Abstract

This study investigated the association of ADH1B (rs1229984) and ALDH2 (rs671) polymorphisms with glucose tolerance status, as determined by a 75-g oral glucose tolerance test, and effect modification of these polymorphisms on the association between alcohol consumption and glucose intolerance in male officials of the Self-Defense Forces. The study subjects included 1520 men with normal glucose tolerance, 553 with prediabetic condition (impaired fasting glucose and impaired glucose tolerance), and 235 men with type 2 diabetes. There was an evident interaction between alcohol consumption and ADH1B polymorphism in relation to type 2 diabetes (interaction ). The ALDH2 Lys allele was associated with a decreased prevalence odds of type 2 diabetes regardless of alcohol consumption. In conclusion, the ADH1B polymorphism modified the association between alcohol consumption and type 2 diabetes. A positive association between alcohol consumption and type 2 diabetes was confounded by ALDH2 polymorphism.

1. Introduction

Moderate alcohol consumption has generally been associated with decreased risk of type 2 diabetes, as summarized in a meta-analysis of 15 prospective cohort studies [1]. However, the results from these studies are not necessarily consistent, especially regarding high alcohol consumption and diabetes risk. While several studies showed a U-shaped relationship between alcohol and diabetes risk [1], others reported increased risks of type 2 diabetes in alcohol consumption categories of 25 g/day [2], 40 g/day [3], and 3 drinks per day [4]. Another study found a progressive decrease in the risk of type 2 diabetes up to a consumption of 50 g of alcohol per day [5]. The inconsistent results may be ascribed to differences in ascertainment of alcohol consumption and diabetes mellitus among studies and different genetic susceptibilities to alcohol exposure among study populations.

Ethanol is first oxidized to acetaldehyde by alcohol dehydrogenase (ADH), and acetaldehyde is further metabolized to acetate by aldehyde dehydrogenase (ALDH). Human ADH exhibits several isoenzymes, and functional polymorphisms are known for the ADH1B and ADH1C genes [6, 7]. The ADH1B Arg47His polymorphism (rs1229984) affects the enzyme activity substantially, and the ADH1B*47His allele (alternatively ADH2*2) is associated with faster oxidation. The ADH1C Ile349Val polymorphism (rs698) influences ADH activity to a lesser extent, and the ADH1C*349Ile allele (alternatively ADH3*1) is associated with moderately faster oxidation [8]. The ADH1B*47His allele is the major allele in Asians and is very rare in Caucasians, while the ADH1C*349Val allele is rare in Asians and fairly common in Caucasians [7, 9]. The ALDH2 gene encodes mitochondrial ALDH which contributes to acetaldehyde oxidation in human liver and contains a functional polymorphism of ALDH2 Glu487Lys (rs671). The variant ALDH2*487Lys allele (alternatively ALDH2*2) results in an inactive form and is almost exclusively found in Asian populations [9, 10].

Few studies have addressed the relation of these genetic polymorphisms to glucose metabolism and type 2 diabetes. Cross-sectional studies in Japan reported higher concentrations of fasting plasma glucose in men with ADH1B*47Arg/Arg genotype who consumed 10 g of alcohol per day [11] and lower concentrations of fasting plasma insulin associated with ADH1B*47Arg allele [12]. On the other hand, the ADH1C*349Val variant allele was shown to attenuate a decreased risk of type 2 diabetes with alcohol consumption in the United States [13]. As regards ALDH2 Glu487Lys polymorphism, ALDH2*487Lys allele was associated with deterioration in glycemic control assessed by hemoglobin concentrations in Japanese patients with type 2 diabetes who had a habitual light to moderate alcohol consumption [14]. In this paper reported here, we examined the relation of ADH1B Arg47His and ALDH2 Glu487Lys polymorphisms to glucose tolerance status determined by a 75-g oral glucose tolerance test (OGTT) in middle-aged Japanese men, focusing on the effect modification of these genetic polymorphisms on the association between alcohol consumption and glucose intolerance.

2. Methods

2.1. Study Population

Study subjects were male officials in the Self-Defense Forces who received a preretirement health examination at the Self-Defense Forces Fukuoka Hospital (Kasuga, Japan) or Kumamoto Hospital (Kumamoto, Japan) during the period from January 1997 to March 2001. The preretirement health examination is a nationwide program offering comprehensive medical examinations including a 75-g OGTT for those retiring from the Self-Defense Forces. Details of the health examination have been described elsewhere [15, 16]. A sample of 7 mL fasting venous blood was obtained for the purpose of medical research with written informed consent. The study was approved by the ethics committee of Kyushu University Faculty of Medical Sciences.

