Intratumoral homogeneity of MGMT promoter hypermethylation as demonstrated in serial stereotactic specimens from anaplastic astrocytomas and glioblastomas



Hypermethylation of the DNA repair gene O6-methyl-guanine DNA methyltransferase (MGMT) has been linked to prolonged survival in glioblastoma patients treated with alkylating agents. It was aimed to analyze prospectively whether the MGMT status of malignant gliomas could be determined from small-sized stereotactic biopsies (maximum volume: 1 mm3). Special attention was directed towards the intratumoral distribution of the MGMT promoter methylation, the MGMT protein expression and potential correlations between both. Twenty-five adult patients were included (20 patients with primary World Health Organisation (WHO) Grade III or IV malignant gliomas, 5 patients with secondary malignant gliomas). About 2–4 biopsy specimens per tumor were collected from different sites within the tumor. Promoter methylation of the MGMT gene was assessed by methylation-specific PCR (MSP) and sodium bisulfite sequencing in each of the collected specimens (overall number of specimens: 69). Both methods were validated for application in small-sized tissue samples (1 mm3). The MGMT protein expression was analyzed by immunohistochemistry. The overall MGMT promoter methylation rate was 30% in the de novo group and 80% in the tumor progression group. The success rates of MSP and sequencing were 100% and 80%, respectively. Sequence analysis and MSP exhibited 100% concordant findings. No differences in MGMT promoter methylation were detected between the different samples of each individual tumor in 24 of 25 patients. One false negative result was obtained due to the contamination of the biopsy specimen by necrotic tissue. Tissue samples taken from different sites of each individual tumor (13 tumors investigated) exhibited equal or highly similar MGMT protein expression. No correlation between MGMT protein expression and MGMT promoter methylation was observed. The MGMT promoter methylation status of malignant gliomas can be reliably determined from small-sized stereotactic biopsies. The methylation profile, as defined by MSP and sodium bisulfite sequencing, constitutes a homogeneous marker throughout malignant gliomas. The lack of correlation between MGMT status and MGMT protein expression needs further evaluation. © 2007 Wiley-Liss, Inc.

MGMT is a ubiquitous DNA-repair enzyme in normal human tissues. It removes and accepts alkyl groups from the O6 position of methylguanine without affecting DNA integrity. MGMT exerts its reparative function in a stochiometric fashion, without requiring any additional cofactor and by finally inactivating itself. Overexpression of the MGMT gene in brain tumors confers resistance to alkylating chemotherapeutic agents such as temozolomide by removing chemotherapy-induced cytotoxic DNA lesions.1

Epigenetic silencing of the MGMT gene by promoter methylation has been shown to be an independent predictor of response to alkylating chemotherapy and prognosis in patients with newly diagnosed glioblastomas treated with radiotherapy as well as concomitant and adjuvant temozolomide.2 However, it is not known whether and to which extent MGMT promoter methylation varies throughout the same tumor. Findings of heterogeneity will possibly influence diagnostic methods, treatment strategies and outcome scores.

Stereotactic serial biopsy is considered a powerful and minimal invasive tool for obtaining tissue samples for histopathological analysis and is routinely performed in patients with glioblastomas not suitable for open gross total tumor resection. The highly controlled sampling procedure in combination with intraoperative evaluation of smear preparations allows collecting solid tumor samples from different sites of the tumor, which are not or only minimally contaminated with tumor necrosis or non-neoplastic cells. Thus, the combination of stereotactic and molecular genetic techniques might be an ideal approach for analyzing the intratumoral profile of MGMT methylation in extremely heterogeneous tumors such as glioblastomas. Moreover, it would provide important data for treatment planning and overall prognosis in those patients not suitable for open tumor resection. However, difficulties in determining the methylation status in small-sized tissue samples from biopsies could limit this approach. The aim of the present prospective study was 3-fold, i.e., (i) to establish and validate MSP and sequencing analysis for frozen, small-sized tumor biopsy specimens (maximum volume: 1 mm3) by blinded re-evaluation of 10 brain tumor samples (obtained by open tumor resection) with a known MGMT promoter profile; (ii) to analyze the frequency of MGMT promoter methylation in frozen stereotactic biopsy samples from patients with untreated primary or secondary malignant gliomas and to check for correlations between the promoter methylation status and the actual MGMT protein expression as defined by immunohistochemistry; (iii) to investigate the intratumoral distribution of both MGMT promoter methylation and MGMT protein expression of each patient.

