ST段抬高型心肌梗死患者介入术后发生慢血流/无复流的预测模型分析

A prediction model for the occurrence of slow-flow/no-reflow after percutaneous coronary intervention in ST-segment elevation myocardial infarction patients

目的:探讨ST段抬高型心肌梗死(ST-segment elevation myocardial infarction,STEMI)在PCI治疗后发生慢血流/无复流会造的严重不良心血管事件并进行预测并评估预测效能。方法:回顾性纳入于2018年1月1日至2023年7月1日,吉林大学中日联谊医院行PCI的STEMI患者,根据慢血流/无复流发生情况分为未出现组(n=318)和出现组(n=61)。比较两组基线资料,运用Logistic逐步回归法筛选预测变量,分析慢血流/无复流的相关影响因素,并绘制预测慢血流/无复流的列线图。采用Bootstrap法绘制校正曲线进行内部验证,采用受试者工作特征曲线(ROC)分析和决策曲线分析(DCA)对预测模型进行评价。结果:本研究共纳入379例患者,慢血流/无复流的发生率为16.1%(61/379)。基线组间存在差异有统计学意义的变量为:Killip分级、GRACE评分、GRACE评分风险程度、发病至球囊开通时间、LVEF、中性粒细胞百分比、LDL-C、全身炎症反应指数(systemic inflammatory response index, SIRI)评分,通过Logistic逐步回归法进一步筛选后,选出Killip分级、GRACE评分风险程度、发病至球囊开通时间、LVEF、LDL-C、SIRI评分,均是慢血流/无复流的影响因素。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0.799,预测慢血流/无复流的ROC下面积(AUC)为0.799,敏感度为80.3%,特异度为72.3%;内部验证的重抽样校准曲线表明模型理想曲线和实际曲线拟合良好;临床决策曲线显示该预测模型具有较好的临床净效益。结论:基于SIRI、Killip分级、GRACE评分风险程度、发病至球囊开通时间、LVEF、LDL-C构建了STEMI患者PCI后慢血流/无复流的列线图预测模型,该预测模型具有较高区分度、校准度、预测效能以及临床净获益,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:The occurrence of slow-flow/no-reflow after PCI in ST-segment elevation myocardial infarction(STEMI)results in a serious adverse cardiovascular prognosis.The aim of this study was to construct a covariate prediction model to predict the risk of slow-flow/no-reflow after PCI in STEMI patients and evaluate the predictive efficacy.Methods:This study retrospectively recruited STEMI patients after PCI admitted to the China-Japan Union Hospital of Jilin University from January 1,2018 to July 1,2023 and divided them into non-presenting group(n=318)and presenting group(n=61)according to the occurrence of slow-flow/no-reflow.The baseline information of the two groups was compared.Logistic stepwise regression was used to screen predictor variables,and multivariate Logistic regression was used to construct related influencing factors of slow-flow/no-reflow.R software was used to plot predictive column chart of slow-flow/no-reflow.Bootstrap method was used to plot calibration curves for internal validation.Receiver operating characteristic(ROC)analysis and decision curve analysis(DCA)were used to assess the prediction model.Results:A total of 379 STEMI patients after PCI were included in this study,and the incidence of slow-flow/no re-reflow was 16.1%(61/379).Variables that differed between the baseline groups were Killip classification,global registry of acute coronary events(GRACE)score,degree of risk of GRACE score,time from onset of balloon,LVEF,percentage of neutrophils,LDL-C,and systemic inflammatory response index(SIRI)and other baseline indices score.6 predictor variables were selected as independent influences on slow-flow/no-reflow after further screening by Logistic stepwise regression:Killip classification,degree of risk of GRACE score,time from onset of balloon,LVEF,LDL-C,and SIRI score.The predictive column chart model of slow-flow/no-reflow plotted by R software showed a predictive risk capability index(C-index)of 0.799,an area under the ROC(AUC)of 0.799 for predicting slow-flow/no-reflow,a sensitivity of 80.3%,and a specificity of 72.3%.The resampling calibration curve of internal validation indicates that the predicted curve of the model fitted well with the actual curve.Clinical decision curve showed that the model was of good clinical benefit.Conclusions:In this study,we constructed a predictive column chart model of slow-flow/no-reflow after PCI in patients with STEMI based on SIRI score as well as Killip classification,degree of risk of GRACE score,time from onset of balloon,LVEF,and LDL-C.This prediction model,which has a high degree of discrimination,calibration,predictive efficacy,and net clinical benefit,can be used as a prediction model of slow-flow/no-reflow to support the decision-making in clinical management.