Treatment with Antidepressants

The modern history of antidepressants began with the introduction of the tricyclic antidepressants (TCAs) and monoamine oxidase inhibitors (MAOIs) in the 1950s. Although these agents are now usually fallback options when newer generation drugs fail, the study of their pharmacology led to the catecholamine and indoleamine theories of depression, which are still the basis for most antidepressant development.

The therapeutic properties of TCAs and MAOIs were discovered by chance and careful clinical observation. Imipramine is a derivative of chlorpromazine and was originally synthesized as a possible antipsychotic, but Kuhn (1) found that it benefited only depressed schizophrenic patients. This observation prompted him to successfully test it in patients suffering from melancholia. Iproniazid was developed as an antitubercular drug, and the observation of euphoria as an adverse effect led Crane (2) to conduct successful trials in depressed patients. A year later, Kline (3) followed up on this observation and reported positive results in another early antidepressant trial.

Paralleling these clinical developments were basic pharmacological studies noting that reserpine (4,5 and 6) and α-methyldopa produced depression in patients treated for hypertension (7,8 and 9). The fact that the TCAs and MAOIs functionally increased norepinephrine (NE) activity, while reserpine lowered its activity leading to the NE hypothesis of depression (10,11). This same line of reasoning was also applied to serotonin (5-HT) (12,13).

Given two effective classes of antidepressants, pharmacologists developed animal models to screen new compounds in an attempt to predict efficacy. Everett (14) developed the dopa test, and others found that antidepressants reversed reserpine- or tetrabenazine-induced sedation in rodents (15). The learned helplessness test is another paradigm in which an animal is put in an impossible situation and eventually gives up (16). For example, a dog repeatedly shocked and unable to escape gives up trying, even though an exit is subsequently made available to him. A similar model is a test in which animals are dropped into a tank of water (17). At first they actively try to escape by swimming for a time, but then they give up and just float. Antidepressants cause the animals in both paradigms to struggle longer before capitulating. NE and 5-HT were hypothesized to be the mediating neurotransmitters in these behavioral models.

Because many have suggested that the beneficial effect of antidepressants is based on their ability to block the transport mechanisms of NE and 5-HT, companies screen potential candidates for this ability. Partially as a result of this paradigm, the pharmaceutical industry has developed agents that can specifically block NE uptake, 5-HT uptake, or both. Most selective NE reuptake inhibitors, however, have failed in clinical trials (e.g., reboxetine), calling into question whether this mechanism alone can produce a clinically relevant antidepressant effect.

The next section provides a brief review of our understanding of the pathophysiology of major depression. The goal is to provide a basis for conceptualizing the mechanism of action of currently available antidepressants (Table 7-1).

TABLE 7-1 ANTIDEPRESSANT AGENTS

Class/Generic Name		Trade Name	Usual Daily Dosage (mg/day)
SSRI
	Citalopram	Celexa	20-40
	Escitalopram	Lexapro	10-20
	Fluoxetine	Prozac	10-60
	Fluvoxamine^a	Luvox	100-300
	Paroxetine	Paxil	10-50
	Sertraline	Zoloft	50-200
Selective NRI
	Atomoxetine^a	Strattera	60-120
SNRI
	Duloxetine	Cymbalta	30-60
	Milnacipran^b		100-200
	Venlafaxine	Effexor	75-375
	Desvenlafaxine	Pristiq	50-100
Aminoketone
	Bupropion	Wellbutrin	150-450
Triazolopyridine
	Nefazodone	Serzone^c	100-600
	Trazodone	Desyrel	150-600
Tetracyclic
	Amoxapine	Ascendin	200-600
	Maprotiline	Ludiomil	75-225
	Mirtazapine	Remeron	15-45
TCA
	Amitriptyline	Elavil	75-300
	Clomipramine	Anafranil	100-250
	Desipramine	Norpramine	75-300
	Doxepin	Sinequan	75-300
	Imipramine	Tofranil	75-300
	Nortriptyline	Pamelor	75-300
	Protriptyline	Vivactil	20-60
	Trimipramine	Surmontil	75-300
MAOI
	Isocarboxazid	Marplan	40-60
	Phenelzine	Nardil	30-90
	Selegiline^d	Emsam	20 mg/20 cm² patch
	Tranylcypromine	Parnate	30-60
^aNot approved by the FDA for depression.^bNot available in the United States.^cSerzone no longer available; generic form of nefazodone still available.^dTransdermal system approved for depression.
MAOI, monoamine oxidase inhibitor; SNRI, selective norepinephrine reuptake inhibitor; SSRI, selective serotonin reuptake inhibitor; TCA, tricyclic antidepressant

Pharmacodynamics

NEUROBIOLOGY OF DEPRESSION

Our understanding of the neurobiology of depression (e.g., anatomy, chemistry, endocrinology, immunology, and genetics) parallels our increased understanding of antidepressant pharmacology. Indeed, hypotheses regarding the biological mechanisms subserving mood disorders are developed from observations on the clinical effects of drug and other therapies in humans, as well as drug-induced behavioral changes in animals (18). For the purpose of discussion, current theories can be categorized as

Genetic
- Pharmacogenetics
Neuroanatomical
Neurotransmitter and related hypotheses
- Monoamine
- Interactional
- Second messenger
Membrane and cation
Biological rhythms
Neuroendocrine
Immunological

We emphasize that these theories are not mutually exclusive. Thus, a genetically determined membrane defect could produce a dysregulation in the neurotransmitter-receptor interaction. This, in turn, may impact subsequent neuronal messenger systems within specific neural networks, resulting in a disturbance in relevant neurocircuitry and biological rhythms such as neuroendocrine function.

Genetic Hypothesis

The evidence available from family, twin, and adoption studies supports genetic factors in the development of primary mood disorders. There is clearly a higher incidence of both bipolar and apparent unipolar disorders in first-degree relatives of bipolar patients. We say “apparent” because “unipolar” patients from bipolar families may not yet have experienced a manic phase. Families of unipolar patients, however, show a higher incidence of unipolar but not bipolar disease. That unipolar and bipolar disorders are inherited separately suggesting there are at least two variations of mood disorder. We do not yet know where the genes for these disorders are located or how they translate into affective dysregulation. Linkage studies using recombinant DNA techniques and genomewide association studies (GWAS) can examine several possible loci for susceptibility to mood disorders (19,20).

One approach to finding a biological process that may identify a specific genotype is to look for differences in various enzymes (e.g., MAO, catechol-O-methyltransferase, and dopamine β-hydroxylase [DBH]). For example, single nucleotide polymorphisms (SNPs) in COMT were associated with response in duloxetinetreated patients in one recent trial (21). In another example, DBH is critical to the conversion of dopamine (DA) to NE. In this context, the homozygous GG genotype for DBH may be more protective against psychosis in major depressive disorder (MDD) than the AA or AG genotype. The assumption is that the GG genotype may be more efficient in converting DA to NE, thus diminishing the risk for a hyperdopaminergic state (22). Alternatively, nongenetic factors could lower DBH expression, predisposing to psychosis in unipolar MDD (23). Genetic variants of brain-derived neurotropic factor (BDNF) are also associated with susceptibility to MDD (24,25). Further, there are data indicating the agents such as venlafaxine may exert antidepressant effects by stimulating BDNF expression (26).

Both unipolar and bipolar disorders are often recurrent and progressive. Post (27) postulated that early episode, stress-related, alterations in gene expression may subserve long-lasting changes in stress responsivity, episode sensitization, and differences in pharmacosensitivity as a function of the longitudinal course of an illness. He further proposes that adequate drug prophylaxis may interrupt the phenomenon of “episodes begetting episodes,” but that subsequent exacerbations may overwhelm or circumvent previously effective treatments.

The potential interaction between genetics and environment was highlighted in an important paper by Caspi et al. (28). They found that a functional polymorphism in the promoter region of the serotonin transporter (5-HTT) gene could alter mood response to stressful life events (SLEs). Specifically, patients with two short alleles (SS) for the 5-HTT promoter polymorphism manifested more depressive symptoms, diagnosable depression, and suicidal features compared to those who were homozygous for the long allele. This discrepancy, however, was only evident when there was also a recent history of significant external stressors. They concluded that genetic makeup interacts with the environment to determine affective response to stress. This finding was subsequently replicated using different analytic methods, different measures of SLEs, and with a greater degree of temporal resolution (29). The authors note that the findings from these two studies suggest that the 5-HTT gene influences “major depression not by a main effect on risk but rather by control of sensitivity to the pathogenic effects of the environment.”

More recently, however, a meta-analysis did not support an association between this 5-HT transporter genotype (alone or in interaction with SLEs) and an increased risk of depression (30).

Pharmacogenetics. As noted in Chapter 2, pharmacogenetics is an important variable determining the biological state of a given patient, which in turn can affect response to a drug (i.e., efficacy, safety, and/or tolerability). For example, there is early evidence that individualized dose recommendations for antidepressants can be based on whether a patient is an extensive, intermediate, or poor metabolizer as determined by one’s genotype for the cytochrome P450 (CYP) 2D6 and 2C19 isoenzymes (31). Such data may also complement therapeutic drug monitoring (TDM) to improve outcome (32). In another example, genotype variants in the 5-HT_2A locus (HTR2A) may produce pharmacodynamically mediated differences in adverse effects related to paroxetine-treated MDD in elderly patients (33). The short allele in the promoter region of the serotonin transporter gene (SLC6A4) may predict more severe adverse effects in patients treated with paroxetine (34), and NE transporter gene polymorphisms (but not 5-HTT polymorphisms) may predict response to drugs such as citalopram (35,36). GWAS studies may also help predict response to specific antidepressants. A recent example from the Sequenced Treatment Alternatives to Relieve Depression (STAR*D) data identified three possible SNPs associated with response and remission to citalopram (36). Another example comes from the Genome-Based Therapeutic Drugs for Depression (GENDEP) project. One report from this group indicates that a genetic variation in noradrenergic signaling predisposes males on nortriptyline to experience suicidality (37,38).

A recent meta-analysis suggests that up to six candidate genes may modulate antidepressant response (39).

Dantz (2009) notes, however, that clinical application of such pharmacogenetic findings requires more than statistical association (40). Thus, study populations must be homogeneous; positive and negative predictive values must be sufficiently high; and the results justify the costs of testing.

Neuroanatomical Hypotheses

Increasingly, there is an emphasis on dysregulation in neurocircuits, which constitute substrates for various psychiatric disorders, including depression (41). In this context, there are important advances in anatomical localization with imaging studies as well as focused postmortem studies of the brains of patients with unipolar and bipolar depression. For example, neuroimaging studies of patients with familial pure major depression identify neurophysiological abnormalities in multiple areas of the orbital and medial prefrontal cortex (PFC), the amygdala, and related parts of the striatum and thalamus. Alternatively, structural magnetic resonance imaging (MRI) demonstrating increased grey matter volume in the ACC may predict antidepressant response (42). In another approach, diffusion tensor imaging (DTI) may identify white matter abnormalities, particularly in frontal and temporal lobes and tracts in affective disorders versus controls (43,44). Some of these abnormalities appear to be state dependent (i.e., present only when the patient is clinically depressed), whereas others appear to be trait dependent (i.e., present whether the patient is depressed or not) (45). Further, one positron emission tomography (PET) imaging study correlated discrete symptom components of depression (i.e., psychic depression, sleep disturbance, loss of motivated behavior) with different brain regions (46).

It is possible that the state-dependent changes mediate the emotional, cognitive, and behavioral manifestations of a major depressive episode, whereas trait-dependent changes play a more fundamental role in the pathogenesis of the illness. Based on this work, a critical neural circuit has been proposed, including

Prefrontal cortex (PFC)
Striatum
Hippocampus
Other related areas

The areas in the PFC include the ventrolateral, orbital and dorsomedial/dorsoanterolateral PFC, the anterior (agranular) insula, and the anterior cingulate gyrus. The areas in the striatum include the ventromedial caudate and the accumbens projecting to the ventral pallidum. Hippocampal (total and gray matter) volumes are also reported to be reduced in depressed patients (47,48 and 49). Further neurogenesis and neuronal plasticity (i.e., dendritic remodeling and synaptic contacts) in the hippocampus are two proposed mechanisms for the behavioral effects of antidepressants (50). Other areas include the ventral tegmentum, the bed nucleus of the stria terminalis, the nucleus tractus solitarius, the periaqueductal gray, and the locus coeruleus (51,52 and 53).

The identification of these areas prompted other studies looking for microanatomical abnormalities in these areas as well as changes in receptor physiology. For example, astroglia are responsible for a number of important neural processes, including regulation of extracellular potassium, glucose storage and metabolism, and glutamate uptake. All of these processes are, in turn, crucial for normal neuronal function. One study of the subgenual part of Brodmann’s area revealed marked reductions in the number of astroglial cells in patients with familial major depression and in those with bipolar disorder (54). Of interest, there was no change in either the number or the size of neurons in these areas. Glial density and the glia-to-neuron ratio were reported to be reduced in the amygdala of subjects with MDD (55). These findings suggest a primary dysfunction in glial cells as a causative factor in the pathogenesis of major depression and could shift the focus of antidepressant therapy from mechanisms focused solely on neurons to those involving glial cells.

In summary, developing a better anatomical understanding of major depression can provide a structure upon which to integrate disparate neurochemical and neurophysiological findings. At the same time, it identifies new areas, such as the role of glial dysfunction in at least some forms of major depression.

Neurotransmitter and Related Hypotheses

Monoamine Theories of Depression

Catecholamine Hypothesis. This theory postulates diminished activity of catecholamines in the central nervous system (CNS) (e.g., NE) (10,11). Conversely, mania involves a relative increase in their activity.

Norepinephrine. The ascending NE pathway in the CNS begins with projections from the locus coeruleus, an anatomical site encompassing neurons (containing 85% to 90% of central NE stores) that project to the following:

Hippocampus
Cerebral cortex
Amygdala
Lower brainstem center (which controls sympathetic output)

The effect of antidepressants on this system may subserve their efficacy. The TCAs and MAOIs increase the activity of this transmitter by two different mechanisms. TCAs block the reuptake pump that recovers NE from the synaptic cleft shortly after its release from the presynaptic neuron. Thus, reuptake inhibition is occurring during both the acute and the chronic phases of therapy. Interestingly, antidepressant response usually occurs during the chronic phase. MAOIs interfere with enzymatic deamination. In either case, the outcome is increased synaptic NE concentrations.

The earliest investigations of this hypothesis measured the major metabolites of NE (e.g., 3-methoxy-4-hydroxyphenylacetic acid [MHPG]) in cerebrospinal fluid (CSF), plasma, and urine. The purpose was to elucidate the biological mechanisms subserving mood disorders, to develop potential markers, to facilitate diagnosis, and to aid in the prediction of treatment response. Although initially promising, this line of inquiry was impeded by various methodological obstacles (e.g., the relative contribution of peripheral vs. central sources) and conflicting results. For example, most studies found that CSF MHPG concentrations in depressed patients were identical to normal control subjects. Because it is in equilibrium with plasma MHPG, however, the failure to find low CSF or plasma MHPG does not negate the NE hypothesis. This is in part because CSF MHPG may not be an accurate reflection of NE activity in the CNS.

The Depression-Type (D-type) score was developed as a predictive tool by Mooney et al. (56) and exemplifies one attempt to pursue this line of investigation. Janicak et al. (57) summarized this issue while reporting negative results on the predictive value of urinary MHPG in unipolar depressed patients treated with standard antidepressants.