Patients and methods


Adult patients with an untreated primary or a secondary supratentorial anaplastic astrocytoma (WHO Grade III) or glioblastoma (WHO Grade IV) were included; histopathological examination of stereotactic biopsy specimens was performed in all patients. Reasons to decide for stereotactic serial biopsy were (i) an eloquent tumor location not allowing gross total tumor resection, (ii) a significant comorbidity and/or (iii) uncertainties concerning the diagnosis of a malignant glioma. The aim was to include at least 20 consecutively treated patients during a time period of 3 months. The study was conducted in the framework of the German Glioma Network. All enrolled patients gave informed consent, and the study protocol was reviewed and approved by the institutional review board of the Ludwig-Maximilians-Universität (AZ 216/14), Klinikum Grosshadern, Munich, Germany.

Stereotactic biopsy technique

The technique of multimodal imaging guided stereotactic serial biopsies has been described in detail before.3 In brief, all stereotactic procedures were performed using a modification of the Riechert headring and a workstation for multiplanar trajectory planning (target@1.19®, BrainLAB AG, Feldkirchen, Germany). On the basis of the multimodal planning following coregistration of computerized tomography (CT), magnetic resonance imaging (MRI) and O-(2-[(18)F]fluoroethyl)-l-tyrosine positron emission tomography (FET-PET) data (i-plan stereotaxy®, BrainLAB AG, Feldkirchen, Germany) for better visualization of the tumor and the tumor/vessel interface, the 3D workstation allowed the simulation of any given trajectory. Multiplanar reconstructions were used in every patient for choosing the best trajectory. A serial biopsy was taken along a trajectory representative of the solid tumor tissue (including the area with highest FET uptake (“hotspot”), if present). The maximum amount of the tissue per biopsy specimen was 1 mm3. The number of specimens taken in one millimeter steps was in the range of 12 samples per patient. Intraoperatively, smear preparations were routinely performed by a neuropathologist (K.B.).


Samples were taken for routine neuropathological examination and, in parallel, stored at −20°C until further molecular genetic analysis. From at least 2 of these fresh-frozen biopsy samples, DNA was isolated as described below. To verify the presence of tumor tissue in these samples and to check for correlations between MGMT promoter methylation and MGMT protein expression, a corresponding sample (level + 1 mm) was taken for paraffin embedding and histopathological examination including MGMT immunohistochemistry. For the detection of a potential intratumoral heterogeneity, between 2 and 4 biopsy specimens were collected from different sites along the chosen trajectory.

Isolation of nucleic acids

DNA was isolated from each biopsy specimen using the QIAamp® DNA Micro Kit (Qiagen, Hilden, Germany). DNA isolation from normal blood lymphocytes was performed either using magnet-based techniques (QuickPick™ gDNA, Bionobile, Turku, Finland) or the QIAmp DNA Blood Kit (Qiagen, Hilden, Germany). The quantity and purity of DNA were assessed using the NanoDrop® ND-1000 Spectrophotometer (NanoDrop, Wilmington, NC). DNA recovery from each stereotactic biopsy sample amounted to around 30–60 ng/μl.

Bisulfite modification of DNA

The bisulfite conversion reaction was performed with a total of 200–400 ng DNA by use of the EpiTect® Bisulfite Kit (Qiagen, Hilden, Germany). In this reaction, all cytosines except for their methylated counterparts are converted to uracil.

