Identification of tolerogenic dendritic cells-related prognostic biomarkers in Wilms tumor via machine learning integration

Wilms tumor (WT) is the most common pediatric kidney cancer. Tolerogenic dendritic cells (TolDCs) promote tumor immune evasion in the tumor microenvironment. Therefore, establishing a TolDC-based prognostic model for WT holds significant clinical value. We analyzed WT-related genes from The Cancer Genome Atlas and TolDC-associated datasets to identify shared differentially expressed genes using Venn analysis. Protein-protein interaction network analysis and machine learning algorithms (Boruta and Support Vector Machine Recursive Feature Elimination, SVM-RFE) were performed to screen candidate hub genes. A prognostic risk model was constructed using univariate Cox proportional hazards regression, with predictive performance evaluated by Kaplan-Meier survival analysis and receiver operating characteristic curves. Immune infiltration analysis, gene set enrichment analysis, and BioGRID were conducted to elucidate biological functions. Drug-gene interaction analysis was performed using the Drug Signature Database. A total of 181 co-expressed genes were identified. Among these, MSH2, CDH2, ALDH1A1, AURKA, CD274, FOSL2, IL15RA, GADD45B, TGM2, CXCR4, SOD2, and MT1E were selected as TolDC-associated biomarkers for WT. The prognostic model ultimately pinpointed ALDH1A1, CXCR4, and FOSL2 as key diagnostic biomarkers, supported by Kaplan-Meier survival analysis and ROC curves, which confirmed the model's robust predictive capacity for survival risk. Drug-gene interaction analysis predicted 335 potential therapeutic compounds targeting ALDH1A1, CXCR4, and FOSL2. Comprehensive bioinformatics analysis identified the prognostic biomarkers of WT related to TolDCs, providing new insights for personalized WT treatment.