Directory UMM :Data Elmu:jurnal:B:Biological Psichatry:Vol48.Issue10.2000:
Altered Salivary Dehydroepiandrosterone Levels in
Major Depression in Adults
Albert Michael, Alison Jenaway, Eugene S. Paykel, and Joe Herbert
Background: The authors sought to examine whether
levels of dehydroepiandrosterone are abnormal in
depression.
Methods: Three groups of subjects aged 20 – 64 were
studied: 44 major depressives, 35 subjects with partially
or completely remitted depression, matched as far as
possible for age and drug treatment, and 41 normal
control subjects. Dehydroepiandrosterone and cortisol in
saliva were determined from specimens taken at 8:00 AM
and 8:00 PM on 4 days.
Results: The mean age of the three groups did not differ.
Dehydroepiandrosterone was lowered at 8:00 AM and 8:00
PM compared with control subjects. Values for the remitted
group were intermediate. Dehydroepiandrosterone levels
at 8:00 AM correlated negatively with severity of depression and were not related to drug treatment or smoking,
but decreased with age (as expected). Cortisol was elevated in depression in the evening. The molar cortisol/
dehydroepiandrosterone ratio also differentiated those
with depression from the control group.
Conclusions: Lowered dehydroepiandrosterone levels are
an additional state abnormality in adult depression. Adrenal steroid changes are thus not limited to cortisol.
Because dehydroepiandrosterone may antagonize some
effects of cortisol and may have mood improving properties, these findings may have significant implications for
the pathophysiology of depression. Biol Psychiatry
2000;48:989 –995 © 2000 Society of Biological
Psychiatry
Key Words: DHEA, cortisol, depression, adults, saliva,
rhythm
Introduction
T
he adrenal cortex secretes a wide spectrum of steroid
hormones, but nearly all attention on endocrine abnormalities associated with affective disorder has been
reserved for cortisol. Hypersecretion of cortisol occurs in
From the Departments of Psychiatry (AM, AJ, ESP) and Anatomy (JH), University
of Cambridge, Cambridge, United Kingdom.
Address reprint requests to Joe Herbert, University of Cambridge, Department of
Anatomy, Cambridge CB2 3DY, United Kingdom.
Received March 6, 2000; revised June 13, 2000; accepted June 13, 2000.
© 2000 Society of Biological Psychiatry
about 50% of cases, whether assessed by increased basal
levels, or by heightened resistance to the suppressive
feedback effects of dexamethasone (Butler and Besser
1968; Carroll and Mendels 1976; Murphy 1991a). Abnormal corticoid secretion may mark either the course of
depression or predict subsequent relapse following apparent recovery (Holsboer 1992). Increased levels of glucocorticoids, deriving either from endogenous sources (as
in Cushing’s syndrome) or exogenous ones (as during
steroid therapy) are known to induce abnormal affective
states, particularly depression (Dorn et al 1997; Lewis and
Smith 1983; Starkman et al 1981). Administration of
metyrapone, a compound that reduces cortisol levels, has
been reported to be beneficial in depression (O’Dwyer et
al 1995). In a recent study of depressed children and
adolescents, we have identified those with comorbid
dysthymia (i.e., “chronic depression”) as being most likely
to show evening hypercortisolaemia (Goodyer et al 1996;
Herbert et al 1996).
There are, however, other adrenal steroids that may be
equally as significant for understanding the psychopathology of depression. Dehydroepiandrosterone (DHEA) is
particularly interesting. The sulphated form of DHEA
(DHEAS) is the most abundant circulating steroid in man,
although information on its function has been elusive until
recently. Blood DHEA and DHEAS increase markedly
during early childhood and again during puberty (Parker et
al 1978). Levels peak in early adulthood and decrease
equally markedly thereafter (2–3% per year), so that by
age 60 they are about one third of those at age 20
(Orentreich et al 1992). Chronic physical ill-health or
more acute events, such as imminent surgery, result in an
elevated cortisol level but lowered DHEA and altered
cortisol/DHEA ratios (e.g., Gray et al 1991; Hedman et al
1992; Spratt et al 1993). One important feature of
DHEA(S) is thus that it can alter independently of cortisol
both during normal life and in illness. A second one is that
there is increasing evidence that DHEA can act as an
antagonist to glucocorticoids (Blauer et al 1991; May et al
1991; Shafagoj et al 1992).
The relevance of this to psychiatric disorder has been
shown recently. An extensive study of depression in
adolescents showed that about 30% have abnormally low
0006-3223/00/$20.00
PII S0006-3223(00)00955-0
990
A. Michael et al
BIOL PSYCHIATRY
2000;48:989 –995
DHEA levels; however, this occurred at a different time
(in the morning) than high (evening) cortisol (Goodyer et
al 1996; Herbert et al 1996). Levels of DHEAS were not
significantly altered by depression (Goodyer et al 1996).
Depressed adolescents with high cortisol/DHEA ratios and
experiencing recent adverse life events were unlikely to
have recovered 36 weeks later (Goodyer et al 1998).
Whether the new findings on DHEA in adolescents apply
to adults with major depressive disorder is not known. One
report of DHEA levels in adult depressives was carried out
on only nine depressed men compared with nine control
subjects, and showed that DHEA levels at 8:00 AM or 4:00
PM were no different in the two groups (Osran et al 1993);
however, the diurnal decrease in DHEA was less in the
depressed group than the control subjects. More recently,
the daily minima in DHEA levels have been reported to be
increased in depression (26 cases; Heuser et al 1998).
There is also a report of decreased cortisol/DHEA ratios in
anorexia nervosa (Devesa et al 1988; Zumoff et al 1983),
which return to normal upon restoration of normal weight
(Winterer et al 1985). “Well-being” was increased by
administration of DHEA to middle-aged men and women,
and plasma DHEA (but not DHEAS) correlated positively
with well-being in postmenopausal women (Cawood and
Bancroft 1996; Morales et al 1994). Recent preliminary
reports in depressed adults noted that ingestion of oral
DHEA over 4 weeks improved mental state and memory
function (Wolkowitz et al 1995, 1999). In patients with
Addison’s disease who lack DHEA(S), supplementation
improved well-being and reduced depression and anxiety
scores (Arlt et al 1999). These findings on DHEA suggest
that endocrine abnormalities in depression may extend to
steroids other than cortisol, and that the psychopathology
of this condition may be more complex than is presently
thought.
This article reports, for the first time, lowered (morning)
DHEA in major depression in adults compared with a
control group of age- and gender-matched subjects. Because concurrent drug treatment is a confounding factor in
such studies, this conclusion is reinforced by the inclusion
of an additional control group of subjects whose depression had undergone remission, most of whom were still
taking anti-depressant treatment.
Methods and Materials
Subjects
Three groups aged 20 – 64 years were studied.
MAJOR DEPRESSIVE DISORDER. There were 44 subjects
with depression (12 male, 32 female), aged 20 – 64 years,
recruited either as inpatients (36) or from an outpatient (8) clinic.
Subjects were accepted only if free from other disorders (see
below). All subjects met DSM-IV criteria (American Psychiatric
Association 1994) for unipolar major depressive disorder, single
or recurrent episodes (moderate or severe) without psychotic
symptoms. Smoking, alcohol intake, and current drug therapy
were recorded.
REMITTED DEPRESSION. Thirty-five subjects (14 male, 21
female) with a previous episode of depression but who had
recovered (i.e., no longer met DSM-IV criteria for major depressive disorder), were collected from follow-up outpatients. These
subjects were either in complete or partial remission. The
17-item Hamilton Rating Scale (HAM-D; Hamilton 1967) scales
were completed for each subject. Smoking, alcohol intake, and
current drug therapy were recorded
NORMAL CONTROL SUBJECTS. Forty-one normal control
subjects (14 male, 27 female) were recruited from hospital staff
and age matched with the depressed group. Saliva was collected
as for the other two groups. These subjects did not complete the
psychiatric rating scales, and were apparently free of current
illness (assessed from history). Smoking and alcohol intake were
recorded. In all groups, liver function tests and urine screening
were carried out if drug or alcohol abuse was suspected.
All subjects provided written informed consent following
complete description of the study, which was approved by the
Cambridge Local Ethics Committee.
Procedures
RATINGS. Subjects in the first two groups were assessed
using the Brief Psychiatric Rating Scale (BPRS), the HAM-D,
and the Clinical Interview for Depression (CID); the latter was
divided into depression and anxiety total scores (Overall and
Gorham 1962; Paykel 1985).
