Abstract
Human sleep is characterised by an electrophysiological component, which is recorded by the sleep EEG, and by distinct patterns of the secretion of various hormones. A bidirectional interaction exists between these two components of sleep. During disturbed sleep, changes of sleep EEG and of hormone secretion occur. For example during an episode of depression and during normal ageing, slow wave sleep and growth hormone (GH) secretion decrease whereas wakefulness increases and the activity of the hypothalamo-pituitary-adrenocortical (HPA) system is changed. During depression and during primary insomnia, elevated HPA activity is mirrored by increased cortisol levels. There is much evidence from preclinical and clinical studies that various neuropeptides and steroids participate in sleep regulation, and that changes in their activity contribute to disturbed sleep. The reciprocal interaction of the peptides growth hormone-releasing hormone (GHRH) and corticotropin-releasing hormone (CRH) plays a keyrole in sleep regulation. In young normal male subjects, GHRH promotes slow wave sleep and GH secretion, whereas CRH exerts opposite effects. Changes in the GHRH/CRH ratio in favour of CRH are thought to result in disturbed sleep, particularly in insomnia-related depression (CRH overactivity) and in normal ageing (reduced GHRH activity). Treatment with a CRH-1 receptor antagonist was shown to improve sleep in patients with depression. The menopause is a major turnpoint of sleep quality in women. In postmenopausal women the levels of circulating estrogens and progesterone are low. Replacement therapy with these steroids improved sleep in postmenopausal women.
Introduction
Sleep is a time of considerable activity in various endocrine systems. A bidirectional interaction exists between the sleep electroencephalogram (EEG) and neuroendocrine activity during sleep. Changes of the sleep structure result in changes of hormone secretion and vice versa. Certain hormones, particularly neuropeptides and neurosteroids were identified as common regulators of sleep EEG and peripheral hormone secretion. There is now much evidence that changes in their activity contribute to impaired sleep including insomnia.
In young normal subjects during the first half of the night, the major amounts of slow wave sleep (SWS) and the growth hormone (GH) surge occur, whereas the levels of corticotropin (ACTH) and cortisol reach the nadir. In contrast during the second half of the night, rapid eye movement (REM) sleep, ACTH and cortisol preponderate whereas GH release is low (Weitzman, 1976). This pattern of sleep-endocrine activity suggests that (i) a reciprocal interaction exists between the hypothalamo-pituitary-somatotrophic (HPS) and the hypothalamo-pituitary-adrenocortical (HPA) systems and (ii) there exist common factors regulating sleep EEG and the nocturnal hormone secretion as well. It appears likely that the key hormones of the HPS and HPA systems, GH-releasing hormone (GHRH) and corticotropin-releasing hormone (CRH) are such factors and that their reciprocal interaction plays a major role in the physiological regulation of sleep. Finally changes in the balance between GHRH and CRH appear to contribute to impaired sleep.
This topic aims to summarize the state of the art in the field of endocrine mechanisms of insomnia and the related basic research.
