HPA Axis Alterations in Mental Disorders: Impact on Memory and its Relevance for Therapeutic Interventions

SUMMARY

Dysfunctions in hypothalamic–pituitary–adrenal (HPA) axis have been reported for several mental disorders that are also often characterized by memory disturbances. It is now well established that glucocorticoids influence cognitive processes by enhancing memory consolidation and impairing memory retrieval. There is further evidence for an association between HPA axis related disturbances and memory function in mental disorders. The present selective review provides a brief overview of HPA axis dysfunction and its impact on memory function in major depressive disorder, posttraumatic stress disorder, and borderline personality disorder. Furthermore, the relevance of these findings for therapeutic intervention is discussed.

Introduction

Following the biopsychosocial model of mental disorders, many studies have investigated the functioning of the hypothalamic–pituitary–adrenal (HPA) axis, which is an important part of the neuroendocrine system involved in the coordination of the stress response. Briefly, upon stress exposure, corticotropin‐releasing factor (CRF) is released from the hypothalamus and is transported to the anterior pituitary, where it stimulates the secretion of adrenocorticotropin (ACTH), which in turn stimulates the synthesis and release of glucocorticoids (GCs) from the adrenal cortex. The neuroendocrine stress response is counter‐regulated by circulating GCs via negative feedback mechanisms targeting the pituitary, hypothalamus, and hippocampus. This negative feedback loop is essential for the regulation of the HPA axis and, therefore, for the regulation of the stress response [1]. GCs mediate their effects by binding to two subtypes of intracellular receptors, the mineralocorticoid receptor (MR), and the glucocorticoid receptor (GR). These two receptors differ in their affinity and distribution within the CNS [2]. In addition, a membrane‐bound MR has recently been characterized [3]. Because most of the effects associated with GCs–especially those that are related to the CNS–have been attributed to GR, we will focus here on this particular receptor type. GR have their highest density in the hippocampus [4] but are also prominent in the prefrontal cortex [5], which are both brain regions of substantial importance for memory function.

In healthy subjects, multiple previous investigations have found that acute administration of GCs impairs long‐term memory retrieval [6, 7, 8]. Similar effects have been obtained using psychosocial laboratory stressors [9, 10]. Memory consolidation, on the other hand, seems to be positively influenced by cortisol [7, 11]. In addition, working memory is also impaired by GCs, whereas simple emotional learning (fear conditioning) as well as rigid stimulus response or habit learning is enhanced [5, 12, 13].

Several mental disorders are characterized by memory disturbances, which will be described later. These alterations are not just secondary symptoms but must be regarded as key features of these disorders. In major depressive disorder (MDD), memory disturbances are one of the symptoms defined in the Diagnostic and Statistical Manual of Mental Disorders (DSM‐IV), and memory bias to negative information has been studied extensively [14]. In contrast, posttraumatic stress disorder (PTSD) is characterized by intense memories in which patients re‐experience their traumatic experiences [15]. Furthermore, borderline personality disorder (BPD) is also characterized by neuropsychological disturbances [16], but comorbid axis I disorders might contribute to the clinical picture and perhaps explain some of the observed neuropsychological deficits [17]. HPA axis dysregulations are not only a correlate of the these mental disorders but also predict symptom development [18, 19], treatment resistance [20], and risk for suicide [21], which points to the importance of hormonal stress‐regulation systems in psychopathology.

This review aims to integrate findings on the relationships between HPA axis and cognitive functioning (i.e., memory) in selected mental disorders, including MDD, PTSD, and BPD.

Major Depressive Disorder

MDD is one of the most prevalent mental disorders, with a 12‐month prevalence rate of up to 10%[22]. A major depressive episode is characterized by depressed mood and/or loss of interest or pleasure accompanied by sleep disturbances, psychomotor agitation or retardation, fatigue and loss of energy, feelings of worthlessness or excessive or inappropriate guilt, diminished ability to think or concentrate, as well as recurrent thoughts of death or suicide or even suicide attempt [23]. Biological, psychological, and social factors are known to play a role in the development of MDD, suggesting that depression results when a pre‐existing vulnerability, or diathesis, is activated by stressful life events [24].

