Pituitary–adrenal reactivity in a child psychiatric population: salivary cortisol response to stressors

European Neuropsychopharmacology 9 (1999) 67–75

Pituitary–adrenal reactivity in a child psychiatric population: salivary cortisol response to stressors Lucres M.C. Jansen*, Christine C. Gispen-de Wied, Maurits A. Jansen, Rutger-Jan van der Gaag, Walter Matthys, Herman van Engeland Rudolf Magnus Institute of Neuroscience, Department of Child and Adolescent Psychiatry, Utrecht University, Academic Hospital, Utrecht, The Netherlands Received 15 April 1997; received in revised form 15 January 1998; accepted 20 January 1998

Abstract The aim of this explorative study was to investigate whether physical and psychological challenges are effective in inducing a cortisol response in psychiatric and control children, and if so whether the cortisol response can discriminate between diagnostic groups and is related to psychiatric symptoms. Fifty-two patients, including children with dysthymia, oppositional defiant disorder / conduct disorder, pervasive developmental disorder, not otherwise specified (PDDNOS) and attention deficit hyperactivity disorder, were compared to 15 healthy control children. Symptomatology was scored using the Child Behaviour Checklist. The response to both psychological and physical challenges was assessed by measuring salivary cortisol and heart rate. Physical challenge, but not psychological challenge, resulted in an overall increase in heart rate and saliva cortisol. Dysthymic and PDDNOS patients showed a diminished cortisol response, in spite of a significant increase in heart rate. These groups scored highest on the symptom factor withdrawal. Withdrawal was negatively correlated with the cortisol response. Thus, physical exercise is effective in inducing a salivary cortisol response in children. Dysthymic and PDDNOS patients have a disturbed pituitary-adrenal function in relation to physical stress, that may be associated with withdrawal.  1999 Elsevier Science B.V. / ECNP. All rights reserved. Keywords: Child psychiatry; HPA; Cortisol; Stress, psychological; Exercise

1. Introduction The endocrine response to stress has been the subject of many studies since it was first described by Selye, 1936. Overall, stressful stimuli are capable of activating the hypothalamic pituitary adrenal (HPA) system, resulting in an increase in glucocorticoid secretion. This glucocorticoid response is the final step in the normal stress response. Autonomic activation, for instance an increase in heart rate, is seen first. The neuroendocrine response, provided that the stressor is strong enough, occurs 15 to 20 min after the onset of the stressor (Scedlowski et al., 1993; Jorgensen et al., 1990). An adequate glucocorticoid stress re-

*Corresponding author. Tel.: 31 30 2508352; fax: 31 30 2505443; e-mail: [email protected]

sponse consists of rapid glucocorticoid secretion, followed by inhibition via feedback systems (De Kloet et al., 1988). Recently, the ability to measure cortisol in saliva has led to renewed interest in the HPA response to various stressors in humans. Saliva sampling has the advantage that it is non-invasive, making multiple sampling easy and stress-free (Kirschbaum and Hellhammer, 1989). Saliva cortisol concentrations are closely correlated to serum free cortisol concentrations (free cortisol being the biologically active fraction). This correlation between serum and saliva cortisol concentrations is also highly significant in children (Woodside et al., 1991). Moreover, saliva cortisol concentrations show a small intra-individual variance (Walker et al., 1984). Therefore, research on HPA reactivity has become more accessible, especially in children. Studies on the cortisol response to stressors have mainly focused on healthy adult populations. Kirschbaum and

0924-977X / 99 / $ – see front matter  1999 Elsevier Science B.V. / ECNP. All rights reserved. PII: S0924-977X( 98 )00003-0