Sleep-Endocrine Changes in Patients with Insomnia
In an early study Adam et al. (1986) compared poor and good sleepers, selected on the basis of their stated opinion about the sleep. According to sleep EEG the poor sleepers woke up more often and achieved half an hour less sleep. They tended to have higher urinary cortisol and adrenalin excretion. In a preliminary study Vgontzas et al. (1998) found an association between chronic insomnia and the activity of the stress system. The participants of the study were 15 younger adult patients with insomnia, less than 50 years old. Each patient was recorded in the sleep laboratory for three consecutive nights. The 24-hour levels of urinary free cortisol were positively correlated to the total wake time. Furthermore the 24hour urinary levels of the catecholamine metabolites dihydroxyphenylglycol (DHPG) and dihydroxyphenylacetic acid (DOPAC) showed a positive correlation with the percentage of sleep stage 1 and intermittent wakefulness. Also norepinephrine tended to correlate positively with these sleep-EEG variables. Urinary GH excretion was detectable in only three of the subjects. The authors concluded that in chronic insomnia the activity of both limbs of the stress system, the HPA and the sympathetic system as well relate positively to the degree of objective disturbance. The same authors (Vgontzas et al., 2001a) investigated the sleep EEG of young patients with insomnia and normal controls during four consecutive nights. During the fourth day plasma measures of the HPA hormones ACTH and cortisol were performed. During baseline nights sleep latency and wakefulness were more extended or longer respectively in the patients than in the controls. The 24 hours ACTH and cortisol levels were significantly elevated in the patients. Within 24 hours the greatest elevations of these hormones were observed in the evening and in the first half of the night. Insomniac patients with high degree of objective disturbance secreted a higher amount of cortisol compared to those with a low degree of sleep disturbance. Pulsatile analysis showed a significantly higher number of peaks per 24 hours in the patients than in the control subjects, whereas no difference in the temporal pattern of ACTH or cortisol secretion were found by cosinor analysis. The authors concluded that insomnia is associated with an overall increase of ACTH and cortisol secretion, which, however, retains a normal circadian pattern. They suggested that these findings are consistent with a disorder of CNS hyperarousal rather than one of sleep loss. In a group of seven male patients with a severe chronic primary insomnia, evening and nocturnal cortisol levels were significantly increased compared to matched controls. Evening cortisol correlated with the number of nocturnal awakenings in patients and controls. Furthermore the patients showed significant correlations between several sleep-EEG parameters and the cortisol secretion during the first four hours of the night (Rodenbeck et al., 2002). In contrast, in another study nocturnal cortisol levels of patients with insomnia did not differ from those of normal controls, whereas the nocturnal melatonin production was significantly diminished in the patients (Riemann et al., 2002). After awakening salivary cortisol decreased in patients with primary insomnia (Backhaus et al., 2004). The authors suggest that its decrease may be related to nocturnal cortisol activation after an increased number of nocturnal awakenings.
Insomnia is a frequent symptom in patients with depression. Accordingly their sleep EEG shows characteristical changes consisting of (Armitage, 2007, Benca et al., 1997, review: Reynolds & Kupfer, 1987):
(i) impaired sleep continuity (prolonged sleep latency, increased number of intermittent awakenings, early morning awakenings),
(ii) changes of nonREM sleep (decreases of SWS, slow wave activity [SWA] and sleep stage 2, a shift of SWS and SWA from the first to the second sleep cycle in younger patients),
(iii) REM sleep desinhibition: a shortened REM latency or sleep onset REM periods (SOREMPs), REM latency (0-20 min), prolonged first REM period, enhanced REM density (measure of the frequency of rapid eye movements) particularly during the first REM period.
A robust finding in depressed patients is elevated levels of the hormones of the hypothalamo-pituitary-adrenocortical (HPA) system cortisol and corticotropin (ACTH) throughout the night (Antonijevic et al., 2000c, Steiger et al., 1989, Steiger, 2007) or throughout 24 hours (Linkowski et al., 1987), respectively, in comparison to normal control subjects. Nocturnal GH secretion is blunted in most studies in depressed patients (Jarrett et al., 1990, Steiger et al., 1989, Voderholzer et al., 1993). These finding suggest a crucial relationship between shallow sleep, HPA overactivity and low GH levels in patients with depression.
In a longitudinal study sleep EEG and hormone secretion were compared in acute depression and recovery (Steiger et al., 1989). During acute depression characteristical sleep-EEG changes were found. These persisted after recovery and the time spent in sleep stage 4 even decreased. Also GH levels were low at both examinations. In contrast cortisol levels decreased after recovery. This finding is in line with various observations showing that HPA overactivity is a state marker of depression. Obviously cortisol normalizes independently from sleep. Therefore hypercortisolism in depressed patients is unlikely to be a consequence of shallow sleep. The persistence of most sleep-EEG (Kupfer et al., 1993) and GH changes (Jarrett et al., 1990) in remitted patients has been confirmed over a period of three years. It appears likely that the metabolic aberrances during acute depression result in a biological scar which is mirrored by the persistence of sleep-EEG and GH changes after recovery. This hypothesis is further supported by a study in male patients who survived severe brain injury (Frieboes et al., 1999). Several months after the injury, cortisol levels of these patients were in the range of normal controls. In contrast, the time spent in sleep stage 2 was reduced and GH levels were blunted. Whereas cortisol concentrations were normal at the time of examination in this study, it appears likely that either HPA overactivity due to stress under the intensive care situation after brain injury or treatment with glucocorticoids in some patients contribute to the changes of sleep EEG and of GH levels.