HPA Axis

Dysregulation of the HPA axis is a prominent finding in MDD. Several studies [25, 26, 27, 28, 29] but not all [26, 30, 31] found an enhanced basal and stimulated cortisol release in MDD. Furthermore, high cortisol levels after dexamethasone (DEX) administration (so‐called Dex nonsuppression) has been reported in MDD [32], but again not all studies confirmed this effect [33]. These diverging findings might be due to differences in study populations, type of depression, and factors which are known to influence HPA axis, for example, psychotic versus nonpsychotic patients [31], melancholic versus atypical depression [34], comorbid anxiety [35], age [36], gender [37], and history of childhood trauma [24, 38].

The finding of reduced feedback sensitivity has been interpreted as reflecting an exaggerated CRF drive [39] and/or as a reduction of functioning of GRs [40, 41]. One of the most sensitive measurements of HPA axis feedback sensitivity is the combined DEX/CRF test. In that test, HPA axis activity is initially suppressed by dexamethasone treatment before exogenous CRF is given the following day. In depressed patients as well as in persons with childhood trauma, a pronounced escape from this suppression has been found with elevated ACTH and cortisol after CRF administration, further supporting the idea of reduced GR functioning [32]. In support of this hypothesis, post‐mortem studies have reported a reduced GR mRNA in depressed patients [42]. Furthermore, an increased methylation of the GR gene promoter inhibiting GR expression [43] has been reported. GR gene polymorphisms are also discussed to be associated with depression [44, 45]. Furthermore, recent findings suggest that single nucleotide polymorphisms of the GR gene, namely ER22/23EK, BclI, and 9beta A/G may be associated with an increased risk for major depression [29, 30]. Traditionally, the GR has been at the focus of most studies examining neuroendocrine pathways to depression. However, recent evidence suggests that MR dysfunction might also play a role [2, 46, 47]. In light of the MR/GR balance model of depression [4], combined investigations of both receptors are now required.

The results of abnormal HPA axis functioning in patients with MDD have generally been interpreted as reflecting an exaggerated CRH drive and/or a reduced functioning of GRs. A possible shift of MR/GR balance seems to play an important role in this process [28]. Although it is still a matter of debate, GR and/or MR function and genetic variations of them appear to be important factors in the pathogenesis of MDD.

Memory

Memory disturbances are frequent in MDD. One of the most thoroughly investigated cognitive functions is hippocampal‐based episodic/declarative memory. Cognitive performance in episodic memory tasks, including paragraph delayed recall, learning, and retrieval of word lists is impaired in MDD patients [48, 49, 50]. Disturbances in information processing, including a negative bias, is also prominent in MDD, although not all studies agree [51]. Furthermore, autobiographical memory seems to be less specific in these patients [52].

Surprisingly, only few studies have investigated the association between HPA axis functioning and memory performance in depression. Some studies found associations between cortisol levels and cognitive impairment in depressed patients [53, 54, 55, 56, 57], or predominantly in depressed patients with psychotic symptoms [58], but other studies failed to find such associations [59, 60, 61, 62, 63]. However, the cross‐sectional and correlational design of these studies renders them inconclusive with regard to causal directionality.

To our knowledge, until recently only one study has investigated the effect of GC administration on memory in MDD [64]. Bremner et al. found that 2 days of 2 mg DEX treatment improved episodic memory in patients with MDD and suggested that a reduction of cortisol after DEX might have led to the observed memory improvement. In a recently published study by our working group, we investigated the effect of acute cortisol administration on autobiographic memory and found memory impairment in healthy subjects after cortisol intake [65]. In patients with MDD, no further reduction of autobiographical memory performance after cortisol could be observed but memory performance was worse compared to the control group (Figure 1). We hypothesized that the lack of an effect of acute cortisol administration on memory performance might be due to reduced functioning of hippocampal and/or prefrontal GRs. However, these two studies are not comparable with regard to the chosen GC (DEX vs. hydrocortisone) and with regard to the study design (repeated GC treatment vs. single treatment), which complicates the interpretation of these contrasting results.

Figure 1.

