Biologically Effective Dose and Prediction of Obliteration of Arteriovenous Malformations in Pediatric Patients Treated by Gamma Knife Radiosurgery

Abstract

BACKGROUND AND OBJECTIVES: Stereotactic radiosurgery (SRS) represents an effective treatment for pediatric arteriovenous malformations (AVMs). Biological effective dose (BED) has shown promising results in 2 previous studies as a predictive variable for outcomes in adults, but its role has never been studied in pediatric outcomes. METHODS: Retrospective data for patients 18 years or younger treated with a single-session SRS for AVMs were collected from 1989 to 2019. BED calculations were performed using an α/β ratio of 2.47. Kaplan-Meier analysis was used to evaluate obliteration, new hemorrhage, and radiation-induced changes (RIC). Cox-regression analysis was used for obliteration prediction using 2 models (margin dose vs BED). RESULTS: One hundred ninety-seven patients (median age = 13.1 years, IQR = 5.2) were included; 72.6% (143/197) of them presented initially with spontaneous hemorrhage. A median margin dose of 22 Gy (IQR = 4.0) with a median BED of 183.2 Gy (IQR = 70.54) was used to treat AVM with a median volume of 2.8 cm3 (IQR = 2.9). After SRS, obliteration was confirmed in 115/197 patients (58.4%) using magnetic resonance imaging and angiography at a median follow-up of 2.85 years (IQR = 2.26). The cumulative obliteration probability was 43.6% (95% CI = 36.1-50.3), 60.5% (95% CI+ = 2.2-67.4), and 66.0% (95% CI = 56.0-73.7) at 3, 5, and 10 years, respectively. In Cox multivariate analysis, a BED >180 Gy (hazard ratio [HR] = 2.11, 95% CI = 1.30-3.40, P = .002) in model 1 and a margin dose >20 Gy (HR = 1.90, 95% CI = 1.15-3.13, P = .019) in model 2 were associated with obliteration. An AVM nidus volume >4 cm3 was associated with lower obliteration rates in both models. The probability of symptomatic RIC at 10 years was 8.6% (95% CI = 3.5-13.4). Neither BED nor margin dose was associated with RIC occurrence, with the only predictive factor being deep AVM location (HR = 3, 95% CI = 1-9.1, P = .048). CONCLUSION: This study confirms BED as a predictor for pediatric AVM obliteration. Optimization of BED in pediatric AVM SRS planning may improve cumulative obliteration rates.