Academic performance significantly influences educational advancement, career opportunities, and life outcomes. The extent to which adolescent substance use and brain morphology predict academic achievement has not been extensively explored. We examined grade point average (GPA) at the time alcohol and cannabis use often starts (7th - 9th grade) and subsequently during 11th and 12th grade in a 170 physically healthy adolescents in a longitudinal study. Covariance analysis examined predictive features from 36 metrics of middle school academic performance and initiation of alcohol and cannabis use. Using a machine learning approach, GPA from 7th, 8th, and 9th grade strongly predicted 11th and 12th grade GPA, followed in predictive power by alcohol use age of onset. A machine learning approach determined 16 (from 336) baseline neuroimaging features that reflected lower thickness, area, or volume in average high school GPA drinkers compared to nondrinkers. Features that distinguished average performing drinkers from nondrinkers suggested accelerated gray matter loss during adolescence for drinkers, while high performing drinkers compared to nondrinkers may have attenuated gray matter maturation. Additional possibilities are discussed.
Keywords: Academic performance; Adolescence; Alcohol use; Cannabis use; Machine learning; Magnetic resonance imaging.