The MGMT gene encodes a DNA repair protein that removes alkyl groups from the O6 position of guanine, an important site of DNA alkylation [1]. Chemotherapy-induced alkylation at this site triggers cytotoxicity and apoptosis. Tumor cells that express high levels of the MGMT repair protein may thereby counteract the therapeutic effect of alkylating agents, including nitrosourea compounds and temozolomide that are most commonly used for the treatment of malignant gliomas. MGMT is epigenetically inactivated via hypermethylation of the 5′-CpG island in approximately 40% of primary glioblastomas and over 70% of secondary glioblastomas (Fig. 1) [2]. MGMT promoter methylation is also found in half of the diffuse and anaplastic astrocytomas as well as approximately two thirds of the oligodendroglial and mixed tumors [2]. CpG islands are defined as genomic regions that contain a higher than average frequency of CG dinucleotides (CpG sites) and are involved in regulation of gene transcription. CpG islands, including the one associated with the MGMT gene, often span the transcription start site of genes and contain crucial transcription factor binding sites. Aberrant methylation of CpG islands may impair gene transcription, which consequently leads to reduced or even complete loss of expression of the respective gene product [3, 4]. In case of the MGMT 5′-CpG island, a region covering an MGMT enhancer element appears to be most critical for loss of transcription and MGMT protein expression, as suggested by luciferase reporter assays [5, 6]. In malignant gliomas, however, the MGMT promoter methylation patterns are highly heterogeneous from tumor to tumor and it is unknown which particular CpG sites or combinations thereof need to be methylated for silencing the gene and conveying benefit from alkyating agent therapy. The various assays that are in current use to evaluate MGMT status assess different numbers of CpGs at distinct locations within the MGMT promoter, typically covering between 3 and 20 of a total of 97 CpGs. Although in most instances results are expected to overlap, one has to be aware of the fact that different laboratories may obtain different results in individual patients. Such differences in test results do not need to be due to improper testing in one or the other laboratory, but may reflect a heterogeneous methylation pattern in the investigated tumor.

Fig. 1
figure 1

Association of favorable prognostic factors with MGMT methylation in malignant glioma. The incidences of MGMT methylation, IDH1 or IDH2 mutations, and co-deletions of 1p/19 are displayed for different types of malignant gliomas in adults. MGMT methylation is associated with IDH1 or IDH/2 mutations in all types of malignant gliomas, with the exception of primary glioblastoma [2, 17, 27–29]

Full size image

Methodological aspects of MGMT promoter methylation testing

The most commonly employed method, and also the technique originally described to convey the relevant clinical information, is methylation-specific PCR (MSP) analysis [7, 8]. This technique employs primers that specifically amplify fragments from either the methylated or the unmethylated sodium bisulfite-modified DNA sequence. In order to make the primers discriminative between both sequences they are designed to contain a maximum number of CpG sites that differ between methylated and unmethylated bisulfite-modified DNA. PCR products can be evaluated by gel-based approaches or quantitatively using real-time PCR assays that define a threshold for detecting methylation levels and thus permit a higher degree of standardization [9]. Such cut-offs, however, are just technically substantiated to date, and there is a need to validate cut-off points prospectively to establish clinically relevant methylation thresholds. A major advantage of MGMT methylation testing compared to testing at the gene or protein expression levels is that any methylation signal detected is representative of neoplastic glial cells only. RT-PCR analysis, western blotting or immunohistochemistry do not only detect MGMT expression from neoplastic glial cells but also from contaminating non-neoplastic cell populations, which can be very prominent in malignant gliomas and include microglia/macrophages, lymphocytes, reactive astrocytes, residual brain parenchymal elements and vascular cells. Without doubt, MGMT immunohistochemistry would clearly facilitate the diagnostic procedure, however, due to the described methodical limitations it does not uniformly reflect a tumor’s MGMT methylation status and conveys a degree of uncertainty that is not acceptable for clinical decision-making [10, 11]. In fact, even when the extent of microglial contamination was considered in the evaluation of MGMT immunostaining, there was still no significant association of the immunohistochemical results with MGMT promoter methylation and patient survival [10]. Thus, either specificity and sensitivity of the antibodies used for immunohistochemical detection of MGMT is limited or other, yet to be identified molecular mechanisms contribute to the regulation of MGMT protein expression in glioblastomas in addition to promoter methylation [10]. Further research is needed to dissect possible reasons that may account for the lack of association of MGMT immunoexpression with MGMT promoter methylation and patient survival in glioblastomas.

