OUP user menu

Rational site-directed pharmacotherapy for major depressive disorder

Pierre Blier
DOI: http://dx.doi.org/10.1017/S1461145713000400 997-1008 First published online: 1 July 2014


It is now accepted that major depressive disorder (MDD) is not a single pathophysiological entity. It is therefore not surprising that remission rates to a first antidepressant trial are low. In addition, antidepressants may target various neuronal elements for which there are gene polymorphisms, such as the serotonin (5-HT) reuptake transporter, which may modulate response. Acting on a single monoaminergic target, such as inhibiting the 5-HT transporter, may confer efficacy in MDD, but other targets may be used and/or combined in treatment-resistant patients. These include the blockade of norepinephrine transporters, monoamine oxidase, 5-HT2A, 5-HT1B and 5-HT7 receptors, and the activation of 5-HT1A and dopamine 2 receptors. While antidepressants may have more than one of these properties, so do atypical antipsychotics. When using the latter medications, however, their regimens should be below those effective in treating psychosis to avoid dopamine 2 antagonism, which could be counter-productive in MDD. In some patients, combining medications from treatment initiation may also provide additional therapeutic benefits.

Key words
  • Antidepressants
  • atypical antipsychotic
  • drug combination
  • monoamines
  • polypharmacy


Unipolar depression carries the largest burden of all diseases in middle to high income countries, as determined by the World Health Organization (2008). This burden is estimated using disability adjusted life years and unipolar depression scores immediately ahead of ischaemic heart disease. The same report predicts that, by the year 2030, unipolar depression will rank first worldwide, again ahead of ischaemic heart disease. Despite an ‘epidemic’ of obesity, the burden of unipolar depression will grow more than that of ischaemic heart disease: 50 vs. 30%, respectively. These statistics indicate the severity of major depressive disorder (MDD) and a growing trend that needs to be tackled effectively.

Currently, any first-line treatment for MDD yields a remission rate of about 35% (Trivedi et al., 2006a; Lam et al., 2009). This most often consists of a single antidepressant medication given at an adequate dose for a sufficient time, anywhere from 6–12 wk. This is a very poor outcome for such a serious illness. Most guidelines recommend a medication switch after a first failed trial. This second step commonly requires another 6–12 wk and leads to an even smaller success rate than the first trial.

Possible reasons for low remission rates with a first antidepressant trials

One needs to consider the possible reasons for such low success rates. A first possible explanation is the heterogeneity of the clinical presentations. Indeed, when using the criteria of the Diagnostic and Statistical Manual-IV of the American Psychiatric Association (2000), it is possible to have two patients meeting the criteria for MDD and not have any sign or symptom in common (Fig. 1). With such diametrically opposed symptoms for sleep, appetite and psychomotor activity and other symptoms, it is evident that the pathophysiology of patients with MDD must be different, implicate different brain structures and/or anomalies of neurotransmitter functions. For instance, elevated levels of cortisol/corticotropin releasing factor and substance P have been reported in depressed patients (Holsboer, 2000; McLean, 2005). Altered functions of serotonin (5-HT), norepinephrine (NE) or dopamine (DA) have also been reported in some but not all patients (Delgado, 2000; Dunlop and Nemeroff, 2007). It is therefore difficult to imagine that an antidepressant with a single primary target, such as the 5-HT transporter (5-HTT), could produce a therapeutic remission in the majority of patients.

Fig. 1

Signs and symptoms of two patients each having depressive mood or anhedonia, plus five additional symptoms to meet the criteria for major depressive disorder according to the DSM-IV. Note that sleep, appetite and psychomotor activity are in opposite directions.

It is now well established that there are functionally important polymorphisms for a variety of receptors and transporters directly and indirectly targeted by antidepressant medications. This may contribute further to low remission rates with antidepressants used in monotherapy. In the case of selective serotonin reuptake inhibitors (SSRIs), for instance, several studies have shown that the response rate can be lower in patients having one or two short alleles vs. two long alleles in the promoter region of the 5-HTT gene (Porcelli et al., 2012).