68

L.M.C. Jansen et al. / European Neuropsychopharmacology 9 (1999) 67 – 75

Hellhammer (1994) have reviewed research on the effects of physical and psychological challenges on saliva cortisol concentrations. Most physical challenges, such as swimming, running marathons, or bicycle ergometry, are able to produce significant rises in saliva cortisol concentrations in adults. Studies which used psychological challenges, like film stimuli, public speaking, or mental arithmetics, have shown that the cortisol response is highly dependent on the intensity, controllability, and type of psychological stimulus. Serum ACTH and cortisol concentrations increase significantly in situations with high ego-involvement, low predictability, low controllability, and novelty (Kirschbaum and Hellhammer, 1994, review). However, most studies indicate that there is considerable individual variation in cortisol responsiveness. Overall, sex differences have been found with respect to psychological stress: men tend to have higher cortisol responses than do women (Kirschbaum et al., 1992). Furthermore, anxiety is an important psychological factor that determines the cortisol response: high trait anxious subjects have higher cortisol responses than less anxious subjects (Benjamins et al., 1992). In children, the only study on physical exercise and cortisol revealed an increase in both serum and saliva cortisol concentrations in healthy boys (Del Corral et al., 1994). Studies on psychological stressors in healthy children mainly consist of studies in newborn infants. The HPA system in infants is already responsive to various stressors, such as inoculation and separation (Gunnar, 1992). Behavioural reactions are an important factor in the cortisol response: infants with low behavioural responses tend to have higher cortisol responses (Kagan, 1994). In older children, a similar reaction pattern is found: an inadequate behavioural stress response (inadequate coping) appears to be related to an increased cortisol response (Tennes and Kreye, 1985; Gunnar, 1992). Some studies have been performed on the HPA stress response of child psychiatric patients, using only psychological stressors. In these studies, the cortisol response to stressors appeared to be associated with symptoms like anxiety, aggression, and withdrawal rather than with the diagnosis. Hubert and Jong-Meyer (1992) used unpleasant film fragments as stressful stimuli. They found that high trait anxious children showed a diminished salivary cortisol response as compared to low trait anxious subjects. Kirschbaum et al. (1989) also used film stimuli. However, they showed a significant increase in salivary cortisol in high trait anxious children as compared to low trait anxious children. In fact, high trait anxious children responded to stressful and control (non-stressful) film stimuli, whereas low trait anxious children did not respond to either film stimulus. It seems that it is not the film stimulus itself, but the test setting in general that is challenging for high trait anxious children. Prepubertal boys at risk of substance abuse (i.e., sons of fathers with either substance abuse or antisocial behaviour) show

cortisol hyporesponsiveness in anticipation of a stressful task (Event Related Potential task), the anticipation being the stressful stimulus (Moss et al., 1995). Moss and coworkers found that cortisol hyporesponsiveness was associated with higher scores for aggressive delinquency and impulsiveness. In another study, a more psychosocial stimulus (i.e., a parent-child conflict task) was used in a general population of child psychiatric patients. In this study, social dysfunctions such as withdrawal and social anxiety were associated with high cortisol responses (Granger et al., 1994). Thus, factors that generally influence the HPA stress response in healthy subjects are, apart from the type of stressor, gender, adequacy of behavioural response, and anxiety. Anxiety is also an important factor in the stress response in child psychiatric patients, as are symptom scores on aggression, impulsivity and withdrawal. In this explorative study, all children admitted to the child psychiatric in-patient clinic during the study period took part in both physical and psychological tests to challenge the HPA system. Patients were compared to a group of normal control children. Our aim was to explore whether these tests were able to produce a salivary cortisol response in children and, if so, whether HPA reactivity could distinguish between diagnostic groups and whether it was related to psychiatric symptoms. We hypothesized that highly anxious and withdrawn children (as is seen in dysthymia and pervasive developmental disorder, not otherwise specified) would be highly responsive to the challenges, whereas children with more aggressive and impulsive characteristics (as is seen in oppositional defiant disorder / conduct disorder and attention deficit hyperactivity disorder) would be less responsive than control children.

2. Experimental procedures

2.1. Subjects All children admitted to the child psychiatric clinic between October 1992 and January 1995 were asked to participate in the study. In total, 52 children between 6 and 12 years old (38 boys, 14 girls, mean age 10.061.5 years) took part. Diagnostic evaluation included an extensive psychiatric evaluation consisting of a developmental history, medical examination, semi-structured psychiatric interview and 8 weeks of clinical observation. The Child Behaviour Checklist (CBCL, T.M. Achenbach, Translation in Dutch: F.C. Verhulst, Academic Hospital Rotterdam / Erasmus University Rotterdam) was filled out by the parents of each child. Subjects were classified according to DSM-IV criteria. When consensus was reached between three psychiatrists (RJG, WM, and HE), subjects were assigned to 4 different diagnostic groups on the basis of their primary diagnosis: dysthymia (Dysth) (n58), oppositional

L.M.C. Jansen et al. / European Neuropsychopharmacology 9 (1999) 67 – 75

defiant disorder / conduct disorder (ODD/ CD) (n514), pervasive developmental disorder, not otherwise specified (PDDNOS) (n512), attention deficit hyperactivity disorder (ADHD) (n510) and a rest group (n58). Co-morbidity

69

was taken into account upon decision making (see Table 1). As a control group, 15 children from two regular primary schools were tested (13 boys, 2 girls, mean age

