|More Like This|
Approximately 32 to 35 million adults in the United States have an episode of major depression sometime during their lifetime, and many of them do not respond to initial treatment.1 The results of an analysis undertaken by Fava and Davidson2 suggest that between 29% and 46% of depressed patients fail to respond fully to antidepressant treatment of adequate dose and duration; about 15% of patients fail to respond to multiple treatment trials.3 Fagiolini and Kupfer4 have suggested that patients with treatment-resistant depression (TRD) may represent a biologically unique subtype of depressed patients. Unfortunately, the chances for full recovery diminish the longer a patient remains depressed—a fact that lends a sense of urgency for appropriate therapy.5
Before discussing the effectiveness of individual agents in TRD, it is important to review recent evidence regarding the principles of using these agents to maximize the chance of a response.6-9 Some of the most salient findings from the STAR*D trial on this topic are summarized in Table 1.
The standard strategy for managing patients who are depressed includes an adequate clinical trial; when the response is unsatisfactory, the clinician’s options are:
• Switch to another antidepressant
• Combine a second antidepressant with the first
• Add a second drug that is not approved as an antidepressant
• Start psychotherapy
Antidepressants can be grouped into 6 major categories: tricyclic antidepressants (TCAs), SSRIs, serotonin norepinephrine reuptake inhibitors (SNRIs), monoamine oxidase inhibitors (MAOIs), serotonin neurotransmitter (5-HT2)-receptor antagonists (eg, nefazodone, trazodone), and novel agents (eg, mirtazapine, bupropion). If a patient fails to respond to one antidepressant class, it makes sense (at least conceptually) to switch to another, although most guidelines acknowledge that 2 failed trials of SSRIs may be justifiable before switching classes.7,8 When examining switching studies, be certain that the reason for switching was a failure to respond to an adequate antidepressant trial rather than a lack of tolerability, as these are fundamentally different issues.
The concept of switching classes of antidepressants is only as valid as the theoretical basis by which the antidepressants are grouped. As noted, some treatment guidelines allow 2 SSRI trials before switching classes. For example, in a study by Thase and colleagues,10 58 patients failed
a trial of fluoxetine; however, there was a 76% response rate to citalopram among completers. Although this study was not blinded, it is interesting that citalopram (which is believed to be the most selective of the SSRIs) was effective when another SSRI was not. It is hoped that as our knowledge of the mechanism of action of antidepressants grows more sophisticated, methods of classifying these agents will evolve to better guide clinicians in switching and combination strategies.
The clinical trial data on switching are surprisingly sparse—a recent meta-analysis identified only 4 randomized controlled trials.11 These trials include data from level 2 of the STAR*D study in which patients who were switched from citalopram to sertraline, venlafaxine XR, or bupropion SR showed remission rates of 27%, 25%, and 26% (respectively) based on the Quick Inventory of Depressive Symptomatology, Self-Report (QIDS-SR-16).12 The meta-analysis found a modest but statistically significant advantage in switching from SSRIs to non-SSRIs. Of the 5 comparisons, 3 used venlafaxine, which is predominately a serotonergic agent (the serotonin-norepinephrine Ki ratio is 30), although it is classified as an SNRI. This may account for the relative lack of separation. One wonders if duloxetine, with a Ki ratio of 9 and relatively greater noradrenergic potency, might show greater separation.13 However, only one open-label switching study with duloxetine has been published to date.14
In switching antidepressants, 2 pharmacokinetic issues are important:
• Drugs with potent cytochrome P-450 (CYP450) effects—particularly fluoxetine (because of its long half-life)—may produce toxic levels of medications taken concurrently, such as TCAs that are also metabolized by these systems. (This effect is even more of an issue with combinations.)
• Antidepressants that are associated with discontinuation symptoms (eg, paroxetine, venlafaxine, duloxetine) may produce discontinuation syndromes when stopped that can be erroneously attributed to adverse effects of the new drug, particularly if the new drug does not possess a significant degree of serotonin reuptake inhibition.
What we need are agents with novel mechanisms of action. As indicated in Table 2, there is a rich pipeline of candidates. Some of these agents, such as agomelatine, are relatively far along in the clinical trial process with human subjects. Others, such as amibegron and sipatrigine, have been shown to be efficacious only in animal models of depression and their benefit is therefore highly speculative.15-17 Some of the drugs have been tried as augmentation agents as well (eg, riluzole). Agents such as pramipexole, memantine, riluzole, and ketoconazole are already available in the United States.
