Construction and validation of four-metabolism related-long non-coding RNAs as potential signature in prognosis of colon cancer

BACKGROUND: Emerging evidence suggests that metabolism plays important roles in the initiation and progression of colon cancer (CC) and the outcomes of CC patients. Long non-coding RNAs (lncRNA) are key regulators of regulatory molecules linking to a wide variety of cancer cellular functions. This study aims to develop a metabolic lncRNA signature to help better predict prognosis for CC patients. METHODS: In the current study, the transcriptome data and clinical data of CC was downloaded from The Cancer Genome Atlas (TCGA). Metabolism-related gene sets were downloaded from the Molecular Signatures Database (MSigDB). Differential lncRNAs related to metabolism was obtained by performing the correlations between differential expression profile of metabolic genes and lncRNAs. To construct a prognostic model of CC based on metabolism-related lncRNAs, we divided patients, whose clinical data were available, into a training set and a validation set at a ratio of 7:3. The prognostic metabolism related-lncRNA signature was established using the training set by univariate and multivariate Cox regression analysis, and the validation set was used to test the capacity of the prognostic model. The correlation between risk score and clinicopathological features, immune function GO and KEGG analysis was investigated using the entire set. Finally, GSEA pathway enrichment analysis was carried out on the entire set samples for the high- and low- risk groups. RESULTS: We identified 604 differential lncRNAs and 252 genes related to metabolism. After univariate and multivariate Cox regression analysis, four lncRNAs were finally identified to build a signature, which was verified the effectiveness by the TCGA validation set. The multivariate Cox regression analysis showed that the risk score, age of diagnosis and T stage were independent prognostic factor for CC patients. It is shown that some immunopathogenesis, GO items and KEGG pathways demonstrated difference between high- and low- risk group. CONCLUSIONS: We developed a four-metabolism related-lncRNA signature for prognostic prediction of CC, which may help select high-risk subpopulation patients who require more aggressive therapy or intervention.