• Head & Neck Cancers Genomics Clinical Trials

• Head & Neck Cancers
• Oncology Genomics
• Oncology Clinical Trials
• Oncology

The results of a correlative study in the Phase 3 Temozolomide and Radiation Therapy With or Without Bevacizumab in Treating Patients With Newly Diagnosed Glioblastoma trial [Radiation Therapy Oncology Group (RTOG) 0825; NCT00884741; Sulman EP et al. J Clin Oncol 2013 (suppl; abstr LBA2010)] examining the molecular predictors of outcome and response to bevacizumab (BEV) added to standard chemoradiation for patients with newly diagnosed gliobastoma were discussed by Erik P. Sulman, MD, PhD, University of Texas MD Anderson Cancer Center, Houston, Texas, USA.

The study aimed to identify patients likely to respond to BEV during initial treatment for glioblastoma using a gene biomarker detectable in formalin-fixed, paraffin-embedded (FFPE) tissue.

The study focused on the mesenchymal signature, a set of genes upregulated in glioblastomas that are associated with invasive, angiogenesis functions and poor patient survival [Colman H et al. Neuro Oncol 2010]. BEV was hypothesized to beneficially affect patients (ie, prolonged overall survival [OS] and progression-free survival [PFS]) whose tumors exhibit the mesenchymal gene signature. A multigene signature that approximates mesenchymal enrichment was used for patient stratification in the trial. This multigene signature did predict modest improvement in PFS and OS in newly diagnosed glioblastoma patients treated with BEV compared with patients treated with a standard regimen. However, patients with more mesenchymal tumors did worse, which was the opposite of what was anticipated.

This predictive response was consistent with a whole genome transcriptome analysis in a subset of 114 cases, which detected a subgroup of mesenchymal genes expressing tumors that, when compared with other molecular subtypes, correlated with worse survival of patients treated with BEV.

Because of the survival differences observed in mesenchymal-expressing tumors, real-time polymerase chain reaction assays of a set of mesenchymal genes that were validated for use with FFPE tissue were used to build a predictive model with an optimal set of genes in a training/validation approach. A subset of 234 patients (out of a total of 637 randomized in the trial) was used for the analysis. The resulting predictor of response to BEV for patients with newly diagnosed glioblastoma (PRoB-GBM) separated patients into BEV-responsive and unresponsive groups. In the training set, PRoB-GBM predicted BEV response for PFS (p<0.0001) and OS (p<0.0001). The validation set also showed favorable responders with a significant response for PFS (p=0.0385) and OS (p=0.0014). Patients in the control arm displayed no difference, indicating that the biomarker is predictive and not prognostic.

In the subgroup of 234 patients, use of PRoB-GBM predicted BEV responders. For PFS, the target group of patients predicted to benefit from BEV treatment did display enhanced PFS (13.2 months) as compared with 7.2 months in the other patient group (p<0.0001). Similarly, OS was enhanced in the predicted target group (20.3 months) versus 10.4 months in the other group (p<0.0001).

The PRoB-GBM unfavorable group (ie, predicted to respond poorly to BEV) correlated with both the unfavorable (ie, more mesenchymal) multigene assay used to stratify patients in the trial as well as to the mesenchymal class identified by transcriptome analysis, suggesting that the predictive responder group was low in mesenchymal gene expression. In multivariate analyses, PRoB-GBM showed a strong interaction with treatment arm (placebo vs BEV) and, within the BEV arm only, was predictive of OS independent of prognostic factors including methylated or unmethylated MGMT gene, and RTOG Recursive Partitioning Analysis of Glioma class.

Within the trial, patients who experienced tumor recurrence (including those in the control arm) could be given salvage treatment with BEV. In these patients, PRoB-GBM was not predictive for salvage treatment, indicating that the PRoB-GBM biomarker may be useful only in the newly diagnosed setting.

Thus, the developed biomarkers represent a molecular diagnostic tool that can be used to identify patients for BEV treatment for newly diagnosed glioblastoma using FFPE tissue.

• © 2013 MD Conference Express®