Distinct Patterns of Weight Gain, Age, and Subcortical Microstructure in Early Adolescence

IMPORTANCE Associations between childhood obesity and brain microstructural differences have been observed. It remains unknown whether these associations are driven by sex-specific excessive weight gain. Restriction spectrum imaging characterizes brain tissue microstructural health via water diffusion, where the restricted normalized isotropic (RNI) compartment assesses neuronal and glial cellularity, which may reflect neuroinflammation, synaptic pruning, or both. OBJECTIVE To identify associations among RNI scaling factor values, normal neurodevelopment, and weight gain during early adolescence. DESIGN, SETTING, AND PARTICIPANTS This cohort study used data from the Adolescent Brain Cognitive Development (ABCD) Study, a 10-year, ongoing, multisite longitudinal cohort study conducted among youths aged 9 to 20 years. The analyses focused on data from baseline (collected in 2016-2018) and the 2-year follow-up (collected in 2018-2020). Participants who initially had healthy weight (body mass index [BMI] percentile <85th) at age 9 to 10 years were eligible for this study. Data analysis was performed between March 2024 and March 2025. MAIN OUTCOMES AND MEASURES Linear mixed-effects models were used to examine bidirectional associations among RNI, age, and BMI (a proxy for weight gain) across 16 appetite-controlling brain regions. First, analysis was performed among youths with healthy-weight, weight-stable (HW-WS) status, stratified by sex. Second, an evaluation was conducted to determine how these associations changed among all youths (eg, conversion to healthy-weight, non–weight-stable [HW-NS] status). RESULTS At baseline, data were available for 3110 youths (mean [SD] baseline age, 119.2 [7.5] months); at year 2, only 1855 youths had complete data. Of these 1855 youths, 1072 (592 males [55.2%]) had HW-WS and 773 (445 females [57.2%]) had HW-NS. Among youths with HW-WS, RNI values were associated with age but not BMI. Among youths with varying weight gain, RNI values had bidirectional associations with BMI across many subcortical regions independent of age. In females, but not in males, higher RNI values had robust associations with greater increases in BMI over time. CONCLUSIONS AND RELEVANCE In this cohort study, RNI values were associated with age and BMI, and greater RNI values beyond normal developmental processes may suggest neuroinflammation. Thus, higher RNI values may signal neuroinflammatory processes associated with unhealthy weight gain, suggesting potential for the RNI scaling factor as an early indicator of obesity-related neurodevelopmental changes in adolescence. Future mechanistic studies are needed to determine the specific cellular changes underlying these associations.