Multi-Omics Integration Identifies Epithelial-Stromal-Immune Co-Regulators Bridging Ulcerative Colitis and Colorectal Cancer

What To Know

This study used multi-omics integration (microarray, TCGA tumor data, machine learning, Mendelian randomization, immune deconvolution, and Visium spatial transcriptomics) to identify four genes—CXCL1, S100P, THY1, and TRIM29—that are shared between inflamed ulcerative colitis mucosa and colorectal cancer tissue and that form a spatially structured epithelial–stromal–immune module.

The authors treated the four-gene panel as hypothesis-generating rather than definitive. Machine-learning selection was not highly stable on resampling, but independent replication in another UC cohort showed the four genes moved in the same direction and a gene-expression model separated inflamed from control mucosa in that cohort.

Additional analyses included two-sample Mendelian randomization (which returned null estimates for tested genes where instruments existed), immune-cell correlation patterns in TCGA-COAD, survival modelling (no robust prognostic model emerged), and spatial transcriptomics showing compartmental localization of the genes across epithelial, stromal, and immune regions.

Overall, the paper proposes candidate molecular links between chronic UC inflammation and colorectal cancer risk but emphasizes that findings are preliminary and not ready for clinical prognostic or treatment use.