REVIEW ARTICLE
`
`Drugs 2011; 71 (1): 43-64
`0012-6667/11/0001-0043/$55.55/0
`ª 2011 Adis Data Information BV. All rights reserved.
`
`If at First You Don’t Succeed
`A Review of the Evidence for Antidepressant Augmentation,
`Combination and Switching Strategies
`
`K. Ryan Connolly1 and Michael E. Thase1,2
`
`1 University of Pennsylvania School of Medicine, Philadelphia VA Medical Center, Philadelphia,
`Pennsylvania, USA
`2 University of Pittsburgh Medical Center, Pittsburgh, Pennsylvania, USA
`
`Contents
`
`Abstract. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
`1. Methods of Literature Review . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
`2. Clinical Guidelines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
`3. Augmentation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
`3.1 Lithium . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
`3.2 Triiodothyronine . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
`3.3 Second-Generation Antipsychotics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
`3.3.1 Olanzapine . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
`3.3.2 Quetiapine . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
`3.3.3 Risperidone . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
`3.3.4 Ziprasidone . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
`3.3.5 Aripiprazole . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
`3.3.6 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
`3.4 Buspirone . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
`3.5 Pindolol . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
`3.6 Psychostimulants . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
`3.6.1 Traditional Psychostimulants . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
`3.6.2 Modafinil . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
`3.6.3 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
`Switching Antidepressant Medication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
`4.1 Changing to a Serotonin and Norepinephrine Reuptake Inhibitor versus Another
`Selective Serotonin Reuptake Inhibitor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
`4.2 Changing to Bupropion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
`4.3 Changing to Mirtazapine . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
`4.4 Changing to a Heterocyclic Antidepressant. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
`4.5 Changing to a Monoamine Oxidase Inhibitor. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
`5. Combination Therapy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
`5.1 Mirtazapine plus Newer Antidepressants . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
`5.2 Mianserin plus Newer Antidepressants . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
`5.3 Desipramine plus Newer Antidepressants . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
`6. Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
`6.1 Augmentation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
`6.2 Switching . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
`6.3 Combination Therapy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
`7. Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
`
`4.
`
`Merck 2003
`Argentum v. Merck
`IPR2018-00423
`
`

`

`44
`
`Abstract
`
`Connolly & Thase
`
`Major depressive disorder is a common and disabling illness that leads to
`significant reductions in quality of life and considerable cost to society. De-
`spite numerous advances in the pharmacological treatment of depression,
