Prediction of the evolution of endocrine pancreatic tumors remains difficult based on histological criteria alone. We have previously demonstrated that epigenetic changes are an early event in a mouse model developing insulinomas. Particularly, overexpression of the imprinted IGF2 was caused by the hypermethylation of CpGs in the differentially methylated region 2 (DMR2). Here, we investigated whether IGF2 hypermethylation is also observed in human insulinomas and whether this alteration is common to other human endocrine tumors of the pancreas and the digestive tract. We analyzed the methylation status of 40 CpGs located in the DMR0 and DMR2 of the IGF2 as well as in the H19 DMR by pyrosequencing in a cohort of 62 patients with pancreatic or small intestine endocrine tumors. Altered methylation patterns were observed in all tumor types for the different regions of IGF2, but not for H19. However, hypermethylation of the IGF2 DMR2 was specific for insulinomas and did not occur in any of the other types of tumors which were characterized by a loss of methylation in this region. Gain of methylation in the IGF2 DMR2 in insulinomas correlated with loss-of-imprinting and promoter 4 mediated overexpression of IGF2 at the RNA and protein level. Furthermore, a decreasing degree of methylation in the different regions of IGF2 correlated well with increasing degree of malignancy according to the WHO classification of pancreatic endocrine tumors (PETs), suggesting that methylation of IGF2 might be a useful biomarker for classification and staging of PETs.
Emelyne Dejeux, Robert Olaso, Bertrand Dousset, Anne Audebourg, Ivo G Gut, Benoit Terris, and Jörg Tost
Marc Diedisheim, Solène Dermine, Anne Jouinot, Amandine Septier, Sébastien Gaujoux, Bertrand Dousset, Guillaume Cadiot, Etienne Larger, Jerome Bertherat, Raphael Scharfmann, Benoit Terris, Romain Coriat, and Guillaume Assié
Duodenopancreatic neuroendocrine tumors (DPNETs) aggressiveness is heterogeneous. Tumor grade and extension are commonly used for prognostic determination. Yet, grade classes are empirically defined, with regular up-dates changing the definition of classes. Genomic screening may provide more objective classes, and reflect tumor biology. The aim of this study was to provide a transcriptome classification of DPNETs. We included 66 DPNETs, covering the entire clinical spectrum of the disease in terms of secretion, grade, and stage. Three distinct molecular groups were identified, associated with distinct outcome (log-rank p<0.01): (i) better-outcome DPNETs with pancreatic beta-cell signature. This group was mainly composed of well-differentiated, grade 1 insulinomas; (ii) poor-outcome DPNETs with pancreatic alpha-cell and hepatic signature. This group included all neuroendocrine carcinomas and grade 3 DPNETs, but also some grade 1 and grade 2 DPNETs; and (iii) intermediate-outcome DPNETs with pancreatic exocrine and progenitor signature. This group included grade 1 and grade 2 DPNETs, with some insulinomas. Fibrinogen gene FGA expression was one of the top most expressed liver gene. FGA expression was associated with disease-free survival (HR=1.13, p=0.005), and could be validated on two independent cohorts. This original pathophysiologic insight provides new prognostic classification perspectives.