Alcohol consumption is associated with glaucoma severity regardless of ALDH2 polymorphism

The present study investigated the effect of aldehyde dehydrogenase2 (ALDH2) rs671 polymorphism and alcohol consumption on the severity of primary open-angle glaucoma (POAG). The questionnaire for alcohol consumption pattern and targeted genotyping for ALDH2 rs671 polymorphism was performed from 445 Korean POAG patients. Retinal nerve fiber layer (RNFL) and ganglion cell-inner plexiform layer (GCIPL) thicknesses were measured and compared according to alcohol consumption and ALDH2 rs671 genotype. Heavy drinking group eyes had thinner RNFL thickness than did abstinence group eyes (65.0 ± 10.9 vs. 70.9 ± 11.5 µm, P = 0.023). Both mild (65.8 ± 9.6 µm) and heavy (63.8 ± 8.4 µm) drinking group eyes had significantly thinner macular GCIPL thickness than did abstinence group eyes (68.1 ± 8.2 µm, P = 0.003). However, ALDH2 rs671 polymorphism did not show any significant association with RNFL or GCIPL thickness. Alcohol consumption was significantly associated with GCIPL thinning (β = –0.446, P = 0.035) after adjustment for multiple confounding factors. As excessive alcohol consumption was significantly associated with thinner GCIPL thickness while ALDH2 polymorphism had no significant effect on RNFL or GCIPL thickness, glaucoma patients should avoid excessive alcohol consumption regardless of ALDH2 polymorphism.


Results
Subject demographics. Of the 445 POAG patients, 265 reported that they did not drink (the abstinence group) and another 180 that they drank (the drinking group). Of these 180 patients for whom alcohol consumption information was available, 147 were classified as mild drinkers and 33 as heavy drinkers. According to the binge drinking criteria, 141 patients were non-binge drinkers and 39 were binge drinkers. Patients in the drinking group were significantly younger (52.5 ± 13.6 years old) and comprised of more males (62.8%) than was the abstinence group (55.1 ± 12.8 years old and 40.4%, respectively, all P < 0.05). Although there was no significant difference in the prevalence of diabetes or hypertension between the two groups, the patients from the drinking group had significantly more smoking pack-years (2.8 ± 7.3 pack-years) than did the abstinence group (0.7 ± 3.3 pack-years, P < 0.001). The eyes from the drinking group were more myopic (-3.1 ± 3.8 vs. -2.1 ± 3.2 D, P = 0.003) and had greater AXL (25.1 ± 1.6 vs. 24.3 ± 1.5 mm, P < 0.001) than those from the abstinence group. The patients in the drinking group showed a greater proportion of the major allele (GG) for ALDH2 rs671 polymorphism. Detailed information on the subject demographics is provided in Table 1.
The study population was subdivided into four groups according to drinking status and rs671 polymorphism. In terms of peripapillary RNFL thickness, there was no significant difference among the subgroups (Fig. 2). In terms of macular GCIPL thickness, eyes with the major allele for rs671 showed significantly thinner macular GCIPL thickness in the drinking group (65.5 ± 9.7 µm) than in the abstinence group (68.9 ± 8.1 µm, P = 0.009, Fig. 2). The eyes with the minor allele for rs671 in both the abstinence (66.6 ± 8.0 µm) and drinking In the subgroup analysis, only females reached marginal significance (r = − 0.126, P = 0.06); males did not (r = − 0.088, P = 0.19). Likewise, the alcohol consumption score demonstrated a significant negative correlation with macular ganglion cell-inner plexiform layer (GCIPL) thickness (r = − 0.139, P = 0.003). In the subgroup analysis, only females (r = − 0.150, P = 0.024), not males (r = − 0.104, P = 0.12), showed a significant negative correlation. Table 3. Comparison of structural and functional parameters according to ALDH2 rs671 polymorphism. Mean ± standard deviation, Comparison was performed using one-way analysis of variance with post hoc Scheffe's multiple comparison testing. RNFLT, retinal nerve fiber layer thickness; GCIPL, ganglion cell-inner plexiform layer; MD, mean deviation; PSD, pattern standard deviation; VFI, visual field index.  www.nature.com/scientificreports/ (65.1 ± 8.3 µm) groups tended to have thinner macular GCIPL thickness than did major allele eyes in the abstinence group (68.9 ± 8.1 µm), but the difference did not represent statistical significance.

Discussion
The present study found a negative correlation between alcohol consumption and macular GCIPL thickness, which association was more remarkable in females. Greater alcohol consumption score was significantly associated with thinner macular GCIPL thickness after controlling for confounding factors. However, the present study did not find any significant association of ALDH2 rs671 polymorphism with glaucoma severity marked by peripapillary RNFL or macular GCIPL thickness.