`many patients remain ill despite initial treatment. Beyond first-line treatment,
`current guidelines recommend either augmentation or switching of the initial
`antidepressant. In this narrative review, we summarize the data from ran-
`domized controlled trials and meta-analyses in order to concisely discuss how
`the impact of current research can be translated into clinical practice and,
`ultimately, into lasting improvements in patient outcomes. The augmentation
`strategies reviewed are lithium, thyroid hormone, pindolol, psychostimulants
`and second-generation antipsychotics. The data on switching from first-line
`antidepressants to other antidepressants are also reviewed, and include
`switching within the same class, switching to other first-line antidepressant
`classes and switching to less commonly prescribed antidepressants. Finally,
`the strategy of antidepressant combinations is examined. Overall, the strength
`of evidence supporting a trial of augmentation or a switch to a new agent is
`very similar, with remission rates between 25% and 50% in both cases.
`Our review of the evidence suggests several conclusions. First, although it
`is true that adjunctive lithium and thyroid hormone have established efficacy,
`we can only be confident that this is true for use in combination with tricyclic
`antidepressants (TCAs), and the trials were done in less treatment-resistant
`patients than those who typically receive TCAs today. Of these two options,
`triiodothyronine augmentation seems to offer the best benefit/risk ratio for
`augmentation of modern antidepressants. After failure of a first-line selective
`serotonin reuptake inhibitor (SSRI), neither a switch within class nor a switch
`to a different class of antidepressant is unequivocally supported by the data,
`although switching from an SSRI to venlafaxine or mirtazapine may poten-
`tially offer greater benefits. Interestingly, switching from a newer antide-
`pressant to a TCA after a poor response to the former is not supported by
`strong evidence. Of all strategies to augment response to new-generation
`antidepressants, quetiapine and aripiprazole are best supported by the evi-
`dence, although neither the cost effectiveness nor the longer-term benefit of
`these strategies has been established.
`The data to guide later steps in the treatment of resistant depression are
`sparse. Given the wide variety of options for the treatment of major depres-
`sive disorder, and the demonstrated importance of truly adequate treatment
`to the long-term outcomes of patients facing this illness, it is clear that further
`well conducted studies are needed.
`
`Major depressive disorder is a common and
`disabling illness that affects up to 15% of people
`over the course of their lives.[1] Those affected
`face significantly reduced quality of life, impaired
`ability to work and poorer overall health, while
`society incurs considerable economic costs.[1,2] The
`introduction of newer-generation antidepressants
`has improved our ability to treat depression, al-
`though only about 50–60% of patients will re-
`
`spond to first-line treatment and only 35–40%
`will experience a remission of symptoms during
`an initial 8-week trial.[3] Clearly, evidence-based
`second steps are needed if patients are to begin to
`recover from this serious illness.
`The question of how to proceed with the next
`step in depression treatment after an initially
`unsuccessful trial is a vital one; a good choice
`can improve outcomes and resolve illness, while
`
`ª 2011 Adis Data Information BV. All rights reserved.
`
`Drugs 2011; 71 (1)
`
`

