This study was conducted to examine whether there are quantitative or qualitative differences in the connectome between psychiatric patients and healthy controls and to delineate the connectome features of major depressive disorder (MDD), schizophrenia (SCZ) and bipolar disorder (BD), as well as the severity of these disorders. Toward this end, we performed an effective connectivity analysis of resting state functional MRI data in these three patient groups and healthy controls. We used spectral Dynamic Causal Modeling (spDCM), and the derived connectome features were further subjected to machine learning. The results outlined a model of five connections, which discriminated patients from controls, comprising major nodes of the limbic system (amygdala (AMY), hippocampus (HPC) and anterior cingulate cortex (ACC)), the salience network (anterior insula (AI), and the frontoparietal and dorsal attention network (middle frontal gyrus (MFG), corresponding to the dorsolateral prefrontal cortex, and frontal eye field (FEF)). Notably, the alterations in the self-inhibitory connection of the anterior insula emerged as a feature of both mood disorders and SCZ. Moreover, four out of the five connectome features that discriminate mental illness from controls are features of mood disorders (both MDD and BD), namely the MFG→FEF, HPC→FEF, AI→AMY, and MFG→AMY connections, whereas one connection is a feature of SCZ, namely the AMY→SPL connectivity. A large part of the variance in the severity of depression (31.6%) and SCZ (40.6%) was explained by connectivity features. In conclusion, dysfunctions in the self-regulation of the salience network may underpin major mental disorders, while other key connectome features shape differences between mood disorders and SCZ, and can be used as potential imaging biomarkers.
Keywords: depression; effective connectivity; mood disorders; psychiatry; salience network; schizophrenia.