J Neural Transm DOI 10.1007/s00702-015-1460-y

PSYCHIATRY AND PRECLINICAL PSYCHIATRIC STUDIES - ORIGINAL ARTICLE

Successful physical exercise-induced weight loss is modulated by habitual sleep duration in the elderly: results of a pilot study Monique Goerke1,2 • Uwe Sobieray1 • Andreas Becke3,4 • Emrah Du¨zel1,4 Stefan Cohrs2 • Notger G. Mu¨ller1,3

•

Received: 5 May 2015 / Accepted: 11 September 2015  Springer-Verlag Wien 2015

Abstract Although it is widely accepted that physical exercise promotes weight loss, physical exercise alone had been found to result in only marginal weight loss compared to no treatment. Interestingly, both subjective and objective sleep duration have been shown to be negatively correlated to the body mass index (BMI). Despite this growing evidence of a relation between sleep duration and body weight, the role of habitual sleep duration in physical exercise-induced weight loss has not been studied so far. Twenty-two healthy elderly good sleepers aged 61–76 years (mean 68.36 years, 55 % female, BMI mean 25.15 kg/m2) either took part in a 12-week aerobic endurance training (3 9 30 min/week) or in a relaxation control (2 9 45 min/week). The BMI was assessed prior to and after intervention. Subjects maintained sleep logs every morning/evening during the training period, allowing for calculation of habitual sleep duration. Besides a significant main effect of the type of training, a significant interaction of type of training and habitual sleep duration was observed: while after treadmill training subjects who slept less than 7.5 h/night during intervention reduced their BMI by nearly 4 %, a comparable decrease in the BMI was

& Monique Goerke [email protected] 1

German Center for Neurodegenerative Diseases (DZNE) Magdeburg, Leipziger Str. 44, 39120 Magdeburg, Germany

2

Department of Psychiatry and Psychotherapy, University of Rostock, Gehlsheimer Str. 20, 18147 Rostock, Germany

3

Medical Faculty, Otto-von-Guericke University Magdeburg, Leipziger Str. 44, 39120 Magdeburg, Germany

4

Institute of Cognitive Neurology and Dementia Research, Otto-von-Guericke University Magdeburg, Leipziger Str. 44, 39120 Magdeburg, Germany

found neither in subjects who slept more than 7.5 h nor after relaxation training independent of sleep duration. Sleep duration itself did not change in any group. Although results should be interpreted with caution due to the small sample size, this is the first study to indicate that physical exercise might compensate for disturbed body weight regulation associated with short sleep duration. Keywords Sleep duration
Body weight
Body mass index
Physical exercise
Weight reduction

Introduction Although it is widely accepted that physical exercise promotes weight loss, evidence for a weight-reducing effect of physical exercise alone is limited. In a systematic Cochrane review of 41 randomized controlled trials with 3476 overweight or obese subjects aged 20–75 years, physical exercise alone resulted in only marginally weight losses compared to no treatment (Shaw et al. 2006). Congruently, Swift and colleagues concluded in a very recent review that with physical exercise consistent with public health recommendations [225–420 min/week (Donnelly et al. 2009)] only a modest weight loss can be expected (Swift et al. 2014). A newly identified factor which seems to influence body weight is sleep duration: In epidemiological cross-sectional studies, both subjective and objective sleep duration have been shown to be negatively correlated with the body mass index (BMI) in adult populations (Xiao et al. 2013; Mezick et al. 2014; Ford et al. 2014). Moreover, short sleepers (\6 h/night) were more likely to be obese in a middle-aged and older population (Xiao et al. 2013). Since cross-sectional data make causal inferences difficult, prospective studies are of interest. A systematic review of 20

123

M. Goerke et al.

longitudinal studies found short sleep duration consistently associated with subsequent weight gain in children, while in adults inconsistent results emerged (Magee and Hale 2012). However, a recent study with 83,377 subjects aged 51–72 years free of chronic conditions which could have been confounders in earlier studies showed short sleep (\6 h/night) to be associated with more weight gain in the 7.5-year follow-up compared with sleeping 7–8 h/night (Xiao et al. 2013). Furthermore, among subjects who were not obese at baseline, those who reported less than 5 h of sleep per night had an approximately 40 % higher risk of developing obesity than did those who reported to sleep 7–8 h/night (Xiao et al. 2013). Another recent study with 10,532 Mediterranean middle-aged subjects free of chronic conditions and obesity at baseline found short sleep (\5 h/ night) to be associated with a higher risk of becoming obese in a 6.5-year follow-up compared with sleeping 7–8 h/night (Sayon-Orea et al. 2013). Additionally, those who took a siesta for 30 min/day had a 33 % lower risk of developing obesity compared with those who never or almost never took a siesta (Sayon-Orea et al. 2013). Although this growing evidence for a relation between body weight and sleep duration, to date, the role of habitual sleep duration in physical exercise-induced weight loss has not been studied yet. We, therefore, addressed this issue in a sample of healthy elderly subjects participating in a 12-week physical exercise training program, which was part of a larger study exploring the effect of physical exercise on hippocampal plasticity. Because Vgontzas et al. (2013) found the longitudinal association between subjective short sleep duration and incident obesity was mediated by poor sleep in older adults, only healthy sleepers (Pittsburgh Sleep Quality Index at baseline B5) were included into our study. We hypothesized that (1) 12 weeks of physical exercise would result in a moderate BMI reduction, and (2) this BMI reduction might be modulated by the habitual sleep duration. Based on the finding that sleep duration is inversely related to the BMI, we expected short sleepers to lose less weight compared to long sleepers. Because a meta-analysis revealed an association between sleep duration and metabolic syndrome (Ju and Choi 2013) which is, amongst others, characterized by dyslipidemia and impaired glucose tolerance, we additionally explored the effect of physical exercise and sleep duration on lipid and glycosylated hemoglobin A1c (HbA1c) levels.

