RESUMO
BACKGROUND: Neurocognitive impairment is a well-known phenomenon in schizophrenia that begins prior to psychosis onset. Connectome-wide association studies have inconsistently linked cognitive performance to resting-state fMRI. We hypothesized a carefully selected cognitive instrument and refined population would allow identification of reliable brain-behavior associations with connectome-wide association studies. To test this hypothesis, we first identified brain-cognition correlations via a connectome-wide association study in early psychosis. We then asked, in an independent dataset, if these brain-cognition relationships would generalize to individuals who develop psychosis in the future. METHODS: The Seidman Auditory Continuous Performance Task (ACPT) effectively differentiates healthy participants from those with psychosis. Our connectome-wide association study used the Human Connectome Project for Early Psychosis (n=183) to identify links between connectivity and ACPT performance. We then analyzed the North American Prodrome Longitudinal Study 2 (n=345), a multi-site prospective study of individuals at risk for psychosis. We tested the connectome-wide association study-identified cognition-connectivity relationship in both individuals at risk for psychosis and controls. RESULTS: Our connectome-wide association study in early-course psychosis identified robust associations between better ACPT performance and higher prefrontal-somatomotor connectivity (p<.005). Prefrontal-somatomotor connectivity was also related to ACPT performance in at-risk individuals who would develop psychosis (n=17). This finding was not observed in nonconverters (n=196) or controls (n=132). CONCLUSIONS: This connectome-wide association study identified reproducible links between connectivity and cognition in separate samples of psychosis and at-risk individuals who would later develop psychosis. A carefully selected task and population improves the ability of connectome-wide association studies to identify reliable brain-phenotype relationships.