Stefanoli Michele, La Rosa Stefano, Sahnane Nora, Romualdi Chiara, Pastorino Roberta, Marando Alessandro, Capella Carlo, Sessa Fausto, Furlan Daniela
Section of Anatomic Pathology, Department of Surgical and Morphological Sciences, University of Insubria and Ospedale di Circolo, Varese, Italy.
Neuroendocrinology. 2014;100(1):26-34. doi: 10.1159/000365449. Epub 2014 Jul 5.
BACKGROUND/AIMS: The occurrence and clinical relevance of DNA hypermethylation and global hypomethylation in pancreatic neuroendocrine tumours (PanNETs) are still unknown. We evaluated the frequency of both epigenetic alterations in PanNETs to assess the relationship between methylation profiles and chromosomal instability, tumour phenotypes and prognosis.
In a well-characterized series of 56 sporadic G1 and G2 PanNETs, methylation-sensitive multiple ligation-dependent probe amplification was performed to assess hypermethylayion of 33 genes and copy number alterations (CNAs) of 53 chromosomal regions. Long interspersed nucleotide element-1 (LINE-1) hypomethylation was quantified by pyrosequencing.
Unsupervised hierarchical clustering allowed to identify a subset of 22 PanNETs (39%) exhibiting high frequency of gene-specific methylation and low CNA percentages. This tumour cluster was significantly associated with stage IV (p = 0.04) and with poor prognosis in univariable analysis (p = 0.004). LINE-1 methylation levels in PanNETs were significantly lower than in normal samples (p < 0.01) and were approximately normally distributed. 12 tumours (21%) were highly hypomethylated, showing variable levels of CNA. Interestingly, only 5 PanNETs (9%) were observed to show simultaneously LINE-1 hypomethylation and high frequency of gene-specific methylation. LINE-1 hypomethylation was strongly correlated with advanced stage (p = 0.002) and with poor prognosis (p < 0.0001). In the multivariable analysis, low LINE-1 methylation status and methylation clusters were the only independent significant predictors of outcome (p = 0.034 and p = 0.029, respectively).
The combination of global DNA hypomethylation and gene hypermethylation analyses may be useful to define distinct subsets of PanNETs. Both alterations are common in PanNETs and could be directly correlated with tumour progression.
背景/目的：胰腺神经内分泌肿瘤（PanNETs）中DNA高甲基化和整体低甲基化的发生情况及其临床相关性仍不清楚。我们评估了PanNETs中这两种表观遗传改变的频率，以评估甲基化谱与染色体不稳定性、肿瘤表型及预后之间的关系。
在一组特征明确的56例散发性G1和G2级PanNETs中，进行了甲基化敏感的多重连接依赖探针扩增，以评估33个基因的高甲基化和53个染色体区域的拷贝数改变（CNA）。通过焦磷酸测序对长散在核苷酸元件1（LINE-1）的低甲基化进行定量。
无监督层次聚类可识别出22例PanNETs（占39%），这些肿瘤表现出基因特异性甲基化频率高且CNA百分比低。该肿瘤簇与IV期显著相关（p = 0.04），在单变量分析中与预后不良相关（p = 0.004）。PanNETs中LINE-1的甲基化水平显著低于正常样本（p < 0.01），且大致呈正态分布。12例肿瘤（占21%）高度低甲基化，CNA水平各异。有趣的是，仅观察到5例PanNETs（占9%）同时出现LINE-1低甲基化和基因特异性甲基化频率高 的情况。LINE-1低甲基化与晚期显著相关（p = 0.002），与预后不良相关（p < 0.0001）。在多变量分析中，低LINE-1甲基化状态和甲基化簇是仅有的独立且显著的预后预测指标（分别为p = 0.034和p = 0.029）。
整体DNA低甲基化和基因高甲基化分析相结合可能有助于定义PanNETs的不同亚组。这两种改变在PanNETs中都很常见，且可能与肿瘤进展直接相关。
