Background: Management of unruptured BAVM remains controversial despite recent clinical trial results demonstrating a 3-fold increased risk of stroke and death in treated versus medically managed patients. We compared outcomes between treated and untreated patients with unruptured BAVMs using observational data from the Multicenter AVM Research Study (MARS).
Methods: MARS consists of 3 unruptured cohorts: UCSF (n=416), Columbia (n=223), and the Scottish Audit of Intracranial Vascular Malformations (n=70). We used a propensity score adjusted analysis with Cox proportional hazards regression of time to adverse event (stroke or death) from date of diagnosis or treatment, censoring at last follow-up (up to 5 years) to compare the group receiving any interventional treatment of embolization, gamma knife, or surgical resection (‘treated’) to the group receiving medical management (‘untreated’). The propensity score for treatment was computed using logistic regression with predictors of age, age2, sex, Caucasian race, cohort, Spetzler-Martin score, and reason for presentation.
Results: There were 709 unruptured BAVM patients (565 treated/144 untreated). A total of 81 adverse events (68 strokes/13 deaths) occurred during 1632 patient-years of follow-up for an annual rate of 5.0% (95% CI: 4.0%-6.2%) overall, 5.6% (4.4%-7.2%) in the treated group, and 4.6% (2.0%-5.4%) in the untreated group. There was a trend for increased risk of adverse events in the treated group (HR=1.5, 95% CI:0.8-2.7, p=0.180). In stratified analysis, subjects in the highest propensity quintile (i.e., those most expected to receive treatment) seemed to have lower rates of adverse events in the treated versus untreated group (HR=0.4, 95% CI:0.1-1.5, p=0.178). An analysis using all patients except those in the highest quintile suggested outcomes were worse in the treated group (HR=1.81, 95% CI:0.96-3.4, p=0.065).
Conclusion: Our results suggest an increased risk of adverse events in unruptured BAVMs that are treated. However, taking into account propensity for treatment revealed a possible subgroup that may benefit from treatment. Larger sample sizes are needed to identify risk factors and develop risk stratification models for unruptured BAVMs.