Joint Effects of Alcohol Consumption and Polymorphisms in Alcohol and Oxidative Stress Metabolism Genes on Risk of Head and Neck Cancer

Head and neck cancers typically include tumors of the oral cavity, pharynx, larynx, nose, nasal cavity and sinuses, and esophagus. This study focuses specifically on squamous cell cancers of the oral cavity, pharynx, and larynx.

There were an estimated 49,260 new cases and 11,480 deaths from oropharyngeal and laryngeal cancer in the United States in 2010 (1). Globally in 2008, oral cavity tumors were among the top 10 incident cancers in men world wide, and the top 10 fatal cancers in men in developing countries (2).

Squamous cell carcinoma of the head and neck (SCCHN) incidence is higher in men than women, and in the United States, in African-Americans and the poor. Much of this disparity is due to higher incidence of laryngeal tumors among African-American men (3).

SCCHN is strongly associated with smoking tobacco products and drinking alcoholic beverages, and recently with human papillomavirus infection. It is estimated that 75% of SCCHN in the United States is due to cigarette smoking and alcohol consumption (4). The effect of these exposures varies by anatomic subsite, with smoking more associated with laryngeal tumors, and drinking with oral cavity tumors. However, only a small fraction of people exposed to these carcinogens will develop SCCHN, suggesting that other factors, including genetic, must be considered. Inherited genetic variation in alcohol metabolism has been suggested as a potentially important contributor to SCCHN risk. Investigation of the association between single-nucleotide polymorphisms (SNP) and SCCHN may help to identify high-risk groups and clarify carcinogenesis pathways.

Many studies of genes in the alcohol metabolism pathway (ADH family, ALDH2, CYP2E1) have been limited by sample size, and with one exception (5), none have included a significant percentage of African-Americans. Furthermore, few studies have examined the influence of genetic variation in oxidative stress pathways (SOD, GPx, CAT). We examined the association between SNPs and haplotypes of genes in the alcohol metabolism and oxidative stress pathways and SNP–alcohol interactions using data from a large NC population–based case–control study of SCCHN including 22% African-Americans.

The Carolina Head and Neck Cancer Epidemiology Study (CHANCE) is a population-based case–control study upon which these analyses are based (6).

All cases of squamous cell carcinoma of the oral cavity, pharynx, and larynx diagnosed in 46 NC counties between January 1, 2002, and February 28, 2006, were eligible for enrollment. Rapid case identification was conducted by the NC Central Cancer Registry. CHANCE cases included ICD-O-3 topography codes C0.00-C14.8 and C32.0-C32.9, excluding salivary gland (C07.9, C08.0-C08.9), nasopharynx (C11.0-C11.9), nasal cavity (C30.0), and nasal sinuses (C31.0-C31.9). ICD-O-3 morphology codes included were 8010/3, 8051/3, 8083/3, 8071/3, 8072/3, 8073/3, 8074/3, and 8076/3. Benign tumors, carcinomas in situ, papillary carcinomas, and adenoid carcinomas were excluded. We further excluded 21 lip cancers (C00.3-C00.9, C14.2), 46 of “other” race, and 96 without genotyping data, producing a study composition of 1,227 cases and 1,325 controls.

Potentially eligible controls from the same counties as cases were identified through NC Department of Motor Vehicles records. Controls were frequency matched to cases using random sampling with stratification on age, race, and sex.

Trained nurse-interviewers conducted an in-person interview with each subject. For this analysis only self-reported, nonproxy data were included. Questions were asked about demographics, tobacco use, drinking of alcoholic beverages, diet, oral health, medical history, and family history of cancer.

Blood samples were obtained by nurse-interviewers trained in phlebotomy. If the subject was not willing or able to consent to the blood draw, they were asked to contribute a buccal cell sample via mouth rinse.

Written informed consent was obtained from all subjects. The study was approved by the Biomedical Institutional Review Board at the University of North Carolina at Chapel Hill.

