Genetic, Environmental, and Social Interactions Shaping Early Cannabis Use (GENESIS): Decoding Predictive Factors Among U.S. Youth

About This Grant

PROJECT SUMMARY Early initiation of cannabis use (<16 years of age) increases risk for cannabis use disorder (CUD), mental illness, cognitive impairment, later unemployment, and poor social relationships. Prevention of early initiation is critical for improving social and health outcomes. Precision prevention programs have reduced youth substance use, but no approaches have specifically targeted cannabis use. Furthermore, no studies have comprehensively considered risk factors for early cannabis initiation (genetic, social, behavioral, environmental, and cognitive) to enhance the prediction of early use and inform precision prevention approaches. Comprehensive multivariable prediction models for early cannabis initiation that include genetics and social/environmental factors are needed. Cannabis use is polygenic, influenced by multiple genetic variants with weak-to-moderate effects, and polygenicity makes it difficult to translate genetics for clinical application. One method for clinically applying genetics is through the development of polygenic risk scores (PRS) that are composite scores representative of overall genetic risk. Prior PRS have typically lacked portability to non-European populations; however, a state- of-the-art method has been developed to build PRS with significantly improved risk prediction (34% improvement) across ancestries. There is a need to apply this method to develop cross-ancestry PRS for cannabis use for inclusion of overall genetic risk in comprehensive prediction models. Furthermore, given the complex interplay between genetics and social/environmental factors, research is needed to understand gene by environment (GxE) interactions in which social/environmental factors synergistically impact the risk conferred by genetics. Research into GxE interactions is statistically and computationally challenging, and traditional single-variant and more recent polygenic approaches focus on lower order 2-way interactions. Our logic forest (LF) algorithm efficiently searches all possible interactions up to 8 variables without a priori specification. This study will apply these state-of-the-art computational methods to the Adolescent Brain Cognitive Development (ABCD) Study, which examines childhood risk factors and initiation of substance use from ages 9-10 years to early adulthood in a population demographically reflective of the U.S. Nearly all youth had not used cannabis at recruitment, enabling the prospective measurement of initiation and the development of prediction models integrating genetics with pre-substance use measurements of cognitive, social, and environmental factors. This research will 1) develop cross-ancestry PRS for inclusion in prediction models that comprehensively consider genetic, sociodemographic, behavioral, cognitive, and environmental factors, and 2) apply LF to gene-sets within known biological pathways across the whole genome to identify pathway-specific GxE interactions. Comprehensive models coupled with a more complete understanding of GxE factors influencing early cannabis initiation can identify 1) high risk youth populations for targeted prevention, 2) targetable factors present among high-risk clusters for tailored interventions, and 3) biological pathways for therapeutic development.