Interaction between alcohol consumption and methylenetetrahydrofolate reductase polymorphisms in thyroid cancer risk: National Cancer Center cohort in Korea

The effect of alcohol intake on thyroid cancer is unestablished, and its interaction effects with genetic susceptibility are unclear. In this case-control study, the relationship among alcohol intake, the methylenetetrahydrofolate reductase (MTHFR) gene, and thyroid cancer risk has been evaluated. In total, 642 cases and 642 controls of Korean origin were included, and the genetic variants C677T and A1298C of the MTHFR gene were analysed. The interactions between alcohol-consumption behaviour and genetic variants were analysed with a likelihood ratio test, wherein a multiplicative interaction term was added to a logistic regression model. There was an independent association between the C677T polymorphism and thyroid cancer risk but not for drinking history. For C677T C/C homozygotes, individuals with a history of alcohol consumption showed a protective OR (95% CI) of 0.42 (0.15–1.13) when never drinkers were used as the reference. However, this protective association was not observed among individuals with a T+ allele with an OR (95% CI) of 1.27 (0.89–1.82), showing different directions for the association between genotypes with a significant interaction (Pinteraction = 0.009). Based on the genetic characteristics of individuals included, an interaction between alcohol intake and MTHFR C677T may modify the risk of thyroid cancer.

habits, alcohol consumption is a well-known avoidable risk factor for many types of cancer, and the International Agency for Research on Cancer (IARC) has classified ethanol in alcoholic beverages as a Group 1 carcinogen in humans [7][8][9][10] . The adverse effect of alcohol on cancer risk has been studied broadly in accordance with folate deficiency, oxidative stress, and carcinogenesis on the molecular and genetic levels 11,12 . Interestingly however, several epidemiologic studies across diverse populations have suggested that alcohol consumption might have a protective effect on thyroid cancer [13][14][15][16] . A meta-analysis by Bagnardi et al. 9 that included 9 independent studies reported the protective odds ratio (OR) for light and moderate drinkers compared to that of non-drinkers. However, there is no established conclusion regarding the thyroid cancer risk and the intake of alcoholic beverages, with no sound biological mechanism to explain the protective association due to discrepancies in the reported relationships and the scarcity of studies. The inverse relationship between alcohol and thyroid cancer may be caused by effect modification by other complex risk factors such as smoking, diet, and genetic characteristics. To explore this inverse trend of risk association, we hypothesized that an individual's genetic characteristics might affect thyroid cancer risk and alcohol consumption differently. Heavy alcohol intake is known to cause folate deficiency due to intestinal malabsorption, decreased liver uptake, and increased excretion of folic acid via urine 17 . However, in this study, we did not explore the effect of folate itself but rather the genetic susceptibility of the methylenetetrahydrofolate reductase (MTHFR) gene, which not only plays a crucial role in folate metabolism affected by alcohol consumption but also is involved in DNA synthesis, the methylation cycle, and epigenetic regulation [18][19][20][21] and is a widely studied one-carbon metabolism-related gene with reported associations with all types of cancer, including thyroid cancer 20,22-24 . In this case-control study, we examined whether the independent risk association between either alcohol consumption or MTHFR genetic variants and thyroid cancer is present in a Korean population. The study also aimed to assess the combined effect of alcohol consumption and MTHFR genetic polymorphism on thyroid cancer risk.