Alcohol is primarily metabolized in the liver but is also metabolized in extrahepatic tissues such as the brain 21 . Alcohol is oxidized to acetaldehyde, a highly toxic byproduct that can cause tissue damage, by three enzymes: and macular ganglion cell-inner plexiform layer (GCIPL) thicknesses were compared according to rs671 polymorphism and alcohol consumption. No significant differences were observed in peripapillary RNFL thickness. In terms of macular GCIPL thickness, eyes with the major allele for rs671 showed significantly thinner macular GCIPL thickness in the drinking group (65.5 ± 9.7 µm) than in the abstinence group (68.9 ± 8.1 µm, P = 0.009). The eyes with the minor allele for rs671 in both the abstinence (66.6 ± 8.0 µm) and drinking (65.1 ± 8.3 µm) groups tended to have thinner macular GCIPL thickness than did the major allele eyes in the abstinence group (68.9 ± 8.1 µm), but the difference was not statistically significant. www.nature.com/scientificreports/ alcohol dehydrogenase (ADH), catalase, and cytochrome P450 isoenzymes 22 . During this process, reactive oxygen species (ROS) are also generated, making the neural tissues vulnerable to oxidative stress. Acetaldehyde is then further metabolized to acetate by mitochondrial aldehyde dehydrogenase2 (ALDH2). Since acetaldehyde is capable of binding to proteins (i.e. enzymes, microsomal proteins, and microtubules) and forming adducts with DNA, which ultimately results in impaired target-organ function 21,22 , ALDH2′s acetaldehyde-removal activity has attracted clinical interest. The genetic diversity of ALDH2, on chromosome 12, is well established elsewhere 2,17,23-25 . The G-to-A missense mutation in exon 12, where glutamate at position 504 is replaced by lysine (Glu504Lys), is one of the representative SNPs (rs671) in ALDH2. Interestingly, the minor alleles of rs671 are very common in East Asians (being present in up to 50% of a given population), but less common in Caucasians 26,27 . The present study revealed an MAF for rs671 of 15.6%, which is quite consistent with previous findings from Korean populations 28,29 . The heterozygote and minor homozygotes for this SNP lead to decreased activity of ALDH2, affecting 17-35% and 1-3% of ALDH2 activities, respectively 30 . Increased acetaldehyde concentration after alcohol consumption has been further confirmed in subjects with rs671 minor alleles 31 .
It has been well documented that subjects with the rs671 minor allele are not likely to have alcohol dependence and more likely to be reluctant to drink, due to the acute effect of facial flushing or palpitation after alcohol  www.nature.com/scientificreports/ consumption, which fact was also confirmed in the present study 32 . Contrary to our expectation, however, rs671 polymorphism did not affect the peripapillary RNFL or macular GCIPL thicknesses in POAG eyes. The present study demonstrated the tendency to thinner macular GCIPL thickness in eyes with the minor allele for rs671 than in eyes with the major alleles, but the difference was not statistically significant. The post hoc power was 0.999 for RNFL thickness (effect size f = 0.281, α error probability = 0.05, total sample size = 445) and 1.000 for GCIPL thickness (effect size f = 0.512, α error probability = 0.05, total sample size = 445). We could obtain sufficient power for the association analysis between rs671 polymorphism and POAG severity. The present study found that greater alcohol consumption was significantly associated with thinner macular GCIPL thickness after compensation for several confounding factors including baseline IOP, AXL, and optic nerve head parameters. This is consistent with a recent large UK cohort study reporting that frequent alcohol consumption is associated with thinner macular RNFL, GCC and GCIPL 16 . Speculative explanations for the present finding are as follows. First, alcohol metabolism itself can aggravate oxidative stress in the central nervous system (CNS) as the result of increased production of ROS by catalase and cytochrome P450 isoenzymes 21,22 . Alcohol is mainly metabolized in the liver, but catalase has emerged as the main alcohol-metabolizing enzyme in the brain 33 . Cytochrome P450 2E1 (CYP2E1) also converts alcohol to acetaldehyde and generates ROS that cause neuronal injuries 34 . As the retina is one of the most energy-consuming organs for maintenance of vision, the retinal ganglion cells might be vulnerable to alcohol-induced oxidative stress. Second, chronic alcohol consumption can lead to thiamine deficiency due to inadequate nutritional intake, reduced uptake of thiamine from gastrointestinal tracts, or impaired utilization of thiamine in the cells 35 . Thiamine deficiency can cause impaired cerebral glucose metabolism, and exacerbate CNS dysfunction, a typical example being Wernicke encephalopathy. In line with this idea, previous reports have shown that thiamine levels in glaucoma patients are low 36 and that thiamine protects against glaucoma 37 . The thiamine concentration associated with alcohol consumption in POAG patients needs to be further verified in future studies. Third, alcohol consumption can modify the gastrointestinal microbiome through the acute and chronic effects of alcohol on gastric motility dysfunction, changes in gastric acid output, and direct mucosal damage 38,39 . A recent study from Chen et al. 40 raised the issue of the role of commensal microflora-induced T cell responses in glaucomatous progressive neurodegeneration. They postulated that pre-sensitized T cells that had been exposed to commensal microflora mediated glaucomatous neurodegeneration in response to IOP increase in mouse eyes. Recent evidence showing a significant association between Helicobacter pylori infection and open-angle glaucoma can further support this hypothesis 41,42 .
