Combining MGMT promoter pyrosequencing and protein expression to optimize prognosis stratification in glioblastoma.

Pyrosequencing (PSQ) represents the golden standard for MGMT promoter status determination. Binary interpretation of results based on the threshold from the average of several CpGs tested would neglect the existence of the "gray zone". How to define the gray zone and reclassify patients in this subgroup remains to be elucidated. A consecutive cohort of 312 primary glioblastoma patients were enrolled. CpGs 74-81 in the promoter region of MGMT were tested by PSQ and the protein expression was assessed by immunohistochemistry (IHC). Receiver operating characteristic curves were constructed to calculate the area under the curves (AUC). Kaplan-Meier plots were used to estimate the survival rate of patients compared by the log-rank test. The optimal threshold of each individual CpG differed from 5% to 11%. Patients could be separated into the hypomethylated subgroup (all CpGs tested below the corresponding optimal thresholds, n = 126, 40.4%), hypermethylated subgroup (all CpGs tested above the corresponding optimal thresholds, n = 108, 34.6%), and the gray zone subgroup (remaining patients, n = 78, 25.0%). Patients in the gray zone harbored an intermediate prognosis. The IHC score instead of the average methylation levels could successfully predict the prognosis for the gray zone (AUC for overall survival, 0.653 and 0.519, respectively). Combining PSQ and IHC significantly improved the efficiency of survival prediction (AUC: 0.662, 0.648, and 0.720 for PSQ, IHC, and combined, respectively). Immunohistochemistry is a robust method to predict prognosis for patients in the gray zone defined by PSQ. Combining PSQ and IHC could significantly improve the predictive ability for clinical outcomes.