Sex differences of salivary cortisol secretion in patients with major depression

Abstract

Depression is associated with increased cortisol secretion and occurs more often in women than in men. Thus, it has been hypothesized that differences in cortisol secretion might, in part, be responsible for the greater risk of developing depression in women. However, only few studies have examined sex differences in baseline cortisol secretion in depressed patients and healthy controls. We examined sex effects on cortisol secretion in 52 medication-free patients with major depression (37 women, 15 men, mean ± SD age 35 ± 11 years, Hamilton Depression Scale mean score 27 ± 5) and 50 healthy age- and sex-matched control subjects. Salivary cortisol concentrations were measured at 8:00, 12:00, 16:00, and 22:00 h. Repeated measures analysis of covariance revealed a group × sex interaction (p = 0.05). Post hoc tests revealed higher cortisol concentrations in depressed compared to healthy men [F(1;29) = 7.5, p = 0.01]. No differences were found between depressed and non-depressed women. Our results do not support the hypothesis that differences in cortisol secretion between depressed and non-depressed subjects are more pronounced in women than in men. Study characteristics and methods as well as sex-specific confounding variables such as menstrual cycle, menopause and the use of oral contraceptives may account for inconclusive results across studies.

Keywords::

• Cortisol
• depression
• HPA axis
• psychoneuroendocrinology
• sex effects
• stress

Introduction

Increased activity of the hypothalamus–pituitary–adrenal axis (HPA) leading to elevated cortisol is a characteristic feature of depression (de Kloet et al. 2005; Holsboer and Ising 2010). Women have a greater lifetime risk than men to become depressed (Kessler et al. 2003). It has been hypothesized that sex differences in the HPA system may contribute to higher risk of depressive symptoms in females (Young and Korszun 2010).

However, sex differences in baseline cortisol secretion in depressed patients are rarely reported. The few existing studies have come to inconclusive results. Recently, Young and Korszun (2010) have summarized data regarding sex effects in the HPA axis and depression. They and other (Halbreich et al. 1984) reported that depressed female patients showed significantly higher mean cortisol concentration than their matched controls, whereas male patients and matched controls showed no differences regarding baseline cortisol. However, a study without a healthy control group (Grant et al. 2007) found higher cortisol secretion in depressed men compared to depressed women, whereas other studies did not find sex differences in baseline cortisol level (Deuschle et al. 1998; Weber et al. 2000). In summary, studies of sex-specific effects on HPA activity in depressed patients are equivocal.

We have previously reported increased HPA activity in depressed patients compared to healthy controls (Hinkelmann et al. 2009). Because of the hypothesized sex differences in HPA activity among depressed patients (Young and Korszun 2010), we conducted a post hoc analysis and aimed to investigate sex differences in baseline cortisol secretion in 52 non-psychotic, drug-free patients with major depression and 50 healthy matched subjects. We hypothesized that differences in salivary cortisol levels between depressed and non-depressed participants would be more pronounced in females than in males.

Methods

Subjects

We recruited 52 inpatients and outpatients (mean age 35, SD 11 years, 15 men and 37 women) from a specialized depression clinic at the Department of Psychiatry and Psychotherapy, University Hospital Hamburg, as described in Hinkelmann et al. (2009). Inclusion criteria were (1) a diagnosis of major depressive disorder, single or recurrent according to DSM-IV criteria, according to the MINI-interview (Sheehan et al. 1998) by two experienced psychiatrists (K.H. and C.O.); (2) a minimum baseline score of 18 points on the Hamilton Rating Scale for Depression, 17-item version (HAMD-17); (3) age from 18 to 70 years; (4) a period of at least 5 days free from antidepressants, antipsychotics, mood stabilizers, and other medications influencing HPA activity (benzodiazepines and zolpidem were allowed). Criteria for exclusion were (1) dementia, schizophrenia spectrum disorder, bipolar disorder, substance and alcohol dependence, and abuse within the last 6 months, (2) serious medical conditions, especially those associated with adrenal dysfunction, steroid use, or a condition with well-known impact on HPA activity (e.g. diabetes mellitus), and (3) pregnancy and nursing.