In animals, subchronic treatment (i.e., several days or weeks) with TCAs, MAOIs, and electroconvulsive shock (ECS) coincides more closely with the time to maximal clinical response (58,59). This time frame also correlates with the most consistent adaptive change (i.e., a reduced sensitivity of postsynaptic receptors), leading to diminished adenylate-cyclase activity. In the original hypothesis, depression was postulated to be secondary to decreased NE levels, decreased NE release, or subsensitive NE receptors. A downregulation (or reverse catecholamine hypothesis) subsequently proposed a decreased number of postsynaptic β₂-adrenergic receptors in peripheral tissues (e.g., leukocytes) after chronic antidepressant treatment (59). Thus, depression may be the result of a hyperadrenergic rather than a hypoadrenergic state (i.e., increased levels or release, or supersensitive receptors are the critical characteristics). This hypothesis is further supported by the neuropharmacological effects of chronic antidepressant treatment, which decrease the following:

Brain tyrosine hydroxylase and NE
Postsynaptic β-adrenergic receptor sensitivity and density
The basal firing rate of NE neurons in the locus coeruleus

Within this context, the original theory may still be valid because a defect in presynaptic neurotransmission should result in a compensatory upregulation of postsynaptic receptors. Thus, normalization of presynaptic activity should downregulate (or “normalize”) postsynaptic receptor function.

In addition, β-adrenergic receptors are increased in the brain of suicide victims, as are the number of α₂-adrenergic receptor binding sites in the brains of suicide completers and in the platelets of depressed patients (60). The implication is that the pathological increased activity of these autoreceptors may reduce NE output secondary to a short loop, negative feedback mechanism. Further, Crews and Smith (61) found the α₂-adrenergic receptors adapted (i.e., downregulated) after 3 weeks of treatment with desipramine, ultimately enhancing NE transmission.

On balance, these actions could support a decrease rather than an increase in the functional state of brain NE transmission if depression is conceptualized as a state of supersensitive catecholamine receptors secondary to decreased NE availability. This reasoning is consistent with the original hypothesis of diminished NE functioning, with antidepressants returning receptors to a more normal state of sensitivity. Siever and Davis (62) further elaborated on this concept by suggesting the possibility of dysregulation in the homeostatic mechanisms of one or more neurotransmitter systems, culminating in an unstable or erratic output.

Dopamine. In part, based on the role DA plays in the reward system circuitry, this neurotransmitter is also postulated to be important in the pathophysiology of depression (63). In 1975, Randrup et al. (64) postulated a role for DA in depressive disorders. A reanalysis of the data from several groups indicated a bimodal distribution of CSF homovanillic acid (HVA) levels in depressed patients, with one group comparable to normal control subjects and the other with decreased levels (65). In addition, Roy et al. (66) reported on the potential predictive value of lower urinary HVA output in depressed patients who attempted suicide versus those who did not. These reports indicate a decreased turnover in DA in at least a subset of depressed patients.

Consistent with earlier studies, Muscat et al. (67) reported on chronic exposure to mild unpredictable stress in rats as a model to study the antidepressant-reversible decreases in the consumption of palatable sweets. They found that certain DA agonists (i.e., quinpirole, bromocriptine) administered intermittently had the same positive effects as TCAs. They further postulated that the infrequent, intermittent administration of DA agonists (e.g., psychostimulants) avoided problems with tolerance and abuse while providing a clinically relevant antidepressant strategy. In this context, other DA agonists with potential antidepressant properties include

Amineptine (68)
Bupropion (69)
Nomifensine (70,71)
Pramipexole (72)

A subsequent report by Kapur and Mann (73) reviewed the role of DA in depressive disorders. They discussed several lines of evidence, including

The lower CSF HVA levels in some depressed patients
An increased incidence of depression in Parkinson disease as well as in patients receiving DA-depleting or antagonistic agents
The antidepressant effect of agents that enhance DA transmission
The ability of various classes of antidepressants and ECS to enhance DA effects in animal models

This last point is supported by autoradiographic evidence indicating that chronic treatment with several different antidepressants modulates postsynaptic DA function. This increases the density of D₂ and D₃ receptors, particularly in the nucleus accumbens and striatum (74,75,76 and 77). Lammers et al. (78) also found that chronic antidepressant treatment with several TCAs, MAOIs, and ECS produced a selective increase in D₃ receptor gene expression in the shell of the nucleus accumbens. Although fluoxetine decreased D₃ messenger RNA (mRNA) when given alone, fluoxetine and imipramine prevented downregulation of D₃ receptors caused by handling stress. These findings support impairment in the nucleus accumbens associated with the experience of pleasure, a hallmark of depression.

Indolamine Hypothesis. The second neurotransmitter implicated in the monoamine theory is serotonin (5-HT) (79). This neurotransmitter is contained in a few pathways, with those in the midbrain raphe nuclei projecting to the limbic-septal area where the hippocampus and amygdala may be of particular importance. Serotonin abnormalities are widely reported in patients with depression. Chapter 6 notes several of these in the section on “Suicide.” Other abnormalities include

Decreased 5-HT uptake in the platelets (V_max) of depressed patients is linked to a decrease in the number of platelet imipramine binding sites (80).
Blunting of the maximal prolactin response to intravenous tryptophan (the precursor of 5-HT) is found in depressed patients. Similar results have also been observed with fenfluramine and m-chlorophenylpiperazine (mCPP) (81).
p-Chlorophenylalanine, which decreases 5-HT synthesis, reverses the clinical efficacy of antidepressants (82).
Depletion of plasma tryptophan precursors may reverse antidepressant-induced remissions (83).
Tryptophan and 5-hydroxy’tryptophan, the precursors of 5-HT, may have antidepressant effects, alone or in combination with other drugs (84).
Antidepressants reduce 5-HT receptor number but not their affinity (85), in a manner analogous to β-adrenergic receptor downregulation.
Electroconvulsive therapy (ECT) potentiates prolactin response to thyrotropin-releasing hormone (TRH), which is mediated by serotonin (86).
ECS enhances 5-HT₂ receptor functional activity and binding characteristics in postmortem studies on animals (87).
Frontolimbic 5-HT_2A receptor binding may identify vulnerability to mood disorders (88).
5-HTT binding potential in the amygdala and midbrain is lower during a major depressive episode (MDE) (89).

From a therapeutic perspective, the most successful application is the clinical efficacy of agents that impact this system by

Blocking the reuptake transport mechanism (90)
Antagonizing specific receptor subtypes (91)
Modulating presynaptic receptors

Regarding this last point, there is interest in partial 5-HT agonists that stimulate their receptor target in the absence of the natural, full agonist (i.e., endogenous neurotransmitter). For example, in the early 1990s, there was considerable interest in the azapirones (i.e., 5-HT_1A partial agonists) as possible antidepressants and anxiolytics. Eison (92) argued for a common underlying pathology for anxiety and depression based on early evidence that azapirones (e.g., buspirone, gepirone, ipsapirone, and tandospirone) appeared to have both anxiolytic and antidepressant effects. Eison postulated that this is a result of the ability of the azapirones to modulate central 5-HT function through their partial agonism of 5-HT _1A receptors. In support of their possible antidepressant effects, Eison noted that azapirones downregulate 5-HT₂ receptors, as well as desensitize presynaptic 5-HT_1A autoreceptors, and may also affect postsynaptic 5-HT_1A receptors differently than their presynaptic counterparts (i.e., buspirone is a partial agonist of only postsynaptic receptors). Thus, he postulated that azapirones could normalize central 5-HT neurotransmission either up or down depending on the baseline level of activity (i.e., excess—anxiety; deficit—depression). Agents were then developed to block the 5-HT transporter and also act as 5-HT_1A receptor agonists (93).

Other Neurotransmitter Hypotheses

Glutamate. Evidence from preclinical studies, postmortem studies, and clinical trials suggests that increases in glutamate and its ionotropic receptors (e.g., N-methyl-D-aspartate [NMDA]; α-amino-3-hydroxy-5-methyl-4-isoxazole propionic acid [AMPA]) may play an important role in the pathophysiology of depression (94). For example, some NMDA modulators, including 2-aminophosphonoheptanoic acid (a competitive NMDA antagonist), 1-aminocyclopropanecarboxylic acid (ACP, a glycine partial agonist), and dizocilpine (a use-dependent channel blocker), reduce immobility time in the forced swim test (95). Chronic antidepressant treatment also reduces high-affinity glycine sites in the cerebral cortex but not in other brain regions and causes more widespread alterations in mRNA levels coding for NMDA receptor subunits (95). Postmortem studies also find a decrease in glycine binding sites on the NMDA receptor in the frontal cortex of suicide victims (96). One small clinical trial reported that intravenous ketamine improved depressive symptoms for up to 72 hours in patients not responsive to standard antidepressants (97). A subsequent trial also suggests that repeateddose IV ketamine could benefit treatmentresistant depression (TRD) with acceptable tolerability (98). Other investigations also implicate AMPA glutamate receptors with the rapid onset and longer lasting antidepressant effects of ketamine versus other NMDA antagonists (99). The activation of ligand-gated NMDA receptors results in an influx of Ca²⁺ into the cell, which activates nitric oxidase synthase (NOS) and the formation of nitric oxide. In this context, there is preclinical evidence that NOS inhibitors demonstrate antidepressant-like activity. Thus, the antidepressant effect of NMDA receptor antagonists may be mediated in part through its effects on NOS (100,101).

γ-Aminobutyric Acid. Since glutamate and γ-aminobutyric acid (GABA) systems have reciprocal links throughout the brain, GABA may play a role in the pathophysiology of depression. Injections of bicolline (a GABA-A receptor antagonist) into the hippocampus promoted learned helplessness in rats, whereas injections of GABA into the hippocampus reversed or prevented its development (102). Martin et al. (103) found that GABA-B receptors are downregulated in the frontal cortex of animals with learned helplessness and are normalized by successful treatment with chronic TCA administration. A recent study in depressed patients using proton magnetic resonance spectroscopy (MRS) indicated that reduced GABA in the occipital cortex may predict treatment resistance (104). Other human data include

GABA concentrations are reduced in the CSF, blood, and postmortem brains of depressed patients when compared to controls (105)
Antidepressant treatment reduces the ability of 3H-flunitrazepam to bind to the benzodiazepine (BZD) binding site
Selective serotonin reuptake inhibitors (SSRIs) and ECT increase cortical GABA in depressed subjects (106,107)
TCAs inhibit GABA transporters (108,109)
Valproate (an indirect GABA agonist), via the inhibition of GABA transaminase, may be effective in treating bipolar depression (110)
Fengabine (α-GABA agonist) may benefit unipolar depression (111)

Histamine. Histamine (H) interacts with both monoamines and acetylcholine (ACh). Specifically, the H₃ receptors can act as heteroreceptors to modulate the release of monoamines and ACh in the brain (112). Stress decreases H₃ receptors in the rat cortex, while chronic amitriptyline treatment reverses this effect (113). Finally, thioperamide (an H₃ antagonist) and iodophenpropit (a very selective H₃ antagonist) demonstrate antidepressant-like activity in the forced swim test in rats (79).

Others. Data support the potential role of the muscarinic cholinergic antagonist, scopolamine, which demonstrated rapid and robust antidepressant effects in one trial (114).

At one time, opioids were first-line antidepressants, but they are no longer clinically acceptable, in part because of the severity of their adverse effects. Clinical data, as well as basic science observations, suggest a role for opioids in the pathophysiology of depression. For example, SSRIs and TCAs increase the concentration of enkephalins in the brainstem and µ-opioid receptors in the rat forebrain (115). Opioids can inhibit the adrenocorticotropic hormone and cortisol overactivity often associated with depressed patients. Finally, cholecystokinin (CCK) is a neuropeptide that colocalizes with opioids in a number of brain regions. In one report, CCK-B (but not CCK-A) receptors enhanced the antidepressant-like activity of opioids (116).

Interactional Theories of Depression.

A single neurotransmitter theory does not sufficiently explain all known evidence. As a result, models that include two or more systems can consider their modulatory interactions.

Permissive Hypothesis. The “permissive” hypothesis proposes that decreased function in central serotonin transmission sets the stage for either a depressive or a manic phase (117). Although not sufficient to produce a mood disturbance, when superimposed on aberrations in NE function, serotonin changes may determine the phase of an affective episode (i.e., decreased 5-HT and decreased NE subserves depression; decreased 5-HT and increased NE subserves mania). Data from animal studies to support this theory include

5,6-Dihydroxytryptamine lesions of the serotonin nuclei are known to attenuate the reduction in β-noradrenergic receptor binding induced by chronic tricyclic treatment (118)
6-Hydroxydopamine lesions of the NE nuclei in the dorsal and ventral bundles, as well as in the locus coeruleus, block the enhanced locomotor responses to quipazine after repeated ECS

Adrenergic-Cholinergic Balance Hypothesis. A second interactional theory postulates an imbalance between the cholinergic and the noradrenergic systems (119). The central cholinergic system consists of projections primarily from the nucleus basalis. A relative increase in this system’s activity in comparison with central NE activity may play a role in producing depression. Conversely, a decrease relative to central NE activity may play a role in producing mania. Clinically, agents with cholinomimetic effects (e.g., precursors, cholinergic agonists, cholinesterase inhibitors) have shown some benefit in mania (see Chapter 10). Cholinergic abnormalities are also thought to underlie some of the abnormal sleep patterns (e.g., decreased rapid eye movement [REM] latency; increased REM density) found in depression. Consistent with this theory is evidence that ECT

Decreases brain ACh levels
Increases activity of choline acetyltransferase, the enzyme most prominently involved in ACh breakdown
Causes release of CSF ACh
Produces cholinergically mediated electroencephalographic (EEG) slowing following a series of treatments

The numerous links between smoking and clinical depression (or schizophrenia) also suggest a role for the cholinergic system in the pathophysiology of depression. Thus, patients with depression have a higher incidence of smoking compared with the general population (120). Depression can also emerge when someone stops smoking (121) and can be reversed by either resuming smoking or with antidepressant treatment (122). Nicotine increases the concentration of all three biogenic amine neurotransmitters (i.e., DA, NE, and 5-HT) (123,124) and can reverse learned helplessness in rats (125).

Bidimensional Model Hypothesis. Proponents of this hypothesis identify three types of abnormal neurochemistry:

Causative
Phenomenological (expressive)
Epiphenomenological (possible useful state markers)

As others before them, Emrich and Wolf (126), propose that a single neurochemical imbalance is not sufficient to explain many of the inconsistencies and contradictions in studies with various mood stabilizers. Instead, they speculate that several neurotransmitter imbalances relating to different brain areas should be anticipated.

Assuming differing mechanisms of action for mood stabilizers such as lithium, valproate, and carbamazepine (CBZ), a bidimensional model of mood regulation postulates two “gating zones” (one for depression and one for mania). These zones are subserved by different neurochemical abnormalities, leading to a situation in which both could be impacted by certain agents (i.e., mood stabilizers) or, alternatively, could individually be affected by unidirectional compounds (e.g., TCAs).

Second Messenger Dysbalance Hypothesis. Receptors are glycoproteins imbedded in the lipid bilayer of neuronal membranes and can detect minute amounts of specific ligands (e.g., neurotransmitters, hormones). The ligand-receptor interaction sets in motion a transduction system (e.g., an enzyme, an ion channel) that orchestrates various intracellular biochemical events. Further, investigators look beyond the receptor-ligand binding relationship to study intraneuronal events stimulated by this interaction. Two primary areas are the adenylate cyclase (AC) and the phosphoinositol (PI) second messenger systems.

Such investigations led to the postulation that functional disturbances in intraneuronal signal transmission distal to the receptors of classic neurotransmitters (i.e., the first messengers) are pathogenetically important in mood disorders (127). Further, it suggests that these disorders are caused by a dysfunction in the “crosstalk” between major intraneuronal signal amplification systems (e.g., AC and the phospholipase C systems). Thus, depression may result from a diminished functioning in cyclic adenosine monophosphate (cAMP)-mediated effector cell responses, together with an absolute or relative dominance of the inositol triphosphate diacylglycerol-mediated responses. In the context of bipolar disorder, mania is conceptualized as resulting from the reverse circumstances.

