Epigenetic variations in the O6-methylguanine-methyltransferase (MGMT) gene had been widely associated with a favorable impact on survival in patients (pts) affected by glioblastoma (GBM). MGMT includes 98 CpG islands (CpGi) and patterns of methylation are rather heterogeneous. Aim of this study is to explore a scoring system based on the gene promoter methylation in order to predict pts' prognosis.
The study analyzed a series of 121 pts with GBM treated at the University Hospital of Udine between 2008 and 2014. The methylation level of CpGi from 74 to 83 was analyzed through pyrosequencing. In accordance to previous literature, each island was assigned with 1 point if the corresponding methylation level was higher than 9%. The sum consisted in a score that went from 0 (all CpGi = 9%). A training set of 75 pts was randomly generated. A threshold capable to detect a favorable outcome (OS > 24 months) was identified by ROC analysis. The prognostic impact was explored through Cox regression. The results were verified on a validation set of 46 pts.
Median OS was 14 months. Among the total population 35% of the pts had a score of 0, while 29% had a score of 10. The score's prognostic impact was confirmed also by comparison with the methylation mean and median through stepwise Cox regression (P= 0.0002). The threshold identified was 6 (AUC 0.74). On univariate analysis, a score > 6 was associated with a favorable prognosis both in the training and in the validation set (HR 0.42, 95%CI 0.23-0.77, P= 0.0046; HR 0.37, 95%CI 0.18-0.77, P = 0.0078; respectively). The result was maintained also in multivariate analysis of the whole population (HR 0.43, 95%CI 0.27-0.67, P = 0.0002) when corrected for age (>70 vs ≤ 70 years HR 2.19, 95%CI 1.30-3.69, P = 0.0032) and ECOG performance status (0-1 vs 2-3 HR 2.20, 95%CI 1.36-3.54, P = 0.0012). Similar results were observed also in terms of PFS.
The present study explored a novel scoring system capable to take into consideration the methylation status of single CpGi. Since the limited prognostic significance of each CpGi, combining the information from multiple CpGi is crucial in order to better predict prognosis in GBM patients.
Clinical trial identification
Legal entity responsible for the study
University Hospital of Udine
All authors have declared no conflicts of interest.