A control group of 50 healthy subjects (15 males and 35 females) matched for age and sex were enrolled in the study. Subjects were free of former and present DSM-IV axis I disorders according to the MINI-interview, had no physical illness, and had been free of any medication for at least 3 months.

The study was approved by the local ethics committee from the Hamburg Medical Council (Ethikkommission der Ärztekammer Hamburg). After complete description and explanation of the study to the subjects, written informed consent was obtained.

Cortisol assessment

Saliva for cortisol measurement was collected at 08:00, 12:00, 16:00, and 22:00 h. All participants received oral and written instructions on the correct use of the Salivette saliva collection device (Sarstedt, Germany). As inpatients were woken up by ward staff between 06:30 and 07:00 h, all participants were instructed to wake up between 06:00 and 07:00 h as well. Wake-up time was verbally confirmed, and participants were required to document the collection times on a form. Saliva collection was limited to weekdays. Participants were advised not to eat, drink, smoke, brush their teeth, or use mouthwash in the 30 min prior to saliva collection. Cortisol concentration was determined by radioimmunoassay (DRG, Marburg, Germany). Inter- and intra-assays coefficients of variation were below 8%. Detection limits were 0.5 ng/ml for cortisol.

Statistical analyses

Differences in demographic characteristics between patients and healthy controls for each sex were compared using t-tests for continuous variables and chi-square tests for dichotomous variables. As female patients and healthy controls differed significantly in their smoking habits, all analyses were adjusted for smoking.

Mixed analyses of covariance (ANCOVAs) were conducted to investigate differences in cortisol concentrations with group (healthy vs. depressed) and sex as between-subjects factor and time (four collection time points) as within-subjects factor.

In all analyses, two-sided tests were used and as nominal level of significance, α = 0.05 was accepted.

Results

Participants

Patients were moderately to severely depressed with a mean Hamilton Score of 27.5 (SD 4.5). Out of 52 patients, 35 (67.3%) were inpatients. The number of inpatients and outpatients did not differ between male and female patients (chi-square, p = 0.75). There were no significant differences between patients and healthy controls in demographic variables except smoking, which was more frequent in female patients compared to female controls (see Table I).

Table I.  Demographic variables of study participants.

	Men			Women
	Men DEP N = 15	Men HC N = 15	p	Women DEP N = 37	Women HC N = 35	p
Age (SD)	36.7 (10.2)	36.3 (9.7)	n.s.	34.9 (12.2)	35.0 (12.4)	n.s.
Education (years)	11.4	11.3	n.s.	11.2	11.5	n.s.
BMI (SD)	24.8 (6.5)	25.1 (4.0)	n.s.	24.0 (6.6)	22.3 (3.2)	n.s.
Smokers	53%	40%	n.s.	49%	23%	0.02
BDI (SD)	28.1 (13.1)	2.6 (2.5)	< 0.01	32.0 (8.3)	3.5 (2.8)	< 0.01
HAMD (SD)	26.8 (5.4)	n.a.	n.a.	27.4 (4.1)	n.a.	n.a.
Oral contraceptives	n.a.	n.a.	n.a.	6	8	n.s.

Note: BDI, Beck Depression Inventory; BMI, body mass index; HAMD, Hamilton Depression Scale; DEP, depressed patients; HC, healthy control subjects; n.a., not applicable; n.s., not significant.

Salivary cortisol concentrations

Complete sets of saliva for cortisol measurement (sufficient material at all time points) were available from 33 patients and 43 healthy controls. Missing values were imputed by linear interpolation. The areas under the curve (AUC) for cortisol measurements of completers and non-completers were not significantly different.

