Role of aldehyde dehydrogenases, alcohol dehydrogenase 1B genotype, alcohol consumption, and their combination in breast cancer in East-Asian women

The associations between genetic polymorphisms in ADH1B (rs1229984) and ALDH2 (rs671), alcohol consumption, the effect of a combination of the two polymorphisms, and breast cancer risk were studied in a population of East-Asian women. In this study, 623 breast cancer cases and 1845 controls, aged 40 or above, were included. The association between ALDH2 polymorphism and breast cancer risk was validated in 2143 breast cancer cases and 3977 controls. Alcohol consumption increased the risk of breast cancer regardless of ADH1B and ALDH2 genotypes. The rs671 polymorphism of ALDH2 was independently associated with increased breast cancer risk (OR = 1.27, 95% CI = 1.02–1.58 per increment of A). The ADH1B rs1229984 polymorphism, and combined effects of the rs671 and rs1229984 polymorphisms, did not reveal any significant association with breast cancer. Stratification by menopausal status revealed that rs671 gene polymorphisms were significantly associated with breast cancer only in postmenopausal women (OR = 1.45, 95% CI = 1.03–2.05 per increment of A). This is the first study to demonstrate an independent association between ALDH2 gene variants and breast cancer in Asian women. Further studies are warranted to further elucidate the etiology of breast cancer as it relates to alcohol consumption in Asian women.

Alcohol consumption is a major risk factor that adversely affects the health of people worldwide 1 . In 2007, the International Agency for Research on Cancer (IARC) classified alcoholic beverages as group 1 carcinogens for the pharynx, larynx, esophagus, liver, colorectum, and female breast based on evidence on the toxicity of ethanol from epidemiological and animal studies 2 . Following alcohol intake, the enzyme alcohol dehydrogenase (ADH) oxidizes alcohol to acetaldehyde, which is highly toxic and carcinogenic to cells and tissues, and is detoxified to acetate by aldehyde dehydrogenases (ALDH) 3 . Thus, gene variants involved in alcohol metabolism and detoxification are also thought to affect the genetic susceptibility to alcohol-associated cancers. Among these, polymorphisms in the ADH1B and ALDH2 genes are known to play a pivotal role in cancer development in combination with alcohol consumption 4 . The active ADH1B allele results in approximately 40 fold higher metabolism of alcohol to acetaldehyde compared to the inactive allele 5 . In cases with mutant ALDH2 genes, ALDH2 enzyme activity is suppressed and acetaldehyde metabolism in the liver is impaired; subsequent toxic reactions cause symptoms including facial flushing, nausea, headache, or palpitations after alcohol intake; those with deficient variants of ALDH2 genes are likely to either consume less alcohol or be highly addicted 2,3,6,7 .
In general, epidemiological and animal studies have consistently correlated high alcohol intake with increased breast cancer risk in women 2 ; however, the gene variants involved in alcohol metabolism or detoxification have only been inconclusively associated with breast cancer risk 8,9 . Considering increases in both alcohol consumption in women 10 and incidence of breast cancer and related mortality in Asian countries 11 , in combination with the higher distribution of variant genes involved in alcohol metabolism or detoxification in the Asian

Results
The general characteristics of the patients with breast cancer from the National Cancer Center (NCC) and age-matched controls from the Korean Genome Epidemiology Study (KoGES) are described in Table 1. Except age distribution and mean age, which were matched between cases and controls, the distribution of the other variables differed significantly between the cases and controls. In particular, the proportion of ever drinkers and those who drink >12 g ethanol/day was higher in the breast cancer group. Table 2 presents the association between alcohol consumption, ALDH2 rs671 polymorphism, ADH1B rs1229984 