Structural Brain Correlates of Bulimia Nervosa Diagnosis and Symptom Severity: A Coordinated ENIGMA Eating Disorders Working Group Analysis.

Abstract

Bulimia nervosa (BN) ranks second among eating disorders in both prevalence and disease burden, but its neurobiology is poorly understood. Progress has been limited by neuroimaging studies with inconsistent findings across small samples. To identify reliable structural brain correlates of BN diagnosis and severity across multiple cohorts worldwide. This case-control study used 3-dimensional T1-weighted magnetic resonance imaging (MRI) and clinical data. Prospective meta-analyses assessed group differences in cortical thickness (CT), surface area (SA), and subcortical volumes (SV) and associations with symptom severity. Data were analyzed between February 1, 2022, and October 1, 2025. Thirteen international sites (17 cohorts) participating in the Enhancing Neuroimaging Genetics through Meta-Analysis (ENIGMA) Eating Disorders Working Group were included, yielding a total of 369 participants with BN and 417 healthy control individuals. Regional CT, SA, and SV extracted from T1-weighted MRI scans (uniformly preprocessed using ENIGMA-standardized pipelines and quality control procedures); binge eating severity; and compensatory behavior severity. Participants were all female with a mean (range) age of 23.6 (12.0-53.6) years. Relative to control individuals, the BN group had lower nucleus accumbens volume (Cohen d, -0.20; 95% CI, -0.34 to -0.06) and lower cortical SA of the superior temporal (d, -0.26; 95% CI, -0.41 to -0.12) and transverse temporal cortices (d, -0.24; 95% CI, -0.38 to -0.10). We did not observe significant group differences in CT. Most differences remained significant when controlling for cohort-level variation in body mass index, illness duration, depressive symptoms, psychotropic medication use, and purging as a compensatory behavior. Binge-eating frequency was associated with lower SA of the superior temporal cortex, insula, pars orbitalis, and medial orbitofrontal, rostral middle frontal, and isthmus and posterior cingulate gyri (r value range, -0.20 to -0.14) but not CT or SV. No anatomical characteristics were associated with total compensatory behaviors. This case-control study found that localized brain structural alterations, mostly in SA, were associated with BN and BN symptoms. These findings provide evidence implicating reward, cognitive control, interoceptive, and social cognition circuits in the pathophysiology of BN, advancing neurobiological models of a disorder with a poorly characterized brain basis.