Weighted gene co-expression network analysis identifies the prognosis-related models of left- and right-sided colon cancer

Abstract

Left-sided colon cancer (LC) and right-sided colon cancer (RC) are 2 essentially different diseases, and the potential mechanisms regulating them remain unidentified. In this study, we applied weighted gene co-expression network analysis (WGCNA) to confirm a yellow module, mainly enriched in metabolism-related signaling pathways related to LC and RC. Based on the RNA-seq data of colon cancer in The Cancer Genome Atlas (TCGA) and GSE41258 dataset with their corresponding clinical information, a training set (TCGA: LC: n = 171; RC: n = 260) and a validation set (GSE41258: LC: n = 94; RC: n = 77) were divided. Least absolute shrinkage and selection operator (LASSO) penalized COX regression analysis identified 20 prognosis-related genes (PRGs) and helped constructed 2 risk (LC-R and RC-R) models in LC and RC, respectively. The model-based risk scores accurately performed in risk stratification for colon cancer patients. The high-risk group of the LC-R model showed associations with ECM-receptor interaction, focal adhesion, and PI3K-AKT signaling pathway. Interestingly, the low-risk group of the LC-R model showed associations with immune-related signaling pathways like antigen processing and presentation. On the other hand, the high-risk group of the RC-R model showed enrichment for cell adhesion molecules and axon guidance signaling pathways. Furthermore, we identified 20 differentially expressed PRGs between LC and RC. Our findings provide new insights into the difference between LC and RC, and uncover the potential biomarkers for the treatment of LC and RC.