What Can Make the Difference Between Chronotypes in Sleep Duration? Testing the Similarity of Their Homeostatic Processes

Knowledge of individual differences in the timing of human behavior and physiology seems to be of importance for the optimization of times for taking medications and for attending school, management of fatigue in occupational settings, prediction of performance and risk of accidents on the roads and in unsafe workplaces, diagnosis, and treatment of sleep and circadian rhythm disorders, etc. The recommendations about optimal timing for human activity are often challenged by the lack of information about the causes of the observed profound differences between individuals in the phase characteristics of their circadian rhythms. Additionally, the phase characteristics of such rhythms as the sleep-wake cycle demonstrate notable changes across the lifespan. These changes might cause conflicts between social and biological clocks. For instance, a dramatic weekday sleep loss occurs in late adolescents due to the confrontation between early school start times and their tendency to delay the timing of their sleep on weekends, which, at least partly, is determined by their biology (Crowley et al., 2014).

The publication of Kleitman’s seminal book “Sleep and Wakefulness” (1939) initiated a search for the biological underpinnings of two extremes of the timing of human behavior, known as morningness and eveningness. Given the complexity of the biological mechanisms controlling the phase characteristics of human behavioral and physiological rhythms, valuable insights into the causes of these inter- and intra-individual differences can be provided by mathematical modeling. Model-based simulations can numerically predict the dynamics of the regulatory processes underlying the observed differences between individuals, including the differences between morning and evening (M- and E-) types in these processes. The two-process conceptualization of the mechanisms of sleep-wake regulation (Borbély, 1982) seems to be the major contributor to our current understanding of the causes of differences between individuals in their 24-h sleep-wake pattern. It postulates that the cyclicity of sleep and wakefulness is determined by two regulation processes, a circadian process and a homeostatic process (Borbély, 1982; Daan et al., 1984). The circadian process can be linked to the entrained phase of the circadian pacemaker. It can be measured by tracing such markers of the circadian phase as the times for body temperature minimum and onset of melatonin secretion. The homeostatic process adjusts the intensity and duration of sleep as a function of the duration of prior wakefulness. It can be traced by measuring several spectral electroencephalographic (EEG) indexes during sleep and wakefulness. In particular, Slow Wave Activity (SWA: EEG power density in the 0.75–4.5 Hz range) during non-rapid eye movement sleep was utilized as the spectral EEG marker of the kinetics of sleep homeostasis. It was demonstrated that the exponential function might be applied for the quantitative description of the homeostatic process as the alternations of buildups and decays of SWA in the course of wakefulness and sleep, respectively (Daan et al., 1984; Duffy et al., 1999).

Evidence of the involvement of the circadian process in producing inter- and intra-individual differences in sleep times cannot be questioned. For instance, the results of measurement of body temperature minimum and the onset of melatonin secretion always suggested a 2–3-h difference in the positions of the circadian phase of M- and E-types (Foret, 1982; Kerkhof and van Dongen, 1996; Duffy et al., 1999, 2001; Baehr et al., 2000; Bailey and Heitkemper, 2001; Lack et al., 2009; Paine and Gander, 2016).

Research also provided solid evidence for the contribution of the homeostatic process to the age-associated (intraindividual) differences in sleep timing. For instance, both experiments and simulations confirmed that the kinetics of homeostatic process can be slower in mature adolescents compared to the kinetics in prepubescent adolescents (Jenni et al., 2005; Crowley et al., 2014; Putilov and Verevkin, 2018). Model-based simulations of sleep times showed that the typical changes in sleep timing and duration, from adolescence to old age, can be understood as a consequence of changes in the kinetics of the homeostatic process (Skeldon et al., 2016). Such understanding was supported by experimental research indicating that the age-associated differences cannot be explained by the change in the free-running circadian period (e.g., Czeisler et al., 1999; Kendall et al., 2001; Crowley and Eastman, 2018).

As for the experimental studies of two chronotypes of similar age (intraindividual difference), they suggested that one of the possible sources of a later sleep phase in E-types compared to M-types might be slower kinetics of the homeostatic process (Kerkhof and Lancel, 1991; Lancel and Kerkhof, 1991; Taillard et al., 1999, 2003; Mongrain et al., 2006; Mongrain and Dumont, 2007). However, can such a difference between chronotypes in the kinetics of sleep homeostasis lead to the difference between them in sleep duration? If the result of experimental research shows that the duration of sleep is practically identical in M- and E-types, this implies that the chronotypes are also identical in achieving the ultimate goal of homeostatic sleep regulation, irrespective of chronotype, which is an adequate amount of sleep. Indeed, in accordance with this goal of homeostatic regulation, the vast majority of cited above experiments, M- and E-types were found to be similar to one another in terms of sleep duration, despite significant differences between them in the kinetics of the homeostatic process (e.g., Lancel and Kerkhof, 1991; Taillard et al., 2003; Mongrain and Dumont, 2007).

