The effect of the association between CETP variant type and alcohol consumption on cholesterol level differs according to the ALDH2 variant type

Alcohol consumption is associated with a high increased lipid profile and this association may depend on genetic risk factors. In this study, we aimed to assess the effects of genetic variation associated with alcohol consumption on lipid profiles using data from two Korean population studies. We performed a genotype association study using the HEXA (n = 51,349) and KNHANES (n = 9158) data. Genotype analyses of the two sets of Korean population data showed associations of increased total cholesterol and high-density lipoprotein (HDL)-cholesterol with CETP rs708272. The HEXA and KNHANES populations revealed differences in HDL cholesterol according to the presence of CETP rs708272, independent of ALDH2 rs671 and alcohol consumption. In contrast, total cholesterol levels were associated with alcohol consumption and ALDH2 rs671 in men with CETP rs708272 (CT and TT genotypes). Furthermore, in drinkers with ALDH2 rs671 (GA and AA genotypes), higher total cholesterol was associated with the CETP rs708272 TT minor homozygous genotype based on both HEXA and KNHANES data. Our findings demonstrated that alcohol consumption and genetic variation in either CETP or ALDH2 may be associated with cholesterol levels. We hope these findings will provide a better understanding of the relationship between alcohol consumption and cholesterol according to each individual’s genetic background.


Results
Data were obtained from the Health Examinees Study (HEXA) and Korean National Health and Nutrition Examination Survey (KNHANES). The characteristics of the study participants, according to sex, are presented in Table 1. Age, systolic/diastolic blood pressure, fasting glucose level, total cholesterol level, HDL-C level, TG level, aspartate transaminase (AST) level, alanine transaminase (ALT) level, and alcohol consumption rate differed by sex in both the HEXA and KNHANES study populations (p < 0.05). The frequency of ALDH2 rs671 differed significantly by sex in the HEXA population, but not in the KNHANES population. The general characteristics of age, systolic/diastolic blood pressure, HDL-C level, TG level, and AST and ALT levels differed according to alcohol consumption in both study populations (Online Supplementary Table S1).
The participants' clinical characteristics are summarized according to CETP rs708272 [CC, CT, or TT] status in Table 2. The total cholesterol and HDL-C levels of CETP rs708272 risk T allele carriers were significantly higher than others in the HEXA and KNHANES populations (p < 0.05). In addition, the risk allele T carriers had a higher LDL-C levels in the KNHANES study (p < 0.05). When categorized by sex, LDL-C levels were significantly different between men in both the HEXA and KNHANES studies. (Online Supplementary Tables S2  and S3). Men carrying CETP rs708272 TT allele had higher LDL-C levels than those carrying the CT of the TT genotype. On the other hand, almost all general characteristics according to the ALDH2 rs671 genotype, except for sex and total cholesterol, showed differences in both study populations (Online Supplementary Table S4). Table 3 shows the total cholesterol and HDL-C levels according to alcohol consumption in carriers of CETP rs708272. HDL-C levels differed significantly between CETP rs708272 by alcohol consumption and sex in the HEXA and KNHANES study populations. In the HEXA population, CETP rs708272 was associated with a higher HDL-C level in drinkers than in non-drinkers of both sexes. However, HDL-C level did not differ with the association between the CETP rs708272 TT minor homozygous genotype and alcohol consumption in both sexes in the KNHANES population. In the HEXA population, total cholesterol varied significantly with alcohol consumption in men with CETP rs708272, whereas in women, there was no association between drinking and CETP rs708272. In addition, in the KNHANES population, total cholesterol was not associated with CETP rs708272 among non-drinkers and drinkers. In the HEXA population, total cholesterol differed significantly among men with CETP rs708272 depending on alcohol consumption. In contrast, the KNHANES population, there was no association between drinking status and CETP rs708272. In addition, CETP rs708272 was associated with higher total cholesterol levels in male drinkers than in non-drinkers in the HEXA population, whereas no association was found in the KNHANES population. In women, total cholesterol differed significantly among CETP rs708272 in non-drinkers in both the HEXA and KNHANES populations. Total cholesterol was inversely associated with alcohol consumption according to CETP rs708272 in the HEXA population. However, the only risk TT minor homozygous genotype of CETP rs708272 was inversely associated with higher total cholesterol in drinkers in the KNHANES population. Conversely, the association of cholesterol with alcohol consumption in the CETP rs708272 major homozygous genotype group disappeared.
