Pronounced and sustained central hypernoradrenergic
function in major depression with melancholic
features: Relation to hypercortisolism and
Ma-Li Wonga,b, Mitchel A. Klingb,d, Peter J. Munsone, Samuel Listwaka, Julio Licinioa, Paolo Proloa, Brian Karpa,
Ian E. McCutcheonf, Thomas D. Geracioti, Jr.g, Michael D. DeBellish, Kenner C. Ricei, David S. Goldsteinj,
Johannes D. Veldhuisk, George P. Chrousosl, Edward H. Oldfieldf, Samuel M. McCannm, and Philip W. Golda,c
aClinical Neuroendocrinology Branch, National Institute of Mental Health, andeAnalytical Biostatistics Section, Laboratory of Structural Biology, Division of
Computer Research and Technology, andiLaboratory of Medicinal Chemistry, National Institute of Diabetes and Digestive and Kidney Diseases, andjClinical
Neuroscience Branch andfSurgical Neurology Branch, National Institute of Neurological Disorders and Stroke, andlDevelopmental Endocrinology Branch,
National Institute of Child Health and Human Development, National Institutes of Health, Bethesda, MD 20892;dDepartment of Psychiatry, Baltimore
Veterans Affairs Medical Center and University of Maryland School of Medicine, Baltimore, MD 21201;gDepartment of Psychiatry, University of Cincinnati
Medical Center, Cincinnati, OH 45267;hDepartment of Child and Adolescent Psychiatry, University of Pittsburgh Medical Center, Pittsburgh, PA 15213;
kDepartment of Internal Medicine and National Science Foundation Center for Biological Timing, University of Virginia Health Sciences Center,
Charlottesville, VA 22908; andmPennington Biomedical Research Center, Louisiana State University, Baton Rouge, LA 70808-4124
Contributed by Samuel M. McCann, November 9, 1999
Both stress-system activation and melancholic depression are char-
changes in autonomic and neuroendocrine function. Because norepi-
nephrine (NE) and corticotropin-releasing hormone (CRH) can pro-
duce these physiological and behavioral changes, we measured the
cerebrospinal fluid (CSF) levels each hour for 30 consecutive hours in
controls and in patients with melancholic depression. Plasma adre-
nocorticotropic hormone (ACTH) and cortisol levels were obtained
every 30 min. Depressed patients had significantly higher CSF NE and
plasma cortisol levels that were increased around the clock. Diurnal
variations in CSF NE and plasma cortisol levels were virtually super-
imposable and positively correlated with each other in both patients
and controls. Despite their hypercortisolism, depressed patients had
normal levels of plasma ACTH and CSF CRH. However, plasma ACTH
and CSF CRH levels in depressed patients were inappropriately high,
considering the degree of their hypercortisolism. In contrast to the
relationship between these parameters. These data indicate that
persistent stress-system dysfunction in melancholic depression is
independent of the conscious stress of the disorder. These data also
suggest mutually reinforcing bidirectional links between a central
CRH pathways that each are driven and sustained by hypercorti-
solism. We postulate that ?-noradrenergic blockade, CRH antago-
alone or in combination, in the treatment of major depression with
Depression is the main cause of suicide: ?70% of all suicides are
attributed to untreated depression. Studies in the United States
suggest that, at any given time, ?2–3% of the population is
hospitalized or seriously impaired by affective illness. The World
Health Organization has declared major depression as the single
largest cause of morbidity for women and the leading cause of
The current standard diagnostic instrument for psychiatry, the
Diagnostic and Statistical Manual of Mental Disorders, 4th Ed.
(DSM-IV) (2), lists two subtypes of major depression, melan-
cholic and atypical. The features of melancholic depression
include insomnia (most often early morning awakening), loss of
appetite, weight loss, inappropriate guilt, and lack of pleasure
ajor depression is a complex disorder with an estimated
lifetime prevalence of 15% in women and 8% in men (1).
(anhedonia). The second major subtype is major depression with
atypical features, characterized in part by hypersomnia, hy-
perphagia, lethargy, and fatigue. The subclassification of depres-
sion provides direction for the appropriate choice of antidepres-
sant medication. Studies in identical twins show a significantly
higher concordance for either melancholic or atypical depres-
sion in a given set of twins (see ref. 50 for review).
and feeling, major depression with melancholic features belies the
term depression in that it is a state of hyperarousal and fear, often
stemming from a profound sense of personal unworthiness and
pessimism about the future (3, 4). Although patients with melan-
cholic depression are often clear about the deficiencies that they
attribute to themselves, they find it more difficult than usual to
concentrate or to fashion complex plans of action. Patients with
severe melancholia also seem to have preferential access to pain-
fully charged memories of past losses and failures (4). Inseparable
from their loss of both cognitive and emotional ranges is a loss of
pleasure in everyday activities. It is now clear that the hyperarousal
of melancholia is matched by indices of physiological hyperarousal,
including early morning awakening; sustained hypothalamic–
pituitary–adrenal (HPA) axis and sympathetic nervous system
activation (5, 6); and inhibition of appetite, libido, and endocrine
programs for growth and reproduction (3).
