HPA regulation

Cells in the PVN secrete CRH, which is the driving force of the HPA axis; AVP only weakly stimulates ACTH by itself, but markedly potentiates the effects of CRH. The secretory cells in the PVN receive neuronal inputs from many brain regions, including amygdala, hippocampus and nuclei of the midbrain. Excitatory and inhibitory afferents, including those containing 5-hydroxytryptamine (5-HT), noradrenaline, acetylcholine, and both excitatory and inhibitory amino acids, have been demonstrated (Reference Jones, Gillham and CampbellJones et al, 1987). In addition to these extrinsic regulatory inputs to the HPA axis, a crucial intrinsic autoregulatory mechanism also exists. Thus, endogenous cortisol, by binding to glucocorticoid receptors in the HPA axis tissues and the hippocampus, acts as a potent negative regulator of HPA activity (Reference Sapolsky, Krey and McEwenSapolsky et al, 1986). The relative contribution of the two glucocorticoid receptor subtypes (GR and MR) in the regulation of HPA activity is as yet unclear. However, the high-affinity MRs putatively control low, basal levels of circulating cortisol, while the lower-affinity GRs come into play during circadian and stress-related peaks in cortisol (Reference Jacobson and SapolskyJacobson & Sapolsky, 1991). This autoregulatory role of endogenous cortisol is crucial to the maintenance of the intrinsic homoeostasis of the HPA axis (Reference Sapolsky, Krey and McEwenSapolsky et al, 1986; Reference Jacobson and SapolskyJacobson & Sapolsky, 1991). Furthermore, it is possible that by subtle changes in glucocorticoid receptor number or function the homoeostatic ‘setpoint’ can be shifted and can contribute to the characteristic diurnal rhythm in circulating cortisol levels in healthy individuals.

HPA dysfunction and psychiatric disorders

In the late 1950s Board and colleagues made the seminal observation that basal plasma cortisol levels of patients with depression are higher than those of healthy control subjects (Reference Michael and GibbonsMichael & Gibbons, 1963). This finding was replicated by Gibbons (Reference Gibbons1964), who further reported that, in the majority of patients, cortisol concentrations are lower during remission than they are during an acute illness episode. These initial observations prompted much further research which has confirmed that hyperactivity of the HPA axis is associated with mood disorders. This HPA hyperactivity is revealed by increased levels of cortisol in plasma, urine and cerebrospinal fluid, exaggerated cortisol responses to ACTH and enlarged pituitary and adrenal glands in individuals suffering from severe mood disorders (Reference Owens and NemeroffOwens & Nemeroff, 1993). Measurement of plasma cortisol over 24 hours reveals that, compared with controls, people with depression exhibit a loss in the characteristic temporal rhythms in circulating cortisol secretion and show elevated nadir and flattening of circadian rhythm (Reference Deuschle, Schweiger and WeberDeuschle et al, 1997). Evidence of increased concentrations of CRH in the cerebrospinal fluid in patients with depression indicates that the hypercortisolaemia seen in mood disorders can be linked to over-secretion of CRH (Reference Nemeroff, Widerlov and BissetteNemeroff et al, 1984). This CRH hypersecretion in mood disorders is further suggested by the blunted ACTH responses to a CRH challenge in people with depression and down-regulation of CRH receptors in post-mortem frontal cortices of suicide victims (Reference Owens and NemeroffOwens & Nemeroff, 1993; Reference Holsboer and BardenHolsboer & Barden, 1996). The above evidence strongly indicates that the hypercortisolaemia associated with mood disorders is a result of CRH hypersecretion. Although AVP hypersecretion has also been postulated, as yet there is little direct evidence to implicate AVP abnormalities in mood disorders.