Another issue to consider for the low remission rates with antidepressants is the fact that, upon attempting to enhance the function of a single monoamine system, there may be an ensuing decrease in the function of other neurotransmitter systems (see Fig. 2). In the case of SSRIs, their sustained administration has consistently been shown to decrease the firing activity of NE neurons in the rat locus coeruleus (Szabo et al., 2000; West et al., 2009). Furthermore, such a prolonged regimen of the SSRI citalopram has been reported to produce a corresponding decrease of extracellular levels of NE in the rat amygdala (Kawahara et al., 2007). Since an antidepressant response can be obtained with medications that selectively block NE reuptake, such as the non-tricyclic agent reboxetine, it therefore appears highly likely that, in some patients who are not responding to a SSRI, the resulting attenuation in NE neuronal activity may contribute to the lack of response (Szabo and Blier, 2001a). Such a decreased activity of NE neurons may also contribute to decreased energy and interest in some patients after long-term treatment with SSRIs, given that the latter two symptoms are controlled in part by the NE system (Stahl, 2002).

Fig. 2

Reciprocal norepinephrine (NE)×serotonin (5-HT)×dopamine (DA) neuron interactions showing their functional connectivity. Note that some of these pathways are monosynaptic while others utilize γ-aminobutyric acid interneurons. The circles with the arrows represent the reuptake transporters. The dots represent the neurotransmitters. The + and − symbols within the circles describe the excitatory or the excitatory effects, respectively, on the firing activity of the neurons depicted by the zig-zags on the axons.

Obviously, patients with MDD may respond fully to SSRIs and do not present signs or symptoms of lowered NE activity. It is conceivable that the manifestations of such symptoms may be dependent on the functional status of the NE system at baseline before the treatment was initiated and/or on the polymorphisms of receptors in the pathway from 5-HT to NE neurons. More specifically, 5-HT neurons project to γ-aminobutyric acid GABA) neurons where they activate an excitatory 5-HT2A receptor, thereby increasing GABA release that inhibits the firing activity of NE neurons (Szabo and Blier, 2001b; Fig. 2). For instance, a polymorphism for the 5-HT2A receptor was linked to the antidepressant response to citalopram in the initial phase of the STAR*D project (McMahon et al., 2006). That a dampened NE activity can lead to lower energy is supported by the same problem that often arose during the treatment of hypertension with the α2-adrenergic agonist clonidine. This anti-hypertensive decreases NE availability by activating the inhibitory α2-adrenergic autoreceptors located on the cell body and terminals of NE neurons (Fig. 2). Clonidine may lower mood in certain individuals, possibly because it decreases the NE activation of 5-HT neurons and directly decreases 5-HT release from the activation of inhibitory α2-adrenergic receptors on 5-HT terminals (Lacroix et al., 1991; Mongeau et al., 1993).

A similar inhibitory 5-HT pathway exists between 5-HT and the DA neurons of the ventral tegmental area (Guiard et al., 2008). These DA neurons are known to play an important role in drive and reward. Serotonin neurons project to GABA neurons where they activate an excitatory 5-HT2C receptor, thereby increasing GABA release, which inhibits the firing activity of DA neurons (Fig. 2). A decreased firing rate of DA neurons has not been reported with all SSRIs, but the most potent SSRI, escitalopram, decreases both the firing rate and the burst activity of DA neurons (Dremencov et al., 2009). Again, the differential susceptibility of some patients to side-effects of SSRIs potentially mediated by a decreased dopaminergic activity may be linked to the multiple forms of the 5-HT2C receptor, which undergoes mRNA editing (Drago and Serretti, 2009).

Taken together, these anatomical and functional observations indicate that even selective drugs can have profound repercussions on the function of other transmitter systems to the detriment, or the benefit, of the clinical response. For instance, the attenuation of the firing rate and reactivity of NE neurons by SSRIs may contribute to dampen the intensity of a panic attack because in such a state there is a hyper-adrenergic drive that would then be under serotonergic inhibition. Indeed, the levels of the main metabolite of NE are decreased following fluoxetine treatment of panic disorder patients (Coplan et al., 1997). On the other hand, depressed patients not responding to a SSRI may be attributable in some cases to an attenuated noradrenergic and/or dopaminergic function.

