Xiang Sha, Wei Wang, Jian Wang, Ruzhu Wang

Department of Cardiology, The Affiliated Taizhou People’s Hospital of Nanjing Medical University, Taizhou School of Clinical Medicine, Nanjing Medical University, Taizhou, 225300, People’s Republic of China

Correspondence: Ruzhu Wang, Email 15052800790@163.com

Background: Coronary microvascular dysfunction (CMD) significantly impacts outcomes in patients with acute ST-segment elevation myocardial infarction (STEMI) undergoing percutaneous coronary intervention (PCI). The cardiometabolic index (CMI), an indicator combining lipid and anthropometric parameters, has been linked to cardiovascular risk, but its association with CMD remains unclear. This study aims to investigate the relationship between CMI and the occurrence of CMD following PCI in STEMI patients and to assess its predictive value using Least Absolute Shrinkage and Selection Operator (LASSO)-based feature selection and multiple machine learning algorithms.
Methods: This retrospective cohort study enrolled STEMI patients who underwent primary PCI with stent implantation and post-procedural coronary microvascular function assessment between January 2021 and December 2024. Patients were categorized into CMD and non-CMD groups based on noninvasive microvascular resistance indices. Logistic regression, restricted cubic spline analysis, and machine learning models (Random Forest (RF), LightGBM, XGboost and K-Nearest Neighbors) were employed to evaluate the predictive value of CMI for post-PCI CMD.
Results: A total of 702 STEMI patients were included, and CMD was observed in 52.1% of patients. Compared to the first CMI tertile (T1) group, T2 and T3 group had increased odds of CMD (T2: adjusted odds ratio (aOR) 2.41, 95% confidence interval (CI) 1.60– 3.63; T3: aOR 3.40, 95% CI 2.17– 5.32). There was a non-linear relationship between CMI and CMD (P < 0.001). The area under the curve (AUC) for CMI predicting CMD was 0.627 (95% CI: 0.586– 0.666). Seven variables were screened by LASSO-Logistic regression for model development. Comparing four models’ performances, the RF model achieved the best performance (AUC = 0.772). SHapley analysis revealed that CMI had the highest predictive value for CMD.
Conclusion: A higher CMI level is an independent risk factor for CMD of STEMI patients after PCI, and its predictive value enhanced when integrated into RF model.

Keywords: cardiacmetabolic index, acute st elevation myocardial infarction, percutaneous coronary intervention therapy, quantitative blood flow fraction