CORTISOL AND DHEA IN THE SALIVA. Four salivary
samples were collected at 8:00 AM, and another four at 8:00 PM,
usually on consecutive days. Subjects were instructed not to
clean their teeth within 2 hours of collecting saliva, to wash out
their mouths with clean water immediately before collection, and
to obtain about 5 mL of saliva by spitting into a plastic tube; no
salivation aids were used. Samples were taken individually into
prelabeled tubes supplied by the laboratory. They were stored in
the freezing compartments (220°C) of the subjects’ refrigerators
before being transferred to the laboratory.
ASSAYS. Salivary cortisol was measured by an enzymelinked immunosorbent assay technique, using unextracted duplicates from each sample. The final value at each time point for
each subject was the mean of the four samples collected at that
point. Variations: intra-assay, 4.5%; inter-assay, 4.4% (based on
pooled quality control subjects incorporated into each assay run).
Salivary DHEA was measured by radioimmunoassay, following
extraction with hexane ether. Variations: intra-assay, 5.8%;
inter-assay, 13.3%. Both assays have been extensively validated
(Goodyer et al 1996). Molar cortisol/DHEA ratios were calculated for each of the two time points.
DHEA in Major Depression
991
BIOL PSYCHIATRY
2000;48:989 –995
Table 1. Characteristics of the Sample
Depressed
Remitted
Control subjects
M (n)
F (n)
Age (years)
HAM-D
CID
12
14
14
32
21
27
43.2 6 11.17 (20 – 64)
41.4 6 9.25 (22–59)
40.6 6 9.62 (27– 62)
24.1 6 3.99 (17–33)
2.6 6 2.25 (0 – 8)
56.2 6 9.75
5.5 6 4.81
Values are mean 6 SD. Ranges in parentheses. M, male; F, female; HAM-D, Hamilton Rating Scale (Hamilton 1967); CID, Clinical Interview for Depression.
STATISTICAL ANALYSIS. Between-variable correlations
and between-group comparisons were made using two-way, univariate, or repeated-measures analyses of variance (ANOVAs; with
appropriate covariances) or nonparametric statistics (Kruskal–
Wallis [K-W] and Mann–Whitney [M-W] tests). All tests are
two tailed.
Results
Subjects
Table 1 shows the age and gender distribution of the three
samples. Neither age (K-W x2 5 0.57, p 5 ns) nor gender
(x2 5 1.31, p 5 ns) differed significantly between the
groups; however, smoking (defined as more than five
cigarettes per day) was significantly more frequent in the
depressed group (K-W x2 5 8.96, p , .01). There were 12
smokers in the depressed group (27.3%), compared with
seven in the remitted group (20.0%) and only two (4.9%)
in the control subjects group. The mean HAM-D scores in
the depressed group were no different in smokers and
nonsmokers [F(1,42) 5 2.6, p 5 ns]. Alcohol intake
(reported units per week) did not differ between the groups
(K-W x2 5 0.72, p 5 ns).
The HAM-D scores for the depressed group were
significantly greater than the remitted group (M-W U 5
7.5, p , .001; Table 1). All the depressed group had
HAM-D scores of greater than 17; the highest score in the
recovered group was 8. In the depressed group, there were
highly significant correlations between the HAM-D and
both CID depression (Spearman r 5 .63, p , .001) and
CID anxiety scores (r 5 .50, p , .001).
Drugs
Of the depressed group, 41 were taking psychotropic
drugs; three were not. Of the recovered group, 26 were
also taking drugs, and nine were drug free.
In the depressed group, 31 were taking a single drug,
nine were taking two, and one was taking three; 14 were
on tricyclic antidepressants, 18 on selective serotonin
reuptake inhibitors (SSRIs), and nine on monoamine
oxidase inhibitors (MAOIs). In addition to their primary
medication, 10 were also taking lithium (as an adjunct to
antidepressant treatment), one was taking benzodiazepines, and one sodium valproate (as an antidepressant). In
the recovered group, 10 were on TCAs, 14 were on SSRIs,
and one was on MAOIs. Six were also taking lithium.
Salivary Steroids
DHEA. At both 8:00 AM and 8:00 PM, DHEA was
negatively correlated with age (Spearman: AM r 5 2.36,
p , .0001; PM r 5 2.33, p , .0001). In male subjects,
DHEA (but not cortisol) was also higher (t tests; p , .02);
there was no significant difference in age.
A two-way ANOVA (group and smoking [yes/no] as
factors, age as covariate) showed that both morning and
evening levels of salivary DHEA were significantly different between the groups [DHEA AM F(6,98) 5 3.5, p ,
.02; PM F(6,98) 5 3.4, p , .02; Figure 1); there was no
main effect of smoking and no interaction between smoking and group. Further univariate ANOVAs showed that
morning DHEA was significantly lower in depressed than
control groups, whereas evening DHEA was lower in
depressed compared to remitted groups (Bonferroni test,
p , .05; Figure 1, top). A repeated-measures multivariate
ANOVA (MANOVA) on both morning and evening
DHEA (age as covariate) also showed significant group
effects [F(2,105) 5 3.55, p 5 .03].
Within the depressed group, there was a highly significant negative correlation between HAM-D scores and
morning DHEA (Spearman r 5 2.49, p , .01; Figure 2).
Similar associations were found with CID total depression
(r 5 2.38, p , .01) and total anxiety scores (r 5 2.37,
p , .05; Table 2). Evening DHEA correlated significantly
only with the CID anxiety score (Table 2).
In the depressed group, morning DHEA was not significantly lower in smokers than nonsmokers [F(1,38) 5 3.9,
p 5 .06]. There was no association between age and
HAM-D scores. In the remitted group, those currently on
medication had higher HAM-D scores than those who
were drug-free (M-W U 5 3.7, p , .001) and also higher
PM cortisol levels (U 5 2.3, p , .02), but no differences in
DHEA.
CORTISOL. There were no correlations between cortisol and age at either time point. Cortisol (8:00 AM,
8:00 PM) was significantly higher in the saliva of depressed
subjects than the other two groups at both time points
992
A. Michael et al
BIOL PSYCHIATRY
2000;48:989 –995
Figure 2. The relation between salivary dehydroepiandrosterone
(DHEA) levels at 8:00 AM and Hamilton Rating Scale (Hamilton
1967) scores in the depressed subjects.
ASSOCIATIONS BETWEEN SALIVARY CORTISOL AND
Overall, there were highly significant correlations between morning and evening levels of both cortisol
and DHEA. There were also significant associations between morning cortisol and DHEA. In the depressed
group, there were similar associations between morning
and evening levels of both cortisol and DHEA but not for
morning cortisol and DHEA (Table 3).
DHEA.
Figure 1. Depression is associated with lowered dehydroepiandrosterone (DHEA), raised salivary cortisol, and elevated
cortisol/DHEA ratios. The mean (6 SD) salivary levels of
DHEA, cortisol, and the molar cortisol/DHEA ratio at either
8:00 AM (gray bars) or 8:00 PM (patterned bars) in three groups
of subjects (depressed, recovered depressive, and normal control)
are shown. Significance levels given in text.
[AM F(2,117) 5 4.9, p , .01; PM F(2,117) 5 5.4, p , .01;
Figure 1, middle). Salivary cortisol in the depressed
subjects differed significantly from the control subjects
and remitted groups (AM p , .05, PM p , .05; Bonferroni
test). There were no significant correlations within the
depressed group between cortisol at either time point and
HAM-D, CID depression, or CID anxiety scores. A
repeated-measures MANOVA, taking morning and
evening levels of each hormone as dependent variables
and age and smoking as covariates, showed significant
group effects for cortisol [F(2,105) 5 5.96, p 5 .01).
MOLAR CORTISOL/DHEA RATIOS. Overall, there
were significant differences between the groups for both
morning (K-W x2 5 31.0, p , .0001) and evening (x2 5
28.1, p , .0001) molar cortisol/DHEA ratios (Figure 1,
bottom). Group comparisons showed that the depressed
group differed significantly from the recovered subjects
(AM x2 5 13.1, p 5 .0003; PM x2 5 20.9, p , .0001),
whereas the control subjects and the recovered groups
were indistinguishable (x2 5 2.2 and 0.2; ps 5 ns).
Table 2. Associations between Salivary Steroids and
Psychometric Scores in the Depressed Group (n 5 44)
HAM-D
DHEA (8:00 AM)
DHEA (8:00 PM)
Cortisol (8:00 AM)
Cortisol (8:00 PM)
a
20.48
20.08
20.20
0.13
CIDd
CIDa
a
20.37a
20.28
20.12
20.04
20.38
20.09
20.06
20.12
HAM-D, Hamilton Rating Scale (Hamilton 1967); CIDd, Clinical Interview for
Depression (Depression scale); CIDa, Clinical Interview for Depression (Anxiety
scale); DHEA, dehydroepiandrosterone.
a
p , .01.