In the case of ALDH2 rs671, HDL-C showed a significant difference according to the ALDH2 rs671 genotype in non-drinkers and drinkers in the HEXA population. Nevertheless, only drinkers showed differences in the KNHANES population. Moreover, ALDH2 rs671 [GA + AA] genotype was associated with lower HDL-C levels in drinkers than those in non-drinkers in the HEXA population. Total cholesterol in the HEXA and KNHANES populations differed according to the ADLH2 rs671 genotype in drinkers (Online Supplementary Table S5).
The effects of CETP rs708272 and alcohol consumption on HDL-C according to ALDH2 rs671 are shown in Table 4 and Online Supplementary Fig. S1. HDL-C differed significantly according to CETP rs708272 independent of ALDH2 rs671 and alcohol consumption, but CETP rs708272 and HDL-C were slightly associated with the ALDH2 rs671 [GA + AA] genotypes. In addition, those with the ALDH2 rs671 [GA + AA] genotype exhibited trends toward lower HDL-C compared with those with the ALDH2 rs671 GG major homozygous genotype. Still, overall, the HDL-C level was higher in men drinkers than in non-drinkers, regardless of the ALDH2 rs671 genotype. www.nature.com/scientificreports/ Table 5 shows the association of total cholesterol with alcohol consumption and ALDH2 rs671 according to CETP rs708272. In men, total cholesterol was associated with alcohol consumption and ADLH2 rs671 when CETP rs708272 CT heterozygous and TT minor homozygous genotypes were present in both HEXA and KNHANES populations. Additionally, drinkers with ALDH2 rs671 exhibited the highest total cholesterol level with the CETP rs708272 TT risk minor homozygous genotype in both the HEXA and KNHANES populations. On the other hand, as a result of analyzing the association between the CETP rs708272 genotype and alcohol consumption according to ALDH2 variation in women, the total cholesterol level was high in the presence of the CETP rs708272 risk T allele, but there was no significant difference, unlike in men. In particular, male drinkers with the CETP rs708272 TT risk minor homozygous genotype had the highest total cholesterol level when they had the ALDH2 rs671 [GA + AA] genotype. Table 1. General characteristics of Korean population. All data except alcohol consumption, CETP and ALDH2 genotype are represented as means ± standard deviation. Student's t-test and chi-square tests were used to determine differences between men and women. HDL high-density lipoprotein, AST aspartate aminotransferase, ALT alanine aminotransferase.

Discussion
Several studies have reported associations between CETP rs708272, alcohol consumption, and cholesterol levels.
In addition, ALDH2 rs671, which is closely related to alcohol consumption in East Asians, is well known to be associated with HDL-c according to alcohol consumption 17 , but studies involving CETP variants have not been reported. We investigated the association between alcohol consumption and cholesterol levels according to CETP and ALDH2 variants and their combinations in two large-scale Korean populations. Previous studies have shown that CETP rs708272 increases HDL-C levels. In a meta-analysis of 98 published studies 18 , CETP rs708272 was associated with higher mean HDL-C levels. In that study, HDL-C level was higher in the CETP rs708272 risk allele (TT and CT genotype) carriers than in the CC major homozygous genotype carriers. In 2018, Cai et. al. reported that the HDL level of participants with CETP rs708272 TT genotype was higher than those of CC genotype, and the HDL level of male participants with only T allele was significantly higher 19 . Also, some studies have shown a gene × alcohol consumption interaction for CETP. Based on the limited evidence, alcohol consumption appears to be associated with lower CETP activity 20 . In the association between alcohol consumption and HDL-C, changes in other metabolic pathways including an increased transport rate of apolipoproteins, reduced hepatic lipase activity, and so on may be likely [21][22][23][24][25] . However, our study found no interaction between CETP rs708272 and alcohol consumption in determining HDL-C. This is in line with previous studies carried out in a Mediterranean population 26 , healthy men 27 , and insulin-dependent men 28 . The metabolic mechanisms underlying an increase in HDL-C following the interaction between CETP and alcohol consumption are poorly understood, and more research is needed.
