冠心病血瘀证PCI术后1年MACEs风险临床预测模型的建立与验证

Establishment and Validation of Clinical Prediction Model for 1-year MACEs Risk After PCI in CHD Patients with Blood Stasis Syndrome

目的:构建并验证冠心病血瘀证PCI术后1年主要不良心血管事件(MACEs)风险临床预测模型。方法:回顾性收集中日友好医院中西医结合心内科2019年9月1日至2021年3月31日确诊冠心病血瘀证并首次接受经皮冠状动脉介入治疗(PCI)的连续病例,统一采集基本临床特征和相关检验检查指标,按7∶3比例分为训练集和验证集,并根据1年内是否发生MACEs进一步把各组分为MACE组和非MACE组。采用最小绝对收缩选择算子(Lasso)回归分析筛选冠心病血瘀证PCI术后1年MACEs预测因素,构建多因素Logistic回归临床预测模型并筛选独立影响因素。通过Hosmer-Lemeshow检验评价模型拟合优度,采用受试者工作特征(ROC)曲线下面积(AUC)、校准曲线、策曲线分析(DCA)和临床影响曲线(CIC)评估模型的区分度、校准度及临床影响力。结果:共筛选连续病例731例,最终纳入404例,训练集283例,验证集121例。Lasso回归分析筛选出年龄、性别、空腹血糖(FPG)、甘油三酯(TG)、低密度脂蛋白胆固醇(LDL-C)、同型半胱氨酸(Hcy)、肱踝动脉脉搏波速度(baPWV)、血管舒张功能(FMD)、左室射血分数(LVEF)、Gensini积分共10个变量为影响结局的因素。通过多因素Logistic回归分析,初步判断男性、高FMD、高LVEF是保护因素,年龄、高FPG、高TG、高LDL-C、高Hcy、高baPWV、高Gensini积分是危险因素。其中,年龄、FPG、TG、Hcy、LDL-C、LVEF和Gensini积分为独立影响因素。构建出冠心病血瘀证PCI术后1年MACEs临床预测模型,Hosmer-Lemeshow检验得到χ^(2)=12.371,P=0.14;AUC为0.90;阈值概率>10%时,使用该预测模型预测冠心病血瘀证PCI术后1年MACEs风险比对所有患者实施干预方案更有利,阈值>60%时,阳性估计值越接近真实患病数。结论:此模型有助于早期识别冠心病血瘀证PCI术后1年MACEs风险,便于进行早期临床决策和干预,对改善患者预后有重要指导意义。同时期待更多中心、更大样本的研究对该模型进行进一步验证、完善和更新。

Objective:To establish and validate a clinical prediction model for 1-year major adverse cardiovascular events(MACEs)risk after percutaneous coronary intervention(PCI)in coronary heart disease(CHD)patients with blood stasis syndrome.Method:The consecutive CHD patients diagnosed with blood stasis syndrome in the Department of Integrative Cardiology at China-Japan Friendship Hospital from September 1,2019 to March 31,2021 were selected for a retrospective study,and basic clinical features and relevant indicators were collected.Eligible patients were classified into a derivation set and a validation set at a ratio of 7∶3,and each set was further divided into a MACEs group and a non-MACEs group.The factors affecting the outcomes were screened out by least absolute shrinkage and selection operator(Lasso)and used to establish a logistic regression model and identify independent prediction variables.The goodness-of-fit of the model was evaluated by the Hosmer-Lemeshow test,and the area under curve(AUC)of the receiver operating characteristic(ROC)curve,calibration curve,decision curve analysis(DCA),and clinical impact curve(CIC)were employed to evaluate the discrimination,calibration,and clinical impact of the model.Result:A total of 731 consecutive patients were assessed and 404 eligible patients were enrolled,including 283 patients in the derivation set and 121 patients in the validation set.Lasso identified ten variables influencing outcomes,which included age,sex,fasting plasma glucose(FPG),triglyceride(TG),low-density lipoprotein cholesterol(LDL-C),homocysteine(Hcy),brachial-ankle pulse wave velocity(baPWV),flow-mediated dilatation(FMD),left ventricular ejection fraction(LVEF),and Gensini score.The multivariate Logistic regression preliminarily identified age,FPG,TG,Hcy,LDL-C,LVEF,and Gensini score as the independent variables that influenced the outcomes.Of these variables,male,high FMD and high LVEF were protective factors,and the rest were risk factors.The prediction model for 1-year MACEs risk after PCI in CHD patients with blood stasis syndrome showedχ^(2)=12.371(P=0.14)in Hosmer-Lemeshow test and the AUC of 0.90.With the threshold probability>10%,the model showed better prediction performance for 1-year MACEs risk after PCI in CHD patients with blood stasis syndrome than for that in all the patients.With the threshold probability>60%,the estimated value was much closer to the real number of patients.Conclusion:The established clinical prediction model facilitates the early prediction of 1-year MACEs risk after PCI in CHD patients with blood stasis syndrome,which can provide ideas for the precise treatment of CHD patients after PCI and has guiding significance for improving the prognosis of the patients.Meanwhile,multi-center studies with larger sample sizes are expected to further validate,improve,and update the model.