Defining the prognostic role of MGMT methylation value by pyrosequencing assay in glioblastoma patients: A large Italian multicenter study.

Abstract

2539 Background: MGMT methylation (MGMTmet) status represents an important prognostic factor for glioblastoma (GBM) patients (PTS). Quantitative pyrosequencing approach has proven to be feasible for MGMTmet testing but its value is still unclear. We performed a large, multicentre, retrospective study to identify the association between MGMTmet values and clinical outcome. Methods: from 9 Italian neuro-oncology centres, we collected consecutive GBM PTS with assessment of MGMTmet by pyrosequencing approach evaluating CpG islands from 75 to 84. Other inclusion criteria were: histological diagnosis of GBM, ECOG PS ≤2, therapy with RT+TMZ. Kaplan-Meier method was used to estimate the survival curves, time-dependent ROC curve for defining the optimal cut-off value of mean percentage of MGMTmet in terms of 2y-OS, Cox regression for multivariable analysis, and restricted cubic spline to investigate the non-linear association between methylation values and OS. Results: 681 PTS were enrolled; median age was 60 ys; ECOG PS was 0 in 292 PTS, 1 in 306 PTS, 2 in 83 PTS; 391 PTS (58%) had a complete resection. 8% of PTS received a second surgery. IDH was mutated in 6%. 2y-OS was 31.6%, median OS was 17.4 ms. Median MGMTmet was 3.5% (IQR 0-22%). ROC curve identified a cutoff of 15% of MGMTmet in terms of 2y-OS (sens 78%, spec 57%, AUC = 0.67). 2y-OS was 19.7% and 53.7% for PTS with MGMTmet < and ≥15%, respectively (p < 0.0001). At multivariable analysis, MGMTmet < 15% was associated with impaired survival (HR 2.7, 95% CI 2.1-3.4; p < 0.00001), adjusting for age, KPS, type of surgery and second surgery. A non-linear association between MGMT methylation and survival was identified (non-linear term: p < 0.0001), with lower values of MGMT methylation associated with lower survival; indeed, estimated median OS was lowest (14 months, 2ys-OS: 17.4%) with MGMTmet of 4%, 21ms (2yr-OS: 40.9%) with MGMTmet of 20%, 27ms (2yr-OS: 40.9%) when MGMTmet was 40%, then leveled around 30ms (2yr-OS: 54.5-59.8%) when MGMTmet was > 40%. Conclusions: this study represents one of the largest trials analyzing MGMTmet by pyrosequencing approach. Lower values of MGMTmet were associated with impaired survival and the relationship was non-linear. Noteworthy, we identified a strong prognostic value of MGMTmet which could be used as stratification factor in prospective clinical trials