A05 - HPB Interrogating the spatial biology of pancreatic cancer-associated fibroblasts

Abstract Background Cancer-associated fibroblasts (CAFs) are key players in the fibrotic tumour microenvironment (TME) of pancreatic cancer. Recent evidence has demonstrated the existence of multiple CAF subtypes, however there is relatively little evidence validating and corroborating these CAF subtypes in resected human pancreatic cancer. Furthermore, the spatial interactions between CAF subtypes and the epithelial and immune compartments of the pancreatic cancer TME is not well understood. Method We designed and optimised a bespoke 8-plex immunofluorescence panel for the purposes of CAF subtype profiling and elucidation of CAF spatial biology: panCK, FAP, SMA, CTGF, S100A4, IL6, LIF, PDGFR. This was applied to a cohort of n=215 resected pancreatic cancer, well characterised with highly granular clinico-pathological and transcriptomic data through the Australian Pancreatic Cancer Genome Initiative. Spatial analysis was performed with novel computational biology techniques. Results We demonstrate seven CAF subtypes which cluster to form distinct, highly prognostic “enrichment profiles” (EPs). The cold EP represents quiescent CAFs in close proximity to immune cells (multivariate HR 0.45, 95% CI 0.30-0.68, p<0.001), and the fibrotic EP represents activated CAFs in close proximity to epithelial cells (multivariate HR 2.22, 95% CI 1.48-3.35, p<0.001). We also demonstrate the prognostic power of high spatial entropy (loss of tissue architecture, correlating with the fibrotic EP) vs low spatial entropy (preservation of tissue architecture, correlating with the cold EP) (HR 2.22, 95% CI 1.47-3.35, p<0.001). Conclusion We begin to unravel the elusive spatial biology of pancreatic CAFs in a large resected human cohort. We see changes in immune cell infiltrate, modulation of cellular relationships, and greater tissue disorganisation as a result of CAF activation. These features are highly prognostic and we hypothesise they are driven by changes in crosstalk between TME compartments. Current work focuses on application of these techniques to a large resected neoadjuvantly-treated cohort, allowing further interrogation of the notorious fibrotic post-neoadjuvant pancreatic cancer TME. If accepted for presentation we look forward to presenting both cohorts simultaneously.