Submitted in: Journal of Psychophysiology
Authors: Katja Weber1, Harald Engler2, Barbara Awiszus1, Carsten Riether3, Manfred Schedlowski2, and Brigitte Rockstroh1*
1Department of Psychology, University of Konstanz, Germany
2Division of Medical Psychology and Behavioral Immunobiology, Medical Faculty, University of Duisburg-Essen, Germany
3Institute for Behavioral Sciences, Switzerland Swiss Federal Institute of Technology (ETH), Zürich, Switzerland
Abstract
Animal and human studies indicate lasting effects of early life stress on the hypothalamic-pituitary-adrenal axis and on psychopathology. The present study examined these effects in 95 psychiatric inpatients with diagnoses of major depressive disorder, schizophrenia, drug addiction, or borderline personality disorders, and in 30 healthy subjects. Diurnal and reactive salivary cortisol measures were related to stress history (assessed by interviews, Early Trauma Inventory and the Posttraumatic Stress Diagnostic Scale). The diurnal cortisol profile was determined from three salivary samples after awakening and one sample in the evening; the cortisol response was determined from one sample before and two after an affective picture-monitoring task. In patients compared to healthy subjects, high stress load before the age of 6 years was related to higher diurnal salivary cortisol and to larger and longer cortisol response to affective picture viewing. Effects were independent of diagnosis or gender. Both endocrine measures varied with Posttraumatic Stress Disorder (PTSD) symptom severity in patients with comorbid PTSD (n=27). Results support the impact of early life stress on the neuroendocrine
Early life stress, Diurnal and Reactive Salivary Cortisol in Psychiatric Patients system in individuals suffering from a psychiatric disorder and suggest that adverse effects on the developing brain, including endocrine systems, may interact with other factors in their impact on the progress of a psychiatric disorder.
Introduction
Many studies have addressed the impact of stress on neuro-endocrine responses and mental health with various perspectives, concepts and methods. An impact of early life stress (ELS) on adult psychopathology has been specified for different psychiatric disorders, most frequent for major depressive disorders (MDD, e.g., Heim, Owens, Plotsky, & Nemeroff, 1997; Heim, Plotsky, & Nemeroff, 2004) and posttraumatic stress disorder (PTSD, e.g., van der Kolk, Roth, Pelcovitz, Sunday, &
Spinazzola, 2005; Scheller-Gilkey, Moynes, Cooper, Kant, & Miller, 2004), but also for schizophrenia (e.g., Thompson, Pogue-Geile, & Grace, 2004; Corcoran, et al., 2003), and borderline personality disorder (BPD, e.g., Goodman, New, & Siever, 2004). This impact of early life stress has been linked to the particular sensitivity of the developing brain including the neuroendocrine system in childhood (Charmandari, Kino, Souvatzoglou, & Chrousos, 2003). Animal studies demonstrate more severe brain alterations when experiences of pre-weaning maternal separation were later followed by exposure to an open elevated platform (Stewart, Petrie, Balfour, Matthews, & Reid, 2004). In animal and human studies, ELS has been found to exert lasting effects on the dysregulation of corticotropin-releasing hormone (CRH) and the hypothalamus-pituitary-adrenal (HPA)-axis with consequences for the development of anxiety and depression (Matsuzaki, Takamatsu, & Moroji, 1989;
Makino, Gold, & Schulkin, 1994 a,b; Shepard, Barron, & Myers, 2000; Strome, et al., 2002; Caldji, et al., 1998; Meaney, et al., 1996; Teicher, Andersen, Polcari, Anderson, & Navalta, 2002; Nemeroff, 1996). Such evidence gave rise to the hypothesis that adverse early experiences might also exert some influence on the evolution of psychopathology in humans.
