Measuring HPA axis activity

Assessing HPA activity can be challenging, as the system is highly reactive and cortisol levels fluctuate markedly; the increase in biomarkers of the HPA axis in response to stress can occur in situations such as in blood sampling for endocrine biomarkers, particularly in psychiatric populations. Fortunately, cortisol can be measured with less invasive methods, such as saliva or urine sampling. However, these are still transitory assessments, which are limited in that they do not accurately reflect chronic HPA function (or dysfunction), even when measured at many points over time. Recently, it has been discovered that cortisol can be measured in hair and nails. This has the benefit of providing a cortisol measurement over a period of months (with hair growth rate approximately 1 cm per month) prior to the time of testing, and a review has now addressed the assessment of chronic cortisol in mood and anxiety disorders (Herane Vives et al., 2015).

Cognition and HPA axis function

Specific effects of the HPA axis in response to physiological or psychological stressors on cognitive functioning manifest via a number of mechanisms. When released, corticosteroids cross the blood–brain barrier and exert some control over the activation of neural networks related to learning and memory, including the hippocampus and prefrontal cortical areas. Glucocorticoid receptors are densely populated in these regions which regulate cognition, and there is now a wealth of evidence demonstrating that cognitive processing and glucocorticoid function are directly and bi-directionally related. A causal link has also been demonstrated between chronic glucocorticoid administration and atrophy of neurons in the hippocampus (see Lupien et al., 2005). These findings are augmented by evidence demonstrating that administration of glucocorticoids impairs learning and memory in animals (Starkman & Schteingart, 1981) and reversibly in healthy people (Young, Sahakian, Robbins, & Cowen, 1999).

Both GRs and MRs are represented widely in brain areas important to cognition; these include the hippocampus, medial prefrontal cortex, and amygdala. MR has a binding affinity for cortisol ten times higher than GR affinity, though while MR binding at high levels occurs continually, GR binding primarily increases in response to stress and is responsible for re-establishing homeostasis following an HPA stress response, as well as having a greater role in memory processing.

Different types of memory have different relationships with HPA activity (the animal literature has been expanded on by Prickaerts & Steckler, 2005). Procedural learning (which is often subconscious; for instance, conditioning or habit forming) has been under-investigated; learning abilities may correlate with corticosterone levels in rats, but the association with cortisol in humans is unknown. Extinction learning (the reduction of a previously conditioned response) may be important in altered HPA activity; high cortisol levels in animals have been positively correlated with extent of extinction. This may be primarily influenced by MR function, while the GR plays an important role in consolidation and retrieval processes of declarative memory, which has been far more widely studied. There is often a negative correlation between HPA function and declarative memory (as well as across other cognitive domains), and this association may be mediated by hippocampal integrity. GR agonists (cortisol in humans; corticosterone in rats) may improve memory consolidation but MRs are less understood and we have contradictory theories and evidence of their role in humans.

While long-term use of glucocorticoids has been consistently linked to impaired cognitive processing across domains (Brunner et al., 2005) there is now some indication that this effect may disappear after glucocorticoid levels are normalized, and may not be associated with structural changes in brain areas with high numbers of GRs (Coluccia et al., 2008).

The content of memory is also important. Although many studies have found that acute glucocorticoid administration impairs learning and memory abilities, there are now well-documented links between emotional stimuli and improvements in cognition with corticosteroids in animals (for a review see Roozendaal, 2000) and (to a lesser extent) in humans: when given hydrocortisone and presented with non-emotional and emotional (both positive and negative valence) stimuli and then tested for long-term (one week) memory recall, healthy participants remembered more emotional (particularly unpleasant rated images) material (Buchanan & Lovallo, 2001). This may be an interesting finding in relation to depression, where negatively emotional past events can be salient (and given the links between childhood trauma and levels of corticosteroids). A positive effect of acute cortisol on emotional memory may be attributed to interactions between the HPA axis and the amygdala: this structure plays a particular stimulatory role in the HPA axis and is not subject to direct negative feedback from the glucocorticoids.

However, similar findings have recently been reported that show improved contextualization of memory for non-emotional as well as emotional recall when healthy participants experienced stress, mediated by cortisol response (Van Ast, Cornelisse, Meeter, & Kindt, 2014).

Exogenous substances that result in HPA activity increases may be related to mood disorders via a disruptive effect on cognitive ability when administered either at high doses or chronically. Though cortisol levels alone may not predict level of cognitive function, it appears that when experience of stress is chronic or extreme, the stress-cognition effect can be mediated by HPA response to cause impairments in long-term memory recall (discussed comprehensively by Finsterwald & Alberini, 2014).

