Sleep and Puberty

Abstract

In the 1970’s, Boyar and colleagues made the seminal observation that during the early stages of puberty, there is a sleep-specific augmentation of pulsatile luteinizing hormone (LH) secretion. Building on this tantalizing association between sleep and the re-awakening of the neuro-reproductive axis, a number of investigators have since mapped the dynamic relationship between sleep and reproductive hormones across the pubertal transition. In this review, we focus on the complex, reciprocal relationship between sleep and reproductive hormones during adolescence as well as the potential effects of melatonin and orexin on gonadotropin-releasing hormone (GnRH) activity in children with chronic insomnia and narcolepsy, respectively. Given the important interaction between the reproductive and somatotropic axes during puberty, we end with a discussion of sleep and growth hormone (GH) secretion in children.

Reciprocal interactions between sleep and the reproductive axis during puberty

Sleep/wake state and pulsatile LH secretion in pubertal children

The sleep and reproductive axis are intimately related during pubertal development and in adulthood. Indeed, the dramatic increase in LH levels, a biomarker of the re-activation of the hypothalamic gonadotropin-releasing hormone (GnRH) neuronal network, that signals the onset of puberty in boys and girls is initially restricted to sleep. This observation was made nearly 50 years ago in pioneering studies by Boyar et al. that included nocturnal polysomnography and frequent blood sampling in healthy pubertal boys [1]. His group convincingly demonstrated that the increase in nocturnal luteinizing hormone (LH) was specifically related to sleep and not to time of day or to light exposure using sleep-wake reversal studies and/or blindfolding subjects during wakefulness [2]. More recent studies by our group in pubertal children demonstrated that LH pulses during sleep occur during or immediately following deep (slowwave) sleep [3], suggesting that entrance into deep sleep may stimulate pulsatile GnRH/LH secretion during puberty. Interestingly, a similar temporal association between deep sleep and LH pulse onset was also observed in women with polycystic ovarian syndrome (PCOS) [4]. A single night of deep sleep disruption via auditory stimuli did not diminish LH secretion in pubertal children [5], yet it is unknown whether chronic sleep disruption or sleep restriction may interfere with LH secretion and delay pubertal development. We found that girls with obstructive sleep apnea (OSA) who have abnormal sleep architecture have relatively delayed thelarche compared to girls without OSA [6], however, additional studies are necessary to investigate the impact of poor sleep hygiene on the developing reproductive axis in boys and girls.

Cross-sectional studies in girls suggest that as puberty progresses, LH pulse frequency remains relatively constant during sleep (~ q120min), whereas LH pulse frequency begins to rise during wakefulness until it exceeds that of sleep in late puberty [7, 8]. A recent study by Kim et al. suggests that the dynamic patterning of LH secretion during wakefulness across puberty may relate to greater hypothalamic sensitivity to progesterone (P4) negative feedback during wakefulness compared with sleep [9]. In a randomized, placebo-controlled, crossover study of low-dose oral P4 in 11 early post-menarchal girls, P4 acutely reduced waking LH pulse frequency by 26% but had no effect on LH secretion during sleep. The investigators propose that as androgens increase during the course of puberty, they antagonize P4 negative feedback activity, and this allows LH pulse frequency to gradually increase during wakefulness. A study of exogenous P4 ± androgen receptor blockade with spironolactone is now underway in pubertal girls to test this very hypothesis [10].

A final LH secretory pattern emerges after menarche that continues through the reproductive years: sleep-specific slowing of LH pulse frequency during the early/mid-follicular phase (FP) of the cycle (~q120min during sleep vs. q90min during wake) [11–14]. In contrast to early/mid-puberty, when LH pulses appear to be tied to deep sleep, in adult women, LH pulses are more likely to occur during brief awakenings from sleep whereas deep sleep appears to be inhibitory [15]. It has been proposed that this sleep-related slowing of LH pulse frequency in the FP reflects the waning influence of P4 from the preceding luteal phase [16]. Note that during the luteal phase, under the influence of high P4 levels, LH pulse frequency is slow (~ q4hr in mid-luteal phase) during both wake and sleep. In our study of 23 early post-menarchal girls who had a gradation of endogenous P4 exposure (from high levels in normal ovulatory cycles, to intermediate levels in cycles with evidence of luteal insufficiency, to low levels in anovulatory cycles), we observed a strong inverse correlation between P4 exposure and LH pulse frequency during sleep in the FP [17]. Importantly, however, the observation that sleep-related slowing of LH pulses persists in post-menopausal women [18], suggests that P4 exposure alone may not entirely explain LH pulse patterning during sleep and wake during late puberty and adulthood. Other potential mediators include androgens, insulin sensitivity, and/or energy availability [4, 19, 20].