The hypersecretion of CRH and resultant hypercortisolaemia are though to be a consequence of impaired feedback inhibition by endogenous corticosteroids (see Fig. 1). Since GRs in HPA axis tissues are integral to this feedback inhibition, this receptor is the prime suspect with respect to the genesis of HPA hyperactivity (Reference Sapolsky, Krey and McEwenSapolsky et al, 1986). The dexamethasone suppression test (DST) is a measure of the functional integrity of the GR-mediated negative feedback mechanism. Here, the cortisol-suppressing activity of the synthetic glucocorticoid, dexamethasone, is an approximate indicator of GR status. Reports of cortisol non-suppression in response to dexamethasone in both unipolar and bipolar disorders do indeed suggest a primary GR abnormality in these disease states (Reference Zhou, Shen and ShuZhou et al, 1987; Reference Swann, Stokes and CasperSwann et al, 1992). The methodology of the DST has been criticised, however, in that it fails to take into account the regulatory role of CRH, and subsequent modifications have resulted in the combined dexamethasone/CRH (dex/CRH) test (Reference Heuser, Yassouridis and HolsboerHeuser et al, 1994). The DST and the dex/CRH test consistently indicate dysfunction in HPA axis auto-regulation in both unipolar and bipolar affective disorders. Furthermore, normalisation of responses in these tests on remission of affective symptoms suggests that reinstating GR-mediated HPA auto-regulation may be at least a correlate, if not the mechanism of action, of drugs used in the treatment of mood disorders (Reference Zobel, Yassouridis and FrieboesZobel et al, 1999).

Although abnormalities downstream of the receptor could conceivably underlie the impaired GR function seen in mood disorders, evidence suggests that the dysfunction may simply be due to decreased receptor number. Thus, evidence of an abnormality of GRs in severe mood disorders has been directly shown by the recent post-mortem demonstration of reduced GR messenger RNA (mRNA) in the hippocampi of individuals suffering from unipolar and bipolar affective disorders (Reference Webster, O'Grady and OrthmannWebster et al, 1999). In addition, there is growing evidence that antidepressants and mood stabilisers stimulate GR mRNA expression in the brain, and in so doing, enhance the HPA autoregulation leading to lowered levels of CRH and cortisol.

GR-compromised mice

With the genetic engineering technologies now available to the medical researcher it has been possible to develop a transgenic mouse line with reduced GR number. In common with patients with depression, these mice exhibit HPA disturbances and cognitive deficits which are partially normalised after antidepressant treatments (Reference Pepin, Pothier and BardenPepin et al, 1992b ; Reference Montkowski, Barden and WotjakMontkowski et al, 1995). Furthermore, initial reports suggest that these transgenic mice exhibit attenuated 5-HT_1A-mediated hypothermia (Reference Man, McAllister-Williams and StecklerMan et al, 1999), a deficit observed in drug-free people with depression (Reference Cowen, Power and WareCowen et al, 1994). It is of particular note that in the study of Cowen et al the hypothermia was more pronounced in people with depression of melancholic type, a group strongly associated with hypercortisolaemia.

Although the evidence is less clear, there are reports that MRs (type I glucocorticoid receptors) are also susceptible to regulation by antidepressants (Reference Seckl and FinkSeckl & Fink, 1992). These high-affinity, low-abundance receptors are extensively occupied by low, circulating levels of cortisol, and their role in the feedback inhibition has been established (Reference Jacobson and SapolskyJacobson & Sapolsky, 1991). Clearly, any modulation of MR number by antidepressants would also have important consequences for HPA activity.

Corticosteroids and cognition

One of the most interesting findings of relevance to mood disorders is the association between hypercortisolaemia, cognitive deficits and hippocampal atrophy. Thus, it is now established that in conditions in which there are raised endogenous or exogenous corticosteroids (including Cushing's disease and severe mood disorders) there is also a significant degree of cognitive impairment (Reference Weingartner and SilbermanWeingartner & Silberman, 1982; Reference Wolkowitz, Reus and WeingartnerWolkowitz et al, 1990). In line with these clinical data, studies in experimental animals have indicated that both acute and chronic administration of glucocorticoids result in deficits in learning and memory (Reference White-Gbadebo and HammWhite-Gbadebo & Hamm, 1993; Reference Lupien and McEwenLupien & McEwen, 1997). Chronic administration of glucocorticoids to rodents has also been shown to cause marked atrophy of neurons in the hippocampal formation, and it has been postulated that a similar neurodegenerative effect of cortisol may underlie some of the cognitive deficits observed in humans suffering from mood disorders (Reference Sapolsky, Krey and McEwenSapolsky et al, 1986; Reference Brown, Rush and McEwenBrown et al, 1999). Indeed, recent studies report that cortisol treatment induces cognitive deficits in healthy humans (Reference Young, Sahakian and RobbinsYoung et al, 1999). Furthermore, the cognitive deficits appear to be mediated in part via the frontal lobe, suggesting that this brain area may also be sensitive to the neurodegenerative effects of cortisol (Reference Young, Sahakian and RobbinsYoung et al, 1999).

