Major Depressive Disorder

T h e n e w e ng l a n d j o u r na l o f m e dic i n e
review article
Mechanisms of Disease
Major Depressive Disorder
R.H. Belmaker, M.D., and Galila Agam, Ph.D.
From Ben Gurion University of the Negev,
Beersheba, Israel. Address reprint requests
Depression is related to the normal emotions of sadness and
bereavement, but it does not remit when the external cause of these emo- to Dr. Belmaker at Beersheba Mental
tions dissipates, and it is disproportionate to their cause. Classic severe states Health Center, P.O. Box 4600, Beersheba,
of depression often have no external precipitating cause. It is difficult, however, to Israel, or at belmaker@bgu.ac.il.
draw clear distinctions between depressions with and those without psychosocial N Engl J Med 2008;358:55-68.
precipitating events.1 The diagnosis of major depressive disorder requires a distinct Copyright (c) 2008 Massachusetts Medical Society.
change of mood, characterized by sadness or irritability and accompanied by at least
several psychophysiological changes, such as disturbances in sleep, appetite, or sex-
ual desire; constipation; loss of the ability to experience pleasure in work or with
friends; crying; suicidal thoughts; and slowing of speech and action. These chang-
es must last a minimum of 2 weeks and interfere considerably with work and fam-
ily relations. On the basis of this broad definition, the lifetime incidence of depres-
sion in the United States is more than 12% in men and 20% in women.2 Some have
advocated a much narrower definition of severe depression, which they call melan-
cholia or vital depression.3
A small percentage of patients with major depression have had or will have manic
episodes consisting of hyperactivity, euphoria, and an increase in pleasure seeking.
Although some pathogenetic mechanisms in these cases and in cases of major depres-
sive disorder overlap, a history of mania defines a distinct illness termed bipolar dis-
order.4
Depression is a heterogeneous disorder with a highly variable course, an inconsis-
tent response to treatment, and no established mechanism. This review presents the
major current approaches to understanding the biologic mechanisms of major de-
pression.
Genetics
Studies comparing concordance rates for major depression between monozygotic and
dizygotic twins suggest a heritability of about 37%,5 which is much lower than the
heritability of bipolar disorder or schizophrenia. Some aspects of the normal person-
ality, such as avoidance of harm, anxiousness, and pessimism, are also partly heritable.6
Kendler et al.7 showed that although depression is due in part to heritable depression-
prone personality traits, it is also the result of heritable factors that are independent
of personality. Early-onset, severe, and recurrent depression may have a higher heri-
tability than other forms of depression.8 It is clear from studies of families that major
depression is not caused by any single gene but is a disease with complex genetic fea-
tures. Studies of pedigrees with multiple cases of major depression have identified
chromosomal regions with linkage to the disorder, and some of these loci have been
replicated in more than one study, although no single chromosomal region has been
replicated in every family study of genetic linkage in depression. Holmans et al.9 found
n engl j med 358;1 www.nejm.org january 3, 2008
55
Downloaded from www.nejm.org by MARY G. MERCURIO MD on January 27, 2008 .
Copyright (c) 2008 Massachusetts Medical Society. All rights reserved.

