Radiosurgery for unruptured cerebral arteriovenous malformations in pediatric patients

Acta Neurochir (2015) 157:281 291 DOI 10.1007/s00701-014-2305-4 CLINICAL ARTICLE - VASCULAR Radiosurgery for unruptured cerebral arteriovenous malformations in pediatric patients Dale Ding & Zhiyuan Xu & Chun-Po Yen & Robert M. Starke & Jason P. Sheehan Received: 13 November 2014 /Accepted: 3 December 2014 /Published online: 17 December 2014 # Springer-Verlag Wien 2014 Abstract Background Unruptured cerebral arteriovenous malformations (AVMs) in pediatric patients (age <18 years) were excluded from A Randomized Trial of Unruptured AVMs (ARUBA). Therefore, the efficacy of radiosurgery for unruptured pediatric AVMs is poorly understood. The goal of this study is to determine the outcomes and define the predictors of obliteration following radiosurgery for unruptured AVMs in pediatric patients. Methods We evaluated a prospective database, from 1989 to 2013, of AVM patients treated with radiosurgery at our institution. Patients with age less than 18 years at the time of radiosurgery, unruptured nidi, and at least 2 years of radiologic follow-up or AVM obliteration were selected for analysis. Statistical analyses were performed to determine actuarial obliteration rates and identify factors associated with obliteration. Results In the 51 unruptured pediatric AVM patients included for analysis, the median age was 13 years, and the most common presentation was seizure in 53 %. The median nidus volume and radiosurgical margin dose were 3.2 cm 3 and 21.5 Gy, respectively. The median radiologic follow-up was 45 months. The actuarial AVM obliteration rates at 3, 5, and 10 years were 29 %, 54 %, and 72 %, respectively. In the multivariate Cox proportional hazards regression analysis, higher margin dose (P=0.002), fewer draining veins (P= 0.038), and lower Virginia Radiosurgery AVM Scale (P= 0.003) were independent predictors of obliteration. Obliteration rates were significantly higher with a margin dose of at least 22 Gy (P=0.003) and for nidi with 2 or fewer draining veins (P=0.001). The incidences of radiologically evident, symptomatic, and permanent radiation-induced changes were D. Ding: Z. Xu : C.<P. Yen : R. M. Starke : J. P. Sheehan (*) Department of Neurosurgery, University of Virginia, P.O. Box 800212, Charlottesville, VA 22908, USA e-mail: jps2f@virginia.edu 55 %, 16 %, and 2 %, respectively. The annual postradiosurgery hemorrhage rate was 1.3 %, and the incidence of post-radiosurgery cyst formation was 2 %. Conclusion Radiosurgery affords a favorable risk to benefit profile for unruptured pediatric AVMs. Pediatric patients with unruptured AVMs merit further study to define an optimal management approach. Keywords Gamma knife. Intracranial arteriovenous malformation. Pediatric. Radiosurgery. Vascular Malformations Introduction Cerebral arteriovenous malformations (AVMs) are congenital lesions which typically do not clinically manifest until the 3rd or 4th decades of life [1, 2]. A Randomized Trial of Unruptured Brain AVMs (ARUBA) was a prospective, randomized controlled trial which showed superior outcomes with medical management compared to intervention for patients with unruptured AVMs [29]. However, pediatric patients (age less than 18 years) were specifically excluded from ARUBA. Although the upper age limit for the pediatric population differs across regions throughout the world, we used an upper age limit of 18 years, since this is widely considered the upper age boundary for the classification of a pediatric patient in the United States (the location of our study) and the lower age boundary of ARUBA was 18 years. Given the relatively longer time period over which pediatric AVM patients are exposed to the risk of hemorrhage compared with their adult counterparts, it is unclear if the results of ARUBA are generalizable to the pediatric population. Stereotactic radiosurgery is an effective treatment for AVMs and provides a minimally invasive alternative to surgical resection, especially for deep-seated or eloquent nidi [13,

282 Acta Neurochir (2015) 157:281 291 17, 18, 30, 37]. Previous studies of pediatric AVMs treated with radiosurgery have included a significant proportion of nidi with prior hemorrhage, which makes it difficult to distinguish the outcomes of unruptured pediatric AVMs from ruptured ones [5, 25, 40, 54]. Given that prior AVM hemorrhage may affect radiosurgery outcomes, a study evaluating the outcomes following radiosurgery for unruptured AVMs in pediatric patients appears warranted [10, 15, 30, 52]. In this retrospective cohort study, our aims are: (1) to determine the outcomes following treatment of unruptured pediatric AVMs with radiosurgery, and (2) to define the predictors of obliteration and complications following radiosurgery for unruptured pediatric AVMs. Methods Patient selection We performed a retrospective evaluation of a prospectively collected, institutional review board (IRB) approved, database of over 1,400 AVM patients who were treated with Gamma Knife radiosurgery at the University of Virginia from 1989 to 2013. The inclusion criteria were: (1) AVMs without prior hemorrhage, (2) patient age less than 18 years at the time of radiosurgery, (3) sufficient data regarding baseline patient and AVM characteristics as well as follow-up clinical and radiologic outcomes, (4) complete nidus obliteration on either angiography or magnetic resonance imaging (MRI), and (5) minimum follow-up duration of 2 years for patients without nidus obliteration. Patients treated with volume- or dosestaged radiosurgery were excluded. Baseline data and variables The following baseline data were extracted from directed chart review: (1) patient characteristics, (2) AVM angioarchitectural features, and (3) radiosurgery parameters. Patient variables included gender, age, and presenting symptoms. AVM angioarchitectural features included prior interventions (surgical resection and/or embolization), size (maximum diameter and volume), location (eloquent versus non-eloquent, superficial versus deep), venous anatomy (number of draining veins, superficial only versus deep component), and presence of associated aneurysms. Eloquent location was