The present study included 553 cases of prediabetic condition, 235 cases of type 2 diabetes, and 1520 controls of normal glucose tolerance. In the consecutive series of 2454 men aged 46–59 years, 121 men were excluded for the following reasons: chronic hepatitis or liver cirrhosis (n = 49), use of steroids (n = 6), past history of gastrectomy (n = 38), history of insulin treatment for type 2 diabetes (n = 11), and undetermined glucose tolerance status (n = 19). Of the remaining 2333 men, 25 men were further excluded due to lack of DNA sample (n = 23) and unsuccessful genotyping for both of the two polymorphisms (n = 2). Thus, a total of 2308 men remained in the analysis.

2.2. Determination of Glucose Tolerance Status

After an overnight fast, a 75-g OGTT was performed. Plasma glucose levels were assayed by the glucose oxidase method using commercial reagents (Shino Test Co. Ltd., Tokyo) at each laboratory of the two hospitals. Subjects were classified as having normal glucose tolerance, impaired fasting glucose (IFG), impaired glucose tolerance (IGT), or type 2 diabetes in accordance with the World Health Organization criteria in 1998 [17]. Men under dietary or drug treatment for type 2 diabetes were regarded as having type 2 diabetes, irrespective of their glucose levels. The number of men with normal glucose tolerance, IFG, IGT, and type 2 diabetes were 1520, 137, 416, and 235, respectively. IFG and IGT were combined as prediabetic condition.

2.3. Lifestyle Questionnaire

A self-administered questionnaire was used to ascertain alcohol use, smoking habits, coffee intake, and other lifestyle characteristics. Details have been described previously [15, 16]. In brief, alcohol drinkers were defined as those who had drunk once a week or more over a period of one year or longer. Past alcohol drinkers were separated from lifelong nondrinkers. Daily ethanol intake was estimated for current alcohol drinkers, based on consumption frequencies and amounts of five different types of alcoholic beverages (sake, shochu, beer, whisky/brandy, and wine) on average in the past year. Estimated alcohol intake was highly valid in comparison with the intake derived from 7-day diet records in four seasons [18]. Cumulative exposure to cigarette smoking was expressed as cigarette-years, which were calculated by multiplying the average number of cigarettes per day by total years of smoking. Weekly frequency of coffee consumption was inquired about, and the number of cups of coffee consumed per day was ascertained among daily coffee users. As regards physical activity, men with a regular participation in leisure-time physical activity in the past year (one or more times a week) reported at most three types of physical activities together with frequency per week and time spent per occasion for each activity. Type of physical activity was classified into light, moderate, heavy, or very heavy activity in terms of metabolic equivalent (MET). The time spent in recreational exercise was multiplied by the corresponding MET value (light 2, moderate 4, heavy 6, and very heavy 8) to yield a MET-hour score per week. Parental history of diabetes mellitus was also elicited.

2.4. Genotyping

DNA was extracted from the buffy coat using a commercial kit (QIAGEN GmbH, Hilden, Germany) and genotyping was performed with the polymerase chain reaction (PCR) restriction fragment length polymorphism method. The PCR was performed in a reaction mixture of 10  L containing 0.5 units of Taq and 1  L of template DNA with a concentration of approximately 50–150 ng/ L. The ADH1B Arg47His genotypes were determined according to the method described by Osier et al. [7], and the ALDH2 Glu487Lys genotypes were determined by the method described by Goedde et al. [19].

2.5. Statistical Analysis

Departure from the Hardy-Weinberg equilibrium of the genotype distribution was tested by χ2 test with 1 degree of freedom. Associations of the genetic polymorphisms with prediabetic condition and type 2 diabetes were evaluated using logistic regression analysis, with those of normal glucose tolerance as controls. Odds ratio (OR) and 95% confidence interval (CI) were obtained from the logistic regression coefficient and standard error for the corresponding indicator variables. Statistical adjustment was made for age (continuous variable), hospital (Fukuoka or Kumamoto hospital), rank in the Self-Defense Forces (low, middle, or high), body mass index ( 22.5, 22.5–24.9, 25.0–27.4, and 27.5 kg/m2), cigaretteyears (0, 1–399, 400–799, and 800), alcohol consumption (never, past use, and current use with consumption of 30, 30–59, or 60 mL/day), coffee intake ( 1, 1-2, 3-4, and 5 cups per day), physical activity (categorized at quartiles of MET-hours per week), and parental diabetes. The trend of an association was tested by the Wald statistic, using an ordinal score for a variable of interest. The interaction was evaluated by the likelihood ratio test, comparing models with and without interaction terms. Statistical significance was declared if the two-sided P-value was less than  .05 or if the 95% CI did not include unity. All statistical analyses were performed using SAS version 9.1 (SAS Institute Inc., Cary, NC).

