Glutathione S-Transferase Polymorphisms and the Synergy of Alcohol and Tobacco in Oral, Pharyngeal, and Laryngeal Carcinoma

Head and neck squamous cell carcinoma (HNSCC) is the 8th most common cancer worldwide for both sexes, the 6th most common cancer in developing nations, and is the 10th most common cancer in the United States (1, 2). The use of tobacco and alcohol accounts for ∼75% of all HNSCCs in the United States (3, 4). Rothman (5) and Blot et al. (3) have shown that alcohol and tobacco have both independent and synergistic roles in the genesis of HNSCC, as strong dose-response relationships for alcohol and for tobacco consumption as well as a multiplicative interaction of both are observed in virtually all of the HNSCC studies published to date. Tobacco smoke is well recognized as a complete carcinogen, but the mechanism by which alcohol exerts its carcinogenicity remains unclear. Alcohol may act as a solvent for other carcinogens or perhaps generate and exacerbate coincident inflammation, producing significant reactive oxygen species (6, 7). Although the combined exposures of tobacco and alcohol are primarily responsible for the genesis of this disease, interindividual genetic variability may play an important role in modifying the potency of these carcinogens and mediating potential susceptibility to the action of both tobacco and alcohol either individually or as they interact to induce cancer (8, 9). Indeed, genetic polymorphisms affecting the expression or activity of metabolic enzymes responsible for the detoxification of tobacco and alcohol are thought to influence an individual's susceptibility to oral cancer (10, 11).

Glutathione S-transferases (GST) are phase II xenobiotic metabolizing enzymes directly involved in catalyzing the conjugation of reactive intermediates of electrophilic xenobiotics with glutathione (8, 11, 12). The GSTs are composed of four major groups: GSTA (α), GSTT1 (𝛉), GSTM1 (μ), and GSTP1 (π). The 𝛉 class of glutathione transferases is also well known to activate haloalkanes (13), such as those found in the chlorination by-products in treated drinking water (14, 15), and methyl chloride, which is found in tobacco smoke (16). GST expression varies among individuals, and this variation is tissue and gender specific (17). GSTM1, GSTT1, and GSTP1 are well known to be polymorphic, and deleted variants of the GSTM1 and GSTT1 loci result in loss of functional activity (18-20). Deletion of the GSTM1 and GSTT1 genes has been reported in approximately 50% and 20% of the Caucasian population, respectively (21-23). Two polymorphisms in the GSTP1 gene, one at codon 105 with an A to G transition resulting in an isoleucine to valine amino acid change and another at codon 114 (alanine to valine substitution), have been reported to cause differences in catalytic activity (9).

Various studies have shown positive associations between the GSTM1 and GSTT1 null genotypes and increased risk for skin, lung, bladder, and oral cancer (9, 24-29). Despite several independent studies and subsequent meta-analysis, the reports about the association of HNSCC and the GST variants remain inconclusive (11, 30). The GSTs play an important role in the metabolism of chemical carcinogens, especially with regard to those present in tobacco smoke. Further, although tobacco and alcohol are major independent risk factors for HNSCC, it is in their interaction that the most profound risk of HNSCC is incurred. The mechanism of this interaction is not known, and there has been little attention paid to the possible modification of this interaction by the GST polymorphisms.

We hypothesize that if the synergy of tobacco and alcohol simply derives from an increase in the potency of tobacco carcinogens and that the potency of the tobacco carcinogens can be further enhanced by polymorphisms in the GST enzyme, then these synergistic effects should be observable by examining the modification of the tobacco-alcohol interaction on HNSCC risk by GST polymorphism (in essence, a three-way interaction). Overall, it is not clear if the absence of the GST catalytic enzymes alters the potency of tobacco and/or alcohol or modifies their interaction, but doing this novel examination of synergistic interaction may aid in elucidating the true carcinogenic mechanism of all three of these potential risk factors. Hence, we have examined the associations of the GSTM1, GSTT1, and GSTP1 polymorphisms with HNSCC and investigated whether they alter cancer risk associated with smoking, drinking, or the combination of both in a population-based case-control study.

We conducted a case-control study from December 1999 to December 2003 in the Greater Boston Metropolitan Area. This region of Massachusetts includes a population of ∼3.5 million people in 249 cities and towns within a 1-hour drive of Boston. The institutional review boards at all participating institutions approved this study, and all volunteer participants provided informed consent. Details of the case-control ascertainment have been presented elsewhere and are detailed briefly below (31).

