Identification of biomarkers for the diagnosis of chronic kidney disease (CKD) with dilated cardiomyopathy (DCM) by bioinformatics analysis and machine learning.

Background: Chronic kidney disease (CKD) is a globally prevalent and highly lethal condition, often accompanied by dilated cardiomyopathy (DCM), which increases the risk of cardiac complications. Early detection of DCM in CKD patients remains challenging, despite established research demonstrating the relationship between CKD and cardiac abnormalities.

Methods: We retrieved expression matrices for DCM (GSE57338, GSE29819) and CKD (GSE104954) from GEO and a DCM scRNA-seq dataset (GSE145154). These were analyzed for differential gene expression and WGCNA. KEGG and GO analyses were performed on shared differentially expressed genes in DCM and CKD. Potential drugs for DCM were identified using CMAP. Machine learning methods LASSO, SVM-RFE, and RF were used to find biomarkers and develop a diagnostic nomogram for CKD-associated DCM, validated with external datasets. Single-gene GSEA was conducted to understand model gene mechanisms in CKD-associated DCM. Immune cell infiltration was analyzed with CIBERSORT, and single-cell sequencing examined model gene distribution and expression in the heart.

Results: Our examination of the expression matrix datasets associated with DCM and CKD revealed 115 key model genes that are shared by the two disorders as well as 47 genes that are differently expressed. These 47 differentially expressed genes were primarily linked to immune regulation and inflammation, according to enrichment analysis. CMAP analysis suggested withaferin-a, droxinostat, fluorometholone, and others as potential DCM treatments. Machine learning identified MNS1 and HERC6 as significant CKD-associated DCM biomarkers. A diagnostic nomogram using these genes was developed, showing strong discriminative power and clinical utility. MNS1 and HERC6 are implicated in metabolism, inflammation, immunity, and heart function. Immune cell infiltration analysis indicated dysregulation in DCM, with MNS1 and HERC6 correlating with immune cells. Single-cell sequencing showed MNS1 and HERC6 expression in endothelial cells and fibroblasts, respectively.

Conclusion: We identified MNS1 and HERC6 as biomarkers and developed a new diagnostic nomogram based on them for the timely diagnosis of CKD patients presenting with DCM complications. This study's findings offer novel insights into potential diagnostic methods and therapeutic strategies regarding the coexistence of CKD and DCM.