Genetic overlap and causality among major depressive disorder, alcohol dependence, and alcohol consumption: Findings from the Psychiatric Genomics Consortium

Major depressive disorder (MDD) and alcohol dependence (AD) are common psychiatric disorders, contribute substantially to global morbidity, and often co-occur in the same individual. Epidemiological studies report that... [ view full abstract ]

Major depressive disorder (MDD) and alcohol dependence (AD) are common psychiatric disorders, contribute substantially to global morbidity, and often co-occur in the same individual. Epidemiological studies report that individuals with MDD are at increased risk for AD and vice versa. Leading hypotheses suggest these associations may be due to shared risk factors (e.g., genetic, environmental) or causal processes of one disorder leading to the other, such as the self-medication hypothesis. However, the mechanisms at the basis of the MDD-AD dual diagnosis, contemporaneously, or across the lifetime, remain unclear. Twin studies show genetic factors influence individual differences in susceptibility to MDD, AD, and alcohol consumption, with heritabilities on the order of 0.37, 0.49, and 0.43 respectively. Large-scale genome-wide association studies (GWAS) have identified putative risk variants for MDD, AD, and alcohol consumption quantity (AC-Quantity) and, like other complex diseases, have revealed polygenic architectures with multiple loci of small effect. Bivariate twin studies report moderate shared genetic liability between MDD and AD, estimating the genetic correlation from 0.3 to 0.6. However, questions remain whether these traits show genetic correlation because of shared genetic effects independently on each trait (i.e., pleiotropy) or because of causal processes indexed by genetic instruments. Here, we leverage summary statistics from genome-wide findings to estimate genetic correlations between MDD, AD, AC-Quantity, and alcohol consumption frequency (AC-Frequency). We obtained GWAS summary statistics of MDD and AD from the Psychiatric Genomics Consortium and conducted GWAS of alcohol consumption measures in the recently released UK biobank data. Further, we investigate support for causal mechanisms via two-sample Mendelian randomization (MR). Our results are consistent with twin studies, as MDD and AD showed moderate overlap of genetic risk factors (rg_MDD-AD=+0.47, P=6.6×10^-10). A significant genetic correlation between AC-Quantity and AC-Frequency was observed (rg=+0.52, P=1.3×10^-149), but MDD showed significant correlations with these traits in opposite directions (rg_MDD-ACQ=+0.14, P=2.9×10^-7; rg_MDD-ACF=-0.17, P=1.5×10^-10). The high genetic correlation between AD and AC-Quantity (rg_{AD-AC-Quantity}=0.75, P=1.8×10^-14) suggests that these phenotypes capture overlapping constructs and that quantity of consumption is an indicator of problematic alcohol use. The MR analysis indicated that most of these correlations are best explained by shared genetic mechanisms, but a strong causal relationship is present between MDD and AD as individuals with a lifetime MDD diagnosis have a 32% increased risk of having a lifetime AD diagnosis (beta=0.28, p=1.29×10^-6). Results indicate that comorbidity among MDD, AD, and alcohol consumption are partially due to shared genetic liability, and partially to causal mechanisms of MDD predisposing to AD. The current findings have important implications for both MDD and AD treatment and prevention efforts as well as understanding mechanisms involved in the etiology of psychiatric comorbidities.