The Metabolic Rheostat Index Uncovers Pathway-Level Coordination, Immunosuppressive Microenvironment Remodeling, and Therapeutic Co-Vulnerabilities in Pancreatic Ductal Adenocarcinoma

Pancreatic ductal adenocarcinoma (PDAC) is characterized by extensive metabolic reprogramming, yet how major metabolic pathways are organized at single-cell resolution, and how this relates to the tumor microenvironment and therapeutic vulnerabilities, remains incompletely understood. We analyzed 32,227 cells from the GSE155698 discovery cohort and 32,627 cells from an independent healthy-pancreas reference (GSE229413; n = 5 donors) and introduce the Metabolic Rheostat Index (MRI), the standard deviation across per-cell glycolysis, OXPHOS, and hypoxia pathway scores. A pseudobulk analysis, hallmark gene-set validation, unified-threshold classification, a GDSC2 drug-sensitivity analysis with robust regression, EPIC tumor-purity-adjusted deconvolution, and a survival analysis of 183 TCGA-PAAD samples were performed. The pseudobulk analysis revealed a monotonic MRI gradient from healthy pancreata (mean = 0.259) to adjacent-normal tissue (0.322) to PDAC tumor tissue (0.352; p = 0.0039, Cohen’s d = 1.437), indicating progressive metabolic polarization during malignant transformation driven predominantly by cancer-associated fibroblasts rather than malignant epithelial cells. The glycolysis–hypoxia correlation was stronger in tumor cells (r = 0.121 vs. 0.019), and ERK/MAPK inhibitors showed the greatest drug sensitivity (r = 0.597, p = 0.0008). The Warburg Index correlated with fibroblast infiltration (partial r = 0.312, p = 1.76 × 10−5) but was not an independent prognostic marker (HR = 0.78, p = 0.321). These findings indicate that PDAC tissue undergoes progressive, stroma-driven metabolic polarization associated with an immunosuppressive microenvironment, providing a multi-dimensional single-cell framework for future metabolic-targeted therapeutic strategies.