Metabolic characterization of the tumor microenvironment orchestrates therapeutic strategies and clinical outcomes in pancreatic cancer.

Metabolic reprogramming and immunosuppressive tumor microenvironment (TME) are hallmark features driving pancreatic ductal adenocarcinoma (PDAC) progression. Despite the therapeutic potential of targeting immunometabolism, effective strategies remain scarce in clinical practice, likely due to cell-specific metabolic heterogeneity within PDAC TME. Here, we show integration of three algorithms to estimate metabolic fluxomes and pathways using scRNA-seq data, generating a comprehensive cell type-specific metabolic atlas. Leveraging 460 PDAC samples, we establish a TME-metabolism subtyping system, classifying PDAC into three subtypes (TMS1-3) with distinct immune-metabolic profiles and clinical outcomes. TMS1, characterized by low immune infiltrates, is susceptible to ferroptosis inducers. TMS2, enriched in macrophages, responds to chemoimmunotherapy with inhibition of glutamine synthetase. TMS3, characterized by matrix remodeling, responds to glycolysis inhibitors and albumin-paclitaxel. Finally, we develop a computational classifier for subtype discrimination. Together, this study delineates the metabolic heterogeneity of the PDAC TME and proposes a classification system that suggests promising therapeutic targets.