In the present study, the effect of alcohol consumption on macular GCIPL thickness was more remarkable in females than in males. This finding is consistent with those for African-American women 14 . Generally, females are likely to be more vulnerable to alcohol than males, as they have a lower proportion of body water than males, resulting in higher blood concentrations of alcohol after drinking 43 . ADH activity is also known to be lower in women than in men 44 . Taken together, the present findings indicate that excessive alcohol consumption can be more harmful in women than in men, in terms of its effects on glaucoma susceptibility.
The present study has several limitations. First, as it was based on a questionnaire for alcohol consumption, it was difficult to acquire the exact alcohol consumption of each participant. Thus, this study design might be limited in its suitability for investigation of the precise dose-dependent effects of alcohol on glaucoma severity. However, the relatively large number of subjects in the present study along with the alcohol consumption scoring system employed was helpful in revealing the negative relationship between alcohol consumption and macular GCIPL thickness. Second, as the present study was cross-sectional, it was difficult to evaluate the risk of excessive alcohol consumption with respect to the incidence or progression of glaucoma. Additional case-control or longitudinal cohort studies on alcohol consumption and glaucoma risk are essential to gain more practical information. Third, the present study recruited only POAG patients but not non-glaucomatous subjects. This may have caused the selection bias on the correlation analysis as the POAG eyes have thin RNFL and GCIPL that belong to the pathogenic end of the spectrum. However, the present findings still have clinical significance in warning POAG patients to avoid excessive alcohol consumption regardless of ALDH2 rs671 polymorphism, as it was significantly associated with thinner macular GCIPL thickness. Lastly, the present study investigated only rs671 polymorphism but not the whole tag SNPs in ALDH2. Full sequencing of the ALDH2 gene may reveal additional significant association results that deserve further exploration in the near future.
In conclusion, excessive alcohol consumption was negatively correlated with macular GCIPL thickness in POAG patients. ALDH2 rs671 polymorphism, which is prevalent in East Asians, can alter ALDH2 activity during alcohol metabolism, but did not show any significant association with peripapillary RNFL or macular GCIPL thickness. However, it should be noted that subjects with the rs671 minor allele tend to drink less alcohol, which fact might be related to the present insignificant findings with regard to glaucoma severity. Our data suggest that glaucoma patients should avoid excessive alcohol consumption regardless of ALDH2 polymorphism.

Methods
The present study was initiated as a part of the GLAU-GENDISK (GLAUcoma GENeDIscovery Study in Korea) project, which is an ongoing prospective study designed in 2011 45 . The primary objective of the GLAU-GENDISK project is to investigate and identify novel genetic susceptibility loci for various types of glaucoma in a Korean population. The secondary objectives included establishing the genotype-phenotype relationships in glaucoma patients and constructing new disease prediction models. The present study was approved by the Seoul National University Hospital Institutional Review Board and followed the tenets of the Declaration of Helsinki (1964). The subjects who met the eligibility criteria provided written informed consent to participate.
