2.6.3 The endocrine response to stress

In the face of any threat or challenge, either real or perceived, an organism must mount a series of coordinated and specific hormonal, autonomic, immune, and behavioural responses that allow it to either escape or adapt (1–3). To be successful, the characteristics and intensity of the response must match that posed by the threat itself and should last no longer than is necessary. A response that is either inadequate or excessive in terms of its specificity, intensity or duration may result in one or more of a multitude of psychological or physical pathologies (2–5). This concept of threat and the organism’s response to it is frequently recognized and understood as ‘stress’ but is so diverse that it lacks a universally accepted definition (2) and thus is difficult to investigate or study (6).

In the early 1900s, Walter Cannon introduced the concept of homoeostasis (4)—an ideal steady state for all physiological processes. Stress has been defined as the state where this ideal is threatened. More easily appreciated, however, are those factors, both intrinsic and extrinsic, which represent a challenge to homoeostasis (termed stressors) and the complex physiological, hormonal, and behavioural responses that occur to restore the balance, the stress response (1). The importance of endocrine systems in this stress response was emphasized by Hans Selye (7), who described the need for multiple, integrated systems to respond in a coordinated fashion following exposure to a particular stressor. Nonspecific activation of the hypothalamic–pituitary–adrenal (HPA) and sympatho-adrenomedullary (SAM) axes occurred following initial exposure to a noxious stimulus. Continued exposure to the same agent has been shown to have lasting and damaging effects on various endocrine, immune, and other systems, although recovery from this state was possible provided the stress was terminated (7). In addition to various noxious agents, numerous potential stressors exist including exertion, physical extremes, trauma, injury, and psychological stress. Indeed, psychological stressors are some of the most potent stimuli of the endocrine stress response particularly when they involve elements of novelty, uncertainty, and unpredictability. This has been highlighted by the observation that anticipating an event can be as potent an activator of the stress response as the event itself (7).

The HPA and SAM axes are the principal endocrine effector arms of the stress response (Fig. 2.6.3.1). However, a number of other hormone axes and neurotransmitter systems are either directly stress responsive themselves, or modulate these other hormone systems.

Corticotropin-releasing hormone (CRH), identified by Vale and others (8) in 1981, is a 41 amino acid peptide responsible for promoting the synthesis and release of anterior pituitary adrenocorticotropin (ACTH). Hypophyseotropic CRH neurons project to the median eminence from the paraventricular hypothalamic nucleus (PVN). CRH is also widely distributed throughout the CNS, being found within the cortex where it has important effects on behaviour and cognitive processing. Within the brainstem interactions with sympathetic and parasympathetic centres influence autonomic functioning while within limbic and paralimbic regions such as the amygdala, CRH influences the expression of mood and anxiety-type behaviours (9). Arginine vasopressin (AVP), synthesized in parvocellular cells of the PVN, acts synergistically with CRH to stimulate the release of ACTH (9).

ACTH release from the anterior pituitary acts directly on the adrenal cortex to promote the release of adrenal glucocorticoids into the circulation (1, 3, 9). Glucocorticoids, in general, have two fundamental roles in the stress response. First, during stress-free periods, basal levels have a role in preparing the organism for future stress exposure. The circadian rise in glucocorticoids actually occurs prior to activity and thus in humans starts at about 03.00 h and peaks around 09.00 h (see below). This anticipatory activity results in energy storage and conservation by promoting glucose and fat uptake and opposing energy utilization, and prepares the organism for the activities of the next waking day. The glucocorticoids also prime the immune system for future activation and promote memory formation of previous stressors so that future exposure to the same or similar stressor may facilitate a more rapid and efficient response (10).

