A Novel Model of Pancreatic Cancer Dormancy Reveals Mechanistic Insights and a Dormancy Gene Signature with Human Relevance

Abstract

Latent recurrences following curative-intent pancreatic cancer surgery is a major clinical problem thought to be due to the reactivation of dormant tumor cells that disseminate before the primary tumor has been removed. How dormancy is established and ultimately reversed to drive recurrence is poorly understood. Here we introduce a novel model of pancreatic cancer dormancy that mimics early and latent survival outcomes of resected patients. Using single-cell transcriptomics we compared primary, dormant, and reactivated tumor cells and found the primary and reactivated tumor cell transcriptomes clustered together with and away from the dormant tumor cells. Using a chromatin accessibility assay we found dormancy exhibits large scale changes in chromatin remodeling. Dormant tumor cells express cancer stem cell markers that are lost upon reactivation and are chemotherapy resistant. We identified a dormancy gene signature and investigated this in patients undergoing surgery for localized PC by isolating cells from the primary tumor and liver disseminated tumor cells (DTCs) for single-cell transcriptomics. We found the signature correlated with DTCs indicating that these cells are dormant at the time of surgery. The signature also identified CCL5 as a novel dormancy marker in PC. Mechanisms of PC dormancy include upregulation of the transcriptional repressor Dec2 which drives quiescence, monoallelic suppression of the mutant KRAS allele by DNA methylation, and immunoregulation. We conclude that PC dormancy is a highly plastic and heterogeneous cellular state governed by tumor cell autonomous and non-autonomous mechanisms. One Sentence SummaryA novel model of resectable pancreatic cancer reveals pancreatic cancer dormancy is characterized by significant cellular plasticity, heterogeneity and chromatin remodeling