Structural equation modeling and principal component analysis of gray matter volumes in major depressive and bipolar disorders: differences in latent volumetric structure.

ABSTRACT

Abnormalities of the cortico-striatal-thalamic-cortical (CSTC) and the limbic-cortico-striatal-thalamic-cortical (LCSTC) circuits have been hypothesized in mood disorders. We performed principal component analysis (PCA) to identify latent volumetric systems on regional brain volumes and correlated these patterns with clinical characteristics; further, we performed exploratory structural equation modeling (SEM) to test a priori hypotheses about the organization among the structures comprising the CSTC and LCSTC circuits, and to investigate differences among subjects with bipolar disorder (BD), major depressive disorder (MDD), and healthy controls (HC). Participants included 45 BD and 31 MDD patients, and 72 HC. Regional MR brain volumes were used to calculate patterns of volumetric covariance. The identified latent volumetric systems were related to the depression severity and the duration of illness. BD differed from HC on the estimated parameters describing the paths of cortico-striatal, thalamo-striatal and intrastriatal loops of the CSTC circuit, and the paths between anterior and posterior cingulate cortex (PCC), and hippocampus to amygdala of the LCSTC circuit. MDD differed from HC on the paths between putamen and thalamus, and PCC to hippocampus. This study provides evidence to suggest different organizational patterns among structures within the CSTC and LCSTC circuits for BD, MDD, and HC, which may point to structural abnormalities underlying mood disorders.