Between December 1, 1999 and December 1, 2003, incident cases of HNSCC were identified through multidisciplinary head and neck clinics, otolaryngology, and radiation oncology departments at nine medical facilities located in Boston, MA. We defined HNSCC as including International Classification of Disease, Ninth Revision codes 141, 143-6, 148, 149, and 161. All patients with carcinoma in situ, lip, salivary gland, or nasopharyngeal cancer or with recurrent cancer of the head and neck region were excluded. Histologic classification of malignancy was based on that reported by pathology at the participating hospitals.

Population-based controls were drawn from the specified greater Boston population. The controls were frequency matched (1:1) to cases by age (±3 years), gender, and town of residence. These controls were identified through random selection from the Resident Lists for the 249 cities and towns within the study area using the cases' addresses as reference.

Participating cases and controls were given a self-administered questionnaire to collect medical history, demographic information, as well as information on tobacco and alcohol consumption. Each questionnaire was reviewed with each participant by a trained research coordinator. Smoking history was ascertained with a standardized instrument that assesses the number of years smoked, the number of cigarettes smoked daily, age at which an individual started smoking, number of years since quitting, and the duration of smoking. Similar information was obtained about lifetime consumption of beer, wine, and liquor. Questionnaires were given to case participants during an initial clinic visit and subsequently retrieved in person. Control participants received their questionnaires in the mail and returned them in person to the research assistant.

DNA was obtained either from whole blood or from exfoliated buccal cells. Genotyping was carried out by PCR amplification using previously described methods (32, 33). Positive and negative controls were included in each batch of samples. Blood-derived DNA from individuals of known genotype (including all possible genotypes) served as a positive control, whereas additional controls for detection of contamination (blanks) were routinely included in each genotype run.

Tests for Hardy-Weinberg equilibrium among all controls were conducted. Exposures were classified as follows: pack-years of smoking were calculated as the product of current or former daily cigarette use (packs/day) and the duration of cigarette smoking (years). The number of alcohol drinks consumed weekly was derived from the questionnaire for average lifetime weekly beer, wine, and liquor consumption. The distribution of cigarette use was compared between the cases and controls and across genotype levels. Odds ratios (OR) were calculated to measure the independent and synergistic effects of tobacco, alcohol, and genotype on the risk of HNSCC. Variables for the joint effects were coded using a common referent group, and interactions between smoking and genotype and smoking and alcohol by genotype were evaluated by including interaction terms in the regression model. Cut points were selected to reflect the published literature (3, 34-37). Unconditional logistic regression was done in Statistical Analysis System version 9.1 (SAS Institute, Cary, NC) to calculate the ORs and 95% confidence intervals (95% CI) for the association between each factor of interest and the case status while controlling for age, gender, and race (38). Multivariate models included the matching terms: age (categorical), gender, and race. Frequency matching allows the use of unconditional regression with inclusion of the matching variables in the model (39). To test for the significance of the observed interactions, we conducted likelihood ratio tests using a χ² distribution.

Eight hundred twenty-three eligible cases were invited to participate; of these, 57 refused to participate. Among the 766 consented subjects, another 44 did not complete the questionnaire. Of the 722, complete genotype and questionnaire data were available on 692 participants. Similarly, 1,623 subjects were identified and eligible for participation as controls. Eight hundred twenty-eight subjects refused to participate and 815 subjects were consented with 771 finally enrolled in the study. Six of the controls were withdrawn as they were matched to a case that subsequently became ineligible, such that 765 controls were enrolled and completed. Complete genotype and questionnaire data were available on 753 people. The characteristics of the final study population are described in Table 1.

Among the 692 cases and 753 controls, the mean age was 60 and 61 years, respectively. Twenty-eight percent of controls and 27% of cases were female. Among controls, 91% reported their race to be Caucasian compared with 89% of the cases. The overall mean weekly alcohol consumption among the cases was 26 drinks weekly, whereas controls reported a mean of 13 alcoholic drinks weekly. Cases were more likely to consume >40 drinks of alcohol (heavy drinking) weekly than controls (21% versus 7%). Heavy alcohol consumption (>40 drinks weekly) was associated with a >4-fold increased risk for HNSCC compared with nondrinkers, adjusted for age, race, gender, and tobacco use (OR, 4.1; 95% CI, 2.8-6.0). The mean pack-years of smoking among cases was 36 pack-years compared with 20 pack-years among controls. In addition, cases were more likely to have heavy (≥45 pack-years) cigarette consumption than were controls (36% versus 16%). Similarly, heavy tobacco use was associated with a >3-fold elevation in HNSCC risk, adjusted for age, race, gender, and alcohol consumption (OR, 3.4; 95% CI, 2.4-4.7). When exposure was modeled in this fashion, there was a monotonic dose response for HNSCC risk associated with both tobacco and alcohol use. Table 1 also shows the ORs for tobacco and alcohol exposure by cancer site (oral cavity, pharynx, and larynx). The magnitude of these risks generally parallels that for all sites, with the exception of suggestion of a more marked elevation of risk for laryngeal cancer associated with smoking.