Independent of ELS, dysregulation of the HPA-axis has been described for different psychiatric disorders: Elevated basal levels of plasma cortisol, increased CRH in the cerebrospinal fluid, an altered circadian cortisol rhythm, or increased responses to endocrine challenge tests (e.g. after dexamethasone administration, DEX) have been reported for MDD (Heim, Owens, Plotsky, & Nemeroff, 1997; Heim, Plotsky, & Nemeroff, 2004; Nemeroff, 1996, 2002; Bremner, et al., 1997; Baker, et
Early life stress, Diurnal and Reactive Salivary Cortisol in Psychiatric Patients al., 1999; Belmaker & Agam, 2008). In individuals suffering from PTSD, reduced plasma and urinary cortisol, or blunted HPA responses to DEX/CRF challenge were found (Heim, Mletzko, Purselle, Musselman, & Nemeroff, 2008; Bremner, et al., 2003; Yehuda, et al., 1990, 1996; overview Wessa & Rohleder, 2007). Moreover, PTSD patients, but also trauma-exposed individuals without PTSD diagnosis (Wessa, Rohleder, Kirschbaum, & Flor, 2006), frequently failed to show the normal (salivary) cortisol response after awakening, and presented lower daytime cortisol or a flattened diurnal profile (Rohleder, Joksimovic, Wolf, & Kirschbaum, 2004; Yehuda, 2005). Abnormal responses to DEX/CRH challenge tests have also been reported in patients with BPD (Carrasco, et al., 2007; Rinne, et al., 2002; Lieb, et al., 2004;
Wingenfeld, Hill, Adam, & Driessen, 2007). Only few studies directly addressed the relationship between HPA-axis dysregulation and early (traumatic) stress: Heim, Newport, Bonsall, Miller, & Nemeroff (2001) described larger ACTH responses to endocrinological challenge and lower baseline and stimulated plasma cortisol in abused women without MDD in contrast to blunted responses in abused women with MDD or MDD patients without stress load. For BPD, Simeon, Knutelska, Smith, Baker, & Hollander (2007) report an inverse relationship between basal urinary cortisol and childhood trauma, while Rinne, et al. (2002) found hyperresponsiveness to DEX/CRH tests in female subjects with a history of childhood abuse.
The cortisol response to emotional or mental challenge serves as additional measure in the study of stress effects on HPA-axis. Activation of the HPA-axis by acute experimental psychological stressors (e.g., the ‘Trier Social Stress Test’) has been documented in healthy subjects (Kirschbaum, Pirke, & Hellhammer, 1993).
Subjects with a history of childhood abuse and with or without current PTSD diagnosis, showed elevated (salivary) cortisol levels before and during psychological stress tests, and delayed and reduced recovery after exposure to traumatic reminders (Elzinga, Schmahl, Vermetten, van Dyck, & Bremner, 2003; Bremner, et al., 2003; Santa Ana, et al., 2006).
As part of a project investigating the impact of stress on the course of illness in adult psychiatric patients (see Weber, et al., 2008), the present study explored the relationship between stressful experiences early in life and measures of HPA-axis activity. Salivary diurnal cortisol was assessed as an indicator of basal activity-level of the HPA-axis (Kirschbaum & Hellhammer, 1989, 1994). Only the free fraction of released cortisol (2%-15%, Kirschbaum & Hellhammer, 2000) is biologically active
Early life stress, Diurnal and Reactive Salivary Cortisol in Psychiatric Patients and able to diffuse passively into the saliva. In addition, salivary cortisol response was assessed in a magnetoencephalographic (MEG) study involving affective picture viewing6 with the assumption that the processing of high-arousing pleasant and unpleasant pictures would be experienced as a mild stressor.
We hypothesized a relationship between ELS and salivary cortisol measures within the entire sample indicating that stress load in sensitive developmental periods affects HPA-axis regulation. Moreover, we hypothesized a prominent impact of ELS relative to stress load in other periods of life on neuroendocrine functions, which should become evident in relationships between ELS, but not between acute stress load in adulthood, and salivary cortisol measures. Finally, if ELS interacts with vulnerability factors for a psychiatric disorder in their impact on neuroendocrine functions, cortisol profiles should differ between patients with different diagnoses and relationships with adult stress load were to be expected.
