the a/g allele of rs16906252 predicts for mgmt methylation … · (vsc) began recruitment of...

Predictive Biomarkers and Personalized Medicine

The A/G Allele of Rs16906252 Predicts for MGMT Methylationand Is Selectively Silenced in Premalignant Lesions fromSmokers and in Lung Adenocarcinomas

Shuguang Leng1, Amanda M. Bernauer1, Chibo Hong4, Kieu C. Do1, Christin M. Yingling1,Kristina G. Flores2, Mathewos Tessema1, Carmen S. Tellez1, Randall P. Willink1,Elizabeth A. Burki1, Maria A. Picchi1, Christine A. Stidley2, Michael D. Prados4,Joseph F. Costello4, Frank D. Gilliland5, Richard E. Crowell3, and Steven A. Belinsky1

AbstractPurpose: To address the association between sequence variants within the MGMT (O6-methylguanine-

DNA methyltransferase) promoter–enhancer region and methylation of MGMT in premalignant lesions

from smokers and lung adenocarcinomas, their biological effects on gene regulation, and targetingMGMT

for therapy.

Experimental Design: Single nucleotide polymorphisms (SNP) identified through sequencing a 1.9 kb

fragment 50 ofMGMTwere examined in relation toMGMTmethylation in 169 lung adenocarcinomas and

1,731 sputum samples from smokers. The effect of promoter haplotypes on MGMT expression was tested

using a luciferase reporter assay and cDNA expression analysis along with allele-specific sequencing for

methylation. The response of MGMT methylated lung cancer cell lines to the alkylating agent temozo-

lomide (TMZ) was assessed.

Results: The A allele of rs16906252 and the haplotype containing this SNPwere strongly associated with

increased risk forMGMTmethylation in adenocarcinomas (ORs � 94). This association was observed to a

lesser extent in sputum samples in both smoker cohorts. The A allele was selectively methylated in primary

lung tumors and cell lines heterozygous for rs16906252. With the most common haplotype as the

reference, a 20 to 41% reduction in promoter activity was seen for the haplotype carrying the A allele that

correlated with lower MGMT expression. The sensitivity of lung cancer cell lines to TMZ was strongly

correlated with levels of MGMT methylation and expression.

Conclusions: These studies provide strong evidence that the A allele of aMGMTpromoter–enhancer SNP

is a key determinant forMGMTmethylation in lung carcinogenesis.Moreover, TMZ treatmentmay benefit a

subset of lung cancer patients methylated for MGMT. Clin Cancer Res; 17(7); 2014–23. �2011 AACR.

Introduction

The fact that detection of gene promoter methylation oftumor suppressor genes in sputum that contains exfoliatedcells from premalignant lesions from smokers’ lungs is apromising marker for early stage lung cancer makes iden-

tifying factors that influence the propensity for this epige-netic process throughout the respiratory epithelium a highpriority. Such knowledge could impact strategies for screen-ing to improve risk models, and identify persons whowould benefit most from chemoprevention. The genesand the pathways they modulate underlying the interindi-vidual susceptibility to DNA methylation remain largelyelusive. Accumulating evidence from our group and othersimplicate DNA damage as an important step in the acquisi-tion of de novo methylation (1–4). This premise was sup-ported in vivo through a community-based study in whichwe showed a significant association between DNA repaircapacity and germ line sequence variants within DNArepair genes and gene promoter methylation in sputumfrom smokers (5). The impact of this finding for preventionwas addressed by asking whether dietary nutrients andvitamin supplements that may influence DNA damageand repair would modify the extent of gene methylationin the aerodigestive tract. Significant protection againstmethylation was observed in subjects consuming adiet rich in leafy green vegetables, or folate, or taking a

Authors' Affiliations: 1Lung Cancer Program, Lovelace RespiratoryResearch Institute, 2Department of Internal Medicine, University ofNew Mexico, and 3New Mexico VA Health Care System, Albuquerque,New Mexico; 4Brain Tumor Research Center, Department of Neuro-logical Surgery, University of California-San Francisco, San Francisco,California; and 5Norris Cancer Center and the Keck School ofMedicine, University of Southern California, Los Angeles, California

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

Corresponding Author: Steven A. Belinsky, Lung Cancer Program, Love-lace Respiratory Research Institute, Albuquerque, NM 87108. Phone: 505-348-9465; Fax: 505-348-4990; E-mail: [email protected]

doi: 10.1158/1078-0432.CCR-10-3026

�2011 American Association for Cancer Research.

ClinicalCancer

Research

Clin Cancer Res; 17(7) April 1, 20112014

on June 2, 2020. © 2011 American Association for Cancer Research. clincancerres.aacrjournals.org Downloaded from

Published OnlineFirst February 25, 2011; DOI: 10.1158/1078-0432.CCR-10-3026

http://clincancerres.aacrjournals.org/

multivitamin, supporting the concept that diet and supple-ments can affect reprogramming of the epigenome andthus have the potential to modify risk for lung cancer (6).Specific sequence patterns within gene promoters and

embryonic targets of polycomb-repressive complex 2 arepredictive for identifying specific genesmethylated inmanycancers, but cannot discriminate individual susceptibilityfor gene silencing (7–10). However, other local sequencefeatures that are polymorphic among people may contri-bute to the susceptibility of a specific CpG island tobecoming methylated (11). A genome-wide survey in nor-mal tissues found differential DNA methylation andexpression between alleles at multiple nonimprinted locithat was sequence dependent (12). An additional studyassessing 16 CpG rich gene promoters of chromosome 21in leukocyte DNA identified 4 genes with allele-specificmethylation (ASM) associated with single nucleotide poly-morphisms (SNPs; ref.13). Therefore, sequence dependentASM may represent an important mechanism connectinggenetic polymorphisms and phenotypic variability toimpact cancer susceptibility.O6-methylguanine-DNA methyltransferase (MGMT) is

a DNA repair enzyme that protects cells from the carci-nogenic effects of alkylating agents by removing adductsfrom the O6 position of guanine (14). The protectiveeffect of MGMT activity against the cytotoxic effects ofalkylating agents has been demonstrated in human celllines, xenograft models, andMGMT transgenic mice (14).The predominant mechanism for MGMT gene inactiva-tion is methylation of a CpG island within its promoter–enhancer region that leads to transcriptional silencing inbrain, colorectal, lung, lymphoma, and head and neckcancers (15). Silencing of this gene in cancer is associatedwith an increased prevalence for transition mutations inthe K-ras and p53 genes, consistent with the importantrole for this gene in removing alkyl DNA adducts (16, 17).

