LAUSR

Background Prior investigation of the genetic architecture of educational attainment and schizophrenia has reported a counter-intuitive positive genetic correlation between these phenotypes. Yet, it has been widely reported that debilitating cognitive impairments are a primary feature of the illness, and schizophrenia has been shown to have an inverse genetic correlation with general cognitive ability. In this study, we sought to understand the differential genetic correlation that exists between those of cognition and education on schizophrenia, despite the former phenotypes having largely overlapping genetic architecture. Methods LD-Score Regression (LDSC) was conducted to examine the relationship between schizophrenia (PGC2-SczWGp, 2014, Nature, PMID: 25056061), and subsets of SNPs derived from GWAS for education (Okbay et al., 2016, Nature, PMID: 27225129) and cognition. GWAS summary statistics from Sniekers et al., 2017 (Nature Genetics, PMID: 28530673) and independent cohorts reported in Trampush et al., 2017 (Molecular Psychiatry, PMID: 28093568) were meta-analyzed resulting N=107,207 individuals for cognition. SNPs related to either education or cognition, but were completely unrelated to the other phenotype were identified. Thus, SNPs nominally significantly (p .50) were selected; there were 294,319 such SNPs. We also examined 386,664 SNPs nominally associated with cognition (p .50). Follow-up pathway analysis was conducted to identify potential biological pathways underlying each set of markers. Results We confirmed significant inverse genetic correlations between cognition and schizophrenia (rg=-.192, p=2.85e-10) and positive correlations between education and schizophrenia (rg=.097, p=3.91e-5). LDSC results for filtered markers demonstrated that the counter-intuitive positive correlation between education and schizophrenia was entirely driven by SNPs unrelated to cognition (rg[scz]=.55, p=1.06e-7). Much of the inverse correlation between cognition and schizophrenia was accounted for by the cognition-specific SNPs (rg[scz]=-.11, p=4.65e-2). By contrast, SNPs nominally significant (p Discussion While the genetic architecture between cognition and education is largely concordant; the discordant association with schizophrenia reveals a critical set of SNPs and pathways that differentiate the two. In schizophrenia, ion transport and ion channel regulation related pathways appear dysregulated, explaining the strong negative genetic correlations with cognition. However, it was a subset of pathways related to cell adhesion and neuronal development underlying education genetic architecture that appears to drive the counter-intuitive positive genetic correlation with schizophrenia. These findings suggest that a subset of the neurodevelopmental anomalies implicated in schizophrenia may be associated with educational success in the general population, perhaps providing a mechanism of balancing selection for these seemingly deleterious alleles.