Molecular Stratification Assay for Glioblastoma using Isoform-Level Gene Signatures

Each year in the United States 18,000 people are diagnosed with Glioblastoma multiforme (GBM) (1). Of those diagnosed, there is a 3% five-year survival rating as compared to breast cancer that has an 89% five-year survival rating (2). Glioblastoma prognosis is dependent on several factors that directly influence course of treatment. These factors are age of diagnosis, location, and the genetic and molecular profile of the tumor (3).

However, GBM has fallen victim to the limitations of current tumor subtyping using gene-level expression profiling. For GBM, an aggressive brain cancer with a median survival of 15 months post-diagnosis, this has led to inaccurate classification with weak prognostic value. Importantly, GBMs are heterogeneous in molecular nature adding greater complication to both diagnosis and treatment response. Our strategy is to produce molecular subtyping with improved prognosis significance for the realization of personalized medicine and successful patient outcomes for GBM.

Wistar Institute researchers have produced a molecular stratification system using the expression of transcript variants (isoform-level gene expression) as a prognostic and diagnostic tool for GBM (4). This technology holds promise as a robust diagnostic assay for patient stratification and prognosis assignment in a number of cancers. In addition to prognostic value, the transcript/isoform-based classifier has identified novel targets for pharmacotherapies and tailored treatments for GBM.

Genome-wide cellular studies suggest the majority of human genes produce transcript variants and protein isoforms that are functionally discrete. Moreover, altered expression of specific isoforms for numerous genes is linked to cancer and its prognosis (5). Cancer associated alterations in alternative exons and splicing machinery have been identified in breast, lung, brain, and several other cancers (6-8). Even with this knowledge, GBM profiling techniques have been limited to gene-level expression analysis and isoform profiling has been largely overlooked.

The research team at Wistar classified expression levels of gene isoforms in GBM samples from the Cancer Genome Atlas consortium and identified transcript-based molecular signatures for GBM characterization that cluster into four subtypes. These subtypes are combinations of unique transcripts discretely dysregulated in GBM tumor clusters. Critically, the markers identified from a high-throughput RNA-Seq/exon-array were translated to a clinically relevant and widely accessible diagnostic platform (RT-qPCR) without the loss of analytical precision determined through validation using over 200 unique GBM samples obtained from the Hospital of the University of Pennsylvania. Specifically, RT-qPCR is performed on 121 distinct transcripts that are used to stratify GBM patients with 93.6% accuracy.

A novel computational approach was employed to derive GBM subtype gene-isoform signatures and then translated to a clinically relevant diagnostic platform. The use of isoform-level expression profiling is a breakthrough in such a diagnostic used to accurately identify GBM subtypes with greater prognostic significance predicted between the subtypes. In addition, the isoform-level analysis is able to produce substantially better characterization predictions for GBM classification with greater accuracy using fewer variables as compared to a gene-based classifier.

Pending patent applications covering the assay and methods of use. PCT/US2014/32966.

The Wistar Institute has exclusively licensed this technology to ISOMA Diagnostics, LLC.

1. Schwartzbaum et al. Nat Clin Pract Neurol. 2006 Sep;2(9):494-503 PMID: 16932614
   
   
2. http://www.iarc.fr/en/publications/pdfs-online/wcr/2003/WorldCancerReport.pdf
   
   
3. Walid. Perm J. 2008 Fall;12(4):45-8. PMID: 21339920
   
   
4. Pal et al. Nucleic Acids Res. 2014 Apr;42(8) PMID: 24503249
   
   
5. Pal et al. Pharmacol Ther. 2012 Dec;136(3):283-94. PMID: 22909788
   
   
6. Ebert & Bernard. N Engl J Med. 2011 Dec 29;365(26):2534-5. PMID: 22150007
   
   
7. Lapuk et al. Mol Cancer Res. 2010 Jul;8(7):961-74. PMID: 20605923
   
   
8. Misquitta-Ali et al. Mol Cell Biol. 2011 Jan;31(1):138-50. PMID: 21041478