Multicohort Construction, Immune Landscape Analysis, and PCOLCE2 Functional Validation of an Inflammation-Related Prognostic Model in Colorectal Cancer.

Colorectal cancer (CRC) prognosis remains challenging due to tumor heterogeneity and immune evasion driven by aberrant inflammatory signaling, and robust multicohort validated prognostic tools with clear therapeutic implications are still lacking. To address this gap, we integrated five independent CRC cohorts (TCGA, GSE17536, GSE17537, GSE29621, and GSE38832; n = 931) and applied single-sample gene set enrichment analysis (ssGSEA) across 15 inflammation-related signaling pathways to construct a prognostic risk score model via multicohort Cox regression and LASSO-Cox regression. The immune landscape was systematically characterized using immune infiltration algorithms, Cancer Immunity Cycle scoring, CellChat analysis, and single-cell RNA sequencing (scRNA-seq) data (GSE166555). The functional role of the candidate gene PCOLCE2 was further validated through in vitro migration and adhesion assays, and confirmed in a syngeneic C57BL/6 mouse subcutaneous tumor model in which MC38-shPCOLCE2 cells were inoculated and treated with anti-PD-1 antibody (200 μg per mouse, every 3 days) alone or in combination to evaluate the synergistic antitumor effect.

The resulting 12-gene inflammation-related risk model achieved robust survival stratification across all five independent cohorts, with AUC values of 0.70-0.85 and C-indices consistently outperforming existing prognostic models, demonstrating strong generalizability as a clinical prognostic tool. High-risk patients exhibited significantly worse overall survival (p