`

`Treatment of Resistant Depression
`
`45
`
`persistent depression may lead to a more chronic
`and morbid course.[4,5] A number of options exist
`and are recommended in clinical guidelines, but
`guidance as to which option is best remains limited.
`This review aims to synthesize results from ran-
`domized clinical trials, meta-analyses and evidence-
`based clinical guidelines to make recommendations
`for the next step in the pharmacological treatment
`of adults with major depressive disorder who have
`not had an adequate clinical response to their first
`antidepressant medication treatment.
`
`1. Methods of Literature Review
`
`For this review, a search for evidence-based
`guidelines for the treatment of adults with major
`depressive disorder was performed on 1 March
`2010, using the National Guideline Clearinghouse
`database, the Agency for Healthcare Research
`and Quality Evidence Reports database and the
`Cochrane Database of Systematic Reviews. In
`addition, a ‘clinical query’ of the PubMed data-
`base and searches of drug manufacturers’ web-
`sites (for unpublished trials) were performed to
`identify randomized controlled trials (RCTs) and
`meta-analyses evaluating strategies to treat re-
`sistant depression.
`
`These guidelines provide the following evidence-
`supported first-line recommendations after the
`failure of an initial antidepressant trial that has
`been optimized in dose and duration:
`
` the addition of psychotherapy;[6,9]
` lithium augmentation;[6-9]
` augmentation with second-generation anti-
`psychotics (SGAs);[8,9]
` augmentation with triiodothyronine (T3);[6-9] and
` switching to another antidepressant.[6-9]
`
`We summarize the evidence for each of the
`medication options noted in order to help clini-
`cians decide which option is the best next step.
`These guidelines note that the strategies of com-
`bination therapy with two antidepressants and
`augmentation with stimulants or buspirone are
`commonly practiced but relatively unsupported
`by evidence; combination therapy is understudied,
`while there is evidence against the efficacy of bus-
`pirone for this purpose. The use of neurostimula-
`tion therapies (electroconvulsive therapy and
`transcranial magnetic stimulation) is recommend-
`ed later in the treatment course.[8] We note that the
`addition of psychotherapy is a helpful option, and
`one that is not incompatible with further medica-
`tion changes, but that a discussion of psychother-
`apy is beyond the scope of this review.
`
`2. Clinical Guidelines
`
`3. Augmentation
`
`Although many guidelines exist to aid in the
`initial management of major depressive disorder,
`recommendations for treatment-resistant depres-
`sion are more limited. The following depression
`guidelines were identified and reviewed:
`
` American Psychiatric Association Practice Guide-
`
`line for Treatment of Patients with Major Depres-
`sive Disorder (2000; Guideline Watch Update
`2005);[6]
` Clinical Practice Recommendations for Depres-
`sion (2009);[7]
` Canadian Network for Mood and Anxiety Treat-
`
`ments (CANMAT) Clinical Guidelines for the
`Management of Major Depressive Disorder in
`Adults (2009);[8] and
` Institute for Clinical Systems Improvement
`
`Augmentation is the addition of an agent – not
`thought to be an antidepressant itself – to an
`antidepressant regimen in order to improve efficacy.
`Also referred to as adjunctive therapy, currently
`recommended agents include lithium, SGAs and
`T3. These agents are recommended as adjuncts to
`first-line antidepressants, which, as per the guide-
`lines in section 2,
`include selective serotonin
`reuptake inhibitors (SSRIs), serotonin and nor-
`epinephrine reuptake inhibitors (SNRIs), bupro-
`pion and mirtazapine.[7,8] Stimulants and buspirone
`are also used for adjunctive therapy, although they
`are not considered first line in current guidelines
`due to a lack of strong supporting evidence.
`
`3.1 Lithium
`
`(ICSI) Healthcare Guideline for Major De-
`pression in Adults in Primary Care.[9]
`
`Lithium salts are among the oldest drugs used
`in psychiatry and have been used since the 1960s
`
`ª 2011 Adis Data Information BV. All rights reserved.
`
`Drugs 2011; 71 (1)
`
`