Work in this area has yielded results that link the changes in receptor physiology to changes in second messenger systems. For example, blunted β-adrenergic receptor responsivity of noradrenergic receptor-coupled AC occurs after repeated doses of most, but not all, antidepressants (57). Yet, an increase in AC activity was demonstrated in the hippocampus and cortex following chronic antidepressant treatment and ECS (128,129). This suggests that, even though there is a relative decrease in β-adrenergic receptors after chronic as compared with acute antidepressant administration, levels of cAMP remain elevated compared with the no treatment condition. Thus, despite downregulation of β-adrenergic receptors, there is an overall increase in the activity of the cAMP system because of an increase in biogenic amine levels in the synapse as a result of antidepressant action on amine reuptake mechanisms. Thus, current antidepressants, including NE and serotonin reuptake inhibitors, may exert their effects through activation of the cAMP pathway, which in turn leads to regulation of cAMP-dependent protein kinase and subsequently to activation of the cAMP response element binding protein (CREB).

The latter is a downstream component of the cAMP cascade system and, in its phosphorylated form (pCREB), induces the expression of neuroprotective factors such as BDNF (25,129). BDNF in turn mediates neuronal genesis, survival, and plasticity (130). BDNF is the most widespread growth factor in the brain and is activated by a number of stimuli in addition to pCREB. BNDF infusions into the adult rat neocortex result in 5-HT nerve terminal growth and regrowth after parachloroamphetamineinduced destruction, especially in the CA3 region of the hippocampus (131). CREB, the BDNF transmembrane receptor tyrosine kinase B, and BDNF itself are all elevated following chronic antidepressant and ECS exposure (132,133,134 and 135). Stress decreases BDNF mRNA expression in rat brain consistent with stress as a precursor to the onset of depression in humans (136).

The administration of BDNF into the brain produces antidepressant-like activity in both the forced swim and the learned helplessness animal model. In one study, BDNF was continuously infused into the midbrain by an osmotic minipump over 14 days (137). In another, BDNF was administered as a single injection into the dentate gyrus or CA3 region of the hippocampus. In both of these animal models, BDNF produced an antidepressant effect lasting over 10 days (138). Although the postmortem findings from the brains of depressed patients are inconsistent (139,140), BDNF levels were low in the serum of depressed patients and increased in the hippocampus and cerebral cortex of postmortem brains of patients treated with antidepressants (141).

Taken together, the effect of increased neurotrophins could mitigate hippocampal changes associated with exposure to stress. Although theoretical, this model is supported by empirical studies of the pathophysiology of depression in patients (142,143) as well as in animal studies (144,145). These theories are also consistent with the neuroanatomical findings of cell loss and decreased hippocampal volume, as well as increased ventricular volume consistent with cortical atrophy, reported in the brains of depressed patients (146,147).

Membrane and Cation Hypothesis

The resting membrane potential, monoamine transport and reuptake, and other functions of the cellular membrane are partially related to cation transport mechanisms. This led to a hypothesis proposing a deficiency in one or more of these transport functions that such a deficit is genetically determined and that the resulting membrane dysfunction predisposes the person to a mood disorder.

The steady state distribution of lithium provides some evidence to support this hypothesis. For example, patients with bipolar disorder have higher mean intracellular and extracellular red blood cell to lithium ratios, as do their firstdegree relatives (148). These findings suggest genetic control of the lithium ratio and its role in the pathogenesis, as well as the pharmacotherapy, of mood disorders.

There is also evidence that blockade of potassium (K⁺) channels produces antidepressant effects. For example, pretreatment with glyburide (an ATP-sensitive K⁺ channel blocker) potentiated the reduction in immobility seen in the forced swim test in mice following treatment with a number of different classes of antidepressants. Other K⁺ channel blockers such as apamin, charybdotoxin, and gliquidone also show antidepressant properties in this model, while ATP-sensitive K⁺ channel activators such as cromakalim, minoxidil, and pinacidil increase the amount of time mice remain immobile (149). Finally, fluoxetine at clinically relevant concentrations inhibits voltage-activated K⁺ channels (150).

Biological Rhythm Hypothesis

Halberg (151) postulated a desynchronization of circadian rhythms in depression. In this context, Goodwin et al. (152) found a phase advance in the rhythms of depressed patients, and Schulz and Lund (153) found a diminished amplitude. Perhaps, most interesting is the ability of antidepressants to alter these rhythms, possibly by binding to receptor sites in the suprachiasmatic nucleus (154).

A clear example is the disturbance in the sleep-wake cycle that constitutes one of the hallmarks of depression. Thus, several components are disrupted, including

A decrease in total sleep time
An increase in sleep onset latency
A decrease in the arousal threshold
An increase in wakefulness
Terminal insomnia (early morning awakening)
REM-related phenomena
A decrease in REM onset latency An increase in REM density
A redistribution of REM sleep to earlier in the sleep phase

These last three items constitute “REM pressure,” which can be viewed as a state of hyperarousal.

Theories explaining the mechanism of action of phototherapy and the biological basis of seasonal affective disorder (SAD) are also related to this area of investigation (i.e., infraradian rhythm disturbances) (155). For example, Skewerer et al. (156) reported that plasma NE levels in SAD were inversely related to the level of depression, and these levels increased proportionally to the degree of therapeutic improvement. Depue et al. (157), in reporting their work on the possible role of DA in SAD, noted that basal serum prolactin values did not change as a function of season or after successful phototherapy. Further, these values remained significantly lower in comparison with those of control subjects, suggesting they could serve as a trait marker for this disorder (see later section).

These findings do not exclude the possibility that the central serotonin system is also involved, given its influence on prolactin levels. In this regard, some data show an antidepressant response to dietary L-tryptophan and D-fenfluramine, an indirect serotonin agonist (158). In this context, Jacobsen et al. (159) reported a euphoric, energized reaction in 10 SAD patients given infusions of the serotonin agonist mCPP in an open trial. Finally, Lacoste and Wirz-Justice (160) found evidence for seasonal rhythms in serotonin levels in their healthy control subjects, with the winter values being significantly lower than their summer counterparts.

Neuroendocrine Hypothesis

Cortisol hypersecretion; blunted growth hormone and prolactin responses; blunted thyroidstimulating hormone (TSH) response to TRH; reduced luteinizing hormone secretion; and disturbances in β-endorphin, vasopressin, and calcitonin have all been associated with depression.

As summarized by Gold et al. (161), acute behavioral and physiological changes that occur in the general adaptational syndrome are almost identical to those seen in the depressive syndrome. They suggest that melancholia may be an acute generalized stress response that escapes the normal regulatory restraints. Further, they implicate a dysregulation in glucocorticoid activity, which typically antagonizes corticotropinreleasing hormone (CRH) neurons, as well as the locus coeruleus-NE system, in addition to mediating immunosuppression. This activity would normally restrain or counterregulate the effectors of the stress response, precluding excessive or extended activation.

Dysregulation of the hypothalamic-pituitary-adrenal axis is thought to cause a disturbance in the circadian rhythm of cortisol. In this context, depression may be associated with failure of feedback mechanisms to regulate cortisol secretion, resulting in high cortisol levels (162). In addition, preliminary data suggest that associated cognitive impairments may persist independent of mood improvement (163). The relationship between cortisol secretion and depression was investigated through the use of the dexamethasone suppression test (DST) because approximately 50% of patients with symptoms of MDD show nonsuppression of cortisol. Further, patients who improve but continue to have an abnormal DST may be at higher risk for relapse or suicidal behavior (see Chapter 1). More recently, the combination of DST and CRH testing was proposed as a potentially more sensitive biomarker to predict clinical outcome in major depression (164).

Dysregulation of the hypothalamic-pituitary-thyroid axis causes a reduction in thyroid function. There may be a relationship between an abnormal TSH response to TRH and depressive symptoms. Thus, unipolar patients undergoing the TRH-TSH test (which measures the difference between baseline TSH and peak postinfusion TSH after they are given synthetic TRH) reportedly have a blunted response, whereas bipolar depressed patients have an elevated response (see Chapter 1).

Immunological Hypothesis

Two hypotheses about the relationship between mood and the immune system include

Depression may alter immunological function
An unidentified infectious process (e.g., viral) may induce affective disturbances

In support of the former postulate, Bartrop et al. (165) and Schleifer et al. (166) both reported suppression of the immune system following a period of bereavement.

Subsequently, Schleifer et al. (167) found significant age-related immunological differences between their group of 91 unipolar depressed patients and a group of matched control subjects. Specifically, mitogen responses and the number of T4 lymphocytes did not increase in the depressed group with advancing age, as was the case with the normal control subjects. The impact of elevated cortisol levels on this phenomenon needs further clarification.

Theoharides et al. and Miller et al. (168,169) reviewed the role of cytokines in various CNS activities and concluded that these immunological chemical messengers may help further our understanding of pathological conditions such as depression, provide for possible novel treatments, and, in addition to behavioral changes, help to monitor efficacy at the immune system level.

Although the results of several lines of investigation are inconclusive, there may be subtypes of depressive disorders that affect immune function. Furthermore, the question of a possible causative infectious process has not been adequately addressed. As discussed in Chapter 6, there is also data to implicate proinflammatory cytokines in the pathophysiology of depression. In this context, excess levels of other inflammatory mediators may produce inappropriate glutamate receptor and microglia activation, promoting loss of astroglia, which may then predispose to depression (170).

CONCLUSION

Development of new psychotropics now involves selecting a target of interest based on existing knowledge about how that mechanism affects the pathophysiology underlying a specific disorder (89,90). Examples include

The neuronal reuptake pump for a neurotransmitter
A specific neurotransmitter receptor subtype
A subunit of an ion channel

A new molecular entity is then developed to stereospecifically interact with one of these targets and at the same time modified structurally so that it does not affect other sites that mediate undesired effects.

At times, a drug that affects more than one target simultaneously may be preferable for specific indications because of desired consequences, such as enhanced potency, wider therapeutic efficacy (e.g., greater spectrum of activity in a heterogeneous condition such as major depression), faster onset of desired activity, or better tolerability. The goals of new drug development involve

Maintain or improve efficacy
Widen the therapeutic index
Improve tolerability profiles
Reduce pharmacodynamically or pharmacokinetically mediated drug-drug interactions

Since the original catecholamine hypothesis, there have been multiple attempts to elucidate the biological mechanisms subserving mood disorders as well as the mechanisms of antidepressant action. In this context, there is a gradual evolution from consideration of a single neurotransmitter system to the modulating interactions of various neurotransmitters. Next, the effects these “first messenger” systems have upon receptor subtypes and the intraneuronal “second messenger” or signal amplification system in postsynaptic cells were explored. Most recently, the subsequent cascade of intraneuronal events culminating in altered gene expression is increasingly the focus of investigations of pathophysiology as well as potential targets for drug therapy.

Further, many of these neurotransmitters subserve or intimately interact with a number of neuroendocrine, circadian rhythm, and neurophysiological activities, which may be dysregulated in mood disorders. Genetic defects moderated by environmental factors are likely the basis for such pathology.

Pharmacokinetics

SELECTIVE SEROTONIN REUPTAKE INHIBITORS

Although SSRIs are quite similar in terms of their pharmacodynamics, they are quite different in terms of their pharmacokinetics. The most important differences are

CYP isoenzyme(s) responsible for their metabolism
Their ability to inhibit CYP isoenzyme activity
Elimination half-lives
Linear versus nonlinear pharmacokinetics
Changes in plasma drug levels as a function of age and gender

CYP Isoenzymes

Given the significant structural differences between the SSRIs, it follows that different CYP isoenzymes mediate their metabolism. For example, the demethylation of citalopram and escitalopram is principally dependent on CYP 2C19 and 3A3/4 (171), an important point when one considers that 20% of Asians are genetically deficient in CYP 2C19. Thus, higher levels of citalopram and escitalopram can occur in such individuals on a standard dose, so starting at a lower dose and adjusting more gradually is prudent.

Fluvoxamine and fluoxetine inhibit multiple CYP isoenzymes, complicating the study of their metabolism and an important consideration when using them in patients on other medications. Although some evidence suggests that CYP 1A2 may be important in the metabolism of fluvoxamine, another report is not consistent with that conclusion (172).

The potential role of other CYP isoenzymes in the metabolism of fluoxetine and norfluoxetine remains controversial. There is some evidence suggesting that R- and S-fluoxetine and S-norfluoxetine, but not R-norfluoxetine, are substrates of CYP 2D6 (173). This, coupled with the substantial inhibition of this isoenzyme by fluoxetine and norfluoxetine, may account for their long half-lives.

Paroxetine clearance at low concentrations is dependent on CYP 2D6, which is almost completely saturated at these levels. This accounts for the drug’s nonlinear pharmacokinetics and the increase in its half-life from 10 to 20 hours when the dose is increased from 10 to 20 mg/day. At higher concentrations, paroxetine is probably dependent on CYP 3A3/4 for its clearance. This dose-dependent change in the clearance may account for the higher incidence of withdrawal reactions than might otherwise be expected with a half-life of 20 hours at steady state on 20 mg/day (174).

The N-demethylation of sertraline is catalyzed by multiple CYP isoenzymes (e.g., 3A3/4, 2C19, 2C9, 2D6, and 2B6) (175). In most patients, no one isoform contributes to more than 40% of its overall metabolism, making it relatively immune to harmful interactions with other drugs that inhibit or induce CYP isoenzymes. Nevertheless, a dose adjustment may be prudent when starting or stopping drugs that are substantial inducers or inhibitors of CYP 3A3/4 (176).

Elimination Half-Life

All of the SSRIs except fluoxetine have half-lives of 15 to 30 hours. Fluoxetine and its equipotent metabolite, norfluoxetine, have half-lives of 2 to 4 and 7 to 15 days, respectively. Thus, it can take several weeks to achieve steady state and a comparable interval to fully clear the drug once it is discontinued. This means it can take several weeks to reach maximal effect, which can persist for many weeks after its cessation (making it in essence an oral depot medication). Although this property led to the marketing of an enteric-coated, long-acting, oral formulation of fluoxetine (177), it also complicates its use, since the risk of an adverse drug-drug interaction can persist for weeks after fluoxetine discontinuation. Coupled with its ability to substantially inhibit more than one CYP isoenzyme, fluoxetine should be reserved for cases in which the advantages of its long half-life outweigh its disadvantages and avoided in older patients since there are other SSRIs (e.g., citalopram, escitalopram, and sertraline) available that do not pose these problems.

Linear versus Nonlinear Kinetics

Fluvoxamine, fluoxetine, and paroxetine have nonlinear pharmacokinetics, which means that dose increases lead to disproportionately greater increases in plasma drug levels. For these reasons, dose increases with fluvoxamine, fluoxetine, and paroxetine can lead to greater than proportional increases in concentration-dependent effects such as serotonin-mediated adverse effects (e.g., nausea) and inhibition of specific CYP isoenzymes. In contrast, citalopram, escitalopram, and sertraline have linear pharmacokinetics

Age and Gender Issues

Paroxetine has the largest age-related change in plasma drug levels in comparison with other SSRIs. Its levels can be up to 100% greater in physically healthy individuals older than 65 years than in younger individuals. For this reason, the recommendation is to start paroxetine at half the usual dose and adjust upward more slowly in elderly patients. Age-related changes in SSRI plasma levels are important to monitor because elderly patients are likely to be on concomitant medications and effects on specific CYP isoenzymes are concentration dependent (see Chapter 15).

Although there is no age-related change in its plasma levels, fluvoxamine has nonlinear and gender-dependent pharmacokinetics (178). A doubling of the dose from 50 to 100 mg twice a day causes, on average, a 340% increase in fluvoxamine plasma levels, which is more pronounced in men (460% increase) than in women (240% increase).

Sertraline also has an age by gender interaction, but it is relatively modest (179). Young men (mean = 30.6, range = 21 to 43) develop levels that are lower than those of young women (mean = 34.4, range = 20 to 45) or of older men (mean = 72.0, range = 67 to 82) or older women (mean = 69.7, range = 65 to 82).

SEROTONIN AND NOREPINEPHRINE REUPTAKE INHIBITORS

Venlafaxine

The biotransformation of venlafaxine to its active metabolite O-desmethylvenlafaxine (ODV) is dependent on CYP 2D6. The further elimination of ODV is dependent on CYP 3A3/4. Venlafaxine and ODV have approximate half-lives of 5 and 11 hours, respectively. Because venlafaxine and ODV have virtually identical pharmacological profiles, both are believed to contribute equally to efficacy and adverse effects. Given the 11-hour half-life of ODV, the immediate-release formulation of venlafaxine is efficacious when administered twice daily.

