Extent and Patterns of MGMT Promoter Methylation in ... · The initial MGMT status of GS was...

Human Cancer Biology

Extent and Patterns of MGMT Promoter Methylation in Glioblastoma- andRespective Glioblastoma-Derived Spheres

Davide Sciuscio1–3, Annie-Claire Diserens1,2, Kristof van Dommelen1,2,4, Danielle Martinet5,Greg Jones6, Robert-Charles Janzer7, Claudio Pollo2, Marie-France Hamou1,2, Bernd Kaina8,Roger Stupp2, Marc Levivier2, and Monika E. Hegi1–3

AbstractPurpose: Quantitative methylation-specific tests suggest that not all cells in a glioblastoma with

detectable promoter methylation of the O6-methylguanine DNA methyltransferase (MGMT) gene carry

a methylatedMGMT allele. This observation may indicate cell subpopulations with distinct MGMT status,

raising the question of the clinically relevant cutoff ofMGMTmethylation therapy. Epigenetic silencing of

theMGMT gene by promoter methylation blunts repair of O6-methyl guanine and has been shown to be a

predictive factor for benefit from alkylating agent therapy in glioblastoma.

Experimental Design: Ten paired samples of glioblastoma and respective glioblastoma-derived spheres

(GS), cultured under stem cell conditions, were analyzed for the degree and pattern of MGMT promoter

methylation by methylation-specific clone sequencing, MGMT gene dosage, chromatin status, and

respective effects on MGMT expression and MGMT activity.

Results: In glioblastoma, MGMT-methylated alleles ranged from 10% to 90%. In contrast, methylated

alleles were highly enriched (100% of clones) in respective GS, even when 2 MGMT alleles were present,

with 1 exception (<50%). The CpG methylation patterns were characteristic for each glioblastoma

exhibiting 25% to 90% methylated CpGs of 28 sites interrogated. Furthermore, MGMT promoter

methylation was associated with a nonpermissive chromatin status in accordance with very low MGMT

transcript levels and undetectable MGMT activity.

Conclusions: In MGMT-methylated glioblastoma, MGMT promoter methylation is highly enriched in

GS that supposedly comprise glioma-initiating cells. Thus, even a low percentage of MGMT methylation

measured in a glioblastoma sample may be relevant and predict benefit from an alkylating agent therapy.

Clin Cancer Res; 17(2); 255–66. �2010 AACR.

Introduction

O6-Methylguanine-DNAmethyltransferase (MGMT) is acellular DNA repair protein ubiquitously expressed innormal human tissues (1). The MGMT protein rapidlyreverses alkylation (e.g., methylation) at the O6 position

of guanine, thereby neutralizing the cytotoxic effect ofalkylating agents (2). Hence, expression ofMGMT in tumorcells represents a keymechanism of resistance for alkylatingagent therapy (3).

Epigenetic silencing by aberrant promoter methyla-tion of MGMT (4) has become the first predictivemarker for benefit from alkylating agent therapy inglioblastoma (5–7). As combined chemoradiotherapycomprising the alkylating agent, temozolomide, hasbecome the new standard of care (8), there has beengrowing interest to use MGMT promoter methylationstatus for individual patient management, and forpatient stratification or selection in clinical trials, andnovel methods to assess MGMT status have been devel-oped (9). Most methylation-specific (MS) assays arebased on bisulfite conversion of the tumor-derivedDNA that converts unmethylated cytosine, but not 5-methyl-cytosine to uracil, followed by qualitative orquantitative determination using methylation-specificPCR (MSP) or sequencing (10–12).

Quantitative evaluation of methylated MGMT intumors; for example, by quantitative MSP or MS clone se-quencing suggests that not all cells in a given "methylated

Authors' Affiliations: 1Laboratory of Brain Tumor Biology and Genetics ofthe 2Service of Neurosurgery, Department of Clinical Neurosciences,Centre Hospitalier Universitaire Vaudois (CHUV) and University of Lau-sanne (UNIL), 3National Center of Competence in Research MolecularOncology, ISREC-SV-EPFL, Lausanne,4Service of Neurosurgery, CentreHospitalier du Centre du Valais, Sion, and 5Laboratory of Cytogenetics,Service of Medical Genetics, CHUV, UNIL, Lausanne, Switzerland;6MDxHealth Inc. Li�ege; Belgium; 7Division of Neuropathology, CHUVUNIL, Lausanne, Switzerland; and 8Department of Toxicology, UniversityMedical Center, Mainz, Germany.

Note: Supplementary data for this article are available at Clinical CancerResearch Online (http://clincancerres.aacrjournals.org/).

Corresponding Author: Monika E. Hegi, Laboratory of Brain TumorBiology and Genetics, Department of Neurosurgery, Centre HospitalierUniversitaire Vaudois (CHUV BH19-110), 46, rue du Bugnon, Lausanne1011, Switzerland. Phone: 41-21-314-2582; Fax: 41-21-314-2587; E-mail:[email protected]

doi: 10.1158/1078-0432.CCR-10-1931

�2010 American Association for Cancer Research.

ClinicalCancer

Research

www.aacrjournals.org 255

Cancer Research. on February 15, 2021. © 2011 American Association forclincancerres.aacrjournals.org Downloaded from

Published OnlineFirst November 19, 2010; DOI: 10.1158/1078-0432.CCR-10-1931

http://clincancerres.aacrjournals.org/

tumor" (glioblastoma) carry a methylated MGMT allele(11, 12). This raises the question of the clinicallyrelevant evaluation of MGMT methylation for indi-vidualized therapy. Several potential reasons could explainthese observations: (i) Tumor heterogeneity, a particularfeature of glioblastoma. However, a recent report demon-strated consistent results among multiple stereotacticbiopsies from various glioblastoma regions in a cohortof patients (13). Other possible reasons are (ii) contam-inating normal tissue, (iii) a heterogeneous methylationpattern reducing the efficiency of MSP, or finally(iv) presence of MGMT methylation in distinct tumorsubpopulations of cells, for example glioma-initiating cells(GIC). Of note, most glioblastoma exhibit monosomyof chromosome 10, home of the MGMT gene (10q26).Consequently, most glioblastoma need to acquire methy-lation of only 1 MGMT allele for complete loss of MGMTactivity in the tumor cells.

Here, we aimed at elucidating these distinct possibili-ties that are relevant for clinical interpretation of testresults for patient management. We investigated thedegree and pattern of MGMT promoter methylation inpaired samples of glioblastoma-derived spheres (GS) thatare enriched for GICs and depleted for nontumoral cells,and their respective original glioblastoma tissues usingMS clone sequencing and qMSP. The degree and densityof MGMT methylation was then compared with thechromatin structure of MGMT, gene dosage, gene expres-sion and enzyme activity, and the tumor cell content ofthe patient samples.

