Oral Oncology (2007) 43, 515–522

ava i lab le a t www.sc iencedi rec t . com

journal homepage: ht tp : / / in t l .e lsevierheal th .com/ journals /oron/

MGMT expression in oral precancerous andcancerous lesions: Correlation with progression,nodal metastasis and poor prognosis

Meenakshi Sawhney a,1, Nidhi Rohatgi a,1, Jatinder Kaur a,Siddhartha D. Gupta b, Suryanaryana V.S. Deo c,Nootan K. Shukla c, Ranju Ralhan a,*

a Department of Biochemistry, All India Institute for Medical Sciences, Ansari Nagar,New Delhi 110 029, Indiab Department of Pathology, All India Institute for Medical Sciences, Ansari Nagar,New Delhi 110 029, Indiac Department of Surgical Oncology, Institute of Rotary Cancer Hospital, All India Institute for Medical Sciences,Ansari Nagar, New Delhi 110 029, India

Received 15 April 2006; received in revised form 6 May 2006; accepted 9 May 2006Available online 25 September 2006

Summary Alkylation of DNA at the O6 position of guanine is a critical step in the induction ofmutations by carcinogenic and chemotherapeutic alkylating agents. O6-methylguanine–DNAmethyltransferase (MGMT) is an enzyme that removes mutagenic adducts from the O6 positionof guanine, thereby protecting the genome against guanine to adenine transitions. We hypoth-esized that alteration in MGMT expression might occur in early stages of development of oralcancer and be associated with disease progression. Immunohistochemical analysis of MGMTexpression was carried out in 107 oral squamous cell carcinomas (OSCCs), 78 oral precancerouslesions (OPLs) (58 hyperplasias and 20 dysplasias) and 30 histologically normal oral tissuesand correlated with clinicopathological parameters as well as major risk factors. DecreasedMGMT expression was observed as early as in hyperplasia (p = 0.003; Odd’s Ratio (OR) = 5.0).

KEYWORDSMGMT;Oral Squamous cellcarcinomas;Oral precancerous lesions;Smokeless tobacco

1d

nWn

368-8375/$ - see front matter �c 2006 Elsevier Ltd. All rights reserved.oi:10.1016/j.oraloncology.2006.05.007

Abbreviations: MGMT, O6-methyl-DNA methyltransferase; NNN, N-nitrosonornicotine; NNK, 4-(methylnitrosamino)-1-(3-pyridyl)-1-buta-one; OPL, oral precancerous lesion; OSCC, oral squamous cell carcinoma; ST, smokeless tobacco; TSNA, tobacco specific nitrosamine;DSCC, well differentiated squamous cell carcinoma; MDSCC, moderately differentiated squamous cell carcinoma; PDSCC, poorly differe-tiated squamous cell carcinoma; BQT, betel quid with tobacco.* Corresponding author. Tel.: +91 11 26593478; fax: +91 11 26588663.E-mail addresses: ralhanr@hotmail.com, ralhanr@rediffmail.com (R. Ralhan).

1 These authors have contributed equally.

516 M. Sawhney et al.

Significant loss of MGMT expression was observed from hyperplasia to dysplasia (p = 0.034;OR = 4.0). Loss of MGMT expression was associated with late clinical stage of OSCCs(p = 0.027, OR = 2.0) and nodal metastasis (p = 0.031, OR = 2.5). Decreased MGMT expressionwas associated with smokeless tobacco (ST) consumption in patients with OPLs (p = 0.017,OR = 3.6) and OSCCs (p = 0.031, OR = 2.8). Significant association was also observed betweenloss of MGMT expression and poor prognosis of OSCC patients (p = 0.02; OR = 5.2). Thedecreased MGMT expression in OPLs suggested that deregulation of MGMT expression is an earlyevent in the development of oral cancer. In OSCCs, its correlation with late clinical stage, andnodal metastasis suggests association with aggressive tumor behavior and cancer progression,underscoring its potential as a candidate predictive marker for nodal metastasis and diseaseprognosis. Correlation of loss of MGMT expression with ST consumption underscored its signif-icance in ST-associated oral carcinogenesis.

�c 2006 Elsevier Ltd. All rights reserved.