`

`46
`
`Connolly & Thase
`
`in attempts to augment antidepressants.[10] Since
`then, of all of the strategies used to augment an-
`tidepressants, lithium remains one of the most
`extensively studied in RCTs.[11] It was initially
`proposed to act by increasing serotonergic neu-
`rotransmission; 30 years of subsequent research
`has not produced a more definitive answer, and
`this remains a viable hypothesis.[12,13]
`A recent meta-analysis by Crossley and
`Bauer[14] (2007) found five studies of lithium as
`an ‘accelerator’ of antidepressant response and ten
`studies of lithium augmentation of antidepressant
`medications that warranted inclusion in their meta-
`analysis. No significant effect of lithium on the
`speed of antidepressant response was noted.
`By contrast, the efficacy of lithium augmen-
`tation was confirmed, with an overall odds ratio
`for response of 3.1 (1.8–5.4) favouring lithium.[14]
`Eight of the ten RCTs included in the meta-anal-
`ysis showed a significant benefit of lithium added
`to an antidepressant; pooling the results of the
`ten trials, the number needed to treat (NNT) to
`achieve one clinical response can be calculated as 4.
`Of the two studies that failed to observe a
`statistically significant effect favouring lithium
`augmentation,
`it is plausible that one failed
`to do so because of the low doses of lithium used
`(the mean plasma concentration was only
`0.25 mEq/L).[15] However, this was unlikely in
`the second study.[16]
`In that study, depressed patients with a history
`of significant treatment resistance (defined as
`at least one, but no more than five, unsuccessful
`antidepressant treatment trials in the current
`episode) were first treated for 6 weeks with nor-
`triptyline alone (mean dose 116.7 mg/day, plasma
`concentration 95.7 ng/mL), with those who did
`not respond (n = 92) randomized to 6 weeks of ad-
`ditional double-blind augmentation therapy with
`either lithium or placebo. There was no evidence
`of efficacy whatsoever; intent-to-treat response
`rates were 11.1% and 17.6%, respectively. A pos-
`sible explanation is that most of the patients in
`this study had already received one or more ade-
`quate trials of therapy with SSRIs before the
`sequenced combination of nortriptyline and lithi-
`um, and, therefore, had little potential to benefit
`from lithium, with its comparatively weaker ef-
`
`fects on serotonergic neurotransmission; that is,
`if enhancing serotonergic neurotransmission were
`to be effective for these patients, then responses
`would likely have already been seen with SSRIs.
`If this is true, one might hypothesize a lower
`efficacy for lithium as an augmenter of SSRIs,
`the most commonly recommended first-line
`antidepressants.[7,8]
`Lithium effectively augments response to tri-
`cyclic antidepressants (TCAs). However, since
`that first generation of research, the TCAs have
`been replaced as the first line of antidepressant
`therapy by the SSRIs and several other antide-
`pressants, and surprisingly few data exist on these
`combinations. Thus, it is unclear that lithium
`provides similar benefits when added to other
`antidepressant medications. Nevertheless,
`the
`majority of open-label or uncontrolled trials
`suggest a beneficial effect of this combination and
`the small randomized, placebo-controlled trial by
`Baumann and colleagues[17] found a large benefit
`of this combination in 24 patients for whom ci-
`talopram monotherapy had proved ineffective;
`60% versus 14% response with lithium and pla-
`cebo augmentation, respectively.[17-20]
`This dramatic effect was not replicated in the
`more recent STAR*D (Sequenced Treatment
`Alternatives to Relieve Depression) trial, which
`compared lithium and T3 augmentation in 142 pa-
`tients who failed to respond to two sequential
`trials with newer antidepressants (i.e. citalopram,
`followed by sertraline, bupropion or venlafaxine
`extended release [XR]).[21] This study used a ran-
`domized but open-label design, with the primary
`outcome assessment completed by an indepen-
`dent evaluator without knowledge of treatment
`assignment. In this trial, only 15.9% of the pa-
`tients treated with lithium augmentation remit-
`ted, compared with 24.7% of the patients treated
`with T3 augmentation; this difference was not
`statistically significant. Four caveats are note-
`worthy. First, the primary outcome of remission
`(specified as £7 on the 17-item Hamilton De-
`pression Rating Scale [HDRS]) is far more strin-
`gent than that of response (50% improvement),
`which had been the typical outcome measure in
`prior studies. Second, lithium doses were rela-
`tively low and indices of tolerability, including
`
`ª 2011 Adis Data Information BV. All rights reserved.
`
`Drugs 2011; 71 (1)
`
`

