Screening of Sepsis Diagnostic Biomarkers Based on Fumarate Metabolism-Related Genes with Analysis of Immune Infiltration and Subtype Identification

BACKGROUND: Sepsis is a major global health challenge characterized by a complex pathogenesis involving an early hyperinflammatory phase followed by a subsequent immunosuppressive state. Recent studies have revealed that dysregulation of fumarate metabolism plays a central role in immune dysregulation during sepsis, making related genes promising candidates as novel diagnostic biomarkers and therapeutic targets. METHODS: Three sepsis datasets (GSE65682, GSE95233, GSE131761) were analyzed to identify differentially expressed fumarate metabolism-related genes. Key genes were selected via machine learning (Least Absolute Shrinkage and Selection Operator, Support Vector Machine, Boruta) to construct a diagnostic model, validated by Receiver Operating Characteristic, nomogram, and Decision Curve Analysis. Immune infiltration, functional enrichment, subtype analysis, and a ceRNA network were further explored. RESULTS: T cells. Molecular subtyping based on these genes identified two sepsis subtypes with distinct immune characteristics, and a relevant ceRNA regulatory network was constructed. CONCLUSION: This study constructs a diagnostic model for sepsis based on fumarate metabolism-related genes, linking metabolic reprogramming to immune dysregulation and offering biomarkers and theoretical support for personalized treatment.