Accuracy of 'At Risk' Tissue Predictions Using CT Perfusion in Acute Large Vessel Occlusions

BACKGROUND AND PURPOSE: The validity of CT perfusion (CTP) predictions of expected infarction volume (“at risk” tissue) without rapid recanalization remains poorly characterized. METHODS: From the MR CLEAN trial, we included patients who underwent CTP without successful recanalization. “At risk” volume was defined as T max > 6 seconds and ischemic core as relative CBF < 30 (Olea Sphere). Coprimary outcomes were follow-up infarct volume (FIV) on CT at 1-5 days and 90-day mRS. Data are presented as median [IQR] or OR [95% CI] unless otherwise specified. RESULTS: Among 37 patients who met criteria, 14 (38%) were women, median age was 61 years [52-69], NIHSS was 19 [15-21], ASPECTS was 8 [7-9], and onset to imaging was 160 minutes [39-200]. Occlusion location was M1 for 22 (59%), ICA-T in 10 (27%), and M2 in 4 (11%). In univariable analysis, “at risk” volume correlated poorly with FIV (r =.06, P =.77). Among patients with predicted “at risk” volume < 100 mL, 36% had FIV > 200 mL. In adjusted linear regression, NIHSS but not “at risk” volume was associated with FIV (Coef 12, P =.045; Coef –.15, P =.8). In adjusted logistic regression, NIHSS but not “at risk” volume was associated with mRS 0-2 at 90 days (OR.7 [.5-.99]; OR 1.0 [.99-1.04]). CONCLUSION: Predictions of “at-risk” tissue using CTP may underestimate the natural history of infarction from acute large vessel occlusions. NIHSS may perform better as a predictor of clinical outcomes in patients without rapid recanalization.