Second-generation antipsychotics

The value of antipsychotic medications for a select subgroup of depressed patients was recognized shortly following the introduction of medications such as chlorpromazine and thioridazine; the potential for additive benefit in combination with antidepressants was explicitly recognized by the marketing of several proprietary formulations, including amitriptyline plus perphenazine and tranylcypromine plus trifluoperazine. However, by the late 1980s, use of these “first-generation” antipsychotics was largely delimited to treatment of more severely depressed patients, particularly those experiencing psychotic features. This practice changed within a few years of the introduction of risperidone and olanzapine, the first members of the SGA class. The SGAs are now the best proven option for adjunctive therapy following inadequate response to antidepressant therapy of MDD ( ; ; ). In the United States, three SGAs (aripiprazole, quetiapine, and brexpiprazole) are approved by the Food and Drug Administration (FDA) for the broader indication of inadequate response to antidepressants, and a fourth—olanzapine, specifically in combination with fluoxetine—has FDA approval for the stricter classification of treatment-resistant depression (TRD). Yet another SGA, risperidone, has established adjunctive efficacy in controlled trials ( ) even though the manufacturer opted not to seek FDA approval for this indication. As there are positive placebo controlled studies of adjunctive therapy with a number of other SGAs, including ziprasidone ( ), cariprazine ( ) and pimavanserin ( ), it may well be that most if not all of the SGAs has efficacy as adjuncts to antidepressants.

Looking across studies, there are several key clinical observations. First, significant drug placebo differences are routinely observed within the first 2 weeks of adjunctive therapy with SGAs. The SGAs thus have a faster onset of effect than conventional antidepressants. Second, the dose needed to exert such a rapid effect tends to be lower than the minimal dose needed for antipsychotic efficacy in patients with schizophrenia or mania. For aripiprazole, for example, the modal dose for antidepressant responders is 5 mg/day, whereas 15 mg/day is generally considered to be the minimal effective dose for treatment of schizophrenia or mania ( ). Third, unlike conventional antidepressants, the SGAs appear to be comparably effective for patients with or without prominent anxiety symptoms. This observation does not mean that adjunctive SGA therapy is specifically indicated for patients with high levels of anxiety, but does reflect the clinical observation that they are useful for a broad range of patients who do not benefit from antidepressant (see, for example, ). And fourth, although there are marked differences between the various members of the SGA class in terms of side effects and risks of weight gain and other metabolic complications ( ; ), their efficacy—as determined by drug versus placebo differences in RCTs—appears to be roughly comparable.

It may also be that some—if not most—of the SGAs could be antidepressants if prescribed as monotherapy. In this respect, it is noteworthy that the efficacy of quetiapine monotherapy in MDD has been established by multiple RCTs in doses ranging from 50 to 300 mg/day ( ). Although approved for this indication in some countries, the US FDA did not recommend approval of quetiapine monotherapy of MDD because of concerns about side effects and the potential for longer term complications, including tardive dyskinesia. The potential role of SGA monotherapy for a carefully selected subset of depressed patients is underscored by the fact that four members of the class—olanzapine, quetiapine, lurasidone, and cariprazine—have established efficacy for treatment of bipolar depression ( ). In the case of lurasidone, monotherapy efficacy also has been shown in one placebo controlled study of patients with MDD with mixed features ( ).

Although inspection of effect sizes strongly supports the notion that these medications have similar efficacy as adjunctive therapies, it must also be noted that there is a dearth of data from comparative RCTs. Indeed, there is only one placebo controlled RCT in the published literature comparing two SGAs as adjunctive therapies for antidepressant nonresponders ( ). In this trial, 503 MDD outpatients with a history of nonresponse to antidepressant therapy in the current depressive episode as well as nonresponse to a prospective, 8–10 week trial of antidepressant monotherapy were randomized to 6 weeks of double blind adjunctive therapy with brexpiprazole, quetiapine or placebo. At study endpoint, they found that brexpiprazole (titrated to 1–3 mg/day) was efficacious (versus an adjunctive placebo; mean MADRS difference 1.5 points, P < .008), whereas quetiapine (titrated to 150–300 mg/day) was not (mean MADRS difference 0.3 points). One important limitation in the interpretability of this study is that there were no planned comparisons between the two SGAs. In fact, the quetiapine group was included in the design to determine “assay sensitivity” and was only one half the size of the brexpiprazole and placebo groups. Thus, it cannot be said that adjunctive brexpiprazole therapy was significantly more effective than adjunctive quetiapine therapy in this study.

There is also a dearth of studies of the comparative efficacy of SGAs and other clinically relevant options. Only two large scale studies exist, one of quetiapine and the other of aripiprazole, contrasting adjunctive SGA therapy against proven strategies. In the first, randomized 681 outpatients with MDD with a history of antidepressant nonresponse to 6 weeks of open label therapy; about one-third of the sample was switched quetiapine (300 mg/day), one-third received quetiapine as an adjunct (300 mg/day), and one-third received adjunctive therapy with lithium salts (target blood level: 0.6–1.0 mEq/L). The latter intervention was chosen as the existing standard of adjunctive therapy. They found that adjunctive therapy with quetiapine was significantly more effective than the other two options after 4 days of treatment ( P < .01), which is consistent with the notion that this strategy has a more rapid onset of benefit. After 6 weeks of therapy, the adjunctive quetiapine group continued to show a larger effect than the adjunctive lithium group (difference in MADRS scores of 2.32; nominal [unadjusted] P < .05). However, as the primary hypothesis of this study was a strict test of noninferiority, the conclusion was that adjunctive therapy with quetiapine was not inferior to adjunctive lithium therapy. There also did not appear to be a meaningful difference in tolerability between the adjunctive interventions.

