Pharmacotherapies for Treatment-Resistant Depression: How Antipsychotics Fit in the Rapidly Evolving Therapeutic Landscape

A recent systematic review found that no consensus definition for TRD exists (13). In the context of MDD, most studies define the presence of TRD as nonresponse to either one or two antidepressant medication treatments, delivered at an adequate dosage and duration. However, these studies vary in the definition of nonresponse, adequate dosage and duration of prior treatment, and whether these treatments were in the current depressive episode (13). Even when similar criteria are used for dosage and duration of antidepressant treatment in the current depressive episode, studies have differed in the definition of nonresponse (≤25% improvement for phase 3 esketamine studies [14, 15] vs. <50% improvement for an NIMH-funded study of ketamine for TRD [16]). Additionally, definitions for nonresponse to treatment do not incorporate the substantial burden of commonly co-occurring symptoms, such as anxiety (17), irritability (18), and impairments in day-to-day functioning (19) that patients with MDD often experience. The focus on acute-phase treatment outcomes to define TRD has overlooked the chronic and recurrent nature of MDD and has led to calls for shifting the terminology toward “difficult-to-treat” depression (20), which may help incorporate response persistence, burden of side effects, and impact of therapies on daily functioning and quality of life (21). In the absence of a consensus definition for TRD (13), we have taken the broader approach of considering inadequate response to one course of antidepressant medication as meeting an operationalization definition of TRD. This more liberal definition is relevant to evaluating the evidence for atypical antipsychotics, given that studies of aripiprazole and brexpiprazole have included among inclusion criteria a history of nonresponse to at least one antidepressant and inadequate response to a prospective trial with another antidepressant during the current depressive episode, whereas studies of extended-release quetiapine and cariprazine required a history of inadequate response to a minimum of one antidepressant in the current episode.

Confirming the presence of TRD can be facilitated by systematic assessments of symptom severity (such as the 9-item Patient Health Questionnaire [22]), side effects, and adherence using a measurement-based care approach (23). This approach allows for prospective assessments of treatment response that incorporate changes in antidepressant medication dosages as well as duration of treatment. In the absence of structured symptom assessments, tools such as the Massachusetts General Hospital Antidepressant Treatment Response Questionnaire can be used to establish the presence of TRD (24). Because unrecognized bipolar depression or presence of subthreshold symptoms may account for reduced likelihood of response to commonly used antidepressants (25, 26), a careful diagnostic assessment is needed to establish the presence of MDD and potential comorbid conditions. Additionally, medical conditions such as obstructive sleep apnea that are frequently comorbid in individuals with TRD and can exacerbate symptoms should be systematically assessed (11, 27). Similarly, systematic assessment of pain in individuals with MDD (28) may be warranted given the comorbidity of chronic pain conditions with MDD and their association with treatment nonresponse (29, 30).

Aripiprazole (31), brexpiprazole (32), extended-release quetiapine (33), and most recently cariprazine (34) have received approval from the FDA as augmentation agents after inadequate response to an antidepressant, and olanzapine is approved in combination with fluoxetine (35) for the TRD indication specifically per the prescribing label (i.e., patients who did not respond to two different antidepressants after at least 6 weeks at or above the minimally effective labeled dosage in their current episode) (35).

Cariprazine is a partial agonist of both dopamine D₂ and D₃ receptors, with preferential binding to D₃ over D₂, especially at lower dosages (60). An initial phase 2 study demonstrated no significant difference from placebo for cariprazine augmentation at low dosages (0.1–0.3 mg or 1–2 mg) (61). A subsequent RCT enrolled patients with MDD who had not responded to one or two antidepressant treatments and randomized them to double-blind augmentation with placebo, cariprazine at 1–2 mg/day, or cariprazine at 2–4.5 mg/day (62). The titration protocols differed between the two cariprazine dosages, with the target dosage of 1 mg/day achieved by week 1 in the 1–2 mg/day group and the target of 2 mg/day achieved by week 1 in the 2–4.5 mg/day group (62). Compared with placebo, there was significantly greater reduction in depression severity with both cariprazine dosages by week 2, but it was maintained until week 8 only in the 2–4.5 mg/day arm (62). Rates of remission were similar in the three arms: 29.9% with placebo, 31.9% with cariprazine at 1–2 mg/day, and 32.1% with cariprazine at 2–4.5 mg/day (62). A subsequent phase 3 study that enrolled patients with MDD in an 8-week prospective treatment with open-label antidepressant treatment and then randomized those with inadequate response to 8 weeks of augmentation with cariprazine (1.5–4.5 mg/day) or placebo did not find any significant improvement in depressive symptoms with cariprazine (63).

