Abstract
BACKGROUND AND HYPOTHESIS: Shared clinical features and genetic factors in schizophrenia (SCZ), bipolar disorder (BD), and major depressive disorder (MDD) have led to the hypothesis of common pathophysiological mechanisms. This study aims to elucidate aberrant transdiagnostic structural covariance patterns across these disorders employing a multivariate analytical approach. STUDY DESIGN: Structural magnetic resonance imaging data were acquired from a sample of 704 subjects, comprising 244 healthy controls, 119 first-episode treatment-naïve SCZ individuals, 159 BD individuals, and 182 treatment-naïve MDD individuals. Seed-based partial least squares correlation analysis was applied to construct structural covariance networks (SCNs) across 6 predefined functional networks: the default mode network (DMN), dorsal attention network (DAN), frontoparietal control network (FPCN), somatomotor network (SMN), ventral attention network (VAN), and visual network. Network seeds were selected based on functional network definitions. Spatial distributions of SCNs were calculated, and individual network integrity indices were derived as measures of SCN strength. Group comparisons of network integrity were performed using multiple t-tests to identify network-specific alterations across the diagnostic groups. STUDY RESULTS: Structural covariance patterns exhibited spatial distributions akin to those of functional networks. Network integrity showed common reductions across all 3 disorders in DMN, DAN, and FPCN, while BD showed specific reductions in the SMN, and both BD and MDD showed reductions in the VAN. Furthermore, there was a significant correlation between individualized network integrity and clinical and cognitive manifestations. CONCLUSIONS: Our results highlight the potential of the integrity of SCNs as transdiagnostic biomarkers.