polymorphism, and breast cancer risk. Compared to the non-drinkers, the ever drinkers and those who consumed 12 g alcohol or more per day exhibited increased association with breast cancer (OR = 1.27, 95% CI = 1.01-1.61, P-value = 0.042; OR = 2.15, 95% CI = 1.28-3.59, P-value = 0.004). The rs671 genotype, although unassociated with breast cancer in an adjusted model, was significantly associated with breast cancer risk after the model was additionally adjusted for daily alcohol intake in patients with a GA or AA genotype compared with those presenting a GG homozygote (OR = 1.29, 95% CI = 1.00-1.65, P-value = 0.045) and per increment of A (OR = 1.27, 95% CI = 1.02-1.58, P-value = 0.032). The ADH1B rs1229984 polymorphism did not exhibit any significant association. The combined effects of the ALDH2 rs671 and ADH1B rs1229984 polymorphisms are presented in Appendix Table 2. No significant association was observed between the combination of the two polymorphisms and breast cancer risk. Table 3 lists the effect of alcohol consumption according to ALDH2/ADH1B polymorphism. In those homozygous for the GG allele of rs671, the effects of ever consuming alcohol or daily alcohol intake were associated with increased risk of breast cancer (OR = 1.54, 95% CI = 1.17-2.04, P-value = 0.002; OR = 2.57, 95% CI = 1.45-4.53, P-value = 0.001). In terms of the rs1229984 variant of the ADH1B gene, those homozygous for TT exhibited a significantly increased risk of breast cancer in both ever drinkers and those who daily consumed >12 g day alcohol (OR = 1.58, 95% CI = 1.15-2.17, P-value = 0.005; OR = 2.97, 95% CI = 1.49-5.94, P-value = 0.002); however, in women with a TC or CC genotype, no association was observed between alcohol consumption and breast cancer. When we combined the ALDH2/ADH1B polymorphisms, those homozygous for the major allele for both rs671 and rs1229984 (GG and TT) presented increased risk of breast cancer in ever drinkers (OR = 1.76, 95% CI = 1.22-2.57, P-value 0.003) and those who consumed 12 g alcohol or more per day (OR = 3.37, 95% CI = 1.59-7.22, P-value = 0.002).
After investigating 2143 breast cancer cases and 3977 controls, the polymorphism in the ALDH2 gene showed increased breast cancer risk in all genetic models (Appendix Table 3). The OR of GA/AA compared with that of GG was 1.13 (95% CI = 1.01-1.27, P-value = 0.034) and an increment in the A allele increased breast cancer risk by 1.14-fold (95% CI = 1.03-1.26, P-value = 0.013).

Discussion
This is the first study suggesting that variants of the ALDH2 gene (rs671) increase the risk of breast cancer independently, after adjusting for other covariates such as alcohol consumption, in Asian women, and in particular in postmenopausal women. In this study, both ever consuming alcohol and high dose alcohol consumption (>12 g/ day) increased the risk of developing breast cancer. However, the association between alcohol consumption and breast cancer was more prominent in those with ADH1B His/His or those that were homozygous for the major allele in both ADH1B and ALDH2.
The association among ever consuming alcohol, high alcohol consumption (>12 g/day), and breast cancer in all women observed in the present study was in accordance with that in previous studies and reviews 8,13 . Previous studies have reported inconsistent results, wherein they evaluated whether menopausal status modifies the effect of the association between alcohol consumption and breast cancer risk 8,14 . In the present study, this association was observed only in premenopausal women. A possible explanation for the nonsignificant result in postmenopausal women could be that fewer women ever consumed alcohol or high alcohol content drinks; thus, the number was insufficient to statistically power our results. Considering that several studies suggest a modified effect of known genetic and environmental factors such as MPO gene polymorphism, family history, obesity, or alcohol use by menopausal status 14,15 , a detailed approach stratified by menopausal status with larger sample size is warranted.