Materials and methods Subjects A total of 22 community-dwelling healthy elderly aged 61–76 years (mean age 68.36 years, 55 % female) were

123

included in the present study. Inclusion criterion was age 60–80 years. Exclusion criteria were (a) a score [5 on the Pittsburgh Sleep Quality Index (Buysse et al. 1989) reflecting altered sleep quality, and according to the larger study the project was implemented in, (b) a score [13 on the Beck Depression Inventory-II (Beck et al. 1996; Kuhner et al. 2007) reflecting possible depression, (c) a score B27 on the Mini-Mental State Examination (Folstein et al. 1975), (d) low visual acuity despite correction, (e) a history of neurological diseases or infarcts (e.g., stroke, medically treated diabetes mellitus, Parkinson’s disease, Alzheimer’s disease), (f) joint or muscle complaints whilst walking, (g) history of major cardiovascular disease and presently uncontrolled hypertension, (h) self-report of regular exercise of more than 1 h of endurance-related exercise per week in the last 3 months, (i) medication which interacts with brain function, and (j) contraindications against magnetic resonance imaging. The study protocol was approved by the ethics committee of the University of Magdeburg. Prior to the study, all subjects provided written informed consent. Subjects were compensated for their study participation. Experimental design and procedure Body mass index (BMI), fasting lipid levels [total cholesterol, low-density lipoprotein (LDL-) cholesterol, and highdensity lipoprotein (HDL-) cholesterol], glycosylated hemoglobin A1c (HbA1c) levels, and sleep quality/mean sleep duration during the past 4 weeks were assessed all at the same day, once at baseline and again after the 12-week intervention period. The BMI is a measure for body composition based on an individual’s weight and height that is commonly used to classify overweight and obesity in adults (World Health Organization 1998). It is defined as the weight in kilograms divided by the square of the height in meters (World Health Organization 1998). Body weight was measured via a calibrated electronic scale. Fasting blood samples were all drawn around the same time (9.30 am). Lipid levels were measured in serum samples, HbA1c levels were measured in whole blood added ethylenediaminetetraacetic acid (EDTA), performed by routine methods at the central laboratory of the University Hospital Magdeburg. Sleep quality/mean sleep duration over the past 4 weeks was assessed by the Pittsburgh Sleep Quality Index (PSQI). In this self-rated questionnaire nineteen individual items generate seven sub-scores: (1) subjective sleep quality, (2) sleep latency, (3) sleep duration, (4) habitual sleep efficiency, (5) sleep disturbances, (6) use of sleeping medication, and (7) daytime dysfunction. The sum of scores for these seven components yields one global score, with a

Successful physical exercise-induced weight loss is modulated by habitual sleep duration in…

score [5 indicating poor sleep (Buysse et al. 1989). Mean sleep duration was assessed by question 4 of the PSQI: ‘‘during the past month, how many hours of actual sleep did you get at night?’’. Intervention Physical exercise training The aerobic endurance training was carried out on stationary treadmills (HP Cosmos; http://www.h-p-cosmos.com/en/ running-machines/mercury_med/specifications.htm). Sport scientists supervised the intervention. Subjects received three times/week individually optimized 30 min interval training for 12 weeks plus 5 min warm up and 5 min stretching at the end of each training session. Training intensity was determined by target heart rate [Karvonen method (Karvonen et al. 1957)], starting at 65 % and increased by 5 % steps for 4 weeks. Target heart rate was based on the maximum heart rate during the assessment of oxygen consumption at the ventilatory anaerobic threshold (VO2VAT) prior to intervention in addition to the felt exertion [CR10 scale (Borg and Kaijser 2006)] during intervention. Oxygen consumption at the ventilatory anaerobic threshold was used instead of maximal oxygen consumption (VO2MAX) to account for the fact that some subjects did not perform up to maximal exhaustion to avoid any cardiovascular risk [for details see (Maass et al. 2014)]. Individually preferred walking speed and steepness of the treadmill was controlled via treadmill heart rate monitors. Walking/running interval duration was increased from 5 min, with slow walking brakes of 2 min at the beginning of the training period, up to 30 min continuously walking/running periods at the end of the intervention. There were no drop-outs during the 12-week training period. Progressive muscle relaxation training Subjects in the relaxation group came in twice a week and received 45 min of supervised progressive muscle relaxation training (Jacobson 1938) for 12 weeks. They were asked to tense and then relax specific muscle groups with closed eyes in supine position, following the instructions of a trained exercise leader. There were no drop-outs during the 12-week training period. Progressive muscle relaxation was chosen as control intervention to hold social interactions, schedule, and motivation as similar as possible to the exercise training without affecting cardiovascular fitness. The different time regimes (3 9 30 versus 2 9 45 min/week) for the two intervention groups were chosen for organizational reasons; however, both intervention groups received a total duration of 90 min of training per week.

Assessment during the 12-week training period Sleep Subjects maintained sleep logs every morning and every evening during the 12-week training period. In the sleep logs, bedtime, lights-off time, estimated sleep onset latency, estimated time spent awake after sleep onset, time of final awakening, and duration of daytime naps were recorded. For every single night during the training period total sleep time was calculated. Total sleep time was defined as time from lights-off to the final awakening time minus estimated sleep onset latency and estimated time spent awake after sleep onset. Physical activity During the 12-week training period, the International Physical Activity Questionnaire (IPAQ) was given by the end of each week to assess activities beyond intervention. The IPAQ (long version) captures 1-week retrospective estimates of physical activity including moderate workrelated activities, transportation, household, leisure activities as well as time spent sitting (Booth et al. 1996). The IPAQ-metabolic equivalent (MET) scores were calculated by comparing the energy consumption of different activities, based on the calories burned at each activity or oxygen consumption, multiplied with turnover at rest (1 MET = 1 kcal/kg/h or 3.5 ml/kg/min). Statistical analysis Statistical analysis was done using IBM SPSS Statistics version 21 (IBM Corp., Armonk, United States). Data were tested for normal distribution by the Shapiro–Wilk test. Because the HDL-cholesterol data were not normally distributed, a log-transformation was applied. Since age and gender were not of direct interest, correlational analyses were performed as partial correlations controlling for age and gender. Repeated-measures analyses of variance (ANOVA) were conducted with time (baseline versus post intervention) as inner-subject factor, type of training (treadmill versus relaxation training) and habitual sleep duration (long versus short sleepers) as between-subjects factors and age, gender and IPAQ physical activity (due to the fact, that during the training period the relaxation group was more active beyond intervention) as covariates. Furthermore, a univariate ANOVA with the percentage change in the BMI across the training period as dependent variable, type of training and habitual sleep duration as between-subjects factors and age, gender and IPAQ physical activity as covariates was performed. Analyses were adjusted for multiple comparisons using Bonferroni

