Skip to main content
Premium Trial:

Request an Annual Quote

Integrated Glioma Analysis Uncovers Epigenetics-based Subtypes

NEW YORK (GenomeWeb) – In a study published online today in Cell, an international research team led by investigators in the US and Brazil proposed a new classification scheme for diffuse glioma that takes into account genetic and epigenetic cues from the central nervous system tumors.

By combining clinical information on patient progression patterns and outcomes, information on diffuse glioma-associated genes in more than 1,000 grade II, III, or IV grade tumors, and new methylation data, the researchers identified seven glioma categories that show promise for classifying lower- and high-grade tumors, and, in some cases, predicting patient outcomes.

"We discovered low-grade and high-grade gliomas mixed together within these different epigenetic subtypes," co-senior author Houtan Noushmehr, director of OMICs and Bioinformatics at the University of São Paulo's Ribeirão Preto Medical School, said in a statement. "This was an unexpected finding and allowed us to further understand the progression of gliomas within the different subtypes."

Traditionally, most adult diffuse gliomas have been catalogued as oligodendroglioma, astrocytoma, glioblastoma, and other types based largely on histological features, Noushmehr and his co-authors noted.

Last spring, members of the Cancer Genome Atlas described three diffuse lower-grade glioma subtypes based on genome, exome, microRNA, and/or RNA profiling on almost 300 matched tumor and normal samples. And there is mounting evidence that molecular features such as IDH1/2 mutation status and chromosome 1 and 19 co-deletions can provide clues to glioma patient outcomes.

For the new study, the researchers brought together TCGA data for 1,122 diffuse gliomas spanning grades II, III, and IV, including 290 lower-grade gliomas that the TCGA had previously analyzed, 226 newly assessed lower-grade glioma tumors, and samples from 606 individuals with more advanced gliomas.

More than half of the tumors were grade IV gliobastomas, though the sample set also included oligodendrogliomas, astrocytomas, and oligoastrocytomas, the team noted. It scrutinized gene expression and DNA copy number data for more than 1,000 of the tumors, while exome sequencing, DNA methylation, and protein expression data was available for 820 tumors, 932 tumors, and 473 tumors, respectively.

The team identified 75 significantly mutated genes based on point mutations and small insertions and deletions data — a set that included 45 genes not linked to glioma in the past. With the help of CNV profiles and fusions found with RNA sequence data, the researchers also verified glioma driver mutations described previously and tracked down new driver candidate genes in pathways related to chromatin regulation and Ras-Raf-MEK-ERK signaling.

Meanwhile, their combined copy number and mutation data uncovered cohesin complex gene glitches in around 16 percent of the gliomas.

Along with analyses focused on co-occurring alterations in the glioma tumors, the team clustered the tumors based on overlapping gene mutations, gene expression, copy number, and DNA methylation marks.

For example, the analysis pointed to the presence of three IDH1/2-mutated glioma subtypes with distinct epigenetic patterns. Among them: tumors containing IDH1/2 mutations alongside chromosome 1 and 9 co-deletions and two clusters of IDH1/2-mutated tumors without the chromosome co-deletions.

Tumors with wild type IDH1/2 made up four more methylation- and gene expression-based clusters — patterns the researchers used to develop a methylation-based classifier that they tested in hundreds more IDH1/2 wild type gliomas.

Finally, by incorporating clinical information for the glioma cases, the researchers were able to find survival differences that seemed to coincide with their newly identified epigenetic classifiers, as well as clues to the molecular events that underlie patient progression from lower-grade glioma to glioblastoma.

"The epigenomic data defined by profiling DNA methylation levels for each of our samples allowed us to determine with accuracy which patient will present the best clinical outcome and the worst," Noushmehr explained.