Methylation of MGMT is also a promising biomarker forlung cancer detection and its silencing is prognostic forresponse of glioblastoma patients to the alkylating agenttemozolomide (TMZ; refs.18 and 19). Transcriptionalregulation of MGMT occurs largely in a 1.2 kb fragment50 of the gene which contains the first untranslated exon(20). A 59 bp enhancer required for efficient MGMTpromoter function was identified between the exon 1and intron 1 boundary along with a SNP (rs16906252)located at the boundary between the enhancer and exon 1(20, 21). Colorectal cancer patients heterozygous orhomozygous for the A allele (vs. G allele) of this SNPhad a 32-fold increased odds for methylation of theMGMT gene in their tumors, a finding replicated in asecond, independent study (21, 22).

The purpose of this study was to address the associationbetween sequence variants within the MGMT promoter–enhancer region and methylation of MGMT in lung ade-nocarcinomas and exfoliated epithelial cells within sputumfrom smokers, and their biological effects on gene regula-tion. Genetic and epigenetic variations that affect MGMTexpression may impact personalized therapy for lung can-cer. Therefore, the potential of TMZ as a therapeutic optionfor lung cancer was evaluated in lung cancer cell lines withdifferent levels of methylation and expression of MGMT.

Materials and Methods

Study populations and sample collectionThe Lovelace Smokers Cohort (LSC) began recruitment

of female smokers in 2001 and expanded to include malesmokers in 2004 (5). Enrollment is restricted to current andformer smokers age 40 to 74 years with a minimum of 15pack-years of smoking. At study entry, all cohort memberscompleted a detailed questionnaire that collected informa-tion on medical, smoking, exposure history, socioeco-nomic status, and quality of life. Pulmonary functionwas tested to assess lung function, blood was drawn forcollection of plasma and lymphoctyes, and sputum wascollected by induction. The Veterans Smokers Cohort(VSC) began recruitment of smokers in 2000. Similarenrollment criteria and procedures as used in the LSC werefollowed. Most participants in the VSC are males and haveover 20 pack-years smoking history. In total, 1,731 lungcancer-free subjects (1,256 subjects from LSC and 475subjects from VSC) were included in this study.

Lung adenocarcinomas (n ¼ 169) were obtained fromthe tumor banks at Johns Hopkins and the Mayo Clinic(23). Demographic characteristics of the patients have beendescribed (23). Stage I lung cancer patients (n ¼ 98)enrolled in the New Mexico lung cancer cohort were alsoused in this study because of the availability of frozen tissuefor assessing MGMT expression. A detailed description ofthe enrollment and characteristics of the cohort membershas been provided previously (24). All samples were col-lected with informed consent and the institutional reviewboards of the Lovelace Respiratory Research Institute (Wes-tern IRB) and New Mexico VA Health Care System

Translational Relevance

Epigenetic silencing of MGMT (O6-methylguanine-DNA methyltransferase) by promoter methylation is abiomarker for identifying persons at high risk for lungcancer and for predicting therapeutic response to alky-lating agents in glioblastoma. The A/G allele ofrs16906252 in MGMT promoter–enhancer region pre-dicts for gene methylation in sputum from cancer freesmokers. This association was increased substantially inprimary lung tumors supporting this variant as a keydeterminant for epigenetic silencing of MGMT. Impor-tantly, the sensitivity of lung cancer cell lines to temo-zolomide (TMZ) was strongly correlated with levels ofMGMT methylation and expression, suggesting TMZtreatment may benefit a subset of lung cancer patientsmethylated for MGMT, a hypothesis that warrantsfurther preclinical studies prior to initiating clinicaltrials.

Selective Silencing of the A/G Allele of MGMT in Lung Cancer

www.aacrjournals.org Clin Cancer Res; 17(7) April 1, 2011 2015




approved all investigations using human tissues and clin-ical data.

Cell linesNormal human bronchial epithelial cells (NHBEC) iso-

lated from bronchoscopy of cancer-free smokers (n ¼ 20),19 lung cancer-derived cell lines (A549, Calu-3, Calu-6,HCC4006, HCC827-1, H23, H358, H460, H522, H1299,H1435, H1568, H1795, H1993, H2009, H2023, H2085,H441, and H2228), 1 immortalized normal human bron-chial epithelial cell line (HBEC2), and 2 glioblastoma celllines (U251 and T98G) were used in this study. HBEC2 wasreceived from Drs. Shay and Minna, Southwestern MedicalCenter, Dallas, TX. All the other cell lines were fromAmerican Type Culture Collection. Cells were maintainedin recommendedmedias in a 37�C incubator supplied with5% CO2. All experiments were conducted using cell lineswhich were passaged in the laboratory for fewer than6 months after resuscitation.

Sputum processing and MGMT promoter methylationA detailed procedure for sputum processing has been

introduced (18). Briefly, sputum samples were stored inSaccomanno’s fixative. Sputum adequacy defined as thepresence of deep lungmacrophages or Curschmann’s spiral(18) was assessed by cytotechnologists and subjects withcancer cells detected in sputum slides were removed fromthis study. An aliquot from sputum samples for each studysubject was used for DNA isolation. Promoter methylationof MGMT was studied in NHBECs and cell lines usingcombined bisulfite modification and restriction analysis(COBRA) and methylation-specific PCR (MSP) asdescribed (25). Promoter methylation of MGMT in pri-mary adenocarcinoma and sputum samples was evaluatedusing MSP and nested MSP, respectively (25). The COBRAandMSP assays are designed to interrogate the methylationstatus of CpGs around the transcriptional start site ofMGMT. COBRA and MSP primer sequences are listed inSupplementary Table S1, respectively.