In an even larger study ( ), 1522 Veterans with MDD who had not responded to at least one adequate trial of antidepressant monotherapy in the current episode were randomized to up to 36 weeks of treatment with one of three strategies: switch to bupropion, combined therapy with bupropion and the current therapy (typically an SSRI or SNRI) or adjunctive therapy with aripiprazole (dosage titrated from 2 to 15 mg/day). They found adjunctive aripiprazole to be significantly more effective than switching to antidepressant monotherapy with bupropion (week 12 response rates: 74% vs 62%, respectively); comparisons between adjunctive aripiprazole arm and combination antidepressant therapy yielded intermediate differences that generally did not reach statistical significance.

Longer term considerations . Although we are now have more than two decades of experience with some of the older SGAs, there are still some important questions about their use as adjunctive therapies for patients with MDD and it seems likely that these questions will not be answered by clinical trial data ( ). Perhaps most important is the uncertainty about just how long an effective adjunctive therapy be continued. Clinical experience informs us that many patients who respond to adjunctive SGA therapy will relapse if the adjunct is withdrawn during the first few months after response, so a standard of practice has emerged for patients to routinely receive at least 4 to 6 months of adjunctive therapy. The need for a continuation phase of treatment forces clinicians to be vigilant to minimize the hazards of weight gain and other metabolic side effects, which are likely to be cumulative across at least the first 6 months of therapy. At some point of continuation therapy, clinicians also must begin to monitor for signs of treatment-emergent dyskinetic movements. In fact, in one study in which patients who responded to adjunctive aripiprazole (mean dose 10 mg/day) could receive up to 1 year of continuation therapy, 4 of 987 patients (about 0.4%) developed dyskinesias ( ). Importantly, in all four cases the dyskinetic movements first diminished and then ceased after withdrawal of aripiprazole, which suggests that clinician vigilance and ongoing monitoring can have rewarding outcomes. The risk of weight gain in this study (4.4 kg) also was not trivial, with about 37% among the 303 patients who completed a year of adjunctive therapy gaining at least 7% body weight.

There are simply not enough data on longer term adjunctive treatment with SGAs in MDD patients. The study of Brunner and colleagues ( ), a double-blind relapse prevention study that enrolled 444 outpatients with TRD who responded to an 8-week open-label course of olanzapine-fluoxetine combination and who did not relapse across 12 weeks of continuation therapy, is the largest of its kind. They found that, after 20 weeks of treatment, the risk of relapse over the next 6 months of double blind therapy was about 32% for the patients who were withdrawn from olanzapine (double blind) and maintained on fluoxetine monotherapy, compared to 16% among those who continued to take the combination of olanzapine and fluoxetine. One patient in the combined treatment arm developed dyskinetic movements during the 6 months of continued therapy, which suggests a risk of treatment-emergent dyskinesia during longer term olanzapine therapy that is roughly comparable to that reported for aripiprazole. Weight gain was a much larger problem in this trial, with about a 4 kg difference in mean weight at the end of the 6 month trial, accompanied by and subtler changes in selected metabolic parameters, including fasting triglycerides. Moreover, looking across treatment phases, 56% of the patients who received 10 + months of therapy with OFC experienced a weight gain of at least 7% of their pretreatment weight. Although open label data collected during ongoing adjunctive therapy with other SGAs suggests somewhat lower risks ( ), it is clear that careful monitoring of weight and metabolic parameters is indicated during longer therapy.

Lithium salts

The therapeutic effects of lithium salts, principally lithium carbonate and lithium citrate, were recognized for patients with recurrent mood disorders from that outset of the first generation of psychopharmacology. Moreover, lithium—typically in doses of 600–1200 mg/day—was the first adjunctive medication found to be efficacious for antidepressant nonresponders on the basis of multiple placebo controlled studies ( ; ; ). Originally described to have a remarkably rapid onset of benefit (i.e., within 48 h ( )) in patients who had not responded to 3–4 weeks of TCA therapy, subsequent reports suggested that many patients needed a more conventional 6–8 weeks of treatment with serum levels (e.g., 0.5–0.8 mEq/L; see, for example, ). In the first metaanalysis of placebo controlled trials, identified 10 eligible studies and confirmed a meaningful advantage for adjunctive lithium salts as compared to the double blind placebo. With an average benefit estimated as an odds ratio of about 3, Crossley and Bauer’s findings suggested that one might expect up to a 20% advantage over and above placebo-expectancy effects, which would correspond to a number needed to treat of about 5. It is noteworthy that only two of the RCTs included in Crossley and Bauer’s metaanalysis studied patients who were taking SSRIs. Thus, their findings and conclusions may be more relevant to adjunctive therapy of TCAs.