The clinical and biochemical manifestations of melancholic
depression that often persist for months closely resemble those
that occur reflexively during acutely stressful or threatening
situations (3, 4). Both melancholic depression and acute stress
are characterized by arousal, sustained focus on the threatening
stimulus, fear-related behaviors, and relatively stereotyped
states of cognition and affect (3, 4). These symptoms are
associated with activation of the HPA axis and the sympathetic
nervous system, and with inhibition of neurovegetative functions
that might be counterproductive during a life-threatening situ-
Abbreviations: NE, norepinephrine; CRH, corticotropin-releasing hormone; CSF, cerebro-
spinal fluid; ACTH, adrenocorticotropic hormone; HPA, hypothalamic–pituitary–adrenal;
LC, locus ceruleus.
bM.-L.W. and M.A.K. contributed equally to this work.
cTo whom reprint requests should be addressed at: Clinical Neuroendocrinology Branch,
National Institute of Mental Health, National Institutes of Health, Building 10?2D46, 10
Center Drive, MSC 1284, Bethesda, MD 20892-1284. E-mail: email@example.com.
The publication costs of this article were defrayed in part by page charge payment. This
article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C.
§1734 solely to indicate this fact.
January 4, 2000 ?
vol. 97 ?
no. 1 ?
ation (e.g., feeding, sleeping, growth, sexual behavior, reproduc-
tion, etc.) (3, 4).
The close clinical and biochemical resemblance between the
symptom complex of melancholic depression and the phenome-
nology of the stress response led us to postulate that melancholic
depression represents activation of the principal effectors of the
stress response [the locus-ceruleus–norepinephrine (LC-NE) and
the corticotropin-releasing hormone (CRH) systems] (3, 4). Cen-
trally, NE acts as a major alarm-producing neurotransmitter in the
brain and inhibits feeding, grooming, and sleeping. In addition, NE
plays an important modulatory role by activating the amygdala (4,
7) and, by so doing, enhances the eventual coding of emotionally
laden memories of disturbing or stressful events. In contrast, NE
to shift mood from one state to another, to promote novel over
well-rehearsed behaviors, and to inhibit HPA axis and brainstem
HPA axis and, during chronic stress, mediates direct sympathetic
hypercortisolism (9). Thus, activation of noradrenergic neurons
increases fear-related behaviors, promotes relatively stereotyped
behaviors and affects, and may both activate hypothalamic and
brainstem neuroendocrine and arousal centers and release them
from cortical inhibition (4).
CRH was first isolated as a principal hypothalamic releasing
factor for the HPA axis. CRH has also been shown to play a
significant role in the acquisition and storage of classically condi-
tioned fear responses in the amygdala. Descending hypothalamic
promotes activation of the HPA axis, but also sets into motion a
sequence of behavioral and physiological events that closely resem-
ble those in melancholic depression and stress (3, 4, 10–14).
The purpose of the present study was to evaluate centrally
directed NE and CRH secretion in medication-free patients with
melancholic depression by measuring the NE and CRH levels in
lumbar cerebrospinal fluid (CSF) sampled continuously for 30-h
by an indwelling lumbar catheter. We also measured the 30-h
pattern of plasma adrenocorticotropic hormone (ACTH) and
cortisol secretion in the same subjects. Cortisol is not only
hypersecreted in melancholic depression but also significantly
influences behavior and affects the LC-NE and CRH systems, as
well as the activities of the reproductive and growth hormone
axes. By evaluating CSF NE and CRH levels at night while
patients slept, as well as during the daytime, we also hoped to
determine whether the activation of stress mediators in major
depression represented a potential contributor to its clinical and
biochemical manifestations, or whether the stress mediators
were merely artifacts of the conscious distress of the illness.
Specifically, this study asked four major questions: (i) Is either
CSF NE or CRH hypersecreted in melancholic depression? (ii) If
so, do abnormalities persist while patients sleep? (iii) Are abnor-
(iv) Does hypercortisolism seem to influence or to reinforce
potential abnormalities in CSF NE or CSF CRH secretion?
Clinical Research Protocol. Under an Institutional Review Board-
approved clinical protocol (National Institute of Mental Health
depressive episodes with melancholic features [age (mean ? SD)
40.9 ? 2.7 yr; range, 28.2–53.7 yr], who were medication-free for at
37.7 ? 2.2 yr; range, 27.4–53.0 yr] participated in this study.
