Genome-wide association study results for educational attainment aid in identifying genetic heterogeneity of schizophrenia

Abstract

Higher educational attainment (EA) is negatively associated with schizophrenia (SZ). However, recent studies found a positive genetic correlation between EA and SZ. We investigate possible causes of this counterintuitive finding using genome-wide association study results for EA and SZ (N = 443,581) and a replication cohort (1169 controls; 1067 cases) with deeply phenotyped SZ patients. We find strong genetic dependence between EA and SZ that cannot be explained by chance, linkage disequilibrium, or assortative mating. Instead, several genes seem to have pleiotropic effects on EA and SZ, but without a clear pattern of sign concordance. Using EA as a proxy phenotype, we isolate FOXO6 and SLITRK1 as novel candidate genes for SZ. Our results reveal that current SZ diagnoses aggregate over at least two disease subtypes: one part resembles high intelligence and bipolar disorder (BIP), while the other part is a cognitive disorder that is independent of BIP.

Fig. 1

Workflow of the proxy-phenotype analyses. Notes: Educational attainment (EA) and schizophrenia (SZ) GWAS results are based on the analyses reported in refs.^, . All cohorts that were part of the SZ GWAS were excluded from the meta-analysis on EA. The GRAS data collection was not included in either the SZ or the EA meta-analysis. Proxy-phenotype analyses were conducted using 8,240,280 autosomal SNPs that passed quality control. Genetic outliers of non-European descent (N = 13 cases) were excluded from the analysis in the GRAS data collection

Fig. 2

Results of the proxy-phenotype analyses. Notes: a Manhattan plot for educational attainment (EA) associations (n = 363,502). The x axis is the chromosomal position, and the y axis is the significance on a −log₁₀ scale (two-sided). The black dashed line shows the suggestive significance level of 10⁻⁵ that we specified in our preregistered analysis plan. Black and magenta crosses identify EA-associated lead-SNPs that are also associated with SZ at nominal or Bonferroni-adjusted significance levels, respectively. b Q–Q plot of the 506 EA-associated SNPs for schizophrenia (SZ) (n = 34,409 cases and n = 45,670 controls). SNPs with concordant effects on both phenotypes are pink, and SNPs with discordant effects are blue. SNPs outside the grey area (21 SNPs) pass the Bonferroni-corrected significance threshold that corrects for the total number of SNPs we tested (P < 0.05/506 = 9.88 × 10⁻⁵) and are labelled with their rs numbers. Observed and expected P values are on a −log₁₀ scale. For the sign concordance test: P = 0.40, two-sided

Fig. 3

Genetic correlations of GWAS and GWIS results. Notes: The heatmap displays the genetic correlations across seven sets of GWAS or GWIS summary statistics. Genetic correlations were estimated with LD score regression ^. The colour scale represents the genetic correlations ranging from –1 (red) to 1 (blue). Asterisks denote significant genetic correlations at P value < 0.01