`

`Treatment of Resistant Depression
`
`47
`
`weeks ‘on therapy’, favoured the group treated
`with T3 augmentation. Third, the randomization
`strategy used in STAR*D, an equipoise strati-
`fied randomization that took patient preference
`into account, skewed the population of the aug-
`mentation study to include a disproportionate
`number of patients who were partial respon-
`ders to antidepressant therapy. If response to
`lithium augmentation is better among patients
`with more severe symptoms, the randomization
`procedure may have biased the trial in favour of
`T3 augmentation. Finally, as in the study of
`Nierenberg et al.,[21] which failed to detect a
`significant drug versus placebo difference in fa-
`vour of lithium augmentation, all patients in the
`STAR*D trial had not obtained adequate ben-
`efit from at least two courses of antidepressant
`therapy.
`Lithium is one of the oldest and most studied
`agents used to improve outcomes in patients who
`experience inadequate results from an antide-
`pressant
`trial. The evidence is strongest
`for
`patients whose depression is inadequately treated
`by an initial trial of a TCA. However, the data for
`augmentation of newer antidepressants are weak,
`and significant benefit has not been observed in
`the two most recent trials of this strategy (table I).
`As the TCAs are now typically reserved for
`patients with more advanced levels of treatment
`resistance, the relevance of lithium augmentation
`in contemporary practice is less clear than might
`be assumed from the strong recommendations
`found in current guidelines; given the adverse ef-
`fects, relatively low therapeutic index and longer-
`term risks of thyroid and renal compromise
`associated with this agent, it should be used to
`augment SSRIs with caution.
`
`Table I. Lithium and triiodothyronine (T3) augmentation in ran-
`domized, placebo-controlled trials (RCTs)
`
`Agents
`
`N in RCT NNT
`
`Lithium for TCA nonresponse[14]
`
`269
`
`4
`
`3.2 Triiodothyronine
`
`Thyroid hormones have been less extensively
`studied than lithium as augmenters of antidepres-
`sant efficacy. In contrast to practice guidelines
`for the treatment of hypothyroidism, which rec-
`ommend levothyroxine (T4), the preferred form
`of thyroid hormone for adjunctive use in combi-
`nation with antidepressants is T3 because of the-
`ories of its activity within the CNS and perceived
`rapid onset of action.[24] Like lithium, T3 was in-
`itially proposed as a strategy to accelerate TCA
`response. The first placebo-controlled study was
`conducted by Prange and colleagues[25] in 1969,
`and reported that T3 was effective at improving and
`accelerating antidepressant response to imipramine.
`Since this classic study, numerous others have in-
`vestigated the potential role of T3 in improving
`antidepressant response.
`A meta-analysis by Aronson et al.[22] (1996)
`focused on the efficacy of T3 augmentation in
`studies of patients who had not responded to
`TCAs. Compared with placebo-treated patients,
`those who received augmentation with T3 were
`twice as likely to respond; the absolute increase in
`response rate was 23.2%, for an NNT of 4.3.[22]
`As with lithium augmentation, an important
`question remains as to whether these findings can
`be generalized to treatment with newer antide-
`pressants. Though the STAR*D results discussed
`in section 3.1 suggest that T3 may have an ad-
`vantage compared with lithium in this regard,
`this study did not include a placebo control and
`the difference between the two augmentation
`groups was not statistically significant on the
`primary dependent measure.[21]
`Five RCTs have been reviewed by Cooper-
`Kazaz and Lerer[26] to evaluate the ability of T3 to
`‘enhance’ (improve response by initial coadmin-
`istration of SSRI and T3) or augment response to
`SSRIs. Of note, the single trial that demonstrated
`a statistically significant benefit over placebo ac-
`tually did not test T3 augmentation, but rather
`evaluated coadministration of T3 from the outset
`of therapy with a relatively modest dose of ser-
`traline (maximum dose 100 mg/day).[27] Remitters
`showed lower baseline T3 and a greater suppres-
`sion of thyroid-stimulating hormone than did
`
`24
`
`292
`
`Lithium for SSRI nonresponse[17]
`T3 for TCA nonresponse[22]
`T3 for SSRI nonresponse[23]
`36
`No advantage found
`N = number of subjects; NNT = number needed to treat
`for one
`response; SSRI = selective serotonin reuptake inhibitor;
`clinical
`TCA = tricyclic antidepressant.
`
`2.2
`
`4.3
`
`ª 2011 Adis Data Information BV. All rights reserved.
`
`Drugs 2011; 71 (1)
`
`

