[Preliminary drawing of very far-out conclusions from research that hasn’t even been 100% confirmed yet]
A few weeks ago, Nature published a bombshell study showing that ketamine’s antidepressant effects were actually caused by a metabolite, 2S,6S;2R,6R-hydroxynorketamine (don’t worry about the name; within ten years it’ll be called JOYVIVA™®© and you’ll catch yourself humming advertising jingles about it in the shower). Unlike ketamine, which is addictive and produces scary dissociative experiences, the metabolite is pretty safe. This is a big deal clinically, because it makes it easier and safer to prescribe to depressed people.
It’s also a big deal scientifically. Ketamine is a strong NMDA receptor antagonist; the metabolite is an AMPA agonist – they have different mechanisms of action. Knowing the real story behind why ketamine works will hopefully speed efforts to understand the nature of depression.
But I’m also interested in it from another angle. For the last ten years, everyone has been excited about ketamine. In a field that gets mocked for not having any really useful clinical discoveries in the last thirty years, ketamine was proof that progress was possible. It was the Exciting New Thing that everybody wanted to do research about.
Given the whole replication crisis thing, I wondered. You’ve got a community of people who think that NMDA antagonism and dissociation are somehow related to depression. If the latest study is true, all that was false. This is good; science is supposed to be self-correcting. But what about before it self-corrected? Did researchers virtuously say “I know the paradigm says NMDA is essential to depression, and nobody’s come up with a better idea yet, but there are some troubling inconsistencies in that picture”? Or did they tinker with their studies until they got the results they expected, then triumphantly declare that they had confirmed the dominant paradigm was right about everything all along?
This is too complicated an issue for me to be really sure, but overall the picture I found was mixed.
A big review of ketamine and NDMA antagonism came out last year. In this case, I was most interested in the section on other NMDA antagonists – if ketamine’s efficacy is unrelated to its NMDA antagonism, then we shouldn’t expect other NMDA antagonists to be antidepressants like ketamine. So if the review found that other NMDA antagonists worked great, that would be a sign that something fishy was going on. But in fact the abstract says:
The antidepressant efficacy of ketamine, and perhaps D-cycloserine and rapastinel, holds promise for future glutamate-modulating strategies; however, the ineffectiveness of other NMDA antagonists suggests that any forthcoming advances will depend on improving out understanding of ketamine’s mechanism of action.
This is pretty impressive; they basically admit that other NMDA antagonists don’t work and that maybe this means they don’t really understand ketamine.
But dig deeper, and you find a less sanguine picture. The body of the paper lists notes five NMDA antagonists as confirmed ineffective – memantine, lanicemine, nitrous oxide, traxoprodil, and MK-0657. But the paper itself notes that all of these were effective on some endpoints and not others, and the decision that they were ineffective was sort of a judgment call by the reviewers. Just to give an example, there’s only ever been one study done on traxoprodil. Since the reviewers reviewed this one study and declared it ineffective, you might expect the study to be negative. But here’s the abstract of the study itself:
On the prespecified main outcome measure (change from baseline in the Montgomery-Asberg Depression Rating Scale total score at day 5 of period 2), CP-101,606 produced a greater decrease than did placebo (mean difference, 8.6; 80% confidence interval, -12.3 to -4.5) (P < 0.10). Hamilton Depression Rating Scale response rate was 60% for CP-101,606 versus 20% for placebo. Seventy-eight percent of CP-101,606-treated responders maintained response status for at least 1 week after the infusion. CP-101,606 was safe, generally well tolerated, and capable of producing an antidepressant response without also producing a dissociative reaction. Antagonism of the NR2B subtype of the N-methyl-D-aspartate receptor may be a fruitful target for the development of a new antidepressant with more robust effects and a faster onset compared with those currently available and capable of working when existing antidepressants do not.
