Psychological reactivity to laboratory stress is associated with hormonal responses in postmenopausal women

Abstract

Objective

The present study examined associations between psychological reactivity and hormonal responses to a standardized laboratory stressor (Trier Social Stress Test, TSST) in postmenopausal women.

Methods

Postmenopausal women aged 50–74 years undertook anxiety and mood assessments prior to and following the TSST. Blood samples were drawn at multiple timepoints for assessment of cortisol, adrenocorticotrophic hormone (ACTH) and dehydroepiandrosterone (DHEA).

Results

Forty postmenopausal women completed the assessments. As expected, significant increases in anxiety and negative affect and decreases in positive affect were observed after the TSST; however, the magnitude of change in anxiety and mood varied considerably across individuals. Analyses indicated that greater increases in anxiety and negative affect after the TSST were associated with higher levels of cortisol, ACTH and DHEA after controlling for race, age, body mass index and smoking status. Changes in positive affect were not associated with cortisol, ACTH or DHEA.

Conclusions

These findings suggest that enhanced reactivity to stress is associated with higher hormone levels among postmenopausal women, which could have potential implications for health.

Introduction

The neuroendocrine stress response has been extensively studied using both naturalistic and laboratory stress paradigms. ¹ As a result, it is well established that stress exposure leads to corticotrophin-releasing hormone (CRH) and arginine vasopressin (AVP) discharge from the hypothalamus. ² CRH and AVP synergize to stimulate pituitary adrenocorticotrophic hormone (ACTH) production, which increases adrenal production of cortisol and various androgens, including dehydroepiandrosterone (DHEA) and its sulphate (DHEAS). ³ Stress-related alterations in hypothalamic–pituitary–adrenal (HPA) axis function and its effects on ACTH, cortisol and DHEA have been described in many studies and across various populations. ⁴

The effects of stress on DHEA and DHEAS may be particularly relevant for breast cancer risk in postmenopausal women. Several prospective studies have reported that elevated levels of these adrenal androgens are associated with increased risk of breast cancer among postmenopausal women.^5–9 In a pooled analysis of prospective studies, ¹⁰ postmenopausal women with elevated DHEA and DHEAS, which are almost exclusively derived from the adrenals, ¹¹ were twice as likely to develop breast cancer compared with women with low levels. The cortisol level has also been reported to be elevated in women with breast cancer ¹² and may increase risk indirectly through its effects on metabolism: for example, cortisol increases adiposity, ¹³ an established risk factor for breast cancer in postmenopausal women. ¹⁴

Despite the well-documented effects of stress on HPA axis function and its downstream hormonal pathways, studies of stress and breast cancer risk have yielded mixed results. ¹⁵ Some studies have noted a positive association between stress and breast cancer risk,^16–18 whereas others have not.^19–22 The mixed findings may be due to a variety of factors, including behavioral, biological and reproductive factors, and differences in psychological reactivity to stress. ¹⁵ Studies have noted that variations in psychological stress reactivity lead to differential physiological responses to stress. ²³ Indeed, measures of sympathetic nervous system activity (e.g. cardiac activation and systemic vascular resistance) have been found to vary by participants’ perceptions of the stressor.^24,25

The Trier Social Stress Test (TSST) is a standardized protocol consisting of evaluative speech and mental arithmetic tasks, which has been commonly used to induce social stress in a laboratory setting. The TSST has been found to reliably stimulate ACTH, cortisol and DHEA in healthy men and women,^4,26,27 although considerable variation in the magnitude of response exists. ²⁸ Variation in neuroendocrine reactivity to the TSST has been investigated with respect to a variety of factors, including race, ²⁹ sex, ³⁰ age ³¹ and various clinical conditions,^32–34 but data also suggest that subjective psychological responses (e.g. affect, mood, perceived stress) are associated with cortisol reactivity.^28,35 For example, increases in stress and negative affect were positively associated with increased cortisol reactivity in studies of adult men^36,37 and adolescents. ³⁸ Studies have also reported stress-induced increases in DHEA during the TSST, including large variations in the magnitude of DHEA response.^26,27

Thus, the purpose of the present study was to examine differences in psychological reactivity to a standardized laboratory stressor (TSST) in relation to hormonal reactivity among postmenopausal women. Specifically, it was hypothesized that greater increases in anxiety and negative affect, and decreases in positive affect, after the TSST would be associated with greater ACTH, cortisol and DHEA reactivity to the TSST.