This issue of the Journal includes a report on a phase 3 trial of cariprazine augmentation in MDD (64). The trial enrolled 757 patients with MDD with inadequate response (<50% improvement in depressive symptoms) to one to three antidepressants in the current episode (of adequate dosage per prescribing label and of at least 6 weeks’ duration) and randomized them in a 1:1:1 ratio to 6 weeks of augmentation with placebo, cariprazine at 1.5 mg/day, or cariprazine at 3 mg/day. This study used fixed-dose arms, whereas previous studies had allowed for flexible dosing within specified dosage ranges. The study found significantly greater improvement in depressive symptoms with cariprazine at 1.5 mg/day (−14.1 points in Montgomery-Åsberg Depression Rating Scale [MADRS] score) compared with placebo (−11.5 points in MADRS score) but not with 3 mg/day (−13.1 points in MADRS score). Remission rates at week 6 did not differ significantly among the treatment arms (placebo, 23.3%; cariprazine 1.5 mg/day, 25.2%; cariprazine 3 mg/day, 16.7%). Over 85% of patients enrolled in the study had only one prior antidepressant failure, which is a limitation for the generalizability of these findings to the broader group of patients with TRD.

The average weight gain associated with cariprazine in the study was 0.68 kg (SD=2.4) with the 1.5 mg/day dosage and 0.78 kg (SD=2.8) with the 3 mg/day dosage, compared with 0.11 kg (SD=2.8) with placebo. The proportion of those with ≥7% weight gain was 0.8% in the placebo group and 4.0% and 1.2%, respectively, in the groups receiving 1.5 mg/day and 3 mg/day cariprazine. The two cariprazine dosages were similar to placebo in rates of clinically significant treatment-emergent changes in cholesterol, glucose, and triglyceride levels (64). Akathisia was the most common extrapyramidal symptom among treatment-emergent adverse effects, with rates of 0.8%, 5.2%, and 7.9%, respectively, in the placebo, cariprazine 1.5 mg/day, and cariprazine 3 mg/day groups. Notably, discontinuation due to an adverse effect was more common with cariprazine at 3 mg/day (7.1%) than with placebo (2.4%) and cariprazine at 1.5 mg/day (1.2%).

Cariprazine is similar to aripiprazole and brexpiprazole in its partial agonist effect on the dopamine D₂ receptor, signifying that in the absence of full agonist effects, it has functional agonist activity (albeit at lower levels), while in the presence of a full agonist, it serves as an antagonist (65). Cariprazine has a 10-fold greater affinity for D₃ than D₂ receptors in in vitro studies (66). Greater selectivity of cariprazine for D₃ than D₂ receptors has also been demonstrated in a human positron emission tomography study; the mean D₃ and D₂ receptor occupancies with a low dosage of cariprazine (1 mg/day) were 76% and 45%, respectively (60). With higher dosages of cariprazine (12 mg/day), near 100% occupancies of both receptors were observed (60). Cariprazine and its two major active metabolites, desmethyl cariprazine and didesmethyl cariprazine, have long half-lives, and the steady state of these metabolites is attained in 2–4 weeks (67, 68).

With the recent (December 2022) approval of cariprazine as adjunctive treatment for MDD (34), head-to-head trials are needed to understand the effectiveness and tolerability of other antipsychotics with a similar indication (aripiprazole, brexpiprazole, extended-release quetiapine, and OFC). The acute-phase studies suggest lower weight gain potential with cariprazine compared with extended-release quetiapine and OFC, and similar to or lower than with aripiprazole and brexpiprazole. Rates of akathisia with cariprazine are also similar to or lower than those with aripiprazole and brexpiprazole.