HPA abnormalities in unipolar depression

There are long-established links between depression and many illnesses of the endocrine system (covered by Musselman & Nemeroff, 1996). Illnesses that share symptoms with depression also have frequently been related. As has been reviewed widely, HPA axis function is found to be blunted in chronic fatigue syndrome (CFS) which shares many similarities with some forms of depression (Cleare & Wessely, 1996), and lower cortisol in this illness has also been associated with poorer response to cognitive behavioral therapy (CBT) (Roberts et al., 2010). Tak et al. (2011) have meta-analyzed the association between cortisol and somatic disorders (such as CFS, irritable bowel syndrome (IBS), fibromyalgia, and chronic pain), finding that many patients (mainly those with CFS) show hypocortisolism. However, some studies have identified elevated cortisol in IBS and fibromyalgia, and many endocrine-related disorders are also associated with cognitive impairments (e.g. Addison’s disease; Cushing’s syndrome, Musselman & Nemeroff, 1996).

Abnormal cortisol has also been implicated in psychiatric disorders that are associated with cognitive difficulties. Findings include hypocortisolemia in depersonalization disorder (Stanton et al., 2001) and anxiety, but elevated HPA activity in mania and depression (Yehuda, Boisoneau, Mason, & Giller, 1993). These endocrinological differences between psychiatric disorders may be compared with Cushingoid and Addisonian phenotypes (which demonstrate hypo and hyper HPA activity, respectively), which can both show reduced depression alongside abnormal cortisol levels albeit in opposite ways. These findings suggest that dysregulation of the system in either direction can show similar effects of chronic stress.

In vulnerable individuals, repeated exposure to stress can affect the body’s ability to maintain homeostasis and contribute to the onset of depression. Mechanisms by which this can occur involve neural, endocrine, and immune physiological mediation of stress, leading to the increased allostatic load which may precipitate the onset of depression. Specifically, theories of HPA axis dysregulation in depression are derived from many sources, including some of those outlined above and there is now an established association between major depressive disorder (MDD) and abnormalities in HPA activity (McQuade & Young, 2000).

HPA axis function in bipolar disorder

In bipolar disorder, the evidence illustrating HPA dysfunction is less extensive but similar in nature. As Bond and Young (2007) report, HPA dysfunction appears in all mood states in individuals with bipolar disorder.

Corticosteroid levels: Similarly to unipolar depression, results observed consistently include hypersecretion of cortisol, as well as different diurnal patterns of corticosterone release. While hypercortisolemia was demonstrated by Gallagher and colleagues (Gallagher, Watson, Smith, Young, & Ferrier, 2007), bipolar patients in this study did not show abnormal DHEA, or cortisol–DHEA ratio; this may provide a biomarker for differentiating bipolar from unipolar depression.

HPA axis receptor function: Abnormal suppression of dexamethasone (also applicable for the DEX/CRF combination test), and as in unipolar depression can show changes alongside remission of affective symptoms (Daban and colleagues review these findings in further detail (Daban, Vieta, Mackin, & Young, 2005)). These abnormalities appear to persist through euthymic state, as well as in mania and depression: Watson and colleagues (Watson, Gallagher, Ritchie, Ferrier, & Young, 2004) have identified that patients continue to show heightened cortisol in response to the DEX/CRF test between mood and euthymic states. Post-DST levels of AVP may also be altered in both bipolar and unipolar depression, and this may persist into remission representing a trait in these patients (Watson, Gallagher, Ferrier, & Young, 2006a).

Recently it has been suggested that some people with bipolar disorder demonstrate hypocortisolism in response to a low dose of DST (Maripuu, Wikgren, Karling, Adolfsson, & Norrback, 2014). In this study, the groups classed as hyper- or hypo-cortisolemic mostly comprised bipolar patients, while healthy control levels tended to be in the normal range. Low levels of cortisol appeared most pronounced in those with more severe depressive symptoms and with poorer quality of life. It is possible that psychiatric comorbidity may account for part of this observation; however, there remains an important question to be addressed as to whether these groups may demonstrate distinct clinical populations in some respect.

HPA genetics: Emergent findings have reported a putative relationship between diagnosis of bipolar disorder and the GR gene, NR3C1, and it has been posited that NR3C1 polymorphism variants that are implicated in major depression might also be associated with more dominant depressive symptoms in patients with bipolar disorder (Szczepankiewicz et al., 2011). RNA expression of FKBP5 may be increased and also correlated with GR gene signaling. Reduced RNA expression of the GR has been demonstrated in bipolar as well as unipolar depressive disorder, in the hippocampi and frontal regions of post-mortem samples (Webster, Knable, O’Grady, Orthmann, & Weickert, 2002). In addition, there appear to be epigenetic effects of traumatic experiences in childhood on the HPA axis genes (specifically increasing methylation on the NR3C1 gene), which are related to both the perceived severity and number of events experienced early in life (Perroud et al., 2014).