Taking into consideration that MDD may be more than one illness, implying that its manifestation in the form that we currently define it is multi-factorial, and that some of the current treatments exert effects other than their primary action because of the functional connectivity of neurotransmitter systems, it would seem logical to use more than one single mechanism to achieve remission in a larger percentage of patients. This is indeed the norm in medicine. When treating HIV/AIDS, tritherapy has improved success rates. Asthma is always tackled with at least a β2-adrenoceptor agonist and a steroid inhaler and sometimes a leukotriene receptor antagonist as well. Septicaemia is treated with two, sometimes three antibiotics with different action spectra, until the anti-biogram is available. Unfortunately, we do not have ‘antidepressant-grams’ for MDD.

Mechanisms that may contribute to the antidepressant response

First-line treatment for MDD often consists of blocking the 5-HTT, all the more so as most SSRIs are now generic. Serotonin and NE reuptake inhibitors (SNRIs) are also widely used, as at least one has become generic. These medications, however, need to be used in the upper range of their approved range to significantly inhibit NE reuptake. Venlafaxine at its minimal effective dose of 75 mg/d acts as a SSRI, whereas at 225 mg/d it produces a physiologically significant interference with the NE reuptake process (Debonnel et al., 2007). Since its active metabolite desvenlafaxine taken at its minimal effective dose of 50 mg/d produces plasma levels identical to those produced by venlafaxine at 75 mg/d (Nichols et al., 2011), it is evident that desvenlafaxine like venlafaxine needs to be titrated up to significantly inhibit NE reuptake. A similar titration of duloxetine from its minimal needs to be implemented to obtain optimal NE reuptake inhibition (Turcotte et al., 2001; Vincent et al., 2004). It is nevertheless difficult to demonstrate that SNRIs by having an additional mechanism of action are more effective than SSRIs if they are compared from treatment initiation. Indeed, analysis of the studies comparing SSRIs and venlafaxine in such a manner only yielded a 7% superior remission rate (Nemeroff et al., 2008). In contrast, when treatment-resistant patients were randomized to the SSRI paroxetine or high-dose venlafaxine, the remission rate with venlafaxine was twice that obtained with paroxetine (i.e. 42 vs. 18%; Poirier and Boyer, 1999). In the STAR*D switch from citalopram to sertraline, the remission rate was also 18% but that to venlafaxine was only 25% (Rush et al., 2006). The mean of the daily doses of venlafaxine in that study was, however, only 190 mg/d, whereas in the former it was 272 mg/d, being clearly in the SNRI range for the majority of patients. Although there is no such double-study supporting the greater effectiveness of duloxetine at 120 mg/d, there was significant additional benefit when 60 mg/d initial non-responders received the higher regimen (Sagman et al., 2011). It is noteworthy that these patients had initially been treated with a SSRI or venlafaxine prior to study inclusion.

Another mechanism to achieve an antidepressant response is to antagonize α2-adrenergic receptors. Importantly, these receptors are located presynaptically on the cell body of NE neurons and inhibit firing on NE terminals and inhibit NE release and on 5-HT terminals, where they also inhibit 5-HT release (Fig. 2; Mongeau et al., 1993). Consequently, blocking these receptors enhances the synaptic availability of both NE and 5-HT. The antidepressant mirtazapine antagonizes these three types of presynaptic α2-adrenergic receptors while leaving unaffected the responsiveness of post-synaptic α2-adrenergic receptors, at least in the hippocampus (Haddjeri et al., 1997). Several atypical antipsychotics, which can be used in treatment-resistant MDD, also block to some extent α2-adrenoceptors; these include asenapine, clozapine, paliperidone, quetiapine, risperidone (Fig. 3; Schotte et al., 1996). Officially, however, only quetiapine and aripiprazole have an indication for treatment-resistant MDD, although there are controlled studies for risperidone (Nelson and Papakostas, 2009). The degree of α2-adrenergic antagonism at the four above-mentioned adrenoceptors can vary among these medications. For instance, asenapine partially blocks post-synaptic α2-adrenergic receptors in the hippocampus while quetiapine does not (Chernoloz et al., 2012a, b; Osterhoff et al., 2012). The heterogeneity of response of these receptors, as well as other monoaminergic properties, may explain whether some atypical antipsychotics may act as antidepressants when used alone. For instance, quetiapine has been shown to be effective in monotherapy for MDD (Maneeton et al., 2012); however, an important metabolite, desalkyl-quetiapine, blocks NE reuptake to a significant extent (Jensen et al., 2008; Chernoloz et al., 2012a, b). There is a synergistic action on synaptic NE availability when the NE transporter and the α2-adrenergic autoreceptors are blocked simultaneously (L'Heureux et al., 1986). This potentiation is presumably at play with the combination of high doses of venlafaxine and therapeutic doses of mirtazapine (Blier et al., 2010).

