Predicting hemorrhagic transformation after thrombectomy in acute ischemic stroke: a multimodal score of the regional pial collateral.

Abstract

This study aims to analyze the multimodal score of the regional pial collateral in predicting hemorrhagic transformation (HT) after mechanical thrombectomy in acute ischemic stroke (AIS). On the basis of different brain regions and multiphase computed tomography angiography (mCTA), we evaluated the pial arterial filling status in extent, delay, and contrast washout. The prediction models of HT and symptomatic intracerebral hemorrhage (sICH) were established using mCTA (model-H1 and model-S1), CT perfusion (CTP, model-H2 and model-S2), and comprehensive parameters (model-H3 and model-S3). The receiver operating characteristic curve was used to analyze the prediction performance of each model. Among the 102 patients with AIS who received thrombectomy, 36 (35.3%) developed HT, and 15 (14.7%) of whom had sICH. In model-H1 and model-S1, washout independently influenced HT (OR, 95%CI 0.398, 0.249-0.634) and sICH (OR, 95%CI 0.552, 0.342-0.892). In model-H2, the relative surface permeability independently influenced HT (OR, 95%CI 1.217, 1.082-1.370). Model-H3 and model-S3 improved the prediction performance (areas under the curve: HT, 0.957; sICH, 0.938). The correlation coefficients between relative cerebral blood volume and the three modes of pial arterial filling status were higher than those of other CTP parameters. The 90-day modified Rankin scale score in the sICH group was significantly increased (P < 0.05). The multimodal regional pial collateral score has good value in the risk assessment of HT and sICH in patients with AIS after mechanical thrombectomy. The combination of multiple parameters can improve diagnostic performance.