Comparison of Hemorrhage on CT Versus MRI After Thrombectomy: The HECATE Study

Abstract

BACKGROUND The characterization of hemorrhage following acute stroke intervention has largely been computed tomography (CT) based. We sought to compare magnetic resonance imaging‐ (MRI‐) and CT‐based scoring of hemorrhage after acute endovascular therapy (EVT) applying the Heidelberg Bleeding Classification (HBC) to assess intermodal agreement and quantify interrater agreement. METHODS Consecutive patients with acute stroke were included in this retrospective study if they (1) had MRI and CT ≤12 hours of each other OR (2) had CT bracketed by MRI pre‐ and post‐CT (ie, MRI‐CT‐MRI) ≤7 days post‐EVT. The concordance of the HBC ratings by consensus panel were compared between CT and T2 * gradient recalled echo MRI. RESULTS For the 87 EVT‐treated patients included, median age was 68 years [60–74], admission National Institutes of Health Stroke Scale score 18 [13–23], 47% were treated with intravenous/intraarterial thrombolytics, and 93% were successfully recanalized (modified Thrombolysis in Cerebral Infarction 2b/3). Hemorrhage was detected on at least 1 modality in 60% (52/87) of patients. We found a 68% (59/87, 95% CI [57%–77%]) agreement overall between CT and MRI for hemorrhage classification post‐EVT. MRI had the best interrater agreement for HBC 0 (no hemorrhage) with excellent concordance (ĸ = 0.882), compared with CT (ĸ = 0.683). T2 * gradient recalled echo MRI tended to have increased sensitivity to scattered petechial hemorrhage (HBC 1a) as compared with CT with 17% (2/12) intermodal agreement. The interrater agreement of HBC class 2 (ie, parenchymal hematoma grade 2 was substantial for MRI (ĸ = 0.781) and excellent in CT (ĸ = 0.951), with 67% (8/12) intermodal agreement. Subarachnoid hemorrhage was detected in 24% (21/87) of patients on CT and/or MRI with 29% (6/21) intermodal agreement. CONCLUSION With the exception of subarachnoid hemorrhage and minor petechial hemorrhagic transformation, we found that MRI and CT are overall interchangeable for detecting and classifying hemorrhage after EVT. These findings are reassuring for both clinical decision‐making and research application. Given the complexity of hemorrhage subtypes post‐EVT, work to further refine a post‐EVT hemorrhage classification scale with clinical correlation would be beneficial.