Materials and Methods

Glioblastoma samplesTumor tissues were collected from patients operated in

the Service of Neurosurgery at the University HospitalLausanne, with written informed consent of the patientsand approval of the local ethics committee (protocol F25/99). Tumors were histologically classified according to theWHO classification. Patient characteristics are detailed inTable 1.

Glioblastoma-derived spheres (GS)Fresh glioblastoma tissue was mechanically and enzy-

matically dissociated into a single cell suspension usingpapain 30 U/mL, (Worthington Biochem.; LS003119)and DNAse I 40 mg/mL (Roche, 1284932) and subjectedto magnetic bead cell sorting based on CD133 accordingto the manufacturer’s instructions (Miltenyi Biotec.).Cells were cultured under stem cell conditions usingDMEM-F12 medium (Invitrogen; 10565-018) supple-mented with human recombinant EGF (epidermalgrowth factor) and human recombinant basic FGF (fibro-blast growth factor; Peprotech; AF-100-15 and 100-18B),20 ng/mL of each, and 2% of B27 (Invitrogen; 17504). Ifnot otherwise stated, 50% of the medium was substitutedtwice weekly. The tumorigenicity of GS was evaluated byorthotopic injection into the brain of five 6- to 9-week-old female Swiss nu/nu mice as described previously(14). GS-induced tumors were recovered at sacrifice, apart was preserved for histologic evaluation, and GS wererederived without selection in stem cell medium. Theprotocol was approved by the local veterinary authorities(VD1181-3).

DNA and RNA isolationDNA and RNA were isolated from frozen tissue or cells

using the AllPrep DNA/RNA Kit (Qiagen; 80204). DNAfrom paraffin sections were isolated as described before(11). Nucleic acids were quantified by UV spectrophoto-metric analysis using Nanodrop 1000 Spectrophotometer(Thermo Scientific).

Bisulfite treatment and gel-based and quantitativeMethylation-Specific PCR

A total of 600 ng of genomic DNA was treated withsodium bisulfite using the EZ DNA Methylation Kit (ZymoResearch; D5001). After the bisulfite treatment step, pur-ified DNA was subjected to gel-based MSP using a 2-stepapproach with nested primers as previously described (15,16), using presence or absence of PCR products derivedfrommethylated and unmethylatedMGMT on agarose gelsas readout. The initial MGMT status of GS was determinedbetween 2 and 6 months in culture, with the exception of1871 that was first tested at 12 months. All other GS wereretested between 2 and 4 times over the observation periodof over 12 months. Quantitative MSP was performedby MDxHealth Inc. (former Oncomethylome Sciences)

Translational Relevance

Glioblastoma is the most common malignant pri-mary brain tumor with a dismal prognosis of 15months. Combined chemoradiotherapy comprisingthe alkylating agent, temozolomide, has led to a mod-est increase of survival. The demonstration that benefitfrom the alkylating agent, temozolomide, is largelyrestricted to patients whose glioblastoma contains amethylated O6-methylguanine DNA methyltransfer-ase (MGMT) promoter has stirred up the field. Con-sequently, clinical trials now stratify or selectglioblastoma patients according to the MGMT methy-lation status. However, the best technique of MGMTstatus determination, cutoff definition betweenmethylated and unmethylated, and the relevance ofextent of methylation are all subject to ongoing con-troversy. Our study provides molecular data to disen-tangle these issues, suggesting that in MGMT-methylated glioblastoma, cells with MGMT methyla-tion are enriched in the glioblastoma-derived spherefraction that comprises glioma-initiating cells. Hence,measurement of low levels ofMGMTmethylation maybe relevant for potential benefit from alkylating agenttherapy.

Sciuscio et al.

Clin Cancer Res; 17(2) January 15, 2011 Clinical Cancer Research256




as described (11). The copy number of methylatedMGMT,the copy number of beta-actin (ACTB) that serves as refer-ence gene for normalization, and the ratio "methylatedMGMT*1000/ACTB" were reported.

Methylation-specific clone sequencingPCR products from the first round of nested MSP (þ75–

þ364 from the transcription start site, TSS; Fig. 1B) encom-passing 28 CpG sites were cloned into a pCR2.1-TOPOvector according to the manufacturer’s instructions (TOPOTA Cloning; Invitrogen). The plasmid was then used totransform TOP10 competent cells. Ten to twenty clonedfragments per sample were sequenced using M13 primers(Sanger method, Microsynth CH).

Real-time quantitative PCRReal-time quantitative PCR was performed with Fast

SyBr Green Master Mix (Applied Biosystem) using RotorGene 6000 Real-Time PCR system (Corbett Life Science).

PCR reactions were run as triplicates. The temperatureprofile was as follows: 95�C (100 seconds) followed by40 cycles at 95�C (3 seconds) to 60�C (20 seconds). Thequality of the products was controlled by a meltingcurve. For quantification, standard curves were estab-lished by amplification of serial dilutions of cDNA forboth, the target gene and the endogenous reference(GAPDH). The primer sequences have been publishedpreviously (17, 18) and are listed in SupplementaryTable S2.

MGMT activityMGMT activity was measured in frozen samples on the

basis of a radioactive assay that measures the transfer oftritium-labeled methyl groups from the O6 position ofguanine to the MGMT protein as detailed elsewhere (19,20). The mean of 3 independent measurements werereported. The limit of detection was at 1 fmol/mg ofprotein. Extracts of HeLa S3 (750 fmol/mg of protein)

Primers

2081

2207

2288

2683

2459

2638

2540

2669

GBM

GS

GBM

GBM

GBM

GBM

GBM

GBM

GBM

GS

GS

GS

GS

GS

GS

GS

mGS

mGS

100 bp

minimal promoter enhancer

EXON 1

Complete CpG island (97 CpGs)

Fragment MS clone sequenced

Promoter with maximal activity

28 CpGs

A

B

CpG 1 97

MGMT

Figure 1. Methylation pattern of the MGMT promoter in glioblastoma and respective GS. A, the methylation patterns of individual MGMT alleles wereestablished in paired samples of glioblastoma (GBM) and respective GS byMS clone sequencing of 28 CpG sites (see B for location). Each line corresponds toa different clone and each column to a different CpG site. Black squares represent methylated sites, white squares represent unmethylated, and graysquares represent undefined status (unreadable sequence). The frequency of densely methylated alleles is enriched in the GS fraction. GBM_2638 shows nodense methylation (unmethylated). Patterns of methylation are specific for a given glioblastoma and are maintained in respective GS and even preserved inGS rederived from tumors initiated from respective GS in the brain of nude mice (mGS; mGS_2540, mGS_2669). The CpGs interrogated by the MS primersof the qualitative MSP (grey/grey) or the qMSP (grey/black) assay are indicated schematically in the header of each column. B, schematic representation ofthe CpG island located in the 50-region of theMGMT gene that includes 97 CpGs [adapted from Nakagawachi et al. (24)]. The location of the minimal promoterand the enhancer region are indicated. The region amplified for MS clone sequencing (þ75–þ364, reference to TSS (genome build V37.1) is indicatedat the bottom with a bar.

