Clear, engaging, and practical updates on clinical psychiatry.

Earn CME for listening to the podcast with a multimedia subscription. Listen for free here or using Apple Podcasts, Android, or Stitcher.

Previous Post
Next Post

What Works Best in Treatment Resistant Depression Part 2

Podcast, Volume , Number ,

Print Friendly, PDF & Email

A new study looked at what works best after two antidepressants have failed, and it claims to have found some surprising winners. We take a closer look at the data in the final episode of our two-part series on treatment resistant depression (see March 8 2021 for part 1).

Published on: 3/3/21

Duration: 19 minutes 49 seconds


A new study looked at what works best after two antidepressants have failed, and it claims to have found some surprising winners. We take a closer look at the data in the final episode of our two-part series on treatment resistant depression (see March 8 2021 for part 1). What do you do after a single antidepressant fails? We covered that question in March 8ths podcast, and in this episode we’ll look at what works best after two antidepressant failures.

Dr. Aiken: Welcome to the Carlat Psychiatry Podcast, keeping psychiatry honest since 2003. I’m Chris Aiken, the editor in chief of The Carlat Psychiatry Report.

Kellie Newsome: And I’m Kellie Newsome, a psychiatric NP and a dedicated reader of every issue.

Dr. Aiken: It used to be one of the great unsolved questions in psychiatry: When an antidepressant fails to work, should you switch to a different antidepressant or augment? The old answer was that nobody knew, but that has changed in the past decade as pharmaceutical companies have realized their antidepressants only bring about a full recovery in 1 out of 3 patients, leaving a huge market to augment their modest drugs with newer medications. 

The result has been an explosion of research on antidepressant augmentation, and those studies have tipped the scales in favor of augmentation and away from antidepressant switch strategies. Now, maybe there’s some industry influence going on there, but let’s not jump to that conclusion so quickly. Last March we looked at the augmentation strategies that work best after one failure of an antidepressant, and two of those options have long been generic. To recap, they were: Lithium, thyroid, and the atypical antipsychotics, particularly aripiprazole, quetiapine, risperidone, olanzapine, brexpiprazole (Rexulti), cariprazine (Vraylar), and lurasidone (Latuda), and possibly buspirone. And remember treatment failure doesn’t mean there was no response – it just means the medication didn’t finish the job – it only worked part of the way or didn’t work at all. 

And those are just the medication options. Psychotherapy also works at this stage, as does ECT, TMS, aerobic exercise 30 minutes a day and a few complimentary treatments, like augmentation with lightbox, aerobic exercise, methylfolate 15mg a day, or SAMe 800-1600 mg a day. These don’t get used as much as they should because psychiatric education has become synonymous with psychotropic education – so we have a detailed guide to lighttherapy and methylfolate on our website, and we ought to do one on SAMe too. If SAMe was copyrighted it would be classified as an antidepressant – it enhances dopamine, serotonin and norepinephrine transmission much like the fabled triple reuptake inhibitor that companies have been trying to develop for years. And it can cause mania in bipolar disorder.

So that’s what we know works after one failed antidepressant trial. But one failed trial is not really treatment resistance; you need two failures to reach that level. So what works best when 2 antidepressant trials have failed? ECT and TMS definitely fit in here, and today we’ll report on a new study that ranks all the other strategies for this full level of treatment resistant depression.

Kellie Newsome: A group of British and Canadian researchers lead by Allan Young published a new analysis of therapies for treatment resistant depression (they appear as two articles: one and two). What makes their analysis unique is that they focused only on patients with true treatment resistant depression – at least two failed antidepressant trials. They did not look at TMS or ECT, but they did include ketamine and psychotherapy. All in all they gathered 27 medication trials and 3 psychotherapy trials, and analyzed the data two different ways in two papers. First they looked at how each treatment changed depressive symptoms – comparing the pre- and post-treatment depression for every treatment including placebo. The second paper was a network metaanalysis that looked at how each treatment fared against a common placebo arm. 

Dr. Aiken: Here’s what they found. First, the atypical antipsychotics worked – specifically aripiprazole (Abilify) (5 studies), brexpiprazole (Rexulti) (5 studies), cariprazine (Vraylar) (1 study), olanzapine (as Symbyax in combination with fluoxetine) (1 study), and quetiapine (Seroquel) (2 studies). It’s hard to tell about lithium because they lumped it in with lamotrigine, and these – as a group – worked in one of the analyses but not the other.  

Kellie Newsome: OK so that’s not too surprising – so far we see the same line-up for 2 antidepressant failures as we saw for 1 antidepressant failure. 

