Identification of diabetes related phenotype and diagnostic biomarkers in coronary artery disease.

BACKGROUND: Diabetes mellitus (DM) and coronary artery disease (CAD) are closely interrelated clinical conditions. However, the combination analysis based on DM related CAD diagnostic model remains a gap. The primary objective of this study was to identify diagnostic models and diagnostic markers for CAD based on the association of diabetic phenotypes and attempt to explore them further in a mouse model. METHODS: We used data integration as well as multiple datasets for both coronary artery disease and diabetes to exclude bias as well as to improve reliability. We employed the least absolute shrinkage and selection operator (LASSO) regression algorithms to construct the CAD diagnostic model. Furthermore, we established mouse CAD model (low-density lipoprotein receptor deficient mice with high fat diet) to explore the crosstalk between the screened biomarkers and severe CAD progress. RESULTS: The intersecting genes from differential analysis and weighted correlation network analysis (WGCNA) results yielded 32 diabetes-related biomarkers. We then identified two diabetes-related phenotypes through the consensus clustering in CAD patients. Microenvironmental analysis revealed that phenotype 1 exhibited higher expression of most cytokines, inflammatory factors, interleukins, and related receptors. Immune cell composition in phenotype 1 showed increased infiltration compared to phenotype 2. The LASSO regression identified 16 diabetes-related genes and we further constructed a diagnostic model based on these genes, which the area under the curve (AUC) reached 0.8. Additionally, single cell immune analysis exhibited the location of these genes. KCNQ1, ATP6V1B1, MTDH, and ITPK1 were predominantly located in macrophages, indicating their potential in regulating macrophage during myocardial injury. Furthermore, We elucidated that KCNQ1 and ITPK1 exhibited high expression level in mouse CAD model in tissue level. exhibited similar expression trends with macrophage biomarkers (CD31 and CD68). The result of qPCR also indicated the elevated level of KCNQ1 and ITPK1, which exhibited crosstalk with CD31 and CD68 in mouse CAD model. CONCLUSION: This study delves into the microenvironmental characteristics of diabetes-related phenotypes in CAD, constructing an optimal diagnostic model and validated the significance of diagnostic markers in mouse CAD model, which may offer insights that could be beneficial for clinical management in the near future.