Psychiatry Research, 37:23?-244
237
Elsevier
Cortisol Hypersecretion in Depressed School-Aged Children and Adolescents Ian Goodyer. Received April24,
Joe Herbert,
Stephanie
Moor, and Patricia Altham
1990; revised version received November 7, 1990; accepted January 4, 1991.
Abstract. A prospective longitudinal study has been carried out to determine the secretory pattern of cortisol in children (n = 10) with major depressive disorder. Salivary cortisol samples were collected at 4-hourly intervals over 24 hours when the subjects were depressed and again when they were recovered. Group comparison indicated that significant increases in mean cortisol output occurred during illness as compared with recovery. This difference occurred only at three points (midnight, 4 a.m., 8 a.m.) of six measured. Not all cases were showed hypersecretion, but when hypersection was present, it occurred in cases with more severe symptoms. In addition, marked the depressed state vs. the recovered associated with a significant alteration cases. The degree of cortisol responsivity
differences existed within individuals in state. Hypersecretion appeared to be in diurnal rhythm in some, but not all,
and the shape of the curve over 24 hours during the depressed state deserve further investigation and may have implications for the course and outcome of major depression in this age group.
Key Words. Adolescents,
depression,
cortisol.
Major depressive disorder in children and adolescents is more common than previously considered, with a prevalence rate of l-496 reported from epidemiological and clinical studies (Angold, 1988a, 1988b; Goodyer et al., 1988). Disturbances of the hypothalamic-pituitary-adrenal axis have also been confirmed by the finding of elevated levels of cortisol in both plasma and saliva during the depressive episode in some 40-50% of adolescents and lo-12Yc of prepubertal children (Foreman and Goodyer, 1988; Puig-Antich et al., 1989). The majority of studies have reported findings from case-control comparisons in which control groups have been either other psychiatric cases or normals. One recent study, which compared cortisol secretion during a major depressive episode with basal adrenal activity following recovery over 24 hours from venous samples obtained at 20-min intervals, showed that cortisol levels were significantly lower when mental state improved (Puig-Antich et al., 1989). These findings, together with other studies of cortisol activity in depressed young persons, suggest that (as in adults) cortisol hypersecretion may be a Ian Goodyer, M.D., F.R.C.Psych., D.C.H., is University Lecturer in Psychiatry and Head, Section of Developmental Psychiatry, Department of Psychiatry, University of Cambridge, England. Joe Herbert, M.A., B.M., Ch.B., Ph.D., is Reader in Neuroendocrinology, Department of Anatomy, University of Cambridge, England. Stephanie Moor, M.B., B.S., M.R.C.Psych., is Research Psychiatrist, Section of Developmental Psychiatry, Department of Psychiatry, University of Cambridge, England. Patricia Altham, M.A., Ph.D., is University Lecturer, Statistical Laboratory, University of Cambridge, England. (Reprint requests to Dr. Ian Goodyer, Dept. of Psychiatry, University of Cambridge, Addenbrooke’s Hospital, Cambridge, CB2 2QQ. United Kingdom.) 0165-1781/91/$03.50 @ 1991 Elsevier Scientific Publishers Ireland Ltd.