MGMT Promoter Methylation in Glioma Derived Spheres

www.aacrjournals.org Clin Cancer Res; 17(2) January 15, 2011 257




and HeLaMR (no detectable MGMT protein) cells served aspositive and negative control, respectively.

Array CGH analysisArray comparative genomic hybridization (CGH) was

performed using the Agilent Human Genome CGH Micro-array Kit 44A as described (21). The array was analyzedwith the Agilent scanner and the Feature Extraction soft-ware (v.9.5.3.1). Graphical overview was obtained usingthe CGH analytics software (v3.5.14) according to Hg18genome assembly (March 2006).

Quantitative chromatin immunoprecipitationChromatin immunoprecipitation (ChIP) assay was per-

formed using the MAGnify Chromatin Immunoprecipita-tion System (Invitrogen) according the manufacturer’sprotocols. Briefly, proteins from cell extracts of 3 � 106

cells were cross-linked to DNA by addition of formalde-hyde to a final concentration of 1% for 10minutes at roomtemperature. Cells were sonicated to yield fragments withan average size of 200 to 600 bp using the sonicatorsonoplus mini20 (Bandelin electronic). The soluble chro-matin fraction was collected, and 10% of the supernatantwas used for input normalization. Equivalent amounts ofeither anti-Histone-H3 (Positive control; Abcam; AB1791),anti-Trimethyl-Histone H3 (Lys 4; Cell Signaling Technol-ogy; C42D8), anti-Trimethyl-Histone H3 (Lys 27; CellSignaling Technology; C36B11) and normal rabbit IgG(negative control, Invitrogen) were added and incubatedaccording to the protocol. Purified eluted DNA was quan-tified by quantitative real-time PCR. Primers for MGMT(22), and the 2 controls,GAPDH as representative of a genewith an open chromatin state (EZ ChIP. Chromatin Immu-noprecipitation Kit) and MYOD1 as representative of agene with a closed chromatin state (23) are listed inSupplementary Table S1. The experiments have beenrepeated 3 times with 2 biological replicates.

Immunohistochemistry and estimation of tumorcontent

Formalin-fixed, paraffin-embedded tissue sections wereimmunostained for MGMT (MT3.1, dilution 1:50; Neo-Markers; Fremont), glial fibrillary acidic protein (GFAP),(G3893; Sigma; 1:1,000), CD45 (M0701; DAKO), CD68(ref. M0814; 1:100; DakoCytomation), and p53 (RM9105-SO; dilution 1:500; NeoMarkers). Heat-induced epitoperetrieval (HIER) was applied using citrate buffer of pH ¼6.0 (MGMT, 15 minutes; CD68, 5 minutes; CD45, p53, 10minutes). The primary antibody was incubated overnightat 4�C. Secondary antibodies were applied using the EliteVectastain Kit (Vector Laboratories) according to the man-ufacturer’s instructions. Immunoreactivity was visualizedwith 3030-diaminobenzidine as the chromogen. All sectionswere counterstained with hematoxylin. Negative controlswere carried out by omission of the primary antibody.

The tumor cell content was estimated semi-quantita-tively by the neuropathologist, Robert Janzer, by evaluating4 representative high-power fields within the area macro-

dissected for molecular analyses based on H&E (hematox-ylin and eosin) staining, and in a second step, including theimmunohistochemistry (IHC) for GFAP, identifying reac-tive astrocytes by their stellate and fine cell processes, andCD45 and CD68, visualizing microglia and macrophages.All cells expressing CD45 or CD68 were considered asnontumoral. Overall, cells excluded as non tumoral com-prised vascular and intravascular cells, lymphocytes,macrophages, microglia, and reactive astrocytes.

Results

Ten paired samples of GS and their corresponding ori-ginal glioblastoma tissues were evaluated for the extent anddensity of methylation by clone sequencing 28 of 97 CpGsin the CpG-island of the MGMT promoter (Fig. 1B). Thisregion of the promoter encompasses the enhancer elementand according to reporter assays is associated with com-plete silencing of the gene when fully methylated (24).Hence, most assays interrogate CpG methylation in thisregion (9).

These 10 samples investigated are highly selected inthe sense that they were obtained from patients forwhom we received large amounts of tissue (>1 g) andwho had mostly complete resections (Table 1). In ourhands, 50% of glioblastoma subjected to the CD133selection procedure to enrich for GICs eventually yieldglioma-derived spheres that can be maintained in stemcell culture for further experiments. CD133-positiveglioma cells have been shown to be enriched for GICs(25), although it has become clear that not all GICs areCD133 positive (26). More recently, other markers orspecifc phenotypic properties have been proposed for theisolation of GICs (27).

Frequency and methylation density of methylatedalleles in GBM and respective GS.

Using qualitative MSP visualized on a gel, 7 of the 10samples were considered MGMT methylated (Supplemen-tary Fig. S1 and Table 1). For all glioblastoma with MGMTmethylation, a band for MGMT unmethylated alleles wasalso detectable. In accordance, subsequent MS clonesequencing of the original glioblastoma tissue revealedthat 10% to 90% of all clones sequenced showed densemethylation (Figs. 1 and 2B), arbitrarily defined as at least 4consecutive CpGsmethylated in a given interrogated clone.In contrast, in the respective GS, all alleles were methylatedregardless of the gene copy number with 1 exception (Figs.1and 2B). In this sample, GS_2683, the methylated alleleswere also enriched, however, exhibitedmethylation in only40% of the clones. This is in accordance with a normalchromosome 10 copy number as determined by aCGH(Supplementary Fig. S4), comprising a methylated and anunmethylated allele, respectively. However, we cannotexclude to have different populations with a distinctmethy-lation status.

The density of MGMT promoter methylation, defined asthe number of methylated CpGs over 28 interrogated

Sciuscio et al.





CpGs, ranged from 25% to 90%, never reaching 100%, andshowing a characteristic pattern in a given tumor (Figs. 1and 2C). The pattern was retained in the respective GScultured under stem cell conditions. Furthermore, thepattern was even maintained after spheres were rederivedfrom respective tumors induced orthotopically in nudemice exemplified for GS_2669 and GS_2540 (Fig. 1A).

Characteristic features of the GS-derived tumors indicativefor the presence of GICs, such as highly invasive growthand resemblance to the original glioblastoma is displayedin Supplementary Figure S5. These are properties that havebeen attributed to GIC-derived tumors in previous reportsand are in stark contrast to glioblastoma cell line–derivedintracranial tumors (28, 29).