The sum of the concentrations of venlafaxine and ODV is probably more important than their relative ratio. Thus, CYP 2D6 deficiency, which occurs in approximately 7% of Caucasians, has fewer clinical implications for venlafaxine than for drugs that are biotransformed by this isoenzyme to either centrally inactive metabolites (e.g., paroxetine) or metabolites that have a different pharmacological profile than the parent drug (e.g., TCAs). Nevertheless, substantial inhibition of CYP 3A3/4 could result in a meaningful increase in both venlafaxine and ODV plasma levels, particularly with CYP 2D6 deficiency. Such an increase would be expected to result in an increase in the incidence or severity of dose-dependent adverse effects.

Desvenlafaxine

Recently, the active metabolite of venlafaxine (i.e., desvenlafaxine [DVEN] succinate) was approved by the FDA for treatment of major depression. This agent follows linear pharmacokinetics and has a half-life of about 11 hours. It is primarily metabolized by conjugation and to a minor extent through CYP 3A4 oxidative metabolism. With renal impairment, the maximum dose is 50 mg/day, and with severe or end-stage renal disease (ESRD), it is 50 mg every other day.

Duloxetine

The issue of absorption of duloxetine is more complicated than for other antidepressants. Duloxetine is rapidly hydrolyzed in acidic media to naphthol, which has no known antidepressant activity. For this reason, the marketed product has an enteric coating that resists dissolution until it reaches a segment of the gastrointestinal tract where the pH exceeds 5.5. This delivery system explains why there is a median 2-hour lag before absorption begins and why maximum plasma concentrations (C_max) usually do not occur until 6 hours postdose.

For these reasons, the integrity of the enteric coating is critical to the usual and effective absorption of duloxetine. Thus,

The capsules should not be chewed or divided
The contents should not be removed and sprinkled on food or dissolved in juice
Conditions that significantly delay gastric outflow could cause erratic and incomplete absorption
Conditions that appreciably increase the pH of the stomach could lead to earlier breakdown of the enteric coating and quicker absorption than usual

Ingestion with food does not affect C_max but delays the time to C_max from 6 to 10 hours. When taken in the morning with food, the extent of absorption is decreased by approximately 10%. When taken in the evening with food, there is a 3-hour delay in absorption and a one-third decrease in the AUC, or conversely a one-third increase in an apparent clearance compared to the morning dose. To put these numbers in perspective, the evening effect of food is comparable to the difference permitted between generic formulations of the same drug.

The plasma half-life of duloxetine is 12 hours (range: 8 to 17 hours). Thus, steady state plasma concentrations are typically reached within 3 days of stable dosing. The apparent volume of distribution is about 1,640 L. Duloxetine is highly protein bound in human plasma, primarily to albumin and α₁-acid glycoprotein (180).

There are several lines of in vitro, ex vivo, and in vivo evidence suggesting that duloxetine dissociates slowly from (i.e., binds tightly to) the serotonin reuptake pump (181). This finding, along with its large volume of distribution and concentration in the brain, suggests that the plasma half-life underestimates the half-life in the brain and may account for its demonstrated efficacy as an antidepressant when given once a day

Duloxetine represents approximately 3% of the total radiolabeled material in plasma, indicating that it undergoes extensive biotransformation prior to elimination. The major pathway involves oxidation of the naphthyl ring to form 4-hydroxy duloxetine, which is then glucuronidated to 5-hydroxy, 6-methoxy duloxetine, which is then sulfated. These reactions are principally catalyzed by CYP 1A2 and 2D6. Many additional metabolites have also been identified in urine but usually represent minor pathways of elimination. Only 1% of the total dose appears in the urine or feces as unchanged duloxetine, while 70% and 20% of the dose is excreted as metabolites in either the urine or the feces, respectively. None of the metabolites of duloxetine are known to have antidepressant activity.

Duloxetine is a moderate inhibitor of CYP 2D6 and could inhibit its own metabolism but at doses above those usually recommended for the treatment of depression (i.e., 40 to 60 mg/day). It is also capable of inhibiting CYP 1A2 in vitro and could theoretically inhibit its own metabolism via this pathway. However, its potency for CYP 1A2 inhibition is even less than that for CYP 2D6, making this unlikely in most patients on recommended doses.

The half-life of duloxetine is similar in men and women. The AUC is approximately 25% higher and the half-life 4 hours longer in elderly (65 to 77 years old) compared to middle-aged (32 to 50 years old) females.

Duloxetine is not recommended for patients with ESRD. After a single 60-mg dose of duloxetine, the elimination half-life of the parent drug was the same, but the C_max and AUC values were approximately 100% greater in patients with ESRD receiving chronic intermittent hemodialysis compared to individuals with normal renal function. Moreover, the AUC of the major metabolites, 4-hydroxy duloxetine glucuronide and 5-hydroxy, 6-methoxy duloxetine sulfate, were seven to nine times higher. This indicates they are largely excreted in the urine. Population pharmacokinetic data suggest that laboratory evidence (e.g., modestly elevated serum creatinine levels) in individuals not considered to have clinically significant renal disease does not affect duloxetine clearance and dose adjustment is not required.

Duloxetine is also not recommended for patients with hepatic insufficiency. After a single 20-mg dose, the half-life of duloxetine was three times longer, the mean AUC was five times greater, and the mean plasma clearance was six to seven times lower in six cirrhotic patients with moderate liver impairment (Child-Pugh Class B) compared with values seen in age- and gendermatched healthy controls.

AMINOKETONES

Bupropion

There is a dearth of information about the metabolism of bupropion. That may initially seem surprising, since this drug is one of the oldest of the newer antidepressants, entering clinical trials in the mid-1970s and approved before fluoxetine. Ironically, its marketing was delayed after its approval because of the risk of seizures, which, in turn, is almost undoubtedly a consequence of its complicated pharmacokinetics (182,183).

The parent drug has a half-life of 8 to 10 hours and is biotransformed by oxidative metabolism to three active metabolites:

Hydroxybupropion
Threohydrobupropion
Erythrohydrobupropion

These metabolites all have half-lives of 24 hours or more and thus accumulate to a greater extent than the parent drug. Although preclinical testing demonstrates they are pharmacologically active, their beneficial or adverse effects were not tested beyond one clinical study that showed that patients in whom higher levels of these metabolites developed had a poorer outcome than those with lower levels (184).

Substantial interindividual variability is seen in the plasma levels of bupropion as well as in its three metabolites. There is a correlation between levels of threohydrobupropion and erythrohydrobupropion but not between the levels of these metabolites and those of either the parent drug or hydroxybupropion. Thus, there can be substantial differences in plasma levels among patients on the same dose of bupropion.

CYP 2B6 is responsible for the conversion of bupropion to hydroxybupropion (185,186). The mechanisms responsible for the conversion of bupropion to its other two major metabolites are not known, nor are the mechanisms that mediate the eventual elimination of hydroxybupropion from the body. Moreover, little is known about CYP 2B6, including any potential genetic polymorphisms, whether it is inhibited or induced by other drugs or substances, and whether genetic polymorphisms or pharmacokinetically mediated interactions are risk factors for seizures during bupropion treatment. For the same reason, it is unknown whether coadministration of specific CYP isoenzyme inducers (e.g., CBZ) or inhibitors (e.g., fluoxetine) could alter the clearance of bupropion or its metabolites in a clinically meaningful way. Conversely, bupropion is a potent inhibitor of CYP 2D6 (187).

TRIAZOLOPYRIDINES

Nefazodone and trazodone have complicated pharmacokinetics. Both parent drugs have short half-lives of approximately 4 hours and are primarily metabolized by CYP 3A3/4 to form the active metabolite, mCPP, which has a pharmacological profile different from that of either nefazodone or trazodone. This metabolite is a 5-HT_2C agonist and is anxiogenic when administered alone, in contrast to the parent drugs (188,189). Thus, a significant shift in the relative ratio of mCPP to the parent drug could paradoxically cause anxiety and stimulation instead of anxiety reduction and sedation. Although mCPP is formed by CYP 3A3/4, its elimination is dependent on CYP 2D6. Such a shift in the ratio of the parent drug to this metabolite could occur as a result of either genetically determined or drug-induced deficiency of CYP 2D6 activity. For example, drug-induced deficiency in CYP 2D6 activity and the resultant accumulation of mCPP may account for paradoxical reactions when switching from the long-lived fluoxetine to trazodone. This is all that is known about the pharmacokinetics of trazodone, reflecting the age of and limited experience with this drug.

Nefazodone

Serzone is no longer produced, partly because of economics and partly due to concerns about hepatotoxicity. Generic formulations of nefazodone, however, are available.

Nefazodone is biotransformed into two other metabolites, hydroxynefazodone (OH-NEF) and triazolodione (190). These latter two metabolites cannot be formed from trazodone because of structural differences between it and nefazodone. OH-NEF is an active metabolite with a pharmacological profile similar to that of the parent drug and is believed to contribute comparably to the overall clinical response. Its half-life is also approximately 4 hours. Triazolodione has only part of the pharmacological activity of nefazodone, being a relatively pure 5-HT_2A antagonist blocker. Thus, it would be expected to produce only part of the overall pharmacological effects of the parent drug. Nonetheless, this metabolite may be clinically important because it accumulates in concentrations 10 times that of nefazodone because of its considerably longer half-life (i.e., 18 to 33 hours).

Nefazodone has appreciable nonlinear pharmacokinetics because of its metabolism by and inhibition of CYP 3A3/4 (191). At doses of 200 mg/day, it undergoes an extensive first-pass effect such that its bioavailability is only approximately 20%. At doses of 400 mg/day, its bioavailability is appreciably higher, as are its plasma drug levels. This phenomenon is most likely due to inhibition of its own first-pass metabolism by CYP 3A/4. For this reason, dose-dependent effects of nefazodone can increase nonlinearly with higher levels.

TETRACYCLICS

Mirtazapine

Mirtazapine has a half-life of 20 to 40 hours (192). Its elimination is principally dependent on CYP isoenzyme-mediated biotransformation as a necessary step. Three CYP isoenzymes, 1A2, 2D6, and 3A3/4, mediate mirtazapine biotransformation to approximately an equal extent. Mirtazapine is also about 25% dependent on elimination by way of a phase II conjugation reaction with glucuronic acid.

Mirtazapine exhibits linear pharmacokinetics over its clinically relevant dosing range. This suggests that it does not inhibit the three CYP isoenzymes that mediate its biotransformation, and coadministration will not alter the elimination of other drugs dependent on these specific CYP isoenzymes. Nonetheless, a definitive statement on this matter awaits appropriately designed in vivo pharmacokinetic studies testing the potential effect of mirtazapine on model substrates for these CYP isoenzymes.

TRICYCLICS

TCAs are pharmacologically complex and so are their pharmacokinetics. The use of these drugs has diminished in large measure due to their narrow therapeutic index (i.e., the difference between the dose or concentration needed for efficacy versus the dose or concentration that causes serious toxicity). This is why understanding their pharmacokinetics is critical to their safe and effective use.

TCAs are slowly, but usually completely, absorbed from the small bowel, enter the portal circulation, and pass through the liver, where there is significant first-pass metabolism mediated by CYP 3A3/4 (40% to 70%). They then enter the systemic circulation for distribution. These agents are usually highly protein bound (75% to 95%), as well as highly lipophilic, with a large volume of distribution. Their half-lives range from 16 to 126 hours but are usually 24 to 30 hours.

Prolonged half-lives occur in individuals who are genetically deficient in CYP 2D6 or who have significant hepatic, renal, or left ventricular cardiac dysfunction. In this context, TCAs are metabolized in the liver by three oxidative pathways:

N-demethylation (CYP 1A2, 2C19, and 3A3/4)
N-oxidation (CYP 2D6)
Aromatic hydroxylation (CYP 2D6)

Aromatic hydroxylation is the most important of these because it is the principal pathway mediating the elimination of these drugs. This is true for all patients except those deficient in CYP 2D6, either because of genetics or because of inhibition by a coprescribed drug (e.g., fluoxetine, paroxetine).

The ratio of parent drug to demethylated metabolite at steady state ranges from 0.47 to 0.70 for imipramine and desipramine and from 0.83 to 1.16 for amitriptyline and nortriptyline. These typical ratios help distinguish between an acute overdose (increased ratios) versus steady state (normal ratios).

MONOAMINE OXIDASE INHIBITORS

Only minimal information is available about the pharmacokinetics of the traditional MAOIs (e.g., phenelzine, tranylcypromine). Such data are probably less critical because MAOIs are consumed by their mechanism of action (i.e., irreversible inhibition of MAO by covalent binding to the enzyme) (193). This mechanism accounts for the fact that the traditional MAOIs have halflives of only 2 to 4 hours, but effects persist for an extended period because of the irreversible inactivation of their target. These MAOIs undergo presystemic or first-pass degradation; thus, genetic or acquired alterations in metabolism could alter their bioavailability

Therapeutic Drug Monitoring

As discussed in Chapters 1 and 2, the basic concept underlying TDM is that a relationship must exist between drug concentration and the magnitude of its effect. Drug dose is simply the first approximation of the concentration achieved in an average individual. TDM is a refinement of this approximation, establishing the concentration achieved in a specific individual on a given dose. Since concentration is determined by dosing rate divided by clearance, TDM provides a measure of an individual’s ability to clear, or eliminate, a drug. Thus, TDM is principally a tool to determine how interindividual variability in pharmacokinetics accounts for interindividual variability in drug response.

TDM is not needed with most antidepressants but is recommended with TCAs due to their narrow therapeutic index. However, TDM can be used with any antidepressant to ensure adherence or to detect unusual metabolism, which might make the patient either unusually sensitive (i.e., substantially lower than usual clearance) or resistant (i.e., substantially higher than usual clearance) to usual doses of these drugs.

TCAs provide an excellent example of how biological variance in clearance can make certain patients outliers on the standard dose-response curve. Patients fall into three groups with regard to their ability to clear TCAs:

The majority (i.e., greater than 90%) are normal or “extensive” metabolizers in whom levels of 0.5 to 1.5 ng/mL/mg/day develop.
Members of a smaller group (i.e., 5% to 10%) are genetically deficient in CYP 2D6 and, hence, are poor metabolizers in whom plasma drug levels in the range of 4.0 to 6.0 ng/mL/mg/day develop.
An even smaller group of patients (i.e., 0.5%) are ultrarapid metabolizers of TCAs in whom levels of less than 0.5 ng/mL/mg/day develop (194).

There is a linear relationship between dose and plasma drug levels (i.e., linear or first-order pharmacokinetics) in normal and ultrarapid metabolizers. In poor metabolizers, TCAs follow nonlinear pharmacokinetics (i.e., disproportionate increases in plasma drug levels with dose increases) because they lack CYP 2D6 and must use lower affinity isoenzymes to metabolize these drugs.

In summary, TDM enhances the clinician’s ability to adjust the dose of medications rationally to compensate for interindividual differences in elimination rate. It enables more precise monitoring of the dose-response relationship in a given individual and can be seen as a step beyond the traditional but often inefficient and error-prone approach of dose adjustment based on clinical assessment.

SELECTIVE SEROTONIN REUPTAKE INHIBITORS

Given the wide therapeutic index of the SSRIs, the safety issues that make TDM a standard of care with the TCAs are not applicable to this class. Although the risk and severity of their typical adverse effects (e.g., nausea) increase with dose escalation, there is almost no chance of life-threatening toxicity. The aim of TDM with the SSRIs, therefore, is to increase the percentage of patients who will experience an optimal antidepressant response.

Although there are studies with most SSRIs attempting to correlate antidepressant efficacy with plasma drug levels, they have consistently failed to find a relationship (195,196,197,198,199 and 200). However, a relationship has been found between the plasma levels of each SSRI achieved on the lowest usually effective dose and the ability to inhibit approximately 70% to 80% of serotonin uptake (90). Thus, patients who have SSRI plasma levels below this threshold probably have not had an optimal trial of the SSRI as a result of rapid clearance, inadequate dose, or nonadherence.