Proposed mechanisms of action vary widely, but 2 development strategies are generating a great deal of activity:
• Agents are being developed that affect corticosteroid function at various levels. (Glucocorticoids have figured prominently in recent theories as major agents in stress-induced neuronal injury and cell death leading to depression.18)
• Compounds are being developed that induce the synthesis of brain-derived neurotrophic factor (BDNF), particularly through their effects on glutamate receptors.
In a sense, these 2 strategies are related: it has been proposed that BDNF acts as a modulator of neuronal repair and even neurogenesis in response to cortisol-induced brain injury. The findings with ketamine are particularly exciting. A recent study reported that ketamine, through its activity as an N-methyl-D-aspartate (NMDA)-receptor antagonist, increases synaptic glutamate and therefore stimulates a-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) receptors (both NMDA and AMPA receptors are subtypes of glutamate receptors).19 AMPA potentiators are known to induce BDNF production more rapidly than current antidepressants.20 It has been hypothesized that a more rapid induction of BDNF would lead to a faster onset of antidepressant action. In a study by Zarate and colleagues,21 the onset of the antidepressant effect of ketamine occurred within 2 hours. Unfortunately, the drug must be given by intravenous infusion, but because the effect persisted for 7 days, ketamine treatment may be useful if given as a series.
Another NMDA-receptor antagonist now available in the United States is memantine (FDA-approved for Alzheimer disease). Memantine was shown to be effective in an open-label study in major depression. In a double-blind, randomized trial it showed comparable effects to escitalopram in patients with major depression and alcohol dependence.22,23 However memantine failed to separate from placebo in a double-blind, placebo-controlled study in major depression.24 The less robust antidepressant properties of memantine when compared with those of ketamine may be due to differing NMDA receptor binding properties between the 2 compounds.21
In this category there are 2 similar strategies: (1) combination, that is, combining 2 FDA-approved antidepressants with presumably complementary mechanisms of action; and (2) augmentation, in which a drug not approved as an antidepressant is used with an FDA-approved antidepressant. Such pairings can be used either at the beginning of therapy to speed the response (an accelerator strategy) or to improve the overall response, thereby enhancing the odds of achieving remission. As noted, when combining agents, one must be aware of possible CYP450-based pharmocokinetic interactions or other pharmacological interactions, such as the risk of serotonin syndrome when an SSRI is combined with an MAOI.
Combination strategies. Of the 2 strategies, the addition of a second antidepressant is the more intuitively obvious. Randomized controlled trials have supported the superior efficacy of a TCA/SSRI combination, as well as a mirtazapine/SSRI combination; open studies have done the same for TCA/MAOI and SSRI/bupropion combinations.25 A positive retrospective chart review of 10 patients treated with a combination of duloxetine and bupropion has also been published.26
These findings raise the question of whether combining antidepressants is superior to switching. For example, an open-label, nonrandomized study by Lam and colleagues27 found that when 61 patients who were taking citalopram or bupropion SR for TRD were either switched to the other drug or given the combination, the combination group showed statistically greater rates of response and remission. At a minimum, combination treatments ensure that a partial response to the first agent will not be lost.
Augmentation strategies. If the switching and combination literature is conspicuous by its paucity, the augmentation literature is abundant—but the quality of evidence varies tremendously and the reader must maintain a “buyer beware” approach. One recent review concluded that although studies of augmentation were abundant, many were underpowered and uncontrolled.28 Several comprehensive reviews have been published.28-31
Evaluation strategies have been suggested to assess the quality of evidence.28,32 In general, augmentation strategies are recommended in the event of a partial response to an antidepressant. Switching is preferred in the event of a nonresponse.25 One of the few studies to examine this issue systematically found numerically higher response rates among partial responders compared with nonresponders when lithium or desipramine was added to fluoxetine, but these differences were not statistically significant.33Table 3 presents a summary of agents that may be effective. Although the amount and quality of evidence about the agents in category A is meaningful, the differentiation of groups B and C is somewhat arbitrary.
The presumed mechanism of action of augmentation agents varies tremendously and is generally believed to complement the action of the primary antidepressant in some predictable theoretical manner. Questions remain about the validity of our current classification systems for grouping augmentation agents—an issue of crucial importance for clinicians who need to pair them with antidepressants for maximum benefit.
Lithium, the classic augmentation agent, has been the subject of the greatest number of studies. A recent meta-analysis identified 10 randomized, placebo-controlled trials, all of which were relatively small (the largest included 61 subjects).34 The authors of these studies concluded that lithium was significantly more effective than placebo. The vast majority of the studies with lithium did not include SSRIs but instead used TCAs. Remission rates from level 3 of the STAR*D trial using lithium as an augmentation agent with citalopram were only 13.2% (measured by QIDS-SR), whereas T3 augmentation remission rates at this same step were 24.7%.35 Although these differences were not statistically different, there was a significantly higher dropout rate in the lithium group.
