The R2eD Arteriovenous Malformation Score: A Novel Predictive Tool for Arteriovenous Malformation Presentation With Hemorrhage

Abstract

Abstract INTRODUCTION The management of unruptured brain arteriovenous malformations (AVMs) remains unclear. This study aims to develop a predictive tool that could guide hemorrhage risk stratification. METHODS A database of 789 AVM patients presenting to our institution between 1990 and 2017 was used. A hold-out method of model building and validation was employed whereby the data was randomly split in half into training and validation datasets. Factors significantly associated with hemorrhage presentation at the univariable level in the training dataset were used to construct a model based on multivariable logistic regression. Model performance was assessed using receiver operating curves (ROC) on the training, validation, and complete datasets. The model predictors and the complete dataset were then used to derive a formula for risk prediction and a practical scoring system according to the model coefficients. RESULTS In 755 patients with complete data on presentation status, 272 (36%) presented with hemorrhage. After model building and validation, the final model contained the following risk factors: non-white race (odds ratio [OR] = 2.49, P < .01), deep location (OR = 1.68, P = .02), small AVM size (<3 cm, OR = 1.63, P < .01), exclusive deep venous drainage (OR = 1.73, P = .02), and “monoarterial” feeding (OR = 1.58, P = .02). Area under the curve from ROC analysis was 0.702, 0.698, and 0.685 for the training, validation, and complete datasets, respectively. Every risk factor was worth 1 point except race, which was worth 2 points. The factors are summarized by the acronym R2eD AVM (total score varies from 0 to 6). In the entire study population, the predicted probability of hemorrhagic presentation increased in a stepwise manner from 16% for patients with no risk factors (score of 0) to 78% for patients having all the risk factors (score of 6) with around a 10% increase in risk per added point. CONCLUSION This risk can be used as a predictive tool that supplements clinical judgement and aids in patient counselling.