The second role of the HPA response is to modulate events at the time of stress exposure itself. Initially glucocorticoids enhance the cardiovascular effects of catecholamines and AVP, promote energy provision and utilization, influence and enhance appropriate stress-related behaviours, and stimulate certain aspects of the immune response (10). It is perhaps even more important that once the stress response has been initiated, some of the principal actions of glucocorticoids are to suppress and restrain the activity of these systems, in particular the SAM and immune systems. In doing so, glucocorticoids provide an essential regulatory balance to ensure the stress response is appropriate in terms of both its intensity and duration and that all these responses are ‘switched off’ when the stress has been successfully dealt with (2, 10).

Glucocorticoid secretion is precisely controlled by a complex feedback system that involves a direct action on the hypothalamus and anterior pituitary reducing the amount of releasing hormone (CRH and ACTH, respectively) produced, and consequently limiting the amount of further glucocorticoid released into the circulation. In addition, a further level of feedback activity occurs at the level of the hippocampus, a site that is also important in memory formation. A subset of hippocampal neurons that release the neurotransmitter γ-aminobutyric acid (GABA), project to the hypothalamus where GABA inhibits CRH release, thus contributing to the negative feedback effect on cortisol (1).

There are two known glucocorticoid receptors in the brain, the glucocorticoid receptor (GR) and the mineralocorticoid receptor (MR) which are involved in the feedback system. GRs are found throughout the brain but are most abundant in the hypothalamic CRH neurons and pituitary corticotrophs while MR expression is highest in the hippocampus. The low affinity GR is occupied during periods of intermediate to high glucocorticoid secretion (e.g. during the circadian peak and following stress) while the high affinity MR will be extensively bound even during periods of basal secretion (11). Therefore MR is thought to regulate tonic HPA activity while GR (in coordination with MR) mediates the response to stress.

Three time domains of corticosteroid feedback have been described (12). Fast, rate sensitive feedback occurs within seconds to minutes, during the period of increasing plasma corticosteroid concentrations, and probably controls the rate and magnitude of ACTH and corticosteroid response to stimuli. This may be mediated by membrane-associated MRs via rapid, nongenomic mechanisms (13). Disruption of fast feedback has been demonstrated in ageing humans and in depressed patients, and thus may have a role in the maintenance of homoeostasis (13). Intermediate feedback occurs over 2–10 h and may limit the response of the system to repeated stimulation within a relatively short period of time (hours) while slow feedback (over hours to days) may have the same role during prolonged stress (12).

As with virtually all endocrine systems, ACTH and cortisol show fluctuation in their secretory activity. The classic circadian (24-h) rhythm describes the pattern of HPA activity with hormone concentrations reaching a nadir around midnight, commencing to rise about 03.00 h to reach a peak around 09.00 h before gradually falling throughout the day toward the nadir levels. However, this circadian rhythm is subserved by an underlying ultradian (less than 24-h) rhythm of secretory pulses which can only be detected by frequent blood sampling.

The episodic, pulsatile secretion of ACTH and cortisol has been known for some time (14). More recently specific mathematical models such as deconvolution analysis (15) provide quantitative estimates of in vivo hormone secretion such as the number, amplitude, and duration of underlying secretory bursts. Modulation of the amplitude of both ACTH and cortisol secretory pulses gives rise to their respective nyctohemeral rhythms (16). Under physiological conditions ACTH secretion is characterized by episodic pulses of activity separated by intervals of low basal (nonpulsatile) secretion.

Cortisol is synthesized and secreted from zona fasciculata cells of the adrenal cortex in response to ACTH secreted by the corticotroph cells of the anterior pituitary. There is a high temporal concordance between ACTH and cortisol secretion peaks, with the latter lagging those of ACTH by 10 min (14, 16, 17) (Fig. 2.6.3.2). Secretory bursts of both hormones are episodic in that they are independent events produced randomly over time (16, 18). Sexual diergism in ACTH pulsatility has been demonstrated with males showing greater pulse frequency (18 vs 10 per 24 h), mean peak amplitude and area under the 24-h profile (19). Cortisol secretory bursts occur more frequently in the early hours of the morning (shortly before arising from sleep) and least frequently during late afternoon (18).