Methods
Participants: Diurnal cortisol was obtained from 95 inpatients of a local Center for Psychiatry (Reichenau) and 30 individuals without psychiatric diagnoses (see Table 11a for demographic and clinical information). Cortisol response evoked by the affective picture series was obtained from a subset of this sample, including 73 patients and 27 healthy comparison subjects (Table 11b). In both samples, patients and comparison subjects were similar in gender balance and age (see Table 11a, b), whereas comparison subjects had more years of education than patients and consumed less cigarettes/day. Five female patients and seven comparison subjects used oral contraceptives. Within the patient sample, diagnostic subgroups differed in gender distribution and age, but not in educational level or smoking habit (see Table 11a, b).7
6 The picture-viewing task, while the magnetoencephalogram was recorded, was part of the project and aimed at analyzing cortical affective processing in relationship to early life stress and psychiatric diagnoses; for details see Weber et al. (2009).
7 Smoking habit, the use of hormones, as well as demographic data such as age were assessed, as influences of smoking (Kirschbaum et al., 1994, 1992; Rohleder and Kirschbaum, 2006), age (e.g., Kudielka et al., 2004), gender, and particularly the use of oral contraceptives in women (e.g., Kirschbaum et al., 1999), on HPA-axis activity have been emphasized.
Early life stress, Diurnal and Reactive Salivary Cortisol in Psychiatric Patients
Early life stress, Diurnal and Reactive Salivary Cortisol in Psychiatric Patients
Early life stress, Diurnal and Reactive Salivary Cortisol in Psychiatric Patients Patients were recruited from wards specialized for MDD, schizophrenia, DA, and BPD. The respective psychiatrists identified patients meeting the diagnostic criteria and verified diagnoses from the categories of MDD (F31-33), schizophrenia spectrum (F20, F25), DA (F19, F10), and BPD (F60.31) using ICD-10 criteria (International Classification of Diseases, WHO, 1992). Except for DA patients, most patients were on medication, the majority receiving combinations either of antidepressants and neuroleptics, typical and atypical neuroleptics, or tricyclic and SSRI antidepressants Table 11a, b).
Comparison subjects were recruited from hospital staff and community centers by advertisement to be comparable to the patient sample with respect to age, gender, and education. Comparison subjects were screened with the MINI (Mini-International Neuropsychiatric Interview, Sheehan, Lecrubier, Sheehan, Amorim, Janavs, Weiller, et al., 1998) for lifetime diagnoses. Only subjects without any history of psychiatric diagnosis and free of psychoactive medication were included.
Design and Materials: The study protocol was approved by the ethics committee of the University of Konstanz. Participants were informed about the goal of the study and procedures. All subjects gave written informed consent.
Patients’ psychopathology was evaluated with the Brief Psychiatric Rating Scale (BPRS, Lukoff, Liberman, & Nuechterlein, 1986), Beck Depression Inventory (BDI, German version Hautzinger, Bailer, Worall, & Keller, 1995), and the Positive and Negative Symptom Scale (PANAS, Watson, Clark, & Tellegen, 1988).
In all participants, stress history was assessed with the German version (Heim, 2000) of the Early Trauma Inventory (ETI, Bremner, Vermetten, & Mazure, 2000). The interview screens adverse experiences for each year of life in the domains physical punishment, emotional neglect, sexual abuse, and general trauma.
Stress load was defined as the number of reported experiences within the first three domains summed up for the time period before the age of six (labeled Early Life Stress, ELS), for the time period between the age of 6 and the individual onset of puberty (labeled Pre-Pubertal Stress, PPS), and for the time between puberty and the current age (labeled Adulthood Stress, AS). Within the same session, subjects were screened with the Posttraumatic Stress Diagnostic Scale (PDS, Foa, 1995) for traumatic events, PTSD diagnosis, and severity of PTSD symptoms (intrusions, avoidance, and hyperarousal).