Patients’ discontinuation of medication did not take place for the
purpose of participation in this study. Patients were either medi-
cation-free when they were referred to us or taken off medications
that were not clinically effective. All depressed patients were
hospitalized on the clinical research units of the National Institute
of Mental Health. The presence of a current major depressive
of Mental Disorders, 3rd Ed., Revised (DSM-III-R) (15) and the
World Health Organization’s Research Diagnostic Criteria (RDC)
(16). A minimum Hamilton Rating Scale for Depression score (17)
of 15 was required for inclusion in this study. Healthy volunteers
were free of ongoing physical illness, determined by history and by
physical examination, and showed no evidence of major psychiatric
illness, determined by both clinical and structured interviews.
Female subjects were studied during days 1–7 of the menstrual
cycle. Before the actual sampling of CSF and plasma, all subjects
were adapted to the hospital setting for at least 1 night.
We conducted continuous CSF and plasma sampling in
healthy volunteers and in patients with major depression as
previously described (18). All patients and volunteers were on a
low-monoamine diet for at least 3 days before CSF sampling.
CSF sampling began at 09:00–10:00 a.m. with the introduction
of a standard 20-gauge epidural catheter through an 18-gauge
Touhy spinal needle into the subarachnoid space at the L3?4 or
L4?5 level. The catheter was attached to a miniroller pump and
6 ml?h was exfused at a constant rate that represents 25–33% of
the normal CSF production rate (19). Sampling lasted for 30 h,
at which time the catheter was removed; during CSF sampling,
the subject remained flat in bed, but was free to lie in any
position (i.e., prone, supine, or lateral).
Six 1-ml aliquots initially were placed in the refrigerator for 30
min and then placed on dry ice. The 1-ml aliquot taken between
30 and 40 min was assayed for CSF CRH and the 1-ml aliquot
taken from 40 to 50 min was measured for CSF NE. This
procedure previously has been demonstrated to be a safe and
reliable method for sampling CSF continuously (18, 20). Previ-
ous studies have shown human CRH to be stable in CSF under
catheter in the arm, on the half-hour and at half-hour intervals
for 30 h, for measurement of plasma ACTH and cortisol.
Subjects remained on bed rest during the study and at least until
the morning after the withdrawal of the catheters.
Assays. CSF CRH levels were measured in Sep-Pak C18 column-
extracted samples in a pool of two consecutive 1-ml aliquots per
hour, representing a total of 20 min of continuous sampling
beginning at 30 min past the hour. Extracted samples were recon-
stituted in buffer at a concentration 3-fold higher than the original
sample and assayed in duplicate. The CSF CRH concentration was
by RIA with prior extraction by using a standard method (22).
Plasma cortisol was measured by direct RIA using a standard kit
(Diagnostic Products, Los Angeles). CSF NE levels were measured
by reversed-phase high-performance liquid chromatography with
electrochemical detection in the aliquot consecutive to those used
for CRH levels, according to a standard method (23). All samples
from the same subject were analyzed together, and assay runs
included both patients and control subjects. Interassay coefficients
of variation were ?15% for all assays.
Statistical Analysis. For each patient, each series of measurements
was averaged, and the average values were compared. Analyses
were performed by using the entire series of hormone measure-
ments or by cropping data from the initial hour and the last 2 h
of each series. Cropping was done to minimize the effects of
hormonal changes related to the stress of beginning and termi-
nating the CSF-sampling procedure in our analysis. Differences
between control and depressed groups were examined by using
an unpaired, two-tailed Student’s t test, using STATISTICA soft-
ware (Statsoft, Tulsa, OK) on a Macintosh computer.
www.pnas.orgWong et al.
Correlation Analysis. Crosscorrelation analyses with and without
detrending were performed on series of measurements of the
four substances as previously described (24). Crosscorrelation
analysis was computed between the series of measurements for
two informational substances at various time lags covering the
30-h period of study. If the release of a substance Y is regulated
by a substance X (X, releasing substance; Y, effector substance),
then one might expect the concentration time series of substance
Y to follow (lag) quantitatively in time the concentration time
series of substance X. Crosscorrelation was computed after
lagging (shifting) the concentration time series of substance X
rk the coefficient of correlation between two informational-
substance time series at a lag time k (any lag time from 0 to
?1440 min) for one person, then the mean rk of all persons in
each group was considered significant when it exceeded 0 by
more than 2 standard errors (SE). The SE was calculated from
the individual rk values for all persons in that group at a lag time
k. We mitigated the effect of baseline shifts and of the circadian
the impact of the inherent circadian component by detrending
analysis (using a five-point moving average). As the relationship
the short-term, circadian-independent correlation between
plasma ACTH and cortisol around lag 0 was used as positive
control for our correlation analysis.
