Abstract
BACKGROUND: Coronary heart disease (CHD) is the leading cause of death and disability worldwide. Accumulating evidence reveals that atherosclerosis (AS), characterized by systemic, chronic, and multifocal disease, and is the primary pathological basis of cardiovascular diseases, including CHD. However, the molecular underpinnings of CHD are still far from well understood. Our study attempted to identify aberrant plasma exosome-derived circRNAs and key exosomal circRNA biomarkers for CHD. METHODS: The expression profiles of mRNAs, circRNAs, and lncRNAs in the blood exosomes of CHD patients and healthy controls were obtained from the exoRBase database. The corresponding miRNAs of the differentially expressed mRNAs, circRNAs, and lncRNAs were predicted via ENCORI and the miRcode database. LncRNAs/circRNAs and mRNAs with the cotargeted miRNAs were selected to construct an interaction network. Multiple machine learning algorithms have been used to explore potential biomarkers, followed by verification in patients with CHD using real-time quantitative reverse transcription-polymerase chain reaction (RT-qPCR). RESULTS: Based on the cutoff criterion of P < 0.05, we identified 85 differentially expressed circRNAs (4 upregulated and 81 downregulated), 43 differentially expressed lncRNAs (24 upregulated and 19 downregulated), and 312 differentially expressed mRNAs (55 upregulated and 257 downregulated). Functional enrichment analysis revealed that the differentially expressed mRNAs were involved mainly in neutrophil extracellular trap (NET) formation and the nucleotide-binding oligomerization domain- (NOD-) like receptor signaling pathway. Further analysis revealed that the DEGs in the circRNA/lncRNA-miRNA-mRNA interaction network were closely related to lipid and atherosclerotic signaling pathways. Hsa_circ_0001360 and hsa_circ_0000038 were identified as potential biomarkers for CHD based on three machine learning algorithms. The relative expression levels of hsa_circ_0001360 and hsa_circ_0000038 were significantly altered in plasma exosomes from patients with CHD. ROC curve analysis revealed that the areas under the curve (AUCs) were 0.860, 0.870, and 0.940 for hsa_circ_0001360, hsa_circ_0000038, and the two-gene combination, respectively. CONCLUSION: The circRNA/lncRNA-miRNA-mRNA interaction network might help to elucidate the pathogenesis of CHD. Hsa_circ_0001360 combined with hsa_circ_0000038 might be an important diagnostic biomarker.