Abstract A091: Single-cell spatial transcriptomics reveals reprograming molecular hallmarks of PanIN initiation and progression preceding histological changes
Canh Hiep Nguyen, Anders Ohman, Winston R. Becker, Carla Rujana, Tate Jacobson. Weltzin, Sung Chul Hong, Len Taing, Micaela Morgado, Andressa Dias Costa, Sloane C. Furniss, Yasutoshi Takashima, Cibelle F. Lima, Alexander C. Jordan, Dalia Elganainy, Lauren K. Brais, Emma C. Coleman, Seema Chugh, Dan Y. Gui, Julien Dilly, Vasilena Gocheva, Joseph D. Mancias, Andrew J. Aguirre, Brian M. Wolpin, Jonathan A. NowakAbstract
Background:
Recent studies have identified a high prevalence of low grade pancreatic intraepithelial neoplasia (PanIN) in the pancreata of healthy adults. This study aimed to characterize the transcriptomic landscape of human PanINs at single cell resolution, with the goal of identifying hallmarks of initiation and progression to invasive pancreatic ductal adenocarcinoma (PDAC).
Methods:
Whole slide spatial transcriptomic profiling was conducted on 14 primary surgical resected specimens containing 13 PDACs, 1 neuroendocrine tumor, 80 PanIN foci, and adjacent normal pancreas, using the CosMx Human 6k Discovery Panel. After the CosMx runs, tissue sections were automatically segmented into multiple fields of view (FOVs). Histological review was performed by a trained pathologist to annotate FOVs containing PDAC, high-grade (HG) PanIN, low-grade (LG) PanIN, normal duct, and other tissue compartments. After quality control, cells with at least 50 transcripts were retained. Unsupervised clustering was performed, and cell types were annotated by combined evaluation of marker genes and spatial distribution of cells correlated to morphology. Individual gene expression and gene expression programs were compared among histologically normal duct, LG-PanIN, HG-PanIN, and PDAC lesions. Basal-like and classical scores were also generated as per previously defined Moffitt signatures.
Results:
After filtering 8,466 FOVs comprising 10,455,346 cells were obtained. Based on histology, we selected 220 normal duct, 453 LG-PanIN, 48 HG-PanIN, and 1,714 PDAC FOVs for further analyses, excluding FOVs containing both PanIN and PDAC, mainly acinar cells or stroma. We identified molecular programs associated with PanINs compared to normal ducts, LG-PanIN compared to HG-PanIN, and HG-PanIN compared to invasive cancer. Histologically distinct LG-PanIN strongly expressed classical/secretory gene signatures, including TFF1, PIGR, and GOLM1. This transcriptional program, however, was also found in 6% of histologically normal duct epithelial cells. Comparing LG- to HG-PanINs, differentially expressed genes included S100A6, STAT3, PDXK, CST3, and MUC1, which were upregulated in HG-PanIN. Notably, we found spatially discrete cells within morphologically defined LG-PanIN displaying transcriptomic programs consistent with HG-PanIN and PDAC, potentially marking LG-PanINs with a higher likelihood of progression to HG-lesions.
Conclusions:
Employing single cell transcriptomic approaches together with traditional pathology techniques allowed us to identify gene expression profiles that distinguish normal ducts, LG-PanIN, HG-PanIN and invasive PDAC. Remarkably, transcriptomic profiles consistent with HG-PanIN were observed in a subset of morphologically defined LG-PanIN, indicating a subset of early lesions that may be more likely to progress to HG-PanIN before morphologic changes are evident. Thus, these data suggest that single cell transcriptomic profiles can be employed to identify candidate targets for PDAC interception. Further studies are ongoing in human and mouse tissues.
Citation Format:
Canh Hiep Nguyen, Anders Ohman, Winston R. Becker, Carla Rujana, Tate Jacobson. Weltzin, Sung Chul Hong, Len Taing, Micaela Morgado, Andressa Dias Costa, Sloane C. Furniss, Yasutoshi Takashima, Cibelle F. Lima, Alexander C. Jordan, Dalia Elganainy, Lauren K. Brais, Emma C. Coleman, Seema Chugh, Dan Y. Gui, Julien Dilly, Vasilena Gocheva, Joseph D. Mancias, Andrew J. Aguirre, Brian M. Wolpin, Jonathan A. Nowak. Single-cell spatial transcriptomics reveals reprograming molecular hallmarks of PanIN initiation and progression preceding histological changes [abstract]. In: Proceedings of the AACR Conference on Pancreatic Cancer: New Frontiers in Biology and Therapeutic Development; 2026 Sep 25-28; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2026;86(18_Suppl_2):Abstract nr A091.