SNP discovery and haplotype-tagging SNPs selectionA 1,871 bp fragment of theMGMT gene that contains the

well-characterized maximal promoter, the entire CpGisland, and part of intron 1 was carefully selected forSNP discovery because several major regulatory elementsfor MGMT transcription were identified to locate in thisregion (Supplementary Figure; refs. 20 and 26). Primers arelisted in Supplementary Table S1. DNA samples from 49cancer-free non-Hispanic Whites randomly selected fromthe LSC were used for discovery of SNPs. An additional 11samples were selected for sequencing based on theirrs16906252 genotypes (3 A/Gs and 8 A/As) to improvethe probability of finding a SNP in perfect linkage dis-equilibrium (LD)with rs16906252. PCR products were gel-purified and were commercially sequenced to identifyvariants (Sequetech). Pairwise r2 between any 2 SNPswas calculated to identify a SNP in perfect LD withrs16906252 (r2 ¼ 1). A Bayesian statistical method imple-

mented in the program PHASE (Version 2.1) was used toconstruct the haplotypes using the variants discovered inthe 60 subjects (27, 28). Haplotype-tagging SNPs (htSNP)were chosen using the TagSNPs program (29) based on thehaplotype alleles and frequencies generated by the PHASEprogram in 49 randomly selected samples. In this process,the squared correlation (Rh

2) between the true haplotypes(h) and their estimates was calculated. An expectationmaximization algorithm approach was then implementedto find the minimum set of SNPs (within the promoter–enhancer region ofMGMT) that would have Rh

2 > 0.80 forall common haplotypes.

SNP genotypingGenotyping was conducted in DNA samples isolated

from peripheral lymphocytes in subjects from LSC andVSC. DNA isolated from tumor tissues was used for geno-typing in 169 lung adenocarcinomas because no blood wascollected from these patients. It is highly unlikely that DNArecovered from tumor tissue will generate false calls forgerm line SNPs because stromal and inflammatory cellsalso comprise the tumor specimen. Primers amenable todetecting a PCR-restriction fragment length polymorphism(RFLP) were designed to genotype the SNPs in DNA sam-ples isolated from the lung adenocarcinomas (n ¼ 169),subjects (n ¼ 475) from the VSC, NHBECs (n ¼ 20), andcell lines (n ¼ 22). The primers and restriction digestionenzymes are listed in Supplementary Table S1. The 60 DNAsamples used for SNP discovery were regenotyped by PCR-RFLP and 100% concordance of genotypes was observedbetween this method and sequencing. Two DNA sampleswith wild homozygote and variant homozygote wereincluded in each batch of PCR-RFLP assay for qualitycontrol. The Illumina Golden Gate Assay was used togenotype rs16906252 in subjects from the LSC (n ¼1,256). Intraplate and interplate duplicates and 1 par-ent–child trio were included in each 96-well plate forquality control.

Selection of subjects and construction of MGMTpromoter constructs

A 1,528-bp fragment covering the entire maximal pro-moter and CpG island of MGMT with different haplotypeswas amplified by PCR from homozygotes, or heterozygoteswhen a homozygote for a haplotype was not present in the60 subjects used in discovery of the SNPs (SupplementaryFigure). The primers are listed in Supplementary Table S1.The promoter–enhancer fragment was directionally clonedinto the pGL2-basic Luciferase Reporter Vector (Promega)upstream of the luciferase coding sequence. Four to 6clones from each person were commercially sequencedto identify the correct construct with each specific haplo-type (Sequetech).

Transient transfection and reporter gene assaysTransient transfections were conducted in Calu-6,

H1299, and HBEC2. Cells (0.8–1.2 � 106) were platedinto 6-well plates and transfected the following day.

Leng et al.

Clin Cancer Res; 17(7) April 1, 2011 Clinical Cancer Research2016




Plasmid DNA (1 mg) and the pSV-b-Galactosidase controlvector (0.2 mg; Promega) were cotransfected into cells withLipofectamine 2000 transfection reagent (Invitrogen) at aLipofectamine 2000: DNA ratio of 3:1. A promoter-lesspGL2-basic vector and the pGL2-control vector that con-tains the SV40 promoter were used as a negative andpositive control, respectively. Forty-eight hours after trans-fection, cells were harvested and lysed. Immediately afterlysing, cell extracts were assayed in the Lumionskan Ascentluminometer (Thermo Electron) for luciferase activityusing the Luciferase Assay System (Promega). b-galactosi-dase activity in cell lysates was measured using the Galacto-Star Reporter Gene Assay System (Tropix). Promoter activ-ity was calculated as the ratio of activities of luciferase andb-galactosidase. Transfections were done in triplicate in 3independent experiments.

Cytotoxicity of TMZ in tumor cell linesThe stock TMZ (Sigma-Aldrich) solution (129 mmol/L)

was prepared by dissolving 100 mg TMZ in 4 mL dimethylsulfoxide and was frozen at �80�C in aliquots. Beforetreatment, TMZ stock was thawed and added to the culturemedium at a series of final concentrations (0, 10, 30, 90,270, 810, and 2,500 mmol/L). Cells (5 � 103/well) wereseeded into 96-well plates and were incubated with TMZ inthe culture medium for 72 hours. The MTT assay measuredcytotoxicity and an IC50 (the half maximal inhibitoryconcentration) was calculated. Three independent experi-ments were conducted for each cell line and the averageIC50 was presented.

Real-time PCR for measuring MGMT expressionCells were harvested in TRI reagent (Sigma-Aldrich) and

mRNA was isolated following TRI reagent instructions.cDNA was synthesized using the high capacity cDNAreverse transcription kit (Applied Biosystems). A real-timePCR using SYBR Green as the fluorescent dye was devel-oped to measure the MGMT expression level using the DCTmethod. The primers are listed in Supplementary Table S1.

Statistical analysisThe two-sample t-test, Wilcoxon rank sum test, and c2

test were used to compare demographic variables andgenotypes between MGMT methylated and unmethylatedgroups as appropriate. Hardy-Weinberg equilibrium(HWE) was tested for each SNP in the LSC, VSC, andthe lung adenocarcinomas separately and no deviationfrom HWE was identified (not shown). The minor allelefrequency (MAF) for rs16906252 is 7.6% in the 1,900samples. This study is well powered to investigate anassociation between a SNP with MAF of 0.076 and riskfor MGMT methylation in 3 different cohorts. Under theadditive inheritance model that assumes that the effect ofthe A/A genotype is twice that of the G/A genotype, we have99% power to detect an OR of 2.3, 1.9, and 1.6 in lungadenocarcinoma samples, VA cohort, and Lovelace cohort,respectively (30). A logistic regression model was used tocalculate the ORs and 95% CIs for each individual SNP

with adjustment for age, sex, ethnicity, and smoking historyselected a priori.