Several subsequent metaanalyses have permitted estimations of benefit to be updated and refined, in part by including several more recent studies, and, in part, by use of newer analytic methods to estimate relative efficacy ( ; ). Zhou and colleagues ( ), for example, found that the average benefit of adjunctive lithium therapy was smaller than originally estimated by Crossley and Bauer (Odds Ratio = 1.56). In terms of the average benefit over and above placebo, an effect of size magnitude would suggest an average benefit of only about 10% absolute advantage, which corresponds to a number needed to treat of about 10. In their network metaanalysis, estimated that the benefit was generally comparable to the estimated effect sizes of the SGAs. Consistent with the findings of the RCT conducted by Bauer and colleagues ( ) discussed earlier, concluded that the tolerability of adjunctive lithium therapy also was comparable to adjunctive SGA therapy across 4–8 weeks of treatment ( Fig. 11.1 ).

Network metaanalysis of efficacy of SGAs for adjunctive treatment of MDD.

Before the introduction of the SGAs, adjunctive lithium therapy was the best-proven adjunctive therapy for patients with difficult to treat depression and was generally considered to be more effective than the major alternative of the era, adjunctive therapy with thyroid hormones ( ). With this question in mind, the STAR*D research project included a head-to-head trial that directly compared adjunctive therapy with lithium carbonate (up to 900 mg/day) and thyroid hormone (triiodothyronine [T ₃ ], up to 50 μg/day)( ). Like all of the STAR*D comparisons, a randomized, open label and rated-blinded design was used to compare the adjunctive therapies across 12 weeks of open label treatment. The study enrolled 142 MDD patients who had not responded to about 6 months of treatment, including a 12 week course of citalopram monotherapy, followed by either a second course of antidepressant treatment (with sertraline, venlafaxine or bupropion) or “augmentation” of citalopram by either buspirone or bupropion. Neither adjunctive therapy was particularly effective, with intent to treat remission rates of 16% and 25%, for lithium and thyroid respectively, across up to 12 weeks of therapy. By contemporary standards, a 9% nominal difference in remission rates in patients with difficult to treat depression would be strongly suggestive of an advantage for thyroid hormone, the STAR*D study suffered from a higher than expected attrition rate and statistical power was not as high as planned. Nevertheless, tests of several secondary outcomes pertaining to acceptability and side effects did reveal significant advantages for the patients who received adjunctive thyroid hormone. The unexpectedly poor showing of adjunctive lithium therapy in STAR*D was not limited to tolerability: prescribers also had problems implementing lithium therapy, with infrequent monitoring of serum levels and a tendency to prescribe lower than recommended lithium doses.

Perhaps heralded by the STAR*D findings but more likely a consequence of prescribers’ growing confidence in the value of combined antidepressant therapy and adjunctive SGA therapy, the role of lithium salts in the management of patients with difficult to treat forms of MDD has diminished over the past 15 years. Nevertheless, there important reasons to continue to consider this time-tested strategy for selected patients, particularly for patients at more advanced levels of treatment resistance. Beyond efficacy in MDD that is comparable to SGAs ( ; ), other potential reasons to consider lithium salts include its utility across phases of treatment for bipolar affective disorder ( ) and its established efficacy for relapse/recurrence prevention for patients with recurrent (unipolar) MDD ( ; ). Of particular relevance to patients with more advanced levels of TRD is prophylactic efficacy of the combination of lithium and the TCA nortriptyline following successful treatment with ECT ( ). Finally, although evidence of lithium’s “antisuicide” effect is largely derived from studies of bipolar disorder ( ), a review conducted under the auspices of the Cochrane Collaboration concluded that this benefit was relevant to all patients with recurrent mood disorders ( ).

Longer-Term Considerations . If effective, adjunctive lithium may be continued for months for relapse prevention or years for maintenance therapy. Later emerging side effects include skin changes (including acne and psoriasis) and weight gain, which may suggest development of hypothyroidism. For this reason, thyroid stimulating hormone (TSH) levels are monitored periodically and, when clinically indicated, on the basis of newly emergent symptoms or side effects. Onset of symptoms such as excessive thirst or polyuria may reflect an impairment of renal concentrating ability, which in turn is a consequence of lithium-induced interstitial fibrosis of the kidneys. This “not uncommon” consequence of years of lithium therapy is arguably the most serious complication of maintenance treatment and, if not addressed, may eventually cause chronic renal failure. Following standards of care developed for maintenance lithium therapy of bipolar disorder ( ; ), periodic monitoring of blood levels and measures of renal function, including electrolytes, plasma BUN and creatinine levels and 24 h urinary creatinine clearance levels are recommended to identify patients at risk for renal complications of adjunctive lithium therapy.