In this study, we observed that increased HDL-C was associated with alcohol consumption independent of ALDH2 rs671; this result is similar to that of a previous meta-analysis, which revealed an association between alcohol consumption and HDL-C. However, Wakabayashi et al. reported that the effect of alcohol consumption on HDL-C was different in drinkers with an alcohol flushing response 29 . The beneficial effects of an association between alcohol consumption and HDL-C according to ALDH2 rs671 remains controversial because several other studies have shown that HDL-C is influenced by alcohol consumption differently in drinkers with an Table 2. General characteristics of subjects according to CETP genotype in HEXA and KNHANES. All data except alcohol consumption and sex are represented as means ± standard deviation. General linear models and chi-square tests were used to determine differences between CETP genotype. HDL high-density lipoprotein, AST aspartate aminotransferase, ALT alanine aminotransferase.  www.nature.com/scientificreports/ alcohol flushing response. According to some data, the ALDH2 rs671 A allele carriers, GA, and AA genotypes have a 2.6-fold higher risk of hypo-HDL-cholesterolemia than the GG homozygous genotype 30 . These conflicting findings may be due to the small size of the study population. Further studies using larger populations are needed to clarify whether ALDH2 affects the relationship between alcohol consumption and HDL-C. The effects of alcohol consumption on total cholesterol may differ according to the CETP or ALDH2 status. Previous studies have shown that alcohol consumption in subjects with the ALDH2 rs671 risk A allele was associated with increased total cholesterol compared to alcohol consumption with the ALDH2 rs671 GG homozygous genotype 31 . Similar to previous studies, our study found that drinkers with the ALDH2 [GA + AA] genotype had higher total cholesterol levels than non-drinkers with the ALDH2 GG homozygous genotype. CETP variants also showed an association with total cholesterol increase 32 , but no significant difference was found in total cholesterol levels according to alcohol consumption in CETP rs708272 33 . The total cholesterol levels in our study were the highest in both men and women with CETP rs708272 risk TT genotype. Still, there was no difference according to ALDH2 rs671 genotype, while the presence of ALDH2 rs671 risk alleles in the drinking men group with CETP rs708272 risk TT genotype was associated with changes in total cholesterol levels.
This was an observational study of the association between alcohol consumption and cholesterol according to CETP and ALDH2 variants. Residual confounding is a critical issue that should be considered in observational studies. We replicated two Korean population cohorts (HEXA and KNHANES) to address residual confounding. In the present study, we revealed differences in HDL-cholesterol according to the presence of CETP rs708272, independent of ALDH2 rs671 and alcohol consumption, and found that the presence of CETP rs708272 risk allele in men was associated with an increase in HDL cholesterol following alcohol intake. Furthermore, total Table 4. Effects of CETP genotype and alcohol consumption on HDL-cholesterol according to ALDH2 genotype. Data are expressed as means ± standard deviations. 1 General linear models by CETP genotypes in each group, and differences between alcohol consumption and ALDH2 genotype. 2 General linear models were used to assess CETP genotype differences in variables between alcohol consumption and HDL-cholesterol by ALDH2 genotype. www.nature.com/scientificreports/ cholesterol levels were associated with alcohol consumption and ALDH2 rs671 in men with CETP rs708272 CT and TT genotypes with strong genetic effects. This study provides information about genetic variants of CETP and ALDH2 in the Korean population. It clarifies the probable mechanism underlying the differential association between alcohol consumption and cholesterol (total cholesterol and HDL-C) according to CETP and ALDH2 variants. Overall, our work would be helpful in clarifying the clinical implications of improving the translational value for preventive management of public health policies, and further realizing personalized or precision medicines that can prevent individual diseases or realize early detection, treatment decisions, and prognosis. It can also be used as a predictor of population assessment and vulnerable groups, while providing evidence for health policy development and health insurance applications.

