A cute large vessel occlusion strokes (LVOS) account for 25% to 40% of all acute ischemic strokes, resulting in significant morbidity and mortality. Data from a natural history study of patients with LVOS with moderate-severe deficits (National Institutes of Health Stroke Scale [NIHSS] ≥10) also suggested low rates of good outcome (modified Rankin Score [mRS], 0-2) in internal carotid artery (14%) and M1 middle cerebral artery (23%) occlusions.1 Proximal occlusions are also less likely to recanalize with intravenous thrombolysis, and clinical trials are underway to determine whether endovascular therapy improves clinical outcome. 2,3 In anterior circulation LVOS patients, the final infarct volume (FIV) is a strong and independent predictor of clinical outcome.
4,5After endovascular therapy for LVOS, aggressive medical measures including continuation of life support and surgical therapies (hemicraniectomy, gastrostomy, and tracheostomy) are frequently considered. The decision to pursue aggressive treatment versus de-escalation of care often depends on an estimation of the chance of long-term functional independence. Stroke and critical care physicians are the ones guiding families in this complex decision-making process. Results from the Clinician Judgment vs Risk Score to Predict Stroke Outcomes Background and Purpose-Prognostication tools that predict good outcome in patients with anterior circulation large vessel occlusions after endovascular therapy are lacking. We aim to develop a tool that incorporates clinical and imaging data to predict outcomes after endovascular therapy. Methods-In a derivation cohort of anterior circulation large vessel occlusion stroke patients treated with endovascular therapy within 8 hours from time last seen well (n=247), we performed logistic regression to identify independent predictors of good outcome (90-day modified Rankin Scale, 0-2). Factors were weighted based on β-coefficients to derive the Pittsburgh Outcomes After Stroke Thrombectomy (POST) score. POST was validated in an institutional endovascular database (University of Pittsburgh Medical Center, n=393) and the Diffusion-Weighted Imaging Evaluation for Understanding Stroke Evolution Study-2 (DEFUSE-2) data set (n=105), as well as in patients treated beyond 8 hours (n=194) and in octogenarians (n=111). Results-In the derivation cohort, independent predictors (P<0.1) of good outcome included 24-to 72-hour final infarct volume (in cm 3 , P<0.001), age (years, P<0.001), and parenchymal hematoma types 1 and 2 (H, P=0.01). POST was calculated as age+0.5×final infarct volume+15×H. Patients with POST score <60 had a 91% chance of good outcome compared with 4% with POST score ≥120. POST accurately predicted good outcomes in the derivation (area under the curve [AUC]=0.85) and validation cohorts (University of Pittsburgh Medical Center, AUC=0.81; DEFUSE-2, AUC=0.86), as well as in patients treated beyond 8 hours (AUC, 0.85) and octogenarians (AUC=0.76). POST had better predictive accuracy for good and poor outcome than the isch...