defined as sensorimotor, language, and visual cortex, hypothalamus and thalamus, internal capsule, brainstem, cerebellar peduncles, and deep cerebellar nuclei [45]. Deep location was defined as basal ganglia, thalamus, and brainstem [51]. Based on the aforementioned variables, the Spetzler-Martin grade, modified radiosurgery-based AVM score (RBAS), and Virginia Radiosurgery AVM Scale (VRAS) were determined for each nidus [45, 49, 51]. The modified RBAS is the weighted sum of patient age (multiplied by 0.02), nidus volume (multiplied by 0.1), and deep (1 point) versus superficial (0 points) location (multiplied by 0.5) [46]. The VRAS is comprised of nidus volume (<2 cm 3, 0 points; 2 4cm 3 ; 1 point; >4 cm 3, 2 points), eloquent (1 point) versus non-eloquent location (0 points), and prior AVM hemorrhage (1 point) or lack thereof (0 points) [44]. Our Gamma Knife radiosurgery procedure has been previously described [50]. Prior to 1991, magnetic resonance imaging (MRI) was not readily available and was therefore not routinely used in addition to angiography for treatment planning. After 1991, MRI, angiography and, occasionally, computed tomography (CT) were used in order to improve the spatial accuracy of radiosurgical planning. From 1989 to June 1994, the Kula software was used for dose planning, whereas after June 1994, the Gamma Plan software (Elekta, Stockholm, Sweden) was used for dose planning. There are a number of different proposed methods for the calculation of conformity index (CI) [34]. Since some treatment plans date back to earlier versions of Gamma Plan software which are no longer readily available, the conformity and gradient indices could not be computed for these cases. Therefore, routine calculation of CI was not performed in our Gamma Knife planning, particularly prior to 2004. In recent cases, CIs of less than 1.5 but certainly less than 2 were deemed acceptable [22]. At times, conformity adjacent to embolized material was deemed less critical than failing to achieve full coverage of the remaining nidus. Significant interobserver variations exist, even amongst experts in the field, in the delineation of the target AVM volume on catheter digital subtraction angiography, MRI, and MR angiography [7, 8]. Thus, delineation of the target (in this case the AVM nidi), as well as the conformity indices derived from it, are comprised of both an art and science to stereotactic radiosurgery. However, in general, conformity and gradient indices, as well as margin dose to the nidus and dose to critical adjacent structures, were carefully optimized by the collaborative effort of a neurosurgeon, medical physicist, and radiation oncologist. The radiosurgery variables included prescription dose, maximum dose, isodose line, and number of isocenters. Follow-up and outcomes Routine radiologic follow-up after radiosurgery consisted of serial MRIs every 6 months for the first 2 years, followed by MRIs annually after 2 years follow-up. Patients with neurological decline in the post-treatment period underwent additional neuroimaging by CT or MRI. All follow-up imaging was reviewed by a neurosurgeon and neuroradiologist at our institution, regardless of where it was obtained. Due to the relatively increased susceptibility of pediatric patients to radiation-induced neoplasms compared with their adult counterparts and the widespread availability of MRI and catheter angiography for a significant majority of the treatment period,

Acta Neurochir (2015) 157:281 291 283 CT angiography (CTA) was generally avoided for treatment planning or radiologic follow-up in this pediatric population. AVM obliteration was defined by the lack of flow voids on MRI or by the absence of abnormal arteriovenous shunting on angiography. Angiography was performed to confirm obliteration diagnosed by MRI or to evaluate a residual nidus for further intervention(s). Radiation-induced changes (RICs) were defined by perinidal T2-weighted hyperintensities on follow-up MRI. The time interval between radiosurgery and the onset of RICs and the duration of RICs were noted. RIC accompanied by new or worsening neurological status was classified as symptomatic RIC. Symptomatic RIC without resolution of the associated neurological deterioration over the course of follow-up was defined as permanent RIC. Latency period hemorrhage was defined by follow-up CT or MRI as any AVM hemorrhage after radiosurgery, regardless of the patient s neurological status. Post-radiosurgery cyst formation was defined by MRI or CT as the development of a cystic cavity within or adjacent to the initial AVM nidus. Clinical follow-up consisted of clinic and hospital records from our institution, records from outside hospitals and referring institutions, and correspondence with patients local physicians. The patient s neurological condition at the most recent clinical follow-up was compared to the baseline neurological status at the time of radiosurgery and then classified as improved, unchanged, or deteriorated. For patients with seizures at presentation, the seizure status at the most recent clinical follow-up was compared to the baseline seizure status at the time of radiosurgery and then categorized as follows: (1) seizure remission was defined as the complete absence of seizures, (2) seizure improvement was defined as decreased seizure frequency or seizure remission, (3) seizure worsening was defined as increase seizure frequency or intensity, or (4) unchanged seizure status. De novo seizures were defined as the onset of seizures after radiosurgery in patients without seizures at presentation. Statistical analysis Statistical analyses were performed with the IBM SPSS 20 software program. All statistical tests were two-sided. Statistical significance was defined as a P value less than 0.05. Data were presented as frequency for categorical variables and as mean with standard deviation (SD) and median with range for continuous variables. Post-radiosurgery radiologic and clinical outcomes were reported as frequencies. Kaplan-Meier analysis was performed to determine the actuarial obliteration rates over time. The log-rank test was used to compare actuarial obliteration rates between different subgroups. The