Introducing a novel method to assess cumulative steroid concentrations: Increased hair cortisol concentrations over 6 months in medicated patients with depression

Abstract

Depression has been linked to increased cortisol concentrations using point measures taken from urine, blood, or saliva samples. However, with regard to hypercortisolism-induced consequences, long-term cumulative cortisol burden is of relevance. Our objective was to use hair analysis as a new method to assess cortisol exposure over 6 months in depressed patients and healthy controls. We examined 23 depressed patients (8 men and 15 women, mean age: 41.6 years ( ± standard deviation (SD), 13.1 years); mean duration of current depressive episode 9 months ( ± SD, 13 months)) and 64 healthy controls, matched for age and gender. Cortisol concentrations in two 3-cm hair segments from near to the scalp were analyzed, representing cortisol secretion during the 6 months prior to sampling. Compared with healthy individuals, depressed patients had higher hair cortisol concentrations in the first (mean ± SD: 26.7 ± 20.8 vs. 18.7 ± 11.5 pg/mg, p < 0.05) and second hair segment (mean ± SD: 21.9 ± 23.7 vs. 13.4 ± 9.6 pg/mg, p < 0.05). In conclusion, hair cortisol analysis confirmed enhanced cortisol secretion in depressed patients over a prolonged time period. Because of the retrospective and cumulative nature of cortisol in hair, the assessment of hair cortisol concentration may help in addressing unanswered questions regarding hypothalamic–pituitary–adrenal axis overactivity and associated health consequences in psychiatric disorders.

Keywords::

• Cortisol
• depression
• glucocorticoids
• hair analysis
• HPA axis
• stress

Introduction

Cortisol, the end product of the hypothalamus–pituitary–adrenal (HPA) axis, is a potent endogenous glucocorticoid regulating a wide range of bodily functions including metabolism, immunity, neuronal survival, and neurogenesis. Based on its involvement in brain functions, cortisol has been studied abundantly in psychiatric disorders, and in particular in major depression, with largely consistent findings of elevated cortisol secretion in depressed patients than in controls (Nemeroff and Vale 2005; Vreeburg et al. 2009). These findings refer to cortisol concentrations found in saliva, plasma, and urine, all of which are point assessments (saliva and plasma) or reflect cumulative concentrations over a few hours only (urine). However, with regard to supposed hypercortisolism-induced health consequences of depression, including metabolic risk (Weber-Hamann et al. 2002; Muhtz et al. 2009), cognitive impairment (Hinkelmann et al. 2009), and hippocampal atrophy (Vythilingam et al. 2004), long-term cumulative cortisol exposure seems to be of great relevance. More specifically, hypercortisolemia is considered to lead to structural alterations in the brain including cell apoptosis and consequent volume reductions of the prefrontal cortex and the hippocampus. The amygdala, by contrast, evidently increases in size in response to hypercortisolemia (Sharpley and Bitsika 2010). Interestingly, these specific organic changes appear to be associated with the development of depressive symptomatology. Hence, it may well be that hyperactivation of the HPA axis, one of the main endocrine stress axes, is a key element in the link between exposure to stress and subsequent development of depression symptomatology.

Recently, segmental hair analysis, a method long known from forensic science and toxicology to capture cumulative analyte exposure in a retrospective fashion, has been introduced into biomedical research (Sauve et al. 2007; Van Uum et al. 2008; Gao et al. 2010; Stalder et al. 2010; Thomson et al. 2010). Hair cortisol measures, similar to salivary cortisol measures, are considered to reflect free cortisol levels. However, conclusive evidence is lacking and several incorporation pathways have been proposed including diffusion from blood, body secretion, and deep skin compartments (Pragst and Balikova 2006). A separate line of research indicates that hair follicles may function as independent peripheral neuroendocrine organs which not only respond but also synthesize and/or metabolize key molecular players of the HPA axis (Ito et al. 2005). The potential interplay between systemic stress responses and local responses at the hair follicle has been suggested but data on possible feedback loops between the two systems are lacking (Arck et al. 2006). However, the work of Davenport et al. (2006) on rhesus monkeys showed high correlations between hair cortisol levels and eight averaged post-stress salivary cortisol concentrations, indicating that hair cortisol levels may indeed reflect adrenal cortex secretion. Recent work by Sharpley et al. (2009, 2010), however, indicates that the cortisol response in hair to a cold pressor test is (a) faster and briefer than central HPA axis activity, (b) confined to the site of stressor impact (i.e. arm vs. leg), and (c) independent of the central HPA axis response. In sum, it remains unclear whether cortisol concentrations in hair reflect intrafollicular HPA axis secretion, adrenal cortex secretion, or both.

