Plasma Metabolic Profiling and Multiclass Diagnostic Model Development for Stable Angina Pectoris and Acute Myocardial Infarction

Abstract

Coronary heart disease remains a major global health challenge, with a clear need for enhanced early risk assessment. This study aimed to elucidate metabolic signatures across various stages of coronary heart disease and develop an effective multiclass diagnostic model. Using metabolomic approaches, gas chromatography–mass and liquid chromatography–tandem mass spectrometry were used to analyze plasma samples from healthy controls, patients with stable angina pectoris, and those with acute myocardial infarction. Pathway enrichment analysis was conducted on metabolites exhibiting significant differences. The key metabolites were identified using Random Forest and Recursive Feature Elimination strategies to construct a multiclass diagnostic model. The performance of the model was validated through 10-fold cross-validation and evaluated using confusion matrices, receiver operating characteristic curves, and calibration curves. Metabolomics was used to identify 1491 metabolites, with 216, 567, and 295 distinctly present among the healthy controls, patients with stable angina pectoris, and those with acute myocardial infarction, respectively. This implicated pathways such as the glucagon signaling pathway, d-amino acid metabolism, pyruvate metabolism, and amoebiasis across various stages of coronary heart disease. After selection, testosterone isobutyrate, N-acetyl-tryptophan, d-fructose, l-glutamic acid, erythritol, and gluconic acid were identified as core metabolites in the multiclass diagnostic model. Evaluating the diagnostic model demonstrated its high discriminative ability and accuracy. This study revealed metabolic pathway perturbations at different stages of coronary heart disease, and a precise multiclass diagnostic model was established based on these findings. This study provides new insights and tools for the early diagnosis and treatment of coronary heart disease.