NEW YORK (GenomeWeb) – Members of the Cancer Genome Atlas have introduced three molecular subtypes for diffuse lower-grade gliomas, producing tumor groupings that seem to share closer prognostic and clinical features than those defined by tumor histology.
The study, published yesterday in the New England Journal of Medicine, "classifies tumors differently than we have in the past," corresponding author Daniel Brat, a pathology and laboratory medicine researcher at Emory University and neuropathologist with the Winship Cancer Institute, told GenomeWeb.
"These tumors have been diagnosed morphologically under the microscope for more than 100 years," he explained, noting that this strategy tends to be subjective and variable.
For the study, Brat and his TCGA colleagues took a crack at categorizing lower-grade gliomas using an integrated profile of genome, exome, microRNA, and RNA sequences, along with array-based copy number profiles, methylation patterns, and more, on samples from almost 300 individuals with lower-grade gliomas.
The approach established "cohesive disease groups based on molecular profiles," Brat explained, apparently creating more reliable classifications than the traditional histological approach.
In particular, the team's unsupervised clustering of this combined molecular data classified low and intermediate forms of astrocytoma, oligoastrocytoma, and oligodendroglioma into three main molecular groups, largely defined by their IDH1/2 gene mutations, chromosome 1 and 19 co-deletion, and TP53 mutation status. That points to the possibility of classifying lower-grade gliomas in the clinic using a small set of informative biomarkers.
"If we all had these advanced molecular platforms at our disposal, we could all use those," Brat said. "But these biomarkers act as a pretty good surrogate for the integrative analysis."
"It suggests that these diseases can be defined with a handful of biomarkers — at least to begin with," he added.
As in the past, for example, he and his colleagues found recurrent and informative roles for mutations in the IDH1 and IDH2 genes — collectively referred to as IDH mutations — in the glioma tumors. They also detected frequent mutations in TP53 and characteristic deletions of chromosomes 1 and 19, known as the 1p/19q co-deletions, since they typically turn up together.
Integrated molecular profiles for tumor and matched normal samples from 293 individuals with grade II and grade III glioma also provided a clearer look at molecularly defined subtypes containing these alterations.
In particular, tumors containing IDH mutations could further be sub-classified depending on the presence or absence of the chromosome 1 and 19 co-deletion. Oligodendrogliomas almost always contained both IDH mutations and the 1p/19q deletion, Brat noted, suggesting these mutations may molecularly define that subtype.
The most commonly mutated genes in that subtype included NOTCH1, FUBP1, and CIC. The IDH-mutant, 1p/19q-deleted tumors also tended to contain alterations that affected the promoter of the telomerase-coding gene TERT, boosting expression of the enzyme.
On the other hand, tumors from the astrocytoma subtype frequently carried IDH mutations in combination with TP53 and the ATRX gene glitches, while tumors lacking IDH mutations defined an aggressive subtype with some genetic and clinical similarities to glioblastoma.
The researchers hope to track down additional markers in the IDH wild type tumors, Brat noted, since there's concern about the feasibility of relying solely on a negative marker to define a disease subtype.
The TCGA study did not uncover a molecular signature associated with so-called oligoastrocytoma — a subtype that was previously defined by ambiguous histological features, Brat said, noting that most were IDH-mutant, TP53-mutatnt tumors that would be molecularly classified as astrocytomas.
Findings from the TCGA analysis largely coincide with those reported in a related paper appearing in the same issue of NEJM.
There, researchers from the University of California at San Francisco and Mayo Clinic used IDH mutations, chromosome 1p/19q co-deletion, and TERT promoter mutations to define five diffuse glioma subtypes based on data from almost 1,100 malignant glioma cases — including those profiled for TCGA — and 11,590 control individuals.
That team found that those molecularly defined subtypes offered clues to everything from age at diagnosis to overall survival times in those with lower-grade glioma. In cases of the grade IV tumors, though, age of diagnosis was a more informative prognostic indicator.
Brat said the study complements analyses performed by TCGA members and praised the Mayo Clinic and UCSF team for tracking down germline mutations with apparent ties to the risk of different glioma subtypes.
He noted that additional work will be needed to start digging into the clinical feasibility of biomarkers detected in both studies.
For some members of the TCGA team, that will involve follow-up studies focused on lower-grade gliomas in combination with glioblastomas, in the hopes of uncovering additional signaling pathways and molecular features within larger and larger tumor datasets.
In a NEJM editorial, St. Jude Children's Research Hospital pathologist David Ellison wrote that "both studies can justifiably claim that molecular classification captures the biologic features of glioma variants better than does the histopathological evaluation, even though grade remains an independent prognostic indicator."
Ellison added that the new data sets "have the potential to inform how we define and treat the range of adult gliomas at a time when the current edition of the [World Health Organization] classification of nervous system tumors is being revised to include, for the first time, molecular information on the classification of disease."