Fig. 3

Pharmacological properties of atypical antipsychotics that may contribute to their potential antidepressant activity when used in combination with antidepressant medications. Not all of these medications have been reported to be effective in treatment-resistant depression.

When an atypical antipsychotic is used alone or in combination with an antidepressant medication for MDD, the regimen is almost invariably below that used to treat psychosis (Nelson and Papakostas, 2009). This signifies that such doses do not block to a marked extent DA D2 receptors and they are used for their other properties on monoaminergic receptors. Using higher doses may not be effective or lead to a loss of response it diminishes D2 transmission (Blier and Blondeau, 2011). The term atypical antipsychotic is thus a misnomer. When used in mood and anxiety disorders, these drugs maybe better denoted as anxiothymic modulators.

Agonism of 5-HT1A receptors has been shown to be an effective strategy in MDD (Blier and Ward, 2003). Although most 5-HT1A agonists are partial agonists, this annotation does not hold true for all brain regions as most are full agonists on the cell body 5-HT1A autoreceptor. Even on post-synaptic 5-HT1A receptors, the partial agonism of ipsapirone, for example, is not present when the drug is administered systemically (i.e. it does not attenuate the inhibitory effect of 5-HT; Dong et al., 1997). Consequently, 5-HT1A agonists act by enhancing and not decreasing 5-HT transmission (Haddjeri et al., 1998). The 5-HT1A receptor agonist buspirone was shown to have antidepressant activity at high doses (Robinson et al., 1990). This medication is, however, rapidly absorbed and eliminated, giving rise to marked side-effects, thereby often preventing its use at high doses that are antidepressive. Buspirone is better tolerated in patients already on SSRIs since they would have already developed a certain tolerance to serotonergic side-effects. In STAR*D, buspirone addition produced a similar remission rate as bupropion in citalopram-resistant patients (Trivedi et al., 2006b). The buspirone analogue gepirone in an extended release formulation was shown to be significantly superior to placebo in three studies, thus demonstrating the usefulness of this strategy in MDD (Feiger, 1996; Feiger et al., 2003; Bielski et al., 2008). Several atypical antipsychotics can also activate 5-HT1A receptors; these include aripiprazole, asenapine, clozapine, iloperidone, lurasidone and ziprasidone. The affinity of aripiprazole and ziprasidone is markedly greater than that of buspirone (Pou et al., 1997; Wood et al., 2006). It is thus likely that this property can contribute to their beneficial action in SSRI-resistant patients. Similarly, the potent 5-HT1A receptor agonism of the SSRI vilazodone may not only contribute to its antidepressant effect, but also to its side-effect profile (Heinrich et al., 2004).