Introduction

Unquestionably, the use of tobacco products is the majorcause of oral cancer worldwide. Globally over 400,000new Oral cancers are reported per year; of which about100,000 cases occur in India alone.1 There is an alarming in-crease in use of smokeless tobacco (ST) products, especiallyin South Asia.2,3 Recently ST has been classified as ‘Carcin-ogenic to humans’.4 The development of tobacco associatedOSCC is a multistep process involving interactions betweenseveral factors such as tobacco associated intra-oral carcin-ogens, areca nut, betel quid and alcohol consumption, and/or viral infections.5 In India tobacco chewing is prevalent,usually in the form of betel quid with tobacco. Betel quidconsists of leaf of Piper betel vine, coated with lime andcatechu wrapped around areca nut and tobacco. Slakedlime is an essential ingredient; as it lowers the pH andaccelerates the release of the alkaloids from both tobaccoand nut, with enhanced pharmacological effect. OSCCsare often preceded by clinically evident OPLs, often leuko-plakia, and the risk of multiple cancers is 5–10 times great-er in patients with OSCCs preceded by leukoplakia.6 TheseOPLs are more common in chewing tobacco related oralcancer in India. If untreated, 10–15% of leukoplakias willdevelop into cancer over a period of time.7–9 Based on sev-eral studies 5% transformation in five years resulting in anaverage annual transformation rate of 1% has beenproposed.10

Biologically, the most important DNA lesions produced byboth carcinogenic and chemotherapeutic alkylating agentssuch as tobacco specific nitrosamines (TSNAs), nitrosoureasand related compounds, are O6-alkylguanine adducts.11,12

The ability of these adducts to pair with thymine insteadof cytosine during DNA replication is responsible for the in-crease in frequency of transition mutations. O6-methylgua-nine–DNA methyltransferase (MGMT) is a DNA repairenzyme involved in protection of the cellular genome frommutagenic actions of endogenous and environmental carcin-ogens. Abnormal MGMT expression causes O6-methylguanineto accumulate in cellular DNA, and this could result in acti-vation of oncogenes or inactivation of tumor suppressorgenes, contributing to carcinogenesis or tumor progres-sion.13 MGMT functions by a stoichiometric and suicidalreaction mechanism in which the methyl groups bound tothe O6-position of guanine are transferred to a cysteine in

its active site, resulting in direct restoration of the normalbase and self-inactivation of MGMT.13,14

MGMT is ubiquitously expressed in normal human tissues,although its levels vary considerably between organs andindividuals.15 However, tumors are heterogeneous with re-spect to MGMT expression and in a subset of tumors includinghead and neck cancers, its expression is silenced mostly dueto abnormal promoter methylation.14,16 MGMT is reportedlya strong predictor of survival in various solid tumors.17 Tu-mors with little or no MGMT are predictably responsive todrug therapy; conversely, tumors with high MGMT levelsare probably drug resistant. Moreover, depletion of MGMTactivity with the substrate analog inhibitor O6-benzylguaninehas been shown to increase the rate of cell death by alkylat-ing nitrosoureas in human glioma-derived cell lines.18

Recently, our laboratory showed modulation of MGMTprotein expression by ST extracts in oral epithelial cell cul-ture from a precancerous stage (AMOL-III) as an importantevent in the growth promoting effects of ST (khaini), whichmay lead the cells down the carcinogenic pathway.19 Pro-moter methylation of MGMT has been reported in chewingtobacco associated OSCC in Indian population.20 However,there is no report on MGMT protein expression in these can-cers. Immunohistochemistry is a simple method for detec-tion of protein expression and it can be easily done onarchival paraffin-embedded specimens. In addition, it al-lows for the evaluation of not only of the degree of staining,but also localization of the target protein in individual cells.Taking these considerations into account, the aim of thisstudy was to evaluate the clinical significance of alterationin MGMT protein expression in oral precancerous stages andOSCCs using immunohistochemical analysis and determinetheir correlation with tobacco consumption habits of thesepatients and disease prognosis.

Materials and methods

Tissue specimens

This study was approved by Human Ethics Committee of AllIndia Institute of Medical Sciences, New Delhi, India. Forimmunohistochemical analysis, surgically resected tissuesand biopsy specimens from OSCCs, OPLs and histologicallynormal oral tissues were obtained from Department of Surgi-cal Oncology, Dr. B.R. Ambedkar Institute of Rotary Cancer

MGMT expression in oral carcinogenesis 517

Hospital, All India Institute of Medical Sciences, New Delhi,India, with prior informed written consent of the patients.Normal human oral tissues were also collected from theOut Patients Department of Dental Surgery, All India Insti-tute of Medical Sciences, New Delhi, with prior informedwritten consent of the patients. After excision, tissue spec-imens were formalin fixed and paraffin-embedded for histo-pathological examination and immunohistochemicalanalysis. The clinical and pathological data were recordedusing a pre-designed Performa as described previously.21