A unique role for TDM with SSRIs is to determine the risk and possible magnitude of a CYP isoenzyme-mediated drug-drug interaction. As with most drugs, the effect of SSRIs on CYP isoenzymes is concentration dependent. Thus, TDM can determine the likely magnitude of the effect of an SSRI on the metabolism of another drug. This is relevant primarily for fluoxetine, fluvoxamine, and paroxetine because of their substantial inhibitory effect on one or more CYP isoenzymes at their lowest effective dose. This is particularly true for fluoxetine and its active metabolite, which because of their long half-lives can persist for several weeks after discontinuation. If the physician wishes to start another medication that could interact pharmacodynamically or pharmacokinetically with fluoxetine or norfluoxetine, then TDM could determine when the plasma levels have decreased sufficiently so as not to put the patient at substantial risk.

SEROTONIN AND NOREPINEPHRINE REUPTAKE INHIBITORS

Venlafaxine/Desvenlafaxine

As with SSRIs, TDM is not a standard of care issue with these agents. One study demonstrated that the sequential effects of venlafaxine on 5-HT and then NE reuptake pumps are dose and, hence, concentration dependent (200). Thus, TDM can be used with venlafaxine in the same manner as with the SSRIs. There is no data with DVEN, but the general comments here and in Chapters 1 and 2 likely apply.

Duloxetine

There is limited preliminary data on the usefulness of TDM with duloxetine indicating a curvilinear quadratic relationship with optimal anxiolytic effect occurring at intermediate plasma levels (201).

AMINOKETONES

Bupropion

This antidepressant comes the closest to the TCAs in terms of the potential usefulness of TDM. As discussed previously, it has a narrow therapeutic index (i.e., the difference between the antidepressant dose and a dose that carries a significantly increased seizure risk), and considerable interindividual variability in its plasma levels and active metabolites among patients on the same dose.

In addition, studies indicate a relationship between trough steady state plasma concentrations of 50 to 100 ng/mL and optimal response. Moreover, higher levels of the parent compound and its metabolites are associated with a poorer outcome (202).

The following points indicate that pharmacokinetics is a significant contributor to the occurrence of seizures in patients taking bupropion:

The incidence of seizures is dose related and, hence, concentration related.
Seizures typically occur within days of a dose change and a few hours after the last dose, suggesting that peak plasma concentrations play a role.
Individuals with lean body mass (i.e., smaller volume of distribution), such as anorexic-bulimic patients, are at increased risk.

Thus, TDM can guard against the development of unusually high plasma levels of the parent drug or its metabolites, particularly for patients who are medically compromised or on other drugs that could interfere with the clearance of bupropion. In such cases, the laboratory should assay the parent drug and its three major metabolites— hydroxybupropion, threohydrobupropion, and erythrohydrobupropion.

TRIAZOLOPYRIDINES

Nefazodone and Trazodone

The pharmacokinetics of these two phenylpiperazines makes TDM technically difficult. The half-lives of the parent drugs (i.e., trazodone and nefazodone) are approximately 4 hours, which means that modest differences in sample timing (i.e., time after last dose) could cause significant differences in observed plasma levels of the parent compounds (203). These drugs also have nonlinear pharmacokinetics, such that modest differences in adherence could substantially affect plasma drug levels.

Furthermore, they also undergo extensive biotransformation to form several active metabolites (190,191,204). For trazodone, this includes the production of mCPP (188). In the case of nefazodone, one of these metabolites (hydroxynefazodone) has a pharmacological profile similar to the parent drug, another (triazoledione) reproduces only part of the pharmacology of the parent drug, and the third (mCPP) has a pharmacological profile substantially different from the parent drug (205). This further complicates any attempt to relate drug plasma levels to clinical efficacy because each metabolite is likely to contribute to the clinical outcome. Their contributions, however, may be different from and even opposite to that of the parent drug. For these reasons, TDM research would be exceedingly difficult and unlikely to generate clinically useful therapeutic ranges.

TRICYCLICS

Efficacy

Therapeutic ranges associated with optimal antidepressant response and avoidance of toxicity is established for the following TCAs:

Amitriptyline (80 to 150 ng/mL)
Desipramine (110 to 160 ng/mL)
Imipramine (≤265 ng/mL)
Nortriptyline (50 to 150 or 170 ng/mL, depending on the study)

A summary of the plasma concentration-efficacy data with these four TCAs supports the use of TDM, at least once, as a routine aspect of therapy for MDD. The data are consistent across three of these agents that optimal plasma levels are associated with a greater likelihood of full remission after 4 weeks. Translated into clinical terms, Perry et al. found a 1.7- to 3-fold increase in clinical response to TCAs if the depressed patient obtains an optimal plasma level (206).

Safety and Tolerability

At levels above 450 ng/mL, there is both an increased risk and increased severity of CNS effects (i.e., seizures and delirium) and cardiac effects (i.e., conduction disturbances, which can lead to sudden death).

Central Nervous System Toxicity. The relationship between TCA plasma concentration and CNS toxicity is well established (207). Furthermore, these symptoms often evolve insidiously, such that they can mimic the depressive episodes that these medications are being used to treat. An increase in affective symptoms (e.g., deterioration in mood, poor concentration, social withdrawal, and lethargy) may be the earliest warning signs of impending CNS toxicity. Motor symptoms (e.g., tremor and ataxia) often develop next, followed by psychosis (e.g., thought disorder, delusions, and hallucinations). All of these may lead the clinician to erroneously conclude that the depressive episode is worsening, prompting an increase in the TCA dose or the addition of an antipsychotic. The latter strategy, in turn, can increase the TCA plasma levels by inhibiting metabolism, further exacerbating toxicity. The last stage in the evolution of TCA-induced CNS toxicity is delirium (e.g., memory impairment, agitation, disorientation, and confusion) and/or seizures. TDM can detect the slow metabolizer who is at greater risk for developing this scenario.

In patients who are not neurologically compromised, TCA plasma concentrations less than 250 ng/mL rarely produce CNS manifestations. As this threshold concentration is exceeded, asymptomatic, nonspecific EEG abnormalities develop. With concentrations beyond 450 ng/mL, there is a significant increased risk of seizures and delirium (208). Of note, peripheral anticholinergic effects (e.g., blurry vision, dry mouth, and constipation) are not a foolproof means of detecting potentially toxic plasma levels of TCAs.

A TCA-induced seizure typically has no prodrome and is a single-generalized motor seizure that lasts several minutes. The incidence of seizures during treatment with standard doses is estimated at 0.5% in nonepileptic patients (209). Patients who experience TCA-induced seizures during routine therapy and with no other risk factors for seizures generally have plasma TCA levels well in excess of 450 ng/mL (208,210).

Finally, the minimum threshold for development of coma is approximately 1,000 ng/mL (211). This occurs almost exclusively in acute overdoses but nevertheless is consistent with the conclusion that TCA-induced CNS toxicity is concentration dependent.

Cardiovascular Toxicity. The cardiovascular effects of TCAs are well documented, and the mechanisms underlying these effects elucidated by in vitro and in vivo animal studies (212). Because the effects of TCAs on intracardiac conduction are concentration dependent and occur at levels above the upper therapeutic threshold, TDM can help avoid iatrogenic cardiotoxicity

Although conduction disturbances are more likely to occur in persons predisposed to cardiac disease, data demonstrate that these effects also occur in healthy individuals if the appropriate threshold is exceeded. In healthy middle-aged subjects, TCA plasma concentrations of less than 200 ng/mL rarely induce intracardiac conduction defects. At 200 ng/mL, however, asymptomatic clinically nonsignificant slowing of the His bundle-ventricular system routinely occurs (213). At concentrations of more than 350 ng/mL, first-degree atrioventricular (A-V) block was found in 70% of physically healthy patients on desipramine or imipramine (214,215). A number of case reports demonstrate that TCA plasma levels near 1,000 ng/mL, achieved as a result of slow clearance during routine treatment rather than from overdose, can cause fatal arrhythmias (216).

Clinicians should be aware that significant postmortem changes in TCA plasma levels may occur, and interpretation must be made cautiously. Sudden death may only be coincidentally associated with a drug the patient was taking. Nevertheless, if a patient is found to have a toxiclevel postmortem and no anatomical cause of death at autopsy, the clinician may have a legal problem if TDM was not used at least once early in treatment to rationally adjust the dose.

Accurate Assessment of Tricyclic Plasma Levels

To obtain an accurate TCA plasma level, start with a standard TCA dose (e.g., 50 to 75 mg/day of nortriptyline) in a physically healthy adult. After 1 week, most patients are at steady state. A blood sample is then drawn 10 to 12 hours after the last dose to ensure that absorption and distribution of the drug are complete and because virtually all the data on optimal concentration ranges are based on this postdose time interval.

Single-dose prediction strategies can more rapidly predict the dose required to achieve a therapeutic plasma concentration. The utility of this approach is limited, however, because errors in technique (e.g., imprecise timing of blood draw) can be magnified, leading to miscalculation of the required dose.

MONOAMINE OXIDASE INHIBITORS

Phenelzine is used for the treatment of anxiety, phobic, and obsessive-compulsive disorders as well as for typical and atypical depressions. Its antidepressant efficacy is correlated with an 80% to 85% inhibition of the MAO enzyme. Studies indicate that 60 mg/day are usually needed to inhibit MAO by at least 80% (217,218). Although monitoring of platelet MAO inhibition during treatment permits more optimal dosing, the assay is not readily available commercially. This, coupled with the infrequent use of MAOIs, hampers the application of this specialized form of TDM.

CONCLUSION

There are three possible relationships between plasma concentration and efficacy:

None or a poor relationship between plasma level and therapeutic response
A threshold for therapeutic response, such that below this level, there is less likelihood of response and above it, there is a better chance for response
A threshold for toxicity such that above this level, the potential for increased adverse effects outweighs the likelihood of further therapeutic gain

In this context, TDM with antidepressants can be used when

Plasma levels are in the usual optimal range, but there is insufficient clinical response: try another agent, an augmentation strategy, or ECT (see also “Alternative Treatment Strategies” later in this chapter)
Plasma levels are low and there is insufficient response: increase the dose
Plasma levels are high and there is insufficient clinical response or there are severe adverse effects: reduce the dose or try another agent

Management of an Acute Depressive Episode

The treatment of major depression is analogous to the treatment of many medical conditions in that they are benefited by several classes of medications with different mechanisms of action and adverse effects.

In this section, response and remission rates are often quoted. Response is usually defined as a 50% or greater reduction in baseline symptom severity as measured by a standardized rating assessment such as the Hamilton Depression Rating Scale (HDRS) or the Montgomery-Asberg Depression Rating Scale (MADRS). The drawback to this approach is that it does not differentiate between partial and complete response, particularly when the initial symptom severity is high. Thus, a patient could be classified as a responder and still be quite symptomatic. In some instances, a patient is classified as a responder and still meets entry requirements for an antidepressant clinical trial based on end of treatment symptom severity.

Remission means that the symptom severity is below a predetermined cutoff on a standardized rating assessment (219). The cutoff score is generally such that nondepressed or never depressed individuals could reach that level. There is no universally agreed upon cutoff to define remission (220). For example, a cutoff of 5 is sufficiently low that the percentage of patients achieving this score is quite low, whereas a cutoff of 15 is so high that the difference between response and remission rates is virtually nonexistent. Most studies use a cutoff between 7 and 10 (221).

The 17-item HDRS was the most commonly used scale in these antidepressant clinical trials. This version of the HDRS is heavily weighted toward melancholic symptoms. There are also 21-, 24-, and 28-item versions with the additional items assessing nonmelancholic symptoms. The MADRS and the Inventory of Depression Symptomatology (IDS) are other instruments used in antidepressant clinical trials. Increasingly, however, there is concern regarding the adequacy of these scales to assess medication-related changes in depression (222,223). In this context, self-report scales (e.g., BDI; IDS-SR) and functional outcome measures are increasingly used in clinical practice to better assess medication impact (i.e., measurement-based care).

TABLE 7-2 FIRST-GENERATION ANTIDEPRESSANTS VERSUS PLACEBO: ACUTE TREATMENT

			Responders (%)
Number of Studies		Number of Subjects	Drug (%)	Placebo (%)	Difference (%)	Chi Square	p Value
Combined TCAs
	79	5,159	63	36	27	365	<10^-40
Combined MAOIs
	16	1,697	66	32	35	49.9	2 × 10^-12
MAOI, monoamine oxidase inhibitor; TCA, tricyclic antidepressant

Severe depression is another term commonly seen in the literature. Its assessment takes on increasing importance given recent evidence that the magnitude of antidepressant benefit compared with placebo may increase as symptom severity increases (224).

One or more of the following criteria are often used to differentiate nonsevere from severe depression:

Hospitalization
Functional impairment as assessed by a scale such as the Global Assessment Scale
Absolute symptom severity as assessed by a standardized rating assessment
Depressive subtype, particularly psychotic or melancholic as defined in Diagnostic and Statistical Manual of Mental Disorders, 4th edition, Text Revision (DSM-IV-TR)

None of these definitions is ideal. Patients may be hospitalized for a suicide gesture or a comorbid condition such as a severe personality disorder or substance abuse rather than solely on the basis of their depression severity. A score of 25 or higher on the 17-item HDRS is the most commonly used cutoff to distinguish nonsevere from severe depression. This approach, however, is subject to the possibility of inflation of scores to qualify patients for the trial.

A related issue is the onset of antidepressant action. Presently, most agents usually take several weeks to achieve their intended effects. Increasingly, however, the field is attempting to develop paradigms that achieve a more rapid onset (225,226 and 227).

Another concern is the impact that selective publication may have in accurately assessing the efficacy of antidepressants (228,229). These issues must be kept in mind when interpreting claims made on the basis of clinical trial results (see Chapter 3). The variability in scales and definition of terms used in antidepressant trials also confounds attempts to make comparisons across studies and to do meta-analyses. With these caveats in mind, the next several sections review the acute and maintenance efficacy of the available and investigational antidepressant options.

There are several hundred well-controlled studies and many more partially controlled or open trials supporting the efficacy of firstgeneration antidepressants for major depression (Table 7-2). In severely-ill patients, these agents produce a striking improvement in behavior and a marked lessening of depression, generally beginning 3 to 10 days after their initiation. The rate of response is linear over time, with a “half-life to improvement” of about 10 to 20 days. Consequently, patients who do not demonstrate a satisfactory response after an adequate trial for a 4- to 6-week period probably will not. The degree of response in the first 2 weeks of treatment may predict the ultimate outcome, with 60% to 80% of depressed patients substantially benefited (e.g., >50% improvement from baseline rating scores) by marketed antidepressants during this period, in contrast to 20% to 50% of those on placebo.

We calculated or provided other summaries of the results from double-blind, random-assignment studies (usually class I or II designs) comparing SSRIs, other new antidepressants, and older antidepressants (TCAs and MAOIs) with placebo or with each other for the acute treatment of depression.

To complement these controlled trials, we also provide the results of the STAR*D study when relevant to the discussion. This National Institute of Mental Health (NIMH)-sponsored effectiveness study involved 4,041 outpatients with unipolar major depression (230). The primary outcome was remission [i.e., not response as in randomized controlled trials (RCTs)] defined by a final score of ≤7 on the 17-item HDRS. The initial phase was an open-label trial with citalopram (mean exit dose = 41.8 mg/day) for 12 weeks. Those who did not achieve remission or did not tolerate treatment could then participate in three additional sequential levels involving various switch or augmentation options until they met remission criterion.

SELECTIVE SEROTONIN REUPTAKE INHIBITORS

As noted earlier, there is strong evidence that 5-HT neurotransmission is altered during an episode of depression. In this context, drugs that modify 5-HT activity by inhibiting its reuptake transporter are the most clinically productive treatment strategy thus far.

