Venous Stenosis and Hemorrhage After Radiosurgery for Cerebral Arteriovenous Malformations.

Abstract

The risk of hemorrhage remains after radiosurgery for patients with arteriovenous malformations (AVMs), especially during the latency period. The effect of venous outflow stenosis on postradiosurgery AVM hemorrhage has been understudied. The present study sought to clarify the effect of venous stenosis on postradiation hemorrhage. We retrospectively reviewed the records of patients with AVM seen at our institution from 1990 to 2015. Patients who had undergone radiosurgery were included, and those without sufficient data were excluded. We performed multivariable Cox regression to evaluate the predictors of postradiosurgery hemorrhage, with specific emphasis on venous stenosis. Patients were censored from the first radiosurgery to hemorrhage or the last follow-up visit. The baseline and angiographic characteristics were compared between those with venous stenosis and those without to address potential confounders. The present study included 240 patients, of whom 29 (12.1%) had venous stenosis. The venous stenosis cohort included more patients with venous varices (P = 0.009) and fewer with deep venous drainage (P = 0.048) compared with those without venous stenosis. Most patients had grade III or higher AVMs (63.2%), with an obliteration rate of 32.9%. In an all-inclusive multivariable Cox regression, hemorrhage risk was associated with venous stenosis (hazard ratio [HR], 3.70; P = 0.034), age (HR, 1.05; P = 0.002), AVM volume (HR, 1.04; P = 0.004), and hemorrhage before treatment (HR, 4.11; P = 0.014). Male gender was protective (HR, 0.31; P = 0.036) against hemorrhage. We identified statistically significant risk factors for postradiosurgery AVM hemorrhage, which included advanced age, female gender, the presence of venous stenosis, a larger AVM volume, and previous hemorrhage. We recommend cautious selection of patients for radiosurgery with close follow-up after treatment, especially for patients with these risk factors.