Cognition and HPA activity in affective disorders

Evidence directly linking cognition, HPA measurement, and mood disorder is outlined in Table 13.1. Some further indirect links have been described in addition, such as a recent report that when some of the most hard-to-treat patients are treated intensively and strategically, those still showing non-response tend to have more severe anxiety and problems with sleeping, and greater impairments in cognition (Wooderson et al., 2014), and these factors have all been indirectly associated with the HPA axis.

In post-mortem brain tissue analysis, investigators have identified a smaller number of GRs and MRs in areas relating to cognition in those with bipolar and unipolar disorders and there are further well-established findings of brain regions involved closely in cognitive function that are modified in individuals with affective disorder: reduced hippocampal volume may represent the most publicized phenomenon, even found in unmedicated patients (Cole et al., 2010), though few studies have explored this in entirely treatment-naïve patients. Mice that have been engineered to have a reduced number of glucocorticoid receptors have shown substantial changes in HPA and cognitive function which improve in response to exogenous antidepressants, and preliminary similar results have been reported with mineralocorticoid receptors (McQuade & Young, 2000).

As expanded on in Table 13.1, cognitive deficits appear to be related to HPA axis activity: although two studies identified no significant correlation between cortisol and cognition, one of these did identify an effect of mood and six studies did show significant linkages. Some of this evidence suggests that HPA axis abnormalities may be causal in the association, as administration of glucocorticoid modifiers affects mood and cognition. These results are strengthened further by the recent identification that in healthy humans, hydrocortisone administration can improve performance on an executive function task, but cognition did not improve in people with MDD (Schlosser et al., 2013).

The implications of these observations for mood disorder treatments have begun to be explored; a randomized controlled trial for bipolar patients in a depressed state examined the mood effects of mifepristone (also named RU-486; a GR antagonist). In addition to previous findings that this drug can improve spatial working memory in people with bipolar disorder (Young et al., 2004) where patients’ symptoms also improved, Watson et al. (2012) have now demonstrated that with one week of mifepristone treatment in depressed bipolar patients, memory (of digits and spatial parameters) can improve for seven weeks after the treatment ended. This appeared to occur independently of changes in mood, or of cortisol measures at three or six weeks after treatment. The cognitive effect was correlated with cortisol levels in response to RU-486 at post-treatment, however, and this may indicate that the memory enhancement by mifepristone may be contributed to via mineralocorticoid receptor binding in hippocampal regions.

Evidence suggests that cognition and HPA associations may only significantly present in people with depression (and not healthy controls). Despite this, both HPA axis abnormalities (as suggested above) and cognitive deficits appear to persist through remission of symptoms in both unipolar and bipolar disorders (Robinson et al., 2006). From the above summary of the literature incorporating cognition, mood dysregulation, and HPA measurement, it is clear that mood, cognition, and the HPA axis share a complex and important relationship.

Other neurobiological associations with HPA axis, cognition, and mood disorders

While the complexities of these interactions are yet to be fully unraveled, other systems must be considered, which share notable associations with these constructs.

Neurogenesis: It is understood that chronic secretion of cortisol inhibits neuron formation via 5-HT receptors. Evidence showing that BDNF (brain-derived neurotrophic factor; a neurotrophin) is reduced in depressed populations has been consistently replicated (Hashimoto, Shimizu, & Iyo, 2004), and is definitively related to cognitive function. Indeed, it has been posited that an antidepressant mechanism of action may be through up-regulating neurogenesis in limbic regions (particularly the hippocampus), via GR (Anacker et al., 2011).

Monoamines: Serotonin especially has been heavily implicated in mood disorders and described widely elsewhere, but has noteworthy links with the endocrine system, primarily via the hippocampus where serotonin down-regulates CRF (see Feldman, Conforti, & Weidenfeld, 1995 for an overview). The relationship of the serotonergic system with cognition has also been examined comprehensively (Hughes et al., 2003). Although serotonergic medications have represented the majority of pharmacology for depression since its origin, it has been posited that selective serotonin reuptake inhibitors (SSRIs) can exert therapeutic effects directly through GR modulation, or indirectly via other neurobiological pathways (Reus & Wolkowitz, 2001).

Inflammation: Tryptophan (an amino acid that precedes serotonin production) synthesizes kynurenine which has an immune regulatory role. This is just one of numerous mechanisms through which the inflammatory response is linked with the HPA axis, and there remain contrasting theories on the intricacies of these connections (Anisman, Ravindran, Griffiths, & Merali, 1999). Elevated inflammation has now been widely demonstrated in affective disorders, and a greater activation of this response (mostly assessing pro-inflammatory cytokines) has also been associated with a poorer response to treatment (Strawbridge et al., 2015).

This brief outline of other neurobiological phenomena in mood disorders (and their interactions between the systems we are focusing on here) illustrates the multifaceted nature of biological psychiatry and bi-directional relationships involved. It also highlights a number of other potential treatment targets that may assist with improving illnesses of affect, specifically in the realm of stress, endocrinology, pharmacology, and neurocognition.