Until recently the relevance of episodic ultradian signalling has been unclear. It had been postulated that the quiescent interpulse interval period may allow intracellular synthesis, processing, transport, and storage of ACTH by the metabolically replete and unstressed corticotroph, providing readily releasable hormone in the event of an acute stressful stimulus (20). It is now emerging that corticosteroid pulsatility is important in steroid signalling in that it provides scope for a digital, in addition to analogue, signal for tissue glucocorticoid receptors (21). Hippocampal GR and MR receptors have been shown to translocate rapidly from the cytoplasm to the nucleus and bind DNA in response to a corticosteroid pulse (22). Since GR dissociated rapidly from DNA and disappears from the nucleus within a 1-h interpulse interval, in contrast to MR which remains bound to DNA, changes in pulse frequency will have differential effects on MR and GR binding to DNA. Given the presence of different transcription factors and molecular chaperones in cells of different tissues there is scope for multiple cell specific responses to different digital signals (23).

The hallmark sympathetic ‘fight or flight’ response is characterized by global activation of the SAM system and features typical physiological and behavioural activation including accelerated heart rate, increased blood pressure, and rapid breathing. Fear, vigilance, sensory arousal, and motor activation often with trembling, goose bumps, and piloerection also occur. Catecholamines, the effector hormones of this system, act through specific cell surface receptors that are widely distributed and account for the rapid effects these hormones have on multiple physiological processes (1, 3). Release of glucose stores, immune activation, and increased blood flow to essential organs such as the brain while inhibiting nonessential activity such as digestion together produce a ‘state of emergency’, which can rapidly attend to a sudden change in physiological balance (3). This response is characterized by its speed of onset, its ability to begin in anticipation of an event being stressful, and by its interaction with other stress-responsive systems (3). This interaction can occur either through neural connections or through increased blood flow that transports other messengers (such as hormones and cytokines) more rapidly to their respective sites of action (3).

The sympathetic nervous system originates from nuclei in the lower brainstem that use noradrenaline as their principal neurotransmitter (see Fig. 2.6.3.1). These noradrenergic nuclei, centred on the locus coeruleus (LC), project downward to the intermediolateral columns of the spinal cord. Cell bodies from here send preganglionic fibres to the paraspinal ganglia chain from where postganglionic fibres give rise to sympathetic nerves that supply the heart, blood vessels, lungs, gut, kidneys, and other organ systems. These nerves principally release noradrenaline from their terminals close to their site of action. Other preganglionic fibres also innervate the adrenal medulla and regulate the release of adrenaline into the general circulation.

The stress response system has evolved as both an early warning system capable of recognizing potential or existing threats, and as a response system that can initiate and drive the necessary processes required to escape or confront the threat. By its very nature, the response must be dynamic, beginning rapidly with brain and behavioural activation followed quickly by physiological activation. These processes are characterized by positive feedback and feedforward loops that enhance and reinforce themselves as well as recruiting other arms of the stress response. Slower acting hormone systems are recruited into the cascade providing checks and balances to the already active, but energy expensive systems, putting a brake on the whole response to ensure it is kept appropriate to the type of stress faced, to its intensity and duration, and to ensure the response is switched off when the threat has been adequately dealt with (10, 24).

Changes in the internal or external environment that represent either real or potential threats are recognized with the parts of the brain responsible for receiving, integrating, interpreting, and then relaying this information on to those areas responsible for coordinating the necessary response. This brain activation can be detected within milliseconds and proceeds over seconds to minutes as the response continues to unfold. Stereotypical orienting behaviour, initiated within seconds, gradually gives way to more goal-directed behaviour that is specific to the stressor being faced and the environment in which it is occurring (24).

