Abstract 3878: Therapy-associated remodeling of pancreatic cancer revealed by single-cell spatial transcriptomics and optimal transport analysis

Abstract The aggressive nature of pancreatic ductal adenocarcinoma (PDAC) is driven by cell-intrinsic features and cell-extrinsic interactions between tumor cells and the desmoplastic stroma, which is infiltrated with heterogeneous populations of cancer-associated-fibroblasts (CAFs) and immune cells. These features are able to drive emergent properties such as chemoresistance through diverse and not yet fully elucidated mechanisms. Through single-nucleus RNA-seq and whole-transcriptome digital spatial profiling of PDAC patient tumors, we previously reported marked heterogeneity in the malignant cells and CAF composition of PDAC. In this project, we performed high-plex single-cell spatial molecular imaging (990-plex SMI, Nanostring) to dissect the therapy-remodeled multicellular neighborhoods and malignant-stroma cell interactions in primary resected human pancreatic cancer with (n = 6) or without (n = 7) neoadjuvant chemotherapy and radiotherapy. SMI of the PDAC samples yielded ~753,000 high-confidence single cells and captured the heterogeneity of malignant cells and CAFs. We identified three malignant subtypes: classical (CLS), basal-like (BSL), and the recently characterized neural-like progenitor (NRP) malignant subtype that exhibits stem-like and neurodevelopmental features. To leverage the single-cell spatial resolution of SMI to further dissect treatment-associated changes in cell-cell interactions, we developed Spatially Constrained Optimal Transport Interaction Analysis (SCOTIA), an optimal transport model with a cost function that includes both spatial distance and ligand-receptor gene expression. Our results uncovered a marked enrichment in ligand-receptor interactions between CAFs and malignant cells in response to treatment. In combination with several complementary methods, including examination of downstream target gene expression and ex vivo 3D co-culture of CAF and PDAC cells, we found that IL-6 family signaling was the most consistently enriched interaction between CAFs and malignant cells after treatment. This led us to hypothesize that IL-6 signaling between CAFs and malignant cells may induce chemoresistance in cancer cells. To test this hypothesis, we treated three human cancer cell lines (AsPC1, Panc1, Panc0203) with 5FU with or without IL-6 and discovered that IL-6 induced chemoresistance for two of the lines (Panc1, Panc0203). Taken together, we have integrated novel experimental and computational approaches to enable high-resolution, spatially-guided discovery of treatment-associated remodeling in the pancreatic cancer microenvironment. Further studies will be conducted to dissect additional mechanisms by which candidate cell-cell interactions promote disease progression and/or resistance to cytotoxic therapy, with the ultimate goal of guiding novel therapeutic development. Citation Format: JINGYI CAO, Carina Shiau, Dennis Gong, Mark T. Gregory, Xunqin Yin, Jae-Won Cho, Peter L. Wang, Jennifer Su, Jason W. Reeves, Jimmy A. Guo, Nicole A. Lester, Jung Woo Bae, Ryan Zhao, Martin Hemberg, William L. Hwang. Therapy-associated remodeling of pancreatic cancer revealed by single-cell spatial transcriptomics and optimal transport analysis [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 3878.