Sleep and metabolism

Sleep and circadian rhythms have direct impacts on energy metabolism, and represent important mechanisms underlying the major health epidemics of obesity and diabetes⁽ Reference Wolk and Somers ^{3 ,} Reference Laposky, Bass and Kohsaka ^{5 )}. The majority of studies are observational in nature, and a few intervention studies have been reported. For instance, after partial sleep deprivation increased food intake and increased intake of energy from snacks with a higher carbohydrate or fat content has been shown⁽ Reference Nedeltcheva, Kilkus and Imperial ^{46 –} Reference St-Onge, Roberts and Chen ^{49 )}, and in short sleepers, increased respiratory quotient, implying increase in carbohydrate oxidation was observed⁽ Reference Nedeltcheva, Kilkus and Imperial ^{50 )}. Sleep deprivation may affect energy balance through hormone changes⁽ Reference Leproult, Copinschi and Buxton ^{40 ,} Reference Taheri, Lin and Austin ^{51 –} Reference Omisade, Buxton and Rusak ^{53 )}, although this was not observed in all studies⁽ Reference Nedeltcheva, Kilkus and Imperial ^{46 ,} Reference Schmid, Hallschmid and Jauch-Chara ^{54 –} Reference Morselli, Leproult and Balbo ^{56 )}.

In addition to reduced sleep, reduced QS is associated with metabolic disorders. Reduced QS obtained by a single night of fragmented sleep without reducing total sleeping time, induced a shift in insulin concentrations, from being lower in the morning and higher in the afternoon, while glucagon-like peptide-1 (GLP-1) concentrations and fullness scores were decreased. The decreased GLP-1 concentrations and fullness scores in the afternoon were synchronously related with reduced Visual Analogue Scale fullness scores, and increased Visual Analogue Scale desire to eat scores after dinner. This may lead to increased food intake and snacking, thus contributing to a positive energy balance⁽ Reference Gonnissen, Hursel and Rutters ^{43 )}. Also an increased respiratory quotient was observed when sleep was fragmented⁽ Reference Hursel, Rutters and Gonnissen ^{57 )}. Reduced QS affects several neuroendocrine signals involved in the control of substrate utilisation, including cortisol concentrations. The sharp morning rise and the steep fall to lower evening levels are modulated by even a single night of reduced sleep⁽ Reference Leproult, Copinschi and Buxton ^{40 ,} Reference Omisade, Buxton and Rusak ^{53 )}. After a single night of reduced QS, cortisol levels were significantly reduced after awakening and were elevated in the evening⁽ Reference Leproult, Copinschi and Buxton ^{40 ,} Reference Gonnissen, Hursel and Rutters ^{43 ,} Reference Omisade, Buxton and Rusak ^{53 )}. Taken together, a reduced sleep-duration, and reduced QS affect substrate oxidation, leptin- and ghrelin concentrations, sleeping metabolic rate, appetite, food reward, HPA-axis activity, gut-peptide concentrations as such, that a positive energy balance is enhanced, which increases the risk for overweight.

Sleep and body-weight management

Effects of changes in sleep duration during a dietary intervention for body-weight loss was assessed by Nedeltcheva et al.⁽ Reference Nedeltcheva, Kilkus and Imperial ^{46 ,} Reference Nedeltcheva, Kilkus and Imperial ^{50 )}. They showed that sleep restriction to 5·5 h sleep compared with 8·5 h sleep compromised the efficacy of a dietary intervention for weight loss. The combination of energy and sleep restriction in overweight adults resulted in decreased loss of fat and considerably increased loss of fat-free body mass. These results suggest that sleep plays a role in the preservation of human fat-free body mass during periods of reduced energy intake⁽ Reference Nedeltcheva, Kilkus and Imperial ^{46 ,} Reference Nedeltcheva, Kilkus and Imperial ^{50 )}. The effect on sparing fat-free mass was confirmed by Verhoef et al.⁽ Reference Verhoef, Camps and Gonnissen ^{58 )} who assessed whether during a weight-loss weight-maintenance programme in overweight subjects, a possible increase in sleep duration would precede the diet-induced decreases in body weight. They observed a concomitant inverse correlation between changes in sleep duration and in body weight, and respectively fat mass.

In addition, Chaput et al.⁽ Reference Chaput, Despres and Bouchard ^{59 )} observed that short-duration sleepers who maintained their short sleep duration habits experienced a greater increase in BMI and fat mass over a 6-year follow-up period compared with short-duration sleepers who increased their sleep duration, suggesting that shifting sleep duration from a short length to a healthier length is associated with lower adiposity gain⁽ Reference Chaput, Despres and Bouchard ^{59 )}. Moreover, they showed that both sleep duration and sleep quality were significantly related to fat mass loss during dietary interventions in overweight and obese adults⁽ Reference Chaput and Tremblay ^{60 )}. Despite these significant correlations it is not possible to determine any direction of causation.

