Changes in Serum TSH and Free T4 during Human Sleep Restriction

Abstract

Study Objectives:

To examine whether recurrent sleep restriction is accompanied by changes in measures of thyroid function.

Design:

Two-period crossover intervention study.

Setting:

University clinical research center and sleep laboratory.

Participants:

11 healthy volunteers (5F/6M) with a mean (± SD) age of 39 ± 5 y and BMI 26.5 ± 1.5 kg/m².

Intervention:

Randomized exposure to 14 days of sedentary living with ad libitum food intake and 5.5- vs. 8.5-h overnight sleep opportunity.

Measurements and Results:

Serum thyroid-stimulating hormone (TSH) and free thyroxine (T4) were measured at the end of each intervention. Partial sleep restriction was accompanied by a modest but statistically significant reduction in TSH and free T4, seen mainly in the female participants of the study.

Conclusions:

Compared to the well-known rise in TSH and thyroid hormone concentrations during acute sleep loss, tests obtained after 14 days of partial sleep restriction did not show a similar activation of the human thyroid axis.

Citation:

Kessler L; Nedeltcheva A; Imperial J; Penev PD. Changes in serum TSH and free T4 during human sleep restriction. SLEEP 2010;33(8):1115-1118.

THE THYROID HORMONES THYROXINE (T4) AND TRIIODOTHYRONINE (T3) EXERT PHYSIOLOGIC EFFECTS IN ALL TISSUES AFFECTING OXYGEN CONSUMPTION and protein, carbohydrate, lipid, hormone, growth factor, and vitamin metabolism. The release of T4 and T3 by the thyroid is controlled by the activity of the hypothalamic thyrotropin-releasing hormone (TRH) and pituitary thyroid-stimulating hormone (TSH) cascade. At the same time, circulating T4 and T3 exert negative feedback on the activity of this pathway in a classic endocrine fashion.

Today many people report sleeping less than 6 h/day, and such short sleep duration has been associated with disorders of fuel and energy metabolism.¹ Several rodent studies have also established the existence of links between sleep, energy homeostasis, and thyroid function.^2–4 Sleep deprivation in the rat results in increased energy expenditure, weight loss, increased food intake, and initial elevation of body temperature. Despite the similarity of these findings to the signs of hyperthyroidism, sleep deprived animals show biochemical features of central hypothyroxinemia, characterized by reduced circulating concentrations of total and free T4 and T3 without an increase in TSH.^2,4,5

Sleep loss can also affect the function of the human hypothalamo-pituitary-thyroid axis. In contrast to the effects of sleep deprivation in rodents, acute sleep loss in humans is associated with increased TSH, T4, and T3,^6,7 and human sleep is believed to have an acute inhibitory effect on overnight TSH secretion.⁸ Previous human sleep deprivation studies have been acute in nature (lasted < 7 days) and reduced sleep to less than 4 h/day, which is not common even in chronic short sleepers most of whom report average sleep times of 5 to 6 h/day. Whether or not exposure to less severe but more prolonged sleep restriction, which approximates the short sleep times experienced by many individuals in everyday life settings, can modify the function of the human thyroid axis has not been studied. To address this possibility, we used a set of serum samples, collected during an earlier study of the effects of sleep loss on human energy metabolism,⁹ to compare the TSH and free T4 concentrations of healthy middle-aged adults whose sleep opportunity was restricted to time-in-bed of 5.5 vs. 8.5 h/night for 2 weeks on 2 separate occasions.

METHODS

Six men and 5 women with a mean (± SD) age 39 ± 5y, BMI 26.5 ± 1.5 kg/m², and self-reported sleep duration of 7.6 ± 0.7 h/day completed the study. Research volunteers gave written informed consent and were paid for their participation. The study protocol was approved by the Institutional Review Board of the University of Chicago. The inclusion and exclusion criteria, and the experimental design of the study have been published⁹ and are described in detail in the Electronic Supplement, which accompanies this paper. Briefly, each subject completed two 14-day intervention periods with sedentary activity, ad libitum food intake, and scheduled time-in-bed of 5.5 or 8.5 h/night in random order at least 3 months apart. Nightly sleep was recorded polysomnographically and scored in 30-sec epochs according to standard criteria.¹⁰ At the end of each intervention period, participants remained at bed rest for 48 hours with identical caloric intake including oral (day 1; 75 g) or intravenous (day 2; 0.3 g/kg) doses of glucose at 09:00 and identical weight-maintenance carbohydrate-rich meals (62% carbohydrate, 15% protein, and 23% fat) at 14:00 and 19:00. During the last 24 h of this period, blood was sampled every 30 min starting at 20:00. Time-in-bed was 5.5 or 8.5 h/night according to the assigned sleep condition. All data collection in women occurred during the first half of their menstrual cycle.

