Predicting Successful Treatment of Intracranial Aneurysms with the Pipeline Embolization Device Through Meta-Regression.

Abstract

To conduct meta-regression for the identification of clinically meaningful characteristics that can predict successful treatment of intracranial aneurysms with the Pipeline Embolization Device (PED). Using the key word "Pipeline Embolization Device" and prespecified eligibility criteria, we searched PubMed and EMBASE and identified 105 original research articles (10,813 patients) published or in press (April 6, 2011, to December 31, 2016) focused on safety and/or effectiveness of PED. We constructed meta-regression models to identify predictors of 3 measures of PED treatment success, specifically rates of aneurysm occlusion, modified Rankin Scale score, and death. Rates were log_e-transformed, and the adjusted R² statistic was reported for each model. Aneurysm morphology and type accounted for substantial variation in 1-year aneurysm occlusion rate (14 studies with 689 patients; R_adj² = 88.91%). By contrast, predictors of 6-months aneurysm occlusion rate (aneurysm location, aneurysm type; 24 studies with 1321 patients; R_adj² = 31.67%) and aneurysm occlusion rate at unspecified follow-up time (year of publication, mean duration of angiographic follow-up; 34 studies with 1663 patients, R_adj² = 42.82%) accounted for less variability. Few predictors were identified for death rate (number of patients, aneurysm morphology; 34 studies with 6463 patients, R_adj² = 50.69%) or modified Rankin Scale score ≤2 (year of publication, mean age, aneurysm morphology; 14 studies with 556 patients, R_adj² = 100%). Few clinically meaningful characteristics appear to be relevant as predictors of the selected measures of PED treatment success. Future studies should attempt to identify additional predictors while focusing on aneurysm morphology as a key predictor of PED outcomes.