Analysis of 24-h Rhythm. Each series of measurements (CSF NE,
CSF CRH, plasma cortisol, and plasma ACTH) was analyzed for
circadian rhythmicity. The presence of sinusoidally varying
diurnal trends was tested by cosinor analysis, which represents a
linear reduction of sinusoidal regression (25, 26). Cosinor anal-
ysis was done using CHRONOLAB for the Macintosh computer
(Universidade de Vigo, Spain, Bioengineering and Chronobiol-
ogy Laboratory, http:??www.tsc.uvigo.es?BIO?). Each series of
measurements was individually input into CHRONOLAB, and
parameters of the sinusoidal regression such as MESOR (mid-
line-estimating statistic of rhythm or rhythm-adjusted mean),
acrophase [a measure of time, the lag from a defined reference
time point (midnight of the first day of measurement in our
analysis) of the crest time in the cosine-curve-fitted curve to the
data], and amplitude (half the extent of rhythmic change, or the
difference between the maximum concentration and the ME-
SOR of the fitted curve) were obtained for a 24-h period.
Rhythm detection is sought by testing the null hypothesis of zero
amplitude with an F test by the CHRONOLAB program.
Fig. 1 displays the average of the curves of plasma cortisol,
plasma ACTH, CSF NE, and CSF CRH concentrations (mean ?
SE) for control and major depression (melancholic type) groups
of subjects. Unless otherwise specified, the results presented are
from the analysis of the cropped series of measurements. Mean
30-h plasma cortisol levels were significantly elevated in patients
with melancholic depression when compared with controls
[11.6 ? 1.2 (mean ? SE) and 8.7 ? 0.6 ?g?dl, respectively; P ?
0.02] (Fig. 1A). Despite significant increases in basal cortisol
levels, plasma ACTH levels were similar in patients with de-
pression and in controls (10.8 ? 1.6 and 10.2 ? 1.1 pg?ml,
respectively; P ? not significant) (Fig. 1B). Compared with the
ratio in controls, the plasma cortisol-to-ACTH ratio was signif-
icantly higher in patients with melancholic depression, indicating
depression, melancholic type. Curves result from the averaged measurement per time point across a group of subjects by using the cropped hormonal series.
The shaded area represents data recorded with the lights off (23:00–07:00 h). In the right corner Insets under each pair of curves, the bar graphs represent the
average of the mean value for each series of hormonal measurements (mean ? SE).*, P ? 0.02.
Wong et al. PNAS ?
January 4, 2000 ?
vol. 97 ?
no. 1 ?
that patients with major depression showed a relatively greater
plasma cortisol response to a given simultaneous level of plasma
ACTH (1,572,000 ? 164,000 and 4,071,000 ? 778,000, respec-
tively; P ? 0.001). Mean 30-h CSF NE levels were significantly
increased in depressed patients when compared with controls
(137.4 ? 12.7 and 102.3 ? 7.0 pg?ml, respectively; P ? 0.02) (Fig.
1C). CSF NE levels were elevated both at night (including during
sleep) and during the day to the same proportional extent. Mean
30-h CSF CRH levels, despite significant increases in basal
cortisol levels, were also similar in patients with depression and
in controls (48.9 ? 9.8 and 57.1 ? 4.1 pg?ml, respectively; P ?
not significant) (Fig. 1D). Compared with the ratio in controls,
the plasma cortisol to CSF CRH ratio was significantly higher in
patients with melancholic depression (1,558,000 ? 102,000 and
3,032,000 ? 606,000, respectively; P ? 0.01).
Analysis of the 24-h pattern in plasma cortisol, plasma ACTH,
and CSF CRH showed significant diurnal variations. Polar chro-
nograms of CSF NE and plasma cortisol are shown in Fig. 2. CSF
NE showed a 24-h pattern very similar to that of plasma cortisol in
both patients and controls; the highest levels for both substances
occurred in the morning. It is important to note that the detection
of rhythm in CSF NE was virtually as consistent as that for plasma
cortisol. Plasma cortisol levels showed the expected diurnal varia-
tion in both patients and controls, with the mean peak occurring in
between 1:00 and 2:00 a.m. (Fig. 1A) (27).
The results of our crosscorrelation analyses are summarized in
Fig. 3. Crosscorrelation for CSF CRH and plasma cortisol
lag 0 (P ? 0.05); the correlation between CSF CRH and plasma
cortisol was not found to be significant in depressed patients
(Fig. 3B). Crosscorrelation analysis revealed a significant posi-
tive correlation at lag 0 between CSF NE and plasma cortisol in
both patients and controls (Fig. 3 C and D; P ? 0.05). Even after
detrending the hormonal series, a significant correlation (P ?
