Dosimetric predictors of toxicity for brainstem metastases and AVMs treated with stereotactic radiosurgery: An international, multi-institutional analysis

Abstract

Background: There are limited data on dosimetric predictors of radiation-induced changes (RICs) for brainstem metastases or arteriovenous malformations (AVMs) following single-fraction stereotactic radiosurgery (SRS). Methods: We examined a multi-institutional cohort of patients with brainstem metastases or AVMs treated with SRS. We evaluated predictors of RICs graded per CTCAE(Common Terminology Criteria for Adverse Events), including D5%, D95%, D0.03 cc, and D0.5 cc (brainstem minus lesion). Univariate logistic regressions were initially performed with independent variables trending towards significance included on multivariate logistic regression. Results: A total of 124 brainstem lesions treated with SRS were analyzed (21 AVMs and 103 metastases). The median prescription dose was 16 Gy(range: 13-23 Gy), and the median treatment volume was 0.48 cc(range: 0.002-11.19 cc). The incidence of RICs was 9.7 % (with 3/12 being Grade 3-4 and no Grade 5). All cases occurred in brainstem metastases, with no cases among those of the midbrain-pons transition. Treatment volumes >= 1cc were correlated with a higher symptomatic RIC incidence(6/57 vs. 6/65; p = 0.04). Notably, all RIC cases had a D0.5 cc >= 15 Gy(12/87 vs. 0/32). Both D5%>= 6 Gy(9/55 vs. 3/69; p = 0.04) and D95%>= 1 Gy(7/31 vs. 5/93; p = 0.01) were significantly correlated with higher incidence of RICs and D0.03 cc >= 22 Gy was correlated with a lower risk(2/61 vs. 9/63; p = 0.03). On MVA, D0.03 cc >= 22 Gy remained a significant predictor of a lower incidence of RICs(odds ratio = -1.72 (95 % CI: -3.32 to -0.12; p = 0.04). Conclusions: Incidence of RICs was low following SRS for brainstem metastases and AVMs. We recommend optimizing radiosurgical plans for D0.5 cc < 15 Gy with consideration of D5% and D95%, with less emphasis on D0.03 cc to allow to meet the former metrics as feasible.