Although some studies reporting on the relationship between hair cortisol concentrations and depressive symptoms have provided conflicting results (Kalra et al. 2007; Dowlati et al. 2010), to our knowledge no study so far has examined cumulative cortisol exposure measured via hair analysis in clinically depressed patients. Based on a mean hair growth rate of approximately 1 cm/month (Wennig 2000), this method allows for retrospective measurement of cumulative cortisol concentrations over up to 6 months. More extended retrospective measurement is currently limited due to a possible washout effect for hair cortisol, which has been found in some studies (Kirschbaum et al. 2009; Dettenborn et al. 2010), but not all (Thomson et al. 2010; Manenschijn et al. 2011). It may well be that differences in sample preparation between studies were responsible for this effect as variation over time but no overall decline was found in studies that did not apply a washing procedure to the hair to be analyzed (Manenschijn et al. 2011). Here, we hypothesized that depressed individuals would have increased hair cortisol concentrations compared with healthy controls as measured by the first 3-cm hair segments near to the scalp.

Materials and methods

Participants

Participants were 23 depressed patients (8 men and 15 women, mean age ± standard deviation (SD): 41.6 ± 13.1 years) treated at the Department of Psychiatry and Psychotherapy, University Medical Center, Hamburg, Germany, and 64 age- and gender-matched control participants (mean age ± SD: 39.9 ± 12.4 years). All depressed individuals met DSM-IV criteria for major depressive disorder, single or recurrent, according to the Mini-International Neuropsychiatric Interview (MINI; Sheehan et al. 1998) and the Hamilton Depression Rating Scale (HAMD-17). Further, the Beck Depression Inventory (BDI) was employed. Participants were screened for psychiatric co-morbidity with the MINI. For demographic and health-related characteristics (gender, age, body mass index, smoking status, and current medication intake), participants were asked to answer a short self-developed questionnaire.

Exclusion criteria were (i) dementia, schizophrenia spectrum disorder, bipolar disorder, substance dependence < 6 months according to the MINI, (ii) serious medical conditions, including those suspected to be associated with alterations of the HPA axis, and (iii) use of steroid medication. Control participants were recruited as part of a larger study. They were screened for current illness and medication intake. Controls on current antidepressant drugs were excluded from the present study. All participants provided written informed consent prior to study participation. The study protocol was approved by the ethics committees of the Medical Schools in Hamburg and Dresden. All study procedures were in accordance with the Declaration of Helsinki.

Hair cortisol analysis

Hair strands of a total thickness of approximately 3 mm (diameter) were taken from the scalp—near posterior vertex position. Cortisol concentrations were determined from the first 3-cm hair segments most proximal to the scalp and from the adjacent 3-cm segment, if hair length permitted. Based on an average hair growth rate of 1 cm/month (Wennig 2000), hair segments were assumed to represent the hair grown over the 3-month period prior to hair sampling (first segment) and a period of 4–6 months prior to sampling (second segment). Wash and steroid extraction procedures followed the laboratory protocol described in detail in Kirschbaum et al. (2009) with 25 mg of powdered hair being used for analyses in the present study. Cortisol concentrations were determined using a commercially available chemo-luminescence-assay with high sensitivity of 0.16 ng/ml (CLIA, IBL-Hamburg, Germany). Intra and interassay coefficients of variance were below 10%.

Statistical analysis

Hair samples were analyzed from a total of 87 participants (23 depressed individuals and 64 controls). Due to varying hair length, the analysis for the second hair segment was conducted in 19 depressed individuals and 50 controls. Means and SDs are expressed in pg/mg. Group comparisons regarding sociodemographic characteristics were conducted using univariate ANOVAs (continuous variables) or χ² contingency tables (dichotomous variables). When differences between groups occurred, the respective variables were checked for associations with hair cortisol measures. In case of significant relationship with hair cortisol concentrations, these variables were introduced as covariates in the subsequent univariate and repeated measures ANOVAs. Following Tabachnick and Fidell (2001), we checked for violations of homogeneity of variances (Levene's test) and applied log transformations, which effectively eliminated the violation. Two separate univariate ANOVAs for group differences in each hair segment were run to conserve the larger sample size for the first hair segment. Additionally, a repeated measures ANOVA was run to validate findings and to explore a possible washout effect seen in previous samples. Pearson correlations were run to check for associations between depression severity, length of current episode, and number of past episodes with hair cortisol concentrations.

