Prognostic prediction of glioblastoma by quantitative assessment of the methylation status of the entire MGMT promoter region
1 Research Institute, Osaka Medical Center for Cancer and Cardiovascular Diseases, 1-3-3 Nakamichi, Higashinari-ku, Osaka, Japan
2 Department of Neurosurgery, Kyoto University Graduate School of Medicine, 54 Kawahara-cho, Shogoin, Sakyo-ku, Kyoto-shi, Kyoto 606-8507, Japan
3 Department of Neuro-Oncology/Neurosurgery, Saitama Medical University International Medical Center, 1397-1 Yamane, Hidaka, Saitama 350-1298, Japan
BMC Cancer 2014, 14:641 doi:10.1186/1471-2407-14-641Published: 30 August 2014
O6-methylguanine-DNA methyltransferase (MGMT) promoter methylation is reported to be a prognostic and predictive factor of alkylating chemotherapy for glioblastoma patients. Methylation specific PCR (MSP) has been most commonly used when the methylation status of MGMT is assessed. However, technical obstacles have hampered the implementation of MSP-based diagnostic tests. We quantitatively analyzed the methylation status of the entire MGMT promoter region and applied this information for prognostic prediction using sequencing technology.
Between 1998 and 2012, the genomic DNA of 85 tumor samples from newly diagnosed glioblastoma patients was subjected to bisulfite treatment and subdivided into a training set, consisting of fifty-three samples, and a test set, consisting of thirty-two samples. The training set was analyzed by deep Sanger sequencing with a sequencing coverage of up to 96 clones per sample. This analysis quantitatively revealed the degree of methylation of each cytidine phosphate guanosine (CpG) site. Based on these data, we constructed a prognostic prediction system for glioblastoma patients using a supervised learning method. We then validated this prediction system by deep sequencing with a next-generation sequencer using a test set of 32 samples.
The methylation status of the MGMT promoter was correlated with progression-free survival (PFS) in our patient population in the training set. The degree of correlation differed among the CpG sites. Using the data from the top twenty CpG sites, we constructed a prediction system for overall survival (OS) and PFS. The system successfully classified patients into good and poor prognosis groups in both the training set (OS, p = 0.0381; PFS, p = 0.00122) and the test set (OS, p = 0.0476; PFS, p = 0.0376). Conventional MSP could not predict the prognosis in either of our sets. (training set: OS; p = 0.993 PFS; p = 0.113, test set: OS; p = 0.326 PFS; p = 0.342).
The prognostic ability of our prediction system using sequencing data was better than that of methylation-specific PCR (MSP). Advances in sequencing technologies will make this approach a plausible option for diagnoses based on MGMT promotor methylation.