Autobiographic memory performance after 10 mg hydrocortisone and placebo. Evidence for reduced central glucocorticoid sensitivity in major depressive disorder (MDD). Although cortisol administration impaired autobiographic memory in healthy controls (n = 16), it had no effect on autobiographical memory in patients with MDD (n = 16). Taken from Schlosser et al. [65] with permission from Elsevier.

Implication for Therapeutic Interventions

With respect to the treatment of depression, it has been shown that remission from depression is associated with normalization of HPA axis dysfunction. Impaired cortisol suppression, for example, is known to normalize after symptom improvement [32] but not within the depressive episode [66]. Thus, several pharmacological approaches aim to reduce HPA axis activity. Possible options are CRF and GR antagonists [67, 68, 69, 70, 71]. Alternatively, future approaches might target the MR or GR receptor (e.g., increasing the expression or sensitivity of the GR). To our knowledge, only one study has investigated the impact of antiglucocorticoids on cognitive function in mood disorder [72]. In this study, 20 patients with bipolar disorder were treated with mifepristone or placebo for 1 week. In contrast to placebo treatment, spatial working memory, verbal fluency, and spatial recognition memory improved significantly with mifepristone, irrespective of improvement in depressive symptoms. The improvement in cognition correlated inversely with basal cortisol levels. This suggests that initially impaired GC receptor function might be restored by the drug and a subsequently appropriate MR/GR balance might then account for the enhancement in cognitive performance. In addition, most antidepressants with known action on the serotonergic system influence HPA axis activity [73] because the HPA axis and the serotonergic system are interconnected [74, 75]. Interestingly, citalopram, a selective serotonin reuptake inhibitor, has been shown not only to affect HPA axis and depression but also to improve working memory [76]. In another study, a 7 months treatment with an SSRI resulted in a significant improvement in memory function and a reduction in UFC excretion, but did not alter hippocampal volume [57]. Thus, it seems that antidepressants may improve hippocampal‐mediated memory function without inducing structural changes.

Future treatment studies should assess HPA alterations as well as memory performances before and after treatment.

Posttraumatic Stress Disorder

PTSD follows exposure to a traumatic stressor, defined as a threat to the life of self or close other, associated with intense fear, horror, or helplessness. Traumatic experiences include childhood abuse, accident, rape, assault, war, and natural disaster. PTSD is characterized by three distinct but co‐occurring symptoms: re‐experiencing the trauma, avoidance, and hyperarousal [23]. The estimated lifetime prevalence of PTSD is 7.8% in the US population [77]. Stress‐induced changes in neurobiological systems are believed to account for PTSD symptoms, such as an enhanced sensitization to stress, enhanced physiological arousal (i.e., hyperactivity of the noradrenergic system), and disturbances in learning and memory [78].

HPA Axis

In contrast to MDD, cortisol findings in PTSD suggest reduced rather than enhanced hormone concentrations [79, 80, 81, 82, 83]. However, these results are not uniformly consistent across all studies, and several factors, such as differences in trauma type, symptom patterns, gender, comorbidity with other mental disorders as well as genetic factors and other predisposing factors have been discussed to contribute to this inconsistency [83]. Furthermore, some studies suggest that comorbid depression plays an important role in HPA axis alteration in PTSD [84, 85]. Beyond this so‐called hypocortisolism, an enhanced suppression after a low dose (0.5 mg) of dexamethasone has been reported repeatedly [80]. This has been interpreted in the context of increased negative feedback regulation of the HPA axis due to increased GR binding [86, 87]. Interestingly, in healthy subjects, enhanced suppression was found to be associated with higher levels of anxiety, interpersonal sensitivity, and avoidant coping strategies [88]. The lower dose of dexamethasone has been used because the standard dose of 1 mg almost completely suppresses cortisol secretion in normal people. To identify increases in negative feedback sensitivity, the dose of dexamethasone must be lowered to 0.5 mg, to avoid a floor effect and maximize distinctions between normal suppressors and hyper‐suppressors. Because of the fact that dexamethasone does not pass the blood–brain barrier and differs in pharmacokinetic and pharmacodynamic features from dexamethasone, a novel suppression test has been developed, that is, the prednisolone suppression test [89]. Further research in PTSD should, thus, move beyond the standard methods. At higher levels of the HPA axis, namely at the central nervous system, increased concentration of CRF has been found [78]. The finding of a blunted ATCH response to exogenous CRF, possibly due to down‐regulation of pituitary CRF receptors, further supports the hypothesis of an enhanced activity of hypothalamic CRF [78].

