Circulating metabolites associated with objectively measured sleep duration and sleep variability in overweight/obese...

To investigate the associations of circulating metabolites with sleep duration and sleep variability. We also assessed the ability of metabolites to discriminate between sleep duration and sleep variability categories.

Cross-sectional analyses were performed on baseline data from 205 participants with overweight/obesity in the "Satiety Innovation" (SATIN) study. A targeted metabolite profiling (n = 159 metabolites) approach was applied using three different platforms (nuclear magnetic resonance, liquid chromatography coupled to mass spectrometry, and gas chromatography coupled to mass spectrometry). Associations between circulating metabolite concentrations and accelerometer-derived sleep duration and variability in sleep duration were assessed using elastic-net regression analysis. Ten-fold cross-validation was used to estimate the discriminative accuracy of metabolites for sleep duration and sleep variability categories.

A metabolite profile, including acyl-carnitines (C11:0/C5:1-DC/iso-C11:0, 2-M-C4:1/3-M-C4:1, C4:0), sphingomyelins (42:1, 33:1), glycerol, stearic acid, 2- and 3- hydroxyl-butyric acid, docosahexaenoic acid, serotonin, and phosphatidylcholine (34:2), was significantly associated with high sleep duration (4th plus 5th quintile). Ten metabolites, including acyl-carnitines (C18:1, C7:0, C6-OH), phosphatidylcholine (40:6, 37:4, 42:5), lyso-phosphatidylcholine (20:1), sucrose, glutamic acid, and triacylglycerol (52:4), were significantly associated with high sleep variability (4th plus 5th quintile). The area under the curve was 0.69 (95% CI: 0.62-0.75) and 0.63 (95% CI: 0.53-0.72) in the multimetabolite score for high sleep duration and sleep variability, respectively. The variance in sleep duration explained by metabolites was 7%. No metabolites were selected for prediction of sleep variability (continuous).

A small set of metabolites within distinct biochemical pathways were associated with high sleep duration and sleep variability. These metabolites appeared to moderately discriminate sleep duration and sleep variability categories.