238 state rather than a trait phenomenon and may reflect a pathophysiological process arising as a consequence of depression (Weller et al., 1985). All studies to date have also reported either absolute or mean differences in cortisol levels between groups. The pattern of cortisol secretion, as well as the magnitude, may indicate abnormal cortisol activity (Halbreich et al., 1985). Some individuals show a disruption of the diurnal rhythm of cortisol secretion, the mechanism of which may differ from that underlying hypersecretion (Healy and Williams, 1988; Kathol et al., 1989). Thus, in some depressed subjects circadian dysregulation or hypersecretion of cortisol may be present. Future studies will need to determine if either or both of these alterations in cortisol activity are present in young patients (1) at the time of disorder and (2) following recovery. The investigation of rhythm alterations is hampered by the vast number of psychological and physical stressors that increase cortisol release, including smoking, alcohol use, school examinations, and pain (Baxter and Tyrrell, 1987). Studies in young persons may be less confounded by some of these factors. The use of noninvasive procedures such as salivary sampling may further decrease the risk of conflating changes in cortisol activity. The salivary cortisol concentration is an indication of plasma free cortisol concentration (Laudat et al., 1988). Due to its lipid solubility, cortisol diffuses freely through the acinar salivary cells and its concentration in saliva is independent of salivary flow rate (Read et al., 1982; Walker et al., 1978). High correlations have been demonstrated between plasma and salivary cortisol in children (Bober et al., 1988). Thus, measurements of salivary cortisol are comparable with studies using blood. The relationship, however, between plasma and cerebrospinal fluid cortisol concentrations is not constant across time (Herbert, 1989). Studies in the rat have suggested that cortisol uptake by the brain is limited to cortisol not bound to corticobinding globulin (CBG) (Partridge et al., 1983). This represents only some 25% of the total cortisol (Baxter and Tyrrell, 1987). Measurement of the free cortisol fraction, as found in saliva, may therefore be a better reflection of the circulating cortisol concentration reaching the brain. The aim of this preliminary study was to determine if (1) cortisol hypersecretion and (2) diurnal rhythm were significantly different in subjects when in the depressed and recovered states. In addition, it was hoped to confirm the flexibility of salivary plasma sampling techniques. The study was therefore confined to ambulatory outpatient cases who would be responsible for collecting salivary samples over 24 hours in their own homes and during school hours. Methods Subjects.
Subjects were obtained from an affective disorders clinic for young persons based at the Child Development Centre, Addenbrooke’s Hospital, Cambridge, England. Referrals to the clinic come from primary care, pediatricians, and other specialists involved in child health. The clinic operates weekly for new assessments and the treatment and management of mood disorders in young persons up to age 18, including physically and mentally handicapped patients. Subjects who entered this preliminary study were chosen only from those attending normal schools and free of physical illness. mental handicap, and learning disabilities. They were between 8 and 16 years of age and were not on any medication or special diet at the time of assessment. Only ambulant outpatients without a history of drug abuse were considered.
239 Procedure. A prospective longitudinal design was used to identify depressed subjects. New clinic attenders over a 6-month period (July to December 1988) were interviewed by the first author (LG.) using the Modified Hamilton Index for Depression (MHID; Foreman and Goodyer, 1988), which was designed for use with young persons. A score of 12 or more has a greater than 95% chance of reliably identifying major depression (ICD-9 and DSM-III-R criteria) in this age group (Foreman and Goodyer, 1988). If a score of 12 or more was obtained, the subject was invited to participate in the study. Of 12 subjects invited, 10 agreed. All 10 met DSM-III-R criteria (American Psychiatric Association, 1987) for major depressive disorder. Salivary cortisols were collected from subjects at 4-hourly intervals over the 24 hours immediately following diagnosis at the first interview. Each subject started and ended the collection at the same time (8 p.m.