Thyroid hormones

The importance of the thyroid axis in the maintenance of well-being and of disturbances of thyroid hormone—both hypothyroidism and hyperthyroidism—in the pathogenesis of mood disorders were known early in the first generation of antidepressant therapy ( ; ). On the basis of small studies by Prange and colleagues ( ; ), it was hypothesized that adjunctive therapy with thyroid hormone could accelerate response to TCAs such as imipramine or even be used as an adjunct to convert nonresponders into remitters. Several smaller RCTs tested this hypothesis and the results of a metaanalysis ( ) suggested that adjunctive therapy with T ₃ (25–50 μg/day) significantly lowered depressive symptoms. However, the overall effect on response rates in the four placebo controlled trials (OR = 1.53) did not reach conventional standards of statistical significance (see Fig. 11.2 ).

Metaanalysis of placebo-controlled studies of adjunctive therapy with lithium salts.

That the efficacy of adjunctive therapy with thyroid hormone was comparable to that of adjunctive therapy with lithium carbonate and SGAs. As noted earlier, results of the STAR*D comparison tended to favor T ₃ over adjunctive lithium therapy, particularly with respect to ease of implementation and better tolerability ( Fig. 11.3 ). However, the most recent systematic review of adjunctive thyroid therapy concluded there were too few studies in well-documented TRD to establish efficacy ( ).

Adjunctive therapy with thyroid hormone in STAR*D: results of level 3 comparison with lithium.

It is noteworthy that most of evidence pertaining to adjunctive therapy with thyroid hormone in MDD is derived from studies utilizing triiodothyronine (T ₃ )instead of the more commonly prescribed thyroxine (T ₄ ); this is because it was known early on that the brain preferentially uses T ₃ ( ). Although a small body of work can be marshaled to support this practice ( ), most people have the enzymatic capacity to readily convert T ₄ to T ₃ ; this underpins the practice of internists to prefer T ₄ for treatment of hypothyroidism ( ) and weakens the rationale for psychiatrists’ preferential use of T ₃ . With this issue in mind, the results of a reanalysis of an RCT ( ) investigating the use of adjunctive T ₃ to accelerate the antidepressant effects of the SSRI sertraline are particularly interesting, albeit preliminary. The reanalysis focused on whether the benefit of adjunctive T ₃ was primarily experienced by patients who had the less functional polymorphism of the type 1 form of the enzyme deiodinase enzyme (DIO1-C785T); this subgroup might be expected to have low brain levels of T ₃ ( ). As hypothesized, they found a significant treatment x genotype interaction such that the entire advantage favoring adjunctive T ₃ over placebo appeared to be explained by subgroup of patients who had at least one copy of DIO1-C785T. If replicated, this finding could lead to a more personalized and efficient use of adjunctive T ₃ therapy for antidepressant nonresponders. In the meantime, evidence pertaining to safety and ease of implementation are reassuring and it is at the least clinically reasonable consider this strategy for patients have a history of thyroid dysfunction and/or who have borderline high or elevated TSH levels ( ; ).

Longer-term considerations . As there have been no longer term studies of adjunctive thyroid hormone in patients with difficult to treat MDD, decisions about whether or not continue treatment must be made on an individual basis, taking into account the magnitude of the benefit and side effects. For patients who have had a good response and have experienced minimal or no side effects, one consideration during continued therapy is to ensure that the thyroid axis is not chronically suppressed, which is a risk factor for development of osteoporosis. In practice, periodic monitoring of TSH values is sufficient to implement this recommendation, with that are below 0.5 iu ( ; ; ).

Psychostimulants

The psychoactive effects of amphetamine salts were first recognized in the 1930s, about 25 years before the discovery of the first antidepressants ( ). However, for most depressed patients the favorable effects that racemic amphetamine, its more potent stereoisomer and the several related drugs that followed had on mood, energy or motivation were often short-lived; even those who obtained more dramatic early responses sometimes developed tolerance within weeks of starting treatment ( ). By the early 1970s, it was clear that, whatever the benefits that psychostimulants might have for the treatment of depression, their effects were less reliable than those of the TCAs or MAOIs. Growing awareness of the potential of psychostimulants for misuse and abuse further marginalized their role in most clinicians’ treatment hierarchies ( ; ). As such, the psychostimulants were largely reserved for third- or fourth-line use, often as adjuncts to ongoing antidepressant therapy. Nevertheless, it has been consistently documented over the last 50 years that a small proportion of highly treatment-resistant depressed patients obtain palliative benefits from psychostimulants, typically as adjuncts to conventional antidepressants ( ; ).

In current practice, clinicians may pick among several psychostimulants, including d -amphetamine, immediate or sustained release formulations of mixed amphetamine salts, and lisdexamfetamine, and methylphenidate, which also is available in immediate and sustained release formulations. These drugs are all considered Schedule II controlled substances by the US Drug Enforcement Agency (DEA), which necessitates a fair amount of oversight by both the prescriber and the pharmacy. Two more recently introduced drugs, modafinil and its stereoisomer armodafinil, are typically grouped with the classical psychostimulants, although these drugs have a more benign side effect profile and less potential for misuse and abuse ( ); both of these alertness promoting drugs are classified as Schedule IV drugs by the DEA.