TSH and free T4 were measured using Immulite 2000 human assays (Siemens Diagnostics, Deerfield, IL) in serum collected at 30 and 120-min intervals, respectively. TSH assay sensitivity was 0.004 uIU/mL with intra- and inter-assay variability of 4.3% and 4.9%. Free T4 assay sensitivity was 0.3 ng/dL with intra- and inter-assay variability of 7.3% and 7.7%. The hormone measurements obtained at regular intervals in each participant at the end of each intervention were averaged to calculate her/his 24-h concentrations of TSH and free T4. In addition to comparing the corresponding 24-h mean TSH and free T4 concentrations at the end of each bedtime condition, we also analyzed their mean daytime (08:00-23:00) and overnight (23:00-08:00) concentrations. In 6 subjects (1F/5M), who had sufficient residual serum from 3-4 pooled fasting morning samples during each intervention, we also measured hormone protein binding and reverse T3. Since short sleep is an increasingly common aspect of the Westernized lifestyle, these experiments were carried out under sedentary conditions with ad libitum access to palatable food (see Electronic Supplement) and, as previously reported,⁹ the energy intake of the subjects exceeded their energy expenditure during both sleep conditions. Because changes in energy balance can affect human thyroid function, we used generalized-estimating-equations-based regression to assess the effect of assigned time-in-bed (5.5 vs. 8.5 h/night) on serum TSH and thyroid hormone concentrations while controlling for the change in body weight during each intervention (expressed as percent of pre-treatment baseline) and order-of-treatment (initial vs. crossover) as time-varying covariates. P-values for the effect of time-in-bed of less than 0.05 were considered statistically significant. All analyses were performed using SPSS Version 16.0 (SPSS Inc., Chicago, IL) and results are reported as group mean ± SD.

RESULTS

As previously reported, subjects went to bed later, 00:31 vs. 23:16 (± 39 min), and got out of bed earlier, 06:02 vs. 07:42 (± 39 min), when assigned time-in-bed changed from 8.5 to 5.5 h.⁹ Mean sleep duration between the two 14-day periods was reduced by 122 ± 25 min from 7 h 13 min ± 26 min to 5 h 11 min ± 7 min (P < 0.01).⁹

Mean daytime, overnight, and 24-h TSH concentrations of the 11 study participants were lower at the end of the 5.5-h time-in-bed condition (Figure 1A). This was due mainly to the decline in TSH concentrations in female subjects (Figure 1B), whereas the observed changes in men were small and not statistically significant (Figure 1C). Independent of these sleep related differences, the 24-h TSH concentrations of men vs. women were not significantly different (P = 0.66).

Figure 1.

Mean (± SE) 24-h profiles of serum TSH at the end of the 8.5-h (open circles) and 5.5-h (solid circles) bedtime condition in the entire group (A; n = 11), and in the subgroup of female (B; n = 5) and male (C; n = 6) study participants. Summary data are group means (± SD). P-values reflect the effect of time-in-bed (5.5 vs. 8.5-h) based on a generalized estimating equations regression model with order-of-treatment and 14-day change in body weight as time-varying covariates.

Serum free T4 did not exhibit 24-h variability. Mean 24-h free T4 concentrations were lower at the end of the 5.5-h than the 8.5-h bedtime condition (1.03 ± 0.04 vs. 1.12 ± 0.04 mcg/dL; P = 0.003; n = 11). When subjects were divided by gender, female study participants had higher 24-h free T4 concentrations (P = 0.03) than males. Short-sleep related declines in free T4 were seen in both men (0.98 ± 0.06 vs. 1.06 ± 0.05 mcg/dL; P = 0.14) and women (1.10 ± 0.03 vs. 1.19 ± 0.05 mcg/dL; P < 0.001), but the results reached statistical significance only in the female subgroup.