Agonism of DA D2 receptors on its own is an effective strategy for MDD. This was first reported with bromocriptine and subsequently with pramipexole in a placebo and fluoxetine controlled study (Waehrens and Gerlach, 1981; Corrigan et al., 2000). Sustained administration of pramipexole enhances D2 transmission in a time-dependent fashion. Initially, the firing rate of DA neurons is decreased because of D2 autoreceptor activation on DA neurons (Chernoloz et al., 2009). These autoreceptors desensitize with treatment prolongation and allow a recovery of the firing rate of DA neurons, at which time there is an increased activation of post-synaptic D2 receptors in the frontal cortex (Chernoloz et al., 2012a, b). D2 transmission is also enhanced at excitatory D2 receptors on 5-HT neurons in the dorsal raphe after long-term pramipexole administration (Chernoloz et al., 2009). This elevates the firing rate of 5-HT neurons and leads to enhanced 5-HT transmission in the hippocampus (Chernoloz et al., 2012a, b). Among the atypical antipsychotics, only aripiprazole is endowed with D2 receptor agonism. It is a partial agonist, but since there is likely a decreased availability of DA in MDD (Dunlop and Nemeroff, 2007), the exogenous agonist may work cooperatively with endogenous DA to enhance the tonic activation of post-synaptic D2 receptors.

There are other mechanisms that may contribute to an antidepressant response and receptors when used in combination with antidepressants. These include monoamine oxidase (MAO) inhibition, monoamine releasers, the terminal 5-HT autoreceptor and 5-HT7 receptors.

MAO inhibitors (MAOIs) enhance the availability of 5-HT, NE and DA and are sometimes useful after several failed trials with a variety of agents. The effects of their sustained administration on the firing activity of these three types of neurons in the rat brain over time have been reported (Blier and de Montigny, 1985; Chenu et al., 2009). They will not be reviewed herein, but a variety of augmentation strategies can be used, as in the case of reuptake inhibitors.

Lithium addition was one of the first documented augmentation strategies. While lithium does have a variety of actions, it is believed to act relatively rapidly in treatment-resistant MDD by enhancing 5-HT release. While on its own, lithium is not sufficient to increase the tonic activation of post-synaptic 5-HT receptors in the hippocampus, it potentiates the effects of antidepressants, namely, the MAOI tranylcypromine, the tricyclic imipramine and the SSRI paroxetine (Haddjeri et al., 2000).

Bupropion is a catecholamine releaser. This is not surprising as it is a derivative of the appetite suppressant diethylpropion. Initially, bupropion was thought to be a DA and NE reuptake inhibitor, despite having very low affinity for these transporters (520 and 52 000 nm, respectively; Tatsumi et al., 1997). Indeed, four positron emission tomographies have shown that clinically effective doses of bupropion occupy about 20% of DA transporters, at times with a variance barely above the control situation (see Meyer et al., 2002; Argyelán et al., 2005). In acute microdialysis studies in rodents, bupropion enhances extracellular DA, but not in all brain regions as would be expected from a reuptake inhibitor (Li et al., 2002). In the striatum, bupropion is without effects (Egerton et al., 2010). In contrast, bupropion enhances NE availability in the vicinity of the cell body of NE neurons and decreases their firing rate through α2-adrenergic autoreceptor activation. With treatment prolongation, these autoreceptors desensitize, as well as those on NE terminals in the forebrain, and enhance NE transmission in the hippocampus (El Mansari et al., 2008; Ghanbari et al., 2011). In contrast, bupropion promptly enhances the firing of 5-HT neurons through enhanced NE release on excitatory α1-adrenergic receptors on 5-HT neurons. This constitutes indirect evidence that bupropion does not block NE reuptake because NE reuptake inhibitors, such as reboxetine, do not alter the firing rate of 5-HT neurons (Szabo and Blier, 2000a). Prolonged bupropion increases 5-HT transmission to a small extent after subacute administration, likely as a result of the increase in firing of 5-HT neurons. After prolonged administration, 5-HT transmission is enhanced to a marked extent in the presence of sustained elevation of 5-HT firing and the desensitization of the α2-adrenergic receptors on 5-HT terminals (Ghanbari et al., 2011).

The terminal 5-HT autoreceptor exerts an inhibitory influence on 5-HT release. When it is antagonized, 5-HT release is enhanced and when it is blocked in the presence of a SSRI, 5-HT levels are increased further (Sharp et al., 1997). In parallel with these observations, the terminal 5-HT autoreceptor gets desensitized with long-term administration of SSRIs (Piñeyro and Blier, 1999). There is no clinically available selective antagonist of the terminal 5-HT autoreceptor, but several atypical antipsychotics can block this autoreceptor; these include risperidone, paliperidone, ziprasidone and asenapine.

