Body mass index is associated with subjective workload and REM sleep timing in young healthy adults

1 Introduction

Sleep is a vital restorative body function that is particularly essential for the functioning of the brain, immune system, musculoskeletal system, and endocrine system [1]. While sleeping, the body is in a predominantly anabolic state, which contributes significantly to the homeostasis of vital body functions [1]. Important hormones are released during sleep, including growth hormone, which not only promotes anabolic metabolism but also mental and emotional well-being [2]. Sleep occurs in repeating cycles in which rapid eye movement (REM) and various non-REM sleep stages alternate. REM sleep appears to particularly contribute to sleep quality, cognitive performance, and mood [3,4]. Internal oscillators consist of complex neural networks, control the timing, duration, and composition of sleep [5,6]. Individuals differ not only in their preferred sleep duration but also in their preferred bedtime which is controlled by the circadian clock [7]. The individual chronotype can be defined by the midpoint of sleep on work-free days [7].

However, in modern society, the morning alarm clock largely determines both the duration and the midpoint of sleep on workdays, resulting in suboptimal sleep duration and poor sleep quality [8]. Especially in late chronotypes, early schedules lead to circadian misalignment due to an advance shift in the midpoint of sleep which is called “social jet lag,” (SJL) and is associated with sleep loss [9]. Shift work leads to a particularly high variability in sleep timing and duration and thus to a strong circadian misalignment [10].

There is increasing evidence that poor sleep as a result of circadian misalignment and sleep loss contribute to the so-called “diseases of civilization,” such as chronic stress, obesity, metabolic disorders, cardiovascular diseases, and immune dysfunction [10]. Various data show an association between very short sleep duration (less than 6 h per day) and higher body mass index (BMI) in the obesity range [11–14] especially in women [12,14]. On the other hand, higher BMI in the obesity range seems to be associated with poor sleep quality [13,15,16] indicating a reciprocal relationship between poor sleep and obesity. In addition, overweight and obese individuals also show greater variability in self-reported sleep duration [17], indicating lower rhythm robustness. On the other hand, there is a complex relationship between sleep and mental and physical health and substance use disorders [18]. In older adults, metabolites are associated with sleep architecture and brain function and possibly also with circadian dysregulation [19]. However, there is hardly any research on the connection between BMI, sleep, psychological stress, and substance use in young individuals in the normal weight range.

Wrist-worn wearables are able to record sleep parameters over long periods of time under real-life conditions, although they still have lower accuracy than polysomnography [20]. In the ChronoSleep study, we used real-life longitudinal Fitbit sleep data in combination with a questionnaire in a small and relatively homogeneous sample of young, healthy adults, first-year medical students. In a previous study, we could show that shorter sleep duration on workdays was associated with a higher subjective workload and higher subjective impact of a high workload on sleep, which, in turn, were associated with higher levels of anxiety and depression [21]. In addition, we identified promising digital markers for objective sleep quality such as REM sleep proportion and consolidation which are associated with the intraindividual phase relationship between total sleep and REM sleep (ICC50REM) [22]. In the present study, we aim to investigate whether there are associations between BMI as a marker for metabolic state and ICC50 REM as a marker for internal circadian synchronization in a cohort of lean, healthy young adults. In addition, we aim to investigate whether psychological stress, sleep times, and lifestyle factors, such as physical activity and substance use, interact with these markers.

2 Materials and methods

3 Results

3.2 Correlation-based network analysis

Relationships between the variables were analyzed by Spearman correlations (Figure 1a) and illustrated in a correlation-based network analysis (Figure 1b). Biological sex was not associated with any other variable (Figure 1a) and thus not included in the network analysis. The only association of age and caffeine consumption was with each other (not shown).

Figure 1

Relationship of BMI with sleep variables and life style factors. (a) Plot of Spearman correlation coefficients for significant (p < 0.05) correlations. The color code and size of circles represent the value of correlation coefficients. (b) Network analysis of the most relevant variables based on significant (p < 0.05) Spearman correlations. Variables that are more highly correlated appear closer together and are joined by stronger paths. Positive and negative correlations are indicated in blue and red paths, respectively. BMI, body mass index; ICC50REM, measure for phase relationship between REM sleep and total sleep; MS, midpoint of sleep on workdays; MS_std, intraindividual variability of MS; SD, sleep duration on workdays; SD_std, intraindividual variability of SD; SJL, social jet lag.

The BMI was negatively correlated with subjective workload (Figure S1), suggesting a relationship between metabolic homeostasis and resilience to psychological stress in our cohort. Moreover, the BMI was positively correlated with the intraindividual variability in midpoint of sleep (MS_std) and sleep duration (SD_std) and the intraindividual strength in the phase relationship between REM sleep and total sleep (ICC50REM) (Figure S1), which are interconnected and connected with the midpoint of sleep. This suggests a relationship between metabolic homeostasis with variability in sleep timing and robustness of REM sleep timing.