ADH1B and ALDH2 are the major enzymes involved in alcohol metabolism in humans. The role of the ALDH2 gene has been widely studied in East-Asian countries because of the low prevalence of its variants in European populations. It has been reported that variants of the ALDH2 gene are correlated with an increased susceptibility to numerous cancers, particularly of the upper digestive organs, such as esophageal or gastric cancer 16,17 ; however, for other cancer types, the results were generally nonsignificant or inconsistent 16 www.nature.com/scientificreports www.nature.com/scientificreports/ cancer, only three studies have investigated the role of the ALDH2 gene (rs671); however, none of them reported a significant association. In all three studies, the controls were recruited from women who visited the same hospital as the patients [19][20][21] . Although controls did not have cancer or other systemic diseases, and the proportion of alcohol consumption was in accordance with that in our study population, their inclusion may not substitute for the general population due to Berkson's bias, a type of selection bias 22 . In our study population, the controls belonged to a population-based cohort, presumably free of Berkson's bias, thus indicating a better reflection of females in general, and even after adjusting for alcohol consumption, polymorphisms in the ALDH2 gene independently increased breast cancer risk. The association between ALDH2 gene variants and breast cancer was validated in independent cases and all sampled control populations. In addition, compared with the previous studies in which the number of breast cancer patients and controls was less than 500 [19][20][21] , the present study has a higher number of patients and 1:3 matching was used to increase the statistical power. In combination with the possible side effects of paclitaxel, a chemotherapeutic agent, due to alcohol's role as a solvent 23 women with the GA/AA genotype of ALDH2 would be highly susceptible to not only breast cancer but also side effects of breast cancer treatment, and are therefore a population with a potentially high risk for delterious health effects.
ADH1B action results in higher blood acetaldehyde levels, which can increase DNA damage and lead to cancer 24 . In people with ADH1B His/His, the activity of ADH1B is enhanced, followed by rapid accumulation of acetaldehyde 25 . In epidemiological studies, ADH1B polymorphism has been associated with tumors of the upper digestive organs 4 including esophageal cancer 26 . In terms of breast cancer, ADH1B itself did not present any association in most previous studies 19,27,28 . One study reported a marginally protective association for ADH1B IVS1 + 896 A > G with breast cancer risk; however, the same SNP considered in this study did not exhibit any association 29 . Another study reported an interactive effect of ADH1B and alcohol consumption on breast cancer risk 27 . In the present study, although we did not identify a significant association between ADH1B polymorphism and breast cancer risk, a decreasing trend was observed in those with the C allele, in particular the homozygotes of the C allele, which was consistent with previous studies 29 . Another study reported that the effects of the ADH1B and ALDH2 genes were mediated via alcohol consumption and decreased hepatocellular carcinoma risk 30 , which was not consistent with a previous meta-analysis 16 . In a study by Lie et al., the population included chronic hepatitis B carriers and the effect of alcohol was higher 30 than seen in the present study or other studies conducted in the Korean population 31 . Considering the subgroup of women with possibly higher alcohol susceptibility, the www.nature.com/scientificreports www.nature.com/scientificreports/ effects of the ADH1B and ALDH2 genes on alcohol drinking behavior and breast cancer risk might require further investigation.
In the present study, the association between alcohol consumption and breast cancer risk stratified by ALDH2 and ADH1B polymorphism status, despite the presence of variants and alcohol intake, specifically a high dose of alcohol intake, increased the risk of breast cancer. Among the variants studied, those with a TT genotype for rs1229984 exhibited higher risk of breast cancer in ever drinkers or in those who consume more than 12 g of ethanol per day. These results are contradictory to those from a study in German women, wherein a reduction in breast cancer risk associated with 12 g or more alcohol consumption per day was observed 27 . The frequency of the His/Arg allele of ADH1B gene varies among populations, specifically the Asian and European populations 32 , and is based on diet or other lifestyle factors; moreover, other interacting genes may presumably lead to different results between studies.

Cases Controls
Odds ratio a P-  Table 3. Association between alcohol consumption and breast cancer risk according to ALDH2 rs671 or ADH1B rs1229984 polymorphism. a Unadjusted odds ratio; b Adjusted for age, age at menarche, age at first full term pregnancy, age at menopause, body mass index, and smoking status.