annual post-radiosurgery hemorrhage rate was calculated by dividing the cumulative number of hemorrhages by the cumulative number of risk years, which was sum of the time intervals between radiosurgery and AVM obliteration or most recent radiologic follow-up (for incompletely obliterated AVMs) for each patient. Univariate Cox proportional hazards regression analysis was performed with the aforementioned patient, AVM, and radiosurgery variables to determine factors significantly associated with obliteration. Univariate logistic regression analysis was performed with the same variables to determine factors significantly associated with RIC. Due to the low number of post-radiosurgery hemorrhage events and cysts, analysis for factors associated with their occurrences was not performed. Interaction and confounding was assessed through stratification and relevant expansion covariates. Factors with a P value less than 0.20 in the univariate analyses were entered into a multivariate analysis to determine independent predictors of obliteration (Cox proportional hazards) and RIC (logistic). Results Patient characteristics and AVM angioarchitectural features A total of 51 unruptured pediatric AVMs met study eligibility criteria and were included for analysis. Table 1 details the patient demographics and clinical characteristics. The median age at the time of radiosurgery was 13.3 years (range, 4.7 17.8 years), and 27 patients were male (52.9 %). Prior AVM interventions included embolization in 14 patients (27.5 %) and surgical resection in three patients (5.9 %). The most common presenting symptoms were seizure in 27 patients (52.9 %). The Table 1 AVMs Clinical characteristics of pediatric patients with unruptured Gender Male 27 (52.9 %), female 24 (47.1 %) Age (years) Mean±SD, 12.6±3.6; median, 13.3; range, 4.7-17.8 Prior embolization 14 (27.5 %) Prior surgical resection 3 (5.9 %) Presenting symptom Seizure 27 (52.9 %) Headache 11 (21.6 %) Asymptomatic 6 (11.8 %) Focal neurological deficit 4 (7.8 %) Other a 3(5.9%) Radiologic follow-up (months) Clinical follow-up (months) Mean±SD, 61.0±41.0; median, 45.1; range, 12.5-167.6 Mean±SD, 78.7±51.8; median, 69.9; range, 8.5-196.4 a Other presenting symptoms were hydrocephalus, altered mental status, and proptosis in one patient each

284 Acta Neurochir (2015) 157:281 291 median radiologic and clinical follow-up durations after radiosurgery were 45.1 months (range, 12.5 167.6 months) and 69.9 months (range, 8.5 196.4 months), respectively. Table 2 details the AVM angioarchitectural features. The median nidus diameter and volume were 2.3 cm (range, 1.0 6.9 cm) and 3.2 cm 3 (range, 0.3 15.6 cm 3 ), respectively. The most common AVM locations were frontal in 11 patients (21.6 %), parietal in 10 patients (19.6 %), and temporal and brainstem each in 7 patients (13.7 %). The nidus location was eloquent in 39 patients (76.5 %) and deep in 16 patients (31.4 %). There were 22 nidi with a single draining vein (43.1 %) and 17 nidi with two draining veins (33.3 %). The median number of draining veins was 2 (range, 1 5), and 25 nidi had venous drainage patterns with a deep component (49.0 %). The Spetzler-Martin grade was III or higher in 25 patients (49.0 %), and the VRAS was 2 or higher in 34 patients (66.7 %). The RBAS was less than 1.00 in 41 patients (80.4 %), with a median of 0.76 (range, 0.23 1.88). Radiosurgery parameters Table 3 details the radiosurgical parameters. The median margin dose was 21.5 Gy (range, 14 27 Gy), and the median number of isocenters was 3 (range, 1 20). Repeat radiosurgery was performed for incompletely obliteration AVMs in five patients (9.8 %). At the time of repeat radiosurgery, the median nidus diameter and volume were 1.4 cm (range, 1.3 4.7 cm) and 0.7 cm 3 (0.4 12.7 cm 3 ), respectively. The median radiosurgical parameters were as follows: margin dose 19 Gy Table 2 Nidus angioarchitectural features of pediatric patients with unruptured AVMs Maximum diameter (cm) Mean±SD, 2.6±1.0; median, 2.3; range, 1.0-6.9 Volume (cm 3 ) Mean±SD, 3.9±3.2; median, 3.2; range, 0.3-15.6 Eloquent location 39 (76.5 %) Deep location 16 (31.4 %) Venous drainage pattern Superficial only, 26 (51.0 %); deep component, 25 (49.0 %) Number of draining veins Mean±SD, 1.9±1.1; median, 2; range, 1-5 Associated aneurysms Intranidal, 1 (2.0 %); perinidal, 1(2.0%) Spetzler-Martin grade I, 5 (9.8 %); II, 21 (41.2 %); III, 17 (33.3 %); IV, 7(13.7%); V, 1(2.0%) Radiosurgery-based AVM score Mean±SD, 0.80±0.39; median, 0.76; range, 0.23-1.88 <1.00, 41 (80.4 %); 1.00-1.50, 6 (11.8 %); 1.51-2.00, 4 (7.8 %); >2.00, 0 Virginia radiosurgery AVM scale 0-1, 17 (33.3 %); 2, 19 (37.3 %); 3, 15 (29.4 %); 4, 0 Table 3 Radiosurgery treatment parameters for pediatric patients with unruptured AVMs Margin dose (Gy) Mean±SD, 21.3±2.8; median, 21.5; range, 14-27 Maximum dose (Gy) Mean±SD, 40.0±5.6; median, 40; range, 27-50 Isodose line (Gy) Mean±SD, 54.1±10.7; median, 50; range, 50-90 Number of isocenters (Gy) Mean±SD, 3.6±3.4; median, 3; range, 1-20 (range, 15 23 Gy), maximum dose 38 Gy (range, 30 46 Gy), isodose line 50 % (same for all cases), and number of isocenters 3 (range, 2 20). AVM obliteration following radiosurgery Total AVM obliteration was achieved in 30 patients (58.8 %), including 10 cases documented only by MRI (19.6 %) and 20 cases confirmed by angiography (39.2 %). Based on Kaplan- Meier analysis, the actuarial obliteration rate was 29 % at 3 years, 54 % at 5 years, and 72 % at 10 years (Fig. 1). The median time to obliteration was 50.6 months (95 % CI, 18.9-82.3 months). If only patients with at least 2 years of radiologic follow-up were included, the rates of cumulative, MRI only, and angiographic obliteration were 48.8 % (20/41 patients), 9.8 % (4/41 patients), and 39.0 % (16/41 patients), respectively. In this cohort, there were no cases in which an AVM remnant was found on angiography after complete nidus obliteration was initially diagnosed by MRI. Table 4 details the univariate and multivariate Cox proportional regression analyses for predictors of obliteration. In the univariate analysis, lower nidus volume (P=0.001), higher radiosurgical margin dose (P<0.0001), fewer draining veins (P=0.005), lower RBAS (P=0.003), lower Spetzler-Martin grade (P=0.004), and lower VRAS (P<0.0001) were significantly associated with obliteration. In the multivariate analysis, higher margin dose (P=0.002), fewer draining veins (P= 0.038), and lower VRAS (P=0.003) were found to be independent predictors of obliteration. Notably, prior embolization was not significantly associated with obliteration (P=0.254). Nidi treated with a radiosurgical margin dose of at least 22 Gy were significantly more likely to undergo obliteration (P=0.003, log-rank test; Fig. 2). For AVMs treated with a margin dose of at least 22 Gy, the actuarial obliteration rate was 46 % at 3 years, 71 % at 5 years, and 86 % at 10 years. For AVMs treated with a margin dose of less than 22 Gy, the obliteration rate was 12 % at 3years,39%at5years,and59%at10years. Nidi with two or fewer draining veins were significantly more likely to achieve obliteration following radiosurgery (P=0.001, log-rank test; Fig. 3). For AVMs with two or fewer draining veins, the actuarial obliteration rate was 39 % at

Acta Neurochir (2015) 157:281 291 285 Fig. 1 Kaplan-Meier plot of AVM obliteration over time following radiosurgery. The obliteration rates at 3, 5, and 10 years were 29 %, 54 %, and 72 %, respectively. The number of patients remaining at each time point is shown under the x-axis 3 years, 66 % at 5 years, and 89 % at 10 years. For AVMs with more than two draining veins, the actuarial obliteration rate was 0 % at 3 years and 18 % at 5 and 10 years. Complications following radiosurgery There was radiologic evidence of RIC in 28 patients (54.9 %). Symptomatic RIC was observed in eight patients (15.7 %), of which seven patients had transient symptoms (13.7 %) one patient had permanent symptoms (2.0 %). The time interval between radiosurgery and the onset of RIC was mean 10.7± 6.8 months and median 7.2 months (range, 2.7 29.6 months). The duration of RIC was mean 22.0±15.0 months and median 15.2 months (range, 5.3 61.5 months). If only patients with at least 2 years of radiologic follow-up were included, the rates of cumulative, symptomatic and permanent RIC were 53.7 % (22/41 patients), 9.8 % (4/41 patients), and 0 % (0/41 patients), respectively. No variables were found to have a P value of less than 0.20 in the univariate logistic regression analysis for factors associated with RIC. Thus, it was not valid to perform a multivariate logistic regression analysis for independent predictors of RIC. Notably, there were no significant associations between prior embolization (P=0.291), nidus volume (P=0.794), Table 4 Univariate and multivariate Cox proportional hazards regression analyses for predictors of obliteration following radiosurgery for unruptured pediatric AVMs. Only factors with P<0.20 in the univariate analysis were listed Factor Univariate Multivariate Hazard Ratio 95 % CI P value Hazard Ratio 95 % CI P value Lower volume 1.449 1.164-1.805 0.001* - - NS Higher margin dose 1.331 1.142-1.552 <0.0001* 1.289 1.099-1.512 0.002* Fewer draining veins 1.838 1.199-2.817 0.005* 1.675 1.030-2.725 0.038* Lower RBAS 6.849 1.949-23.810 0.003* - - NS Lower Spetzler-Martin grade 1.821 1.214-2.732 0.004* - - NS Lower VRAS 1.969 1.389-2.793 <0.0001* 1.852 1.233-2.786 0.003* PNSnot significant (P 0.05), RBASradiosurgery-based AVM score, VRASVirginia Radiosurgery AVM scale *P<0.05, statistically significant

286 Acta Neurochir (2015) 157:281 291 Fig. 2 Kaplan-Meier plots of obliteration over time for AVMs treated with a margin dose of at least 22 Gy compared to AVMs treated with a margin dose of less than 22 Gy. For cases in which the margin dose was at least 22 Gy, the obliteration rates at 3, 5, and 10 years were 46 %, 71 %, and 86 %, respectively. For cases in which the margin dose was less than 22 Gy, the obliteration rates at 3, 5, and 10 years were 12 %, 39 %, and 59 %, respectively. The obliteration rates were significantly higher for nidi treated with a margin dose of at least 22 Gy (P=0.003, log-rank test). The number of patients remaining at each time point is shown under the x-axis number of draining veins (P=0.312), or margin dose (P= 0.361) and the development of RICs. Two patients had a total of three hemorrhages during the latency period (cumulative 232 risk years) following radiosurgery, including one patient with a single hemorrhage and one patient with two hemorrhages, for an annual post-radiosurgery hemorrhage rate of 1.3 %. Post-radiosurgery cyst formation was observed in one patient (2.0 %) but did not require intervention. There were no cases of radiosurgery-induced neoplasia. Fig. 3 Kaplan-Meier plots of obliteration over time for AVMs with 2two or fewer draining veins compared to AVMs with more than two draining veins. For AVMs with two or fewer draining veins, the obliteration rates at 3, 5, and 10 years were 39 %, 66 %, and 89 %, respectively. For AVMs with more than two draining veins, the obliteration rates at 3, 5, and 10 years were 0%,18%,and18%, respectively. The obliteration rates were significantly higher for nidi with two or fewer draining veins (P=0.001, log-rank test). The number of patients remaining at each time point is shown under the x-axis

Acta Neurochir (2015) 157:281 291 287 Clinical outcomes following radiosurgery At the most recent clinical follow-up, overall neurological improvement was observed in 17 patients (33.3 %), whereas neurological deterioration occurred in three patients (5.9 %). Of the 17 patients with neurological improvement, the mechanism was decreased seizure frequency in 13 patients with seizures at presentation (76.5 %) and a reduction in the severity of headache at presentation, mostly likely due to reduced steal phenomenon after radiosurgery, in four patients (23.5 %). The remaining 31 patients were unchanged from their baseline neurological status (60.8 %). For patients with seizures at presentation, the incidence of seizure improvement following radiosurgery was 48.1 % (13/27 patients), including 14.8 % seizure remission (4/27 patients) and 33.3 % decreased seizure frequency (9/27 patients). Baseline seizure status was unchanged in 51.9 % (14/27 patients). There were no cases of worsening seizures (0/27 patients), and, in patients without