Subjects and methods

Criteria for participation

The study included healthy postmenopausal women aged 50–74 years. The exclusion criteria were: having both ovaries removed; history of cancer other than nonmelanoma skin cancer; history of Cushing’s or Addison’s disease, coronary heart disease, stroke, uncontrolled hypertension or other health conditions that could place the participant at risk; Alzheimer’s disease or cognitive deficit; taking any medications that could interfere with assessment of biological outcomes such as oestrogens, progestogens, androgens, prednisone or psychoactive drugs.

Recruitment

Participants were recruited between February 2008 and June 2009, using local newspapers from the Philadelphia, Pennsylvania region, and online classified advertisements. Eligible women who consented to participate provided written informed consent and were scheduled for the TSST.

Ethics

This study was approved by the Institutional Review Boards at Fox Chase Cancer Center (IRB no. 07-840) and the University of Pennsylvania (IRB no. 808042).

Administration of the TSST and blood sampling

Depending on participant location, the TSST procedures were conducted at either the Fox Chase Cancer Center Clinical Research Unit or the University of Pennsylvania Clinical and Translational Research Center. The TSST was administered during a 3-h visit using a modification of the protocol by Singh et al.^39,40 Participants were instructed to abstain from caffeine, alcohol, cigarettes and strenuous activities for 24 h and to fast for between 2 h and 3 h before arriving for the TSST. Participants were also instructed to arrive 75 min before the start of the TSST protocol, which began at ∼13:00 h for all participants. First, participants’ characteristics and mood were recorded. At 60 min before the test, participants drank water (5 ml/kg) to insure uniform hydration, and 40 min before the test, an intravenous (i.v.) catheter (with saline lock) was placed in the forearm vein for blood sampling. The TSST was immediately preceded by a 20-min waiting period, during which participants were instructed to sit quietly, then the TSST was administered. Specific tasks in the TSST included preparing for (5 min) and then delivering (5 min) a speech and a mental arithmetic task (5 min). Following the TSST, participants sat quietly during a recovery period.

Blood samples for ACTH, cortisol and DHEA were collected through the i.v. catheter at 1 min before (−1 min) and 1, 15 and 30 min after the TSST, and additionally for cortisol and DHEA at 60 and 90 min following the TSST. Immediately following the last blood sampling point, participants completed measures of post-TSST anxiety and mood. Blood samples were collected using one 5-ml purple-top (ethylenediaminetetra-acetic acid) plastic tube and one 10-ml red-top (plain) glass tube and held at room temperature for ≥30 min to allow complete clotting. After aliquoting, samples were stored at −80℃. All blood samples were transported to Fox Chase Cancer Center for processing and storage (for ≤18 months) prior to analysis.

Participant characteristics

The height, weight and waist circumference of each participant were measured. For demographic and socio-economic characteristics, participants completed a brief questionnaire that contained items on demographic background, health behavior, age, education, race/ethnicity, marital status and smoking.

Anxiety and mood scales

Levels of anxiety are commonly assessed prior to and following the administration of the TSST. ⁴¹ Self-reported anxiety was measured using the Beck Anxiety Inventory (BAI), which is a 21-item list of anxiety-related symptoms and feelings.^42,43 Items were summed to obtain a total score, with higher scores reflecting greater anxiety. Cronbach’s α in the present sample was 0.90 indicating high internal consistency.

Mood was assessed using the Positive and Negative Affect Scale (PANAS). ⁴⁴ This 20-item scale comprises two mood scales: one assessing positive affect and the other measuring negative affect. This well-validated scale has been widely used in previous studies of stress reactivity, in which increases in negative affect and decreases in positive affect were observed following the TSST.^32,45,46 Each of the 20 items is rated on a 5-point scale ranging from 1 = ‘Very slightly or not at all’ to 5 = ‘Extremely’ and participants are instructed to indicate the extent to which they felt at the time. The 10 positive affect items were summed to create a positive affect score, with higher levels indicating more positive affect. Similarly, the 10 negative affect items were summed to create a negative affect score, with higher levels indicating more negative affect and distress. In the present sample, the Cronbach’s alpha coefficient was 0.86 for Positive Affect and 0.83 for Negative Affect.

Hormone assays

Cortisol, ACTH and DHEA assays were conducted in the Reproductive Endocrine Research Laboratory at the Keck School of Medicine, University of Southern California. All samples were assayed in duplicate. Blood samples were batched at Fox Chase Cancer Center and shipped overnight on dry ice to the laboratory. Each participant’s samples were assayed together in the same batch.

Cortisol was measured by a solid-phase competitive chemiluminescent enzyme immunoassay on an Immulite® 2000 analyser (Siemens Healthcare, Erlangen, Germany). ⁴⁷ The interassay coefficient of variation (CV) was 7.3%. The sensitivity of the assay is reported to be 20 µg/dl.

