Abstract

Background: Brain arteriovenous malformations (BAVM) are a major cause of hemorrhagic stroke in young adults. Microsurgical resection is considered definitive treatment. To gauge surgical risk, the Spetzler-Martin scale (SM) is regularly used. We have recently proposed an alternate grading scheme called the SM-supplementary scale (SM-Supp). Further improvement of surgical risk prediction will require novel risk factors, such as genetic variants. Brain-derived neurotrophic factor (BDNF) encodes a secreted neurotropin known to influence neural proliferation and plasticity. BDNF contains a functions polymorphism, Val66Met, which causes decreased BDNF secretion and is associated with poor outcome in acute neurological insults. Objective: We sought to determine if 1) SM-Supp predicts poor surgical outcome more accurately than SM; 2) Val66Met genotype is associated with surgical outcome; and 3) the addition of Val66Met genotype to SM-Supp improves predictive accuracy. Methods: 483 BAVM-resection patients seen at UCSF formed the study group. SM-Supp was derived in a 300-patient subgroup, the development cohort. The remaining 183 patients formed an independent validation cohort. 341 patients had Val66Met genotype data. Outcome was dichotomous in which poor outcome was defined as worsening on the modified Rankin Scale at last follow-up. Models were compared using area under the ROC curve (AUROC) and Net Reclassification Index (NRI) analyses. Genotype-outcome associations were assessed using logistic regression analysis. BDNF results were stratified by hemorrhage due to significant interaction (p=0.03). Results: SM-Supp had a larger AUROC than SM in both the development and validation cohorts (0.76 vs 0.66 and 0.71 vs 0.65, respectively). Continuous NRI of 64% demonstrated superior risk reclassification by SM-Supp compared with SM. The Met allele of Val66Met was associated with increased risk of poor outcome in unruptured patients (multivariate results: OR=2.15, 95%CI=1.02-4.55, p=0.045, n=173) but not ruptured patients (multivariate results: OR=0.54, 95%CI=0.19-1.53, p=0.25, n=168). Adding Val66Met genotype to the SM-Supp model increased AUROC in unruptured patients compared with SM-Supp alone (0.70 vs 0.66, p=0.06). Discussion: SM-Supp performed better than SM and should be considered for clinical risk prediction. Having the Val66Met variant of BDNF may be a novel risk factor for poor surgical outcome. Combining clinical and genetic risk factors may improve risk prediction and influence treatment decisions.