DHEA in Major Depression
BIOL PSYCHIATRY
2000;48:989 –995
993
Table 3. Spearman Correlations between Salivary Cortisol and Dehydroepiandrosterone (DHEA) in All Subjects (n 5 107) or in
the Depressed Group Alone (n 5 44)
All subjects
DHEA (8:00 PM)
Cortisol (8:00 PM)
Cortisol (8:00 PM)
a
b
DHEA
(8:00 AM)
DHEA
(8:00 PM)
.72a
.24b
2.03
.01
.11
Depressed group only
Cortisol
(8:00 AM)
DHEA
(8:00 AM)
DHEA
(8:00 PM)
Cortisol
(8:00 AM)
.45a
.66a
.24
2.15
.13
2.04
.39a
p , .01.
p , .05.
Taking the 85th percentile morning cortisol/DHEA ratio
of the control group as an arbitrary (post hoc) cut-off,
82.5% of the depressed group had ratios that were equal to
or greater than this value (control 15%). The corresponding result for the ratio at 8:00 PM was 46% (Table 4).
Discussion
The principal new finding of this study is that mean levels
of DHEA in the saliva of depressed adults are lower than
those in normal control subjects. There are a number of
possible confounding variables that need to be considered
before this endocrine parameter can be added to the known
pathophysiology of this condition in adults.
The first, and most obvious, is the coincident effect of
the antidepressant treatment being taken by the depressed
subjects. We attempted to control for this by recruiting a
number of remitted depressives, the majority of whom
were still taking antidepressant treatment. It was noticeable that DHEA levels in this group as a whole fell
between that in the depressed subjects and the normal
control subjects. As might be expected, in the remitted
group the HAM-D scores of those who continued treatment were higher than those who were drug free; however,
separating those on antidepressant drugs failed to show
any significant effect on DHEA levels, though the numbers in each group are not large. The possible role of drug
treatment needs to be re-examined on a much larger
population than we were able to gather, but there is no
Table 4. The Number of Each Group Having Molar Salivary
Cortisol/Dehydroepiandrosterone (DHEA) Ratios in the Upper
85th Percentiles of the Values in Normal Control Subjects
Cortisol/DHEA ratio
Group
Depressed
Recovered
Control subjects
8:00 AM
(85th percentile 5 9.6)
8:00 PM
(85th percentile 5 5.7)
33/40 (82.5%)
11/34 (32.4%)
6/41 (14.2%)
18/39 (46.1%)
2/34 (5.9%)
6/40 (15.0%)
Overall significance: 8:00
.001).
AM
x2 5 40.4 (p , .01); 8:00
PM
x2 5 18.8 (p ,
evidence from our results that drug treatment itself was
responsible for the lower levels of DHEA recorded in
depressed subjects. The concomitant findings of elevated
salivary cortisol in the depressed (compared to the other
two groups) reinforces the conclusion that lowered morning DHEA is a state-dependent feature of at least a
proportion of depressed subjects. Raised cortisol (in both
blood and saliva) is a well-known association with this
condition and is not the result of coincident drug therapy
(Butler and Besser 1968; Gold et al 1988).
The association between DHEA and depression was
strongly supported by the finding of a negative correlation
between morning DHEA and HAM-D scores in the
depressed group. This results both supports the conclusion
that morning DHEA levels may be a state-related feature
of depression, and further argues against a simple drugrelated effect on DHEA. The fact that no such association
was found for cortisol suggests that there may be differences in the way that these two steroids relate to the
depressed state, or may reflect the limited range of severity
within the currently depressed sample.
Age is another possible confounding factor, because
DHEA decreases with age (Orentriech 1992); however, our
groups were well balanced for age and— had this been a
factor— both morning and evening DHEA levels would have
been expected to decline. Smoking was more prevalent in the
depressed subjects; but smoking has been found to raise
DHEA(S) levels (Khaw et al 1988), and so this effect, if
present, would act against the findings we report. Moreover,
statistical analysis failed to show an independent effect of
smoking on DHEA levels in our sample.
We measured levels of DHEA and cortisol twice daily for
4 days. Steroid levels may fluctuate for a number of reasons,
but the 4-day mean would have eliminated most of these
adventitious sources of variation. Salivary levels of both
DHEA and cortisol correlate well with those in the blood
(Goodyer et al 1996; Guazzo et al 1996; Woolston et al
1983); in both cases, levels in the saliva represent about 5%
of those in the blood. There is now a considerable literature
on the reliability of measuring steroids in the saliva, and the
way that such levels track those in the blood (Kirshbaum and
994
A. Michael et al
BIOL PSYCHIATRY
2000;48:989 –995
Hellhammer 1994). It seems highly likely, therefore, that
depressed subjects have lower levels of blood DHEA in the
morning, though the exact shape of the diurnal variation in
DHEA needs to be examined more closely. A major conclusion, however, is that changes in DHEA may not be parallel
to those in cortisol. The latter are more obvious in the evening
(confirming previous reports: e.g., Seckl et al 1990; review
by Murphy 1991a), though morning cortisol is also elevated.
Whether the same subjects show both changes in cortisol or
DHEA, or whether these can occur independently, could not
be reliably determined from our study on about 40 subjects
but should be become apparent when much larger populations are examined. For the same reasons, we cannot reliably
analyze the relation between DHEA and depression in male
and female subjects separately in this study, nor assess the
contribution of individual drugs to the endocrine levels
observed in the depressed or comparison groups.
Levels of cortisol and DHEA in the saliva (as in the
blood) show diurnal changes, with higher values at 8:00
AM than 8:00 PM (Krieger et al 1971); however, the
amplitude of the cortisol rhythm is much greater than
DHEA. Hence, the molar cortisol/DHEA ratio in control
subjects was 8.3 at 8:00 AM but only 2.7 at 8:00 PM (i.e.,
a three-fold change across 12 hours). This indicates that
tissues, including the brain, are exposed to very different
relative concentrations of cortisol and DHEA during the
day. In view of the accumulating evidence that DHEA can
act as an antagonist to corticoids (Blauer et al 1991;
May et al 1991), it was noteworthy that the molar
cortisol/DHEA ratios (particularly at 8:00 AM) clearly
separated depressed and normal subjects. Women, not
currently depressed, but with higher morning salivary
cortisol levels and subsequently experiencing an adverse
life event, have been shown to be at increased risk of
becoming depressed within the following 12 months
(Harris et al, in press); treatments designed to lower
cortisol have been found to decrease levels of depression
(Murphy 1991b; O’Dwyer et al 1995; Reus et al 1997).
Dehydroepiandrosterone antagonizes the damaging actions of corticosterone on cultured hippocampal neurons
(Kimonides et al 1999), and there is some evidence for its
being antidepressive (Wolkowitz et al 1995, 1997, 1999).
Cortisol/DHEA ratios have been found to predict the
course of depression in adolescents (Goodyer et al 1998),
which suggests that measuring both may give a more
complete picture of the steroidal changes associated with
depression. If cortisol does affect either onset or recovery
(O’Toole et al 1997), and if DHEA acts as an effective
antagonist to cortisol in this context, then both may need
to be taken into account.
Comorbid anxiety is common in depression. In this study,
relationships with morning salivary DHEA were similar to
depression as assessed by the CID, but evening DHEA was
associated only with anxiety. This would not necessarily
apply to pure anxiety disorder, in which depression and
anxiety are not so closely related as in this sample.
The findings in this article confirm and extend those
already reported on first-onset depression in 8- to 16-yearolds, on a population largely free of antidepressant treatment (Goodyer et al 1996; Herbert et al 1996). Depressed
subjects in this age group also show raised evening
salivary cortisol and lower morning DHEA, and the fact
that the results on the two groups are so similar increases
the reliance that can placed on either. The main conclusion
from these studies is that alterations in DHEA, a steroid
with known actions on the brain, may be an intrinsic
feature of depression in adults as well as in adolescents,
and this finding may give added insight into the psychopathology of this condition.
We thank Helen Shiers and Sarah Cleary for the assays. Conducted
within the MRC co-operative group in Brain, Behaviour and
Neuropsychiatry.
References
American Psychiatric Association (1994): Diagnostic and Statistical Manual of Mental Disorders, 4th ed. Washington, DC:
American Psychiatric Press.
Arlt W, Callies F, van Vlijmen JC, Koehler I, Reinicke M,
Bildingmaier M, et al (1999): Dehydroepiandrosterone replacement in women with adrenal insufficiency. N Engl
J Med 314:1013–1018.
Blauer KL, Poth M, Rogers WM, Bernton EW (1991): Dehydroepiandrosterone antagonises the suppressive effects of dexamethasone on lymphocytes. Endocrinology 129:3174 –3179.
Butler PWP, Besser GM (1968): Pituitary-adrenal function in
severe depressive illness. Lancet 1(7554):1234 –1236.