For the detection of promoter hypermethylation of the MGMT gene, both sequence analysis and MSP were performed using bisulfite-modified DNA as previously described by Möllemann et al.4

Methlyation-specific PCR

For MSP, 2 pairs of primers, each specific for either the methylated or the unmethylated MGMT promoter region, were used asoriginally described by Esteller et al.5 Primer sequences of MGMT were for the unmethylated reaction 5′-TTTGTGTTTTGATGTTTGTAGGTTTTTGT-3′ (upper primer) and 5′-AACTCCACACTCTTCCAAAAACAAAACA-3′ (lower primer) and for the methylated reaction 5′-TTTCGACGTTCGTAGGTTTTCGC-3′ (upper primer) and 5′-GCACTCTTCCGAAAACGAAACG-3′ (lower primer). The MSP reactions were routinely prepared in a total volume of 25 μl using HotStarTaq® Master Mix Kit (Qiagen, Hilden, Germany). For PCR, 1.8 μl of bisulfite-modified DNA were added and subjected to 36 PCR cycles with denaturation at 94°C for 30 s, primer annealing at 59°C for 40 s, extension at 72°C for 2 min and a final extension for 10 min. PCR products were separated using Spreadex® EL 600 or 800 gels (Elchrom Scientific AG, Cham, Switzerland) and loaded with 3–5 μl of the respective PCR product. Electrophoresis was carried out at 120 V and 55°C for 40 min. Amplicons were visualized using SYBR® Gold (Molecular Probes, OR; cf., Figs. 1 and 2).

Figure 1.

Methylation-specific polymerase chain reaction was performed for each of 3 stereotactic biopsy samples from 2 patients with glioblastomas (Patient ID: 7 [left panel] and 6 [right panel]). Whilst the 93 bp product of the unmethylated reaction can be detected in all six samples, a methylation-specific product of 89 bp can be seen in all 3 biopsy specimens from one patient (Patient ID: 6). Three microliters of the PCR product were loaded onto Spreadex EL800 gels, run at 55°C for 40 min and stained with SYBR gold; 20 bp ladder. In both patients, PCR results were confirmed by sequencing of bisulfite-modified DNA. The graphs depicting the methylation profile of 25 CpG islands (1–25) of each corresponding biopsy sample are shown in the lower panels.

Sequence analysis of bisulfite-modified DNA

Sequencing analysis was performed following bisulfite treatment of single-stranded DNA.6 For sequencing, a 316 base pairs (bp) PCR product encompassing 25 CpG sites between nucleotide numbers 46,922 and 47,124 (GenBank accession no. AL355531) was obtained using the primers described by Möllemann et al.4 The respective primers were sense: 5-ATGTTGGGATAGTTCGCGTTTTTAGA-3 and antisense: 5-CCAATCCACAATCACTACTTC-3. PCR was routinely performed in a total volume of 25 μl containing 2.5–3 μl of bisulfite-treated DNA and using HotStarTaq Master Mix Kit (Qiagen, Hilden, Germany). Forty-five cycles were carried out (94°C, 30 s; 52°C, 1 min; 72°C, 2 min) and 10 min final extension. PCR amplicons were separated on a 1.8% agarose gel prestained with ethidium bromide, extracted using MinElute™ Gel Extraction Kit (Qiagen, Hilden, Germany) and further processed for automated sequencing on an ABI 3130 Genetic Analyzer (Applied Biosystems, CA).

In the forward sequencing reaction, “partial methylation” was defined as the cytosine and thymine peaks being equally sized or the cytosine peak being twice as high as compared to the corresponding thymine peak. Positions with cytosine peaks as small as 10–50% of the thymine peak were considered weakly methylated. Accounting for the fact that sequence analysis per se is not a quantitative method, both completely and partially, but not weakly, methylated positions were considered as “methylated.” According to previous work4 and the described criteria, the MGMT promoter of a biopsy specimen was considered “methylated” when more than half of the CpG sites (13 of 25 CpG sites) were found to be “methylated.” The graphic presentation of the 25 CpG sites was performed by a custom designed Microsoft Excel Makro, which was used to convert the methylation results of each single CpG site into defined nuances on a grayscale as shown in Figures 1 and 2.

Figure 2.

