Parsing Heterogeneity in Mood Disorders: The Challenges of Modeling Stable Mood Disorder-Related Functional Connectomes.

Despite the reported links between mood disorders and neural network abnormalities, discrepancies in findings abound. Given the heterogeneous nature of depression, attempts to use resting-state connectivity to identify different depression subgroups have sometimes failed to be replicated (1,2). Examining functional neural networks during active clinical states of depression involving thousands of combinations of different symptoms across individuals may result in the identification of unstable biomarkers of mood disorders. Focusing on remitted or euthymic clinical phases that reduce symptom variability may elucidate more stable and trait-like mood disorder biomarkers. In the current issue of Biological Psychiatry: Cognitive Neuroscience and Neuroimaging, Langenecker et al. (3) use a graph theory–based approach focusing on resting-state functional network edges in a transdiagnostic mood disorder sample in the remitted or euthymic phase. Notably, Langenecker et al. (3) combine both diagnostic category and the National Institute of Mental Health’s Research Domain Criteria (RDoC) frameworks to examine associations between functional network edges with mood disorder diagnostic status and mood disorder–relevant RDoC constructs of response inhibition and reward responsiveness. The authors report interactions between mood disorder diagnostic status and these RDoC-defined constructs, with better response inhibition or greater reward responsiveness among the mood disorder group being linked to different functional network patterns compared with the healthy control group. These results highlight the value of combining both frameworks to enhance the understanding of mood disorder pathophysiology. There has been growing enthusiasm for the use of restingstate functional magnetic resonance imaging (fMRI) as a potentially powerful tool for identifying biomarkers of psychiatric disorders. Resting-state fMRI is a particularly desirable neuroimaging modality for clinical applications as it is easy to collect, is less burdensome for participants than cognitively demanding tasks, and has been shown to reliably derive largescale intrinsic functional neural networks across both healthy control subjects and clinical populations (4). Mood disorders, including major depressive disorder (MDD) and bipolar disorder (BD), have been linked to abnormalities involving the default mode network (DMN), the cognitive control network (CCN), and the salience and emotion network (SEN) (5,6). The DMN is a functional network that is involved in internal/selfreferential thought processes and has core brain hubs in the medial prefrontal cortex and the posterior cingulate cortex. The