Polygenic risk score analysis reveals shared genetic burden between epilepsy and psychiatric comorbidities

Background One in three people with epilepsy experiences psychiatric comorbidity, with higher rates in people with drug-resistant epilepsy. Despite their high heritabilities, finding genetic links between epilepsy and psychiatric disorders has proven difficult. We used polygenic risk scoring (PRS) to test whether people with epilepsy have an increased polygenic burden of common genetic variants for depression, anxiety, psychosis, and attention deficit/hyperactivity disorder (ADHD), and examined whether such polygenic burden influences the response to pharmacological treatment of epilepsy.

Methods Phenotype data in the UK Biobank were assessed to identify people with 1) epilepsy (n=8 488), 2) depression (n=143 440), 3) psychosis (n=2 357), 4) ADHD (n=89), and 5) anxiety (n=18 222. Using genotype data and restricting to Caucasian-ancestry samples (n=409 634), PRS for each psychiatric trait were calculated and multinomial regression was used to compare 1) population controls, 2) people with epilepsy and no psychiatric illness, 3) people with epilepsy and the psychiatric trait of interest, and 4) people with the psychiatric trait of interest and no epilepsy. Fixed-effect meta-analysis was used to compare psychiatric PRS in drug-resistant and drug-responsive epilepsy samples from the UK Biobank (n=1 640) and the EpiPGX consortium (n=3 449).

Results After correction for multiple testing, people with epilepsy showed elevated PRS for depression (p<2 x10−16), psychosis (p=0.04) and ADHD (p<0.001). Patients with drug-resistant epilepsy had an increased PRS for psychosis (p=0.002) and depression (p=0.0004) relative to responsive cases.

Conclusion We present evidence that the common genetic basis of epilepsy overlaps with that of psychiatric conditions which are frequently comorbid in people with epilepsy. Common genetic variants that drive psychiatric illness are enriched in people with drug-resistant epilepsy. These results further our understanding of the genetic architecture of epilepsy and suggest a potential future role for polygenic interpretation of common variants in prognostic stratification, both for seizure-treatment outcomes and non-seizure comorbidities.