The PHASE program was used to reconstruct the haplo-types and calculate their estimated probabilities from thehtSNPs data in the 50 region of MGMT in lung adenocarci-nomas (27, 28). All possible haplotypes with correspond-ing estimated probabilities (>0.01) for each individualwere generated. The probabilities of the common haplo-types for each individual were used as explanatory variablesin the logistic regression model with adjustment for non-genetic factors to assess the association between the hap-lotypes and the risk for MGMT methylation. Haplotypeswith frequency less than 5% were combined into 1 group.The effect for each common haplotype was estimated usingthe most common haplotype as the reference group. Theeffects for the common haplotypes were also analyzedusing all other haplotypes as the reference group andsimilar results were obtained. All statistical analyses wereconducted in Statistical Analysis System 9.2.

Results

Association of SNPs to MGMT methylationSeveral SNPs were discovered in the promoter–enhancer

region of MGMT in a European population; however, theLD between those SNPs has not been characterized (31).Sequencing of a 1.9 kb fragment of MGMT identified 9SNPs with 1 SNP detected for the first time (SupplementaryTable S2). No 2 SNPs were in perfect LD (Fig. 1). The r2

value between any 2 SNPs in this area was 0.32 or less

Figure 1. The LD map of 9 SNPs identified in the 50 region of MGMT andthe 4 common haplotypes is depicted. The number in each diamondrepresents the r2 between any 2 corresponding SNPs. Three SNPs in grayrectangles (1, 3, and 7) were selected as htSNPs. The base components ofthe 4 common haplotypes and their frequency in 49 subjects are displayedas well.






except for rs1625649 and rs12356501 (r2 ¼ 0.71). FourSNPs had MAF greater than 0.04. A total of 4 haplotypeswith frequency greater than 0.04 were constructed based onthe 4 common SNPs. Three SNPs including rs1711646,rs1625649, and rs16906252 were selected as the htSNPs(29) that accounted for 87% of the haplotype variation inthe promoter–enhancer region of MGMT.

The association between these 3 htSNPs and MGMTmethylation was first tested in 169 lung adenocarcinomas.Rs1625649 was modestly associated with MGMT methyla-tion (Table 1). In contrast, a striking and highly significantassociation was seen between rs16906252 and increasedrisk forMGMTmethylation (OR ¼ 97.0, P < 0.0001). Onlyrs16906252 remained significant after Bonferroni correc-tion. The association between the genetic variation of theentire promoter–enhancer region and MGMT methylationwas then analyzed using a haplotype-based approach thatintegrated the genetic information ofmultiple SNPs. A totalof 7 haplotypes were reconstructed based on the genotypefor the 3 htSNPs including rs1711646, rs1625649, andrs16906252 in the 169 lung adenocarcinomas (27, 28).Four of the 7 haplotypes including HAP1 (GCG), HAP3(GAG), HAP4 (GAA), and HAP7 (TCG) had frequencygreater than 5% with their cumulative frequency of 94%(Supplementary Table S3). Haplotype analysis found thatonly HAP4 defined solely by rs16906252 was associatedwith MGMT methylation (Supplementary Table S3).MGMT is one of a panel of genes whose concomitant

methylation detected in sputum collected from moderateand heavy smokers associates with increased risk for lungcancer in a high-risk cohort (18, 32). Therefore, the asso-ciation between rs16906252 and MGMT methylation wasassessed in sputum samples collected in 2 smoker cohorts.The variant homozygotes and heterozygotes were asso-ciated with 12.4- and 6.3-fold increased risk for MGMTmethylation, respectively in the LSC (Table 1). The strongassociation between rs16906252 and MGMT methylationwas also replicated in smokers from the VSC (Table 1).

The commonality for promoter sequence variants toaffect other frequently methylated genes was addressedby studying the p16 gene that is silenced by promotermethylation in greater than 50% of lung tumors (25).Only 1 common (MAF > 0.05) SNP (rs3814960) wasidentified within the p16 basal promoter and CpG islandby searching the Environmental Genome Project database(http://egp.gs.washington.edu/). No significant associa-tion (P > 0.10) was found between rs3814960 and p16methylation in adenocarcinomas or sputum samples fromthe LSC indicating that not all genes are silenced bysequence dependent ASM.

Influence of the A/G genotype on MGMTtranscription, expression, and methylation

The effect of sequence variants in the promoter–enhan-cer region on transcription of MGMT was assessed inHBEC2, Calu-6, and H1299. The lowest promoter activity

Table 1. Association between SNPs and risk for MGMT methylation in lung adenocarcinoma and insputum from smokers

Samples/cohorts (n) SNPs Genotypes Methylated/unmethylated OR (95% CI)

Lung adenocarcinoma(n ¼ 169) rs1711646a G/G 39/66 1.0

G/T 15/35 0.7 (0.3–1.4)T/T 5/9 0.9 (0.2–3.2)

rs1625649a C/C 17/56 1.0C/A 32/42 2.4 (1.1–4.9)A/A 10/12 2.7 (1.0–7.6)

rs16906252a,b G/G 36/109 1.0G/A 23/1 97.0 (12.1–778.4)A/A Not detected –

SputumLSC rs16906252c G/G 207/853 1.0(n ¼ 1,256) G/A 113/75 6.3 (4.5–8.7)

A/A 6/2 12.4 (2.5–62.8)VSC rs16906252d G/G 82/332 1.0(n ¼ 475) G/A 39/21 8.0 (4.4–14.5)

A/A 1/0 Not calculatede

aAge, gender, race (white vs. nonwhite), and smoking status (ever vs. never) were used as adjustment factors in logistic regression.bDue to small sample size in one cell, exact logistic regression was used.cAge, gender, race, current smoking status, and pack years were used as adjustment factors in logistic regression in the LSC.dAge, race, current smoking status, and pack years were used as adjustment factors in logistic regression in VA smokers cohort.eOR for A/A genotype in the VA cohort was not calculated because no A/A genotype was identified in the unmethylated group.