Metaanalyses have confirmed clinical impressions that psychostimulants have at least short-term antidepressant effects for patients with both MDD and bipolar depression ( ; ; ). However, evidence from well-controlled studies has not established adjunctive efficacy in patients with a documented history of nonresponse to standard antidepressants. This point is underscored by the failure of several well-funded clinical development programs for newer formulations of psychostimulants to obtain regulatory approval for adjunctive therapy in MDD. The discrepancy between metaanalytic evidence of an antidepressant effect and the failure of a number of potentially pivotal phase III trials could result from two possibilities: (1) psychostimulants have a beneficial effect on several depressive symptoms, but not a broad enough effect to show overall efficacy or (2) stimulants may have a meaningful effect for a small proportion of depressed patients, but have little or no benefit for most. The case series of Parker and colleagues ( ), which described a highly resistant group of patients treated at a tertiary care specialty program, suggests that the first explanation may be accurate—only about 20% of their patients were judged to obtain great benefit from open label therapy.

Longer-term considerations . The lack of regulatory approval for use in MDD, coupled and the potential for misuse, abuse and drug dependence, provide a solid argument to minimize the use of psychostimulants as adjunctive therapy for depressed patients. Of course, cotherapy is indicated for a subset of patients with difficult to treat depressive disorders who suffer from comorbid conditions such as attention deficit hyperactivity disorder, sleep apnea, narcolepsy, or binge-eating disorder. And, as illustrated by Parker and colleagues ( ), there is another subgroup of patients with more severe, treatment-resistant depressive disorders who do obtain sustained benefits with these medications. The principal longer term consideration in such cases is to ensure that there is not upward drift in dosing over time, which would reflect tolerance, and that the medication is being responsibly managed. In many parts of the United States, these goals are facilitated by state-sponsored data bases that register prescriptions for all controlled substances.

Nutraceuticals

The so-called nutraceuticals, which are standardized and purified pharmaceutical versions of common nutrients, have broad popularity in many societies, both as alternatives to conventional medications or as adjuncts to standard therapies. Systematic reviews and metaanalyses of treatment research on various nutraceuticals generally provide cautiously optimistic appraisals about the antidepressant effects of a select group of nutraceutical compounds, including S -adenosylmethionine (SAMe), folate and methylfolate, omega-3 fatty acids (primarily EPA or ethyl-EPA), vitamin D and tryptophan (including 5-HTP) ( ; ; ). Whereas evidence of antidepressant efficacy is generally modest, these reviews are largely reassuring about the safety and tolerability of nutraceuticals for depressed patients ( ; ; ).

With specific reference to adjunctive therapy of MDD following inadequate response to antidepressants, Sarris and colleagues ( ) identified 40 RCTs of adjunctive therapy with various nutraceuticals that met minimal standards for inclusion; the best studied strategies in their systematic review were folic acid (nine studies, including two studies of l -methylfolate), omega-3 fatty acids (eight studies), tryptophan (eight studies, including one study of 5-HTP) and SAMe (four studies). Individual metaanalyses of the adjunctive studies of tryptophan and SAMe were not possible: in the former case, the literature was older and did not conform to current standards of reporting and, in the latter case, the studies were too heterogeneous to permit a valid pooling of the data. Among the nutraceuticals that could be assessed by metaanalysis, the studies of adjunctive omega-3 revealed a significant, moderate effect as compared to placebo, although the validity of this finding was weakened by significant heterogeneity across studies. The metaanalysis of RCTs of folic acid likewise revealed significant heterogeneity, although the difference between drug and placebo across studies was not significant ( ).

Given the heterogeneity of antidepressant effects for adjunctive folic acid, it is noteworthy that l -methylfolate, not folate, is used in the so-called one carbon cycle for synthesis of monoamines. Conversion of dietary folic acid to l -methylfolate requires the enzyme methylenetetrahydrofolate reductase (MTHFR), which is under genomic control; about 40% of the population has at least one copy of the less functional polymorphism of this gene ( ). Thus, it is possible that an adequate intake of dietary folic acid or even the impact of adjunctive folic acid might not ensure adequate amounts of l -methylfolate in the brain. Two RCTs—reported in one paper ( )—evaluated adjunctive therapy with l -methylfolate (7.5–15 mg/day) in antidepressant nonresponders with MDD; both studies used a sequential multistage design. In the first trial, 148 outpatients with SSRI-resistant major depressive disorder were enrolled in a 60-day study divided into two 30-day periods. Patients were randomly assigned, in a 2:3:3 ratio, to receive l -methylfolate for 60 days (7.5 mg/day for 30 days followed by 15 mg/day for 30 days), placebo for 30 days followed by l -methylfolate (7.5 mg/day) for 30 days, or placebo for 60 days. SSRI dosages were kept constant throughout the study. In the second trial, with 75 patients, the design was identical to the first, except that the l -methylfolate dosage was 15 mg/day during both 30-day periods. In the first trial, no significant difference was observed in outcomes between the treatment groups. In the second trial, adjunctive l -methylfolate at 15 mg/day showed significantly greater efficacy compared with continued SSRI therapy plus placebo on both primary outcome measures (response rate and degree of change in depression symptom score) and two secondary outcome measures of symptom severity. The number needed to treat for response was approximately six in favor of adjunctive l -methylfolate at 15 mg/day. l -Methylfolate was well tolerated, with rates of adverse events no different from those reported with placebo. Results of secondary analyses of this dataset indicated that the benefit of adjunctive l -methylfolate therapy was largely delimited to those who were carriers of the less functional MTHFR genotype ( ); this effect was further moderated by obesity and a subset of biomarkers of inflammation ( ).

