A genome-wide association study of using substances to cope with depression and anxiety in the uk biobank and polygenic associations with alcohol use disorder severity.

Negative reinforcement effects of substance use (i.e., relief from negative affective states like stress, anxiety, or depressed mood) are associated with substance use disorder (SUD) development. Prominently featured in neurobiological models of addiction, shifts from positive to negative reinforcement are thought to reflect progression through stages of addiction. However, genetic influences underlying negative reinforcement motives, as precursors to developing addiction, are currently understudied in comparison to SUDs and other substance use-related consequences. Here we investigate genetic influences underlying negative reinforcement motives by conducting a genome-wide association study (GWAS) of using substances (i.e., drugs or alcohol) more than once to cope with symptoms of depression and anxiety (using to cope; UTC) among European ancestry individuals endorsing elevated anxious and depressive symptoms in the UK Biobank (ncases = 12,467, ncontrols = 72,457, neffective = 42,547). We also examine associations between UTC polygenic scores (PGSs) and AUD severity in the Collaborative Study on the Genetics of Alcoholism (COGA; N = 7,317). Our GWAS of UTC in the UK Biobank was conducted using a generalized linear mixed model approach controlling for sample relatedness by a sparse genetic relationship matrix (fastGWA) including sex, age, age × sex, age2, age2 × sex, and 20 genetic principal components as fixed effect covariates. LD-score regression was used to examine genetic correlations between UTC and other relevant traits including AUD, AUDIT-P scores, alcohol consumption, depression, anxiety, and neuroticism. Finally, SBayesR was used to calculate UTC and AUD PGSs in COGA, a family-based study ascertained for high prevalence of AUD. PGSs were used in mixed effects regression models, including fixed effect covariates of sex, age, age2, 10 genetic principal components, birth cohort, and genotyping array, and random effect of family membership, to predict AUD severity. Individuals reporting UTC demonstrated elevated AUDIT, PHQ-9, GAD-7 scores, and a higher prevalence of lifetime cannabis use compared to controls. A single variant for the UTC GWAS, rs1229984 in ADH1B, demonstrated genome-wide significance (P = 6.0 × 10-12). Genetic correlation analyses revealed a genetic correlation between UTC and AUD (rg = 0.73, SE = 0.07) that was significantly larger than between UTC and other mood disorder and negative affect traits analyzed (anxiety-rg = 0.39, SE = 0.09; depression-rg = 0.30, SE = 0.08; neuroticism-rg = 0.14, SE = 0.09) and between AUD and alcohol consumption (rg = 0.55, SE = 0.03). Further, the genetic correlation between UTC and AUDIT-P scores in the UK Biobank suggested near complete overlap (rg = 0.90, SE = 0.10). In COGA, UTC PGSs predicted greater AUD severity (β = 0.04, SE = 0.01, P = .006). This effect persisted with the inclusion of AUD PGSs in a multi-PGS model (β = 0.03, SE = 0.01, P = .019), suggesting unique associations between UTC PGSs and AUD severity not accounted for by AUD-specific genetic influences. These analyses provide preliminary evidence that negative reinforcement associated with substance use may be suitable as a partial genetic proxy for AUD. GWAS of specific characteristics or symptoms associated with disordered substance use can serve to refine our understanding of the genetic etiology of addiction. Future research aimed at further disentangling genetic liability reflecting domain-specific addiction risk is needed.