Figure 2. MGMT promotermethylation and MGMTexpression in GBM and GS. Thecorrelation of MGMT methylationand expression is visualized forthe original glioblastoma tissue(GBM, blue) and respective GS(red): A, visualizes the ratio ofmeth_MGMT*1000/ACTB byqMSP; B, the percentage ofclones with dense MGMTmethylation is represented, 10 ormore clones were evaluated byMS clone sequencing; C, themethylation density is defined aspercentage of methylated CpGs of28 CpGs interrogated; D, MGMTexpression as evaluated by real-time quantitative PCR. Themolecular characterization of theGS was performed between 2 and6 months in culture, with theexception of GS_1871 that wasevaluated at 12 months. PBL,peripheral blood lymphocytes.

35

30

25

20

15

10

5

0

800

700

600

500

400

300

200

100

0

100908070605040302010

0

2207

2288

2459

2540

2081

2669

2683

2638

1871

2108

Bra

in

PB

L/Li

ver

A

B

C

D

100908070605040302010

0

Met

h-M

GM

T ×

1,0

00/A

CT

B%

MG

MT

met

h. c

lone

sM

ethy

latio

n de

nsity

(%

)M

GM

T/1

0,00

0 G

AP

DH

Bra

in

PB

L/Li

ver

SW

48

HT

29

336

GBM

GS






For the 3 paired cases with unmethylated MGMT, MSPwas repeated at several time points for the GS fractions torule out the possibility that MGMT promoter methylationis acquired in the GS as an in vitro artifact. This was furtherconfirmed by MS sequencing in late passages (GS_1871,360 days, GS_2108, 165 days; GS_2638, days 120). MSclone sequencing of GS_2638, 1 of the 3 unmethylatedGS_lines, identified individual scattered methylated CpGsites in some clones, confirming the unmethylated status asdetermined by all other methods utilized (Fig. 1A andSupplementary Figs. S1 and S2).

qMSP confirmed the observed "enrichment" of methy-lated alleles in GS as compared with the original glioblas-toma (Fig. 2A). In 2 cases, however, 2207 and 2683, theassay did not adequately reflect the frequency of methy-latedMGMT alleles (ratio ofmeth_MGMT normalized withACTB). Both cases, exhibited a methylation pattern notfavoring hybridization of the primers of the qMSP assay. Infact, in both the cases, a signal forMGMT methylation wasmeasured in the original glioblastoma and was enriched inthe respective GS. The measured meth_MGMT copy num-ber would be visible on a gel [2207: GBM (glioblastoma)34 copies, GS 342 copies; 2683: GBM 22 copies, GS 86copies], as also shown for the gel-based MSP result (Sup-plementary Fig. S1). However, the high copy numbermeasured for ACTB that does not suffer from changes inPCR efficiency neutralized the signal in the normalizationstep.

To estimate the frequency of such false-negative cases,we looked at the raw data from diagnostic MGMT testsperformed by MDxHealth Inc. for glioma patients treatedin Lausanne in the last 3 years. Of 76 tests, only 2 cases

showed a low, but measurable meth_MGMT copy numberwith a normalized ratio (ACTB) corresponding tounmethylated MGMT. In fact, 1 corresponded to thepatient sample GBM_2683 characterized here. The second,glioblastoma 2558, was subsequently subjected to clonesequencing for characterization of the methylation pattern.Indeed, clone sequencing revealed methylated alleles in20% of the clones with a density of 30% (SupplementaryFig. S3). The pattern of methylation was unfavorable fordetection by qMSP and the gel-based assay. The latterrevealed a band in 2 of 4 replicas, thus the test was atthe limit of detection. Only formalin-fixed material wasavailable for this case.

MGMT methylation is associated with no or lowMGMT expression in GS

Evaluation of MGMT RNA expression in the pairedsamples revealed no or very low expression of MGMT inany GS with dense MGMT methylation, regardless of thepattern that is significantly lower than in the unmethylatedGS samples (P ¼ 0.03; Fig. 2D). Several of the respectiveoriginal glioblastoma tissues (GBM_2207, GBM_2288,GBM_2540; Fig. 2D) however, displayed MGMT expres-sion, which in some cases reached similar levels ofunmethylated cases (GBM or GS), normal brain or periph-eral blood lymphocytes. Physiologic expression levels ofMGMT vary in a cell-type–specific manner, with normalliver expressing more than 10� the expression of normalbrain and lymphocytes (Fig. 2D). In accordance with lackof expression, MGMT activity was below the limit ofdetection in the GS of completely methylated cases(Fig. 3). Moderate activity was measured in GS_2683 that

MG

MT

act

ivity

(fm

ol/m

g of

pro

tein

)

GBM GS GBM GS GBM GS GBM GS GBM GS 104 1780 2825 122

2207 2540 2669 2683 2638 BRAIN PBL Liver

100

90

80

70

60

50

40

30

20

10

0

324 574

Figure 3.MGMT activity in glioblastoma and respective GS. MGMT activity (fmol/mg of protein) was determined in GS and the respective original glioblastomatissue (GBM). MGMT activity was at the limit of detection for the MGMT methylated samples GS_2207, GS_2450, and GS_2669, whereas the respectiveoriginal glioblastoma tissues were positive. MGMT activity was detected in the GS_2683, displaying a methylated and an unmethylated MGMT status;and in GS_2638 with an unmethylated MGMT status. MGMT activity is in the same order of magnitude in nonneoplastic brain, whereas activity in peripheralblood lymphocytes (PBL) and particularly, liver is much higher.

Sciuscio et al.





has a methylated and an unmethylated allele. And theunmethylated GS_2638 showed the highest activity.MGMT activity was measurable in all respective originaltissues including the glioblastoma of the methylated cases.The source of MGMT expression and MGMT activity in themethylated cases is likely nontumoral tissue as will bediscussed later.

MGMT methylation is associated with a closedchromatin structureNext, we addressed the question whether the observed

CpG methylation in the MGMT promoter was associatedwith a repressive chromatin structure. Hypermethylationof CpG islands in the promoter region leads to genesilencing either by direct inhibition of transcription fac-tor binding, or by attracting methylated DNA-bindingproteins, recruiting other transcriptional repressors suchas histone acetylases (HDACs) and histone methyl trans-ferases. Respective modifications of histones lead tochanges in the chromatin structure resulting in transcrip-tionally inactive chromatin (reviewed in ref. 30). A markfor transcriptionally repressed chromatin is trimethylationof histone H3 at lysine 27 (H3K27m3), whereas trimethy-lation at lysine 4 (H3K4m3) is associated with transcrip-tionally active chromatin. Indeed, qChIP analysis providedevidence that dense MGMT methylation in GS was asso-ciated with presence of histone marks for inactive chroma-tin (H3K27m3) and absence of histone marks for activechromatin (H3K4m3; Fig. 4). In accordance with lack ofMGMT expression and MGMT activity, this pattern forclosed chromatin was also observed for GS_2207 that

exerted a methylation density of only 35%, and was there-fore falsely classified as unmethylated by qMSP. GS_2683with a methylated and an unmethylated MGMT alleleaccordingly displayed both, marks for active and inactivechromatin. Interestingly, GS_2638 with an exclusivelyunmethylated MGMT promoter revealed marks for active,but also inactive chromatin that may explain the relativelylow MGMT expression observed for this particular case(Figs. 2D and 4). In contrast, lymphocytes expectedlyshowed only marks for active chromatin.