Results of methylation-specific polymerase chain reaction analyses of 3 biopsy specimens from one patient (Patient ID: 8) showing a heterogeneous methylation pattern. Histology of sections neighboring biopsy position −26 showed a highly cellular and highly proliferative glioblastoma (for details see text). In contrast, histological sections from positions adjacent to position −1 and −22 were found to exhibit just necrotic debris and fresh erythrocytes. PCR results were confirmed by sequencing of bisulfite-modified DNA. The graphs depicting the methylation profile of 25 CpG islands (1–25) of each corresponding biopsy sample are shown in the lower panels.

Validation of the methodology adjusted for small sample size

Validation of this technique for stereotactic biopsy specimens was performed in 10 oligodendroglial brain tumor samples obtained from open tumor resection with a known MGMT promoter profile (blinded to the investigator) and previously analyzed by Möllemann et al.4 The amount of starting DNA from each of these samples was reduced to 300–600 ng to adjust these techniques to the average DNA recovery from stereotactic probes. All 10 samples were analyzed by methylation-specific PCR and sequencing analysis. Furthermore, concordance regarding the methylation status of specimens obtained by (i) open tumor resection and (ii) stereotactic biopsy was analyzed in selected patients.

MGMT immunohistochemistry

Whenever possible, 2 biopsy specimens from different sites along the chosen trajectory were collected for MGMT protein expression analysis; these specimens, which were topographically closely related to the tissue samples used for the molecular genetic analysis (level + 1 mm), were studied by immunohistochemistry on routinely processed, formalin-fixed, paraffin-embedded tissue sections (6 μm thick) according to a slightly modified protocol initially reported by Möllemann et al.4 In brief, tissue sections were deparaffinized in xylene for 30 min and rehydrated in decreasing concentrations of ethanol. The DAKO catalyzed signal amplification (CSA) horseradish peroxidase system (DAKO, Copenhagen, Denmark) was used as the detection system according to the manufacturer's protocol, including target retrieval solution (DAKO) and steamer pretreatment. Immunoreactivity was visualized with 3,3′-diaminobenzidine as the chromogen. All sections were counterstained with hematoxylin. The mouse monoclonal antibody MT 3.1 (DAKO) was used as primary antibody at a final dilution of 1:500 for an incubation period of 15 min at room temperature. Negative controls were carried out by omission of the primary antibody. Two observers blind to the clinical or molecular information scored each slide. Only nuclear staining was regarded as indicative of positive staining. The fraction of MGMT immunopositive tumor cells in each sample was evaluated semiquantitatively according to the following scale: (i) less than 10% positive tumor cells; (ii) 10–50% positive tumor cells; (iii) more than 50% positive tumor cells. Staining of vascular endothelial cells and/or lymphocytes was used as internal control for a positive reaction.



The current series comprised 20 patients with a primary supratentorial malignant glioma (16 patients with primary glioblastomas and 4 patients with primary anaplastic astrocytomas), 3 patients with a secondary anaplastic astrocytoma and 2 patients with a secondary glioblastoma. The mean age of the primary (de novo) tumor group was 62 years as compared to 42 years in the tumor progression group. The distribution of relevant clinical, histological and treatment-related data is summarized in Tables I and II. The overall number of biopsy specimens available for MSP and sequence analysis was 69 and the mean number of the analyzed biopsy specimens per patient was 3 (range 2–4); the mean of the maximum distance between the chosen biopsy sites throughout each tumor was 14 mm (range 2–32); 15 patients of the study group showed a maximum distance of at least 10 mm. The overall number of biopsy specimens available for MGMT immunohistochemistry was 36, which were taken from 23 tumors. In 13 of these 23 tumors, 2 tissue samples from distant sites within the tumor could be studied. The mean maximum distance between the chosen biopsy sites was 11 mm (range 2–30 mm). Untreated and previously treated tumors did not differ in terms of number of sample sizes taken per tumor and the maximum distance between the chosen biopsy sites. There was no surgery-related morbidity in this series.

