Abstract
IMPORTANCE: Polysubstance use is common, but substance use associations with neuroimaging measures have largely been investigated within individual drug types. Whether effects are substance-specific or -general, and how predispositional risk and exposure contribute, remains unclear. OBJECTIVE: Identify shared and unique associations between substance use and brain structure, and characterize the contributions of predispositional risk and environmental exposure, in a large sample of young adults in the US. DESIGN: This cross-sectional family-based study used data from the Human Connectome Project (2017 release, collected from 2012-2015). SETTING: Data were collected at Washington University in St. Louis, MO, USA. PARTICIPANTS: Twins, non-twin siblings, and singletons with magnetic resonance imaging (MRI) and substance use self-report were included in the analysis. Data were analyzed in 2025. EXPOSURE: History of substance use was assessed using the Semi-Structured Assessment for the Genetics of Alcoholism. Variables included lifetime use, heavy or past-year hazardous use, and age of use onset for alcohol, marijuana, tobacco, and illicit drugs. Additionally, alcohol and marijuana dependence were assessed. MAIN OUTCOMES AND MEASURES: Linear mixed-effect models examined associations between substance use and brain structure, with an initial focus on past-year hazardous alcohol use, as 95% of the sample endorsed lifetime alcohol use. Analyses then tested associations with other substance use variables, and whether effects were shared or substance-specific. Between-family, within-family, and genetic variance component analyses tested risk and exposure effects. RESULTS: 1,113 participants (N = 445 families; ages 22 - 37; M=28.8, SD=3.7) had no missing data for the primary analyses. Hazardous alcohol use was negatively associated with global brain thickness (β = -0.12, p < 0.001), which explained all other regional and global associations. Of the drugs with a shared-effect on global brain thickness, only lifetime marijuana use explained unique variance over alcohol (β = -0.08, p = 0.013). Within-family analyses found evidence for unique putative exposure effects of both alcohol (β = -0.11, p < 0.001) and marijuana use (β = -0.07, p = 0.002) on global thickness. Marijuana use further showed a predispositional effect, both in between-family comparisons (β = -0.11, p = 0.007) and genetic variance component analyses (ρG = -0.2, p = 0.004), which were not explained by alcohol use. CONCLUSIONS AND RELEVANCE: Brain structural associations with substance use reflect substance-general and -specific effects, as well as a combination of predispositional and exposure effects. Findings suggest that the negative consequences of polysubstance use may reflect the additive effects of multiple unique exposures.