Plasma samples for ACTH were collected on ice and processed within 1 h of collection. ACTH was measured by a solid-phase two-site sequential chemiluminescent immunometric assay on an Immulite® 2000 analyser. ⁴⁸ The interassay CV was 8.7%. The sensitivity of the assay is reported to be 5 pg/ml.

Next, DHEA in serum was extracted and measured by radioimmunoassay. ⁶ Briefly, following extraction using ethyl acetate/hexane (2 : 3) and evaporation of the organic solvents, ⁵ the residue was redissolved in iso-octane and applied on a Celite column impregnated with ethylene glycol. DHEA was eluted using 15% toluene in iso-octane. After evaporation of the eluate, the residue was reconstituted in assay buffer and aliquots were taken for radioimmunoassay. The radioimmunoassay utilized a highly specific antiserum in conjunction with an iodinated radioligand. Following incubation, antibody-bound and unbound DHEA was separated by incubation with a second antibody. After centrifugation for 15–20 min at 1500 g at room temperature, antibody-bound DHEA was then quantified. The interassay CV was 10.5%. The sensitivity of the assay is reported as 0.2 ng/ml.

Statistical analyses

Preliminary analyses were conducted to examine potential socio-economic and behavioral covariates of baseline neuroendocrine measures using Pearson’s correlation coefficient and one-way analysis of variance. Next, to obtain a summary measure for each neuroendocrine marker, the area under the receiver operating characteristic curve (AUC) was computed using the trapezoid formula. ⁴⁹ Peak reactivity (Δ peak) was computed using the difference score between the maximum level of each neuroendocrine marker and the pre-TSST baseline value. ³¹ Changes in anxiety and mood ratings for each participant were computed by subtracting the pre-TSST rating from the post-TSST rating, with higher change scores reflecting greater increases in anxiety, negative affect and positive affect. Correlational analyses were then performed to examine associations among changes in anxiety and mood.

Because changes in anxiety and negative affect were likely to be highly correlated with each other, a composite variable of psychological reactivity was created by computing z-scores for change in anxiety and negative affect. ⁵⁰ The z-scores were then combined into a composite score, where higher values reflected greater increases in psychological reactivity. This composite measure was examined in relation to neuroendocrine reactivity using multivariable regression models. To control for potential confounding socio-economic or behavioral variables that were observed to be associated (P ≤ 0.10) with neuroendocrine response in preliminary analyses, these variables were entered as covariates in regression models. Post-hoc analyses were also conducted to explore whether individual characteristics, such as smoking status and body mass index (BMI), were associated with baseline levels of cortisol, ACTH or DHEA. For BMI post-hoc analyses, we divided participants using a median split into two groups: (1) those with BMI ≤26 kg/m²; (2) those with BMI >26 kg/m². Multivariable regression models were used to explore whether smoking status and BMI were associated with reactivity. All analyses controlled for baseline levels of the neuroendocrine marker. Analyses were conducted using SPSS® Statistics version 21.0 (SPSS Inc., Chicago, IL, USA).

Results

A total of 40 women agreed to participate in this study. The characteristics of the participants are presented in Table 1.

Table 1.

Characteristics of participants recruited for a study of associations between psychological reactivity and hormonal responses to a standardized laboratory stressor in postmenopausal women.

Characteristic	Study population, n = 40
Age, years	56.3 ± 4.69
Race
Caucasian	50.0
African American	42.5
Asian American/Pacific Islander	5.0
American Indian/Alaska Native	2.5
Body mass index, kg/m2	27.6 ± 5.94
Marital status
Single (never married)	20.0
Married/living as married	42.5
Separated/divorced/widowed	37.5
Education
High school or less	17.5
Some college	55.0
College or postgraduate degree	27.5
Smoking history
Never	47.5
Former	30.0
Current	22.5

Data presented as mean ± SD or %.

In preliminary analyses, BMI was positively correlated with higher baseline levels of ACTH (r = 0.32, P < 0.05). Post-hoc analyses revealed that women with BMI >26 kg/m² had significantly greater ACTH reactivity, as measured by AUC and peak reactivity, compared with women with BMI ≤26 kg/m². There was no association between BMI and cortisol or DHEA reactivity. Also in preliminary analyses, never smokers had higher baseline cortisol levels (mean ± standard deviation, 7.52 ± 3.26 µg/dl) than former smokers (5.83 ± 2.40 µg/dl) and current smokers (4.80 ± 2.76 µg/dl; F(2,39) = 2.97, P = 0.06). Caucasians had higher baseline cortisol levels (7.38 ±  3.06 µg/dl) than non-Caucasians (5.42 ± 2.81 µg/dl; F(1,39) = 4.40, P = 0.05). Age was negatively associated with baseline DHEA levels (r = −0.34, P < 0.05). Therefore, BMI, race, smoking status and age were included as covariates in subsequent analyses.