Table 1 DSM-IV Diagnosis and co-morbidity Group

Axis I diagnosis

Dysthymia

Dysthymia, ADHD, enuresis Dysthymia, parent-child problem Dysthymia Dysthymia Dysthymia Dysthymia, ODD Dysthymia, CD, ADHD Dysthymia, CD(NOS), enuresis ODD, encopresis ODD, dev. expr. language disorder ODD ODD, dev. arithm. disorder CD ODD ODD, react. attachm. disorder CD ODD, dev. expr. language disorder ODD, react. attachm. dis. CD ODD CD, react. attachm. dis., enuresis, encopresis PDDNOS, enuresis PDDNOS PDDNOS, ODD, eating disorder (NOS) PDDNOS, enuresis PDDNOS PDDNOS PDDNOS, tic disorder (NOS) PDDNOS PDDNOS PDDNOS PDDNOS PDDNOS ADHD, parent-child problem, dev. articulation disorder ADHD ADHD, induced psychotic disorder ADHD, Gilles de la Tourette ADHD ADHD, ODD, dev. arithm. and expr. writing disorder ADHD, ODD ADHD, CD ADHD, ODD ADHD, tic disorder (NOS) Overanxious disorder, parent-child problem, specific Developmental disorder (NOS) Separation anxiety disorder No diagnosis Autism Encopresis, enuresis Separation anxiety disorder Autism Avoidant disorder

ODD/ CD

PDDNOS

Rest group

ODD/ CD5Oppositional Deviant Disorder / Conduct Disorder. PDDNOS5Pervasive Developmental Disorder, Not Otherwise Specified. ADHD5Attention Deficit Hyperactivity Disorder.

Axis II diagnosis

Axis III diagnosis

Asthmatic bronchitis Myopia

CARA

a-1-antitrypsin-deficiency

Chronic juvenile arthritis Rolandic epilepsy

Mozaic 46 / 47XY Phimosis, allergy Eczema Lactosis hypersensitivity

Eczema EEG abnormalities

Mild mental retardation

Allergic rhinitis

ENT problems Lactosis hypersensitivity

Eczema, scoliosis Chronic constipation Growth retardation

70

L.M.C. Jansen et al. / European Neuropsychopharmacology 9 (1999) 67 – 75

10.062.0 years). Before entering the study, all control subjects were screened for psychiatric problems, by asking their parent to complete the CBCL. All children with a symptom cluster score above the 98th percentile were excluded. Subjects were screened for physical illnesses by filling out a medical checklist. All children with known endocrine, cardiovascular, pulmonary, liver or kidney diseases, or any organic cerebral disorders were excluded from the study. The study was approved by the Ethics Committee of the University Hospital. Informed consent was obtained from parents and children.

2.2. Psychological challenge Two different psychological challenges were used, both focusing on attention and concentration. The first psychological challenge task consisted of a Continuous Performance Task (CPT): a computerized attention task with negative feedback on errors made. Subjects were stimulated to perform as best they could, i.e., to try to make no errors, and to stay highly attentive. This task was performed by 46 subjects (37 patients, 9 controls). Because neither patients nor controls showed a significant increase in salivary cortisol after the test, half-way through the study we extended the challenge by including problem solving and time pressure. The second challenge task consisted of four different tasks from a neuropsychological test battery (D2 test, digit span, Californian Verbal Learning Task, and Tower of London). Subjects had to complete all four tasks within 20 min, which was in general too short. This task was performed by 21 subjects (15 patients, 6 controls). Both tests lasted 20 min. Two baseline saliva samples were taken before the test and four samples were taken after the test, at 20 min intervals (T1 to T6). Heart rate was measured automatically (Omron HEM705), just before and after the psychological challenge tests, as a measure of autonomic arousal.

2.3. Physical challenge The physical challenge consisted of 10 min of exercise on a bicycle ergometer. Subjects were asked to perform with maximal effort. Two baseline saliva samples were taken before the test and four samples were taken after the test, at 20 min intervals (T1 to T6). Heart rate was measured automatically (Omron HEM705), just before and right after the test, as a measure of effort. On the control day no challenge test was performed; only saliva samples were collected at the same times as when they were challenged (T1 to T6). This is referred to as the control test. All tests were carried out in the afternoon, when the HPA-activity is low and stable and therefore more suscep-

tible to stimulation. The challenge tests took place during a two-hour experimental session from 1:00 to 3:00 p.m., on separate days in random order. Each subject performed the three tests in one week. Patients were tested at least 3 weeks after hospital admission to rule out the effects of acute hospitalization stress on the cortisol concentration. Controls were tested at their school.

