-
Molecular Psychiatry (2002) 7, 254–275 2002 Nature Publishing
Group All rights reserved 1359-4184/02 $25.00
www.nature.com/mp
GRAND ROUNDS
Organization of the stress system and its dysregulationin
melancholic and atypical depression: high vs lowCRH/NE statesPW
Gold1 and GP Chrousos2
1Clinical Neuroendocrinology Branch, Intramural Research
Program, NIMH, NIH, Bethesda, MD, USA; 2Pediatric andReproductive
Endocrinology Branch, Intramural Research Program, NICHD, NIH,
Bethesda, MD, USA
Stress precipitates depression and alters its natural history.
Major depression and the stressresponse share similar phenomena,
mediators and circuitries. Thus, many of the features ofmajor
depression potentially reflect dysregulations of the stress
response. The stressresponse itself consists of alterations in
levels of anxiety, a loss of cognitive and affectiveflexibility,
activation of the hypothalamic-pituitary-adrenal (HPA) axis and
autonomic nervoussystem, and inhibition of vegetative processes
that are likely to impede survival during alife-threatening
situation (eg sleep, sexual activity, and endocrine programs for
growth andreproduction). Because depression is a heterogeneous
illness, we studied two diagnostic sub-types, melancholic and
atypical depression. In melancholia, the stress response seems
hyper-active, and patients are anxious, dread the future, lose
responsiveness to the environment,have insomnia, lose their
appetite, and a diurnal variation with depression at its worst in
themorning. They also have an activated CRH system and may have
diminished activities ofthe growth hormone and reproductive axes.
Patients with atypical depression present with asyndrome that seems
the antithesis of melancholia. They are lethargic, fatigued,
hyperphagic,hypersomnic, reactive to the environment, and show
diurnal variation of depression that isat its best in the morning.
In contrast to melancholia, we have advanced several lines
ofevidence of a down-regulated hypothalamic-pituitary adrenal axis
and CRH deficiency in atypi-cal depression, and our data show us
that these are of central origin. Given the diversity ofeffects
exerted by CRH and cortisol, the differences in melancholic and
atypical depressionsuggest that studies of depression should
examine each subtype separately. In the presentpaper, we shall
first review the mediators and circuitries of the stress system to
lay thegroundwork for placing in context physiologic and structural
alterations in depression thatmay occur as part of stress system
dysfunction.Molecular Psychiatry (2002) 7, 254–275. DOI:
10.1038/sj/mp/4001032
Keywords: atypical depression; corticotropin releasing hormone
(CRH); melancholic depression;norepinephrine (NE); stress
Stress precipitates major depression and influences
itsincidence, severity and course.1,2 The stress responseand major
depression share many features because ofsimilar brain circuitries
and mediators (reviewed in 3–5).Each is associated with a
diminution of cognitive andaffective flexibility, alterations in
arousal, and pertur-bations in neuroendocrine and autonomic
function(reviewed in 5). Because major depression is a
hetero-geneous disorder, we focus here on two subtypes,
mel-ancholic and atypical depression. Our data and thoseof others
indicate that the principal arousal producingmediators of the
stress response, such as the corticotro-pin releasing hormone (CRH)
system, are hyperactivein melancholic depression.6 Not
surprisingly, melan-
Correspondence: PW Gold, MD, NIH Clinical Center, Room
2D-46-1284, Bethesda, MD 20892-1284, USA. E-mail:
philgold�codon.nih.govReceived 16 October 2001; accepted 17 October
2001
cholia is associated with anxiety, dread of the future,insomnia,
loss of appetite, and hypothalamic-pituitary-adrenal activation.3
Atypical depression seems to bethe reverse of melancholia, in that
is characterized bylethargy, fatigue, hypersomnia and hyperphagia.5
Wehave advanced several lines of evidence of a downreg-ulated
hypothalamic-pituitary adrenal axis in atypicaldepression, and our
data show us that it is of centralorigin.3,7In the present paper,
we shall first review themediators and circuitries of the stress
system to lay thegroundwork for placing in context physiological
andstructural alterations in depression that may occur aspart of
stress system dysfunction. We shall then pro-vide an overview of
critical stress mediators and struc-tures that we postulate lay
significant roles in thepathophysiologies of melancholic and
atypicaldepression.
We would like to emphasize at the outset that themajority of
patients with major depression present
-
High vs low CRH/NE statesPW Gold and GP Chrousos
255with a mixture of cognitive, affective, and
physiologicfeatures that do not fully conform to the
classificationsof melancholic and atypical depression. Moreover,
notall cases of melancholic and atypical depressionresemble one
another. We also do not suggest here thatabnormalities in the
stress system are primary factorsin the pathophysiology of
depression. Rather, we feelthat stress mediators, as likely
downstream elementsin depressive pathophysiology, transduce many of
theclinical and physiological alterations we are currentlyable to
decipher. Therefore, further elucidation ofstress system
dysfunction in patients with majordepression could provide improved
targets for system-atic research, diagnosis, treatment, and
prevention.
Major depression
Major depression is a heritable disorder that
affectsapproximately 8% of men and 15% of women.1 Forover 75% of
patients, major depression is a recurrent,lifetime illness,
characterized by repeated remissionsand exacerbations.8 Over 50% of
patients who recoverfrom a first depressive episode will have a
secondwithin 6 months unless they are given
maintenanceantidepressant treatment.2 For those who never
receivetreatment, as many as 15% will succumb to suicide.9
Depression not only causes great mental anguish butalso intrudes
upon fundamental biological processesthat regulate sleep, appetite,
metabolic activity, auto-nomic function, and neuroendocrine
regulation(reviewed in 4,8). These disturbances are likely to
con-tribute to premature coronary artery disease,10–12
premature osteoporosis,13 and the doubling of mor-tality in
patients with major depression at any ageindependent of suicide,
smoking, or significantphysical illness.10–12 In taking into
account the naturalhistory, mental suffering, and medical
morbidityassociated with major depression, the World
HealthOrganization ranked this disorder as one of the leadingcauses
of disability worldwide.14
It is now clear that a history of childhood traumaincreases the
risk for depression in adulthood. More-over, environmental stress
or internal conflict duringadult life can precipitate major
depression and influ-ence its course and severity.15 Thus,
susceptibility tomajor depression includes burdens of internal
conflictand external stressors, as well as the sum, intensity,and
accessibility of emotional memories that recallpast abandonment,
failure, or abuse.
Classification of depressionThe Diagnostic and Statistical
Manual of Mental Dis-orders IV (DSM-IV) is the principal instrument
for psy-chiatric diagnoses in the United States.16 The DSM-IVlists
two major divisions of depressive subtypes basedon the
phenomenology of recurrent affective episodesrather than the
clinical phenotype of the depression.Bipolar affective illness is
associated with recurrentbouts of both major depression and mania
or hypo-mania, affects 1–2% of the population, and occurs withequal
frequency in men and women. Major depression
Molecular Psychiatry
is characterized by recurrent bouts of major depressionalone,
occurs in approximately 12% of the population,and presents with a
2:1 female preponderance. Bothdisorders are heritable and involve
multiple genes.17,18
Epidemiological studies suggest overlap in geneticand
environmental factors predisposing to bipolar orunipolar disorder.
As an example, the offspring ofbipolar parents have a higher
incidence of both bipolarillness and unipolar illness than the
general popu-lation. First degree relatives of patients with
unipolarillness also have a smaller increase in the incidence
ofmajor depression.19
The DSM-IV lists two distinct clinical depressivesyndromes that
seem the antithesis of one another,melancholic and atypical
depression This distinctionis based on the pattern of psychological
and neuroveg-etative symptoms,20 is independent of the
unipolar-bipolar distinction, and provides direction for
theappropriate choice of antidepressant medication.21
Melancholic depression belies the term depressionin that it is a
state of pathological hyperarousal. Intenseanxiety is often focused
on the self and takes the formof feelings of worthlessness and
recollections of pasttransgressions, failures, and helplessness. As
a cor-ollary, melancholics are beset by dread about futureprospects
for so deficient a self. It matters little thattheir self
assessments and emotional memories are dis-cordant with the facts
of their lives. Rather, their feel-ings of personal deficiency
color and pervade thoughtand affect (reviewed in 5).
Patients with melancholic depression also manifestevidence of
physiological hyperarousal such as hyper-cortisolism, suppression
of the growth hormone andreproductive axes, insomnia (most often
early morningawakening), and loss of appetite. Another
consistentfeature of melancholia is a diurnal variation in
theseverity of depressed mood, which is greatest early inthe
morning (reviewed in 5).
Although both atypical and melancholic depressionare associated
with dysphoria and anhedonia, atypicaldepression is in many ways
the antithesis of melan-cholia. Atypical depression is associated
with a dis-turbing sense of disconnectedness and
emptiness,punctuated by brief emotional reactions to external
cir-cumstances. In contrast to melancholics, who seem tohave ready
access to negatively charged memories,patients with atypical
depression often seem walled offfrom themselves. They may complain
of a cognitiveand mental weariness and avoid others, often with
thesense that contact would be too demanding, tiring, andpoorly
received. Neurovegetative symptoms in atypicaldepression are the
reverse of those in melancholia andconsist of lethargy, fatigue,
excessive sleepiness,increased food intake, weight gain, and
depressivesymptoms that worsen as the day progresses.22
Only 25–30% of patients with major depressionpresent with pure
melancholic features while another15–30% present with pure atypical
features. Thosewith melancholic or atypical features show a
muchmore severe course of illness than those with
mixedneurovegetative features.20 Recent data from identical
-
High vs low CRH/NE statesPW Gold and GP Chrousos
256
Molecular Psychiatry
twin and family studies indicate that melancholic andatypical
features are each heritable entities.23 However,only a few studies
of depression have stratified patientson the basis of clinical
subtype.
We will first describe the stress system prior to ourdiscussion
of depression.
Phenomenology of the stress response
The acute response to danger consists of a relativelystereotyped
series of physiological and behavioral pro-grams that promote
survival during threatening situ-ations. Physiological changes
include increases inheart rate and blood pressure, shifts in blood
flow tothe brain and to the stressed body site, and breakdownof
tissue in the mobilization of fuel. In addition, thereis inhibition
of a repertoire of neurovegetative func-tions whose execution would
be likely to diminish thelikelihood of surviving a life threatening
situation (egfeeding, sleep, sexual behavior, and the endocrine
pro-grams for growth and reproduction) (reviewed in 6,24).