3. Results

Selected characteristics of the study subjects by glucose tolerance status are summarized in Table 1. Body mass index was greater in men with prediabetic condition and type 2 diabetes, and alcohol intake was greater in the former than in those with normal glucose tolerance. Parental history of diabetes mellitus was more frequent in individuals with type 2 diabetes and showed little difference between normal glucose tolerance and prediabetic condition. Coffee consumption was lower in those with prediabetic condition. Age, smoking, and leisure-time physical activity did not vary much according glucose tolerance status.

The ADH1B and ALDH2 genotypes were not determined in 21 subjects and 5 men, respectively. Genotype distributions of the ADH1B Arg47His and ALDH2 Glu487Lys polymorphisms in individuals of normal glucose tolerance were each in agreement with the Hardy-Weinberg equilibrium ( for ADH1B and for ALDH2).

While the ADH1B Arg47His polymorphism was not associated with either prediabetic condition or type 2 diabetes, the ALDH2 polymorphism was associated with a substantial decrease in the prevalence odds of type 2 diabetes and with a modest decrease in the prevalence odds of prediabetic condition (Table 2). The OR of type 2 diabetes decreased stepwise in men heterozygous and homozygous for the ALDH2*487Lys allele regardless of adjustment for the covariates.

There was no measurable interaction between the ADH1B and ALDH2 polymorphisms for either prediabetic condition or type 2 diabetes (Table 3). Individuals homozygous for ALDH2*487Lys allele were relatively few, and they were combined with those heterozygous for the polymorphism. The ALDH2*487Lys allele was associated with a statistically significant decrease in the prevalence odds of type 2 diabetes in the ADH1B*47His/His and *47Arg/His genotypes, while such a decrease was not clear in the ADH1B*47Arg/Arg genotype. Decreases in the prevalence odds of prediabetic condition associated with the ALDH2*487Lys allele were modest across the ADH1B genotypes.

Alcohol drinking differed substantially by ALDH2 genotype and slightly so with respect to the ADH1B polymorphism (Figure 1). Both current alcohol drinkers (hatched bar) and heavy alcohol drinkers (black bar) were less frequent with increasing numbers of the ALDH2*487Lys allele and were slightly more frequent with increasing numbers of the ADH1B*47Arg allele. High alcohol consumption was associated with increased prevalence odds of prediabetic condition and type 2 diabetes (Table 4). There was a dose-dependent increase in the OR of prediabetic condition, but no such increase in the OR was noted for type 2 diabetes.

In the analysis on the interaction of alcohol and genotype, past alcohol drinkers (n = 62) were excluded, and men homozygous for the ALDH2*487Lys allele (n = 117) were also excluded in the interaction analysis on ALDH2 genotypes because their alcohol consumption was markedly low. There was a statistically significant interaction between alcohol use and ADH1B polymorphism in relation to type 2 diabetes while a positive association between alcohol consumption and prediabetic condition did not appreciably vary according to ADH1B genotypes (Table 5). A positive association between alcohol consumption and type 2 diabetes was observed in the ADH1B*47His/His genotype, and the association was in an inverse direction in the ADH1B*47Arg/Arg genotype, which was associated with an increased prevalence odds of type 2 diabetes in nondrinkers.

When stratified by ALDH2 genotype, alcohol consumption showed no clear positive association with type 2 diabetes although a suggestive, positive association between alcohol consumption and prediabetic condition was noted in the ALDH2*487Glu/Glu genotype. Decreases in the OR of type 2 diabetes associated with ALDH2*487Lys allele were observed regardless of alcohol consumption although such decreases were less notable for prediabetic condition (Table 6). With further adjustment for ALDH2*487Glu/Lys and Lys/Lys genotypes, adjusted OR (95% CI) of prediabetic condition for lifelong nondrinkers, past drinkers, and current drinkers consuming 30, 30–59, or 60 mL alcohol per day were 1.00 (referent), 0.99 (0.47–2.09), 1.28 (0.87–1.88), 1.38 (0.94–2.04), and 1.52 (1.02–2.27), respectively, and the corresponding values for type 2 diabetes were 1.00 (referent), 1.40 (0.57–3.44), 1.32 (0.75–2.32), 0.95 (0.53–1.69), and 0.99 (0.54–1.77), respectively.

4. Discussion

The present study showed an effect modification of ADH1B Arg47His polymorphism on the association between alcohol consumption and type 2 diabetes. Alcohol consumption was positively associated with type 2 diabetes in individuals harboring the ADH1B*47His/His genotype, but inversely in those with ADH1B*47Arg/Arg genotype. The latter genotype was associated with an increased prevalence odd of type 2 diabetes in the absence of alcohol exposure. ALDH2*487Lys allele was associated with a substantial decrease in the prevalence odds of type 2 diabetes, and a positive association between alcohol consumption and type 2 diabetes disappeared with stratification by ALDH2 genotype.