Even more, if M- and E-types sleep at different circadian phases, there is no way to have identical sleep durations without having different rates of the kinetics of homeostatic processes. The reason for this is because M-types were found to sleep on a later phase of their circadian pacemaker due to a wider phase angle between this pacemaker and sleep (e.g., Duffy et al., 1999, 2001; Baehr et al., 2000; Mongrain et al., 2004; Emens et al., 2009). This was first predicted by simulations (e.g., Putilov, 1995) and then confirmed by the results of experimental research (e.g., Lazar et al., 2015), indicating that the parameters of SWA appear to be modulated by the circadian pacemaker (i.e., by the circadian process). Therefore, to produce similar amounts of sleep in two chronotypes sleeping at different circadian phases, it is necessary to have different rates of homeostatic buildup and decay. For example, when the decay of SWA is quicker on a later circadian phase than on an earlier circadian phase, the duration of decay (sleep) phase on the former phase must be shorter than the duration on the later phase. To make durations identical, the former decay rate must become slower than the later decay rate. Therefore, to clarify whether similar amounts of sleep can be obtained by the chronotypes sleeping at different circadian phases, it is necessary to simulate sleep times in M- and E-types with a model of sleep-wake regulation.

The present study aimed to test the hypothesis of the identity of the homeostatic processes in M- and E-types. We tested whether these processes in two distinct chronotypes are designed for obtaining similar amounts of sleep on free days (i.e., on the days when they are free of any constraints placed on their sleep-wake schedule by the society). The practically important consequence of providing support for the hypothesis of the identity of amounts of sleep in two chronotypes on free days would be a possibility to implement the model-based simulations into the calculation of weekday sleep losses in each of the chronotypes.

The first page ofcontains four sleep times (weekday and weekend bed- and rise times) estimated for 50 paired samples of study participants classified into M- and E-types by the authors of published papers. To select these 50 paired samples listed in, none of the exclusion criteria was applied. In the vast majority of publications, information on four sleep times was found in one of the paper’s tables. If four sleep times were reported for several ages, data on each age were included as separate lines in. Information on mean age in each sample and the methods of data collection was also included in. Objective methods (i.e., actigraphy) were used for estimation of sleep times in only seven paired samples, and sleep times were calculated from sleep diaries in four other paired samples. The second page ofcontains additional information on the sample sizes and the questionnaires used for the classification of study participants into chronotypes (with the number of items for each of the questionnaires and references). The current sleep times were used to classify into chronotypes only 5 pairs of samples, while, to classify 45 remaining pairs, 9 different versions of diurnal preference scales were applied (see the second page of). Supplementary Appendix A Supplementary Appendix A Supplementary Appendix A Supplementary Appendix A Supplementary Appendix A Supplementary Appendix A

The paired samples were assigned to eight age groups to test the significance of age-associated changes in the differences between M- and E-types in sleep times with one-way ANOVAs (Figures 1–3), and paired Student’s t-test was employed for the comparison of mean sleep times obtained for M- and E-types by averaging over 50 samples (Table 1). The statistical tests were performed using the Statistical Package for the Social Sciences (SPSS 23, IBM, Armonk, NY, United States).

The average weekday and weekend sleep times obtained for M- and E-types (Table 1) were simulated with a model of sleep-wake regulating processes (Putilov, 1995) that postulates the modulating influence of the circadian process on the parameters of SWA, a marker of the homeostatic process. In this model, t1 and t2 are the initial times for the buildup and decay phases of the 24-h sleep-wake cycle (i.e., the rise- and bedtimes on free days, respectively). The process of sleep-wake regulation, X(t), is simulated using the following equations:

where

is a periodic function with a period τ assigned to 24 h. This term (2) represents the modulating effect of the circadian process on the parameters of the homeostatic process represented by the time course of relative SWA (Figure 4). All parameters of the model applied for the simulations of sleep times in M- and E-types are listed in Table 2. A difference between chronotypes in the phase angle between the circadian pacemaker and sleep was suggested to reach 3.8 h with M-types sleeping at a later phase of their circadian pacemaker. The difference in their sleep times on free days was proposed to be close to the empirically obtained mean value for weekends, 1.8 h (Tables 1–3). We also proposed a 1-h difference in weekday risetime between M- and E-types that is also close to the empirically obtained mean value (Tables 1–3). Table 3 provides the estimates of the empirically obtained differences between chronotypes and the estimates of discrepancies between these and simulated values.