Leng et al.





was seen in HAP4 carrying the A allele. With the mostcommon haplotype (HAP1) as the reference, a 20 to 41%reduction in promoter activity associated with HAP4 wasseen (Fig. 2A). In addition, expression of the A allele wasreduced 31 to 39% compared with the G allele in 2NHBECs from cancer-free smokers heterozygous forrs16906252 and unmethylated for MGMT (Fig. 2B). Incontrast, a previous study focused on a small fragment ofthe MGMT promoter (between SmaI and SstI in Supple-mentary Figure) and showed a 16 to 64% increase ofMGMT promoter–enhancer activity associated with A allele(33). However that small promoter construct generatedonly 13% promoter activity compared with that seen withthe 1.2 kbMGMT promoter tested by Harris and colleagues(20) and replicated in our studies using a slightly largerconstruct. Furthermore, our findings that are based on thehaplotype of the entire maximal promoter should be morereflective of the in vivo situation by allowing the effect of aSNP along with the complex interaction of the multipletranscription factors binding to the maximal MGMT pro-moter and the LD of multiple SNPs in this area to influencepromoter activity (34).Reduced gene transcription has been proposed as one

mechanism predisposing a gene for epigenetic silencing viapromoter methylation (35). Therefore, studies were con-

ducted to test whether cell lines with the A/G genotype havepreferential methylation of the A allele that in turn shouldlead to monoallelic expression of MGMT. Calu-6 andH1993were heterozygous for rs16906252 and hemimethy-lated forMGMT. Bisulfite sequencing found that only the Aallele is densely methylated and selectively silenced inCalu-6 and H1993 (Figs. 2B and 3A; not shown). TheASM of MGMT associated with rs16906252 was furtherconfirmed in primary lung tumors. An MSP assay targetingthe reverse strand of the MGMT promoter that distin-guishes the G versus the A allele was developed. Six A/Gtumorsmethylated forMGMTwere sequenced and only theA allele was methylated (Fig. 3B).

Targeting MGMT for lung cancer therapyTMZ is an alkylating chemotherapeutic agent that does

not cause severe systemic toxicity and significantlyextends survival in glioblastoma patients with methyla-tion of MGMT (19). If a similar response to TMZ depen-dent on MGMT methylation status could be replicated inlung cancer, this agent could potentially benefit a subsetof patients. This premise was tested by determining if lungtumor cell lines with differential MGMT methylation andexpression would have different sensitivity toward TMZtreatment. Expression of MGMT was quantified in 9 G/G

Figure 2. The A allele confersreduced transcription andexpression of MGMT. A,haplotype 4 (HAP4) carrying the Aallele of rs16906252 shows a 20to 40% reduction in promoteractivity in 3 lung cell linescompared with HAP1. Values aremean � SD from 3 experiments.B, a 153 bp fragment surroundingthe rs16906252 was amplifiedfrom cDNA isolated from these 2NHBECs and then digested byFspI. Densitometry analysisshowed that the expression ofthe A allele was reduced 31 to39% compared with the G allele,whereas rs16906252 dependentmethylation of MGMT leads toselective loss of A alleleexpression in Calu-6 and H1993.As expected, expression of onlythe G allele is seen in cell lines(HCCC4006, H23, and H2009)with G/G genotype thatare unmethylated for MGMT.






lung cancer cell lines with unmethylated MGMT andcompared to the average expression seen in 6 G/GNHBEC lines. Two unmethylated cell lines with highMGMT expression were selected to serve as controls(Table 2). All 3 of the G/G cancer cell lines with MGMTmethylation showed hemimethylation based on theCOBRA assay (not shown). Two G/G methylated celllines and 1 A/G cell line were selected and expressionlevels of MGMT were quantified. Two cell lines derivedfrom glioblastoma, 1 G/G fully methylated for MGMTshowing no expression (U251) and the second (T98G) G/G was sparsely methylated showing no reduction inexpression compared to NHBECs were included for com-parison of sensitivity to TMZ. The IC50 for TMZ wassimilar in the unmethylated glioblastoma and lung cancercell lines (1,590–2,500 mmol/L). As expected the fullymethylated glioblastoma cell line showed heightened

sensitivity to TMZ with an 83% reduction in IC50 com-pared to unmethylated cell lines (Table 2). The responseof the hemimethylated lung cancer cell lines was inter-mediate between fully methylated and unmethylatedtumor cell lines. A high correlation between TMZ IC50sand MGMT expression was observed for the lung cancercell lines (r ¼ 0.99; Table 2 and Fig. 4).

Discussion

This study provides strong evidence that the A allele of aMGMT promoter–enhancer SNP (rs16906252) is a keydeterminant in the acquisition of MGMT methylation inlung carcinogenesis. This association was found in primaryadenocarcinoma and in exfoliated cells in sputum fromcancer-free smokers, a finding that reflects the ongoing fieldcancerization within the aerodigestive tract. Themagnitudeof the association between rs16906252 andMGMTmethy-lation in premalignant lesions from lung cancer-free smo-kers was much smaller than observed from the lungadenocarcinoma patients, consistent with the increase inprevalence of MGMT methylation of 25% in premalignantlesions from smokers to 43% in stage I lung cancer and61% in stage II to IV lung cancer (36). The reducedtranscription and expression seen for the A allele ofMGMTmay be one predisposing factor for methylation. Thishypothesis is supported by the selective methylation andaccompanied silencing of the A allele in tumors from A/Gheterozygotes. Thus, some of the MGMT unmethylatedlung cancer-free smokers who carry the A allele shouldbe more prone to become methylated as their field cancer-ization progresses over time (32).