Table I. Clinical Parameters and Biopsy Protocols for 25 Patients
Patient IDSexAge (years)Histological diagnosis (WHO°)MSPSEQMGMT expression
  1. Clinical parameters and biopsy protocols for all 25 patients with primary (n = 20) or secondary (n = 5) malignant gliomas (GBM = glioblastoma multiforme, AA = anaplastic astrocytoma) that were analyzed for MGMT promoter methylation (M = methylated, U = unmethylated) using methylation-specific PCR (MSP) and sodium bisulfite sequence analysis (SEQ; n.a. = not analyzed). MGMT immunopositive tumor cells in each sample were evaluated semiquantitatively according to the following scale: 1, less than 10% positive tumor cells; 2, 10–50% positive tumor cells; 3, more than 50% positive tumor cells.

Primary malignant gliomas (n = 20)
1m68GBM (IV)UU2
2m40GBM (IV)UU3
3m64GBM (IV)UU3
4m63GBM (IV)UU1
5m69GBM (IV)UU1
6f73GBM (IV)MM2
7f74GBM (IV)UU3
8m79GBM (IV)U & MU & M1
9m73GBM (IV)MM1
10m59GBM (IV)UU2
11f70GBM (IV)MM1
12m63GBM (IV)UUn.a.
13f69GBM (IV)UU3
14f76GBM (IV)UU3
15f82GBM (IV)Mn.a.2
16m64GBM (IV)Un.a.1
17m34AA (III)Un.a.1
18f50AA (III)UU3
19m46AA (III)MM3
20m32AA (III)Un.a.n.a.
Secondary malignant gliomas (n = 5)
21m44GBM (IV)MM3
22m68GBM (IV)MM2
23m34AA (III)MM1
24m35AA (III)Un.a.3
25m30AA (III)MM3

Validation experiments

DNA from 10 oligodendroglial brain tumor samples, including 5 samples with and 5 samples without a previously demonstrated MGMT promoter methylation (Möllemann et al.4), were analyzed with MSP and sequencing analysis in a blinded fashion using conditions suitable for small amounts of template DNA as obtained on average from stereotactic biopsy samples. These analyses revealed in all cases findings identical to the results independently obtained by Möllemann et al.4

Methylation-specific PCR

DNA recovery from each of the 69 stereotactically obtained tissue samples was in the range of 30–60 ng/μl (total amount of DNA extracted from one biopsy sample: 1 μg). The determination of the MGMT methylation status by MSP was successful in all of these samples: In 5 of 16 patients with a primary glioblastoma and in 1 of 4 patients with a primary anaplastic astrocytoma, a methylation-positive PCR product was detected, whilst the remaining patients showed an unmethylated MGMT promoter. The 2 patients harboring a secondary glioblastoma and 2 of 3 patients with a secondary WHO Grade III anaplastic astrocytoma exhibited a promoter hypermethylation (Table II). One patient of the tumor progression group was presented with a positive methylation status after extensive chemotherapy with temozolomide (16 cycles), which was interrupted at the time of tumor progression, whereas the remaining 3 patients did not receive chemotherapeutical treatment at all (Table II). The overall MGMT promoter methylation rate was 30% in the de novo group and 80% in the tumor progression group. The difference was not statistically significant (p = 0.1, Fishers' exact test).

Table II. List of Patients with Secondary Malignant Gliomas
Pat. IDInitial tumor histologyTreatmentActual tumor histologyMGMT promoter methylation status
  1. GBM, glioblastoma; AA, anaplastic astrocytoma; TR, gross total tumor resection; IRS, interstitial radiosurgery with transient Iodine-125 seeds; TMZ, temozolomide; c, cycles completed.

  2. All patients were initially diagnosed with diffuse astrocytoma of WHO grade II and had undergone different treatment modalities before progression. The MGMT promoter methylation status was determined at the time of malignant transformation in all these patients.

