In silico development and clinical validation of novel 8 gene signature based on lipid metabolism related genes in colon adenocarcinoma.

Background Changes in lipid metabolism pathways play a major role in colon carcinogenesis and development. Hence, we conducted a systematic analysis of lipid metabolism-related genes to explore new markers that predict the prognosis of colon adenocarcinoma (COAD). Methods The non-negative Matrix Factorization (NMF) algorithm was applied to identify the molecular subtypes based on lipid metabolism-related genes. A weighted correlation network analysis (WCGNA) was used to identify co-expressed genes, and Lasso multivariate Cox analysis was performed to build a risk prognosis model. A timer database was used to analyze the immune infiltration of the gene signature and the GSCALite database was used for genome-wide analysis of the gene signature. Results TCGA-COAD samples were divided into 3 subtypes based on lipid metabolism-related genes. 2739 genes were identified by WGCNA analysis. Finally, an 8-gene signature (RTN2, FYN, HEYL, FAM69A, FBXL5, HMGN2, LGALS4, STOX1) was constructed that demonstrated good robustness in different datasets, as well as an independent risk factor for colon cancer patients' prognosis. In addition, our model's predictive efficacy overall was higher than that of the other published models, and the 8 genes' expression analysis indicated that RTN2, HEYL, and STOX1 were all expressed highly significantly in COAD, while FAM69A, FBXL5, LGALS4, FYN and HMGN2 were expressed significantly poorly in cancer tissues, which was confirmed in immunohistochemistry. The 8 genes were expressed significantly differently in COAD immune subtypes and correlated with clinical variables. Genome-wide analysis revealed that the STOX1 mutation frequency was the highest, and genome methylation influenced HEYL, FAM69A, and STOX1 gene expression significantly; further, the expression of HEYL and FBXL5 was correlated positively with Copy number variation (CNV) and was regulated significantly by CNV in most cancers. FBXL5 was correlated significantly with austocystin d and bafilomycin and played an important role in anti-tumor and immunotherapy. The HEYL, FYN, FAM69A, and RTN2 genes' expression was associated with the EMT pathway's activation, while LGALS4 and STOX1 were associated significantly with the EMT pathway's inhibition. Conclusion This study constructed an 8-gene signature as a novel marker to predict colon cancer patients' survival.