Early life stress, Diurnal and Reactive Salivary Cortisol in Psychiatric Patients Diurnal salivary cortisol sampling considered the individuals’ circadian rhythms in a natural setting on a weekday. Three salivary samples were obtained in the morning, the first one immediately after individual awakening (which varied between 5:40 a.m. and 7:30 a.m.), the second one 20 min later, and the third 45 min after the first one. Between awakening and the third sample, subjects stayed awake but were not allowed to drink (except for water), eat, smoke, or brush their teeth. The fourth salivary measurement (which varied between 9 p.m. and 11:30 p.m.) was obtained in the evening 20 min before going to bed (see Figure 8).
Figure 8. Schematic representation of diurnal sampling protocol.
Again, subjects refrained from eating, drinking, smoking, or teeth brushing one hour before the evening sample. Moreover, participants refrained from drinking alcohol the day before and on day of measurement as well as from any heavy-duty physical exercise.8 The staff on the respective ward was trained to guarantee that inpatients followed the sampling protocol and study team members controlled compliance with the rules. Healthy comparison subjects were instructed how to collect the cortisol samples at home and team members controlled for compliance by phone calls. Sleep-wake habits were assessed before including a subject into the study; shift workers were excluded.
For cortisol response, salivary samples were obtained before and after a MEG protocol, in which subjects monitored altogether 600 affective colored photographs from the International Affective Picture System (IAPS, Center for the Study of Emotion and Attention, 2004). Photographs varied in their arousal and hedonic valence according to normative valence and arousal ratings (Lang, Bradley, &
Cuthbert, 1999). No active task was related to picture viewing (for details of this
8 Confinement to only one day of measurement was a compromise between methodological requirements and compliance of psychiatric inpatients. We preferred a stable getting-up and going-to-bed schedule within the individual over a strict time-locked sampling protocol.
+ 20 min + 45 min - 20 min t
before bedtime immediately
after awakening
Early life stress, Diurnal and Reactive Salivary Cortisol in Psychiatric Patients study, see Weber, Miller, Schupp, Borgelt, Awiszus, Popov, Elbert, & Rockstroh, 2009).
Figure 9. Schematic representation of response sampling protocol.
Cortisol assessment was standardized starting for each subject at 1 p.m. with 250 ml of grape juice (to adjust the level of blood sugar; see Figure 9). The first salivary sample was obtained 45 min later, after preparation for the MEG recording and right before the onset of the picture viewing period. After MEG-measurement, subjects rated the valence and arousal of a subset of pictures in order to validate stimulus selection. The second cortisol sample was taken immediately after this rating (42 min after the first sample); the third sample was obtained after a subsequent 20 min resting period.
Data analyses: Commercial collection devices (Salivette, Sarstedt, Numbrecht, Germany) were used for saliva sampling. Samples were centrifuged immediately after collection and stored at -20°C until analyses. Saliva cortisol levels were measured using a competitive bead-based assay. Undiluted saliva or cortisol standard dilutions were incubated overnight in 96-well round-bottom plates with appropriate amounts of cortisol-BSA-conjugated polystyrene beads and fluorescein isothiocyanate (FITC)-conjugated rabbit anti-cortisol antibody (HTB192, Chromaprobe, Maryland Heights, MO, USA). After incubation, beads were washed and resuspended in phosphate-buffered saline, and analyzed on a flow cytometer (LSR II, BD Immunocytometry Systems, San Jose, CA, USA). The median fluorescence intensity is inversely proportional to the amount of cortisol in the sample. Intra- and interassay variance were 5.4% and 10.7%, respectively. Antibody
1:45 pm 2:40 pm 3 pm t
1 pm
MEG scan
Resting period Resting period
250 ml grape juice
Early life stress, Diurnal and Reactive Salivary Cortisol in Psychiatric Patients cross-reactivity with other relevant steroids was 4.0% (testosterone) and 0.9%
(progesterone), respectively.