Origin of MGMT expression in methylated GBMIn contrast to the original clinical tumor samples, we

found good accordance in the GS between loss of MGMTexpression, lack of MGMT activity, and marks for a repres-sive chromatin structure on one hand, and MGMT pro-moter methylation on the other. The detection ofunmethylated MGMT alleles and expression and activityof MGMT in an otherwise methylated tumor can beexplained by the presence of contaminating/infiltratingnontumoral cells that we and others have shown to expressMGMT RNA, protein or displayMGMT activity, like normalbrain (Figs. 2 and 3), tumor blood vessels, tumor-infiltrat-ing lymphocytes, and tumor-associated macrophages ormicroglia (9, 31). For macrophages or activated microgliathat are morphologically not always easy to recognize, wehave shown MGMT expression by double staining withCD68 (9). Furthermore, presence of a nonmethylated allelein a glioblastoma balanced for chromosome 10 mayexplain MGMT expression. Semi-quantitative estimationof the tumor cell content based on morphology on

H3K

4m3

GS 2207 GS 2540 GS 2669 GS 2683 GS 2638

2.5

2

1.5

1

0.5

0

PBL

H3K

27m

3

H3K

4m3

H3K

27m

3

H3K

4m3

H3K

27m

3

H3K

4m3

H3K

27m

3

H3K

4m3

H3K

27m

3

H3K

4m3

H3K

27m

3

Per

cent

age

of in

put

Figure 4. Histone modifications at the promoter region of the MGMT gene. GS were investigated for marks of active or inactive chromatin as determined byqChIP using anti-H3K4m3 and anti-H3K27m3 antibodies, respectively, quantified by qPCR usingMGMT specific primers. Percentage of input is reported. The3 completely MGMT methylated cases, GS_2207, GS_2540, and GS_2669, exhibit only marks for inactive chromatin, whereas lymphocytes that areunmethylated and express MGMT only show marks for activate chromatin. GS_2683 with a methylated and an unmethylated MGMT allele shows marks forboth active and inactive chromatin. GS_2638 with an unmethylated promoter shows both marks, in agreement with modest expression of MGMT.






H&E-stained slides ranged from 62% to 96% (Fig. 5A).However, subsequent evaluation taking into account IHCfor several markers, including GFAP and CD45, and inparticular CD68 (Fig 5B) that recognizes activated macro-phages and microglia, suggested that the tumor contentwas dramatically overestimated in some cases for example,GBM_2669. Normal gene dosage at the MGMT locus, withpresence of a methylated and an unmethylated allele,explained why less than 50% of methylated clones weredetected in GS_2683. However, in the other 3 cases with abalanced MGMT locus, both alleles were methylated, indi-cative for complete silencing of MGMT.

Taken together, the percentage of methylated alleles in aglioblastoma with a methylated MGMT status was in gen-eral in good accordance with the tumor content adjustedfor CD68-positive cells and gene dosage at the MGMTlocus, with the exception of GBM_2459. In this tumor,only 10% of the clones were methylated, in contrast to100% of the clones in the respective GS. The discrepancycould not be explained by the tumor cell content, which

was estimated at 60%. This suggests that a tumor cellsubpopulation overlapping with the GS fraction is MGMTmethylated in this glioblastoma.

Our experiments demonstrate that the unavoidable pre-sence of variable amounts of nontumoral cells in clinicalsamples of glioblastoma precludes measurement ofMGMTexpression or MGMT activity as biomarker. As the smallnumbers of cases prohibited statistical evaluation of differ-ences in MGMT expression between methylated andunmethylated samples in the present study, we comparedMGMT mRNA expression in a larger set of glioblastomafrom the EORTC-NCIC study for which gene expressionprofiles and the MGMT methylation status were available(32). Although, we found a significant difference ofMGMTexpression between MGMT methylated and unmethylatedsamples (P < 0.002; Mann–Whitney test), there was nosignificant association with outcome when either consider-ing only the 42 patients in the TMZ/RT arm or all 68patients (Cox analysis, corrected for age, P > 0.4 or P >0.2, respectively). This is not surprising given the large

100

90

80

70

60

50

40

30

20

10

0

2207

2288

2459

2540

2081

2669

2683

2638

1871

2108

H&E CD68

Per

cent

age

of tu

mor

al c

ells

A

B2540 MGMT Meth 2540 MGMT Meth

MGMT CD68

Figure 5. Estimation of tumor cellfraction. A, the tumor cell fractionwas estimated on an H&E sectionin the tumor area designated formacrodisection for molecularanalysis; the mean of 4 high-power fields is reported. Theevaluation was repeated withCD68 staining. CD68 is a markerfor activated microglia andmacrophages. B, immunostainingof GBM_2540 for MGMT andCD68, respectively. Of note,GBM_2540 is MGMT methylatedand its GS fraction does neitherexpressMGMT nor display MGMTactivity (Figs. 2 and 3), suggestingthat nontumoral cells such asthose identified by CD68 may bethe origin of MGMT expressiondisplayed here.

Sciuscio et al.





Tab

le1.

Sum

maryof

patient

inform

ation

IDGen

der

Tum

or

loca

lization

Exten

tof

rese

ction

MGMT

status

Age,

yMMS

(<27

,27

–30

)Perform

status

OS,mo

Aliv

eFirst-lin

etrea

tmen

t

1871

FLe

fttemporal

STR

Unm

eth

73<2

72

50

TMZ

2081

MRight

parietal

PR

Meth

5827

–30

09

0Cileng

itide/TM

Z/RT

2108

MLe

ftfron

totemporal

STR

bUnm

eth

8427

–30

05

0RT(protoco

lNordic)

2207

MLe

ftparietal

MTR

Meth

4227

–30

116

0Cileng

itide/TM

Z/RT

2288

MLe

ftfron

tal

STR

Meth

4927

–30

140

1PTK

787þT

MZ/RT–

TMZ

2459

MRight

temporal

pos

t.STR

Meth

5727

–30

138

0TM

Z/RT–

TMZ

2540

MLe

ftfron

tal

MTR

Meth

5127

–30

135

1TM

Z/RT–

TMZ

2638

MLe

fttemporal

STR

bUnm

eth

59<2

71

90

TMZ/RT–

TMZ

2669

MRight

fron

tal

STR

Meth

70<2

72

130

RThy

pofractiona

tedd

2683

aM

Right

parietal

MTR

Meth

5127

–30

152

c0

TMZ/RT–

TMZ

Abbreviations

:MTR

,mac

rosc

opicallytotalres

ectio

n(>98

%);STR

,sub

totalres

ectio

n(90%

–98

%);PR,p

artia

lres

ectio

n(50%

–90

%);MMS,m

inim

entalstatus;

