急性心肌梗死患者经皮冠状动脉介入术后冠状动脉慢血流/无复流预测模型的构建研究

Construction of prediction model of coronary slow flow/no reflow after percutaneous coronary intervention in patients with acute myocardial infarction

目的构建急性心肌梗死患者经皮冠状动脉介入术(PCI)术后冠状动脉慢血流/无复流预测模型,为临床早期预测、干预提供参考。方法选取2019年1月~2022年12月芜湖市第一人民医院收治的90例行PCI治疗急性心肌梗死患者,根据是否出现慢血流/无复流分为出现组(n=27)、未出现组(n=63),比较两组临床资料、脂蛋白(a)[LP(a)]、胱抑素C(Cys C)、视黄醇结合蛋白(RBP)水平,运用LASSO回归分析初筛慢血流/无复流的预测因素,运用logistic回归分析慢血流/无复流的相关影响因素,运用R语言绘制预测慢血流/无复流的列线图,运用受试者工作特征曲线(ROC)评价列线图的区分度,运用校准曲线分析列线图的校准度。结果出现组吸烟史、ST段抬高型心肌梗死、发病至PCI时间≥6 h患者占比高于未出现组,Killip分级、全球急性冠状动脉事件注册(GRACE)评分、低密度脂蛋白、血肌酐、LP(a)、Cys C、RBP高于未出现组,收缩压低于未出现组(P<0.05);LASSO回归分析显示,可使模型性能优良且影响因素最少的最佳惩罚系数λ下预测慢血流/无复流的变量数为6个:发病至PCI时间≥6 h、Killip分级、GRACE评分、LP(a)、Cys C、RBP;logistic回归分析显示,发病至PCI时间≥6 h、Killip分级、GRACE评分、LP(a)、Cys C、RBP是慢血流/无复流的相关危险因素(P<0.05);运用R语言绘制预测慢血流/无复流列线图显示,其预测风险能力指数(C-index)为0.958,预测慢血流/无复流的ROC下面积(AUC)为0.958,敏感度为93.70%,特异度为96.30%;校准曲线显示,列线图的校准度为0.879,校正曲线与理想曲线拟合良好,模型与实际观测结果有较好的一致性。结论PCI时间、Killip分级、GRACE评分、LP(a)、Cys C、RBP与急性心肌梗死PCI术后冠状动脉慢血流/无复流有关,基于以上指标所构建的预测模型具有较高区分度、校准度以及预测效能,可作为慢血流/无复流的一个预测方案,为临床管理提供决策支持。

Objective To construct a coronary slow flow/no-reflow prediction model after percutaneous coronary intervention(PCI)for acute myocardial infarction(AMI)containing commonly used clinical information,lipoprotein(a)[LP(a)],cystatin C(Cys C),and retinol-binding protein(RBP),and to provide a reference for early clinical prediction and intervention.Methods Ninety patients admitted to our hospital for acute myocardial infarction treated with PCI from January 2019 to December 2022 were selected and divided into presenting group(n=27)and non-presenting group(n=63)according to whether they presented with slow flow/no recurrent flow,comparing the clinical data,levels of lipoprotein(a)[LP(a)],cystatin C(Cys C),and retinol binding protein(RBP)between the two groups,analyzing the predictors of initial screening slow flow/no recurrent flow using LASSO regression,the correlation factors of slow flow/no recurrent flow were analyzed by using logistic regression,and the column line graphs predicting slow flow/no recurrent flow were plotted by using R language,and the differentiation of the column line graphs was evaluated by using subject operating characteristic curve(ROC),and the calibration of the column line graphs was analyzed by using calibration curve.Results History of smoking,ST-segment elevation myocardial infarction,and time from onset to PCI≥6 h were higher in the presenting group than in the non-presenting group,and Killip score,Global Acute Coronary Event Registration(GRACE)score,low-density lipoprotein,blood creatinine,LP(a),Cys C,RBP were all higher than those in the non-presenting group,and systolic blood pressure was lower than those in the non-presenting group(P<0.05).LASSO regression analysis shows that there were 6 variables that can predict slow blood flow/no reflow under the optimal penalty coefficientλ,which can improve model performance and minimize influencing factors:time from onset to PCI≥6 h,Killip classification,GRACE score,LP(a),Cys C,and RBP;logistic regression analysis showed that time from onset to PCI≥6 h,Killip grade,GRACE score,LP(a),Cys C,and RBP were risk factors associated with slow flow/no recurrent flow(P<0.05);the use of R language to draw a column chart for predicting slow blood flow/no reflow shows that its predictive risk index(C-index)was 0.958,and the area under the ROC(AUC)for predicting slow blood flow/no reflow was 0.958,the sensitivity was 93.70%,and the specificity was 96.30%;the calibration curve shows that the calibration degree of the column chart was 0.879,and the calibration curve fits well with the ideal curve,the model has good consistency with the actual observation results.Conclusion PCI time,Killip classification,GRACE score,LP(a),Cys C,and RBP are associated with coronary slow flow/no regurgitation after PCI for acute myocardial infarction,and the prediction model constructed on the basis of the above indexes has high differentiation,calibration,and prediction efficacy,and can be used as a prediction scheme for slow flow/no regurgitation to provide decision support for clinical management.