Abstract PO-111: A human single-cell RNA sequencing atlas of pancreatic ductal adenocarcinoma enables harmonized cell type calling and comprehensive analyses of potential intercellular signaling

Abstract

Abstract Purpose: Pancreatic ductal adenocarcinoma (PDAC) is characterized by a dismal prognosis, low ductal cancer cellularity and a dominant tumor microenvironment (TME) in response to malignant degeneration. Modern single-cell RNA sequencing (scRNA-seq) platforms fundamentally improved the opportunities to analyze PDAC biology through isolation of diverse cell types including ductal, mesenchymal, myeloid and lymphoid populations. Published scRNA-seq data of PDAC patients provide innovative and astounding insights, but are limited by cohort size and intrinsically vulnerable to internal biases. Herein, we present a human single-cell PDAC atlas which links the previous sequencing efforts aiming for increasing depth and robustness with the combined analysis of transcriptomes. Methods: We selected scRNA-seq data of all patients with PDAC from six publicly available datasets (published between 2019-2020). Altogether, 61 different human samples with 142,807 cells were integrated into one dataset leveraging 15,219 genes, which were conclusively identical between all datasets based on the utilized nomenclature in the provided raw data. In addition, we extracted 16 samples with 31,587 cells from control pancreas specimens which were included in three out of the six datasets. The analyses were performed using the R statistics and Python environments utilizing established software including Monocle3 and Seurat. Results: After computational preprocessing of the integrated dataset, cell types were identified based on differentially expressed and canonical markers. The generated PDAC atlas consists of 26% ductal cancer, 2% ductal normal, 12% mesenchymal (stellate cells and cancer-associated fibroblasts), 18% myeloid, 19% lymphoid and 23% other cells (including acinar and endocrine cells). Copy number variation analyses confirmed the discrimination between cancer and normal ductal cells. Certain subpopulations within cell types were mapped based on the expression of supervised gene sets. Within the ductal cancer cell population, the Classical and Basal-like Moffitt signatures coexisted in the majority of patients with distinct ratios and predominance, which were associated with differences in the TME composition. Furthermore, the presence of myofibroblasts and inflammatory fibroblasts could be quantified at the patient-level. The reconstruction of intercellular signaling between ductal cancer cells and several TME components revealed potential ligands, receptors and transcription factors that may functionally define routes and polarity of cross-talk in PDAC. Conclusion: This human scRNA-seq atlas is the largest available dataset of patients with PDAC while harmonizing previously published data. It is engineered to analyze current knowledge gaps with increased rigor and, most importantly, overcomes obstacles related to bulk transcriptome sequencing data. Molecular characteristics of the ductal cancer cells and TME components inferred from the presented framework are promising to identify disease- and patient-specific signaling, key regulators, and therapeutic targets. Citation Format: Benedict Kinny-Köster, Melissa R. Lyman, Dimitrios N. Sidiropoulos, Melanie Loth, Alexandra B. Puscek, Laura D. Wood, Jin He, Jun Yu, Richard A. Burkhart, Elizabeth M. Jaffee, Jacquelyn W. Zimmerman, Elana J. Fertig. A human single-cell RNA sequencing atlas of pancreatic ductal adenocarcinoma enables harmonized cell type calling and comprehensive analyses of potential intercellular signaling [abstract]. In: Proceedings of the AACR Virtual Special Conference on Pancreatic Cancer; 2021 Sep 29-30. Philadelphia (PA): AACR; Cancer Res 2021;81(22 Suppl):Abstract nr PO-111.