OS,o

verallsu

rvival;

meth,

methy

lated;an

dun

meth,

unmethy

lated.

aRec

urrent

glioblastom

a.bNoMRIwith

in24

hours.

cOve

rallsu

rvival

from

first

surgery47

mon

ths,

from

seco

ndsu

rgery5mon

ths.

dPatient

rece

ived

TMZas

seco

nd-linetrea

tmen

t.






overlap of expression levels observed between the 2 popu-lations (Supplementary Fig. S6). In contrast, the MGMTmethylation status was significantly (P < 0.0001) asso-ciated with outcome in either of the evaluated subpopula-tions (32).

Discussion

The clinical utility of the MGMT methylation status as abiomarker for benefit from alkylating agent therapy ingliomas has led to an ongoing debate on how the effectof MGMT should be assessed and which specific procedurewould be the best for routine clinical application.

Comparison of the methylated allele frequency esti-mated byMS clone sequencing in the original glioblastomatissues and respective GS and evaluation of the tumor cellcontent of the respective macrodissected paraffin sectionssuggest that inMGMT-methylated glioblastoma, the appar-ent level of methylation is strongly influenced by thepresence of nontumoral tissue. However, the methylationpattern determined in the original tumor tissue was alwaysretained in the respective GS, and dense methylation wasassociated with very low MGMT expression and lack ofMGMT activity. This was in concordance with the presenceof histone marks for inactive chromatin (H3K27m3) andabsence of histone marks for active chromatin (H3K4m3).As postulated, these patients with a methylated MGMTgene had longer survival (Table 1; P ¼ 0.004; log-ranktest), although the small numbers preclude proper statis-tical evaluation. In contrast, estimates ofMGMT expressionand activity levels in the original tumor tissue were con-founded by prominent "contamination" of the tumortissue with CD68þ microglial cells and macrophagesexpressing MGMT (9) and maybe responsible for false-positive results. This might explain the lack of associationof MGMT expression in glioblastoma, as measured by IHC,with survival of patients treated in a phase III trial withtemozolomide, whereas the MGMT methylation status inthe same tumor samples was a strong predictor of outcome(n ¼ 122; P ¼ 0.0001; ref. 33). Hence, this as well asprevious studies certainly discourage the use of proteinexpression or enzyme activity in clinical samples of glio-blastoma to evaluate the MGMT status for patient selectionor stratification (9, 31, 33). For tests evaluating the pro-moter methylation status, a particularly tricky point for alltests is the definition of both a technically and clinicallyrelevant cutoff for MGMT promoter methylation. Despitesome uncertainty of the perfect cutoff, the majority ofsamples will likely be classified similarly by most tests.Individual samples may be "misclassified" due to incom-plete methylation of the CpGs interrogated by a given testor presence of undue amounts of nontumoral tissue dilut-ing the signal in a quantitative test normalizing for input(e.g., ACTB).

The extent of methylation required and the sets of CpGsthat are crucial for complete silencing are unknown due tothe complexity of gene regulation (34). Methyl-CpG bind-ing proteins, such as methyl CpG binding protein 2

(MeCP2) and methyl-CpG binding domain protein 2(MBD2), bind to aberrantly methylated sequences, leadingto alterations of chromatin structure and preventing bind-ing of transcription factors, thereby silencing the gene (24).

The density and pattern ofMGMTmethylation may leadto underestimation of methylated alleles; this is due tolocation of the primers or the fidelity requirements of thepolymerase with the respective methylation-specific pri-mers. On the other hand, the evaluation of a set of CpGs, asopposed to individual CpGs, allows identification of densemethylation required for silencing. Thus, in qMSP, speci-ficity to recognize dense methylationmay sometimes be onthe cost of sensitivity. Conversely, MS-pyro-sequencingrecognizes the integral of methylation at individual CpGsites (35). In extreme, this could lead to a call of "methy-lated" in a case with only scattered methylated CpGs thatnever reaches the necessary density in a given allele to resultin silencing, hence, a false call. A similar test result wouldbe obtained with few densely methylated alleles comingfrom a fraction of cells. Identification of cases presenting amethylated and an unmethylated allele would be a chal-lenge for any test on tumor tissue due to the confoundingsignal from normal tissue. Interestingly, we found in 3 of 4GS with a normal copy number of the MGMT locus thatboth alleles were methylated.

Given the complexity of the biological relationshipbetween promoter methylation and gene silencing andthe difficulties to integrate these features into a test thatis in addition complicated by presence of nontumoraltissue, there are unavoidable drawbacks for any technologyattempting to predict loss of MGMT expression for poten-tial benefit from alkylating agent therapy. Key to introdu-cing tests for diagnostics is their prospective validation. Atpresent, most clinical trials use qMSP for prospectivepatient stratification or patient selection (36, 37). Resultsfor validation are expected in early 2011 from theRadiotherapy Oncology Group (RTOG) 0525/EORTC26052-22053 Intergroup trial (http://clinicaltrials.gov;NCT00304031) investigating 2 temozolomide chemother-apy regimens. In this study, all patients’ tumors were testedbefore stratified randomization for the MGMT statusaccording to a technical cutoff based on qMSP includinga gray zone around the cutoff termed "indetermined" (11,36). Expected from the bimodal distribution of methylatedand unmethylated cases, only a small percentage of caseswere close to the technical cutoff (indeterminate; n ¼ 62,6.2%; ref. 36). On the basis of our present study, we expectthat the samples classified as "indeterminate" comprisesamples with distinct features, such as "difficult" methyla-tion patterns inefficiently recognized by the test, or a smallsubpopulation of cells displaying a methylated promoter.Either group may or may not respond to alkylating agenttherapy giving the impression of a dose-dependent effect ofMGMT methylation on prognosis of glioblastoma patientstreated with temozolomide and radiotherapy as proposedby Dunn et al. (35). In the first, it is not clear if themethylation pattern leads to silencing; in the latter group,the question is if the subpopulation with a methylated

Sciuscio et al.





promoter is relevant to the malignant behavior of thetumor for example, overlaps with GICs. These considera-tions do not take into account other possibly prevailingmolecular and clinical factors for resistance or tumorgrowth.In conclusion, although most methylation-specific tests

will be able to discriminate clearly methylated fromclearly unmethylated cases that have an association withgene silencing, there is a gray zone comprising sampleswith different features recognized to a different degree bydistinct technologies. The frequency of these difficult toclassify subgroup is hard to estimate and may comprisearound 5% of glioblastoma patients; but, most impor-tantly may contain a fraction of patients that will benefitfrom alkylating agent chemotherapy. This view is sup-ported by our finding that methylatedMGMT alleles wereenriched in GS, a tumor-relevant cell subpopulation, evenwhen the original glioblastoma exerted only a minority ofmethylated alleles. For these gray-zone patients, standardtemozolomide chemotherapy or concomitant chemora-diotherapy should not be withheld until better treatmentoptions emerge. On the basis of these considerations, theEORTC 26082/22081 trial (http://clinicaltrials.gov,NCT01019434), selecting for MGMT unmethylatedpatients based on a qMSP assay, has introduced a respec-

tive safety margin (lower bound of the 95% confidenceinterval of the technical cutoff) in order not to withholdtemozolomide from a patient group potentially benefit-ing from it.