---

T h e n e w e ng l a n d j o u r na l o f m e dic i n e
evidence of linkage of recurrent, early-onset de- purported deficiency reliably. However, a newly dis-
pression to chromosome 15q25-q26, but the pop- covered form of the enzyme tryptophan hydroxy-
ulation attributable risk was small.
lase, designated TPH-2, is specific to the brain15
No specific molecular risk factor has been reli- and could explain why previous postmortem stud-
ably identified. One common polymorphic variant ies of total enzyme activity did not show differ-
of the serotonin-transporter-linked polymorphic ences in tryptophan hydroxylase activity between
region (5-HTTLPR), which affects the promoter of patients with depression and controls.16 A recent
the serotonin-transporter gene, causes reduced positron-emission tomographic study using a li-
uptake of the neurotransmitter serotonin into the gand for brain monoamine oxidase showed a 30%
presynaptic cells in the brain.10 Some studies have increase of the enzyme in a subgroup of patients
shown that this polymorphism confers a predis- with depression.17 A study measuring differences
position to depression,11 but it also confers a pre- in monoamine metabolites between the internal
disposition to an anxious and pessimistic person- jugular vein and the brachial artery showed lower
ality.10 Brain imaging reveals functional differences production by the brain of norepinephrine metabo-
in emotion-related areas of the brain among car- lites in patients with depression than in controls.18
riers of the different common polymorphisms of The monoamine-deficiency hypothesis continues
5-HTTLPR,12 although a direct relation to depres- to stimulate research whenever a new technical
sion is unclear. In a large, prospective epidemio- window into the brain is opened.
logic study, Caspi et al.13 found that 5-HTTLPR
Serotonin and norepinephrine can be depleted
predicted depression only in association with de- experimentally in humans by oral treatments.19
fined life stresses. Some environmental factors A drink containing all amino acids except trypto-
could confer a predisposition to depression by af- phan stimulates the liver to synthesize proteins
fecting the genome epigenetically -- for example, and rapidly depletes the plasma (and therefore the
increased maternal care in rodents causes an epi- brain) of tryptophan. Tryptophan is rate-limiting
genetic change in the promoter region of the glu- for serotonin synthesis in the brain. Such oral
cocorticoid-receptor gene.14
tryptophan depletion does not induce depression
in healthy subjects but will cause a relapse of de-
The Monoa mine-Deficiency
pression in patients who have been successfully
Hypothesis
treated with a serotonin-reuptake inhibitor.19 Sim-
ilarly, -methyl paratyrosine inhibits tyrosine hy-
The noradrenergic and serotonergic systems orig- droxylase, the rate-limiting step in catecholamine
inate deep in the brain and fan out over almost synthesis. Treatment with -methyl paratyrosine
the entire brain, suggesting a system capable of does not induce depression in normal subjects but
modulating many areas of feeling, thinking, and will induce a relapse in patients who have been
behaving. The early antidepressants blocked the treated successfully with a norepinephrine-reup-
reuptake of norepinephrine and serotonin by the take inhibitor.19 These findings suggest that nor-
presynaptic neuron. The immediate effects of this epinephrine and serotonin have critical roles in the
pharmacologic action are to increase the availabil- mechanisms of these treatments of depression but
ity of norepinephrine and serotonin in the synapse that additional neurochemical factors are neces-
and to increase stimulation of the postsynaptic sary to cause depression.
neuron. Inhibitors of the enzyme monoamine oxi-
Because direct measurements of monoamine
dase were also discovered to have antidepressant neurotransmission did not yield definitive findings
properties. This enzyme catabolizes norepineph- in relation to depression, the downstream effects
rine and serotonin in their respective presynaptic of monoamine neurotransmission were explored
neurons, and such inhibition could be expected to (Fig. 1). The serotonin-1B receptor is located pre-
increase the availability of neurotransmitters. These synaptically and regulates the release of serotonin
discoveries led to a major theory of depression by feedback inhibition. Postmortem studies show
known as the monoamine-deficiency hypothesis. that the levels of p11, a protein that enhances the
Numerous studies of norepinephrine and serotonin efficiency of serotonin-1B receptor signaling, are
metabolites in plasma, urine, and cerebrospinal decreased in the brains of patients with depres-
fluid, as well as postmortem studies of the brains sion.20 The serotonin-1A receptor is located both
of patients with depression, have yet to identify the presynaptically and postsynaptically to regulate
56
n engl j med 358;1 www.nejm.org january 3, 2008
Downloaded from www.nejm.org by MARY G. MERCURIO MD on January 27, 2008 .
Copyright (c) 2008 Massachusetts Medical Society. All rights reserved.