Results

Although the two groups were well matched on demographic characteristics, there were more smokers in the depressed group than in the control group (see Table I). However, ANOVA revealed that smoking status was unrelated to hair cortisol concentrations (p = 0.92). Inherent to the nature of the study, all but one depressed patient was on at least one antidepressant.

Table I.  Comparisons of sociodemographic characteristics of depressed individuals and healthy controls.

Sociodemographic characteristics	Depressed (n = 23)	Controls (n = 64)	p value
Age	41.6 ± 13.1	39.9 ± 12.4	p = 0.58
Gender: females	15 (65.2%)	47 (73.4%)	p = 0.46
Body mass index	23.8 ± 3.9	23.2 ± 2.8	p = 0.46
Habitual smokers	8 (36.4%)	9 (14.1%)	p < 0.05
Non-psychotropic medication*	3 (13.0%)	14 (21.9%)	p = 0.36
Psychoactive medication†	22 (95.7%)
SSRI	12
Tricyclic AD	1
Dual AD	7
Other AD	2
Atypical antipsychotic	7
Mood stabilizer/lithium	4
Benzos/sleep meds	4
Inpatient	17 (77.3%)
Recurrent depression	15 (65.2%)
Mean duration of current depressive episode (months)	9.1 ± 13.8
BDI	24.3 ± 10.4
HAMD-17	17.5 ± 5.6
Psychiatric comorbidities	12 (52.3%)
Personality disorder	5
Anxiety disorder	5
Eating disorder	1

Note: Values are means ± SDs (range) and numbers of participants (percentage); BDI, Beck Depression Inventory; HAMD-17; Hamilton Depression Rating Scale; SSRI, selective serotonin reuptake inhibitor; AD, antidepressant; Dual AD, venlafaxine, mirtazapine, or agomelatine

*These included mainly thyroid medication, beta blocker, and pain medication

†Total numbers in antidepressant classes are greater than number of participants due to intake of more than one type of antidepressant in most cases.

Hair cortisol concentrations of both 3-cm segments were found to differ between the two groups (F_1,85 = 5.0, p < 0.05, η² = 0.06 and F_1,67 = 5.6, p < 0.05, η² = 0.08, see Figure 1). Depressed patients had approximately 50% higher hair cortisol concentrations in both hair segments (segment 1: 26.7 ± 20.8 vs. 18.7 ± 11.5 pg/mg; segment 2: 21.9 ± 23.7 vs. 13.4 ± 9.6 pg/mg, means ± SD). Repeated measures ANOVA revealed that cortisol concentrations decreased from the first to the second hair segment (p>0.001) and confirmed the larger cortisol concentrations in depressed patients across hair segments (p < 0.05). Results were identical when conducting the analyses without participants on non-psychotropic medications (all p < 0.05). Depression severity, number of past episodes, or length of current episode did not correlate with hair cortisol concentrations (all p>0.3).

Figure 1.  Hair cortisol concentrations in depressed and healthy control participants in the near scalp (segment 1, 23 depressed vs. 64 controls) and adjacent (segment 2, 19 depressed vs. 50 controls) 3-cm segment (means and SEs).

Discussion

This is the first study to report increased cortisol concentrations in hair samples of clinically depressed patients compared with controls, indicating increased cumulative cortisol exposure over the 6 months period prior to sampling. Our results suggest that the hair analysis for cortisol is a valuable tool for examining long-term dysregulation of the HPA axis in a retrospective fashion.

Our findings of HPA axis hyperactivity extend previous results on increased cortisol concentrations captured via saliva, blood, and urine sampling (Nemeroff and Vale 2005; Vreeburg et al. 2009). Interestingly, most consistent results have been found in reports studying medicated inpatients with more severe melancholic or psychotic depression (Maes et al. 1994; Nelson and Davis 1997) as is the case in the present study. These findings may appear contradictory to the assumption that antidepressants dampen HPA axis activity, implying that cortisol concentrations in medicated patients are somewhat comparable with those in healthy controls. Studies of antidepressant effects on cortisol release, however, have produced inconsistent results (Bschor et al. 2002; Adli et al. 2009; Manthey et al. 2010; Bicikova et al. 2011). Some argue that the HPA system-dampening effect of antidepressants found in some studies but not all depends on (a) the class of antidepressant, (b) the clinical response to treatment, and (c) the type of HPA measure (Kauffman et al. 2005; Schüle et al. 2006; Scharnholz et al. 2010). As hair cortisol measurement is a novel method, studies investigating the possible effects of antidepressants on this particular HPA axis measure are lacking. Future studies should address antidepressant effects on hair cortisol more carefully by, for example, studying medicated vs. unmedicated depressed patients.