Disclosure of Potential Conflicts of Interest

ME Hegi is a paid consultant for MDxHealth Inc.

Acknowlegments

We thank the patients and their families for allowing the use of tumortissue for research and the neurosurgeons for providing fresh tumor tissuefor the experiments. We are grateful to Drs. Migliavacca and Bady for helpwith statistics, Dr. Nagel for cell extracts and MGMT activity assay, and Dr.Beckman for critically reading the manuscript.

Grant Support

This work was supported by the National Center of Competence in ResearchMolecular Oncology (NCCR), the Swiss National Science Foundation (31003A_122557), the Brain Tumor Funders Collaborative, and MDxHealth Inc.

The costs of publication of this article were defrayed in part by thepayment of page charges. This article must therefore be hereby markedadvertisement in accordance with 18 U.S.C. Section 1734 solely to indicatethis fact.

Received July 20, 2010; revised November 3, 2010; accepted November13, 2010; published OnlineFirst November 19, 2010.

References1. Gerson SL. MGMT: its role in cancer aetiology and cancer therapeu-

tics. Nat Rev Cancer 2004;4:296–307.2. Pegg AE. Repair of O(6)-alkylguanine by alkyltransferases. Mutat Res

2000;462:83–100.3. Kaina B, Christmann M, Naumann S, Roos WP. MGMT: key node in

the battle against genotoxicity, carcinogenicity and apoptosisinduced by alkylating agents. DNA Repair (Amst) 2007;6:1079–99.

4. Watts GS, Pieper RO, Costello JF, Peng YM, Dalton WS, FutscherBW. Methylation of discrete regions of the O6-methylguanine DNAmethyltransferase (MGMT) CpG island is associated with heterochro-matinization of the MGMT transcription start site and silencing of thegene. Mol Cell Biol 1997;17:5612–9.

5. Hegi ME, Diserens AC, Gorlia T, Hamou MF, de Tribolet N, Weller M,et al. MGMT gene silencing and benefit from temozolomide in glio-blastoma. New Engl J Med 2005;352:997–1003.

6. Gorlia T, Van Den Bent MJ, Hegi ME, Mirimanoff RO, Weller M,Cairncross JG, et al. Nomograms for predicting survival of patientswith newly diagnosed glioblastoma: prognostic factor analysis ofEORTC and NCIC trial 26981–22981/CE.3. Lancet Oncol 2008;9:29–38.

7. Esteller M, Garcia-Foncillas J, Andion E, Goodman SN, Hidalgo OF,Vanaclocha V, et al. Inactivation of the DNA-repair gene MGMT andthe clinical response of gliomas to alkylating agents. N Engl J Med2000;343:1350–4.

8. Stupp R, Hegi ME,MasonWP, van den Bent MJ, TaphoornMJ, JanzerRC, et al. Effects of radiotherapy with concomitant and adjuvanttemozolomide versus radiotherapy alone on survival in glioblastomain a randomised phase III study: 5-year analysis of the EORTC-NCICtrial. Lancet Oncol 2009;10:459–66.

9. Weller M, Stupp R, Reifenberger G, Brandes AA, van den Bent MJ,Wick W, et al. MGMT promoter methylation in malignant gliomas:ready for personalized medicine?Nat Rev Neurol 2010;6:39–51.

10. Esteller M, Hamilton SR, Burger PC, Baylin SB, Herman JG. Inactiva-tion of the DNA repair gene O6-methylguanine-DNA methyltransfer-

ase by promoter hypermethylation is a common event in primaryhuman neoplasia. Cancer Res 1999;59:793–7.

11. Vlassenbroeck I, Califice S, Diserens AC, Migliavacca E, Straub J, DiStefano I, et al. Validation of real-time methylation-specific PCR todetermine O6-methylguanine-DNA methyltransferase gene promotermethylation in glioma. J Mol Diagn 2008;10:332–7.

12. Mikeska T, Bock C, El-Maarri O, Hubner A, Ehrentraut D, SchrammJ, et al. Optimization of quantitative MGMT promoter methylationanalysis using pyrosequencing and combined bisulfite restrictionanalysis. J Mol Diagn 2007;9:1525–78.

13. Grasbon-Frodl EM, Kreth FW, Ruiter M, Schnell O, Bise K, Felsberg J,et al. Intratumoral homogeneity of MGMT promoter hypermethylationas demonstrated in serial stereotactic specimens from anaplasticastrocytomas and glioblastomas. Int J Cancer 2007;121:2458–64.

14. Lachat Y, Diserens AC, Nozaki M, Kobayashi H, Hamou MF, GodardS, et al. INK4a/Arf is required for suppression of EGFR/DeltaEGFR(2–7)-dependent ERK activation in mouse astrocytes and glioma. Onco-gene 2004;23:6854–63.

15. Stupp R, Hegi ME, Neyns B, Goldbrunner R, Schlegel U, ClementPMJ, et al. Phase I/IIa study of cilengitide and temozolomide withconcomitant radiotherapy followed by cilengitide and temozolomidemaintenance therapy in patients with newly diagnosed glioblastoma. JClin Oncol 2010:2712–8.

16. Palmisano WA, Divine KK, Saccomanno G, Gilliland FD, Baylin SB,Herman JG, et al. Predicting lung cancer by detecting aberrantpromoter methylation in sputum. Cancer Res 2000;60:5954–8.

17. Rood BR, Zhang H, Cogen PH. Intercellular heterogeneity of expres-sion of the MGMT DNA repair gene in pediatric medulloblastoma.Neuro Oncol 2004;6:200–7.

18. Andreeff M, Ruvolo V, Gadgil S, Zeng C, Coombes K, Chen W, et al.HOX expression patterns identify a common signature for favorableAML. Leukemia 2008;22:2041–7.

19. Voelter V, Diserens AC, Moulin A, Nagel G, Yan P, Migliavacca E, et al.Infrequent promoter methylation of the MGMT gene in liver metas-tases from uveal melanoma. Int J Cancer 2008;123:1215–8.