---

mechanisms of disease
serotonin function (Fig. 1). The receptor can be texes of patients who had a major depressive
evaluated in patients with depression by injecting disorder and had not taken antidepressants, as
specific agonists and measuring specific neuro- compared with controls.26,28 Many studies of sec-
endocrine responses, such as elevation of the pro- ond-messenger systems and transcription factors
lactin level.21 Results suggest that the sensitivity in depression were inspired by the belief that it
of this receptor is reduced in patients with depres- takes several weeks before antidepressant treat-
sion.21 The 2-noradrenergic receptor, which is ment has an effect; consequently, the studies were
usually presynaptic, modulates norepinephrine re- designed to detect time-dependent biochemical
lease by feedback inhibition (Fig. 1). Heightened changes in the cell. New meta-analyses suggest
receptor sensitivity has been described in patients that antidepressant effects begin rapidly, howev-
with depression,22 which is consistent with re- er,29 thereby supporting the classic monoamine-
duced norepinephrine release.
deficiency hypothesis.
It is conceivable that the second-messenger sys-
A strong point of the monoamine theory has
tems for serotonergic and noradrenergic neuro- been its predictive power. Almost every compound
transmission malfunction in depression, and for that has been synthesized for the purpose of in-
this reason the phosphatidylinositol and cyclic hibiting norepinephrine or serotonin reuptake has
AMP second-messenger systems have been exten- been proved to be a clinically effective antidepres-
sively evaluated. Reduced inositol levels have been sant. A behavioral model of depression has been
found in postmortem studies of the brains of per- developed in which a rodent is placed in a glass
sons who have died by suicide23 and in magnetic cylinder filled with water, the sheer wall offering
resonance spectroscopic studies of the frontal cor- no chance of escape. The animal struggles for a
tex in patients with depression.24 A blunted cyclic while and then floats passively (the forced swim
AMP response to stimulation was found in post- test). A single prior injection of antidepressant in-
mortem studies of the brains of patients with de- creases the struggling time; results in this model
pression.25 These reductions in second-messenger have excellent predictive validity for new antide-
function may impair neurotransmitter function pressants. Other animal models have been devel-
even without changes in monoamine levels or re- oped by selective breeding of rats for depression-
ceptor numbers. These data indirectly support like behavior, and these genetically susceptible
elaborations of the original monoamine-deficiency rodents also have a response to antidepressants.30
hypothesis of depression (Fig. 1).
Still other models that can be studied biochemi-
G proteins that mediate signaling between re- cally induce depression with the use of long-term
ceptors and second-messenger systems have also mild stress or learned helplessness. However, no
been investigated in patients with depression, both animal model of depression captures the periodic
in postmortem studies of the brain26 and in stud- change of behavior into and out of depression that
ies of peripheral-blood cells.27 Although these is seen in patients with depression.
systems are clearly affected, no consistent picture
Molecular techniques such as gene knockout
has emerged because there are numerous forms partially support the monoamine theory of depres-
of G proteins that vary in different areas of the sion. The serotonin-reuptake-transporter knock-
brain. The cyclic AMP response element-binding out mouse is excessively anxious and characterized
protein (CREB) is a transcription factor affected by by increased immobility in the forced swim test.31
cyclic AMP in the cell. In an animal model of de- This effect is similar to that of the low-activity
pression, rats with overexpression of CREB in the polymorphic variant of the serotonin receptor on
dentate gyrus behaved similarly to rats treated human personality10 but is the opposite of the ex-
with antidepressants, but the opposite effect was pected effects of serotonin-reuptake-inhibitor an-
found when CREB was overexpressed in the nu- tidepressants. However, this inconsistency could
cleus accumbens.26,28 Thus, the role of CREB in be explained by the difference between a chron-
depression is specific to the region of the brain. ic monoamine abnormality during brain develop-
Most but not all studies show that long-term treat- ment31 and the hypothesized acute monoamine
ment with antidepressants stimulates CREB func- depletion in an adult with depression. Table 1
tion, possibly depending on the type of drug and shows the effects in mice of knocking out genes
the dosage.28 Levels of CREB and phospho-CREB related to monoamine neurotransmitters.
were reduced in postmortem studies of the cor-
The effects of stimulants on mood indirectly
n engl j med 358;1 www.nejm.org january 3, 2008
57
Downloaded from www.nejm.org by MARY G. MERCURIO MD on January 27, 2008 .
Copyright (c) 2008 Massachusetts Medical Society. All rights reserved.

