Sodium Overload-Related Molecular Subtypes and a Four-Gene Prognostic Signature Predict Survival, Immune Landscape,...

Sodium overload has recently emerged as a critical metabolic stressor involved in cancer progression; however, its molecular characteristics and clinical relevance in acute myeloid leukemia (AML) remain unexplored. RNA-seq data sets, clinical annotations, and mutational profiles of AML patients were annotations from The Cancer Genome Atlas and integrated with Genotype-Tissue Expression normal samples. Sodium overload-related genes (SORGs) were obtained from GeneCards. Differentially expressed SORGs (DESORGs) screened by applying the limma statistical model, followed by univariate Cox proportional hazards regression, consensus clustering, functional enrichment, immune infiltration analysis, and pathway evaluation. A prognostic signature was developed through least absolute shrinkage and selection operator regression followed by multivariate Cox modeling. The model's performance was further verified in two external GEO data sets (GSE71014 and GSE37642). Nomogram construction, subgroup analysis, tumor mutational burden (TMB) assessment, drug sensitivity prediction, transcription factor (TF) analysis, and competing endogenous RNA (ceRNA) network analyses were also performed. A total of 57 DESORGs were identified, and 2 sodium overload-related molecular subtypes exhibited distinct survival, immune infiltration, and inflammatory pathway activation. A robust four-gene signature (DOCK1, GABRE, HTR7, ACSM1) stratified patients into high- and low-risk categories with significantly different survival across training and validation cohorts. High-risk patients displayed increased immune infiltration, higher TMB, reduced sensitivity to multiple chemotherapeutic drugs, and inferior predicted response to PD-L1 blockade. TF and ceRNA networks revealed multilayered transcriptional and post-transcriptional regulation of the signature genes. This study identifies sodium overload-related molecular heterogeneity in AML and establishes a validated four-gene prognostic signature that integrates genomic, immunologic, and therapeutic features, offering potential utility for personalized risk assessment and treatment optimization.