Accurately balancing the competing risks of interventional treatment and natural history ? primarily intracranial hemorrhage (ICH) ? is necessary for optimal management of patients harboring unruptured brain arteriovenous malformations (bAVMs). As a class, unruptured patients are at higher risk for treatment-related injury but at lower risk for ICH. Accordingly, treatment of unruptured bAVMs?half of all cases? is becoming increasingly controversial. Results from A Randomized Trial of Unruptured Brain AVMs (ARUBA, NCT00389181) showed higher stroke and death rates in patients receiving intervention (either surgery, embolization and/or radiosurgery) compared to non-intervention. However, the ARUBA trial was stopped early and has generated strong criticisms from the interventional community expressing doubts about the generalizability of findings due to the potential for selection bias in a randomized trial, underrepresentation of cases from the US, atypical treatment patterns, and short follow-up of 33 months that leaves the question unanswered over patients' entire lifetimes. Observational cohort studies of bAVM clinical course afford an opportunity to address important gaps and previous criticisms of randomized data to help guide treatment decisions. The overarching goal of this project is to identify novel risk factors for ICH in the untreated course of unruptured bAVMs, to estimate risk with precision, and to create personalized risk prediction models for patients. We propose to conduct a large, multicenter study of bAVM cohorts with longitudinal data to identify robust risk factors for ICH in the natural history course and to causally test the effect of treatment on long-term outcomes in unruptured bAVM.
Aim 1 will identify predictors of ICH and poor outcome in unruptured bAVM patients using a systematic approach to identify and combine data from collaborating cohorts. We will perform a large individual patient data meta-analysis to more accurately quantify traditional and novel risk factors.
Aim 2 will test whether long-term outcomes differ by treatment using observational data and sophisticated epidemiological approaches for causal inference and unbiased estimates, including Cox regression with time- varying covariate, propensity-score analysis, and marginal structural models.
Aim 3 will test whether long- term outcomes differ by treatment using randomized data and will also compare results to observational data to address low external validity of randomized data due to provider biases, exclusion criteria, or atypical treatment patterns.
Aim 4 will build and validate risk prediction models and provide a novel tool for calculating individualized risks. The proposed multicenter consortium project will provide important and comprehensive characterization of the untreated and treated course for unruptured bAVM patients that will be a useful resource for clinicians planning management and assessing current and future interventions. Observational approaches using large datasets and sophisticated statistical approaches are the only feasible way of informing and improving unruptured bAVM patient outcomes in the near future.

Public Health Relevance

Brain arteriovenous malformations (bAVMs) are an important cause of hemorrhagic stroke in young adults. Interventional therapy can be curative, though at high costs, and the inherent morbidity and mortality associated with treatment may outweigh the benefits, especially for patients with unruptured bAVMs. This project will identify risk factors for ICH in the untreated course, causally test whether treating unruptured lesions is better than leaving them alone, identify possible sub-groups at higher risk to aid rational choice of therapy to maximize benefit and minimize risk, and provide a novel tool for individualized risk prediction to help guide management decisions.

National Institute of Health (NIH)
National Institute of Neurological Disorders and Stroke (NINDS)
Research Project (R01)
Project #
Application #
Study Section
Neurological, Aging and Musculoskeletal Epidemiology (NAME)
Program Officer
Moy, Claudia S
Project Start
Project End
Budget Start
Budget End
Support Year
Fiscal Year
Total Cost
Indirect Cost
University of California San Francisco
Schools of Medicine
San Francisco
United States
Zip Code