Unlike atypical antipsychotics, which were developed initially for other indications, clinical trials of the repurposed anesthetic agent ketamine and its S-enantiomer esketamine started with a focus on their rapid antidepressant effects. In fact, TRD specifically was the first indication pursued for the esketamine nasal spray. A panel of international experts synthesized the available evidence for ketamine and esketamine in a recent report in the Journal (69), and we refer readers additionally to recent meta-analyses of ketamine and esketamine effects (70–73). These medications represent a paradigm shift in the management of TRD. Esketamine nasal spray is the first non-monoaminergic antidepressant that has been approved by the FDA, and it has rapid antidepressant effects (69). In parallel with the FDA approval of esketamine nasal spray in 2019, there has been a rapid increase in clinics offering ketamine (primarily intravenously) in an off-label fashion (74).

Several key issues related to use of ketamine and esketamine remain unanswered. One issue is the optimal duration of acute-phase treatment. Early studies of intravenous ketamine deemed individuals who did not respond to the first infusion as nonresponders (75, 76). However, subsequent studies have shown that intermittent, repeated doses increase response rates (77, 78). In the phase 3 esketamine program, clinical benefit continued to accrue beyond the first 4 weeks of treatment (when administered twice weekly), and a recent report of ketamine in community practices found increasing rates of response and remission with continued infusions over a 6-week period (79, 80). Longer-term treatment options after initial improvement with ketamine or esketamine remain a challenge. While continuation-phase administration of esketamine reduces the likelihood of relapse (14), other pharmacological treatments have proven to be of little utility in prolonging the transient improvement with ketamine or esketamine.

A study of individuals who had at least a partial response to a single infusion of ketamine and received three additional ketamine infusions along with 4 weeks of double-blind treatment with either lithium or placebo did not find a significant difference between the two treatment arms in maintenance of initial response (81). Studies of drugs targeting glutamate neurotransmission have proven to be of limited benefit in prolonging the antidepressant effect of ketamine. A study of individuals with TRD who responded to ketamine and were randomized to 6 weeks of double-blind treatment with either d-cycloserine, a partial agonist of the glycine co-agonist of the N-methyl-d-aspartate (NMDA) glutamate receptor and NMDA receptor antagonist (at higher dosages), or placebo did not find any significant difference between the two groups in overall depression severity (82). Two placebo-controlled studies of riluzole, a glutamate-modulating agent with neuroprotective and synaptic-plasticity-enhancing effects, for relapse prevention after improvement with ketamine were negative (83, 84). Based on preclinical studies that suggested a key role of mechanistic target of rapamycin complex 1 (mTORC1) in the antidepressant-like effects of ketamine, a recent study used a single-dose pretreatment with rapamycin, an inhibitor of mTORC1, to block the antidepressant effect of ketamine infusion. However, contrary to their hypotheses, the authors found that rapamycin unexpectedly prolonged the antidepressant effect of ketamine (85).

Apart from pharmacological approaches, various therapeutic interventions have been investigated with regard to their potential for preventing early relapse following treatment with ketamine. In an initial open-label study, Wilkinson et al. (86) demonstrated the feasibility of adding a 10-week course of CBT to prolong the antidepressant effect of ketamine (four infusions over 2 weeks). A subsequent RCT of acute-phase responders to ketamine (six infusions over 3 weeks) demonstrated greater sustained improvement with 14 additional weeks of CBT compared with treatment as usual (87). In a recent study, a type of self-training paradigm that promotes positive self-associations, “automated self-association training” (ASAT), was paired with either a single ketamine infusion or a placebo infusion (normal saline) (88). The study found that ketamine in conjunction with ASAT kept depression measures significantly and stably low 30 days after a single infusion, while ketamine paired with sham treatment culminated in depression scores increasing linearly to approach those of the placebo plus ASAT group. Further longer-term and multicenter studies are needed to investigate the potential role of digital therapeutics and psychotherapies in enhancing or maintaining response to ketamine and esketamine.