The overall lack of an association of the ADH1B Arg47His polymorphism with prediabetic condition or type 2 diabetes in the present study is in agreement with the previous cross-sectional observation regarding type 2 diabetes in Japan [12]. The effect modification of the ADH1B Arg47His polymorphism on the association between alcohol consumption and type 2 diabetes is a notable finding. However, the inverse association between alcohol consumption and type 2 diabetes in men with the ADH1B*47Arg/Arg genotype is in disagreement with previous findings [11, 12]. A cross-sectional study reported that fasting plasma glucose concentrations were higher in men, but not in women, with ADH1B*47Arg/Arg genotype than those with the His/His and His/Arg genotype combined when alcohol intake was high ( 10 g/day) [11] whereas another study found no difference in fasting plasma glucose or hemoglobin concentrations according to ADH1B genotypes [12]. In the latter, fasting plasma insulin concentrations were lower in both men and women with ADH1B*47His/Arg genotype than those with the His/His genotype while there was no difference in alcohol consumption between the His/His and His/Arg genotypes [12]. None of these studies have suggested adverse effects of ADH1B*47Arg allele on glucose metabolism in the absence of alcohol exposure. It should be noted that the estimated OR associated with ADH1B*47Arg/Arg genotype among neveralcohol drinkers was unstable because of very small numbers (2 cases and 7 controls).

The ADH1C*349Ile allele associated with fast oxidation was shown to be related to greater decrease in the risk of type 2 diabetes associated with alcohol consumption in the United States, suggesting an involvement of acetate, the end product of alcohol oxidation, in the protective association between alcohol and type 2 diabetes [13]. The ADH1C Ile349Val polymorphism was not assessed in the present study because the *349Val allele is extremely rare in the Japanese [9]. The ADH1B Arg47His polymorphisms are in linkage disequilibrium with the ADH1C Ile349val polymorphism in Asian and Caucasian [20, 21]. The present findings do not support the hypothesis that fast alcohol oxidation confers a greater decrease in the risk of type 2 diabetes associated with alcohol consumption. We do not have a prompt explanation for the discrepancy in the role of fast alcohol oxidation in the effect of alcohol on glucose metabolism.

Previously the ALDH2*487Lys allele was associated with deterioration in glycemic control in patients with type 2 diabetes who had a habitual light to moderate alcohol consumption [14], and the ALDH2*487Lys allele was associated with higher concentrations of fasting plasma glucose when alcohol consumption was high ( 5 g/day) [11]. Contrary to these observations, the present study suggests that individuals with the ALDH2*487Lys may have a lower risk of type 2 diabetes. It should be noted that a decreased risk of type 2 diabetes associated with ALDH2*487Lys allele was observed even in lifelong nondrinkers of alcohol. It is possible that individuals with ALDH2*487Lys allele may have had favorable ways of living in addition to abstinence from alcohol use. In the present study population, men with ALDH2*487Lys allele had a higher consumption of coffee, which has been related to decreased risk of type 2 diabetes [22], although they had a lower physical activity in leisure time (data not shown). A positive association between alcohol consumption and type 2 diabetes was totally ascribed to a confounding effect of the ALDH2 genotype. The present study does not support either a decreased risk of type 2 diabetes with moderate alcohol consumption or an increased risk associated with heavy alcohol consumption.

The effect modification of the ADH1B genotype on the association with alcohol consumption as well as the association with ALDH2 genotype was evident for type 2 diabetes, but not for prediabetic condition. We have no clear explanation for these differential effects. Further studies are needed to confirm the present findings.

This is the first large study regarding the ADH1B Arg47His or ALDH2 Glu487Lys polymorphism and glucose tolerance status, which was determined by the standard 75-g OGTT. However, there were several weaknesses to be discussed. The study subjects were not representative of Japanese men in the general population, but selection bias was unlikely to exist as to the genetic polymorphisms under study. The frequencies of ADH1B*47Arg (25%) and ALDH2*487Lys (24%) alleles were similar to those reported in Japanese populations elsewhere [9, 23]. An attrition bias is always possible in cross-sectional studies. A decreased risk of type 2 diabetes associated with ALDH2*487Lys allele would be observed if men carrying ALDH2*487Lys allele were more likely to leave the Self-Defense Forces earlier than the age of retirement when they developed type 2 diabetes.

5. Conclusions

In a cross-sectional study of middle-aged Japanese men, the ADH1B Arg47His polymorphism modified the association between alcohol consumption and type 2 diabetes. A positive association between alcohol consumption and type 2 diabetes was confounded by ALDH2 polymorphism.

Abbreviations

ADH:alcohol dehydrogenase
ALDH:aldehyde dehydrogenase.

Acknowledgments

The authors are grateful to the ward nurses of the Self-Defense Forces Fukuoka and Kumamoto Hospitals for their cooperation. This work was supported by a Grant-in-Aid for Scientific Research (A) (21249044) from the Japan Society for the Promotion of Science.