Carroll BJ, Mendels J (1976): Neuroendocrine regulation in
depression. In: Sachar EJ, editor. Hormones Behavior and
Psychopathology. New York: Raven.
Cawood EHH, Bancroft J (1996): Steroid hormones, the menopause, sexuality and well-being of women. Psychol Med
26:925–936.
Devesa J, Perez-Fernandez R, Bokser L, Casanueva FF (1988):
Adrenal androgen secretion and dopaminergic activity in
anorexia nervosa. Horm Metab Res 20:57– 60.
Dorn LD, Burgess ES, Friedman TC, Dubbert B, Gold PW,
Chrousoos GP (1997): The longitudinal course of psychopathology in Cushing’s syndrome after correction of hypercortisolism. J Clin Endocrinol Metab 82:912–919.
Gold PW, Goodwin FK, Chrousos GP (1988): Clinical and
biochemical manifestations of depression. Relation to neurobiology of stress. N Engl J Med 319:413– 420.
Goodyer IM, Herbert J, Altham PME (1998): Adrenal secretion
during major depression in 8 to 16 year olds III. Influence of
cortisol/DHEA ratio at presentation on subsequent rates of
disappointing life events and persistent major depression.
Psychol Med 28:265–273.
DHEA in Major Depression
Goodyer IM, Herbert J, Altham PME, Secher S, Shiers HM
(1996): Adrenal secretion during major depression in 8 to 16
year olds I. Altered diurnal rhythms in salivary cortisol and
dehydroepiandrosterone (DHEA): At presentation. Psychol
Med 26:245–256.
Gray A, Feldman HA, McKinley JB, Longcope C (1991): Age,
disease and changing sex hormone levels in middle-aged
men. Results of the Massachusetts male aging study. J Clin
Endocrinol Metab 73:1016 –1025.
Guazzo EP, Kirkpatrick PJ, Goodyer IM, Shiers HM, Herbert J
(1996): Cortisol, dehydroepiandrosterone (DHEA) and
DHEA sulfate in the cerebrospinal fluid of man: Relation to
blood levels and the effects of age. J Clin Endocrinol Metab
81:3951–3960.
Hamilton M (1967): Development of a rating scale for primary
depressive illness. Br J Soc Clin Psychol 6:278 –296.
Harris TO, Borsanyi S, Messari S, Stanford K, Cleary SE, Shiers
HM, et al (in press): Morning cortisol as a risk factor for
subsequent major depressive disorder in adult women. Br J
Psychiatry.
Hedman M, Nilsson E, de la Torre B (1992): Low blood and
synovial fluid levels of sulpho-conjugated steroids in rheumatoid arthritis. Clin Exp Rheumatol 10:25–30.
Herbert J, Goodyer IM, Altham PME, Secher S, Shiers HM
(1996): Adrenal steroid secretion and major depression in 8 to
16 year olds II. Influence of comorbidity at presentation.
Psychol Med 26:257–263.
Heuser I, Deuschle M, Luppa P, Schweiger H, Weber B (1998):
Increased diurnal plasma dehydroepiandrosterone in depressed patients. J Clin Endocrinol Metab 83:3130 –3133.
Holsboer F (1992): The hypothalamo-pituitary-adrenocortical
system. In: Paykel ES, editor. Handbook of Affective Disorders. Edinburgh, UK: Churchill, 267–287.
Khaw K-T, Tazuke S, Barret-Connor E (1988): Cigarette smoking and levels of adrenal androgens in postmenopausal
women. N Engl J Med 318:1705–1709.
Kimonides VG, Spillantini MG, Sofroniew MV, Fawcett JW,
Herbert J (1999): Dehydroepiandrosterone antagonizes the
neurotoxic effects of corticosterone and translocation of
stress-activated protein kinase 3 in hippocampal primary
cultures. Neuroscience 89:429 – 436.
Kirschbaum C, Hellhammer DH (1994): Salivary cortisol in
psychoendocrine research: Recent developments and applications. Psychoneuroendocrinology 19:313–333.
Krieger DT, Allen W, Rizzo R, Krieger HP (1971): Characterisation of the normal temporal pattern of plasma corticosteroid
levels. J Clin Endocrinol 32:266 –284.
Lewis DA, Smith RE (1983): Steroid-induced psychiatric syndromes. J Affect Disord 5:319 –332.
May M, Holmes E, Rogers W, Poth M (1991): Protection from
glucocorticoid induced thymic involution by dehydroepiandrosterone. Life Sci 46:1627–1631.
Morales AJ, Nolan JJ, Nelson JC, Yen SSC (1994): Effects of
replacement dose of dehydrepiandrosterone in men and women
of advancing age. J Clin Endocrinol Metab 78:1360 –1367.
Murphy BEP (1991a): Steroids and depression. J Steroid Mol
Biol 38:537–559.
Murphy BEP (1991b): Treatment of major depression with
steroid suppressive drugs. J Steroid Biochem Mol Biol 39:
239 –244.
BIOL PSYCHIATRY
2000;48:989 –995
995
O’Dwyer A-M, Lightman SL, Marks MN, Checkley S (1995):
Treatment of major depression with metyrapone and hydrocortisone. J Affect Disord 33:123–128.
Orentreich N, Brind JL, Vogelman JH, Andres R, Baldwin H
(1992): Long-term longitudinal measurements of plasma
dehydroepiandrosterone sulfate in normal men. J Clin Endocrinol Metab 75:1002–1004.
Osran H, Reist C, Chen CC, Lifrak ET, Chicz-DeMet A, Parker
LN (1993): Adrenal androgens and cortisol in major depression. Am J Psychiatry 150:806 – 809.
O’Toole SM, Sekula LK, Rubin RT (1997): Pituitary-adrenalcortical axis measures as predictors of sustained remission in
major depression. Biol Psychiatry 42:85– 89.
Overall JE, Gorham DR (1962): The brief psychiatric rating
scale. Psychol Rep 10:799 – 812.
Parker LN, Sack J, Fisher DA, Odell WD (1978): The adrenarche: Prolactin, gonadotropins, adrenal androgens and cortisol. J Clin Endocrinol Metab 46:396 – 401.
Paykel ES (1985): The clinical interview for depression: Development, reliability and validity. J Affect Disord 9:85–96.
Reus VI, Wolkowitz OM, Frederick S (1997): Antiglucocorticoid treatments in pschiatry. Psychoneuroendocrinology 22:
S121–S124.
Seckl J, Campbell JC, Edwards CRW, Christie JE, Whalley LJ,
Goodwin GM, Fink G (1990): Diurnal variation of plasma
corticosterone in depression. Psychoneuroendocrinology 15:
485– 488.
Shafagoj Y, Opuku J, Quereshi D, Regelson W, Kalimi M
(1992): Dehydroepiandrosterone prevents dexamethasoneinduced hypertension in rats. Am J Physiol 263:E210 –E213.
Spratt DI, Longcope C, Cox PM, Bigos ST, Wilbur-Welling C
(1993): Differential changes in serum concentrations of androgens and estrogens (in relation with cortisol in postmenopausal women with acute illness). J Clin Endocrinol Metab
76:1542–1547.
Starkman MN, Schteingart DE, Schork MA (1981): Depressed
mood and other psychiatric manifestatiuons of Cushing’s syndrome: Relationship to hormone levels. Psychosom Med 43:3–18.
Winterer J, Gwirtsman DR, George DT, Cutler B (1985):
Adrenocorticotropin-stimulated adrenal secretion in anorexia
nervosa: Impaired secretion at low weight with normalization
after recovery. J Clin Endocrinol Metab 61:693– 697.
Wolkowitz OM, Reus VI, Keebler A, Nelson N, Friedland M,
Brizendine L, Roberts E (1999): Double-blind treatment of
major depression with dehydroepiandrosterone. Am J Psychiatry 156:646 – 649.
Wolkowitz OM, Reus VI, Roberts E (1995): Antidepressant and
cognition enhancing effects of DHEA in major depression.
Ann N Y Acad Sci 774:337–339.
Wolkowitz OM, Reus VI, Roberts E, Manfredi F, Chan T, Raum
WJ, et al (1997): Dehydroepiandrosterone (DHEA): Treatment of depression. Biol Psychiatry 41:311–318.
Woolston JL, Gianfredi S, Gertner JM, Paugus JA, Mason JW
(1983): Salivary cortisol: A non-traumatic sampling technique for assaying cortisol dynamics J Am Acad Child
Psychiatry 22:474 – 476.
Zumoff B, Walsh BT, Katz JL, Levin J, Rosenfeld RS, Kreqm J,
Weiner H (1983): Subnormal plasma dehydroisoandrosterone to
cortisol ratio in anorexia nervosa: A second hormonal parameter
of ontogenic regression. J Clin Endocrinol Metab 56:668 – 672.