123

M. Goerke et al.

correction. Two-tailed p values \.05 were considered significant. Results are expressed as mean ± standard error of the mean. Because lipid-/HbA1c data from one subject were lacking, data from an uneven number of subjects (11 treadmill training, 10 relaxation training) were included in the respective analyses. Effect sizes and post hoc calculated achieved power were determined using G*Power 3.1.7 (Faul et al. 2007).

Results Prior to training Prior to training subject’s Body mass index (BMI) ranged from 18.18 to 31.31 kg/m2 (mean 25.15 ± 0.66 kg/m2) reflecting the fact that this exercise training was part of a larger study which has not been set up as a weight reduction program. Mean sleep duration per night during the last 4 weeks prior to training as assessed by the PSQI differed between 360 and 570 min (mean 452.73 ± 11.99 min). The BMI measured at baseline was inversely correlated with mean sleep duration during the last 4 weeks prior to training (r = -0.444, p = .050). A closer look revealed (i) overweight subjects (BMI [ 25 kg/m2, n = 10) to report a 50 min shorter sleep duration than normal weight subjects (BMI \ 25 kg/m2, n = 12): 426.00 ± 17.78 min versus 475.00 ± 13.73 min, T(20) = 2.216, p = .038; and (2) a higher baseline-BMI in subjects who slept less than 7.5 h per night during the last 4 weeks prior to training (n = 12, mean sleep duration 412.50 ± 9.86 min) compared to subjects who slept more than 7.5 h per night (n = 10, mean sleep duration 501 ± 11.00 min): 26.58 ± 0.85 versus 23.45 ± 0.77 kg/m2, T(20) -2.677, p = .015 (cut-off of 7.5 h was based on a median split). Detailed data on sleep duration and physical activity beyond intervention assessed across the 12-week training period Mean sleep duration per night across the 12-week training period assessed by sleep logs ranged from 344.76 to 525.06 min (mean 452.47 ± 10.14 min). Mean sleep duration per night across the 12-week training period and mean sleep duration per night during the last 4 weeks prior to training assessed by the PSQI were significantly correlated (r = 0.712, p \.0001). Treadmill and relaxation groups did not differ significantly in their habitual sleep duration across the 12-week training period (p = .505). There was a wide range in physical activity beyond intervention across the 12-week training period assessed by the IPAQ spreading from 2178.90 to 16,236.82 MET (mean 7586.04 ± 872.13 MET). Unfortunately, there was

123

a significant difference between the treadmill and relaxation groups with the relaxation group being more active beyond intervention [treadmill training: 6033.43 ± 620.99 MET, relaxation training: 9138.64 ± 1525.38 MET; F(1,16) = 5.521, p = .032]. We, therefore, controlled for mean IPAQ physical activity by including it as a covariate in the analyses of the effects of the training. However, habitual sleep duration during the training period in terms of grouping subjects by less or more than 7.5 h sleep per night during the training period had no significant effect on their physical activity level (p = .775). Effects of a 12-week treadmill versus relaxation training on the BMI A repeated-measures ANOVA revealed that changes in the BMI across the training period were significantly modulated by the type of training [F(1,15) = 5.138, p = .039]: While BMI decreased from 25.38 ± 0.77 to 24.74 ± 0.77 kg/m2 after a 12-week treadmill training, no decrease was found after a relaxation training of equivalent length (24.93 ± 1.11 to 25.06 ± 1.21 kg/m2). Moreover, a significant three-way interaction indicated the decrease in BMI induced by training to depend on sleep duration [F(1,15) = 6.047, p = .027]: a reduction in the BMI by treadmill training was only observable in the group of subjects who slept less than 7.5 h per night during the training period (‘‘habitual short sleepers’’, habitual sleep duration 416.13 ± 10.18 min), whereas in subjects who slept more than 7.5 h per night during the training period (‘‘habitual long sleepers’’, habitual sleep duration 480.72 ± 12.72 min) the BMI remained unchanged as it did in both habitual short and long sleepers in the relaxation control (habitual short sleepers 402.34 ± 20.27 min, habitual long sleepers 492.08 ± 8.88 min; for details see Table 1). Because the BMI measured prior to training—as well as the BMI measured after the training period—was significantly higher in habitual short sleepers than in habitual long sleepers [BMI prior to training: habitual short sleepers 27.00 ± .93 kg/m2, habitual long sleepers 23.62 ± .69 kg/ m2, T(20) = 2.973, p = .008; BMI after training: habitual short sleepers 26.55 ± 1.13 kg/m2, habitual long sleepers 23.52 ± .68 kg/m2, T(20) = 2.381, p = .027], a univariate ANOVA with the percentage change in the BMI as dependent variable was performed. There was a significant main effect of the type of training [F(1,15) = 5.761, p = .030, effect size f = 0.62, power = 0.79] indicating a percentage reduction in the BMI after treadmill but not after relaxation training (treadmill training: -2.48 ± 1.12 %; relaxation training: 0.40 ± 0.64 %; see Fig. 1a). Additionally, a significant interaction effect of type of training x habitual sleep duration was found

