Acute Sleep Disruption Does Not Diminish Pulsatile Growth Hormone Secretion in Pubertal Children

Abstract

Context

In children, growth hormone (GH) pulses occur after sleep onset in association with slow-wave sleep (SWS). There have been no studies in children to quantify the effect of disrupted sleep on GH secretion.

Objective

This study aimed to investigate the effect of acute sleep disruption on GH secretion in pubertal children.

Methods

Fourteen healthy individuals (aged 11.3-14.1 years) were randomly assigned to 2 overnight polysomnographic studies, 1 with and 1 without SWS disruption via auditory stimuli, with frequent blood sampling to measure GH.

Results

Auditory stimuli delivered during the disrupted sleep night caused a 40.0 ± 7.8% decrease in SWS. On SWS-disrupted sleep nights, the rate of GH pulses during N2 sleep was significantly lower than during SWS (IRR = 0.56; 95% CI, 0.32-0.97). There were no differences in GH pulse rates during the various sleep stages or wakefulness in disrupted compared with undisrupted sleep nights. SWS disruption had no effect on GH pulse amplitude and frequency or basal GH secretion.

Conclusion

In pubertal children, GH pulses were temporally associated with episodes of SWS. Acute disruption of sleep via auditory tones during SWS did not alter GH secretion. These results indicate that SWS may not be a direct stimulus of GH secretion.

During childhood, growth hormone (GH) plays a critical role in glucose homeostasis and linear growth. GH also contributes to bone mineral density, lean body mass, lipoprotein metabolism, and cardiovascular health (reviewed in [1]) across the lifespan. The neuroendocrine control of GH secretion is complex and involves the interplay of multiple neurotransmitters and hormones, including growth hormone–releasing hormone (GHRH), somatostatin, ghrelin, catecholamines, dopamine, and excitatory amino acids [2].

Sleep is also a powerful stimulus for GH secretion. In the 1960s, Takahashi et al [3] reported an association between the onset of deep sleep (also known as slow-wave sleep; SWS) and GH pulse initiation in a small group of adult men; this temporal relationship was confirmed in more recent studies with frequent (every 30-second) blood sampling in healthy men that showed that maximal GH release occurs within 5 minutes of the onset of SWS [4]. Sleep-associated GH pulses also occur in women, but GH in women is predominantly secreted during the daytime [5, 6]. Studies in children have similarly demonstrated GH secretory episodes during bouts of SWS [7], suggesting that sleep-state specific augmentation of GH secretion is also a feature of normal development.

While the temporal coincidence of bouts of SWS and GH pulses has now been firmly established, the few studies that have investigated the possibility of a functional relationship between SWS and GH secretion have been limited to male adults and have produced mixed results. For example, while pharmacological augmentation of SWS with the gamma-aminobutyric acid (GABA)-agonist gamma-hydroxybutyrate (GHB) caused a proportionate and contemporaneous increase in GH levels [8], SWS deprivation studies in adults have not consistently produced deficits in GH secretion [9–12].

Given the rising incidence of restricted sleep among adolescents [13] as well as sleep disorders such as obstructive sleep apnea [14] that cause sleep disruption, it is critical to determine whether sleep disruption during childhood may dysregulate the somatotropic axis and possibly alter linear growth and body composition. To investigate whether undisrupted sleep is critical for GH secretion in children, we studied healthy children with and without SWS disruption using controlled auditory stimuli.

Materials and Methods

Participants

The 14 participants were pubertal children (7 male, 7 female), aged 11.3 to 14.1 years, whose clinical characteristics and reproductive and metabolic profiles have been previously reported [15, 16]. Participants were euthyroid and were not on any medication known to interfere with sleep, growth, or puberty. The participants had no known sleep disorder and underwent 2 sleep studies spaced 2 months apart, 1 with and 1 without SWS disruption, in random order (28 studies in total). GH secretion was a secondary outcome in these studies that were primarily powered to detect a statistically significant change in luteinizing hormone secretion after sleep disruption.

The studies were approved by the Partners Human Research Committee at Massachusetts General Hospital. Signed informed assent and consent was obtained from each participant and one parent, respectively. These studies were not registered at clinicaltrials.gov because they do not meet the definition of an “applicable clinical trial.”

