Background Past studies have found a strong relationship between alcohol drinking and human health.
Methods In this study, we first tested the association of rs671 with alcohol use in 2349 participants in southeast China. We then evaluated the causal impact between alcohol use and cardiovascular traits through a Mendelian randomisation (MR) analysis.
Results We found strong evidence for the association of rs671 in the ALDH2 gene with alcohol drinking (p=6.08×10-47; ORadj G=4.50, 95% CI 3.67 to 5.52). We found that female G carriers of rs671 had a higher proportion of non-drinkers than male G carriers (88.01% vs 38.70%). In non-drinkers, the female G allele frequency was higher than the male G allele frequency (71.1% vs 55.2%). MR analysis suggested that alcohol use had a causal effect on blood pressure (increasing 9.46 mm Hg for systolic blood pressure (p=9.67×10-4) and 7.50 mm Hg for diastolic blood pressure (p=9.62×10-5)), and on hypertension in men (p=0.011; OR =1.19, 95% CI 1.04 to 1.36) and in pooled samples (p=0.013; OR =1.20, 95% CI 1.04 to 1.39), but not in women. We did not observe a causal effect of alcohol use on body mass index and lipid levels; further studies are needed to clarify the non-causal relationship.
Conclusions Compared to never-drinkers, current and previous alcohol use had a causal effect on blood pressure and hypertension in pooled samples and in men. These results reflect Chinese culture which does not encourage women to drink.
- blood pressure
- alcohol use
- mendelian randomization
- chinese population
Statistics from Altmetric.com
Observational studies have demonstrated that blood pressure is lower among individuals with light-to-moderate alcohol consumption than in non-drinkers, while an increase in blood pressure only occurs at higher levels of consumption.1 2 Epidemiological studies also show that alcohol consumption is relevant to obesity, high-density lipoprotein cholesterol (HDL-C) concentration3 and hypertriglyceridaemia.4 However, other factors, such as diet, smoking, exercise level, socioeconomic position and population structure, might confound the observational associations between alcohol consumption and cardiovascular traits. Aldehyde dehydrogenase-2, encoded by the ALDH2 gene on 12q24.2, is one of the important enzymes in the major oxidative pathway of alcohol metabolism. A mutation on the 12th exon on ALDH2 (NP_000681.2:p.Glu504Lys, rs671) could lead to the inactivation of this gene, while inactive ALDH2 has an inhibitory effect on alcohol drinking.5 There has been only one genome-wide association study (GWAS) of an isolated rural Chinese population sample to demonstrate that ALDH2 (rs671) was associated with alcohol dependence, flushing response and daily maximum drinks.6
Mendelian randomisation (MR) is an approach to infer the causality from associations between genetic variants that mimic the influence of environmental exposure and the outcome of interest.7 This technique exploits the idea that genotypes are distributed randomly at conception, facilitating their use as instrumental variables (IV) for investigating causality.8 Therefore, MR could be a better approach for investigating the role of alcohol use in cardiometabolic traits and vice versa, through strong genetic instruments without pleiotropic effects and sufficient statistical power.
In the present study, we first tested the association of rs671 with alcohol use within a Chinese population sample. Then we evaluated the causal impact among alcohol use and cardiovascular traits through an MR analysis. We tested whether these causal estimates from MR analyses were consistent with findings from observational analyses.
The study samples were recruited from a community in Zhejiang Province in southeast China, and included 2349 participants. Among the study participants, only unrelated samples were included. All study participants provided written informed consent. The study was approved by the Institutional Ethical Committee of Hangzhou Normal University, and was conducted according to the principles of the Declaration of Helsinki. The demographic characteristics and cardiovascular traits included age, sex, occupation, education, height, weight, waist, systolic blood pressure (SBP), diastolic blood pressure (DBP), total cholesterol (TC), triglyceride (TG), HDL-C and low-density lipoprotein cholesterol (LDL-C). All subjects were asked to be under fasting conditions in the morning and to not take any antihypertensive and lipid-lowering drugs. The nurse collected the peripheral venous blood samples. All the blood sample tube wall and the questionnaire were posted a consistent stream code. Behavioural measurements such as smoking, alcohol use and exercise were also included. Alcohol drinking was estimated by white wine and rice wine 25 g per cup, red wine 50 g per cup, and beer 100 g per cup. An alcohol drinker was defined as someone who consumed alcohol for ≥1 year with an average daily consumption of rice wine ≥50 g or red wine ≥150 g or beer ≥500 g. Alcohol use was defined as a dichotomous variable: drinker (including current drinker and previous drinker) and non-drinker (never drink). The other traits were continuous variables. The baseline survey was conducted from April to July 2013 and genetic testing was done in October 2013. Descriptive statistics of these exposures and outcomes were assessed by the Stata software package (version 14.1, StataCorp, College Station, Texas, USA).