2.4. Psychometrics The CBCL for ages 4 to 18 years was completed to obtain a total problem score and a symptom profile for each child. The CBCL deals with both internalizing and externalizing problems and consists of nine symptom clusters: withdrawn, somatic complaints, anxious / depressed, social problems, thought problems, attention problems, delinquent behaviour, aggressive behaviour, and sex problems (CBC1 to CBC9). At the beginning of each test session, children used a visual analogue scale (colouring a thermometer with a scale from 0 to 4) to score anxiety, in order to account for anticipation fear for the tests. Just before and immediately after the actual challenge, the Von Zerssen scale (modified for children) for the assessment of fear, anxiety, and aggression was completed by the children (Von Zerssen, 1986). A difference between scores before and after the tests was taken as a measure of mood change.

2.5. Cortisol analyses Saliva samples were collected in plastic vials after saliva production was stimulated with citric acid and were stored at 2208C until analysis. Saliva cortisol concentrations were measured without extraction using an in-house competitive radio immunoassay with a polyclonal anticortisol-antibody (K7348). [1,2- 3 H(N)]-Hydrocortisone (NET185, NEN- DUPONT, Dreiech, Germany) was used as a tracer after chromatographic verification of its purity. The lower limit of detection was 0.5 nmol / l and interassay variation was 11.0; 8.2; and 7.6% at 4.7; 9.7; and 14.0 nmol / l respectively (n520).

2.6. Data reduction and statistical analyses For analysis of the cortisol response to each test, the area under the cortisol response curve for both the psychological challenges and the physical challenge was computed relative to the control test for each subject. Since the cortisol response occurs only 15 to 20 min after the onset of a stressor, the area under the curve (AUC) was computed from T3 to T6 for all tests. Then, the delta area under the curve was computed by subtracting the AUC for the control test from the AUC for the challenge test:

L.M.C. Jansen et al. / European Neuropsychopharmacology 9 (1999) 67 – 75

71

DAUC 5 area under the curve after the challenge test 2 area under the curve after the control test ( AUC 5(T4 1 T5 2 T3 2 T6) test 2 (T4 1 T5 2 T3 2 T6) control ). Overall effects for the challenge tests were computed by one-sample t-tests for the delta AUC and per cent increase in heart rate (% increase). Differences between patients and controls and between sexes were computed using students’ t-tests. Differences in basal cortisol concentrations (cortisol concentration at T2) between the challenge tests and the control test were computed using one-way ANOVA. If any significant differences for sex or age are found, all further analyses will be performed with this factor as a covariate. ANOVAs were used to test for differences in DAUC, basal cortisol concentrations (T2), per cent increase in heart rate, and age between diagnostic groups. Differences in sex ratio between groups were tested by using chisquare analysis. Differences in psychometric (anticipation anxiety and Von Zerssen scores) and symptom cluster scores (CBCL) between the diagnostic groups were tested by using ANOVA’s. Controls were not included in the analyses of symptom scores, since they were selected on the basis of their low CBCL scores. All subjects were divided into responders and nonresponders in terms of the cortisol response to the physical challenge test in order to evaluate factors that may be responsible for response or non-response to the challenge. If a subject’s DAUC was larger than 1 standard deviation of the mean AUC for the control test for the total population, the subject was regarded a responder. The difference in the percentage of responders and non-responders per diagnostic group was tested with chi-square tests. ANOVA’s were used to test for differences in basal cortisol concentrations (T2), per cent increase in heart rate (% increase), age and psychometric scores between responders and non-responders. Finally, partial correlation coefficients were calculated between DAUC and basal cortisol concentration, percentage increase in heart rate, age, psychometric scores, and symptom scores. All tests were two-tailed with an a of 0.05.

3. Results Patient characteristics are given in Table 1.

3.1. Psychological challenge Mean cortisol concentration curves for both the computer and the neuropsychological tasks and the control tests are given in Fig. 1. These are based on data from all subjects.

Fig. 1. Overall cortisol curves for both psychological challenges and control test. CPT: computer task (n545), Psych: neuropsychological task (n520).

Basal cortisol concentrations (T2) were not different between the psychological challenge tests and the control test. Mean DAUC for the computer task was 0.21 (s.d. 2.8) and for the neuropsychological task 0.15 (s.d. 1.5). Neither of these DAUC values was significantly different from zero (resp. t50.51, P50.61 and t50.42, P50.68). No significant increase in heart rate was seen. Heart rates were 84.2 bpm (s.d. 11.2) before the computer task and 84.1 bpm (s.d. 12.2) after the task (paired t-tests: t50.07, P50.95). Heart rate was 83.4 bpm (s.d. 12.1) before and 85.4 bpm (s.d. 11.9) after the neuropsychological task (t5 20.71, P50.49). No significant differences in DAUC between patients and controls were found (t50.21, P50.84 for the computer task and t5 20.81, P50.43 for the neuropsychological task) nor between diagnostic groups for either task (F 50.82, P50.52 for the computer task and F 50.14, P50.93 for the neuropsychological task). As neither of the psychological challenges elicited an endocrine response, these data were not analysed further.

