Features predictive of brain arteriovenous malformation hemorrhage: extrapolation to a physiologic model.

Background And Purpose: Although there is generally thought to be a 2% to 4% per annum rupture risk for brain arteriovenous malformations (bAVMs), there is no way to estimate risk for an individual patient.

Methods: In this retrospective study, patients were eligible who had nidiform bAVMs and underwent detailed pretreatment diagnostic cerebral angiography at our medical center from 1996 to 2006. All patients had superselective microcatheter angiography, and films were reviewed for the purpose of this project. Patient demographics, clinical presentation, and angioarchitectural characteristics were analyzed. A univariate analysis was performed, and angioarchitectural features with potential physiological significance that showed at least a trend toward significance were added to a multivariate logistic regression model.

Results: One hundred twenty-two bAVMs met criteria for study entry. bAVMs with single venous drainage anatomy were more likely to present with hemorrhage. In addition, patients with multiple draining veins and a venous stenosis reverted to a risk similar to those with 1 draining vein, whereas those with multiple draining veins and without stenosis had diminished association with hemorrhage presentation. Those bAVMs with associated aneurysms were more likely to present with hemorrhage. These findings were robust in both univariate and multivariate models.

Conclusions: The results of this article lead to the first physiological, internally consistent model of individual bAVM hemorrhage risk, where 1 draining vein, venous stenosis, and associated aneurysms increase risk.