LAUSR

Background Excessive alcohol use is a leading contributor to morbidity and mortality. 12% of drinkers meet criteria for Alcohol Dependence (AD), a serious psychiatric disorder characterized by tolerance, withdrawal, loss of control over drinking and excessive alcohol consumption despite negative consequences. Alcohol Dependence (AD) is heritable and previous studies have implicated functional variants in alcohol dehydrogenase genes ADH1B and ADH1C and loci elsewhere (with less support), but relatively little is known about the full polygenic architecture of AD. Methods Genotype data were collected for 12,798 individuals with alcohol dependence (DSM-IV diagnosis or equivalent) and 29,087 controls from 22 studies. Each study was strictly QCed, stratified by ancestry (N=35,605 European ancestry, 6,208 African ancestry), and imputed using standardized pipelines for case/control (N=16,296) and family-based (N=22,912) cohorts. GWAS were performed for each cohort, with family-based studies analyzed using generalized estimating equations (GEE) or logistic mixed models to control for family structure as appropriate. Results were then aggregated in genome-wide meta-analysis along with summary statistics from 5 additional studies (total 14,519 cases, 37,295 controls for meta-analysis). We evaluated heterogeneity by sex, ethnicity (European vs. African ancestry), and study ascertainment (case/control vs. family-based). In addition to the primary GWAS, we characterize the genome-wide pattern of results through analysis of partitioned heritability and genetic correlation with AD-related traits using LD score regression, using the European ancestry cohorts. Results In initial results, one locus reached genome-wide significance (ADH1B rs1229984, p=1.8e-11), replicating previous work. We applied trans-ethnic meta-analysis models to improve fine-mapping of variants in the region. Different lead variants in the ADH1B locus between the European and African ancestry analyses are observed and are consistent with differences in allele frequency and linkage disequilibrium rather than effect size. There was little evidence for genome-wide heterogeneity of effects by sex, ethnicity, or study design. Comparing the GWAS of AD to related traits identified significant genetic correlation with several psychiatric disorders, including schizophrenia (rg=.36, p=2.4e-10), ADHD (rg=.44, p=.0003), and depressive symptoms (rg=.60, p=6.1e-7). The results are also strongly correlated with GWAS of alcohol consumption (rg=.71, p=6.8e-6), and use of cigarettes (rg=.58, p=2.5e-6) and cannabis (rg=.73, p=.003). Similar to other psychiatric disorders, partitioned heritability analysis suggests that common variant risk for AD is strongly enriched in evolutionarily conserved regions of the genome (28-fold enrichment, p=.008). Discussion To our knowledge, the current results represent the largest GWAS of DSM-IV AD to date. In addition to replicating previous associations with ADH1B, suggestive evidence for association of AD with a novel intergenic locus on chromosome 3 is awaiting replication. The genetic correlation results strongly suggest substantial overlap in genetic risk with other psychiatric disorders and with use and abuse of other substances.