Case tumors were classified into anatomic subsites according to the following 5 ICD-O categories used by the International Head and Neck Cancer Epidemiology Consortium (ref. 7): (i) oral cavity: C02.0-C02.3, C03.0, C03.1, C03.9, C04.0, C04.1, C04.8, C04.9, C05.0, C06.0-C06.2, C06.8, and C06.9; (ii) oropharynx: C01.9, C02.4, C05.1, C05.2, C09.0, C09.1, C09.8, C09.9, C10.0-C10.4, C10.8, and C10.9; (iii) oral cavity–oropharynx–hypopharynx NOS: C02.8, C02.9, C05.8, C05.9, C14.0, C14.2, and C14.8; (iv) hypopharynx: C12.9, C13.0-C13.2, C13.8, and C13.9; and (v) larynx: C32.0-C32.3, and C32.8-C32.9.

Questions about alcohol use were designed to estimate lifetime history of consumption, and usual consumption of each beverage type, prior to the year before diagnosis. Questions asked about beer, wine, and hard liquor separately as follows: (i) Did you drink (beer/wine/hard liquor)? (ii) At what age did you start? (iii) At what age did you stop? (iv) For how many years did you drink (beer/wine/hard liquor) during this period? (v) How much (beer/wine/hard liquor) did you usually drink? Per day/week/month/year? and (vi) What size did you usually drink?

As frequency of drinking has shown stronger associations with SCCHN than duration (8), a single frequency measure that included all types of alcoholic beverages would have been optimal for estimating alcohol interaction with SNPs. Because this was unavailable in CHANCE, we instead derived a lifetime measure of alcohol intake, in milliliters, for beer, wine, and liquor combined. Using splines, we confirmed that tertiles best represented the risk associated with alcohol intake.

The primary tobacco exposure covariate selected was continuous duration of cigarette smoking. Dichotomous variables representing additional potential tobacco confounders were: ever use of non-cigarette tobacco, and ever-exposed to environmental tobacco smoke at work or at home.

Seventy-five SNPs (69 tag SNPs, and 6 candidate SNPs found in prior studies to be associated with cancer incidence or survival, or alcohol dependence) were selected in 12 genes that are part of 2 metabolic pathways: ADH1B, ADH1C, ADH4, ADH7, ALDH2, and CYP2E1 in the alcohol metabolism pathway in the upper aerodigestive tract; and CAT, SOD1, SOD2, GPX1, GPX2, and GPX4 in the oxidative stress pathway. Tag SNPs, chosen to represent the genetic variation within each of the 12 candidate genes (gene and 2,000 bp upstream and downstream) were selected using the Genome Variation Server (9), using SNPs that were polymorphic in either CEU or YRI HapMap Release 2 (unrelated only), with the following parameters: allele frequency cutoff value of 10%, 0.8 R² threshold minimum for variations to belong to the same cluster, 85% minimum data coverage for tag SNPs, 70% minimal data coverage for a variation to be potentially clustered with others.

To control for potential population stratification, we selected 157 ancestry informative markers (AIM) to maximize (i) the difference in allele frequencies (delta) between European and African populations in the HapMap data (CEU vs. YRI), and (ii) the Fisher's information criterion (FIC). AIMs were prioritized on the basis of having the highest delta and FIC values in the following order: 90% European/10% African, 10% European/90% African, and 50% European/50% African. This allowed AIMs to represent the entire expected ancestral distribution of the study population. Individual estimates of percentage African ancestry were calculated from 145 successfully genotyped AIMs using maximum likelihood estimation methods previously described (10–12). AIMs were chosen to differentiate only between African and European ancestry, so individual ancestry estimates for the 2 groups sum to 1.0.

DNA was extracted from blood or buccal samples collected at time of interview. Genotyping was done by the University of North Carolina at Chapel Hill, Mammalian Genotyping Core Facility, using the Illumina GoldenGate Genotyping Assay with Sentrix Array matrix and 96-well standard microtiter plates.

Haplotypes using SNP data were constructed separately for African- and European-Americans using default D' blocks in Haploview 4.2. The algorithm (13) constructs 95% confidence limits on D' and each comparison is defined as “strong LD,” “inconclusive,” or “strong recombination.” A block is created if 95% of informative comparisons are in “strong LD.” Markers with minor allele frequency less than 5% are ignored. Assignment of most likely haplotype for individuals with ambiguous haplotype was done using an EM algorithm in haplo.stats (14), with minimum counts set to 10.

