Logo image
Back
Abstract A078: Development of Prognostic Models and Nomograms for Independent Ambulation After Endovascular Thrombectomy in the SELECT2 Trial
Abstract   Peer reviewed

Abstract A078: Development of Prognostic Models and Nomograms for Independent Ambulation After Endovascular Thrombectomy in the SELECT2 Trial

Santiago Ortega-Gutierrez, Muhammad Hussain, Michael Chen, Scott Kasner, Spiros Blackburn, Juan Arenillas, Fawaz Al-Mufti, Thanh Nguyen, Hannah Johns, Alhassin Al Mostaneer, …
Stroke (1970), Vol.57(Suppl_1), A078
02/2026
DOI: 10.1161/str.57.suppl_1.A078

View Online

Abstract

Background: While endovascular thrombectomy(EVT)improved outcomes for large ischemic strokes in randomized trials, only one in three EVT patients achieved mRS 0-3. Accurate prognostication for patient selection, counseling, and management is difficult with currently available tools and not readily applicable to clinical practice. Methods: From SELECT2 randomized trial, EVT patients were analyzed. A pre-thrombectomy model using baseline, and a post-thrombectomy model with both baseline and post-procedural variables were developed through backward stepwise regression with relevant clinical and imaging variables to predict mRS 0-3. Discrimination, Hosmer Lameshow goodness of fit, calibration analyzed, 5-fold crossvalidation performed for internal validation and nomograms were developed. Results: Tissue volume with <26HU(strongest predictor), CT hypodensity and CTP/MRI core did not show significant collinearity (VIFs <2), and were selected for pre-thrombectomy model, along with age and presentation NIHSS and demonstrated strong prognostic association with mRS 0-3. The model showed good discrimination - AUC=0.83 - fig1a/b, calibration and fit(GOF p=0.78), with equivalent crossvalidation performance(AUC=0.81-fig1c). The post-thrombectomy model included 24h NIHSS(strongest predictor), age and mTICI2c-3, along with CTP/MRI core, tissue volume with <26HU and follow-up lesion volume. The resultant model showed excellent discrimination(AUC=0.92-fig2a/b), calibration, and fit(GOF p=0.99); and equivalent crossvalidation performance(AUC=0.90-fig2c). Nomograms (figures 3a/b) had scores ranging from 0-35, with score <15 associated with <5% probability and score >30 associated with >99% probability of mRS 0-3 for pre-; and scores <18 associated with <5% and >28 associated with >99% probability of mRS 0-3 for post-thrombectomy model. These nomograms will undergo external validation using prospective data to be presented at ISC 2026. Conclusions: Prognostication scales on pre- and post-EVT nomograms demonstrated that patients with larger ischemic lesions, older age and higher stroke severity had lower likelihood of independent ambulation. These nomograms will be externally validated through prospective data to assist in pre- and early post-procedure prognostication and represent an important advance toward individualized, evidence-based decision-making with significant implications in large-core stroke.
 Stroke Neuroendovascular (for stroke) Prognosis

Details

Metrics

1 Record Views
Logo image