Investigating Causal Relations Between Sleep-Related Traits and Risk of Type 2 Diabetes Mellitus: A Mendelian Randomization Study

OBJECTIVE: Extensive literature put forward the link between sleep and type 2 diabetes mellitus (T2DM), however, little is known about the underlying causality of the associations. Here we aim to assess the causal relationships between five major sleep-related traits and T2DM. DESIGN, SETTING, AND PARTICIPANTS: Two-sample Mendelian randomization (MR) was utilized to investigate the potential causal relations. Independent genetic variants associated with five sleep-related phenotypes—insomnia, sleep duration, short sleep duration, long sleep duration, and morningness—were chosen as instrumental variables to estimate the causal associations with T2DM. Summary statistics were acquired from the genome-wide association studies of UK Biobank and 23andMe (for sleep-related measures), the DIAbetes Genetics Replication And Meta-analysis and the FinnGen (for T2DM). MAIN METHODS: Individual Cochran’s Q statistic was applied to remove the pleiotropic instruments, global Q statistics and MR-Egger regression were adopted to test for the global heterogeneity and horizontal pleiotropy of the screened instruments, respectively. Two T2DM cohorts were selected to analyze their associations with sleep traits. A modified inverse variance weighted (IVW) estimate was performed to combine the ratio estimators from each instrument and acquire the causal estimate, alternative methods including IVW with first-order weights, simple and weighted median estimations, and MR-Egger regression were conducted as sensitivity analyses, to ensure the robustness and solidity of the findings. RESULTS: Two-sample MR supported findings for an adverse effect of genetically predicted insomnia on T2DM risk (odds ratio [OR] = 1.14, 95% confidence interval [CI]: 1.09–1.19, p = 1.29E–08) at the Bonferroni-adjusted level of significance (p < 0.005). We further investigated the causal role of T2DM on insomnia but obtained a non-significant estimation. There was also little evidence for the causal effect of other sleep-related measures on T2DM. Results were largely consistent when leveraging two different T2DM cohorts, and were robust among various sensitivity analyses. CONCLUSION: Findings provide significant evidence for an adverse effect of insomnia on T2DM risk. The study extends fundamental knowledge to further understanding of the pathophysiological mechanisms of T2DM, and points out the non-negligible role of insomnia on epidemiologic intervention and clinical therapeutics of T2DM.