Biology Top Ten: For Glioblastoma, Genetic Signatures Help to Optimize Treatment

Glioblastoma multiforme (GBM) is the most common form of brain tumor in adults, responsible for more than half of all brain tumors. Prognosis is generally very poor, with survival measured in months rather than years. As a result, brain tumors are probably the most feared of all cancers. And as a further result, many recent studies have focused on understanding the molecular basis of GBM.

A breakthrough paper, published in January 2010, clearly identified four different subtypes of GBM, each with distinct molecular characteristics. That paper, from a group led by D. Neil Hayes, of the Broad Institute in Cambridge, Massachusetts, and the University of North Carolina at Chapel Hill, first came to ScienceWatch’s attention in March 2011, when its citation rate was increasing more rapidly than all other recent reports being indexed in the Clarivate Analytics main category of Biology & Biochemistry.


Biology Top Ten
Katherine Hoadley and Hayes told ScienceWatch at the time that “the opportunity to define a distinct subset of GBMs promises to open new avenues in understanding disease progression and treatment.” Early interest in the results has continued to build, with the result that the paper is currently at #3 in the latest Biology Top Ten (see attached table).

Looking overall at 1,740 genes associated with GBM, the team found evidence that patients reliably fell into only one of four clusters.

Hayes and his co-authors made use of data compiled by The Cancer Genome Atlas (TCGA []), which was established by the National Institutes of Health to catalog genetic abnormalities associated with a chosen set of cancers, among them GBM. The data confirmed mutations in TP53 and RB1 and identified others in many more genes. Looking overall at 1,740 genes associated with GBM, the team found evidence that patients reliably fell into only one of four clusters. From this, it was possible to derive a subset of genes associated with each class that could be considered a genetic signature of the GBM subtype.


As is common with broad genetic analyses of this type, the team next validated the genetic signatures with independent datasets, which recapitulated both the four subtypes and the overall frequencies of each subtype among the dataset populations. Based on genetic signature and previously assigned names, the researchers called the subtypes Proneural, Neural, Classical, and Mesenchymal, and set out to investigate their similarities and differences.

The Classical subtype, for example, always had an amplification in chromosome 7 and a deletion in chromosome 10, associated with a large increase in the expression of EGFR (epidermal growth factor receptor). By contrast, TP53, which is generally mutated in GBM, is never mutated in the Classical subtype, but is mutated in 54% of the Proneural subtype. Each of the subtypes had its characteristic genetic peculiarities, which the team used to infer the biological meaning of the categories by looking at gene-expression profiles for four different types of brain cell: neurons, and three kinds of support cells—oligodendrocytes, astrocytes, and astroglial cells. Mesenchymal GBM shared a profile with astroglia, Classical with astrocytes, Proneural with oligodendrocytes, and Neural with neurons. Of the four, Proneural subtypes were more common in younger patients.


It had been thought that perhaps GBM arose from mutations in a common and unique precursor neural stem cell, and that the different subtypes might then be caused by different pathways of differentiation. However there are also other types of stem-like precursor cells in the brain, and these do share some markers that are affected in the different GBM subtypes. This suggests that the different subtypes reflect different origins in separate cell lines that have already differentiated to some extent, although further conclusive evidence is still needed. The study also found no evidence that subtype changed in recurrence of the disease, which is an unfortunate characteristic of GBM, further indicating that each subtype of GBM is, effectively, a different disease.

All of which is doubtless fascinating, but does little to address the fundamental question: does the efficacy of treatment depend on GBM subtype?


Aggressive treatment—meaning concurrent chemotherapy and radiation or more than four cycles of chemotherapy—reduced mortality in Classical and Mesenchymal subtypes (with a suggestion that it had a smaller effect in Neural subtypes). However, the authors report that “it did not alter survival in the Proneural subtype.” For now, that means that biomarker tests for GBM will be expanded to include diagnostics for the subtypes, which can then be used to select the most appropriate therapy. In the longer term, enhanced understanding of metabolic pathways affected in the different subtypes could be used to devise more effective treatment regimes, which may go some way towards reducing the fear universally associated with brain tumors.

As Headley and Hayes told ScienceWatch, “Many of the genes we found altered happen to be genes that are drug targets.This strongly suggests that we may be able to selectively target each of these types of GBM with a different subset of drugs, in line with the idea of personalized medicine.”

Dr. Jeremy Cherfas is Senior Science Writer at Bioversity International, Rome, Italy.

(See also the ScienceWatch Special Topic on glioblastoma.)

< Back to July issue — main page