SNP included in instrumental variable analysis
Single nucleotide polymorphism (SNP) rs671 is a missense variation on gene ALDH2, and has been identified as a genetic determinant for alcohol dependence, flushing response and daily maximum drinks in Europeans.
SNP genotyping and quality control
Genomic DNA was extracted from peripheral blood leucocytes by standard procedures using Flexi Gene DNA kits (Qiagen). DNA concentration was normalised to 15–20 ng/µl (diluted in 10 mM Tris and 1 mM EDTA) with a Nanodrop Spectrophotometer (ND-1000). Genotypes for SNPs selected were obtained using the Sequenom MassArray System (Sequenom iPLEX assay) according to the manufacturer’s instructions. Approximately 15 ng of genomic DNA was used to genotype each sample. Locus-specific polymerase chain reaction (PCR) and detection primers were designed using the MassARRAY Assay Design 3.0 software (Sequenom). The sample DNAs were amplified by multiplex PCR reactions, and the PCR products were then used for locus-specific single-base extension reaction. The resulting products were desalted and transferred to a 384-element SpectroCHIP array. Allele detection was performed using MALDI-TOF MS. The mass spectrograms were analysed by the MassARRAY TYPER software (Sequenom).
SNP rs671 satisfied the criteria with a call rate larger than 98% and with Hardy-Weinberg equilibrium (p≥0.05) (online supplementary table S1). Furthermore, we checked the cluster patterns of rs671 from the genotyping data from the Sequenom analyses to confirm their good quality.
Logistic regression of rs671
The association statistical analyses of rs671 were undertaken using the PLINK software package.9 We inspected for normality in the distribution of continuous variables in our study population and excluded extreme outliers (3.5 SD from the mean). The SNPs were coded as 0, 1 and 2, as the count of the effect allele. A logistic regression was used to assess the association of rs671 with alcohol use. Results were adjusted using a multivariate model that included the demographic characteristics of age and sex.
Observational analysis and MR analysis
We generated linear regression models for continuous outcomes (body mass index (BMI), waist circumference, SBP, DBP, TC, TG, HDL-C and LDL-C), adjusted for age and sex. Effect estimates were presented per 1 SD increases in continuous variables. Missing data points were assumed to be missing at random; thus, complete case analyses were performed.
We used rs671 as the IV to estimate the causal effect of alcohol use on other traits in MR analysis. We used the two-stage least-squares estimator method that regressed each outcome against predicted values per genotype in the Stata software package. The Wu-Hausman χ2 was used to test endogeneity in a regression estimated via IVs, where the null hypothesis states that an ordinary least-squares estimator of the same equation would yield consistent estimates. The null hypothesis was tested using α=0.05.
Characteristics of the study samples
In total, 2349 participants were included, of which 54.1% were women. The mean (SD) age was 59.23 (11.36) years at the time of baseline assessment, and was higher in men than women (60.34 years vs 58.29 years, p<0.001) (table 1). Height (165.87 cm vs 155.59 cm), weight (63.53 kg vs 56.67 kg) and waist circumference (82.46 cm vs 80.45 cm) were greater in men than women (p<0.001) (table 1). Weight and BMI were highly correlated with waist circumference (r=0.7104 and r=0.7222) (online supplementary figure S1). DBP was higher in men than women (82.67 mm Hg vs 81.05 mm Hg, p=0.006), while SBP did not differ between the sexes (p=0.429). TC and LDL-C values were higher in women than men (p<0.001), while HDL-C concentration did not differ between the sexes (p=0.311). The correlation between TC and LDL-C values was high (r=0.7719) (online supplementary figure S2).
Association of ALDH2 (rs671) with alcohol use
SNP rs671 is a missense variation on gene ALDH2. We tested the association between rs671 and alcohol use in 752 drinkers and 1574 non-drinkers. The G allele of rs671 had a frequency of 83.9% in drinkers and 66.5% in non-drinkers (table 2). Genotype frequencies were 42.9% (GG), 47.1% (AG) and 10% (AA) in non-drinkers, and 68.6%, 30.6%, and 0.80%, respectively, in drinkers (table 2). Adjusted for age and sex, the logistic regression test demonstrated that individuals carrying the G allele were disposed to drink alcohol (p=6.08×10-47; ORadj =4.50, ORhom =19.97, ORhet =8.12, ORGcarrier =13.78) (table 3). It is worth pointing out that, of the 752 drinkers, 99.20% were G carriers (individuals with GG and AG genotype) (table 2, online supplementary figure S3A), and the non-drinkers had a lower proportion of the GG genotype (online supplementary figure S3A); of the 163 individuals with AA genotype, 96.32% were non-drinkers (online supplementary figure S3B). The number of drinkers decreased when the genotypes of individuals changed from GG to GA to AA (online supplementary figure S3B); individuals who carried the G allele would trend towards being a drinker.