3.2. Physical challenge 3.2.1. Overall effects Mean cortisol concentration curves for the physical challenge and the control test are given in Fig. 2. These are based on data from all subjects. No difference in basal cortisol concentrations (T2) was found between the physical challenge test and the control test. Mean DAUC for the physical exercise test was 2.85 (s.d. 5.6). This was significantly different from zero (t54.20, P,0.001). Heart rate increased significantly from 85.1 bpm (s.d. 13.0) to 141.8 bpm (s.d. 25.9) (Paired t-test: t5 218.38, P,0.001, mean increase was 70%). Overall, boys appeared to have higher DAUC values (mean DAUC values were resp. 0.5164.7 for girls and

72

L.M.C. Jansen et al. / European Neuropsychopharmacology 9 (1999) 67 – 75

Fig. 2. Overall cortisol curves for the physical challenge and the control test (n567). AUC: area under the cortisol response curve.

3.5865.6 for boys, t5 21.97, P50.05) as well as a larger increase in score on the Von Zerssen scale (21.80 (s.d. 4.6) for girls and 1.46 (s.d. 3.5) for boys, U 5183, P,0.01). Therefore, all further analyses for the physical challenge test were performed with sex as a covariate.

3.2.2. Differences between diagnostic groups and normal controls The results for diagnostic groups are given in Table 2. Comparison of the DAUCs showed a significant difference between diagnostic groups (F 52.94, P,0.05). Contrast analyses revealed that this difference was determined by the dysthymic and the PDDNOS groups: both groups showed significantly smaller cortisol responses than the control group (21.14 vs. 4.63 nmol / l, t5 22.22, P,0.05 and 0.36 vs. 4.63 nmol / l, t5 21.98, P50.05).

Basal cortisol concentration (T2) was different between groups (F 52.53, P,0.05), but only the restgroup had a higher basal concentration (t52.38, P,0.05) the other groups were not different. There were no differences in per cent increase in heart rate, age, and psychometric scores (anticipation anxiety and Von Zerssen score) between diagnostic groups. No significant difference in sex ratio between the groups was found (chi-square56.77, P5 0.24). Symptom cluster scores revealed significant differences between patient groups for ‘withdrawal’ (F 53.94, P5 0.01). The dysthymic and PDDNOS patients scored highest on the withdrawal cluster, which consisted of nine items: rather be alone, will not talk, secretive, shy, stares, sulks, underactive, sad, and withdrawn. There were no differences between the number or sort of withdrawal items present in dysthymic and PDDNOS patients. Also, diagnostic groups differed on the CBCL cluster ‘thought problems’ (F 53.25, P,0.05), with the highest score being for the PDDNOS patients.

3.2.3. Responders vs. non-responders The results for responders and non-responders are given in Table 3. When dividing the subjects, independent of diagnosis, into responders and non- responders in terms of the cortisol response to the physical exercise test, a chi-square test showed that the response ratio per diagnostic group was significantly different from the expected values (chisquare511.15, P,0.05). The least responders were found among the dysthymic and PDDNOS patient groups (See Table 2).

Table 2 Differences between diagnostic groups for the physical challenge Measure

Controls (n515)

Dysth. (n58)

ODD/ CD (n514)

PDDNOS (n512)

ADHD (n510)

Boys / girls Age Basal cort (T2) % HR increase DAUC** % responders** Anticipation score Von Zerssen score (difference) Withdrawn** Somatic complaints Anxious / depressed Social problems Thought problems** Attention problems Delinquent behaviour Aggressive behaviour Sex problems

13 / 2 10.0 (2.0) 6.59 (2.5) 66.7 (28.9) 4.63 (4.1) 73.3 1.1 (1.1) 1.5 (2.4)

5/3 9.5 (1.4) 6.29 (2.1) 55.0 (33.6) 21.13 (4.2) 12.5 0.6 (0.8) 20.7 (3.7) 10.7 (2.7) 2.8 (3.3) 14.5 (6.2) 8.2 (3.3) 3.7 (1.8) 11.2 (7.2) 7.5 (7.2) 27.2 (7.4) 1.5 (2.3)

