Prediction of GBM outcome using combined analysis of MGMT protein expression and promoter methylation.

Abstract

2003 Background: The standard of care for patients with newly-diagnosed glioblastoma (GBM) consists of postoperative radiotherapy and temozolomide (TMZ). O6-methylguanine-DNA methyltransferase (MGMT) provides resistance to TMZ, and MGMT promoter methylation correlates with improved treatment outcome presumably through silencing of MGMT protein expression. However, the prognostic significance of MGMT protein expression as assessed by immunohistochemistry (IHC) and its relationship with methylation remain unclear. Methods: We identified 418 newly-diagnosed GBM patients treated at UCLA/Kaiser with available pre-treatment FFPE tissue. We analyzed MGMT expression by IHC and MGMT promoter methylation by both methylation specific PCR (MSP) and bisulfite sequencing (BiSEQ) of 24 contiguous CpG sites containing the MSP region. Classification and regression tree analysis was performed on IHC and BiSEQ data to determine thresholds most predictive of outcome. Survival analyses were performed to assess the predictive value of these MGMT biomarkers separately and in combination. Results: MGMT expression alone was predictive of progression-free survival (PFS) and overall survival (OS) for GBM patients treated with TMZ. Patients with <30% staining on IHC had PFS*=11.3 mo and OS*=21.7 mo compared with PFS=7.3 mo and OS=16.7 mo for patients with ≥30% staining (*p<0.0001). We confirmed that MSP was predictive of outcome in our cohort but found that BiSEQ was superior to MSP for assessment of MGMT promoter methylation. Patients with ≥9 methylated CpG sites showed PFS*=13.3 mo and OS*=29.3 mo compared with PFS=8.1 mo and OS=16.1 mo for patients with <9 methylated CpG sites (*p<0.0001). Lastly, we observed that combined assessment of MGMT IHC and methylation provided improved predictive value (patients with low IHC and methylation had PFS=14.3 mo and OS=32.2 mo). Conclusions: MGMT IHC is a predictive biomarker for GBM patients treated with TMZ, and BiSEQ appears superior to MSP for methylation determination. Combined assessment of protein expression and methylation has better predictive value than either biomarker alone, and outcome for patients with low MGMT expression is dependent on MGMT methylation status.