Fear-related behaviors predominate during stressfulsituations
and are crucial for survival during emerg-encies. For this reason,
an extensive circuitry for gener-ating and modulating fear has
evolved.25 Depending onthe context and constitutional factors (eg
gender, stresssystem set point), fear leads to either
defensivebehavior that protects from harm or stimulates a strug-gle
for survival. Speed and simplicity are essential,leading to a rapid
deployment of simple, well-rehearsed behavioral and cognitive
responses. At thesame time, there is an inhibition of more
complex,novel, or untested responses that require considerabletime
to assemble.26
Consistency is also essential for surviving stressfulsituations
and is most apparent in the inhibition ofmood shifts from one state
to the another. Thus, affectis often confined to a distressed,
fearful mode. Asnoted, cognitive and behavioral repertoires are
alsorelatively stereotyped during stressful situations. Dur-ing the
acute crisis, the mesolimbic dopaminergicreward system is
stimulated to help maintain morale.27
Different stressors activate different components of thestress
system. The response to a physiological stressorlike hypoxia may
require the involvement of onlyhypothalamus and brainstem;
structures such as theamygdala and prefrontal cortex must be
recruited toeffectively respond to somewhat more
complexenvironmental danger.28
The neurobiology of the stress response
The core stress systemFor the purpose of this review, the core
stress systemconsists of the (CRH) system and the locus
ceruleus-norepinephrine (LC-NE) systems and their
peripheralmediators, NE and cortisol. These systems play keyroles
in physiological responses to stressful situations,promoting
arousal, essential for identifying a givensituation as important,
as well as for maintaining thelimbic system and the cortex in
states that most favor
survival during stressful situations. The core compo-nent also
serves as a homeostat for the overall stresssystem, utilizing
inputs from many areas in the brainand periphery in contributing to
the modulation of theintensity and duration of the stress
response.
The CRH systemCRH was first isolated as the principal
hypothalamichormone that releases corticotropin (ACTH), which
inturn activates adrenocorticosteroid secretion. Over theyears, a
series of painstaking studies in rodents hasestablished roles for
CRH in the stress response otherthan that of HPA axis regulation.
These include acti-vation of the locus ceruleus, the sympathetic
nervoussystem and the adrenal medulla, as well as inhibitionof a
variety of neurovegetative functions such as foodintake, sexual
activity, and the endocrine programs forgrowth and reproduction
(reviewed in 3,6,24). Extrahy-pothalamic CRH-containing neurons in
the amygdala,though technically outside of the core stress
system,also play a key role in the stress response by
activatingfear-related behaviors while inhibiting
exploration(reviewed in 3,6,24). Taken together, CRH in the rat
par-ticipates in virtually the entire cascade of the physiol-ogic
and behavioral alterations occurring in responseto stressors.
CRH-mediated glucocorticoid secretion has an abun-dance of
adaptive and adverse effects. Acute glucocort-icoid secretion
during stress serves several roles,including enhancement of
cardiovascular function andmobilization of fuel. Cortisol (along
with CRH) also sig-nificantly contributes to the inhibition of
programs forgrowth and reproduction via inhibition of the
growthhormone and gonadal axes, as well as to feedbackrestraint
upon an activated immune system.
For the most part, the adaptive advantages conferredby cortisol
secretion during stress are limited to itsacute rather than chronic
release. Chronic cortisolexcess is almost always deleterious and
includesexcessive fear, insulin resistance/visceral fat depo-sition
and their many pro-atherogenic sequelae,osteopenia/osteoporosis,
sarcopenia, inhibition of Thelper-1 directed cellular immunity, and
chronicsuppression of the mesolimbic dopaminergic
rewardsystem.24,29 Glucocorticoid receptors are widelydistributed
in brain. Acutely, activation of glucocort-icoid receptors located
in the prefrontal cortex, hippo-campus, amygdala, and the
hypothalamus, inhibit theHPA axis. McEwen, Sapolsky, and their
colleaguesfound that chronic activation of glucocorticoid
recep-tors located in the hippocampous can damage hippo-campal
neurons containing glucocorticoid receptors,potentially leading to
more severe hypercortisolism.30
Not all glucocorticoid receptors transduce inhibitoryeffects. We
found that activation of glucocorticoidreceptors located in the
central nucleus of the amyg-dala and the bed nucleus of the stria
terminalisincrease rather than decrease CRH mRNA. Glucocort-icoids
also raise CRH mRNA levels located in a distinctpopulation of PVN
neurons that send descending ter-minals to brainstem noradrenergic
neurons.31
-
High vs low CRH/NE statesPW Gold and GP Chrousos
257Corticotropin releasing hormone and its receptorsin brain
In addition to the PVN CRH pathway to the medianeminence, as
noted, a separate pathway emanatingfrom a distinct population of
PVN CRH neuronsdescends for activation of brainstem
noradrenergicneurons.32 An intrahypothalamic pathway for
trans-synaptic release of CRH33 was shown to inhibit thegrowth
hormone34 and reproductive axes35 (in concertwith cortisol) and to
inhibit feeding36 and sexualbehavior.37 An extrahypothalamic CRH
system in theamygdala was subsequently shown to play a key rolein
classical fear conditioning.38,39 Thus, CRH wasshown to participate
in the behavioral, neuroendo-crine, neurovegetative, and autonomic
components ofthe stress response. The CRH receptor type 1 (CRHR
1)is widely distributed in brain to transduce its effectsduring
stress and other situations.40
While the CRHR-1 knockout mice show decreasedanxiety,41 CRH type
2 receptor knockout mice showaccentuation of arousal and anxiety,
suggesting thatthis receptor may counter-regulate the
anxiogeniceffects mediated by type 1 receptor activation.42,43
Type2 receptors also mediate diminished food intake. ACRH binding
protein parallels CRH receptors in brainand functions as an
endogenous CRH antagonist bycomplexing with CRH; its antagonism
promotes arou-sal and diminishes feeding.44
We have recently shown in rhesus macaques that theoral
administration of a non-peptide CRH type 1receptor antagonist
(antalarmin) that penetrates theblood–brain barrier significantly
inhibited stress-induced anxiety-like responses while promoting
explo-ration (Figure 1). We also found that antalarmin
sig-nificantly inhibited increases in plasma ACTH, NE,epinephrine
and cortisol (Figure 1). These data indi-cate that CRH plays a
tonic role in the comprehensivemodulation of the stress response
not only in rodents,but also in primates.45 In rodent studies, we
found thatantalarmin not only blocked the expression of
con-ditioned fear, but also its development and consoli-dation
(Figure 2).46 These data, if applicable tohumans, suggest that a
CRH antagonist could be help-ful after an acute traumatic event or
in preventing theadverse secondary CNS changes that occur
duringchronic stress (Figure 2). We have also found that
anta-larmin significantly reduces stress ulcer in the rat.47 Inthe
light of the important processes transduced by thetype 1 CRH, many
laboratories, including ours, areattempting to synthesize a small
CRH antagonist withoptimal lipophilicity that would be suitable as
a PETligand.48 In an effort to develop such a ligand, in
col-laboration with Dr Kenner Rice, we have synthesizedover 60
analogs of antalarmin.49
The LC-NE system
The LC-NE system resides in the mid-pons and con-tains the
highest concentration of noradrenergic cellbodies in the brain. A
single LC neuron can have as
Molecular Psychiatry
many as 100 000 nerve terminals and can innervatecells in
several different portions of the brain. At nor-mal firing rates,
the LC is thought to increase the signalto noise ratio at disparate
sites in brain by specificallyenhancing responses to either
excitatory or inhibitorystimuli. At faster LC rates, the general
enhancement ofsignal to noise ratio decreases and the LC becomes
thebrain’s alarm system. In addition, activation of the
LCcontributes to sympathetic nervous system and HPAaxis
stimulation. At the same time, LC activationinhibits the
parasympathetic nervous system as wellas neurovegetative functions
such as feeding and sleep(reviewed in 50).
During stress, the LC enhances the role of the amyg-dala and
other structures involved in the encoding ofaversively charged
memories. Thus, the LC not onlypromotes survival during an acute
crisis, but helps inpreparing for subsequent dangers as well.
Arnsten etal have recently found another important role of theLC-NE
during stress, namely in the inhibition of theprefrontal cortex,
thereby favoring rapid instinctualresponses over more complex ones
in the service ofsurviving acute life-threatening situations (see
below51).
Taken together, at fast firing rates, the LC, like theCRH
system, plays a role in promoting arousal,inhibiting several
vegetative functions, and biasingtowards a loss of affective and
cognitive flexibility.
The central role of the amygdala as a fear generatorBecause fear
is essential for surviving serious threats,the stress system must
be capable of producing theexperience of being afraid. The amygdala
is a key struc-ture that transforms experiences into feeling.25
Toaccomplish this task, the amygdala provides workingmemory with
information about whether something isgood or bad and, along with
the core stress system,activates disparate arousal centers to
maintain focusupon the current danger.25 Like the core stress
system,the amygdala evolved relatively early compared tohigher
cortical centers.
The amygdala is responsible for acquiring and stor-ing classic
fear conditioned responses that can beimmediately mobilized even
though they remain out-side of conscious awareness.26 Because the
amygdalacannot store complex, explicit, aversively chargedemotional
memories, it relays them to areas such as thehippocampus and
striatum for retrieval during sub-sequent emergencies.52
Like the core stress system, the amygdala is thoughtto inhibit
key functions of the prefrontal cortex. Theamygdala also stimulates
hypothalamic CRH releaseand brainstem autonomic centers resulting
inincreased HPA and LC activity. These core stress sys-tem
mediators not only confer adaptive physiologicaladvantages, but are
also thought to encode visceralresponses that provide bodily
feedback as part of theoverall affective experience. In addition,
both norepi-nephrine and cortisol significantly enhance the
relayand encoding of aversively charged emotional memor-ies from
the amygdala to elsewhere in brain.26 Taken
-
High vs low CRH/NE statesPW Gold and GP Chrousos
258
Molecular Psychiatry
Figure 1 Effects of antalarmin on behavior and neuroendocrine
responses in rhesus macaques. (a) pharmacokinetics in plasma;(b)
effects on anxiety; (c) effects on exploration; (d) effects on CSF
CRH; (e) effects on plasma norepinephrine; (f) antalarmineffects on
dose response curve for arousal vs CSF CRH. At a given level of CSF
CRH, arousal levels are lower for antalarmintreated macaques. From
Habib et al. Proc Natl Acad Sci 2000; 97: 6079–6084.
together, there are multiple feed forward loops amongthe
amygdala, the hypothalamus, and brainstem norad-renergic neurons.