The 5-HT7 receptor has been identified as potential target for an antidepressant response. A selective 5-HT7 receptor antagonist is active in behavioural and animal models of antidepressant action. The following atypical antipsychotics are antagonists of this receptor: clozapine; risperidone; amisulpiride (Rauly-Lestienne et al., 2007; Abbas et al., 2009). Furthermore, in mice lacking the 5-HT7 receptor, amisulpiride does not exert its usual antidepressant-like action in two animal models. In Europe, low doses of amisulpiride are used in treatment-resistant MDD. This clinical benefit may result from a preferential blockade of presynaptic D2 receptors (Schoemaker et al., 1997), but possibly as well from the antagonism 5-HT7 receptors since amisulpiride has a 10 nm affinity for these receptors (Abbas et al., 2009).

All the above-mentioned strategies have been reported to be effective in treatment-resistant depression with various level of clinical evidence, ranging from double-blind randomized to case series (Fig. 4). It is sometimes difficult for some strategies to attribute a clinical success to a single mechanism because drugs can have more than one action known to contribute to an antidepressant response. The multiplicity of actions of some old antidepressants indirectly supports the notion that more mechanisms may lead to superior effectiveness. Nevertheless, it is important to follow what can be termed a neurobiological algorithm, that is, to avoid sequential steps that solicit the same mechanism(s) (Zigman and Blier, 2012). There follows some examples of such irrational polypharmacy: (1) switching from venlafaxine to duloxetine, or vice versa, when the first drug is well tolerated but there is no response at an adequate dose; (2) adding a SSRI to a SNRI, knowing that SNRI at doses that inhibit NE reuptake have an optimal effect on 5-HT reuptake; (3) using mirtazapine either alone or in combination following a failed trial of extended release quetiapine, as both drugs are α2-adrenergic, 5-HT2A/2C and histamine type 1 receptor antagonists, as well as 5-HT1A agonists; (4) using pramipexole or buspirone addition after a failed trial with the D2 and 5-HT1A agonist aripiprazole.

Fig. 4

Range of options for add-on therapy to standard antidepressants. The regimens mentioned are those generally reported to be effective. Most, but not all, of these strategies have been shown to be effective in double-blind studies.

How to choose logically an augmentation strategy

  1. Primum non nocere. Medications that would be relatively contraindicated should be avoided. Combining a SSRI/SNRI with a MAOI is an absolute contraindication because it would block the only two inactivating pathways for 5-HT, thus triggering a potentially lethal 5-HT syndrome. In contrast, using a selective NE reuptake inhibitor with a MAOI would not create such a catastrophe because NE can also be inactivated by catecholamine-O-methyl transferase. In fact, it would protect the patient from a hypertensive crisis following the ingestion of tyramine as it needs to penetrate NE terminals through the NE reuptake pump to release NE (Debonnel et al., 2007). Lithium addition should not be a first-line augmentation strategy in patients with hypothyroidism because of its potential to diminish thyroid function; thyroid stimulating hormone levels would have to be closely monitored for an eventual adjustment of thyroid supplementation.

  2. Choosing a medication that would target a cumbersome symptom could produce rapid relief before a full antidepressant response is achieved. For instance, in patients with insomnia and/or high anxiety levels, mirtazapine or quetiapine XR could resolve insomnia from day 1. A patient with low energy could potentially benefit more from an activating drug such as bupropion rather than a sedative one such olanzapine.

  3. Choosing a strategy that would treat a medical or psychiatric co-morbidity could minimize the number of medications prescribed. For instance, a patient with restless leg syndrome could benefit from pramipexole on two fronts, as it is indicated for that pathology and can be effective in treatment-resistant MDD (Hori and Kunugi, 2012). Risperidone augmentation could help patients with urinary flow problems due to prostate hypertrophy, due to its high α1-adrenergic antagonism (Schotte et al., 1996), which is a treatment for that condition. This strategy would be a first choice if a patient also has treatment-resistant obsessive–compulsive disorder since there are several positive augmentation studies with risperidone in this disorder (Dold et al., 2012).