Methods
Study population. Data were obtained from two studies conducted by the Korea Disease Control and Prevention Agency (KDCA) as part of the Korean Genome and Epidemiology Study (KoGES) and the KNHANES. Written informed consent was obtained from all participants, and the research project was approved by KDCA. The study protocol was approved by the Institutional Review Board of the Korean National Institute of Health (2019-03-01-PE-A). HEXA and KNHANES have been previously described in detail 34 . All study protocols were carried out by the approved guidelines.
The HEXA cohort subjects (n = 53,754) were invited to health examination centers in eight regions (metropolitan areas or major cities) and enrolled in 38 health examination centers and hospitals in Korea between Table 5. Effects of CETP genotype and alcohol consumption on total cholesterol according to ALDH2 genotype. Data are expressed as means ± standard deviations. 1 General linear models by CETP genotypes in each group, and differences between alcohol consumption and ALDH2 genotype. 2 General linear models were used to assess CETP genotype differences in variables between alcohol consumption and total cholesterol by ALDH2 genotype. www.nature.com/scientificreports/ 2004 and 2013. Individuals with missing data for the CETP rs708272 genotype (n = 149) or alcohol consumption (n = 2256) were excluded. A total of 51,349 participants were included in this cross-sectional study. KHNANES is a nationally representative survey of the health and nutritional status of the Korean population. We used data from a population of 15,000 people aged 20-60 years collected during the 2009-2015 period. Individuals with missing data for alcohol consumption (n = 401), cholesterol (HDL and total cholesterol, n = 153), and those under 40 years of age (n = 5288) were excluded. In total, 9158 participants were included in this crosssectional study.
Anthropometric and biochemical analyses. In HEXA, age and smoking status information was obtained using interview-based questionnaires. Body mass index (BMI) was calculated as weight divided by height squared. The lipid profile (total cholesterol, TGs, and HDL-C) and fasting glucose, ALT, and AST levels were measured using a Hitachi 747 chemistry analyzer (Hitachi, Ltd., Tokyo, Japan). Systolic/diastolic blood pressure was measured twice using a standard mercury sphygmomanometer, and the results were averaged.
In the KNHANES, data were collected by various means, including interview-based and self-reported questionnaires, physical examinations, and assessments of nutritional status. The calculation of BMI and blood pressure (systolic and diastolic) was the same as that for HEXA. The lipid profile (total cholesterol, TGs, and HDL-C) and plasma glucose, ALT, and AST levels were measured using a Hitachi Automatic Analyzer 7600 (Hitachi, Tokyo, Japan).
Alcohol consumption data were collected using an interview-based questionnaire in HEXA and KNHNES. Subjects were asked whether they had ever consumed at least one alcoholic drink every month, and if they had, whether they were former-drinker or current drinkers 35 . Genotyping. HEXA cohort subjects were genotyped using a Korean chip designed by the Center for Genome Science at the Korean National Institute of Health. The chip was provided and its use was approved by the National Biobank of Korea, the Centers for Disease Control and Prevention, Republic of Korea (No. 2019-022). The genotyping protocol and quality control process has been described in detail previously 36 .
We preliminarily analyzed the association between cholesterol and genes using 128 SNPs related to alcohol metabolism which were selected from the KNHANES by TaqMan genotyping assay as previously described 37 .
Only CETP rs708272 and ALDH2 rs671 were related to cholesterol in two sets of Korean population data, which led to the use of this study.
Statistical analysis. Statistical analyses were performed using SAS software package (ver. 9.4; SAS Institute Inc., Cary, NC, USA). Data are presented as the mean ± standard deviation or number (%). Logarithmic transformation was applied to variables with a non-Gaussian distribution. Student's t-tests were conducted to compare clinical characteristics according to sex, and chi-square tests were used for categorical variables (ALDH2, CETP, and alcohol consumption). Analysis of variance was conducted to compare alcohol consumption according to CETP and ALDH2 genotypes. Multivariable linear regression models were used to assess differences in variables between CETP and ALDH2 genotypes and the combined effects of CETP and ALDH2 on the association between alcohol consumption and cholesterol, adjusted for age, BMI, and smoking. Statistical tests were two sided, and values of p < 0.05 were considered indicative of statistical significance.