Machine learning based identification of anoikis related gene classification patterns and immunoinfiltration characteristics in diabetic nephropathy.

Anoikis and immune cell infiltration are pivotal factors in the pathophysiological mechanism of diabetic nephropathy (DN), yet a comprehensive understanding of the mechanism is lacking. This work aimed to pinpoint distinctive anoikis-related genes (ARGs) in DN and delve into their impact on the immune landscape. Three datasets (GSE30528, GSE47184, and GSE96804) were downloaded from the gene expression omnibus (GEO) dataset. Differentially expressed genes (DEGs) were identified using the "limma" package, while ARGs were obtained from GSEA, GeneCard, and Harmonizome datasets. The intersection of DEGs and ARGs was analyzed for Gene Ontology (GO) function and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment. The CIBERSORT algorithm was employed to estimate the infiltration percentage of 22 immune cell types in DN renal tissue. Subsequently, the least absolute shrinkage and selection operator (LASSO), support vector machine recursive feature elimination (SVM-RFE), and random forest (RF) algorithms were adopted to screen key ARGs related to DN. After that, receiver operating characteristic (ROC) analysis was employed to assess the diagnostic accuracy of each gene and the real-time quantitative polymerase chain reaction (RT-qPCR) was adopted to quantitatively detect the expression of biomarkers in DN cell models. Finally, correlations between key genes and immune cell infiltration were analyzed, and a competitive endogenous ribonucleic acid (RNA) (ceRNA) network based on key genes was constructed. A total of 59 DEARGs were identified. GO functional annotation enrichment analysis revealed their involvement in kidney development, extracellular matrix (ECM), cytoplasmic vesicle cavity, immunoinflammatory response, and cytokine effect. KEGG pathway analysis indicated that MAPK, PI3K -Akt, IL -17, TNF, and HIF- 1 signaling pathways are critical for DN. In addition, seven key genes, including PDK4, S100A8, HTRA1, CHI3L1, WT1, CDKN1B, and EGF, were screened by machine learning algorithm. Most of these genes exhibited low expression in renal tissue of DN patients and positive correlation with neutrophils, and their expressions were verified in an external dataset cell model. The ceRNA analysis suggested potential regulatory pathways (H19/miR-15b-5p/PDK4 and KCNQ1T1/miR-1207-3p/WT1) influencing early DN progression. This work provided a comprehensive analysis of the role of DEARGs in DN for the first time, offering valuable insights for further understanding the disease mechanism and guiding clinical diagnosis, treatment, and research of DN.