It is not clear what role nutraceuticals should play in the management of treatment-resistant depression. On the one hand, omega-3 fatty acids, SAMe, and methylfolate are safe and well-tolerated compounds; they also are generally not particularly expensive, even if purchased over the counter. On the other hand, none of the nutraceuticals have the kind of replicated efficacy in well-controlled studies of patients with carefully documented treatment-resistant depression that leads to high rankings in practice guidelines or systematic reviews (see, for example, or ). One potential area for study is a potential link between the putative mechanisms of action of several of the nutraceuticals and the role of inflammation in both chronicity of depression and nonresponse to antidepressants ( ).

Antiinflammatory strategies

There has been growing evidence over the past decade that the notorious heterogeneity of the depressive disorders may be partly parsed by the impact of inflammation or, more likely, the pathogenic impact of the mediators of inflammation ( ; ). Up to one half of depressed patients may show one or more of the so-called inflammatory biomarkers, and both cytokines and nonspecific biomarkers such as C-Reactive Protein (C-RP) have been associated with both chronicity and nonresponse to antidepressant medications ( ; ). Moreover, although the literature is far from conclusive, several recent metaanalyses of effects of a broad range of interventions, patient populations and durations of treatment suggest that medications that are known to suppress or attenuate inflammation have a modest effect on improving depressive symptoms ( ; ). For example, the metaanalysis of Köhler-Forsberg and colleagues ( ) identified eight placebo controlled RCTs of cytokine inhibitors in patients ( n = 3345) with depressive symptoms ( ). The participants in these studies generally suffered from illnesses such as psoriasis or rheumatoid arthritis and the drugs studied, which included adalimumab, ixekizumab, or methotrexate, were typically prescribed without concomitant antidepressant medication. They found that therapy with cytokine inhibitors resulted in a modest improvement of depressive symptoms (as compared to placebo). However, confidence in the overall effect was diminished by significant heterogeneity across studies. With respect to tolerability, these drugs are certainly not as innocuous as an Omega-3 fatty acid or l -methylfolate, but are considered to have a favorable benefit/risk ratio for treatment of significant medical illnesses.

The handfuls of RCTs that have examined the impact of antiinflammatory medications in depressed patients with documented nonresponse to antidepressant therapy have yielded mixed results ( ; ; ; ). For example, the single-site study of Raison and colleagues ( ), which evaluated the antidepressant effect of three infusions with the tumor necrosis factor (TNF) antagonist infliximab (5 mg/kg) across 12 weeks, found no effect overall, but did observe a significant improvement in depressive symptoms in the subset of patients with C-RP levels of 5 mg/L or higher. In this subgroup, which comprised slightly more than one-third of the sample, almost twice as many of the infliximab-treated patients responded as compared the placebo treated patients (62% [8/13] vs 33% [3/9]). Of note, baseline concentrations of TNF and its soluble receptors were significantly higher in infliximab-treated responders, who also showed significantly greater decreases in CRP levels at week 12 compared with placebo-treated responders.

A similar result was observed of placebo controlled RCT of the minocycline ( ), an antibiotic descendent of tetracycline thought to have immune suppressing and potentially neuroprotective effects. An earlier metaanalysis of three small placebo controlled RCTs ( n = 158) suggested that minocycline (100–400 mg/day) may have moderate to large antidepressant effects ( ). The study of Nettis and colleagues ( ) randomized 39 MDD patients who had not responded to at least one adequate course of antidepressant therapy in the current episode to 4 weeks of treatment with either minocycline (200 mg/day) or placebo; all patients had C-RP levels of at least 1 mg/L. At study endpoint, the groups did not differ significantly with respect to symptom scores or response rates. A post hoc stratification on the basis of CRP levels (< 3 mg/L versus ≥ 3 mg/L), revealed that minocycline was effective for patients with higher CRP levels (mean change in HAM-D-17 scores 12.00 points [sd = 6.45] compared with 3.50 points [sd = 4.34] among those treated with placebo. By contrast, adjunctive minocycline was of no benefit compared to placebo among those with lower CRP levels; both groups showed minimal improvement on the HAMD.

Antiinflammatory strategies also failed to show clear-cut efficacy in two well-designed placebo controlled RCTs of antidepressant-resistant bipolar depression ( ; ); these trials infliximab ( ) and celecoxib and minocycline, singly and in combination ( ). Although the primary test of efficacy failed, the study of infliximab infusion therapy (three infusions of 5 mg/kg across 12 weeks) did observe a significant treatment × time × subgroup interaction among the patients with a past history of abuse or trauma during childhood.

Antiinflammatory therapies cannot be recommended at this time for adjunctive treatment of MDD. There is enough evidence to suggest that this approach may be helpful for a subset of patients with laboratory indicators such as elevated CRP. It may be that further research on this general approach will identify important, pathophysiologically relevant treatment options for small subsets of antidepressant nonresponders.

