Abstract DP025: Differential Actionable Information Yield and Prognostic Value of Perfusion Imaging Ischemic Core Volumes Among EVT Patients with Good and Poor Collaterals – A Secondary Analysis of SELECT2 and SELECT Trial

Le résumé fourni par la source

Introduction: Collateral status and perfusion imaging both capture downstream flow in the occluded territory for patients with acute ischemic stroke. Collateral status can be obtained from CT/MR Angiography, which is part of the standard workflow. Perfusion imaging provides quantitative measures of blood flow and volume restrictions but requires specific set-up and commercial post-processing software. Both suffer from failure to capture changes that happened over time. Objective: We assessed if actionable information yield and prognostic utility of ischemic core measures vary based on collateral status among patients receiving Endovascular Thrombectomy(EVT) in SELECT2 with independent validation in SELECT. Methods: EVT SELECT 2 patients were stratified into those with good(Tan collateral score[TCS] of 2-3) and poor collaterals(TCS of 0-1). Ischemic core volumes were obtained from CTP processed with RAPID using rCBF<30% threshold. Multivariable logistic regression models were created using age, stroke severity and time from last-known-well to procedure, without (base) and with(base + CTP core) ischemic core measures. Discrimination (AUC) and model fit using Bayesian Information Criteria(BIC) were compared, BIC reductions of 2–6, 6–10, and >10 were taken as positive, strong, and very strong evidence of superior fit, respectively. SELECT trial data provided independent validation. Results: Among SELECT 2 patients, 121/178 (68%) had poor collaterals . 73/121 (61%) poor collaterals demonstrated CTP core ≥70 ml and 46(38%) ≥100 ml, whereas only 19/57 (33%) good collaterals had core ≥70 ml and 8 (14%) ≥100 ml. In poor collaterals, adding CTP core increased AUC by 0.05–0.10 and reduced BIC by 5–12 points (strong to very strong improvement) across mRS 0–2, mRS 0–3, and mortality. In good collaterals, AUC gains were minimal (0.01–0.03) (Fig 1), and BIC changes were negligible or worsened, indicating little incremental value of core. (Fig 3). SELECT showed qualitatively similar effects: poor collaterals benefited (BIC −6 to −13; AUC +0.04-0.05), while good collaterals showed ≤1-point BIC change and trivial AUC gains.(fig 2-3). Conclusions: Actionable information yield and prognostic value of CTP core volume appeared to be higher among patients with poor collaterals as compared to those with good collaterals. These findings need to be confirmed in further studies and may help with CTP acquisition decision-making for optimizing workflow and/or in limited resource settings.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.