Only few studies investigated prospectively the relationship between cortisol release and traumatization. These investigations found that lower cortisol measured shortly after the occurrence of a trauma was associated with the development of PTSD, suggesting that hypocortisolism might be a pre‐existing risk factor that might lead to a maladaptive stress response [80]. Interestingly, the finding of a reduced hippocampal size [90], which has been formerly interpreted as result of enhanced cortisol release in response to the trauma, seems to be also a pre‐existent risk factor as suggested by a twin study [91]. As mentioned earlier, the hippocampus is not only rich in GR but also important for memory, including autobiographical memory function. Of note, autobiographical memory is implicated in PTSD (e.g., in terms of intrusive memories).

Memory

Neuropsychological alterations are an important feature of the clinical presentation of PTSD. Several studies revealed problems with learning and memory, including deficits in verbal declarative memory functions and attention [92] as well as reduced autobiographical memory specificity and overgeneralized memory [93]. Patients with overgeneralized memory have difficulties in retrieving specific autobiographical events; instead, they tend to reply with abstract or general memory content (e.g., they summarize several different events) [94].

Studies that have investigated the effects of GC treatment on learning and memory in PTSD have yielded inconclusive results. One study reported a stronger negative effect of hydrocortisone on declarative memory in PTSD, compared to controls. Furthermore, in contrast to the control group patients showed impairments in working memory after pharmacological treatment [95]. In older PTSD patients, further evidence for a more pronounced effect of cortisol was obtained, but this time enhanced working memory performance after injection of hydrocortisone was observed [96]. Another study reported that hydrocortisone led to an impaired hippocampal‐dependent trace eye‐blink conditioning, which is a simple form of associative learning, only in PTSD patients but not in healthy control participants [97]. These findings are in line with the hypothesis of an enhanced GC sensitivity in these patients, which results in an exaggerated effect of GCs on neuropsychological functioning [47]. Contrary to these results, another study reported blunted effects of dexamethasone on declarative memory in PTSD [98]. However, in this experiment not only the pharmacological agent but also the treatment regime differed from the other studies, that is, dexamethasone was given for 2 days before memory testing.

With regard to PTSD symptoms, one might suggest that lower cortisol (in concert with an increased noradrenergic output) leads to a less well‐integrated memory trace and integration of the traumatic event into autobiographical memory. Because cortisol is known to have inhibitory effects on memory retrieval, hypocortisolism might be associated with intrusive memories, flashbacks, and nightmares [13, 79].

Implication for Therapeutic Interventions

Up to now only few studies have aimed to transfer neuroendocrine findings into treatment strategies. In one study, for example, patients were treated after cardiac surgery with hydrocortisone or standard therapy during the perioperative period. The authors found hydrocortisone treatment to be associated with a lower intensity of chronic stress and PTSD symptoms after 6 months [99] (Figure 2). This is in line with a former study of this group, in which the protecting effects of hydrocortisone during septic shock for the development of PTSD had been shown [100]. Thus, such a treatment immediately after the trauma might be an effective secondary‐prevention strategy. In addition, in three patients with chronic PTSD, a reduction of symptoms has been reported after 10 mg/day cortisol over 1 month [101]. However, the results of this pilot study have yet to be replicated in a larger sample. Other intervention ideas, such as treatment with GR blocker, such as mifepristone, also need empirical support [102].

Figure 2.

Posttraumatic stress disorder (PTSD) symptom scores in patients from the hydrocortisone group (N = 26) and patients from the control group (N = 22) with no or one category of traumatic memory and more than one category of traumatic memory, respectively. Evidence for the preventing effect of hydrocortisone treatment on PTSD symptoms. Taken from Schelling et al. [68] with permission from Elsevier. Note: The broken black line shows the 35‐point cut‐off value for PTSD diagnosis by the PTSD screening Questionnaire. Black dots indicate outliers. * indicates a significant reduction in PTSD scores in the hydrocortisone group (with multiple categories of traumatic memory) compared to the control group.