Major Depression in Adults
Albert Michael, Alison Jenaway, Eugene S. Paykel, and Joe Herbert
Background: The authors sought to examine whether
levels of dehydroepiandrosterone are abnormal in
depression.
Methods: Three groups of subjects aged 20 – 64 were
studied: 44 major depressives, 35 subjects with partially
or completely remitted depression, matched as far as
possible for age and drug treatment, and 41 normal
control subjects. Dehydroepiandrosterone and cortisol in
saliva were determined from specimens taken at 8:00 AM
and 8:00 PM on 4 days.
Results: The mean age of the three groups did not differ.
Dehydroepiandrosterone was lowered at 8:00 AM and 8:00
PM compared with control subjects. Values for the remitted
group were intermediate. Dehydroepiandrosterone levels
at 8:00 AM correlated negatively with severity of depression and were not related to drug treatment or smoking,
but decreased with age (as expected). Cortisol was elevated in depression in the evening. The molar cortisol/
dehydroepiandrosterone ratio also differentiated those
with depression from the control group.
Conclusions: Lowered dehydroepiandrosterone levels are
an additional state abnormality in adult depression. Adrenal steroid changes are thus not limited to cortisol.
Because dehydroepiandrosterone may antagonize some
effects of cortisol and may have mood improving properties, these findings may have significant implications for
the pathophysiology of depression. Biol Psychiatry
2000;48:989 –995 © 2000 Society of Biological
Psychiatry
Key Words: DHEA, cortisol, depression, adults, saliva,
rhythm
Introduction
T
he adrenal cortex secretes a wide spectrum of steroid
hormones, but nearly all attention on endocrine abnormalities associated with affective disorder has been
reserved for cortisol. Hypersecretion of cortisol occurs in
From the Departments of Psychiatry (AM, AJ, ESP) and Anatomy (JH), University
of Cambridge, Cambridge, United Kingdom.
Address reprint requests to Joe Herbert, University of Cambridge, Department of
Anatomy, Cambridge CB2 3DY, United Kingdom.
Received March 6, 2000; revised June 13, 2000; accepted June 13, 2000.
© 2000 Society of Biological Psychiatry
about 50% of cases, whether assessed by increased basal
levels, or by heightened resistance to the suppressive
feedback effects of dexamethasone (Butler and Besser
1968; Carroll and Mendels 1976; Murphy 1991a). Abnormal corticoid secretion may mark either the course of
depression or predict subsequent relapse following apparent recovery (Holsboer 1992). Increased levels of glucocorticoids, deriving either from endogenous sources (as
in Cushing’s syndrome) or exogenous ones (as during
steroid therapy) are known to induce abnormal affective
states, particularly depression (Dorn et al 1997; Lewis and
Smith 1983; Starkman et al 1981). Administration of
metyrapone, a compound that reduces cortisol levels, has
been reported to be beneficial in depression (O’Dwyer et
al 1995). In a recent study of depressed children and
adolescents, we have identified those with comorbid
dysthymia (i.e., “chronic depression”) as being most likely
to show evening hypercortisolaemia (Goodyer et al 1996;
Herbert et al 1996).
There are, however, other adrenal steroids that may be
equally as significant for understanding the psychopathology of depression. Dehydroepiandrosterone (DHEA) is
particularly interesting. The sulphated form of DHEA
(DHEAS) is the most abundant circulating steroid in man,
although information on its function has been elusive until
recently. Blood DHEA and DHEAS increase markedly
during early childhood and again during puberty (Parker et
al 1978). Levels peak in early adulthood and decrease
equally markedly thereafter (2–3% per year), so that by
age 60 they are about one third of those at age 20
(Orentreich et al 1992). Chronic physical ill-health or
more acute events, such as imminent surgery, result in an
elevated cortisol level but lowered DHEA and altered
cortisol/DHEA ratios (e.g., Gray et al 1991; Hedman et al
1992; Spratt et al 1993). One important feature of
DHEA(S) is thus that it can alter independently of cortisol
both during normal life and in illness. A second one is that
there is increasing evidence that DHEA can act as an
antagonist to glucocorticoids (Blauer et al 1991; May et al
1991; Shafagoj et al 1992).
The relevance of this to psychiatric disorder has been
shown recently. An extensive study of depression in
adolescents showed that about 30% have abnormally low
0006-3223/00/$20.00
PII S0006-3223(00)00955-0
990
A. Michael et al
BIOL PSYCHIATRY
2000;48:989 –995
DHEA levels; however, this occurred at a different time
(in the morning) than high (evening) cortisol (Goodyer et
al 1996; Herbert et al 1996). Levels of DHEAS were not
significantly altered by depression (Goodyer et al 1996).
Depressed adolescents with high cortisol/DHEA ratios and
experiencing recent adverse life events were unlikely to
have recovered 36 weeks later (Goodyer et al 1998).
Whether the new findings on DHEA in adolescents apply
to adults with major depressive disorder is not known. One
report of DHEA levels in adult depressives was carried out
on only nine depressed men compared with nine control
subjects, and showed that DHEA levels at 8:00 AM or 4:00
PM were no different in the two groups (Osran et al 1993);
however, the diurnal decrease in DHEA was less in the
depressed group than the control subjects. More recently,
the daily minima in DHEA levels have been reported to be
increased in depression (26 cases; Heuser et al 1998).
There is also a report of decreased cortisol/DHEA ratios in
anorexia nervosa (Devesa et al 1988; Zumoff et al 1983),
which return to normal upon restoration of normal weight
(Winterer et al 1985). “Well-being” was increased by
administration of DHEA to middle-aged men and women,
and plasma DHEA (but not DHEAS) correlated positively
with well-being in postmenopausal women (Cawood and
Bancroft 1996; Morales et al 1994). Recent preliminary
reports in depressed adults noted that ingestion of oral
DHEA over 4 weeks improved mental state and memory
function (Wolkowitz et al 1995, 1999). In patients with
Addison’s disease who lack DHEA(S), supplementation
improved well-being and reduced depression and anxiety
scores (Arlt et al 1999). These findings on DHEA suggest
that endocrine abnormalities in depression may extend to
steroids other than cortisol, and that the psychopathology
of this condition may be more complex than is presently
thought.
This article reports, for the first time, lowered (morning)
DHEA in major depression in adults compared with a
control group of age- and gender-matched subjects. Because concurrent drug treatment is a confounding factor in
such studies, this conclusion is reinforced by the inclusion
of an additional control group of subjects whose depression had undergone remission, most of whom were still
taking anti-depressant treatment.
Methods and Materials
Subjects
Three groups aged 20 – 64 years were studied.
MAJOR DEPRESSIVE DISORDER. There were 44 subjects
with depression (12 male, 32 female), aged 20 – 64 years,
recruited either as inpatients (36) or from an outpatient (8) clinic.
Subjects were accepted only if free from other disorders (see
below). All subjects met DSM-IV criteria (American Psychiatric
Association 1994) for unipolar major depressive disorder, single
or recurrent episodes (moderate or severe) without psychotic
symptoms. Smoking, alcohol intake, and current drug therapy
were recorded.
REMITTED DEPRESSION. Thirty-five subjects (14 male, 21
female) with a previous episode of depression but who had
recovered (i.e., no longer met DSM-IV criteria for major depressive disorder), were collected from follow-up outpatients. These
subjects were either in complete or partial remission. The
17-item Hamilton Rating Scale (HAM-D; Hamilton 1967) scales
were completed for each subject. Smoking, alcohol intake, and
current drug therapy were recorded
NORMAL CONTROL SUBJECTS. Forty-one normal control
subjects (14 male, 27 female) were recruited from hospital staff
and age matched with the depressed group. Saliva was collected
as for the other two groups. These subjects did not complete the
psychiatric rating scales, and were apparently free of current
illness (assessed from history). Smoking and alcohol intake were
recorded. In all groups, liver function tests and urine screening
were carried out if drug or alcohol abuse was suspected.
All subjects provided written informed consent following
complete description of the study, which was approved by the
Cambridge Local Ethics Committee.
Procedures
RATINGS. Subjects in the first two groups were assessed
using the Brief Psychiatric Rating Scale (BPRS), the HAM-D,
and the Clinical Interview for Depression (CID); the latter was
divided into depression and anxiety total scores (Overall and
Gorham 1962; Paykel 1985).
CORTISOL AND DHEA IN THE SALIVA. Four salivary
samples were collected at 8:00 AM, and another four at 8:00 PM,
usually on consecutive days. Subjects were instructed not to
clean their teeth within 2 hours of collecting saliva, to wash out
their mouths with clean water immediately before collection, and
to obtain about 5 mL of saliva by spitting into a plastic tube; no
salivation aids were used. Samples were taken individually into
prelabeled tubes supplied by the laboratory. They were stored in
the freezing compartments (220°C) of the subjects’ refrigerators
before being transferred to the laboratory.