Read this quickly, and it looks like they’ve confirmed traxoprodil is pretty great. The reviewers say it isn’t. They argue that p < 0.10 isn't good enough (I think the study was trying to use a one-sided t-test or something?), and that of five different days when responses were measured (day 2, 5, 8, 12, and 15 after the infusion), there was only a difference on day 5. Apparently this isn't good enough for the reviewers. On the other hand, take rapastinel, one of the NMDA antagonists the reviewers say "holds promise".
No statistically significant differences were observed in rates of treatment response or symptom remission associated with placebo (64% and 42%, respectively) versus rapastinel at any dose (up to 70% and 53%, respectively). However, statistically significant differences in the reduction of the 17-item HAM-D scores were observed for the 5-mg/kg dose at all intervals except day 14) and the 10-mg/kg dose at day 1 and day 3. Neither the low nor the high rapastinel doses were associated with significant greater 17-item HAM-D score reduction than placebo, leading the authors to posit an inverted U-shape dose-response curve.
Sometimes things do have inverted U-shaped dose-response curves – for some discussion of why, read the Last Psychiatrist’s Most Important Article On Psychiatry. But a study that shows no treatment response or symptom response, and the test score response is only on a medium dose but not a high or a low dose – that makes me kind of suspicious.
Why is the review so much more accepting of these ambiguous results than of the last set of ambiguous results? Psych blog 1BoringOldMan points out that the original study was done by the company making rapastinel and two authors of the review article I’m citing were affiliated with the companies that are developing rapastinel. And that at least one of them has a “legendary” history of conflicts of interest.
I don’t want to say for sure this is what’s going on. For all anybody knows, rapastinel might work – the NMDA and AMPA systems are really connected, and the base rate of a randomly chosen compound being an antidepressant is higher than you’d think. But I think it’s at least one possible explanation.
This review article also gets into the nitty-gritty of mechanisms of action:
That other NMDA channel blockers have yet to replicate ketamine’s rapid antidepressant effects has led to speculation that ketamine’s antidepressant properties may not be mediated via the NMDA receptor at all…additional evidence indicates that activation of glutamatergic AMPA receptors is necessary for ketamine’s antidepressant effects. Specifically, coadministration of an AMPA receptor antagonist has been shown to block ketamine’s antidepressant-like behavioral effects.
So that’s neat.
Two other relevant studies: Do The Dissociative Side Effects Of Ketamine Mediate Its Antidepressant Effects finds that they do, which contradicts the recent metabolite-related findings. On the other hand, the two papers share some authors, so I’m tempted to say it was an honest mistake. This paper incidentally finds that the dissociative effects of ketamine are not related to its antidepressant effects, which I think makes more sense now.
The other studies I found were mostly compatible with the new results, with a lot of people expressing doubt about whether NMDA really mediated ketamine, a lot of people finding null results for other NMDA antagonist medications, and a lot of people saying there were weird hints that AMPA was involved somewhere.
I feel kind of premature doing this, because as much as I think it’s elegant the discovery about the metabolite hasn’t been totally confirmed yet. But assuming it’s right, psychiatry comes out of this looking sort of okay. There were a lot of early results with a lot of hype. But the big review articles mostly put these in their place and were able to come up with the right results and fit the pieces together.
The one place this wasn’t so clear was when there were conflicts of interest. If we assume rapastinel doesn’t really work (which right now would be very preliminary and I’m not actually saying this, but these latest findings do seem to imply that), various teams made up of people affiliated with rapastinel’s manufacturers were unable to determine this (neither was the FDA, who just just gave rapastinel “breakthrough drug” status, apparently on the strength of industry studies).
A big reason I’m concerned about this is that I want to know how much to trust the rest of the psychiatric literature – for example, those claims about SSRIs being mostly ineffective. An answer of “you can trust it a lot, except in cases of conflicts of interest” would be a mixed bag. Almost every drug was originally researched and promoted by people with conflicts of interest, and then we trust the academics to catch up with them later and keep them honest. I don’t think this system has failed us too terribly yet. But it’s important to remember that that is the system.