Recent approvals by the FDA of dextromethorphan-bupropion combination (where inhibition of CYP2D6 by bupropion increases the bioavailability of dextromethorphan) for treatment of MDD (89) and brexanolone for postpartum depression (90) represent examples of drugs with non-monoaminergic mechanisms that possess antidepressant properties. Phase 2 clinical trials suggest efficacy of inhaled nitrous oxide (91) and MIJ821, a novel NMDA receptor subtype antagonist (92), for TRD. Esmethadone (an NMDA receptor antagonist) and seltorexant (an orexin-2 receptor antagonist) are two examples of novel-mechanism compounds that demonstrated superiority over placebo in phase 2 studies as augmentation agents in individuals with MDD who experienced inadequate response to one to three antidepressants of adequate dosage and duration in their current episode (93, 94). Drugs targeting kappa opioid receptors represent another potential novel mechanism as adjunctive treatment after inadequate response to antidepressant medications (95, 96). Pimavanserin, an antipsychotic medication that is approved by the FDA for Parkinson’s disease psychosis, demonstrated initial efficacy compared with placebo as an augmentation agent in patients with MDD who had inadequate response to one or two antidepressants of adequate dosage and duration in the current episode (97). However, two subsequent phase 3 trials did not find any significant difference between augmentation with pimavanserin and placebo (98). While an initial study suggested targeting inflammation as a novel treatment for patients with TRD (99), subsequent trials of drugs targeting inflammation, including minocycline (100) and sirukumab (NCT02473289), did not demonstrate any significant difference compared with placebo. Despite antidepressant effects of ketamine, esketamine, and dextromethorphan-bupropion combination, there is more limited evidence for antidepressant effects of other modulators for glutamate receptors (70). According to sponsor reports, two phase 3 trials of esmethadone (as monotherapy and as an augmentation agent) did not attain their primary outcome of significant reduction in depression symptoms compared with placebo (101, 102). Lanicemine, a parenterally administered low-trapping NMDA channel blocker, did not significantly improve depression compared with placebo in a large study (N=302) of individuals with MDD with a history of inadequate response to three or more antidepressants (including one antidepressant in the current depressive episode) (103). Rapastinel, another parenterally administered NMDA receptor modulator with glycine-site partial agonist properties (104), was evaluated in a large phase 3 program as adjunctive treatment and did not differ significantly from placebo (acute-phase studies: NCT02932943, N=457; NCT02943564, N=658; and NCT02943577, N=429; relapse prevention study: NCT02951988, N=604).

Psychedelic drugs that act as agonists of 5-HT_2A receptors have emerged as novel treatment options for patients with TRD. In an initial open-label study of patients with TRD (N=20) who received two sessions of psilocybin (10 mg and 25 mg, 7 days apart, plus psychological support), there was significant improvement in depressive symptoms that persisted for 6 months (105). A randomized waiting-list-controlled trial of two sessions of psilocybin (20 mg/70 kg and 30 mg/70 kg, 7 days apart, plus supportive psychotherapy) also found significant improvement in individuals with MDD (106). A subsequent double-blind RCT (N=59) compared psilocybin at 25 mg (two sessions administered 3 weeks apart) plus daily placebo pills for 6 weeks versus psilocybin at 1 mg (two sessions administered 3 weeks apart) plus escitalopram (10 mg/day for 3 weeks followed by 20 mg/day for the next 3 weeks) in patients with TRD. The two treatment arms did not differ significantly in the primary outcome of reduction in self-reported depressive symptoms (107). However, rates of remission were higher with psilocybin at 25 mg (57%) than with psilocybin at 1 mg plus escitalopram (28%), with an NNT of 4 (107). A recent phase 2 double-blind RCT compared a single administration of psilocybin at a dose of 25 mg, 10 mg, or 1 mg (control), along with psychological support in patients with TRD (N=233). Three weeks after the treatment, there was significantly greater improvement in depressive symptoms and higher rates of remission with 25 mg (29%), but not with 10 mg (9%), compared with 1 mg (8%), with an NNT of 5 for psilocybin 25 mg compared with 1 mg (108). Finally, a small RCT of ayahuasca (N=29), a psychedelic mixture that contains N,N-dimethyltryptamine and the monoamine oxidase inhibitors harmine and harmaline, in patients with TRD found significantly greater improvement in depressive symptoms compared with placebo, with an NNT of 4 for remission (36% vs. 7%) (109).