Activation of the autonomic nervous system occurs within seconds, mediated by the release of catecholamines from sympathetic nerves and the adrenal medulla and enhanced by a withdrawal of parasympathetic activity. These systems promote the immediate physiological, motor, and behavioural responses needed in the face of acute physical or psychological stress. Within minutes of the onset of this cascade of events occurring, hypothalamic-releasing hormones stimulate the release of pituitary hormones with the appearance of ACTH signalling the recruitment of the HPA axis into the process (1, 9). Cortisol levels begin to rise within 2–5 min (25), with peak levels not seen for 15–20 min after the onset of the stress (26) (Fig. 2.6.3.3). Early actions of the HPA system provide additional energy resources for the stress response, while slower gene-related effects over the next few minutes to hours serve to restrain ongoing actions of the stress response which, if left unchecked, may prove to be unsustainable for the individual (1, 3).

Stress is an ambiguous term with many connotations and does not distinguish between the experiences of daily life and major life events such as abuse or trauma (27). The term ‘allostasis’ was therefore introduced to define the active process by which the body responds to daily events and maintains homoeostasis; literally, achieving stability through change. Since a chronic increase or dysregulation of allostasis may lead to disease, the term ‘allostatic load or overload’ was coined to describe the ‘wear and tear’ that results from either too much stress or the inefficient management of allostasis (27). Four situations are associated with allostatic load (2): (1) frequent stress; (2) lack of adaptation to a homotypic (same) stressor; (3) inability to shut off allostatic responses after a stress is terminated; and (4) inadequate response by one allostatic system triggering a compensatory increase in another. The advantage of this terminology arises from the fact that behavioural changes (such as poor sleep, eating/drinking too much, smoking, lack of physical activity) that are part of the allostatic load/overload concept are not obvious in the use of the word ‘stress’ (27). With the superimposition of unpredictable events in the environment, disease, human disturbance, and social interactions then allostatic load can significantly increase, becoming allostatic overload and predisposing the individual to disease (28).

There is a marked change in the hypothalamic response to chronic stress with a greater role for AVP (23). In the hypothalamus there is an increase in AVP synthesis, in the proportion of CRH neurons coexpressing AVP, and in the ratio of AVP to CRH immunoreactivity in neurosecretory vesicles as well as colocalization of AVP and CRH in neurosecretory axon terminals (29). Furthermore, AVP stimulation of ACTH secretion is less sensitive to glucocorticoid feedback than is CRH (30). Pituitary changes with chronic stress paradigms include a reduction in CRH receptor numbers and sustained elevations in V1b (AVP) receptor mRNA (29). It appears in some chronic stress paradigms that CRH has a permissive role whereas AVP is the dynamic mediator of ACTH secretion.

It is important for survival that the HPA axis responds adequately during chronic stress. Rodent stress models reveal three basic patterns of response, depending on the type of stress (31): (1) desensitization of ACTH responses to the sustained stimulus, but hyperresponsiveness to a novel stress despite elevated plasma glucocorticoid levels; (2) corticotroph hyperresponsiveness to a novel stimulus, with no desensitization to the primary repeated stress; and (3) small and transient increases in basal ACTH, followed by marked hyporesponsiveness to novel stimuli. The level of response is determined by the differential regulation of CRH and AVP. The increase in AVP during chronic stress (where glucocorticoid negative feedback down-regulates CRH and ACTH responses) appears to be an important mediator of ACTH release upon new demand. Decreased sensitivity of glucocorticoid feedback is critical for the maintenance of ACTH responses in the presence of increased plasma glucocorticoid levels during chronic stress. It appears that the increase in number of pituitary V1b receptors is the main determining factor for the responsiveness of the corticotroph during adaptation to chronic stress (31).

Involvement of the limbic system in HPA axis regulation is complex (see Fig. 2.6.3.1). The role of limbic structures is both region- and stimulus-specific, they all express both GR and MR and they all exert their effects via subcortical intermediaries (32). Typically, the hippocampus and anterior cingulate/prelimbic cortex inhibit stress-induced HPA axis activation, whereas the amygdala and possibly the infralimbic cortex may enhance glucocorticoid secretion (32). Furthermore, the HPA axis is also subject to glucocorticoid-independent inhibition from neuronal sources. For example, the PVN is richly innervated by GABAergic neurons from the bed nucleus of the stria terminalis, medial pre-optic area, dorsomedial hypothalamus and lateral hypothalamic area. However, the degree to which these GABAergic inhibitory circuits respond to neural vs. glucocorticoid inhibition has not been fully elucidated (32).

