Construction of a prognostic model of colon cancer patients based on metabolism-related lncRNAs.

Background: Many studies have shown that metabolism-related lncRNAs may play an important role in the pathogenesis of colon cancer. In this study, a prognostic model for colon cancer patients was constructed based on metabolism-related lncRNAs. Methods: Both transcriptome data and clinical data of colon cancer patients were downloaded from the TCGA database, and metabolism-related genes were downloaded from the GSEA database. Through differential expression analysis and Pearson correlation analysis, long non-coding RNAs (lncRNAs) related to colon cancer metabolism were obtained. CRC patients were divided into training set and verification set at the ratio of 2:1. Based on the training set, univariate Cox regression analysis was utilized to determine the prognostic differential expression of metabolic-related lncRNAs. The Optimal lncRNAs were obtain by Lasso regression analysis, and a risk model was built to predict the prognosis of CRC patients. Meanwhile, patients were divided into high-risk and low-risk groups and a survival curve was drawn accordingly to determine whether the survival rate differs between the two groups. At the same time, subgroup analysis evaluated the predictive performance of the model. We combined clinical indicators with independent prognostic significance and risk scores to construct a nomogram. C index and the calibration curve, DCA clinical decision curve and ROC curve were obtained as well. The above results were all verified using the validation set. Finally, based on the CIBERSORT analysis method, the correlation between lncRNAs and 22 tumor-infiltrated lymphocytes was explored. Results: By difference analysis, 2491 differential lncRNAs were obtained, of which 226 were metabolic-related lncRNAs. Based on Cox regression analysis and Lasso results, a multi-factor prognostic risk prediction model with 13 lncRNAs was constructed. Survival curve results suggested that patients with high scores and have a poorer prognosis than patients with low scores (P<0.05). The area under the ROC curve (AUC) for the 3-year survival and 5-year survival were 0.768 and 0.735, respectively. Cox regression analysis showed that age, distant metastasis and risk scores can be used as independent prognostic factors. Then, a nomogram including age, distant metastasis and risk scores was built. The C index was 0.743, and the ROC curve was drawn to obtain the AUC of the 3-year survival and the 5-year survival, which were 0.802 and 0.832, respectively. The above results indicated that the nomogram has a good predictive effect. Enrichment analysis of KEGG pathway revealed that differential lncRNAs may be related to chemokines, amino acid and sugar metabolism, NOD-like receptor and Toll-like receptor activation as well as other pathways. Finally, the analysis results based on the CIBERSORT algorithm showed that the lncRNAs used to construct the model had a strong polarized correlation with B cells, CD8+T cells and M0 macrophages. Conclusion: 13 metabolic-related lncRNAs affecting the prognosis of CRC were screened by bioinformatics methods, and a prognostic risk model was constructed, laying a solid foundation for the research of metabolic-related lncRNAs in CRC.