Several theories have been posited about the neurobiology of depressive disorder. The monoaminergic hypothesis has much supporting evidence, such as relapse induced by serotonin depletion (Reference Aan, Rot, Mathew and Charney1), contributing to the foundation for the development of several antidepressant drugs. However, this theory has some drawbacks, such as the unexpected antidepressant effect of tianeptine, which reduces extracellular serotonin (Reference McEwen, Chattarji and Diamond2). Furthermore, this theory has not led to the development of antidepressant drugs with greater efficacy or a faster onset of action compared with older drugs. Thus, alternative hypotheses have been proposed, including those that involve molecular neurotrophy (Reference Duman, Heninger and Nestler3), glutamatergic hyperactivity (Reference Berton and Nestler4) and glucocorticoid dysfunction (Reference Barden5–Reference Oitzl, Champagne, van der Veen and De Kloet7). This latter hypothesis is the subject of the review by Von Werne Baes et al. (Reference Von Werne Baes, Tofoli, Martins and Juruena8) published in the current issue of Acta Neuropsychiatrica. Their review highlights two important but (relatively) poorly evaluated points in hypothalamic-pituitary-adrenal (HPA) axis-depression relationships, specifically early life stress and mineralocorticoid receptors (MRs).

The glucocorticoid dysfunction hypothesis proposes that depression is related to HPA axis dysfunction, leading to increased plasma cortisol levels, and antidepressant drugs exert their effects through HPA normalisation (Reference Barden5,Reference Holsboer9). Following the initial studies of Carroll et al. (Reference Carroll Curtis and Mendels10), numerous clinical studies correlated some aspects of changes in the HPA axis, such as increased plasma cortisol levels, with major depression. Preclinical findings also supported this association. Glucocorticoid administration appeared to induce depression-like behaviour in rodents (Reference Gregus, Wintink, Davis and Kalynchuk11), and antiglucocorticoids exert an antidepressant-like effect in animal models (Reference Wulsin, Herman and Solomon12).

Functional neuroendocrine tests have been one of the most frequent strategies used to study HPA axis function in depression. The dexamethasone suppression test (DST) has been used in several studies and is based on the physiological negative feedback of glucocorticoids on the HPA axis. Normally, after dexamethasone [a synthetic corticoid that binds to glucocorticoid receptor (GRs)] administration, plasma cortisol levels decrease (i.e. are suppressed), but this suppression does not occur in some depressed patients (Reference Ribeiro, Tandon, Grunhaus and Greden13).

The non-normalisation of altered HPA axis function after antidepressant treatment has been associated with non-responses to antidepressant treatment or a high percentage of relapse. For example, non-suppression in the DST after treatment is correlated with a poor prognosis (Reference Ribeiro, Tandon, Grunhaus and Greden13). Antidepressant treatment leads to the reinstatement of corticosteroid receptors, which would result in the normalisation of HPA function. Thus, the degree of normalisation may be a biological marker for the clinical efficacy of antidepressant treatments.

Glucocorticoids bind to two different receptors: MRs (or type I receptors) and GRs (or type II receptors). These receptor types have some important differences. For example, MRs bind glucocorticoids with 10-fold higher affinity than GRs (Reference Oitzl, Champagne, van der Veen and De Kloet7). Furthermore, GRs are widely distributed in the brain, but MRs are concentrated in specific brain areas, such as the hippocampus, where MRs and GRs may be co-expressed (Reference Holsboer9). MRs in the hippocampus exert inhibitory control over the HPA axis (Reference Kellner and Wiedemann14). MRs were initially proposed to be completely occupied at basal plasma corticosteroid levels, but more recently, these receptors have been suggested to mediate the initial phase of the stress response (Reference Joëls, Karst, Derijk and De Kloet15). GRs and MRs may interact synergistically because they can form heterodimers that act in gene regulation (Reference Young, Lopez, Murphy-Weinberg, Watson and Akil16). These receptors also have opposing effects. 5-Hydroxytryptamine transmission is decreased by MR activity but increased by GR activity (Reference Young, Lopez, Murphy-Weinberg, Watson and Akil16).

