Abstract
The connection between metabolic reprogramming and tumor progression has been demonstrated in an increasing number of researches. However, further research is required to identify how metabolic reprogramming affects interpatient heterogeneity and prognosis in clear cell renal cell carcinoma (ccRCC). In this work, Single-cell RNA sequencing (scRNA-seq) based deconvolution was utilized to create a malignant cell hierarchy with metabolic differences and to investigate the relationship between metabolic biomarkers and prognosis. Simultaneously, we created a machine learning-based approach for creating metabolism related prognostic signature (MRPS). Following the administration of immune checkpoint inhibitors (ICIs), MRPS shows consistent and strong performance and is an independent risk factor for overall survival. Furthermore, compared to 51 published signatures and conventional clinical features, MRPS showed substantially higher accuracy. GGT6, an essential metabolic indicator and component of MRPS, has been proven to support proliferation and invasion in ccRCC. In conclusion, MRPS has the potential to be a highly effective tool in improving the clinical results of patients with ccRCC.