Distinct DNA methylation signatures in neuroendocrine tumors specific for primary site and inherited predisposition

Purpose: To compare the deoxyribonucleic acid (DNA) methylation signature of neuroendocrine tumors (NETs) by primary tumor site and inherited predisposition syndromes von Hippel–Lindau disease (VHL) and multiple endocrine neoplasia type 1 (MEN1). Methods: Genome-wide DNA methylation (835 424 CpGs) of 96 NET samples. Principal components analysis (PCA) and unsupervised hierarchical clustering analyses were used to determine DNA methylome signatures. Results: Hypomethylated CpGs were significantly more common in VHL-related versus sporadic and MEN1-related NETs (P < .001 for both comparisons). Small-intestinal NETs (SINETs) had the most differentially methylated CpGs, either hyper- or hypomethylated, followed by duodenal NETs (DNETs) and pancreatic NETs (PNETs, P < .001 for all comparisons). There was complete separation of SINETs on PCA, and 3 NETs of unknown origin clustered with the SINET samples. Sporadic, VHL-related, and MEN1-related PNETs formed distinct groups on PCA, and VHL clustered separately, showing pronounced DNA hypomethylation, while sporadic and MEN1-related NETs clustered together. MEN1-related PNETs, DNETs, and gastric NETs each had a distinct DNA methylome signature, with complete separation by PCA and unsupervised clustering. Finally, we identified 12 hypermethylated CpGs in the 1A promoter of the APC (adenomatous polyposis coli) gene, with higher methylation levels in MEN1-related NETs versus VHL-related and sporadic NETs (P < .001 for both comparisons). Conclusions: DNA CpG methylation profiles are unique in different primary NET types even when occurring in MEN1-related NETs. This tumor DNA methylome signature may be utilized for noninvasive molecular characterization of NETs, through DNA methylation profiling of biopsy samples or even circulating tumor DNA in the near future.