Abstract
Background: Pancreatic ductal adenocarcinoma (PDAC) is a highly aggressive cancer with poor prognosis, and its progression is driven by epithelial plasticity and tumor microenvironment remodeling. Finding biomarkers that are responsible for the turning point from the early stage to the aggressive phase would facilitate clinical management. Method: In this study, we employed single-cell RNA sequencing to characterize a distinct subpopulation of highly proliferative epithelial cells undergoing a transitional phase during PDAC progression. By linking to cell cycle dysregulation, epithelial differentiation, and clinical staging, we constructed a gene-based risk score model using Lasso Cox regression. The expression of selected genes within the model was further validated using qPCR. Results: The model demonstrated robust predictive power for patient prognosis, TNM staging, and chemotherapy sensitivity. Further analysis of the tumor microenvironment revealed intensified crosstalk between a specific fibroblast subpopulation and transitional epithelial cells, mediated largely by collagen signaling. This stromal-epithelial interaction was found to contribute to the fibrotic barrier characteristic of PDAC. Additionally, immune profiling uncovered altered infiltration patterns, particularly involving natural killer (NK) cells in high-risk patients, suggesting mechanisms of immune tolerance and inhibition. Conclusions: These findings offer potential avenues for early detection, risk stratification, and targeted therapeutic strategies in PDAC.