There were no significant differences in reverse T3 (176 ± 46 vs. 178 ± 64 pg/mL; P = 0.92) and protein binding (resin uptake ratio: 1.33 ± 0.11 vs. 1.31 ± 0.12; P = 0.73) at the end of the 5.5- and 8.5-h time-in-bed condition. The reverse T3 and protein binding data of the single female subject in this analysis did not differ from those of the 5 men.

DISCUSSION

Using a protocol of recurrent restriction of the overnight time-in-bed, we modified the sleep duration of our study participants from over 7 h/day, which in epidemiological studies corresponds to the sleep category with lowest morbidity and mortality, to less than 6 h/day, which falls in a sleep category associated with increased risk of chronic illness.¹ In contrast to the increase in TSH and thyroid hormone levels in studies of acute sleep loss,^6–8 in the present experiment bedtime restriction was accompanied by modestly reduced TSH and free T4 concentrations. These changes were readily apparent in female study participants, whereas those seen in men did not reach statistical significance.

Previous studies have not specifically examined gender as a modifier of the impact of sleep loss on the human thyroid axis. Indeed, acute sleep deprivation has been shown to have comparable effects on TSH and thyroid hormone concentrations in both men and women,^6–8 and our brief report is the first to highlight the potential for such gender-based differences. We studied women during the follicular phase of their menstrual cycle, which is a time of heightened pituitary sensitivity to TRH.¹¹ Furthermore, rodent data indicate that sleep deprivation is accompanied by reduced hypothalamic TRH mRNA and suggest that sleep loss can interfere with the synthesis and secretion of the peptide and its ability to stimulate the pituitary.^2,5 Taken together, these observations raise the possibility that decreased release and/or action of hypothalamic TRH may potentiate the effect of recurrent sleep restriction on TSH and free T4 in women.

The mechanisms of sleep-loss-related changes in human thyroid function have not been elucidated. Type II 5'-deiodinase (5'D-II) activity is a major determinant of T3 concentrations in rodent and human pituitary and upregulation of brain 5'D-II may contribute to the lack of increased TSH release in response to hypothyroxinemia. Thus, despite the marked reduction in serum T4 in sleep deprived rats, maintenance of relatively stable T3 levels in their pituitary may cause the appearance of inappropriately normal TSH concentrations.^2,5 Such changes have been described in the setting of uremia, another catabolic state in the rat, where TSH and pituitary nuclear T3 were normal when serum thyroid hormone levels were reduced.¹² These data raise the possibility that the decrease in serum free T4 during recurrent partial sleep loss may reduce the catabolic threat to peripheral tissues posed by the extended metabolic demands of the waking brain. In agreement with this hypothesis, the magnitude of the observed changes in thyroid function (very large vs. more modest) seems to match the severity of the catabolic effect of sleep loss in rodents vs. humans.^3,9

An alternative explanation for our findings is based on the well-documented disinhibitory effect of acute sleep loss on TSH release, which is accompanied by a rise in thyroid hormone levels.^6–8 A previous study of bedtime restriction to 4 h/night for 6 days found elevated T4 concentrations and decreased TSH.¹³ It was hypothesized that this was the result of negative feedback inhibition of the pituitary by the acute rise in TSH and thyroid hormone levels during the first few days of experimental sleep deprivation. If this is correct, 14 days may not be enough for pituitary thyrotroph function to recover from such inhibition and the ongoing reduction in TSH release may explain the decrease in free T4 at the end of our 5.5-h bedtime condition. Thus, given enough time, both TSH and free T4 may return to their pre-exposure levels. If our current findings are confirmed by studies with larger sample size, longer sleep restriction experiments will be needed to determine whether the observed changes in TSH and free T4 represent a step in the normal process of re-equilibration after a transient shift due to acute sleep loss, or the result of central changes in the hypothalamo-pituitary-thyroid axis similar to those seen in sleep deprived rodents. Furthermore, this experiment did not examine the potential impact of extended exposure to indoor lighting on the 24-h melatonin profiles of the participants during the 5.5-h time-in-bed condition, and future studies should consider the effects of such photoperiodic signals on the human hypothalamo-pituitary-thyroid axis.¹⁴

