Association of Fetal Gene Regulatory Gene Deletions With Poor Cognition in Schizophrenia and Community-Based Samples

OBJECTIVE: Schizophrenia is a neurodevelopmental disorder involving clinical and genetic heterogeneity. Multiple recurrent copy number variants (CNVs) increase risk for schizophrenia spectrum disorders (SSDs). However, it is unclear how known risk CNVs and broader genome-wide CNVs influence clinical variability. Furthermore, whether biological annotation of CNV scores can improve power for patient stratification is unknown. This study therefore investigated the relationships between severe SSD-related phenotypes and varied CNV metrics, including CNV burden affecting genes involved in different aspects of neurodevelopment. METHODS: This study of 617 individuals with SSDs examined associations of two severe phenotypes-childhood-onset psychosis and borderline intellectual functioning (IQ)-with 1) known risk CNVs, 2) genome-wide deletion burden scores, and 3) novel scores capturing deletion burden in 18 previously validated and mutually exclusive gene sets, representing distinct aspects of neurodevelopment. Associations with borderline IQ were assessed for replicability in 233 relatives of SSD patients, 581 control subjects, and 9,930 youths from the Adolescent Brain Cognitive Development (ABCD) Study. RESULTS: Known SSD-risk CNVs (odds ratio=7.07, 95% CI=1.60, 31.32) and neurodevelopmental disorder (NDD)-risk CNVs (odds ratio=4.56, 95% CI=1.48, 14.10) were associated with borderline IQ in SSDs. Furthermore, beyond effects of known NDD-risk CNVs, deletion of genes involved in regulating gene expression during fetal brain development was associated with borderline IQ across SSD cases and noncases (odds ratio=2.57, 95% CI=1.44, 4.60) and in the ABCD cohort (odds ratio=1.33, 95% CI=1.00, 1.76). Exploratory structural MRI-based analyses showed associations between fetal gene regulatory gene deletions and altered gray matter volume (b=0.09, 95% CI=0.004, 0.17) and cortical thickness (b=0.14, 95% CI=0.05, 0.24) across SSD cases and noncases. CONCLUSIONS: The study results confirm contributions of known risk CNVs to severe phenotypes in SSDs, implicate disrupted fetal brain development in poor cognition, and demonstrate the utility of a neurodevelopmental framework for identifying mechanisms underlying severe SSD-relevant phenotypes.