Polyamine Metabolism Drives Immune Evasion in Malignant Epithelial Cells and Defines a Six-Gene Prognostic Signature for Colorectal Cancer

Background Dysregulated polyamine metabolism (PAM) is implicated in colorectal cancer (CRC), yet its comprehensive prognostic value and underlying mechanisms at a single-cell resolution remain poorly elucidated. Methods A PAM-related gene (PAMRG) score was developed using ssGSEA in TCGA-CRC cohort. Weighted gene co-expression network analysis (WGCNA) was employed to pinpoint genes correlated with the PAMRG score. By integrating these genes with those differentially expressed between CRC and healthy control samples, a prognostic model leveraging machine learning was established and subsequently validated by applying GSE39582 dataset. The difference in tumor immune microenvironment was analyzed based on risk stratification. Single-cell RNA sequencing (scRNA-seq, GSE166555) was employed to dissect cellular heterogeneity, and prognostic genes were confirmed by reverse transcription-quantitative PCR (RT-qPCR) and western blot in CRC cell lines. Results The PAMRG score was elevated in CRC tissues and associated with poor survival. A six-gene prognostic signature (CCNF, ESCO2, MRPL37, NOP14, TAMM41, TFB1M) stratified patients into two risk groups with distinct survival, immune features, and immunotherapy responses. scRNA-seq showed differential expression of these genes in epithelial cells. High-PAMRG malignant epithelial cells displayed immune evasion, enhanced DNA repair, increased stemness, and endothelial communication. CCNF, MRPL37, NOP14, and TFB1M were upregulated in CRC cell lines. Conclusion This study defined a novel PAM-related prognostic signature for CRC and uncovered, at the single-cell level, a distinct malignant epithelial cell subpopulation with pro-tumorigenic features, providing new insights for risk stratification and potential therapeutic targeting.