The lack of an association between rs16906252 with riskfor lung cancer based on the similar percentage of subjectscarrying the A allele in lung adenocarcinoma cases andcancer-free controls in the Lovelace and Veteran Cohorts(14.2%, 15.6%, and 12.8%, respectively) is not surprising.Our previous studies demonstrate that concomitant methy-lation of a panel of tumor suppressor genes includingMGMT in sputum could prospectively predict risk for lungcancer in moderate and heavy smokers (18, 32). MGMTmethylation alone in sputum only contributes to a portionof the predicted increased cancer risk. Thus, a large case-control study would be needed to more precisely estimatethe association of this SNP to risk for lung cancer. More-over, approximately 20% of subjects with G/G genotypealso have MGMT methylated in sputum, suggesting thatother genetic or environmental factors may also contributeto MGMT methylation in lung carcinogenesis, albeit prob-ably to a lesser degree than rs16906252. Thus, the predic-tion for lung cancer risk by MGMT methylation in sputumcould also vary by rs16906252 genotype. Testing thishypothesis will require long-term followup in our smokercohorts to accumulate incident lung cancer cases withsputum samples collected before cancer diagnosis.

The sequence dependent ASM of MGMT is notlung tumor-specific because similar associations werealso reported in colorectal cancer (21, 22). Studies in

Figure 3. ASM ofMGMT. A, bisulfite sequencing of COBRA product fromCalu-6 encompassing CpGs with locations from �5 to þ148 relative to thetranslation start site shows methylation of only the A allele. B, bisulfitesequencing of the MSP product in a primary lung tumor sampleencompassing CpGs with locations from þ27 to þ181 relative to thetranslation start site heterozygous for rs16906252 shows selectivemethylation of the A allele.

Leng et al.





glioblastoma did not identify an association betweenrs16906252 and risk for methylation (37). However, theprimers used for genotyping had several incorrect basesthat likely influenced the accuracy for identifying thevariant allele (37). Recently, methylation of GSTP1, RIL,andMSH2were found to be allele-specific in several tumortypes (11, 38, 39). Low-level methylation of MGMT wasalso seen in normal colon mucosa and peripheral lympho-cytes from heterozygote subjects, corroborating a likelyimportant role for this SNP in the susceptibility of the Aallele of MGMT for silencing (22, 40). Zhang and collea-gues studying methylation of gene promoters on chromo-some 21 estimated that ASM may affect 10% of humangenes in leukocytes (13). Thus, genetic differences caninfluence interindividual variation for DNA methylationof many genes in normal tissues and tumors.Evidence is accumulating that reduced transcription may

be one trigger for initiation and spreading of heterochro-

matin and methylation along a gene promoter (34). Songand colleagues (41) and Stirzaker and colleagues (42)showed that reduced transcription and seeds of methyla-tion within theGSTP1 promoter served as a stimulus for thespread of methylation and chromatin remodeling. In vivostudies found that breast cancer patients carrying theGSTP1 promoter haplotype E comprised of 4 SNPs andthe shortest ATAAA repeat had increased methylation (38).The polymorphisms within haplotype E disrupted bindingof c-Myb leading to reduced transcription. Our studies alsosupport reduced transcription as a mechanism for de novopromoter methylation based on the strong associationbetween reduced promoter activity associated with the Aallele in NHBEC and the targeting of this allele for silencingin lung tumors from heterozygote patients. Other mechan-isms independent of effects on gene transcription may alsoinfluence de novo promoter methylation. For example, a 12bp insertion polymorphism in the RIL promoter creates anSp1/Sp3 binding site that protects against methylation incancer and does not change expression in lymphoblastoidcell lines (11). The Sp1/Sp3 binding site created by thispolymorphism may through protein-DNA interactionsprevent spreading of DNA methylation and heterochro-matin from neighboring "methylation centers." In silicoprediction of transcription factor-binding sites usingTFSEARCH and MATCH programs did not identify anytranscription factors with core motifs binding rs16906252.In addition, resequencing the 1.9 Kb 50 region of theMGMTgene did not identify any SNP in perfect LD withrs16906252. Therefore, rs16906252 may influence thebinding of an unknown transcription factor or unlistedtranscription factor in these 2 databases. Recently, Hawkinsand colleagues examined 6 gene-associated CpG islandsflanking MGMT in colorectal tumors and found thatthe EBF3 gene located 500 kb downstream of MGMThad a similar methylation pattern to MGMT (22). Thus,it is possible that rs16906252 is in perfect LD withother polymorphisms present in the euchromatin and

Figure 4.MGMT expression and TMZ IC50s in 5 lung tumor cell lines (D) ishighly correlated (r ¼ 0.99).

Table 2. MGMT methylation status, mRNA expression, rs16906252 genotype, and TMZ IC50s

Cell line MGMT methylationa Genotype % Expressionb TMZ IC50c

LungHCC4006 Unmethylated G/G 162 1,655H2085 Unmethylated G/G 232 2,500Calu-6 Hemimethylated G/A 50 680Calu3 Hemimethylated G/G 64 693A549 Hemimethylated G/G 70 500

BrainU251 Fully methylated G/G 0 270T98G Sparsely methylated G/G 100 1,590

aMGMT methylation status was measured by COBRA.b% expression was the average from 3 cDNAs independently made from 1 mRNA sample and was relative to the average expressionseen in the NHBEC lines.cTMZ IC50 was the average from 3 independent experiments.






heterochromatin boundaries surrounding theMGMT and/or EBF3 which could affect gene regulation. Currently,testing this hypothesis will necessitate the completion ofthe 1,000 Genome Project that will provide a dense map ofgenetic variation by sequencing the genomes of 1,200people (43).

The refractory nature of unresected lung cancer to che-motherapeutic combinations tested over 3 decades neces-sitates developing targeted therapies based on underlyinggenetic and epigenetic tumor profiles. Responses have beenextremely positive for patients with mutations in the epi-dermal growth factor receptor receiving the tyrosine kinaseinhibitors gefitinib or erlotinib (44). Alkylating agents suchas 1,3-bis(2-chloroethyl)-1-nitrosourea were tested as che-motherapeutics for lung cancer decades ago with responserates limited by systemic toxicity (35, 45, 46). TMZ hasbeen evaluated in a phase II study for efficacy in pretreatedlung cancer, melanoma, and breast cancer patients withbrain metastasis. Treatment was provided without knowl-edge ofMGMTmethylation status, and 26%of patients hada clinical response with median overall survival greatest forlung cancer patients (47). Our in vitro studies in lung cancercell lines demonstrating sensitivity to TMZ that is highlycorrelated with levels of MGMT methylation suggest that a

subset of lung cancer patients with relatively low MGMTexpression could respond to this drug, a hypothesis thatrequires further preclinical studies prior to initiating clin-ical trials.