Following Pruessner, Kirschbaum, Meinlschmid, & Hellhammer (2003), the area under the curve with reference to ground (AUCg) resulting from four salivary samples across the day served as measure of diurnal cortisol. The cortisol response was determined as the area under the curve with reference to increment (AUCi) resulting from three salivary samples before and after the experimental period. AUCg
has been related to the individual hyper- or hypocortisolemic trait (Pruessner, et al., 2003), AUCi has proven sensitive for momentary states of the HPA-axis (Hellhammer, et al., 2007). Group differences in diurnal cortisol (patients versus comparison subjects and between diagnostic subgroups) were statistically verified by nonparametric analyses (Mann-Whitney-U-test and Kruskal-Wallis-χ²-ANOVA).
Significant main effects or interactions were gradually decomposed with Bonferroni-corrected pairwise comparisons. The impact of stress experiences on cortisol measures was evaluated for the developmental periods by correlation (Spearman rho, rs) and linear regression analyses. ELS effect, in particular, was further scrutinized by comparing subgroups of individuals with high- and low-ELS: Patients with ETI-scores exceeding 3 standard deviations of the mean of the comparison group were assigned to a ‚high-ELS’ group and patients with ETI-scores below the mean of the comparison group to a ‚low-ELS’ group. Statistical significance for all tests was evaluated at the .05 level.
Early life stress, Diurnal and Reactive Salivary Cortisol in Psychiatric Patients
Results
Cortisol and Stress Load:
All measures of stress load were higher in patients than in healthy comparison subjects (Table 12). A comorbid diagnosis of PTSD was found in 28% of patients.
None of the healthy subjects met the diagnostic criteria and no subject was excluded from the comparison sample because of a PTSD diagnosis (Table 11a, b).
Diurnal Salivary Cortisol: AUCg, which varied between 712 and 20780.5 nmol/l (M±SEM: 6576.8±331.04) did not differ between patients and comparison subjects (patients: 6535.3±397.3, healthy subjects: 6708.1±575.8; U(1,124)= 1296, p> .1) or between diagnostic subgroups (χ2(3,92)= 2.68, p> .1). Neither age (rs= .02, p> .1), nor oral contraceptives (rs= .01, p> .1), or smoking habits (rs= .004, p> .1) were related to diurnal salivary cortisol, and measures did not differ between gender (F(1,124)= 1.55, p> .1) or medication type (F(8,90)= .80, p> .1).
Higher ELS varied with elevated cortisol across the day (AUCg) in patients (rs= .37, p< .01) but not in comparison subjects (rs= .10, p> .1; total sample: rs= .26, p< .05). AUCg was neither related to PPS or AS, nor to traumatic events (PDS) or PTSD-diagnosis.
Early life stress, Diurnal and Reactive Salivary Cortisol in Psychiatric Patients
Early life stress, Diurnal and Reactive Salivary Cortisol in Psychiatric Patients Linear regression models determined ELS as best predictor of diurnal cortisol (AUCg, total sample: R²= .05; patient group: R²= .07; Table 13a). Analyses of the stress domains disclosed relationships between AUCg and emotional neglect (total sample: rs= .28, ß= .19, p< .05, R²= .029; patient group: rs= .36, ß= .22, p< .05, R²= .039). No additional variance was explained by physical abuse (total sample:
rs= .21, p< .05; patient group: rs= .30, p< .05), and sexual abuse (total sample: rs= .16, p= .07; patient group: rs= .24, p< .05).9
Table 13a. Relationship between Early Life Stress & Diurnal Salivary Cortisol (AUCg).