20. Wiewrodt D, Nagel G, Dreimuller N, Hundsberger T, Perneczky A,Kaina B. MGMT in primary and recurrent human glioblastomas afterradiation and chemotherapy and comparison with p53 status andclinical outcome. Int J Cancer 2008;122:1391–9.

21. Martinet D, Filges I, Besuchet SchmutzN,MorrisMA,GaideAC,DahounS, et al. Subtelomeric 6p deletion: clinical and array-CGH characteriza-tion in two patients. Am J Med Genet A 2008;146A:2094–102.

22. Ji W, Yang L, Yu L, Yuan J, Hu D, Zhang W, et al. Epigenetic silencingof O6-methylguanine DNA methyltransferase gene in NiS-trans-formed cells. Carcinogenesis 2008;29:1267–75.

23. Taniguchi H, Yamamoto H, Akutsu N, Nosho K, Adachi Y, Imai K,et al. Transcriptional silencing of hedgehog-interacting protein byCpG hypermethylation and chromatic structure in human gastro-intestinal cancer. J Pathol 2007;213:131–9.

24. Nakagawachi T, Soejima H, Urano T, Zhao W, Higashimoto K, SatohY, et al. Silencing effect of CpG island hypermethylation and histonemodifications on O6-methylguanine-DNA methyltransferase (MGMT)gene expression in human cancer. Oncogene 2003;22:8835–44.

25. Singh SK, Hawkins C, Clarke ID, Squire JA, Bayani J, Hide T, et al.Identification of human brain tumour initiating cells. Nature2004;432:396–401.

26. Beier D, Hau P, Proescholdt M, Lohmeier A, Wischhusen J, Oefner PJ,et al. CD133þ and CD133- glioblastoma-derived cancer stem cellsshow differential growth characteristics and molecular profiles. Can-cer Res 2007;67:4010–5.

27. Cl�ement V, Marino D, Cudalbu C, Hamou MF, Mlynarik V, de TriboletN, et al. Marker-independent identification of glioma-initiating cells.Nat Methods 2010;7:224–8.

28. Ricci-Vitiani L, Pallini R, Larocca LM, Lombardi DG, Signore M,Pierconti F, et al. Mesenchymal differentiation of glioblastoma stemcells. Cell Death Differ 2008;15:1491–8.

29. Gunther HS, Schmidt NO, Phillips HS, Kemming D, Kharbanda S,Soriano R, et al. Glioblastoma-derived stem cell-enriched cultures

form distinct subgroups according to molecular and phenotypiccriteria. Oncogene 2007;27:2897–909.

30. Esteller M. Epigenetics in cancer. N Engl J Med 2008;358:1148–59.31. Hegi ME, Liu L, Herman JG, Stupp R, Wick W, Weller M, et al.

Correlation of O6-methylguaninemethyltransferase (MGMT) promotermethylation with clinical outcomes in glioblastoma and clinical stra-tegies to modulate MGMT activity. J Clin Oncol 2008;26:4189–99.

32. Murat A, Migliavacca E, Gorlia T, LambivWL, Shay T, HamouMF, et al.Stem cell-related "self-renewal" signature and high epidermal growthfactor receptor expression associated with resistance to concomitantchemoradiotherapy in glioblastoma. J Clin Oncol 2008;26:3015–24.

33. Preusser M, Janzer CR, Felsberg J, Reifenberger G, Hamou MF,Diserens AC, et al. Anti-O6-methylguanine-methyltransferase (MGMT)immunohistochemistry in glioblastomamultiforme: observer variabilityand lack of association with patient survival impede its use as clinicalbiomarker. Brain Pathol 2008;18:520–32.

34. Bhakat KK, Mitra S. CpG methylation-dependent repressionof the human O6-methylguanine-DNA methyltransferase genelinked to chromatin structure alteration. Carcinogenesis 2003;24:1337–45.

35. Dunn J, Baborie A, Alam F, Joyce K, Moxham M, Sibson R, et al.Extent of MGMT promoter methylation correlates with outcome inglioblastomas given temozolomide and radiotherapy. Br J Cancer2009;101:124–31.

36. Aldape KD, Jones G, Wang M, Hegi M, Janzer RC, Stupp R, et al.MGMT methylation testing in RTOG 0525: a phase III trial of newlydiagnosed glioblastoma. J Clin Oncol 2009;27 Suppl:15s.Abstract2051.

37. Stupp R, Van Den Bent MJ, Erridge SC, Reardon DA, Hong Y, WheelerH, et al. Cilengitide in newly diagnosed glioblastoma with MGMTpromoter methylation: protocol of a multicenter, randomized, open-label, controlled phase III trial (CENTRIC). J Clin Oncol 2010;28Suppl:15s. Abstract TPS152.

Sciuscio et al.





2011;17:255-266. Published OnlineFirst November 19, 2010.Clin Cancer Res Davide Sciuscio, Annie-Claire Diserens, Kristof van Dommelen, et al.

SpheresGlioblastoma- and Respective Glioblastoma-Derived Promoter Methylation inMGMTExtent and Patterns of

Updated version

10.1158/1078-0432.CCR-10-1931doi:

Access the most recent version of this article at:

Material

Supplementary

http://clincancerres.aacrjournals.org/content/suppl/2011/03/21/1078-0432.CCR-10-1931.DC1Access the most recent supplemental material at:

Cited articles

http://clincancerres.aacrjournals.org/content/17/2/255.full#ref-list-1

This article cites 34 articles, 6 of which you can access for free at:

Citing articles

http://clincancerres.aacrjournals.org/content/17/2/255.full#related-urls

This article has been cited by 7 HighWire-hosted articles. Access the articles at:

E-mail alerts related to this article or journal.Sign up to receive free email-alerts

SubscriptionsReprints and

[email protected] at

To order reprints of this article or to subscribe to the journal, contact the AACR Publications

Permissions

Rightslink site. (CCC)Click on "Request Permissions" which will take you to the Copyright Clearance Center's

.http://clincancerres.aacrjournals.org/content/17/2/255To request permission to re-use all or part of this article, use this link



http://clincancerres.aacrjournals.org/lookup/doi/10.1158/1078-0432.CCR-10-1931

http://clincancerres.aacrjournals.org/content/suppl/2011/03/21/1078-0432.CCR-10-1931.DC1

http://clincancerres.aacrjournals.org/content/17/2/255.full#ref-list-1

http://clincancerres.aacrjournals.org/content/17/2/255.full#related-urls

http://clincancerres.aacrjournals.org/cgi/alerts

mailto:[email protected]

http://clincancerres.aacrjournals.org/content/17/2/255


Extent and Patterns of MGMT Promoter Methylation in ... · The initial MGMT status of GS was...

Documents

Transcript of Extent and Patterns of MGMT Promoter Methylation in ... · The initial MGMT status of GS was...