Patients exhibited significantly higher salivary cortisol concentrations compared to healthy controls as reported previously (Hinkelmann et al. 2009). In this post hoc analysis, we conducted a repeated measures ANCOVA with the four collection time points as within-subject factor and group and sex as between-subjects factors and found a group × sex interaction [repeated measures ANCOVA, adjusted for smoking, F(1;101) = 3.8, p = 0.05] on cortisol secretion.

Therefore, we conducted analyses separately in each sex: repeated measures ANCOVA (adjusted for smoking) with the four collecting times as within-subject factor, and group (healthy vs. depressed) as between-subjects factor revealed significantly higher cortisol concentrations in depressed men than in healthy men (main effect for group with F(1;29) = 7.5, p = 0.01, see Figure 1). In addition, cortisol AUC was significantly higher in depressed compared to healthy men (univariate ANOVA, adjusted for smoking, effect for group F(1;29) = 4.8, p = 0.04). Cortisol concentrations of depressed women were not significantly different from those of matched female controls [F(1;71) = 1.2, p = 0.26].

Figure 1.  Salivary cortisol secretion in depressed patients (37 women and 15 men) and healthy controls (35 women and 15 men). Repeated measures ANCOVA (adjusted for smoking) revealed a group × sex interaction [p = 0.05]. Post hoc tests revealed significantly higher cortisol in depressed men than in healthy matched male controls [p = 0.01], whereas no significant differences were found between depressed and non-depressed women. DEP, depressed patients; HC, healthy controls. Lines represent group means, and error bars represent SEM.

Discussion

In this study, we examined sex effects on baseline salivary cortisol concentrations in medication-free depressed patients compared to healthy controls. In our study sample, depressed men showed significantly higher cortisol secretion than their age-matched controls. There were no significant differences in baseline cortisol concentrations in depressed women compared to healthy women. Comparing all four groups, depressed men exhibited the highest and healthy men the lowest salivary cortisol concentrations.

Our results do not support the hypothesis that differences in basal HPA-axis activity between depressed and non-depressed individuals are more pronounced in women than in men. In contrast, in our study, we found a larger effect of depression on cortisol secretion in men. Our results are in line with and extend results from an earlier study (Grant et al. 2007), which found higher 24-h urinary cortisol levels in depressed men compared to depressed women and an association of cortisol and symptom severity in depressed men but not in depressed women. However, that study did not include a healthy control group limiting the conclusions that could be drawn from the results. Our results are also consistent with Binder et al. who reported that a measure of HPA-axis dysregulation, the dexamethasone–corticotropin-releasing hormone test (Dex–CRH test), at admission might serve as a predictor of response to antidepressant treatment in male but not in female patients with major depression (Binder et al. 2009).

Our findings are contrary to other studies (Halbreich et al. 1984; Young and Korszun 2010), which report that depressed female patients showed higher cortisol levels compared to both depressed men and non-depressed female controls, whereas cortisol concentrations in depressed vs. non-depressed men were not significantly different. Yet other studies did not find a sex effect regarding cortisol secretion in depressed patients (Deuschle et al. 1998; Weber et al. 2000).

Consideration of sample characteristics and methodological differences may shed some light onto these inconclusive results across studies. Age has a great impact on study results when investigating HPA-axis activity (Otte et al. 2005). Plasma cortisol concentrations increase significantly with age in depressed patients of both sexes (Halbreich et al. 1984). In healthy subjects, however, cortisol concentrations seem to increase age dependently in women during reproductive age (Halbreich and Lumley 1993). Other studies suggest that cortisol concentrations increase gradually with age in both men and women (Van Cauter et al. 1996), but in women especially during the menopause transition (Woods et al. 2009).

Our study sample was rather young (mean age of 35 years) with a restricted age range, comparable to the sample Grant et al. (2007) and Young and Korszun (2010) studied. Other studies investigated older patient samples (mean age 47 years), and it seems noteworthy that sex differences did not emerge in these studies (Deuschle et al. 1998; Weber et al. 2000). Also, Halbreich et al. (1984) found an effect of sex only comparing the younger subgroup (the sex effect was significant for premenopausal women only).