Zimelidine, the first serotonin reuptake inhibitor available for clinical use, was withdrawn worldwide in 1982 because of its toxicity (231). Despite this initial setback, several members of this class are presently marketed:

Citalopram. (Celexa)
Escitalopram (Lexapro)
Fluoxetine (Prozac)
Fluvoxamine (Luvox)
Paroxetine (Paxil)
Sertraline (Zoloft)

All have indications in the United States for the treatment of major depression except fluvoxamine. This agent is approved only for the treatment of obsessive-compulsive disorder; however, it is marketed in other countries for major depression.

The development of enantiomers of racemic agents or the active metabolite of any agent is not unique to psychiatric medications but is a general strategy of pharmaceutical companies to extend product life. For example, escitalopram is the active S-enantiomer of citalopram. Active refers to having high potency for inhibiting the 5-HT transporter protein. In fact, all of the SSRIs with the exception of fluvoxamine are chiral (i.e., have an asymmetrical carbon) and thus are natural racemates. In the case of paroxetine and sertraline, only the active enantiomers were originally developed and marketed. In the case of citalopram and fluoxetine, the racemates were initially developed and marketed. With patent loss looming, however, efforts were undertaken by Eli Lilly and Lundbeck Pharmaceuticals to develop their active enantiomers. That effort was unsuccessful with fluoxetine because the active enantiomer blocks K⁺ channels and prolongs the QTc interval. Efforts with citalopram, however, were successful.

A parallel strategy is to develop the active metabolite of a product (e.g., DVEN).

The development of enantiomers and metabolites has several advantages over the development of completely new molecules:

Less expense
Less time
Less risk of developing a nonmarketable product at the end of the process
The pharmaceutical company is able to maintain and perhaps grow its franchise in the marketplace

The issue is always whether the enantiomer or active metabolite is a sufficient advance to warrant the cost differential when the original agent goes generic. This is often difficult to prove unequivocally with psychiatric medications because of the heterogeneity in drug- placebo response (232).

Efficacy for Acute Treatment

The overall efficacy and effectiveness among the SSRIs for MDD is remarkably similar, consistent with the hypothesis that they have the same mechanism of action (i.e., serotonin reuptake transporter inhibition). Differences in onset of action, adverse effects, pharmacokinetics, and potential for drug interactions may help dictate the choice of a specific agent (233). Of interest, recent meta-analyses suggest that sertraline may be the SSRI of first choice based on relative efficacy, acceptability, and acquisition cost (234,235).

Most clinical trials demonstrate that the various SSRIs are superior to placebo (Table 7-3) (236,237,238,239,240,241,242,243,244,245,246,247,248,249,250,251,252,253,254,255,256,257,258,259,260,261,262 and 263). Comparative studies also show that the SSRIs are equal in effectiveness to each other as well as to other classes of antidepressants (Table 7-4) (255,264,265,266,267,268,269 and 270). Even though the large number of patients studied and the failure to find a significant difference strongly suggest that SSRIs are equal to each other and comparative antidepressants, one important limitation is that the studies were generally based on depressed outpatients.

TABLE 7-3 SSRIS VERSUS PLACEBO: ACUTE TREATMENT

			Responders (%)
Drug	Number of Studies	Number of Subjects	SSRI (%)	Placebo (%)	Difference (%)
Citalopram	5	1,352	54	37	+17
Escitalopram	3	1,055	57	37	+20
Fluoxetine	9	1,365	65	41	+24
Fluvoxamine	3	125	67	42	+25
Paroxetine	9	649	65	36	+29
Sertraline	3	575	78	48	+30
SSRI, selective serotonin reuptake inhibitor

All of the SSRIs also show a flat dose-response curve, meaning that there is usually no advantage to increasing the dose beyond the usually effective minimum amount. Of interest, the SSRIs at their usual effective therapeutic dose (i.e., 40 mg/day for citalopram; 20 mg/day for escitalopram, fluoxetine, and paroxetine; and 50 mg/day for sertraline) produce comparable effects on either plasma serotonin levels or the serotonin reuptake pump in platelets (89), consistent with the conclusion that reuptake inhibition is the mechanism responsible for their antidepressant efficacy.

Rate of Onset of Efficacy. There are several randomized, double-blind studies suggesting that fluoxetine has a slower onset of antidepressant activity than other antidepressants, including citalopram (266), moclobemide (267), paroxetine (268), and venlafaxine (269). That may be because effective levels are not usually achieved in a comparable interval of time as in other SSRIs when the patient is started on the lowest effective dose (i.e., 20 mg/day). Nevertheless, although the magnitude of the difference in early response was statistically greater for the comparative drug in these studies, the absolute magnitude of the difference in effective dose was modest (245,269).

TABLE 7-4 SSRIS VERSUS FIRST-GENERATION ANTIDEPRESSANTS: ACUTE TREATMENT

			Responders (%)
Drug	Number of Studies	Number of Subjects	SSRI (%)	FG ADs (%)	Difference (%)
Citalopram	6	347	73	74	-1
Escitalopram	4	1,167	62	57	+5
Fluoxetine	16	1,549	63	64	-1
Fluvoxamine	4	137	70	66	+4
Paroxetine	16	1,322	62	60	+2
Sertraline	3	682	68	65	+3
AD, antidepressant; FG, first generation; SSRI, selective serotonin reuptake inhibitor

Unique Spectrum of Efficacy. Although there is ample evidence for efficacy in minor depression or mild-to-moderate depression, there is some debate as to how effective SSRIs are for more severe episodes, particularly in those who require hospitalization (245). This may in part be due to the vast majority of studies with SSRIs involved outpatients. As mentioned earlier, hospitalization is not simply determined by severity of an episode but also driven by other variables such as concomitant substance abuse or medical illness.

There have been five, double-blind studies comparing the antidepressant efficacy of different SSRIs versus different TCAs in patients with HDRS scores of 25 or more (246,247,248,249 and 250). Three of these studies permitted inclusion of both inpatients and outpatients (246,247 and 248), whereas the other two included only outpatients (249,250). The three, placebo-controlled (247,249,250) studies found that the SSRI (i.e., fluvoxamine, paroxetine, or sertraline) was either superior to both the TCA and placebo or was comparable with the TCA and superior to placebo. In the other two studies, the SSRI was not different from the TCA, and there was no placebo control. There also have been four studies and one meta-analysis of European clinical trials that found no difference in antidepressant efficacy between several different SSRIs and several different tertiary amine TCAs in patients hospitalized for major depression (251,252,253,254 and 255). Finally, there are two relatively small studies showing that fluoxetine and fluvoxamine both had antidepressant efficacy superior to placebo in patients with melancholia (256,258). Another larger study failed to find a difference between paroxetine and amitriptyline in treating such patients (258).

Studies that do not support comparable efficacy include two double-blind, active-controlled trials that found clomipramine produced a superior response to either paroxetine or citalopram in the treatment of patients hospitalized for major depression (259,260). Two double-blind trials also reported that venlafaxine and mirtazapine were more effective than fluoxetine in patients hospitalized with depression (261,262,263,264,265,266,267,268 and 269). Finally, there are studies showing that the addition of desipramine (one of the most selective NE reuptake inhibitors) to an SSRI can convert nonresponders or partial responders to full response (263,271).

There is also some debate as what course of action to follow if a patient does not respond to the first trial of an SSRI. If the reason for nonresponse is poor tolerability, then many clinicians try a second SSRI. There is also a modest amount of evidence that nonresponders to TCA monotherapy, principally desipramine or imipramine, may respond to an SSRI alone and vice versa (272). Based on a survey, the course of action preferred by most psychiatrists is to switch to a drug with a different mechanism of action or a dual mechanism of action when a patient experiences inadequate efficacy from an adequate trial of an SSRI (273). Results of the STAR*D study, however, found that a similar percentage of initial nonremitters to citalopram achieved remission when switched to an alternate SSRI (i.e., sertraline), bupropion-SR, or venlafaxine-XR (230,274).

SEROTONIN AND NOREPINEPHRINE REUPTAKE INHIBITORS

Venlafaxine

Venlafaxine is a phenylethylamine that sequentially inhibits the neuronal uptake pump for 5-HT and then NE (200). In contrast to tertiary amine TCAs, which also inhibit 5-HT and NE uptake pumps, venlafaxine has low binding for most other neuroreceptors and does not inhibit Na⁺ ion fast channels, making it relatively safe in overdoses compared with TCAs. CYP 2D6 converts venlafaxine into ODV, which has pharmacological activity comparable with the parent drug.

Efficacy for Acute Treatment. In contrast to the SSRIs, venlafaxine has an ascending dose-response curve consistent with its sequential, dose-dependent effects on 5-HT and NE uptake pumps. At 225 mg/day, the magnitude of the antidepressant effect is 50% higher than that seen with the SSRIs. Also consistent with its dual mechanism of action at higher concentrations, venlafaxine at a dose of 225 mg/day produced an antidepressant response in hospitalized patients with melancholia superior to both placebo and fluoxetine (269). In contrast to the flat dose-response curve seen with the SSRIs, the ascending dose-response curve seen with venlafaxine provides a stronger rationale to use higher doses in patients who have not responded or only partially responded to an initial lower starting dose (e.g., 75 mg/day).

A close examination of the efficacy data from the fixed dose study with venlafaxine affords an opportunity to comment on how study design can affect results. In this study, the magnitude of the antidepressant effect of venlafaxine was smaller at 375 mg/day than at 225 mg/day. Although possibly a chance result, this is readily understood based on the design and data analysis used for this study. In fixed dose studies, patients are randomly assigned to predetermined doses, regardless of need, and dosage is adjusted rapidly to that predetermined dose. This approach is in contrast to clinical practice, wherein patients are typically started on the lowest usually effective dose and only adjusted upward as needed and tolerated. Such adjustment is often done much more gradually than in a fixed dose study, and thus the patient may adjust to dose-dependent adverse effects. Most patients, for example, will adapt to the nausea produced by drugs that potentiate 5-HT, such as the SSRIs and venlafaxine.

This study, like most, was analyzed using a lastobservation-carried-forward (LOCF) approach, wherein the last observed data are carried forward if a subject drops out before the final assessment. The reason is to avoid missing data and biasing the results by looking only at patients who completed the study and thus benefited from and tolerated the treatment, whether placebo, investigational drug, or standard comparator drug. Based on this, a fixed dose design will often produce a curvilinear dose-response curve, with a better outcome in the middle than at either end. Response is decreased at the lowest dose because it is insufficient to produce therapeutic concentrations of drug at the necessary site of action in most patients. The blunted response at the upper end reflects the early dropouts who could not tolerate the therapy. Thus, the results in such studies may well underestimate the efficacy of the drug at the highest doses, in contrast to a design in which only patients who need the higher dose are advanced. These are common problems in the interpretation of all fixed dose studies.

Rate of Onset of Efficacy. There is a statistically significant early separation between venlafaxine and placebo in terms of average reduction in depressive symptoms over several double-blind, random-assignment clinical trials (275,276 and 277). These data are consistent with the observation from basic studies showing that combined administration of drugs that individually inhibit either 5-HT or NE uptake pumps, or the administration of high-dose venlafaxine, produces a more rapid β-adrenergic receptor subsensitivity (175). In addition, data from clinical studies using combined administration of NE and 5-HT reuptake inhibitors demonstrate a more rapid antidepressant action (269). These onset-of-response data provide physicians with clinically meaningful information when faced with severely-ill patients for whom shortening the time to onset of antidepressant action is particularly important. In this context, it would be ideal to have similar data on all available antidepressants so that more comparative statements about the relative onset of action can be made.

Unique Spectrum of Efficacy. An almost universal question when a new drug is marketed is whether it helps patients who have not responded to other medications. However, for the following reasons, there is rarely a good answer to this question:

Such studies are technically difficult
Such studies involve high risk (i.e., answers may not be favorable to the sponsor)
Trials during clinical development are designed to meet registration requirements rather than to answer such questions

As with other drugs, a methodologically rigorous study addressing this issue has not been done with venlafaxine.

Summary

Potential advantages of venlafaxine:

An ascending dose-antidepressant response curve, with possible greater overall efficacy at the higher doses
Lack of affinity for muscarinic, histaminergic, α-adrenergic, dopaminergic, and opioid receptors
Minimal effects on any of the major CYP isoenzymes in contrast to antidepressants such as fluoxetine, fluvoxamine, or paroxetine

Potential disadvantages of venlafaxine:

Possible increase in blood pressure
Sexual dysfunction
Wide variability in effective dose

The marketing of a once-a-day sustainedrelease (SR) form of venlafaxine effectively dealt with the need for a twice-a-day dosing schedule with the original immediate-release (IR) version. That the venlafaxine dose can be adjusted is a pharmacological advantage, but it may be a marketing disadvantage because many physicians may prefer the simplicity of a “one dose fits all” concept as seen with the SSRIs and possibly DVEN.

Desvenalfaxine

Efficacy for Acute Treatment. As noted earlier, the active metabolite of venlafaxine has the same mechanism of action as the parent compound. Several 8-week, double-blind, controlled trials generally, but not always, demonstrated superior acute efficacy for DVEN over placebo. DeMartinis et al. found that the 100 and 400 mg fixed dose of DVEN, but not a 200 mg fixed dose, produced significantly greater change scores on the HDRS-17 and response rates compared with placebo (278). Further, only the 400-mg dose produced significantly greater remission rates. By contrast, Liebowitz et al. did not observe a significant difference between DVEN (179 to 195 mg/day) and placebo on their primary or key secondary outcome measures (279). Septien-Velez et al. reported that 200 and 400 mg of DVEN were superior to placebo in change scores on the HDRS-17 and on various secondary measures. Only the 200-mg dose, however, produced a significantly greater remission rate versus placebo (280). Feiger et al. did not find a difference between DVEN (200 to 400 mg/day) and placebo (281). They noted that discontinuations due to AEs in the active treatment arm may have contributed to this outcome. Finally, Tourian et al. compared DVEN (50 or 100 mg/day) with placebo in a fifth trial and pooled the results from two other studies (282). Although this trial failed to meet its primary efficacy end point, a post hoc pooled analysis of the three trials using DVEN (50 and 100 mg fixed doses) found this agent effective relative to placebo.

Clayton et al. pooled the safety and tolerability data from nine trials using DVEN in doses ranging from 50 to 400 mg/day for MDD (283). They concluded that its adverse effect profile was similar to other selective norepinephrine reuptake inhibitors (SNRIs) with nausea the most commonly reported AE as well as the primary or secondary reason for drug discontinuation. Most AEs were dose related with the 50-mg dose comparable to placebo in terms of discontinuation due to adverse events.

Duloxetine

Duloxetine is a chiral compound [(+/—)-Nmethyl-3-(1-napthalenyloxy)-3-(2-thiophene) propanamine] formed from the building blocks of (S)-3-chloro-1-(2-thienyl)-1-propanil and the corresponding (R)-butanoate (284). Based on both in vitro and in vivo preclinical pharmacological evidence, duloxetine may be a more balanced inhibitor of 5-HT and NE reuptake pumps than venlafaxine (i.e., the ratio of its affinity constants for these two pumps is closer to 1). Like venlafaxine and DVEN, duloxetine has low affinity for muscarinic, H_1, adrenergic, DA, and serotonin receptors (285). This suggests a low potential for causing adverse effects mediated by these mechanisms, and was confirmed in clinical trials.

The ability of duloxetine to block both uptake pumps in vivo was also demonstrated in a number of animal models of depression (286,287,288,289,290,291,292,293,294,295,296 and 297). In addition to an indication for the treatment of major depression and diabetic peripheral neuropathic pain, duloxetine is also approved for

Stress urinary incontinence
Fibromyalgia
Generalized anxiety disorder

Efficacy for Acute Treatment. The acute efficacy of duloxetine in the treatment of adult patients (18 to 83 years of age) with major depression was established by five double-blind, placebo-controlled, random-assignment studies lasting 8 to 9 weeks (298,299,300,301,302 and 303). In two studies, patients were randomized to either duloxetine 60 mg once a day (n = 123 and 128, respectively) or placebo (n = 122 and 139, respectively) for 9 weeks. In a third study, patients were randomized to either 20 or 40 mg of duloxetine twice daily (n = 86 and 91, respectively) or placebo (n = 89) for 8 weeks. In the fourth study, patients were assigned to either 40 or 60 mg of duloxetine twice daily (n = 95 and 93, respectively) or placebo (n = 93) for 8 weeks. In the fifth study, subjects who entered the 8-week trial were assigned to placebo (n = 70) or duloxetine (n = 70) given in a forced titration regimen (i.e., 20 µg twice a day to 60 µg twice a day over the first 3 weeks). In all five studies, duloxetine was superior to placebo (and comparable to SSRIs used in three of these trials) in reducing depressive symptoms as measured by the 17-item HDRS. Analysis of these studies did not reveal any relationship between antidepressant efficacy and age, gender, or race, although the vast majority of patients in these trials were Caucasians.