Clinicopathological characteristics of patients

One hundred and seven primary OSCC patients ranging inage from 29 to 75 years (median age 53 years) diagnosedduring 2002–2005 were enrolled in this study. The diagnosiswas based on clinical examination and histopathologicalanalysis of the tissue specimens. The TNM staging was donefollowing the UICC staging system.22 The tumors were histo-logically graded as well, moderately or poorly differentiatedSCCs. Seventy-eight OPLs that were histologically confirmedas hyperplasia (58 cases) or dysplasia (20 cases) were in-cluded in this study. Of the 30 normal tissues, 15 non-malig-nant oral tissues were collected from an area adjacent tothe site of tumor and histologically confirmed to have nor-mal epithelium. Fifteen normal oral tissues were collectedfrom patients with no clinically evident oral premalignantor malignant lesions attending the Dental Clinic and histo-logically confirmed to have normal oral epithelium.

Immunohistochemistry

Paraffin embedded sections (5 lm thickness) of human oraltissue specimens were collected on gelatin-coated slidesand formalin fixed. For histopathological analysis, the rep-resentative sections were stained with hematoxylin and eo-sin, whereas immunostaining was done on serial sections asdescribed previously21. Briefly, the sections were deparaff-inized and rehydrated. Thereafter, antigen retrieval wascarried out using a microwave oven in 0.01 M-citrate buffer,pH 6.0. Endogenous peroxidase activity was blocked byincubating sections in methanol containing hydrogen perox-ide (0.3%, v/v) for 20 min. Non-specific binding was blockedwith 1% (w/v) BSA in PBS for 1 h followed by incubation withanti-MGMT polyclonal antibody (0.2 lg/ml) (Santa Cruz Bio-technology, CA) for 16 h at 4 �C. The anti-MGMT antibodywas detected using biotinylated secondary antibody andperoxidase labelled Streptavidin complex using Dako LSABplus kit (Dako Co., Denmark). The color was developed usingthe chromogen, diaminobenzidine (DAB). Finally, the slideswere counter stained with Mayer’s hematoxylin andmounted with D.P.X. mountant. Parallel sections in whichthe primary antibody was replaced by non-immune mouseIgG of the same isotype were examined to ensure specificityand exclude cross reactivity between the antibodies andconjugates used (data not shown).

Positive criterion for immunohistochemical staining

The immunopositive staining was evaluated in five areas ofthe tissue sections. For MGMT protein expression, specific

staining in the tumor cell nucleus was defined as positivestaining. The immunoreactivity of MGMT protein was evalu-ated semiquantitatively by counting the number of positivecells and <10% was regarded as negative. Thus P10% tumorcells showing immunoreactivity were graded as immunopos-itive. The immunohistochemical investigation was blind,i.e., the slides were coded and pathologist did not haveprior knowledge of the local tumor burden, lymphonodularspread and grading of the OSCCs while scoring theimmunoreactivity.

Follow-up study

Forty nine of 107 OSCC patients could be followed up. Sur-vival status of patients was verified and regularly updatedfrom Tumor Registry records of hospital as of January2006. Patients were monitored for a period of median 18months and a maximum of 42 months. As per our protocolOSCC patients with T1 and T2 tumors were treated with rad-ical radiotherapy (RT) or Surgery (Sx) alone, where as major-ity of patients with T3 and T4 disease were treated using acombination of radical Sx followed by postoperative RT. Allpatients undergoing surgery had tumor free margins (Roresection). Radical Sx consisted of radical resection of thetumor with 1–2 cm three dimensional tumor free marginsand an appropriate neck dissection based on nodal status.The RT regime comprised of a total dose of 60–70 Gy in25–30 fractions over a period of 5–6 weeks. The chemother-apy regimen included two cycles of infusion of five fluoroura-cil 500 mg/m2 on day one and cisplatin 30 mg/m2 days 1–4.The cycle was repeated after three weeks. The patientswere followed up periodically and time to recurrence was re-corded. A new cancer developing after more than threeyears of treatment of the primary tumor in a different ana-tomical location was considered as a second primary tumor(SPT). A local recurrence was defined as any tumor of similarhistology appearing within 2 cm or three years of the primarytumor. Disease relapsing in the treated neck was taken as aregional relapse. No distant metastasis was observed in thisstudy. If a patient died during the follow-up, patient survivaltime was censored at the time of death. Medical history,clinical examination and radiological evaluation were usedto determine whether death resulted from recurrent cancer(relapsing patients) or from any other cause. Disease freesurvivors were defined as patients free from clinical andradiological evidence of local, regional or distant relapseat the time of the last follow up. Loco regional relapse/death was observed in 15/49 (31%) patients monitored in thisstudy. Patients who did not show recurrence were alive tillthe end of the follow up period. Among the 58 patients thatwere lost to follow up, the number of deaths could not beascertained, therefore overall survival could not be consid-ered as a separate parameter in our study. Only disease freesurvival of the patients was studied. Disease free survivalwas expressed as the number of months from the date of sur-gery to the loco regional relapse.21

