Abstracts

Rationale: The UK Biobank is a collection of half a million adults aged 40-69 recruited between 2006-2010. Current and retrospective self-reported measures of lifestyle and health were prospectively collected alongside physical and morbidity-related data. The sample were SNP genotyped on two Affymetrix arrays with 95% content overlap. We have previously described the bioinformatic pipeline for reliably identifying copy number variation (CNVs) from these data. Methods: We identified a cohort of people with epilepsy, filtered for comorbid conditions likely to predict an acquired epilepsy such as brain tumours and dementia. Diagnoses were self-declared or came through hospital discharge diagnoses and death certificates. Controls were people without epilepsy, who were also not taking an anti-epilepsy drug. Analyses were restricted to individuals of white British or Irish self-reported ethnicity and stringent QC checks were used during CNV calling. We performed logistic regression with age, sex and array type as covariates. We describe the presence and frequency of 49 rare pathogenic CNVs in our cohort. Results: We identified 169 pathogenic CNVs in 3,291 cases with epilepsy (5.14%) and 16,075 in 416,185 controls (3.86%). Pathogenic CNVs were not evenly distributed. 15q13.3del (RR 9.68, p=0.001), the TAR duplication at 1q21.1 (RR 3.87, p<0.0001) and 3q29del (RR 43, p<0.0001) were seen more commonly in people with epilepsy.  A further seven CNVs showed evidence for enrichment in cases compared to controls.  In keeping with the lack of childhood onset neurodevelopmental syndromes in the UK Biobank we had no cases with a 22q11.2 deletion (9 seen in controls) but 6 cases had 22q11.2 duplications (RR 2.77 P=0.018). 35 of the pathogenic CNVs were seen at least once in cases and having a qualifying CNV increased the chance of having epilepsy 1.7 fold (p<0.0001). Covarying analyses for age, sex and array type did not substantially change the results.   Conclusions: We report the frequency of rare pathogenic CNVs in a large community sample of adults with epilepsy. The evidence provided for the increased frequency of 3q29 del in people with epilepsy is novel; in the 3q29del patient registry less than 5% of individuals report seizures. The phenotype of 3q29 microdeletion syndrome is pleomorphic but intellectual disability and facial dysmorphism are core features– in DECIPHER only one of 78 reported cases have seizures.  We do not know the proportion of the cohort with a genetic generalised epilepsy, nor the age of onset.  We also do not know whether these CNVs were de novo or inherited. Our enriched CNVs are all associated with neurodevelopmental disorders, but the 1q21.1dup is more commonly associated with autism and intellectual disability than with epilepsy. Harbouring a rare pathogenic CNV is a strong risk factor for developing epilepsy but is only one of many aetiological factors and many people with a pathogenic CNV do not have seizures. Funding: This work was not directly funded.