A machine learning-based prognostic model integrating mRNA stemness index, hypoxia, and glycolysis‑related biomarkers for colorectal cancer.

Background: Cancer stemness, hypoxia, and glycolysis collectively influence colorectal cancer (CRC) progression. However, the intricate relationships among these factors remain incompletely understood.

Methods: This study (1) explored hypoxia and glycolysis-related genes (HGRGs) in CRC by mRNA stemness index (mRNAsi), analyzed the gene expression profiles from Gene Expression Omnibus and The Cancer Genome Atlas (TCGA) databases, (2) established a Cox-prognostic model based on single-sample gene set enrichment analysis, differentially expressed gene analysis, weighted gene co-expression network analysis, and Least Absolute Shrinkage and Selection Operator (LASSO) and Cox regression analyses, and (3) assessed the predictive accuracy of the model. Decision curve analysis (DCA) was employed to determine the clinical utility of the model.

Results: Ten HGRGs were selected based on mRNAsi to create the LASSO model. High-risk CRC patients in the TCGA dataset displayed unfavorable clinical outcomes and responses to immunotherapy. Consensus cluster analysis revealed two distinct colon adenocarcinoma/rectal adenocarcinoma clusters, with patients in cluster 2 having a worse prognosis compared to patients in cluster 1. A five-gene prognostic nomogram was developed through univariate and multivariate Cox regression analyses, with DCA confirming its accuracy.

Conclusions: This innovative prognostic model, incorporating ALDOB, AQP1, IL1A, PHGDH, and PTGIS, is highly accurate in predicting patient survival.