-4 p.m. the next day). Where necessary, collections were made before meals. Plasma levels were not recorded. Therefore, neither the salivary/plasma cortisol ratio nor the level of CBG was estimated. The interviews and salivary collections were repeated when patients were clinically recovered. Subjects were followed up through the clinic, and when they were deemed to be free of depressive symptoms by the first interview, were reinterviewed in their homes by a second interviewer (S.M.) who was unaware of the original MHlD profile and total score. Salivary collection was then repeated as described above. At the time of the second interview, all subjects were ambulant, attending normal school, and free of physical illness, medication (including psychotropic drugs), or special diets. Analysis
of Cortisols. The cortisols were analyzed in the neuroendocrine research laboratory in Cambridge. Saliva samples were kept frozen at -20 “C until assay. Aliquots of 200 ,uI were extracted in IO vol ethanol, centrifuged at IOOOgfor 10 min, and decanted. They were then evaporated and resuspended in 400 ,ul phosphate gelatine buffer (PH 7.4). Samples (100 ~1) of this were assayed in duplicate. Antibody (100 ~1) (1: 1000; Bioclin) and 12,500 counts per min labeled cortisol were added and incubated overnight at 4 “C. Free and bound cortisol were separated by adding 500 ,ul dextran-charcoal mixture (0.25% charcoal + 0.025Yc dextran) for 15 min on ice. Samples were decanted into scintillation vials, 4 ml Ecoscint added, and counted along with appropriate standards. Detection limit was 15 pg/ tube, intra-assay variation was 6.7%. and interassay variation was 6.6%. Statistical Analysis. To confirm recovery, Student’s t test was used to compare total MHID scores between the first and second interviews. Comparisons were then made for cortisol secretion, with the subjects being used as their own controls, over the two time periods in four ways: (I) the mean of all samples, (2) the value for each of the six time periods, (3) the curve of cortisol secretion over 24 hours, and (4) the association between MHID scores and change in cortisol activity between depressed and recovered patients. For comparisons 1 and 2, a nonparametric test (Wilcoxon matched sign rank test) was used; for analysis 3, regression analysis (Aitkin et al., 1989) fitted transformed cortisol data to predict the likelihood of significant differences in curvature between the two conditions (depressed vs. nondepressed); for comparison 4, regression analysis was also used to investigate the relationship between MHID and cortisol.
Results of the Study Group. The 10 subjects consisted of 7 girls (3 prepubertal and 4 postpubertal) and 3 boys (1 prepubertal and 2 postpubertal). Mean age at entry into the project was 12.68 years (SD = 1.85; range 10.2-15.9). At the first interview the mean MHID score was 23.5 (SD = 7.49; range 15-36). All cases received active treatment: eight received a tricyclic antidepressant and Characteristics
240 psychological treatment combined, and two received psychological treatment alone. The mean time between first and second interview was 8.2 months (SD = 2.44; range 5-12 months). At the second interview, the mean MHID score was 4.8 (SD = 4.6; range O-1 1). The mean MHID score was significantly lower at the second interview (t = 7.13, df= 9,p
237
Elsevier
Cortisol Hypersecretion in Depressed School-Aged Children and Adolescents Ian Goodyer. Received April24,
Joe Herbert,
Stephanie
Moor, and Patricia Altham
1990; revised version received November 7, 1990; accepted January 4, 1991.
Abstract. A prospective longitudinal study has been carried out to determine the secretory pattern of cortisol in children (n = 10) with major depressive disorder. Salivary cortisol samples were collected at 4-hourly intervals over 24 hours when the subjects were depressed and again when they were recovered. Group comparison indicated that significant increases in mean cortisol output occurred during illness as compared with recovery. This difference occurred only at three points (midnight, 4 a.m., 8 a.m.) of six measured. Not all cases were showed hypersecretion, but when hypersection was present, it occurred in cases with more severe symptoms. In addition, marked the depressed state vs. the recovered associated with a significant alteration cases. The degree of cortisol responsivity
differences existed within individuals in state. Hypersecretion appeared to be in diurnal rhythm in some, but not all,
and the shape of the curve over 24 hours during the depressed state deserve further investigation and may have implications for the course and outcome of major depression in this age group.
Key Words. Adolescents,
depression,
cortisol.
