Psychiatric comorbidities in epilepsy: population co-occurrence, genetic correlations and causal effects

Abstract

Background: Psychiatric comorbidities are common in patients with epilepsy. Reasons for the co-occurrence of psychiatric conditions and epilepsy remain poorly understood.

Aim: We aimed to triangulate the relationship between epilepsy and psychiatric conditions to determine the extent and possible origins of these conditions.

Methods: Using nationwide Swedish health registries, we quantified the lifetime prevalence of psychiatric disorders in patients with epilepsy. We then used summary data from genome-wide association studies to investigate whether the identified observational associations could be attributed to a shared underlying genetic aetiology using cross-trait linkage disequilibrium score regression. Finally, we assessed the potential bidirectional relationships using two-sample Mendelian randomisation.

Results: In a cohort of 7 628 495 individuals, we found that almost half of the 94 435 individuals diagnosed with epilepsy were also diagnosed with a psychiatric condition in their lifetime (adjusted lifetime prevalence, 44.09%; 95% confidence interval (CI) 43.78% to 44.39%). We found evidence for a genetic correlation between epilepsy and some neurodevelopmental and psychiatric conditions. For example, we observed a genetic correlation between epilepsy and attention-deficit/hyperactivity disorder (r_g=0.18, 95% CI 0.09 to 0.27, p<0.001)-a correlation that was more pronounced in focal epilepsy (r_g=0.23, 95% CI 0.09 to 0.36, p<0.001). Findings from Mendelian randomisation using common genetic variants did not support bidirectional effects between epilepsy and neurodevelopmental or psychiatric conditions.

Conclusions: Psychiatric comorbidities are common in patients with epilepsy. Genetic correlations may partially explain some comorbidities; however, there is little evidence of a bidirectional relationship between the genetic liability of epilepsy and psychiatric conditions. These findings highlight the need to understand the role of environmental factors or rare genetic variations in the origins of psychiatric comorbidities in epilepsy.

Keywords: neuropsychiatry; psychiatry.

Figure 1

Flowchart depicting the derivation of the analytical sample from the Swedish registry data. No individuals were excluded from the analysis. Those missing disposable income were imputed, as they were either too young to have accumulated any disposable income, or this information was missing. Further details are available in the text.

Figure 2

The population-wide co-occurrence of epilepsy and psychiatric morbidities. (A) The adjusted (marginal) lifetime prevalence of psychiatric diagnoses among those with (n=94 435) and without an epilepsy diagnosis (n=7 534 060). (B) The adjusted (conditional) OR of psychiatric diagnoses comparing those with and without an epilepsy diagnosis. All analyses are adjusted for birth year, sex, highest achieved disposable income and immigration status. Both panels are sorted according to the adjusted lifetime prevalence among those with an epilepsy diagnosis. Black lines indicate 95% CIs in both panels (CIs are plotted in (B) but they are smaller than the marker size, reflecting our large sample size). CI, confidence interval; OR, odds ratio.

Figure 3

The genetic correlation between epilepsy (all epilepsies, generalised and focal) and common psychiatric comorbidities of epilepsy. Estimates represent the genetic correlation obtained from cross-trait linkage disequilibrium score regression. Asterisks reflect a p value lower than the Bonferroni-corrected threshold (α=0.001515).

Figure 4

The bidirectional relationship between genetic liability to epilepsy (all epilepsies, generalised and focal) and genetic liability to common psychiatric comorbidities of epilepsy. The estimates are ORs obtained from an inverse-variance weighted regression, which assumes that all variants are valid instruments. Findings for intelligence are presented in-text, as intelligence is measured on a continuous scale and therefore not scale-comparable to the other traits. Both x-axes are plotted on the natural logarithmic scale. CI, confidence interval; OR, odds ratio.