ASSAYS. Salivary cortisol was measured by an enzymelinked immunosorbent assay technique, using unextracted duplicates from each sample. The final value at each time point for
each subject was the mean of the four samples collected at that
point. Variations: intra-assay, 4.5%; inter-assay, 4.4% (based on
pooled quality control subjects incorporated into each assay run).
Salivary DHEA was measured by radioimmunoassay, following
extraction with hexane ether. Variations: intra-assay, 5.8%;
inter-assay, 13.3%. Both assays have been extensively validated
(Goodyer et al 1996). Molar cortisol/DHEA ratios were calculated for each of the two time points.
DHEA in Major Depression
991
BIOL PSYCHIATRY
2000;48:989 –995
Table 1. Characteristics of the Sample
Depressed
Remitted
Control subjects
M (n)
F (n)
Age (years)
HAM-D
CID
12
14
14
32
21
27
43.2 6 11.17 (20 – 64)
41.4 6 9.25 (22–59)
40.6 6 9.62 (27– 62)
24.1 6 3.99 (17–33)
2.6 6 2.25 (0 – 8)
56.2 6 9.75
5.5 6 4.81
Values are mean 6 SD. Ranges in parentheses. M, male; F, female; HAM-D, Hamilton Rating Scale (Hamilton 1967); CID, Clinical Interview for Depression.
STATISTICAL ANALYSIS. Between-variable correlations
and between-group comparisons were made using two-way, univariate, or repeated-measures analyses of variance (ANOVAs; with
appropriate covariances) or nonparametric statistics (Kruskal–
Wallis [K-W] and Mann–Whitney [M-W] tests). All tests are
two tailed.
Results
Subjects
Table 1 shows the age and gender distribution of the three
samples. Neither age (K-W x2 5 0.57, p 5 ns) nor gender
(x2 5 1.31, p 5 ns) differed significantly between the
groups; however, smoking (defined as more than five
cigarettes per day) was significantly more frequent in the
depressed group (K-W x2 5 8.96, p , .01). There were 12
smokers in the depressed group (27.3%), compared with
seven in the remitted group (20.0%) and only two (4.9%)
in the control subjects group. The mean HAM-D scores in
the depressed group were no different in smokers and
nonsmokers [F(1,42) 5 2.6, p 5 ns]. Alcohol intake
(reported units per week) did not differ between the groups
(K-W x2 5 0.72, p 5 ns).
The HAM-D scores for the depressed group were
significantly greater than the remitted group (M-W U 5
7.5, p , .001; Table 1). All the depressed group had
HAM-D scores of greater than 17; the highest score in the
recovered group was 8. In the depressed group, there were
highly significant correlations between the HAM-D and
both CID depression (Spearman r 5 .63, p , .001) and
CID anxiety scores (r 5 .50, p , .001).
Drugs
Of the depressed group, 41 were taking psychotropic
drugs; three were not. Of the recovered group, 26 were
also taking drugs, and nine were drug free.
In the depressed group, 31 were taking a single drug,
nine were taking two, and one was taking three; 14 were
on tricyclic antidepressants, 18 on selective serotonin
reuptake inhibitors (SSRIs), and nine on monoamine
oxidase inhibitors (MAOIs). In addition to their primary
medication, 10 were also taking lithium (as an adjunct to
antidepressant treatment), one was taking benzodiazepines, and one sodium valproate (as an antidepressant). In
the recovered group, 10 were on TCAs, 14 were on SSRIs,
and one was on MAOIs. Six were also taking lithium.
Salivary Steroids
DHEA. At both 8:00 AM and 8:00 PM, DHEA was
negatively correlated with age (Spearman: AM r 5 2.36,
p , .0001; PM r 5 2.33, p , .0001). In male subjects,
DHEA (but not cortisol) was also higher (t tests; p , .02);
there was no significant difference in age.
A two-way ANOVA (group and smoking [yes/no] as
factors, age as covariate) showed that both morning and
evening levels of salivary DHEA were significantly different between the groups [DHEA AM F(6,98) 5 3.5, p ,
.02; PM F(6,98) 5 3.4, p , .02; Figure 1); there was no
main effect of smoking and no interaction between smoking and group. Further univariate ANOVAs showed that
morning DHEA was significantly lower in depressed than
control groups, whereas evening DHEA was lower in
depressed compared to remitted groups (Bonferroni test,
p , .05; Figure 1, top). A repeated-measures multivariate
ANOVA (MANOVA) on both morning and evening
DHEA (age as covariate) also showed significant group
effects [F(2,105) 5 3.55, p 5 .03].
Within the depressed group, there was a highly significant negative correlation between HAM-D scores and
morning DHEA (Spearman r 5 2.49, p , .01; Figure 2).
Similar associations were found with CID total depression
(r 5 2.38, p , .01) and total anxiety scores (r 5 2.37,
p , .05; Table 2). Evening DHEA correlated significantly
only with the CID anxiety score (Table 2).
In the depressed group, morning DHEA was not significantly lower in smokers than nonsmokers [F(1,38) 5 3.9,
p 5 .06]. There was no association between age and
HAM-D scores. In the remitted group, those currently on
medication had higher HAM-D scores than those who
were drug-free (M-W U 5 3.7, p , .001) and also higher
PM cortisol levels (U 5 2.3, p , .02), but no differences in
DHEA.
CORTISOL. There were no correlations between cortisol and age at either time point. Cortisol (8:00 AM,
8:00 PM) was significantly higher in the saliva of depressed
subjects than the other two groups at both time points
992
A. Michael et al
BIOL PSYCHIATRY
2000;48:989 –995
Figure 2. The relation between salivary dehydroepiandrosterone
(DHEA) levels at 8:00 AM and Hamilton Rating Scale (Hamilton
1967) scores in the depressed subjects.
ASSOCIATIONS BETWEEN SALIVARY CORTISOL AND
Overall, there were highly significant correlations between morning and evening levels of both cortisol
and DHEA. There were also significant associations between morning cortisol and DHEA. In the depressed
group, there were similar associations between morning
and evening levels of both cortisol and DHEA but not for
morning cortisol and DHEA (Table 3).
DHEA.
Figure 1. Depression is associated with lowered dehydroepiandrosterone (DHEA), raised salivary cortisol, and elevated
cortisol/DHEA ratios. The mean (6 SD) salivary levels of
DHEA, cortisol, and the molar cortisol/DHEA ratio at either
8:00 AM (gray bars) or 8:00 PM (patterned bars) in three groups
of subjects (depressed, recovered depressive, and normal control)
are shown. Significance levels given in text.
[AM F(2,117) 5 4.9, p , .01; PM F(2,117) 5 5.4, p , .01;
Figure 1, middle). Salivary cortisol in the depressed
subjects differed significantly from the control subjects
and remitted groups (AM p , .05, PM p , .05; Bonferroni
test). There were no significant correlations within the
depressed group between cortisol at either time point and
HAM-D, CID depression, or CID anxiety scores. A
repeated-measures MANOVA, taking morning and
evening levels of each hormone as dependent variables
and age and smoking as covariates, showed significant
group effects for cortisol [F(2,105) 5 5.96, p 5 .01).
MOLAR CORTISOL/DHEA RATIOS. Overall, there
were significant differences between the groups for both
morning (K-W x2 5 31.0, p , .0001) and evening (x2 5
28.1, p , .0001) molar cortisol/DHEA ratios (Figure 1,
bottom). Group comparisons showed that the depressed
group differed significantly from the recovered subjects
(AM x2 5 13.1, p 5 .0003; PM x2 5 20.9, p , .0001),
whereas the control subjects and the recovered groups
were indistinguishable (x2 5 2.2 and 0.2; ps 5 ns).
Table 2. Associations between Salivary Steroids and
Psychometric Scores in the Depressed Group (n 5 44)
HAM-D
DHEA (8:00 AM)
DHEA (8:00 PM)
Cortisol (8:00 AM)
Cortisol (8:00 PM)
a
20.48
20.08
20.20
0.13
CIDd
CIDa
a
20.37a
20.28
20.12
20.04
20.38
20.09
20.06
20.12
HAM-D, Hamilton Rating Scale (Hamilton 1967); CIDd, Clinical Interview for
Depression (Depression scale); CIDa, Clinical Interview for Depression (Anxiety
scale); DHEA, dehydroepiandrosterone.
a
p , .01.