Experimental Protocol

All overnight studies were conducted at the Clinical Research Center of the Massachusetts General Hospital. They included frequent blood sampling and polysomnography (PSG) [15–17]. PSG was performed according to standard methodology using an electroencephalogram (EEG; total of 6 frontal, central, and occipital leads), electro-oculogram, electrocardiogram, and pulse oximetry recordings (ALICE LE PSG system, Sleepware software, Phillips Respironics). An intravenous catheter with a long line was inserted on admission to allow for blood sampling from outside the room while the individual slept. All participants ate dinner before lights out. Caffeine was prohibited. PSG recording began before lights out and continued until spontaneous awakening the following morning. Lights were turned off between 9 and 10:30 Pm, based on participant and parent reports of habitual bedtime. Blood samples were drawn every 10 minutes for 8 hours starting at or just before sleep onset. During SWS disruption nights, auditory stimuli (3 second, 1500-Hz tones, 40-100 dB followed by 18 seconds of a 75-dB noise simulating a knock on the door) were delivered via a bedside speaker whenever the individual entered SWS (defined as ≥ 2 delta waves on EEG in a 15-second recording interval) as determined by a registered polysomnographic sleep technician (rPSGT).

All samples were analyzed for GH using a quantitative sandwich enzyme immunoassay (Quantikine, R&D Systems, RRID: AB_2923238) that uses antibodies raised against the full-length 22-kDa GH isoform. The minimum detectable concentration was 0.64 pg/mL with an interassay coefficient of variation of 7.1% for quality control serum containing 1.3 ng/mL.

Data Analysis

PSG recordings were visually scored by an rPSGT according to American Academy of Sleep Medicine criteria [18] in 30-second epochs as stages of non–rapid eye movement (REM) ([N1, N2, and N3 (ie, SWS)]), REM, or Wake. Sleep onset was defined as the first appearance of 2 consecutive 30-second sleep epochs after Wake.

The basal secretory rates and timing and amplitude of individual GH pulses were determined using a sparse deconvolution method that assumes a second-order pharmacokinetic model for GH dynamics [19].

Four statistical analyses were performed: (i) Differences in sleep stages and GH secretion between the 2 study nights (with vs without SWS disruption) were compared using paired t tests or the Wilcoxon signed rank test. (ii) The associations of age, sex, and body mass index (BMI) percentile to GH pulse frequency and amplitude was determined using analysis of variance (sex) or Pearson correlation coefficients (age, BMI percentile). (iii) The relationship between GH pulse rate per hour and sleep stage was evaluated using Poisson generalized estimating equations with an unstructured covariance matrix for undisrupted and disrupted sleep nights separately to account for the repeated nature of the observations. These models were estimated using pulse count as the outcome and included an offset for time in each sleep stage to produce incidence rate ratio (IRR) estimates. Due to the small sample size, these models were not adjusted for potential confounders. (iv) The effect of SWS disruption on the relationship between GH pulse onset and sleep stage was determined for each sleep stage or during wake after sleep onset (WASO) using Poisson generalized estimating equation models restricted to each respective sleep stage. Data are expressed as mean ± SE unless otherwise indicated, and P less than .05 was considered significant.

Results

Baseline Characteristics

Seven pubertal girls (mean age 12.5 years ± 0.6 SD; Tanner II-IV breasts; all premenarchal) and 7 pubertal boys (mean age 12.9 years ± 1.0 SD; testicular volumes 4-15 cc) participated (Table 1). The individuals were predominantly (71.4%) non-Hispanic White and 28.6% were overweight or obese. There were no changes in participants’ Tanner stages or BMI percentiles between study visits.

Table 1.

Participant demographics

Participant	Age	Sex	BMI	BMI percentilea	Pubertal stageb
1	11.6	F	16.5	29	III
2	12.3	F	19.2	70	II
3	12.3	F	19.8	73	III
4	12.3	F	17.9	44	III
5	12.5	F	18.6	53	III/IV
6	13.2	F	25.0	94	III
7	13.4	F	20.5	68	III
8	11.3	M	26.1	97	5 cc
9	12.0	M	22.9	92	4 cc
10	12.2	M	29.3	97	8 cc/6 cc
11	13.4	M	17.9	37	15cc
12	13.4	M	17.9	37	15 cc
13	13.8	M	18.7	46	15 cc
14	14.1	M	21.6	78	15 cc

Abbreviations: BMI, body mass index; F, female; M, male.