---

T h e n e w e ng l a n d j o u r na l o f m e dic i n e
support the monoamine-deficiency hypothesis of
Figure 1 (facing page). The Monoamine-Deficiency
depression and show that mood can be altered
Hypothesis Extended.
rapidly. Cocaine and amphetamines are powerful
The monoamine hypothesis of depression postulates
releasers of monoamines into the synapse as well
a deficiency in serotonin or norepinephrine neurotrans-
as inhibitors of reuptake. Their mood-elevating
mission in the brain. Monoaminergic neurotransmis-
effects are immediate, but in patients with severe
sion is mediated by serotonin (5-hydroxytryptamine
depression they have often been reported to cause
1A [5-HT1A] and 5-hydroxytryptamine 1B [5-HT1B]) or
norepinephrine (noradrenaline) released from presyn-
agitation rather than relief of depression. This
aptic neurons (serotonergic neuron, shown on the left
finding could reflect the ability of these stimulants
side, and noradrenergic neuron, shown on the right
to deplete the presynapse of monoamines and thus
side [condensed virtually]). Serotonin is synthesized
cause a "crash" into depression. Recent studies
from tryptophan, with the first step in the synthetic
support the theory that an acute response to a
pathway catalyzed by tryptophan hydroxylase; norepi-
nephrine is synthesized from tyrosine, with the first
single dose of amphetamine predicts a patient's
step catalyzed by tyrosine hydroxylase. Both mono-
longer-term response to monoamine-reuptake in-
amine transmitters are stored in vesicles in the presyn-
hibitors.46
aptic neuron and released into the synaptic cleft, there-
The role of dopamine deficiency in depression
by affecting both presynaptic and postsynaptic
is suggested by the frequency of depression in pa-
neurons. Cessation of the synaptic action of the neu-
rotransmitters occurs by means of both reuptake
tients with Parkinson's disease and the effect of
through the specific serotonin and norepinephrine
reserpine, which depletes serotonin, norepineph-
transporters and feedback control of release through
rine, and dopamine, causing a hypoactive state in
the presynaptic 5-HT1A and 5-HT1B regulatory autore-
animals. The antidepressant agent buproprion in-
ceptors for serotonin and the 2-noradrenergic autore-
hibits the reuptake of dopamine. Some direct do-
ceptors for norepinephrine. Monoamine oxidase A
(MAO-A) catabolizes monoamines presynaptically and
pamine-receptor agonists, such as pramipexole,
thereby indirectly regulates vesicular content. The pro-
have been reported to be efficacious in the treat-
tein p11, which interacts with 5-HT1B receptors, in-
ment of depression, even though they were devel-
creases their function. Postsynaptically, both serotonin
oped for Parkinson's disease.47
and norepinephrine bind two kinds of guanine nucleo-
A major liability of the monoamine-deficiency
tide triphosphate-binding protein (G protein)-coupled
receptors: cyclic AMP (cAMP)-coupled receptors,
hypothesis is its derivation from the mechanism
which activate adenylate cyclase (AC) to generate
of currently available antidepressants. Approxi-
cAMP, and phosphatidylinositol (PI)-coupled recep-
mately two thirds of patients have a clinical re-
tors, which activate phospholipase C (PLC). PLC gener-
sponse to these agents, whereas one third have a
ates inositol triphosphate (IP3) and diacylglycerol
response to placebo.48 Perhaps the mechanism of
(DAG); cAMP activates protein kinase A (PKA), and IP3
and DAG activate protein kinase C (PKC). The two pro-
depression is not related to monoamines in two
tein kinases affect the cAMP response element-bind-
of three cases.
ing protein (CREB). Findings in patients with depres-
sion that support the monoamine-deficiency
hypothesis include a relapse of depression with inhibi-
S tr ess, the Hypothal a mic -
tion of tyrosine hydroxylase or depletion of dietary
Pituitary-Adr enal A xis,
tryptophan, an increased frequency of a mutation af-
and Grow th Factor s
fecting the brain-specific form of tryptophan hydroxy-
lase (TPH-2), increased specific ligand binding to
Stress49 is perceived by the cortex of the brain and
MAO-A, subsensitive 5-HT1A receptors, malfunction-
transmitted to the hypothalamus, where cortico-
ing 5-HT1B receptors, decreased levels of p11, poly-
tropin-releasing hormone (CRH) is released onto
morphisms of the serotonin-reuptake transporter asso-
ciated with depression, an inadequate response of G
pituitary receptors. This stimulus results in the se-
proteins to neurotransmitter signals, and reduced lev-
cretion of corticotropin into plasma, stimulation
els of cAMP, inositol, and CREB in postmortem brains.
of corticotropin receptors in the adrenal cortex,
and release of cortisol into the blood. Hypothalam- CRH levels in cerebrospinal fluid,50 and increased
ic cortisol receptors respond by decreasing CRH levels of CRH messenger RNA and protein in
production to maintain homeostasis (Fig. 2).
limbic brain regions.50 In studies using dexa-
There is considerable evidence that cortisol and methasone to evaluate the sensitivity of the hypo-
its central releasing factor, CRH, are involved in thalamus to feedback signals for the shutdown of
depression.50,51 Patients with depression may have CRH release, the normal cortisol-suppression re-
elevated cortisol levels in plasma,38 elevated sponse is absent in about half of the most se-
58
n engl j med 358;1 www.nejm.org january 3, 2008
Downloaded from www.nejm.org by MARY G. MERCURIO MD on January 27, 2008 .
Copyright (c) 2008 Massachusetts Medical Society. All rights reserved.

---

mechanisms of disease
verely depressed patients.52 Antidepressant-induced activity of the hypothalamic-pituitary-adrenal axis
clinical remission is accompanied by reversal of and increased immobility in the forced swim test,
some of these abnormalities.52
both of which are reversed by antidepressants.55
Adults with a history of physical or sexual Increased levels of monoamines in the synapse
abuse as children have increased levels of CRH in affect the hypothalamic-pituitary-adrenal axis56
cerebrospinal fluid.53 Adult rodents that were sepa- and reverse some of the long-term effects of
rated from their mothers or abused as pups show stress.56 It is possible that antidepressants relieve
increased immobility in the forced swim test, depression by reducing the secondary stress caused
which is reversed by antidepressant treatment.54 by a painfully dispirited mood rather than by di-
Mice with region-specific knockout of the gluco- rectly elevating mood. An antistress mechanism
corticoid receptor at an adult age have increased could explain the general usefulness of antidepres-
n engl j med 358;1 www.nejm.org january 3, 2008
59
Downloaded from www.nejm.org by MARY G. MERCURIO MD on January 27, 2008 .
Copyright (c) 2008 Massachusetts Medical Society. All rights reserved.

