Establishment and validation of predictive model for major adverse cardiovascular events after percutaneous coronary intervention in patients with coronary artery disease based on quantitative parameters of ambulatory electrocardiography

MEN Li; LIU Rui; YANG Long; CHEN Bing-xin; YIN Ting-ting; SHI Jiang-rong; Bilali PAIZULA; FAN Ping

doi:10.3969/j.issn.1672-5301.2025.06.006

Abstract：Objective To develop and validate a predictive model based on 24-hour Holter electrocardiogram parameters for major adverse cardiovascular events(MACEs)following percutaneous coronary intervention(PCI)in patients with coronary artery disease,providing a basis for early clinical risk assessment.Methods This multicenter retrospective study included 388 coronary artery disease patients who received treatment at the First Affiliated Hospital of Xinjiang Medical University and four collaborating centers from January 2023 to December 2024.A total of 281 patients from the First Affiliated Hospital of Xinjiang Medical University were assigned to the training cohort,while 107 patients from the four collaborating centers were included in the validation cohort.The clinical variables such as sex,age,smoking,body mass index(BMI),blood pressure,C-reactive protein(CRP),interleukin-6(IL-6),N-terminal pro-brain natriuretic peptide(NT-proBNP),triglycerides(TG),total cholesterol(TC),high-density lipoprotein cholesterol(HDL-C)and left ventricular ejection fraction(LVEF)were recorded.All the patients were instructed to wear Holter devices within 48 hours after PCI;and the quantitative parameters such as average heart rate,standard deviation of NN intervals(SDNN),triangular index,total power and heart rate deceleration capacity were collected.The patients were followed up for one year after discharge;and MACEs were defined as cardiovascular death,rehospitalization for heart failure or nonfatal myocardial infarction,malignant arrhythmia,stroke,gastrointestinal bleeding and clinically indicated repeat revascularization.The least absolute shrinkage and selection operator(LASSO)regression algorithm was applied in the training cohort to reduce the dimensionality of predictive variables.Multivariate Logistic regression analysis was performed to validate the predictive value of the selected variables,and a nomogram model was constructed based on the predictive probabilities.The receiver operating characteristic(ROC)curves and calibration curves were generated to evaluate the predictive performance and consistency of the model in the training and validation cohorts.Results Using LASSO regression,the optimal penalty coefficient(λ)was identified,and the number of predictive variables was reduced from 42 to 6.The final predictive model included the triangular index,heart rate deceleration capacity,BMI,LDL-C,NT-proBNP and smoking history.Multivariate Logistic regression analysis demonstrated that the triangular index(Oβ=0.942,95％CI 0.907-0.973,P＜0.001),heart rate deceleration capacity(Oβ=0.703,95％CI 0.508-0.950,P=0.027),BMI(Oβ=1.231,95％CI 1.082-1.423,P=0.003),LDL-C(Oβ=2.120,95％CI 1.302-3.562,P=0.003),NT-proBNP(Oβ=1.626,95％CI 1.202-2.268,P=0.003)and smoking history(OR=8.404,95％CI 3.100-25.049,P＜0.001)were significantly associated with the occurrence of MACEs.A nomogram model was constructed based on these six variables,and its predictive performance was evaluated.The ROC curve analysis showed that the area under the curve(AUC)for predicting MACEs was 0.933(95％CI 0.899-0.967,P＜0.001)in the training cohort and 0.901(95％CI 0.830-0.971,P＜0.001)in the validation cohort.Calibration curves for both cohorts were close to the ideal curve,with Hosmer-Lemeshow P-values of 0.287 and 0.855,indicating no significant difference between the predicted and observed values.Conclusion The trigonometric index,heart rate deceleration capacity,BMI,LDL-C,NT-proBNP,and smoking history are influential factors in the development of MACEs after PCI in patients with CAD.The simplified nomogram model that integrates traditional risk factors with quantitative Holter parameters,demonstrating high discriminative and calibration performance would provide robust support for early clinical risk stratification and intervention planning.