The concept of a central stress response network recruited by glucocorticoids and chronic stress has recently been described (25). There is a critical role for extrahypothalamic CRH neuronal cell groups, in particular the amygdala. Elevated glucocorticoids acting in a feedforward manner at the amygdala increase CRH expression and secretion, and this increased amygdalar CRH expression is tightly coupled to hypersensitivity of the HPA axis to stressors. The CRH acts on receptors in structures throughout the brain, in particular monoaminergic cell groups that widely innervate the forebrain, resulting in behavioural changes (e.g. more cautious, more ready to be diverted from tasks at hand, adopt alternative strategies, enjoy rewards and remember fearful situations) that make the organism chronically exposed to stress more capable of adapting to the stressful conditions (see Fig. 2.6.3.1).

Reciprocal neural connections exist between CRH and the locus coerulus/noradrenergic neurons of the central stress system, with each one stimulating the other (4) (see Fig. 2.6.3.1). Chronic stress increases CRH content in the locus coerulus. Thus, CRH may induce mechanisms that result in HPA axis facilitation via increased catecholaminergic input to CRH cells, preparing the organism for the capacity to maintain CRH responses to acute stress during periods of chronic stress when the corticosteroid feedback signal is high (30).

Throughout the history of medicine, reference has been made to the influence of stress, particularly in the form of negative emotions and psychological distress, on physical health (6). Relevant examples include psychiatric conditions such as depression and post-traumatic stress disorder (1), vascular disease such as coronary heart disease, immune-mediated conditions including asthma, and other conditions such as osteoporosis, diabetes, dementia, and premature death (1, 6). Why some individuals manifest stress as psychiatric illness, whilst others are more prone to physical disease and yet others seem resistant to the effects of stress exposure is not well understood.

The implication from these associations is that all stress is ultimately damaging with negative consequences for the individual in whom it is occurring. It is clear, however, that there is a protective role for the stress response in the short term (2), and the associated learning and adaptation (a process that requires plasticity of brain responses) that follows stress exposure is critical to the longer term health and survival of the individual. It is only when these responses occur in excess of the body’s requirements, or continue for longer than is necessary that damaging effects occur (2).

The importance of the concept of allostatic load can be seen in the fact that there is an association between socioeconomic status and health at every level of the socioeconomic status hierarchy (33). This was classically demonstrated in the Whitehall study of coronary heart disease (CHD) mortality (34) which classified 17 530 UK civil servants according to employment grade and recorded their CHD mortality over 7.5 years. Employment grade was a stronger predictor of subsequent risk of CHD death than any other major coronary risk factor. Depression and depressive symptoms are both inversely related to socioeconomic status and depression is linked to health outcomes, particularly CHD (33). As explained by Adler et al. (33) there are two mechanisms by which higher placement in the socioeconomic status hierarchy can reduce stress and its somatic consequences: (1) by diminishing the likelihood that individuals will experience negative events; and (2) through greater social and psychological resources to cope with stressful life events, therefore being less susceptible to the subjective experience of stress.