Circadian alignment and energy balance

The significance of circadian alignment and energy balance implies assessment of the significance of circadian alignment for sleep, and sleep architecture, food-intake regulation, and physical activity. This is mainly assessed by circadian misalignment experiments. In the following sections, effects of circadian misalignment on sleep architecture and food-intake regulation are considered. In addition the significance of physical activity for circadian alignment will be highlighted.

Circadian misalignment and sleep

Circadian misalignment may reduce total sleep time, but mainly affects sleep architecture. The circadian phase at which sleep occurs affects the distribution of sleep stages. The preferential distribution of REM sleep towards the latter part of the night is linked to a circadian oscillator, while the preferential distribution of SWS towards the beginning of a sleep episode is mediated by homeostatic processes, i.e. the length of prior wakefulness⁽ Reference Carskadon, Dement, Roth and Dement ^{61 )}. Circadian misalignment resulted in disruption of the normal phase relationship between SWS and REM sleep, so that REM sleep is relatively phase advanced to SWS⁽ Reference Gonnissen, Mazuy and Rutters ^{44 )}. This abnormal circadian sequencing results in shortening of REM sleep latency and increasing REM sleep duration in a phase advanced stage. This short latency to REM sleep is typical of narcoleptic and depressive patients⁽ Reference Lee, Swanson and de la Iglesia ^{62 )}. Mood disorders, especially unipolar depression and seasonal affective disorder, have been linked to circadian rhythm abnormalities⁽ Reference Lee, Swanson and de la Iglesia ^{62 )}. Dysregulation in the HPA-axis, implying an overall increased cortisol secretion with a phase advance of the cortisol circadian rhythm is extremely frequent in individuals with depression⁽ Reference Monteleone and Maj ^{63 ,} Reference Van Cauter, Leproult and Plat ^{64 )}. Misalignment between timing of the clock and the timing of sleep, in either direction, has been associated with depression in vulnerable individuals⁽ Reference Emens, Lewy and Kinzie ^{65 )}.

Increased REM sleep during both a phase advance and a phase delay is not favourable⁽ Reference Nedeltcheva, Kilkus and Imperial ^{46 ,} Reference Nedeltcheva, Kilkus and Imperial ^{50 ,} Reference Peterson and Benca ^{66 –} Reference Brass and Auerbach ^{72 )}, because this results in a relatively shorter REM sleep duration during the second part of the night, associated with higher cortisol concentrations, higher fasting insulin concentrations, and a higher HOMA-IR index⁽ Reference Gonnissen, Mazuy and Rutters ^{44 ,} Reference Van Cauter, Leproult and Plat ^{64 ,} Reference Knutson, Van Cauter and Zee ^{73 ,} Reference Wu, Stone and Hsi ^{74 )}. Sleep during the circadian nadir (03.00–06.00 hours), is important in protecting normal physiological rhythms and function of the HPA-axis⁽ Reference Wu, Stone and Hsi ^{74 )}. Circadian misalignment, both a phase advance and a phase delay results in dysregulation of the HPA-axis. All in all, circadian misalignment appears to affect sleep-architecture, namely the distribution of sleep stages. REM sleep then becomes phase advanced to SWS with reduced REM sleep latency. REM sleep duration increases during phase advance, and during phase delay, resulting in a shorter REM sleep duration during the second part of the night.

Circadian misalignment, endocrinology, energy homeostasis and meal patterns

The daily patterns of feeding, energy utilisation, and energy storage across the daily 24-h cycle, is based upon a neuro-endocrinological system⁽ Reference Koren, Levitt Katz and Brar ^{75 ,} Reference Huang, Ramsey and Marcheva ^{76 )}. Metabolically relevant hormones show circadian oscillation with different daily patterns. Cortisol secretion has a circadian rhythm with the nadir during the early biological night (i.e. a time according to the original circadian rhythm associated with the start of behavioural inactivity) and the peak in the biological morning (i.e. a time according to the circadian rhythm associated with the start of behavioural activity)⁽ Reference Czeisler and Klerman ^{77 )}. Glucose and insulin levels peak during the late biological night⁽ Reference Kalsbeek and Strubbe ^{78 ,} Reference Morgan, Hampton and Gibbs ^{79 )}. Leptin, which suppresses appetite, is secreted in a circadian manner⁽ Reference van Aggel-Leijssen, van Baak and Tenenbaum ^{80 –} Reference Wong, Licinio and Yildiz ^{83 )}. In human subjects, night-time plasma leptin levels are high when appetite decreases, favouring fasting and nocturnal rest, and low during the day, when hunger increases. Gastric leptin levels oscillate in a circadian manner where leptin levels are high at night but low during the day⁽ Reference Cinti, Matteis and Pico ^{84 –} Reference Cinti, de Matteis and Ceresi ^{86 )}.