The GSTM1, GSTT1, and the GSTP1 genotype frequencies among cases and controls are shown in Table 1. The prevalence of the GSTM1 deletion was 58% among cases and 54% among controls, and the GSTT1 deletion was observed at 18% among cases and 22% among controls. The prevalence of the GSTP105 AA, GSTP105 AG, and GSTP105 GG genotypes was 44%, 45%, and 11% among cases and 45%, 43%, and 12% among controls, respectively. The GSTP1 AG and GSTP1 GG genotypes were combined in subsequent analyses.

The GSTM1 deletion was an independent risk factor for HNSCC, with the deletion genotype exhibiting an OR of 1.3 (95% CI, 1.0-1.6) compared with those with at least one wild-type allele in a model controlled for tobacco and alcohol use, age, race, and gender. The frequency of the GSTT1 deleted genotype was significantly higher in controls compared with cases, and after adjustment for tobacco, alcohol, age, and gender, there was an association of the gene deletion with diminished HNSCC risk (OR, 0.8; 95% CI, 0.6-1.0). There was no association between HNSCC and the GSTP1 polymorphism. We also found no association between medication use or household income associated with GST genotype.

The interactions between the glutathione transferase polymorphisms and tobacco use with HNSCC risk are shown in Table 2. We observed evidence of effect modification of the GSTM1 deletion on the association between tobacco use and HNSCC. Among the individuals with at least one GSTM1 allele, the OR for heavy smokers (>45 pack-years) was 2.6 (95% CI, 1.6-4.3), in contrast to the GSTM1 deleted individuals who had an increased OR of 4.2 (95% CI, 2.6-6.7). The interaction term for GSTM1 deletion and pack-years (as a continuous variable) was marginally significant (P = 0.1).

The GSTT1 and GSTP1 polymorphisms did not significantly alter the association between tobacco use and HNSCC. The GSTT1 deletion was observed to diminish the strength of the association between heavy tobacco use and HNSCC (OR, 2.8; 95% CI, 1.6-5.0) relative to the presence of GSTT1 (OR, 3.4; 95% CI, 2.4-4.9). We consequently combined the GSTM1 deleted with the GSTT1 present individuals (the at-risk genotypes in the analysis) and observed that, among heavy smokers, in individuals with the GSTM1 deletion and at least one GSTT1 allele there was a 140% increase in risk compared with the other genotype combinations (OR, 4.1; 95% CI, 2.6-6.6 versus OR, 2.9; 95% CI, 1.9-4.4), although this difference did not attain statistical significance.

When the association of alcohol with HNSCC risk was examined by GSTM1, or GSTT1 deletion, there was no appreciable change in estimated ORs. Likewise, we did not observe that the GSTP1 polymorphism altered the ORs of HNSCC due to alcohol consumption (data not shown).

As expected, the combination of both smoking and drinking increased an individual's risk for developing HNSCC (Table 3). Nonsmoking individuals (≤1 pack-year) who consumed >40 alcohol drinks weekly had a >4-fold elevated risk for HNSCC, adjusted for age, race, and gender, compared with those who did not drink (OR, 4.6; 95% CI, 1.5-14.7). Similarly, light-drinking individuals (<10 drinks weekly) with a history of >45 pack-years of cigarette consumption had a 3-fold greater risk for HNSCC compared with nonsmokers (OR, 3.0; 95% CI, 1.9-4.6). When both tobacco and alcohol were evaluated jointly, consumption of >40 drinks of alcohol weekly and >45 pack-years of cigarette use was associated with a significant 14-fold increased risk of HNSCC (OR, 13.9; 95% CI, 7.8-25.0).

In an effort to determine whether any effect modification was associated with the independent effects of alcohol or tobacco or if it altered their interactive effects, we examined the association of the GSTM1 polymorphism with the cancer risk associated with the interaction of tobacco and alcohol (Table 3). Interestingly, we observed the greatest difference in the disease risk estimates among those individuals consuming <20 drinks weekly who were heavy smokers. Those individuals with a homozygous GSTM1 deletion had a substantial elevation in estimated cancer risk at higher levels of smoking with lighter alcohol consumption compared with individuals with at least one copy of the GSTM1 gene (Table 3). The combination of 10 to 20 drinks weekly and smoking >45 pack-years was associated with an ∼13-fold elevated risk (OR, 12.6; 95% CI, 4.0-40.2) among the GSTM1 deleted subjects compared with an OR of 3.6 (95% CI, 1.5-8.7) among the GSTM1 present individuals. Neither GSTT1 nor GSTP1 seemed to alter the tobacco-alcohol interaction on the risk of HNSCC.