Changes in sleep patterns during adolescence: do sex steroids play a role?

There are significant changes in sleep/wake patterns during puberty including a delayed sleep phase (tendency for later bedtimes), shorter sleep duration, irregular sleep schedules (sleeping longer during weekends compared to weekdays), and a greater ability to maintain wakefulness, indicating a tolerance to the homeostatic drive for sleep (or “sleep pressure”). Pubertal changes in sleep behavior coincide with a steep decline in deep sleep, and its EEG correlate, delta spectral power, that begins at age 11–12 years and continues until 16.5 years [21]. The decline in delta power not only coincides with the average age of normal puberty, but, akin to sex differences in pubertal timing, occurs approximately 1 year earlier in girls than in boys. These observations have raised the possibility that changes in the sleep EEG and associated sleep behaviors during puberty may be driven by sex steroids.

Indeed, nearly 30 years ago, Carskadon et al. reported an association between advanced pubertal stage and delayed sleep phase in a cross-sectional study of 6^th grade girls [22]. Following up on this early finding, a number of longitudinal cohort studies have investigated the temporal association between pubertal milestones and sleep/wake using a variety of data collection methods. Pubertal status, for example, has been determined by a trained physician (Tanner staging) or by self-report; no studies have included reproductive hormone levels. Sleep measures have been either objective (in-home actigraphy or EEG) or subjective (sleep diary, questionnaire). A study of approximately 100 Israeli school-children, aged 9.9–11.2 years, demonstrated that maturational changes in sleep/wake patterns, including delayed sleep onset and decreased sleep efficiency, occurred before any physical signs of puberty [23], a time when estradiol and testosterone levels are typically undetectable. Of note, this study did not consider pubic hair and genital/breast development separately; rather, the two scores were averaged for analysis. In a longitudinal study of 67 children that included biannual Tanner staging by a physician and EEG studies over the course of 6–7 years, Campbell et al. found that the timing of the decline in delta power was strongly related to the timing of pubic hair development in boys and girls; the relationship to breast and genital development was less robust [24]. Two recent studies of the relationship between pubertal timing and sleep behavior also found an association between earlier pubic hair development and either later bedtimes [25] or shorter sleep duration [26] in girls. Taken together, these studies suggest that adrenal androgens may in fact play a role in sleep maturational changes.

These observational study designs, however, cannot distinguish the effects of pubertal maturation from age per se on sleep. Thus, an alternative approach has been to study sleep architecture and behavior in children with precocious or delayed puberty compared to controls. We conducted in-hospital sleep and hormone tests in girls with central precocious puberty (CPP) and age-matched pre-pubertal controls. We found that premature estrogen exposure was not associated with a premature decline in delta power, and hormonal suppression did not affect the natural decline in delta power over time [27]. Further, there was no association between pubertal status and morningness/eveningness preference (a surrogate for chronotype) or between dehydroepiandrosterone sulfate (DHEAS) or estradiol levels and delta power. Jessen et al. [28] investigated the association between pubertal status, reproductive hormones, and chronotype in girls with early puberty (CPP or premature adrenarche), boys with delayed puberty, and historic age- and sex-matched controls. Girls with earlier puberty tended to have delayed sleep timing whereas boys with delayed puberty had earlier sleep timing compared with controls; there were no group differences in sleep duration. In a combined analysis of boys and girls with pubertal disorders, there was a positive (r=0.5) correlation between 17-hydroxy-progesterone (17OHP) levels and later sleep times but no relationship between estradiol, testosterone, LH, follicle-stimulating hormone (FSH), or DHEAS and sleep preferences. While these studies separated the effects of pubertal hormones and age on sleep, they both suffered from small sample sizes, and in the latter study, control girls were younger than girls with CPP and pubertal status was unknown in controls (assumed to have normal pubertal development for age). Thus, these pilot studies require replication in a larger group of carefully phenotyped children with disorders of pubertal timing to investigate further the relationship between sex steroids, particularly those associated with adrenarche, and sleep during puberty.

