Analysis of novel sleep variable highlights shared genetics of sleep and psychiatric disorders

Insomnia is prevalent worldwide, and its associations with negative health outcomes make it a prominent public health concern [1]. A large body of literature has highlighted the bidirectional relationship between sleep disturbances, including insomnia, and psychiatric disorders, an area with significant clinical impact [2, 3]. Disturbed sleep is part of the diagnostic criteria for a range of psychiatric disorders, including depression, post-traumatic stress disorder, and bipolar disorder [4]. It frequently accompanies substance use disorders as well, as sleep is directly impacted by use of substances like alcohol and cannabis [5]. Longitudinal relationships have also been demonstrated; an updated meta-analysis replicates insomnia as a predictor of depression [6]. Thus, a better understanding of the etiology of sleep problems is essential and has the potential to help a large number of patients, including those with psychiatric disorders [2, 3].

Both genetic and environmental factors influence poor sleep, with the early literature utilizing known relationships among twins and other family members to calculate heritability [7, 8]. A recent meta-analysis of twin studies estimates the heritability of insomnia at 40% (moderate), in line with related psychiatric traits like depression and anxiety [9]. Twin studies have also demonstrated substantial genetic overlap between insomnia and psychopathology (e.g. [10]). The sequencing of the human genome and increasing accessibility of molecular genetic data have led to advances in genetic methodology over the past few decades [11–13]. We can now identify specific genes that contribute to a phenotype using data from unrelated individuals, in contrast to aggregate measures of heritability estimated from traditional twin and family studies [7, 8, 12].

The field has since moved from a candidate gene approach—association studies where specific genetic variants were chosen a priori, often based on biological plausibility, to hypothesis-free genome-wide association studies (GWAS) which examine millions of genetic variants across the entire genome, or available set of genetic markers, to identify associations. This requires large datasets of unrelated individuals in order to have sufficient power, and identified genes may not correlate with function [7, 8, 11–13]. Current methods now include a number of post-association analyses, many of which are based on the best of our knowledge of the genetic architecture of complex traits, which is that they are polygenic, in that many individual genes with a small effect size have a larger effect when combined. Functional analysis of identified variants is possible as well [11, 13]. There are even methods that leverage genetics to identify causal relationships, such as Mendelian randomization (MR) [14]. The reader is encouraged to refer to recent reviews for a more thorough overview of genetic methodology [7, 8, 11–13].

The past decade has also seen an increase in published research related to the genetics of sleep [8, 15]. Genome-wide association studies utilizing large datasets, such as the UK Biobank cohort [16] and 23andMe [17] have identified genetic variants contributing to insomnia. Researchers have also conducted post-GWAS analyses using large datasets to examine shared etiology between sleep phenotypes and other medical and psychiatric disorders, frequently identifying overlap [18–22]. However, the current literature is limited by the populations used for analysis, heterogeneous phenotypes, and small effect sizes of identified variants. Furthermore, the current ability to utilize genetic data for prevention and precision treatment is limited. In moving forward, Madrid-Valero and Barclay identify the need for additional large-scale GWAS, as well as the importance of incorporating both genes and the environment for prediction [10].

In this issue of SLEEP, Broberg et al. [23] utilize a large population-based registry from Finland to conduct GWAS of a novel sleep phenotype, sleep medication purchases (i.e. prescriptions for non-benzodiazepine Z-drugs). This appears to be a proxy for insomnia, as insomnia is the primary indication for these medications in Finland. The authors present GWAS results as well as results from an array of post-association analyses, spanning from polygenic risk score prediction to MR examining causality. They are able to replicate several previously identified variants from other GWAS of sleep and psychiatric disorders, including MEIS1, PAX2, and CACNA1C. The authors also present gene expression results for top variants using expression quantitative trait loci analysis. Association results are robust across different variations of the sleep medication phenotype (quantitative trait and binary traits). Eight of the twenty-seven identified genes were replicated within the UK Biobank cohort, including previously identified sleep genes, despite differences in the baseline prevalence of sleep medication use within the populations.

Post-GWAS analyses add support to the growing literature that sleep and psychiatric disorders are related on a genetic level, with results consistent across different methodologies. Significant genetic correlations were found between the sleep medication phenotype and psychiatric disorders, with correlations with internalizing disorders (e.g. depression and anxiety) having the highest magnitudes (see Table 2 of [23] for details). These relationships were then replicated within polygenic risk score analyses, where a risk score created for insomnia from variants identified in the GWAS results predicted other psychiatric phenotypes. Finally, MR analyses indicated a causal link between sleep medication use and experiencing a psychiatric disorder, further supporting this relationship. Results are in line with (and add to) the extant literature, where others have identified genetic overlap and/or casual relationships between sleep phenotypes and other psychiatric disorders [18–22].

In summary, this manuscript by Broberg et al. [23] adds to the current body of literature investigating the genetics of sleep disturbances, demonstrating that identified variants and genetic overlap with psychiatric disorders are robust to different phenotypes and populations. The authors identify several specific genes that are promising targets for sleep, given their reappearance across studies. Results further highlight the importance of the relationship between sleep and psychiatric disorders, providing evidence through genetic correlation, risk prediction, and a causal link. Limitations to consider are those common to large-scale GWAS studies, including limited generalizability and phenotype definition. Future research involving more diverse populations and well-defined phenotypes will be essential moving forward [12].

As genetic methods continue to advance and we better understand the genetics of insomnia and other sleep disorders, the potential for results to be used in a clinically significant way for risk stratification, prevention, and treatment decisions increases. Polygenic risk scores, ideally combined with environmental risk factors and considering the possibility of gene-environment interactions (GxE) (i.e. that different genotypes may influence response to the environment in different ways, thus differentially affecting the risk of an outcome such as psychiatric diagnosis) represent a potential approach [8, 11–13]. The environment plays a clear role in sleep, with many, though not all, individuals developing insomnia following a significant stressor [24] and should be considered in future research. There are also types of genetic variation beyond the typical common variants used in GWAS, as well as rare variants, which may also shed more light on genetic pathways [8, 11–13, 15]. Some researchers suggest that a focus on the functional effects of variants and well-defined phenotypes is especially important for clinical applications [25]. Ultimately, a better understanding of the genetics of sleep has the potential to inform our treatment of sleep disorders, as well as the development, maintenance, and treatment of comorbid psychiatric illness. Longitudinal research implicates sleep as a potential area of intervention for psychiatric disorders, and more research is needed across a wide variety of disciplines, including genetics, to optimize the treatment of the many individuals who struggle with poor sleep and/or comorbid psychiatric diagnoses [8, 15].

Disclosure Statement

Financial Disclosure: The author has no financial disclosures. Nonfinancial Disclosure: The author has no potential conflicts of interest to declare in relation to this work. This content is solely the responsibility of the author and does not necessarily represent the official views of the Department of Veterans Affairs, the U.S. government, or the University of Washington.