Abstract 7493: Spatially resolved insights into epithelial-stromal co-evolution in pancreatic ductal adenocarcinoma

Abstract Pancreatic ductal adenocarcinoma (PDAC) remains one of the deadliest cancers, with a five-year survival rate of 13%. The tumor microenvironment (TME) in PDAC includes a heterogeneous population of cells. In genetically engineered mouse models, the TME evolves alongside the formation and progression of epithelial precursor lesions to cancer. However, the evolution of the TME in human samples has not been thoroughly characterized, primarily due to limited access to pre-malignant pancreatic tissue. Leveraging a collaboration with Gift of Life Michigan, an organ donor organization, we previously reported that precursor lesions are abundant in the normal human pancreas (Carpenter et al., Cancer Discovery, 2023). To date, we have collected over 100 human pancreata, a dataset that enables us to map the characteristics of these lesions and their surrounding microenvironment across variables such as age, sex, and race. Here, we aim to model the co-evolution of the stroma and neoplastic epithelium during tumor progression. We utilized spatial transcriptomics technology with the 10x Visium platform, analyzing over 20 human pancreata during the discovery phase. We applied a spatially aware clustering approach to delineate distinct tissue domains across normal, tumor-adjacent, and tumor samples. We leveraged a previously established single-cell RNA-seq atlas (n=41) to deconvolute cell-type fractions and cell-type-specific gene expression profiles within each spatial transcriptomics spot. As a result, we identified distinct epithelial structures (e.g., Acinar, Ducts, PanINs, tumor) as well as stromal domains (e.g., tertiary lymphoid structures (TLS), fibrotic regions). We hypothesized that stromal gene expression is influenced by the proximity to and the density of different epithelial structures. Therefore, we employed a kernel density estimation approach to model gene expression in stromal cells, specifically fibroblasts and macrophages, as a function of epithelial-specific density. This analysis uncovered distinct characteristics of the stroma surrounding precursor lesions compared to tumor stroma, emphasizing spatially regulated gene expression programs that likely drive TME remodeling in PDAC. Future directions include performing cell-cell interaction analysis to further elucidate epithelial-stromal crosstalk, complemented by in vitro functional experiments to validate the pro- or anti-tumorigenic roles of the identified markers. This work lays the foundation for spatially resolved insights into PDAC biology, informing therapeutic strategies targeting the TME. Citation Format: Ahmed M. Elhossiny, Padma Kadiyala, Alexander Bray, Jamie Mills, Hannah Watkoske, Arvind Rao, Jude Okoye, Atul Deshpande, Jiaqi Shi, Elana J. Fertig, Filip Bednar, Yaqing Zhang, Eileen S. Carpenter, Timothy Frankel, Marina Pasca di Magliano. Spatially resolved insights into epithelial-stromal co-evolution in pancreatic ductal adenocarcinoma [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2025; Part 1 (Regular Abstracts); 2025 Apr 25-30; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2025;85(8_Suppl_1):Abstract nr 7493.