It has been reported that both ADH1B and ALDH2 present strong association with alcohol dependence. People with the A allele of rs671 display less alcohol dependence, possibly owing to the unpleasant effects of a higher flush response 33 , and those with the rs1229984 His allele also show a decreased risk of alcohol dependence, especially in Asian populations 32,34 . A genome-wide association study for alcohol consumption revealed that the rs671 A allele was associated with decreased prevalence of drinkers and number of drinks, and that the rs1229984 T allele was associated with fewer drinks but not with drinking status 35 . In this study population, women who carried the GG allele of rs671 consumed alcohol with >12 g ethanol/day more often compared to those with the GA/AA genotype (P-value < 0.001), as in previous studies 33,35 ; however, their distribution was similar regardless of rs1229984 genotype. Previous studies included both men and women 32,35 , and sex-differences in terms of alcohol intake pattern 35 and alcohol metabolism may affect inconsistencies in the results.
Women who carried both the GG and TT genotypes for rs671 and rs1229984, respectively, presented a higher risk of breast cancer when they had ever consumed alcohol or more than 12 g ethanol per day, suggesting that for those with increased activity of the enzymes involved in alcohol metabolism and detoxification, the effects of alcohol consumption on breast cancer risk are increased. In these populations, alcohol would be eliminated faster than in other populations. Although highly active ADH1B and ALDH2 may reduce the effects of alcohol and its toxic metabolites such as acetaldehyde, the effects of alcohol drinking such as intoxication and the recovery period may result in increased alcohol exposure.
Although a study conducted in Korea revealed marginally decreased association between the Lys allele of the ALDH2 gene and breast cancer in postmenopausal women and no association in premenopausal women 20 Table 4. Association among alcohol consumption, ALDH2 rs671 polymorphism, ADH1B rs1229984 polymorphism, and breast cancer risk based on menopausal status. a Adjusted for age, age at menarche, age at first full term pregnancy, age at menopause, body mass index, smoking status, and alcohol consumption (for the association with ALDH2 or ADH1B) or ALDH2/ADH1B (for the association with alcohol consumption).
www.nature.com/scientificreports www.nature.com/scientificreports/ This study has certain limitations. First, the existence of selection bias should be considered. To increase comparability, cases and controls were individually matched by age. Although external validity may be limited, the prevalence of variants in ALDH2 and ADH1B genes is comparable with that in the general population 37 . Second, recall-bias could affect the results for measurements of alcohol consumption and other covariates. Some patients were recruited while receiving treatment, which may also affect the results. Third, the effects of other genes related to alcohol metabolism, such as Cytochrome P450 2E1, were not considered. Fourth, although this study included a larger sample size compared with previous studies on the association between ALDH2, ADH1B, alcohol consumption, and breast cancer, the statistical power may still be limited for elucidating gene-gene and gene-alcohol consumption interactions, and stratified analysis. In addition, some of the non-significant results from our subgroup analysis could be attributed to the low power of this study. Fifth, some variables that may affect both alcohol intake and breast cancer such as diet and exercise, were not considered because of limited available information and may have been incompletely adjusted for.
This study is the first to reveal an independent association between variants of the ALDH2 gene and breast cancer in Asian women with results validated from a large number of cases and controls. Alcohol consumption increased breast cancer risk, but for cases homozygous for the major alleles of ADH1B or/and ALDH2 polymorphisms, the increment was more prominent, suggesting that this was a high-risk population for breast cancers associated with alcohol consumption. Further studies with a larger sample size could help confirm the association between alcohol intake and gene interaction and breast cancer risk.

