Abstract
Few studies had investigated genome-wide methylation in glioblastoma multiforme (GBM). Our goals were to study differential methylation across the genome in gene promoters using an array-based method, as well as repetitive elements using surrogate global methylation markers. The discovery sample set for this study consisted of 54 GBM from Columbia University and Case Western Reserve University, and 24 brain controls from the New York Brain Bank. We assembled a validation dataset using methylation data of 162 TCGA GBM and 140 brain controls from dbGAP. HumanMethylation27 Analysis Bead-Chips (Illumina) were used to interrogate 26,486 informative CpG sites in both the discovery and validation datasets. Global methylation levels were assessed by analysis of L1 retrotransposon (LINE1), 5 methyl-deoxycytidine (5m-dC) and 5 hydroxylmethyl-deoxycytidine (5hm-dC) in the discovery dataset. We validated a total of 1548 CpG sites (1307 genes) that were differentially methylated in GBM compared to controls. There were more than twice as many hypomethylated genes as hypermethylated ones. Both the discovery and validation datasets found 5 tumor methylation classes. Pathway analyses showed that the top ten pathways in hypomethylated genes were all related to functions of innate and acquired immunities. Among hypermethylated pathways, transcriptional regulatory network in embryonic stem cells was the most significant. In the study of global methylation markers, 5m-dC level was the best discriminant among methylation classes, whereas in survival analyses, high level of LINE1 methylation was an independent, favorable prognostic factor in the discovery dataset. Based on a pathway approach, hypermethylation in genes that control stem cell differentiation were significant, poor prognostic factors of overall survival in both the discovery and validation datasets. Approaches that targeted these methylated genes may be a future therapeutic goal.
Highlights
Cancers are recognized as driven as much by epigenetic as well as genetic changes [1]
glioblastoma multiforme (GBM) and Brain Control Tissues in the Discovery Dataset This study was approved by the institutional review boards (IRBs) of Columbia University (CUMC) and Case Western Reserve University (CWRU)
We investigated the value of methylation biomarkers other than MGMT or Glioma CpG Island Methylator Phenotype (GCIMP), such as LINE1, 5 methyl-deoxycytidine (5m-dC) and 5 hydroxylmethyl-deoxycytidine (5hm-dC), as potential independent prognostic factors
Summary
Cancers are recognized as driven as much by epigenetic as well as genetic changes [1]. There have been few studies that evaluated differential promoter methylation across the entire genome in glioblastoma multiforme (GBM), which is the most common type of malignant brain tumors in adults [2,3,4,5]. The primary goal of some studies, such as the Cancer Genome Altas Project (TCGA), was to characterize methylation patterns in tumors and to correlate with other genomic alterations such as gene mutations, copy number alterations and expression [6]. The investigation of differential methylation poses a challenge, because unlike colon, breast or prostate cancers, it is not possible to obtain matching ‘‘normal’’ tissues during surgery for GBM. Previous reports on genome-wide methylation in normal brain tissues showed methylation patterns varied between neuro-anatomically distinct regions, and methylation level may change in the brain with increasing age [7,8,9]. An accurate profile of differential methylation will require appropriate control tissues with age and neuro-anatomical distribution matching those of glioma subjects
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