Disclosure of Potential Conflicts of Interest

S.A. Belinsky is a consultant to Oncomethylome Sciences. Under alicensing agreement between Lovelace Respiratory Research Instituteand Oncomethylome Sciences, nested MSP was licensed to Oncomethy-lome Sciences, and the author is entitled to a share of the royalties receivedby the Institute from sales of the licensed technology. The Institute, inaccordance with its conflict-of-interest policies, is managing the terms ofthese arrangements.

Grant Support

This work was supported by National Cancer Institute (R01 CA097356and R01 ES008801 to S.A. Belinsky and 1K01CA128823-01 to K.G. Flores)and the State of New Mexico as a direct appropriation from the TobaccoSettlement Fund to S.A. Belinsky.

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 November 11, 2010; revised January 13, 2011; acceptedFebruary 4, 2011; published OnlineFirst February 25, 2011.

References1. Belinsky SA. Silencing of genes by promoter hypermethylation: key

event in rodent and human lung cancer. Carcinogenesis2005;26:1481–7.

2. Belinsky SA, Klinge DM, Liechty KC, March TH, Kang T, Gilliland FD,et al. Plutonium targets the p16 gene for inactivation by promoterhypermethylation in human lung adenocarcinoma. Carcinogenesis2004;25:1063–7.

3. Cuozzo C, Porcellini A, Angrisano T, Morano A, Lee B, Di Pardo A,et al. DNA damage, homology-directed repair, and DNA methylation.PLoS Genet 2007;3:e110.

4. O’Hagan HM, Mohammad HP, Baylin SB. Double strand breaks caninitiate gene silencing and SIRT1-dependent onset of DNA methyla-tion in an exogenous promoter CpG island. PLoS Genet 2008;4:e1000155.

5. Leng S, Stidley C, Willink R, Bernauer A, Do K, Picchi MA, et al.Double-strand break damage and associated DNA repair genes pre-dispose smokers to gene methylation. Cancer Res 2008;68:3049–56.

6. Stidley CA, Picchi MA, Leng S, Willink R, Crowell RE, Flores KG, et al.Multivitamins, folate, and green vegetables protect against genepromoter methylation in the aerodigestive tract of smokers. CancerRes 2010;70:568–74.

7. McCabe MT, Lee EK, Vertino PM. A multifactorial signature of DNAsequence and polycomb binding predicts aberrant CpG islandmethy-lation. Cancer Res 2009;69:282–91.

8. Feltus FA, Lee EK, Costello JF, Plass C, Vertino PM. Predictingaberrant CpG island methylation. Proc Natl Acad Sci U S A2003;100:12253–8.

9. Lee TI, Jenner RG, Boyer LA, Guenther MG, Levine SS, Kumar RM,et al. Control of developmental regulators by polycomb in humanembryonic stem cells. Cell 2006;125:301–13.

10. Widschwendter M, Fiegl H, Egle D, Mueller-Holzner E, Spizzo G,Marth C, et al. Epigenetic stem cell signature in cancer. Nat Genet2007;39:157–8.

11. Boumber YA, Kondo Y, Chen X, Shen L, Guo Y, Tellez C, et al. An Sp1/Sp3 binding polymorphism confers methylation protection. PLoSGenet 2008;4:e1000162.

12. Kerkel K, Spadola A, Yuan E, Kosek J, Jiang L, Hod E, et al. Genomicsurveys by methylation-sensitive SNP analysis identify sequence-dependent allele-specific DNA methylation. Nat Genet2008;40:904–8.

13. Zhang Y, Rohde C, Reinhardt R, Voelcker-Rehage C, Jeltsch A. Non-imprinted allele-specific DNA methylation on human autosomes.Genome Biol 2009;10:R138.

14. 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.

15. 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.

16. Esteller M, Toyota M, Sanchez-Cespedes M, Capella G, Peinado MA,Watkins DN, et al. Inactivation of the DNA repair gene O6-methylgua-nine-DNA methyltransferase by promoter hypermethylation is asso-ciated with G to A mutations in K-ras in colorectal tumorigenesis.Cancer Res 2000;60:2368–71.

17. Yin D, Xie D, Hofmann WK, Zhang W, Asotra K, Wong R, et al. DNArepair gene O6-methylguanine-DNA methyltransferase: promoterhypermethylation associated with decreased expression and G:Cto A:T mutations of p53 in brain tumors. Mol Carcinog 2003;36:23–31.

18. Belinsky SA, Liechty KC, Gentry FD, Wolf HJ, Rogers J, Vu K, et al.Promoter hypermethylation of multiple genes in sputum precedeslung cancer incidence in a high-risk cohort. Cancer Res2006;66:3338–44.

19. 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.

20. Harris LC, Remack JS, Brent TP. Identification of a 59 bp enhancerlocated at the first exon/intron boundary of the human O6-methyl-guanine DNA methyltransferase gene. Nucleic Acids Res1994;22:4614–9.

Leng et al.





21. Ogino S, Hazra A, Tranah GJ, Kirkner GJ, Kawasaki T, Nosho K, et al.MGMT germline polymorphism is associated with somatic MGMTpromoter methylation and gene silencing in colorectal cancer. Carci-nogenesis 2007;28:1985–90.

22. Hawkins NJ, Lee JH, Wong JJ, Kwok CT, Ward RL, Hitchins MP.MGMTmethylation is associated primarily with the germline C>T SNP(rs16906252) in colorectal cancer and normal colonic mucosa. ModPathol 2009;22:1588–99.

23. TessemaM,Willink R, Do K, Yu YY, YuW,Machida EO, et al. Promotermethylation of genes in and around the candidate lung cancer sus-ceptibility locus 6q23–25. Cancer Res 2008;68:1707–14.

24. Chin LJ, Ratner E, Leng S, Zhai R, Nallur S, Babar I, et al. A SNP in alet-7 microRNA complementary site in the KRAS 3’ untranslatedregion increases non-small cell lung cancer risk. Cancer Res2008;68:8535–40.

25. TessemaM, Yu YY, Stidley CA, Machida EO, Schuebel KE, Baylin SB,et al. Concomitant promoter methylation of multiple genes in lungadenocarcinomas from current, former and never smokers. Carcino-genesis 2009;30:1132–8.