Successful physical exercise-induced weight loss is modulated by habitual sleep duration in… Table 1 Body mass index (BMI) prior to and after 12 weeks of intervention (treadmill training versus relaxation control) in subjects with a habitual sleep duration \7.5 and [7.5 h/night during the training period (habitual short versus habitual long sleepers) BMI at baseline (kg/m2)

BMI post training (kg/m2)

Treadmill training Habitual short sleepers

26.38 ± 1.19

25.34 ± 1.27

Habitual long sleepers

24.17 ± 0.67

24.02 ± 0.81

Relaxation control Habitual short sleepers

27.92 ± 1.55

28.37 ± 1.92

Habitual long sleepers

23.23 ± 1.11

23.17 ± 1.05

Mean and SEMs are given

Effects of a 12-week treadmill versus relaxation training on sleep To evaluate the effects of treadmill and relaxation training on sleep, we compared PSQI-assessed mean sleep duration during the last 4 weeks prior to training with mean sleep duration during the last 4 weeks of the training period. However, there was no evidence pointing towards a sleeppromoting effect of either both interventions or one of them (p values [.161). Additionally, we analyzed whether subjective sleep quality was changed by the type of training. However, treadmill and relaxation groups did not differ in their PSQI global score change across the 12-week training period (p = .740). BMI, sleep duration, lipid and HbA1c levels

Fig. 1 Percentage change in body mass index (BMI) after 12 weeks of intervention. a Effect of type of training. b Effect of type of training 9 habitual sleep duration during the training period (\7.5 h/ night habitual short sleepers, [7.5 h/night habitual long sleepers). Means and SEMs are given

[F(1,15) = 5.220, p = .037, effect size f = 0.59, power = 0.74], strengthening the above-reported finding: While after treadmill training habitual short sleepers reduced their BMI by 3.98 ± 1.83 %, neither a comparable decrease in the BMI was found in habitual long sleepers nor after relaxation training (treadmill training: habitual long sleepers -0.68 ± 0.60 %; relaxation training: habitual short sleepers 1.43 ± 1.42 %, habitual long sleepers -0.19 ± 0.58 %; see Fig. 1b). Please note, that treadmill and relaxation groups did not differ significantly in their BMI prior to the training period (p = .748); the distribution of overweight and normal weight subjects at baseline was similar in both groups (overweight: five subjects, normal weight: six subjects); and there was no significant difference in percentage BMI reduction between overweight and normal weight subjects after treadmill training (overweight subjects -2.63 ± 1.05 %, normal weight subjects -2.30 ± 2.29 %, p = .890).

Prior to intervention, BMI was significantly correlated with the HbA1c levels (NGSP1: r = 0.530, p = .020; IFCC2: r = 0.543, p = .016). Additionally, there was a significant negative relation between BMI and HDL-cholesterol level (r = -0.510, p = .026). No significant relations were found between BMI and the total cholesterol or the LDLcholesterol levels (p [ .336). Additionally, significant negative correlations between the PSQI sleep duration during the last 4 weeks prior to training and baseline HbA1c levels were found (NGSP: r = -0.522, p = .022; IFCC: r = -0.471, p = .042). However, there were no significant correlations between the PSQI sleep duration and lipid levels (all p values [.130). Type of training significantly modulated changes in the total cholesterol [F(1,14) = 10.646, p = .006] and in the LDL-cholesterol level [F(1,14) = 11.089, p = .005]: Whereas both levels were found to be reduced after treadmill training (total cholesterol: 5.76 ± 0.28–5.45 ± 0.27 mmol/l; LDL-cholesterol: 3.53 ± 0.23–3.29 ± 0.23 mmol/l), no reduction was observed after relaxation training (total cholesterol: 6.48 ± 0.28–6.69 ± 0.23 mmol/l; LDL-cholesterol: 4.02 ± 0.21–4.29 ± 0.19 mmol/l). However, there was no evidence for an effect of sleep duration on those parameters (both p values [.245). Moreover, there was no effect of training on the HDLcholesterol level (p = .933). Treadmill and relaxation groups did not differ significantly in their lipid levels prior to the training period (p = .084). Concerning the HbA1c levels, no significant interaction effects of time 9 type of training were found (both p values [.115). Treadmill and relaxation groups did not differ 1

National Glycohemoglobin Standardization Program. International Federation of Clinical Chemistry and Laboratory Medicine.

2

123

M. Goerke et al.

significantly in their HbA1c levels prior to the training period (both p values [.227).

Discussion With the current pilot study we aimed to shed light on the role of habitual sleep duration in physical exercise-induced weight loss in a sample of healthy normal (BMI \ 25 kg/ m2) to overweight (BMI [ 25 kg/m2) subjects participating in a 12-week physical exercise training program. In accordance with growing evidence from large epidemiological studies which point to a relation between body weight and sleep duration (Ford et al. 2014; Mezick et al. 2014; Xiao et al. 2013), we found an inverse association between BMI and mean sleep duration during the last 4 weeks prior to training. Specifically, overweight subjects reported a 50-min shorter sleep duration than normal weight subjects; and subjects who slept less than 7.5 h per night during the last 4 weeks prior to training had a 3.13 kg/m2 higher BMI compared to subjects who slept more than 7.5 h per night. The cut-off of 7.5 h was based on a median split and had predominantly been chosen to ensure comparable sample sizes of short and long sleepers. This relatively high median sleep duration might represent a sample of healthy sleepers—remember that subjects with a score [5 on the PSQI reflecting possible sleep disturbance were excluded from the study. In other studies short sleep has been defined as sleep \5 or \6 h per night and the absence of a common definition of short sleep has been repeatedly criticized before (Horne 2011; Grandner et al. 2010). In this context it is worth noticing that a quite large difference in BMI is observable although our short sleepers slept only a little less than 7 h per night which is not an extreme sleep shortage compared to the recommended duration of 7–8 h. Our long sleepers, in contrast, on average slept nearly 90 min more compared to our short sleepers. These data suggest that even a moderate sleep shortage entails an elevated risk of becoming overweight. This is in line with data from a huge cross-sectional survey of 56,507 subjects aged 18–85 years which revealed that less than 7 h sleep per night was associated with obesity (Buxton and Marcelli 2010). We further observed that the BMI was moderately (-2.5 %) reduced after a 12-week treadmill training in our normal to overweight subjects. This is consistent with previous studies in overweight or obese subjects [for an overview see (Shaw et al. 2006; Swift et al. 2014)]. However, previous studies did not consider sleep duration as a possible factor modulating the extent of weight loss. Based on the finding that sleep duration is inversely related to BMI (Ford et al. 2014; Mezick et al. 2014; Xiao et al. 2013), we expected treadmill training to have a larger weight-reducing