Importantly, the thyroid profiles of sleep deprived rats and our study participants are distinct from the euthyroid sick state, which is characterized by central suppression of TSH release, low T4 levels associated with changes in T4 binding, decreased T4 to T3 conversion with reciprocal rise in reverse T3, and increased concentrations of serum corticosteroids. In sleep deprived rats, TRH administration produces a normal rise in TSH and appropriate increase in circulating T4 and T3 concentrations, reverse T3 concentrations are not increased to indicate T4 inactivation, and low T4 cannot be explained by changes in protein binding.^4,5 Likewise, in the present study there were no significant differences in protein binding and reverse T3 between the two bedtime conditions; however, more such data in women are needed. As already reported,¹⁵ sleep loss was not accompanied by increased amplitude, peak levels, or mean concentrations of the 24-h rhythm of serum cortisol to account for the observed pattern of central hypothyroxinemia. Similarly, since increased sympathetic activity has been associated with stimulation rather than inhibition of the thyroid gland, the previously reported modest increase in plasma epinephrine and norepinephrine concentrations¹⁵ could not explain the reduction in TSH and free T4 during the 5.5-h time-in-bed condition.

In summary, compared to the well-known rise in serum TSH and thyroid hormone concentrations during acute sleep loss, tests obtained after 2 weeks of experimental sleep restriction did not show a similar activation of the human thyroid axis. Instead, partial sleep loss was accompanied by modest but statistically significant declines in TSH and free T4, which were seen mainly in female study participants. These results suggest that recurrent sleep restriction, designed to approximate the short sleep times of a growing number of people in everyday life, can affect the function of the human thyroid axis. Whether or not such changes can contribute to the association of reduced sleep duration with chronic metabolic or psychiatric disorders, especially in women, warrants further exploration.

DISCLOSURE STATEMENT

This was not an industry supported study. The authors have indicated no financial conflicts of interest.

SUPPLEMENTAL METHODS

Participants

Sedentary men and women ages 34 to 49 with a body mass index between 24 and 29 kg/m² and self-reported sleep duration of 6.5 to 8.5 h/day were recruited through local newspaper advertisements and by word of mouth. Volunteers were excluded from participation if they had: self-reported sleep problems (Pittsburgh Sleep Quality Index, PSQI, score > 10), night work, variable sleep habits, or habitual daytime naps; physically demanding occupations or regular exercise; depressed mood (Center for Epidemiologic Studies of Depression, CES-D, score > 15); excessive intake of alcohol (> 14 drinks/week for men; > 7 for women) or caffeine (> 300 mg/day); smoking; use of prescription medications or over-the-counter drugs affecting sleep or metabolism; and abnormal findings on medical history, physical exam, and laboratory screening tests (including a 75 g oral glucose challenge and one night of full polysomnography). Only non-pregnant women were studied and data collection was scheduled during the first half of their menstrual cycle. Eleven subjects (5F/6M) including 5 Caucasian, 4 African American, and 2 Hispanic individuals completed the study.

Study Protocol

Each subject completed two 14-day study periods with scheduled time-in-bed of 5.5 or 8.5 h per night in random order at least 3 months apart. All participants were studied in the controlled environment of the University of Chicago sleep research laboratory, which offers individual accommodations similar to those of a comfortable hotel room with a queen-size bed and has built-in infrastructure for video monitoring and sleep recording. To achieve sleep opportunities with time-in-bed of 5.5 vs. 8.5 h without shifts in circadian phase, the usual lights-off and wake up times of the subjects were moved proportionally closer together or further apart. Six subjects started with the 5.5-h time-in-bed condition, and 5 subjects were studied in the 8.5-h time-in-bed condition first. Participants spent most waking hours indoors engaged in leisure activities or home-office-type work and had free access to a telephone, desktop computer, TV, videos, reading material, and the internet, and were exposed to the same mixed room lighting from incandescent and fluorescent light sources. On average, subjects spent no more than 30 min/day outside of the laboratory on the university campus during the late morning or afternoon hours of the day. No naps were allowed and individual safety and compliance were monitored continuously by our research staff.