Statistical analysis

Immunohistochemical data were subjected to statisticalanalysis using SPSS 10.0 software. The relationship between

518 M. Sawhney et al.

MGMT expression and clinicopathological parameters wastested by Chi-Square or Fisher’s exact test. Follow-up stud-ies were analyzed by Kaplan–Meier, and Cox’s ProportionalHazards test. The association between patient outcome andvariables was assessed by log–rank test. Two sided p valueswere calculated and p 6 0.05 was considered to besignificant.

Results

Immunohistochemical analysis of MGMTexpression in normal oral tissues and OPLs andClinical correlations

Loss of MGMT expression was observed in different stages oforal carcinogenesis (normal to precancer to invasive cancer)(p = 0.000) (Table 1). MGMT protein was localized predomi-nantly in nuclei of epithelial cells, through occasionallycytoplasmic staining was also observed. MGMT immunoposi-tivity was observed in 14/15 (93%) normal oral tissues ob-tained from individuals with no evidence of OPLs or OSCCs(Fig. 1A) and in 11/15 (73%) of histologically normal oral tis-sues taken from an area adjacent to the site of OSCC(Fig. 1B). In the oral epithelium, nuclei were positivelystained particularly in the actively proliferating basal andparabasal cells, indicative of the presence of endogenousMGMT protein. Overall 45/78 (58%) OPLs did not showdetectable MGMT immunostaining (Table 1). Significant lossof MGMT expression was observed from normal oral tissuesto OPLs (p = 0.000; OR = 6.8) and from normal oral tissues(25/30 (17%)) to hyperplastic lesions (29/58 (50%))(p = 0.003; OR = 5.0). Figure 1C shows nuclear immunoposi-tivity of MGMT in a hyperplastic lesion while, Figure 1D de-picts loss of MGMT in a representative hyperplastic lesion.Significant loss of MGMT expression was also observed fromhyperplasia to dysplasia (16/20 (80%)) (p = 0.034; OR = 4.0).Figure 1E and 1F show nuclear MGMT expression and nodetectable MGMT expression in representative dysplastic le-sions respectively.

Interestingly, loss of MGMT expression was significantlyassociated with tobacco consumption in OPLs patients(42/67 cases, 63%; p = 0.046, OR = 4.5; Table 1). Whenstratified further for consumption of different forms of to-bacco, use of ST products only was found to be significantlyassociated with loss of MGMT expression; 20/26, 77% pa-tients showed the loss the of protein (p = 0.017; OR = 3.6;Fig. 2). Among the different ST products used, loss of MGMTexpression in OPLs significantly correlated with khaini con-sumption habits (p = 0.02, OR = 4.4) (Fig. 2).

MGMT expression in OSCCs: Correlation withclinicopathological parameters and habits

Loss of MGMT expression was observed in 56/107 (52%)OSCCs (Table 1). Figure 1G shows nuclear MGMT expressionin a well differentiated OSCC. Figure 1H depicts loss ofMGMT in a well differentiated OSCC. Loss of MGMT expres-sion was associated with late clinical stage of OSCCs(p = 0.027, OR = 2.0) (Table 1). Loss of MGMT expressionwas also associated with nodal metastasis (p = 0.031,OR = 2.5). Loss of MGMT expression was significantly associ-

ated with ST consumption in OSCC patients; (21/30, 70%)(p = 0.031; OR = 2.8) (Fig. 2). Among the different ST prod-ucts used; consumption of betel quid with tobacco (BQT)emerged to be an important risk factor, since the habitcould be significantly correlated with loss of MGMT expres-sion (p = 0.004, OR = 5.9) in OSCC patients (Fig. 2).