^a Age- and sex-adjusted BMI percentile greater than 85 is classified as overweight and greater than 95 is classified as obese in children.

^b Tanner breast stage or testicular volume. All girls were premenarchal.

Effect of Slow-Wave Sleep (N3) Disruption on Sleep Stages

An average of 77.5 ± 13.6 (range, 23-190) auditory stimuli were delivered to interrupt SWS (N3) during the night of sleep disruption (n = 14 studies). By design, this intervention caused a 40.0 ± 7.8% decrease in the amount of time spent in SWS (N3) and a 30.7 ± 7.4% increase in the amount of time spent in lighter sleep stages (N1 + N2) relative to the night of undisrupted sleep, whereas there was no change in percentage WASO or percentage REM sleep (Figs. 1 and 2).

Figure 1.

Percentage of time spent in each sleep stage in studies with or without slow-wave sleep (SWS) disruption. REM, rapid eye movement; WASO, wake after sleep onset. Box plots indicate median, interquartile range, minimum, and maximum.

Figure 2.

Polysomnogram (top) and pulsatile growth hormone (GH) secretion (bottom) in 2 individuals, A and B, studied with (left) or without (right) slow-wave sleep (SWS) disruption. GH levels represent model estimates. Vertical lines with circles indicate model-estimated timing and amplitude of GH pulses.

Pulsatile Growth Hormone Secretion During Sleep and Effects of Slow-Wave Sleep Disruption

During undisrupted sleep nights, participants had 7.36 ± 0.58 GH pulses during the 8-hour sampling interval with an average pulse amplitude of 10.55 ± 1.22 ng/mL and average basal secretion rate of 0.29 ± 0.14 ng/mL per min. During SWS-disrupted sleep nights, participants had 6.29 ± 0.54 pulses with an average pulse amplitude of 9.69 ± 1.08 ng/mL and average basal secretion rate of 0.39 ± 0.11 ng/mL per minute. There were no differences in GH secretory parameters in participants studied with vs without SWS disruption (Fig. 3). Participant age (undisrupted r = 0.30, P = .29; disrupted r = 0.36, P = .21), sex (undisrupted F = 0.78, degree of freedom [df] = 1, P = .39; disrupted F = 0.10, df = 1, P = .76), and BMI percentile (undisrupted r = −0.003, P = .99; disrupted r = −0.15, P = .61) were not related to GH pulse frequency in the undisrupted or SWS-disrupted studies. Age (undisrupted r = 0.44, P = .11; disrupted r = 0.3, P = .29) and BMI percentile (undisrupted r = 0.19, P = .53; disrupted r = −0.06, P = .84) were also not related to average GH pulse amplitude. Sex was significantly related to average GH pulse amplitude on undisrupted (F = 12.3, df = 1, P = .004) but not SWS-disrupted nights (F = 2.35, df= 1, P = .15), with boys having a higher average GH pulse amplitude (15.29 ng/mL; 95% CI, 8.30-28.18) than girls (3.58 ng/mL; 95% CI, 1.94-6.59) on the undisrupted nights.

Figure 3.

Growth hormone (GH) secretory parameters in participants studied with vs without slow-wave sleep (SWS) disruption. Box plots indicate median, interquartile range, minimum, and maximum values for GH pulse frequency.

Growth Hormone Secretion and Relationship to Sleep Stage

On nights without SWS disruption, GH pulses occurred at a rate of 1.4 (95% CI, 0.82-2.40) pulses/hour of SWS, which was not significantly different from during any other sleep stages or bouts of WASO (all P > .05) (Table 2). During SWS disruption nights, GH pulses occurred at a rate of 1.14 (95% CI, 0.82-1.59) pulses/hour of SWS, but this was only statistically significant in comparison to the rate during N2 sleep (IRR = 0.56; 95% CI, 0.32-0.97) (see Table 2). SWS disruption nights (compared with undisrupted nights) did not have different GH pulse rates during the various sleep stages or WASO (Table 3).

