Integrated analysis of shared gene expression signatures and immune microenvironment heterogeneity in type 2 diabetes mellitus and colorectal cancer

Abstract

Emerging evidence suggests a bidirectional relationship between colorectal cancer (CRC) and type 2 diabetes mellitus (T2DM), yet the shared molecular mechanisms and prognostic biomarkers remain poorly characterized. This study aimed to identify novel biomarkers linking CRC and T2DM pathogenesis and evaluate their clinical utility in predicting therapeutic responses and survival outcomes. By integrating multi-omics data from public repositories and applying machine learning-driven feature selection, we identified three core biomarkers—FABP4,CDR2L,and FSTL3 that independently predicted overall survival in CRC patients with diabetes. A prognostic nomogram combining these biomarkers with clinicopathological variables (tumor stage, grade, and age) achieved high accuracy for 1-, 3-, and 5-year survival prediction. Functional characterization revealed strong associations between biomarker overexpression and tumor microenvironment remodeling, particularly through fibroblast-mediated immune cell recruitment and cross-talk with lymphocytes. Critically, low expression of these genes correlated with improved anti-PD-1 immunotherapy responses in an independent clinical cohort. Our findings establish FABP4, CDR2L, and FSTL3 as pivotal regulators at the CRC-diabetes interface, with dual utility as prognostic indicators and predictors of immunotherapy efficacy.