Predicting Endovascular Treatment Outcomes in Acute Vertebrobasilar Artery Occlusion: A Model to Aid Patient Selection from the ASIAN KR Registry

Abstract

Background The decision to perform endovascular treatment (EVT) for stroke related to vertebrobasilar occlusion (VBO) remains controversial. Purpose To identify preprocedural predictors of good outcomes and to develop a model to aid patient selection for VBO. Materials and Methods For this retrospective study using a Korean multicenter registry, a predictive model for good outcomes (modified Rankin scale score, 0-2) was generated based on a derivation sample of patients with VBO (January 2011-February 2016). Preprocedural parameters, including onset-to-puncture time, infarct volume, occlusion type as a surrogate marker of intracranial atherosclerotic stenosis-related occlusion or embolic occlusion (truncal-type occlusion vs branching site occlusion), and collateral status, were analyzed. Continuous variables were dichotomized based on receiver operating characteristic analysis. Multiple logistic regression analysis was performed to generate a predictive model. The model was internally validated with the bootstrap method and was externally validated with a single-center sample (April 2016-December 2018). Results A predictive model was generated from 71 patients (mean age, 67 years +/- 11 [standard deviation]; 41 [58%] men) and was externally validated in 32 patients (mean age, 72 years +/- 13; 19 [59%] men). The composite of initial DW imaging volume of less than 10 mL (odds ratio [OR], 19.3; 95% confidence interval [CI]: 3.0, 126.4; P = .002), onset-to-puncture time of less than 8 hours (OR, 8.7; 95% CI: 1.8, 42.0; P = .007), and branching-site occlusion (OR, 6.1; 95% CI: 1.5, 26.0; P = .01) could be used to predict good outcomes, with a median area under the receiver operating characteristic curve of 0.86 (interquartile range [IQR], 0.77-0.95; bootstrap optimism-corrected C statistic, 0.837) in the derivation sample and 0.78 (IQR, 0.62-0.95) in the validation sample. Results failed to show an association between collateral status and outcome (P = .67). Conclusion When selecting patients with vertebrobasilar occlusion for endovascular treatment, the combination of onset-to-puncture time of less than 8 hours, initial infarct volume of less than 10 mL, and presence of branching-site occlusions is indicative of a good outcome.