Thus, the stress system contains theelements for a sustained and
powerful stress response.
The prefrontal cortexThe prefrontal cortex accounts for
approximately one-third of human brain volume. In many respects the
pre-frontal cortex exerts cognitive, behavioral, affective,and
physiological responses that are the virtual antith-esis of those
set into motion during stress. At the sametime, the prefrontal
cortex and the stress system inhibiteach other’s activity.53–55
The dorsolateral prefrontal cortex plays key roles incomplex,
time consuming planning and problem solv-ing (reviewed in 53–55) in
part, by sequentially schedul-ing complex tasks by switching
focused attentionbetween tasks.56 The dorsal prefrontal cortex also
pro-vides a perspective on whether a given task is proceed-ing
satisfactorily.56 In contrast, successful responses todanger depend
upon simplicity and speed, generallyantithetical to complex
planning and problem solving.Indeed, Arnsten et al have shown that
an activatedLC-NE inhibits many key functions of the
prefrontalcortex.51,57 Therefore, optimal functioning of the
dorso-
lateral prefrontal cortex requires a relatively quiescentstress
system.
The progression from dorsolateral to ventromedialprefrontal
cortex is associated with a progressive shiftfrom
attention/cognitive matters to the modulation ofaffect,
neuroendocrine regulation, and autonomicactivity. The ventral
prefrontal cortex (especially theorbital cortex) promotes
extinction of responses tostimuli that are not reinforced,
including the extinctionof conditioned fear responses encoded in
the amyg-dala.51,57–60 Humans with lesions of the orbital
cortex,like endangered individuals, seem driven and disin-clined or
unable to shift intellectual strategies andaffect on the basis of
changing demands.51,57–60
Flexibility in affect and cognition requires not only
anactivated prefrontal cortex, but also an inhibited stresssystem
(and vice versa).
Another component of the ventral prefrontal cortex,the subgenual
prefrontal cortex, participates indetermining whether a given
situation is likely to resultin punishment or reward and in the
adjustment ofaffect based on changes in the environment.61
Thiscapacity contrasts to the unconditional maintenance offear
during stress, even if there is preliminary indi-cation that the
danger is about to subside. Similarly, it
-
High vs low CRH/NE statesPW Gold and GP Chrousos
259
Figure 2 Antalarmin inhibits both the (a) development and(b)
expression of conditioned fear. Antalarmin given bothbefore and
after conditioning produces a greater effect thaneither. From Deak
et al. Endocrinology 1999; 140: 79–86.
is adaptive during stress to expect the worst. An effec-tive
prediction about whether punishment or reward ison the way requires
not only an intact prefrontal cor-tex, but also lack of
interference by an activatedstress system.
The ventral and prefrontal and subgenual prefrontalcortex also
exert cortical inhibition upon the HPA axisand the sympathetic
nervous system. Humans withlesions that include the anterior
cingulate gyrus and/orsubgenual prefrontal cortex show exaggerated
auto-nomic and endocrine responses even when in appar-ently
non-stressful situations.56 In the rat, bilaterallesioning of the
infralimbic region disinhibits the HPAaxis.62 Further study in the
rat revealed that lesioningof the left infralimbic cortex
disinhibited the core stresssystem, while lesioning of the right
resulted in subnor-mal activity of the HPA axis and the LC-NE
systems.62
Thus, in the rat, the left prefrontal cortex inhibits theright.
We have found evidence of the lateralization ofneuroendocrine
function as well.63 It is of potentialinterest that the most
replicated neuroimaging resultsin depression are those showing
abnormalities in theleft amygdala and left subgenual prefrontal
cortex. Fig-ure 3 provides a schematic diagram of the
variousreciprocal positive reinforcing loops among stress sys-tem
components, hypothesizing defects on the left formelancholia (see
below).
Mayberg has advanced data that nicely illustrate
bi-directionally the reciprocity between cortical and sub-cortical
sites and suggest that the reciprocity betweenthese sites is
immediate and obligatory.64 The capacityof the stress system to
inhibit prefrontal cortex functionis but one of several mechanisms
that insure a vigorousand sustained stress response. Figure 3
schematically
Molecular Psychiatry
illustrates the many potential positive feedback loopsthat
emerge during activation of the stress response inwhich each
component activates the other.
Role of the stress system in the pathophysiologyof melancholic
and atypical depression
We shall next discuss the role of the stress system inthe
clinical, biochemical, and structural alterationsdocumented in
patients with major depression. Weshall begin with core system
abnormalities and thenreview abnormalities of amygdala and
prefrontal cortexas well. We focus not only on the role of stress
systemdysregulation in the classic symptoms of depression,but also
on the long-term medical consequences ofthis disorder.
The hypercortisolism of depression is one of themost frequent
findings in biological psychiatry, thoughmany papers cited normal
cortisol levels as well. It isgenerally accepted that hypothalamic
CRH is elevatedin depression. We were the first to report a CRH
abnor-mality in patients with depression. In our first
originalarticle, we showed that hypercortisolemic patients
hadsignificantly blunted plasma ACTH response to ovineCRH, in
association with a substantial cortisolresponse.65 These data were
subsequently replicated byHolsboer and colleagues, and were
published in a let-ter.66 These data indicated that the
hypercortisolism ofdepression appropriately restrained the HPA
axis, sug-gesting a defect above the hypothalamus. Thus, thehigher
the basal cortisol, the lower the plasma ACTHresponse. The
substantial cortisol response to a bluntedACTH response indicated
that the adrenals had beenchronically overstimulated, and therefore
hypertro-phied and hyperresponded to ACTH. Our studies inpatients
with Cushing’s disease, a peripherally(pituitary) mediated form of
hypercortisolism, helpedsubstantiate the central origin of the
hypercortisolismof major depression. In contrast to patients with
majordepression, patients with Cushing’s disease showedprofound
ACTH and cortisol responses to CRH, indi-cating that the pituitary
itself was resistant to cortisolnegative feedback. Our subsequent
studies further con-firmed that the hypothalamic component of
Cushing’sdisease responded normally to glucocorticoid
negativefeedback. The pronounced differences of the responsesto CRH
in depression and Cushing’s disease, based ontheir distinct
pathophysiology, proved to be clinicallyuseful in the often
difficult differential diagnosisbetween major depression with
pronounced hypercort-isolism and early or mild Cushing’s
disease.67
Many other lines of evidence support a role for
thehypersecretion of CRH in the pathophysiology of
hyp-ercortisolism. Nemeroff et al found that CRH receptornumbers
were reduced in frontal cortex in post mortemsamples taken from
patients who had died by suicide.68
Nemeroff also found that CSF CRH levels in depressedpatients69
were elevated and later showed that CSFCRH levels in patients fell
significantly after treat-ment.70 In our group, DeBellis found that
fluoxetinesignificantly lowered CSF CRH levels when
-
High vs low CRH/NE statesPW Gold and GP Chrousos
260
Molecular Psychiatry
Figure 3 Schematic diagram of the interrelation of stress system
mediators and circuitries in melancholic and atypicaldepression.
(middle) Normal. In the absence of stressful stimuli, the stress
system is not quiescent, but rather resides in adynamic state of
bidirectional interactions among stress mediators. Such a
homeostatic equilibrium can react flexibly to a rangeof different
stimuli that may preferentially affect one component over another.
Available data in primates suggest that underordinary
circumstances: (1) the prefrontal cortex inhibits the amygdala, HPA
axis, and LC-NE system; (2) an activated amygdalainhibits the
prefrontal cortex and stimulates both the HPA axis and the LC-NE.
In the reverse direction: (3) the LC-NE activatesthe amygdala and
HPA axis and inhibits the prefrontal cortex; (4) the HPA axis
activates the LC-NE and the amygdala. Dottedlines inhibitory, solid
lines excitatory. Schematically, in the normal state, the relative
strength of each component is similar,denoted by circles of
identical diameter. (left) Melancholic depression can be
conceptualized as a prolonged and intensifiedstress response that
does not yield to its ordinary counter-regulatory restraints. The
net effect is a pronounced shift in equilib-rium with the following
results: (1) diminished activity of the prefrontal cortex; (2)
activation of the amygdala; (3) activationof the core stress
system. The primary defect could arise from any of the structures
pictured in the schematic diagram orcircuits in which they
participate. Note reciprocal relations between prefrontal cortex
and subcortical stress components. Alsonote that the amygdala, LC,
and CRH system are all excitatory to one another so that an
increase in the activation of onecomponent could set off a
reverberate sequence of further activations unless overtaken by
inhibitory stimuli. Similarly, theprefrontal cortex and the
components of the stress system exhibit bi-directional inhibition
on one another. (right) Atypicaldepression can be conceptualized as
a state of stress system hypoactivity that has yielded too readily
to its counter-regulatoryrestraints. The net effect is a pronounced
shift in equilibrium with hypoactivity of each of the components of
the stress system.Theoretically, the prefrontal cortex could be
disinhibited or primarily hyperactive. Abbreviations: PFC,
prefrontal cortex;AMYG, amygdala.
