Metabolic genes show excellent prognostic ability for clear cell renal cell carcinoma

Abstract Background Renal cell carcinoma (RCC) is one of the major malignant tumors of the urinary system, with a high mortality rate and a poor prognosis. Clear cell renal cell carcinoma (ccRCC) is the most common subtype of RCC. Although the diagnosis and treatment methods have been significantly improved, the incidence and mortality of ccRCC are high and still increasing. The occurrence and development of ccRCC are closely related to the changes of classic metabolic pathways. This article aims to explore the relationship between metabolic genes and the prognosis of patients with ccRCC. Patients and methods: Gene expression profiles of 63 normal kidney tissues and 446 ccRCC tissues from TCGA database and gene expression profiles of 39 ccRCC tissues from GEO database were used to obtain differentially expressed genes (DEGs) in ccRCC. Through the the KEGG gene sets of GSEA database, we obtained metabolic genes (MGs). Univariate Cox regression analysis was used to identify prognostic MGs. Lasso regression analysis was used to eliminate false positives because of over-fitting. Multivariate Cox regression analysis was used to established a prognostic model. Gene expression data and related survival data of 101 ccRCC patients from ArrayExpress database were used for external validation. Survival analysis, ROC curve analysis, independent prognostic analysis and clinical correlation analysis were performed to evaluate this model. Results We found that there were 479 abnormally expressed MGs in ccRCC tissues. Through univariate Cox regression analysis, Lasso regression analysis and multivariate Cox regression analysis, we identified 4 prognostic MGs (P4HA3, ETNK2, PAFAH2 and ALAD) and established a prognostic model (riskScore). Whether in the training cohort, the testing cohort or the entire cohort, this model could accurately stratify patients with different survival outcomes. The prognostic value of riskScore and 4 MGs was also confirmed in the ArrayExpress database. Results of GSEA analysis show that DEGs in patients with better prognosis were enriched in metabolic pathways. Then, a new Nomogram with higher prognostic value was constructed to better predict the 1-year OS, 3-year OS and 5-year OS of ccRCC patients. In addition, we successfully established a ceRNA network to further explain the differences in the expression of these MGs between high-risk patients and low-risk patients Conclusion We have successfully established a risk model (riskScore) based on 4 MGs, which could accurately predict the prognosis of patients with ccRCC. Our research may shed new light on ccRCC patients' prognosis and treatment management.