These results may have important implications for future studies linking elevated cortisol concentrations in depressed patients to health risks, including cardiovascular morbidity (Weber-Hamann et al. 2002; Muhtz et al. 2009), cognitive impairment (Hinkelmann et al. 2009), and hippocampal atrophy (Vythilingam et al. 2004), among others. So far, few studies have investigated cortisol concentrations as a mediating factor between depression and disease risk, with mostly inconsistent results. For example, while some studies reported associations between increased cortisol concentrations and cognitive impairment in depressed patients (Gomez et al. 2006; Hinkelmann et al. 2009), others failed to find an association (O'Brien et al. 2004; Reppermund et al. 2007; Michopoulos et al. 2008). Such inconsistencies may be due to methodological limitations related to point assessments of cortisol level, which are known to be biased by situational factors resulting in high intraindividual variability. Long-term cumulative cortisol measures, as captured via hair analysis, may be better suited to study hypercortisolism-induced health consequences. Further, important implications are related to findings indicating that higher basal cortisol concentrations are associated with recurrence and a longer time to recovery in depressed individuals (Rao et al. 2010). Hair cortisol concentrations may be a better predictor for the longitudinal clinical course of depression. A better understanding and assessment of the factors that predict a chronic and recurrent course of depression are key to developing and implementing more effective treatment and prevention strategies.

Currently, it remains unclear whether cortisol concentrations in hair reflect intrafollicular production, adrenal cortex secretion, or both. Therefore, we cannot rule out that the increased cortisol levels in the hair of depressed patients found in this work reflect follicle-based production only. However, support for the idea that hair cortisol levels may indeed reflect adrenal production stems from the data indicating (a) high correlations between hair cortisol levels and salivary cortisol levels (Davenport et al. 2006; D'Anna-Hernandez et al. 2011), (b) high hair cortisol levels in known hypercortisolemic states (i.e. Cushing's Syndrome, Thomson et al. 2010), (c) correlations between hair cortisol and waist-to-hip-ratio (Manenschijn et al. 2011), indicating the effects of cortisol at the tissue level, and (d) high hair cortisol levels among stressed individuals (Kalra et al. 2007; Van Uum et al. 2008; Dettenborn et al. 2010). Nevertheless, these studies only provide indirect evidence for adrenal sources of hair cortisol. Therefore, microbiological studies are needed to assure and possibly quantify sources of hair cortisol levels.

The present study set out to investigate a wide range of depressed patients seeking help at the Department of Psychiatry and Psychotherapy, University Medical Center, Hamburg, Germany, hence avoiding overly restrictive inclusion criteria. While we consider this as a strength of our study, we did not have the power to analyze hair cortisol concentrations with regard to different disease or medication characteristics. Mainly medicated, treatment-seeking patients were included, the majority of whom had already been treated as inpatients for a number of weeks before study inclusion. Consequently, some of the patients had relatively low Hamilton scores at enrollment. However, one particular strength of hair analysis is its retrospective nature allowing the capture of cortisol burden over several months, thus covering time periods of more severe depressed mood states before treatment. This also explains our lack of association between symptom severity scores and hair cortisol concentrations. The BDI covers symptom severity over the last 2 weeks, whereas the hair cortisol measures reported here represent net cortisol output over the 6 months prior to hair collection. This retrospectivity of hair cortisol assessment also applies when considering that at the time of hair collection most patients were on antidepressive medications that have been shown to inhibit HPA axis activity (Schüle 2007). The present hair cortisol assessments reflect net cortisol output over a time frame going beyond medication intake, thus, medication-induced HPA axis dampening should be a negligible influence on our results. If at all, the inclusion of medicated patients in our study represents a bias toward the null hypothesis. Future studies should analyze the role of atypical and melancholic subtypes, influence of comorbidities, and medicated vs. unmedicated patients on hair cortisol level.

In sum, this is the first report of increased cortisol burden over 6 months in depressed patients compared with healthy controls by applying hair steroid analysis. Due to its cumulative and retrospective nature, this novel method provides the opportunity to address research questions that have been difficult to study using established endocrine point assessments.

Acknowledgement

We thank the participants for their valuable time and effort to the study.

Declaration of interest: Funding for this study was provided by the German Research Foundation (DFG, DE 1162/3-1). The DFG had no further role in study design, collection, analysis, interpretation of data, writing of this report, and in the decision to submit this paper for publication. Christian Otte is on the speaker's board of Servier, AstraZeneca, and Lundbeck. All other authors report no conflicts of interest. The authors alone are responsible for the content and writing of the paper.