Association of MGMT expression with diseaseoutcome

Forty nine of 107 OSCC patients could be followed up for amaximum period of 42 months (median 18 months). Signifi-cant association was observed between loss of MGMT immu-nopositivity in OSCCs and reduced disease free-survival(p = 0.02; OR = 5.2; 95% Confidence Intervals = 0.04–0.8)(Fig. 3).

Discussion

This study, to our knowledge, is the first report demonstrat-ing reduced MGMT expression in OPLs, early stages of devel-opment of oral cancer and its correlation with ST use.Furthermore, in OSCCs, its loss is associated with advancedstage, nodal metastasis, poor prognosis as well as ST con-sumption. Reduced MGMT expression was observed as earlyas in hyperplasia stage as compared to normal oral tissues(NPN) obtained from subjects with no evidence of OPLs orOSCCs. However, reduced MGMT expression was observedin 4/15 (27%) paired histologically normal (PN) tissues thatmay be attributed to the field cancerization in the oral mu-cosa adjacent to tobacco consuming patients. Our observa-tions are also supported by another study in OPLs and inchewing tobacco associated OSCCs that reported promotermethylation of MGMT in Indian population in 26.7% normalmucosa of OSCC patients.20 Interestingly, in OPLs, signifi-cant loss of MGMT expression was also observed from hyper-plasia to dysplasia, suggesting that alteration in MGMTexpression is an early event in oral tumorigenesis. However,the role of MGMT in oral cancer development and progres-sion remains to be determined. Nevertheless, our clinicalfindings demonstrate the association of loss of MGMTexpression in OSCCs with late clinical stage, and nodalmetastasis, suggesting that deregulation of MGMT expres-sion is sustained during carcinogenic pathway. The presenceof lymph node metastasis is the most important prognosticfactor in head and neck cancer; survival rates may decreaseby about 50%, and the possibility of distant metastasis mayincrease with cervical node involvement.23 Reduced MGMTexpression has been found to be associated with progressionand nodal metastasis in other human cancers as well,including gastric and bilary tract cancers.24–26. Zuoet al.16 observed loss of MGMT expression in 20.4% headand neck squamous cell carcinomas but did not observeany association between loss of MGMT expression and clini-copathological parameters. In comparison, higher MGMTloss observed in our study may be attributed to the habitof ST consumption in this population.

Our results are also supported by similar differences infrequency of hypermethylation of MGMT gene observed be-tween western (18.2%) and Indian population (51.7%).16,20

Table 1 Analysis of MGMT protein expression in Normal oral tissues, OPLs and OSCCs: Correlation with clinicopathological parameters and tobacco consumption habits

Clinicopath-ologicalfeatures

Totalcases N

MGMTnuclearloss

p value O.R 95% C.I.Lower–Upper

Totalcases N

MGMTnuclearloss

pvalue

O.R 95% C.I.Lower–Upper

n (%) n (%)

Normal (N) 30 5 (17) N/OPLs 0.000 6.8 1.7–14.9 N/OSCCs 0.001 5.5 2.0–15.5NPN 15 1 (7) NPN/OPL 0.000 19.1 2.4–152 NPN/OSCCs 0.001 15.4 1.9–121PN 15 4 (27) PN/OPLs 0.046 3.6 1.1–12.8 PN/OSCCs 0.096 3.0 0.9–10.1

OPLs OSCCs

78 45 (58) 107 56 (52)Age (years)\40 23 11 (48) 12 6 (50)P40 55 34 (62) 95 50 (53)

GenderMale 66 41 (62) 85 44 (52)Female 12 4 (33) 22 12 (55)

Differentiation DifferentiationHyperplasia (H) 58 29 (50) N/H 0.003 5.0 1.7–14.9 WDSCC 63 32 (51)

NPN/H 0.003 14.0 1.7–114 MDSCC + PDSCC 44 24 (55)Dysplasia (D) 20 16 (80) PN/H 0.148 2.7 0.7–9.6 Tumor Stage

H/D 0.034 4.0 1.2–13.4 T1 + T2 50 23 (46)T3 + T4 57 33 (58)Nodal StatusN0 61 26 (43) N0/N1 0.031 2.5 1.1–5.6N1 46 30 (65)StageI + II 40 15 (38) I + II/III + IVa 0.027 2.0 1.2–5.9III + IVa 67 41 (61)

SiteBuccalMucosa

53 29 (55) 36 19 (53)

Tongue 12 6 (50) 35 14 (40)Alveolus 5 2 (40) 12 9 (75)Lip 6 6 6 5 (83)Othersa 2 2 18 9 (50)