Dr. Aiken: Where things get interesting is what they found with a group of meds they classified as glutamatergic. These are also called N-methyl-D-aspartate (NMDA) antagonists, and they included ketamine and two repurposed antibiotics, minocycline and d-cycloserine. Note that the FDA-approved esketamine (Spravato) was not included, just ketamine.

In both analyses, these meds had a much bigger effect than the others – the effect size was large (0.9), compared to the small effect (0.2) of the atypical antipsychotics. These meds were not included in the earlier analysis by Dr. Zhou’s group – the one from 2015 that looked at what works after 1 failed antidepressant – mainly because the studies were not out then. 

Kellie Newsome: It’s rare to see any treatment in psychiatry stand out as more effective than the others. We can say that about clozapine in schizophrenia, lithium in some forms of bipolar and suicidality, and ECT in depression. In Dr. Zhou’s analysis none of the meds really stood out as more potent than the others, so this new finding about the glutamatergic meds really stands out. But can we trust it? 

Dr. Aiken: No, these network meta-analyses give us a signal; but they aren’t the truth. We’d need to see head-to-head trials of these strategies before we hang our hat on them. But if what Dr. Young is saying holds up, we may need to rethink what we are doing, because the most effective treatments – ketamine, minocycline, and d-cycloserine – are rarely ever used. While we await more research to clarify the role of these NMDA antagonists, here’s what we know so far.

  1. The big effect of this class was not simply due to ketamine. Minocycline and d-cycloserine also had large effects on their own, though there were fewer studies supporting them. Dr. Young’ analysis was based on 3 ketamine trials, 1 minocycline trial, and 1 d-cycloserine trial.
  2. Ketamine is well known in psychiatry and no doubt has a large and rapid effect in depression. The only problem is that most of its studies are short term, often after 1 dose or 1 week of treatment, and its isomeric cousin esketamine (Spravato) may fade out after its initial antidepressant burst. In a meta-analysis of its clinical trials, esketamine did not work any better than placebo after a month of use.

Kellie Newsome: We can’t assume that the robust effects of ketamine seen here will translate to esketamine. When people say “Ketamine” they mean the racemic mixture of ketamine, which is made of 50% esketamine and 50% r-ketamine. Even though esketamine is FDA approved, there are animal studies suggesting the R- isomer is the more potent of the two, and there are studies of this R-ketamine underway that may earn it FDA approval. 

Dr. Aiken: Another limitation of Dr. Young’s study is that we don’t really know the mechanism of these drugs, so it’s a bit arbitrary to lump them together as NMDA antagonists. Let’s turn now to two of them that are new to the psychiatric scene: minocycline and D-cycloserine. 


Dr. Aiken: Minocycline does have glutamatergic effects, but it also has antiinflammatory effects and neuroprotective benefits that may explain its antidepressant action. There’s a new randomized controlled trial of minocycline where it worked best in patients with high levels of inflammation – which you can measure by testing CRP – see our January 2020 issue for more on that. The dose was 200 mg/day, and though this new study uses the term “treatment-resistant

Depression” in the title, we should point out they allowed in patients who had failed only one antidepressant trial…. It seems there’s a lot of papers that are loosening the meaning of this term. Dr. Young counted only one trial of minocycline in true treatment resistant depression, but there are about 5 randomized controlled trials of this antibiotic in depression, and taken together it does have a large effect size – around 0.7. It has a few studies in bipolar depression, but some of those are negative. The main drawback to minocycline is that most of these are small studies. There’s a larger trial underway, so we’ll wait for those results before jumping in on this promising therapy. For now, the data is good enough that I have changed one aspect of my practice. When patients with depression develop acne, such as can happen on lithium, I recommend minocycline first-line to cover both bases, whereas in the past they might have started similar antibiotics for acne like doxycycline or tetracycline.


Kellie Newsome: D-cycloserine is an old antibiotic that was used for tuberculosis in the 1950’s. You may have heard about it in psychiatry because it’s being used to enhance the effects of exposure therapy. Patients take a low dose (100-200 mg) right after an exposure exercise – such as for phobias, OCD, or PTSD – and some but not all studies suggest it makes the therapy work better and faster. The thinking is that D-cycloserine enhances learning. But it’s not until you reach the higher dose – 1,000 mg daily – that D-cycloserine starts to affect the NMDA receptor that gives it antidepressant potential. There is only one controlled trial that we know of for this, and indeed it was conducted in true treatment resistant depression and it did have a large effect size, but it was a very small study – 26 patients – so you can’t trust that effect size any more than you can predict the winner of a presidential election by looking at the first 3 states that tally their votes.