The change in delta power during adolescence is posited to be a biomarker of the decline in cortical synaptic density (so-called ‘synaptic pruning’) and cerebral metabolic rate that are part of a profound structural reorganization of the adolescent brain [29]. Recent studies suggest further that this natural rewiring of the cerebral cortex is tightly regulated by an interaction between microglia (the resident macrophages of the brain) and complement proteins [30]. Of note, microglia express estrogen, androgen, and progesterone receptors, although expression varies by developmental stage (reviewed in [31]). While it is possible that changes in sex steroids modulate synaptic pruning by microglia during adolescence, and thereby influence cortical electrophysiology and even sleep behavior, studies in the mouse would suggest that sex differences in neural patterning are programmed much earlier in development (fetal and early postnatal periods) [32–34]. Thus, the potential association between prenatal exposures, such as hormones and the microbiome [35], and adolescent sleep/wake patterns deserves further study.

Sleep and puberty - hot topics

Melatonin as a sleep aid and potential effects on pubertal timing

The sleep-promoting effect of melatonin, a hormone secreted by the pineal gland during darkness, has made it an important treatment option for insomnia among children with neurological disorders (autism spectrum disorder [ASD], ADHD), blindness, or delayed sleep-wake phase disorder [36]. Melatonin’s role in suppressing the reproductive axis during the non-breeding season in mammals [37, 38], however, has raised concerns that long-term treatment may disrupt pubertal maturation in treated children. Cross-sectional studies in children have shown that melatonin levels decrease across puberty [39, 40], are inversely correlated with serum LH levels [40], and are lower in children with CPP than age-matched controls [41]; these associations are consistent with an inhibitory function of melatonin. Studies in the ewe suggest that melatonin acts in the medial basal hypothalamus [38], upstream of kisspeptin neurons (which lack the melatonin receptor 1A), possibly by influencing local triiodothyronine (T3) output by the pituitary pars tuberalis [42].

Very few studies have investigated pubertal timing in children prescribed melatonin for chronic insomnia. In a follow-up study of 41 children (68% male, mean age 9.9 years) with neurodevelopmental disorders who had been treated with melatonin for an average of 4.3 years, the majority of caregivers reported normally-timed puberty with the exception of 5 children with a history of PP that preceded treatment [43]. Malow et al also reported no delays in pubertal development, as determined by a physician using Tanner pubertal staging and by age at menarche, relative to Dutch standards in a group of 31 children with ASD after 2 years of continuous melatonin use [44]. The same group reported no effect of melatonin on serum gonadotropins and sex steroids in a 14-week dose-finding study in 24 children with ASD, however, all participants in this study were pre-pubertal by study design [45]. A follow-up study by Zwart et al. [46] of 33 otherwise healthy children with insomnia (mean age 19 years) who had received melatonin for an average of 10.8 years found that 56% reported normal pubertal timing. However, nearly one-third perceived their pubertal timing to be later than their peers, which was significantly higher than the 17% of participants in a population-based study of nearly 9000 Norwegian students who reported delayed development relative to peers. In addition to self-ratings of puberty, this study was limited by a <50% response rate and subjects were older than controls at the time of the pubertal self-assessment. Taken together, these data provide some reassurance that melatonin is unlikely to harm the developing reproductive system in humans but confirmatory studies in larger cohorts are necessary.

Narcolepsy and central precocious puberty

Narcolepsy is a chronic sleep disorder with an estimated US prevalence of 0.05–0.1% [47]. It is defined by excessive daily sleepiness with or without cataplexy (abrupt muscle weakness) and typically manifests in childhood or adolescence. Narcolepsy is caused by the selective destruction of orexin-producing neurons in the hypothalamus via an interplay of environmental triggers (e.g., H1N1 virus), autoimmunity, and a genetic predisposition [48]. Orexin, or hypocretin, is a neurotransmitter that plays an important role in the regulation of sleep/wakefulness and of multiple neuroendocrine pathways (e.g., stress, arousal, feeding).