DHEA in Major Depression
BIOL PSYCHIATRY
2000;48:989 –995
993
Table 3. Spearman Correlations between Salivary Cortisol and Dehydroepiandrosterone (DHEA) in All Subjects (n 5 107) or in
the Depressed Group Alone (n 5 44)
All subjects
DHEA (8:00 PM)
Cortisol (8:00 PM)
Cortisol (8:00 PM)
a
b
DHEA
(8:00 AM)
DHEA
(8:00 PM)
.72a
.24b
2.03
.01
.11
Depressed group only
Cortisol
(8:00 AM)
DHEA
(8:00 AM)
DHEA
(8:00 PM)
Cortisol
(8:00 AM)
.45a
.66a
.24
2.15
.13
2.04
.39a
p , .01.
p , .05.
Taking the 85th percentile morning cortisol/DHEA ratio
of the control group as an arbitrary (post hoc) cut-off,
82.5% of the depressed group had ratios that were equal to
or greater than this value (control 15%). The corresponding result for the ratio at 8:00 PM was 46% (Table 4).
Discussion
The principal new finding of this study is that mean levels
of DHEA in the saliva of depressed adults are lower than
those in normal control subjects. There are a number of
possible confounding variables that need to be considered
before this endocrine parameter can be added to the known
pathophysiology of this condition in adults.
The first, and most obvious, is the coincident effect of
the antidepressant treatment being taken by the depressed
subjects. We attempted to control for this by recruiting a
number of remitted depressives, the majority of whom
were still taking antidepressant treatment. It was noticeable that DHEA levels in this group as a whole fell
between that in the depressed subjects and the normal
control subjects. As might be expected, in the remitted
group the HAM-D scores of those who continued treatment were higher than those who were drug free; however,
separating those on antidepressant drugs failed to show
any significant effect on DHEA levels, though the numbers in each group are not large. The possible role of drug
treatment needs to be re-examined on a much larger
population than we were able to gather, but there is no
Table 4. The Number of Each Group Having Molar Salivary
Cortisol/Dehydroepiandrosterone (DHEA) Ratios in the Upper
85th Percentiles of the Values in Normal Control Subjects
Cortisol/DHEA ratio
Group
Depressed
Recovered
Control subjects
8:00 AM
(85th percentile 5 9.6)
8:00 PM
(85th percentile 5 5.7)
33/40 (82.5%)
11/34 (32.4%)
6/41 (14.2%)
18/39 (46.1%)
2/34 (5.9%)
6/40 (15.0%)
Overall significance: 8:00
.001).
AM
x2 5 40.4 (p , .01); 8:00
PM
x2 5 18.8 (p ,
evidence from our results that drug treatment itself was
responsible for the lower levels of DHEA recorded in
depressed subjects. The concomitant findings of elevated
salivary cortisol in the depressed (compared to the other
two groups) reinforces the conclusion that lowered morning DHEA is a state-dependent feature of at least a
proportion of depressed subjects. Raised cortisol (in both
blood and saliva) is a well-known association with this
condition and is not the result of coincident drug therapy
(Butler and Besser 1968; Gold et al 1988).
The association between DHEA and depression was
strongly supported by the finding of a negative correlation
between morning DHEA and HAM-D scores in the
depressed group. This results both supports the conclusion
that morning DHEA levels may be a state-related feature
of depression, and further argues against a simple drugrelated effect on DHEA. The fact that no such association
was found for cortisol suggests that there may be differences in the way that these two steroids relate to the
depressed state, or may reflect the limited range of severity
within the currently depressed sample.
Age is another possible confounding factor, because
DHEA decreases with age (Orentriech 1992); however, our
groups were well balanced for age and— had this been a
factor— both morning and evening DHEA levels would have
been expected to decline. Smoking was more prevalent in the
depressed subjects; but smoking has been found to raise
DHEA(S) levels (Khaw et al 1988), and so this effect, if
present, would act against the findings we report. Moreover,
statistical analysis failed to show an independent effect of
smoking on DHEA levels in our sample.
We measured levels of DHEA and cortisol twice daily for
4 days. Steroid levels may fluctuate for a number of reasons,
but the 4-day mean would have eliminated most of these
adventitious sources of variation. Salivary levels of both
DHEA and cortisol correlate well with those in the blood
(Goodyer et al 1996; Guazzo et al 1996; Woolston et al
1983); in both cases, levels in the saliva represent about 5%
of those in the blood. There is now a considerable literature
on the reliability of measuring steroids in the saliva, and the
way that such levels track those in the blood (Kirshbaum and
994
A. Michael et al
BIOL PSYCHIATRY
2000;48:989 –995
Hellhammer 1994). It seems highly likely, therefore, that
depressed subjects have lower levels of blood DHEA in the
morning, though the exact shape of the diurnal variation in
DHEA needs to be examined more closely. A major conclusion, however, is that changes in DHEA may not be parallel
to those in cortisol. The latter are more obvious in the evening
(confirming previous reports: e.g., Seckl et al 1990; review
by Murphy 1991a), though morning cortisol is also elevated.
Whether the same subjects show both changes in cortisol or
DHEA, or whether these can occur independently, could not
be reliably determined from our study on about 40 subjects
but should be become apparent when much larger populations are examined. For the same reasons, we cannot reliably
analyze the relation between DHEA and depression in male
and female subjects separately in this study, nor assess the
contribution of individual drugs to the endocrine levels
observed in the depressed or comparison groups.
Levels of cortisol and DHEA in the saliva (as in the
blood) show diurnal changes, with higher values at 8:00
AM than 8:00 PM (Krieger et al 1971); however, the
amplitude of the cortisol rhythm is much greater than
DHEA. Hence, the molar cortisol/DHEA ratio in control
subjects was 8.3 at 8:00 AM but only 2.7 at 8:00 PM (i.e.,
a three-fold change across 12 hours). This indicates that
tissues, including the brain, are exposed to very different
relative concentrations of cortisol and DHEA during the
day. In view of the accumulating evidence that DHEA can
act as an antagonist to corticoids (Blauer et al 1991;
May et al 1991), it was noteworthy that the molar
cortisol/DHEA ratios (particularly at 8:00 AM) clearly
separated depressed and normal subjects. Women, not
currently depressed, but with higher morning salivary
cortisol levels and subsequently experiencing an adverse
life event, have been shown to be at increased risk of
becoming depressed within the following 12 months
(Harris et al, in press); treatments designed to lower
cortisol have been found to decrease levels of depression
(Murphy 1991b; O’Dwyer et al 1995; Reus et al 1997).
Dehydroepiandrosterone antagonizes the damaging actions of corticosterone on cultured hippocampal neurons
(Kimonides et al 1999), and there is some evidence for its
being antidepressive (Wolkowitz et al 1995, 1997, 1999).
Cortisol/DHEA ratios have been found to predict the
course of depression in adolescents (Goodyer et al 1998),
which suggests that measuring both may give a more
complete picture of the steroidal changes associated with
depression. If cortisol does affect either onset or recovery
(O’Toole et al 1997), and if DHEA acts as an effective
antagonist to cortisol in this context, then both may need
to be taken into account.
Comorbid anxiety is common in depression. In this study,
relationships with morning salivary DHEA were similar to
depression as assessed by the CID, but evening DHEA was
associated only with anxiety. This would not necessarily
apply to pure anxiety disorder, in which depression and
anxiety are not so closely related as in this sample.
The findings in this article confirm and extend those
already reported on first-onset depression in 8- to 16-yearolds, on a population largely free of antidepressant treatment (Goodyer et al 1996; Herbert et al 1996). Depressed
subjects in this age group also show raised evening
salivary cortisol and lower morning DHEA, and the fact
that the results on the two groups are so similar increases
the reliance that can placed on either. The main conclusion
from these studies is that alterations in DHEA, a steroid
with known actions on the brain, may be an intrinsic
feature of depression in adults as well as in adolescents,
and this finding may give added insight into the psychopathology of this condition.
We thank Helen Shiers and Sarah Cleary for the assays. Conducted
within the MRC co-operative group in Brain, Behaviour and
Neuropsychiatry.
References
American Psychiatric Association (1994): Diagnostic and Statistical Manual of Mental Disorders, 4th ed. Washington, DC:
American Psychiatric Press.
Arlt W, Callies F, van Vlijmen JC, Koehler I, Reinicke M,
Bildingmaier M, et al (1999): Dehydroepiandrosterone replacement in women with adrenal insufficiency. N Engl
J Med 314:1013–1018.
Blauer KL, Poth M, Rogers WM, Bernton EW (1991): Dehydroepiandrosterone antagonises the suppressive effects of dexamethasone on lymphocytes. Endocrinology 129:3174 –3179.
Butler PWP, Besser GM (1968): Pituitary-adrenal function in
severe depressive illness. Lancet 1(7554):1234 –1236.