21Diffuse astrocytomaTR TMZ (17 c)GBMM
22Diffuse astrocytoma IRS (54Gy) GBMM
23Diffuse astrocytoma 2 × IRS (54Gy) AAM
24Diffuse astrocytomaTR  AAU
25Diffuse astrocytomaTR  AAM

The MGMT promoter methylation profile was homogeneous in all patients except one (Figs. 1 and 2). In this patient (Patient ID: 8), MSP showed an MGMT promoter hypermethylation in one sample taken from the periphery of the tumor, whereas MSP did not reveal PCR amplicons in the methylated reaction at additional biopsy positions (3 and 24 mm apart from the initial biopsy site, Fig. 2). These divergent findings were confirmed in a second MSP reaction and by subsequent sequencing analysis. Detailed analysis of the adjacent histological sections showed a highly cellular and pleomorphic glioblastoma at the initial biopsy position characterized by marked endothelial proliferation, occasional mitoses and apoptoses. The proliferation index amounted to ∼30%. In contrast, at biopsy positions 4, 22 and 23 mm apart from the initial biopsy site, only necrotic debris and erythrocytes were seen.

Three patients (Patient IDs 16, 20, 24; Table I) underwent subsequent open tumor resection: MSP analysis of the specimens obtained by microsurgery revealed the same MGMT status as initially defined by MSP of small-sized biopsy tissue samples (the promoter of the MGMT gene was unmethylated in these 3 cases).

Sequence analysis

Sequence analysis was successfully performed in 38 biopsy samples derived from 20 patients (cf. Table I). In 5 patients, sequencing data were not analyzable mainly due to technical problems. Sequence analysis and MSP exhibited 100% concordant findings. No heterogeneity between the different samples of each individual tumor was detected.

MGMT immunohistochemistry

MGMT immunohistochemistry was successfully performed in 23 tumors; in 2 patients, it was not initiated due to a limited amount of available tissue. Immunohistochemical studies on tissue samples obtained from distant sites within the same tumor revealed identical results with regard to the MGMT expression score in all 13 patients investigated. A high MGMT protein expression (≥10% positive tumor cells) was seen in 9 unmethylated tumors, whereas a low fraction (<10%) of MGMT positive tumor cells was seen in the remaining four. A methylated MGMT promoter was more often associated with an increased MGMT protein expression (MGMT scale 2 or 3: 6 patients) than with a low expression score (MGMT scale 1: 4 patients). The former concerned 3 of 6 methylated primary malignant gliomas, and 3 of 4 methylated secondary malignant gliomas. In summary, no significant correlation between the MGMT promoter methylation status and the fraction of MGMT protein positive tumor cells (<10% positive cells versus ≥10% positive tumor cells) could be detected in this series (Table I, Fig. 3).

Figure 3.

Examples of immunohistochemical stainings for MGMT protein expression demonstrating partly discrepant results as compared to the findings of MSP and sequence analysis: (a) less then 10% MGMT protein positive tumor cells, the MGMT promoter is methylated (patient ID: 9, Table I); (b) less then 10% positive tumor cells, the MGMT promoter is unmethylated (patient ID: 17); (c) 10–50% positive tumor cells, the MGMT promoter is methylated (patient ID: 22); (d) More than 50% positive tumor cells, the MGMT promoter is unmethylated (patient ID: 7).


MGMT promoter methylation has been shown to be positively correlated with treatment response and survival in patients with glioblastoma multiforme undergoing external beam radiation and concomitant or delayed chemotherapy with alkylating agents such as nitrosourea compounds and temozolomide. Furthermore, preliminary retrospective data let assume a similar prognostic role of the MGMT status for patients with supratentorial WHO grade III astrocytomas. Individualized chemotherapy protocols are currently under investigation: They are either applied to prospectively test the full advantage of a methylated promoter or to deplete MGMT activity using O6-benzylguanine in case of an unmethylated status and render thereby a primarily resistant malignant glioma chemosensitive. In this context, it is important to elucidate whether the MGMT promoter methylation is homogeneously distributed within the same tumor, or whether heterogeneity exists. Findings of heterogeneity might significantly influence the treatment response and prognosis of the patients concerned. For clarification, the current prospective study was conducted. Minimal invasive stereotactic serial biopsy technique (for accurate sampling of tumor tissue from different sites throughout the tumor) was combined with MSP and sequencing analysis for the determination of the intratumoral MGMT promoter methylation status in each of these samples. Additional imunohistochemistry was performed to determine the fraction of MGMT immunopositive tumor cells semiquantitatively throughout the individual tumor and its correlation with the corresponding MGMT status. Adult patients with primary or secondary supratentorial malignant gliomas (WHO Grade III or IV) were considered eligible for this study.