seizures at presentation, there were no occurrences of de novo seizures (0/24 patients). Discussion The optimal management of unruptured AVMs in pediatric patients is incompletely understood. Prior analyses of the AVM natural history have shown that unruptured nidi are less prone to hemorrhage than ruptured ones [23, 27, 36, 46]. Based on the findings from ARUBA and the prospective AVM study from the Scottish Audit of Intracranial Vascular Malformations (SAIVM), intervention remains controversial for unruptured AVMs [16, 39, 47]. However, ARUBA and the SAIVM AVM study excluded patients under the age of 18 and 16 years, respectively. Thus, the risk to benefit profile for the treatment of unruptured pediatric AVMs is not well defined. Two additional management dilemmas are faced in the pediatric AVM population. First, previous studies have shown that the hemorrhage risk of AVMs increases with increasing age. ApSimon et al. [4] reported the incidences of initial hemorrhage in untreated AVMs was 4.6 % for patients aged 0 9, compared to 21 % for patients aged 30 39 and 40 % for patients aged 60 69. Kim et al. [27] performed a multicenter meta-analysis of approximately 2,500 AVM patients and found a correlation between hemorrhagic presentation and increasing age (HR 1.34 per decade, 95 % CI 1.17-1.53). Second, the risk of AVM hemorrhage accumulates throughout a patient s lifetime. Therefore, while the pediatric age group is at the lowest risk of initial hemorrhage, it is also exposed to the highest cumulative lifetime risk of hemorrhage. These aspects of AVM natural history further complicate the decision to observe or treat unruptured pediatric AVMs. In order to guide decision-making for pediatric AVM patients, analysis of nidus angioarchitecture may provide additional insights. Kellner et al. [26] found that pediatric AVMs with a single draining vein (P=0.04) and deep venous drainage (P=0.02) were significantly more likely to rupture. We performed the first analysis of the treatment outcomes following radiosurgery for unruptured AVMs in pediatric patients (Fig. 4). Although radiosurgery has been previously evaluated as a treatment option for pediatric AVMs, prior studies have included a significant proportion of ruptured AVMs, which confounds the generalizability of their findings to unruptured pediatric AVMs. Table 5 summarizes the major pediatric AVM radiosurgery series [5, 11, 25, 28, 32, 35, 40, 41, 44, 54]. The median size of the ten previous major series was 100 patients (range, 53 363 patients), and the median proportion of AVMs with prior hemorrhage was 70.5 % (range, 41 80 %). The mean or median nidus volume and margin dose were 1.6 11.7 cm 3 and 17.5 23 Gy, respectively. Prior embolization was performed in a median 19 % (range, 9-44 %) of AVMs. The median outcomes were obliteration in 73 % (range, 34 83 %), symptomatic post-radiosurgery complications in 6.5 % (range, 2 28 %), and annual postradiosurgery hemorrhage rate of 2 % (range, 0 5 %). In the context of the existing pediatric AVM radiosurgery literature, our outcomes fall within previously reported limits, albeit with exclusively unruptured nidi prior to radiosurgery. From our review of the major series, it is evident that there is significant variability in radiosurgery outcomes for pediatric AVMs. Furthermore, the factors determined to be associated with outcomes are inconsistent across different reports. Borcek et al. [5] found nidus volume less than 3 cm 3 (P= 0.018), RBAS less than 1 (P=0.018), prior hemorrhage (P= 0.049), and compact nidus anatomy (P=0.012) to be predictors of excellent outcome (nidus obliteration without new neurological deficits). Yen et al. [54] found lack of prior embolization (P=0.042), smaller nidus volume (P=0.001), and higher margin dose (P=0.025) to be independent predictors of obliteration. In comparison, our multivariate analysis determined higher margin dose (P=0.002), fewer draining veins (P=0.038), and lower VRAS (P=0.003) to be independent predictors of obliteration. Prior AVM embolization has been associated with lower radiosurgical obliteration rates in numerous studies, although the mechanisms underlying this phenomenon are poorly substantiated and largely hypothetical [3, 6, 12, 14, 24, 31]. In our cohort, prior embolization was not a predictor of obliteration (P=0.254), suggesting that prior embolization may not affect radiosurgery outcomes in unruptured pediatric AVMs to the same extent as in those with ruptured AVMs. Alternatively, our study may be statistically underpowered to detect a difference in obliteration rates between embolized and non-embolized AVMs. Our Kaplan-Meier analysis determined that a margin dose of 22 Gy or higher yielded significantly higher AVM obliteration rates over time (P=0.003, Fig. 2), which is consistent

288 Acta Neurochir (2015) 157:281 291 Fig. 4 Case example of a pediatric patient with an unruptured AVM who was treated with radiosurgery. A 13-year-old boy had an incidentally diagnosed AVM. a T1-weighted, post-contrast MRI, axial section, showed a 2.4 1.3 1.2-cm (volume 1.9 cm 3 ), Spetzler-Martin grade III AVM in the left posterior frontal lobe. Cerebral angiography, (b)apand (c) lateral views of a left internal carotid artery (ICA) injection, showed that the nidus was fed by distal branches of the posterior division of the left middle cerebral artery with predominantly deep venous drainage into the Vein of Galen. There were no intranidal or prenidal aneurysms, and there was no venous outflow obstruction. The nidus was treated with a margin dose of 20 Gy to the 50 % isodose line using ten isocenters. The patient had transient, asymptomatic RIC after radiosurgery without latency period hemorrhage. At 50 months follow-up, (d) T1-weighted, postcontrast MRI, axial section, showed complete AVM obliteration, with a residual subtle hyperintensity and linear contrast enhancement at the site of the original nidus. Subsequent cerebral angiography at 51 months follow-up, (e) AP and (f) lateral views of a left ICA injection, confirmed nidus obliteration with the study by Yen et al. [54]. Similarly, Potts et al. [40] found that a margin dose of at least 18 Gy yielded significantly higher obliteration rates (52 % for 18 Gy versus 16 % for <18 Gy; P=0.015) and lower post-radiosurgery hemorrhage rates. Kano et al. [25] found a margin dose of at least 20 Gy yielded significantly higher obliteration rates. In contrast, Dinca et al. [11] did not find a significantly different obliteration rates between patients treated with 20 Gy (83 %) versus 25 Gy (86 %, P =0.43). Given the mechanisms of radiosurgery-induced AVM occlusion and the previously described dose response relationships for AVM radiosurgery, it seems reasonable that a higher margin dose would yield greater obliteration rates [9, 21, 42]. However, based on the heterogeneity of optimal margin doses determined by the present and prior analyses, further studies are necessary to define the radiobiological association between dose and obliteration in unruptured and ruptured pediatric AVMs. We also found that AVMs with two or fewer draining had significantly higher obliteration rates over time (P=0.001, Fig. 3), which is comparable to the finding by Borcek et al. [5] that excellent outcome was more likely in patients with compact nidi. This suggests that nidus morphology may affect obliteration to a greater extent than nidus volume. Prior analyses have suggested that radiosurgery-induced complications are directly related to target volume [19, 20]. RIC was not significantly associated with nidus volume (P=0.794) or

Acta Neurochir (2015) 157:281 291 289 Table 5 Summary of major series of radiosurgery for pediatric AVMs Series (year) Patients (n) Mean/median age (years) Prior hem. (%) Prior embo. (%) Mean/median nidus volume (cm 3 ) Mean/median prescription dose (Gy) Mean/median FU (months) OR (%) RIC (%) ALPHR (%) Current series 52 13.3 0 27 3.2 21.5 45 59 16 1.3 Borcek at al. (2014) 58 12 41 19 3.5 22 32 69 16 1.8 Potts et al. (2014) 80 12.7 56 13 3.1 17.5 46 34 28 5 Dinca et al. (2012) 363 12 80 16 1.6 22.7 NR 83 4 2.2 a Kano et al. (2012) 135 12 64 19 2.5 20 71 72 b 6 1.8 Yen et al. (2010) 186 12.7 72 20 3.2 21.9 80 69 7 2.4 Pan et al. (2008) 105 12 78 NR 11.7 18.5 25 81 8 1.9 Reyns et al. (2007) 100 12 69 44 2.8 23 26 70 7 2.0 a Nicolato et al. (2005) 63 11.7 79 27 3.8 21.6 33 79 4 0 Shin et al. (2002) 100 15 79 20 1.8 20 71 75 4 1.5 Levy et al. (2000) 53 12 64 9 1.7 20 36 74 2 2.5 Major series included those with more than 50 patients. The median value of each column was preferentially reported. When the median value was not available, the mean value was used hem.hemorrhage, embo.embolization, FUradiologic follow-up, ORobliteration rate, RICsymptomatic radiosurgery-induced complications (excluded latency period hemorrhage), ALPHRannual latency period hemorrhage rate, NRnot reported a Hemorrhage rate only reported as cumulative rate, not annual rate b Cumulative obliteration rate not reported, this is the actuarial obliteration rate at 10 years margin dose (P=0.361) in our analysis. The relatively high rate of radiologically evident RICs (55 %) in our study supports prior conclusions that healthy brain is particularly predisposed to RICs [53]. However, the incidence of symptomatic RICs (16 %) was modest, and the rate of permanent RIC (2 %) was similar to previously reports [16]. Thus, although pediatric patients with unruptured AVMs may be more susceptible to RIC, they also likely possess a more robust reserve for neurological recovery than adult AVM patients. The rate of post-radiosurgery hemorrhage was 1.3 %, which is not significantly than the natural history of unruptured AVMs in the pediatric population [4, 27]. Notably, there were no cases of radiosurgery-induced neoplasia, which we have previously documented in two pediatric AVM patients, both with a history of hemorrhage [43, 48]. Study limitations This study is associated with a number of limitations. First, this was a single cohort analysis of patients who all underwent treatment with radiosurgery. Without a similar cohort of patients treated with another modality or conservatively managed, we are unable to assess the relative safety and efficacy of radiosurgery compared with alternate management strategies, which limits the generalizability of our findings. Thus, we are unable to definitively address if intervention is superior to medical management for unruptured AVMs in pediatric patients. Additionally, the single-center, retrospective design of this study subjects our findings to the selection and treatment biases of our institution and physicians. Furthermore, the nature of being a tertiary referral center for radiosurgery had the ramification of a lack of detailed clinical follow-up and thus, the measurement of clinical neurological improvement mayhavebeenimprecise. Next, the inclusion criteria of the study, wherein patients with less than 2 years of radiologic follow-up were included for analysis if they had AVM obliteration, may have biased the results toward more favorable outcomes. In order to account for this bias, we also reported the outcomes of only the patients with at least 2 years of radiologic follow-up, for whom the rates of obliteration and symptomatic RICs were lower and the rates of radiologically evidenced RICs and permanent RICs were similar. We also acknowledge that, in many European countries, only patients under 16 years of age are considered pediatric. In our cohort, 13 patients (25.5 %) were 16 18 years of age. If only patients less than 16 years old were included in this study, the rates of obliteration, RIC, and annual post-radiosurgery hemorrhage were 58 % (22/38 patients), 47 % (18/38 patients), and 1.7 % (three hemorrhages over 172 risk years), respectively. Therefore, we do not believe that the main findings or conclusions of this study would be significantly affected by lowering the upper age limit for inclusion from 18 to 16 years. Finally, 20 % of patients did not have angiographic documentation of obliteration. However, prior studies have shown that MRI has a diagnostic capability to determine obliteration which is close to that of angiography [33, 38]. Pollock et al. [38] reported the specificity, sensitivity and negative