In healthy volunteer subjects the cognitive deficits induced by cortisol administration are reversible, but this may not be the case with the deficits induced by hypercortisolaemia associated with mood disorders (Reference Ferrier, Stanton and KellyFerrier et al, 1999; Reference Young, Sahakian and RobbinsYoung et al, 1999). Thus, in unipolar and bipolar affective disorders, although cognitive deficits do show some improvement on remission of affective symptoms (paralleling the return of normal HPA function), this improvement is not complete, suggesting permanent and irreversible hypercortisolaemia-induced damage to crucial neuronal circuits. An early re-establishment of normal HPA activity in mood disorders before permanent deficits in cognitive function occur may therefore be an important therapeutic goal.

The argument that in mood disorders, decreased GR function underlies excessive cortisol secretion and that a high cortisol level, in turn, induces deleterious effects on mood and cognition via an action on glucocorticoid receptors might appear contradictory. However, three possible explanations for the deleterious effects of high cortisol levels in the face of reduced GR function present themselves. First, elevated levels of cortisol may be sufficient to overcome the reduction in GR function and so produce an overall increase in effect. Second, it is possible that, while GRs in the hippocampus and hypothalamus associated with autoregulation of the HPA axis are reduced in function, those in other brain regions are normal. Thus, increased cortisol levels combined with normosensitive GRs might result in an increase in the (deleterious) effects of cortisol in some regions. Third, the deleterious effects of high cortisol may, in part at least, be mediated via MR (or a change in the balance of activation of MRs and GRs) or via nonreceptor-mediated events.

Dehydroepiandrosterone

One strategy for counteracting the effects of hypercortisolaemia, used with some success in the treatment of depression, makes use of the adrenal steroid dehydroepiandrosterone (DHEA; Reference Bloch, Schmidt and DanaceauBloch et al, 1999; Reference Wolkowitz, Reus and KeeblerWolkowitz et al, 1999a ). Although the physiological function of DHEA and its sulphated metabolite (DHEA-S) is unclear, these circulating corticosteroids have been shown to possess antiglucocorticoid properties, and high cortisol/DHEA ratios are reported to be associated with persistent depression (Reference Goodyer, Herbert and AlthamGoodyer et al, 1998). However, although DHEA does possess antiglucocorticoid activity, it is partially metabolised to testosterone and oestrogen, which have mood effects of their own and may contribute to DHEA's antidepressant effect (Reference Wolkowitz, Reus and KeeblerWolkowitz et al, 1999a ).

Steroid synthesis inhibitors

Raised levels of cortisol can be lowered pharmacologically by inhibitors of steroid synthesis, and drugs of this class have been used with some success in the treatment of unipolar depression. Ravaris et al (Reference Ravaris, Sateia and Beroza1988) were the first to report that daily doses of ketoconazole reduced both cortisol levels and depressive symptoms within 72 hours in a case of treatment-resistant depression. Since then there have been a number of systematic studies and case reports investigating the use of not only ketoconazole but also metyrapone and aminoglutethimide as antidepressant therapies; as yet, however, the results of these studies are inconsistent (see Reference MurphyMurphy, 1997, for review). This inconsistency is emphasised by two recent double-blind studies. Thus, while Wolkowitz et al (Reference Wolkowitz, Reus and Chan1999b ) found a marked reduction in depressive symptoms following ketoconazole treatment in patients suffering from major depression, Malison et al (Reference Malison, Anand and Pelton1999) found no therapeutic effect of the drug in a similar patient group. It is worth noting that in the former study, ketoconazole was effective in hypercortisolaemic but not normocortisolaemic patients with depression (Reference Wolkowitz, Reus and ChanWolkowitz et al, 1999b ). Unfortunately, one of the main factors which limits the use of steroid biosynthesis inhibitors as antidepressive therapy is the high incidence of side-effects resulting from generalised steroid biosynthesis inhibition.