9/5 10.6 (1.1) 5.27 (2.3) 80.3 (27.3) 2.97 (5.5) 50.0 1.0 (1.2) 1.3 (5.9) 4.2 (2.7) 2.5 (3.3) 7.2 (5.6) 7.5 (2.2) 2.4 (1.1) 8.3 (4.1) 8.5 (3.8) 28.5 (6.1) 2.9 (5.0)

9/3 9.3 (1.8) 6.35 (1.8) 63.6 (46.0) 0.36 (1.6) 25.0 0.8 (1.1) 21.4 (4.1) 8.4 (3.3) 1.6 (1.8) 11.1 (6.1) 6.7 (1.1) 6.1 (3.4) 13.7 (5.1) 6.0 (3.1) 22.9 (8.1) 0.4 (0.9)

10 / 0 9.8 (1.5) 6.78 (4.5) 71.0 (38.8) 2.91 (3.5) 50.0 0.8 (1.3) 0.8 (1.9) 3.0 (2.2) 2.8 (3.5) 6.6 (4.5) 6.7 (2.3) 2.4 (1.9) 10.2 (2.9) 5.2 (4.1) 23.2 (9.5) 1.4 (2.9)

Dysth.5Dysthymia; ODD/ CD5Oppositional Deviant Disorder / Conduct Disorder. PDDNOS5Pervasive Developmental Disorder, Not Otherwise Specified. ADHD5Attention Deficit Hyperactivity Disorder. ** P,0.05.

L.M.C. Jansen et al. / European Neuropsychopharmacology 9 (1999) 67 – 75 Table 3 Responders vs. non-responders Measure

Non-responders (n535)

Responders (n535)

Boys / girls Age** Basal cort (T2) % HR increase Anticipation score Von Zerssen score (difference) Withdrawn** Somatic complaints Anxious / depressed Social problems Thought problems Attention problems Delinquent behaviour Aggressive behaviour Sex problems**

24 / 11 9.6 (1.6) 6.4 (2.5) 63.6 (38.1) 0.8 (1.4) 0.3 (3.2) 6.8 (6.1) 3.1 (3.4) 9.0 (7.1) 5.7 (3.1) 3.5 (3.3) 8.5 (5.7) 5.8 (5.0) 22.2 (9.1) 2.6 (4.0)

27 / 5 10.5 (1.6) 6.9 (3.2) 75.5 (32.2) 1.2 (1.1) 1.0 (4.7) 3.2 (2.8) 1.7 (2.1) 5.7 (5.4) 5.0 (3.7) 2.2 (3.6) 8.0 (5.4) 4.2 (4.2) 16.9 (12.6) 0.4 (1.3)

** P,0.05.

Basal cortisol concentrations (T2) and the per cent increase in heart rate were no different between responders and non-responders, nor was the sex ratio (chi-square5 2.29, P50.13). Non-responders were younger than responders (F 56.02, P,0.05). Anticipation anxiety on the day of the physical challenge nor Von Zerssen difference scores differed between responders and non-responders. Non-responders scored significantly higher on the CBCL symptom clusters ‘withdrawn’ (U 5210.0, P,0.05) and ‘sex problems‘ (U 5135.5, P,0.01). (See Table 3).

3.2.4. Correlations with D AUC Basal cortisol concentrations and per cent increase in heart rate were not correlated with DAUC (correlation coefficients were 20.03, P50.81 and 0.19, P50.12 respectively). However, age was significantly correlated with DAUC (correlation coefficient 0.31, P50.01), but was not a significant covariate in the DAUC analysis. Age can therefore not be responsible for the differences in DAUC between groups. Neither anticipation anxiety for the physical exercise test nor the increase in Von Zerssen score was correlated with DAUC (correlation coefficients were 0.03, P50.83 and 0.17, P50.19 respectively). CBCL clusters ‘withdrawn’ and ’sex problems’ were negatively correlated with DAUC (correlation coefficients were 20.32, P,0.05 and 20.29, P50.05 respectively).