290 Acta Neurochir (2015) 157:281 291 predictive value of MRI compared with angiography were 100 %, 80 %, and 91 %, respectively, for determining AVM obliteration. O Connor and Friedman [33] found MRI to have 82 % accuracy compared with angiography for determining obliteration. Conclusions Radiosurgery affords a favorable risk to benefit profile for unruptured AVMs in pediatric patients (age <18 years). Nidi with simpler venous anatomy, as measured by the number of draining veins, were more likely to achieve obliteration following radiosurgery. Additionally, radiosurgical efficacy was improved when a margin dose of at least 22 Gy was delivered to the nidus. Although the rate of radiologically evident RIC was relatively higher in unruptured pediatric AVMs compared with the adult population, the incidence of symptomatic RIC was modest. Furthermore, the rate of latency period hemorrhage was very low. Thus, radiosurgery provides an effective management option for appropriately selected unruptured AVMs in the pediatric population. Acknowledgments We would like to acknowledge the late Professor Steiner and Professors Kassell, Jensen, and Evans for their roles in the care and treatment of some of the patients in this study. Conflicts of interest References None. 1. Al-Shahi R, Bhattacharya JJ, Currie DG, Papanastassiou V, Ritchie V, Roberts RC, Sellar RJ, Warlow CP (2003) Prospective, population-based detection of intracranial vascular malformations in adults: the Scottish Intracranial Vascular Malformation Study (SIVMS). Stroke 34:1163 1169 2. Al-Shahi R, Warlow C (2001) A systematic review of the frequency and prognosis of arteriovenous malformations of the brain in adults. Brain 124:1900 1926 3. Andrade-Souza YM, Ramani M, Scora D, Tsao MN, terbrugge K, Schwartz ML (2007) Embolization before radiosurgery reduces the obliteration rate of arteriovenous malformations. Neurosurgery 60: 443 451, discussion 451 442 4. ApSimon HT, Reef H, Phadke RV, Popovic EA (2002) A populationbased study of brain arteriovenous malformation: long-term treatment outcomes. Stroke 33:2794 2800 5. Borcek AO, Emmez H, Akkan KM, Ocal O, Kurt G, Aykol S, Karahaciogli E, Baykaner KM (2014) Gamma Knife radiosurgery for arteriovenous malformations in pediatric patients. Childs Nerv Syst 30:1485 1492 6. Buell TJ, Ding D, Starke RM, Webster Crowley R, Liu KC (2014) Embolization-induced angiogenesis in cerebral arteriovenous malformations. J Clin Neurosci. doi:10.1016/j.jocn.2014.04.010 7. Buis DR, Lagerwaard FJ, Barkhof F, Dirven CM, Lycklama GJ, Meijer OW, van den Berg R, Langendijk HA, Slotman BJ, Vandertop WP (2005) Stereotactic radiosurgery for brain AVMs: role of interobserver variation in target definition on digital subtraction angiography. Int J Radiat Oncol Biol Phys 62:246 252 8. Buis DR, Lagerwaard FJ, Dirven CM, Barkhof F, Knol DL, van den Berg R, Slotman BJ, Vandertop WP (2007) Delineation of brain AVMs on MR-Angiography for the purpose of stereotactic radiosurgery. Int J Radiat Oncol Biol Phys 67:308 316 9. Chang SD, Shuster DL, Steinberg GK, Levy RP, Frankel K (1997) Stereotactic radiosurgery of arteriovenous malformations: pathologic changes in resected tissue. Clin Neuropathol 16:111 116 10. Chen CJ, Chivukula S, Ding D, Starke RM, Lee CC, Yen CP, Xu Z, Sheehan JP (2014) Seizure outcomes following radiosurgery for cerebral arteriovenous malformations. Neurosurg Focus 37:E17 11. Dinca EB, de Lacy P, Yianni J, Rowe J, Radatz MW, Preotiuc-Pietro D, Kemeny AA (2012) Gamma knife surgery for pediatric arteriovenous malformations: a 25-year retrospective study. J Neurosurg Pediatr 10:445 450 12. Ding D, Starke RM, Yen CP, Sheehan JP (2014) Radiosurgery for cerebellar arteriovenous malformations: does infratentorial location affect outcome? World Neurosurg 82:e209 e217 13. Ding D, Yen CP, Starke RM, Xu Z, Sheehan JP (2014) Radiosurgery for ruptured intracranial arteriovenous malformations. J Neurosurg 121:470 481 14. Ding D, Yen CP, Starke RM, Xu Z, Sun X, Sheehan JP (2013) Outcomes following single-session radiosurgery for high-grade intracranial arteriovenous malformations. Br J Neurosurg 28:666 674 15. Ding D, Yen CP, Starke RM, Xu Z, Sun X, Sheehan JP (2014) Radiosurgery for Spetzler-Martin Grade III arteriovenous malformations. J Neurosurg 120:959 969 16. Ding D, Yen CP, Xu Z, Starke RM, Sheehan JP (2013) Radiosurgery for patients with unruptured intracranial arteriovenous malformations. J Neurosurg 118:958 966 17. Ding D, Yen CP, Xu Z, Starke RM, Sheehan JP (2013) Radiosurgery for primary motor and sensory cortex arteriovenous malformations: outcomes and the effect of eloquent location. Neurosurgery 73:816 824 18. Ding D, Yen CP, Xu Z, Starke RM, Sheehan JP (2014) Radiosurgery for low-grade intracranial arteriovenous malformations. J Neurosurg 121:457 467 19. Flickinger JC, Kondziolka D, Lunsford LD, Kassam A, Phuong LK, Liscak R, Pollock B (2000) 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 46:1143 1148 20. Flickinger JC, Kondziolka D, Pollock BE, Maitz AH, Lunsford LD (1997) Complications from arteriovenous malformation radiosurgery: multivariate analysis and risk modeling. Int J Radiat Oncol Biol Phys 38:485 490 21. Flickinger JC, Pollock BE, Kondziolka D, Lunsford LD (1996) A dose response analysis of arteriovenous malformation obliteration after radiosurgery. Int J Radiat Oncol Biol Phys 36:873 879 22. Ganz JC, Reda WA, Abdelkarim K, Hafez A (2005) A simple method for predicting imaging-based complications following gamma knife surgery for cerebral arteriovenous malformations. J Neurosurg 102(Suppl):4 7 23. Gross BA, Du R (2013) Natural history of cerebral arteriovenous malformations: a meta-analysis. J Neurosurg 118:437 443 24. Kano H, Kondziolka D, Flickinger JC, Park KJ, Iyer A, Yang HC, Liu X, Monaco EA 3rd, Niranjan A, Lunsford LD (2012) Stereotactic radiosurgery for arteriovenous malformations after embolization: a case control study. J Neurosurg 117:265 275 25. Kano H, Kondziolka D, Flickinger JC, Yang HC, Flannery TJ, Awan NR, Niranjan A, Novotny J, Lunsford LD (2012) Stereotactic radiosurgery for arteriovenous malformations, part 2: management of pediatric patients. J Neurosurg Pediatr 9:1 10 26. Kellner CP, McDowell MM, Phan MQ, Connolly ES, Lavine SD, Meyers PM, Sahlein D, Solomon RA, Feldstein NA, Anderson RC

