A tumor purity-based gene signature predicts prognosis and shapes the immune microenvironment in PAAD

Pancreatic adenocarcinoma (PAAD) is characterized by extreme lethality and a dense stroma that masks tumor-intrinsic biology. Tumor purity—the proportion of malignant cells within tumor tissue—shapes both transcriptomic signals and therapeutic response, yet purity-driven molecular programs in PAAD remain poorly defined. Using TCGA bulk RNA-seq and single-cell RNA-seq dataset GSE155698, we identified purity-associated genes and constructed a prognostic model via differential expression, LASSO, and multivariate Cox regression. Predictive performance was assessed through Kaplan–Meier, ROC, and independent prognostic analyses. Cell–cell communication was used to characterize immune interactions within the tumor microenvironment. Protein-level expression was validated by IHC and multiplex IHC in an independent cohort ( n = 110). A three-gene signature (AMH, PHACTR1, C5AR2) stratified PAAD patients into risk groups with significantly different survival. AMH and PHACTR1 were associated with favorable prognosis, whereas C5AR2 predicted poor outcome. Integration with single-cell data revealed that low-risk tumors exhibited enhanced antigen presentation and stronger CD8⁺ T-cell activation, implicating tumor purity in shaping immune surveillance. Protein validation confirmed consistent expression patterns in clinical samples. We identify a tumor purity–linked three-gene signature that robustly predicts patient survival and reveals an immune modulating axis involving CD8⁺ T-cell activation. These biomarkers provide mechanistic insight into how tumor purity influences PAAD progression and may guide personalized therapy.