Energy Intake

A registered dietitian interviewed all participants to determine their food preferences and exclude the presence of any eating disorders, and developed nutritionally balanced individual meal plans consisting of breakfast, lunch, dinner, and a selection of palatable snacks and soft drinks. During each bedtime condition, subjects received the same customized 3-day meal sequence including typical Western foods on a rotating basis. Breakfast was served at 08:00 to 09:00 and included such items as eggs, bacon or sausage; toast, bagel, pancakes or waffles; jelly, peanut butter, or cream cheese; cereal, fruit, yogurt, juice, milk, coffee and tea. Lunch was served at 13:00 to 14:00 and could be a hot or cold entrée (e.g., hot or cold sandwich, pizza, pasta, or meat, a vegetable, and starch) along with soup or salad, soft drink, and dessert. Dinner was served at 18:30 to 19:30 and was usually a hot meat, poultry, or fish entrée with a vegetable and starch, in addition to salad, non-caffeinated beverage, and dessert. Subjects were allowed one caffeinated beverage with breakfast and one with lunch as needed to match their usual consumption of caffeine at home. Meals were served in excess to allow ad lib intake of energy. Food was weighed before and after each meal to determine actual consumption. In addition, study participants had unlimited access to a snack bar in their room, which was kept stocked with soft drinks and the same individually customized assortment of 10 snacks during each study period. The snacks included salty snacks (e.g., pretzels, chips and dip, cheese and crackers, popcorn, nuts), sweets (e.g., snack bars, muffins, cookies, pudding, ice cream, candy), fresh and dried fruits, yogurt, raw vegetables and dip, and non-caffeinated beverages (e.g., milk, juice, soda, water). Items consumed from the snack bar were weighed and disappearance from the inventory was recorded twice every 24 hours.

Energy Expenditure

For the measurement of total energy expenditure, study participants ingested ¹⁸O and ²H-labeled water (2.0 and 0.14 g/kg total body water, respectively) on the first morning of each bedtime condition and their next 3 urine voids were collected. Two more urine voids were collected in the morning of day 14. Samples were analyzed by isotope ratio mass spectrometry at the University of Wisconsin in Madison. Resting metabolic rate was measured under basal conditions by indirect calorimetry (Vmax Encore 29, Sensormedics, Yorba Linda, CA) after awakening on day 14 of each intervention. The physical activity level (PAL) of the subjects was calculated as the ratio between their total energy expenditure and resting metabolic rate. As previously reported, measured PAL values were in the sedentary range of ∼1.5 (25% centile) during both the 5.5 and 8.5-h time-in-bed condition.

Measurement of Body Weight

Fasting body weight was measured with a digital medical scale (Scale-Tronix Inc., Wheaton, IL) in the morning before and after each 14-day intervention. As previously reported, there were no statistically significant differences in total energy intake and expenditure, and weight gain between the two bedtime conditions when treatment order and baseline body weight were controlled for.

Sleep Recording

Sleep was recorded using a Neurofax-1100 EEG Acquisition System (Nihon-Kohden). Before enrollment, subjects had a night of full laboratory polysomnography for habituation and to exclude the presence of primary sleep pathology. Records were scored in 30-sec epochs of wake, movement, stage 1, 2, 3, 4, and REM sleep. Respiratory events, periodic leg movements, and arousals were defined according to established clinical criteria; and subjects with a respiratory disturbance index > 10 or any sleep movement disorder were excluded from participation. Only electroencephalographic, electrooculographic, and electromyographic channels were recorded on subsequent study nights. The data from each subject during both study conditions were scored by the same sleep technician. Total sleep time was calculated as the sum of all epochs scored as sleep. Sleep efficiency was calculated as the percent of scheduled time in bed that was scored as sleep. The sleep results of each participant during each 14-day intervention were averaged to derive her/his average sleep parameters during the 5.5 and 8.5-h time-in-bed condition.