Abstract B107: Spatial transcriptomics analysis of PanIN reveals loss of pro-inflammatory signaling and the presence of cancer-associated fibroblasts

• Published in: Cancer research (Chicago, Ill.) Vol. 84; no. 2_Supplement; p. B107
• Main Authors: Bell, Alexander T.F., Mitchell, Jacob T., Kiemen, Ashley L., Lyman, Melissa, Fujikura, Kohei, Lee, Jae W., Coyne, Erin, Shin, Sarah M., Wu, Pei-Hsun, Zimmerman, Jacquelyn W., Wirtz, Denis, Ho, Won J., Zaidi, Neeha, Thompson, Elizabeth, Jaffee, Elizabeth M., Wood, Laura D., Fertig, Elana J., Kagohara, Luciane T.
• Format: Journal Article
• Language: English
• Published: 16-01-2024
• ISSN: 1538-7445; 1538-7445
• DOI: 10.1158/1538-7445.PANCA2023-B107

Pancreatic intraepithelial neoplasia (PanIN) is a premalignant lesion that potentially progresses into pancreatic ductal adenocarcinoma (PDAC). Since PanIN diagnosis is restricted to formalin-fixed and paraffin-embedded (FFPE) samples, little is known about PanIN interactions with the surrounding microenvironment due to the technical limitations to perform genome-wide single-cell analysis. We developed whole transcriptome spatial technology for FFPE to provide the ideal tool to profile PanINs in their microenvironment. We performed FFPE spatial transcriptomics (ST) using the Visium (10x Genomics) approach in a cohort of 14 PanIN lesions (9 low-grade and 5 high-grade). To overcome experimental limitations, we developed an artificial intelligence method to integrate imaging, spatial transcriptomics, and single-cell RNA-seq (scRNA-seq) analyses. Imaging analysis of the same stained sections used for ST was applied before gene expression exploration. Using machine learning to annotate cell types, we were able to deconvolve and quantify the cell types per ST spot using cell morphology instead of a scRNA-seq gene expression reference. Subsequently, we applied transfer learning to a scRNA-seq PDAC atlas that was built from multiple published datasets. The whole transcriptome ST findings were confirmed using targeted single-cell transcriptomics (Xenium, 10x Genomics) and proteomics (imaging mass cytometry, IMC). The analysis revealed that PanINs already exhibit molecular and cellular features of PDAC, including expression of the PDAC classical subtype signature and the presence of the same cancer-associated fibroblast (CAF) subtypes enriched in invasive carcinoma stages. The CAF subtypes, including the recently described antigen-presenting CAFs, were confirmed at single-cell resolution using Xenium and IMC. Integrating multi-omics data with a PDAC scRNA-seq atlas through transfer learning, revealed that an inflammatory pattern gradually decreases during PanIN progression, accompanied by a compensatory increase in proliferation pathways in the PanIN cells relative to normal duct cells. This trend between inflammation and proliferation was confirmed specifically in PanIN cells using Xenium. Our experimental and computational approach provides valuable insights into the dynamics of pancreatic tumor progression from PanIN lesions using innovative methods to provide a unique spatial multi-omics reference of PanINs.
Citation Format: Alexander T.F. Bell, Jacob T. Mitchell, Ashley L. Kiemen, Melissa Lyman, Kohei Fujikura, Jae W. Lee, Erin Coyne, Sarah M. Shin, Pei-Hsun Wu, Jacquelyn W. Zimmerman, Denis Wirtz, Won J. Ho, Neeha Zaidi, Elizabeth Thompson, Elizabeth M. Jaffee, Laura D. Wood, Elana J. Fertig, Luciane T. Kagohara. Spatial transcriptomics analysis of PanIN reveals loss of pro-inflammatory signaling and the presence of cancer-associated fibroblasts [abstract]. In: Proceedings of the AACR Special Conference in Cancer Research: Pancreatic Cancer; 2023 Sep 27-30; Boston, Massachusetts. Philadelphia (PA): AACR; Cancer Res 2024;84(2 Suppl):Abstract nr B107.