Interestingly there are similarities in sleep-endocrine changes during depression and during normal ageing. As early as during the third decade of the lifetime, a continuous decline of SWS, SWA and GH starts. In male subjects near to the onset of the fifth decade, the GH pause occurs. From then nearly no GH release occurs. In females the menopause and the GH pause are related. Whereas in male subjects sleep quality declines continuously throughout the life span, in females the menopause is the major turnpoint in sleep quality (Ehlers & Kupfer, 1997).
There are controversial reports on the influence of age on HPA hormones. Elevated and unchanged cortisol levels as well have been reported in elderly subjects. Age-dependent increases of mean cortisol levels and of cortisol nadir and, selectively in women, of the acrophase were found in the study which includes the largest sample of normal human adult subjects over a lifetime. The pattern of cortisol secretion was preserved in the elderly, whereas the amplitude was dampened and the morning rise appeared advanced (Van Cauter et al., 1996).
From clinical practice, it is well known that changes of sleep-wake behaviour are frequent symptoms of disorders of the thyroid gland. Hyperthyroidism is linked with insomnia whereas fatigue occurs frequently in patients with hypothyroidism. Astonishingly only a few data on sleep EEG in these diseases are available. Reduced SWS was reported in patients with hypothyroidism in comparison to normal controls. After therapy these changes turned to normal (Kales et al., 1967).
In clinical practice sedating antidepressants are often used for treating insomnia. The effects of the tricyclic doxepin on nocturnal sleep and plasma cortisol levels were tested in 10 patients with chronic insomnia between 1700 and 0800 h. Single infusions of placebo and 25 mg doxepin were given according to a double-blind randomized crossover design. Afterwards all patients received 70 mg doxepin orally for three weeks in an open study. Both doxepin administration forms improved sleep EEG and reduced mean cortisol levels. The duration of the quiescent period of the cortisol rhythm was significantly prolonged after both doxepin administrations compared to placebo. The authors suggest that the sleep-improving effects of doxepin are mediated at least partly by a normalization of the HPA system (Rodenbeck et al., 2003). Similar effects were observed after sleep-improving antidepressants in patients with depression. In a four week double blind clinical trial, the effects of the tricyclic antidepressants trimipramine and imipramine on the sleep EEG and on nocturnal secretion of HPA hormones were compared in 20 male inpatients with major depression, whereas both treatments produced rapid significant clinical improvement of depression the two drugs had markedly different influences on sleep-endocrine activity. Trimipramine enhanced REM sleep and SWS, whereas imipramine suppressed REM sleep and showed no effect on SWS. Total sleep time and the sleep-efficiency index increased after trimipramine, but not after imipramine. The nocturnal cortisol secretion decreased with trimipramine but remained unchanged with imipramine. Imipramine, but not trimipramine induced a decrease in GH secretion during the first half of the night (Sonntag et al., 1996). Already at the second day of treatment of depressed patients with the noradrenergic and specific serotoninergic antidepressant mirtazapine, sleep continuity was improved. This effect persisted after four weeks, when SWS, low-delta, theta and alpha activity decreased. Furthermore cortisol and ghrelin levels were reduced, whereas leptin and melatonin levels increased. ACTH and prolactin remained unchanged. This study shows a parallel improvement of sleep and a blunting of cortisol levels in patients with depression (Schmid et al., 2006).