Dichotomous variables representing additional potential confounders were: had health insurance on reference date, had a routine dental visit in the last 10 years, ever had a loose permanent tooth due to disease, ever used mouthwash, family history of SCCHN, household poverty as defined by federal guidelines, and education level.

ORs for the independent effects of SNPs and alcohol, and their interactive effects, were computed using conditional logistic regression implemented in SAS 9.2. ORs for the main effects of haplotypes were computed using unconditional logistic regression implemented in haplo.stats 1.4.4.

A dominant genetic model (at least one minor allele vs. referent of no minor alleles) was used for SNPs because for many SNPs, the number of subjects homozygous for the minor allele was too small to permit precise effect measurement.

Potential covariates were eliminated using stepwise backward elimination, comparing each reduced model with a full model that included all covariates listed in Table 1. No collinearity was noted between variables in the full model, with one exception as described below. If a covariate did not change the ln(OR) for any SNP by a difference of at least 0.10, it was eliminated from subsequent models. Final models for genetic main effects contained a single SNP or haplotype and duration of smoking as a continuous variable. Models estimating SNP–drinking interaction also included categorized lifetime ethanol consumption. We had insufficient power to detect haplotype–drinking interaction because haplotypes were constructed and analyzed separately for African- and European-Americans. The conditional logistic regression used for SNPs by definition takes into account the matching variables of age, sex, and race. The unconditional logistic regression models used for haplotypes (for each race separately) included, as covariates, sex, age, and their 2-way interaction. Ancestry was not important for the polymorphisms studied, probably because self-reported race was already included (as a matching variable). The ancestry variable also showed evidence of collinearity with race, so for these reasons and for parsimony's sake, ancestry was excluded from final models.

A Bonferroni correction was used to adjust P values and interaction contrast ratios (ICR) confidence intervals (CI) to control for type 1 error introduced by multiple statistical testing, for either 64 tests (for 64 SNPs) or for 12 or 13 tests (for haplotypes).

Departures from additive interaction were evaluated by computing ICRs and Bonferroni-corrected CIs. ICRs were calculated using cancer ORs of subjects in 3 categories: (i) the highest drinking category and no minor allele (OR₀₁); (ii) never-drinkers with at least one minor allele (OR₁₀); and (iii) subjects in the highest drinking category and at least one minor allele (OR₁₁), compared with never-drinkers homozygous for the major allele (i.e., the referent: OR₀₀ = 1.0). ICR is calculated as follows: ICR = OR₁₁ − OR₀₁ − OR₁₀ + 1. ICRs significantly different from zero indicate departure from additive interaction.

Although controls were somewhat older and more likely to be female and European-American than cases (Table 1), the percentages of cases versus controls in each of the 28 age–sex–race cross-categories, as a proportion of the entire study population, differed by less than 2%. Compared with controls, cases smoked and drank more, and were poorer, less likely to have completed high school or have health insurance, less likely to have had a routine dental visit in the past 10 years, and more likely to have lost a permanent tooth to disease. Cases were also more likely to have been exposed to environmental tobacco smoke at home and work. Mean proportion of African ancestry was slightly higher in cases than controls.

Sixty-four of 75 SNPs (45 alcohol metabolism, 19 oxidative stress) were successfully genotyped. Assay intensity data and genotype cluster images for all SNPs were individually reviewed; as a result, 9 of the original 75 tag SNPs and 12 AIMs (9% of SNPs) were excluded due to inadequate signal or indistinguishable genotype clusters. Blind duplicates of 109 samples were genotyped to verify call reliability; none of our SNPs were discrepant. Two of the original 75 tag SNPs were judged to be out of Hardy–Weinberg Equilibrium (SAS PROC ALLELE) in both European- and African-American controls due to an exact P value <0.001, and were eliminated from analysis.

There were no large differences in allele frequencies between cases and controls, when stratified by race (Supplementary Table S1). However there are large allele frequency differences between African- and European-Americans.