Table 2.

Results of generalized estimating equation Poisson models for growth hormone pulse frequency as a function of sleep stage or wake after sleep onset during “undisrupted sleep” and “slow-wave sleep disrupted sleep” studies

Sleep stage	Undisrupted sleep			SWS disrupted sleep
	Incidence rate ratios	CI	P	Incidence rate ratios	CI	P
Intercept	1.40	0.82-2.40	.22	1.14	0.82-1.59	.42
N1 sleep stage	0.59	0.30-1.15	.12	0.71	0.31-1.64	.42
N2 sleep stage	0.49	0.21-1.11	.09	0.56	0.32-0.97	.04
REM sleep stage	0.58	0.32-1.06	.08	1.07	0.78-1.46	.69
WASO	0.32	0.09-1.13	.08	0.66	0.36-1.20	.17

The intercept represents the GH pulse frequency per hour of SWS (ie, N3). The other estimates in the table represent incident rate ratios of the GH pulse rate during that sleep stage in comparison to SWS. For example, on the undisrupted night, the GH pulse rate during N1 sleep is 0.59 times the rate of GH pulses during SWS. P-values < 0.05 are in bold.

Abbreviations: GH, growth hormone; REM, rapid eye movement; SWS, slow-wave sleep; WASO, wake after sleep onset.

Table 3.

Tests of sleep disruption's effect on the growth hormone pulse rate for each specific sleep stage and wake after sleep onset in the minimally adjusted models, shown with incidence rate ratios comparing the disrupted night to the undisrupted night and corresponding 95% CI

Sleep stage	IRR (95% CI)	P
SWS	0.71 (0.42-1.23)	.22
N1	1.13 (0.47-2.69)	.79
N2	1.02 (0.71-1.46)	.92
REM	1.02 (0.65-1.58)	.95
WASO	0.86 (0.36-2.01)	.72

Abbreviations: IRR, incidence rate ratio; REM, rapid eye movement; SWS, slow-wave sleep; WASO, wake after sleep onset.

Discussion

The present studies demonstrate that substantial SWS disruption during one night does not diminish GH secretion in pubertal children; a 40% reduction in SWS did not affect basal or pulsatile GH amplitude or frequency in boys or girls. While this study is not without limitations (discussed later), this finding was unexpected given the traditional view that unperturbed SWS (ie, deep sleep) is an important physiologic stimulus for GH secretion. Indeed, clinical research studies dating back to the 1960s have demonstrated that there is a large amount of GH secreted during the first entry into SWS following sleep onset and that the GH peak is delayed when sleep onset is delayed [3, 12, 20, 21]. Further, more recent sophisticated frequent (every 30-second) sampling studies with deconvolution of GH secretion and simultaneous PSG measurement [4] confirmed the temporal association between GH pulsatile secretion and episodes of SWS during undisrupted sleep in adults.

The large body of literature reporting a coincidence of GH secretion with SWS has led to the assumption that SWS is a direct stimulus of GH secretion. This concept has been well received perhaps because it is consistent with the hypothesis that one of sleep's major functions is to facilitate anabolic processes such as linear growth. However, studies designed to demonstrate a functional relationship between SWS and GH secretion have produced mixed results. In relatively small studies conducted in men (n = 2-10 per study) [9–12], SWS deprivation using either auditory or electrical stimuli did not consistently produce a decrement in GH secretion compared with baseline nor an increase in GH secretion during recovery nights, a time when there is typically a rebound in the amount of SWS. The present studies were also limited by a relatively small sample size, which prevented the adjustment for potential confounders, with heterogeneous characteristics (ie, boys and girls, different stages of puberty), and the study may have been underpowered to detect a small change in GH parameters since GH secretion was not the primary outcome of the study. In addition, the study design did not include an adaptation night and did not prevent all SWS, rather it considerably shortened each episode of SWS. Given these limitations and the exploratory nature of this study, further investigation in a larger population is needed to confirm our findings.