As expected, anxiety and negative affect significantly increased from pre- to post-TSST, whereas positive affect significantly decreased (see Table 2). However, the extent of change was quite variable across participants. For example, with respect to anxiety, change scores ranged from –7 to 37, indicating that some participants reported a slight decrease in anxiety (as reflected by a negative change score) from pre- to post-TSST.

Table 2.

Overall self-reported anxiety and mood before and after administration of the Trier Social Stress Test in a study of associations between psychological reactivity and hormonal responses to a standardized laboratory stressor in postmenopausal women.

Parameter	Before	After	t-value	Change
Anxiety a	3.33 ± 5.96	11.02 ± 10.21	4.60***	7.68 ± 10.43 (−7 to 37)
Negative affect b	13.04 ± 5.53	16.97 ± 6.65	4.15***	3.93 ± 5.99 (−6 to 23)
Positive affect b	32.53 ± 8.26	28.14 ± 8.21	−4.24***	−4.53 ± 6.67 (−18 to 7)

Data presented as mean ± SD or mean ± SD (range).

***

P < 0.001 (Student’s t-test).

a

BAI, Beck Anxiety Inventory, maximum score = 63.

b

PANAS, Positive and Negative Affect Scale, maximum score = 10.

After controlling for BMI, race, smoking status and age, the correlation analyses indicated that changes in anxiety and negative affect were significantly associated with cortisol, ACTH and DHEA reactivity (Table 3). Specifically, increases in anxiety and negative affect were positively correlated with cortisol AUC and peak reactivity, and ACTH AUC and peak reactivity. Increase in anxiety was positively associated with DHEA AUC and peak reactivity. Negative affect was positively associated with DHEA peak reactivity but not with DHEA AUC (P = 0.056).

Table 3.

Correlations between psychological and hormonal responses in a study of association between psychological reactivity and hormonal responses to a standardized laboratory stressor in postmenopausal women.

Change	Cortisol		ACTH		DHEA
	AUC	Peak reactivity	AUC	Peak reactivity	AUC	Peak reactivity
Anxiety a	0.43**	0.36*	0.50**	0.52**	0.35*	0.37*
Positive affect b	0.22	0.19	0.14	0.14	0.13	0.15
Negative affect b	0.52**	0.55**	0.41*	0.43**	0.32	0.36*

*

P < 0.05; **P < 0.01 (Pearson’s correlation coefficient).

a

Beck Anxiety Inventory.

b

Positive and Negative Affect Scale.

ACTH, adrenocorticotrophic hormone; AUC, area under the curve; DHEA, dehydroepiandrosterone.

Changes in anxiety and negative affect were significantly correlated with each other (r = 0.55, P < 0.001), but neither change in anxiety nor negative affect was associated with change in positive affect (r = −0.06 and −0.10, both P > 0.57). Changes in positive affect were not correlated with neuroendocrine reactivity, therefore positive affect was excluded from subsequent regression analyses.

Using standardized scores, a composite measure of psychological reactivity was computed. Multiple linear regression analyses were then performed to examine associations between psychological reactivity and neuroendocrine response. With respect to cortisol, regression analyses indicated that larger increases in psychological reactivity were significantly associated with greater cortisol reactivity, as measured by AUC (β = 0.55, P = 0.001) and peak reactivity (β = 0.52, P = 0.002), controlling for relevant covariates (i.e. BMI, race, age, smoking status and baseline cortisol levels). Similarly, greater increases in psychological reactivity were significantly associated with greater ACTH reactivity, as measured by AUC (β = 0.50, P = 0.004) and peak reactivity (β = 0.53, P = 0.002), controlling for relevant covariates. Finally, greater increases in psychological reactivity were significantly associated with greater changes in DHEA peak reactivity (β = 0.33, P < 0.04), whereas the association between psychological reactivity and DHEA AUC did not reach statistical significance.

To illustrate such associations, participants were categorized as high or low psychological responders, based on a median split of their composite psychological reactivity change scores. ³⁸ Levels of cortisol, ACTH and DHEA were then plotted according to psychological response status (Figure 1). For high responders, a more pronounced peak was observed for each neuroendocrine marker compared with low responders. Further, among high responders, the peak level of ACTH was reached at ∼1 min post TSST, with levels declining rapidly thereafter. In contrast, peak levels of cortisol and DHEA were highest at 1–15 min post TSST.

Figure 1.

