0034 Genetic Variants for Obstructive Sleep Apnea Identified after Modeling Interactions with Daytime Sleepiness

Abstract

Abstract Introduction Obstructive sleep apnea (OSA) is a complex sleep disorder that presents with a range of symptoms, including excessive daytime sleepiness (EDS), postulated to represent a more severe subtype. Despite significant heritability and past GWASs, genetic mechanisms of the disorder remain unclear. Given that EDS often reflects chronic sleep insufficiency and a pro-inflammatory state, we reasoned it may interact with genetic variants to modify risk for OSA. We tested this hypothesis by investigating gene by sleepiness interaction analysis of OSA. Methods Whole-genome sequencing data of 11,619 individuals of diverse race and ethnic backgrounds from the NHLBI Trans-Omics for Precision Medicine program was used, with the outcome being Apnea-Hypopnea Index (AHI) and the exposure Epworth Sleepiness Scale (>10: EDS, <=10: non-EDS). GENESIS R package was used to perform 2-stage rank normalization, adjusting for age, sex, BMI, ancestral PCs, cohort, and genetic relatedness, and rescaling residuals to a unit variance, separately in each study/ancestry group. GEM software was used to perform gene-EDS interaction analyses, including 1 df tests of marginal SNP effect and interaction effect of SNPxEDS on AHI, respectively, and 2 df test of joint significance of both. Robust standard errors were used, with a significance threshold of p< 5x10-8. Results The 1 df test of marginal SNP effect identified one novel locus (rs35370454) that mapped to IFRD1, whose overexpression has been associated with sleep restriction, and hypomethylation with narcolepsy. The 1 df test of interaction effect identified two novel loci (rs13118183, rs281851) that mapped to genes MARCHF1 and CCDC3. The 2 df joint test did not reveal additional novel loci. CCDC3 has shown to repress inflammatory activity of TNF-alpha, a cytokine postulated to be involved in sleep regulation. MARCHF1 has shown to inhibit cellular insulin sensitivity as well as promote NF-KB, with degree of activation positively correlating with OSA severity. Conclusion By modeling the modification effect of EDS on OSA we identified 3 novel loci. This approach may be useful in understanding the genetic factors that vary across OSA subtypes. For next steps, we seek to conduct replication analysis using independent imputed samples. Support (if any) NHLBI, R01HL153814 (H.W.), R35HL135818 (S.R.).