiological follow-up were included in this study of adverse radiation effects. Approval for this study was granted by the University Health Network research ethics board. All treatments were planned using stereotactic CT angiography, magnetic resonance angiography and two-dimensional digital subtraction angiogram. In our institution, a single dose of 25 Gy is prescribed to the edge of the nidus, but is reduced to 20 Gy for treatment volumes >4 cm3 and further reduced to 15 Gy for lesions in eloquent cortex. Clinical follow-up is undertaken at 3 months. Thereafter, clinical and magnetic resonance imaging follow-up are obtained 6 months, 12 months, and then yearly.
All follow-up imaging was reviewed for the presence of T2 signal change after treatment. Where new T2 signal change occurred, Table 3. Dosimetry variables and univariate analysis Dosimetry variables
Value (median/range) 3
Target volume (cm ) 12 Gy volume Prescription dose Isocenter number D max V100 CI (Radiation Therapy Oncology Group)
3
2.210 cm (0.1–14.6) 6.79 cm3 (0.38–41.01) 20 Gy (15–25 Gy) 10 (1–27) 40 (30–71) 98.59 (88.31–100) 1.54 (1.20–4.56)
p value <0.001 <0.001 0.062 0.04 0.20 0.187 0.41
Predictors of Radiation Affects on Arteriovenous Malformation d C. HAYHURST et al.
the time to onset in months was recorded and the volume of T2 signal change calculated on DICOM FLAIR images using ITKSNAP software (Fig. 1) (12). The time to maximal volume of T2 signal change was recorded.
Adverse radiation effects Clinical data were recorded for all patients. Clinical adverse radiation effects were defined as severe headaches requiring steroid administration, new or worsening seizures, and new focal neurological deficit. No patients experienced new or repeat hemorrhage after radiosurgery in our series. Adverse radiation effects were classified as transient or permanent when still present at last follow-up.
Statistical analysis Univariate logistic regression analysis was used and pretreatment variables with p < 0.10 were then included in a multivariate analysis, after excluding highly collinear variables. Threshold analysis was performed by calculating the area under a receiver operating characteristic curve (AUC) for outcome. All analyses were performed with SPSS version 17 (SPSS Inc., Chicago, IL).
RESULTS Patient characteristics Eighty-eight patients with AVM were treated at our institution between 2005 and 2009. Eight-five patients were included in this study. Three patients were excluded for lack of complete radiological follow-up. There were 66 adults, and 19 children age <18. Table 2 demonstrates the patient’s pretreatment characteristics. Forty-eight (56.5%) had presented with hemorrhage. 22 (25.9%) patients had undergone previous treatment, including surgery, embolization, or linac radiosurgery. The median duration of follow-up is 26 months (range, 12–49 months). No patients experienced posttreatment hemorrhage in the follow-up period. Twenty-three (27%) patients developed adverse radiation effects. Fourteen (16.5%) developed transient complications including headaches, new seizures, and new focal neurological deficit. Nine (10.6%) patients had permanent neurological deficit after treatment: 4 patients had mild hemiparesis
805
and 5 (5.8%) patients had fixed visual field defects. Of those with visual field defects, only one AVM was located within the calcarine cortex. The median time to onset of ARE was 6 months (0.5–36 months). T2 signal change Forty-two (49.4%) patients developed areas of new T2 signal change after treatment. The median time to onset of T2 signal change was 12 months (range, 5–36 months). The median volume of T2 signal change was 11.66 cm3 (0.53–127.3). Increasing volume of T2 signal change is associated with ARE. Those who developed ARE had a median T2 signal change volume of 34.4 cm3 vs. 5.68 cm3 in those without ARE (p = 0.001, OR 1.05). The only significant predictor of the volume of T2 signal change is target volume (p = 0.002). The presence of deep venous drainage, deep perforator arterial supply, and location were not significant. There was no significant association between ARE and early (<12 months) or late (>12 months) onset T2 signal change. Predictors of adverse effects Both transient and permanent ARE were included as the dependent variable for the analysis. On univariate analysis, no prior hemorrhage, target volume, 12 Gy volume, number of isocenters, and Spetzler Martin grade were significant (Tables 2 and 3). However, the significance of the Spetzler Martin grade is related to nidus size because when analyzed separately, deep venous drainage, and eloquent location are not significant. Age at treatment and previous treatment, interestingly including previous radiosurgery, were not significant in this study. The presence of deep perforator arterial supply was also not significant. No additional dosimetric variables contributed to ARE. Location as categorized by the SPIE score or as independent variables were not a significant predictor of ARE in this series (p > 0.12). The category with the highest risk for ARE was SPIE 3, corresponding to occipital and basal ganglia (OR 4.17 [0.91–19.18]).
