Disentangling white matter correlates of symptom severity and general impairment in early psychosis

White matter (WM) alterations are well documented in individuals at clinical high-risk for psychosis (CHR-P) and with recent-onset psychosis (ROP), yet it remains unclear whether they reflect vulnerability, psychosis-related symptom severity, or general impairments shared with other disorders such as depression. To disentangle these correlates, we analyzed diffusion MRI data from 882 individuals (457 females) of the multisite Personalized Prognostic Tools for Early Psychosis Management (PRONIA) study, including 183 CHR-P, 206 ROP, 298 healthy controls, and 195 individuals with depression who were assessed as a clinical comparison group of observed associations. Fractional anisotropy (FA) was extracted from 25 WM regions. Analyses of covariance tested group differences across CHR-P, ROP, and healthy controls. Canonical correlation analysis then identified multivariate correlation components between FA and a broad set of twelve clinical, two cognitive, and two risk-related measures in CHR-P and ROP, restricted to regions with group effects. To identify if these findings are not specific to psychosis, we additionally tested whether the correlation components were expressed in the depression comparison group. Group differences emerged in 17 of 25 regions. Canonical correlation analysis identified two significant correlation components. Component 1 linked widespread lower FA with impaired cognition and functioning, representing general impairment. Component 2 linked focal lower FA with more severe psychosis-related symptoms and absence of familial risk. Component 1, but not Component 2, showed also a significant association in the depression group emphasizing its relevance for general impairment. Disentangling general impairment from psychosis-related symptom severity suggests potentially different underlying processes. Developing imaging biomarkers accounting for these processes may guide early detection strategies and targeted interventions.