(a) Cortisol, (b) ACTH and (c) DHEA levels in postmenopausal women according to psychological response status (low and high responder according to a median split of composite psychological reactivity change scores; both n = 20) at baseline (−1 min), immediately after the Trier Social Stress Test (TSST; 1 min), and during the recovery period (15–90 min). ACTH was measured only up to 30 min. Diamonds, low responders (score ≤0.48); squares, high responders (score >0.48). ACTH, adrenocorticotrophic hormone; DHEA, dehydroepiandrosterone. The color version of this figure is available at: http://imr.sagepub.com.

Discussion

The present study findings suggest that greater psychological reactivity to a laboratory stress is associated with greater hormonal responses among postmenopausal women. Specifically, women who reported larger increases in anxiety and negative affect following TSST exposure had significantly higher levels of cortisol, ACTH and DHEA. These findings are consistent with prior research that reported that stress-induced increases in negative affect are associated with neuroendocrine reactivity. ³⁶

Change in positive affect, however, is not significantly associated with neuroendocrine reactivity. Others have also noted no association between positive affect and neuroendocrine responses to the TSST,^31,45,46 although one study of adult women reported that greater decreases in positive outlook (operationalized as positive affect and cognitions) after the TSST were associated with greater increases in proinflammatory interleukin-1β reactivity to the TSST. ⁵¹ In observational studies, positive affect is widely reported to have health-protective effects.^52,53 Positive affect is generally associated with lower diurnal cortisol levels^54–57 and a smaller rise in the cortisol awakening response.^58–60 Cortisol is a marker of adrenocortical activity that is associated with psychosocial and physical stress. ⁶¹ It may be that sustained positive affect is needed in order to observe any beneficial and protective effects of positive affect on neuroendocrine reactivity, with the mood-induction paradigms and stress tasks employed in the laboratory being too brief to engender any consistent and reliable association between positive affect and physiological outcomes.

Other characteristics, such as smoking status and BMI, were associated with baseline levels of cortisol and ACTH, respectively. However, post-hoc analyses indicated no differences in cortisol, ACTH and DHEA reactivity between smokers and nonsmokers. The observed relationship between BMI and ACTH is consistent with prior studies demonstrating that obese women show altered responsiveness of the HPA axis. ⁶² Specifically, data indicate that obesity is associated with alterations in secretary patterns of ACTH, but not cortisol, throughout the day. ⁶³ Other studies have also noted that cortisol responses to the TSST do not differ between obese and nonobese women. ⁶⁴ Rather, it has been proposed that abnormalities in HPA axis reactivity are more likely to be observed by specific phenotypes of body fat distribution (i.e. visceral versus peripheral), ⁶³ which may be more informative for characterizing the health effects of obesity than BMI alone.

Further, the results suggest that individual differences in psychological reactivity to the TSST may be an important factor to consider when investigating potential differences in HPA axis response between groups. Indeed, the considerable variability in emotional reactivity was manifested by differential hormonal responses: specifically, greater increases in ACTH, cortisol and DHEA in women who were high responders compared with low responders. Given that women do not respond to the same stressor in identical ways, it is not surprising that studies of stress and breast cancer have yielded such diverse results. For example, in a matched case–control study, higher risk of breast cancer was not associated with a history of childhood adversity (i.e. stress condition) per se, but rather with that, coupled with reports of family support. ⁶⁵ As a result, differences in stress reactivity have implications for how future studies of stress and breast cancer risk are carried out and interpreted.

The present study has several limitations. First, the sample size was relatively small. However, given that the study was focused on postmenopausal women, heterogeneity in biologically relevant factors that may be present in other larger studies (such as gender and menstrual cycle timing) was minimized. Second, although the study sample was diverse with respect to racial background, it was homogeneous for age and menopausal status and, therefore, the findings cannot be generalized to younger women. Third, to reduce participant burden during data collection, psychological stress responses were not assessed during the TSST. Because the psychological stress response is dynamic, synchronous and repeated assessments of psychological responses throughout the stress task may be more informative of participants’ stress perceptions during the task. ³⁵ Despite these limitations, the present findings demonstrate that differences in psychological reactivity to stress should not be overlooked when investigating neuroendocrine function among postmenopausal women.

In conclusion, the present study noted that greater increases in negative affect and anxiety in response to a standardized laboratory stressor are associated with higher levels of cortisol, ACTH and DHEA in postmenopausal women. In light of the considerable data supporting an association between elevated DHEA and increased breast cancer risk, these findings point to one potential biological pathway via which stress exposure may be associated with breast cancer risk. Future studies exploring the role of stress in breast cancer risk should consider how variations in stress response may be associated with differential disease risk.