0.05) between CSF NE and plasma cortisol was observed, and
this correlation was nearly as robust as that seen for plasma
ACTH and plasma cortisol (Fig. 3 E and F).
This study shows that around the clock, patients with melan-
cholic depression have elevated levels of CSF NE and plasma
cortisol, but not of CSF CRH or plasma ACTH. The CSF NE
period to detect a rhythm by a confidence region not overlapping the pole by using the cropped hormonal series. Note that in both polar graphs all the acrophases
are within a 12-h time span, and most of them are within a 6-h time span. These characteristics indicate that the rhythm of CSF NE seems to be as tightly regulated as
plasma cortisol is by the circadian clock. The circadian representation for plasma cortisol is shown as a reference of our analysis and population.
Polar representations of the biologic rhythm’s percentage amplitude?MESOR (radius) and acrophase (angle) relations obtained by a single cosinor analysis
and ACTH and cortisol (F), by using the cropped hormonal series. Note that
is lost in patients. A positive correlation exists between NE and cortisol. The
all the subjects in a certain group at the time k and indicates the limits of
significance (P ? 0.05). Crosscorrelation between ACTH and cortisol (F) is
shown as a reference of our analysis and population.
Graphs depicting the crosscorrelation analyses of the mean coeffi-
www.pnas.org Wong et al.
and plasma cortisol changes are tightly linked, but the normal
inverse relation between CSF CRH and plasma cortisol is lost.
This pattern of a central hypernoradrenergic state and adreno-
cortical secretion not only may characterize melancholic depres-
sion but also produce it.
The syndrome of melancholic depression consists of intense
anxiety and arousal, constriction of affect and cognition, acti-
vation of the HPA axis, and inhibition of neurovegetative
functions, including the programs for growth and reproduction
(3, 4). Both NE and CRH each can contribute to these clinical
and biochemical manifestations. Previous studies regarding lev-
els of CSF NE and CSF CRH have reported normal, reduced,
and increased levels (5, 21, 28–30). The differences between
these prior results and those of the current study could reflect
greater resolution with continuous sampling, acute stress-related
changes during a single-time-point study, or differences in the
populations of depressed patients under study.
CSF NE. The continuous elevation in CSF NE levels that persists
during sleep indicates that the activation of the central norad-
renergic system in depression is not related to the conscious
distress of the disorder.
NE serves globally as an emergency or alarm system that leads
to decreases in neurovegetative functions, such as eating and
sleeping, and contributes to accompanying increases in auto-
nomic and neuroendocrine responses to stress, including HPA
axis activation. NE also activates the amygdala, a principal brain
locus for fear-related behaviors. In addition, NE release during
stress inhibits the medial prefrontal cortex (8) and, by so doing,
could interfere with two of its key functions (i.e., the shifting of
mood from one state to the other based on internal and external
cues and the generation of novel, complex behaviors) (4). Thus,
hyperfunctioning of the central noradrenergic system could
contribute to the full spectrum of the clinical and biochemical
manifestations of melancholic depression (3, 4). Moreover, by
enhancing the long-term storage of aversively charged emotional
memories in sites such as the hippocampus and striatum, NE
could also contribute to subsequent episodes of depression by
facilitating the recall of previous ones (4).
Bilateral lesioning of the LC in the rat causes decreases in NE
barrier to peripheral levels of NE has been described (33), and
despite a positive correlation between plasma and CSF levels of
NE (33), CSF levels are not markedly affected by acute alter-
ations in the plasma levels. Thus, the majority of CSF NE is likely
to derive from brainstem noradrenergic nuclei.
Crosscorrelation analysis showed a significant positive corre-
lation between CSF NE and plasma cortisol in depressed
patients and in controls. In controls, the crosscorrelations be-
between plasma ACTH and plasma cortisol levels. Thus, the
positive relationship between CSF NE and plasma cortisol is
virtually as great as one of the most classic relationships in
to the strong positive correlation between CSF NE and plasma
cortisol concentrations. First, NE activates hypothalamic CRH
neurons directly (34). Second, NE activates the amygdala, which
in turn activates the HPA axis (7). Third, NE inhibits the medial
prefrontal cortex, which normally restrains the HPA axis (8).
Fourth, it now clearly is established that the sympathetic nervous
system exerts extrapituitary modulation of adrenocortical cor-
ticosteroid secretion, especially during chronic stress (8, 35–39).
In addition to noradrenergic mediation of hypercortisolism,
hypercortisolism per se activates brainstem noradrenergic neurons
by CRH. Thus, glucocorticoids accentuate a descending hypotha-
lamic CRH pathway that activates brainstem noradrenergic nuclei.