The odds of developing SCCHN increase monotonically as lifetime alcohol consumption increases (Table 2). Subjects in the lowest consumption category experienced reduced SCCHN odds compared with nondrinkers (OR = 0.8; 95% CI, 0.6–1.0), driven largely by laryngeal and oral cavity tumors (OR = 0.7; 95% CI, 0.4–1.1 and OR = 0.4; 95% CI, 0.2–0.9, respectively).

Successively higher levels of alcohol consumption were associated with increasing odds. The middle tertile of lifetime consumption was associated with 30% higher SCCHN odds than never-drinkers, and the highest tertile of consumption with tripled odds. In the highest drinking category, all subsites experienced significantly increased odds: doubled odds of laryngeal cancer, and tripled or greater odds for oropharyngeal and oral cavity tumors.

None of the SNP associations with SCCHN or any of the subsite cancers had a significant Bonferroni-corrected P value, although 5 SNPs in ADH1B, ALDH2, and SOD2 showed evidence of reduced or increased cancer ORs overall and in oral cavity, laryngeal, and hypopharyngeal subsites (Table 3; remaining subsite effects in Supplementary Table S2). In ADH1B, the rs1229984 A allele was associated with 30% decreased SCCHN odds, and the rs17028834 C allele with 50% increased odds of laryngeal tumors. In ALDH2, the rs2238151 C allele was associated with 10% increased odds of SCCHN, driven largely by 20% increased risk of laryngeal tumors. In SOD2, the rs4342445 A allele was associated with 30% greater odds for oral cavity tumors, and the rs5746134 T allele with doubled odds for hypopharyngeal cancer.

Linkage disequilibrium was strong among SNPs within genes, not between genes, so haplotypes included only SNPs within the same gene (Supplementary Table S3). Four haplotypes in ALDH2, CYP2E1, GPX2, and SOD1 were associated with SCCHN, either in European-Americans or African-Americans, or both (Table 4). One GPX2 haplotype was significantly associated with 30% decreased odds of SCCHN in European-Americans. An ALDH2 haplotype was associated with 50% reduced odds in African-Americans and a CYP2E1 haplotype was associated with 30% reduced odds in European-Americans. The SOD1 AGGC haplotype was associated with increased odds in European-Americans and reduced odds in African-Americans.

To examine the potential impact of multiple at-risk alcohol metabolism alleles, we counted the number of previously studied risk alleles (0–4) for each individual, including these alleles: ADH1B rs1229984 ‘G’, ADH1C rs1693482 ‘T’, ADH7 rs1573496 ‘G’, and CYP2E1 rs3813867 ‘C’. The numbers of risk alleles was not associated with an increased or decreased risk of SCCHN (data not shown).

Four SNPs showed evidence of synergistic additive interaction with alcohol consumption (Table 5). All met the following 2 characteristics: (i) statistically significant or near-significant Bonferroni-corrected CI for ICR (64 tests), and (ii) at least 10 cases and 10 controls in each of the 3 comparison groups OR₀₁, OR₁₀, and OR₁₁. For example, heavy drinkers carrying the C allele of rs2238151 in ALDH2 showed statistically significant evidence of synergistic additive interaction. Also the T allele at rs1159918 in ADH1B, the A allele at rs1154460 in ADH7, and the T allele at rs2249695 in CYP2E1 showed some evidence for synergistic additive interaction between alcohol consumption and SNP. (Evaluations of additive interaction with alcohol for remaining SNPs can be found in Supplementary Table S4.)

No interactions with alcohol were detected for anatomic subsites.

SNP effect estimates were similar in European- and African-Americans, with a few exceptions. Two SNPs in SOD1 (rs10432782, rs2070424) were associated with decreased odds of SCCHN in African-Americans and increased odds in European-Americans (Supplementary Table S5; rs10432782, OR = 0.65; 95% CI, 0.42–1.00 in African-Americans, OR = 1.35; 95% CI, 1.07–1.71 in European-Americans; rs2070424, OR = 0.52; 95% CI, 0.33–0.83 in African-Americans, OR = 1.47; 95% CI, 1.10–1.97 in European-Americans). Three additional SNPs, which had sufficient frequency of the minor allele in both races, showed evidence of risk differences by race (ADH1B rs1693457, ADH4 rs10017466, SOD1 rs4998557) though CIs for races overlapped (Supplementary Table S5). The magnitude of the joint effect for the 4 SNPs found to interact additively with alcohol exposure did not differ between races (data not shown).