depressions remitted.71 In addition, we found that thechronic
administration of imipramine to healthy vol-unteers produced
effects compatible with a centraldownregulation of the HPA axis.72
CSF CRH. Finally,in experimental animals, we showed that the
chronic,but not acute, administration of imipramine signifi-cantly
reduced CRH mRNA levels while significantlyincreasing the mRNA
levels of the type I glucocorticoidreceptor in the hippocampus,
thought to be animportant element in the feedback inhibition of
theHPA axis.73
In a study of the 30-h pattern of CSF CRH levels inseverely
depressed inpatient melancholic subjects andcontrols, we found
inappropriately ‘normal’ integrated30-h CSF CRH concentrations
despite significant hyp-ercortisolism and around-the-clock
elevations of CSFNE74 ( Figure 4). Because the overall pool of CSF
CRHand plasma ACTH levels are glucocorticoid suppress-ible,75 we
previously suggested that quantitatively ‘nor-mal’ CSF CRH and
plasma ACTH levels in the face ofhypercortisolism are,
nevertheless, inappropriate forthe patients’ degree of
hypercortisolism.76 Our reason-ing was as follows: we compared
levels of CSF CRHin patients with depression associated with
Cushing’s
disease (a pituitary disorder), who had matchingdegrees of
hypercortisolism. We found extremely lowlevels of CSF CRH in our
patients with Cushing’s dis-ease, whose CNS was normal but in whom
very highlevels of pituitary driven cortisol bombarded the
hypo-thalamic CRH system, profoundly suppressing it.67 Incontrast,
in a group of patients with major depressionassociated with
hypercortisolism of similar magnitudeto the group of patients with
Cushing’s CSF CRH levelswere substantially and significantly higher
indepressed patients than in those with Cushing’s dis-ease. Thus,
cortisol itself has a highly significant sup-pressive effect on the
overall levels of CSF CRH. In con-trast, CSF levels in the
depressed patients were notsuppressed at all. The failure of
hypercortisolism tosuppress CSF CRH levels in depressed patients
sug-gests either resistance to glucocorticoid negative feed-back at
several potential sites or an overdriven HPAaxis whose drive
overcomes normal glucocorticoidfeedback, a possibility that we
favor. This formulationis compatible with the finding that the
significant nega-tive correlation between CSF CRH and plasma
cortisolfound in controls was lost in patients with
melancholicdepression74 (Figure 5). We had previously found
that
-
High vs low CRH/NE statesPW Gold and GP Chrousos
261
Figure 4 Thirty hour levels of CSF CRH, CSF norepinephrine,
plasma ACTH and plasma cortisol. Diurnal curves of (a)
plasmacortisol, (b) plasma ACTH, (c) CSF NE and (d) CSF CRH levels
(mean ± SE) in 14 healthy volunteers and 10 patients with
majordepression, melancholic type. Curves are resultant from the
averaged measurement per time point across a group of subjectsusing
the cropped hormonal series. The shaded area represents lights off
(2300–0700 h). In the right corner insets under eachpair of curves,
the bar graphs represent the average of the mean value for each
series of hormonal measurements (mean ± SE). *P � 0.02. Despite
around-the-clock increases in plasma cortisol and CSF NE levels,
CSF CRH and plasma ACTH are similarto those in controls, though
inappropriately high for the degree of hypercortisolism. Note that
the diurnal rhythms for plasmacortisol and CSF NE are virtually
superimposable. From Wong et al. Proc Natl Acad Sci 2000; 97:
325–330.
CSF CRH levels were normal in a hypercortisolemicgroup of
patients with melancholia, utilizing singlemeasurements of CSF CRH,
while Geracioti found sig-nificant decrements in CSF CRH in a group
of eucorti-solemic depressed patients (see below).
The interpretation of the meanings of CSF CRH indepression is
complicated by the fact that the PVN-median eminence component is
restrained by gluco-corticoid negative feedback. Glucocorticoids,
on theother hand, increase CRH mRNA levels in the amyg-dala, bed
nucleus of the stria terminalis, and in thePVN CRH pathway
descending to brainstem norad-renergic neurons. We have found that
lesioning thePVN in the rat decreases CSF CRH by 50–60%(Mamalaki E,
unpublished observations). Thus, acti-vations of amygdala CRH
neurons and of those thatdescend from the PVN to the brainstem
would be neu-tralized by the potent suppressive effects of
glucocort-icoids on the CRH involved in the HPA axis. Given
thevarious permutations and combinations of multiplesites secreting
CRH into the CSF, we would not be sur-prised by findings of
increased CSF CRH levels in
Molecular Psychiatry
patients vs controls. The pathophysiologic meanings ofthe two
are virtually identical.
Although there are many intriguing lines of infor-mation
implicating CRH in the pathophysiology ofmajor depression,
especially melancholia, it should beemphasized, that this has by no
means been defini-tively substantiated, but merely supported by
circum-stantial evidence. The availability of a CRH type
1antagonist that crosses the blood–brain barrier shouldprovide
further important information about the role ofCRH in
depression.
The locus ceruleus norepinephrine system
Studies of biological factors in major depression havelargely
relied on serendipitous discovery of antide-pressants and the
determination of their mechanismsof action. The most important
hypothesis to emergefrom this work was the catecholamine hypothesis
ofdepression.77 This hypothesis was based on theassumptions that
pharmacologic depletion of NE byreserpine apparently induced major
depression, while
-
High vs low CRH/NE statesPW Gold and GP Chrousos
262
Molecular Psychiatry
Figure 5 Cross correlations of the 30-h levels of CSF CRHand
cortisol and between CSF NE and cortisol. Cross corre-lation
analysis of the mean coefficients of variation betweenCSF CRH and
plasma cortisol (a and b), and between CSF NEand plasma cortisol (c
and d). Note the negative correlationbetween cortisol and CRH in
controls which is lost inpatients. A positive correlation exists
between NE and cor-tisol during standard cross correlational
analysis and in thedetrended analysis as well (e, f). The detrended
CSF analysiscorrects for the effect of diurnal variation and is an
index ofrapid changes in hormone levels between CSF NE and
plasmacortisol level. Note that the positive detrended
correlationbetween CSF NE and plasma cortisol is almost as robust
asthat between plasma ACTH and plasma cortisol. From Wonget al.
Proc Natl Acad Sci 2000; 97: 325–330.
apparent pharmacologic augmentation of norad-renergic activity
by MAO inhibitors and NE uptakeinhibitors (tricyclic
antidepressants) exerted anti-depressant effects.78 By positing
that depression couldbe caused by a deficiency of NE rather than
only byearly psychological trauma or a lifetime of adverseevents,
the catecholamine hypothesis served as a majorimpetus for the
emergence of modern biological psy-chiatry.
Although the original catecholamine hypothesis ofmajor
depression stated that a deficient NE delivery to
its receptors in the CNS was one of the main causes
ofdepression,77 studies of NE or its metabolites in CSF,plasma, or
urine, or of components of the norad-renergic system in post-mortem
brain samples,reported indices suggestive of decreased,79–96
nor-mal,97–105 or increased106–110 delivery of NE to itsintended
receptors in the CNS or periphery. It shouldbe noted that almost
all of the prior studies of CSF NEor its metabolites in depressed
patients were based onsingle time points. Moreover, neither in vivo
nor postmortem studies stratified patients on the basis
ofdepressive symptomatology of melancholic or atypi-cal
subtype.
In an attempt to clarify in vivo central noradrenergicfunction
in major depression, we studied a groupconsisting only of very
severely, drug-free depressedmelancholic patients who were to
receive ECT for thetreatment of their depression. Via an indwelling
lum-bar drain, we measured CSF CRH and NE 30 consecu-tive hours. We
also took half-hourly samples of plasmaACTH and cortisol. We found
unequivocal evidence ofa pronounced central hypernoradrenergic
state in mel-ancholic patients. CSF NE levels were elevated
aroundthe clock, including during sleep74 (Figure 4c).Although
there has been debate about the origin of CSFNE, Goldstein et al
found that patients with theShy–Drager syndrome, a
neurodegenerative diseasethat features loss of central
noradrenergic cells butintact post-ganglionic sympathetic
nerves,111 have adissociation between normal plasma NE and
DHPGlevels and low CSF NE and DHPG levels (D Goldstein,unpublished
observations). These data indicated thatthe hypernoradrenergic
state of depression was notrelated to the conscious distress of the
disorder.
Plasma cortisol levels were also significantlyincreased.
Melancholic patients, like controls, showeddiurnal rhythms of CSF
NE and plasma cortisol levelsthat were virtually superimposable and
positively cor-related (Figure 5 ). These data suggest the
hypothesisthat cortisol stimulates centrally directed NE, and
arecompatible with our in vitro finding that NE stimulatesCRH from
hypothalamus, subsequently replicated byItoi and
colleagues.24,112
In this regard, a post mortem study by Radesheer etal of brains
taken from depressed patients who hadcommitted suicide showed a
significant increase inhypothalamic neurons expressing CRH that was
pre-dominantly found in neurons sending descending pro-jections to
brainstem noradrenergic nuclei113 (Figure6). These data suggest
that a specific, relatively smallhypothalamic CRH pathway which
just goes to thebrainstem may play a disproportionate role in
thepathophysiology of melancholia and implicate a parti-cularly
important way in which CRH and NE mayinteract in this disorder. As
noted earlier, glucocortico-ids increase CRH mRNA levels in the
separate PVN-containing population of CRH neurons that descend
tobrainstem noradrenergic neurons. Therefore, these datasuggest a
specific way in which glucocorticoids canactivate a CRH pathway
which then goes on to activatebrainstem noradrenergic neurons,
providing another
-
High vs low CRH/NE statesPW Gold and GP Chrousos
263
Figure 6 A specific PVN CRH pathway to brainstem noradrenergic
nuclei independent of the HPA axis. Post mortem studies inpatients
who had been diagnosed with major depression reveal a significant
increase in hypothalamic CRH-containing neurons.Surprisingly, this
increase was much more pronounced in hypothalamic CRH-containing
neurons that send descending projec-tions to brainstem
noradrenergic nuclei. The fact that glucocorticoids seem to
activate rather than restrain this pathway intro-duces another
context for a positive feedback loop in which an activated HPA axis
leads to increased CRH secretion, whichin turn activates brainstem
noradrenergic nuclei. This feedback loop may contribute to the
pronounced hypernoradrenergicstate seen in melancholic depression
as well as the positive correlation we found between CSF NE and
plasma cortisol levelsseen in both patients and controls. The
postulated activating effects of glucocorticoids upon hypothalamic
CRH containingneurons that send descending fibers to brainstem
noradrenergic nuclei establishes yet another positive feedback loop
withinthe stress system and in melancholic depression.
context for mutual reverberatory loops between CRH,NE, and
cortisol. These relationships in melancholicdepression are detailed
in Figure 3.