MS was positively correlated with nicotine use (Figure 1), which is positively correlated with alcohol consumption and negatively correlated with sports. Alcohol consumption was also associated with SJL. BMI was not correlated with any of the life style factors.

4 Discussion

Many epidemiological surveys in children and adults have investigated the association between body composition with sleep in the context of sleep deprivation and obesity [11–16,28]. In contrast, this study examines associations between BMI, sleep timing, psychological stress, and lifestyle factors in young, lean adults without sleep deprivation

In our cohort, the mean BMI is 22.03 kg/m² which corresponds to normal weight and it is normally distributed within the sample. None of the subjects were obese. In contrast to cross-sectional studies including overweight and obese subjects [29] physical activity is not related to BMI in our cohort. The mean sleep duration on workdays in our cohort is 7.88 h which is within the range of 7–9 h of sleep per night for healthy adults recommended by the American Academy of Sleep Medicine and Sleep Research Society [30] and it is normally distributed within the sample. None of the subjects had a mean sleep duration of less than 6 h. There is no association between sleep duration and BMI in our cohort, indicating that this association occurs primarily under non-physiological conditions [11–16,28].

However, using the same dataset, we previously reported an association between shorter sleep duration with higher subjective workload [21] and with a stronger temporal alignment between REM sleep and total sleep, which is associated with good sleep quality [24]. Although this seems contradictory, it is consistent with a reciprocal relationship between sleep and psychosocial stress and a balance between sleep quantity and sleep quality. Biological sex, age, and lifestyle factors, such as physical activity or the use of stimulants are not directly associated with BMI, subjective workload, or the PHQ-4 score. Alcohol consumption is associated with SJL, while later midpoint of sleep is associated with nicotine consumption. This is consistent with earlier studies on the effect of chronotype on the substance consumption [9,31]. However, there is an association of BMI with the intraindividual variability in sleep timing and with ICC50REM, a marker for the temporal alignment between REM sleep and total sleep. Although we cannot establish causality, we assume that if sleep timing can be adjusted more flexibly according to an individual’s specific needs, this has a positive effect on the phase adjustment of REM sleep and promotes anabolic metabolic processes during sleep. During sleep, numerous neurotransmitters, metabolites, and hormones are released that contribute to the regulation of metabolism and energy balance. Chronic sleep deprivation [32–35] and high sleep variability [10], e.g., in shift work, can therefore lead to metabolic imbalances with the corresponding risk factors and comorbidities. In addition, acute sleep restriction seems to increase the appetite for high-calorie foods [35]. Taken together, sleep times dictated by the environment that deviate significantly from intrinsically determined sleep time preferences appear to represent a major health problem in our society.

Epidemiological studies have also identified sleep disorders as a risk factor for metabolic and cardiovascular mortality/morbidity [36]. On the other hand, metabolic changes seem to be frequently associated with sleep disorders [36]. Metabolites also appear to play a role in the natural aging process, modulating brain function and sleep architecture with possible involvement of the circadian system [19]. Hence, metabolites may also influence the neuronal networks that control sleep architecture in young individuals, which, in turn, modulates various brain functions. Moreover, chronic inflammation is associated with disturbed sleep patterns and architecture [37,38] and sleep disorders have a negative impact on the immune system [38]. Our own results in mice support the hypothesis that chronodisruption induces a proinflammatory state in the brain [39] and affects neuronal plasticity and brain function [40]. By integrating wearable devices, new biomarkers can be identified to elucidate these interrelationships in humans [38].

Studies combining longitudinal Fitbit sleep data and questionnaire survey suggest a negative impact of short sleep and high sleep variability on mood and depression [41,42]. Using a similar approach, we have previously shown in the same cohort as in this study that short sleep duration on working days is associated with a higher PHQ-4 score, which is mediated by a higher perception of workload under comparable objective workload [21]. The current literature screening suggests that sleep architecture characteristics are endophenotypic traits which might serve to identify increased risk for depressive symptoms and metabolic disturbances [43]. This is consistent with this study showing that ICC50REM is associated with BMI which, in turn, is associated with the perception of workload. Based on these associations, we cannot establish causality. However, we suspect that students with higher BMI may be more resilient to psychosocial stress than students with lower BMI. The relatively small and homogeneous sample limits the generalizability of the results. We suggest to further examine interactions between sleep timing, in general, and REM sleep timing, in particular, with metabolic homeostasis and resilience to psychosocial stress in larger cohorts.

In summary, in our sample of young healthy adults with a mean normal weight and sufficient sleep, BMI correlates with higher individual flexibility in sleep duration and timing, stronger relationship between REM sleep and total sleep, and lower subjective workload. This suggests that greater flexibility in sleep schedule and a stronger temporal alignment between REM sleep and total sleep is associated with anabolic metabolism and resilience to psychosocial stress.