Ghrelin, which is produced in the stomach, pancreas and hypothalamus⁽ Reference Kojima and Kangawa ^{87 ,} Reference Cowley, Smith and Diano ^{88 )}, is involved in stimulating appetite via its action on neuropeptide Y in the lateral hypothalamus⁽ Reference Chen, Trumbauer and Chen ^{89 ,} Reference Hagemann, Meier and Gallwitz ^{90 )} and can also alter clock function in the SCN in vitro ⁽ Reference Yi, Challet and Pevet ^{91 ,} Reference Yannielli, Molyneux and Harrington ^{92 )}. Ghrelin oscillates with feeding⁽ Reference Cummings, Purnell and Frayo ^{93 )}, making this peptide a putative candidate for food-related entraining signals. In addition, elevated levels of ghrelin were found during the early part of the night in sleeping subjects, decreasing in the morning before awakening⁽ Reference Cummings, Purnell and Frayo ^{93 )}. Sleep deprivation can increase circulating ghrelin levels and this is accompanied by heightened hunger sensation⁽ Reference Schmid, Hallschmid and Jauch-Chara ^{54 )}. Thus, ghrelin may be a signal involved in the cross-talk between the peripheral and central circadian clock system.

In parallel to the circadian changes in neuropeptide levels and humoral signals from peripheral tissues, a circadian rhythm in macronutrient selection occurs. In human subjects, a carbohydrate-rich diet is favoured during breakfast and high-fat diets are preferred during evening meals⁽ Reference Westerterp-Plantenga, Miedema and Wijckmans-Duijsens ^{94 )}. Carbohydrates are metabolised better during breakfast because, also in relation to the glucostatic theory, then the body metabolically responds more readily to a glucose stimulus, since the fasting glucose level then is relatively stable, and very clearly indicates the first transient glucose decline⁽ Reference Dos Santos, Aragon and Padovani ^{95 )}.

Disruption of the circadian system affects metabolic and cardiovascular changes⁽ Reference Froy ^{2 ,} Reference Gonnissen, Rutters and Mazuy ^{45 ,} Reference Fletcher ^{96 –} Reference DeSantis, DiezRoux and Hajat ^{98 )}. In relation to appetite, release of some endocrine products shifts with meal patterns, such as glucose, insulin, GLP-1, ghrelin and leptin concentrations. Independently, disruption of the circadian system was associated with a significantly increased insulin response⁽ Reference Gonnissen, Rutters and Mazuy ^{45 )}, possibly related to hyperglycaemia associated with a progressing IR associated with sleep restriction, and to increased sympathetic nervous system activity⁽ Reference Fletcher ^{96 )}. Changes in the magnitudes of glucose and insulin responses indicate a disturbed glucose and insulin metabolism⁽ Reference Cinti, de Matteis and Ceresi ^{86 ,} Reference Ribeiro, Hampton and Morgan ^{97 )} and decreased GLP-1 concentrations indicate decreased satiety⁽ Reference Gonnissen, Rutters and Mazuy ^{45 )}.

Cortisol levels do not show a meal-related pattern during misalignment. During circadian misalignment the cortisol curve is flattened compared to 24-h cycles⁽ Reference Scheer, Hilton and Mantzoros ^{11 ,} Reference Gonnissen, Rutters and Mazuy ^{45 )}. In addition, with sleep loss, cortisol may exert its deleterious metabolic effects through remaining high night-time concentrations, which are associated with IR, suppressed immunity and increased inflammation⁽ Reference DeSantis, DiezRoux and Hajat ^{98 )}.

Moreover, circadian disruption results in increased carbohydrate oxidation⁽ Reference Gonnissen, Rutters and Mazuy ^{45 )}, probably due to both hyperinsulinaemia and hyperglycaemia⁽ Reference Yki-Jarvinen, Bogardus and Howard ^{99 )}. Often the increased carbohydrate oxidation occurs at the cost of protein oxidation, while fat oxidation remains constant⁽ Reference Gonnissen, Rutters and Mazuy ^{45 )}.

