Partial sleep restriction modulates secretory activity of thyrotropic axis in healthy men



Sebastian M. Schmid, Department of Internal Medicine I, University of Luebeck, Ratzeburger Allee 160, House 32, 23538 Luebeck, Germany.

Tel.: 0049-(0)-451-500-2306;

fax: 0049-(0)-451-500-3339;



Sleep and endocrine function are known to be closely related, but studies on the effect of moderate sleep loss on endocrine axes are still sparse. We examined the influence of partial sleep restriction for 2 days on the secretory activity of the thyrotropic axis. Fifteen healthy, normal-weight men were tested in a balanced, cross-over study. Serum concentrations of thyrotrophin (TSH), free triiodothyronine (fT3) and free thyroxine (fT4) were monitored at 1-h intervals during a 15-h daytime period (08:00–23:00 h) following two nights of restricted sleep (bedtime 02:45–07:00 h) and two nights of regular sleep (bedtime 22:45–07:00 h), respectively. Serum concentrations of fT3 (< 0.026) and fT4 (= 0.089) were higher after sleep restriction than regular sleep, with a subsequent blunting of TSH concentrations in the evening hours of the sleep restriction condition (= 0.008). These results indicate profound alterations in the secretory activity of the thyrotropic axis after 2 days of sleep restriction to ~4 h, suggesting that acute partial sleep loss impacts endocrine homeostasis, with potential consequences for health and wellbeing.


The decrease of average nocturnal sleep duration in modern societies over the past century has provoked an ongoing debate on the consequences of sleep restriction for human health and wellbeing (Czeisler, 2011; Schmid and Schultes, 2011). It is well documented that the circadian secretory patterns of distinct endocrine axes are subject to the modulatory influence of sleep which, in some cases, can even be the primary trigger of hormonal activity. For instance, pituitary release of thyrotrophin (TSH) follows a circadian pattern (Brabant et al., 1990a), with its nocturnal rise being inhibited by sleep (Parker et al., 1987). In particular, electroencephalographic delta wave activity (Gronfier et al., 1995), i.e. a characteristic of slow wave sleep (SWS), has been shown to dampen the nocturnal rise in TSH by reducing TSH pulses' amplitude (Goichot et al., 1992). Accordingly, total sleep deprivation has been shown to enhance thyrotropic axis secretory activity until subsequent recovery sleep and the compensatory increased SWS leads to a reduction in TSH concentrations (Brabant et al., 1990b). Total sleep deprivation, however, does not occur on a regular basis in most people. The effects of less severe forms of sleep restriction – as observed more frequently under ‘everyday life’ conditions – on endocrine function thus appear to be more relevant. One night of partial sleep restriction has been shown to induce a moderate stimulation of thyrotropic axis secretory activity (Baumgartner et al., 1993), while reduced thyrotropic axis activity was found after 2 weeks of moderate sleep restriction in women, but not men (Kessler et al., 2010).

To elucidate further the influence of partial sleep restriction on thyrotropic axis secretory activity, we measured respective hormone concentrations in serum samples derived from a cross-over study assessing the effects of sleep restriction on energy homeostasis that included two nights of sleep restriction and regular sleep, respectively (Schmid et al., 2009, 2011).


Experiments were conducted in 15 healthy, normal-weight [body mass index (BMI) 20.0–25.0 kg m−2, mean ± standard error of the mean (SEM): 22.9 ± 0.3 kg m−2] young men aged 20–40 years (27.1 ± 1.3 years). All participants had a regular sleep–wake cycle without shift-working during the last 4 weeks before the experiments, as indicated in a standardized interview on sleep habits. This interview revealed a habitual sleep duration of 459 ± 7 min (range: 450–540 min), with bedtime starting between 22:00 and 00:00 h and wake-up time ranging from 06:00 to 08:00 h. The study protocol was approved by the ethics committee on research involving humans at the University of Luebeck and all participants gave written informed consent prior to participation.

