A burden of rare copy number variants in obsessive-compulsive disorder

Obsessive Compulsive Disorder (OCD) features substantial contributions to risk from genetic variation. As with other neuropsychiatric conditions, current data support a subset of this genetic risk coming from rare variants impacting genes that are more constrained within the general population. Most of the current data supporting this contribution from rare, high-impact variation comes from exome sequencing studies, which are optimized for the detection of small variants in coding regions. Unlike similar studies of other conditions, the field lacks a well-powered case/control study focused on copy number variants (CNVs), which are large deletions and duplications of genomic loci. To fill this gap in the field, we have assembled a comparison of CNV burden in 2248 OCD cases versus 3608 unaffected controls. We generated genotype array data for OCD cases from Sweden and Norway as part of the NIH-funded NORDiC study (R01 MH110427). All cases were genotyped on the Illumina Global Screening Array. We collected previously-generated Swedish and Norwegian control data also genotyped on the Illumina Global Screening array. CNV calls were made using PennCNV and QuantiSNP, across markers present and at sufficiently low missingness in all input data. We only considered CNVs that were large (> 30kb, > =15 probes) and rare (frequency < 0.01 in input data and in the gnomAD v2.1 structural variant callset). OCD cases have a global excess of large CNVs (OR=1.12, p=1.77 × 10-3). This excess appears to be specific to CNVs overlapping at least one protein-coding base (OR=1.19, p=3.08 × 10-4), and is absent when CNVs not overlapping a coding base are considered (OR=1.04, p=0.50). Consistent with exome sequencing studies of OCD, the CNVs that appear to confer the highest amount of relative risk are deletions impacting loss of function intolerant genes (pLI>0.995, OR=4.12, p=2.54 × 10-5). In addition, cases are more likely to have a CNV impacting a neurodevelopmental disorder risk gene (Fu et al. 2022, n=664 genes, OR=2.54, p=1.91 × 10-5). Association tests of gene units clumped by overlapping test statistics with neighboring genes (988 tests total) were well-controlled (lambda=1.02) but did not identify any loci at a level of significance that survived multiple test correction. Our results indicate that rare CNVs contribute to OCD genetic risk, and as seen in prior exome sequencing studies of OCD, converge on genes that are depleted of deleterious variation in the general population. The locus-based association tests we constructed are insufficiently sized for gene discovery. To increase the statistical power of these association tests, we are incorporating new data from over 1000 Scandinavian OCD cases, along with additional unaffected controls.