Skip to main page content
Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
Meta-Analysis
. 2018 Feb;23(2):422-433.
doi: 10.1038/mp.2016.192. Epub 2016 Nov 15.

A DNA Methylation Biomarker of Alcohol Consumption

Affiliations
Free PMC article
Meta-Analysis

A DNA Methylation Biomarker of Alcohol Consumption

C Liu et al. Mol Psychiatry. .
Free PMC article

Abstract

The lack of reliable measures of alcohol intake is a major obstacle to the diagnosis and treatment of alcohol-related diseases. Epigenetic modifications such as DNA methylation may provide novel biomarkers of alcohol use. To examine this possibility, we performed an epigenome-wide association study of methylation of cytosine-phosphate-guanine dinucleotide (CpG) sites in relation to alcohol intake in 13 population-based cohorts (ntotal=13 317; 54% women; mean age across cohorts 42-76 years) using whole blood (9643 European and 2423 African ancestries) or monocyte-derived DNA (588 European, 263 African and 400 Hispanic ancestry) samples. We performed meta-analysis and variable selection in whole-blood samples of people of European ancestry (n=6926) and identified 144 CpGs that provided substantial discrimination (area under the curve=0.90-0.99) for current heavy alcohol intake (⩾42 g per day in men and ⩾28 g per day in women) in four replication cohorts. The ancestry-stratified meta-analysis in whole blood identified 328 (9643 European ancestry samples) and 165 (2423 African ancestry samples) alcohol-related CpGs at Bonferroni-adjusted P<1 × 10-7. Analysis of the monocyte-derived DNA (n=1251) identified 62 alcohol-related CpGs at P<1 × 10-7. In whole-blood samples of people of European ancestry, we detected differential methylation in two neurotransmitter receptor genes, the γ-Aminobutyric acid-A receptor delta and γ-aminobutyric acid B receptor subunit 1; their differential methylation was associated with expression levels of a number of genes involved in immune function. In conclusion, we have identified a robust alcohol-related DNA methylation signature and shown the potential utility of DNA methylation as a clinically useful diagnostic test to detect current heavy alcohol consumption.

Conflict of interest statement

Erik Ingelsson is an advisor and consultant for Precision Wellness Inc (Redwood City, CA). The remaining authors declare no conflict of interest.

Figures

Figure 1
Figure 1
Overview of the study design. ARIC, The Atherosclerosis Risk in Communities study; BMI, body mass index; DNAm, DNA methylation value; FHS, the Framingham Heart Study; I (light/at-risk/heavy drinkers versus non-drinkers), the indicator variable for light drinkers versus non-drinkers, at-risk drinkers versus non-drinkers and heavy drinkers versus non-drinkers; KORA F4, The Cooperative Health Research in the Region of Augsburg study; LASSO, least absolute shrinkage and selection operator; LBC, The Lothian Birth Cohort; MESA, The Multi-Ethnic Study of Atherosclerosis; WBCs, white blood cell counts.
Figure 2
Figure 2
A biomarker of heavy alcohol drinking. Four sets of cytosine-phosphate-guanine dinucleotides (CpGs) were selected at s=0.12 (5 CpGs), s=0.08 (23 CpGs), s=‘lambda.1se’ (78 CpGs) and s=‘lambda.min’ (144 CpGs) using least absolute shrinkage and selection operator (LASSO) in the Framingham Heart Study (FHS) cohort (the training cohort). ROC analysis was performed to classify heavy drinkers versus non-drinkers (left figure) and heavy drinkers versus light drinkers (right figure). ‘Non-drinkers’ were subjects with no alcohol consumption (i.e., g per day=0); ‘light drinkers’ were subjects who consumed 0<g per day⩽28 in men and 0<g per day⩽14 in women; ‘heavy drinkers’ were subjects who consumed ⩾42 g per day in men and ⩾28 g per day in women. ARIC, The Atherosclerosis Risk in Communities study; KORA F4, The Cooperative Health Research in the Region of Augsburg study; LBC1936, The Lothian Birth Cohort 1936; MESA, The Multi-Ethnic Study of Atherosclerosis.
Figure 3
Figure 3
Meta-analysis of epigenome-wide association of alcohol intake in European ancestry (EA) whole-blood samples: the Manhattan plot (top) and the volcano plot (bottom). The DNA methylation proportion was the outcome variable, grams alcohol consumed per day (g per day) was the predictor variable, adjusting for age, sex, body mass index, technical covariates and white blood cell counts. The inverse-variance weighted random-effects model was performed in meta-analysis using all whole blood DNA samples of EA.
Figure 4
Figure 4
Comparison of regression coefficients of the significant cytosine-phosphate-guanine dinucleotides (CpGs) in association analysis of the continuous alcohol trait (g per day): (a) between European and African whole-blood samples; (b) the Forest plot of effect estimates and standard errors of cg11376147 in all study cohorts; and (c) between European whole-blood and CD14+ monocyte samples. (a) Includes a list of CpGs with P<1 × 10−7 in EA or AA whole-blood samples and (c) includes a list of CpGs with P<1 × 10−7 in EA whole-blood samples or in monocyte samples of mixed ancestries. The Pearson’s correlation was r=0.64 between the effect estimates in (a) and r=0.72 in (c). MM, monocyte, mixed ancestries; WB AA, whole blood, African ancestry; WB EA, whole blood, European ancestry.
Figure 5
Figure 5
The γ-aminobutyric acid-A (GABA-A) receptor, delta (GABRD): the associations of the 36 cytosine-phosphate-guanine dinucleotides (CpGs) within GABRD, genomic and regulatory features and correlation of methylation measurements. The results were obtained in meta-analysis of the association analysis of 9643 whole-blood-derived DNA samples of European ancestry (EA) individuals. The correlation of these 36 CpGs was calculated using the methylation measurements at 36 CpGs, adjusting for age, sex, technical covariates and white cell blood counts in the Framingham Heart Study samples.

Comment in

Similar articles

See all similar articles

Cited by 54 articles

See all "Cited by" articles

References

    1. NIAAA. Alcohol Facts and Statistics. Available at: https://www.niaaa.nih.gov/alcohol-health/overview-alcohol-consumption/alcohol-facts-and-statistics.
    1. Rehm J, Baliunas D, Borges GL, Graham K, Irving H, Kehoe T et al. The relation between different dimensions of alcohol consumption and burden of disease: an overview. Addiction 2010; 105: 817–843. - PMC - PubMed
    1. Ogeil RP, Room R, Matthews S, Lloyd B. Alcohol and burden of disease in Australia: the challenge in assessing consumption. Aust NZ J Public Health 2015; 39: 121–123. - PubMed
    1. Rehm J, Taylor B, Roerecke M, Patra J. Alcohol consumption and alcohol-attributable burden of disease in Switzerland, 2002. Int J Public Health 2007; 52: 383–392. - PubMed
    1. Ferreira-Borges C, Rehm J, Dias S, Babor T, Parry CD. The impact of alcohol consumption on African people in 2012: an analysis of burden of disease. Trop Med Int Health 2015; 21: 52–60. - PubMed

Publication types

Grant support

Feedback