Background Young people experiencing psychological or behavioural problems are more likely to engage in poly-substance use. The common liability theory posits a common liability to all substance addictions. Such shared risk factors may therefore explain the observed patterns of comorbid substance use, where a common genetic susceptibility could underlie this co-occurrence. However, to date most research has failed to measure the aforementioned common liability, and the first genome-wide association studies (GWAS) have identified genetic variants specific to each substance rather than common ones. Therefore, the aim of this project is two-fold: firstly, to construct a new longitudinal measurement model of common liability to substance use; secondly, to investigate specific versus common genetic influences on substance use applying polygenic scoring.

Methods The sample includes 6399 young adults from the Avon Longitudinal Study of Parents and Children. Self-reported data on substance use (i.e. tobacco, alcohol, marijuana, and other illicit drugs) was collected at age 17.5, 20, and 22 with validated questionnaires. A Trait-State-Occasion model applied within Structural Equation Modelling (SEM) was used to identify a common trait factor for substance use across the three time points, as well as occasion- and drug-specific latent factors. Polygenic scores were calculated using GWAS results on substance use and related psychopathological, behavioural, and cognitive phenotypes. All analyses were conducted using R and PRSice software.

Results In Study 1, the Trait-State-Occasion model was tested using Robust Maximum Likelihood Estimation, which accounted for the non-normal distribution of the substance abuse measures. The model fitted the data well, χ2 (33)=100.572, p<0.001, CFI=0.987, RMSEA=0.018, SRMR=0.027. The variance of the trait factor was 0.155 [95% CI 0.124; 0.185]. On average, the model R-Squared was 70%, with residuals being 30%. The total variance explained by the model was further decomposed into trait, occasion, and drug-specific variance, which was equal to 16%, 11%, and 44% respectively. Study 2 (ongoing) will test a full SEM model by regressing the polygenic scores on both the common trait and drug-specific factors.

Conclusion The results of Study 1 provided evidence for a common liability to substance use, which is partly shared across different drugs and stable over time. This therefore suggests that both common and drug-specific susceptibility should be considered where examining genetic and environmental risk factors for substance use. Study 2 will reveal the influence of genetic variants associated with various psychopathology related-traits on both common and drug-specific dimensions of substance use.