基于多模态CT净水摄取联合临床评分预测急性缺血脑卒中结局的研究

A Study on the Net Water Uptake Based on Multimodal CT and Clinical Scores for Predicting the Outcome of Acute Ischemic Stroke

目的:CT脑灌注净水摄取定量联合临床评分构建影像模型、临床模型、临床联合影像三个预测模型,寻找最佳预测模型来评估急性缺血性脑卒中患者临床预后的情况。方法:回顾性收集入院行血管内治疗和/或静脉溶栓的急性前循环缺血性脑卒中患者193例。通过多因素逐步Logistic回归确定急性缺血性卒中预后不良发生的独立影响因素,构建临床、影像、临床联合影像的预测模型,应用ROC曲线评估临床、影像、临床联合影像模型的预测效能。结果:多因素Logistic回归结果显示,在临床因素中基线NIHSS评分(OR = 1.261;95% CI: 1.1591.372;P < 0.001)、NLR (OR = 1.593;95% CI: 1.2482.034;P < 0.001)为卒中预后不良的独立预测因子;临床预测预后的模型AUC为0.917 (95% CI: 0.877~0.957),敏感度和特异度分别为86.6%、84.0%。在影像因素中核心梗死体积(OR = 1.050;95% CI: 1.0241.078;P < 0.001)、NWU (OR = 1.231;95% CI: 1.1301.341;P < 0.001)为卒中预后不良的独立预测因子;影像预测预后模型的AUC为0.921 (95% CI: 0.884~0.958),敏感度和特异度分别为81.3%、90.1%。在临床联合影像因素中基线NIHSS评分(OR = 1.181, 95% CI: 1.0581.318, P = 0.003)、NLR (OR = 1.352, 95% CI: 1.0771.696, P = 0.009)、核心梗死体积(OR = 1.042, 95% CI: 1.0081.077, P = 0.014)及NWU (OR = 1.247, 95% CI: 1.1151.393, P < 0.001)均是急性缺血性卒中患者预后不良的独立预测因子;临床联合影像的预测预后模型的AUC最高,达0.964 (95% CI: 0.939~0.989),其敏感度为87.5%、特异度为96.3%。结论:临床联合影像的预测模型相较于单独临床模型或影像模型对于血管内治疗和/或静脉溶栓的急性缺血性卒中患者有更好的预测效能。

Objective: Three prediction models, namely imaging model, clinical model and clinical combined imaging, were constructed by CT cerebral perfusion water uptake quantification and clinical score, and the best prediction model was found to evaluate the clinical prognosis of patients with acute ischemic stroke. Methods: Retrospective collection of 193 patients with acute anterior circulation ischemic stroke admitted for endovascular treatment and/or intravenous thrombolysis. Determine independent influencing factors for poor prognosis of acute ischemic stroke through stepwise lo-gistic regression with multiple factors. Build a predictive model for clinical, imaging, and clinical joint imaging. ROC curve was used to evaluate the prediction efficiency of clinical, imaging and clin-ical combined imaging models. Results: Multivariate logistic regression results showed that among clinical factors, baseline NIHSS score (OR = 1.261;95% CI: 1.159~1.372;P < 0.001) and NLR (OR = 1.593;95% CI: 1.248~2.034;P < 0.001) were independent predictors of poor stroke prognosis among clinical factors;The AUC of the clinical prediction model for prognosis was 0.917 (95% CI: 0.877~0.957), with sensitivity and specificity of 86.6% and 84.0%, respectively. Among imaging factors, core infarction volume (OR = 1.050;95% CI: 1.024~1.078;P < 0.001) and NWU (OR = 1.231;95% CI: 1.130~1.341;P < 0.001) are independent predictors of poor stroke prognosis;The AUC of the imaging prediction prognosis model was 0.921 (95% CI: 0.884~0.958), with sensitivity and specificity of 81.3% and 90.1%, respectively. Among clinical joint imaging factors, baseline NIHSS score (OR = 1.181, 95% CI: 1.058~1.318, P = 0.003), NLR (OR = 1.352, 95% CI: 1.077~1.696, P = 0.009), core infarction volume (OR = 1.042, 95% CI: 1.008~1.077, P = 0.014), and NWU (OR = 1.247, 95% CI: 1.15~1.393, P < 0.001) are independent predictors of poor prognosis in patients with acute ischemic stroke;the AUC of the clinical combined imaging prediction prognosis model is the highest, reaching 0.964 (95% CI: 0.939~0.989), with a sensitivity of 87.5% and a specificity of 96.3%. Con-clusion: The predictive model of clinical joint imaging has better predictive performance compared to individual clinical models or imaging models for acute ischemic stroke patients undergoing in-travascular therapy and/or intravenous thrombolysis.