作者：夏旺旭1，张明2，何永芳1，陈肖波1，黄传芬1

单位：

1.401220重庆市长寿区人民医院放射科　2.637600四川省南充市仪陇县人民医院神经内科

Units：

1.Department of Radiology, Chongqing Changshou District People's Hospital, Chongqing 401220, China2.Department of Neurology, Yilong County People's Hospital, Nanchong 637600, China

关键词：

缺血性卒中；CT灌注成像；预后；列线图

Keywords：

Ischemic stroke; CT perfusion imaging; Prognosis; Nomogram

CLC：

R 743.3

DOI：

10.12114/j.issn.1008-5971.2022.00.239

Funds：

川北医学院2021年度四川省基层卫生事业发展研究中心资助项目（SWFZ 21-Y-35）

摘要：

目的　分析缺血性脑卒中患者预后不良的影响因素，并构建其风险预测列线图模型。方法　选择2019年1月至2021年6月重庆市长寿区人民医院接诊的165例缺血性脑卒中患者作为研究对象，根据随访6个月后格拉斯哥预后量表（GOS）分级将其分为预后良好组（4~5级，n=124）和预后不良组（1~3级，n=41）。比较两组CT灌注成像指标和临床资料。采用LASSO回归分析筛选协变量，采用多因素Logistic回归分析探讨缺血性脑卒中患者预后不良的影响因素。采用R 4.1.3语言“rms”包构建缺血性脑卒中患者预后不良风险预测的列线图模型，绘制ROC曲线以评价该列线图模型的区分度，采用Bootstrap法重复抽样1 000次进行内部验证，计算一致性指数（CI ），采用H-L拟合优度检验、校准曲线评价该列线图模型的校准度，绘制决策曲线以评价该列线图模型的临床有效性。结果　两组相对平均通过时间（rMTT）、相对峰值时间（rTTP）、年龄、入院时美国国立卫生研究院卒中量表（NIHSS）评分、血管闭塞位置、吸烟史、侧支循环情况及改良脑梗死溶栓分级比较，差异有统计学意义（P ＜0.05）。采用LASSO回归模型筛选出5个潜在的影响因素，分别为rMTT、rTTP、年龄、入院时NIHSS评分、侧支循环情况。多因素Logistic回归分析结果显示，rMTT、rTTP、年龄、入院时NIHSS评分、侧支循环情况是缺血性脑卒中患者预后不良的独立影响因素（P ＜0.05）。基于上述5个影响因素构建缺血性脑卒中患者预后不良风险预测的列线图模型，ROC曲线分析结果显示，该列线图模型预测缺血性脑卒中患者预后不良的AUC为0.946〔95%CI （0.904，0.988）〕；采用Bootstrap法重复抽样1 000次，结果显示，CI 为0.913；H-L拟合优度检验结果显示，该列线图模型预测缺血性脑卒中患者预后不良的发生率与患者实际发生率比较，差异无统计学意义（χ2=2.177，P =0.140）；校准曲线分析结果显示，该列线图模型预测缺血性脑卒中患者预后不良的校准曲线接近于理想曲线；决策曲线分析结果显示，当该列线图模型预测缺血性脑卒中患者预后不良的概率阈值为0.15~0.95时，患者的净获益率大于0。结论　rMTT、rTTP、年龄、入院时NIHSS评分、侧支循环情况是缺血性脑卒中患者预后不良的独立影响因素，而基于上述影响因素构建的列线图模型对缺血性脑卒中患者不良预后具有较高的区分度及校准度，有助于临床医生早期识别预后不良的缺血性脑卒中患者。

Abstract：

Objective　To analyze the influencing factors of poor prognosis in patients with ischemic stroke, andconstruct the nomograph model for its risk prediction. Methods　A total of 165 ischemic stroke patients admitted to theChongqing Changshou District People's Hospital from January 2019 to June 2021 were selected as the research subjects. Thepatients were divided into good prognosis group (grade 4-5, n=124) and poor prognosis group (grade 1-3, n=41) according tothe Glasgow Outcome Scale (GOS) grading after 6-month follow-up. The CT perfusion imaging indexes and clinical data were compared between the two groups. LASSO regression analysis was used to screen covariates, and multivariate Logistic regressionmodel was used to analyze the influencing factors of poor prognosis in patients with ischemic stroke. The nomogram model forpredicting the risk of poor prognosis of ischemic stroke patients was constructed by the "rms" package of R 4.1.3 language, andthe ROC curve was drawn to evaluate the discrimination of the nomogram model. Bootstrap method was used to repeatedly sample1 000 times for internal verification, and the consistency index (CI ) was calculated. H-L goodness of fit test and calibration curvewere used to evaluate the calibration of the nomogram model, the decision curve was drawn to evaluate the clinical effectivenessof the nomogram model. Results　There were significant differences in relative mean transit time (rMTT) , relative peak time(rTTP) , age, National Institutes of Health Stroke Scale (NIHSS) score on admission, vascular occlusion location, smoking history,collateral circulation and modified cerebral infarction thrombolysis grade between the two groups (P < 0.05) . LASSO regressionmodel was used to screen out five potential influencing factors, namely rMTT, rTTP, age, NIHSS score on admission, and collateralcirculation. The results of multivariate Logistic regression analysis showed that rMTT, rTTP, age, NIHSS score on admission, andcollateral circulation were independent influencing factors of poor prognosis in patients with ischemic stroke (P < 0.05) . Basedon the above five influencing factors, a nomogram model for predicting the risk of poor prognosis in ischemic stroke patients wasconstructed, the ROC curve analysis showed that the AUC of the nomogram model for predicting poor prognosis in patients withischemic stroke was 0.946 [95%CI (0.904, 0.988) ] . The Bootstrap method was used to sample 1 000 times, and the results showedthat the CI was 0.913; the H-L goodness of fit test showed that, there was no significant difference between the incidence of poorprognosis predicted by the nomogram model and the actual incidence of patients with ischemic stroke (χ2=2.177, P =0.140) ; thecalibration curve results showed that the calibration curve of the nomogram model for predicting poor prognosis in patients withischemic stroke was close to the ideal curve; the results of decision curve analysis showed that when the high risk threshold of thenomogram model for predicting the poor prognosis of patients with ischemic stroke was 0.15-0.95, the standardized net benefit ofpatients was greater than 0. Conclusion　rMTT, rTTP, age, NIHSS score on admission, and collateral circulation are independentinfluencing factors of poor prognosis in patients with ischemic stroke. The nomogram model constructed based on the aboveinfluencing factors has a high degree of discrimination and calibration for the poor prognosis of ischemic stroke patients, which ishelpful for clinicians to identify ischemic stroke patients with poor prognosis early.

ReferenceList：