Spatially revealed malignant and microenvironment ecosystems in PanIN- and IPMN-derived PDAC

Introduction: Pancreatic ductal adenocarcinoma (PDAC) arises from distinct precursor lesions, mainly pancreatic intraepithelial neoplasias (PanINs) and intraductal papillary mucinous neoplasms (IPMNs), which differ in genetics, differentiation ecosystem, and clinical behavior. PanIN-derived PDAC typically demonstrate more molecular and spatial heterogeneity, with pronounced classical-basal divergence. IPMN-derived adenocarcinomas display different histologic types: intestinal-type lesions often develop colloid carcinoma, where adenocarcinomas arising from pancreatobiliary(PB)-type IPMN resemble conventional PanIN-derived PDAC. Defining how precursor origin shapes malignant states, spatial microenvironment ecosystems are essential for PDAC classification. Method: Single-cell spatial transcriptomics was performed by CosMx SMI (Bruker) in whole transcriptome and 6K-panel. Two different Tissue Microarrays (TMAs) cohorts were used (1.5-mm cores) include PanIN-origin PDAC (15 patients) and IPMN-origin PDAC (10 patients). Cell Segmentation performed by CellPose and strict quality control (QC) thresholding yielded 459,195 high-quality cells. Data was log-normalized and integrated with scVI using patient-level correction. Hierarchical annotation was performed by combining expert histopathological alignments with Leiden-based refinements. Downstream spatial analysis used Python based methods including Squidpy. Results: Cross-cohort integration revealed marked origin-dependent divergence in the malignant spatial atlas. Malignant epithelium in PanIN-PDAC exhibited higher heterogeneity with more diverse and dispersed cell states. In contrast, the tumor cells in IPMN-derived PDAC presented obvious differentiation features for each carcinoma type. Intestinal-type IPMN PDAC developed unique malignant program that exhibiting typical goblet-like subtype, which was not observed in either other IPMN-type PDAC or PanIN-derived PDAC. However, the PB-type IPMN-PDAC showed substantial similarity with PanIN-PDAC, sharing multiple malignant programs, primarily including pEMT, progenitor cell-like and ADM-like subtypes. Furthermore, PanIN-PDAC retained specific tumor programs as exocrine-like, squamous, and inflammatory-association basal-like. Notably, although PB-type in IPMN PDAC and PanIN-PDAC share similarities in some tumor states, further analysis revealed significant differences in the spatial patterns and cell communication between tumor and microenvironments (TME). This suggests potentially different evolution of microenvironments. Conclusion: Precursor origin fundamentally shapes PDAC malignant states and stromal-immune ecosystems. Together, this work establishes a spatial framework for understanding PDAC heterogeneity and highlights the importance of origin-aware stratification in clinic and biology.