Do scoring systems help us to estimate prognosis after mechanical thrombectomy? Data from the German Stroke Registry

Abstract

Background Numerous scoring systems have been developed to individualize estimation of functional outcome after endovascular thrombectomy (EVT) of acute ischemic stroke. The aim of our study was to assess their utility for clinical practice based on a large cohort from real-world care of EVT.

Methods For 13 082 patients included in the German Stroke Registry Endovascular Treatment (GSR-ET) (July 2015 to December 2021), we calculated the following prognostic tools: pre-interventional PRE-, Totaled Health Risks in Vascular Events – Endovascular therapy (THRIVE-EVT)- and Computed Tomography for Late Endovascular Reperfusion (CLEAR) scores and post-interventional MR PREDICTS@24 hours and BET-score. Area under the receiver operating characteristic curve (AUC) analyses in the total cohort and pre-defined subgroups were performed to determine each tool’s prognostic value for good functional outcome (modified Rankin Scale (mRS) 0–2) and mortality at 90-day follow-up.

Results All pre-interventional tools achieved a moderate prognostic value for predicting good functional outcome (PRE: AUC (95% confidence interval): 0.757 (0.747–0.768), THRIVE-EVT: 0.751 (0.740–0.761), CLEAR: 0.731 (0.72–0.742)), had a higher predictive value than the admission National Institute of Health Stroke Scale ((NIHSS); 0.705 (0.694–0.716), all P<0.001), but were inferior to the NIHSS 24 hours after EVT (0.864 (0.855–0.872), all P<0.001). Predictive capacity for mortality was less accurate (AUC range: 0.697–0.729). Subgroup analyses revealed that the PRE-score was most robust at predicting good functional outcome, whereas the THRIVE-EVT score was superior in predicting mortality. Post-interventionally, MR PREDICTS@24 hours yielded high predictive accuracy for good functional outcome and mortality (both AUC >0.85), superior to 24-hour NIHSS for all subgroups, except patients <50 years of age.

Conclusion Pre-interventional scoring tools predict functional outcome after EVT better than stroke severity alone. Post-interventionally, the MR PREDICTS@24 hours tool adds predictive value to the 24-hour NIHSS as a single prognostic feature. Multivariate prognostic tools incorporating (post-)procedural information enable individualization of prognosis assessment after EVT under routine-care conditions.