26. 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.

27. Stephens M, Smith NJ, Donnelly P. A new statistical method forhaplotype reconstruction from population data. Am J Hum Genet2001;68:978–89.

28. Stephens M, Donnelly P. A comparison of Bayesian methods forhaplotype reconstruction from population genotype data. Am JHum Genet 2003;73:1162–9.

29. Stram DO, Haiman CA, Hirschhorn JN, Altshuler D, Kolonel LN,Henderson BE, et al. Choosing haplotype-tagging SNPS based onunphased genotype data using a preliminary sample of unrelatedsubjects with an example from the Multiethnic Cohort Study. HumHered 2003;55:27–36.

30. Menashe I, Rosenberg PS, Chen BE. PGA: power calculator for case-control genetic association analyses. BMC Genet 2008;9:36.

31. Egyh�azi S, Ma S, Smoczynski K, Hansson J, Platz A, Ringborg U.Novel O6-methylguanine-DNA methyltransferase SNPs: a frequencycomparison of patients with familial melanoma and healthy individualsin Sweden. Hum Mutat 2002;20:408–9.

32. Belinsky SA. Gene-promoter hypermethylation as a biomarker in lungcancer. Nat Rev Cancer 2004;4:707–17.

33. Krzesniak M, Butkiewicz D, Samojedny A, Chorazy M, Rusin M.Polymorphisms in TDG and MGMT genes—epidemiological andfunctional study in lung cancer patients from Poland. Ann Hum Genet2004;68:300–12.

34. Wu H, Zhang X, Ling G, D'Agostino J, Ding X. Mechanisms ofdifferential expression of the CYP2A13 7520C and 7520G alleles inhuman lung: allelic expression analysis for CYP2A13 heterogeneousnuclear RNA, and evidence for the involvement of multiple Cis-reg-

ulatory single nucleotide polymorphisms. Pharmacogenet Genomics2009;19:852–63.

35. Cross SH, Bird AP. CpG islands and genes. Curr Opin Genet Dev1995;5:309–14.

36. Pulling LC, Divine KK, Klinge DM, Gilliland FD, Kang T, Schwartz AG,et al. Promoter hypermethylation of the O6-methylguanine-DNAmethyltransferase gene: more common in lung adenocarcinomasfrom never-smokers than smokers and associated with tumor pro-gression. Cancer Res 2003;63:4842–8.

37. Zawlik I, Vaccarella S, Kita D, Mittelbronn M, Franceschi S, Ohgaki H.Promoter methylation and polymorphisms of the MGMT gene inglioblastomas: a population-based study. Neuroepidemiology2009;32:21–9.

38. Rønneberg JA, Tost J, Solvang HK, Alnaes GI, Johansen FE, Bren-deford EM, et al. GSTP1 promoter haplotypes affect DNA methylationlevels and promoter activity in breast carcinomas. Cancer Res2008;68:5562–71.

39. Chan TL, Yuen ST, Kong CK, Chan YW, Chan AS, Ng WF, et al.Heritable germline epimutation of MSH2 in a family with heredi-tary nonpolyposis colorectal cancer. Nat Genet 2006;38:1178–83.

40. Candiloro IL, Dobrovic A. Detection of MGMT promoter methylation innormal individuals is strongly associated with the T allele of thers16906252 MGMT promoter single nucleotide polymorphism. Can-cer Prev Res (Phila Pa) 2009;2:862–7.

41. Song JZ, Stirzaker C, Harrison J,Melki JR, Clark SJ. Hypermethylationtrigger of the glutathione-S-transferase gene (GSTP1) in prostatecancer cells. Oncogene 2002;21:1048–61.

42. Stirzaker C, Song JZ, Davidson B, Clark SJ. Transcriptional genesilencing promotes DNA hypermethylation through a sequentialchange in chromatin modifications in cancer cells. Cancer Res2004;64:3871–7.

43. Via M, Gignoux C, Burchard EG. The 1000 Genomes Project: newopportunities for research and social challenges. Genome Med2010;2:3.

44. Rudin CM, Avila-Tang E, Harris CC, Herman JG, Hirsch FR, Pao W,et al. Lung cancer in never smokers: molecular profiles and thera-peutic implications. Clin Cancer Res 2009;15:5646–61.

45. Wolf J. Nitrosoureas as single agents in the treatment of pulmonarycancer. Cancer Treat Rep 1976;60:753–6.

46. Ahmann DL, Carr DT, Coles DT, Hahn RG. Evaluation of cyclopho-sphamide (NSC-26271) and 1,3-bis(2-chloroethyl)-1-nitrosourea(BCNU; NSC-409962) in the treatment of patients with inoperableor disseminated lung cancer. Cancer Chemother Rep 1972;56:401–3.

47. Siena S, Crinò L, Danova M, Del Prete S, Cascinu S, Salvagni S, et al.Dose-dense temozolomide regimen for the treatment of brain metas-tases from melanoma, breast cancer, or lung cancer not amenable tosurgery or radiosurgery: a multicenter phase II study. Ann Oncol2010;21:655–61.






Published OnlineFirst February 25, 2011.Clin Cancer Res Shuguang Leng, Amanda M. Bernauer, Chibo Hong, et al. Lung AdenocarcinomasSelectively Silenced in Premalignant Lesions from Smokers and in

Methylation and IsMGMTThe A/G Allele of Rs16906252 Predicts for

Updated version

10.1158/1078-0432.CCR-10-3026doi:

Access the most recent version of this article at:

Material

Supplementary

http://clincancerres.aacrjournals.org/content/suppl/2011/03/31/1078-0432.CCR-10-3026.DC1

Access the most recent supplemental material at:

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

Subscriptions

Reprints and

[email protected]

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

Permissions

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

.http://clincancerres.aacrjournals.org/content/early/2011/03/25/1078-0432.CCR-10-3026To 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-10-3026

http://clincancerres.aacrjournals.org/content/suppl/2011/03/31/1078-0432.CCR-10-3026.DC1

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

mailto:[email protected]

http://clincancerres.aacrjournals.org/content/early/2011/03/25/1078-0432.CCR-10-3026


the a/g allele of rs16906252 predicts for mgmt methylation … · (vsc) began recruitment of...

Documents