Major depressive disorder in children and adolescents is more common than previously considered, with a prevalence rate of l-496 reported from epidemiological and clinical studies (Angold, 1988a, 1988b; Goodyer et al., 1988). Disturbances of the hypothalamic-pituitary-adrenal axis have also been confirmed by the finding of elevated levels of cortisol in both plasma and saliva during the depressive episode in some 40-50% of adolescents and lo-12Yc of prepubertal children (Foreman and Goodyer, 1988; Puig-Antich et al., 1989). The majority of studies have reported findings from case-control comparisons in which control groups have been either other psychiatric cases or normals. One recent study, which compared cortisol secretion during a major depressive episode with basal adrenal activity following recovery over 24 hours from venous samples obtained at 20-min intervals, showed that cortisol levels were significantly lower when mental state improved (Puig-Antich et al., 1989). These findings, together with other studies of cortisol activity in depressed young persons, suggest that (as in adults) cortisol hypersecretion may be a Ian Goodyer, M.D., F.R.C.Psych., D.C.H., is University Lecturer in Psychiatry and Head, Section of Developmental Psychiatry, Department of Psychiatry, University of Cambridge, England. Joe Herbert, M.A., B.M., Ch.B., Ph.D., is Reader in Neuroendocrinology, Department of Anatomy, University of Cambridge, England. Stephanie Moor, M.B., B.S., M.R.C.Psych., is Research Psychiatrist, Section of Developmental Psychiatry, Department of Psychiatry, University of Cambridge, England. Patricia Altham, M.A., Ph.D., is University Lecturer, Statistical Laboratory, University of Cambridge, England. (Reprint requests to Dr. Ian Goodyer, Dept. of Psychiatry, University of Cambridge, Addenbrooke’s Hospital, Cambridge, CB2 2QQ. United Kingdom.) 0165-1781/91/$03.50 @ 1991 Elsevier Scientific Publishers Ireland Ltd.
238 state rather than a trait phenomenon and may reflect a pathophysiological process arising as a consequence of depression (Weller et al., 1985). All studies to date have also reported either absolute or mean differences in cortisol levels between groups. The pattern of cortisol secretion, as well as the magnitude, may indicate abnormal cortisol activity (Halbreich et al., 1985). Some individuals show a disruption of the diurnal rhythm of cortisol secretion, the mechanism of which may differ from that underlying hypersecretion (Healy and Williams, 1988; Kathol et al., 1989). Thus, in some depressed subjects circadian dysregulation or hypersecretion of cortisol may be present. Future studies will need to determine if either or both of these alterations in cortisol activity are present in young patients (1) at the time of disorder and (2) following recovery. The investigation of rhythm alterations is hampered by the vast number of psychological and physical stressors that increase cortisol release, including smoking, alcohol use, school examinations, and pain (Baxter and Tyrrell, 1987). Studies in young persons may be less confounded by some of these factors. The use of noninvasive procedures such as salivary sampling may further decrease the risk of conflating changes in cortisol activity. The salivary cortisol concentration is an indication of plasma free cortisol concentration (Laudat et al., 1988). Due to its lipid solubility, cortisol diffuses freely through the acinar salivary cells and its concentration in saliva is independent of salivary flow rate (Read et al., 1982; Walker et al., 1978). High correlations have been demonstrated between plasma and salivary cortisol in children (Bober et al., 1988). Thus, measurements of salivary cortisol are comparable with studies using blood. The relationship, however, between plasma and cerebrospinal fluid cortisol concentrations is not constant across time (Herbert, 1989). Studies in the rat have suggested that cortisol uptake by the brain is limited to cortisol not bound to corticobinding globulin (CBG) (Partridge et al., 1983). This represents only some 25% of the total cortisol (Baxter and Tyrrell, 1987). Measurement of the free cortisol fraction, as found in saliva, may therefore be a better reflection of the circulating cortisol concentration reaching the brain. The aim of this preliminary study was to determine if (1) cortisol hypersecretion and (2) diurnal rhythm were significantly different in subjects when in the depressed and recovered states. In addition, it was hoped to confirm the flexibility of salivary plasma sampling techniques. The study was therefore confined to ambulatory outpatient cases who would be responsible for collecting salivary samples over 24 hours in their own homes and during school hours. Methods Subjects.
Subjects were obtained from an affective disorders clinic for young persons based at the Child Development Centre, Addenbrooke’s Hospital, Cambridge, England. Referrals to the clinic come from primary care, pediatricians, and other specialists involved in child health. The clinic operates weekly for new assessments and the treatment and management of mood disorders in young persons up to age 18, including physically and mentally handicapped patients. Subjects who entered this preliminary study were chosen only from those attending normal schools and free of physical illness. mental handicap, and learning disabilities. They were between 8 and 16 years of age and were not on any medication or special diet at the time of assessment. Only ambulant outpatients without a history of drug abuse were considered.