4. Discussion In this study, neither of the psychological challenges was effective in inducing an endocrine response. The physical challenge, however, resulted in a significant increase in salivary cortisol. Boys showed higher endocrine responses to exercise than did girls. Comparison of DAUC values and the number of responders revealed that both the dysthymic and the PDDNOS groups were cortisol

73

hyporesponsive to physical challenge, compared to normal controls. On a symptomatological level, withdrawal appeared to be an important factor in this diminished cortisol response. Overall, saliva cortisol concentrations appeared stable across test days, with basal cortisol concentrations not differing between test days. Also, the basal cortisol concentrations of patients and controls were similar, even though all the patients, but none of the controls were hospitalized. Thus, the stress of hospitalization did not increase basal cortisol concentration 3 weeks after admission. The psychological challenges used in this study did not cause a substantial cortisol response or autonomic arousal induced, as measured by an increase in heart rate. The computer task focused on attention and seemed low in novelty and highly controllable. Obviously, mental exercise is not sufficient to induce a significant stress response. Therefore, a second psychological challenge was introduced, which focused on problem solving to increase ego-involvement. Furthermore, time pressure was introduced as an extra stress factor. In this way, the test was less controllable. However, even so, the task did not affect arousal or evoke an endocrine response. In future studies, an even higher ego-involvement may be achieved by introducing a more social context, for instance, a competition between subjects. The physical challenge test had a significant effect on the saliva cortisol concentration in children. This finding is consistent with the study by Del Corral et al. (1994), who showed an increase in salivary cortisol of 81% above resting level after 30 min of bicycle ergometry, whereas we found an increase of 38% above resting level after 10 min of exercise. The sex difference found is consistent with the findings of Kirschbaum (1992): males appear to be more responsive to challenge than females. This may be associated with increased fear, anxiety, and aggression, since the difference scores on the Von Zerssen were also larger in males. It is therefore important to take the sex composition of groups into account when performing challenge tests. Unexpectedly, we found that the HPA system of patients in two different diagnostic groups (dysthymia and autisticlike disorders), was hyporesponsive to physical challenge. In both groups this hyporesponsiveness was associated with a high score on the psychiatric symptom withdrawal. Partly on the basis of the significant increase in heart rate in all groups, we would have expected to find an altered response instead of hyporesponse, particularly in the patients with dysthymia. If, according to the literature on depression in adults (Croes et al., 1993), one suggests that the HPA system in less inhibited in these patients, a normal or increased cortisol response after challenge would be expected. Disturbances of the HPA axis have been described in autistic patients, but results are rather contradictory and

74

L.M.C. Jansen et al. / European Neuropsychopharmacology 9 (1999) 67 – 75

patient groups are not well defined. Moreover, most studies deal with basal HPA function. Only one study reported an absence of a cortisol response after naltrexone in autistic children (Chamberlain and Herman, 1990). Therefore, it is difficult to interpret our data in the light of findings in the literature. However, from a symptomatological point of view, it is our impression that patients with dysthymia and PDDNOS share a similar psychopathology. Both groups of patients are more internalizing according to CBCL scores, whereas patients with ODD and ADHD are more externalizing. HPA unresponsiveness appears to be related to more internalizing characteristics, especially withdrawal. This is in contrast with the results reported by Moss et al. (1995), who found that cortisol hyporesponsiveness was related to externalizing symptoms such as aggression and impulsiveness. However, our findings are consistent with the literature in so far as withdrawal is often mentioned as a factor in the cortisol response to psychological challenge in children (Granger et al., 1994; Gunnar, 1992). Most studies have found a larger cortisol response to psychological challenge in subjects scoring high on withdrawal, whereas our results suggest a decreased cortisol response to physical challenge in withdrawn subjects. Another internalizing factor that is of influence in the HPA response is anxiety. Although the CBCL score depression / anxiety was not significantly different between groups, the anxiety score, as assessed with the visual analogue scale on the day of the physical challenge, was higher in responders. This supports the findings of Benjamins et al. (1992) in adults and Kirschbaum et al. (1989) in children, suggesting that the HPA system of high trait anxious subjects is more susceptible to psychological stimulation. Our results indicate that this also holds for the physical challenge used in our study. However, a decreased HPA response to stress has also been found in high trait anxious patients (Hubert and Jong-Meyer, 1992). Thus, internalizing factors may influence responsiveness of the HPA system, although the exact impact and the direction of the disturbances in the HPA response are not clear. One must therefore be cautious when interpreting the associations between HPA dysfunctions and psychiatric syndromes. In conclusion, the physical challenge test, but neither one of the psychological challenge tests, was effective in inducing a cortisol response in children between 6 and 12 years of age. The use of challenge tests appeared to reveal HPA dysfunctions in some groups of children. Although the basal cortisol concentration was not different from controls, the HPA system of both PDDNOS and dysthymic patients showed a diminished response to challenge. The hyporesponsiveness of the HPA system in these groups was related to the symptom score for withdrawal. However, one should be careful when interpreting these differences in HPA function at a symptomatological level.