Results
General characteristics of the study population. General demographic characteristics were compared between the thyroid cancer cases and controls ( Table 1). The proportion of individuals with a body mass index (BMI) over 25 was higher among the cases than the controls, while smoking status and education level did not differ between groups. Additionally, the proportion of individuals with a first-degree family history of thyroid cancer was significantly higher among the cases. The alcohol consumption status groups, including never drinkers, ex-drinkers, and current drinkers, showed significant differences based on disease status. Independent associations of thyroid cancer risk with alcohol consumption history and MTHFR polymorphisms. When the participants were classified as never or ever drinkers, alcohol consumption status was not significantly associated with thyroid cancer risk, although the OR estimate for ever drinkers compared to that of never drinkers suggested an inverse association (fully adjusted OR, 0.86; 95% confidence interval (CI), 0.67-1.11) ( Table 2). Additionally, no significant association was observed when stratified by sex. When the groups were stratified by average intake, light drinkers (<4.7 g of alcohol/day) showed a significant OR of 0.71 (95% CI, 0.51-0.98) compared to never drinkers, while moderate/heavy drinkers (≥4.7 g/day) showed an insignificant association. Moreover, this association was apparent only in moderate/heavy drinking females when stratified by sex, likely because there were more study subjects available in this group (Supplemental Table 1). The MAFs of A1298C and C677T were 0.17 and 0.46, respectively. The pair wise linkage disequilibrium (R 2 ) between C677T and A1298C was 0.17 (Supplemental Table 2). The effect of each genetic variant was analysed using the dominant model. Compared to A/A homozygotes of A1298C, individuals with minor allele C showed an OR of 0.80 (95% CI, 0.64-1.01) for thyroid cancer risk. By contrast, individuals with the C677T T+ genotype showed a significantly increased risk (OR, 1.37; 95% CI, 1.08-1.75) compared to individuals with the C/C genotype (Table 3).
Interaction between alcohol exposure and MTHFR variants in thyroid cancer risk. Stratification by alcohol consumption status and MTHFR variants in the dominant model was performed to analyse their interacting effect in association with thyroid cancer risk (Table 4). Compared to never drinkers, ever drinkers with the A/A genotype of A1298C had an insignificant OR of 0.94(95% CI, 0.79-1.76) while individuals with the C+ allele showed an OR of 0.56(95% CI, 0.20-1.60) (P for interaction = 0.559). For C677T C/C homozygotes, individuals with an alcohol consumption history showed a protective OR estimate of 0.42 (95% CI, 0.15-1.13) when never drinkers were used as the reference. However, individuals with a T+ allele of C677T showed an insignificant association with thyroid cancer (OR, 1.27; 95% CI, 0.89-1.82), indicating the opposite trend in the association's direction between different genotypes with a significant interaction (P for interaction = 0.009). When the interaction between alcohol consumption dose and MTHFR variants was evaluated to observe any dose-response effect, no significant interaction was observed (data not shown).

Discussion
Our findings demonstrated that a history of alcohol consumption may reduce thyroid cancer risk with limited evidence, and there was an independent association between the MTHFR genetic polymorphism C677T and thyroid cancer risk. C/C monozygotes of MTHFR C677T with a history of drinking alcohol showed a decreased risk of thyroid cancer, where an increasing trend was observed in individuals with a minor allele T. These associations support the importance of considering alcohol consumption and alcohol-related genetic variations when defining groups at high risk for thyroid cancer in a Korean population.  Higher alcohol intake is a global modifiable risk factor for cancer 25 that affects numerous organs and systems of the human body, including impaired folate metabolism, vitamin deficiency, oxidative stress and DNA damage, which may lead to carcinogenesis 26 . A recent study conducted in Korea reported that threshold effects of acute high-dose and long-term alcohol consumption are associated with an increase in thyroid cancer 27 . However, several prospective studies examining the risk association between alcohol use and thyroid cancer have consistently shown a clear reduction in the risk of thyroid cancer with higher alcohol consumption 15,16,[28][29][30] , supporting the results of our study. Previous studies have indicated that alcohol may affect thyroid cancer development differently by inhibiting the thyroid-specific hormonal response of elevated thyroid-stimulating hormone (TSH), which is related to increased thyroid cell proliferation, although its effect is not clear in humans 13,31 . Further studies are required to explore the negative association between alcohol and thyroid cancer risk.
