Linked dimensions of psychopathology and connectivity in functional brain networks

1 Neurobiological abnormalities associated with psychiatric disorders do not map well to existing diagnostic 2 categories. High co-morbidity and overlapping symptom domains suggest dimensional circuit-level 3 abnormalities that cut across clinical diagnoses. Here we sought to identify brain-based dimensions of 4 psychopathology using multivariate sparse canonical correlation analysis (sCCA) in a sample of 663 youths 5 imaged as part of the Philadelphia Neurodevelopmental Cohort. This analysis revealed highly correlated 6 patterns of functional connectivity and psychiatric symptoms. We found that four dimensions of 7 psychopathology — mood, psychosis, fear, and externalizing behavior — were highly associated 8 (r=0.68-0.71) with distinct patterns of functional dysconnectivity. Loss of network segregation between the 9 default mode network and executive networks (e.g. fronto-parietal and salience) emerged as a common 10 feature across all dimensions. Connectivity patterns linked to mood and psychosis became more prominent 11 with development, and significant sex differences were present for connectivity patterns related to mood 12 and fear. Critically, findings replicated in an independent dataset (n=336). These results delineate 13 connectivity-guided dimensions of psychopathology that cut across traditional diagnostic categories, which 14 could serve as a foundation for developing network-based biomarkers in psychiatry. 15 . CC-BY-NC 4.0 International license peer-reviewed) is the author/funder. It is made available under a The copyright holder for this preprint (which was not . http://dx.doi.org/10.1101/199406 doi: bioRxiv preprint first posted online Oct. 6, 2017;