Construction of the metabolic reprogramming-associated gene signature for clear cell renal cell carcinoma prognosis prediction

Abstract

Background: Metabolism reprogramming is a hallmark that associates tumor growth, metastasis, progressive, and poor prognosis. However, the metabolism-related molecular patterns and mechanism in clear cell renal cell carcinoma (ccRCC) remain unclear. Herein, the purpose of this study was to identify metabolism-related molecular pattern and to investigate the characteristics and prognostic values of the metabolism-related clustering.

Methods: We comprehensively analyzed the differentially expressed genes (DEGs), and metabolism-related genes (MAGs) in ccRCC based on the TCGA database. Consensus clustering was used to construct a metabolism-related molecular pattern. Then, the biological function, molecular characteristics, Estimate/immune/stomal scores, immune cell infiltration, response to immunotherapy, and chemotherapy were analyzed. We also identified the DEGs between subclusters and constructed a poor signature and risk model based on LASSO regression cox analysis and univariable and multivariable cox regression analyses. Then, a predictive nomogram was constructed and validated by calibration curves.

Results: A total of 1942 DEGs (1004 upregulated and 838 downregulated) between ccRCC tumor and normal samples were identified, and 254 MRGs were screened out from those DEGs. Then, 526 ccRCC patients were divided into two subclusters. The 7 metabolism-related pathways enriched in cluster 2. And cluster 2 with high Estimate/immune/stomal scores and poor survival. While, cluster 1 with higher immune cell infiltrating, expression of the immune checkpoint, IFN, HLA, immune activation-related genes, response to anti-CTLA4 treatment, and chemotherapy. Moreover, we identified 295 DEGs between two metabolism-related subclusters and constructed a 15-gene signature and 9 risk factors. Then, a risk score was calculated and the patients into high- and low-risk groups in TCGA-KIRC and E-MTAB-1980 datasets. And the prediction viability of the risk score was validated by ROC curves. Finally, the clinicopathological characteristics (age and stage), risk score, and molecular clustering, were identified as independent prognostic variables, and were used to construct a nomogram for 1-, 3-, 5-year overall survival predicting. The calibration curves were used to verify the performance of the predicted ability of the nomogram.

Conclusion: Our finding identified two metabolism-related molecular subclusters for ccRCC, which facilitates the estimation of response to immunotherapy and chemotherapy, and prognosis after treatment.

Keywords: Clear cell renal cell carcinoma; Gene signature; Metabolic reprogramming; Prognosis; Survival predicting.