Abstract A093: Defining the histologic and molecular landscape of the human pancreas: A clinical, histopathologic, and transcriptomic multi-feature interactive atlas of donor pancreata

Abstract

Pancreatic ductal adenocarcinoma (PDAC) remains among the deadliest cancers, largely due to late diagnosis. To advance early detection, we must define the clinical, histologic, and molecular features of precursor lesions in the general population, and then identify which features drive carcinogenesis. Progress is limited by the lack of truly normal human pancreas tissue with preserved molecular integrity from the general population, and not the routinely captured tumor adjacent normal from surgically resectable PDAC. To address this need, we established a large-scale program to collect and characterize donor pancreata recovered after circulatory (DCD) or brain death (DBD). Specimens are systematically processed, pathologically curated, molecularly characterized, digitized, and integrated with donor-level metadata. For each donor, two tissue samples were collected from each of the major anatomic regions of the pancreas (head, body, and tail), enabling assessment of intrapancreatic heterogeneity. The cohort currently includes 293 donors, 61% are male, with a mean age of 49 years (18–83 years). Self-reported race includes 78% White, 15% Black/African American, 6% other races, and 22% Hispanic ethnicity. Pancreata from 293 unique donors, representing more than 1,700 unique tissue samples and corresponding hematoxylin and eosin (H&E)-stained sections, were digitized at 40X as whole-slide images. All slides underwent expert histopathologic review with digital annotation at the structural level. Histopathologic alterations were identified in more than 80% of donors, including ( i ) pancreatic intraepithelial neoplasia (PanIN) and intraductal papillary mucinous neoplasms (IPMNs), ( ii ) metaplastic changes (i.e., acinar-to-ductal, squamous), and ( iii ) others (i.e., benign cysts). Pancreatic specific cell types (i.e., acinar, ductal, and β-islet cells) from 31 donors were profiled using GeoMx Digital Spatial Profiler with the Human Whole Transcriptome Atlas and next-generation sequencing readout, revealing cell-type-specific molecular signatures and precursor lesion-specific patterns. To integrate clinical, histologic, image, and molecular features, we developed a normal pancreas multi-feature interactive atlas: a virtual workspace linking H&E whole-slide images, histologic annotations, spatial gene-expression data, and donor clinical, medication, and demographic metadata. This platform enables real-time interrogation of integrated histologic, molecular, and clinical datasets. This ongoing program provides an integrated framework for defining the histologic and molecular landscape of the human pancreas in the general population. To our knowledge, it represents the largest fully annotated cohort of normal pancreatic tissue and preneoplastic lesions derived from DCD and DBD donors. Coupled with molecular profiling of major pancreatic cell types and detailed clinical metadata, this resource is uniquely positioned to support all pancreatic cancer and diabetes research, inform early detection strategies, and enable broader studies of pancreatic disease.