Total sample (N= 124) Patients (N=93) Predictor Zero-order correlation β Zero-order correlation β ELS (physical,
emotional, and sexual abuse)
.19* .24** . 24* .29**
Gender .10 .18 (p= .055) .10 .21 (p= .055) Note. Zero-order correlation is represented by Spearman’s rho; β coefficients are standardized and result from a linear regression model of AUCg; full models adjusted R²= .05 for the whole sample and R²= .07 for the patient group. Neither traumatic events and PTSD, nor any other age interval of the Early Trauma Inventory predicted AUCg. None of the variables age, diagnosis, medication, smoking habit, or any of the psychopathology measure did correlate with salivary cortisol. *p< .05, **p< .01
Comparing patient subgroups with high-ELS (n=23), low-ELS (n=26) and comparison subjects (n=30), higher cortisol levels in high-ELS compared to low-ELS patients (U(1,48)= 202, p< .05) were found, whereas healthy subjects did not differ from the low- or the high-ELS patient group (χ2(2,77)= 3.71, p= .15; Figure 10).
9 Sexual abuse was rare in this period of life: 6 patients and 1 comparison subject reported sexual abuse.
Early life stress, Diurnal and Reactive Salivary Cortisol in Psychiatric Patients
Figure 10. (A) Salivary cortisol profile. Abscissa: Three morning saliva samples (right after awakening, +25 min, + 45 min) and one sample before bedtime (-20 min) in patients with high early-life stress (high-ELS, n=23, filled squares) and low early-life stress (low-ELS, n=26, open circles), and in comparison subjects (n=30, open diamonds). Ordinate: Mean ± SEM of cortisol in nmol/l. (B) Area under the curve with reference to ground (AUCg) for diurnal salivary profile in stress groups. Abscissa:
Low-ELS patients: grey bar; high ELS patients: black bar; comparison group: white bar. Ordinate: Mean ± SEM of cortisol in nmol/l.
Salivary Cortisol Response: AUCi varied between -485 and 389 nmol/l (M±SEM: -14.3±13.75), with no difference between the groups (patients: -13.9±16.5, comparison subjects: -15.3±25.3; U(1,99)= 925, p> .1), nor between diagnostic subgroups (χ2(3,70)= 2.28, p> .1). Again, age (rs= -.01, p> .1), oral contraceptives (rs= -.04, p> .1), smoking habits (rs= .19, p> .1), gender (F(1,98)= .11, p> .1), and medication type (F(7,68)= .74, p> .1) were unrelated to the cortisol response.
Patients with higher ELS displayed elevated AUCi in response to the picture viewing and rating task (rs= .25, p< .05), whereas relationships with PPS (rs= .20, p=
.09) and AS (rs= .21, p= .08) only approached significance. In comparison subjects and within the total sample, correlations between AUCi and stress experiences did not reach significance (ELS: rs= .13, p> .1; total sample: rs= .18, p= .07). The experience of the first ELS at an earlier age (age of first remembered stress experience according to ETI) was related to more cortisol release in response to the task (total sample: rs= -.23, p< .05; patient sample: rs= -.24, p< .05; comparison
Early life stress, Diurnal and Reactive Salivary Cortisol in Psychiatric Patients sample: rs= -.25, p> .1). Again, diagnostic groups did not differ and no relationships were found between AUCi and traumatic events (PDS) or current PTSD diagnosis.
Linear regression models determined ELS as best predictor of cortisol response (AUCi, total sample: R²= .07, patient group: R²= .12; Table 13b). Analyses of stress domains disclosed a relationship between AUCi and physical abuse (total sample: rs= .14, ß= .22, p< .05, R²= .04; patient group: rs= .25, ß= .30, p< .05, R²=
.09), while emotional neglect (total sample: rs= .17, p= .09; patient group: rs= .20, p=
.09), and sexual abuse (p> .1) did not explain additional variance.10
Table 13b. Relationship between Life Stress and Reactive Salivary Cortisol (AUCi).
Total sample (N= 100) Patients (N=73) Predictor Zero-order correlation β Zero-order correlation Β
Total sample (N= 100) Patients (N=73) Predictor Zero-order correlation β Zero-order correlation Β