Global and regional cortical thickness associations with general and domain-specific cognition in first-episode psychosis
Background: Psychosis is often characterized by global and regional brain structure alterations as well as general and domain-specific cognitive deficits. Previous brain structure-cognition findings in psychosis generally involve brain regions of interest or individual cognitive domains that may reflect global/general rather than regional/domain-specific effects. Here, we systematically examined whether global cortical thickness and/or general cognition drive thickness-cognition associations in two first-episode psychosis (FEP) samples to examine generalizability. Methods: Magnetic resonance imaging, cognitive, and sociodemographic data were acquired in two samples (sample 1: 114 FEP, 81 control; sample 2: 62 FEP, 53 control). Multivariate global thickness-general cognition associations were examined in patients using partial least squares (PLS) to optimize brain-cognition associations. Then, mean thickness and mean cognition were cumulatively covaried from the thickness data and cognitive data, respectively, to probe associations between (1) regional thickness and general cognition (covarying mean thickness) and (2) regional thickness and domain-specific cognition (covarying mean thickness and mean cognition). Results: A general cognitive impairment was associated with global thinning as well as bidirectional regional alterations (decreased motor, inferior temporal, and increased frontal thickness), which related to symptoms and generalized out-of-sample. Sample- and domain-specific bidirectional alterations were observed for verbal memory, processing speed, visual memory, attention. Conclusions: Examining brain regions-of-interest and/or individual cognitive domains may capture global/general deficits rather than region- or domain-specific associations. Bidirectional thickness alterations point to inefficient cortical reorganization rather than unidimensional decline. Domain-specific associations may represent subgroups of patients with distinct sociodemographic, clinical, and cognitive profiles, who might benefit from more personalized interventions.