Buspirone

Given the association between high levels of anxiety and antidepressant nonresponse and the relevance of the pre- and postsynaptic serotonin receptor sensitivity to response to SSRIs, the possibility that adjunctive therapy with buspirone may have beneficial effects for patients with difficult to treat depression received careful investigation. Buspirone—a high affinity agonist of 5-HT _1A receptors and a partial agonist of presynaptic 5-HT _1A receptors, which are inhibitory autoreceptors ( )—was approved by the FDA in 1986 for treatment of anxiety (more specifically in current context, generalized anxiety disorder). Unlike the benzodiazepines, buspirone has no abuse liability and is minimally sedating, though it also is not useful on an “as needed” basis and has little impact on sleep. Like first-generation antidepressants, buspirone therapy typically must be up-titrated and requires 4–6 weeks to exert therapeutic effects.

Following a number of clinical reports and case series, three RCTs of adjunctive buspirone therapy were completed in patients with difficult to treat depression ( ; ; ). In the first ( ), 117 MDD patients who had not responded to treatment with citalopram (40 mg/day) or paroxetine (30 mg/day) were randomized to 4 weeks of adjunctive therapy with either buspirone (titrated up to 60 mg/day) or placebo. At endpoint, the groups had comparable outcomes; there were no significant differences on measures of depressive symptoms.

The second study ( ) randomized 102 MDD patients who had not responded to a least 6 weeks of SSRI treatment ( N = 102) to an additional 6 weeks of treatment with either buspirone (titrated up to 30 mg/day) or placebo. The group that receive active buspirone showed significantly greater improvement at week 1, but by week 6 the treatment groups had roughly comparable outcomes and did not show statistically significant differences on any outcome measure. Results of a secondary analysis suggested that adjunctive buspirone therapy had a significant effect among a subset of patients with high pretreatment symptom severity scores.

The third trial ( ) was conducted as a component of the second stage of the large STAR*D project. A total of 565 MDD patients participated; all had not responded to an average of 11.9 weeks of citalopram therapy (mean final dose: 55 mg/day). The trial was open patients who were agreed to continue to take citalopram and accept randomization for an additional 12 weeks of treatment with either adjunctive buspirone (titrated up to 60 mg/day) or the addition of bupropion (titrated to up to 400 mg/day). At endpoint, the two groups did not meaningfully differ on the primary outcome: approximately 30% of the patients in each group had a final HAMD score of 7 or less. The group that received treatment with adjunctive buspirone did significantly worse on several secondary outcomes, including higher levels of self-reported depressive symptoms at study endpoint and a greater dropout rate due to side effects (21% vs 13%). A detailed post hoc analysis of these data by Bech and colleagues ( ) focused on a novel concept of benefit of adjunctive therapy, namely a “triangle” of desirable effects defined by reduction of depressive symptoms, fewer side effects and improved social functioning. According to Bech et al.’s reanalysis, adjunctive buspirone was significantly inferior to the combination of bupropion and citalopram.

In summary, the results of three RCTs indicate that buspirone is not a particularly effective adjunct to pharmacotherapy with SSRIs. Interestingly, another drug that modulates 5-HT _1A receptor sensitivity—pindolol—also failed to show adjunctive efficacy despite a clear rationale supported by a large body of preclinical research (see for metaanalytic review). Because buspirone is approved for treatment of anxiety and may have some value for reducing sexual side effects, clinicians continue to prescribe it. Nevertheless, it is not recommended for adjunctive therapy of antidepressant nonresponders.

Dexmecamylamine

Given decades of research and anecdotal evidence suggesting that modulation of nicotinic acetylcholine receptors may have antidepressant effects—perhaps through enhancement of dopaminergic neurotransmission ( )—the therapeutic potential of dexmecamylamine, a nicotinic channel modulator, was evaluated in preclinical studies and showed promising effects ( ). Subsequently, the results of a placebo controlled Phase II study of adjunctive therapy of MDD patients who were citalopram nonresponders were announced in a press release to be unequivocally positive ( https://www.nomura.com/resources/npv/pdfs/15-07-09-targacept.pdf ). In retrospect, it is noteworthy that the actual results of this Phase II study were never published and, even more ominously, it appears that they were never even posted on the US regulatory website ( https://www.clinicaltrials.gov/ct2/show/record/NCT00692445?term=TC-5214&cond=depression&draw=2&rank=1 ). However, the field only had to wait a couple of years for the disappointment engendered by the subsequent series of negative Phase III trials ( ; ; ). The findings of the metaanalysis of succinctly summarize this story: the odds ratio for the likelihood of benefit for adjunctive dexmecamylamine was 1.01, which is about as close to being identical to placebo as mathematically possible.

Selective norepinephrine reuptake inhibitors

By the end of the first decade of the 21st century it was widely believed that response to SSRIs could be enhanced by norepinephrine reuptake inhibitors (NRIs). This belief was directly supported by the work of Nelson and colleagues ( ) on combination therapy with desipramine and fluoxetine and indirectly supported by evidence that tertiary amine TCAs and SNRIs might have greater efficacy than SSRIs ( ; ), perhaps more specifically for patients with more severe depressive symptoms ( ; ). Given the potential cardiovascular side effects of desipramine and other secondary amine TCAs and the growing preference of clinicians for adjunctive strategies (as opposed to switching to an SNRI or tertiary amine TCA), there was interest in determining if one of the novel, more selective NRIs might prove useful for adjunctive therapy of SSRI nonresponders.

