Landscape of co-expressed genes between the myocardium and blood in sepsis and ceRNA network construction: a bioinformatic approach.

Septic cardiomyopathy is a serious complication of sepsis. The mechanism of disease pathogenesis, which is caused by infection, is well researched. Despite ongoing efforts, there are no viable biological markers in the peripheral blood for early detection and diagnosis of septic cardiomyopathy. We aimed to uncover potential biomarkers of septic cardiomyopathy by comparing the covaried genes and pathways in the blood and myocardium of sepsis patients. Gene expression profiling of GSE79962, GSE65682, GSE54514, and GSE134364 was retrieved from the GEO database. Student's t-test was used for differential expression analysis. K-means clustering analysis was applied for subgroup identification. Least absolute shrinkage and selection operator (LASSO) and logistic regression were utilized for screening characteristic genes and model construction. Receiver operating characteristic (ROC) curves were generated for estimating the diagnostic efficacy. For ceRNA information prediction, miWalk and lncBase were applied. Cytoscape was used for ceRNA network construction. Inflammation-associated genes were upregulated, while genes related to mitochondria and aerobic metabolism were downregulated in both blood and the myocardium. Three groups with a significantly different mortality were identified by these covaried genes, using clustering analysis. Five characteristic genes-BCL2A1, CD44, ADGRG1, TGIF1, and ING3-were identified, which enabled the prediction of mortality of sepsis. The pathophysiological changes in the myocardium of patients with sepsis were also reflected in peripheral blood to some extent. The co-occurring pathological processes can affect the prognosis of sepsis. Thus, the genes we identified have the potential to become biomarkers for septic cardiomyopathy.