Developmental enhancer signatures stratify and predict outcomes of non-functional pancreatic neuroendocrine tumors

Most pancreatic neuroendocrine tumors (PNETs) do not produce symptoms of hormonal excess and are hence considered ‘non-functional’. Their clinical behaviors vary widely, emphasizing the need for a robust classification with prognostic power. Using enhancer maps to infer regulatory programs, we find that the large majority of non-functional PNETs fall into two major sub-types –A and B– that reflect alpha and beta endocrine cell ontogeny, respectively. A and B tumors have similar clinical presentations and histology, but express distinct lineage-specifying transcription factors, ARX or PDX1. In immunohistochemistry, 84% of 142 non-functional PNETs expressed one or the other factor, rarely both. Longitudinal data on 104 cases revealed markedly different outcomes, irrespective of MEN1 mutation status: relapse occurred almost exclusively in patients with type A tumors. These findings reveal a robust molecular stratification that provides insight into cellular origins of non-functional PNETs, accurately predicts disease course, and informs clinical decisions and future trial design. RNA-seq, H3K27ac and H3K4me2 profiles profiling in carinoids and pancreatic neuroendocrine tumors Due to patient privacy concerns, raw data have been submitted to dbGAP, accession phs001910.v1.p1.