Re-interpretation of neuropharmacological data ofrelevance to
depression
Scientists were initially convinced that norepinephrineuptake
blockers and MAO inhibitors enhanced norad-renergic function by
either preventing the removal ofNE from the synaptic cleft78 or by
interfering with itsenzymatic degradation. However, the elegant
work ofWeiss and others has shown that tricyclic antidepress-ants,
MAO inhibitors, and specific serotonin uptakeinhibitors
consistently decrease the firing rate of the LCduring
stress.114
Although the purported capacity of reserpine toinduce depression
played a prime role in the originalcatecholamine hypothesis of
depression, a carefully-written review of the cases of
reserpine-induceddepression suggests that, in retrospect, most
patientsmay have experienced a neuroleptic like syndromeconsisting
of sedation, hyperphagia, apathy, and Park-insonism.115
It should be emphasized that we do not believe thatall
melancholics have activated noradrenergicsecretion. The
pathophysiology of this state is, ofcourse very complex and
influenced by multiple genes.There may be many melancholic patients
who havenormal noradrenergic function but combined abnor-malities
in other genes in the context of adverseenvironmental factors that
lead to melancholia. It isalso, of course, clear that
norepinephrine is not theonly or principal neurotransmitter
involved in thepathophysiology of major depression.
Norepinephrineand serotonin uptake inhibitors each exert effects
on
Molecular Psychiatry
both systems, suggesting an important role for sero-tonin as
well.
Over a dozen 5HT receptors have been identified. Insome cases,
multiple receptors are found on the sameneuron that exert
antithetical effects on cell firing. Inthe midst of this diversity
and complexity, the hypo-thesized role for 5HT in the
pathophysiology ofdepression is based largely on data showing that
effec-tive antidepressants of all classes (including NE
uptakeinhibitors) increase the release of 5HT.116 In
addition,effective antidepressants, including specific
norepi-nephrine as well as serotonin uptake inhibitors,increase the
density of post-synaptic 5HT1a receptorsand decrease the density of
5HT 2a receptors.117 As acorollary, post-synaptic 5HT1a are reduced
in thebrains of suicides, while post-synaptic 2a receptors
areincreased in suicide post-mortem brains.118 The inter-dependence
of the LC-NE and serotonergic systems isillustrated by the fact
that activation of 5HT1areceptors leads to a decrease in LC
firing119 and in thedensity of cortical beta adrenergic receptors,
whileactivation of 5HT 2a receptors increases the LC
firingrate.120
The confluence of long-term activation of the CRHand
noradrenergic systems in depression, in associ-ation with
glucocorticoid hypersecretion, is a highlypathologic state that
could readily produce the pro-found hyper-arousal and anxiety that
occurs in melan-cholic depression. The capacity for components of
theCRH and NE systems to activate one another, to leadto
glucocorticoid excess, and for key components ofeach to respond (eg
the amygdala CRH neurons) posi-tively to glucocorticoids,
establishes the context for apernicious cycle of stress-mediator
activation that canbe exceedingly difficult to break (Figure 3).
Excessivesecretion of norepinephrine and cortisol, regardless
of
-
High vs low CRH/NE statesPW Gold and GP Chrousos
264
Molecular Psychiatry
the primary cause, could intensify this pathophysiolog-ical
picture in several ways. By activating the amygdalaand inhibiting
the medial prefrontal cortex, norepi-nephrine would promote well
rehearsed rather thannovel programs of behavior and accentuate the
activityof the amygdala. Glucocorticoid excess could set intomotion
several vicious cycles, including damage tohippocampal
glucocorticoid responsive neurons thatrestrain the HPA axis,
activation of the amygdala andextra-amygdala sites involved in
conditioned fear anddeclarative emotionally-laden memories (that
would inturn lead to more hypercortisolism), and activation
ofdescending hypothalamic CRH pathways to furtherpotentiate
brainstem noradrenergic activity.
To support a role for glucocorticoids acceleratingthis vicious
cycle, are the data of Schatzberg’s on theseveral-day
glucocorticoid antagonist administration inpatients with psychotic
depression. RU 486 induced arapid decrease in depressive symptoms
of at least 50%in the majority of his patients. This study lays
thegroundwork for the potential use of RU 486 as a rapidlyacting
agent for ameliorating very severe depressivesyndromes that require
immediate intervention(Belanoff et al in press, PNAS).
Neuroimaging studies of stress system componentsin major
depression
Neuroimaging studies in patients with majordepression reveal
changes at local synaptic sites in sev-eral areas, most notably the
amygdala and prefrontalcortex. Such regional abnormalities will
ultimatelyprovide the basis for the construction of models
thatplace these abnormalities in the context of the variouscycles
in which these structures partake.
Patients with major depression show increased cer-ebral blood
flow and metabolism in the amygdala.121
Activation in the left amygdala persisted after recoveryfrom
depression. During depression, amygdala acti-vation correlated
positively with depression severityand baseline plasma cortisol
levels.121 The latter find-ing is of interest in the light of the
fact that the amyg-dala activates the HPA axis.31 Glucocorticoids
in turn,accentuate the amygdala CRH system.122 A recentstudy found
that neural activity in several 5-HT-relatedbrain areas, eg dorsal
raphe, habenula, septal region,amygdala, and orbitofrontal cortex,
covaried signifi-cantly with plasma levels of tryptophan and
ratings ofdepressed mood. Antidepressant-treated patients
whorelapsed upon tryptophan depletion had higher base-line amygdala
metabolism than similar subjects whodid not relapse.
A series of studies in patients with major depressionhave
reported significant decreases in activation of thedorsolateral
prefrontal cortex and significant increasesin ventral prefrontal
and paralimbic structures.123
Higher depression ratings correlated negatively withthe activity
of left dorsolateral prefrontal cortex, whileanxiety levels were
positively correlated with paralim-bic system activity. Successful
treatment of depressionwas associated with inhibition of overactive
paralimbic
regions and normalization of hypoactive dorsolateralprefrontal
cortex sites.124 Thus, major depressionseems associated with
hypoactivity of cortical struc-tures and a corresponding
hyperactivity of paralimbic/subcortical loci.
A key finding that has been well replicated is that ofa
significant loss of volume in the left subgenual pre-frontal
cortex, an area that is closely connected to theamygdala and that
contributes to the inhibition of theHPA axis and sympathetic
nervous system.125 Thesepatients had predominantly melancholic
depression(W Drevets, personal communication). Scanning
andexamination of post mortem brain samples taken frompatients who
had committed suicide revealed a 40%decrease in the volume of the
left sub-genual cortex. Itis of interest that the lateralization
found in patientswith depression is compatible with data in the
ratshowing that lesioning of the left infralimbic cortexcauses
activation of the HPA axis and of sympatheticfunction, while
lesioning of the right produced a dec-rement in the activity of
these systems. These dataindicate that the left infralimbic region
inhibits theright.126 We therefore postulate that the
left-sideddefect in melancholic depressed patients leads to
hyp-eractivity of both the amygdala and core stress
systemcomponents. In patients with atypical depression,however, the
left could be hyperactive or hypertro-phied, leading to excessive
restraint of the right andhypoactivity of core stress system
components. Thus,in patients with this subtype of depression, a
primarydefect in the right side may emerge, in contrast to thatseen
in melancholia (Figure 7).
Long-term medical consequences of melancholicdepression
Patients with major depression show a doubling of themortality
rate at any age, independent of suicide.10,127
Premature ischemic heart disease is likely to play animportant
role, and the relative risk for clinically sig-nificant coronary
artery disease in patients with majordepression is 2.0 or more in
studies that independentlycontrolled for risk factors such as
smoking and hyper-tension.10,127 Figure 7 details the potential
mechanismsfor premature ischemic heart disease that includes
avicious spiral between insulin resistance and increasedvisceral
fat, potentially leading to hypertension, dysli-pidemia,
hypercoagulation, and enhanced endothelialinflammation.128,129
Increased sympathetic outflowsseen in both our severely depressed
inpatients and lessseverely depressed outpatients also further add
to car-diac risk in several other ways. Norepinephrine is wellknown
to promote insulin resistance,130 left ventricularhypertrophy,131
and increases in myocyte growth, art-eriolar and ventricular
remodeling,132,133 blood volumeand blood viscosity.134 In addition,
NE also activatesplatelets, cytokine release and is
arrhythmogenic135
(Figure 7).We have also shown that as many as 40% of
premenopausal women (average age 41) with severeaffective
disorder have peak bone density that is two
-
High vs low CRH/NE statesPW Gold and GP Chrousos
265
Figure 7 Factors in major depression that promote susceptibility
to heart disease. Hypercortisolism and a deficiency of sexsteroids
and growth hormone each contribute to increases in the visceral fat
mass, leading to increases in both portal andperipheral free fatty
acids. Hypercortisolism and the increases in portal and peripheral
free fatty acids both contribute to insulinresistance, which
exacerbates the increase in visceral fat mass and promotes
activation of the sympathetic nervous system andhypertension.
Elevated portal free fatty acids lead to dyslipidemia associated
with increased triglycerides (TG) and decreasedHDL cholesterol.
Increased portal and peripheral free fatty acid levels also promote
endothelial inflammation through increasesin tumor necrosis factor
� (TNF-�), interleukin 6 (IL-6), and C-reactive protein (CRP).
