Abstract

Obstructive sleep apnea (OSA) is defined by frequent episodes of reduced or complete cessation of airflow during sleep and is linked to negative health outcomes. Understanding the genetic factors influencing expression of OSA may lead to new treatment strategies. Electronic health records can be leveraged to both validate previously reported OSA-associated genomic variation and detect novel relationships between these variants and comorbidities. We identified candidate single nucleotide polymorphisms (SNPs) via systematic literature review of existing research. Using datasets available at Geisinger (n=39,407) and Vanderbilt University Medical Center (n=24,084), we evaluated associations between 48 SNPs and OSA diagnosis, defined using clinical codes. We also evaluated associations between these SNPs and OSA severity measures obtained from sleep reports at Geisinger (n=6,571). Finally, we used a phenome-wide approach to perform discovery and replication analyses testing associations between OSA candidate SNPs and other clinical codes and laboratory values. Ten SNPs were associated with OSA diagnosis in at least one dataset, and one additional SNP was associated following meta-analysis across all datasets. Three other SNPs were solely associated in subgroups defined by established risk factors (i.e., age, sex, and BMI). Five OSA diagnosis-associated SNPs, and 16 additional SNPs, were associated with OSA severity measures. SNPs associated with OSA diagnosis were also associated with codes reflecting cardiovascular disease, diabetes, celiac disease, peripheral nerve disorders and genitourinary symptoms. Results highlight robust OSA-associated SNPs, and provide evidence of convergent mechanisms influencing risk for co-occurring conditions. This knowledge can lead to more personalized treatments for OSA and related comorbidities.