Genomic analysis and filtration of novel prognostic biomarkers based on metabolic and immune subtypes in pancreatic cancer

Abstract

An increasing number of studies have indicated that patients with pancreatic cancer (PC) can be classi ed into various molecular subtypes and bene t from some precise therapy. Nevertheless, the interaction between metabolic and immune subtypes in the tumor microenvironment (TME) remains unknown. Thus, we utilized unsupervised consensus clustering and ssGSEA analysis respectively to construct molecular subtypes related to metabolism and immunity. Meanwhile, diverse metabolic and immune subtypes were characterized by distinct prognoses and TME. Afterward, we ltrated the overlapped genes based on the differentially expressed genes (DEGs) between the metabolic and immune subtypes by lasso regression and Cox regression, and used them to build risk score signature which led to PC patients was categorized into high-and low-risk groups. Furthermore, high-risk patients have a better response for various chemotherapeutic drugs in the Genomics of Drug Sensitivity in Cancer (GDSC) database. Finally, we built a nomogram with the risk group, age, and the number of positive lymph nodes to predict the survival rates of each PC patient with average 1-year, 2-year, and 3-year areas under the curve (AUCs) equal to 0.792, 0.752, and 0.751. In summary, the risk score signature based on the metabolism and immune molecular subtypes can accurately predict the prognosis and guide treatments of PC, meanwhile, the metabolism-immune biomarkers may provide novel target therapy for PC.