Genetic overlap between mood instability and alcohol-related phenotypes suggests shared biological underpinnings

Abstract

ABSTRACTAlcohol use disorder (AUD) is a pervasive and devastating mental illness with high comorbidity rates with other mental disorders. We aimed to characterize the sharedvs. unique polygenicity of AUD, alcohol consumption (AC) and mood instability (MOOD), a relevant transdiagnostic factor, using large genome-wide association studies (GWASs) data. We hypothesize that cross-analyzing these phenotypes would shed light on their unique and shared polygenicity, increase our knowledge regarding the genetic basis of the comorbidity between AUD and mood disorders, and boost discovery for jointly-associated loci. Summary statistics for MOOD, AC and AUD GWASs (Ns =363,705; 200,680 and 200,004; respectively) were analysed to characterize the cross-phenotype associations between MOOD and AC, MOOD and AUD and AC and AUD, respectively. To do so, we used a newly-established pipeline that combines (i) the bivariate causal mixture model (MiXeR) to quantify the cross-phenotype polygenic overlap and (ii) the conjunctional false discovery rate (conjFDR) to discover specific jointly-associated genomic loci. These loci were functionally characterized and mapped to genes and biological functions. We also performed validation in independent samples and phenotypic analyses. MOOD was highly polygenic (10,400 single nucleotide polymorphisms, SNPs) compared to AC and AUD (4,900 SNPs, SD =600 and 4,300 SNPs, SD =2,000; respectively). The polygenic overlap of MOOD and AC was much larger than that of MOOD and AUD (98%vs. 49%) and genetic correlation was opposite (−0.2vs. 0.23), which was confirmed in independent samples. MOOD&AUD causal SNPs were significantly enriched for brain genes, conversely to MOOD&AC. Among 38 loci identified in the joint analysis, sixteen were novel for MOOD, AC and AUD. Similarly distinct patterns were evidenced for SNP localization, function and previous GWAS associations outside of the phenotypes that were currently studied. MOOD, AC and AUD were also strongly associated at the phenotypic level. Overall, using multilevel polygenic quantification, joint loci discovery and functional annotation methods, we evidenced that the polygenic overlap between MOOD and AC/AUD implicated shared biological underpinnings but clearly distinct functional patterns between MOOD&AC and MOOD&AUD. Using the MOOD endophenotype, the current study suggests new mechanisms for the comorbidity of AUD with mood disorders.