As with lithium, a recent review concluded that the trial data that support the efficacy of T3 augmentation are of better quality with TCAs than with SSRIs.32 Although it has been reported that approximately 50% of patients with unipolar depression given thyroid augmentation show a response, a pooled analysis of 4 randomized double-blind studies of T3 augmentation revealed nonsignificant effects, but the results of 1 of the 4 studies may have accounted for this finding.36,37
By contrast to these relatively small studies with classical augmentation agents, there are a number of large, randomized, placebo-controlled trials of atypical antipsychotics in nonpsychotic, unipolar depression in combination with antidepressants. These studies culminated in the first-ever FDA approval of an agent specifically for antidepressant augmentation—aripiprazole. In their meta-analysis, Papakostas and colleagues38 conclude that the results support the utility of atypical augmentation, even in the absence of double-blind, placebo-controlled trials with aripiprazole (now available) and ziprasidone. Virtually all of the studies to date of atypical augmentation have been conducted with SSRIs, although there are reports that aripiprazole was effective with bupropion, tranylcypromine, and mirtazapine.39-41
The study of tranylcypromine (an MAOI) with aripiprazole suggests the need for cautious use of some combinations. As noted in the package insert, tranylcypromine should not be coadministered with dibenzazepine-related entities because of the risk of severe interactions, including hypertensive crises and seizures.42 Some of the atypical antipsychotics (eg, aripiprazole, ziprasidone) have a high affinity at the 5-HT1A-receptor; agents such as buspirone, which also binds at this receptor site, are not recommended for use with MAOIs. Ziprasidone also has a similar potency to imipramine in terms of blocking norepinephrine and serotonin reuptake (imipramine is al-so relatively contraindicated with MAOIs).43 There is a report of 5 patients treated safely with risperidone and MAOIs, and another of 12 patients exposed to olanzapine with the selegiline patch.44,45
In general, the effective dosage of atypical antipsychotics in this role seems to be lower than that used in psychosis. Data on the long-term risks in depressed patients, including tardive dyskinesia and metabolic syndrome, are urgently needed. The recent FDA approval of quetiapine as monotherapy in bipolar depression raises questions as to whether some of the atypical antipsychotics may be effective (and perhaps better tolerated) when given as a single agent for unipolar TRD.
In regard to the value of switching atypicals, if one agent fails, is it worth attempting trials of another? There are limited data that show that switching to another atypical is worthwhile.46,47 However, results from randomized controlled trials are still needed.
Each of the agents listed in Table 3 may be most effective when used for augmentation purposes. The quality of the evidence varies widely, particularly in the second and third categories of the table. For example, in a pooled analysis, estrogen was effective when added to sertraline in a group of depressed women over 60 years; however, the total sample size was only 127.48 There are small, positive, double-blind studies for agents such as testosterone, L-tryptophan, and omega-3-fatty acids.49-51 In the case of pindolol, there are negative studies as well, and in a recent review, the evidence supporting its efficacy was rated as a “C” (the lowest rating).28,32
Open-label reports for buspirone were positive, but 2 randomized controlled trials were negative.52,53 Buspirone was used as an augmentation agent for citalopram in level 2 of STAR*D: remission rates were 33% by the QIDS-SR compared to 39% with bupropion at the same stage. Although the remission rates between the 2 drugs did not differ significantly, the bupropion group had a significantly greater reduction in depression scores.54
Despite reports that suggest the efficacy of lamotrigine, this agent did not separate significantly from placebo in the only large, multicenter, double-blind, placebo-controlled trial to date.55 Modafinil, used in 2 randomized, placebo-controlled trials did not show significant placebo separation on most mood-scale scores, but it did reduce sleepiness and fatigue.56,57
The unfortunate reality is that conducting large, double-blind, placebo-controlled trials is extremely expensive and time consuming. Definitive evidence of the efficacy of these agents is likely to accrue slowly. Negative studies may not prove that a drug is ineffective because of factors such as high placebo response rates, or the more subtle reality that there may be meaningful subgroups of patients with TRD that we are currently unable to identify.
Drugs Mentioned in This ArticleAgomelatine (Valdoxan, Melitor)
References1. Kessler RC, Berglund P, Demler O, et al. The epidemiology of major depressive disorder. results from the National Comorbidity Survey replication (NCS-R). JAMA. 2003:289:3095-3105.
Evidence-Based ReferencesBarowsky J, Schwartz TL. Part 1: An evidence-based approach to augmentation and combination strategies for treatment-resistant depression. Psychiatry 2008. 2006;3:42-61.