The critical role that DNA methylation plays in gene regulation has been widely studied and implicated in the risk of developing chronic diseases including cancer 32,33 . The MTHFR gene is known to be involved in folate metabolism, playing an essential role in inherited DNA methylation profiles, and MTHFR C677T is a common polymorphism encoding an enzyme whose activity level is affected by the levels of folate 34 . Additionally, it has been reported that individuals with the TT genotype of MTHFR C677T have only approximately 30% of the MTHFR enzyme activity compared to CC homozygotes and 65% of the activity for CT 35 . There are few previous studies assessing the association between MTHFR C677T and thyroid cancer, and the reported findings are inconsistent. According to a case-control study that examined the association between germ-line polymorphism in the MTHFR gene and differentiated thyroid carcinoma (DTC), a 2.33-fold increased risk was observed for DTC in individuals with the 677TT genotype compared to the controls 20 . However, no significant associations between papillary thyroid cancer (PTC) and the T/T genotype of MTHFR C677T were reported in a previous Korean study, showing some discrepancies between different anatomical sites of thyroid carcinoma 36 . The findings of the present study support the presence of a minor allele T being associated with a 1.37-fold increased risk of thyroid cancer. Further replicated large-scale studies are necessary to validate the association between MTHFR C677T and thyroid cancer.
Genetic susceptibility that may account for numerous metabolic reactions, including activation or deactivation of metabolites and DNA repair, is affected by other susceptibility factors, such as demographics and nutritional status 37 . A significant interaction between history of alcohol consumption and the C677T polymorphism affecting thyroid cancer risk was observed in this study. It has been suggested that the cancer risk associated with MTHFR polymorphisms may be modulated by alcohol intake, but only a few previous studies have examined this issue in detail regarding thyroid cancer risk. A similar Brazilian study was recently conducted to determine the interaction between alcohol consumption and MTHFR C677T on thyroid cancer risk, but the interactive effect was not statistically significant (P for interaction = 0.84) 38 . The inconsistency between the Brazilian study and the current study may be due to differences in genetic characteristics of the studied populations, which may affect certain biochemical reactions after alcohol consumption; for instance, the ALDH2 genetic variant is prevalent only in East Asians and is associated with an inability to metabolize a carcinogenic alcohol metabolite, acetaldehyde, into acetate 39 . Moreover, differences in ethnicity, dietary intake, exposure to environmental carcinogens and sample structure may be responsible for the discrepancies in the results. Earlier epidemiological studies have provided evidence that an interaction between alcohol consumption and C677T variation might contribute to modification of the risk for various types of cancer, such as head and neck, oesophageal, and colorectal cancers [40][41][42][43][44][45] . To explain the interactions of alcohol with the C677T genotype in previous studies, folate deficiency in mitochondria modified by alcohol exposure has been proposed to increase the risk of cancer 41,44 . This may induce an imbalance in DNA precursors leading to modified DNA synthesis and repair, which may contribute to the loss of normal control of proto-oncogene expression 40 . Other theoretical hypotheses also relate to the role of alcohol in folate metabolism and MTHFR polymorphism. For instance, alcohol might act as a folate antagonist to modify the function of folate receptors, which might lead to modified DNA synthesis or methylation, which has been proposed to be involved in the carcinogenesis process 7,11,46 . To the best of our knowledge, the underlying mechanisms explaining these interactions are not clearly established, and the findings from alcohol-gene interaction analyses must be interpreted with caution due to several considerations, such as reduced numbers of observations in subgroups and different categories for alcohol consumption in each study, along with different ethnicities 45 . Therefore, further studies at the molecular level are required to conclusively understand the relationship between alcohol and folate mechanism.
There are several strengths of this study. First, the study participants, consisting of case and control groups, were individually matched by age and sex. The use of matching and conditional logistic regression was advantageous for optimizing a case-control study (OR to yield increases in efficiency). Moreover, the study population was relatively larger than those in previous studies exploring the interactive effects between alcohol intake and MTHFR polymorphisms. From this perspective, the present investigation was the first novel study to suggest an interactive effect, with meaningful results that were obtained after adjustment for diverse covariates such as BMI, education, smoking status, and family history of thyroid cancer. By contrast, there were certain limitations in this study, mostly related to the limited number of subjects whose genetic information was available. We focused on analysing the data based on the status of alcohol experience (ever/never) to minimize any recall bias that may have occurred if average alcohol consumption was used. Furthermore, due to limited information, histopathological types were available only for 326 of the total 642 cases. Among them, 315 (96.6%) were PTC cases. However, when the subgroup analysis was performed for only PTC cases, the interaction between MTHFR C677T polymorphism and alcohol consumption was still significant (P = 0.012, table not shown).
