A comprehensive assessment of brain perfusion data in patients with acute ischemic stroke for prediction of hemorrhagic transformation

Abstract

Based on propensity score matching, the authors selected retrospectively 21 case-control pairs from 71 consecutive patients with AIS in the middle cerebral artery territory. Brain computed tomography (CT) perfusion data with blood-brain barrier permeability assessment were obtained in all patients within 12 h after the onset. The infarct core (IC) and penumbra were evaluated with mean transit time (MTT), cerebral blood volume (CBV), cerebral blood flow (CBF), and permeability surface-area product (PS). The outcome was any HT on a follow-up brain CT scan within 2 weeks after the onset. The Firth logistic regression analysis showed that PS was an independent HT predictor in the IC (odds ratio, 8; 95% confidence interval (CI), 1.32-48.4; p=0.023). The cut-off value was 2.88 mL/100 g/min (95% normal-based CI (NB-CI): 2.41-3,34) with sensitivity 0,95 (95% NB-CI: 0.87-1.0), specificity 1 (95% NB-CI: 0.95-1.0), and area under ROC-curve 0.98 (95% NB-CI: 0.94-1.0). However, no independent HT predictor was found in the penumbra. The generalized linear model analysis revealed that the HTI score was a predictor of CBV, CBF, and PS in the IC and penumbra. As the HT risk grew simultaneously with the HTI score increment, the CBV and CBF became low with a substantial PS rise in the IC; the IC size tended to increase as well. In the penumbra, there was a progressive CBF reduction with a significant CBV and PS climb. At the same time, the MTT-CBV mismatch shrank. The IC is more likely to be the site of HT. The PS is an independent HT predictor in the IC. The HTI score can predict the HT probability as well as the brain perfusion data. As the HT risk increases, the perfusion disorders become worse in the IC and penumbra.