Most studies report a strong dose–response relationship between higher levels of drinking, both in lifetime frequency of drinking (e.g., drinks per day) and lifetime alcohol intake (e.g., milliliters of ethanol), and increased SCCHN risk. However, the type of alcohol beverage most strongly associated with cancer risk varies substantially by study, and some studies suggest that the most common alcoholic beverage in the geographic region studied produces the highest cancer risk (15). There is some evidence that moderate levels of wine consumption produce lower risk than beer and liquor (comparing 16–30 ethanol-standardized drinks per week of each type), but above 30 drinks per week, all types are associated with increased risk (15).

We found a general pattern of association with alcohol intake that is consistent with previous studies (7), with monotonically increasing cancer risk as lifetime consumption increases. Beer and liquor accounted for about 90% of lifetime alcohol consumption in our study population, and those beverages were associated with higher cancer risk than wine consumption (data not shown). This is consistent with the hypothesis that the most commonly drunk alcoholic beverages are associated with the highest risk.

Variant ADH and ALDH alleles coding for either superactive or inactive subunits of ADH and ALDH isozymes are common. Numerous studies in Asian populations have reported an association between several presumably functional variants in ADH1B, ADH1C, ADH4, ADH7, and ALDH2 and SCCHN incidence (16–23). However, these studies lacked sufficient power to consistently detect interaction between gene and alcohol drinking. In recent years, these variants and others were investigated in larger studies of Europeans, Latin-Americans, and Indians with similar findings (24–33). However, only a few smaller studies examined risk in European-Americans and African-Americans (5, 34–37), and those included very small numbers of African-Americans.

We discovered an association between rs1229984 in ADH1B and SCCHN odds (OR_{AA + AG vs. GG} = 0.72; 95% CI, 0.57–0.91). It is the same direction of effect for the A allele as reported in a Japanese study (ref. 21; OR_{GG + GA vs. AA} = 2.20; 95% CI, 1.46–3.32) and in European-Caucasians and Latin-Americans (ref. 29; OR_{AA + GA vs. GG} = 0.56; 95% CI, 0.47–0.66), but is the reverse of the effect reported in a few other studies [(18, 27, 28) e.g., (28)]: OR_{GG + GA vs. AA} = 0.36; 95% CI, 0.17–0.77). A recent INHANCE GWAS (38) reported a replicated association of 5 SNPs with SCCHN and esophageal cancer, including rs1229984, for which the A allele under a log-additive genetic model was associated with reduced odds in both the discovery (OR = 0.52; 95% CI, 0.43–0.64) and the replication phases (OR = 0.68; 95% CI, 0.60–0.78). The GWAS replication sample included 2,027 CHANCE subjects as 10% of the replication sample. In our study, only 6 African-American subjects carried the A allele, compared with 104 European-Americans, so most of the effect we observed for rs1229984 occurred in European-Americans.

We found no effect on SCCHN risk of the rs1693482 “slow” allele in ADH1C (OR_{TT + TC vs. CC} = 1.05; 95% CI, 0.95–1.15). The 2 largest studies of this SNP and SCCHN in European-Caucasians (28) and European-Caucasians and Latin-Americans (29) found 20% to 50% increased odds associated with this allele. Also, all 4 studies of rs698 “slow” or G allele in Brazilian, Japanese, European-American, and Latin-American populations (21, 27–29) reported evidence of 16% to 38% increased odds. In CHANCE, rs1631460 is in high LD (r² = 0.95) with rs698 in both CEU and YRI HapMap populations, but we found no association between it and SCCHN.

No ADH4 and ADH7 SNPs were associated with SCCHN, including the rs1573496 C allele in ADH7. This is in contrast to the one study that investigated this allele and found it to be associated with 30% reduced odds in Europeans and Latin-Americans (29).

No ALDH2 SNPs were associated with SCCHN, and a possible haplotype association was present only in African-Americans (OR = 0.5; 95% CI, 0.3–0.8). Previous studies of rs886205, an ALDH2 SNP that is polymorphic in Europeans, found conflicting results of no association and increased association for the G allele (26, 28).

