Clustering bipolar disorder risk variants by their effect on sleep and circadian traits

Abstract Introduction Bipolar disorder (BD) is a mood disorder that causes noticeable, sometimes extreme, changes in mood and behaviour. BD affects nearly 50 million individuals worldwide including 2-4% of the US population. Previous studies have reported sleep and circadian disturbances in patients with BD. Genetic studies have identified 63 risk variants for bipolar disorder (BD). Methods To enable new insights into BD disease causing pathways, we clustered BD risk variants by their effect on sleep and circadian rhythm traits. A Bayesian soft-clustering approach was used to determine if BD variants cluster according to effect on eight subjectively and two objectively measured sleep and circadian traits (using sleep and circadian trait GWAS from the UK Biobank in European ancestry, n = 453,000). Results We found evidence of genetic correlation between BD and sleep/circadian traits (p< 0.05). Bipolar risk variants cluster into two clusters based on their association with sleep and circadian traits. The two clusters are: Cluster 1: late chronotype, shorter time in bed (by actigraphy), difficulty waking up, longer sleep duration (self-report), longer periods of daytime sleepiness, longer and frequent naps and increasing insomnia and Cluster 2: early chronotype, longer time in bed (by actigraphy), longer sleep duration (self-report) and less insomnia. In general, cluster 1 appears to be contributing to sleep insufficiency, while cluster 2 appears to be contributing to hypersomnia or more “typical” sleep. Conclusion In conclusion, sleep and circadian trait clusters indicate bipolar genetic risk may function through two distinct mechanisms. Some genetic risk variants appear to be related to sleep insufficiency, while others are related to hypersomnia, suggesting that it may be possible to genetically subtype BD. This finding may allow classification of BD-patients by distinct genetic pathways linking to their sleep-circadian disturbances, potentially offering a way forward toward genetically informed diagnosis, surveillance and management of BD-patients. Support (if any) N/A