Clinical relevance of the MGMT promoter methylation status in gliomas

As mentioned above, the main use of MGMT as a molecular marker is its predictive value regarding the response of malignant gliomas to alkylating chemotherapy using either nitrosourea compounds [12], temozolomide [8], or a combination of both [13]. In the European Organisation for Research and Treatment of Cancer (EORTC)/National Cancer Institute of Canada (NCIC) 22981/26981 trial [8, 14], patients treated with radiotherapy and temozolomide survived significantly longer when they had a methylated MGMT promoter [8]. While data from this EORTC/NCIC 22981/26981 trial and another large prospective patient cohort [15] found that MGMT promoter methylation was predictive for longer survival only in those patients who received temozolomide, a recent retrospective and single institution analysis reported that MGMT promoter methylation may also be predictive of response to radiotherapy and linked to longer survival in the absence of adjuvant chemotherapy in glioblastoma patients [16]. While the prognostic role of MGMT in glioblastoma patients not treated with chemotherapy is a matter of debate, recent data from anaplastic glioma trials, namely the NOA-04 and the EORTC 26951 trials, both found that MGMT promoter methylation predicted prolonged survival irrespective of the initial treatment, i.e., radiotherapy, chemotherapy or a combination of both [17, 18]. The prognostic role of MGMT promoter methylation in patients with diffusely infiltrating low-grade gliomas who do not receive adjuvant therapy is unclear. In a study of 49 variably treated patients with low grade gliomas, MGMT promoter methylation was reported to be a negative prognostic factor for progression-free survival [19]. In contrast, treatment with temozolomide in a phase II study of low-grade glioma patients reported a better outcome in patients with MGMT promoter-methylated tumors [20].

Coincidence of MGMT promoter methylation with other molecular alterations

The reason for the predictive role of MGMT promoter methylation for response to temozolomide and other alkylating agents can be easily derived from the functional aspects of the MGMT repair protein (see above). In contrast, the prognostic role of MGMT promoter methylation in patients with anaplastic gliomas receiving only radiotherapy is rather unexpected from a functional point of view. However, in contrast to primary glioblastomas, MGMT promoter methylation in diffuse gliomas and secondary glioblastomas is frequently associated with other prognostically favorable genetic alterations (Fig. 1). In oligodendroglial neoplasms, a strong association of MGMT hypermethylation is observed with 1p/19q codeletions [21–23] and IDH1 mutations [24]. In diffusely infiltrating astrocytic gliomas—with the exception of primary glioblastoma—MGMT promoter hypermethylation and IDH1 mutations frequently coincide [24]. Further studies are needed to more closely dissect which of these changes contribute most to the positive prognostic effect in anaplastic gliomas or if even the particular constellation of MGMT methylation with concomitant 1p/19q codeletions and IDH1 mutations is associated with a higher sensitivity to cytotoxic therapy and a more favorable outcome. The situation may be more complex, though, than suggested by merely focusing on a single molecular marker such as MGMT. Noushmehr et al. recently described a glioma CpG island methylator phenotype (G-CIMP) that identified a set of prognostically favorable gliomas showing promoter hypermethylation of a specific set of genes [25]. This G-CIMP correlated to a similarly favorable proneural gene signature described in a preceding study by Phillips et al. [26] and was tightly associated with IDH1 mutation. Interestingly, this G-CIMP was more prevalent among diffuse low-grade and anaplastic gliomas, in particular oligodendroglial tumors. While MGMT promoter methylation was not identified as being part of this hypermethylator phenotype, the exact association of MGMT methylation with this prognostically favorable signature has not yet been tested. However, the study suggests that particularly in anaplastic and low-grade gliomas prognostically favorable effects may be brought about not only by MGMT but by the aberrant methylation of other genes, possibly including yet to be identified candidate genes whose silencing may contribute to increased radiosensitivity of glioma cells. These observations further emphasize a different pathogenesis of diffuse gliomas and secondary glioblastoma as opposed to the majority of primary glioblastoma.

Summary and outlook

MGMT promoter methylation has been established as a clinically important molecular marker in neuro-oncology. While treatment decisions in the routine setting are not yet based on this marker, the MGMT promoter methylation status is now used as an important stratification or selection parameter in ongoing clinical trials. In glioblastoma patients, MGMT promoter methylation is predictive for the response to alkylating chemotherapy and associated with longer survival in patients treated with radiotherapy as well as concurrent and adjuvant temozolomide. Recent data from trials on anaplastic glioma patients indicate that MGMT promoter methylation in this tumor group is a favorable prognostic marker that is independent from the type of therapy, i.e., radiotherapy or alkylating chemotherapy. The high coincidence of MGMT methylation with other genetic aberrations, such as 1p/19q deletions and particularly IDH1 or IDH2 mutations, and potentially epigenetic silencing of yet unknown radioresistance genes may indicate a more complex molecular phenotype of clinical significance and has to be further dissected in subsequent studies.