0010 A Putative Biomarker Fingerprint of Insufficient Sleep Derived from the Human Plasma Metabolome

Abstract Introduction Developing an objective and easily assessed biomarker of insufficient sleep has great potential to enhance clinical assessment of overall sleep health, and improve our diagnostic capacity to identify individuals with poor sleep health. Such a biomarker would also contribute to personalized medicine with the potential of advancing clinical care. We analyzed the plasma metabolome in humans undergoing experimental sleep restriction to identify small molecule biomarkers associated with insufficient sleep. Methods We conducted a randomized cross-over 14–15 day in-laboratory study where 16 (8M/8F) healthy participants aged 22.4 ± 4.8y (mean±SD) completed three baseline days (9h scheduled sleep opportunity/night) followed by five days of insufficient (5h /night), and adequate (9h /night) sleep conditions. Blood was collected every 4h across 24h on the final day of baseline, insufficient, and adequate sleep conditions, and plasma was analyzed by untargeted liquid chromatography/mass-spectrometry. Results After filtering out metabolites not present in at least fifty percent of all samples, we detected 3,995 metabolites. Using variable importance scores derived from partial least squares discriminant modeling with 50 repeated cross-validations where one-third of all samples were randomly withheld, we identified a metabolite biomarker of insufficient sleep that consists of 34 metabolites forming a single biomarker fingerprint. We calculated a receiver operator curve (ROC) to assess the range of sensitivity and specificity of this biomarker fingerprint. The ROC area under the curve for insufficient sleep versus baseline is 0.856 (95% confidence interval 0.788–0.936), suggesting this 34 metabolite biomarker fingerprint has “good” performance as a biomarker of insufficient sleep. Conclusion Under controlled laboratory conditions, we identified a preliminary biomarker fingerprint of insufficient sleep consisting of 34 plasma metabolites measured in parallel. Such a biomarker fingerprint has the potential to provide a more accurate assessment of overall sleep health compared to self-report questionnaires. This preliminary biomarker fingerprint of insufficient sleep sets the stage for future studies to confirm and improve biomarker performance using larger sample sizes and at risk populations. Support (If Any) NIH R01HL085705, R01HL109706, R01HL132150, F32DK111161, and UL1TR000154; and Sleep Research Society Foundation 011-JP-16.