Several case reports have also noted an association between pediatric narcolepsy and CPP, suggesting a role for orexin in modulating the neuro-reproductive axis. A 2010 retrospective case series of 51 US children with narcolepsy in fact observed that patients had earlier pubertal onset than their same-sex sibling(s) and parent [49]. A second cohort study of 43 Italian children with narcolepsy observed a high rate of CPP (17%) compared with age-matched obese controls (1.9%) [50]. Boys and girls were affected in equal proportions, in contrast to the female predominance that typifies CPP. Although the onset of narcolepsy symptoms is frequently associated with weight gain, the investigators found that a younger age at narcolepsy symptom onset - not weight gain - was a significant predictor of CPP. In a follow-up study of this cohort (expanded to 72 cases), the investigators again found a high prevalence of CPP (16%; 6 boys, 6 girls) [51] and overweight/obesity. In addition, menarche occurred about 1 year earlier, on average, in the girls who did not meet formal criteria for CPP than in girls in the general Italian population [51]. Lastly, a recent study of 63 consecutive children referred to a French academic hospital for treatment of CPP similarly reported that 14% of children had comorbid narcolepsy [52].

Studies on the effect of orexin on the reproductive axis have produced conflicting results, most likely due to differences among species and sex steroid milieus. However, two recent studies demonstrated that orexin directly inhibits GnRH electrophysiological activity in brain slices from ovariectomized mice [53] and adult zebrafish [54]. A third study in adult male rats demonstrated that interventricular injection of orexin caused a decrease in serum LH and testosterone, sexual behavior, and expression of GnRH, kiss1, and NKb [55], raising the possibility that orexin suppresses both kisspeptin, neurokinin B, and dynorphin (KNDy) and GnRH neuronal activity.

Sleep and growth hormone secretion during puberty

Pulsatile GH secretion by the anterior pituitary gland primarily reflects the balance of stimulation by growth hormone releasing-hormone (GHRH) and inhibition by somatostatin. GH secretion is also regulated by a number of neuro-transmitters/peptides and is sensitive to stress, hypoglycemia, exercise, sleep stage, sex steroids, ghrelin, and to insulin-like growth factor 1 (IGF-1) negative feedback [56]. During puberty, the rise in estradiol (from the ovary in girls, or via aromatization of testosterone in boys) directly stimulates GH release [57], causing an acceleration of linear growth velocity (i.e., the pubertal growth spurt).

Observational as well as sleep interruption studies in the 1960’s by Takahashi et al. were the first to demonstrate that in young adults, the largest GH peak typically occurs just after sleep onset and that there is a close correlation between GH pulse onset and episodes of deep sleep [58]. These findings were confirmed and extended upon in intensive clinical research studies in adults conducted largely by the Van Cauter [59] and Veldhuis [60] labs. Eastman et al. reported a similar association between pulsatile GH secretion, sleep onset, and deep sleep in a group of children referred for short stature [61].

While these studies have clearly demonstrated a temporal concordance between GH pulses and deep sleep, it remains to be determined whether there is a specific quality and quantity of deep sleep that is required to maintain normal GH levels. Obstructive sleep apnea (OSA) has frequently been used as a model to address this question as it is one of the most common sleep disorders and is associated with abnormal sleep architecture (but also with inflammation and hypoxemia). Indeed, before the pediatric obesity epidemic, OSA was often associated with growth failure [62], and it was posited that the increase in serum IGF-1 and improvement in linear growth that occurred after surgical correction of OSA was due to restoration of normal nocturnal GH secretion. As IGF-1 is also a faithful marker of nutritional status, however, an alternative explanation is that correction of OSA led to improved caloric intake and hence improved linear growth. There have been no studies to investigate changes in nocturnal pulsatile GH secretion after treatment of OSA in children. Of interest, a randomized, double-blind, sham-controlled 3-month study of obese men with moderate to severe OSA demonstrated a significant increase in mean GH levels, GH pulse amplitude, and GH pulse frequency (as well as IGF-1) in the group treated with continuous positive airway pressure (CPAP) [63]. However, the change in GH secretion was inversely correlated with the severity of hypoxemia and was not related to changes in deep sleep, suggesting that OSA may not be the ideal model with which to interrogate the dependency of GH secretion on consolidated deep sleep.

Conclusion

There are major changes in sleep patterns and architecture during adolescence that parallel changes in reproductive hormone and growth hormone secretory profiles. Further studies are needed to investigate the causal interactions that may exist among these systems and to determine the potential role of orexin in normal pubertal development.