Moreover, in a postmortem study of brains taken from depressed
patients who had committed suicide, the significant increase in
hypothalamic neurons expressing CRH was greatest in those neu-
rons sending descending projections to brainstem noradrenergic
nuclei (40). We postulate that hyperactivity of this pathway plays a
role in the pronounced hyperactivity of central noradrenergic
neurons and serves to help maintain this condition for prolonged
periods of time. We further postulate that either ?-noradrenergic
antagonists or glucocorticoid antagonists will prove to exert ther-
apeutic effects in patients with major depression associated with
significant hypercortisolism, including not only patients with mel-
ancholic depression but also those with psychotic depression.
CSF CRH. Several populations of CRH-containing neurons have
been identified that are of potential relevance to the symptom
complex of melancholic depression. The largest of these is a
glucocorticoid-suppressible hypothalamic system that sends CRH
nerve terminals to the median eminence for activation of the
pituitary–adrenal axis. As we noted previously, a separate, smaller
descending hypothalamic pathway that responds positively to glu-
cocorticoids conveys CRH for the activation of brainstem norad-
renergic neurons. We have previously shown that glucocorticoids
also significantly increase the levels of CRH mRNA in an extra-
hypothalamic CRH pathway that is located in the amygdala that
plays an important role in fear-related behaviors (41).
Glucocorticoids not only activate the descending hypotha-
lamic and amygdala CRH systems, but also accentuate CRH-
induced fear-related behaviors and can produce them in their
own right. These findings further accentuate the potentially
important role of glucocorticoid excess in contributing to and
sustaining the pathological hyperarousal of melancholic depres-
sion and support the use of glucocorticoid antagonists in major
depression with melancholic and?or psychotic features.
Despite several lines of data suggesting that CRH is involved in
the hypercortisolism of depression, we found normal 30-h CSF
CRH levels in hypercortisolemic patients with melancholic depres-
sion, analogous to our previous finding based on a single data point
HPA axis is not abnormal. In a previous study of patients with
Cushing’s disease (a rare form of hypercortisolism caused by a
pituitary rather than a central nervous system defect), we found
that in the absence of a central nervous system defect, CSF CRH
levels are suppressed by hypercortisolism (42). Indeed, in a com-
parison of CSF CRH levels in groups of patients with major
depression or Cushing’s disease, individually matched for the
severity of hypercortisolism, we found that depressed patients
showed significantly higher CSF CRH levels than did Cushing’s
disease patients. Thus, CSF CRH levels in patients with melan-
cholic depression, though quantitatively ‘‘normal,’’ are inappropri-
ately elevated for their degree of hypercortisolism.
We found a significant negative correlation between CSF CRH
and plasma cortisol levels in controls that was similar to that which
we had previously reported (20). However, this significant negative
correlation was not found in patients with depression. This loss of
the normal relationship between CSF CRH and plasma cortisol
could mean either a disruption in the integrity of glucocorticoid
negative feedback on hypothalamic CRH-containing neurons that
project to the median eminence or an overriding of glucocorticoid-
negative feedback by excitatory stimuli, such as NE from the LC or
other brainstem loci. In addition, disproportionate glucocorticoid-
mediated activation of descending hypothalamic and amygdala
ACTH pathways in depression could also contribute to the loss of
the negative correlation between plasma cortisol and CSF CRH
seen in healthy volunteers.
In clinical studies exploring hormonal responses to CRH, we
advanced data indicating that the CRH system in major depression
was activated as compared with controls (43). Holsboer et al. (44)
subsequently replicated this finding. Our findings in experimental
Wong et al. PNAS ?
January 4, 2000 ?
vol. 97 ?
no. 1 ?
animals (45, 46) and human subjects (47) that antidepressants Download full-text
consistently down-regulate the CRH system are compatible with
this formulation. Thus, even though overall CSF CRH levels in
is likely to play an important role in the symptom complex of
melancholic depression, including hypercortisolism, inhibition of
vegetative function, and behavioral arousal and anxiety.
As we noted, hypercortisolism fails to suppress adequately CSF
CRH levels in depressed patients. Moreover, compared with con-
trols, CRH function in depression is likely to be relatively more
active in areas in which CRH neurons are activated by glucocor-
ticoids. These areas (the amygdala and descending parvocellular
paraventricular nucleus projections) promote fear-related behav-
that CRH and glucocorticoid excess could exert additive or syner-
gistic effects on fear-related behaviors (48, 49). Glucocorticoids
serve to heighten behavioral responses to CRH, such as acoustic
startle (49), and can themselves mimic the effects of CRH on
acoustic-startle responses (49).