Our findings may differ from those previously reported due to differences in sample size, the specific population studied, and the composition of tumor subsites included.

We discovered evidence of synergistic additive interaction with alcohol of several SNPs in alcohol metabolism pathway genes, although the SNPs we identified were different from those previously reported in the literature. Specifically, we found 2 SNPs in ADH1B and ADH7—rs1159918 and rs1154460, respectively—that seem to interact with alcohol. We also found one previously unstudied ALDH2 SNP, rs2238151, that showed evidence of additive interaction (OR_11actual = 3.3 vs. OR_11expected = 1.4). Whereas previous studies reported that rs1229984 in ADH1B, rs4148887 in ADH4, rs1573496 in ADH7, and rs886205, rs441 (both in high LD with our SNP rs4767939), and rs440 in ALDH2 interacted with alcohol drinking (16, 18, 27–29), we did not find evidence for an interaction with these SNPs, probably because we measured alcohol consumption using lifetime alcohol intake instead of drinking frequency.

We also found evidence for synergistic additive interaction for CYP2E1 rs2249695 with alcohol. A recent linkage and association study (39) identified that SNP, among others, to be associated with “tipsiness,” or quick response to alcohol challenge. In CHANCE, the T allele (Table 5, last row) was protective in never-drinkers (OR₁₀ = 0.6; 95% CI, 0.4–0.9) but was associated with 70% greater than expected risk in the heaviest drinkers.

We found 2 previously unstudied SNPs in SOD2 to be associated with subsite tumors: rs4342445 with oral cavity, and rs5746134 with hypopharynx. One SOD1 haplotype was associated with SCCHN risk in both races, albeit in different directions, due to the effect of multiple individual SNP effects that differed by race in that gene. Finally, we found a GPX2 haplotype to be associated with reduced SCCHN risk in European-Americans only. This may indicate that the haplotype is in high linkage disequilibrium with an unmeasured causal polymorphism in European-Americans but not in African-Americans.

Only one previous study examined effects on SCCHN incidence of any SNPs in oxidative stress pathways (26); it reported that rs2758346 in SOD2 (which we did not study) was not associated with SCCHN.

We found no evidence of interaction with alcohol consumption for any oxidative stress SNP.

Three SNPs in SOD1 that had inverse effects in the 2 races were part of the SOD1 haplotype that was also associated with differential effects by race. The direction of effect for carrying the minor allele of each individual SNP was consistent with the haplotype effect. The same is true for the 2 SNPs in ADH1B and ADH4 that seemed to have different effects in African- and European-Americans.

CHANCE is one of the largest studies of head and neck cancer conducted in both African- and European-Americans. This study examined genetic polymorphisms in genes in the alcohol metabolism and oxidative stress biological pathways and estimated main effects of these polymorphisms along with their interaction with alcohol.

We selected tag SNPs to capture most of the variation in the 12 genes studied, rather than studying only missense SNPs within coding regions. However, an inherent limitation of genotyping common tag SNPs is that the method is likely to miss rare variants.

Because of small numbers of African-Americans compared with European-Americans, we could not definitively evaluate differences in SNP effects between races. We also had insufficient power to detect haplotype-drinking interaction because haplotypes were constructed and analyzed separately for African- and European-Americans. Small numbers overall also precluded precise estimation of interaction between SNPs and alcohol for allele frequencies less than 30%, or in relation to anatomic subsite.

Our study confirms findings of previous studies that the effects of many polymorphisms in alcohol metabolism pathways are modified by alcohol intake. However, most genetic variants in ALDH2 and CYP2E1 have been understudied and warrant additional investigation in light of the new associations that we report.

Our analysis of tag SNPs in GPX2, SOD1, and SOD2 has identified several that are associated with SCCHN, hypopharyngeal, and oral cavity tumors. Confirmation of these findings in diverse populations is warranted.

No potential conflicts of interest were disclosed.

This work was supported in part by the National Cancer Institute (R01-CA90731; 2T32 CA009330-26) and the National Institute of Environmental Health Sciences (P30ES10126).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.