Duloxetine response rates (i.e., rate on drug minus rate on placebo) at doses of 60 and 120 mg/day were 23% and 16%, respectively (298,301). Duloxetine remission rates (i.e., rate on drug minus rate on placebo) at doses of 60 and 120 mg/day were 15% and 24% to 28%, respectively (298,300,301). A more recent analysis indicates doses of 80 to 120 mg may be optimal and that this agent remains inadequately tested against standard alternatives (302). These rates are comparable to those of other marketed antidepressants. Duloxetine-treated patients also experienced a greater reduction in associated painful physical symptoms commonly seen in depressed patients, including overall pain, back pain, shoulder pain, and time in pain while awake (298,300). In this context, the results of trials utilizing duloxetine for treatment of fibromyalgia with or without concomitant depression led to an FDA approval for this indication (304,305). Of note, a recent trial and meta-analysis indicate that these antinociceptive properties may be common with most antidepressants (306,307).

Rate of Onset of Efficacy. There is no evidence to date that duloxetine produces a faster onset of action than other marketed antidepressants.

Summary

Potential advantages of duloxetine:

Equal impact on 5-HT and NE reuptake across dosing range
Improves pain symptoms and fibromyalgia syndrome
Effective for generalized anxiety disorder
Safe in overdose
Does not cause clinically significant elevation in blood pressure or heart rate
Lack of affinity for muscarinic, histaminergic, α-adrenergic, dopaminergic, and opioid receptors

Potential disadvantages of duloxetine:

May worsen narrow-angle glaucoma
May elevate liver enzymes
Gastrointestinal symptoms: nausea, constipation
May cause sexual side effects or urinary hesitancy
May produce discontinuation syndrome
May produce clinically relevant drug interactions (e.g., via CYP 2D6)

AMINOKETONES

Bupropion

Bupropion is an established antidepressant that has gone through several reformulations, so it still remains a patented product. It entered clinical trials in the mid-1970s and gained FDA approval before fluoxetine, but its marketing was delayed due to concerns about seizure risk. Bupropion was originally marketed in the late 1980s in an IR formulation that required three daily doses. Bupropion-IR never gained appreciable market acceptance, largely because of its complicated dosing requirements and ongoing concerns about seizures. In the early 1990s, an SR formulation was developed (but it still required twice daily dosing), and a marketing effort was undertaken to reduce physicians’ concern about seizure risk. That effort, plus its approval as an aid to smoking cessation, led to a sizable increase in its market share. In the early 2000s, a third extended release (XL) formulation was introduced, and market share continued to increase. Bupropion is often used in combination with SSRIs to augment antidepressant response as well as to treat adverse effects caused by excessive 5-HT agonism, particularly sexual dysfunction and sedation.

Bupropion is a weak dual reuptake inhibitor of DA and NE. There are a number of observations in animals that bupropion can in vivo block the uptake of both DA and NE, including the downregulation of postsynaptic β-noradrenergic receptors and protection against the DA neurotoxictiy of 6-hydroxydopamine (308). Bupropion is also self-administered in animals when using the paradigm for amphetamine-like drugs. The combined plasma concentration of bupropion and its three active metabolites is consistent with the conclusion that the inhibition of both of these uptake pumps likely occurs with clinically relevant doses.

TABLE 7-5 SPECIFIC ANTIDEPRESSANTS VERSUS PLACEBO: ACUTE TREATMENT

				Responders (%)
Drug		Number of Studies	Number of Subjects	Drug (%)	Placebo (%)	Difference (%)	Chi Square	p Value
First Generation
	Amitriptyline	8	292	60	26	34	30.9	3 × 10^-8
	Imipramine	50	2,649	68	40	28	184.0	<10^-40
Second Generation
	Amoxapine	10	386	67	49	18	12.4	4 × 10^-4
	Bupropion	4	425	55	29	26	26.6	2 × 10^-7
	Mianserin	5	336	60	28	31	31.0	2 × 10^-8
	Trazodone	13	824	59	28	32	93.3	1 × 10^-22

Efficacy for Acute Treatment. As discussed later, the pivotal antidepressant efficacy data are based on several early, double-blind studies done with the IR formulation (Tables 7-5 and 7-6) (309). Since then, the published efficacy trials with the other bupropion formulations were usually active controlled designs in which the comparison was another marketed antidepressant. A meta-analysis of seven randomized controlled trials (six with the SR formulation) found bupropion equivalent to SSRIs in terms of response and remission rates (i.e., 62% vs. 63% and 47% vs. 47%, respectively) (310). Since these studies are not sufficient to prove efficacy to the standard required by the FDA, the approval of the subsequent formulations was based on pharmacokinetic considerations rather than efficacy data.

The high levels of bupropion and its metabolites achieved with usual therapeutic doses may also account for its narrow therapeutic index in terms of causing seizures. The weak in vitro activity of bupropion is consistent with its marginal evidence of antidepressant activity and should be kept in mind when evaluating the efficacy data. In fact, the FDA concluded that all three double-blind, placebo-controlled studies done to support the approval of the SR formulation failed to show that the SR version of bupropion in the doses used (i.e., ≤400 mg/day) was superior to placebo in the treatment of outpatients with major depression (FDA summary basis for approval of bupropion). As a result, this formulation was approved on the basis of bioequivalence with the IR formulation. The FDA in its approval documents did not specify why it chose to approve this formulation without efficacy data. There is reason to believe that SR of bupropion may reduce the risk of seizures compared to that seen with the IR version at comparable doses. In contrast to these failed studies, two relatively small studies published in 1983 reported that IR bupropion at doses up to 600 mg/day was superior to placebo in the treatment of inpatients hospitalized for major depression (311,312).

TABLE 7-6 SECOND-GENERATION VERSUS FIRST-GENERATION ANTIDEPRESSANTS: ACUTE TREATMENT

			Responders (%)
Drug	Number of Studies	Number of Subjects	New Antidepressant (%)	Standard Antidepressant (%)	Difference (%)
Amoxapine	19	784	79	73	6
Bupropion	4	293	71	72	-1
Clomipramine	6	350	61	62	-1
Maprotiline	20	1,638	73	72	1
Mianserin	15	1,155	58	64	-6
Trazodone	18	913	62	58	4

Summary

Potential advantages of bupropion:

Pharmacologically unique relative to other conventional antidepressants; most closely resembles the action of psychostimulants
May benefit patients who historically have not responded to TCAs (313)
Relatively free of sexual adverse effects, in contrast to the SSRIs or venlafaxine (314)
Useful in conditions not typically responsive to other antidepressants (e.g., adult attention deficit hyperactivity disorder [ADHD] (315) and smoking cessation (316))
Not sedating, low incidence of anticholinergic adverse effects, does not cause weight gain, and minimal effects on intracardiac conduction

Potential disadvantages of bupropion:

Possibly a less robust antidepressant effect
Possible increase in suicidality in younger patients
Increased risk for seizures (e.g., with overdose)
Contraindicated in those with eating disorders

TRIAZOLOPYRIDINES

Nefazodone

Nefazodone has a chemical structure related to trazodone and incorporates both 5-HT reuptake properties plus 5-HT_2A receptor blockade (317,318 and 319). It may be that the antidepressant effect of serotonin agents is due to mediation of 5-HT_1A transmission in the absence (or even the blockade) of 5-HT_2A transmission. As a result of these actions, nefazodone is even more specific in terms of affecting a subtype of serotonin receptor than the SSRIs or SNRIs.

Nefazodone requires divided doses and titration because of complicated pharmacokinetics and dose-dependent adverse effects. Dosing is made even more problematic because arguably more variability exists among patients in terms of the optimal dose of nefazodone with regard to efficacy and tolerability than of virtually any other antidepressant. For this reason, nefazodone never gained widespread market acceptance. In addition, there were reports of rare hepatotoxicity resulting in death or liver failure requiring transplantation, which led the FDA to require a box warning for the drug. These facts, coupled with a pending loss of patent protection, led the manufacturer to stop production of the drug. However, generic formulations are still available.

Efficacy for Acute Treatment. The clinical trials of nefazodone differed from those of the SSRIs and the SNRIs in that the only truly fixed dose study was done early in its development (320). That study had major limitations, such as the doses used were low (50 to 300 mg/day) compared with those later found to be optimally effective. Also, the number of subjects was small. Subsequent studies did not use a fixed dose design but rather a targeted dose range design (i.e., low dose = 50 to 300 mg/day and high dose = 300 to 600 mg/day) (320). Even though patients were randomly assigned to either range, the physician could adjust dosing within these ranges to maximize the treatment outcome. Although the use of fixed dose ranges is reasonable and valid, this design does not permit a straightforward assessment of the dose-response curve as do conventional fixed dose studies. This difference should be kept in mind, particularly when comparing the antidepressant effect size seen with nefazodone with that of other antidepressants that were studied using truly fixed dose designs.

Using this fixed dose range design, a series of double-blind, controlled trials with placebo and standard drug (mainly imipramine or fluoxetine) were made (204,320,321,322,323 and 324). Nefazodone was superior to placebo and equivalent to imipramine and fluoxetine in efficacy. A metaanalysis, including all the efficacy studies with nefazodone, examined the magnitude of the antidepressant response as a function of the average daily dose and found an ascending dose-antidepressant response curve (325). The magnitude of the antidepressant effect at doses of 300 to 400 mg/day is comparable with that seen with the SSRIs and lower doses of venlafaxine. This finding suggests that nefazodone at these doses is effective in the same percentage of patients as the other comparative antidepressants. At doses of 500 mg/day, however, the magnitude of the response exceeds that seen with the SSRIs but is less than that seen with high-dose venlafaxine (i.e., 225 to 375 mg/day).

Interestingly, at 600 mg/day, the response to nefazodone was virtually the same as to placebo. This finding may be an artifact of the study design, however, because the clinician could adjust the dosage within the dose range based on efficacy and tolerability. Hence, dosages for patients who failed to respond should have been adjusted to the highest tolerated dose. Such a design produces a curvilinear dose-response curve with a diminution in response at the highest level because dosages for nonresponders are adjusted upward to the highest tolerated dose. In addition, there is the problem of dropouts, which occurs because of the acute nuisance adverse effects at high doses, as discussed in the sections on SSRIs and venlafaxine. Nonetheless, this result indicates that there is generally no advantage to exceeding 500 mg/day of nefazodone.

Rate of Onset of Efficacy. Nefazodone produced a rapid (i.e., by 1 week) and statistically significant improvement in sleep and anxiety symptoms associated with major depression (326). The full antidepressant response, however, takes an interval of time comparable with that of the SSRIs.

Unique Spectrum of Efficacy. No studies demonstrate unique efficacy for nefazodone in any subgroup of patients with major depression. There is also no rigorous evidence to support the position that nefazodone is effective in patients who did not benefit from an adequate trial of another antidepressant. There is a double-blind, placebo-controlled study, however, showing nefazodone to be superior to placebo in patients hospitalized for major depression.

Summary

Potential advantages of nefazodone:

An ascending dose-antidepressant response curve in contrast to SSRIs
Perhaps a more benign adverse effect profile, particularly with regard to activation and sexual dysfunction
A potentially more rapid onset of anxiolytic effect
A lack of blood pressure elevation, in contrast to venlafaxine

Potential disadvantages of nefazodone:

Need for a twice-a-day dosing schedule
Need for dose adjustment
Rare but serious hepatotoxicity
Produces substantial inhibition of CYP 3A3/4, leading to the potential for adverse pharmacokinetic interactions

Trazodone

Several well-controlled trials found trazodone to be 32% more effective than placebo and a nonsignificant 4% more effective than tertiary TCAs (Tables 7-5 and 7-6). The therapeutic dose is 200 to 600 mg/day, and although it has fewer anticholinergic adverse effects, excessive sedation may limit higher doses. In fact, the most common clinical use of this agent is in lower doses as a sedative/hypnotic alternative to the BZDs. The absolute milligram dose is higher than for TCAs, with some patients needing 400 to 600 mg to achieve adequate response.

TETRACYCLIC AGENTS

Mianserin

Mianserin and its analog, mirtazapine (i.e., 6-azomianserin), are tetracyclic compounds and differ from other antidepressants in terms of the putative mechanism responsible for their antidepressant efficacy. Mianserin is the older drug and is marketed in several countries around the world but not in the United States.

Five double-blind studies found mianserin to be superior to placebo, but 15 studies found it to be slightly less effective than tertiary amine TCAs (Tables 7-5 and 7-6).

Mirtazapine

Mirtazapine entered the US market in August 1996 and is available in several other countries. The putative mechanism of action mediating antidepressant activity is the blockade of several serotonin (i.e., 5-HT_2A and 5-HT_2C) and α₂-adrenergic receptors (327). The latter effect increases NE release by a direct effect on the presynaptic α₂-adrenergic autoreceptor and indirectly increases serotonin release via the tonic effect on the adrenergic input to raphe neurons. In addition to these mechanisms of action, mirtazapine and mianserin also block histamine and specific 5-HT receptors (i.e., 5-HT_2A, 5-HT_2C, 5-HT₃) but have minimal affinity for muscarinic or α₁-adrenergic receptors and do not inhibit either NE or 5-HT uptake pumps.

Efficacy for Acute Treatment. Based on six double-blind, placebo- and amitriptylinecontrolled studies, a meta-analysis found mirtazapine superior to placebo and comparable to amitriptyline (328). In two studies, mirtazapine was efficacious in the treatment of patients hospitalized for major depression. In the first study, mirtazapine was comparable to amitriptyline and superior to placebo (329). In the second study, its antidepressant efficacy was superior to fluoxetine (261).

Rate of Onset of Efficacy. A recent systematic review and meta-analysis concluded that mirtazapine is likely to have a faster onset of action than the SSRIs, but no significant difference was evident after 6 to 12 weeks of treatment (262).

Summary

Potential advantages of mirtazapine:

Sedation; may be useful for agitation and hostility in geriatric patients
Safety in overdose
Minimal anticholinergic effects
No quinidine-like effect
No blood pressure effects, including orthostatic hypotension

Potential disadvantages of mirtazapine:

Daytime sedation
Increased appetite
Weight gain
Transient neutropenia (typically presents and resolves within first 6 weeks of treatment)

Transient mild elevations in liver function tests, particularly serum glutamic-pyruvic transaminase (SGPT; typically presents and resolves within first 6 weeks of treatment)
Rare adverse event of severe neutropenia (3 out of 2,796 patients); uncertain causal relationship to mirtazapine treatment

TRICYCLICS

TCAs include tertiary amines such as amitriptyline, doxepin, and imipramine; secondary amines such as desipramine and nortriptyline; and their analogs (e.g., maprotiline and amoxapine). For decades, these drugs were the cornerstone of treatment for patients suffering from clinical depression. They have since lost that position to the SSRIs and other newer antidepressants, principally because of their lethality in overdose and adverse effect profile. Because of their loss of preeminence as antidepressants, the discussion of these drugs was reduced and more details are in earlier editions of this textbook.

Efficacy for Acute Treatment

Tertiary Amine Agents. Tertiary amine agents include amitriptyline, clomipramine, doxepin, imipramine, and trimipramine. For each of these agents, studies demonstrate superiority over placebo and equivalence to other TCAs (Tables 7-5 and 7-7). In general, these drugs produce a 70% overall response rate compared to 30% to 40% for the parallel placebo condition.