Participants were tested according to a balanced, cross-over design, with two conditions taking place at least 6 weeks apart. Blood samples were obtained after two consecutive nights of 4 h of sleep (‘4-h-sleep’; bedtime 02:45–07:00 h) and after two consecutive nights of 8 h of sleep (‘8-h-sleep’; bedtime 22:45–07:00 h), respectively. Before each experimental night, subjects arrived at the research unit at 20:00 h. After preparation for polysomnographical recordings, subjects went to bed and lights were turned off at 22:45 h in the 8-h sleep condition. In the 4-h-sleep condition, subjects remained awake in a sitting position within the laboratory illuminated by standard room lighting (~200 Lux) until 02:45 h. They were allowed to read and to watch non-arousing movies. Brisk physical activities were avoided and subjects were monitored constantly by the experimenters. In each condition, subjects were woken at 07:00 h. Subjects spent the day between the experimental nights outside the laboratory, having been instructed beforehand not to deviate from their usual eating habits and to avoid intense physical activities (e.g. working out) and naps during the day. Compliance with these instructions was confirmed by accelerometric recordings. Subjects reported back to the laboratory at 20:00 h for the second experimental night, which was identical to the night before.

On the experimental day subjects were awakened at 07:00 h and an intravenous catheter was inserted into a vein of the non-dominant distal forearm to allow for the drawing of blood samples at 07:40 h and, in 1-h intervals, between 08:00 h and the end of the experiment at 23:00 h. During the 15-h daytime assessment period (08:00–23:00 h), subjects spent their time with reading and watching non-arousing movies. They had ad libitum access to a standardized food buffet including breakfast, snacks and main meals. Results on physical activity, food intake and glucose metabolism have been reported previously (Schmid et al., 2009, 2011).

Sleep recordings were performed using a Nihon Kohden amplifier (EEG 4400 series; Nihon Kohden GmbH, Rosbach, Germany) and were scored offline according to standard criteria (Rechtschaffen and Kales, 1968). Concentrations of TSH, free triiodothyronine (fT3) and free thyroxine (fT4) were determined from stored (−80°C) serum samples by enzyme-linked immunoassays (all from DPC Immulite; DPC, Los Angeles, CA, USA), as described previously (Bremer et al., 2006).

All values are expressed as mean ± SEM. Analyses of hormonal data were based on analysis of variance (anova) for repeated measures, including the factors ‘condition’ and ‘time’. Pairwise comparisons of single time-point values were performed using Student's t-test. A P-value < 0.05 was considered significant.


In both nights of the 4-h-sleep condition, subjects as intended slept around 4 h less than in the 8-h-sleep condition (236 ± 2 versus 465 ± 4 and 238 ± 1 versus 467 ± 3 min, < 0.001 for ‘condition’ main effect). The absolute time spent in SWS was identical in both conditions (47 ± 7 versus 49 ± 6 min and 69 ± 7 versus 61 ± 6 min, = 0.84 for ‘condition’ main effect). The longer sleep duration in the 8-h-sleep condition was due primarily to more pronounced shallow sleep, i.e. S1 and S2, as well as rapid eye movement (REM) sleep (Schmid et al., 2011).

Morning TSH concentrations at 07:40 h did not differ between the 4-h-sleep and 8-h-sleep condition (2.28 ± 0.35 versus 2.06 ± 0.27 mU L−1, = 0.22). Across the experimental day, TSH levels decreased markedly to low midday levels in both conditions and started to rise again in the evening hours (< 0.001 for the ‘time’ main effect; Fig. 1a). The evening increase in TSH levels was blunted significantly in the 4-h-sleep condition (= 0.008 for ‘condition’ × ‘time’), resulting in lower levels in the sleep-restriction condition at the end of the experiment (1.94 ± 0.29 versus 2.34 ± 0.36 mU L−1, = 0.028).

Figure 1.

Mean (±standard error of the mean) serum concentrations of thyrotrophin (TSH) (a), free triiodothyronine (fT3) (b) and free thyroxine (fT4) (c) during an experimental day following two nights with 8 h of sleep each (white bars/circles) and two nights containing 4 h of sleep (black bars/circles), respectively; = 15.

In contrast to basal TSH levels, morning serum fT3 and fT4 concentrations were higher in the 4-h- than 8-h-sleep conditions (4.9 ± 0.2 versus 4.5 ± 0.2 pmol L−1, = 0.045; 18.5 ± 0.6 versus 17.6 ± 0.6 pmol L−1, = 0.039, respectively). Subsequently, fT3 and fT4 levels decreased in both conditions (< 0.001 and = 0.03, respectively, for the ‘time’ main effect; Fig. 1b,c), but remained elevated in the 4-h-sleep condition (= 0.026 and = 0.089, respectively, for the ‘condition’ main effect).


