Abstract / Summary
Non-obstructive azoospermia (NOA) is a severe male infertility disorder. This study aims to identify immune-related biomarkers for explore immune-related molecular signatures associated with NOA. Here, we integrated transcriptomic data from four independent cohorts (GSE145467, GSE45885, GSE45887, GSE9210) using an integrative pipeline combining differential expression analysis, weighted gene co-expression network analysis (WGCNA), and consensus-based multi-machine learning to identify an immune-related gene signature associated with NOA. We identified 16 immune-related differentially expressed genes through WGCNA, which were refined into a 12-gene panel— Hub NOA.Sig ( CD14, CD19, CMTM2, CXCL16, HSPA1L, HSPA2, IGF1, IL13, POMC, RBP1, STAB2, TYROBP )—using five feature selection algorithms (Bagged Trees, Bayesian, Boruta, LASSO, LVQ). A Random Forest classifier based on Hub NOA.Sig yielded good classification performance with an AUC of 0.87. This signature offers a transcriptomic reference for characterizing NOA heterogeneity and exploring immune-related molecular features. Single-sample gene set enrichment analysis (ssGSEA) revealed significant alterations in six immune cell populations—including effector memory CD4 ⁺ /CD8 ⁺ T cells, regulatory T cells, and plasmacytoid dendritic cells—and demonstrated strong correlations between hub genes and specific immune infiltrates. Furthermore, drug-gene interaction screening and molecular docking predicted loperamide- POMC , acitretin- RBP1 , and retinol- RBP1 as candidate drug-gene pairs for subsequent experimental validation. This work provides transcriptomic clues and candidate molecular targets for future mechanistic investigations into NOA-related male infertility.