Peripheral free fatty acids increase thehepatic production of
fibrinogen and the level of plasminogen activator inhibitor 1,
leading to increased clotting and deficientfibrinolysis. GH denotes
growth hormone; T, testosterone; E2, estradiol; PAI-1, plasminogen
activator inhibitor. From Gold andChrousos. Proc Assoc Am Phys
1999; 111: 22–34.
standard deviations or (20%) below their peak13 (forbiopsy
sample in a 40-year-old woman, please see Fig-ure 8). We have also
shown that these women showan almost 40% incidence of premature
osteoporosis ateither the hip or spine, which usually occurs in
themid-to-late twenties.13 Ordinarily, bone mineral den-sity does
not fall 20% below peak density until womenare in their 60s. For
every 10% loss below peak den-sity, the fracture rate doubles.136
For 40-year-oldwomen, this is not a great risk. However, for
40-year-old women who were already 20% below peak density,this
means that they have lost 1.5–2% per year sinceattaining peak bone
density at 28–30. At age fifty, theycould potentially have bone
mineral density that is35% below peak density and enter the
menopause withalready very compromised bones.136 It is of interest
thatin our depressed women, in contrast to the usual pres-entation
of osteoporosis or osteopenia, the greater lossof bone mineral
density was at the hip rather than atthe spine. We must emphasize
that this degree of boneloss is likely to reflect the fact that our
patients hadsevere affective illness. Thus, large studies of
patientswith major depression in outpatient settings are likelyto
find a lower incidence of osteopenia/ osteoporosis.
Many factors could contribute to this loss of bonemineral
density in women with past or currentdepression. Hypercortisolism
is an obvious potentialcause.137 In patients given glucocorticoids,
maximalbone loss occurs at 3–4 months after treatment.138,139
Since depressed, hypercortisolemic patients have gluc-ocorticoid
concentrations that are often equivalent to apatient receiving 10
mg of prednisone for 4 months orlonger, the loss of bone in
hypercortisolemic depressedpatients can be quite severe. These data
make a clearplea for the early and effective treatment of
melan-
Molecular Psychiatry
cholic depression. In addition to hypercortisolism,other factors
could also contribute to bone mineraldensity loss in women with
depression, including sup-pression of the growth hormone and
gonadal axes. Thehypersecretion of NE in patients with
melancholiacould also contribute to bone loss via activation of
thesecretion of IL-6. In postmenopausal women, it is
IL-6hypersecretion in the face of falling estrogen levels thatis
primarily responsible for post-menopausal osteo-porosis.
Neuroimaging studies of stress system componentsin major
depression
Neuroimaging studies in patients with majordepression reveal
changes at local synaptic sites in sev-eral areas, most notably the
amygdala and prefrontalcortex. Such regional abnormalities will
ultimatelyprovide the basis for the construction of models
thatplace these abnormalities in the context of the variouscycles
in which these structures partake.
Patients with major depression show increased cer-ebral blood
flow and metabolism in the amygdala.140 Arecent study found that
neural activity in several 5-HT-related brain areas, eg dorsal
raphe, habenula, septalregion, amygdala, and orbitofrontal cortex,
covariedsignificantly with plasma levels of tryptophan and rat-ings
of depressed mood. Antidepressant-treatedpatients who relapsed upon
tryptophan depletion hadhigher baseline amygdala metabolism than
similar sub-jects who did not relapse.
A series of studies in patients with major depressionhave
reported significant decreases in activation of thedorsolateral
prefrontal cortex and significant increasesin ventral prefrontal
and paralimbic structures.141
-
High vs low CRH/NE statesPW Gold and GP Chrousos
266
Molecular Psychiatry
Figure 8 Bone biopsy of the anterior iliac crest in a
40-year-old female with major depression currently in remission
(right)compared to the biopsy of a gender, age and BMI-matched
control (left). There are two striking features. The
trabeculationsare markedly reduced in the depressed patient. These
trabeculae are critical scaffolding for the bone and confer much of
itsstrength. Note that the cortex is also thinner in the depressed
patient. Ordinarily, glucocorticoids have much more effect
ontrabeculae per se than on the cortex. This suggests that factors
other than glucocorticoids are operative in the bone loss
ofdepression. Parenthetically, bone loss in depression is greater
in the hip than in the spine. Glucocorticoid mediated bone
lossoccurs predominantly in the spine.
Higher depression ratings correlated negatively withthe activity
of left dorsolateral prefrontal cortex, whileanxiety levels were
positively correlated with paralim-bic system activity. Successful
treatment of depressionwas associated with inhibition of overactive
paralimbicregions and normalization of hypoactive
dorsolateralprefrontal cortex sites.124 Thus, major depressionseems
associated with hypoactivity of cortical struc-tures and a
corresponding hyperactivity ofparalimbic/subcortical loci. A recent
very importantfinding by Drevets et al using in vivo scanning
andexamination of post mortem brain samples revealed a40% decrease
in the volume of the left sub-genual cor-tex, an area extremely
important in emotion and inregulation of the HPA axis and the LC-NE
system.124
These patients were predominantly patients withmelancholic
depression (W Drevets, personalcommunication). As noted, it is of
interest that the lat-eralization found in patients with depression
is com-patible with data in the rat showing that lesioning ofthe
left infralimbic cortex causes activation of the HPAaxis and of
sympathetic function, while lesioning ofthe right produced a
decrement in the activity of thesesystems. These data indicate that
the left infralimbicregion inhibits the right.126 We therefore
postulate thatthe left-sided defect in melancholic depressed
patientsleads to hyperactivity of both the amygdala and corestress
system components (Figure 3, left). Conversely,in patients with
atypical depression, a primary defect
in the right side may emerge, in contrast to that seenin
melancholia (Figure 3, right).
Whether decrements in the volumes of the subgenualmedial
prefrontal cortex and hippocampus in patientswith major depression
are reversible or representenduring neuropathic change is not yet
clear. However,both glucocorticoids and CRH have been shown to
beneurotoxic in experimental animals, so that core stresssystems
could participate in this neurodegeneration. Inan important
hypothesis by Nestler’s group, and in theexcellent work of Manji et
al, abnormalities in growthfactors and in other intracellular
transductionmediators may play a highly significant role in the
neu-rodegeneration in depressed patients.142,143 The idea
issupported by the fact that antidepressants such as lith-ium can
cause a regrowth of this lost tissue. Theresponse of tissue trophic
factors to stress and theirinteraction among stress mediators has
not yet beenelucidated.
Atypical depression
Rene Spitz made seminal observations regarding devel-opmental
abnormalities that befell infants placed inunderstaffed orphanages
shortly after birth.144,145 Forthe first 5 or 6 months, most of the
infants cried bitterlyfor hours until attended. Subsequently, they
withdrewand ceased crying altogether, even if they were leftalone
or had gone without eating for many hours. In
-
High vs low CRH/NE statesPW Gold and GP Chrousos
267addition, they lost apparent interest in the environ-ment
around them. It was as if the trauma and over-stimulation of their
early deprivation had led to a vir-tual shutdown of their affective
existence to protectthem from unenduarable pain. Subsequent studies
innon-human primates who were abandoned or abusedby their mothers
reveal a similar behavioral with-drawal in association with
hypoactivity of the HPAaxis.
One of the implications of these findings is that theclinical
presentation of melancholic depression mayrelate, in part, to
systematic activation of stressmediators. Prompted many years ago
by the idea thatarousal symptoms paralleled stress system activity
inmelancholia, we hypothesized that the lethargy,fatigue, and
hypersomnia of atypical depression wasassociated with a
pathological reduction of stress sys-tem mediators (reviewed in
3,6,24). This possibility wassupported in the previously cited data
in experimentalanimals showing that specific lesions in
infralimbiccortex resulted in pathological suppression of the
corestress system. The recent seminal findings of Yahudaand her
colleagues showing exaggerated cortisolresponses to low dose
dexamethasone and other datarevealing decreased urinary free
cortisol excretion sug-gest HPA axis hypoactivity in patients with
post-trau-matic stress disorder.146 To date, we have not foundfrank
decreases in 24-hour urinary free cortisol inpatients with atypical
depression or excessiveresponses to dexamethasone, although a much
largerseries is now pending. Therefore, detecting a
centrallymediated decrement in HPA axis function patientswith
atyical depression was initially difficult until wedeveloped an
endocrine paradigm for the differentialdiagnosis between adrenal,
pituitary, and hypothal-amic CRH-mediated hypoactivity of the HPA
axis.76,147
Prior efforts to document a centrally-mediated hypoac-tivity of
the HPA axis were complicated by the factsthat the HPA axis could
be normally quiescent formany hours per day, and that low level
pulses occur-ring many times a day were missing in
samplingstudies.
Because there were no known forms of a non-trau-matically
induced, centrally mediated HPA axishypoactivity in humans, we
studied patients with adre-nal insufficiency as a consequence of
either hypothal-amic tumor or traumatic pituitary stalk section.148
Inthe course of these studies, we identified a unique pat-tern of
delayed plasma ACTH responses to CRH, asso-ciated with very
attenuated cortisol responses to theACTH released during CRH
stimulation (Fig 9a, ii). Wepostulated that the blunted and delayed
plasma ACTHresponses reflected a lack of priming of pituitary
ACTHsecreting cells by endogenous CRH, similar to theblunted and
delayed TSH response to TSH in patientswith centrally mediated
hypothyroidism. Thus, thoughthe pituitary ACTH secreting cells are
making someACTH, they are not releasing it, so that it is kept in
asecondary releasable pool and is more slowly released.In response
to synthetic CRH, they eventually releasesome ACTH, though the
response is delayed, and can
Molecular Psychiatry
be prolonged. The blunted rather than the exaggeratedcortisol
responses to the blunted ACTH responsesmeant that their adrenals
had not been hyperstimul-ated by excessive CRH-driven
hypercortisolism, andindeed, had been hypostimulated. To support
thispremise, we gave these patients repeated priming pul-ses of
human CRH prior to their next CRH stimulationtest. In this context,
they significantly increased theircorticotroph capacity to respond
to CRH challenge.149
This ‘hypothalamic CRH deficiency’ pattern sub-sequently served
as a template for our search of othersyndromes of hypothalamic CRH
deficiency. It shouldbe noted that in contrast to hypothalamic
CRHdeficiency, the characteristic pattern in primary adre-nal
insufficiency is that of low cortisols, a very highACTH response to
CRH, and little cortisol response tothese high levels of ACTH.
Here, the hypothalamusand pituitary are normal, so that can respond
to disin-hibition mediated by an inability of the adrenals to
pro-duce adequate cortisol (Figure 9a, iii).
