2025, Vol. 29 ›› Issue (36): 7909-7920

Machine learning-based analysis of neutrophil-associated potential biomarkers for acute myocardial infarction

Yang Dingyan1, Yu Zhenqiu1, Yang Zhongyu2   

Abstract: BACKGROUND: Accurate early diagnosis and timely reperfusion therapy are important prerequisites for saving the lives and improving the prognosis of patients with acute myocardial infarction. Therefore, it is important to find ideal biomarkers for early diagnosis of acute myocardial infarction.
OBJECTIVE: To analyze key genes associated with neutrophils by acute myocardial infarction through bioinformatics and machine learning to explore new biomarkers.
METHODS: Differentially expressed genes were identified based on the Gene Expression Omnibus (GEO) database and Limma R package. Deconvolution algorithm was used to explore the immune cells infiltration level. Then, acute myocardial infarction and neutrophils-related biomarkers were screened by weighted gene co-expression network analysis (WGCNA), protein-protein interaction (PPI) networks, machine learning, and functional enrichment analysis. Receiver operating characteristic curve analysis was conducted to assess the diagnostic efficacy of biomarkers for acute myocardial infarction. Targeted drugs for biomarkers were screened through the STITCH and Herb database. Finally, the hospitalized patients who were first diagnosed with acute myocardial infarction in the Department of Cardiology of Affiliated Hospital of Guizhou Medical University from March to June 2023 were used as the experimental group, and the hospitalized patients who had no ischemic changes on electrocardiograms and no stenosis on coronary angiograms during the same period were used as the control group. Peripheral blood of the patients in the two groups was collected. The relative expressions of the genes were verified in the human peripheral blood samples by RT-qPCR.
RESULTS AND CONCLUSION: (1) A total of 2 349 differentially expressed genes were obtained, and immune infiltration analysis revealed differences in immune cell scores such as B cells memory, NK cells resting, and Neutrophils between the disease and normal groups. (2) Using WGCNA, two gene modules, ME green and ME turquoise, were found to exhibit the highest correlation with neutrophil fine with acute myocardial infarction. (3) Twenty-four differential module genes were obtained after intersecting with differentially expressed genes. Functional enrichment analysis revealed that they were associated with a variety of processes such as innate immune response and defense response to bacteria. KEGG results showed that they were mainly associated with the tumor necrosis factor signaling pathway. (4) The genes mined by the machine learning algorithm took the intersection to obtain three genes, namely, S100A12, PTCH1, and LOC400499, all of which were greater than 0.7 by the area under the receiver operating characteristic curve in both the GSE48060 and GSE66360 datasets. They were considered as potential biomarkers. (5) Based on the STITCH and Herb databases, 11 target drugs were found for S100A12 and a total of 6 target drugs were found for PTCH1. (6) RT-qPCR results showed that S100A12, PTCH1, and LOC400499 were significantly differentially expressed in acute myocardial infarction patients compared with controls (P < 0.05). (7) S100A12, PTCH1, and LOC400499 may be potential diagnostic biomarkers for acute myocardial infarction, but their specificity in relation to acute myocardial infarction needs to be further investigated, in which S100A12 may be a potential target for regulating acute myocardial infarction.

Key words: acute myocardial infarction, neutrophil, biomarker, differential analysis, WGCNA, machine learning algorithms, immune infiltration analysis, S100A12, PTCH1, LOC400499