`

`48
`
`Connolly & Thase
`
`nonremitters, perhaps suggesting that T3 was
`enhancing suboptimal thyroid function. The sole
`randomized, placebo-controlled trial of T3 aug-
`mentation of SSRI therapy failed to find a sig-
`nificant benefit.[23]
`T3 augmentation has established efficacy for
`patients who have not responded to TCAs, with
`meta-analyses suggesting an NNT that is essen-
`tially equivalent to that of lithium. Given the
`generally better tolerability and greater ease of
`administration of T3 compared with lithium
`augmentation, T3 might be thought of as a better
`option for use in combination with SSRIs and
`other newer antidepressants.[21] Yet, the remis-
`sion rate observed in STAR*D (24.7%) was not
`significantly better than with other options, and
`prolonged use of adjunctive thyroid hormone is
`not without safety concerns (e.g. osteopenia). As
`the only placebo-controlled study of adjunctive
`T3 therapy of SSRIs failed to establish a significant
`benefit, we therefore conclude that additional
`RCTs are needed before T3 can be recommended
`for euthyroid patients who have not responded to
`first-line antidepressants (table I).
`
`3.3 Second-Generation Antipsychotics
`
`In recent years, research on the so-called atyp-
`ical antipsychotics or SGAs has shifted from
`schizophrenia and mania to investigating their
`utility for treatment of depressive disorders, both
`unipolar and bipolar, and both alone (mono-
`therapy) and in combination with antidepres-
`sants. The recent timing of these studies has
`meant that this line of research has focused on
`adjunctive therapy with current first-line treat-
`ment strategies (i.e. SSRIs and SNRIs). Further-
`more, unlike the case for research on adjunctive
`therapy with lithium or T3, the impetus of this
`research has included a strong commercial moti-
`vation to expand the ‘on-label’ use of these med-
`ications (i.e. in the US, a medication cannot be
`promoted for a particular use until there has been
`a formal US FDA indication). As a result of such
`proprietary motivation, many of the studies of
`SGAs have followed the conventions required for
`consideration as pivotal studies (i.e. for review by
`regulatory authorities). Thus, a majority of stud-
`
`ies of adjunctive therapy with SGAs have used
`standardized definitions of treatment resistance,
`have included a placebo control group and are
`relatively large, which affords sufficient statistical
`power to reliably detect even modest differences
`in symptom reduction or response rates (i.e. 2- to
`4-point differences in rating scale scores or
`10–20% differences in response or remission rates).
`Moreover, as agencies such as the FDA typically
`require replication of positive studies before
`granting a formal indication, there are generally
`at least two adequately controlled studies of ad-
`junctive therapy for each SGA. Although all of
`these characteristics are methodological strengths,
`the absence of independent replication (i.e. studies
`not sponsored by the manufacturer) does to some
`degree constitute a countervailing limitation.
`The neuropharmacological rationale for the
`use of SGAs to augment SSRIs or SNRIs is large-
`ly based on the broad receptor binding profiles
`of this class of medications.[28] For example, like
`the antidepressants trazodone, nefazodone or
`mirtazapine, SGAs such as olanzapine, quetia-
`pine, aripiprazole and risperidone bind strongly
`to serotonin 5-HT2A receptors. Ziprasidone and
`aripiprazole are also 5-HT1A partial agonists, a
`pharmacological effect that is shared by buspirone
`(discussed in section 3.4). In addition, ziprasidone
`and quetiapine (via its N-desyalkyl metabolite)
`have been shown to inhibit norepinephrine reup-
`take, similar to TCAs such as nortriptyline and
`desipramine. It is therefore possible that different
`SGAs may be helpful for different patients or
`more useful with different antidepressants. This
`speculation has, to date, eluded definitive study.
`The adjunctive efficacy of the SGAs has been
`the subject of a number of recent reviews.[29,30]
`Before summarizing the data from controlled
`trials, we also wish to comment on concerns over
`their adverse effects. In addition to the uncertain
`longer-term risks of tardive dyskinesia and the
`known risks of acute extrapyramidal syndromes,
`the increasingly larger-scale use of the SGAs ne-
`cessitates that prescribers are well
`informed
`about their particular risks of metabolic side ef-
`fects and complications, including weight gain,
`lipid abnormalities and impaired glucose toler-
`ance, which have been reviewed elsewhere.[31]
`
`ª 2011 Adis Data Information BV. All rights reserved.
`
`Drugs 2011; 71 (1)
`
`