Other covariates influencing the HPA axis that are also in part age related are levels of gonadal steroids and the use of oral contraceptives (Young and Korszun 2010). In our study sample, only few women were postmenopausal (5 women aged >50 years) and the rate of oral contraceptive use was not different in healthy and depressed women. Overall, sex-specific influences on HPA activity such as the changing steroid milieu during the menstrual cycle, use of oral contraceptives, and menopause status may outweigh the effects of depression on HPA activity in women.

Lastly, the two studies that found more pronounced effects of depression in men than in women measured free cortisol, in saliva and urine, whereas the other studies report plasma cortisol level (Halbreich et al. 1984; Deuschle et al. 1998; Weber et al. 2000; Young and Korszun 2010).

Free cortisol is affected by the amount of corticosteroid-binding globulin (CBG). Furthermore, CBG levels are altered by estrogens (Kirschbaum et al. 1999). Kumsta et al. (2007) reported a negative correlation between CBG and salivary cortisol concentrations after stress exposure in women taking oral contraceptives. Therefore, sex-specific influences on HPA activity such as the changing steroid milieu during the menstrual cycle, use of oral contraceptives, and menopause status may have a stronger effect on free cortisol than total plasma cortisol and may diminish the effects of depression on HPA activity in women when measuring free cortisol.

A limitation of our study is that no statement can be made about the influence of hormonal status, because we did not differentiate between luteal and follicular phase in the female groups and did not measure estrogen concentrations. However, the use of oral contraceptives was almost identical in depressed vs. non-depressed women and only a few women in the total sample were postmenopausal. Another limitation is our restricted age range and the relatively small sample size, hampering our ability to analyse interactions among depression status, sex, and age. Furthermore, HPA-axis dysfunction has been reported to depend on the subtype of depression (melancholic vs. atypical) and hospitalization status (inpatients vs. outpatients) (Peeters et al. 2003; Knorr et al. 2010; Stetler and Miller 2011). However, neither inpatients vs. outpatient status nor the proportion of melancholically or atypically depressed patients differed between men and women in our sample. Finally, in this study, we were unable to examine if HPA activity is a preexisting risk factor for the development of major depression. A longitudinal-study with cortisol measures prior to, during, and after a depressive episode in a large sample is needed to disentangle cause–effect relationships between cortisol secretion and depression and putative sex differences in this association.

In summary, our results do not support the hypothesis that differences in the effects of depression on alterations in cortisol secretion are more pronounced in women than in men. On the contrary, the impact of sex-specific variables such as menstrual cycle stage, use of oral contraceptives, and menopause status may diminish differences between depressed and non-depressed women, especially when measuring free cortisol. Future studies should systematically examine the moderating effects of sex and age on free vs. total plasma cortisol levels in depressed vs. non-depressed subjects.

Acknowledgments

We are grateful to the excellent technical assistance of Iris Remmlinger-Marten and Kirsten Huwald. This work was supported by the German Research Foundation (Deutsche Forschungsgemeinschaft, grant OT 209/3-1, 3-2). The German Research Foundation had no role in the collection of data, interpretation of results, or preparation of this manuscript. We have no conflict of interest.

Declaration of interest: Dr Wiedemann served as a consultant to or has been on the speakers boards of AstraZeneca, Bristol-MyersSquibb, Janssen, Pfizer, Servier and Wyeth. Dr Kellner received funding for investigator initiated trials by Lundbeck and Pfizer. He is a member of an advisory board for Wyeth. Dr Otte is on the speaker's board of AstraZeneca, Lundbeck, and Servier. Dr Hinkelmann, Dr Moritz, Dr Muhtz, and Dr Botzenhardt report no conflict of interest.