Melancholic depression has been described as the prototypic example of chronic activation of the stress system (both HPA axis and SAM) (4). Cortisol secretion is increased, the plasma ACTH response to exogenous CRH is decreased and autopsy studies have shown a marked increase in the number of PVN CRH and AVP neurons (1). Depression is also associated with increased pituitary vasopressinergic responsivity and the locus coerulus of depressed patients contains elevated CRH concentrations (35) Repeated stress that causes frequent surges in blood pressure and catecholamine release is associated with accelerated atherosclerosis and an increased risk of myocardial infarction. Patients with melancholic depression develop varying degrees of atherosclerosis and cardiovascular disease (1) and there is evidence that patients with depression that is associated with chronic hyperactivity of the HPA axis have a reduced life expectancy predominantly as a result of an excess of cardiovascular deaths (1, 6, 36). Furthermore, patients with melancholic depression may develop metabolic syndrome, osteoporosis and Th1 immunosuppression (1) consistent with chronic hyperactivation of the stress system. In addition to depression, hypercortisolism is associated with other mood and affective disorders including anorexia nervosa, chronic anxiety, obsessive-compulsive disorder, chronic alcoholism, and other situations such as childhood sexual abuse (1). Hyperactivity of the locus coerulus and other central noradrenergic centres have been shown to influence anxiety and behavioural arousal, with dysregulation of this system postulated as contributing to the pathogenesis of mood disorders particularly depression and with noradrenaline levels being an important predictor of outcome in major depression (1).

Animal experiments of chronic stress provided evidence that glucocorticoid overexposure affects the hippocampus with respect to neuronal viability and function—decreased neurogenesis, degenerative loss in pyramidal neurons, reduced dendritic branching, and atrophy (27, 35). This led to the so-called ‘glucocorticoid cascade hypothesis’ (35) where stress-induced HPA activation and elevated glucocorticoid levels were purported to act in a feedforward manner causing hippocampal damage, resulting in disinhibition of glucocorticoid negative feedback, further rise in glucocorticoid levels and accumulating damage to the hippocampus. This was supported in principle by the fact that patients with Cushing’s disease (resulting in excess adrenal glucocorticoid production) exhibit both hippocampal atrophy and depression, both of which are reversed with treatment. In addition, depressed patients experience cognitive dysfunction consistent with hippocampal damage and most antidepressant treatments enhance neurogenesis (37). However, although reduced hippocampal volumes have been seen on MRI scans of depressed patients, significant histological damage has not been found on postmortem studies (35). Thus, despite compelling animal data linking stress induced hypercortisolism with modulation of neurogenesis in the pathogenesis of depression, evidence for translation to human depression is inconclusive, but is currently an active area of ongoing research. It is also increasingly apparent that HPA dysregulation appears well before clinical symptomatology and is a predictor of treatment resistance in depression. Similarly, failure to normalize HPA axis responses with treatment is a strong predictor of relapse (38).

Chronic stress has been linked to obesity and the metabolic syndrome which is characterized by the combination of central obesity, insulin resistance, dyslipidaemia, and hypertension (39). Glucocorticoids regulate adipocyte differentiation and stress-induced excess cortisol is associated with increased abdominal fat accumulation (39). In humans, chronic stress-induced increases in cortisol, catecholamines, and interleukin (IL)-6 in combination with associated suppression of the growth hormone-, gonadal- and thyroid-axes produces a hormonal milieu conducive to the development of visceral obesity, hypertension, atherosclerosis, osteoporosis, and immune dysfunction (39). Corticosteroids stimulate behaviours that are mediated by dopaminergic mesolimbic ‘reward’ pathways and the central stress response network (25). In fact, glucocorticoids stimulate caloric intake and ‘comfort foods’ may result in a metabolic feedback signal that damp brain stress responses (25). In the current era of chronic social stress and allostatic load, together with the availability of high-calorie palatable foods (acquired with ever-decreasing physical effort), this adaptive mechanism proposed to enable many species to survive may be occurring at a significant (maladaptive) metabolic cost to contemporary humans.

According to McEwen (27) the experience of feeling ‘stressed out’ is associated with elevations in cortisol, sympathetic activity and proinflammatory cytokines that result in an allostatic overload, classically exemplified by sleep deprivation. In animal models with varying degrees of sleep deprivation there has been a consistent pattern of cognitive impairment, namely in learning and retention (40). This has been associated with increased brain levels of proinflammatory cytokines (IL-1β mRNA), and hippocampal oxidative stress and structural changes. Clinical studies have confirmed elevated evening cortisol and day time growth hormone levels with increased sympathetic nervous activity for both total and partial sleep deprivation (41). The resultant increased insulin resistance and reduced glucose tolerance promotes the risk of developing diabetes. This is further compounded by the dysregulation of the neuroendocrine control of appetite promoting obesity.