In conclusion, our study supports an inverse association of alcohol consumption and an association between MTHFR polymorphisms and thyroid cancer risk. Additionally, among C/C homozygotes of the MTHFR C677T polymorphism, ever drinkers showed a suggestive protective effect on thyroid cancer risk, whereas this effect was not observed in individuals with a T+ allele. As there are no sound explanations for the protective effect of alcohol on thyroid cancer risk, it is important to further explore whether this relationship is due to an interaction or is a proxy for other risk factors, which was significantly observed for MTHFR C677T polymorphism in this study.

Subjects and Methods
Study population. The subjects of this hospital based case-control study were selected from the participants of the Cancer Screening Program at the National Cancer Center (NCC) of South Korea 47 between August 2002 and July 2014 with a total of 41,109 individuals. Participants were individuals 30 years or older, who completed health-screening examinations as well as screening for selected cancers. At the baseline evaluation, the participants were asked to complete a self-administered questionnaire, and information on demographic characteristics, medical history, and lifestyle factors was collected. Subjects diagnosed with thyroid cancer (ICD10 code C73) were identified by linking to the Korea Central Cancer Registry (KCCR) database. From the subjects diagnosed with thyroid cancer (n = 1,106), individuals with missing questionnaires and blood biomarkers were excluded. Additionally, of the potential controls (n = 37,236), participants with a missing questionnaire or blood biomarker; those with a history of thyroid-related disease, surgery, or medicine usage; and those with a diagnosis of any type of cancer were excluded from the study. For the final analysis, 642 thyroid cancer cases (201 males, 441 females) and 642 individually age-and-sex matched controls (201 males, 441 females) were selected (Fig. 1). All of the participants in this study provided written informed consent, and the study protocol was approved by the Institutional Review Board of the NCC (#NCC2016-0088). All actual procedures utilized in the present study were performed in accordance with the guidelines and regulations of the IRB of the National Cancer Center.
Alcohol intake measurement. Information on alcohol consumption behaviours was obtained from all participating individuals. The questionnaire asked whether the participants had a history of ever consuming alcoholic beverages regularly. Additionally, they were asked to provide information on the average frequency of drinking occasions, types of alcoholic beverages consumed (i.e., beer, soju, rice wine, wine, and spirits), the average number of drinks consumed on each drinking occasion, and the serving size of each alcoholic drink from a previous year. The average total daily alcohol intake (g/day) was calculated by summing the beverage-specific alcohol consumed using the provided information. The study subjects were classified as a never drinker, light drinker if their average daily intake of alcohol was less than 4.7 g/day (which is the median level of alcohol intake for all drinkers), or moderate or heavy drinker if their average alcohol intake exceeded 4.7 g/day.
Genotype measurement. The study participants' genomic DNA was extracted from peripheral blood leukocytes isolated from the obtained whole-blood samples. Genotyping was performed using the Infinium OncoArray-500K BeadChip (Illumina Inc., San Diego, California) with 533,631 genetic variants. After quality controlling for monomorphic variants, a minor allele frequency (MAF) < 0.01, call rate <95%, and deviation from Hardy-Weinberg equilibrium (P-value < 1 × 10 −6 ), 345,675 single nucleotide polymorphisms (SNPs) were available for further analysis. The genetic variants C677T (rs1801133) and A1298C (rs1801131) of the MTHFR gene, which passed the above quality-control criteria, were selected for the analyses.
Statistical analyses. General characteristics of the participants of this study were compared between thyroid cancer cases and controls using chi-square tests for categorical factors and Student's t-tests for continuous variables. For the association analysis between alcohol consumption status, the MTHFR gene, and thyroid cancer risk, ORs and 95% CIs were calculated using conditional logistic regression for age-and-sex matched cases and controls. The model was further adjusted for BMI, smoking status, education level, and family history of thyroid cancer. Only the individuals with no missing data of confounders were included in the fully adjusted model. To evaluate the interaction between alcohol consumption and MTHFR SNPs, the likelihood ratio test, where the multiplicative term of alcohol consumption and the SNPs was inserted to the logistic regression model, was performed. The datasets generated and/or analysed during the current study are available from the corresponding author upon reasonable request.