Posted: January 4, 2012
Researchers Corroborate the TCGA Subtype Classification of Glioblastomas Using the Rembrandt Glioma Dataset
Pritty Patel Joshi
Gliomas are primary tumors that originate from glial cells, also known as the ‘supportive’ cells of the brain. These tumors can be classified clinically based on their microscopic appearance as well as their grade, or degree of malignancy, with I being the lowest grade and IV being the highest.
Glioblastoma multiforme (GBM) is a grade IV tumor and the most common primary brain tumor. Patients with GBM face a dismal prognosis with a median survival of approximately one year. While individuals diagnosed with low grade gliomas often experience longer survival times of 3-8 years, these tumors are also fatal as they typically progress to GBM.
Using a tumor classification scheme developed by The Cancer Genome Atlas (TCGA), investigators at Emory University offer insights into the progression of this debilitating disease. Their findings, published September 2010 in PLoS ONE, establish the predictive value of TCGA GBM gene expression signatures with implications for targeted treatments.
The TCGA Research Network previously used the GBM gene expression datasets to identify four distinct molecular subtypes. In this work, classification of tumors into the Proneural, Neural, Classic, and Mesenchymal subtypes provided evidence to support the need for targeted therapeutics. However, subtype classification did not predict overall patient survival. Carlos Moreno, Ph.D. and his colleagues believe they know why. They reasoned that the short survival period of GBM hinders efforts to identify gene signatures predictive of patient outcome. Instead, they argue, in order to gain insight into survival, lower-grade gliomas should be subjected to similar classification procedures and subsequently assessed for clinical outcome.
The Proneural Subtype is Enriched in Certain Low Grade Gliomas and Predictive of Survival Outcome
Since TCGA does not currently include low grade gliomas in their analyses, the researchers turned to an alternate publicly available brain tumor dataset, the Repository for Molecular Brain Neoplasia Data (Rembrandt), which, like TCGA, is managed in part by the National Cancer Institute. Rembrandt contains clinical and molecular data from grades II, III, and IV gliomas.
The scientists used multiple bioinformatics tools to analyze the Rembrandt dataset. They found that classification of these data using the TCGA scheme readily unveiled the four subtypes previously identified by the TCGA Research Network.
Analyses of survival outcomes in gliomas of mixed grades showed that tumors classified as Proneural had significantly better outcome. Additionally, the Proneural gene signature was enriched in a specific type of grade II and III gliomas, oligodendrogliomas, and predicted improved survival in this tumor type. The other three subtypes showed no discernable differences in survival.
To better understand the progression of low grade gliomas to GBM, the investigators focused their subsequent analyses on the Proneural subtype. They examined the gene expression and copy number changes in Rembrandt Proneural tumors. The scientists identified a few hundred genes that exhibit expression or copy number changes between low and high grade gliomas implicating pathways such as Notch and Hedgehog in the progression of Proneural tumors. According to the authors, these findings suggest that the molecular subtypes identified by TCGA signify intrinsic differences in tumor biology.
In Silico Analyses Demonstrate the Significance of TCGA Datasets
The researchers in this study aimed to determine the significance of the TCGA gene expression signatures by applying them to a different glioma dataset. They provide evidence to show that the gene expression patterns developed by TCGA have prognostic value for low grade gliomas. Their results lay the framework for the development of targeted therapies that could suppress low grade Proneural tumors prior to their progression to the fatal GBM disease.
These analyses illustrate the broad applications of TCGA findings. Overall, the scientists stress, this study exemplifies the magnitude of publicly available datasets and presents promising implications for future discoveries from TCGA.
Cooper, L.A., Gutman, D.A., Long, Q., Johnson, B.A., Cholleti, S.R., Kurc, T., Saltz, J.H., Brat, D.J. and Moreno, C.S. (2010) The Proneural Molecular Signature is Enriched in Oligodendrogliomas and Predicts Improved Survival Among Diffuse Gliomas. PLoS One. 5(9):e12548. Read the full article.