Despite the different drinking habits between men and women in the Chinese population, we found a strong association between rs671 and alcohol in both men (p=2.52×10-39; ORadj =5.02, ORhom =50.14, ORhet =12.19, ORGcarrier =24.71) and women (p=7.44×10– 10; ORadj =3.39, ORhom =15.94, ORhet =4.89, ORGcarrier =10.76) (table 3). Six hundred and five out of 610 male drinkers (99.18%) and 141 out of 142 female drinkers (99.30%) were G carriers (table 2). Similarly, 78 out of 83 men with AA genotypes (93.98%) and 79 out of 80 women with AA genotypes (98.75%) were non-drinkers (table 2). Female G carriers had a higher proportion of non-drinkers than male G carriers (1035/1176=88.01% vs 382/987=38.70%) (highlighted data in table 2, figure 1B vs figure 1A). In non-drinkers, the female G allele frequency was higher than the male G allele frequency (71.1% vs 55.2%) (bold data in table 2), and women had a higher proportion of G carriers than men (92.9% vs 83%) (table 2, online supplementary figure S4B vs figure S4A). These results reflect Chinese culture, details of which will be discussed below.
While alcohol use, BMI, blood pressure and lipid values were associated with age, sex and other factors, rs671 was not associated with age and sex, or any of the following potential confounders: exercise and smoking status (online supplementary table S2).
Using SNP rs671 for MR analysis, alcohol use was shown to have a causal effect on blood pressure (p=9.67×10-4, 9.46 (95% CI 3.8 to 15.1) mm Hg increase in SBP, and p=9.62×10-5, 7.50 (95% CI 3.7 to 11.3) mm Hg increase in DBP) (table 4). Assessment for endogeneity indicated that SNP rs671 was particularly meaningful for estimating effects of alcohol use on SBP (Wu-Hausman F=10.92, p=0.001) and DBP (Wu-Hausman F=13.13, p=0.0003).
When stratifying by sex, alcohol use (using SNP rs671 as IV) was shown to have a causal effect on blood pressure in men (p=1.37×10-3, 8.28 (95% CI 3.22 to 13.35) mm Hg increase in SBP, and p=1.28×10-4, 6.72 (95% CI 3.28 to 10.15) mm Hg increase in DBP), but not in women (p=0.135 for SBP and p=0.085 for DBP) (table 4). Endogeneity analysis was also performed and showed that SNP rs671 was useful for estimating effects of alcohol use on SBP (Wu-Hausman p=0.0055) and DBP (Wu-Hausman p=0.0008) in men, but not in women (Wu-Hausman p=0.0727 for SBP and Wu-Hausman p=0.0754 for DBP). These results will be discussed below.
According to the diagnostic criteria of hypertension, we considered the individuals with SBP >140 mm Hg or DBP >90 mm Hg as hypertension cases (964 samples), and the others as controls (1385 samples). Then, we performed the IV analysis to estimate the effect of alcohol use on hypertension, with rs671 as an instrumental variable. We found that alcohol use was shown to have a causal effect on hypertension in men (p=0.011; OR =1.19, 95% CI 1.04 to 1.36) and in pooled samples (p=0.013, OR =1.20, 95% CI 1.04 to 1.39), but not in women (p=0.317). Endogeneity was also assessed (Wu-Hausman, p=0.024 in pooled samples, p=0.027 in men and p=0.31 in women).
We did not observe a causal effect of alcohol use on BMI and lipid levels (table 4), suggesting that the associations from the observational regression analyses were non-causal.
In this study, we evaluated the association between rs671 and alcohol use in a Chinese population. We found an association of ALDH2 rs671 with alcohol drinking (p=6.08×10-47, ORadj =4.50) and this association was independent of sex. MR analysis suggested that alcohol use was shown to have a causal effect on blood pressure (9.46 mm Hg increase in SBP and 7.50 mm Hg increase in DBP) and hypertension.
Although alcohol drinking could be largely influenced by non-genetic factors, such as occupation, religion and familial environment, there is also an important genetic component in the population variance of alcohol consumption, with a heritability of ~50%.10 Among the susceptibility genes, the coding SNP rs671 on ALDH2 gene was one of the important culprits; the mutated type (A allele) encoded a form of the aldehyde dehydrogenase 2 protein that was defective at metabolising alcohol. The allele frequency of rs671 is very different among different populations. From the 1000 Genomes data, we could see that Europeans had the highest G frequency, with all people having the GG genotype, whereas East Asians had a G frequency of 82.6% (bold data in table 2). In China the G frequency differed between the south and north of the country; the frequency in Han Chinese in Beijing was higher than in southern Han Chinese (84.0% vs 72.9%) (bold data in table 2). Because north China is colder than the south, people in the north traditionally drink alcohol in order to keep warm. The study population was recruited from Ningbo, a city in the southeast of China, so we could tell that the G frequency was close to that in southern Han Chinese (72.1% vs 72.9%) (bold data in table 2). We also found that the female G frequency in non-drinkers was higher than the male G frequency (71.1% vs 55.2%) (bold data in table 2), and female G carriers had a higher proportion of non-drinkers than male G carriers (1035/1176=88.01% vs 382/987=38.70%) (highlighted data in table 2, figure 1). This is due to the fact that the Chinese culture does not encourage women to drink and smoke, an attitude inherited from an earlier time in China when women had a lower social profile than men.
