T94. polygenic scores and the relationship between cognition and psychosis

Over 80% of psychosis patients experience cognitive impairment. Heritability of both psychosis and cognition means cognitive performance could be an endophenotype for psychosis. In a sample with 4,827 adults and 5,742 children, we investigated the effect of polygenic risk scores (PRSs) for schizophrenia and bipolar disorder on cognitive performance. We also assessed whether intelligence and educational attainment PRSs could predict psychosis presentation. Schizophrenia PRS was negatively associated with visuospatial processing/problem-solving in the adult sample (β: -0.0770; 95% confidence interval [CI]: -0.114, -0.0403) and working memory (β: -0.171; 95% CI: -0.264, -0.0774), processing speed (β: -0.158; 95% CI: -0.250, -0.0658), episodic memory (βs: -0.178 to -0.187; 95% CIs: -0.284 to -0.0807), language (β: -0.171; 95% CI: -0.263, -0.0802), fluid intelligence (β: -0.194; 95% CI: -0.286, -0.101), and total intelligence (β: -0.186; 95% CI: -0.275, -0.0969) in the child sample. Bipolar disorder PRS was not associated with cognitive endophenotypes after Bonferroni correction. Lower intelligence PRS identified psychosis cases from controls in adults (odds ratio [OR]: 0.884; 95% CI: 0.809–0.966) and those experiencing prodromal psychosis symptoms in childhood (OR: 0.786; 95% CI: 0.725–0.852). Lower educational attainment PRS was associated with prodromal symptoms (OR: 0.713; 95% CI: 0.658–0.773). The relationship between psychosis and cognitive impairment can be demonstrated bidirectionally at the neurobiological level. The effect of schizophrenia PRS and bipolar disorder PRS on cognitive performance differ from each other and across cognitive domains. These specific domains may therefore be better endophenotypes than general cognition.