Successful analysis of MGMT promoter methylation in DNA from stereotactic biopsy samples

In the validation set (consisting of DNA from 10 tumors with a known MGMT promoter methylation status), it was shown that both MSP and sequencing analysis of bisulfite-modified DNA for MGMT promoter hypermethylation revealed reproducible and valid results, even for such small amounts of starting DNA as obtained from one stereotactic biopsy sample. This finding was also supported by 3 patients undergoing subsequent open tumor resection after stereotactic biopsy: MSP analysis of specimens obtained by microsurgery revealed the same MGMT status as initially defined by MSP of small-sized biopsy tissue samples.

In the study group, the analysis of the MGMT promoter status and its distribution throughout each tumor was based on 69 tissue samples (on average: 3 specimens per tumor); the success rate of MSP was 100% and of sequencing analysis 80%; both methods have yielded concordant findings in all of the samples tested. This suggests that either one of these 2 techniques are useful for determination of the MGMT promoter methylation status of small-sized cryopreserved stereotactic biopsy specimens from patients with malignant glioma and that the combination of both might increase the validity of the obtained results. The success rate in the current report was higher than in the study of Hegi et al.,2 who have found a failure rate of MSP in the range of 25% using paraffin embedded tumor tissue in all cases. Apparently, the avoidance of fixation-related reduction of tumor DNA quality by use of cryopreserved tumor tissue must be considered the essential step for obtaining high success rates in diagnostic MGMT testing. A more detailed evaluation of the sequencing results exhibited only slight variations in the methylation profile of 25 CpG sites between the different biopsy positions of each individual patient. In most cases, these variations seem to be due to methodological issues. Slight variations in template DNA can hardly be avoided since standardization of the amount of DNA added is more difficult following bisulfite treatment. We have also observed a higher variability in peak heights when lowering the amount of starting material below 100 ng of starting DNA. Alternatively, varying proportions of normal brain tissue, lymphocytes or endothelial cells with intact MGMT expression may be present in the tumor samples. This, in turn, may result in variable peak heights and render analysis of partially methylated CpG sites more difficult.

Frequency of MGMT promoter methylation in malignant gliomas

The overall MGMT promoter methylation rate in the de novo tumor group was 30% and did not differ between anaplastic astrocytomas and glioblastomas. Watanabe et al.7 and Hegi et al.2 have reported promoter methylation rates in the range of 38 and 45%, respectively, which were somewhat higher than in the current report. Whether this finding can be attributed to the higher aged study population in this series (mean age of the de novo group: 62 years) as compared to the Hegi and Watanabe studies (mean age: 55 and 53 years, respectively), remains a matter of speculation. For patients with malignant glioma, an inverse correlation between increased age and the frequency of MGMT promoter methylation has been reported by Nakamura et al.8

In our series, the methylation rate was higher in patients with secondary malignant gliomas as compared to patients with primary malignant gliomas; the difference, however, was not statistically significant, probably due to the small sample size in this study. A similar distribution in favor of secondary glioblastomas has been reported in the literature.8 Nakamura et al.8 have also frequently demonstrated MGMT promoter hypermethylation in diffuse low-grade astrocytomas that are prone to malignant transformation. The synoptic evaluation of these findings suggests that MGMT promoter methylation must be considered a rather early epigenetic event in the genesis of WHO grade II diffusely infiltrating astrocytomas. This view is also supported by a study by Gonzales-Gomez et al., who found that among genes tending to epigenetic silencing, MGMT was among those being hypermethylated in a large proportion of low grade gliomas, whereas other genes such as GSTP1 or p14ARF were mainly found hypermethylated in higher grade gliomas.9