We did not observe the causal effect of alcohol use on blood pressure and hypertension in women. This is because, as our results showed, female rs671_G carriers had a higher proportion of non-drinkers than male G carriers (1035/1176=88.01% vs 382/987=38.70%), meaning that there were lots of potential female drinkers who did not drink at the baseline survey. This could also be explained by the Chinese culture of discouraging women from drinking.
SNP rs671 has not been reported as being associated with SBP and/or DBP by GWAS. But another SNP rs11066280 in ALDH2 has been reported in Chinese samples at a genome-wide significance (p=9.8×10−8 for SBP, p=3.19× 10−10 for DBP).11 Kamatani et al 12 found that SNP rs671 was associated with gamma glutamyl transferase (GGT) (p=4.50×10-9) and alanine amino transferase (ALT) (p=5.99×10-9) in a Japanese population, but the association became weak after excluding subjects with a regular alcohol drinking habit (p=0.026). The association of ALDH2 with GGT and ALT levels had already been reported13 and was explained by the alcohol intolerance of the subjects with the A allele at rs671. SNP rs671 was also previously identified by a Japanese GWAS to be associated with coronary artery disease (CAD) (OR =1.43, p=1.6 ×10-34) and was more prominent with myocardial infarction (MI) than non-MI CAD (OR =1.53, p=6.9×10-40 for MI; OR =1.19, p=9.0×10-5 for non-MI CAD).14 This SNP also showed a significant association with stroke risk in Korean men, but not in Korean women.15 In addition, the polymorphism was identified as being associated with obesity (BMI)16 and cancers.17–21
MR studies provide a means of assessing causal relations without interventions, but require valid genetic instruments. One of the strengths of our study was that ALDH2 rs671 was a credible and very strong genetic instrument in southern Chinese for MR studies assessing the effects of alcohol on blood pressure. However, a potential limitation of our study was that we included single SNP as the instrumental variable in our MR approach.
In summary, we found evidence for the association of rs671 in the ALDH2 gene with a strong susceptibility to alcohol drinking, and the genotype distribution of this SNP reflected the Chinese culture that women are not encouraged to drink. In further MR analysis, alcohol use was shown to have a causal effect on blood pressure, increasing DBP by 7.5 (95% CI 3.7 to 11.3) mm Hg and SBP by 9.5 (95% CI 3.8 to 15.1) mm Hg, and hypertension.
What is already known on this subject
There has been only one genome-wide association study of an isolated rural Chinese population sample which demonstrated that ALDH2 (rs671) was associated with alcohol dependence, flushing response and daily maximum drinks.
Low to moderate alcohol use among men had the expected effects on most cardiovascular disease based on Mendelian randomisation analysis.
What this study adds
The genotype distribution of SNP rs671 differed between men and women in China. Female G carriers accounted for a higher proportion of non-drinkers than male G carriers (88.01% vs 38.70%). In non-drinkers, the female G allele frequency was higher than the male G allele frequency (71.1% vs 55.2%).
These results are due to the differing drinking habits of men and women in China, where the culture does not encourage women to drink and smoke.
Compared with never-drinkers, current and previous alcohol use had a causal effect on blood pressure and hypertension in pooled samples and in men.
P-PZ, L-WX and TS are joint first authors.
LW, X-WZ, LY and H-FZ are joint senior authors.
Contributors Conceived and designed the study: H-FZ. Contributed materials: LY, L-WX, H-YM, Y-YW, LW, X-WZhu, XC, T-TW, T-TZ, Y-CL. Performed the experiments: P-PZ, X-WZhang, TS, ZC, BZ. Wrote the manuscript: H-FZ. All authors reviewed and approved the manuscript before submission.
Funding This work was supported by the Zhejiang Provincial Natural Science Foundation of China (LR17H070001 and LZ13H02001), and the National Natural Science Foundation of China (81871831). The funding agencies had no role in the study design, data collection and analysis, the decision to publish, or preparation of the manuscript. We thank the peer reviewers for their thorough and helpful review of this manuscript.
Competing interests None declared.
Patient consent for publication Parental/guardian consent obtained.
Provenance and peer review Not commissioned; externally peer reviewed.
Data sharing statement Data are available upon reasonable request.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.