---

T h e n e w e ng l a n d j o u r na l o f m e dic i n e
Table 1. Monoamine-Related Gene Knockouts That Affect Depression-Related Behavior in Mice.*
Corroboration
of Monoamine-
Deficiency
Other Behavior Elicited by
Gene or Protein
Function
Depression-Related Changes
Hypothesis
Knockout of Gene
sert
Serotonin transporter
Increased depressive behavior, reduced se-
No
Excessive anxiety32
rotonin level, desensitized postsynaptic
5-HT1AR, and reduced presynaptic
5-HT1AR function32
net
Norepinephrine transporter Reduced depressive behavior, prolonged
Yes
Increased locomotion response
norepinephrine clearance, elevated ex-
to amphetamines and co-
tracellular norepinephrine levels33
caine33
5-ht1ar
Serotonergic 1A receptor
Reduced depressive behavior, normal sero-
No
Excessive anxiety,
(presynaptic autorecep-
tonin level and release, impaired SSRI-
impaired hippocampal learn-
tor and postsynaptic)
induced neurogenesis32
ing32
5-ht1br
Serotonergic 1B receptor
Reduced response to SSRI in forced swim
Yes
Increased aggressiveness, re-
(presynaptic autorecep-
test, reduced serotonin level and in-
duced anxiety, increased ex-
tor and postsynaptic)
creased serotonin release, increased
ploration, increased use of
SSRI-induced serotonin release, de-
cocaine32
creased serotonin-transporter expres-
sion32
p11 (protein)
Interacts with and enhanc-
Increased depressive behavior, increased
No
Not reported20
es signaling efficiency
serotonin turnover20
of 5-HT1BR
5-ht2ar
Serotonergic 2A receptor
No change34
No
Reduced inhibition in conflict-
anxiety paradigms34
5-ht7
Serotonergic 7 receptor
Reduced depressive behavior and REM
No
Normal locomotion35
(possibly presynaptic
sleep duration35
autoreceptor and post-
synaptic)
2aar
2A-Adrenergic receptors
Reduced norepinephrine levels, presynaptic
No
Altered sympathetic regula-
(presynaptic autorecep-
inhibition of release,36 increased depres-
tion,36 impaired motor coor-
tor)
sive behavior37
dination
2car
2c-Adrenergic receptors
Reduced depressive behavior38
Yes
Increased aggressiveness,32
(presynaptic autorecep-

increased locomotion re-
tor restricted to central
sponse to amphetamines36
nervous system)
mao-a
Monoamine oxidase A
Increased brain serotonin and epinephrine
No
Increased aggressiveness and
levels39
response to stress,30 de-
creased exploration32
ac VII (hetero-
Adenylyl cyclase type 7
Reduced depressive behavior40
No
Unchanged anxiety40
zygotes)
impa1
Inositol monophos-
Reduced depressive behavior, unaltered
Yes
Increased hyperactivity and
phatase 1
brain inositol levels41

sensitivity to pilocarpine-

induced seizures41
smit1
Sodium-myo-inositol trans- Reduced depressive behavior and brain ino-
Yes
Increased sensitivity to pilocar-
porter 1
sitol levels42
pine-induced seizures42
creb
Cyclic AMP-response ele-
Reduced depressive behavior, normal anti-
No
No increase in BDNF after long-
ment-binding protein
depressant-induced behavior43
term use of antidepres-
sants43
bdnf
Male mice
Brain-derived neurotrophic No depressive behavior44
No
Increased aggressiveness, hy-
factor
perphagia,45 hyperactivity44
Female mice Brain-derived neurotrophic Increased depressive behavior44
Yes
Increased aggressiveness, hy-
factor
perphagia45
* BDNF denotes brain-derived neurotrophic factor, 5-HT1AR 5-hydroxytryptamine 1A receptor, 5-HT1BR 5-hydroxytryptamine 1B receptor,
REM rapid eye movement, and SSRI selective serotonin-reuptake inhibitor.
60
n engl j med 358;1 www.nejm.org january 3, 2008
Downloaded from www.nejm.org by MARY G. MERCURIO MD on January 27, 2008 .
Copyright (c) 2008 Massachusetts Medical Society. All rights reserved.

