摘要：

目的: 对急性缺血性卒中患者rt-PA静脉溶栓治疗后颅内出血性转化的预测评分系统进行外部验证，评估各评分系统的预测效能。方法: 纳入2010年7月至2023年8月在大连医科大学附属第一医院行rt-PA静脉溶栓的1157例急性缺血性卒中患者，根据治疗36h内影像学检查结果，98例出现颅内出血性转化，34例出现症状性颅内出血，采用受试者工作特征曲线和Hosmer-Lemeshow拟合优度检验评估HAT、MSS、SEDAN、GRASPS和SPAN-100评分预测颅内出血性转化和症状性颅内出血的区分度和校准度。结果: 颅内出血性转化和症状性颅内出血发生率分别为8.47%（98/1157）和2.94%（34/1157）。各评分系统预测颅内出血性转化的曲线下面积为HAT评分0.625（95%CI：0.563~0.687）、MSS评分0.663（95%CI：0.609~0.718）、SEDAN评分0.641（95%CI：0.581~0.702）、GRASPS评分0.701（95%CI：0.644~0.759），预测症状性颅内出血的曲线下面积为HAT评分0.695（95%CI：0.595~0.795）、MSS评分0.691（95%CI：0.609~0.773）、SEDAN评分0.689（95%CI：0.590~0.788）、GRASPS评分0.783（95%CI：0.714~0.852）；HAT、MSS、SEDAN和GRASPS评分预测颅内出血性转化和症状性颅内出血的Hosmer-Lemeshow拟合优度检验均P>0.05，提示校准度良好；SPAN-100评分预测颅内出血性转化和症状性颅内出血的灵敏度分别为14.30%和23.50%，提示其实际应用时漏诊风险较高。结论: GRASPS评分对急性缺血性卒中患者静脉溶栓治疗后颅内出血性转化和症状性颅内出血的预测能力尚可，有待在不同人群中进行大样本多中心外部验证。

关键词: 缺血性卒中, 组织型纤溶酶原激活物, 血栓溶解疗法, 颅内出血, ROC曲线

Abstract:

Objective: To externally validate different predictive scoring systems for intracranial hemorrhagic transformation (HT) after intravenous thrombolysis in patients with acute ischemic stroke, and to evaluate the predictive abilities of the different scoring systems. Methods: A total of 1157 patients with acute ischemic stroke who received rt-PA intravenous thrombolysis within 4.50 h after symptom onset were collected from July 2010 to August 2023 in The First Affiliated Hospital of Dalian Medical University. According to the imaging examination within 36 h after intravenous thrombolysis, 98 cases presented with intracranial HT and 34 cases presented with symptomatic intracranial hemorrhage (sICH). The area under the curve (AUC) of receiver operating characteristic (ROC) curve and the Hosmer-Lemeshow goodness-of-fit test were used to assess the discrimination and calibration of the HAT score, the MSS score, the SEDAN score, the GRASPS score and the SPAN-100 score for the prediction of HT and sICH, respectively. Results: The incidence of intracranial HT and sICH were 8.47% (98/1157) and 2.94% (34/1157), respectively. The AUCs of the scoring systems for predicting intracranial HT were 0.625 (95%CI: 0.563-0.687) for the HAT score, 0.663 (95%CI: 0.609-0.718) for the MSS score, 0.641 (95%CI: 0.581-0.702) for the SEDAN score, and 0.701 (95%CI: 0.644-0.759) for the GRASPS score. In predicting sICH, the AUCs of the scoring systems were 0.695 (95%CI: 0.595-0.795) for the HAT score, 0.691 (95%CI: 0.609-0.773) for the MSS score, 0.689 (95%CI: 0.590-0.788) for the SEDAN score, and 0.783 (95%CI: 0.714-0.852) for the GRASPS score. In addition, the Hosmer-Lemeshow goodness-of-fit tests of the HAT score, the MSS score, the SEDAN score and the GRASPS score were P > 0.05 for both validation of intracranial HT and sICH, suggesting that all scoring systems were well calibrated. The sensitivities of the SPAN-100 score for predicting intracranial HT and sICH were 14.30% and 23.50%, respectively, suggesting that the SPAN-100 score has a high risk of underdiagnosis in clinical practice. Conclusions: The GRASPS score has a fair predictive ability for predicting intracranial HT and sICH after intravenous thrombolysis in acute ischemic stroke patients. Further external validations with large samples in multiple centers will be needed in different populations.

Key words: Ischemic stroke, Tissue plasminogen activator, Thrombolytic therapy, Intracranial hemorrhages, ROC curve