Homogeneous distribution of MGMT promoter methylation

To study the intratumoral distribution of MGMT promoter methylation, we collected tissue samples from distant sites within each tumor. The maximum distance between the chosen biopsy sites was 14 mm (mean). It could be shown that the MGMT promoter methylation was homogeneously distributed in all samples of each individual tumor when solid tumor tissue was investigated. However, misleading results may be obtained in case of significant contamination of the tissue sample by necrotic debris and/or non-neoplastic tissue or blood (1 patient in this series, Patient ID: 8, Table I). The necessity to obtain viable tumor tissue for a reliable analysis and verify the presence of tumor tissue in at least one neighboring section (level + 1 mm) by histopathological examination particularly in those samples showing an unmethylated promoter must be underscored; otherwise false negative results cannot be excluded.

MGMT immunohistochemistry

Data concerning the correlation between MGMT promoter methylation and MGMT protein expression are scarce for malignant gliomas, and the prognostic relevance of the protein is still debated. Some investigators have advocated performing MGMT determination using immunohistochemistry instead of the more complex MSP or sequencing technique. On the other hand, heterogeneous MGMT expression within the tumors, high levels of MGMT protein in infiltrating lymphocytes, microglia and blood vessels, and the upregulation of the MGMT protein expression during the therapy (external beam radiation, chemotherapy, steroids) have been reported to limit the reproducibility of immunohistochemistry.2, 10, 11, 12, 13

In the current series, the MGMT protein expression profile was very similar in tumor tissue samples obtained from distant sites within the same tumor. This finding underscores the reproducibility of the applied method and indicates homogeneous protein expression throughout the viable areas of each tumor investigated. More data are necessary to confirm this preliminary finding. As tissue samples taken for immunohistochemistry were topographically highly related to those specimens used for MSP and sequence analysis, the current data set was predestined for a correlative analysis between the MGMT status and the MGMT protein expression: No correlation was found. For example, an unmethylated MGMT promoter was associated with low protein expression in 4 of 13 tumors, and a methylated status with a high protein expression in 6 of 10 patients. Notwithstanding the described shortcomings of immunohistochemistry, it still remains to be shown whether methylation of the MGMT promoter is the primary mechanism of MGMT downregulation in malignant gliomas. Further, so far unknown and maybe equally important control mechanisms of MGMT protein expression at transcriptional, posttranscriptional and/or translational level may exist.14

Clinical perspective

The MGMT promoter methylation status is the first molecular genetic determinant of survival and treatment response in patients with glioblastomas. The successful implementation of MSP and sequence analysis of bisulfite-modified DNA for the determination of the MGMT promoter methylation status in small biopsy specimens of patients with malignant gliomas will certainly modify management strategies for these patients in the future. The current series has demonstrated that extensive but still minimal invasive stereotactic biopsy procedures throughout mostly vascularized tumors such as malignant gliomas are possible in experienced hands and can be done without morbidity using a multimodally imaging guided approach to the lesion. However, in the absence of heterogeneity, a less extensive biopsy procedure is sufficient for analysis of the MGMT status. Clearly, prospective randomized studies will be needed for further confirmation of the prognostic role of the MGMT promoter methylation status in glioblastomas and particularly in WHO grade III astrocytomas; so far, results of molecular genetic analysis and immunohistochemistry cannot yet be formally considered for day-to-day clinical decision making outside clinical trials.


Information about the MGMT promoter methylation status in malignant gliomas can be reliably obtained from small-sized cryopreserved stereotactic biopsies using MSP and/or sodium bisulfite sequencing analysis. Contamination of the biopsy specimen by non-neoplastic or necrotic tissue can lead to false negative results. The methylation profile constitutes a homogeneous marker throughout malignant gliomas. The lack of correlation between MGMT status and MGMT protein expression needs further evaluation.


Dr. Nan Liang and Dr. Wei Xiang are gratefully acknowledged for expert help in setting up a custom designed “Excel Methylation Makro.” We further thank Peer Schmitz dedicated to the management of the molecular genetic database at the ZNP and Rosi Lederer for help with the manuscript. We also thank A. Henn and U. Jung for expert help in preparing histological specimens.