32374
Predicting Response to Sleep Aids Using Genetic and Comorbidity Data
Individuals with autism spectrum disorder (ASD) are significantly more likely to endorse challenges with sleep than their peers. In the setting of ASD, sleep has been found to modulate numerous other comorbidities, such as causing flares in aggressivity and anxiety. The symptom-burden of individuals with ASD and the challenges for caretakers can increase significantly if relief from sleep challenges is not found. Despite our increasing understanding of medical factors influencing sleep in ASD, a recent analysis was could only explain~19% of the variance in sleep dysfunction using environmental and medical comorbidity factors. Additionally, while many studies incorporate drug-usage features, to our knowledge, no study has modeled risk factors influencing sleep-associated drug response.
Objectives:
To begin to address these shortcomings, we (1) estimated the heritability of sleep dysfunction in ASD, and (2) performed a comorbidity analysis of sleep-associated drug-response.
Methods:
The foundation of this proposal is an unpublished dataset we generated by re-contacting 5,686 families enrolled in the SPARK genetic study of ASD. SPARK is the largest study of its kind, with whole exome sequencing, genotypes, and medical comorbidity data recently released (November 2018) on over 27,000 individuals. Our unpublished dataset expands the SPARK medical comorbidity data to include quantitative measure of sleep dysfunction (CSHQ; children’s sleep habits questionnaire) and questions regarding drug-response to five major classes of drugs used to treat sleep dysfunction in ASD (melatonin, antihistamines, non-benzodiazepines, antidepressants, PM-variant over the counter (OTC) medicines).
Results:
Within our dataset, ~50% of individuals reported sleeping issues, with a further 14% not finding relief of their sleep symptoms through any medicines they tried. Interestingly, we found that melatonin and antidepressant sleep-associated drug response could be significantly modeled using medical comorbidity data (P < 5.6E-06 & 4.4E-02, respectively). These outcomes were influenced by a range of comorbid diagnoses, such as schizophrenia, intellectual disability, and blindness. Through modeling sleep dysfunction, we found a significant SNP-based heritability for the CSHQ score in our cohort (h2 = 0.26; S.E. = 0.15), suggesting that polygenic, common variant risk influences sleep dysfunction in ASD.
Conclusions:
From our analysis of the largest genetic sleep cohort of ASD to date, we find significant evidence for the SNP-based heritability of sleep traits in ASD. Further, by modeling the success of leading therapeutic options for sleep dysfunction in ASD, we gain valuable insights into the ideal medical profile of individuals taking either melatonin or antidepressants to alleviate their sleeping issues. Going forward, this dataset will be used to model the polygenic risk profiles of traits influencing sleep dysfunction and drug response, bringing a novel genetic-based approach to comorbidity analyses. These findings advance our understanding of the etiology of sleep dysfunction in ASD, and provide an early glimpse of factors which could be leveraged in a personalized-medicine treatment strategy for sleep dysfunction.