While the sleep times obtained for any of two chronotypes drastically changed throughout the lifespan (Figure 1), some of the differences between M- and E-types in these times did not change (Figure 3 and Table 3). The statistically significant changes were limited to the weekend risetime and weekend time in bed, as well as to the weekly averaged risetime and weekly averaged time in bed for which the major contributors are the weekday risetime and weekday time in bed (Figures 2, 3 and Table 3). Undoubtedly, the result suggested that the biologically determined differences between chronotypes do not change with advancing age, and significant age-associated changes in the difference in weekday sleep cannot be explained by the biological differences between chronotypes. Instead, they can be explained by the age-associated weakening of the social constraints placed on sleep-wake schedule of E-type participants (i.e., these constraints seem to be strict for school students with E-type but less strict for working adults with E-types).

Irrespective of age, weekend-weekday difference in bedtime, risetime, and time in bed were smaller in M-types than in E-types (Figure 2 and Table 3). Sleep loss was also smaller in M-types than in E-types (Figure 2 and Table 3). Finally, weekday time in bed was longer in M-types, whereas weekend time in bed was longer in E-types (Figure 3 and Table 3). As suggested by the results reported in Table 3 and illustrated in Figures 4, 5, such differences between M- and E-types were predicted by the simulations based on the assumption that sleep durations in these types on free days are equal (i.e., 9.00 h for both in Table 2). Notably, the simulations predicted only a tiny difference in favor of M-types in weekly averaged time in bed (0.020 h), and the analysis of empirical data also provided a similarly small and same-directional difference in weekly averaged time in bed in favor of M-types (0.098 h with standard error of 0.071 h).

The simulations explain why all these differences between chronotypes emerge despite the identity of their time in bed on free days (Table 2). Compared to M-types, E-types slept less on weekdays and more on weekends because sleep reduction on weekdays was larger in E-types to be extended more on weekends (Figures 5C,D). In other words, this was a model-predicted consequence of a smaller difference between E-and M-types in the advancing weekday shift of their bedtime compared to a larger difference between them in the advancing shift of risetime (Figures 5A,B). As a result of the differences between chronotypes caused by the transition to weekdays, the difference between them during the following delaying shift of risetime on weekends has to be, in turn, relatively larger than the difference in in the shift of bedtimes (Figures 5A,B).

Since the pioneering work of Kleitman (1939), a search for the biological mechanisms underlying the differences between M- and E-types has remained an intriguing topic for many studies in the field of sleep science and chronobiology. In accord with the two-process conceptualization of sleep-wake regulation processes (Borbély, 1982; Daan et al., 1984), the biological underpinning of the observed differences between M- and E-types in terms of sleep time might be explained by the difference in either the circadian process or the homeostatic process or both (e.g., Kerkhof and Lancel, 1991; Lancel and Kerkhof, 1991; Taillard et al., 1999, 2003; Mongrain et al., 2006; Mongrain and Dumont, 2007). Previous studies simulating sleep times using two-process models of sleep regulation have concluded that there might not be a need to consider differences in the circadian process in explaining and predicting intraindividual (not interindividual) differences in sleep timing across the lifespan (Skeldon et al., 2016; Putilov and Verevkin, 2018). The sleep times reported in the literature for 50 paired samples of M- and E-types of the same age were analyzed and simulated with a model, suggesting the possibility of circadian modulation of the homeostatic process. Our simulations support the hypothesis of the sleep homeostatic processes in two distinct chronotypes of the same age.

The ultimate goal of homeostatic regulation, as a way of enabling people to have an adequate amount of sleep on free days, was achieved by the study participants irrespective of their chronotype. Overall, the simulations of sleep times in M- and E-type people of the same age indicate that the homeostatic processes of sleep regulation might be identical. This result provides further support for the results of several other experimental studies (e.g., Lancel and Kerkhof, 1991; Taillard et al., 2003; Mongrain and Dumont, 2007), which indicated that M- and E-types were similar one to another in terms of sleep duration despite significant differences in the kinetics of their homeostatic process.

To our knowledge, this is the first analysis of sleep times in a hundred samples of M- and E-types from various age groups, some of the obtained results are of special interest for practical reasons.