Hypothalamo-Pituitary-Adrenocortical (HPA) System
The HPA system mediates the reaction to acute physical and psychological stress and is essential for the individuals’ survival. The cascade of HPA activity starts with the release of CRH from the parvocellular neurones of the paraventricular nucleus of the hypothalamus. This results in the secretion of ACTH from the anterior pituitary gland and finally in the secretion of cortisol (in humans) or corticosterone (in rats) from the adrenocortex.
Nonpharmacological manipulations which help to study the interaction between sleep EEG and hormones include sleep deprivation and nocturnal awakenings. The pioneering work of Weitzman (1976) led to the still valid conclusion that the pattern of cortisol secretion is widely dependent on a circadian rhythm, whereas manipulation of the sleep-wake pattern causes subtle changes of HPA secretion. In a first investigation of this group, control subjects underwent a repetitive three hour sleep-wake cycle for 10 consecutive days. Each of the cycles consisted of two hours of wakefulness and one hour sleep. Cortisol was consistently low during the intervals with sleep and during the wake periods in each cycle, independently of circadian time (Weitzman, 1976). In a four days protocol, a baseline investigation was followed by sleep deprivation. On day 3, the proband was allowed to sleep at a time 12 hours later than usual. On day 4, he slept again at his usual sleeping time. During the first four hours after sleep onset, cortisol was low even when sleep took place during the time when the subject was usually awake (Weitzman et al., 1983). There exist very few reports on HPA activity at several intervals during and after partial and total sleep deprivation. During the night of sleep deprivation, enhanced and unchanged cortisol levels as well were reported (Steiger, 2002). In the recovery night following one night of sleep deprivation, cortisol was unchanged in young and elderly normal subjects compared to the baseline condition (Spiegel et al., 1999). In the evening of the day after partial or total sleep deprivation, cortisol was enhanced (Leproult et al., 1997). Similarly the evening level of cortisol was higher when sleep in normal young men was restricted to four hours per night for six days (Spiegel et al., 1999). In the recovery night after four nights with restricted sleep, cortisol levels were blunted during the second half of the night (Follenius et al., 1992). A delayed sleep onset in controls was followed by a later occurrence of the cortisol rise (Fehm et al., 1993). Patients with depression were studied during three consecutive nights before, during and after sleep deprivation. Saliva samples of cortisol were collected during daytime before and after the sleep deprivation night. During the night of sleep deprivation, cortisol levels were significantly higher than at baseline. Daytime cortisol levels during the first half of the day were higher than at baseline in the patients who responded to sleep deprivation but not in the nonresponders. During recovery sleep, cortisol secretion returned to baseline values. The authors concluded that the data demonstrate a significantly stimulatory effect of one night of sleep deprivation on the HPA system in depressed patients (Voderholzer et al., 2004). In animal studies the effect of sleep deprivation appears to be not only a consequence of sleep loss but also partly a consequence of the procedure that is used for sleep deprivation, which is the arousal resulting from being kept awake (Rechtschaffen et al., 1999). For example, mice that were kept awake by minimal stimulation and sounds have low glucocorticoid levels, whereas animals that kept awake by letting them engage in social activities with other mice have distinctly elevated corticosterone levels (Meerlo & Turek, 2001). In rats 27 hours of sleep deprivation led to increases of CRH levels in the striatum, limbic areas and pituitary, whereas hypothalamic CRH was reduced. Significant decreases in CRH binding were found in the striatum and pituitary (Fadda & Fratta, 1997). An in vivo microdialysis study in rats showed a marked rise in corticosterone levels in the brain during sleep deprivation (Penalva et al., 2003). In all these data suggest that changes of HPA secretion, particularly elevated HPA levels are a frequent consequence of restricted or disrupted sleep.
In the rat CRH gene transcription levels increase during the dark period, when the animals are active, and decrease in the morning and throughout the light period (Watts et al., 2004). In the Lewis rats, the release of CRH is diminished due to hypothalamic gene defect. These rats spend less time awake and more time in SWS than intact strains (Opp, 1997). Vice versa spontaneous wakefulness was reduced in rats by a CRH antisense oligodeoxynucleotide (Chang & Opp, 2004). These data suggest that CRH exerts a physiological role in the maintenance of wakefulness and that it impairs sleep. In homozygous mice overexpressing CRH in the central nervous system, wakefulness and REM sleep are elevated, whereas nonREM sleep is slightly reduced in comparison to the wildtype and control mice (Kimura et al., 2009).