Fig. 1. T2 signal change: (a) T2 weighted magnetic resonance image acquired on day of stereotactic radiosurgery, (b) new onset T2 signal change at 6 months.
806
I. J. Radiation Oncology d Biology d Physics
Volume 82, Number 2, 2012
ARE. No children had permanent ARE. There is no significant difference in the rate of T2 signal change between adults and children or the time to onset of T2 signal change. The median volume of T2 signal change in children is 14.04 cm3, compared with 13.281 cm3 in adults (p = 0.911). The median target volume in children was 1.67 cm3, which is not significantly different to that in adults (p = 0.212). DISCUSSION
Fig. 2. Incidence of adverse radiation effects according to target volume for stereotactic radiosurgery.
The median target volume in those with ARE is 6.49 cm3 and 2.76 cm3 in those without ARE. Using the AUC to determine significant target volume thresholds, there is a significant threshold of 4 cm3 (AUC 0.77, Fig. 2). 55 AVMs had a target volume of less than 4 cm3 and 6 (10.9%) developed ARE, 30 AVMs were >4 cm3 in volume, and 17 (56.7%) developed ARE. On multivariate analysis of pretreatment variables (excluding 12 Gy volume because it can only be calculated at time of treatment and is correlated to treatment volume), only a treatment volume greater than 4 cm3 and the absence of prior hemorrhage are significant (p < 0.001 OR 9.14 [2.90–28.79] and p = 0.036 OR 3.43 [1.09–10.85], respectively; Fig. 3).
Pediatric AVM Nineteen children were included in this study, representing 22% of the study cohort. Seventeen children (89.4%) presented with hemorrhage. Seven children (36.8%) developed new T2 signal change and 4 developed transient
This study is the first to demonstrate that a target volume threshold exists as a predictor of ARE after radiosurgery for AVMs, independent of location, in patients without a history of hemorrhage. The significant volume threshold of 4 cm3 was seen in this series. There were no vascular parameters, such as venous or arterial anatomy, or nidus morphology and no dosimetry parameters that contributed significantly to ARE. Radiosurgical treatment planning for AVM balances the radiation dose required to achieve nidus obliteration with a dose that risks radiation-induced injury to surrounding normal brain. Knowledge of additional predictors of complications allows for more accurate patient counselling regarding management options and potentially alteration of the radiosurgical plan to minimize complications. In an initial singleinstitution analysis of imaging changes after AVM radiosurgery, Flickinger et al. (5) elegantly demonstrated an actuarial risk of developing imaging changes of 31% at 2 years, with the risk of symptomatic imaging changes 14% at 2 years. Symptomatic imaging changes were more frequent in brainstem locations compared with cortical location. In multivariate analysis only treatment volume was significantly associated with symptomatic imaging changes. However, the study concludes that AVM location determines whether imaging changes are symptomatic or not. Subsequent studies by the University of Pittsburgh Group defined the 12-Gy volume as a significant predictor of symptomatic imaging changes (9). A 12-Gy volume has become the standard for complication
Fig. 3. Multivariate analysis: (a) boxplot depicting target volume for patients with and without adverse radiation effects. (b) Incidence of adverse radiation effects in patients with and without prior hemorrhage.
Predictors of Radiation Affects on Arteriovenous Malformation d C. HAYHURST et al.
807
Fig. 4. Postradiosurgery visual field defect: (a) treatment planning magnetic resonance image showing occipital white matter arteriovenous malformations, (b) extent of occipital white matter T2 signal change 6 months after radiosurgery.