The main effect of circadian misalignment, either phase advanced or phase delayed, is a concomitant disturbance of the glucose–insulin metabolism and substrate-oxidation. Chronically eating and sleeping at unusual circadian times may create a health risk through metabolic disturbance⁽ Reference Gonnissen, Rutters and Mazuy ^{45 )}.

Consequently, meals need to be aligned in a circadian fashion. This requires timing and regularity of meals, with respect to food selection, meal frequency, meal intervals and meal size⁽ Reference Mendoza ^{9 ,} Reference Esquirol, Bongard and Mabile ^{10 ,} Reference Huang, Ramsey and Marcheva ^{76 )}. Beneficial metabolic effects of regular meal frequency on dietary thermogenesis, insulin sensitivity and fasting lipid profiles in healthy obese women align with effects of regular circadian patterns⁽ Reference Gonnissen, Hulshof and Westerterp-Plantenga ^{42 )}. After a regular v. an irregular meal pattern energy intake was lower, postprandial thermogenesis higher, and fasting total and LDL-cholesterol was lower⁽ Reference Farshchi, Taylor and Macdonald ^{100 )}. Peak insulin concentrations and area under the curve of insulin responses to the test meal were lower⁽ Reference Farshchi, Taylor and Macdonald ^{100 )}. Regular meal frequency creates more appropriate insulin sensitivity and lipid profiles compared with irregular meal frequency in healthy lean women⁽ Reference Farshchi, Taylor and Macdonald ^{101 )}, and irregular meal frequency led to a lower postprandial energy expenditure compared with the regular meal frequency, while the mean energy intake was not significantly different between the two⁽ Reference Farshchi, Taylor and Macdonald ^{102 )}. The reduced diet-induced thermogenesis with the irregular meal frequency may lead to weight gain in the long term⁽ Reference Farshchi, Taylor and Macdonald ^{102 )}. With a regular meal frequency glucose excursions are blunted, net insulin production is reduced, and LDL-cholesterol concentrations tend to be lowered, mainly due to gastric emptying slowing down, and insulin production being reduced⁽ Reference Arnold, Ball and Duncan ^{103 –} Reference Jones, Leitch and Pederson ^{111 )}. The net result is that lipid oxidation is favoured at the expense of glucose oxidation and lipid storage, and cholesterol synthesis is reduced. This may reduce adiposity and the level of circulating fatty acids, thereby leading to systematic, adaptive changes in both lipid and carbohydrate metabolism. Also examples of long-term responses to a sustained regular meal frequency such as improved glucose tolerance, and moderately reduced fasting plasma total and LDL-cholesterol, and a higher HDL:LDL cholesterol ratio are observed in normolipidaemic free-living subjects, as well as in type-II diabetes patients⁽ Reference Arnold, Ball and Duncan ^{103 –} Reference Jones, Leitch and Pederson ^{111 )}. Also for cholesterol synthesis, meal frequency-dependent control of cholesterogenesis appeared to be mediated via hormonal mechanisms⁽ Reference Ribeiro, Hampton and Morgan ^{97 )}.

Furthermore, circadian alignment including careful and fixed timing of food intake and meal frequency plays a role in substrate utilisation and in energy expenditure. Large metabolic fluctuations in carbohydrate and fat oxidation were shown in a gorging food-intake pattern, while in the nibbling pattern, carbohydrate and fat oxidation remained relatively constant during the active hours of the day⁽ Reference Verboeket-van de Venne and Westerterp ^{112 )}. Also, the diet-induced thermogenic response was related to meal frequency⁽ Reference Verboeket-van de Venne, Westerterp and Kester ^{113 )}. In a series of experimental studies, variation in energy intake was primarily explained by habitual meal frequency, macronutrient composition and number of blood glucose declines⁽ Reference Westerterp-Plantenga, Kovacs and Melanson ^{114 )}. The variation in habitual meal frequency was explained by percentage energy from carbohydrate or from fat in the diet, while the protein in the diet attenuates the metabolic amplitudes⁽ Reference Garrow, Durrant and Blaza ^{115 ,} Reference Westerterp-Plantenga, Nieuwenhuizen and Tome ^{116 )}. Moreover the effect of protein intake on satiety is partly due to the optimal timing of protein intake⁽ Reference Garrow, Durrant and Blaza ^{115 ,} Reference Westerterp-Plantenga, Nieuwenhuizen and Tome ^{116 )}. In healthy young men, habitual meal frequency appeared to be of greater significance in energy-intake regulation than forced meal frequency⁽ Reference Westerterp-Plantenga, Goris and Meijer ^{117 –} Reference Melanson, Westerterp-Plantenga and Campfield ^{120 )}. Adiposity may increase when young lean male subjects switch from a four- to a three-meal pattern by removing their usual afternoon meal, partly mediated by a change in the macronutrient composition of the diet⁽ Reference Chapelot, Marmonier and Aubert ^{121 )}. Assessment of the effect of omitting or adding the third meal, revealing that eating three meals compared with two meals had no effects on 24 h energy expenditure, diet-induced thermogenesis, activity-induced energy expenditure and sleeping metabolic rate. However, eating the same amount of energy divided over three meals compared with over two meals, did increase satiety, particularly during the day, and did increase fat oxidation, particularly during the night in healthy, normal-weight women⁽ Reference Smeets and Westerterp-Plantenga ^{122 )}.