123

effect in habitual long compared to habitual short sleepers (cut-off 7.5 h sleep duration during the training period). However, contrary to our hypothesis, we found a BMIreduction after treadmill training only in habitual short sleepers, while in habitual long sleepers the BMI remained unchanged as it did in the relaxation control. Importantly, although the BMI measured prior to the training period was significantly higher in habitual short sleepers than in habitual long sleepers, due to the fact that there was no significant difference in percentage BMI reduction between overweight (-2.63 %) and normal weight (-2.30 %) subjects after treadmill training, the BMI reduction after treadmill training in the habitual short sleepers cannot be attributed to a ceiling effect in weight loss in the habitual long sleepers. The finding that physical exercise reduced the BMI only in habitual short sleepers challenges the idea of extra effects of physical exercise in addition to long sleep duration on weight loss. A possible mechanism explaining that the BMI reduction after exercise occurred only in habitual short sleepers might include an association between short sleep, stress, and central obesity on the one hand and the stress-buffering effect of physical exercise on the other hand. First, short sleep duration has been found to be inversely associated with stress measured as hyperactivity of the hypothalamic–pituitary–adrenal (HPA) axis. In a study with 2751 middle-aged subjects self-reported short sleep duration was associated with an increased morning rise and a flattened slope in diurnal salivary cortisol secretion (Kumari et al. 2009). Additionally, nocturnal plasma cortisol was reported to be significantly higher in a group of young men sleeping nearly 8 h compared with those sleeping nearly 10 h (Spath-Schwalbe et al. 1992). Furthermore, short sleep, either from sleep restriction, total sleep deprivation, or insomnia, results in increased serum, salivary, or plasma cortisol in the evening when cortisol is usually low (Guyon et al. 2014; Vgontzas et al. 2001; Leproult et al. 1997; Rodenbeck et al. 2002). Second, short sleep duration has been found to be inversely associated with central obesity. Short sleep duration of less than 6 h has been shown to be associated with a higher risk for central obesity compared to those sleeping 7–8 h (Chaput et al. 2013) or to those sleeping 7–9 h (Ford et al. 2014). Thirdly, stress measured as hair cortisol (a marker of longterm cortisol secretion) has been shown to be positively correlated with central obesity (Kuehl et al. 2015). These data together with our present findings suggest an association between short sleep, stress and central obesity with a higher level of stress and a higher degree of central obesity in short sleepers compared to long sleepers. Physical exercise, however, is thought to reduce sensitivity and to confer resilience to stress. In their meta-analysis Crews and Landers (Crews and Landers 1987) concluded that aerobically fit subjects had a reduced psychosocial stress response

Successful physical exercise-induced weight loss is modulated by habitual sleep duration in…

which is mediated by HPA axis negative feedback (Zschucke et al. 2015; Traustadottir et al. 2005). Thus, habitual short sleepers might have a higher stress level associated with a higher degree of central obesity/a higher body weight; physical exercise might reduce stress and at least partially thereby body weight. Future research on this topic should, therefore, include waist circumference and cortisol measurements, a fact we have, unfortunately, neglected in this study. Another possible mechanism explaining that the BMI reduction after exercise occurred only in habitual short sleepers might include growth hormone (GH) secretion. GH is known to activate the mobilization of stored lipids while exercise has been shown to stimulate GH secretion. An interactive effect of sleep duration and exercise on weight seems likely because both complete deprivation and partial sleep restriction (bedtime from 10.30 pm to 3.00 am) have been shown to increase exercise-induced GH secretion (Ritsche et al. 2014; Abedelmalek et al. 2012). Therefore, the habitual short sleepers in our sample may have excreted more GH and, thereby, mobilized and finally burned more lipids consequently losing more weight than the subjects with long sleep duration. We, therefore, suggest that serum GH levels should be considered in future studies on this topic. In addition, several hypotheses of how short sleep might affect body weight have been suggested. One addresses the idea that insufficient sleep leads to increased fatigue and, therefore, a reduced daytime physical activity level and decreased energy expenditure, thereby predisposing to a higher body weight (Taheri 2006). However, we found little evidence supporting this hypothesis, since (1) our habitual short and long sleepers did not differ in their level of physical activity beyond intervention across the 12-week training period assessed by the IPAQ; and (2) we controlled for physical activity beyond intervention by including it as a covariate in the analyses of the effects of the training. This is in line with two other studies which were not able to explain the sleep–weight association by physical activity (Patel et al. 2006; Hasler et al. 2004). Another hypothesis addresses the availability of more waking time to eat in short sleepers (Sivak 2006). Although we did not control for caloric intake, this link seems unlikely, since in our group of habitual short sleepers sleep duration did not change across the training period, and, therefore, their weight loss cannot be attributed to less waking time to eat. However, we cannot exclude that within our treadmill training group habitual short sleepers had a higher caloric intake than habitual long sleepers independently of waking time to eat. Furthermore, as other cross-sectional studies before (Schroder et al. 2003; Wakabayashi 2004), we found the BMI at baseline to be inversely correlated to the HDL-cholesterol level. Contrary to Schroder et al. (2003) who reported associations