HabitsNo Habit 11 3 (27) 21 11 (39)Tobacco#

consumer67 42 (63) T/NC 0.046 4.5 1.1–18.5 79 45 (57)

ST only 26 20 (77) ST/NST 0.017 3.6 1.2–10.4 30 21 (70) ST/NST 0.031 2.8 1.2–6.9S only 24 11 (46) 22 9 (41)ST + S 17 11 (65) 27 15 (56)#Overlapping habits, N: normal oral tissues, NPN: non-paired normal oral tissues, PN : paired normal oral tissues,H: hyperplasia, D: dysplasia, T: tobacco consumer, NC: non-tobaccoconsumer, ST: ST consumers only, NST : non-ST consumers S: tobacco smoking, WDSCC well differentiated squamous cell carcinoma,MDSCC: moderately differentiated squamous cellcarcinoma, PDSCC: poorly differentiated squamous cell carcinoma, 95% C.I.: 95% confidence intervals.v2 trend analysis: N to OPLs to OSCCs p = 0.000; NPN to OPLs to OSCCs p = 0.001; PN to OPLs to OSCCs p = 0.064; N to D to H to OSCCs p = 0.001; NPN to H to D to OSCCs p = 0.002; PN to H to Dto OSCCs p = 0.088.a Others include retromolar trigone, palate, and floor of mouth.

MGMTexp

ressio

nin

oral

carcinoge

nesis

519

Figure 1 Representative oral tissue sections immunostained for MGMT protein: (A) Non paired normal oral tissue showing nuclearimmunoreactivity, (B) paired normal oral tissue depicting nuclear staining, (C) hyperplastic lesion depicting nuclear staining, (D)hyperplastic lesion showing no nuclear staining for MGMT protein, (E) dysplastic lesion depicting nuclear staining, (F) Dysplasticlesion showing no nuclear staining for MGMT protein, (G) OSCC section depicting nuclear staining, (H) OSCC section with nodetectable MGMT levels, (I) positive control showing immunocytochemical analysis in HSC-2 cells. Original magnification X 200.

Figure 2 Graphical representation of correlation of MGMTexpression with Smokeless Tobacco habits.

Figure 3 Kaplan–Meier survival analysis for loss of MGMTexpression. Note the shorter survival time in the MGMT�

patients (33 months) compared with the MGMT+ patients (42months).

520 M. Sawhney et al.

Promoter hypermethylation mediated silencing of MGMT hasbeen reported in tobacco associated OSCCs20,27,28 and oralrinses of leukoplakia patients.29 Kulkarni and Saranath dem-onstrated that epigenetic alteration of MGMT is a frequentevent in chewing tobacco (ST) associated oral cancer(51.7%), and is an early event observed in normal oral muco-sa (26.7%) of the patients in the Indian population.20 In con-trast, no methylation was detected in buccal scrapings fromnormal healthy individuals with no chewing tobacco habits.Russo et al.28 suggested that targeted DNA methylation

mediated silencing of MGMT occurs in later stages in tobac-co smoke-induced neoplastic progression of lung cancer. Incontrast, our immunohistochemical data and promoter

MGMT expression in oral carcinogenesis 521

methylation data from another Indian laboratory 20 sug-gested that loss of MGMT expression is an early event inST associated oral carcinogenesis. We propose that this dif-ference may occur due to consumption of different forms oftobacco, predominantly chewing forms of tobacco in Indianpopulation, in comparison with smoking in the Western pop-ulation. This rationale is further supported by the findings inour study that loss of MGMT expression in OPLs and OSCCswas associated with ST consumption. Furthermore, thein vitro experimental evidence was provided in a previousreport, wherein we showed that exposure of OPL cells(AMOL-III) to ST (khaini) extracts enhanced cell proliferationat low doses and showed reduced expression of MGMT ascompared to the untreated cells, corroborating the clinicaldata of the present study.19