Like hypothalamic CRH deficiency, patients withmelancholia have
a blunted ACTH response to CRH.However, the response is not delayed
in melancholia.A key feature of melancholia is that the
cortisolresponse to the diminished amount of ACTH releasedduring
CRH stimulation was very robust rather thanalso blunted. Thus, the
adrenals had become hyperres-ponsive to ACTH as a consequence of a
hyperactivehypothalamic CRH neuron (Figure 9b, i).
We next studied patients with Cushing’s disease, anillness
frequently associated with depression. We hadshown that in
comparison to patients with melancholiawho had centrally mediated
hypercortisolism, the hyp-ercortisolism of Cushing’s disease is
pituitarymediated150 Figure 9ci). We had also shown that morethan
80% of their depressions were of the atypical pat-tern.151 We chose
to study these patients after surgerywhen they were without the
complication of the pitu-itary microadenoma, but left with a full
complementof previously normal pituitary ACTH secreting
cells.Often, these subjects are adrenally insufficient
because,after the microadenoma is removed, their normalresidual
ACTH secreting cells are silent. This adrenalinsufficiency could
thus reflect long-term suppressionof hypothalamic CRH or pituitary
ACTH secretion inthe context of many years of pre-existing
hypercortisol-ism, much as patients on supraphysiologic doses
ofglucocorticoids, who must be given steroids after taper-ing. To
study the source of the adrenal insufficiencyin our Cushing’s
patients, we stimulated these patientswith CRH. They responded with
delayed and bluntedplasma ACTH response and very reduced
cortisolresponses (the hypothalamic CRH deficiency pattern,Figure
9c, ii). We surmised that since their pituitaryACTH secreting cells
could respond to exogenous CRH,they would have had levels of plasma
ACTH postoper-atively, if their own hypothalamic CRH cells were
stillsecreting some CRH. These patients showed the
clear‘hypothalamic’ pattern in response to CRH stimulation(Figure
9c, ii). In addition, after many priming dosesof synthetic human
CRH, the ACTH response to CRH
-
High vs low CRH/NE statesPW Gold and GP Chrousos
268
Molecular Psychiatry
was partially restored, indicating successful priming
ofpituitary ACTH secreting cells. To further support thepremise of
a decrement in hypothalamic ACTHsecretion in Cushing’s disease due
to longstandinghypercortisolism, we found that, compared to
controls,their CSF CRH levels were profoundly suppressed.75
We were then ready to study patients with affectivedisorder. To
determine whether patients with atypicaldepression show evidence of
a centrally mediatedhypoactivity of the HPA axis, we first studied
patientswith depression of seasonal affective disorder, a syn-drome
virtually identical to that of atypicaldepression.152,153 Patients
with seasonal affective dis-order responded to ovine CRH with all
of the featuresof the ‘CRH deficient’ phenotype (Figure 3,
right).154
Blunted delayed plasma ACTH responses were associa-ted with
small rather than exaggerated cortisol
responses to CRH. Moreover, the ‘CRH deficient’ pat-tern
resolved after non-pharmacologically inducedremission from
depression.154 We have recently foundthat depressed children of
mothers who also had his-tories of major depression whose parental
style washostile and overbearing showed the CRH
deficiencyphenotype. These individuals had participated in theNIMH
Longitudinal Study of depressed mothers andtheir children, started
over 20 years ago. Many of thechildren had been followed since
infancy.155
Almost simultaneously, we studied patients with aclassic fatigue
state, chronic fatigue syndrome. This ill-ness is by no means
synonymous with depressionthough the two disorders share some
features. We alsoshowed that these subjects had blunted ACTH and
cor-tisol responses to CRH. They also had bimodalresponse in their
ACTH/cortisol dose response
-
High vs low CRH/NE statesPW Gold and GP Chrousos
269Figure 9 Contrasting plasma ACTH and cortisol responsesto
synthetic corticotropin-releasing hormone in various states.Graphs
on the left show the plasma ACTH and cortisolresponses to synthetic
ovine corticotropin releasing hormone(CRH). The gray shaded area
denotes the normal range. Dia-grams on right show pathways from
hypothalamus to pitu-itary to adrenals (triangles). Line thickness
denotes relativeamounts of corticotropin-releasing hormone or
cortisolsecreted. Negative signs denote feedback inhibition.(a)
Normal (i), hypothalamic CRH deficiency (ii) , and primaryadrenal
insufficiency (iii). Compared to controls, patientswith known
hypothalamic CRH deficiency have an unusualcombination of a low
basal plasma ACTH and cortisol levelswith blunted (and delayed)
plasma ACTH and cortisolresponses to synthetic CRH. This pattern
reflects the lack ofpriming of pituitary ACTH-secreting cells by
endogenousCRH, and contrasts markedly with the brisk and
exaggeratedplasma ACTH responses seen in primary adrenal
insuf-ficiency, in which primed pituitary ACTH-secreting
cellsrespond to synthetic CRH unrestrained by the
glucocorticoidnegative feedback. (b) Melancholic depression (i),
and atypi-cal depression (ii). In melancholic depression,
elevatedplasma cortisol levels appropriately restrain pituitary
ACTHsecreting cells in their response to synthetic CRH. Due
tochronic centrally-mediated stimulation of the adrenals, theyshow
an exaggerated plasma cortisol response during syn-thetic CRH
stimulation. In chronic fatigue syndrome, seasonalaffective
disorder, and postpartum depression, blunted andoften delayed
plasma ACTH responses to synthetic CRHoccur in the context of
either normal or significantly reducedbasal cortisol levels. Plasma
cortisol responses are also sig-nificantly reduced, indicating
relatively long-term hyposti-mulation of the adrenals by ACTH. (c)
Cushing’s disease:untreated (i), post-surgery (ii), and primed
post-surgery (iii).The profound basal hypercortisolism of Cushing’s
disease isassociated with exaggerated rather than restrained
plasmaACTH responses (as in melancholia), indicating a gross lackof
glucocorticoid negative feedback at the pituitary. Low
basalcortisol levels after surgery reflect either residual
suppressionof remaining pituitary ACTH secreting cells or
hypothalamicCRH neurons due to exposure to long-standing and
profoundbasal hypercortisolism. This suppression lasts up to a
year.Postoperative Cushing’s disease patients show a blunted
anddelayed plasma ACTH response to synthetic CRH, suggestingthat
remaining pituitary ACTH secreting cells can respond toCRH when it
is available. Priming with multiple pulses ofsynthetic human CRH
largely restores the plasma ACTHresponse to synthetic CRH. From
Gold et al. Proc Assoc AmPhys 1999; 111: 22–34.
curve.156 In response to low dose ACTH, patients withchronic
fatigue responded with exaggerated plasmacortisol responses to
ACTH. This probably reflectedtwo factors: (1) the presence of
increased sensitivity ofadrenal ACTH receptors to ACTH on account
of long-standing hypostimulation; and (2) the fact that eventhough
their adrenals were small, they could produceaugmented cortisol
responses from an adrenal of lowmass in the context of a very low
dose of ACTH.156 Athigh doses, patients with chronic fatigue
syndrome hadattenuated responses to ACTH. This response presum-ably
reflects the fact that the low adrenal mass inunderstimulated
adrenal cortices in chronic fatiguewas unable to fully respond to a
high dose of ACTH.
To further test the hypothesis of a centrally-mediated
Molecular Psychiatry
HPA axis deficiency in patients with classic atypicaldepression,
we measured plasma ACTH and cortisollevels every 3 minutes for 24
hours in four patientswith atypical depression and in four
controls. We rea-soned that such sampling might provide sufficient
res-olution to detect low level pulses of ACTH and cortisolthat
were otherwise missed with less frequent sam-pling. Our data showed
that patients with atypicaldepression had significant reductions in
plasma ACTHsecretion in the face of normal pituitary and
adrenalcomponents of the HPA axis, suggesting a hypothal-amic CRH
deficiency (Licinio and Gold, unpublishedobservations). Normal
plasma cortisol levels may havebeen maintained by compensatory
mechanisms. Weknow that in the context of unilateral
adrenalectomy,the residual adrenal is stimulated by factors
otherthan ACTH.
Only one study reports reduced CSF CRH levels inpatients with
major depression. This was an exceed-ingly carefully done study
from a highly respectedgroup that utilized an indwelling lumbar
catheter fordetermination of CSF CRH levels over some hours.
Ger-acioti et al noted that many, but not all of the subjectsof
this study had atypical features.157 It is noteworthythat these
patients were eucortisolemic rather thanhypercortisolemic.
Eucortisolism in the context of acentral CRH deficiency would not
be surprising, giventhe multiple redundant mechanisms for
maintainingnormal glucocorticoid secretion such as
enhancedsympathetic stimulation of the adrenal cortex.147
Hypoactivity of one of the core stress system compo-nents that
promotes arousal and diminishes foodintake could contribute to the
lethargy, fatigue, andhyperphagia characteristic of atypical
depression.7
These data indicate that hypercortisolism may not bethe only
abnormality of HPA axis function in depress-ive illness. Thus,
optimal functioning of the CNSrequires that core stress system
components remainwithin a carefully-maintained range, and that
deficitsin CNS function can occur in the context of either hyp-er-
or hypoactive LC-NE and CRH systems.
Recent data indicate that inflammatory mediatorssuch as IL-1
recruit hypothalamic CRH containing neu-rons in a negative feedback
loop in which glucocortico-ids exert immunosuppressive effects to
prevent theimmune response from overshooting.158 If hypo-thalamic
CRH neurons fail to respond adequately tocytokine stimulation, the
resultant failure of adequateglucocorticoid-mediated restraint of
the immune sys-tem results in a hyper-immune state. Esther
Sternbergin our group found rats whose hypothalamic CRH neu-rons
responded insufficiently to inflammatorymediators and developed a
range of autoimmune dis-eases dependent on the trigger
administered. Thus, anintact HPA axis seems necessary for an
immuneresponse of normal magnitude. In this regard, patientswith
chronic fatigue syndrome not only present withfatigue, but also
with inflammatory symptoms such asmuscle ache and joint pain and
feverishness.159 It iswell known that many patients visit primary
carefacilities with complaints of fatigue and low-grade
-
High vs low CRH/NE statesPW Gold and GP Chrousos
270
Molecular Psychiatry
inflammatory symptoms that have no apparent patho-physiologic
basis. Many of these patients could havedisinhibited immune systems
on the basis of a hypoac-tive HPA axis.