BPD: Psychopathology and Clinical Features

BPD is characterized by intense and rapidly changing mood states as well as by impulsivity, self‐injurious behaviors, fear of abandonment, unstable relationships, and unstable self‐image [23]. Patients with BPD often suffer from comorbid axis I disorders, with mood disorders (96.3%) and anxiety disorders (88.4%) being the most prominent [103].

Patients with BPD frequently report early, multiple, and chronic adverse or even traumatic experiences, such as repeated sexual or physical abuse or emotional or physical neglect, and it has been suggested that early life stress might be an important risk factor in the development of BPD [104].

HPA Axis

Most studies that investigated the HPA axis in BPD and used the 1 mg dexamethasone suppression test have yielded inconclusive results. However, most of these results suggested an association of reduced feedback inhibition with affective dysregulation or even with comorbid MDD [17]. In many of these studies, no formal diagnostic procedure (e.g., diagnostic interview, such as the SCID) was applied so that the data are difficult to interpret. More recent studies not only used more appropriate diagnostic procedures but some also used the low‐dose DST to detect hyper‐suppression. In sum, there is now evidence that comorbid disorders, such as MDD and PTSD, play an important role in terms of HPA feedback regulation in BPD [105, 106].

Only a few studies so far have investigated basal cortisol release, suggesting enhanced cortisol concentrations [17]. Furthermore, an exaggerated ACTH and cortisol response in the combined DEX/CRF test has been found, but only among those who reported childhood abuse [106]. Again comorbid disorders, especially PTSD, seem to have an important influence on endocrine reactions [17]. Furthermore, dissociation, which is a prominent symptom in case of childhood trauma, has been shown to influence HPA axis functioning in BPD [107].

To conclude, studies investigating HPA axis functioning in BPD have provided compelling evidence for the impact of comorbid PTSD and depression on HPA axis feedback regulation in BPD (Figure 3). One might hypothesize that there are at least two subgroups of BPD patients with different endocrine patterns: one predominantly characterized by trauma‐associated symptoms with unaltered to enhanced feedback sensitivity and normal to reduced cortisol release, and another subgroup with mood disturbances as core symptoms and HPA axis dysfunction in form of enhanced cortisol release and reduced feedback sensitivity [17].

Figure 3.

The association between HPA axis dysregulation (basal cortisol concentrations and suppression (feedback) after Dexamethasone application) and core psychopathology in BPD: two potential dimensions. Taken from Wingenfeld et al. [17], with permission from Elsevier.

Memory

Since the 1990s, increasing efforts have been made to characterize neuropsychological functioning of BPD patients. The majority of these studies aimed to determine neuropsychological functioning using standard tests and batteries with neutral stimuli. Although the results of many studies suggested a significant impairment concerning episodic memory functioning [16], other studies were unable to detect such deficits [108]. Interestingly, the pattern of results changed when emotional valence was also considered in more sophisticated experimental designs. The outcomes of many of these studies showed deficits among BPD patients regarding the control and inhibition of emotional interference. For example, one study reported an enhanced retrograde and anterograde amnesia in BPD patients in response to the presentation of stimuli with negative valence [109]. Using the directed forgetting task, a reduced inhibition of emotionally negative words in BPD patients has been documented [110]. Accordingly, a recent study also suggests no general impairment of verbal memory functions in BPD but found that control and inhibition of interference of emotionally significant material seem to be disturbed [111]. Furthermore, comorbid disorders, such as PTSD seem to play an important role in explaining the attentional bias found in these patients [112].

The findings of impairments in the control and inhibition of emotionally negative information might be interpreted in light of psychobiological research. Neuroimaging studies have revealed evidence for a hyperactive responsiveness of the amygdala in response to emotional stimuli and a reduced size of the ACC and hippocampus in BPD. Thus, a dysfunctional network of brain regions that are involved in the regulation of emotions and response inhibition might underlie these effects [17].

Although memory dysfunctions and HPA dysregulation are prominent in BPD, these associations have attracted little scientific attention. For example, there is not a single study published as of today that investigated the effects of cortisol administration on cognition in BPD. Thus, studies that integrate neuroendocrinology and neuropsychology are urgently needed.

