Comparison of predicting scales for symptomatic intracranial hemorrhage after stroke thrombolysis with recombinant tissue plasminogen activator

Objective Symptomatic intracranial hemorrhage (sICH) is one of the severe complications of ischemic stroke thrombolysis. Several prognostic scales have been developed to predict the risk of sICH. The performance of seven scales was compared in a single center cohort. Methods Data of patients with consecutive ischemic stroke who received 0.9 mg/kg intravenous recombinant tissue plasminogen activator (rt-PA) thrombolysis within 4.5 h time window from stroke onset were collected. Seven scales that can provide an estimate of risk of sICH were identified and evaluated: Hemorrhage After Thrombolysis (HAT), blood Sugar, Early infarct signs, (hyper) Dense cerebral artery sign, Age, National Institutes of Health (NIH) Stroke Scale (SEDAN), Stroke Prognostication using Age and NIH Stroke Scale (SPAN)-100, Safe Implementation of Thrombolysis in Stroke (SITS), Total Health Risks In Vascular Events (THRIVE), Glucose at presentation, Race (Asia), Age, Sex (male), systolic blood Pressure at presentation, and Severity of stroke at presentation (NIH Stroke Scale; GRASPS) and Multicenter Stroke Survey (MSS). The area under the receiver operating characteristic curve (AUROC) was calculated and Logistic regression and the Hosmer-Lemeshow test were also performed. Results The current study included 293 patients, of whom 7.85% (23/293) had sICH by National Institute of Neurological Disorders and Stroke (SICHNINDS), 5.46% (16/293) by Europe Cooperative Acute Stroke Study Ⅱ (SICHECASSⅡ) and 4.44% (13/293) by Safe Implementation of Thrombolysis in Stroke (SICHSITS) criteria. SEDAN had the highest AUROC for predicting sICH: sICHNINDS: AUROC=0.843, OR=3.167, 95%CI 2.106-4.762, P<0.01; sICHECASSⅡ: AUROC=0.797, OR=2.509, 95%CI 1.652-3.812, P<0.01; sICHSITS: AUROC=0.784, OR=2.172, 95%CI 1.405-3.357, P<0.01. And SPAN-100 had the lowest AUROC among all the seven scales and was only associated with risk of SICHNINDS in regression analysis. Furthermore, when sub-grouped the cohort into anterior circulation infarction and posterior circulation infarction, regression analysis suggested that all the seven scales were however not associated with sICH risk in patients with posterior circulation infarction. Conclusions SEDAN constantly had the highest predictive power, SPAN-100 had the worst. The seven scales studied could not predict sICH in posterior circulation infarction. Key words: Stroke; Brain ischemia; Tissue plasminogen activator; Thrombolytic therapy; Cerebral hemorrhage