---

mechanisms of disease
Figure 2. The Hypothalamic-Pituitary-Cortisol System in Depression.
The hypothalamic-pituitary-cortisol hypothesis of depression postulates that abnormalities in the cortisol response
to stress may underlie depression. The black arrows show that in response to stress, which is perceived by the brain
cortex and the amygdala and transmitted to the hypothalamus, corticotropin-releasing hormone (CRH) is released,
inducing the anterior pituitary gland to secrete corticotropin into the bloodstream. Corticotropin stimulates the ad-
renal cortexes to secrete the glucocorticoid hormone cortisol. The red lines show that cortisol, in turn, induces feed-
back inhibition in the hypothalamus and the pituitary, suppressing the production of CRH and corticotropin, respec-
tively. Findings in patients with depression that support the hypothalamic-pituitary-cortisol hypothesis include the
following: cortisol levels are sometimes increased in severe depression, the size of the anterior pituitary and adrenal
cortex is increased, and CRH levels in the cerebrospinal fluid and CRH expression in the limbic brain regions are in-
creased. Hippocampal size and the numbers of neurons and glia are decreased, possibly reflecting reduced neuro-
genesis due to elevated cortisol levels or due to reduced brain-derived neurotrophic factor.
sants for a wide variety of psychiatric conditions, be efficacious in depression, but only the most
including panic disorder, post-traumatic stress severe and psychotic type.58
disorder, bulimia, premenstrual syndrome, and
A single test for the cortisol level in blood does
obsessive-compulsive disorder. CRH-receptor an- not contribute to the diagnosis of depression, since
tagonists show antidepressant activity in animal levels of cortisol vary markedly in a circadian
models,57 but the results of large clinical trials rhythm38 and because the overlap between values
have been disappointing. A compound that blocks in patients and those in controls is considerable.
the glucocorticoid receptor has been reported to Mild stress induced in the laboratory, such as
n engl j med 358;1 www.nejm.org january 3, 2008
61
Downloaded from www.nejm.org by MARY G. MERCURIO MD on January 27, 2008 .
Copyright (c) 2008 Massachusetts Medical Society. All rights reserved.

---

T h e n e w e ng l a n d j o u r na l o f m e dic i n e
-
-
-
97
-
de
in
de
-
symp
brain
in

lacking
the
in
treating
MRS
inconsis
neurotrans
replicated
in
nonspecificity
in are
reduced
95
on
is
not
anxiety
trials
levels
of
effective
and
92,93
modulate
receptors
receptors
suggesting
be
depression
that
antagonist,
GABA
sleep
depression
AMPA
may
symptoms
and
in
brain and
randomized
temporary
Evidence
increased
synapses,
cortex
to
107
receptor
are
dopamine
are
affect
genes
the
D2
in
opiates
brain
memory
cortex
related
similar
of
that controlled,
therapeutic
prefrontal
be
are
anticholinergic
symptoms
not
Contradictory
30%
in
steroids affect
116
may
acetylcholine
clock-related
large,
not
are
occipital
antidepressants
than
suggested
113
are
the
mostly
from
high-affinity-state
in
nicotinic
depressive
to
more
difference
between
correlative
schizophrenia
a
(neuroactive
reports data
whether
in
of
in
110
are
117
antagonists
levels
99
depression
binds
lack
receptors)
early
P
clear
a
of
occurs
is
neurotransmission
103
antidepressants
studies
studies
not
pression
depression
association tent
findings
mitter
pression,
toms
all
is
Glutamate
Ketamine
It
GABA
There
GABA
The
The
Neurosteroids
Although
Mecamylamine,
Many
Most
Cytokine-induced
Substance
-
-
to
anti
-
114
98
sleep,
and
91
enzyme
106
-
100
short
secretion
the
physostig
mood,
is

115
of
prefrontal
101,102
rodents
response
of
of
mono
reduced
sustained
108
depression
cortex
in
in
and
depression
cytokine
are
of
effects
daylight
and
rapid,

antidepressants
diseases
96
effects
and
axis
dorsolateral
cortex
during
models
prefrontal
when
depression
decreased
104
is
administration
induces
the 91-93
the
abnormalities
115
transporters reduced
by
in
brain
with
potentiate
are
98
fluid,
109
112
autoimmune
symptoms,
Evidence
animal
antidepressant
the
prefrontal
effects
reduced
in
depressed
binding

circadian secretion
and
patients
brain
humans
and
115
Depression.*
the
have
antagonist,
glutamate
are
function
reduced
in
antidepressant-like the
in inhibitor
of
in
of
is
antagonists
glutamine
effects
have
become
in
depressive
Supporting
to
cortex
plasma
have
levels
NMDA
cerebrospinal
therapy
effects
levels
have
day
in
infectious
induced
depression
an
BDNF
GABAergic
in
light
the
receptor
induces
94
neuroendocrine
low
of
111
be
antiinflammatory
RNA
-opioid-receptor
occipital
agents
depression
agonists
hypothalamic-pituitary-adrenal
glutamate
plasma,
major
affect
and
and
are
can
in
have
the
Pathophysiology
glutamine
effect
in the
immunoreactivity
with
during
levels
sadness
common
ketamine,
cortical
acetylcholinesterase
cytokines
the
and
and
levels
antidepressant
mood
is
an
to
affect 115
of
acetylcholine
messenger converts
GABA
active 105
for
of
neuron
deprivation
patients
has
increased
depressant
that
cortex,
temperature,
ened
up-regulate
sustained
mine,
is
amines
Theories
Glutamate
Intravenous
Cortical
Levels
GABA-modulating
Antidepressants
GABA
Sleep
Some
Rodents
Cholesterol
DHEA
-Opioid-receptor
Capacity
Depressed
Nicotinic
Depression
Exposure
Antidepressants
Cytokines
-

opi
-

the
Biologic
cross- imthe and
GABAergic
circadian
neurosteroid
endogenous
system
glutamatergic
function
between
neurotransmission
neurotransmission
rhythms
synthesis
oid
imbalance
talk mune brain
T
a
b
l
e

2
.