Implications for Therapeutic Interventions

Because of the equivocal findings of HPA axis regulation in BPD, attempts to study pharmacological treatment strategies in this field are rare. However, there is evidence that the use of serotonin reuptake inhibitors, that is, fluvoxamine, may regulate the hyper‐responsiveness of the HPA axis in BPD patients [113], especially in those patients who reported childhood abuse. Interestingly, the magnitude of the reduction was not dependent on the presence of comorbid PTSD and depression. This kind of research approach holds promise to integrate endocrine and clinical approaches of BPD. Another study has investigated the effects of carbamazepin on HPA axis in BPD, which is not only used in the treatment of epilepsy but also of affective disorders, and found increased postdexamethasone plasma cortisol values [114]. However, the underlying mechanisms as well as the clinical benefit are still under discussion.

Summary and Conclusions

There is compelling evidence that stress and GCs affect learning and memory, with impairing effects on memory retrieval and enhancing effects on memory consolidation [115]. Several mental disorders are characterized by dysfunctions of the HPA axis as well as memory disturbances, which might be associated with each other. Thus, recent research has aimed to transfer neuroendocrine findings into treatment strategies. In PTSD, for example, there is some evidence that GC administration might reduce the retrieval of aversive memories [116], but more research is needed to draw final conclusions about effectiveness and underlying mechanisms. In major depression, much effort is currently being directed at the development of pharmacological agents that may normalize HPA axis activity, with CRF and GR antagonists being of particularly great interest [68, 69]. Of note, there is evidence for subgroups, with different findings on HPA axis regulation not only in BPD [17] but also in depression, as found by the studies of Heim [24], suggesting an important impact of early trauma on HPA axis functioning. Furthermore, it seems that different treatment strategies might be needed for patients with and without trauma, with psychotherapeutic strategies being of great importance [117]. Thus, an enhanced recognition of interindividual differences appears indicated. Accordingly, the development of new treatment strategies should take these findings into account to provide a more individually tailored therapy.

The HPA axis is not an isolated system in the regulation of the stress response. There is a close association between the locus coeruleus‐noradrenergic system and HPA axis [118]. Together they interact at multiple levels in the periphery and the brain and influence memory function [13]. Several brain regions are involved in these processes, such as, for example, the amygdale and hippocampus as well as prefrontal and temporal areas [119]. The locus coeruleus‐noradrenergic system is especially important in the context of emotional memory and conditioning. There is evidence that noradrenergic activation leads to enhanced memory consolidation as well as enhanced fear conditioning but reduced extinction [119, 120, 121]. Interestingly, adrenergic activation seems to mediate the effects of GC on memory [122, 123, 124]. In fact, the disorders discussed earlier are characterized not only by alterations in emotional memory, learning and information processing but also by disturbances in locus coeruleus‐noradrenergic system as well as dysfunctions in related brain areas [17, 125, 126, 127, 128, 129]. Up to now, clinical studies that investigate the interaction between locus coeruleus‐noradrenergic system and HPA axis and its association with memory in mental disorders are rare. However, in major depression, there is first evidence for a disturbed interaction between these systems [125, 130].

Another important neurotransmitter system in this context is the serotonergic system. In MDD as well as BPD, dysfunctions of the serotonergic system have been reported repeatedly, including reduced 5HT_1A receptor binding in the medial prefrontal cortex, amygdala and hippocampus, or lower cortisol and prolactine responses to meta‐chlorophenylpiperazine [17, 131, 132]. There are strong interconnections among the hippocampus, stress response, and the serotonergic system [133]. Interestingly, in BPD, frontal brain regions as well as parts of the limbic system seem to be associated with dysregulation of the serotonergic system. Symptoms of impulsivity, aggression, and suicidal behavior seem to be strongly mediated by the serotonergic system and are prominent features in patients with BPD [132]. Furthermore, fluvoxamine treatment has been found to reduce the hyperresponsiveness of the HPA axis in BPD patients, especially in those with a history of sustained childhood abuse [134]. Serotonin also mediates the effects of stress on hippocampal GR expression [135, 136]. One might hypothesize that stress‐related alteration of HPA axis regulation in concert with diminished serotonergic functioning may contribute to change in brain structure and metabolism in the disorders discussed here.