Acta Neurochir (2015) 157:281 291 291 (2014) Number and location of draining veins in pediatric arteriovenous malformations: association with hemorrhage. J Neurosurg Pediatr. doi:10.3171/2014.7.peds13563 27. Kim H, Al-Shahi Salman R, McCulloch CE, Stapf C, Young WL, Coinvestigators M (2014) Untreated brain arteriovenous malformation: patient-level meta-analysis of hemorrhage predictors. Neurology 83:590 597 28. Levy EI, Niranjan A, Thompson TP, Scarrow AM, Kondziolka D, Flickinger JC, Lunsford LD (2000) Radiosurgery for childhood intracranial arteriovenous malformations. Neurosurgery 47:834 841, discussion 841 832 29. Mohr JP, Parides MK, Stapf C, Moquete E, Moy CS, Overbey JR, Al- Shahi Salman R, Vicaut E, Young WL, Houdart E, Cordonnier C, Stefani MA, Hartmann A, von Kummer R, Biondi A, Berkefeld J, Klijn CJ, Harkness K, Libman R, Barreau X, Moskowitz AJ (2014) Medical management with or without interventional therapy for unruptured brain arteriovenous malformations (ARUBA): a multicentre, non-blinded, randomised trial. Lancet 383:614 621 30. Moosa S, Chen CJ, Ding D, Lee CC, Chivukula S, Starke RM, Yen CP, Xu Z, Sheehan JP (2014) Volume-staged versus dose-staged radiosurgery outcomes for large intracranial arteriovenous malformations. Neurosurg Focus 37:E18 31. Mouchtouris N, Jabbour PM, Starke RM, Hasan DM, Zanaty M, Theofanis T, Ding D, Tjoumakaris SI, Dumont AS, Ghobrial GM, Kung D, Rosenwasser RH, Chalouhi N (2014) Biology of cerebral arteriovenous malformations with a focus on inflammation. J Cereb Blood Flow Metab. doi:10.1038/jcbfm.2014.179 32. Nicolato A, Foroni R, Seghedoni A, Martines V, Lupidi F, Zampieri P, Sandri MF, Ricci U, Mazza C, Beltramello A, Gerosa M, Bricolo A (2005) Leksell gamma knife radiosurgery for cerebral arteriovenous malformations in pediatric patients. Childs Nerv Syst 21:301 307, discussion 308 33. O Connor TE, Friedman WA (2013) Magnetic resonance imaging assessment of cerebral arteriovenous malformation obliteration after stereotactic radiosurgery. Neurosurgery 73:761 766 34. Paddick I (2000) A simple scoring ratio to index the conformity of radiosurgical treatment plans. Technical note. J Neurosurg 93(Suppl 3):219 222 35. Pan DH, Kuo YH, Guo WY, Chung WY, Wu HM, Liu KD, Chang YC, Wang LW, Wong TT (2008) Gamma Knife surgery for cerebral arteriovenous malformations in children: a 13-year experience. J Neurosurg Pediatr 1:296 304 36. Pollock BE, Flickinger JC, Lunsford LD, Bissonette DJ, Kondziolka D (1996) Factors that predict the bleeding risk of cerebral arteriovenous malformations. Stroke 27:1 6 37. Pollock BE, Flickinger JC, Lunsford LD, Maitz A, Kondziolka D (1998) Factors associated with successful arteriovenous malformation radiosurgery. Neurosurgery 42:1239 1244, discussion 1244 1237 38. Pollock BE, Kondziolka D, Flickinger JC, Patel AK, Bissonette DJ, Lunsford LD (1996) Magnetic resonance imaging: an accurate method to evaluate arteriovenous malformations after stereotactic radiosurgery. J Neurosurg 85:1044 1049 39. Pollock BE, Link MJ, Brown RD (2013) The risk of stroke or clinical impairment after stereotactic radiosurgery for ARUBA-eligible patients. Stroke 44:437 441 40. Potts MB, Sheth SA, Louie J, Smyth MD, Sneed PK, McDermott MW, Lawton MT, Young WL, Hetts SW, Fullerton HJ, Gupta N (2014) Stereotactic radiosurgery at a low marginal dose for the treatment of pediatric arteriovenous malformations: obliteration, complications, and functional outcomes. J Neurosurg Pediatr 14:1 11 41. Reyns N, Blond S, Gauvrit JY, Touzet G, Coche B, Pruvo JP, Dhellemmes P (2007) Role of radiosurgery in the management of cerebral arteriovenous malformations in the pediatric age group: data from a 100-patient series. Neurosurgery 60:268 276, discussion 276 42. Schneider BF, Eberhard DA, Steiner LE (1997) Histopathology of arteriovenous malformations after gamma knife radiosurgery. J Neurosurg 87:352 357 43. Sheehan J, Yen CP, Steiner L (2006) Gamma knife surgery-induced meningioma. Report of two cases and review of the literature. J Neurosurg 105:325 329 44. Shin M, Kawamoto S, Kurita H, Tago M, Sasaki T, Morita A, Ueki K, Kirino T (2002) Retrospective analysis of a 10-year experience of stereotactic radio surgery for arteriovenous malformations in children and adolescents. J Neurosurg 97:779 784 45. Spetzler RF, Martin NA (1986) A proposed grading system for arteriovenous malformations. J Neurosurg 65:476 483 46. Stapf C, Mast H, Sciacca RR, Choi JH, Khaw AV, Connolly ES, Pile- Spellman J, Mohr JP (2006) Predictors of hemorrhage in patients with untreated brain arteriovenous malformation. Neurology 66: 1350 1355 47. Starke RM, Sheehan JP, Ding D, Liu KC, Kondziolka D, Crowley RW, Lunsford LD, Kassell NF (2014) Conservative management or intervention for unruptured brain arteriovenous Malformations. World Neurosurg. doi:10.1016/j.wneu.2014.07.001 48. Starke RM, Yen CP, Chen CJ, Ding D, Mohila CA, Jensen ME, Kassell NF, Sheehan JP (2014) An updated assessment of the risk of radiation induced neoplasia following radiosurgery of arteriovenous malformations. World Neurosurg 82:395 401 49. Starke RM, Yen CP, Ding D, Sheehan JP (2013) A practical grading scale for predicting outcome after radiosurgery for arteriovenous malformations: analysis of 1012 treated patients. J Neurosurg 119: 981 987 50. Steiner L, Lindquist C, Adler JR, Torner JC, Alves W, Steiner M (1992) Clinical outcome of radiosurgery for cerebral arteriovenous malformations. J Neurosurg 77:1 8 51. Wegner RE, Oysul K, Pollock BE, Sirin S, Kondziolka D, Niranjan A, Lunsford LD, Flickinger JC (2011) A modified radiosurgerybased arteriovenous malformation grading scale and its correlation with outcomes. Int J Radiat Oncol Biol Phys 79:1147 1150 52. Yen CP, Ding D, Cheng CH, Starke RM, Shaffrey M, Sheehan J (2014) Gamma Knife surgery for incidental cerebral arteriovenous malformations. J Neurosurg. doi:10.3171/2014.7.jns131397 53. Yen CP, Matsumoto JA, Wintermark M, Schwyzer L, Evans AJ, Jensen ME, Shaffrey ME, Sheehan JP (2013) Radiation-induced imaging changes following Gamma Knife surgery for cerebral arteriovenous malformations. J Neurosurg 118:63 73 54. Yen CP, Monteith SJ, Nguyen JH, Rainey J, Schlesinger DJ, Sheehan JP (2010) Gamma Knife surgery for arteriovenous malformations in children. J Neurosurg Pediatr 6:426 434