The hyperphagia that is a defining characteristic ofatypical
depression is likely to lead either to obesity ora cycle of weight
gain and weight loss occurringthroughout recurrent episodes of
depression. Becauseweight that is regained after weight loss may be
prefer-entially distributed as intra-abdominal fat, either
sus-tained obesity or weight cycling could result in adverse
Table 1 Postulated clinical differences and differential
long-term medical consequences of melancholic and
atypicaldepression
Melancholic AtypicalClinical PhenotypeLevel of arousal
Hyperaroused Hypoaroused, apatheticAnxiety level Anxious Generally
not anxiousReactivity Relatively unreactive to environment Reactive
to environmentAffect Stereotyped affect Stereotyped affect,
intermittent
reactivityEmotional memory Predominance of painful emotional
Relatively out of touch with past
memoryCapacity for pleasure Anhedonic AnhedonicCognition
Decreased concentration, perseveration Loss of focusBehavior Shift
to relatively well-rehearsed Unmotivated, inactive
behaviors
NeurovegetativeSleep Decreased sleep; poor quality Increased
sleep; poor qualityAppetite Decreased food intake, weight loss
Increased food intake, weight gainEnergy level Overt energy level
variable Marked lethargy and fatigueLibido Diminished
Diminished
NeuroendocrineHPA axis Centrally-activated Centrally-mediated
hypoactivityGH axis Suppressed SuppressedReproductive axis
Suppressed Suppressed
AutonomicSympathetic activity Increased Decreased
Body CompositionBody Mass Index (BMI) Normal HighLean body mass
Decreased (sarcopenia) NormalTotal body fat Normal or increased
IncreasedVisceral fat Increased Increased
MetabolicInsulin sensitivity Decreased DecreasedLipid metabolism
Dyslipidemia DyslipidemiaCoagulation Hypercoagulable/decreased
fibrinolysis Hypercoagulable/decreased fibrinolysis
Immune Function Relatively immunosuppressed Relatively
immunoenhanced
Medical SequelaeHeart disease Premature ischemic heart disease
Premature ischemic heart diseaseOsteoporosis Premature osteoporosis
Normal boneInfection/inflammation Increased susceptibility to
infection Increased susceptibility to
inflammationNeurodegeneration Hippocampus/medial prefrontal
cortex ?
Melancholic depressed patients, in many instances, should have
different medical complications than patients with
atypicaldepression. We suggest that the common medical
complications we predict might be the same, occur because of
differentmechanisms. For instance, we predict that melancholic
patients will have decreased insulin sensitivity because of
hypercortisol-ism and increased NE secretion, while atypical
patients will be insulin-resistant because of low growth hormone
secretion anda non-cortisol mediated increase in visceral fat.
metabolic consequences conducive to coronary arterydisease. As
noted, the visceral fat mass is metabolicallyvery active and can
lead to several problems such aspremature cardiac disease. On the
other hand, our datain patients with the atypical depression of
seasonalaffective disorder suggest that bone mineral density isnot
reduced (Gold et al, unpublished observations), incontrast to the
group of depressed patients in whommean 24-hour urinary free
cortisol excretion was elev-ated.
For a systematic comparison of the potential long-
-
High vs low CRH/NE statesPW Gold and GP Chrousos
271term medical complications of melancholic and atypi-cal
depression, please see Table 1.
Summary and conclusions
In summary, the available data suggest that there isconcomitant
activation of the CRH and LC-NE systemsin melancholic depression,
and that CRH, NE, and cor-tisol participate in mutually reinforcing
positive feed-back loops that can generate a tremendous and
pro-longed response involving many brain areas. Both CRHand NE
activation may potentiate the intensity offuture stress responses
by enhancing the encoding ofadverse emotional memory and by
sensitizing specificsubstrates so that subsequent responses are
enhanced.Because CRH and the glucocorticoids are neurotoxic,
aprogressive loss of critical tissue may theoreticallyoccur. It is
of interest that as early as 1984, investi-gators postulated a role
for CRH that included not onlytransducing symptoms of melancholic
depression butalso in sensitizing the stress system (Figure 10). We
arenow, of course, armed with the revolution in moleculargenetics,
have gotten inside the cell, and have manynew mediators to
understand, including CRHR-1 andCRHR-2, their endogenous ligands,
and the CRH bind-ing protein.
The neurobiology of major depression with melan-cholic features
suggests a syndrome reflecting a dysreg-ulation of an essential
adaptive response system calledinto play frequently in all of us,
rather than exoticpathophysiologic changes that are otherwise
rarelyseen. The dysregulation of stress system function
thatinvolves genetic, constitutional, and environmentalfactors
allows the interposition of multiple reinforcingfeedback loops that
are likely to contribute to the clini-cal and biochemical
manifestations of majordepression.
The association of a syndrome characterized by leth-
Figure 10 Early formulation of relation between CRH
anddepression. From Gold et al. Am J Psychiat 1984; 311: 1127.
Molecular Psychiatry
argy, fatigue, apathy, hyperphagia, and hypersomnia,with a
pathologic downregulation of a critical stress-responsive
arousal-producing component of the stresssystem, is of both
practical and theoretical significance.This finding provides new
targets for diagnosis andtherapeutic intervention. In addition, the
definitive dif-ferentiation of depressive syndromes into distinct
bio-chemical phenotypes has implications for moleculargenetic
studies and the search for additional long-termmedical
consequences. In particular, failure to effec-tively differentiate
biochemically distinct subgroups ofpatients with melancholic and
atypical majordepression, on the basis of factors that
differentiallyinfluence susceptibility to medical illness, could
com-plicate the systematic identification of individualswith
depressive illness at risk for cardiovascular dis-ease,
osteoporosis, and inflammatory disease. Inaddition, failure to
stratify depression on these, andhopefully more precisely
identified abnormalities inthe future, could impede scientific
inquiry.
The enhanced susceptibility of patients with majordepression to
the premature onset of complex diseasesthat are frequently seen as
we age suggests that thepathophysiological changes of major
depression,especially with melancholic features, leads to a formof
premature aging. Melancholic patients have thepremature onset of
complex disorders such as coronaryartery disease and osteoporosis
and are likely to diesooner. Other stigmata of aging such as the
more fre-quent awakenings during sleep in the elderly are
alsopresent in melancholic depression. We postulate thatpatients
with melancholic depressive episodes alsohave premature progressive
decrements in growth hor-mone and DHEA secretion.
In a recent screen for gene mutations that extend lifein
drosophila, the mutant line methuselah displayedan approximately
35% increase in average life spanand enhanced resistance to various
forms of stressincluding starvation, high temperature, and free
radicalgeneration.160 Preliminary analysis of the methuselahgene
predicted a protein with homology to the well-known family of seven
transmembrane receptorsinvolved in neurotransmission, endocrine
regulation,and metabolism. In C. elegans, life span and stress
arealso closely associated, and organisms selected forpostponed
senescence also show increased tolerance toheat, starvation, and
oxidative damage.161 Thus, thereis likely to be a price for each
activation of the stressresponse, and a higher price for patients
with an illnessthat involves its long-term activation.
Emotional responses call upon disparate sites inbrain for the
integration of previously encoded memor-ies and their significance,
assessment of the presentreality, and initiation of a relatively
well-rehearsedplan of action and reflexive physiological and
meta-bolic changes. Much of the brain is called into play
toaccomplish these tasks.162 Similarly, the pronouncedchanges in
neuroendocrine function and autonomicoutflow associated with
depression potentially influ-ence an enormous number of somatic
cells and neu-rons. Consequently, previous concepts of depression
as
-
High vs low CRH/NE statesPW Gold and GP Chrousos
272
Molecular Psychiatry
a specific disorder affecting mood have now given wayto an
appreciation of this disorder as a systemic illnessthat exerts
enormous effects on the CNS and periphery.
References
1 Kessler RC, McGonagle KA, Zhao S et al. Lifetime and
12-monthprevalence of DSM-III-R psychiatric disorders in the
UnitedStates. Results from the National Comorbidity Survey. Arch
GenPsychiatry 1994; 51: 8–19.
2 Frank EKD, Perel JM et al. Three year outcomes for
maintenancein therapies in recurrent depression. Arch Gen Psychiat
1990; 47:1093–1099.
3 Gold PW, Goodwin FK, Chrousos GP. Clinical and
biochemicalmanifestations of depression: relation to the
neurobiology ofstress (Part 1 of 2 parts). N Engl J Med 1988; 319:
413–420.
4 Gold PW, Kling MA, Whitfield HJ et al. The clinical
implicationsof corticotropin-releasing hormone. Adv Exp Med Biol
1988; 245:507–519.
5 Gold P, Chrousos G. The endocrinology of melancholic and
atypi-cal depression: relation to neurocircuitry and somatic
conse-quences. Proc Assoc Am Phys 1999; 111: 22–34.
6 Gold PW, Goodwin FK, Chrousos GP. Clinical and
biochemicalmanifestations of depression: relation to the
neurobiology ofstress (Part 2 of 2 parts). N Engl J Med 1988; 319:
348–353.
7 Gold PW, Wong ML, Chrousos GP, Licinio J. Stress system
abnor-malities in melancholic and atypical depression:
molecular,pathophysiological, and therapeutic implications. Mol
Psychiatry1996; 1: 257–264.
8 Frank EaT M. Natural history and preventive treatment of
recur-rent mood disorders. Annu Rev Med 1999; 50: 453–468.
9 Guze S, Robins E. Suicide among primary affective disorders.
BritJ Psychiat 1970; 117: 433–438.
10 Barefoot JC, Schroll M. Symptoms of depression, acute
myocardialinfarction, and total mortality in a community sample
Circulation1996; 93: 1976–1980.
11 Penninx BW, Geerlings SW, Deeg DJ, van Eijk JT, van Tilburg
W,Beekman AT. Minor and major depression and the risk of deathin
older persons. Arch Gen Psychiatry 1999; 56: 889–895.
12 Pratt LA, Ford DE, Crum RM, Armenian HK, Gallo JJ, Eaton
WW.Depression, psychotropic medication, and risk of myocardia