239 Procedure. A prospective longitudinal design was used to identify depressed subjects. New clinic attenders over a 6-month period (July to December 1988) were interviewed by the first author (LG.) using the Modified Hamilton Index for Depression (MHID; Foreman and Goodyer, 1988), which was designed for use with young persons. A score of 12 or more has a greater than 95% chance of reliably identifying major depression (ICD-9 and DSM-III-R criteria) in this age group (Foreman and Goodyer, 1988). If a score of 12 or more was obtained, the subject was invited to participate in the study. Of 12 subjects invited, 10 agreed. All 10 met DSM-III-R criteria (American Psychiatric Association, 1987) for major depressive disorder. Salivary cortisols were collected from subjects at 4-hourly intervals over the 24 hours immediately following diagnosis at the first interview. Each subject started and ended the collection at the same time (8 p.m.-4 p.m. the next day). Where necessary, collections were made before meals. Plasma levels were not recorded. Therefore, neither the salivary/plasma cortisol ratio nor the level of CBG was estimated. The interviews and salivary collections were repeated when patients were clinically recovered. Subjects were followed up through the clinic, and when they were deemed to be free of depressive symptoms by the first interview, were reinterviewed in their homes by a second interviewer (S.M.) who was unaware of the original MHlD profile and total score. Salivary collection was then repeated as described above. At the time of the second interview, all subjects were ambulant, attending normal school, and free of physical illness, medication (including psychotropic drugs), or special diets. Analysis
of Cortisols. The cortisols were analyzed in the neuroendocrine research laboratory in Cambridge. Saliva samples were kept frozen at -20 “C until assay. Aliquots of 200 ,uI were extracted in IO vol ethanol, centrifuged at IOOOgfor 10 min, and decanted. They were then evaporated and resuspended in 400 ,ul phosphate gelatine buffer (PH 7.4). Samples (100 ~1) of this were assayed in duplicate. Antibody (100 ~1) (1: 1000; Bioclin) and 12,500 counts per min labeled cortisol were added and incubated overnight at 4 “C. Free and bound cortisol were separated by adding 500 ,ul dextran-charcoal mixture (0.25% charcoal + 0.025Yc dextran) for 15 min on ice. Samples were decanted into scintillation vials, 4 ml Ecoscint added, and counted along with appropriate standards. Detection limit was 15 pg/ tube, intra-assay variation was 6.7%. and interassay variation was 6.6%. Statistical Analysis. To confirm recovery, Student’s t test was used to compare total MHID scores between the first and second interviews. Comparisons were then made for cortisol secretion, with the subjects being used as their own controls, over the two time periods in four ways: (I) the mean of all samples, (2) the value for each of the six time periods, (3) the curve of cortisol secretion over 24 hours, and (4) the association between MHID scores and change in cortisol activity between depressed and recovered patients. For comparisons 1 and 2, a nonparametric test (Wilcoxon matched sign rank test) was used; for analysis 3, regression analysis (Aitkin et al., 1989) fitted transformed cortisol data to predict the likelihood of significant differences in curvature between the two conditions (depressed vs. nondepressed); for comparison 4, regression analysis was also used to investigate the relationship between MHID and cortisol.
Results of the Study Group. The 10 subjects consisted of 7 girls (3 prepubertal and 4 postpubertal) and 3 boys (1 prepubertal and 2 postpubertal). Mean age at entry into the project was 12.68 years (SD = 1.85; range 10.2-15.9). At the first interview the mean MHID score was 23.5 (SD = 7.49; range 15-36). All cases received active treatment: eight received a tricyclic antidepressant and Characteristics
240 psychological treatment combined, and two received psychological treatment alone. The mean time between first and second interview was 8.2 months (SD = 2.44; range 5-12 months). At the second interview, the mean MHID score was 4.8 (SD = 4.6; range O-1 1). The mean MHID score was significantly lower at the second interview (t = 7.13, df= 9,p