Our data, obtained in healthy young men, indicate that sleep restriction to 4 h during two consecutive nights profoundly affects thyrotropic axis activity. Respective hormonal changes are characterized by an elevation of serum levels in the peripheral thyroid hormones fT3 and fT4 throughout the day following sleep restriction and a blunted increase in serum TSH levels in the evening hours.

The finding of increased fT3 and fT4 concentrations after two nights of moderate sleep restriction is in accordance with previous studies showing comparable changes after one night of total or partial sleep deprivation in healthy subjects as well as in patients suffering from depression (Goichot et al., 1994; Kuhs et al., 1996). TSH concentrations have been found previously to be elevated in women during short-term sleep restriction for one night and the subsequent day (Baumgartner et al., 1993). In contrast, we observed a reduction in TSH levels in the evening hours following two consecutive days of sleep restriction, which is in line with observations made after sleep restriction to 4 h for six consecutive nights (Spiegel et al., 1999). In the latter study the free thyroxine index was still increased after six nights of sleep restriction, pointing clearly to a negative feedback mechanism as the main pathway for the observed reduction in TSH. Because participants in our and most other studies experienced sleep restriction while standard room lighting illuminated the laboratory, a shift in circadian secretory rhythm of TSH cannot be excluded. However, completely unchanged patterns of hypothalamic–pituitary–adrenal axis (HPA) axis activity in the present study, as reported earlier (Schmid et al., 2011), speak against a circadian phase shift. Although we cannot report on nocturnal concentrations of TSH, which is a clear limitation of this study, it appears reasonable to conclude that the reduction in evening TSH levels occurring after 2 days of sleep loss represents a negative feedback response to the daytime elevation in free thyroid hormone concentrations (Allan and Czeisler, 1994).

Of note, a recent study (Kessler et al., 2010) found no significant changes in markers of thyrotropic axis activity after 2 weeks of moderate sleep restriction in male subjects, suggesting that thyrotropic axis function may adapt to conditions of subchronic sleep restriction. Interestingly, in that study female subjects showed reduced TSH and fT3 levels at the end of the intervention period (Kessler et al., 2010), suggesting that sex is a factor that modulates the response of the thyroid axis to prolonged sleep restriction. Thus, follow-up studies including both women and men are needed to further elucidate gender as a potential modulator of endocrine effects of sleep loss.

Acutely increased secretory activity of the thyrotropic axis might have contributed to the decrease in insulin sensitivity due to sleep loss, which was found in the same set of experiments (Schmid et al., 2011). This assumption is supported by recent work linking both overt and subclinical hyperthyroidism to insulin resistance and increased lipolysis (Lambadiari et al., 2011; Maratou et al., 2010; Mitrou et al., 2010). However, considering that, at least in men, thyroid hormone concentrations apparently adapt to sleep restriction within a relative short period of time, a major role of thyrotropic axis alterations as a pathophysiological mediator between sleep loss and metabolic disorders seems unlikely (Penev, 2012).

In summary, we demonstrate that moderate short-term sleep restriction induces a biphasic response of the thyrotropic axis that is characterized by short-term up-regulation followed by a delayed and probably feedback-driven down-regulation. While these endocrine changes might contribute to the aversive metabolic effects of short-term sleep restriction (Schmid et al., 2011; Spiegel et al., 2004), their impact under conditions of chronic sleep loss remains to be established.


We are grateful to Elisa Gustke, Claudia Frenzel, Jutta Schwanbohm and Kathleen Kurwahn for their expert and invaluable laboratory assistance. This work was supported by the Deutsche Forschungsgemeinschaft, SFB 654 ‘Plasticity & Sleep’. Funding from the Deutsche Forschungsgemeinschaft had no influence on design and conduct of the study; collection, management, analysis and interpretation of the data; and preparation, review or approval of the manuscript.

Disclosure Statement

None of the authors report financial or other disclosures.

Author Contributions

SMS, MH, KJS and BS designed the study. SMS, MH and BS analysed the data. SMS, MH, KJS, MCK, HL and BS contributed to writing the paper. SMS and MCK collected data or conducted experiments for the study. All authors had full access to all data and take responsibility for the integrity and accuracy of the data analysis.