Abstract
ST-segment elevation myocardial infarction (STEMI) is considered a critical cardiac condition with a poor prognosis. Shortly after STEMI occurs, the increased number of circulating leukocytes including macrophages can lead to the accumulation of more cells in the myocardium, affecting the cardiac immune microenvironment. Identifying serum biomarkers associated with immune infiltration after STEMI is important for diagnosing and treating STEMI. In this work, we aimed to use integrated bioinformatics and machine learning methods to identify new biomarkers. First, candidate genes closely associated with M1 macrophage immune infiltration and STEMI were obtained using the limma package, the CIBERSORTx package, weighted gene coexpression network analysis (WGCNA), and protein‒protein interaction (PPI) networks from the GSE59867 dataset, which comprises peripheral blood mononuclear cell (PBMC) samples. The STEMI patients were subsequently stratified into subtypes using the ConsensusClusterPlus package. Furthermore, using machine learning methods, we identified AKT3, GJC2, HMGCL and RBM17 as the genes with the greatest potential to be associated with STEMI subtypes and with M1 macrophage infiltration during the acute phase of STEMI. Finally, the expression profile and diagnostic value of the four feature genes were validated in the GSE59867 and GSE62646 datasets and in 24 patients using real-time PCR. This study revealed logically and comprehensively that AKT3, GJC2, HMGCL and RBM17, which are derived from PBMCs, could enhance the accuracy of STEMI diagnosis and might provide effective treatment options for STEMI patients.