Abstract B106: Spatially-constrained optimal transport interaction analysis reveals therapy-associated remodeling in the pancreatic cancer microenvironment

Abstract

Abstract Pancreatic ductal adenocarcinoma (PDAC) is one of the most lethal malignancies with a five-year overall survival of only 11%. Treatment failure is ubiquitous, driven by genetic and phenotypic heterogeneity combined with a highly desmoplastic and immunosuppressive microenvironment. Identifying novel therapeutic targets in the tumor microenvironment (TME) is critical to improving patient outcomes and can be advanced through understanding the cell intrinsic states and cell-cell interactions driving clinically-relevant properties. Recent applications of single-cell and spatial proteo-transcriptomic technologies to PDAC have enhanced our understanding of intratumoral heterogeneity, cell state plasticity, and neighborhood composition. However, prior studies were unable to provide high-plex molecular information while preserving in situ spatial relationships at single-cell resolution, limiting detailed analyses of cell-cell interactions and potential therapeutic avenues. Advances in spatial technology have enabled subcellular resolution while profiling up to several thousand transcripts. We performed spatial molecular profiling (SMI; NanoString CosMx) on 13 patient-derived PDAC tumors using a 990-plex mRNA panel, which yielded 753,133 high confidence single-cell profiles. We then developed a novel method for inferring multicellular interactions—Spatially Constrained Optimal Transport Interaction Analysis (SCOTIA), an optimal transport model with a cost function that includes both spatial distance and ligand–receptor (LR) gene expression. We used SCOTIA to dissect the remodeled TME in response to neoadjuvant FOLFIRINOX and chemoradiotherapy. Our results uncovered a marked change in LR interactions between cancer-associated fibroblasts and malignant cells in response to treatment, which was supported by results from experiments using a murine tumoroid co-culture system. We observed in treated specimens a lower abundance of myofibroblastic CAFs (myCAFs) and higher prevalence of inflammatory CAFs (iCAFs), which express high levels of IL6 family cytokines and spatially co-localize with cancer cell subtypes previously associated with therapeutic resistance (i.e., neural-like, basal/mesenchymal). Two of the most consistent treatment-enriched LR interactions between CAFs and cancer cells were CLCF1-CNTFR and LIF-IL6ST, members of the IL6 cytokine family. While CAF-derived IL6 family ligands and downstream JAK/STAT3 and MAPK activation have been linked to induction of a mesenchymal cancer cell phenotype in several malignancies as well as neuroendocrine differentiation in prostate cancer, an association with the neural-like malignant cell state in pancreatic cancer has not been previously established. Overall, this study demonstrates that characterization of the TME using high-plex single-cell spatial transcriptomics allows for identification of molecular interactions that may play a role in the emergence of chemoresistance and establishes a translational spatial biology paradigm that can be broadly applied to other malignancies, diseases, and treatments. Citation Format: Alexander T.F. Bell, Jacob T. Mitchell, Ashley L. Kiemen, Melissa Lyman, Kohei Fujikura, Jae W. Lee, Erin Coyne, Sarah M. Shin, Pei-Hsun Wu, Jacquelyn W. Zimmerman, Denis Wirtz, Won J. Ho, Neeha Zaidi, Elizabeth Thompson, Elizabeth M. Jaffee, Laura D. Wood, Elana J. Fertig, Luciane T. Kagohara. Spatially-constrained optimal transport interaction analysis reveals therapy-associated remodeling in the pancreatic cancer microenvironment [abstract]. In: Proceedings of the AACR Special Conference in Cancer Research: Pancreatic Cancer; 2023 Sep 27-30; Boston, Massachusetts. Philadelphia (PA): AACR; Cancer Res 2024;84(2 Suppl):Abstract nr B106.