A sweet spot for the sleeping brain: Linking human sleep physiology and glucoregulation

Abstract

Vallat et al.¹ demonstrate a positive association between the coupling of slow oscillations and sleep spindles, neurophysiological markers of NREM sleep, and next-morning glucose homeostasis. Extended findings in an independent dataset raise intriguing questions about its directionality and consistency.

Vallat et al. demonstrate a positive association between the coupling of slow oscillations and sleep spindles, neurophysiological markers of NREM sleep, and next-morning glucose homeostasis. Extended findings in an independent dataset raise intriguing questions about its directionality and consistency.

Main text

Epidemiological and experimental studies suggest a beneficial role of nocturnal sleep in the maintenance of glucose homeostasis. According to recent meta-analyses, experimental sleep restriction decreases insulin sensitivity and glucose tolerance in individuals without type 2 diabetes or prediabetes.² In a seminal study in healthy participants,³ Tasali and colleagues focused on the relevance of slow-wave sleep, which is characterized by the intensified occurrence of slow oscillations that originate in the neocortex and display a peak frequency of 0.8 Hz.⁴ Slow-wave sleep is a key element of non-rapid eye movement (NREM) sleep. It turned out that three nights of slow-wave sleep suppression are sufficient to reduce glucose tolerance by 23%.³ However, not much more is known about the relevance of specific neurophysiological properties of human sleep for glycemic control. Vallat and colleagues¹ investigated the relationship between electroencephalographic hallmarks of NREM sleep and measures of glucose homeostasis obtained at 7 a.m. in the subsequent morning from approximately 650 participants of the Cleveland Family Study (CFS). They looked at the coupling between slow oscillations and sleep spindles, i.e., oscillations of 10–15 Hz that originate in the thalamus⁴; coupling was characterized in terms of both quantity (proportion) and the strength of temporal synchrony. Both measures of coupling were negatively associated with next-day fasting blood glucose concentrations, indicating that more pronounced and stronger coupling is associated with improved glucose homeostasis. These relationships remained significant in regression models including variables such as age, sex, race, BMI, hypertension, the sleep apnea-hypopnea index, and sleep duration—but not when diabetes status was included as a covariate. In further analyses, which included many conventional sleep metrics such as sleep duration, the duration of sleep stages, sleep efficiency, the density, frequency and amplitude of slow oscillations and spindles, as well as spectral band power of REM and NREM sleep, coupling quantity was found to be the most significant predictor of next-day fasting glucose levels. The authors point out that going from the 1^st to the 99^th percentile of coupling quantity and, respectively, strength corresponded to decreases in fasting glucose of 13.2 and 9.9 mg/dL.

In the 600 CFS participants without diabetes, the relationship between slow oscillation-spindle coupling and glycemic control was corroborated in analyses that included, instead of fasted glucose levels, glucose concentrations assessed 2 h after the intake of 75 g of glucose, i.e., during an oral glucose tolerance test. This pattern indicates that the observed relationship between NREM sleep neurophysiology and glucoregulation is valid both for fasted (basal) and for dynamic conditions. In terms of statistical significance, the coupling appeared to be more strongly related to insulin sensitivity (reflected by HOMA-IR values) than to pancreatic beta cell function (HOMA-B).

The authors relate their observations in humans to recent findings by Tingley and coworkers,⁵ who demonstrated in freely behaving rats that clusters of hippocampal sharp wave-ripples (SPW-Rs) predict decreases in interstitial glucose concentrations emerging around 10 min later. SPW-Rs comprise fast depolarizing waves and superimposed high-frequency oscillations (ripples) of 120–200 Hz in rodents and 80–140 Hz in humans.⁴ However, while hippocampal ripples have been found to co-occur with slow oscillations and spindles, only a small portion of ripples displays such tri-coupling.⁶ Moreover, the stark difference in the temporal dynamics revealed in the two studies underlines the urgent need for integrative mechanistic investigations. In this regard, further results presented by Vallat and colleagues might lend some clues. In an independent dataset collected from approximately 2,000 participants of the Multi-Ethnic Study of Atherosclerosis (MESA), the authors correlated fasting blood glucose, measured between 7:30 and 10:30 a.m., with sleep parameters derived from polysomnography performed at home. Once more, significant negative associations emerged between both quantity and strength of slow oscillation-spindle coupling and blood glucose concentrations, which also survived adjustment for the factors listed above. In this sample (but not in the CFS sample), heart rate variability significantly mediated these associations, suggesting a contribution of autonomic parasympathetic activity. In line with this assumption, parasympathetic activity is increased in slow-wave sleep.⁷ Also, the suppression of slow wave activity across three nights that impaired glucose homeostasis in the study by Tasali et al. shifted cardiac sympathovagal balance toward higher sympathetic activity.³ Building on these converging results, future investigations may broaden their scope to address the intricate complexity of efferent neurocircuits and hormones that control blood glucose concentrations.

Can we conclude that NREM sleep neurophysiology might be a “prognostic sleep signature”¹ or even a determinant of glucose homeostasis in humans? Two important aspects should be considered. First, glycemic control is not only compromised when sleep is impaired or curtailed but also as an effect of circadian misalignment of sleep,⁸ challenging the idea of a primary contribution to glycemic control of NREM sleep physiology in contrast to the overall integrity of sleep/wake rhythmicity. Second, an important implication of the study by Vallat and coworkers leads to questions about the directionality and consistency of the reported associations. In the participants of the MESA study, i.e., the second cohort under investigation, sleep was recorded, on average, 340 days after the collection of blood samples for the determination of glucose concentrations. In contrast, in the participants of the original study (CFS), blood sampling took place in the morning after polysomnography. It might therefore be argued that rather than slow oscillation-spindle coupling influencing glucose levels, long-term glucose homeostasis might foster the stability of NREM sleep physiology. While the amount of slow-wave sleep is a rather stable individual trait and displays a strong heritable component,⁹ it has been shown to be reduced by high-carbohydrate diets, and participants with poor glycemic control spend less time in slow-wave sleep.¹⁰ These findings demonstrate that glucose fluxes can affect nocturnal electrophysiology—for better or for worse. Therefore, the insights provided in the paper by Vallat and colleagues, and the questions it raises, might keep sleep researchers awake for many a night to come.