A Novel Immune-Escape Signature and Classifier for Predicting Sepsis Constructed by Integrative Bioinformatics

Background: Sepsis is a life-threatening, infection-triggered syndrome of dysregulated inflammation and immune dysfunction. However, the mechanisms underlying immune escape in sepsis remain poorly understood and merit further investigation. Methods: Gene expression data for sepsis were obtained from the Gene Expression Omnibus (GEO) database: GSE65682 was used to construct the training cohort, and GSE95233 served as an independent validation cohort. Feature genes were screened by integrating differential expression profiling with Weighted Gene Co-expression Network Analysis (WGCNA). A diagnostic model was developed and validated using Receiver Operating Characteristic (ROC) analysis, a nomogram, and decision curve analysis (DCA). Functional enrichment, immune infiltration, and competing endogenous RNA (ceRNA) network analyses were also performed. Finally, consensus clustering was applied to identify molecular subtypes of sepsis. Results: 3 feature genes were identified by differential expression analysis and WGCNA. AUC-based screening selected 2 diagnostic genes (NXT1 and UXS1), which demonstrated high diagnostic accuracy (AUC > 0.9). Immune infiltration analysis revealed distinct patterns between the Sepsis and Control groups. Regulatory network analysis identified key miRNAs targeting these 2 genes and lncRNAs that regulate those miRNAs. 2 sepsis subtypes with distinct immune and molecular characteristics were further identified. Conclusion: This study systematically explored the association between immune escape and the pathogenesis of sepsis and, through the integration of multiple bioinformatics approaches, elucidated the immune microenvironmental characteristics and molecular regulatory mechanisms of the disease, providing new insights into understanding its pathophysiology and developing targeted diagnostic and therapeutic strategies.