In Reply In their letter, Goldstein et al bring attention to the interesting idea that the association between body mass index (BMI) and cortical thickness as reported in our study1 may be complicated by an interaction with psychiatric status. Adolescents with significant psychiatric conditions are more likely to be receiving medications that may influence appetite, energy level, weight gain, and potential metabolic dysregulation.2 Goldstein et al note preliminary empirical support for this perspective in 2 small but independent samples of bipolar patients.3,4 We attempted to replicate the referenced finding on frontal lobe thickness in the entire Adolescent Brain Cognitive Development cohort data from release 2.0.1. Among 10 111 individuals with complete data and who passed imaging quality control, 917 had either a current or prior diagnosis of bipolar disorder. We used a linear mixed-effects model, with BMI, bipolar diagnosis, age, sex, puberty status, race/ethnicity, household income, parental education, marital status, and intracranial volume as fixed effects and site and family structure as nested random effects, to test for a main effect of bipolar diagnosis or an interaction of BMI with diagnosis on cortical thickness or volume in the left or right rostral middle frontal, the medial orbitofrontal, and the anterior cingulate cortex. Of these multiple analyses, only the volume of the left medial orbitofrontal cortex produced a significant (P < .05) result, namely a main effect of bipolar diagnosis and an interaction between diagnosis and BMI. To put this in context, the Figure shows an example of cortical thickness from the right rostral middle frontal cortex, which we identified as being the region with the strongest association with BMI. While other investigators may want to pursue this idea further, similar slopes in the Figure illustrate the limited influence that bipolar diagnosis had on our reported findings. We encourage the scientific community to exploit the very large sample size and extensive imaging and nonimaging phenotyping of the ABCD data set, which offers unparalleled opportunities for hypotheses testing and replication in a developmental cohort.