Studies of genetic susceptibility to HNSCC have observed an altered cancer risk associated with polymorphisms of GSTM1 and GSTT1. These studies have almost exclusively investigated potential effect modification of the cancer risk associated with tobacco use, controlling for the confounding effect of alcohol consumption. This approach does not specifically recognize the strong synergistic effect that exists between tobacco and alcohol intake in the development of HNSCC (5, 40). HNSCC is a classic example of biological interaction; however, to our knowledge, no investigations have specifically asked if the GSTM1 or GSTT1 deletion polymorphisms modify the tobacco-alcohol interaction. Thus, we have sought to examine the modification of this interaction by GST genotype.

We observed a main effect of GSTM1 deletion on HNSCC risk, with those having GSTM1 deletion exhibiting a 30% increased risk of HNSSC (OR, 1.3; 95% CI, 1.0-1.6). Among those study participants who reported a smoking history of at least 45 pack-years, the HNSCC risk for individuals who were GSTM1 deficient was almost double that experienced by individuals with at least one allele of the GSTM1 gene; the ORs of HNSCC were 4.19 (95% CI, 2.63-6.68) and 2.64 (95% CI, 1.64-4.25), respectively. This is consistent with the prior work of numerous other investigators (29, 30).

We also observed a protective association of the GSTT1 deletion with HNSCC risk, consistent with the work of both To-Figueras et al. (41), who observed a similar association among laryngeal cancer cases (OR, 0.61; 95% CI, 0.4-1.1), and, most recently, Ophuis et al. (42), who reported a 50% reduction in HNSCC risk associated with the GSTT1 null genotype. Jaskula-Sztul et al. (43) observed that laryngeal cancer risk associated with smoking and the GSTT1 deletion was reduced 20% (OR, 0.8; 95% CI, 0.5-1.3). Gajecka et al. (44), Hanna et al. (45), and Sugimura et al. (46) have also reported risk estimates consistent with our data. Restricting our data to laryngeal cancer showed a protective OR of 0.56 (95% CI, 0.32-0.98) for the modification by the GSTT1 of the association of smoking and cancer. Further, similar associations have been reported for other solid tumors; one recent example is that of Huang et al. (47) who reported a 28% decreased risk for colon cancer among whites (OR, 0.72; 95% CI, 0.53-1.00). Abbas et al. (48) reported that the absence of the GSTT1 deletion was associated with a 13-fold elevated risk of adenocarcinoma of the esophagus, and Kelada et al. (49) observed that individuals with the nondeleted GSTT1 genotype were found to be at a higher risk of prostate cancer. The 𝛉 class of glutathione transferases has been associated with bioactivation and transformation of halogenated compounds, such as dichloromethane and dihaloethanes, to mutagenic intermediates (17, 50). By-products of tobacco smoke include methyl chloride, which is a substrate of GSTT1 detected at appreciable levels in cigarette smoke (16). Methyl chloride undergoes a biotransformation similar to that of dichloromethane (49) and could contribute to any protection from cancer proffered by the absence of the GSTT1 alleles. Lof et al. (51) have shown that the GSTT1 phenotype determines the metabolism of methyl chloride, and GSTT1 null individuals lack the capacity to metabolize methyl chloride.

Perhaps most novel and striking was our observation that polymorphic deletion of the GSTM1 gene seemed to markedly alter the alcohol-tobacco interaction in the generation of cancer risk. This was most evident for individuals with very low (<10 drinks weekly) and low (10-20 drinks weekly) levels of alcohol consumption who were also heavy smokers (>45 pack-years). There was a pronounced difference in the estimates of HNSCC risk associated with combined low alcohol and heavy tobacco use on stratification by GSTM1 variant status. The association between GSTM1 deletion and HNSCC was strongest among those who were both heavy smokers and low alcohol consumers (OR, 12.6; 95% CI, 4.0-40.2). This is in dramatic contrast to the estimated cancer risk among individuals with at least one GSTM1 allele who reported the same combination of alcohol and tobacco use; their estimated risk of disease was ∼4-fold lower (OR, 3.6; 95% CI, 1.5-8.7). This is potentially important evidence of a three-way interaction between homozygous deletion of the GSTM1 gene, tobacco, and alcohol.

