Search

Background: Grade IV glioblastomas exist in two forms, primary (de novo) glioblastomas (pGBM) that arise without precursor lesions, and the less common secondary glioblastomas (sGBM) which develop from earlier lower grade lesions. Genetic heterogeneity between pGBM and sGBM has been documented as have differences in the methylation of individual genes. A hypermethylator phenotype in grade IV GBMs is now well documented however there has been little comparison between global methylation profiles of pGBM and sGBM samples or of methylation profiles between paired early and late sGBM samples. Methods: We performed genome-wide methylation profiling of 20 matched pairs of early and late gliomas using the Infinium HumanMethylation450 BeadChips to assess methylation at >485,000 cytosine positions within the human genome. Results: Clustering of our data demonstrated a frequent hypermethylator phenotype that associated with IDH1 mutation in sGBM tumors. In 80% of cases, the hypermethylator status was retained in both the early and late tumor of the same patient, indicating limited alterations to genome-wide methylation during progression and that the CIMP phenotype is an early event. Analysis of hypermethylated loci identified 218 genes frequently methylated across grade II, III and IV tumors indicating a possible role in sGBM tumorigenesis. Comparison of our sGBM data with TCGA pGBM data indicate that IDH1 mutated GBM samples have very similar hypermethylator phenotypes, however the methylation profiles of the majority of samples with WT IDH1 that do not demonstrate a hypermethylator phenotype cluster separately from sGBM samples, indicating underlying differences in methylation profiles. We also identified 180 genes that were methylated only in sGBM. Further analysis of these genes may lead to a better understanding of the pathology of sGBM vs pGBM. Conclusion: This is the first study to have documented genome-wide methylation changes within paired early/late astrocytic gliomas on such a large CpG probe set, revealing a number of genes that maybe relevant to secondary gliomagenesis. [ABSTRACT FROM AUTHOR]

Background: There are several high throughput approaches to identify methylated genes in cancer. We utilized one such recently developed approach, MIRA (methylated-CpG island recovery assay) combined with CpG island arrays to identify novel genes that are epigenetically inactivated in breast cancer. Results: Using this approach we identified numerous CpG islands that demonstrated aberrant DNA methylation in breast cancer cell lines. Using a combination of COBRA and sequencing of bisulphite modified DNA, we confirmed 5 novel genes frequently methylated in breast tumours; EMILIN2, SALL1, DBC1, FBLN2 and CIDE-A. Methylation frequencies ranged from between 25% and 63% in primary breast tumours, whilst matched normal breast tissue DNA was either unmethylated or demonstrated a much lower frequency of methylation compared to malignant breast tissue DNA. Furthermore expression of the above 5 genes was shown to be restored following treatment with a demethylating agent in methylated breast cancer cell lines. We have expanded this analysis across three other common epithelial cancers (lung, colorectal, prostate). We demonstrate that the above genes show varying levels of methylation in these cancers. Lastly and most importantly methylation of EMILIN2 was associated with poorer clinical outcome in breast cancer and was strongly associated with estrogen receptor as well as progesterone receptor positive breast cancers. Conclusion: The combination of the MIRA assay with CpG island arrays is a very useful technique for identifying epigenetically inactivated genes in cancer genomes and can provide molecular markers for early cancer diagnosis, prognosis and epigenetic therapy. [ABSTRACT FROM AUTHOR]