Longitudinal Associations between Physical Activity and Structural Brain Volume in Mid-Adolescence: Evaluating Metabolic Activity in the ABCD Study

Importance: Despite the importance placed on physical activity throughout childhood to adolescence, the extent to which physical activity is associated with trajectories of brain structure development remains unclear.Objective: To investigate the associations between physical activity on volumetric gray matter morphometry throughout adolescence.Design: The Adolescent Brain Cognitive Development (ABCD) Study is a cohort-based longitudinal design of over 11,880 participants from ages 9-10 followed for 10 years.Setting: Multi-site study design including 21 U.S.-based sites within the continental United States.Participants: 9,291 participants (ages 9-17; 49% female; 55% white/non-Hispanic) with usable neuroimaging and sports activity involvement questionnaire data were analyzed across four longitudinal follow-up visits (i.e., year-0 to year-6).Exposures: Past-year amount of physical activity, from sports and activities, converted to a common metric (metabolic equivalent of task; METs) across late childhood to mid-adolescence.Main Outcomes and Measures: Longitudinal linear mixed effect models evaluated brain structure measured across 84 gray matter volumetric regions-of-interest. The interaction between METs and age was the primary effect investigated; non-linear fits of age were evaluated. Covariates included: total intracranial volume, sex, household income, body mass index, pubertal stage, MRI software version, genetic principal components, and random effects of participant identification, family identification, and MRI serial number.Results: Across 24,929 total observations, participants engaged in an average of 13,858 past-year MET-minutes (i.e., 266 MET-minutes/week) from sports/activities with stepwise increases across time. A quadratic effect of age resulted in optimal model fit. Linear mixed-effect regressions identified significant effects of total past-year METs*age-squared across 23 regions-of-interest with more pronounced trajectories of changes in volume observed for youth with higher past-year METs; standardized beta estimates range from -.005 to -.01 (all FDR-correct p