Lack of GSTM1 enzyme activity is thought to increase cancer susceptibility as a result of a decreased ability to detoxify reactive intermediates of tobacco carcinogens (52, 53). One carcinogenic role for alcohol is believed to be its ability to act as a solvent, facilitating the entry of tobacco carcinogens into oral tissues. Evidence supports the hypothesis that the local permeabilizing effects of alcohol that facilitates the penetration of tobacco-specific carcinogens across oral mucosa may explain the synergy between tobacco and alcohol. Squier et al. (54) showed that the penetration of nitrosonornicotine across the oral mucosa in the presence of nicotine is enhanced by alcohol. Du et al. (55) and Howie et al. (56) have shown that approximately 15% to 25% ethanol alone increased the permeability of the oral mucosa to tobacco carcinogens, such as N-nitrosonornicotine. However, at concentrations >50% ethanol, no further permeabilization is noted probably due to a fixative effect of ethanol on the mucosa. These studies provide evidence that the role of alcohol in oral cancer may in part be due to the topical effect of alcohol on the oral mucosa.

The effect of GSTM1 being present or deleted is more noticeable among heavy smokers and light drinkers than among heavy smokers and heavy drinkers. Higher smoking and alcohol usage levels establish a certain degree of risk for HNSCC. The presence or absence of GSTM1 has little discernable association with risk when both alcohol and smoking are high. When alcohol consumption is minimal, the role of GSTM1 seems to be more influential. Our data are consistent with the hypothesis that deletion of the gene results in relatively higher biologically active levels of tobacco carcinogens at low levels of alcohol intake. Therefore, at lower alcohol intake levels, the deletion of GSTM1 increases the carcinogenic potency of tobacco compared with the GSTM1 intact subjects.

We found no overall association between GSTP1 polymorphisms and HNSCC risk and no apparent interaction with tobacco. There are conflicting reports in the literature on the association of GSTP1 polymorphisms and HNSCC. Park et al. (57) also examined the GSTP1 polymorphisms and found a 2-fold elevated risk for any homozygous combination of the GSTP1 variants. Jourenkova-Mironova et al. (28) and Olshan et al. (29) found no association between the variant GSTP1 genotype and oral and pharyngeal cancers. Our observations are consistent with a meta-analysis of the GSTP1 polymorphisms and head and neck cancer that found no association between the presence of the variant allele and HNSCC (30).

A concern expressed about previous studies of GST and HNSCC was the potential for selection bias (11). Studies often did not adequately define and sample their study base. This could result in a selection bias if the controls do not represent the exposure and genotype distribution in the source population. The presence of selection bias in our study should be diminished through the use of population-based controls. At the same time, only 87% of all eligible cases and 49% of eligible controls fully participated. It remains possible that the participating controls differ with respect to their tobacco and alcohol use relative to the included cases. It is important to note that it is unlikely with respect to genotype that there was appreciable selection bias. In addition, cases that sought care at participating institutions may be different than all cases from the study area. Enrolled cases may be more educated and of higher income than cases that refused to participate. Therefore, it is possible that the selection of cases and controls was influenced by unmeasured factors related to socioeconomic status. Any selection bias in our study could lead to the cases and controls being more similar with respect to their tobacco and alcohol use. Consequently, the effects of tobacco and alcohol may be diminished, which in turn could lead to a weakening of any possible gene-environment interaction. Additionally, misclassification of the true level of tobacco or alcohol consumption reported by cases in our study may alter the magnitude and the direction of their association with HNSCC, thereby resulting in attenuated power to detect effect modification. Although every attempt has been made to administer the questionnaire in an identical fashion to both cases and controls, there still remains the possibility that some degree of misclassification may arise. However, these biases are not likely to segregate by genotype and hence are less likely to affect the noted association of gene-environment interaction. The prevalence of at-risk GST genotypes among the controls was similar to those reported in the literature (30).

Our study provides evidence to support a three-way interaction between GSTM1, tobacco, and alcohol; this interaction is greater among individuals with light alcohol intake who are heavy smokers. As the robust detection of a three-way interaction requires very large sample sizes, it is important to continue the study of genetic susceptibility to this disease, posing these questions in meta-analysis or in the continuing work of investigator-driven study consortia. Data that might address this question may exist in the compilation of the work of prior studies; this may be a potentially productive avenue of investigation for a consortium of investigators. Regardless, our data suggest that alcohol enhances to potency of tobacco carcinogens in the head and neck at low dose, providing additional support for this as the mechanism of action of alcohol in this well-known, important synergistic enhancement of disease risk.

We thank Dr. Heather Nelson for her assistance in the study and article preparation and the study participants, the collaborating clinicians, and the research staff involved throughout the study.