Mapping cell type-resolved transcriptomic profiles to patient survival in pancreatic cancer.

Traditional bulk-level transcriptomic sequencing cannot link cell type-specific gene expression to patient survival. In this study, we integrate single-nucleus RNA-seq and longitudinal data from 152 patients with pancreatic ductal adenocarcinoma (PDAC), profiling 1.2 million cells to construct a prognostic map connecting cell type-resolved gene expression with overall survival. Using a single-cell-resolved spatial transcriptomic platform, we further analyze 3.1 million cells and correlate their spatial distribution with therapeutic response. To empower the translational research community, we develop ctPANDA, an interactive platform offering cell-type-level prognostic insights. This atlas identifies PLOD2 as a promising target, with elevated expression predicting poorer outcomes across eight cell types. We have developed a proof-of-concept compound that can effectively degrade PLOD2 and inhibit PDAC progression in vivo. Collectively, this study advances the prognostic analysis from bulk-level to cell-type resolution, establishing a framework linking gene expression to clinical outcomes and providing actionable insights for target discovery and precision oncology.