TCPR: You’ve been skeptical of antipsychotics in mood disorders. What led to that?

Dr. Kavirajan: It’s mainly their adverse effects. The thing I’m most fearful of is tardive dyskinesia (TD) since it can be irreversible. Typically, I do not even consider using an antipsychotic until I have tried at least one antidepressant from each class of the modern, easier-to-use ones that have emerged since the 1980s. If the person is still significantly depressed, I will consider augmentation, but lithium is usually first on my list. If that does not work, I will often consider a monoamine oxidase inhibitor (MAOI), transcranial magnetic stimulation, ECT, or the ketamines before an antipsychotic (Kavirajan H, N Engl J Med 2023;388(21):2012–2013).

TCPR: What is the risk of tardive dyskinesia? 

Dr. Kavirajan: It’s 3.9% per year with second-generation antipsychotics, which is a little less than the rate with first-generation drugs: 5.5%. But that difference starts to collapse in older patients, where the risk is 5% per year with both generations (Correll CU and Schenk EM, Curr Opin Psychiatry 2008;21(2):151–156). Older patients are more at risk, so this is particularly salient for me as geriatric psychiatry is my subspecialty. (Editor’s note: For more information, see the table “Risk Factors for Tardive Dyskinesia.”)

TCPR: Some depression guidelines consider antipsychotics to have favorable tolerability. 

Dr. Kavirajan: Some of these meds do look favorable in short-term trials, but their major risks, like TD and metabolic problems, are delayed. Even if those risks are rare, I give them more weight, as they are quite serious and debilitating. 

TCPR: One reason that antipsychotics rank so high in treatment guidelines is that the studies are large and numerous.

Dr. Kavirajan: Yes, but in my opinion the efficacy has been exaggerated. The main issue is the magnitude of the effect. Across the board, antipsychotics have only a small effect size, around 0.3, for augmenting antidepressants. That’s equal to three or four points of difference from placebo on the Montgomery-Åsberg Depression Rating Scale, which is like 6% of this 60-point scale (Wang J et al, Medicine (Baltimore) 2023;102(38):e34670). When a drug has a small benefit and serious risks, I’m cautious.

TCPR: That’s a small effect size. Are any of them more effective than the rest?

Dr. Kavirajan: Not really. Aripiprazole has a larger effect size in some meta-analyses, and it also has more studies supporting its efficacy than the others. Quetiapine also stands out as it is the only one with good evidence as monotherapy in major depressive disorder.

TCPR: Monotherapy? But it’s not indicated for that, right?

Dr. Kavirajan: No, it’s not. The manufacturers tried to get approval as a monotherapy for major depression, and they had the efficacy trials to support it. The number needed to treat was 7 for doses of 150–300 mg QHS as a first-line “antidepressant” (Maneeton N et al, BMC Psychiatry 2012;12:160). But the FDA had concerns about exposing 5% of the population to an antipsychotic, so they declined it as monotherapy. My concern with the FDA’s decision to approve quetiapine in cases of one failed antidepressant was that the decision would still permit mass exposure to antipsychotics. It only requires failure of one antidepressant trial, which captures about 70% of people with depression.

TCPR: Quetiapine does form a metabolite (norquetiapine) with properties that resemble a serotonin/norepinephrine reuptake inhibitor (SNRI), so it may still contain an antidepressant even as monotherapy.

Dr. Kavirajan: Yes, that’s true. Quetiapine also has good data as monotherapy in generalized anxiety disorder, and here the effective dose range was lower, 50–150 mg QHS (Maneeton N et al, Drug Des Devel Ther 2016;10:259–276). It came close to approval there, but again the FDA had concerns about exposing a large population to those risks.

TCPR: You mentioned aripiprazole has the best evidence of efficacy. What about others? 

Dr. Kavirajan: It pretty much goes downhill from there. Brexpiprazole had two pivotal studies that were used to justify the approval by the FDA, but only one was positive, and even then, there was no statistically significant difference between drug and placebo on remission. Cariprazine had a very small effect size (0.12) across five trials (Xie M et al, J Psychiatr Res 2024;172:71–80). Olanzapine-fluoxetine combination (OFC) looks less effective in some analyses, but that’s not a fair comparison. OFC is the only one that is approved in treatment-resistant depression (TRD), defined as lack of meaningful response to two adequate antidepressant trials.

TCPR: What about other risks with antipsychotics?

Dr. Kavirajan: Their metabolic risks are concerning. That risk is greater in children but is similar for adults and geriatric patients. Falls are another risk that is more of a problem in older adults, but this is also a risk with antidepressants in general, and two head-to-head studies suggest antipsychotics may have an advantage there, at least with aripiprazole. These were two trials that compared bupropion and aripiprazole as augmentation agents in TRD. One, the PROSPECT trial, was strictly geriatric, and the other, the VAST-D, was from the VA. In these studies, bupropion and aripiprazole had similar benefits, but there was a higher rate of falls with bupropion compared to aripiprazole (Ji M et al, PLoS One 2024;19(4):e0299020). That is consistent with early trials showing a fall risk with bupropion as well as the SSRIs, SNRIs, and trazodone.

TCPR: Antipsychotics were commonly used for depression in the 1970s. What can we learn from that history? 

Dr. Kavirajan: Yes, this actually goes back to 1961 when an MAOI-antipsychotic combo pill was released: tranylcypromine/trifluoperazine (Parstelin). More popular was amitriptyline-perphenazine (Triavil, Etrafon), which is still available. These had support from controlled trials where they were compared to placebo, benzodiazepine augmentation, or antidepressants. There was some effort to identify ideal subtypes for these combos, and patients with anxiety and agitation seemed more responsive. These combinations fell out of favor as we became aware of the risk of TD. 

