Substance use disorders exhibit unique and disorder specific genetic associations with externalizing and internalizing psychopathology

Background and AimsSubstance use disorders (SUDs) are heritable and share genetic variance with externalizing and internalizing psychopathology. Although recent gene identification efforts have demonstrated the value of modeling the shared genetic architecture among SUDs and externalizing, most research has thus far failed to account for overlap with internalizing. In this study, we aim to characterize the genetic relationships of both externalizing and internalizing with SUDs. Design and settingWe used genome-wide association study (GWAS) summary statistics derived from previously published studies of externalizing, internalizing, and SUD outcomes to quantify the genetic overlap between these phenotypes. We characterize this overlap using omnibus, partial, and local genetic correlations, estimates of their shared polygenic effects, genetic causality models, polygenic score (PGS) analyses, and estimates of each SUDs residual variance derived from models in Genomic SEM. ParticipantsWe used GWAS summary statistics from individuals whose genomes were most similar to those from reference panels sampled from Europe (Ns ranged from 45,395 to 1,565,618) and Africa (Ns ranged from 30,000 to 122,571). For polygenic scores analyses, we used data from individuals of European and African ancestry groups available in the Collaborative Study on the Genetics of Alcoholism (COGA) sample (NMaximum = 7,394 for European-like genomes and 3,238 for African-like genomes) MeasurementsMeasurements in this study include GWAS summary statistics for externalizing, internalizing, and four substance use disorders: problematic alcohol use (PAU), cannabis use disorder (CUD), opioid use disorder (OUD), and tobacco use disorder (TUD). SUD outcomes in COGA were DSM-IV symptom counts of AUD, CUD, and OUD and scores on the Fagerstrom Test for Nicotine Dependence. FindingsWe found strong genetic relationships of externalizing and, to a lesser extent, internalizing with all SUDs across methods. Despite their more modest associations, internalizing emerged as an important genetic correlate of SUDs. After accounting for variance shared with externalizing, partial genetic correlations between internalizing and SUDs were attenuated but, with the exception of TUD, still significant. Similarly, the PGSINT accounted for a statistically significant increase in variance over and above PGSEXT. Two SUD specific patterns emerged such that TUD was least associated with both psychopathology spectra and OUD was most strongly related to internalizing relative to other SUDs. ConclusionsFrom these findings we conclude that shared genetic influences may explain comorbidity observed between SUDs and internalizing disorders and suggest that genetic risk for internalizing should be incorporated into SUD identification and prevention efforts. Future gene identification efforts should study SUDs in the context of both externalizing and internalizing psychopathology.