Conversely, pharmacologic augmentation of SWS with GHB, a metabolite of GABA, caused a proportional increase in GH secretion, raising the possibility that SWS directly stimulates GH secretion [8]. However, as noted by those authors, GHB acts on multiple neurotransmitter systems (GABAergic, cholinergic, serotoninergic, and dopaminergic) and thus, it is possible that co-induction of SWS and GH secretion occurred through 2, independent neural pathways. The observation that aging is associated with proportional declines in SWS and GH secretion has also been put forth as evidence of a functional relationship between SWS and GH [22]. However, as noted by Feinberg [23], this relationship does not hold true during adolescence when there is an inverse relationship between GH and SWS: GH secretion is highest during puberty whereas there is a dramatic decline in delta frequency power in the EEG during puberty (large amplitude delta frequency waves in the EEG are a component of SWS). Finally, Van Cauter et al [24] reported that the longer an episode of SWS, the more likely it was to be associated with GH secretion. However, this analysis did not control for the greater opportunity for GH secretion to occur over a longer duration of time, regardless of the sleep stage.

The ability of the present and previous studies to dissociate GH secretion from SWS suggests that SWS itself may not directly stimulate GH secretion. Rather, there may be a common upstream neural or hormonal signal that can, on occasion, simultaneously induce SWS and GH secretion. Therefore, interventions to disrupt SWS, as in the present protocol, would be acting downstream of this signal and would be incapable of interfering with signal transduction to the somatotrophs. There is anatomic and physiologic evidence from the rodent to suggest that GHRH is the signal that temporally links SWS and GH: GHRH neurons in the paraventricular nucleus increase GHRH expression in response to sleep deprivation [25]; GHRH directly activates GABA-ergic, sleep-active neurons in the preoptic area [26]; and intrapreoptic injection of GHRH or a GHRH antagonist to rats increases or decreases SWS, respectively [27]. GHRH neurons in the arcuate nucleus provide the primary input to the median eminence [28], driving pituitary GH secretion. Thus, one possibility is that GHRH neurons in the arcuate and paraventricular nucleus can, at times, coordinate GHRH secretion, resulting in the simultaneous induction of SWS and GH secretion.

An alternative interpretation of the present findings is that SWS does, in fact, stimulate pulsatile GH secretion but that there is no specific requirement that SWS be consolidated into discrete, long bouts, as typically occurs during normal sleep. That is, brief, fragmented bouts of SWS may be sufficient to trigger GH secretion during childhood. This hypothesis would predict that children with disturbed sleep from internal (eg, untreated sleep apnea) or external (eg, noise pollution) reasons should have normal linear growth; further studies are necessary to confirm that SWS disruption, when present chronically, does not diminish GH secretion during childhood. The present studies, however, demonstrate that SWS disruption, when it occurs acutely via auditory stimuli, does not interfere with normal pulsatile GH secretion in pubertal children.

Abbreviations

BMI

body mass index

EEG

electroencephalogram

GH

growth hormone

GHB

gamma-hydroxybutyrate

GHRH

growth hormone–releasing hormone

IRR

incidence rate ratio

PSG

polysomnography

REM

rapid eye movement

rPSGT

registered polysomnographic sleep technician

SWS

slow-wave sleep

WASO

wake after sleep onset

Financial Support

This work was supported, in part, by the Intramural Research Program of the National Institutes of Health, National Institute of Environmental Health Sciences (Z01-ES103315). N.D.S. is also supported as a Lasker Clinical Research Scholar (1SI2ES025429-01). She also received extramural support from the NIH (grant K23HD073304–02) and support from the Pediatric Endocrine Society and Harvard Catalyst (The Harvard Clinical and Translational Science Center [Award 1UL1TR001102-01]) to conduct these studies. E.B.K. is supported by U54AG062322 and the Leducq Foundation. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Center for Research Resources, the National Center for Advancing Translational Science, or the National Institutes of Health. R.T.F. received funding from the US National Science Foundation (CAREER Award 1942585) and the US National Science Foundation (CRII Award 1755780).

Disclosures

E.B.K. reports consulting with Circadian Therapeutics, The National Sleep Foundation, the American Academy of Sleep Medicine Foundation, the Sleep Research Society Foundation, Sanofi-Genzyme, and Yale University Press; her partner owns Chronsulting. The other authors have nothing to disclose.

Data Availability

Some or all data sets generated during and/or analyzed during the present study are not publicly available but are available from the corresponding author on reasonable request.