Bioinformatics mining and experimental validation of prognostic biomarkers in colorectal cancer

Colorectal cancer (CRC) is a prevalent condition with increasing incidence and mortality rates. The identification of robust prognostic gene signatures remains an unmet clinical need in CRC treatment. In this study, data from the GEO and TCGA databases were utilized to identify 2,779 upregulated and 2,629 downregulated genes in CRC tissues compared to adjacent normal tissues. WGCNA analysis highlighted the MEbrown module, which comprised 1,639 genes that exhibited strong correlations with CRC progression. Subsequently, an intersection analysis was conducted to further refine the candidate gene set, resulting in the selection of 926 differentially expressed CRC-related genes for subsequent analysis. Through univariate Cox regression, LASSO regularization, and multivariate Cox regression, a five-gene prognostic signature (TIMP1, PCOLCE2, MEIS2, HDC, CXCL13) was established, demonstrating consistent predictive accuracy in external (GSE32323) and internal validation cohorts. Mutational profiling showed predominant missense mutations across signature genes, with TIMP1 exhibiting the highest variant allele frequency. Functional enrichment analysis linked TIMP1 to critical CRC pathways including type I interferon receptor binding, oxidative phosphorylation, and Notch signaling pathways. High expression of TIMP1 was associated with poor prognosis in patients with CRC. Additionally, using siRNA technology, the impact of TIMP1 on cellular proliferation, metastasis and apoptosis in CRC cell lines (HCT116 and HT29) was investigated, showing that TIMP1 knockdown significantly inhibited CRC cell proliferation, metastasis, and promoted apoptosis. These experimental results were consistent with the conclusions drawn from the bioinformatics analysis. This research presents a prognostic risk model for CRC, further highlights TIMP1 as a potential biomarker and therapeutic target for the disease.