reporting in the radiosurgical literature and, although it is significant in our study, it is of less relevance when considering pretreatment evaluation and risk stratification. Flickinger et al. subsequently devised and validated in a large multiinstitutional series, a model of 12-Gy volume and location for predicting ARE (7, 8, 13). In our series, we were unable to correlate location to the development of ARE, although we have included both transient and permanent ARE in our analysis. In this series, six brainstem AVMs were treated with target volumes between 0.6 cm3 and 2.6 cm3 and none developed complications. Eloquent location was not found to be associated with ARE in our series, although 47.1% of the AVMs treated involved eloquent cortex. A critical finding is that of the nine occipital AVMs treated, five (55.5%) developed visual field defects which are permanent at last follow-up. Only one of these involved eloquent cortex and the remainder are located in deep white matter, suggesting that the optic radiation is highly radiosensitive. Pollock et al. (14) reported 34 patients with AVM within the optic radiation or striate cortex, 27 of whom had normal vision before radiosurgery. Two (6%) developed documented visual field defects after treatment. Consequently, there is increasing interest to integrate diffusion tensor tractography into radiosurgical planning. Maruyama et al. (15) retrospectively integrated optic radiation tractography to the treatment plan of 10 patients treated for AVM located in the postgeniculate visual pathways and demonstrated a maximum dose of less than 12 Gy to the optic radiation did not cause any new visual symptoms. In a study of integrated tractography in cyberknife radiosurgery planning, Pantelis et al. (16) incorporated optic radiation tractography into the optimization process, demonstrating a dose reduction from 2,000 cGy to less than 1,200 cGy to the optic radiation when included in the treatment plan. Despite our initial hypothesis that additional dosimetry or vascular parameters would provide a more accurate prediction
model for ARE, we were unable to demonstrate in a multivariate analysis any additional significant variables in radiosurgical treatment planning. Flickinger et al. did not demonstrate any dosimetry characteristics predictive of ARE, including Dmax, isodose, and dose rate (5, 7–9, 13). Interestingly, prior hemorrhage was not associated with a reduced risk of complications in these studies, in contrast to previous reports and our data (17). An analysis of ARE by Friedman et al. (18), demonstrated a reduced rate of transient complications with improved conformality. The addition of vascular anatomical parameters to our analysis did not show any association with the development of ARE. Isolated case reports have suggested early occlusion of dominant draining veins contributes to perinidal edema and complications (19, 20). Van den Berg et al. (21) reported a higher volume of T2 signal change in those with AVMs drained by a single vein; however, there is no comparison to those without new T2 signal change or association with clinical ARE. It is clear that AVM nidus volume is a determinant of ARE after treatment. In our study, a target volume threshold of 4 cm3 means a large number of AVMs would be considered high risk. However, dose reduction to achieve lower complication rates is associated with significantly lower obliteration rates. Han et al. (22) reported a 12.5% obliteration rate for AVM >14 cm3 treated with a lowest dose of 10 Gy, with a rate of 9.5% postradiosurgery imaging changes. An alternative to balance an effective dose for obliteration with an acceptable risk of complications is staged radiosurgical treatment. Sirin et al. (23), published results of 28 patients with very large volume AVM (>15 mL) treated with 2 or 3 stage radiosurgery. Fourteen patients had more than 36 months’ follow-up with a 50% complete obliteration rate; however, 14% suffered hemorrhage in the follow-up period, appearing to confirm the theoretical risk of increased hemorrhage rates with partial obliteration. Further reports on the outcome of staged
808
I. J. Radiation Oncology d Biology d Physics
radiosurgery for large AVM are required to elucidate the efficacy of this treatment modality. Limitations of the study Our data represent a single-institution retrospective analysis and require validation in a prospective series. However, we would have expected to validate the University of Pittsburgh model regarding location in our patients. We included both transient and permanent neurological sequelae in our study, but it is possible that the location effects are different when permanent deficits alone are studied. Our study numbers remain relatively small, so our analyses might have failed to demonstrate significance for certain predictor variables simply because of lack of statistical power.
Volume 82, Number 2, 2012
CONCLUSION Increasing AVM nidus volume, without a history of prior hemorrhage, is a significant predictor of ARE after radiosurgery, with a volume threshold of 4 cm3, above which complications are more likely to occur. This volume effect appears to be independent of eloquent location, but deep white matter tracts, particularly the optic radiation, are likely to be important in the development of ARE. The optic radiation is most likely more sensitive to radiation injury and a better understanding of the radiation response in the white matter tracts is needed. Further studies integrating tractography into radiosurgical treatment planning will further clarify the role of location in the development of ARE after radiosurgery for AVM.