Additional
Theory
Altered
Reduced
Abnormal
Deficient
Impaired
Monoamine-acetylcholine
Cytokine-mediated
62
n engl j med 358;1 www.nejm.org january 3, 2008
Downloaded from www.nejm.org by MARY G. MERCURIO MD on January 27, 2008 .
Copyright (c) 2008 Massachusetts Medical Society. All rights reserved.

---

mechanisms of disease
stress associated with mental arithmetic calcula-
-
tions or simulated public speaking, results in
greater changes in plasma cortisol levels than
utiliza
resonance
most reported differences between the values in
patients with depression and those in controls.38
glucose
magnetic
It is possible that chronic mild elevations of cor-
depression
tisol, especially at night, when cortisol levels in
MRS
with
normal subjects are very low, have a pathogenic
volumetric,
acid,
role in depression. It is also possible that periph-
eral cortisol elevations are only a reflection of
patients
study
flow,
to
63,124,126
central disturbances in CRH signaling, which me-
most
blood
diate the effects of environmental stress on mood.59
in
study
to
-aminobutyric
A major liability of the hypothalamic-pituitary-
BA
adrenal axis theory of depression is the difficulty
from
GA
ineffective
respect methodologies
of defining the relationship of stress to depres-
is
manifested
differ
sion. Some patients have a single lifetime depres-
with
not
sive episode, whereas a larger proportion have a
is
areas
recurrent or even chronic course. Various types
findings postmortem
brain
of acute stress, early childhood trauma, or long-
monotherapy
and
term psychosocial problems may be involved and
dehydroepiandrosterone,
tion,
may lead to different responses of the stress sys-
Thyroxine
Hypothyroidism
Implicated
Inconsistent
tem. Stress may be causative in some cases and
DHEA
secondary to depressed mood in others.
1
2
7
with
Severe stress in rodents does not necessarily
factor,
in
121
model the common stresses of childhood. The
119
deep-brain
prefrontal
association of abuse in childhood with psycho-
patients
in

and
pathologic disorders, including depression, in
brain
thyroxine
of
122
neurotrophic
adulthood could be due to common factors link-
fluid
the
depression
o
d
e
l

o
f

d
e
p
r
e
s
s
i
o
n
in
subgenual
ing family perpetrators of abuse and their victims,
cortex 123

and
including not only shared genes but also a shared
during
system
mood
124
brain-derived
environment of poverty, poor nutrition, and poor
prenatal care. Depression is not uncommon in
cerebrospinal
administration
prefrontal
cortex
increased
affect
BDNF
people with no psychosocial risk factors. Most
the
the
is
the
in
of
acid,
patients treated for depression have no evidence
serotonergic
after 120
p
u
s

a
r
e

a
l
t
e
r
e
d

i
n

a

r
a
t

m
of hypothalamic-pituitary-adrenal dysfunction,
the
cingulate
prefrontal
just as most such patients have no direct evidence
reduced
the
propionic
in
of brain monoamine deficiency.
are
decreased
stimulation anterior
The classic teaching is that neurons do not di-
modulate
is
triiodothyronine
hypothyroidism
to
the
vide in the adult mammalian brain, but studies
of
118
reduced
with
magnetic
is
i
c
s

i
n

t
h
e

h
i
p
p
o
c
a
m
acid.
have shown that neurogenesis occurs in several
transthyretin
rats
125
areas of the brain, especially the hippocampus.
hormones
use
of
response
neurogenesis
Neurogenesis is more prominent in rodents than
of
-aspartic
depression
adult
stimulation
cortex
d
in primates,60 and some have questioned whether
i
r
c
u
i
t

d
y
n
a
m
Levels
Thyroid
Brain
Rate
Transcranial
Glucose
C
it occurs in the human cortex.61 Elevated levels of
-methyl-
N
glucocorticoids can reduce neurogenesis, and this
and
has been suggested as a mechanism for the de-
NMDA
creased size of the hippocampus on magnetic
specific
of
alpha-amino-3-hydroxy-5-methyl-4-isoxazole and
resonance images of the brain in many patients
abnormalities
structures
with depression.62 In postmortem studies of pa-
denotes
tients with depression, cell loss in the subgenual
brain circuits
prefrontal cortex, atrophy in the dorsolateral pre-
Thyroxine
Dysfunction
AMPA spectroscopy,
*
frontal cortex and the orbitofrontal cortex, and
n engl j med 358;1 www.nejm.org january 3, 2008
63
Downloaded from www.nejm.org by MARY G. MERCURIO MD on January 27, 2008 .
Copyright (c) 2008 Massachusetts Medical Society. All rights reserved.