between BMI and total/LDL-cholesterol in men, we observed no relations between the baseline-BMI and the total cholesterol or the LDL-cholesterol levels. However, our sample size might have been too small to detect sexspecific relations. Additionally, we did not find a relation between PSQI sleep duration during the last 4 weeks prior to training and lipid levels. For total and LDL-cholesterol this is in line with a study in elderly Koreans (Choi et al. 2014). For HDL-cholesterol, in contrast, it conflicts with the findings of Choi and colleagues who reported an inverse association between sleep duration and HDL-cholesterol in women (Choi et al. 2014). However, again, as mentioned above, this discrepancy might be based on our sample size which might have been too small to detect sexspecific relations. Congruently to our baseline data, across the 12-week training period type of training but not habitual sleep duration had an influence on total cholesterol and LDL-cholesterol levels but not on HDL-cholesterol. While after treadmill training total cholesterol and LDLcholesterol levels were found to be reduced, there was no effect of training on the HDL-cholesterol level. Reductions in total cholesterol and LDL-cholesterol levels of elderly subjects have been found before after a 16-week exercise training (Martins et al. 2010); however, Martins and colleagues also found an increase in HDL-cholesterol after training. In contrast, our results are at odds with findings of a meta-analysis examining the effect of exercise on lipids in adults 50 years of age and older which reported physical exercise to significantly increase the HDL-cholesterol level whereas total cholesterol and LDL-cholesterol levels were only non-significantly reduced (Kelley et al. 2005). Like in epidemiological studies showing an increase in HbA1c levels to be associated with an increase in the BMI (Gulliford and Ukoumunne 2001; Boeing et al. 2000), in our pilot study BMI at baseline was positively correlated with HbA1c levels. Additionally, inverse correlations between the PSQI sleep duration during the last 4 weeks prior to training and baseline HbA1c levels were found, which is in line with previous studies which observed such a relation (Hancox and Landhuis 2012; Nakajima et al. 2008). In contrast, across the 12-week training period neither type of training nor habitual sleep duration affected the HbA1c levels. However, since the HbA1c levels are affected by the glycaemic history over a period of the previous 120 days, a longer training might be necessary to display changes in glycosylated hemoglobin. This pilot study has several limitations. First, size of the entire sample is small, leading to very small sample sizes at the type of training/habitual sleep duration group level. However, concerning the main results effect sizes and achieved power are good. Larger sample sizes would allow for further analyses, e.g., analyses of sex differences. As mentioned above, some findings on lipids and HbA1c

123

M. Goerke et al.

levels differ among men and women, thus separate analyses would have been recommendable. Because our sample size was too small to further subdivide the sample, we controlled for gender in terms of a covariate. Future studies, however, should allow for sex-specific analyses. Second, we studied only healthy elderly subjects. At the one hand, this is an advantage because sleep duration of retired people is not systematically affected by work schedules. Indeed, a huge epidemiological study found sleep duration on workdays to be significantly shorter than on free days in younger adults (Roenneberg et al. 2012). On the other hand, it is well known that sleep changes quantitatively and qualitatively with age (Ohayon et al. 2004). Furthermore, there is an age-related decline in GH secretion, which might be a potential mediator in the interaction of sleepand exercise-related body weight regulation (Chertman et al. 2000). It may, therefore, be the case that our findings cannot be extrapolated to all ages. Future studies should include a wider age-range and consider work schedules of younger subjects as a factor which might affect sleep duration. Third, our data on sleep duration are based on self-reports. Self-reported sleep duration has repeatedly been shown to correlate with objective actigraphy-based measures of sleep duration (Lauderdale et al. 2008; Mezick et al. 2013). However, in both studies subjects overestimated their sleep duration relative to actigraphy by 0.5–0.8 h (Lauderdale et al. 2008; Mezick et al. 2013). Therefore, our data on sleep duration might be overestimated too. To better record objective sleep duration future studies should be based on actigraphy measurements rather than on self-report. In conclusion, we found weight loss after a 12-week physical exercise training to be mediated by habitual sleep duration. This indicates that physical exercise might compensate for disturbed body weight regulation associated with short sleep duration. The fact that only habitual short sleepers showed a BMI reduction after treadmill training raises several questions. With a suggested association between short sleep, stress, and central obesity on the one hand and the stress-buffering effects of physical exercise on the other hand or, alternatively, altered exercise-induced GH secretion in habitual short sleepers compared to habitual long sleepers two potential mechanisms await scientific proof in future studies. Compliance with ethical standards Conflict of interest

The authors declare no conflict of interest.

References Abedelmalek S, Chtourou H, Aloui A, Aouichaoui C, Souissi N, Tabka Z (2012) Effect of time of day and partial sleep