Acknowledgements We would like to thank Dr. Herman Wynne for giving us valuable statistical advice. We are also greatly indebted to Mrs.Achterberg and Mr.van de Giessen for their help in recruiting the control children and for making it possible to test these control children at their school.

References Benjamins, C., Asscheman, H., Schuurs, A.H.B., 1992. Increased salivary cortisol in severe dental anxiety. Psychophysiology 29, 302–305. Chamberlain, R.S., Herman, B.H., 1990. A novel biochemical model linking disfunctions in brain melatonin, propriomelanocortin peptides, and serotonin in autism. Biol. Psychiatry 28, 773–793. Croes, S., Merz, P., Netter, P., 1993. Cortisol reaction in success and failure condition in endogenous depressed patients and controls. Psychoneuroendocrinology 18 (1), 23–35. De Kloet, E.R., Rosenfeld, P., Van Eekelen, A.M., et al., 1988. Stress, glucocorticoids and development. Prog. Brain Res. 73, 101–120. Del Corral, P., Mahon, A.D., Duncan, G.E., et al., 1994. The effect of exercise on serum and salivary cortisol in male children. Med. Sci. Sports Exercise 26, 1297–1301. Granger, D.A., Weisz, J.R., Kauneckis, D., 1994. Neuroendocrine reactivity, internalizing behaviour problems, and control-related cognitions in clinic-referred children and adolescents. J. Abnorm. Psychol. 103, 267–276. Gunnar, M.R., 1992. Reactivity of the hypothalamic pituitary adrenocortical system to stressors in normal infants and children. Pediatrics 90 (3), 491–497. Hubert, W., Jong-Meyer, R.D., 1992. Saliva cortisol responses to unpleasant film stimuli differ between high and low trait anxious subjects. Neuropsychobiology 25, 115–120. Jorgensen, L.S., Christiansen, P., Raundahl, U., et al., 1990. Autonomic response to an experimental psychological stressor in healthy subjects: measurement of sympathetic, parasympathetic, and pituitary-adrenal parameters: test-retest reliability. Scand. J. Clin. Lab. Invest. 50, 823–829. Kagan, J., 1994. On the nature of emotion. Monogr. Soc. Res. Child Dev. 59, 7–24. Kirschbaum, C., Hellhammer, D.H., 1989. Salivary cortisol in psychobiological research: an overview. Neuropsychobiology 22, 150–169. Kirschbaum, C., Hellhammer, D.H., 1994. Salivary cortisol in psychoneuroendocrine research: recent developments and applications. Psychoneuroendocrinology 19, 313–333. Kirschbaum, C., Hellhammer, D.H., Strasburger, C.J. et al., 1989. Relationships between salivary cortisol, electodermal activity and anxiety under mild experimental stress in children. In: Weiner, H., Florin, I., Mursion, R. (Eds.), Frontiers of Stress Research. Huber, Toronto, pp. 383–387. ¨ S., Hellhammer, D., 1992. Consistent sex differKirschbaum, C., Wust, ences in cortisol responses to psychological stress. Psychosom. Med. 54, 648–657. Moss, H.B., Vankuyov, M.M., Martin, C.S., 1995. Salivary cortisol responses and the risk for substance abuse in prepubertal boys. Biol. Psychiatry 38, 547–555. Scedlowski, M., Jacobs, R., Alker, J., et al., 1993. Psychophysiological, neuroendocrine and cellular immune reactions under psychological stress. Neuropsychobiology 28, 87–90. Selye, H., 1936. A syndrome produced by diverse nocuous agents. Nature 38, 32. Tennes, K., Kreye, M., 1985. Children’s adrenocortical responses to

L.M.C. Jansen et al. / European Neuropsychopharmacology 9 (1999) 67 – 75 classroom activities and tests in elementary school. Psychosom. Med. 47, 451–460. Von Zerssen, D., 1986. Clinical self-ratings scales (CSRS) of the Munich Psychiatric Information System. In: Sartorius, N., Ban, B.A. (Eds.), Assessment of Depression. Springer Verlag, Berlin, pp. 270–303. Walker, R.F., Joyce, B.G., Dyas, J., Riad-Fahmy, D., 1984. Salivary

75

cortisol: I. monitoring changes in normal adrenal activity. In: Read, G.F., Riad-fahmy, D., Walker, R.F., et al (Eds.), Immunoassays of Steroids in Saliva. Alpha Omega, Cardiff, pp. 308–316. Woodside, D.B., Winter, K., Fisman, S., 1991. Salivary cortisol in children: correlations with serum values and effect of psychotropic drug administration. Can. J. Psychiatry 36, 746–748.