`

`Treatment of Resistant Depression
`
`49
`
`Clinicians must also be well versed on the best
`strategies to minimize these risks and, when such
`metabolic complications develop, to ensure that
`they are properly managed. As such, the potential
`therapeutic benefits of these medications must be
`balanced against the known risks and there is an
`even greater need for longer-term studies to fully
`describe the benefit-risk profiles of these medi-
`cations for depressed patients.
`
`3.3.1 Olanzapine
`
`An indirect descendent of the prototypical
`SGA clozapine, olanzapine was the first member
`of this class to be systematically studied as an
`adjunct to SSRI treatment. The first such trial, by
`Shelton et al.[32] in 2001, was a small but well con-
`trolled trial of patients with a history of treatment-
`resistant depression, which is conventionally defined
`as a failure of a depressive episode to respond to
`two consecutive trials of antidepressant therapy
`given at an adequate dose for at least 4 weeks; in
`this case, patients experienced one failure each of
`an SSRI and a non-SSRI antidepressant. Patients
`were next started on a prospective trial of fluoxetine.
`After 8 weeks of this therapy, 28 nonresponders
`were randomized to 8 weeks of additional ther-
`apy with one of three strategies: continued
`fluoxetine plus placebo, olanzapine plus place-
`bo (i.e. fluoxetine was discontinued) or fluox-
`etine plus olanzapine (OFC). At the end of the
`study period, remission rates were substantially
`higher in the OFC group than either mono-
`therapy group: 60% remission was seen with
`OFC, 25% with olanzapine alone and 20% with
`fluoxetine alone.
`Following these encouraging results, four ad-
`ditional trials were undertaken to evaluate the
`efficacy of OFC as a strategy for treatment-
`resistant depression. To date, there have been no
`large-scale studies of olanzapine in combination
`with other antidepressants.
`The first of these trials to be published, an
`8-week, double-blind study, enrolled patients with
`a history of nonresponse to one SSRI.[33] Patients
`were first treated with open-label nortriptyline,
`with dose escalation up to 175 mg/day. 500 non-
`responders to a prospective 8-week, open-label
`trial of nortriptyline monotherapy were next
`
`randomized to switch to olanzapine, fluoxetine or
`OFC or to continue on nortriptyline for 8 weeks
`of double-blind therapy. The authors suggested
`that
`the disappointing performance of OFC
`might be due to overall lower doses of olanzapine
`than in the initial study. It may also be the case
`that the demand characteristics of the study un-
`dermined internal validity. Specifically, the un-
`expectedly strong performance of the group that
`received ongoing therapy with nortriptyline raises
`the possibility that the sequential study design in-
`advertently incentivized an undervaluation of the
`first, open-label course of nortriptyline therapy.
`A second large trial recruited 483 depressed
`patients who reported a history of not responding
`to one trial of SSRI therapy.[34] Participants first
`received a 7-week course of therapy with venla-
`faxine immediate release (mean 226 mg/day), with
`nonresponders randomly assigned to 12 weeks of
`additional double therapy with a switch to OFC,
`olanzapine or fluoxetine, or one of two control
`groups: ongoing therapy with venlafaxine or a
`low dose of OFC (olanzapine/fluoxetine 1 mg/5 mg).
`As was the case in the previous trial, the group
`that received ongoing therapy with venlafaxine
`was intended to control for the effect of additio-
`nal time on study therapy. By contrast, the group
`that received low doses of OFC was intended to
`serve as an active placebo-control condition. The
`only significant difference in efficacy seen was
`between OFC and olanzapine alone. As was the
`case with nortriptyline in the previous trial, the
`patients who continued to receive venlafaxine
`had a surprisingly high response, which was not
`appreciably different from the group switched to
`full doses of OFC. Perhaps even more note-
`worthy was the observation that the patients who
`were randomly assigned to receive low-dose OFC
`also did nearly as well as those who received full
`doses of OFC. Again, the demand characteristics
`of the study, which may have promoted an un-
`dervaluation of the initial course of venlafaxine
`therapy, could have unintentionally undercut the
`internal validity of the experiment.
`The final two large trials were identical in de-
`sign and mirrored the original protocol used by
`Shelton et al.[32] As such, these studies were true
`augmentation studies, rather than switch studies.
`
`ª 2011 Adis Data Information BV. All rights reserved.
`
`Drugs 2011; 71 (1)
`
`

`

`50
`
`Connolly & Thase
`
`By plan, these trials were pooled and analysed
`together to provide greater statistical power.[35]
`This pair of studies enrolled depressed patients
`with a history of one prior unsuccessful course of
`antidepressant therapy and first treated them
`with a prospective 8-week course of fluoxetine
`(mean dose 47.4 mg/day). Patients who did not
`respond to fluoxetine were randomized to 8 addi-
`tional weeks of double-blind therapy with fluox-
`etine alone, olanzapine alone or OFC. The first
`trial, which randomized 638 patients, showed no
`benefit of OFC compared with ongoing therapy
`with fluoxetine al