Dialog Box


Using big data to understand low grade brain tumours

Roel Verhaak Ph.D, MD Anderson Cancer Centre


By doing a comprehensive genomic analysis of low grade brain tumours, researchers at The University of Texas MD Anderson Cancer Center have established three molecular categories which low grade tumours fall into. Importantly, they have found that one category bears molecular resemblance to the most lethal brain tumour, glioblastoma multiforme. 


“The immediate clinical implication is that a group of patients with tumors previously categorized as lower grade should actually be treated as glioblastoma patients and receive that standard of care - temozolomide chemotherapy and irradiation” 

- Roel Verhaak, Ph.D, Lead Author 


Verhaak and the team analysed data from The Cancer Genome Atlas (TCGA), which compiles data from around the world on cancer genomics and makes it available to researchers to help them make new discoveries. Big data projects like TCGA are great examples of what Cure Brain Cancer believe will be necessary to make significant inroads in brain cancer research. By collecting more patient samples in a standardised way the data can be compared and analysed on a larger scale, leading to new discoveries.


The researchers analysed 254 lower-grade gliomas for gene, protein and micro RNA expression, DNA methylation and gene copy profiles. They then clustered the cases by category. Then they conducted a “cluster of clusters” analysis that brought together all the data. Their research findings have been reported at the American Association for Cancer Research Annual Meeting 2014.


The three groups they’ve identified are defined by mutational status of the IDH1 and IDH2 genes and loss of chromosome arms 1p and 19q. The researchers note that these biomarkers are already routinely checked in clinical care so it would be easy to implement the new categories.


Each molecular group includes a range of tumours, including astrocytoma, oligodendrocytoma and oligo-astrocytoma.


The three molecular clusters have either:


  • Wild-type IDHI1 and IDH2 with no mutations. (Similar to glioblastoma, with a median survival of 18 months.)
  • IDH1/IDH2 mutations and chromosome arms 1p/19q intact. (No dominant histology or grade type, median survival about seven years.)
  • IDH1/IDH2 mutations and co-deletion of 1p/19q. (Mainly oligodendrogliomas, median survival about eight years.)


Verhaak says classifying low grade tumours in these three groups is a more accurate way of doing so than the current methods to group and grade tumours. 


Previous research involving The Cancer Genome Atlas shows that glioblastoma patients with mutations of IDH1/IDH2 have an improved prognosis. Verhaak says we don’t yet know whether these mutations are markers of good prognosis or actually have a role in slowing the progression of the tumour. It has previously been shown that co-deletion of 1p19q leads to increased tumour sensitivity to chemotherapy and longer survival for oligodendroglioma patients.


This study illustrates the importance of both genomic profiling and bioinformatics to brain cancer research. Big data projects like TGCA give researchers access to a larger amount of information and therefore a more complete picture of brain cancer. Big data is critical for brain cancer research as it looks ever more likely that genetics will play a huge role in the disease, and because it will generate access to large volumes of both brain tumour and control data, which is vital in research into a low-incidence cancer. 


Sign up for the latest news and updates