Mega-Analysis Finds Shared Brain Structure Differences Across Youth ADHD, Conduct Disorder, Anxiety, and Depression

Key Takeaways

• A mega-analysis from the ENIGMA Consortium compared brain structure in youths with ADHD, conduct disorder (CD), anxiety disorders, and depression against healthy controls, using MRI data from 4,037 patients and 4,743 controls.
• All four disorder groups showed lower surface area in the insula, entorhinal cortex, and middle temporal gyrus, and lower amygdala volume. Effects were small, and the authors link these regions to the brain's salience network.
• Only ADHD and CD shared a further pattern of reduced surface area in frontoparietal regions. No internalizing-specific alterations were found, and none were specific to depression.

Researchers from the ENIGMA Antisocial Behavior, ADHD, Major Depressive Disorder, and Anxiety Working Groups, led by Sophie Townend and Graeme Fairchild at the University of Bath, report in Biological Psychiatry the results of a cross-disorder analysis of brain structure in youth. The authors say it is the largest mega-analysis to compare the most common internalizing disorders (anxiety and depression) and externalizing disorders (ADHD and CD) in young people.

The team pooled individual participant-level 3D T1-weighted MRI data from youths aged 4 to 21, including 1,317 with ADHD, 1,172 with CD, 1,044 with anxiety disorders, and 504 with depression, plus 4,743 healthy controls. Scans were processed with FreeSurfer using standardized ENIGMA protocols, yielding cortical thickness and surface area for 34 regions and volumes for 7 subcortical regions. Site effects were adjusted with ComBat. Linear models compared each diagnostic group with controls, adjusting for site, age, and sex, plus total intracranial volume for surface area and subcortical volume. Results were corrected for false discovery rate. Associations found in all four disorders were called transdiagnostic. Those found in at least two, across internalizing and externalizing groups, were called shared. Those found in one group only were called disorder-specific. The project was preregistered, and sensitivity analyses adjusted for IQ and psychotropic medication, where data were available.

The transdiagnostic findings were lower surface area in the insula, entorhinal cortex, and middle temporal gyrus, along with lower amygdala volume, total surface area, and intracranial volume relative to controls. Effect sizes were small, with Cohen's d values for the regional and amygdala findings between about −0.07 and −0.24.

Other findings were shared across at least one internalizing and one externalizing disorder. Frontal pole and pars orbitalis surface area and pallidum volume were lower in CD, anxiety, and depression. Rostral middle frontal surface area was lower in ADHD, CD, and anxiety, and precentral cortical thickness was lower in ADHD, anxiety, and depression. ADHD and CD alone shared reduced surface area in frontoparietal regions, including the caudal middle frontal gyrus, superior frontal gyrus, lateral orbitofrontal cortex, and inferior parietal cortex. No internalizing-specific associations were found.

Several alterations were specific to a single disorder, though not to depression. ADHD was linked to lower posterior cingulate surface area and lower fusiform and temporal pole thickness. CD showed the most widespread differences, with lower surface area in 21 of 34 cortical regions, seven of them specific to CD, and lower volume in the hippocampus, nucleus accumbens, and thalamus. Anxiety showed specific frontal, occipital, and cingulate differences and was the only group with higher putamen volume and higher cortical thickness (in the isthmus cingulate). Differences in surface area and subcortical volume were more widespread and larger than differences in cortical thickness. In some regions, such as the parahippocampal gyrus, the CD and anxiety groups differed from controls in opposite directions.

Several age-by-diagnosis interactions emerged for ADHD, anxiety, and depression. Effects were larger in children under 12 than in adolescents, but only eight youths with depression were under 12. There were no significant sex-by-diagnosis interactions.

The authors note that the data are cross-sectional, so they can't confirm altered developmental trajectories. Comorbidity could not be examined systematically because data weren't consistent across sites. An exploratory check in CD suggested a limited effect, but it couldn't be done for ADHD. The diagnostic groups differed in age, with the depression group older on average. "Disorder-specific" meant a significant difference from controls in one group only, not proven exclusivity. The study covered only four disorders, so the labels transdiagnostic, internalizing-specific, and externalizing-specific apply only to these four. Depression was the smallest group, with high co-occurring anxiety (56%), which may have limited power. Diagnostic subtypes within groups were not examined, which may have contributed to the small effect sizes. Most included studies used clinical case-control designs, which may limit generalizability to the wider population.

Sensitivity analyses also showed that many findings were sensitive to medication status. Ninety percent of the disorder-specific associations survived IQ adjustment, but only 42% survived medication adjustment, and adjusted analyses were limited by missing data. The authors say medication may reflect a real effect or be a proxy for disorder severity, which they could not assess.

The authors conclude that youths with different diagnoses share reductions in cortical surface area and subcortical volume, alongside disorder-specific patterns, with CD showing the most widespread changes. They suggest the shared insula and amygdala findings point to the salience network as a target for research into common mechanisms across youth psychiatric disorders. They also say the stronger effects in children highlight the need to study psychopathology early in development.