Zhongguo quanke yixue (Aug 2022)
Predictive Model for Long-term Major Adverse Cardiovascular Events in Patients with Acute Myocardial Infarction Undergoing Percutaneous Coronary Intervention
Abstract
Background Risk stratification for acute myocardial infarction (AMI) is important for clinical decision-making and prognosis evaluation. As changes have been found in clinical characteristics and management of AMI, the current existing clinical risk score for AMI may be inapplicable to clinical practice. To effectively implement strategies of individualized management for AMI patients, it is necessary to improve the prediction accuracy of long-term major adverse cardiovascular events (MACEs) in AMI after percutaneous coronary intervention (PCI) . Objective To develop a predictive model for long-term MACEs in AMI patients after PCI. Methods Among the 1 130 AMI patients treated with PCI in Beijing Anzhen Hospital from January 1 to July 31, 2019, 962 eligible cases were enrolled, and their clinical data and laboratory examination indices were collected. Follow-up of the patients was performed via telephone interviews at a median of 2.4 years. The primary endpoint was a composite of all-cause mortality, non-fatal myocardial infarction, non-fatal stroke, malignant arrhythmia, new heart failure or readmission due to exacerbated heart failure, and unplanned revascularization. Patients were divided into event (122 cases) and non-event (840 cases) groups according to the prevalence of MACEs during the follow-up period. Lasso regression was conducted to identify candidate risk factors of long-term MACEs. Multivariate Logistic regression analysis was used to construct the prediction model and the nomograms. The receiver operating characteristic curve was used to evaluate the discrimination ability of the prediction model. The efficacy of the predictive model was assessed by comparing with that of the Global Registry of Acute Coronary Events (GRACE) score in terms of the net reclassification improvement (NRI) and the integrated discrimination improvement (IDI) . Results The prevalence of MACEs was 12.7% (122/962) . Five predictive variables were identified by Lasso regression, which included ST-segment deviation, diabetes history, hemoglobin (Hb) , left ventricular ejection fraction (LVEF) , and estimated glomerular filtration rate (eGFR) . The algorithm of the prediction model developed using multivariate Logistic regression was: logit (P) =3.596-0.023×X1-0.014×X2-0.036×X3+0.726×X4+1.372×X5 (X1-X5 indicate Hb, eGFR, LVEF, diabetes, and ST-segment deviation, respectively) . ST-segment deviation, diabetes, LVEF, and Hb were associated with MACEs in AMI patients after PCI (P<0.05) . ST-segment deviation, diabetes, eGFR and Hb were associated with MACEs in ST-segment elevation myocardial infarction (STEMI) patients after PCI (P<0.05) . ST-segment deviation, diabetes, and Hb were associated with MACEs in non-STEMI patients after PCI (P<0.05) . The prediction model exhibited an area under the curve (AUC) of 0.774〔95%CI (0.710, 0.834) 〕 for the training cohort, and an AUC of 0.751〔95%CI (0.686, 0.815) 〕for the testing cohort. The NRI estimated by the predictive model in AMI, STEMI, and non-STEMI patients was 0.493〔95%CI (0.303, 0.682) 〕, 0.459〔95%CI (0.195, 0.724) 〕, and 0.455〔95%CI (0.181, 0.728〕, respectively. The IDI estimated by the predictive model in AMI, STEMI, and non-STEMI patients was 0.055〔95%CI (0.028, 0.081) 〕, 0.042〔95%CI (0.015, 0.070〕, and 0.069〔95%CI (0.022, 0.116) 〕, respectively. The predictive efficiency of the predictive model in the three groups was significantly better than that of the GRACE score (P<0.05) . The predictive model was significantly better than the GRACE score in all participants 〔ΔAUC=0.050, P=0.015; IDI=0.055, 95%CI (0.028, 0.081) , P<0.001; NRI=0.493, 95%CI (0.303, 0.682) , P<0.001) 〕. Conclusion Our predictive model containing five factors (ST-segment deviation, diabetes, LVEF, eGFR and Hb) may be useful for early risk stratification and long-term prognosis prediction in patients with AMI after PCI.
Keywords