The most striking findings of this study were the pronounced
activation of the central noradrenergic system in patients with
melancholic depression and the data indicating that cortisol is
likely to play several roles in producing and sustaining the
depressive syndrome. Several implications for future therapeutic
intervention in major depression with melancholic features or in
psychotic depression emerge from these findings. First, the
potential efficacy of an ?-noradrenergic blocker should be
investigated; ?-noradrenergic blockers given to patients with
pheochromocytoma are known to reduce the anxiety associated
with this disorder. Second, given the positive correlation be-
tween plasma cortisol and NE levels and the capacity of glu-
cocorticoids to also potentiate the activity of the descending
hypothalamic and amygdala CRH pathways, further trials of
glucocorticoid antagonism should be conducted in patients with
major depression with melancholic features or with psychotic
depression, including newly available, relatively pure glucocor-
ticoid antagonists. Third, the application of the new nonpeptide,
orally absorbed CRH type 1 receptor antagonists that cross the
blood–brain barrier could mitigate the hypercortisolism of de-
pression and interfere with the CRH-mediated transduction of
fear-related behaviors and the activation of brainstem norad-
This work was partially supported by the National Alliance for Research
on Schizophrenia and Depression (M.-L.W.), the University of Virginia
National Center for Research Resources Grant M01 RR-00847 (J.D.V.),
and the National Institutes of Health Grant MH 51853 (S.M.M.).
1. Kessler, R. C., McGonagle, K. A., Zhao, S., Nelson, C. B., Hughes, M.,
Eshleman, S., Wittchen, H. U. & Kendler, K. S. (1994) Arch. Gen. Psychiatry
2. American Psychiatric Association (1994) Diagnostic and Statistical Manual of
Mental Disorders (Am. Psychiatr. Assoc., Washington, DC), 4th Ed.
3. Gold, P. W., Goodwin, F. K. & Chrousos, G. P. (1988) N. Engl. J. Med. 319,
4. Gold, P. W. & Chrousos, G. P. (1999) Proc. Assoc. Am. Phys. 111, 22–34.
5. Post, R. M. & Ballenger, J. C. (1984) Neurobiology of Mood Disorders (Williams
and Wilkins, Baltimore).
6. Carroll, B. J., Curtis, G. C., Davies, B. M., Mendels, J. & Sugerman, A. A.
(1976) Psychol. Med. 6, 43–50.
7. Goldstein, L. E., Rasmusson, A. M., Bunney, B. S. & Roth, R. H. (1996)
J. Neurosci. 16, 4787–4798.
8. Arnsten, A. F., Mathew, R., Ubriani, R., Taylor, J. R. & Li, B. M. (1999) Biol.
Psychiatry 45, 26–31.
9. Bornstein, S. R. & Chrousos, G. P. (1999) J. Clin. Endocrinol. Metab. 84,
10. Brown, M., Fisher, L., Spiess, J., Rivier, C., Rivier, J. & Vale, W. (1982)
Endocrinology 111, 928–931.
11. Sutton, R. E., Koob, G. F., LeMoal, M., Rivier, J. & Vale, W. (1982) Nature
(London) 297, 331–333.
12. Britton, D. R., Koob, G. F., Rivier, J. & Vale, W. (1982) Life Sci. 31, 363–367.
13. Sirinathsinghji, D. J. S., Rees, L. H., Rivier, J. & Vale, W. (1983) Nature
(London) 305, 232–235.
14. Rivier, C. & Vale, W. (1984) Endocrinology 114, 914–921.
15. American Psychiatric Association (1987) Diagnostic and Statistical Manual of
Mental Disorders (Am. Psychiatr. Assoc., Washington, DC), 3rd Ed., Revised.
16. World Health Organization (1993) The ICD-10 Classification of Mental and
Behavioral Disorders: Diagnostic Criteria for Research (W.H.O., Geneva).
17. Hamilton, M. (1960) J. Neurol. Neurosurg. Psychiatry 23, 56–62.
18. Bruce, J. N. & Oldfield, E. H. (1988) Neurosurgery 23, 788–790.
19. Wood, J. H. (1980) in Neurobiology of Cerebrospinal Fluid, ed. Wood, J. H.
(Plenum, New York), pp. 1–16.
20. Kling, M. A., DeBellis, M. D., O’Rourke, D. K., Listwak, S. J., Geracioti, T. D.,
Jr., McCutcheon, I. E., Kalogeras, K. T., Oldfield, E. H. & Gold, P. W. (1994)
J. Clin. Endocrinol. Metab. 79, 233–239.
21. Roy, A., Pickar, D., Paul, S., Doran, A., Chrousos, G. P. & Gold, P. W. (1987)
Am. J. Psychiat. 144, 641–645.
22. Gold, P. W., Loriaux, D. L., Roy, A., Kling, M. A., Calabrese, J. R., Kellner,
C. H., Nieman, L. K., Post, R. M., Pickar, D. & Gallucci, W. (1986) N. Engl.
J. Med. 314, 1329–1335.
23. Kaler, S. G., Goldstein, D. S., Holmes, C., Salerno, J. A. & Gahl, W. A. (1993)
Ann. Neurol. 33, 171–175.