Methods
Study population. In total, we selected 750 women recently diagnosed with either invasive or in situ breast cancer, as confirmed by histology, at the NCC in Korea between February 2013 and October 2016. Healthy controls were selected from the KoGES, a population-based cohort constructed by the National Research Institute of Health (NIH), and the Centers for Disease Control and Prevention of Korea. Full details of the KoGES are described elsewhere 38 . Female control participants with a history of cancer were excluded, and 3986 women were considered for the study. Among these, 3977 controls provided information on alcohol consumption. All participants were from Korea, and written informed consent was obtained before enrolment. This study was approved by the National Cancer Center Institutional Review Board (IRB No. NCC 2015-0177). All procedures were performed in accordance with the Declaration of Helsinki.
Data collection. All cases and controls were asked to complete a set of interviewer-assisted questionnaires on demographics, medical history, reproductive factors, and lifestyle (including alcohol intake). The subjects were questioned about alcohol consumption, their present drinking status, as well as the total duration of alcohol consumption for both present drinkers and those with a history of alcohol intake. In addition, the type of alcoholic beverage, weekly intake frequency, amount of alcohol consumed on each occasion, and serving sizes were evaluated. The subjects were classified according to drinking status (never and ever). Among ever drinkers, subjects were further classified according to alcohol intake: ≤12 g pure alcohol/day and >12 g pure alcohol/day based on the amount of ethanol in a standard drink.
Genotyping and final study population. Genomic DNA was extracted from peripheral blood leukocytes and genotyped using the Korea Biobank Array with 833,000 SNPs. The Korea Biobank Array is a chip optimized for genetic studies in the Korean population. Full details on Korea Biobank Array are described elsewhere 39,40 . Of the 750 breast cancer cases and 3977 controls, 5 cases were excluded owing to a low call rate of <97% (n = 1) and family relationships (n = 4); after quality control, 745 cases and 3977 controls were evaluated. SNPs were excluded for low quality based on SNPolisher analysis (n = 36,816), low call rate <95% in either the cases or controls (n = 371), deviation from Hardy-Weinberg equilibrium (p-value <1 × 10 −6 ) in the controls (n = 713), and monomorphic characteristics in our population (n = 107,204). In total, 688,196 markers were evaluated for quality control.
The SNPs rs671 and rs1229984, in the ALDH2 and ADH1B genes, respectively, were analyzed. The cases and controls were individually matched by age at a 1:3 ratio, and ultimately 623 breast cancer cases and 1845 controls aged 40 or above were included in the final analysis. Discrepancies between the evaluated cases and controls were 0% and 0% for rs671 and 0% and 0.6% for rs1229984, with a P-value of 0.922. These SNPs had minor allelic frequencies of 15.5% and 25.2% and met the Hardy-Weinberg equilibrium in the control samples.
Validation of the association between ALDH2 polymorphism and breast cancer. Data from 2143 breast cancer patients who were recruited as part of the KoGES were evaluated according to their sex, age, and genotype to validate the association between ALDH2 polymorphism and breast cancer. Demographic and lifestyle information, including on alcohol intake, was unavailable for these patients. Peripheral blood DNA was genotyped using an Affymetrix Genome-wide Human SNP Array 6.0 (Affymetrix Inc., Santa Clara, CA, USA). As ALDH2 (rs671) was not included in the array, genotype imputation for rs671 was performed. The SHAPEIT (v2.r837) was used for prephasing, and imputation was performed using IMPUTE2 (2.3.2). Data from the 1000 Genome Project phase 3 East-Asian Ancestry sample (n = 504) were applied as a reference panel. The quality control process and imputation was conducted as follows: among all subjects, those with a sex mismatch (n = 3) were excluded. No subjects showed a low call rate or family relationships. SNPs were excluded in cases of low call rate <95% (n = 151,323), deviation from Hardy-Weinberg equilibrium (p-value <1 × 10 −6 ) in the controls (n = 897), or monomorphic traits in the population (n = 1710). In total, 513,738 markers were subjected to quality control. After imputation, the quality score was R 2 = 0.57. The age distribution of the 2143 breast cancer patients and 3977 controls is presented in Appendix Table 1.