Abstract A028: Multiomic modelling of pancreatic IPMN stroma reveals distinct tertiary lymphoid structure distribution: Mapping the transcriptomic landscape via regional bulk, single-cell and subcellular approaches

Abstract

Abstract Background Intraductal Papillary Mucinous Neoplasms (IPMN) remain the largest subtype of cystic pancreatic cancer precursors. Cancer risk has been observed to vary between histological grades and histological subtypes. Spatial characterisation has revealed variation in immune cell distribution disease subtypes. Methods A cohort of 14 surgically resected IPMN tumors across a range of histological grades (Low-grade LG, high-grade HG and invasive IPMN cancers IPMN-PDAC) and subtypes (gastric, intestinal, pancreaticobiliary) underwent multimodal spatial interrogation. After expert histopathological annotation; tissue sections underwent regional whole transcriptome analysis with TempO-seq and 10x Visium. Extensive region analysis with NanoString GeoMx employing segmentation using staining for epithelium (PanCK) fibroblast stroma (aSMA) was performed. A tissue microarray was then constructed, and representative cores underwent single-cell and subcellular spatial transcriptomic analysis with NanoString CosMx. Raw count data and digital images were exported and analysed using Seurat, Giotto, SPATA2 and custom R pipelines. Region deconvolution, Harmony integration, Leiden Clustering, Gene Set Enrichment Analysis, Trajectory Mapping, and Moran’s I Analysis were performed. Results TempO-seq revealed up-regulation of canonical oncogene expression in all IPMN-PDAC cases as compared to LG and HG IPMN. There were no statistically significant differences in expression comparing LG and HG IPMN lesions. Visium differential gene expression and clustering models identified cancer epithelium, stroma and lymphatic components that correlated with histopathological annotations. Immune cell rich spots were identified via gene ontology of top 50 marker genes. The spatial trajectory of B-cell expression signatures in IPMN-PDAC showed greater concentration and less variability than HG tumors, Moran’s I -0.34 vs 0.12 (P = 0.021). B-cells in HG IPMN were showed greater concentration and less variability than LG IPMN, Moran’s I -0.15 vs 0.02 (P=0.05). Much of this concentration was in identified lymphoid aggregates or tertiary lymphoid structures. The spatial trajectory of T-cell expression. The spatial trajectory of macrophage expression showed greater concentration and less variability than in LG and HG tumors. GeoMx analysis of stromal regions identified up regulation of B cell signatures in IPMN PDAC and HG IPMN when compared to LG IPMN. Markers associated with TLS formation were found to be significantly upregulated including CXCL13, CXCR5 and LAMP3. Pancreaticobiliary subtypes were found to have the least concentrated B-cell distribution across subtypes. Conclusions Immune cell composition was found to vary across histological grades and subtypes of IPMN. A paradoxical relationship was observed between T-cell infiltrates and B-cell and macrophage populations. The greater proportion of tertiary lymphoid structures, and high levels of TLS specific gene expression, may indicate a propensity for b-cell activation in IPMN tumors which progress to malignancy. Citation Format: Andrew J. Cameron, Assya Legrini, Colin S. Wood, Craig Nourse, Yoana Doncheva, Claire Kennedy Dietrich, Colin Nixon, Jennifer Hay, Fraser Duthie, Pawel Herzyk, Jennifer Morton, Nigel B. Jamieson. Multiomic modelling of pancreatic IPMN stroma reveals distinct tertiary lymphoid structure distribution: Mapping the transcriptomic landscape via regional bulk, single-cell and subcellular approaches [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 A028.