Confirmation bias and esketamine approval

This review, published on line by the FDA reviewers who voted (without unanimity) to approve esketamine for treatment-resistant depression (Jean Kim and colleagues, New England Journal of Medicine, published on line, ahead of print, on July 4, 2019), illustrates potential adversity that may devolve from confirmation bias with good intentions. We do not know whether the authors submitted it unaware of or in spite of the former, but their submission seems destined to fuel controversy about whether esketamine’s bioavailability-dependent emergent diversity of actions may confer more risk than benefit in the care of persons with treatment-resistant depression.

The premise upon which the authors’ review appears to be based is that personal and economic adversity of disease may justify important irregularities in the drug approval process. Unfortunately, according to their exposition, the irregularities derive importance from substituting intuited for documented efficacy and avoiding comprehensive consideration of complex risk.

The FDA reviewers unselfconsciously describe the process that culminated in approval of esketamine for treatment-resistant depression after it had failed to meet FDA’s usual and customary standards of effectiveness, and after its manifold activity and corresponding adversities were taken lightly or dismissed.

In only one of three acute efficacy studies did esketamine prove superior to placebo (both coadministered with an antidepressant, and, sporadically in some cases, a benzodiazepine). The primary outcome criterion was change on the Montgomery-Asberg Depression Rating Scale (MDRS): the difference in rate of improvement was statistically significant in one of three efficacy trials, and that difference was of small effect size. The first efficacy trial was immediately followed by a discontinuation study, wherein a statistically significant difference in depressive relapse favors those who continue to take an experimental drug. It is usually done last, after the first and second trials have shown a drug candidate superior to placebo. That was not done with esketamine, because the clinical trialists were (so they report) confident that esketamine would proceed to evince superiority over placebo in the trials to follow.

After four months of treatment with the maximal dosage of esketamine in the first acute efficacy trial cum discontinuation study, esketamine-treated subjects were started on a dosage nearly 70 percent lower for two weeks, and then were abruptly withdrawn from esketamine (while antidepressant was continued at constant dosage).  What the authors consider depressive relapse occurred in a significant proportion of subjects four to twenty days later (a briefer latency preceding depressive relapse than reported generally and closer to that consistent with emergence and worsening of a drug withdrawal syndrome). Three suicides occurred among the relapsers, which the FDA authors do not mention in their review. Nor do they mention that esketamine, a congener of phencyclidine (PCP) that, like phencyclidine, may induce dissociation, sedation, delusionality, and hallucinosis dose-dependently; has opioid activity (by virtue of noncompetitive mu receptor antagonism) that is accompanied by clinically significant analgesia and mood-elevation blocked by the opioid receptor inhibitor naltrexone. Clinical anecdotage corroborates esketamine abuse, dependence, and opioid-like withdrawal symptoms, a factor documented to contribute to risk of suicide.

 Besides the authors’ failure to mention that suicides occurred in the discontinuation-relapse group, they fail to acknowledge the similarity (and association, causal and not) of opioid withdrawal symptoms and depressive relapse, and the association of both with suicide. They moreover confide nothing about how rapidly relapsers remitted after resumption of esketamine—significant with respect to differentiating depressive relapse and drug withdrawal.

That confirmation bias appears to have contributed to esketamine’s approval is revealed by the authors’ admission that the order of the esketamine efficacy trials was reversed: Typically conducted last, after a drug’s efficacy has been established, the discontinuation study was conducted after only one trial (the one of three that showed statistically significant superiority over placebo, as analyzed by a one-tailed T-test (which assumes that the directionality of an effect has already been established)). The next two mandatory efficacy trials found no statistically significant difference in efficacy between esketamine and placebo, using the same primary outcome criterion as the first (improvement, rated with the Montgomery-Asberg Depression Rating Scale (MDRS)).

The authors were aware on the occasion of voting for its approval that esketamine had proved ineffective in two of three short-term efficacy trials—a result that ordinarily would cause a New Drug Application to be rejected. Moreover, they seem unaware (or at least, do not mention in their review) that esketamine has opioid-like analgesic and mood elevating properties; that its opioid-like analgesic, behavioral and subjective effects (but not its dissociative, anesthetic effects) are blocked by the opioid receptor antagonist naltrexone; that it is sought, liked, and abused to an extent that has made it one of the most popular of the so-called party drugs in the U.S.; and that, since its approval in 1970 as an anesthetic, esketamine abuse has been associated with opioid-like withdrawal symptoms after abrupt discontinuation.

Thus was esketamine, ineffective for treatment-resistant depression by FDA’s standards and potentially conducive to suicide risk by virtue of opioid properties, including opioid withdrawal symptoms, approved for treatment-resistant depression and intended for coadministration with an antidepressant.  

(Obiter: The authors offer in their own defense a citation, reporting that up to fifty percent of short-term trials of long-established antidepressants fail to show statistically significant superiority to placebo (as in comparison studies, for example) (N.A. Khin and colleagues, Journal of Clinical Psychiatry 72(4):464-472,2011), yet they do not discuss the cited study’s  relevance to a purported antidepressant augmentation in a not necessarily comparable patient sample assessed by different outcome criteria. Nor do they discuss the complex relationship between the bioavailability of esketamine and the variability of response of non-anesthetized humans to multiple, distinctive activities that emerge and predominate at different exposures. But even had the cited review comprised studies comparable to the esketamine preapproval trials, the relevance of its finding is obscure.

 Now that FDA has approved esketamine for coadministration with antidepressants for persons with treatment-resistant depression, and esketamine’s manufacturer, Johnson and Johnson, has launched it, a new increment of risk of iatrogenic depressive relapse and of suicide may have been added pursuant to FDA’s approval of an antidepressant augmentation with a low likelihood of efficacy.

Leave a Reply

Your email address will not be published. Required fields are marked *