Dr. Aiken: Well put. There is also a negative study of d-cycloserine but it used the low dose, and there are a few trials where they tried to use d-cycloserine as a maintenance therapy after ketamine, but those results were inconclusive, so…. We’d conclude that d-cycloserine is promising but not at all conclusive – if you had to choose between the two minocycline has the better data.

Kellie Newsome: And another thing – Dr. Young’ analysis assumed that all of these had glutamatergic effects, and they do, but it’s not clear that they treat depression through the glutamatergic pathway. Minocycline as you said has antiinflammatory and neuroprotective effects; d-cycloserine enhances learning, and some evidence suggests that ketamine works through the opioid system – not the glutamate system. 

Dr. Aiken: So perhaps Dr. Young raised more questions than answers here, but he does show us that this glutamatergic group is worth keeping an eye on.

Kellie Newsome: And that antipsychotic augmentation is not as effective as we’d like. Sure, they are FDA approved and have a lot of studies behind them, but they their effect size is on the small size. And that gives me pause, particular considering all the risks they carry. Our March issue adds one more serious risk to that list – in a research update on mortality risks with antipsychotic augmentation in depression.

Dr. Aiken: And before we end, let’s not forget about psychotherapy.

Kellie Newsome: No, Dr. Young did not.

Dr. Aiken: That’s right, his team included 3 psychotherapy studies in their analysis – and they worked. Not only that, psychotherapy had a slightly larger effect than the overall effect of pharmacotherapy in treatment resistant depression. Although, we should point out that effect size is based on difference from placebo – and it’s not fair to compare studies that use different kinds of placebo – I mean, in therapy studies the placebo is often treatment as usual, while in medication trials it’s a sugar pill.

Kellie Newsome: Still, just the idea that psychotherapy works in treatment resistant depression challenges conventional notions. We usually think of treatment resistant cases as more biological and needing medication, but I guess there’s really no evidence for that notion, and who’s to say that good psychotherapy does not have biological effects. 

Dr. Aiken: Yes I remember the old APA treatment guidelines recommended medication for severe depression and psychotherapy for milder cases, but since then there have been good studies of psychotherapy for severe and chronic depression, and now for treatment resistant depression. The specific therapies that worked in Jone’s analysis were cognitive behavioral therapy (CBT) for 6 weeks, mindfulness based CBT for 8 weeks, and psychoanalytic psychotherapy, delivered weekly over 1.5 years using a manualized adaptation of for depression called the Tavistock method. But note that these are either intensive – as in 1.5 years of psychanalytic therapy – or have a strong behavioral component. That’s true in severe depression as well – the therapies that worked often emphasized behavior change outside the room or they had a strong focus on the therapeutic relationship – to shake up the rigidity of severe depression and motivate change. So I wouldn’t conclude from this that any psychotherapy will work in these cases.

Kellie Newsome: Do your patients ever ask if you can prescribe something to help their memory and concentration? And are you a little weary of giving every adult with cognitive problems a psychostimulant? Today, in the patient edition of the Carlat Podcast, we explain how chronic psychiatric disorders can cause cognitive problems that look like ADHD, and detail behavioral steps to sharpen cognition. Find it in your podcast store as The Pocket Psychiatrist. 

Kellie Newsome: And now for the word of the day…. Budeprion

Dr. Aiken: 

Bupropion is not a medication you want to take too much of. Seizures are associated with peak serum levels, and the antidepressant has been wrapped in various modified release mechanisms since it first came out in 1985 to soften those blows. Those changes have also eased the dosing for patients, from three times a day with bupropion instant release, to twice a day with bupropion sustained released, and finally in 2003 patients could take a full dose of 300-450 mg all at once with bupropion XL, and later with Forfivo, and Aplenzin. 

But those branded XLs were expensive, so it was welcome news when Teva pharmaceutical partnered with Impax Laboratories to launch the first generic XL in December 2006. But one thing seemed odd – the generic was no longer called bupropion XL and instead was labeled budeprion XL. Behind the different spelling was a fact that most physicians were not made aware of: The two products were not identical. Budeprion XL used a matrix for slow release, while the branded Wellbutrin XL used a membrane. And worse, the company was able to gain approval by showing therapeutic equivalence of the 150 mg XL doses – the 300 mg were never tested, but they were about to be as 20 million Americans were switched to the generic drug.

Tune in next Monday to hear the rest of that story, and more sordid tales of generic mishaps when we interview investigative journalist Katherine Eban.

Kellie Newsome: Do reality TV shows like The Biggest Loser help the people who watch them lose weight, or do they just demoralize them with body-shaming guilt and make them want to eat even more? Find out in today’s @CarlatPsych tweet, where you can follow us for daily research updates.

Got feedback? Take the podcast survey.


Leave A Comment