In this study, we established and verified an eight-gene metabolic risk model based on TCGA and GEO, to predict the prognosis of colon cancer. Firstly, we identified differentially expressed metabolic genes from TCGA and constructed a prognostic model by LASSO. Then, we verified the model by Kaplan–Meier curves, risk score, ROC curves, univariate and multivariate cox regression based on TCGA and GEO. Finally, we performed KEGG and GO analyses by GSEA.
Colon cancer is a malignant tumor with poor prognosis, so novel prognostic biomarkers is urgently needed. The prognostic models based on metabolic genes has been established in rectal cancer{#12}, lung adenocarcinoma{#13}, hepatocellular carcinoma{#14}, and acute myelogenous leukemia{#15}. However, it has not been applied on colon cancer. In this study, we identified differentially expressed metabolic genes and constructed a prognostic model firstly. In this model, the expression of ENOPH1, ACAA2, PAFAH1B3, CPT2, and ACOX1 predicted better prognosis, whereas the expression of PKM, GPX3 and PTGDS predicted worse prognosis. Most of the genes in our model have been proved to take part in pathogenesis, progression and prognosis of cancers.
Enolase-phosphatase 1 (ENOPH1), a bifunctional enolase-dephosphorylase enzyme{#16}, is required for polyamine biosynthesis {#17}. Previous studies showed that the overexpression of ENOPH1 remarkably promoted cell migration and invasiveness, whereas the downregulation of ENOPH1 significantly impaired cell migration and invasiveness{#18}. Acetyl-Coenzyme A acyltransferase 2 (ACAA2) exerts an effect on β-oxidation of fatty acid, which then provide energy{#19}. Previous showed that ACAA2 could abolished the apoptosis in human hepatocellular carcinoma{#20}, and played a vital role in the the metabolism processes in gliomas{#21}. The higher expression of ACAA2 was associated with better prognosis of colorectal cancer{#22}. Platelet activating factor acetyl hydrolase 1B3 (PAFAH1B3) is reported to play an important role in tumorigenesis and aggressiveness in many different cancerss{#23}. Blockers of PAFAH1B3 could heightened levels of tumor-suppressing lipids, then impairs pathogenicity of different cancers, including breast, ovarian, melanoma, and prostate cancer{#24}. Carnitine palmitoyl transferase 2 (CPT2), a rate-limiting enzyme for mitochondrial fatty acid transportation, had been proved to be a protective prognostic gene for colorectal cancer{#25}, which is in accordance with our results. Peroxisomal Acyl-Coa Oxidase 1 (ACOX1) is the rate-limiting enzyme in fatty acid β-oxidation. The inhibitor of ACOX1 is SIRT1, which has been proved to prevent oxidative damage and is downregulated in liver cancerr{#26}, suppresses colorectal cancer metastasis by transcriptional repression of miR-15b-5p{#27}.
Pyruvate kinase M (PKM), a metabolic regulator, participated in both glycolytic and non-glycolytic pathways. PKM also acts as protein kinase, which shifts the glucose metabolism from the respiratory chain to anaerobic glycolysis in tumor cells{#28}. PKM2 is upregulated in most cancer types, and contributes to tumorigenesis{#29}, which suggested that it could act as a remarkable therapeutic target{#30}. The glutathione peroxidases-3 (GPX3), a selenocysteine-containing redox enzyme, took part in reactive oxygen species signaling and immunomodulatory{#31}. Previous studies suggested that GPX3 prevented the colitis-associated carcinoma by immunomodulation{#32}. But the correlation between GPX and prognosis of colon cancer is not clear. The tumor suppressive effect of prostaglandin D2 (PGD2) on testicular cancer and gastric cancer has been confirmed{#33}. However, the correlation between PGD2 and colon cancer is unclear.
The results of GSEA revealed many metabolic pathways related to prognosis of colon cancer. In one hand, the results validated the close association between metabolic systems and colon cancer. Most significantly enriched pathways focused on metabolism, including butanoate metabolism, fatty acid metabolism, propanoate metabolism and pyruvate metabolism pathways. In the other hand, most of the metabolic pathways were enriched in the low-risk patients, while the non-metabolic pathways were enriched in the high-risk patients. These results uncovered the underlying molecular mechanisms and potential therapy target. In the future, colon cancer patients might benefit from metabolic-related therapy and management.
However, there were some limitations in this study. Firstly, there was no functional experiment in the real world. Moreover, it is relatively weak to take only metabolic genes into prognostic model because much more complicated mechanisms together contributed to the development and progression of colon cancer.