Carroll BJ, Mendels J (1976): Neuroendocrine regulation in
depression. In: Sachar EJ, editor. Hormones Behavior and
Psychopathology. New York: Raven.
Cawood EHH, Bancroft J (1996): Steroid hormones, the menopause, sexuality and well-being of women. Psychol Med
26:925–936.
Devesa J, Perez-Fernandez R, Bokser L, Casanueva FF (1988):
Adrenal androgen secretion and dopaminergic activity in
anorexia nervosa. Horm Metab Res 20:57– 60.
Dorn LD, Burgess ES, Friedman TC, Dubbert B, Gold PW,
Chrousoos GP (1997): The longitudinal course of psychopathology in Cushing’s syndrome after correction of hypercortisolism. J Clin Endocrinol Metab 82:912–919.
Gold PW, Goodwin FK, Chrousos GP (1988): Clinical and
biochemical manifestations of depression. Relation to neurobiology of stress. N Engl J Med 319:413– 420.
Goodyer IM, Herbert J, Altham PME (1998): Adrenal secretion
during major depression in 8 to 16 year olds III. Influence of
cortisol/DHEA ratio at presentation on subsequent rates of
disappointing life events and persistent major depression.
Psychol Med 28:265–273.
DHEA in Major Depression
Goodyer IM, Herbert J, Altham PME, Secher S, Shiers HM
(1996): Adrenal secretion during major depression in 8 to 16
year olds I. Altered diurnal rhythms in salivary cortisol and
dehydroepiandrosterone (DHEA): At presentation. Psychol
Med 26:245–256.
Gray A, Feldman HA, McKinley JB, Longcope C (1991): Age,
disease and changing sex hormone levels in middle-aged
men. Results of the Massachusetts male aging study. J Clin
Endocrinol Metab 73:1016 –1025.
Guazzo EP, Kirkpatrick PJ, Goodyer IM, Shiers HM, Herbert J
(1996): Cortisol, dehydroepiandrosterone (DHEA) and
DHEA sulfate in the cerebrospinal fluid of man: Relation to
blood levels and the effects of age. J Clin Endocrinol Metab
81:3951–3960.
Hamilton M (1967): Development of a rating scale for primary
depressive illness. Br J Soc Clin Psychol 6:278 –296.
Harris TO, Borsanyi S, Messari S, Stanford K, Cleary SE, Shiers
HM, et al (in press): Morning cortisol as a risk factor for
subsequent major depressive disorder in adult women. Br J
Psychiatry.
Hedman M, Nilsson E, de la Torre B (1992): Low blood and
synovial fluid levels of sulpho-conjugated steroids in rheumatoid arthritis. Clin Exp Rheumatol 10:25–30.
Herbert J, Goodyer IM, Altham PME, Secher S, Shiers HM
(1996): Adrenal steroid secretion and major depression in 8 to
16 year olds II. Influence of comorbidity at presentation.
Psychol Med 26:257–263.
Heuser I, Deuschle M, Luppa P, Schweiger H, Weber B (1998):
Increased diurnal plasma dehydroepiandrosterone in depressed patients. J Clin Endocrinol Metab 83:3130 –3133.
Holsboer F (1992): The hypothalamo-pituitary-adrenocortical
system. In: Paykel ES, editor. Handbook of Affective Disorders. Edinburgh, UK: Churchill, 267–287.
Khaw K-T, Tazuke S, Barret-Connor E (1988): Cigarette smoking and levels of adrenal androgens in postmenopausal
women. N Engl J Med 318:1705–1709.
Kimonides VG, Spillantini MG, Sofroniew MV, Fawcett JW,
Herbert J (1999): Dehydroepiandrosterone antagonizes the
neurotoxic effects of corticosterone and translocation of
stress-activated protein kinase 3 in hippocampal primary
cultures. Neuroscience 89:429 – 436.
Kirschbaum C, Hellhammer DH (1994): Salivary cortisol in
psychoendocrine research: Recent developments and applications. Psychoneuroendocrinology 19:313–333.
Krieger DT, Allen W, Rizzo R, Krieger HP (1971): Characterisation of the normal temporal pattern of plasma corticosteroid
levels. J Clin Endocrinol 32:266 –284.
Lewis DA, Smith RE (1983): Steroid-induced psychiatric syndromes. J Affect Disord 5:319 –332.
May M, Holmes E, Rogers W, Poth M (1991): Protection from
glucocorticoid induced thymic involution by dehydroepiandrosterone. Life Sci 46:1627–1631.
Morales AJ, Nolan JJ, Nelson JC, Yen SSC (1994): Effects of
replacement dose of dehydrepiandrosterone in men and women
of advancing age. J Clin Endocrinol Metab 78:1360 –1367.
Murphy BEP (1991a): Steroids and depression. J Steroid Mol
Biol 38:537–559.
Murphy BEP (1991b): Treatment of major depression with
steroid suppressive drugs. J Steroid Biochem Mol Biol 39:
239 –244.
BIOL PSYCHIATRY
2000;48:989 –995
995
O’Dwyer A-M, Lightman SL, Marks MN, Checkley S (1995):
Treatment of major depression with metyrapone and hydrocortisone. J Affect Disord 33:123–128.
Orentreich N, Brind JL, Vogelman JH, Andres R, Baldwin H
(1992): Long-term longitudinal measurements of plasma
dehydroepiandrosterone sulfate in normal men. J Clin Endocrinol Metab 75:1002–1004.
Osran H, Reist C, Chen CC, Lifrak ET, Chicz-DeMet A, Parker
LN (1993): Adrenal androgens and cortisol in major depression. Am J Psychiatry 150:806 – 809.
O’Toole SM, Sekula LK, Rubin RT (1997): Pituitary-adrenalcortical axis measures as predictors of sustained remission in
major depression. Biol Psychiatry 42:85– 89.
Overall JE, Gorham DR (1962): The brief psychiatric rating
scale. Psychol Rep 10:799 – 812.
Parker LN, Sack J, Fisher DA, Odell WD (1978): The adrenarche: Prolactin, gonadotropins, adrenal androgens and cortisol. J Clin Endocrinol Metab 46:396 – 401.
Paykel ES (1985): The clinical interview for depression: Development, reliability and validity. J Affect Disord 9:85–96.
Reus VI, Wolkowitz OM, Frederick S (1997): Antiglucocorticoid treatments in pschiatry. Psychoneuroendocrinology 22:
S121–S124.
Seckl J, Campbell JC, Edwards CRW, Christie JE, Whalley LJ,
Goodwin GM, Fink G (1990): Diurnal variation of plasma
corticosterone in depression. Psychoneuroendocrinology 15:
485– 488.
Shafagoj Y, Opuku J, Quereshi D, Regelson W, Kalimi M
(1992): Dehydroepiandrosterone prevents dexamethasoneinduced hypertension in rats. Am J Physiol 263:E210 –E213.
Spratt DI, Longcope C, Cox PM, Bigos ST, Wilbur-Welling C
(1993): Differential changes in serum concentrations of androgens and estrogens (in relation with cortisol in postmenopausal women with acute illness). J Clin Endocrinol Metab
76:1542–1547.
Starkman MN, Schteingart DE, Schork MA (1981): Depressed
mood and other psychiatric manifestatiuons of Cushing’s syndrome: Relationship to hormone levels. Psychosom Med 43:3–18.
Winterer J, Gwirtsman DR, George DT, Cutler B (1985):
Adrenocorticotropin-stimulated adrenal secretion in anorexia
nervosa: Impaired secretion at low weight with normalization
after recovery. J Clin Endocrinol Metab 61:693– 697.
Wolkowitz OM, Reus VI, Keebler A, Nelson N, Friedland M,
Brizendine L, Roberts E (1999): Double-blind treatment of
major depression with dehydroepiandrosterone. Am J Psychiatry 156:646 – 649.
Wolkowitz OM, Reus VI, Roberts E (1995): Antidepressant and
cognition enhancing effects of DHEA in major depression.
Ann N Y Acad Sci 774:337–339.
Wolkowitz OM, Reus VI, Roberts E, Manfredi F, Chan T, Raum
WJ, et al (1997): Dehydroepiandrosterone (DHEA): Treatment of depression. Biol Psychiatry 41:311–318.
Woolston JL, Gianfredi S, Gertner JM, Paugus JA, Mason JW
(1983): Salivary cortisol: A non-traumatic sampling technique for assaying cortisol dynamics J Am Acad Child
Psychiatry 22:474 – 476.
Zumoff B, Walsh BT, Katz JL, Levin J, Rosenfeld RS, Kreqm J,
Weiner H (1983): Subnormal plasma dehydroisoandrosterone to
cortisol ratio in anorexia nervosa: A second hormonal parameter
of ontogenic regression. J Clin Endocrinol Metab 56:668 – 672.