REFERENCES 1. Karlsson B, Lindquist C, Steiner L. Prediction of obliteration after gamma knife surgery for cerebral arteriovenous malformations. Neurosurgery 1997;40:425–430. discussion 430–421. 2. Flickinger JC, Pollock BE, Kondziolka D, et al. A dose-response analysis of arteriovenous malformation obliteration after radiosurgery. Int J Radiat Oncol Biol Phys 1996;36:873–879. 3. Sun DQ, Carson KA, Raza SM, et al. The radiosurgical treatment of arteriovenous malformations: Obliteration, morbidities, and performance status. Int J Radiat Oncol Biol Phys 2010 Apr 16. [Epub ahead of print]. 4. Flickinger JC, Kondziolka D, Kalend AM, et al. Radiosurgeryrelated imaging changes in surrounding brain: Multivariate analysis and model evaluation. Radiosurgery 1996;1:229–236. 5. Flickinger JC, Lunsford LD, Kondziolka D, et al. Radiosurgery and brain tolerance: An analysis of neurodiagnostic imaging changes after gamma knife radiosurgery for arteriovenous malformations. Int J Radiat Oncol Biol Phys 1992;23:19–26. 6. Ganz JC, Reda WA, Abdelkarim K. Adverse radiation effects after Gamma Knife Surgery in relation to dose and volume. Acta Neurochir (Wien) 2009;151:9–19. 7. Flickinger JC, Kondziolka D, Maitz AH, et al. Analysis of neurological sequelae from radiosurgery of arteriovenous malformations: How location affects outcome. Int J Radiat Oncol Biol Phys 1998;40:273–278. 8. Flickinger JC, Kondziolka D, Lunsford LD, et al. Development of a model to predict permanent symptomatic postradiosurgery injury for arteriovenous malformation patients. Arteriovenous Malformation Radiosurgery Study Group. Int J Radiat Oncol Biol Phys 2000;46:1143–1148. 9. Flickinger JC, Kondziolka D, Pollock BE, et al. Complications from arteriovenous malformation radiosurgery: Multivariate analysis and risk modeling. Int J Radiat Oncol Biol Phys 1997;38:485–490. 10. Pollock BE, Flickinger JC. A proposed radiosurgery-based grading system for arteriovenous malformations. J Neurosurg 2002;96:79–85. 11. Andrade-Souza YM, Zadeh G, Ramani M, et al. Testing the radiosurgery-based arteriovenous malformation score and the modified Spetzler-Martin grading system to predict radiosurgical outcome. J Neurosurg 2005;103:642–648. 12. Yushkevich PA, Piven J, Hazlett HC, et al. User-guided 3D active contour segmentation of anatomical structures: Significantly improved efficiency and reliability. Neuroimage 2006; 31:1116–1128.
13. Flickinger JC, Kondziolka D, Lunsford LD, et al. A multiinstitutional analysis of complication outcomes after arteriovenous malformation radiosurgery. Int J Radiat Oncol Biol Phys 1999;44:67–74. 14. Pollock BE, Lunsford LD, Kondziolka D, et al. Stereotactic radiosurgery for postgeniculate visual pathway arteriovenous malformations. J Neurosurg 1996;84:437–441. 15. Maruyama K, Kamada K, Shin M, et al. Optic radiation tractography integrated into simulated treatment planning for Gamma Knife surgery. J Neurosurg 2007;107:721– 726. 16. Pantelis E, Papadakis N, Verigos K, et al. Integration of functional MRI and white matter tractography in stereotactic radiosurgery clinical practice. Int J Radiat Oncol Biol Phys 2010; 78:257-67. 17. Karlsson B, Lax I, Soderman M. Factors influencing the risk for complications following Gamma Knife radiosurgery of cerebral arteriovenous malformations. Radiother Oncol 1997;43: 275–280. 18. Friedman WA, Bova FJ, Bollampally S, et al. Analysis of factors predictive of success or complications in arteriovenous malformation radiosurgery. Neurosurgery 2003;52:296–307. discussion 307–298. 19. Pollock BE. Occlusive hyperemia: A radiosurgical phenomenon? Neurosurgery 2000;47:1178–1182. discussion 1182–1174. 20. Chapman PH, Ogilvy CS, Loeffler JS. The relationship between occlusive hyperemia and complications associated with the radiosurgical treatment of arteriovenous malformations: Report of two cases. Neurosurgery 2004;55:228–233. discussion 233–224. 21. van den Berg R, Buis DR, Lagerwaard FJ, et al. Extensive white matter changes after stereotactic radiosurgery for brain arteriovenous malformations: A prognostic sign for obliteration? Neurosurgery 2008;63:1064–1069. discussion 1069– 1070. 22. Han JH, Kim DG, Chung HT, et al. Clinical and neuroimaging outcome of cerebral arteriovenous malformations after Gamma Knife surgery: Analysis of the radiation injury rate depending on the arteriovenous malformation volume. J Neurosurg 2008; 109:191–198. 23. Sirin S, Kondziolka D, Niranjan A, et al. Prospective staged volume radiosurgery for large arteriovenous malformations: Indications and outcomes in otherwise untreatable patients. Neurosurgery 2006;58:17–27. discussion 17–27.