123

deprivation on plasma concentrations of IL-6 during a shortterm maximal performance. Eur J Appl Physiol 113(1):241–248. doi:10.1007/s00421-012-2432-7 Beck AT, Steer RA, Brown GK (1996) Manual for the beck depression inventory—II. Psychological Corporation, San Antonio Boeing H, Weisgerber UM, Jeckel A, Rose HJ, Kroke A (2000) Association between glycated hemoglobin and diet and other lifestyle factors in a nondiabetic population: cross-sectional evaluation of data from the Potsdam cohort of the European Prospective Investigation into Cancer and Nutrition Study. Am J Clin Nutr 71(5):1115–1122 Booth ML, Owen N, Bauman AE, Gore JC (1996) Retest reliability of recall measures of leisure-time physical activity in Australian adults. Int J Epidemiol 25:153–159 Borg E, Kaijser L (2006) A comparison between three rating scales for perceived exertion and two different work tests. Scand J Med Sci Sports 16(1):57–69. doi:10.1111/j.1600-0838.2005.00448.x Buxton OM, Marcelli E (2010) Short and long sleep are positively associated with obesity, diabetes, hypertension, and cardiovascular disease among adults in the United States. Soc Sci Med 71(5):1027–1036. doi:10.1016/j.socscimed.2010.05.041 Buysse DJ, Reynolds CF 3rd, Monk TH, Berman SR, Kupfer DJ (1989) The Pittsburgh Sleep Quality Index: a new instrument for psychiatric practice and research. Psychiatry Res 28(2):193–213 Chaput JP, McNeil J, Despres JP, Bouchard C, Tremblay A (2013) Short sleep duration as a risk factor for the development of the metabolic syndrome in adults. Prev Med 57(6):872–877. doi:10. 1016/j.ypmed.2013.09.022 Chertman LS, Merriam GR, Kargi AY (2000) Growth hormone in aging. In: De Groot LJ, Beck-Peccoz P, Chrousos G et al (eds) Endotext [Internet]. MDText.com, Inc., South Dartmouth, MA [Last Updated 4 May 2015] Choi H, Lee J-M, Youm Y, Rhee Y, Kim CO, Kim HC (2014) Association between sleep duration and serum lipid levels in an elderly Korean population: The Korean Urban and Rural Elderly (KURE) Study. Circulation 129(Suppl. 1):AP 112 Crews DJ, Landers DM (1987) A meta-analytic review of aerobic fitness and reactivity to psychosocial stressors. Med Sci Sports Exerc 19(5 Suppl):S114–S120 Donnelly JE, Blair SN, Jakicic JM, Manore MM, Rankin JW, Smith BK (2009) American College of Sports Medicine Position Stand. Appropriate physical activity intervention strategies for weight loss and prevention of weight regain for adults. Med Sci Sports Exerc 41(2):459–471. doi:10.1249/MSS.0b013e3181949333 Faul F, Erdfelder E, Lang AG, Buchner A (2007) G*Power 3: a flexible statistical power analysis program for the social, behavioral, and biomedical sciences. Behav Res Methods 39(2):175–191 Folstein MF, Folstein SE, McHugh PR (1975) ‘‘Mini-mental state’’. A practical method for grading the cognitive state of patients for the clinician. J Psychiatr Res 12(3):189–198 Ford ES, Li C, Wheaton AG, Chapman DP, Perry GS, Croft JB (2014) Sleep duration and body mass index and waist circumference among US adults. Obesity (Silver Spring) 22(2):598–607. doi:10.1002/oby.20558 Grandner MA, Patel NP, Gehrman PR, Perlis ML, Pack AI (2010) Problems associated with short sleep: bridging the gap between laboratory and epidemiological studies. Sleep Med Rev 14(4):239–247. doi:10.1016/j.smrv.2009.08.001 Gulliford MC, Ukoumunne OC (2001) Determinants of glycated haemoglobin in the general population: associations with diet, alcohol and cigarette smoking. Eur J Clin Nutr 55(7):615–623. doi:10.1038/sj.ejcn.1601233 Guyon A, Balbo M, Morselli LL, Tasali E, Leproult R, L’HermiteBaleriaux M, Van Cauter E, Spiegel K (2014) Adverse effects of

Successful physical exercise-induced weight loss is modulated by habitual sleep duration in… two nights of sleep restriction on the hypothalamic-pituitaryadrenal axis in healthy men. J Clin Endocrinol Metab 99(8):2861–2868. doi:10.1210/jc.2013-4254 Hancox RJ, Landhuis CE (2012) Association between sleep duration and haemoglobin A1c in young adults. J Epidemiol Community Health 66(10):957–961. doi:10.1136/jech-2011-200217 Hasler G, Buysse DJ, Klaghofer R, Gamma A, Ajdacic V, Eich D, Rossler W, Angst J (2004) The association between short sleep duration and obesity in young adults: a 13-years prospective study. Sleep 27(4):661–666 Horne J (2011) Obesity and short sleep: unlikely bedfellows? Obes Rev 12(5):e84–e94. doi:10.1111/j.1467-789X.2010.00847.x Jacobson E (1938) Progressive relaxation. University of Chicago Press, Chicago Ju SY, Choi WS (2013) Sleep duration and metabolic syndrome in adult populations: a meta-analysis of observational studies. Nutr Diabetes 3:e65. doi:10.1038/nutd.2013.8 Karvonen MJ, Kentala E, Mustala O (1957) The effects of training on heart rate; a longitudinal study. Ann Med Exp Biol Fenn 35(3):307–315 Kelley GA, Kelley KS, Tran ZV (2005) Exercise, lipids, and lipoproteins in older adults: a meta-analysis. Prev Cardiol 8(4):206–214 Kuehl LK, Hinkelmann K, Muhtz C, Dettenborn L, Wingenfeld K, Spitzer C, Kirschbaum C, Wiedemann K, Otte C (2015) Hair cortisol and cortisol awakening response are associated with criteria of the metabolic syndrome in opposite directions. Psychoneuroendocrinology 51:365–370. doi:10.1016/j.psy neuen.2014.09.012 Kuhner C, Burger C, Keller F, Hautzinger M (2007) Reliability and validity of the Revised Beck Depression Inventory (BDI-II). Results from German samples. Nervenarzt 78(6):651–656. doi:10.1007/s00115-006-2098-7 Kumari M, Badrick E, Ferrie J, Perski A, Marmot M, Chandola T (2009) Self-reported sleep duration and sleep disturbance are independently associated with cortisol secretion in the Whitehall II study. J Clin Endocrinol Metab 94(12):4801–4809. doi:10. 1210/jc.2009-0555 Lauderdale DS, Knutson KL, Yan LL, Liu K, Rathouz PJ (2008) Selfreported and measured sleep duration: how similar are they? Epidemiology 19(6):838–845. doi:10.1097/EDE. 0b013e318187a7b000001648-200811000-00015 Leproult R, Copinschi G, Buxton O, Van Cauter E (1997) Sleep loss results in an elevation of cortisol levels the next evening. Sleep 20(10):865–870 Maass A, Duzel S, Goerke M, Becke A, Sobieray U, Neumann K, Lovden M, Lindenberger U, Backman L, Braun-Dullaeus R, Ahrens D, Heinze HJ, Muller NG, Duzel E (2014) Vascular hippocampal plasticity after aerobic exercise in older adults. Mol Psychiatry. doi:10.1038/mp.2014.114 Magee L, Hale L (2012) Longitudinal associations between sleep duration and subsequent weight gain: a systematic review. Sleep Med Rev 16(3):231–241. doi:10.1016/j.smrv.2011.05.005 Martins RA, Neves AP, Coelho-Silva MJ, Verissimo MT, Teixeira AM (2010) The effect of aerobic versus strength-based training on high-sensitivity C-reactive protein in older adults. Eur J Appl Physiol 110(1):161–169. doi:10.1007/s00421-010-1488-5 Mezick EJ, Wing RR, McCaffery JM (2013) Associations of selfreported and actigraphy-assessed sleep characteristics with body mass index and waist circumference in adults: moderation by gender. Sleep Med 15(1):64–70. doi:10.1016/j.sleep.2013.08. 784 Mezick EJ, Wing RR, McCaffery JM (2014) Associations of selfreported and actigraphy-assessed sleep characteristics with body mass index and waist circumference in adults: moderation by

