Identification Of New Diagnostic Gene Biomarkers In Patients With Diabetic Kidney Disease Using Machine Learning Strategies And Bioinformatic Analysis

Objectives:Diabetic kidney disease(DKD)is the major cause of chronic kidney disease and end-stage renal disease all over the world.Early diagnosis is essential to prevent progression.Our aim was to identify potential diagnostic biomarkers for DKD,illustrate the biological processes related to the biomarkers and investigate the relationship between them and immune cell infiltration.Methods:1、 Gene expression profiles(GSE30528,GSE96804,and GSE99339)for samples obtained from DKD and controls were downloaded from the Gene Expression Omnibus database as a training set,and the gene expression profiles(GSE47185 and GSE30122)were downloaded as a validation set.Differentially expressed genes(DEGs)were identified using the training set.2、 The least absolute shrinkage and selection operator(LASSO),random forests(RF)and support vector machine-recursive feature elimination(SVM-RFE)were performed to identify potential diagnostic biomarkers.3、 To evaluate the diagnostic efficacy of these potential biomarkers,receiver operating characteristic(ROC)curves were plotted separately for the training and validation sets,and immunohistochemical(IHC)staining for biomarkers was performed in the DKD and control kidney tissues.4、 In addition,the CIBERSORT,XCELL and TIMER algorithms were employed to assess the infiltration of immune cells in DKD,and the relationships between the biomarkers and infiltrating immune cells were also investigated.Results:1、 A total of 95 differentially expressed genes in DKD renal tissues were identified using adjusted P-value < 0.05 and a |log fold change(FC)| >1 as screening criteria.Among them,39 genes were highly expressed in DKD renal tissues,and 56 genes were lowly expressed in DKD renal tissues.2、 Using LASSO,SVM-RFE and RF three machine learning algorithms,DUSP1 and PRKAR2 B were identified as potential biomarker genes for the diagnosis of DKD.3、 The diagnostic efficacy of DUSP1 and PRKAR2 B was evaluated by the areas under the curves in the ROC analysis of the training set(0.945 and 0.932,respectively)and validation set(0.789 and 0.709,respectively).IHC staining suggested that the expression levels of DUSP1 and PRKAR2 B were significantly lower in DKD patients compared to normal.4、 Immune cell infiltration analysis showed that B memory cells,gamma delta T cells,macrophages,and neutrophils may be involved in the development of DKD.Furthermore,both of the candidate genes are associated with these immune cell subtypes to varying extents.Conclusions:DUSP1 and PRKAR2 B are potential diagnostic markers of DKD,and they are closely associated with immune cell infiltration.