Such results might have practical implications. For instance, for any individual of certain age and chronotype, the quantitative predictions of adequate amount of sleep required for avoiding adverse health consequences of weekday sleep loss can be made.

There are several limitations affecting these simulations of the sleep-wake regulation processes, which relied exclusively on bed- and rise-times as model inputs. One of the major limitations is the absence of any additional data on the markers of sleep intensity and circadian phases, such as changes in SWA levels during sleep and clock times for body temperature minimum or onset of melatonin secretion. Moreover, the objective methods (i.e., actigraphy) were applied for measuring sleep times only in a minor fraction of the analyzed samples (see). An additional disadvantage of our simulations was in using bed- and rise-times as the input of the model instead of times for sleep onset and offset. However, since sleep onset and offset were mostly provided through self-reporting, this calculation included some other subjective reports, such as the self-assessment of sleep latency, which are of even more questionable accuracy than the self-reporting of bed- and rise-times. It should also be mentioned that there were several other sources of variation in the estimates of sleep times that cannot be excluded or accounted for in the present analysis, i.e., different questionnaire tools were used for the classification of chronotypes by the authors of the published studies. Their studies were conducted in different countries with different sleep-related customs, during different seasons, and under different natural and artificial light-dark conditions. Furthermore, we cannot account in our simulations for the interplay between the biological sleep regulators and various psychological factors contributing to differences between chronotypes in sleep timing. Only average data were simulated because the number of samples in each of the 8 age groups was not big enough to test whether the model predictions remain practically the same for any of these groups. Supplementary Appendix A

Moreover, we simulated only sleep-wake cycles and did not consider other chronobiological differences between the two chronotypes. In particular, the simulations did not account for the differences between M- and E-types in the alertness-sleepiness rhythm. Despite a common-sense view that daily fluctuations of alertness-sleepiness level are simply a reflection of the human sleep-wake cycle, these fluctuations can be regarded as an example of a “strong” (rigid) circadian rhythm, such as the diurnal patterns of core body temperature and melatonin secretion. A “strong” circadian rhythm is characterized by a narrow range of entrainment that is the range of synchronizer’s periods to which a given rhythm can entrain. Unlike a weaker (lax) rhythm (i.e., the sleep-wake cycle), such a “strong” rhythm cannot be entrained by a time giver with a much longer or much shorter period (Pittendrigh and Daan, 1976; Granada et al., 2013). The experimental results indicate that the range of entrainment of alertness-sleepiness rhythm is even narrower than that of the core body temperature rhythm (Folkard et al., 1985). Therefore, the simulations of the alertness-sleepiness rhythm could require somewhat different and more advanced models than simulations of the sleep-wake cycle. At least one additional process is usually included in a model of fluctuations of alertness-sleepiness level (e.g., Åkerstedt and Folkard, 1997). For instance, since alertness demonstrates a gradual declining trend from one day to another in the course of prolonged wakefulness, this trend might be accounted for by postulating an additional change in, at least, one of the asymptotes that cannot be, as it is suggested in Eq. 1, a constant throughout the buildup (wake) phase of the cycle (Putilov et al., 2014, 2015, 2019). While the number of samples with sleep times is already sufficient for performing the present statistical analysis and simulations, the published data on “strong” rhythms in M- and E-types remains scarce. The accumulation of such data in future studies would allow the inclusion of simulations (1,2) and empirically derived (not hypothetical) estimates of the differences between chronotypes in the circadian phase and phase angle between this phase and sleep times on free days.

We tested the hypothesis of the general similarity of the homeostatic processes in morning and evening chronotypes. The homeostatic processes in the two chronotypes were found to be similar, at least, in terms of achieving the ultimate goal of homeostatic regulation, such as getting an adequate amount of sleep on free days. Therefore, in future studies of the mechanisms underlying the difference between chronotypes, it would be sufficient to evaluate the difference between them in the circadian process. In practical terms, the model-based simulations can be applied for the quantitative prediction of sleep losses associated with early weekday wakeups.

The original contributions presented in the study are included in the article/, further inquiries can be directed to the corresponding author. Supplementary Material

The studies involving human participants were reviewed and approved by the Ethic Committee of the FRC FTM. The patients/participants provided their written informed consent to participate in this study.

AP: conceptualization, data curation, formal analysis, funding acquisition, investigation, methodology, project administration, resources, software, supervision, validation, visualization, writing – original draft, and writing – review and editing. OD: funding acquisition, project administration, resources, writing – review and editing. Both authors contributed to the article and approved the submitted version.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fnins.2022.832807/full#supplementary-material↗