Among potential candidates, edivoxetine received a thorough evaluation for adjunctive efficacy of MDD across Phase II ( ) and Phase III ( ; ; ; ) programs. However, despite glimpses of benefit on particular depressive symptoms or aspects of social functioning ( ; ), efficacy as an adjunctive therapy on primary outcome measures could not be established. Perhaps most discouraging was the consistent documentation of side effects that are characteristic of NRIs, which indicates that the drug failed to exert antidepressant effects even though patients received adequate drug exposure ( ; ; ). This would suggest that there also is little reason to try the other FDA approved selective NRI, atomoxetine, for adjunctive therapy of SSRI nonresponders, unless of course the patient had comorbid ADHD. For patients who appear to warrant treatment with the combination of an SSRI and a norepinephrine inhibitor, at this moment, it would appear that switching to venlafaxine, duloxetine, or a tertiary amine TCA or combining therapy with nortriptyline or desipramine would be better justified by the literature.

Buprenorphine + samidorphan combination

Although some people experience as an emotionally painful state and opioids have known antidepressant activity, their use in major depressive disorder (MDD) has been severely limited by risks of misuse, abuse, and addiction ( ). Clinical experience with the μ and κ opioid partial agonist buprenorphine (BUP) suggested that beneficial effects of opioids on mood might be retained with a compound that had a lower potential for abuse liability ( ; ; ). The availability of samidorphan (SAM), a potent μ and κ opioid antagonist, created the opportunity to evaluate whether the putative antidepressant effects of BUP could be harnessed without any risks for misuse or abuse ( ).

The first RCT of adjunctive therapy with BUP + SAM (also known as ALKS 5461) was conducted in 141 MDD patients with a history of an inadequate response to one or two trials of antidepressant therapy in the current depressive episode ( ). This study used a two-stage sequential parallel comparison design; each phase consisted of 4 weeks of double blind treatment ( ). Participants were randomly assigned to receive adjunctive treatment with 2 mg/2 mg of BUP + SAM, 8 mg/8 mg of BUP + SAM or placebo/placebo. In this design, nonresponders to placebo in the first phase were rerandomized to the three options in the second stage. Compared with the placebo group, the 2 mg/2 mg dosage group showed a significant reduction in depressive symptoms; the magnitude of improvement in the 8 mg/8 mg dosage group was smaller and did not differ significantly from the placebo group. Both doses of BUP + SAM were judged to be well tolerated and, at the end of up to 8 weeks of therapy with BUP + SAM there was no evidence of opioid withdrawal.

Subsequent work focused on the 2 mg/2 mg dosage of ALKS 5461 and provided further evidence that this formulation had low abuse liability in both healthy volunteers ( ) and MDD ( ; ); no withdrawal symptoms were noted even after months of continuation phase therapy ( ). However, the adjunctive efficacy of ALKS 5461 (2 mg/2 mg) was not consistently established by a series of placebo controlled Phase III trials ( ; ). Two of these studies, known as FORWARD-4 and FORWARD-5 ( ), employed the same sequential parallel-comparison design as used in Phase II. The FORWARD-5 trial found a significant effect on the primary endpoint favoring the adjunctive BUP + SAM group, although the drug versus placebo difference was relatively small (1.9 points [ P = .026]). The results of the FORWARD-4 study were very similar, although in this case a relatively modest drug versus placebo difference did not meet conventional criteria for statistical significance (1.8 points [ P = .109]). In both studies, secondary analyses of other depression endpoints were supportive of efficacy. A pooled analysis of the two studies demonstrated an overall effect size (Hedges’ g = 0.22) at the end of treatment that also indicated that this intervention was efficacious, albeit with a modest effect. Common side effects of ALKS 5461 included nausea, constipation, dizziness, vomiting, somnolence, fatigue, and sedation.

Despite evidence of efficacy, the FDA reviewers of the New Drug Application for ALKS 5461 for adjunctive therapy of MDD did not recommend approval, citing the inconsistency of results in the Phase III program and, with respect to the positive studies, the relative small signal for benefit ( ). The FDA reviewer’s concerns were heightened by a number of unknowns about the safety of this medication related to its, mechanism of action, including the lack of experience with patients with cooccurring alcoholism or substance use disorders. The reviewers strongly encouraged additional research to better establish efficacy and provide additional information about safety in potentially at risk populations.

The FDA’s appraisal that ALKS 5461 had insufficient efficacy was supported by a recent metaanalysis of controlled studies of BUP, which included five of the RCTs from the ALKS 5461 program ( ). In the aftermath of the FDA decision, the manufacturer suspended further research on ALKS 5461 and turned its attention to other indications for samidorphan. Although the combination of BUP and SAM may never be commercially available, there remains an unmet need and, at this time, off label prescription of buprenorphine probably is the best option (see, for example, ). Until options with established efficacy are available, BUP at low doses is clearly the best studied opioid for use in the treatment of depressed patients and has a reasonable safety profile.