  4. Considering properties of medications that would favour compliance is also desirable. These include undesired side-effects for individual patients, once per day regimens and cost of the drug.

  5. Finally, the weight of the clinical evidence for the effectiveness of a given strategy should help the clinician to make a final decision about the order of the preferred strategies being used.

When a monotherapy is implemented, if there is not a 20% improvement after 2 wk, the likelihood of achieving remission by the end of the trial is about 4% (Szegedi et al., 2009). In a drug addition to an ongoing treatment, the same 2-wk rule should also apply. Indeed, it is not because a medication has not produced a clinical change that it did not induce a biological change, it is just that a threshold may not have been reached to produce an improvement. It may thus be wise to maintain the first drug, first, to avoid discontinuation, second, to maintain a possible small improvement and, third, to try to potentiate it. This potentiation was readily observed when lithium augmentation was first used, that is, if patients were to respond to lithium addition even in the absence of any response, an improvement generally occurred within the first 2 or 3 wk if it were to work. As an illustration of this, if the non-responders are removed from the usual gradual improvement curves on the rating scales, the curves obviously become much steeper. The sequential combinations thus appear much more time-efficient (Fig. 5).

Fig. 5

Time-course of possible interventions to tackle major depressive disorder after a first failed trial. This figure is meant to show that within the time it generally takes to attempt two consecutive trials, with or without a washout period, it is possible to carry out at least two, if not three, consecutive combinations. In most cases, a substitution of the first added medication for the next one should be done to avoid giving three or more medications concomitantly.

As emphasized in the Introduction, there should be no hesitation in using a combination of two effective medications early in the treatment of MDD because of the low success rate of single agents (Blier et al., 2009, 2010). Physicians often do not hesitate to prescribe from treatment initiation an antidepressant and a benzodiazepine to help with insomnia and/or anxiety. The problem with this approach is that benzodiazepine is not an antidepressant and may cause cognitive dysfunctions, especially in elderly patients, can be problematic if alcohol is consumed and produce rebound anxiety and/or insomnia when it is time to discontinue it. One combination has been reported to be efficacious both from treatment initiation and as an augmentation strategy in more than one case series: venlafaxine and mirtazapine (see Hannan et al., 2007; Blier et al., 2010). In contrast, there have also been reports, both double-blind and single-blind, not showing a superior effectiveness of this combination (McGrath et al., 2006; Rush et al., 2011). In the latter studies, the doses of venlafaxine were not in the noradrenergic range for venlafaxine in all patients and, with regard to mirtazapine, most patients did not receive its therapeutic dose of 30 mg/d throughout the study.

In addition, in the CO-MED study by Rush et al. (2011), there was no difference in the clinical effectiveness of escitalopram monotherapy vs. escitalopram plus bupropion. It is noteworthy that the CO-MED was not a double-blind trial with clinicians titrating drug regimens openly throughout the study. This may help understand why more patients on escitalopram alone ended up on the 20 mg/d dose than in the combination arm. When using drug combinations for depression, the full potential of each medication should be exploited in much the same way, for instance, when using two antibiotics at full doses when treating septicaemia in the emergency room or a β2-agonist plus a steroid for asthma.

Subsequent analyses of the CO-MED data revealed that melancholia, chronic depression, co-morbid anxiety, medical co-morbidity, race and ethnicity and the level of anxiety symptoms did not affect the response to combination treatments (Bobo et al., 2011; Lesser et al., 2011; Chan et al., 2012; Davis et al., 2012; Morris et al., 2012; Sung et al., 2012). In contrast, combinations were more effective in reducing suicidal ideation (Zisook et al., 2011).

In conclusion, it is important to proceed methodically when treating MDD; using medications with complementary mechanisms of action at adequate doses, attempting to minimize side-effects or using them judiciously and implementing such strategies in a time-efficient fashion. When such a rational approach is utilized, it provides clinicians with a structured framework allowing them to proceed efficiently with confidence that will also minimize hopelessness for the treatment-resistant patient.


View Abstract