Intracerebroventricular (icv) administration of CRH decreases SWS in rats (Ehlers et al., 1986) and rabbits (Obal et al., 1989), and both nonREM and REM sleep in mice (Romanowski et al, 2006; Sanford et al., 2008). SWS is reduced even after 92 hours of sleep deprivation in rats. Furthermore sleep latency and REM sleep increase (Marrosu et al., 1990). When CRH was injected in a pulsatile fashion each hour between 2200 and 0100 hours to young normal men, SWS, REM sleep and GH decreased and cortisol increased (Holsboer et al., 1988). In young normal women in an analogue study protocol, CRH prompted an increase of wakefulness in the second half of the night and a decrease of sleep stage 3 (Schussler et al., 2008a). The role of age in the sleep-impairing effect of CRH is illustrated by a study comparing young versus middle-aged men. A dosage of CRH which was not effective in young men increased wakefulness and decreased SWS in middle-aged men (Vgontzas et al., 2001b).
Treatment of depressed patients with the CRH-1 receptor antagonist R121919 counteracted the characteristical sleep-EEG changes. The time spent in SWS increased compared to baseline after one week and after four weeks of treatment. The number of awakenings and REM density showed a trend towards a decrease during the same time period. Separate evaluation of these changes for two panels receiving different dosages showed no significant effect at lower doses, whereas in the higher doses after the antagonist REM density decreased and SWS increased significantly between baseline and the end of the trial (Held et al., 2004). These results suggest that (i) CRH participates in the pathophysiology of sleep-EEG changes during depression and (ii) CRH-1 receptor antagonism is capable to treat impaired sleep in depressed patients.
The neuropeptide vasopressin is a major cofactor with ACTH in the activation of the stress reaction. In rats wakefulness increases after icv vasopressin (Arnauld et al., 1989). Chronic intranasal vasopressin administration however improved sleep in normal elderly subjects (Perras et al., 1999b).
The synthetic ACTH (4-9) analogue ebiratide shares several behavioural effects of ACTH whereas it does not affect peripheral hormone secretion. Accordingly GH and cortisol levels were not influenced by repetitive iv administration of the substance whereas sleep latency increased, and during the first third of the night wakefulness increased and SWS decreased (Steiger et al., 1991).
Continuous nocturnal (2300 to 0700 h) infusion (Born et al., 1991) and pulsatile iv administration of cortisol (hourly from 1700 to 0700 h) increased SWS (Friess et al., 1994) and SWA (Friess et al., 2004) and decreased REM sleep in young normal control subjects. In the latter study GH levels increased after cortisol. Similarly increases of SWS, SWA and GH and decreases of REM sleep were found in analogue protocols with iv administration of cortisol in elderly men (Bohlhalter et al., 1997) and in patients with depression (Schmid et al., 2008). Since CRH (Holsboer et al., 1988) and cortisol exert opposite effects on SWS (Born et al., 1991, Friess et al., 1994) and GH (Bohlhalter et al., 1997, Friess et al., 1994), it is unlikely that these effects are mediated by increased cortisol levels. In contrast, these changes appear to be the consequence of negative feedback inhibition of endogenous CRH.
The mixed glucocorticoid receptor and progesterone receptor antagonist mifepristone disrupted sleep distinctly in a single case study (Wiedemann et al., 1992). The effects of acute and chronic glucocorticoid administration on sleep appear to differ. Subchronic treatment of female patients suffering from multiple sclerosis with the glucocorticoid receptor agonist methylprednisolone resulted in shortened REM latency, increased REM density and a shift of the major portion of SWS from the first to the second sleep cycle. These changes resemble the sleep-EEG disturbances in depression (Antonijevic & Steiger, 2003).
