Transdiagnostic and Illness-Specific Functional Dysconnectivity Across Schizophrenia, Bipolar Disorder, and Major Depressive Disorder

BACKGROUND: Mental disorders are typically defined as distinct diagnostic entities, but similar patterns of clinical and cognitive impairments are frequently found across diagnostic groups. We investigated whether these transdiagnostic deficits result from common neural substrates across disorders or various illness-specific mechanisms, or a combination of both. METHODS: Functional magnetic resonance imaging data were collected from clinically stable patients with major depressive disorder (n = 53), bipolar disorder (n = 78), or schizophrenia (n = 100) and matched healthy control subjects (n = 109) using a single scanner. Group comparisons were conducted to identify transdiagnostic and illness-specific features, and possible confounding effects of medication were considered. A multivariate approach with cross-validation was used to associate dysconnectivity features with shared cognitive deficits. RESULTS: Transdiagnostic dysconnectivities were identified within somatomotor (Cohen's d = 0.50-0.58) and salience (Cohen's d = 0.52-0.58) networks and between subcortical-limbic (Cohen's d = 0.55-0.69) and subcortical-dorsal attention (Cohen's d = 0.56-0.61) networks. The executive control network was found to be illness-specifically disconnected from the prefrontal-limbic-pallidal circuit in major depressive disorder (Cohen's d = 0.57-0.58), prefronto-striato-parietal circuit in bipolar disorder (Cohen's d = 0.48-0.53), and default mode network in schizophrenia (Cohen's d = 0.47-0.56). Working memory deficits were associated with a linear combination of 11 transdiagnostic and 5 illness-specific dysconnectivities (r = .322, p = 9.7 x 10(-4), n = 340). The associations of the identified dysconnectivities with medication dosage were nonsignificant. CONCLUSIONS: Disconnectivity in the somatomotor network was a common transdiagnostic profile, while there were illness-specific patterns in different parts of the prefrontal cortex for different disorders. These findings suggest that prominent psychiatric disorders share common impairments, possibly linked to perception and motor output, as well as unique dysconnectivity profiles that hypothetically mediate the more distinctive features of the disorder-specific psychopathology.