---

T h e n e w e ng l a n d j o u r na l o f m e dic i n e
increased numbers of cells in the hypothalamus related to depression.45 Reduced BDNF levels in
and the dorsal raphe nucleus have been reported.63 the peripheral blood of patients with depression
These effects resemble the atrophic changes in the seem to derive almost entirely from blood plate-
brain in patients with Cushing's disease64 and in lets,80 and many artifacts must therefore be con-
rodents treated with glucocorticoids.65 However, sidered in interpreting these findings. Inflamma-
cortisol elevations in depression are much lower tion in the brain and some neurotoxins increase
than in Cushing's disease.
brain BDNF levels, suggesting that the actions of
Restraint in a small container induces stress in BDNF are not uniformly therapeutic.81 Castren82
rodents, suppressing neurogenesis, and this effect has proposed that antidepressant treatments may
is countered by antidepressant treatment.66 An- increase synaptic sprouting and allow the brain
tidepressants also enhance neurogenesis in non- to use input from the environment more effec-
human primates.67 Santarelli et al.68 irradiated the tively to recover from depression. This hypothe-
hippocampus in mice and abolished neurogene- sis highlights the role that cognition may play in
sis. They found that the radiation also abolished depression and suggests that biochemical mech-
the ability of the animals to respond behaviorally anisms may be nonspecific.
to antidepressant treatment in the forced swim
Strong epidemiologic data point to an associa-
test, but this phenomenon does not occur in every tion between major depressive disorder and in-
mouse strain studied.69 Henn and Vollmayr sum- creased cardiovascular morbidity and mortality.83
marized other studies providing evidence that de- In many patients, cardiovascular disorders precede
creased neurogenesis is a result of stress and depression, and in others, depression precedes the
anxiety but may not be behaviorally relevant.70 The cardiovascular disorder. Both n-3 fatty acid defi-
relevance of animal models of neurogenesis to ciency84 and elevated plasma homocysteine levels85
clinical studies of depression has been questioned have been implicated in cardiovascular disease and
by analogy with studies of neuroprotection strat- in depression. Elevated cortisol levels in depres-
egies in stroke, for which numerous findings in sion could increase the risk of coronary artery
animal models have not been replicated in hu- disease, since cortisol increases visceral fat.64,86
man studies.71
Antidepressant treatment increases the survival
Brain-derived neurotrophic factor (BDNF), rate among patients who become depressed after
a neurotrophic peptide, is critical for axonal coronary occlusion.86 Endothelial-cell signaling
growth, neuronal survival, and synaptic plastic- plays a crucial role in brain neurogenesis,87 and
ity,72 and its levels are affected by stress73 and these cells secrete BDNF; thus, both depression
cortisol.74 A postmortem study of patients with and cardiovascular disease could be examples of
depression who had committed suicide showed an endothelial disorder. Signs of inflammatory
that BDNF was reduced in the hippocampus.75 processes have been described in major depres-
Antidepressant drugs and electroconvulsive thera- sion88 and in cardiovascular disease. Some data
py up-regulate BDNF and other neurotrophic and suggest that exercise has protective or therapeu-
growth factors75,76; a single bilateral infusion of tic effects in depression.89 Rodent models support
BDNF into the dentate gyrus has antidepressant- this possibility.90
like effects.77 One study showed that the hippo-
campus was smaller than normal in patients with
Other Possible Mechanisms
depression who carried a met166 BDNF allele.78
In an animal model of depression, epigenetic his- Table 2 summarizes possible pathophysiological
tone methylation mediated down-regulation of mechanisms of depression other than those based
BDNF transcripts and antidepressant treatment on the monoamine-deficiency hypothesis or the
reversed this effect.79 These studies suggest that roles of stress, cortisol, and neurogenesis. Many
BDNF is the link among stress, neurogenesis, and of these other proposed mechanisms have also
hippocampal atrophy in depression. However, been implicated in psychiatric and neurologic dis-
a genetic association of the BDNF val166met poly- orders other than depression. Since the compo-
morphism with depression has not been replicated nents of the brain are highly interconnected, it is
in most studies,74 and BDNF may be related not not difficult to find possible integrative frame-
only to depression but to multiple psychiatric dis- works between two or more of the various theo-
orders.74 BDNF-knockout mice have behaviors un- ries. Testing the theories in a manner that can re-
64
n engl j med 358;1 www.nejm.org january 3, 2008
Downloaded from www.nejm.org by MARY G. MERCURIO MD on January 27, 2008 .
Copyright (c) 2008 Massachusetts Medical Society. All rights reserved.

---