Results from neuroimaging studies might help to better understand the set of presented findings. For depressive disorders, Drevets et al. have suggested that an extended “visceromotor network,” including orbital and prefrontal brain regions interfere with those systems that modulate emotional behavior [137]. In fact, an increased activation of the amygdala as well as a reduced activation of the cingulate gyrus and reduced hippocampal volume has been reported for MDD patients [138, 139]. Interestingly, deficits to activate hippocampus and anterior cingulate cortex (ACC) during a verbal memory encoding task have been demonstrated in patients with MDD [140]. However, these findings are not specific for depression. In BPD, a reduced hippocampal size has been also reported as well as volume reductions in the ACC and the amygdala. Accordingly, functional imaging studies have found a decreased activation in these regions, except for the amygdala, where enhanced activation has typically been reported [17]. Furthermore, prefronatal areas are also involved, which leads to the assumption of a disturbed fronto‐limbic network in BPD [141], which shows some concordance with those brain region which seem to be involved in MDD. In PTSD, were most HPA axis finding are contrary to those found in MDD and BPD, neuroimaging studies also revealed evidence of dysfunctions in the hippocampus (mostly reduced size and activity) and the amygdala (enhanced activity). Furthermore, prefrontal regions as well as the ACC seem to play an important role [128]. However, on the central level of the HPA axis, that is, CRF release, PTSD is also characterized by an enhanced activation, which has also been suggested for MDD [142]. Possibly, the finding of hypocortisolism at the periphery might be an adaptation to a chronically activated HPA axis.

As mentioned previously, the hippocampus plays an important role in memory function and has a high density of GR. Therefore, it is thought to be a stress‐sensitive brain region and, thus, sensitive for the damaging effects of stress and GCs [133]. Repeated stress seems to result, for example, in a disruption of neurogenesis and dendritic atrophy of the hippocampus [143]. However, whether these alterations have to be interpreted as a consequence of GC exposure or have to be understood as a pre‐existing risk factor is still a matter of debate [144]. Other brain regions also seem to be negatively influenced by chronic stress (e.g., the prefrontal cortex). In contrast chronic stress has been found to increase dendritic growth in the amygdala, which seems to be associated with greater anxiety [145, 146]. These observations fit with clinical studies suggesting enhanced amygdala activation in several mental disorders, such as PTSD or BPD.

In sum, the mental disorders discussed here appear to be characterized by distinct patterns of dysregulation in stress regulation systems, with similarities at central levels of the HPA axis but, in part, distinctions at the periphery. Furthermore, they show several cognitive impairments, with memory disturbances being prominent and part of the diagnostic criteria. The discussed alterations in several brain regions, which can be described in simplified fashion as disturbed fronto‐limbic networks, might contribute to these memory problems. Possibly, stress hormones are one reason for neural alterations, such as atrophy of specific brain regions. Furthermore, stress hormones have a direct impact on memory function and, thus, may be directly associated with related memory disturbances in mental disorders. Within the last years, novel approaches in pharmacotherapy and psychotherapy have emerged. There is some evidence, for example, that antiglucocorticoids have an antidepressant effect, but controlled trials are needed. Furthermore, there is also preliminary evidence that HPA axis dysfunction in MDD patients can be altered via psychotherapy [147]. This is of importance for patients who are known to respond less to pharmacotherapy, that is, MDD patients with a history of early trauma [117]. Studies in healthy human subjects indicate that the effect is partly mediated by the anticipatory appraisal of the stressor. Evidently, the relationship between GCs and cognition is not one‐sided but reciprocal. Future studies in patient populations including measures of HPA axis and cognitive function are needed to replicate the preliminary positive results in the treatment of cognitive impairment with either pharmacotherapy or psychotherapy. Furthermore, more integrative research is needed, which combines, for example, endocrine and imaging methods, focussing on the dynamic interplay of GCs with brain structures involved in cognitive performance.

Conflicts of Interest

The authors declare no conflict of interests.