gender. Sleep Med 15(1):64–70. doi:10.1016/j.sleep.2013.08. 784 Nakajima H, Kaneita Y, Yokoyama E, Harano S, Tamaki T, Ibuka E, Kaneko A, Takahashi I, Umeda T, Nakaji S, Ohida T (2008) Association between sleep duration and hemoglobin A1c level. Sleep Med 9(7):745–752. doi:10.1016/j.sleep.2007.07.017 Ohayon MM, Carskadon MA, Guilleminault C, Vitiello MV (2004) Meta-analysis of quantitative sleep parameters from childhood to old age in healthy individuals: developing normative sleep values across the human lifespan. Sleep 27(7):1255–1273 Patel SR, Malhotra A, White DP, Gottlieb DJ, Hu FB (2006) Association between reduced sleep and weight gain in women. Am J Epidemiol 164(10):947–954. doi:10.1093/aje/kwj280 Ritsche K, Nindl BC, Wideman L (2014) Exercise-induced growth hormone during acute sleep deprivation. Physiol Rep. doi:10. 14814/phy2.12166 Rodenbeck A, Huether G, Ruther E, Hajak G (2002) Interactions between evening and nocturnal cortisol secretion and sleep parameters in patients with severe chronic primary insomnia. Neurosci Lett 324(2):159–163 Roenneberg T, Allebrandt KV, Merrow M, Vetter C (2012) Social jetlag and obesity. Curr Biol 22(10):939–943. doi:10.1016/j.cub. 2012.03.038 Sayon-Orea C, Bes-Rastrollo M, Carlos S, Beunza JJ, BasterraGortari FJ, Martinez-Gonzalez MA (2013) Association between sleeping hours and siesta and the risk of obesity: the SUN Mediterranean cohort. Obes Facts 6(4):337–347. doi:10.1159/ 000354746 Schroder H, Marrugat J, Elosua R, Covas MI (2003) Relationship between body mass index, serum cholesterol, leisure-time physical activity, and diet in a Mediterranean Southern-Europe population. Br J Nutr 90(2):431–439 Shaw K, Gennat H, O’Rourke P, Del Mar C (2006) Exercise for overweight or obesity. Cochrane Database Syst Rev. doi:10. 1002/14651858.CD003817.pub3 Sivak M (2006) Sleeping more as a way to lose weight. Obes Rev 7(3):295–296. doi:10.1111/j.1467-789X.2006.00262.x Spath-Schwalbe E, Scholler T, Kern W, Fehm HL, Born J (1992) Nocturnal adrenocorticotropin and cortisol secretion depends on sleep duration and decreases in association with spontaneous awakening in the morning. J Clin Endocrinol Metab 75(6):1431–1435. doi:10.1210/jcem.75.6.1334495 Swift DL, Johannsen NM, Lavie CJ, Earnest CP, Church TS (2014) The role of exercise and physical activity in weight loss and maintenance. Prog Cardiovasc Dis 56(4):441–447 Taheri S (2006) The link between short sleep duration and obesity: we should recommend more sleep to prevent obesity. Arch Dis Child 91(11):881–884. doi:10.1136/adc.2005.093013 Traustadottir T, Bosch PR, Matt KS (2005) The HPA axis response to stress in women: effects of aging and fitness. Psychoneuroendocrinology 30(4):392–402. doi:10.1016/j.psyneuen.2004.11. 002 Vgontzas AN, Bixler EO, Lin HM, Prolo P, Mastorakos G, VelaBueno A, Kales A, Chrousos GP (2001) Chronic insomnia is associated with nyctohemeral activation of the hypothalamicpituitary-adrenal axis: clinical implications. J Clin Endocrinol Metab 86(8):3787–3794. doi:10.1210/jcem.86.8.7778 Vgontzas AN, Fernandez-Mendoza J, Miksiewicz T, Kritikou I, Shaffer ML, Liao D, Basta M, Bixler EO (2013) Unveiling the longitudinal association between short sleep duration and the incidence of obesity: the Penn State cohort. Int J Obes (Lond). doi:10.1038/ijo.2013.172 Wakabayashi I (2004) Relationships of body mass index with blood pressure and serum cholesterol concentrations at different ages. Aging Clin Exp Res 16(6):461–466

123

M. Goerke et al. World Health Organization (1998) Obesity. Preventing and managing the global epidemic. World Health Organization, Geneva Xiao Q, Arem H, Moore SC, Hollenbeck AR, Matthews CE (2013) A large prospective investigation of sleep duration, weight change, and obesity in the NIH-AARP diet and health study cohort. Am J Epidemiol 178(11):1600–1610

123

Zschucke E, Renneberg B, Dimeo F, Wustenberg T, Strohle A (2015) The stress-buffering effect of acute exercise: evidence for HPA axis negative feedback. Psychoneuroendocrinology 51:414–425. doi:10.1016/j.psyneuen.2014.10.019