Current practice guidelines recommend the use of atypical antipsychotics after inadequate response to initial pharmacotherapy with a first-line antidepressant treatment (110, 111). Besides the FDA-approved atypical antipsychotics discussed above, lumateperone is currently in phase 3 trials for MDD (NCT05061706 and NCT04985942), lurasidone has been shown to be efficacious in MDD with mixed features (112), and there are positive RCTs for ziprasidone and risperidone as adjunctive treatment after inadequate response to antidepressant treatment in patients with MDD (8, 113). As there is no clear evidence for superior efficacy of one atypical antipsychotic over another for treatment of MDD, shared decision-making approaches that incorporate information regarding tolerability, anticipated side effects, and costs may be helpful in selecting the specific atypical antipsychotic to use as an augmentation agent. Because of concerns about weight gain and metabolic dysfunction, use of extended-release quetiapine and OFC should generally be avoided. Aripiprazole appears to have rates of weight gain and metabolic dysfunction similar to those of cariprazine but has higher rates of akathisia, which may often be a dose-limiting side effect. Finding the optimal dosage of these medications can be a challenge. While rates of side effects increase with higher dosages of atypical antipsychotics, there is no clear evidence for improved efficacy in symptom reduction with higher dosages. When atypical antipsychotics are effective, patients and clinicians face substantial uncertainty regarding the advisability of both the long-term use of these medications in the absence of continuation-phase data (while incurring the risk of weight gain and tardive dyskinesia) and of discontinuing the medication, with the potential risk of relapse or recurrence. Additional practical challenges with atypical antipsychotics include monitoring of metabolic dysfunction and extrapyramidal symptoms while managing patients with virtual visits (telehealth).

Potential alternatives to atypical antipsychotics in clinical practice, where their use is often preceded by at least two prior trials of first-line antidepressants (9), include TRD-specific treatments such as intranasal esketamine (NCT05554627, a Veterans Affairs study comparing esketamine and aripiprazole) and TMS (NCT02977299, a study comparing switch to another antidepressant and augmentation with aripiprazole or TMS). Minimal or no separation from placebo in attainment of remission with the FDA-approved atypical antipsychotics raises the question of whether these statistically significant findings in large phase 3 trials are clinically robust and meaningful and whether the incremental improvements with the medications are worth the risks. Furthermore, the strength of evidence is limited in individuals who have not sufficiently improved with two or more adequate trials of antidepressants, given that a substantial majority of patients in phase 3 adjunctive trials of atypical antipsychotics had experienced only one prior failed trial. Therefore, large-scale multistage pragmatic clinical trials are needed to understand the real-world comparative effectiveness of medications for TRD and how to sequence them to optimize the care of individual patients (114).

In this era of newly FDA-approved antidepressant medications (e.g., esketamine nasal spray and dextromethorphan-bupropion combination) that are in widespread clinical use (off-label ketamine) or are currently in late-stage clinical development (e.g., zuranolone and psilocybin) and have rapid onset (days to weeks) of antidepressant effects, it remains to be seen whether patient and clinician expectations have fundamentally transformed, such that the use of drugs with traditionally slower onset of action (weeks to months) will decrease. In this regard, the prescribing patterns for atypical antipsychotics may go the way of older classes of antidepressants (e.g., tricyclics and monoamine oxidase inhibitors), where therapies with fewer long-term side effects and more rapid onset of effect may be preferred. Alternatively, precision medicine approaches that allow for the rational selection of an atypical antipsychotic for TRD, based on biological markers and/or clinical features, may help identify the subgroup of patients for whom atypical antipsychotics would be a preferred option. Ongoing development of novel non-dopamine D₂ receptor antipsychotics and promising early-phase trials for drugs targeting trace amine-associated receptor 1 (115) and muscarinic cholinergic receptors (116, 117) may herald a new era for the drug treatment of schizophrenia. However, studies will be needed to evaluate how these safe and generally well-tolerated antipsychotic medications may be effectively repurposed for individuals with TRD.

In summary, four atypical antipsychotics (aripiprazole, brexpiprazole, extended-release quetiapine, and most recently cariprazine) are approved by the FDA as acute-phase augmentation agents after inadequate response to an initial antidepressant. However, their long-term safety in patients with MDD is not well established, and they are potentially concerning regarding weight gain, metabolic dysfunction, extrapyramidal symptoms, and tardive dyskinesia. With the rapidly evolving landscape of therapies for TRD, studies are needed to compare the efficacy and safety of atypical antipsychotic augmentation strategies with other evidence-based and emerging treatments and to determine how to sequence these approaches for the optimal treatment of patients with MDD.