Transcriptomic Analyses of Normal Human Pancreata Reveal the Presence of Cancer Subtypes that Correlate with Acinar...

Molecular subtypes exist in pancreatic ductal adenocarcinoma (PDAC); however, comparable subtypes are not known to exist in normal pancreas. A large, racially diverse cohort of normal, human pancreatic acinar cells was obtained from islet transplantation centers. RNA isolated from uncultured acini and those acini that had undergone acinar ductal metaplasia (ADM) was sequenced. Data analysis from 69 uncultured, normal acini showed two clusters of samples, one that aligns with exocrine-resembling tissue (ERT) signatures and another with the classical and basal (C/B) subtypes. Gene expression associated with the ERT signature decreased following 6 days of ADM transdifferentiation, whereas those assigned to the C/B subtype increased. Subtype classification findings were confirmed in two independent cohorts of transcriptomic data of normal pancreas. A direct correlation exists between ancestral admixture and the molecular index of ADM; those specimens with higher African ancestral admixture showed a greater propensity for ADM compared with those with a higher Ameridigenous admixture. The extent of morphologic ADM decreased in the order of Black > White > Hispanic race. Analysis of associated metadata revealed that potential confounders such as diabetes, obesity, age, gender, or specimen handling and processing variables did not influence the distribution of samples across subtype classification. We report that normal human pancreatic acinar tissues group into molecular subtypes like those used to classify PDAC and that subtype assignment and ADM are influenced by the donor's race.

Human tissue models provide direct insights into human pancreatic biology and preserve the donor-to-donor heterogeneity present in the general population. By utilizing these models, our study suggests that a subset of normal pancreata exhibits a preexisting permissive state that renders acinar cells more susceptible to early reprogramming and ADM.