Several studies have implicated GR dysfunction in depression. However, some inconsistencies have emerged in clinical and preclinical studies. For example, exogenous administration of glucocorticoids (e.g. corticosterone, the main glucocorticoid in rodents) generally induces depressive-like behaviour (Reference Gregus, Wintink, Davis and Kalynchuk11), but an antidepressant-like effect has also been observed (Reference Stone and Lin17). Furthermore, mifepristone, a glucocorticoid and progesterone receptor antagonist, exerted antidepressant-like effects in animal models (Reference Wulsin, Herman and Solomon12). However, although mifepristone showed a clinical antidepressant effect in some studies (Reference Debattista, Belanoff and Glass18), these results have been criticised (Reference Carroll and Rubin19). Mifepristone may be effective specifically in psychotic depression, which has been associated with hypercortisolaemia.

Mifepristone treatment increases cortisol and decreases brain-derived neurotrophic factor levels in bipolar patients (Reference Mackin, Gallagher, Watson, Young and Ferrier20), and thus, the conflicting data about the clinical antidepressant effect of mifepristone may be attributable to its selective antagonism of GRs, which leads to increased cortisol that can activate MRs or act through non-genomic mechanisms. Alternatively, hypercortisolaemia may represent a consequence of a reduction in the effects of corticosteroid in the brain (Reference Pariante6).

Homeostasis is related to a balance between MRs and GRs, with hippocampal MRs mediating the initial response to stress and GRs related to later phases of the stress response (Reference Joëls, Karst, Derijk and De Kloet15). Additionally, low-affinity membrane MRs may mediate glutamate release in the stress response (Reference Joëls, Karst, Derijk and De Kloet15).

It is interesting to note the progression of neuroendocrine tests: the evaluation of GRs in the DST, the evaluation of pituitary function in the dexamethasone/corticotropin-releasing factor test and prednisolone suppression test (PST), a drug which possesses similar pharmacokinetic parameters to cortisol and binds to both GRs and MRs. The PST results suggest the maintenance of MR activity and impaired GR sensitivity (Reference Juruena, Cleare, Papadopoulos, Poon, Lightman and Pariante21). Thus, an imbalance between these two corticoid-sensitive receptors may play an important role in the pathophysiology of major depression (Reference Oitzl, Champagne, van der Veen and De Kloet7,Reference Young, Lopez, Murphy-Weinberg, Watson and Akil16,Reference Juruena, Cleare, Papadopoulos, Poon, Lightman and Pariante21). The review published in the present issue of Acta Neuropsychiatrica is timely. Von Werne Baes et al. (Reference Von Werne Baes, Tofoli, Martins and Juruena8) provide a detailed and updated overview on this topic, suggesting that MRs can play an important role in the pathophysiology of depression.

In a case study series, lithium was successfully replaced by spironolactone, an MR antagonist, in bipolar patients (Reference Hendler22). In a small-sample case series, spironolactone appeared to effectively improve residual symptoms in euthymic bipolar patients (Reference Juruena, Gama, Berk and Belmonte-de-Abreu23). However, a controlled add-on study found that fludrocortisone, an MR agonist, but not spironolactone accelerated the antidepressant response to escitalopram, although it did not increase the effectiveness of escitalopram (Reference Otte, Hinkelmann and Moritz24).

Early life stress can trigger long-term changes in the HPA axis, resulting in an altered response to stress in adulthood (Reference Faturi, Tiba, Kawakami, Catallani, Kerstens and Suchecki25) and alteration in the biological vulnerability to depression. Alternatively, stressful life events in childhood can be continuously actualised by some reminders, resulting in the maintenance of changes in HPA stress reactivity.

This is consistent with a preclinical study, in which early life stress decreased hippocampal GR levels and increased corticosterone levels after stress in adulthood (Reference Uys, Muller, Marais, Harvey, Stein and Daniels26). Glucocorticoids may mediate changes in hippocampal neurogenesis induced by stressful life events. Von Werne Baes et al. (Reference Von Werne Baes, Tofoli, Martins and Juruena8) found that early life stress likely affects the HPA axis, but the results are not conclusive regarding the direction of change.

In summary, the article by Von Werne Baes et al. (Reference Von Werne Baes, Tofoli, Martins and Juruena8) in the present issue highlights some interesting lines of research on the HPA axis and depressive disorder. However, a significant proportion of depressed patients do not exhibit hypercortisolaemia or HPA axis alterations, indicating that other neurobiological changes are also related to depressive symptoms.