In our study, among the different forms of ST products,chewing of khaini and betel quid with tobacco emerged tobe significantly associated with MGMT loss in OPLs andOSCCs respectively. Khaini, a mixture of tobacco and limehas been reported to contain the highest levels of Tobac-co-specific nitrosamines (TSNAs) among the 32 Indian STproducts analyzed.30 How tobacco affects MGMT expressionremains largely unknown. ST products contain carcinogens,such as N-nitrosonornicotine (NNN), 4-(methylnitrosamino)-1-(3-pyridyl)-1-butanone (NNK), alkaloids and precursors ofreactive oxygen species. ST carcinogens form adducts withDNA and hemoglobin, which enhance the risk of cancerdevelopment.31–34 Significant association between reducedexpression of MGMT and ST use observed in our study sug-gests that defects in DNA repair mechanisms may partly ac-count for accumulation of DNA adducts and increasedgenomic instability observed in ST associated OPLs andOSCCs. Liu et al.34 showed inhibition of MGMT activityin vitro by complex mixtures present in the saliva of tobac-co and betel nut chewers. Exposure of buccal cell culturesto various organic and water based extracts of products re-lated to the use of tobacco and betel quid was shown to de-crease MGMT activity.34

In conclusion, our study showed decrease in MGMTexpression to be an early event in oral tumorigenesis, occur-ring as early as in hyperplasia. Futhermore, loss of MGMTexpression is sustained during the development and progres-sion of OSCC and is associated with reduced disease freesurvival. These findings underscore its validation as a candi-date predictive prognostic biomarker in oral cancer.

Acknowledgements

This work was supported by a research grant from theDepartment of Biotechnology, India. M.S, N.R and J.K. arethe recipients of Fellowships of the Council of Scientificand Industrial Research, India.

References

1. GLOBOCAN 2002. Cancer incidence, Mortality and prevalenceworldwide, Version 1.0 IARC Cancer Base No.5, Lyon, IARCPress, 2005.

2. Warnakulasuriya S. Smokeless tobacco and oral cancer. Oral Dis2004;10:1–4.

3. Gupta PC, Ray CS. Smokeless tobacco and health in India andSouth Asia. Respirology 2003;8:419–31.

4. Cogliano V, Straif K, Baan R, Grosse Y, Secretan B, Ghissassi F.Smokeless tobacco and related nitrosamines. Lancet Oncol2004;5:708.

5. Francesschi S, Bidoli E, Herrero R, Munoz N. Comparison ofcancers of the oral cavity and pharynx worldwide: etiologicalclues. Oral Oncol 2000;36:106–15.

6. Scully C, Porter S. ABC of oral health. Swellings and red, white,and pigmented lesions. BMJ 2000;321:225–8.

7. Gupta PC, Mehta FS, Pindborg JJ, Bhonsle RB, Murti PR, DaftaryDK, et al. Primary prevention trial of oral cancer in India: a 10-year follow-up study. J Oral Pathol Med 1992;21:433–9.

8. Tradati N, Grigolat R, Calabrese L, Costa L, Giugliano G, MorelliF, et al. Oral leukoplakias: to treat or not? Oral Oncol1997;33:317–21.

9. Gupta PC, Mehta FS, Daftary DK, Pindborg JJ, Bhonsle RB,Jalnawalla PN, et al. Incidence rates of oral cancer and naturalhistory of oral precancerous lesions in a 10-year follow-up studyof Indian villagers. Community Dent Oral Epidemiol1980;8:283–333.

10. Hunter KD, Parkinson EK, Harrison PR. Profiling early head andneck cancer. Nat Rev Cancer 2005;5:127–35.

11. Saffhill R, Margison GP, O’Connor PJ. Mechanisms of carcino-genesis induced by alkylating agents. Biochim Biophys Acta1985;823:111–45.

12. Singer B. Alkylation of the O6 of guanine is only one of manychemical events that may initiate carcinogenesis. Cancer Invest1984;2:233–8.

13. Pegg AE. Mammalian O6-alkylguanine-DNA methyltransferaseregulation and importance in response to alkylating carcino-genic and therapeutic agents. Cancer Res 1990;50:6119–29.

14. Soejima H, Zhao W, Mukai T. Epigenetic silencing of the MGMTgene in cancer. Biochem Cell Biol 2005;83:429–37.

15. Margison GP, Povey AC, Kaina B, Santibanez Koref MF.Variability and regulation of O6-alkylguanine-DNA alkyltrans-ferase. Carcinogenesis 2003;24:625–35.

16. Zuo C, Ai L, Ratliff P, Suen JY, Hanna E, Brent TP, et al. O6-methylguanine-DNA methyltransferase gene: epigenetic silenc-ing and prognostic value in head and neck squamous cellcarcinoma. Cancer Epidemiol Biomarkers Prev 2004;13:967–75.

17. Madhusudan S, Middleton MR. The emerging role of DNA repairproteins as predictive, prognostic and therapeutic targets incancer. Cancer Treat Rev 2005;31:603–17.