Study population. The detailed information for the GLAU-GENDISK study population is reported elsewhere 45  www.nature.com/scientificreports/ in Seoul National University Hospital. POAG was defined as the presence of glaucomatous optic disc changes with corresponding glaucomatous visual field (VF) defects and an open-angle confirmed by gonioscopy. Glaucomatous optic disc changes were defined as typical neuroretinal rim thinning, notching, excavation, or RNFL defects. Glaucomatous VF defects were defined as (1) glaucoma hemifield test values outside the normal limits, (2) three or more abnormal points with a probability of being normal of P < 5%, of which at least one point has a pattern deviation of P < 1%, or (3) a pattern standard deviation of P < 5%. The VF defects were confirmed on two consecutive reliable tests (fixation loss rate ≤ 20%, false-positive and false-negative error rates ≤ 25%). The subjects underwent a complete ophthalmic examination, including a visual acuity assessment, slit-lamp biomicroscopy, intraocular pressure (IOP) measurement by Goldmann applanation tonometry, gonioscopy, refractions, dilated fundus examination, disc stereophotography, and red-free fundus photography using a digital fundus camera (VX-10; Kowa, Nagoya, Japan) and standard automated perimetry (Humphrey Survey on alcohol consumption. Alcohol consumption was investigated by questionnaire. The subjects were asked about the amount and frequency of drinking during the week. The amount of drinking was based on the traditional Korean liquor, Soju. One bottle of Soju (18% alcohol, 360 mL) contains 50.86 g of alcohol. The Centers for Disease Control and Prevention (CDC) and National Institute on Alcohol Abuse and Alcoholism (NIAAA) 46,47 define excessive alcohol consumption (per occasion or per week) as follows: binge drinking (≥ 4 standard drinks per occasion for a woman, and ≥ 5 standard drinks per occasion for a man) or heavy drinking (≥ 8 standard drinks per week for a woman, and ≥ 15 standard drinks per week for a man), where a standard drink is defined as containing 14.0 g of pure alcohol. Based on these criteria, subjects were categorized into an abstinence group (those who reported never drinking), a non-heavy-drinking group (those who drink < 8 standard drinks per week for a woman, and < 15 standard drinks per week for a man), and a heavy drinking group (those who drink ≥ 8 standard drinks per week for a woman, and ≥ 15 standard drinks per week for a man). In terms of binge drinking, subjects who reported drinking were categorized into a non-binge-drinking group (< 4 standard drinks per occasion for a woman, and < 5 standard drinks per occasion for a man) or a binge-drinking group (≥ 4 standard drinks per occasion for a woman, and ≥ 5 standard drinks per occasion for a man). Additionally, the alcohol consumption score was recorded as the cumulative amount of drinking, as calculated by multiplying the amount of drinking (the number of standard drinks) by the frequency of alcohol consumption per week.
SNP genotyping. The detailed information for the SNP genotyping is reported elsewhere 45 . Blood samples were collected from all of the subjects, and the genomic DNA was extracted. Among the 551 study samples, 309 were analyzed by exome chip (Illumina, Inc.; San Diego, CA, USA) analysis and 242 were analyzed by TaqMan assay (Applied Biosystems, Carlsbad, CA, USA). The exome-chip analysis utilized the Human Exome Bead-Chip 12v1-1 system (Illumina, Inc.), which included 244,651 markers focused on protein-altering variants. Details regarding SNP content and selection strategies can be found at the exome array design webpage (https ://genom e.sph.umich .edu/wiki/Exome _Chip_Desig n). Genotype calling was performed using Illumina's GenTrain version 2.0 clustering algorithm with GenomeStudio software (V2011.1). Cluster boundaries were determined using Illumina's standard cluster file. After additional visual inspection of SNPs with call rates of less than 0.99 and of those with minor allele frequencies (MAFs) less than 0.002, 244, and 552 of 244,651 (99.96%) attempted markers were successfully genotyped with call rates greater than 98% (average call rate: 99.92%).
The TaqMan assay (Applied Biosystems, Carlsbad, CA, USA) for the SNP rs671 was conducted by the following steps: 1) preparation of approximately 20 ng of purified genomic DNA; 2) preparation of a genotyping mixture consisting of 2X genotyping master mix, 20X SNP genotyping assay, DNAse-free water and template DNA, and 3) a polymerase chain reaction (PCR) entailing 40 cycles of denaturation and annealing/extension steps. When the PCR was completed, the genotypes of the DNA samples were analyzed on the ABI prism 7900HT sequence detection system (Applied Biosystems, Foster City, CA, USA).

Statistical analysis.
Continuous variables were compared among the three groups by one-way analysis of variance (ANOVA) with Scheffe's post hoc analysis. Categorical variables were compared using a chi-square test. Pearson's correlation test was used to identify the correlation between alcohol consumption score and peripapillary RNFL and macular GCIPL thicknesses, respectively. The generalized linear model (GLM) was used to determine the factors (i.e. age, sex, diabetes mellitus, hypertension, baseline IOP, AXL, spherical equivalence [SE], CCT, vertical cup-to-disc ratio (C/D), disc rim area, disc area, smoking history (pack-year), alcohol con- www.nature.com/scientificreports/ sumption score, and rs671 polymorphism) associated with average RNFL or macular GCIPL thickness, first with a univariate model, and then with a multivariable model that included the univariate model variables with P < 0.10. All of the statistical analyses were performed with R version 3.6.1. (available at: https ://www.r-proje ct.org; accessed April 2019). Except where stated otherwise, the data are presented as mean ± standard deviations, and the level of statistical significance was set at P < 0.05.

Data availability
The datasets generated and/or analyzed in the current study are available from the corresponding author on reasonable request. The sequencing dataset is held at Seoul National University Hospital (SNUH), access to it requiring approval of the Institutional Review Board of SNUH.