Abstract B121: Morphospatial profiling of cancer-associated fibroblasts reveals architectural subtypes of pancreatic ductal adenocarcinoma

Abstract Introduction: Pancreatic ductal adenocarcinoma (PDAC) cancer associated fibroblasts (CAFs) are a complex and heterogenous group of cells and are a key component of the tumour microenvironment (TME). Our understanding of CAF functional diversity has increased in recent years, however certain aspects of their biology remain elusive, such as the spatial organisation of CAFs, their interactions with other TME compartments, and their contribution to global TME architecture. Better understanding of these will significantly accelerate therapeutic development. Methods: We performed multiplex immunofluorescence on tissue microarrays of resected human pancreatic cancer (n=227) with an 8-plex CAF panel and a molecular subtyping panel. After assigning bulk cell type (CAF – FAP+CK-, Epithelium – FAP-CK+, Other – FAP-CK-), we integrated CAF morphology, spatial pseudotime analysis, and TME architectural scoring in a hierarchical manner (pixel > cell > neighborhood > tissue module > patient) to define architectural subtypes of PDAC with internal validation. Results: Architectural subtypes are defined as follows (name, dominant CAF, median disease-specific survival): Cohesive subtype (iCAF, 40 months) Rich in dense, cohesive, and tightly-packed epithelium. Sparse, loose stroma with a heterogenous population of iCAF-like CAFs which exhibit migration away from epithelium on spatial pseudotime modelling. Inflammatory subtype (roundCAF, 31 months) Dominated by a rich infiltrate of presumed immune cells (FAP-CK-). Highly organised with low spatial entropy. Associated with the classical subtype. CAFs have “cold” marker expression, are round in shape with a smooth surface, and exhibit migration towards epithelium on spatial pseudotime modelling. Activated subtype (Enlarged myCAF, 17 months) Dense stroma infiltrated by large myCAF-like CAFs. High spatial entropy with loss of TME compartmentalisation. Immune excluded. Strongly associated with the squamous/basal-like subtype. Fragmented subtype (Stellate myCAF, 13 months) Loose, dis-cohesive epithelium infiltrated by a CAF-dense stroma rich in myCAF-like CAFs with stellate morphology. High CAF-epithelial cell spatial contact. Associated with poor histological differentiation and poorest prognosis. Hypothesised to represent impending or established (occult) metastasis. Conclusion: Morphospatial profiling reveals striking associations between CAF shape, function, and spatial niche, and adds crucial perspective to our understanding of CAF biology beyond current methods. Combined with spatial pseudotime analysis, it offers exciting insight into CAFs in transitory states between subtypes, allowing maximum leverage of datasets which lack a true temporal dimension. In addition, architectural subtyping reveals the complex and heterogenous spatial organisation of the pancreatic TME, and the co-evolution of CAFs with these organisational changes. Integrating this with histopathological analysis lays foundations for the development of novel spatial biomarkers of prognosis, therapeutic development, and treatment response. Citation Format: Adam Bryce, Silvia Martinelli, Rachel Pennie, Leah Officer-Jones, Leonor Schubert Santana, Fiona Ballantyne, Catherine Ficken, Ian Powley, Australian Pancreatic Cancer Genome Initiative (APGI), John Le Quesne, Fieke Froeling, Stephan Dreyer, David Chang. Morphospatial profiling of cancer-associated fibroblasts reveals architectural subtypes of pancreatic ductal adenocarcinoma [abstract]. In: Proceedings of the AACR Special Conference in Cancer Research: Advances in Pancreatic Cancer Research—Emerging Science Driving Transformative Solutions; Boston, MA; 2025 Sep 28-Oct 1; Boston, MA. Philadelphia (PA): AACR; Cancer Res 2025;85(18_Suppl_3):Abstract nr B121.