Circadian alignment and physical activity

To include circadian rhythm in the assessment of circadian alignment and physical activity, wrist skin temperature has been used as a valid method of assessing circadian rhythms in human subjects⁽ Reference Tranel, Schroder and England ^{123 )}. As a measure for circadianity, circadian temperature amplitude and stability have been used as variables. Tranel et al.⁽ Reference Tranel, Schroder and England ^{123 )} identified physiological and behavioural measures that were significantly associated with the circadianity of these temperature parameters. Moreover, they tested the hypothesis that circadian temperature amplitude and stability would significantly differ among groups of healthy young men of varying adiposities⁽ Reference Tranel, Schroder and England ^{123 )}. Wrist skin temperatures taken at 10 min intervals for seven consecutive days were determined in eighteen optimal, twenty fair and twenty-one poor % body fat grouped young men and subsequently analysed using available validated software. Body composition, cardiorespiratory fitness, actigraphy, daily nutritional and sleep data, and fasting lipid, insulin and glucose concentration measures were also determined⁽ Reference Tranel, Schroder and England ^{123 )}. Subjects with one or more health problems had significantly lower temperature amplitude and stability⁽ Reference Tranel, Schroder and England ^{123 )}. This occurred in subjects with a single metabolic syndrome risk factor compared to those with no metabolic syndrome risk factors, and in subjects with a poor % body fat⁽ Reference Tranel, Schroder and England ^{123 )}. In addition, stepwise multivariate regression analyses showed that 50 % of the variance in temperature amplitude was explained by actigraphy (mean steps taken per day), cardiorespiratory fitness, and late night eating per week; and 57 % in temperature stability by mean steps taken per day, time spent in moderate-to-vigorous activity per day, fat mass, and late night eating per week⁽ Reference Tranel, Schroder and England ^{123 )}. Physical activity was the most important measure associated with the differences in circadian rhythm parameters⁽ Reference Tranel, Schroder and England ^{123 )}.

Genetic background for the amplitude and stability of circadian rhythm

In order to shed light on the genetic component of the circadian marker described above, relative genetic and environmental influences on wrist temperature (WT) was determined using classical twin models⁽ Reference Lopez-Minguez, Ordoñana and Sánchez-Romera ^{124 )}. A study was performed in fifty-three pairs of female twins (twenty-eight monozygotic and twenty-five dizygotic), with a BMI 25·9 (sd 3·78) and mean age 52 (sd 6) years. The sample was selected from the Murcia Twin Register. Circadian patterns were studied by analysing WT during one week every 10 min using Circadianware^{® (} Reference Lopez-Minguez, Ordoñana and Sánchez-Romera ^{124 )}. Genetic influences on WT variability were estimated by comparing correlations of monozygotic and dizygotic twin pairs and fitting genetic structural equation models to measured variables⁽ Reference Lopez-Minguez, Ordoñana and Sánchez-Romera ^{124 )}. Monozygotic twins showed higher intra-pair correlations than dizygotic twins for most of the parameters. Genetic factors were responsible for between 46 and 70 % of variance (broad sense heritability) in parameters such as mean temperature, mesor, acrophase, Rayleigh test, percentage of rhythmicity and 5 h of maximum temperature⁽ Reference Lopez-Minguez, Ordoñana and Sánchez-Romera ^{124 )}. The pattern of correlations and the genetic models point to moderate-to-high heritability for most of the WT parameters, suggesting a relevant genetic influence. The presence of these genetic factors points to endogenicity as the main cause of the coincidence of the WT rhythms⁽ Reference Lopez-Minguez, Ordoñana and Sánchez-Romera ^{124 )}. However, some WT parameters are still dependent on environment to a relevant extent and, hence, more amenable to change through external interventions⁽ Reference Lopez-Minguez, Ordoñana and Sánchez-Romera ^{124 )}.