Corticotropin-releasing hormone antagonists

On the basis of the evidence for oversecretion of CRH in mood disorders, blockade of CRH receptors has been proposed as an approach to normalising hypercortisolaemia. Preclinical studies do indeed indicate that CRH antagonists will be of use in clinical conditions related to HPA hyperactivity, particularly anxiety disorders. Clinical investigations into the use of these compounds in many psychiatric conditions are presently underway and we await their results (see Reference HolsboerHolsboer, 1999, for review).

Type II glucocorticoid receptor (GR) agonists

An alternative strategy for lowering circulating cortisol is activation of the GR-mediated negative-feedback mechanism that regulates cortisol levels. Sub-acute doses of dexamethasone (e.g. 3-4 mg daily for 4 days) have been reported to show antidepressant efficacy (Reference Arana, Santos and LaraiaArana et al, 1995; Reference Bodani, Sheehan and PhilpotBodani et al, 1999). At this dose dexamethasone is thought not to enter the central nervous system and consequently central GRs are spared activation by this exogenous glucocorticoid. Activation of GRs at the level of the pituitary does occur, leading to a lowering of endogenous cortisol. If dexamethasone treatment does indeed act by lowering endogenous cortisol, then it would be interesting to correlate its therapeutic efficacy with the response of patients in the DST. Finally, it should be added that an advantage of the brief course of administration advocated by these studies is a reduction of the side-effects associated with longer-term dexamethasone treatment.

Type II glucocorticoid receptor (GR) antagonists

Paradoxically, as well as GR agonists, GR antagonists have also been advocated as potentially of therapeutic benefit for mood disorders. This is based on the ability of the GR antagonist to block any detrimental effects of the raised levels of circulating cortisol and on the ability of an antagonist to up-regulate its receptor. Thus, administration of a GR antagonist might be predicted to have an acute antiglucocorticoid activity, while also causing a compensatory up-regulation of GR number, leading to enhanced negative feedback on the HPA axis. Preliminary clinical studies using the GR antagonist RU-486 (mifepristone) have been encouraging, even though some clinical efficacy may have been obscured by the prolonged administration of the drug (Reference Murphy, Filipini and GhadirianMurphy et al, 1993). However, animal work suggests that GR numbers can be increased rapidly (within hours) and it is possible that normal feedback is maintained after administration of the antagonist has ceased. This indicates that a brief (i.e. a few days) period of administration of antagonist may be adequate to increase number and normalise HPA function. This might reduce problems of non-compliance and side-effects associated with longer-term administration (Reference Laue, Lotze and ChrousosLaue et al, 1990). A number of new, selective GR antagonists are currently being developed, although preliminary reports suggest that some of these drugs may lack ability to up-regulate the receptors (Reference Bachmann, Linthorst and ReulBachmann et al, 1999).

An evaluation and critique of case reports and clinical trials of some of the treatments outlined above has recently been reported (Reference Wolkowitz and ReusWolkowitz & Reus, 1999). The cumulative evidence makes a strong case implicating GRs in the pathogenesis of affective disorders and suggests targeting these receptors in development of new therapies. We predict that drugs designed specifically to up-regulate GRs will be integral to future therapeutic strategies and may provide a long-awaited paradigm shift in the treatment of unipolar and bipolar mood disorders.

CLINICAL IMPLICATIONS

• ▪ Hypothalamic—pituitary—adrenal axis function is abnormal in mood disorders.

• ▪ The type II glucocorticoid receptor subtype (GR) can be modulated by antidepressants and mood stabilisers.

• ▪ Direct antagonism of GRs may be a future therapeutic strategy in the treatment of mood disorders.

LIMITATIONS

• ▪ We have selectively reviewed a complex area of neuroendocrinology.

• ▪ Evidence that antidepressants and mood stabilisers modulate GRs is mainly taken from animal studies and there is little direct evidence of such a modulation in humans.

• ▪ GR antagonists have not yet been categorically shown to be effective in the treatment of mood disorders.