24. Magiakou, M. A., Mastorakos, G., Rabin, D., Margioris, A. N., Dubbert, B.,
Calogero, A. E., Tsigos, C., Munson, P. J. & Chrousos, G. P. (1996) Clin.
Endocrinol. (Oxford) 44, 419–428.
25. Bingham, C., Arbogast, B., Guillaume, G. C., Lee, J. K. & Halberg, F. (1982)
Chronobiologia 9, 397–439.
26. Fernandez, J. R. & Hermida, R. C. (1998) Chronobiol. Int. 15, 191–204.
27. Krieger, D. T. (1978) Med. Clin. N. Am. 62, 251–259.
28. Potter, W. Z. & Manji, H. K. (1994) Clin. Chem. 40, 279–287.
29. Nemeroff, C. B., Wilderlov, E. & Bisette, G. (1984) Science 226, 1342–1344.
30. Geracioti, T. D., Jr., Loosen, P. T. & Orth, D. N. (1997) Biol. Psychiatry 42,
31. Anselmo, F. J., Franci, C. R., Krulich, L., Antunes-Rodrigues, J. & McCann,
S. M. (1997) Brain Res. 767, 289–296.
32. Fornai, F., Alessandri, M. G., Torracca, M. T., Bassi, L. & Corsini, G. U. (1997)
J. Pharmacol. Exp. Ther. 283, 100–107.
33. Ziegler, M. G., Wood, J. H., Lake, R. & Kopin, I. J. (1977) Am. J. Psychiatry
34. Calogero, A. E., Gallucci, W. T., Chrousos, G. P. & Gold, P. W. (1988) J. Clin.
Invest. 82, 839–846.
35. Goldstein, D. S., Zimlichman, R., Kelly, G. D., Stull, R., Bacher, J. D. & Keiser,
H. R. (1987) J. Neurochem. 49, 1484–1490.
36. Holst, J. J., Ehrhart, B. M., Messell, T., Poulsen, S. S. & Harling, H. (1991)
Am. J. Physiol. 261, E31–E40.
37. Ehrhart, B. M., Bornstein, S. R., Guse, B. H., Stromeyer, H. G., Rasmussen,
T. N., Scherbaum, W. A., Adler, G. & Holst, J. J. (1994) Neuroendocrinology
38. Ehrhart, B. M., Hinson, J. P., Bornstein, S. R., Scherbaum, W. A. & Vinson,
G. P. (1998) Endocr. Rev. 19, 101–143.
39. Dallman, M. F., Engeland, W. C. & McBride, M. H. (1977) Ann. N.Y. Acad.
Sci. 297, 373–392.
40. Raadsheer, F. C., Hoogendijk, W. J. G., Stam, F. C., Tilders, F. J. H. & Swaab,
D. F. (1994) Neuroendocrinology 60, 436–444.
41. Makino, S., Gold, P. W. & Schulkin, J. (1994) Brain Res. 64, 105–112.
42. Kling, M. A., Roy, A., Doran, A. R., Calabrese, J. R., Rubinow, D. R.,
Whitfield, H. J., Jr., May, C., Post, R. M., Chrousos, G. P. & Gold, P. W. (1991)
J. Clin. Endocrinol. Metab. 72, 260–271.
43. Gold, P. W., Chrousos, G. P., Kellner, C. H., Post, R. M., Roy, A., Avgerinos,
P. C., Schulte, H. M., Oldfield, E. H. & Loriaux, D. L. (1981) Am. J. Psychiatry
44. Holsboer, F., Gerken, A. & Steiger, A. (1984) N. Engl. J. Med. 311, 1127 (lett.).
45. Brady, L. S., Whitfield, H. J., Fox., R. J., Gold, P. W. & Herkenham, M. (1988)
J. Clin. Invest. 87, 831–837.
46. Brady, L. S., Gold, P. W., Herkenham, M., Lynn, A. & Whitfield, H. (1992)
Brain Res. 572, 117–125.
47. Michelson, D., Stratakis, C., Hill, L., Reynolds, J., Galliven, E., Chousos, G. P.
& Gold, P. W. (1996) N. Engl. J. Med. 335, 1176–1181.
48. Corodimas, K. P., LeDoux, J. E., Gold, P. W. & Schulkin, J. (1994) Ann. N.Y.
Acad. Sci. 746, 392–393.
49. Lee, Y., Schulkin, J. & Davis, M. (1994) Brain Res. 666, 93–98.
R. C. (1996) Arch. Gen. Psychiatry 53, 391–399.
www.pnas.org Wong et al.