Reduced grey matter volume in adolescents with conduct disorder: a region-of-interest analysis using multivariate generalized linear modeling

Abstract

Abstract Background Conduct disorder (CD) involves a group of behavioral and emotional problems that usually begins during childhood or adolescence. Structural brain alterations have been observed in CD, including the amygdala, insula, ventrolateral and medial prefrontal cortex, anterior cingulate cortex, and fusiform gyrus. The current study developed a multivariate generalized linear model (GLM) to differentiate adolescents with CD from typically developing (TD) adolescents in terms of grey matter volume (GMV). Methods The whole‐brain structural MRI data were collected from 96 adolescents with CD (mean age = $$16.188\pm 1.259$$ 16.188 ± 1.259 years; mean IQ = $$104.292\pm 8.107$$ 104.292 ± 8.107 ; 63 males) and 90 TD individuals (mean age = $$15.956\pm 1.506$$ 15.956 ± 1.506 years; mean IQ = $$106.622\pm 9.076$$ 106.622 ± 9.076 ; 59 males) matched on age, IQ, and sex. Region-wise GMV was extracted following whole-brain parcellation into 68 cortical and 14 subcortical regions for each participant. A multivariate GLM was developed to predict the GMV of the pre-hypothesized regions-of-interest (ROIs) based on CD diagnosis, with intracranial volume, age, sex, and IQ serving as the covariate. Results A diagnosis of CD was a significant predictor for GMV in the right pars orbitalis, right insula, right superior temporal gyrus, left fusiform gyrus, and left amygdala (F (1, 180) = 5.460–10.317, p < 0.05, partial eta squared = 0.029–0.054). The CD participants had smaller GMV in these regions than the TD participants (MCD–MTD = [− 614.898] mm3–[− 53.461] mm3). Conclusions Altered GMV within specific regions may serve as a biomarker for the development of CD in adolescents. Clinical work can potentially target these biomarkers to treat adolescents with CD.