454Relationships between sleep traits and metabolic profiles: evidence from multivariable regression and Mendelian randomization analyses

Background Sleep traits are associated with cardiometabolic disease. The aim of this study was to explore the causal effect of sleep traits (duration and insomnia) on multiple metabolic traits. Methods We used age, sex and BMI adjusted multivariable regression (N = 17,370) and two-sample summary statistic Mendelian randomization (MR) to examine effects of sleep duration and insomnia symptoms on ∼150 NMR metabolites. Multivariable analyses were conducted on data from nine European cohorts and meta-analysed. MR analyses utilised summary statistics from published genome-wide association studies (GWAS) of self-reported sleep traits (sample 1; N = 446,118 to 1,331,010) and from GWAS on NMR serum metabolites (sample 2; N = 38,618). We used inverse-variance weighted (IVW) for the main MR analyses and weighed median (WM) and MR-Egger to explore bias due to pleiotropy. Results MR IVW and multivariable analyses both suggest a positive effect of insomnia symptoms on glycoprotein acetyls (MR: 0.06 s.d. increase in mean concentration comparing any symptoms to none; p = 5.9e-4) and between total sleep duration and creatinine (MR: 0.16 s.d. increase per additional hour; p = 0.03). WM and MR-Egger analyses show consistent results. There was evidence for thirteen and eight effects of insomnia and duration in multivariable only and three and one, respectively, in MR only. Conclusions Insomnia symptoms lead to higher levels of an inflammatory marker (glycoprotein acetyls) and longer sleep duration leads to higher creatinine levels. Key messages We found no evidence of widespread metabolic disruption by sleep traits.