TCPR: So 40 years ago psychiatry dropped antipsychotic-antidepressant combos, and now they are back. Why is that?

Dr. Kavirajan: Guidelines and meta-analyses tend to limit their view to high-quality, double-blind, placebo-controlled trials. Those are expensive to run, which means they are usually industry sponsored and—importantly—short term, so they will miss longer-term adverse effects like TD and metabolic syndrome. The adverse effects associated with long-term treatment are especially important in major depression, where maintenance treatment usually extends for several months if not years.

TCPR: None of the antipsychotics are approved for long-term prevention in major depression.

Dr. Kavirajan: Yes, but that is how they tend to get used. In terms of evidence, we don’t have much to guide us there. It’s reasonable to attempt to taper off after a stable recovery, like if the patient has been well for six to 12 months and doesn’t have a lot of stress going on or other factors that would increase their risk of relapse. But even then, the tapers are not always successful, and there is some evidence the risk of TD is greater for patients who come on and off antipsychotics than it is with continuous use.

TCPR: Do you see a lot of industry bias in the guidelines?

Dr. Kavirajan: I did notice a pattern there and published on that recently. Treatment guidelines that had better protocols to reduce industry influence did a better job of warning about adverse effects on antipsychotics. An earlier study had put forth criteria for treatment guidelines, and I looked at the 11 guidelines that met their criteria for quality. Of those 11, only four even mentioned TD, and only about two-thirds of them mentioned the need to monitor for metabolic adverse effects (Kavirajan H, Am J Psychiatry 2024;181(4):342–345).

TCPR: How do you talk to patients who request an antipsychotic after seeing an advertisement or hearing about a friend’s recovery on one?

Dr. Kavirajan: I tell them, “These are very dirty drugs. I do use them in selected patients if I think the risks outweigh the benefits, but what troubles me is their risk of weight gain, diabetes, and a potentially irreversible movement disorder called TD.” I will demonstrate what some of the oral movements in TD look like. Usually, when I educate patients about these risks, they no longer want to try antipsychotics.

TCPR: You mentioned lithium is high on your list for augmentation, but lithium is also risky for older patients.

Dr. Kavirajan: Yes, but we can test for kidney and thyroid function on lithium. With antipsychotics, we can monitor for TD, but by the time you see it, it may have reached a point where it’s irreversible. I’ve found lithium reasonably well tolerated in the elderly, but you have to aim for a lower serum level. As people age, more lithium passes into the brain, so you can get the same efficacy—and fewer side effects—with serum levels 20%–30% below the typical targets of 0.6–0.8 for depression. Lithium also lacks metabolic risks. In recent meta-analyses, there was no statistically detectable weight gain with lithium. That was surprising. I’d guess that some patients gain weight, but on the whole, it is a lot less than we’ve assumed (Gomes-da-Costa S et al, Neurosci Biobehav Rev 2022;134:104266).

TCPR: You also mentioned using MAOIs in TRD. Which one do you start with?

Dr. Kavirajan: I tend to use tranylcypromine (Parnate), which is a little more favorable in terms of side effects, particularly weight gain and fatigue. If the patient is very anxious, I will choose phenelzine (Nardil), which has better evidence in anxiety and GABAergic effects. One downside to MAOIs is that the titration is slow—you need to do that to avoid orthostatic hypotension. I check blood pressure in the office and have the patient check while at home (sitting and then standing after two to three minutes, recording blood pressure and heart rate). Depending on the risk for that, I will start low and raise every two weeks toward the target, typically 30–100 mg/day for tranylcypromine and 45–60 mg/day for phenelzine. 

TCPR: What about isocarboxazid and selegiline (Emsam)?

Dr. Kavirajan: Isocarboxazid is the least studied of the MAOIs. Emsam—unlike oral MAOIs—does not have studies in TRD. Estimates from studies of oral selegiline suggest that a high dose—such as 18+ mg/day—would be needed to achieve efficacy in TRD. This dose is not achievable with the patch and would require the same dietary restrictions as oral MAOIs anyway.

TCPR: What about dietary restrictions? 

Dr. Kavirajan: The impact on diet is actually quite small, but the person needs to be aware of them. And then, of course, there’s the drug-drug interactions.

TCPR: What else do you use for augmentation in older patients?

Dr. Kavirajan: I tend to use low-dose stimulants, usually methylphenidate but sometimes dextroamphetamine. The doses for both would be in the 10–20 mg/day range. Sometimes I use thyroid augmentation with T3. (Editor’s note: See our interview with Dr. Tammas Kelly, “Thyroid Augmentation in Bipolar Disorder,” in the April 2022 TCPR.)

TCPR: Dextroamphetamine (as Vyvanse) failed as antidepressant augmentation for adults in a large, industry-sponsored trial. Do the data look different in geriatrics?

Dr. Kavirajan: Yes, there was at least one good placebo-controlled augmentation study of citalopram with methylphenidate (mean dose 16 mg/day), and then there are smaller studies (Lavretsky H et al, Am J Psychiatry 2015;172(6):561–569). One reason it may work better in older adults is that the phenomenology of depression looks a little bit different in some patients. I tend to use stimulants in patients who seem very flat. They are lethargic, apathetic, and may even look a bit parkinsonian. Some of these patients may have “vascular depression,” which refers to a subgroup where depression may be related to subcortical vascular disease (Taylor WD et al, Am J Psychiatry 2018;175(12):1169–1175).

TCPR: Thank you for your time, Dr. Kavirajan.