Evidence is emerging that obstructive sleep apnoea (OSA) represents a chronically stressed state. OSA is characterized by intermittent upper airway obstruction and subsequent hypoxia during sleep. A cyclical sequence of events consisting of upper airway obstruction, progressive hypoxaemia, autonomic, and EEG arousal occurs. This is sufficient to prompt the individual to open and clear the airway to reverse the asphyxia, followed by successive relaxation of the airway and subsequent constriction (42). This results in fragmented sleep which in turn results in daytime sleepiness and fatigue. Other associated symptoms include morning headache, poor concentration, irritability, depression, forgetfulness, overweight, and sexual dysfunction. Morbidity and mortality from OSA is primarily due to cardiovascular disease. It has also been associated with significant metabolic dysfunction including insulin resistance and the metabolic syndrome.

We have found evidence of HPA axis dysfunction in OSA that is altered with continuous positive airways pressure (CPAP) therapy. Obese male subjects with moderately severe or severe OSA had ultradian ACTH and cortisol measured every 10 min over 24 h pre- and 3 months post-CPAP under basal conditions using an automated blood sampling system (17). Hormone secretory characteristics were estimated using multi-parameter deconvolution analysis. There was no change in the number of predicted secretory episodes, secretion pulse height or frequency, however, there was a significant reduction in pulsatile and total ACTH and cortisol production post-CPAP (Henley et al., unpublished data). There was an increased mean pulse mass pre-treatment and this was due to a longer duration of the individual secretory episodes. This is consistent with impaired fast feedback affecting pulse duration (13). This may be due to metabolic/hypoxic insults on the hippocampus, alterations in hippocampal MR expression due to SAM hyperactivity or result from an AVP effect on ACTH pulse duration. Further evidence of HPA axis hyperresponsiveness in untreated OSA is provided by the single breath 35% CO₂ stress test, a validated method for evaluating the stress response in humans (26). There was a markedly exaggerated response to CO₂ pre-CPAP which was reduced to normal levels after treatment (Henley et al., unpublished data) (43). It is therefore likely that the activation of the stress system in OSA contributes to the metabolic complications of this condition.

CRH hypersecretion and HPA axis activation has also been shown to influence the activity of other systems and may have a role in producing some of the other clinical manifestations of stress. CRH hyperactivity is associated with gastrointestinal symptoms such as pain, increased gut motility and diarrhoea—typical features of the irritable bowel syndrome that is commonly associated with stress (1). Similarly, glucocorticoids inhibit the growth axis and it has been postulated that the severe growth retardation associated with psychosocial abuse or deprivation during childhood is, in part, related to chronic HPA axis activation (1).

Chronic hypoactivation of the HPA axis in contrast is also associated with specific disease states. Post-traumatic stress disorder, chronic fatigue syndrome and atypical depression (1, 36) are associated with CRH hypoactivity and reduced cortisol production. Similarly, immune dysregulation is an important consequence of altered HPA axis activity. Differential levels of hypothalamic CRH in the high CRH Fischer and lower CRH Lewis rats are associated with enhanced immune response and resistance to infections and tumours in the Lewis rats, but also an increased susceptibility to some autoimmune conditions (6). In human studies, rheumatoid arthritis appears to be associated with HPA axis hypoactivation (44) with blunted cortisol diurnal rhythms and reduced ACTH and cortisol levels.

Stress may be considered as a real or perceived threat to homoeostasis. The two primary arms of the stress response are the HPA axis and the SAM systems. These two systems are interlinked and regulated by complex feedback and feedforward processes. The acute stress response is protective and promotes survival in the short term. However, prolonged activation of the stress response is implicated in the pathogenesis of illness, in particular mood and affective disorders, and also obesity, the metabolic syndrome, and more recently obstructive sleep apnoea.