Genome-Wide Association Study of Obsessive-Compulsive Symptoms including 33,943 individuals from the general population

Abstract

While 1-2% of individuals meet the criteria for a clinical diagnosis of obsessive-compulsive disorder (OCD), many more (~13-38%) experience subclinical obsessive-compulsive symptoms (OCS) during their life. To characterize the genetic underpinnings of OCS and its genetic relationship to OCD, we conducted the largest genome-wide association study (GWAS) meta-analysis of parent- or self-reported OCS to date (N = 33,943 with complete phenotypic and genome-wide data), combining the results from seven large-scale population-based cohorts from Sweden, the Netherlands, England, and Canada (including six twin cohorts and one cohort of unrelated individuals). We found no genome-wide significant associations at the single-nucleotide polymorphism (SNP) or gene-level, but a polygenic risk score (PRS) based on the OCD GWAS previously published by the Psychiatric Genetics Consortium (PGC-OCD) was significantly associated with OCS (P_fixed = 3.06 × 10^-5). Also, one curated gene set (Mootha Gluconeogenesis) reached Bonferroni-corrected significance (N_genes = 28, Beta = 0.79, SE = 0.16, P_bon = 0.008). Expression of genes in this set is high at sites of insulin mediated glucose disposal. Dysregulated insulin signaling in the etiology of OCS has been suggested by a previous study describing a genetic overlap of OCS with insulin signaling-related traits in children and adolescents. We report a SNP heritability of 4.1% (P = 0.0044) in the meta-analyzed GWAS, and heritability estimates based on the twin cohorts of 33-43%. Genetic correlation analysis showed that OCS were most strongly associated with OCD (r_G = 0.72, p = 0.0007) among all tested psychiatric disorders (N = 11). Of all 97 tested phenotypes, 24 showed a significant genetic correlation with OCS, and 66 traits showed concordant directions of effect with OCS and OCD. OCS have a significant polygenic contribution and share genetic risk with diagnosed OCD, supporting the hypothesis that OCD represents the extreme end of widely distributed OCS in the population.

Fig. 1. Miami plot of the association results from the GWAS meta-analysis (upper panel) and of the gene-wide association analysis (lower panel) of OCS.

The y-axes represent -log10 p-values for the association of SNPs/genes with OCS. The x-axis represents chromosomes 1 to 22. In the upper plot, the p-value threshold for genome-wide significance (p = 5 × 10⁻⁸) is represented by the horizontal red line, suggestive significance (p = 1 × 10⁻⁵) by the blue line. In the lower panel, Bonferroni-corrected gene-wide significance (p = 2.708 × 10⁻⁶) is represented by the horizontal red line, suggestive gene-wide significance (p = 1 × 10⁻³) is indicated by the blue horizontal line.

Fig. 2. Genetic correlations between OCS and a variety of other traits and disorders.

Genetic correlations (rg) between OCS and a broad range (N = 97) of other phenotypes, assembled into 11 groups (psychiatric, substance, cognition/socioeconomic status (SES), personality, psychological, neurological, autoimmune, cardiovascular, anthropomorphic, fertility, and other). Error bars represent 95% confidence intervals; red circles indicate significant association after FDR correction for multiple testing.

Fig. 3. Comparing concordance and discordance between genetic correlation estimates of OCS and OCD.

Comparison between genetic correlation estimates (r_G) of OCS and OCD for 97 other phenotypes, color-coded according to 11 groups (psychiatric, substance, cognition/socioeconomic status (SES), personality, psychological, neurological, autoimmune, cardiovascular, anthropomorphic, fertility, and other). On the x-axis the genetic correlation estimates for OCD are displayed, on the y-axis for OCS. Green quadrants indicate concordance, red quadrants discordance in the direction of genetic correlations between OCS and OCD.