18. Bobola MS, Berger MS, Silber JR. Contribution of O6-methyl-guanine-DNA methyltransferase to resistance to 1,3-(2-chloro-ethyl)-1-nitrosourea in human brain tumor-derived cell lines.Mol Carcinogen 1995;13:81–8.

19. Rohatgi N, Kaur J, Srivastava A, Ralhan R. Smokeless tobacco(khaini) extracts modulate gene expression in epithelial cellculture from an oral hyperplasia. Oral Oncol 2005;41:806–20.

20. Kulkarni V, Saranath D. Concurrent hypermethylation of multi-ple regulatory genes in chewing tobacco associated oralsquamous cell carcinomas and adjacent normal tissues. OralOncol 2004;40:145–53.

21. Arora S, Kaur J, Sharma C, Mathur M, Bahadur S, Shukla NK,et al. Stromelysin 3, Ets – 1 and vascular endothelial growithfactor expression in oral precancerous and cancerous lesions:correlation with microvessel density, progression and progno-sis. Clin Cancer Res 2005;11:2272–84.

22. Union Internationale Centre Le Cancer. TNM classification ofmalignant tumors. third ed. Geneva: UICC; 1988.

23. Myers JN, Greenberg JS, Mo V, Roberts D. Extracapsular spread:a significant predictor of treatment failure in patients withsquamous cell carcinoma of the tongue. Cancer (Phila)2001;92:3030–6.

24. Kitajima Y, Miyazaki K, Matsukura S, Tanaka M, Sekiguchi M.Loss of expression of DNA repair enzymes MGMT, hMLH1 and

522 M. Sawhney et al.

hMSH2 during tumor progression in gastric cancer. GastricCancer 2003;2:86–95.

25. Matsukura S, Miyazaki K, Yakushiji H, Ogawa A, Harimaya K,Nakabeppu Y, et al. Expression and prognostic significance ofO6-methylguanine-DNA methyltransferase in hepatocellular,gastric and breast cancers. Ann Surg Oncol 2001;8:807–16.

26. Kohya N, Miyazaki K, Matsukura S, Yakushiji H, Kitajima Y,Kitahara K, et al. Deficient expression of O(6)-methylguanine-DNA methyltransferase combined with mismatch-repair pro-teins hMLH1 and hMSH2 is related to poor prognosis in humanbiliary tract carcinoma. Ann Surg Oncol 2002;9:371–9.

27. Viswanathan M, Tsuchida N, Shanmugam G. Promoter hyper-methylation profile of tumor-associated genes p16, p15,hMLH1, MGMT and E-cadherin in oral squamous cell carcinoma.Int J Cancer 2003;105:41–6.

28. Russo AL, Thiagalingam A, Pan H, Califano J, Cheng KH, PonteJF, et al. Differential DNA hypermethylation of critical genesmediates the stage-specific tobacco smoke-induced neoplasticprogression of lung cancer. Clin Cancer Res 2005;11:2466–70.

29. Lopez M, Aguirre JM, Cuevas N, Anzola M, Videgain J,Aguirregaviria J, et al. Gene promoter hypermethylation in

oral rinses of leukoplakia patients a diagnostic and/or prog-nostic tool? Eur J Cancer 2003;39:2306–9.

30. Stepanov I, Hechtt SS, Ramakrishnan S, Gupta PC. Tobaccospecific nitrosamines in smokeless tobacco products marketedin India. Int J Cancer 2005;116:16–9.

31. Hecht SS, Chen CB, Hirota N, Ornaf RM, Tso TC, Hoffmann D.Tobacco specific nitrosamines: formation from nicotine in vitroand during tobacco curing and carcinogenicity in strain A mice.J Natl Cancer Inst 1978;60:819–24.

32. Hecht SS, Rivenson A, Braley J, DiBello J, Adams JD,Hoffmann D. Induction of oral cavity tumors in F344 rats bytobacco-specific nitrosamines and snuff. Cancer Res 1986;46:4162–6.

33. Hecht SS, Hoffmann D. Tobacco-specific nitrosamines, animportant group of carcinogens in tobacco and tobacco smoke.Carcinogenesis 1988;9:875–84.

34. Liu Y, Egyhazi S, Hansson J, Bhide SV, Kulkarni PS, GrafstromRC. 06-methylguanine-DNA methyltransferase activity in humanbuccal mucosal tissue and cell cultures. Complex mixturesrelated to habitual use of tobacco and betel quid inhibit theactivity in vitro. Carcinogenesis 1997;18:1889–95.