Amitriptyline is still extensively used in the United States and other countries, often for other indications, such as

Chronic pain
Migraines
Insomnia

Clomipramine is approved in the United States only for the treatment of obsessive-compulsive disorder, but it is also used in most of the world as an antidepressant. Among TCAs, clomipramine is one of the more selective reuptake inhibitors of serotonin. Since plasma levels of its demethylated metabolite that primarily works through the NE system can exceed those of the parent compound, clomipramine’s therapeutic action cannot be solely attributed to serotonin reuptake inhibition.

Secondary Amine Agents. The secondary amine TCAs are more NE-selective antidepressants, while the hallmark of tertiary amine TCAs is their effects on multiple neurotransmitters.

Desipramine and nortriptyline are the secondary amine metabolites of imipramine and amitriptyline, respectively. They are comparable in efficacy to their tertiary amine parent compounds and clearly superior to placebo (Table 7-7).

TABLE 7-7 FIRST-GENERATION ANTIDEPRESSANTS: CONTROLLED, DOUBLE-BLIND STUDIES

	Number of Studies in Which the Effect Was
Drug	More than Placebo	Equal to Placebo	More than Imipramine	Equal to Imipramine	Less than Imipramine
Amitriptyline	9	2	2	5	0
Desipramine	3	2	0	6	1
Doxepin	2	0	0	3	0
Imipramine	30	14	—	—	—
Maprotiline	2	2	0	10	0
Nortriptyline	3	0	—	—	—
Protriptyline	2	0	0	2	0
Trimipramine	1	0	2	0	0
Adapted from Klein DF, Davis JM. Diagnosis and Drug Treatment of Psychiatric Disorders.Baltimore, MD: Williams & Wilkins, 1969:193-194.

Clinically, they are often preferred to the tertiary amine compounds because of less bothersome adverse effect profiles. The introduction of newer compounds (e.g., SSRIs) with a wider therapeutic index, however, makes this distinction less clinically relevant. Protriptyline is superior to placebo and equivalent to standard TCAs in outpatient populations (Table 7-7). It is more potent on a per milligram basis (i.e., average daily dose is 20 to 60 mg) due to its low first-pass effect and long half-life.

Other Agents. Amoxapine and maprotiline are structural analogs of TCAs. As shown in Tables 7-5, 7-6, and 7-7, clinical trials support the antidepressant efficacy of all of these drugs.

Amoxapine is a dibenzoxazepine derivative that has both NE and 5-HT reuptake-inhibiting properties. It is converted into 8-hydroxyamoxapine, which has considerable DA receptor binding properties (i.e., radioreceptor bioassays have found activity levels similar to those of firstgeneration antipsychotics (FGAs), a chemical structure similar to loxapine, and effects similar to antipsychotics, including acute and chronic extrapyramidal side effects (EPSs) and elevated prolactin levels) (330,331). For these reasons, amoxapine may have unique beneficial effects in psychotically depressed patients, but this has not been clearly established.

Like TCAs, amoxapine and maprotiline can be lethal in overdoses. Maprotiline also has a dosedependent risk of seizures.

MONOAMINE OXIDASE INHIBITORS

Interest in the MAOIs is spurred by controlled studies that document their efficacy and their benefit in the treatment of

Atypical depression (331)
Mixed anxiety and depressive disorders
Panic disorder, with or without agoraphobia
Eating disorders, particularly bulimia (because of required dietary restrictions, however, this may not be a feasible therapy in many)

Iproniazid was the first widely prescribed MAOI. Realization that it produces rare but dangerous liver toxicity led to the synthesis of the other hydrazine MAOIs, such as isocarboxazid, nialamide, and phenelzine as well as the nonhydrazine MAOIs, tranylcypromine, and pargyline.

There are two types of MAOs (i.e., A and B), which represent different proteins. MAO-A preferentially deaminates 5-HT and NE, whereas MAO-B preferentially deaminates DA, benzylamine, and phenylethylamine. Certain substrates (e.g., tyramine and tryptamine) are comparably deaminated by both types.

Important intraspecies differences are found in the relative proportions of MAO-A or MAO-B in tissues (e.g., the human brain has more MAO-B [about 70%] activity; rat brain has more MAO-A). After administration of an MAOI, intracellular levels of endogenous amines (e.g., NE) increase, but levels of amines not usually found in humans (tryptamine and phenylethylamine) also increase, followed by a compensatory decrease in amine synthesis because of feedback mechanisms. Levels of other amines or their metabolites (i.e., false transmitters) increase in storage vesicles and may displace true transmitters, whereas presynaptic neuronal firing rates decrease. After 3 to 6 weeks, brain 5-HT may return to normal levels, and NE levels may decrease. There is a compensatory decrease in the number of α₂ and β-receptors, including β-adrenergic receptor-related functions (e.g., NE-stimulated adenyl cyclase).

Most available MAOIs are irreversible inhibitors, forming a chemical bond with part of the enzyme or the flavin adenine dinucleotide cofactor. When treatment is stopped, inhibition continues for a time until MAO levels return to normal as the new enzyme is synthesized. Thus, phenelzine, isocarboxazid, and tranylcypromine are all irreversible, nonselective MAOIs. Clorgyline, however, is an irreversible, selective MAO-A inhibitor; moclobemide is a reversible, selective MAO-A inhibitor; and selegiline and pargyline are relatively selective, irreversible MAO-B inhibitors.

Although the half-life of an MAOI is short (hours), the half-life of MAO inhibition is about 2 weeks because it takes that long for a new enzyme to be synthesized. Some speculate that phenelzine is metabolized by acetylation and that there are two hereditary types (i.e., slow and fast acetylators), with slow acetylators presumably having a greater degree of MAO inhibition. There is limited support for the theory that slow acetylators have a better response, whereas other investigators find no difference (332). More importantly, there is no evidence that phenelzine is indeed acetylated.

TABLE 7-8 MAOIS VS. PLACEBO: ACUTE TREATMENT

			Responders (%)
Drug	Number of Studies	Number of Subjects	MAOI (%)	Placebo (%)	Difference (%)
Moclobemide	6	535	65	24	41
Phenelzine	8	429	56	43	14
Selegiline (TS)	2	466	33	21	12
MAOI, monoamine oxidase inhibitor; TS, transdermal system

Efficacy for Acute Treatment

At one time, TCAs were thought to be more effective than the MAOIs, but Greenblatt et al. (333) reported that these two classes were equally effective. The poorer showing in some of the earlier studies was the result of subtherapeutic doses of MAOIs administered to treatment-resistant populations (e.g., psychotic depressions).

The efficacy of phenelzine, tranylcypromine, and selegiline-TS were established in several large, double-blind studies, which found them equal to tertiary amine TCAs and/or clearly superior to placebo (334,335) (Table 7-8). Other studies indicate that atypical depression may respond better to MAOIs and typical depressions to TCAs; however, most find that their similarities are more obvious than their differences (336). One research group suggested that anergic, bipolar patients respond particularly well to tranylcypromine and other MAOIs (337). Extensive clinical experience indicates the MAOIs may be effective when TCAs or SSRIs have failed (338).

New Approaches

Given the unique efficacy of MAOIs, there are attempts to develop safer versions. A transdermal form of selegiline is one promising approach. Another approach is the development of selective and reversible MAOIs (339).

Selegiline Transdermal System. The concept behind the transdermal patch is to preferentially deliver more selegiline to the brain by bypassing the gut and the liver initially. The goal is to have antidepressant efficacy without the risk of tyramine-induced hypertensive crisis. Selegiline is an irreversible but relatively selective inhibitor of MAO-B. An oral preparation, currently marketed as a treatment for Parkinson’s disease, produces meaningful inhibition of MAO-B but not MAOA. Studies suggest that the selegiline transdermal system (STS) may minimize drug-food interactions. Normal volunteers given STS (20 mg/20 cm²) patch daily for 13 days were able to consume more than 300 mg of tyramine without experiencing any clinically relevant increases in systolic or diastolic blood pressure (340). Two, double-blind trials found that the same STS regimen was statistically superior to placebo in the treatment of outpatients with major depression, but effect sizes were modest (334,341). The only adverse effect greater with the STS than placebo was application-site reactions. Thus, physicians using this MAOI formulation for depression may appreciably lower risk of hypertensive crisis by

TABLE 7-9 CLINICAL CHARACTERISTICS OF MAOIS

MAOI	Selectivity	Substrate	Reversibility
Brofaromine (NA)	MAO-A	Serotonin and norepinephrine	Yes
Cimoxatone (NA)			Yes
Clorgyline (NA)			No
Moclobemide (NA)			Yes
Toloxatone (NA)			Yes
Pargyline (NA)	MAO-B	Phenylethylamine and benzylamine	No
Selegiline			No
Isocarboxazid (NA)	MAO-A, -B	Tyramine, dopamine, and tryptamine	No
Phenelzine			No
Tranylcypromine			No
MAOI, monoamine oxidase inhibitor; NA, not available in the United States

Minimizing inhibition of gastrointestinal MAO-A activity
Circumventing first-pass hepatic metabolism, which increases the parent compound to metabolite ratio

Selective and Reversible Agents. Another approach involves selective and reversible MAOIs, with a minimal risk of tyramine reactions and thus a diminished need for the dietary restrictions that plague the use of nonselective and irreversible A and B inhibitors. Collaborative clinical trials of the reversible inhibitors of MAOA (RIMAs) in Europe included more than 2,000 patients, many hospitalized for more severe, endogenous depressive episodes (342).

The best studied of this group is moclobemide, which demonstrated comparable efficacy to tertiary amine TCAs and superiority to placebo (Tables 7-8 and 7-9). Despite these promising results, the development of moclobemide in the United States was discontinued, presumably because of failure to adequately separate from placebo in double-blind studies. Further, its clinical use in other countries is limited due to a perceived less robust efficacy.

Other RIMAs under investigation include

Brofaromine
Cimoxatone
Toloxatone

If selegiline TS and possibly type A inhibitors eventually prove sufficiently efficacious, clinicians should prescribe them with increasing frequency. In addition, if there is minimal risk of adverse interactions with tyramine and other substances (e.g., sympathomimetics), medical and legal concerns about their use will also be reduced.

Table 7-9 summarizes the characteristics of several newer MAOIs that are either clinically available or under study.

Summary

Potential advantages of MAOIs:

Impact three neurotransmitters (i.e., 5-HT, NE, and DA)
With newer agents, antidepressant effect attributed to enzyme MAO-A or MAO-B only; therefore, propensity to cause tyramine potentiation is reduced
Possibly more effective in endogenous and atypical depression
No nonspecific biochemical or pharmacological actions

Potential disadvantages of MAOIs:

Drug-drug and drug-food interactions
Effective dose is variable
Adverse effects—loss of appetite, nausea, and other gastrointestinal disturbances; insomnia
Shorter duration of action
Idiosyncratic responses at times

OTHER DRUG THERAPIES

Stimulants

Such varied agents as amphetamine, dextroamphetamine, methylphenidate, modafinil, atomoxetine, and pemoline can be considered psychomotor stimulants, some of which produce an acute euphoria in control subjects as well as a wide variety of responses in psychiatric patients. These stimulants are also effective in postponing the deterioration in psychomotor performance that often accompanies extreme fatigue, a property that may be useful in some carefully selected cases. The pharmacology of these drugs is discussed in greater detail in Chapter 14.

Although amphetamines or other psychomotor stimulants induce an initial euphoria, there is considerable doubt that they produce a long-lasting antidepressant effect. For example, cocaine produces a euphoria almost immediately after intravenous injection and within a few minutes after intranasal administration, but the euphoria, as well as the tachycardia, decrease at a slightly faster rate than the level of plasma cocaine. A second dose given 1 hour later fails to produce a similar level of euphoria or tachycardia, suggesting a rapid tachyphylaxis.

Amphetamines. One early British study found that amphetamine did not differ from placebo in the treatment of depressed outpatients (343). A second study found amphetamine less effective than phenelzine and no better than placebo (344). A third Veteran’s Administration (VA) study found dextroamphetamine no more effective than placebo in hospitalized depressed patients (345). Uncontrolled clinical evidence indicates that amphetamine may occasionally be of value, but except for a mild, early, transient benefit, there is no evidence that amphetamine can ameliorate moderate-to-severe depressive episodes.

Atomoxetine. Atomoxetine is approved in the United States for the treatment of ADHD. Its NE reuptake transport blockade has also generated interest as a potential treatment for depression. In this context, studies used this agent as an adjunct, usually in patients with comorbid ADHD. A placebo-controlled trial augmented sertraline with atomoxetine in 276 adult patients with major depression who were incompletely responsive (346). After 8 weeks, atomoxetine (40-120 mg/day) did not separate from placebo on the primary or secondary outcome measures.

Methylphenidate. Some patients refractory to all other therapies have experienced a dramatic, full, rapid, and sustained response to methylphenidate. One study found methylphenidate effective in treating mildly depressed outpatients, particularly those who drank three or more cups of coffee a day (347). A replication study did not find a drug-placebo difference on the physician’s ratings, but it did demonstrate one for the patients’ subjective assessment of improvement (348). Another blind study of methylphenidate found improvement in an outpatient group (349). Finally, two of three trials in apathetic, senile geriatric patients showed this drug produced more improvement than placebo (350). To date, however, no evidence indicates that it is beneficial in cases of moderateto-severe or treatment-resistant depression (351).

Modafinil. This agent has some stimulating properties, possibly through increased dopaminergic activity. Although approved for narcolepsy and idiopathic somnolence, open trials also report benefit with modafinil in depressed nonresponders, particularly for fatigue. Doubleblind, placebo-controlled trials, however, failed to demonstrate a significant effect with this agent on the primary outcome measures (352,353,354 and 355).

Dopamine Agonists. Agents that stimulate DA receptors (e.g., pramipexole) have demonstrated potential benefit in TRD. They may, however, produce rare, serious adverse effects, including sleep attacks, compulsive behaviors, and pathological gambling, and psychoses (356).

Summary. Five studies of amphetamine for depression were clearly negative, with none finding it more effective than placebo. Although there were occasional hints of efficacy, in one study, amphetamine was less effective than placebo.

The results with methylphenidate, however, are more encouraging. Two of three studies found a significant effect, and the third found improvement on the patients’ subjective evaluation. Although amphetamine and methylphenidate are similar in their pharmacology, they differ in some respects. Amphetamine releases DA from newly synthesized pools (α-methyl-p-tyrosine-sensitive pool), whereas methylphenidate releases DA from storage sites (reserpine-sensitive sites). This pharmacological difference could explain the apparent greater efficacy of methylphenidate.

Data for atomoxetine and modafinil are less promising, but they may be potential augmentation strategies for treating symptoms such as lack of alertness, fatigue, and sleepiness.

Psychostimulants can cause jitteriness, palpitations, and psychic dependence. Depression may arise after their discontinuance, and high doses can produce a florid psychosis. On occasion, even small doses of amphetamine can precipitate psychotic episodes in those with an underlying predisposition (e.g., schizophrenic disorder).

In conclusion, there are clinical reports of depressed patients who fail to respond to standard antidepressants but do well on low doses of stimulants. Although limited, the best trial data support efficacy for methylphenidate.

Lithium

Lithium is superior to placebo, particularly as an acute treatment for the depressed phase of a bipolar disorder (Table 7-10). Mendels et al. (357) reviewed and Souza and Goodwin (358) statistically combined the data in a meta-analysis from many of the same trials. Their results indicated that the best evidence for the acute antidepressant effect of lithium was in the bipolar depressed patient group.

The use of lithium as an augmentation to standard antidepressants may also be an effective strategy in partially responsive unipolar depressive episodes (see “Alternative Treatment Strategies” later in this chapter).

TABLE 7-10 LITHIUM VS. PLACEBO: ACUTE TREATMENT

			Responders (%)
Disorder	Number of Studies	Number of Subjects	Lithium (%)	Placebo (%)	Difference (%)	Chi Square	p Value
Unipolar depression	4 (uncontrolled)	79	39	27	12	0.5	NS
Unipolar depression	1 (controlled)	27	57	38	19	0.9	NS
Bipolar depression	2 (controlled)	38	76	35	41	5.5	0.02