The role of regulators, investigators, and patient participants in the rise of the placebo response in major depressive disorder

Khan and Brown’s comprehensive overview (1) provides numerous opportunities for reflection on the role of regulators, investigators, and patient participants in the progressive rise of the placebo response in major depressive disorder (MDD). They rightfully argue that the DSM-III and subsequent iterations have led to an expansion of the population meeting criteria for that disorder, and that some of the patients entering MDD trials may have questionable forms of this condition. This is certainly consistent with our experience of independently interviewing patients considered to meet criteria for MDD by site investigators (2).

Khan and Brown also review the robust evidence for investigators’ bias influencing the magnitude of symptom reduction across all treatments, including placebo. Their arguments are consistent with a review from our group (3), which has shown smaller effect sizes in antidepressant trials using active placebos compared to those observed in the presumably less blinded trials with inert placebo, and has pointed out that a trial of quetiapine in bipolar depression found no difference in efficacy between active treatment and placebo in the groups of patients reporting sedation as an adverse event.

While I fully agree with Khan and Brown’s views on the role of patient participants and investigators in the rise of placebo response in antidepressant trials, I differ on their perspectives on the regulatory contributions to this problem. In particular, they argue that U.S. Food and Drug Administration (FDA)’s decisions have, at times, negatively influenced the design and execution of antidepressant trials in the past three decades by using conservative approaches to both design and analysis of trials. I would argue the opposite: the FDA, in my opinion, has shown a great deal of openness to innovation and to novel designs and methods, but we, as a field, have justified our conservative stances by using the anticipation of a negative FDA response as a reason for holding on to obsolete standard designs. Let me offer a few examples.

Khan and Brown mention the regulatory burden of the required use of an active control to establish assay sensitivity, an approach shown to lead to higher placebo response rates due to expectations of increased odds of receiving a form of active treatment (4). As far as I can tell, this has never been a regulatory requirement in adult MDD trials (though often suggested), but it became a popular approach among sponsors with the goal of de-risking investments in novel therapies of uncertain efficacy. In fact, the FDA has approved vilazodone as an antidepressant in the absence of any data involving an active control (5).

Similarly, Khan and Brown state that dose-response data are a regulatory requirement that has led to inflated placebo response rates because of the expectations of increased odds of receiving a form of active treatment (4). Once again, although the FDA likes dose-response data in a submission, there is no official requirement in phase III and, in fact, vilazodone was approved based on two studies that evaluated only one dose (5).

Khan and Brown also state that the FDA staff has never accepted the concept of therapeutic response as a valid primary outcome and that the FDA considers valid only the use of the change in total scores of scales such as the Hamilton Depression Rating Scale (HAM-D) or the Montgomery-Asberg Depression Rating Scale (MADRS) as primary outcome measure. Once again, sponsors have favored this approach because of the greater sensitivity to detect changes of a continuous measure compared to a dichotomous measure, but, in fact, the FDA has accepted proposed registration studies using time to response as the primary outcome.

Khan and Brown add that the FDA does not accept, contrary to the European Medicines Agency (EMA), randomized withdrawal study designs. Actually, S. Borges and other colleagues from the Division of Psychiatry Products of the FDA usually include a post-marketing requirement for such type of maintenance studies in the initial approval, and have in fact published a paper (6) strongly endorsing this design as having good sensitivity to detect signals with antidepressants. Indeed, lamotrigine was approved for maintenance treatment of bipolar disorder by the FDA only based on two relapse prevention trials (7), i.e., in the absence of acute treatment data.

Another example of using the anticipation of a negative FDA response as a reason for holding on to obsolete standard designs is the fact that a number of sponsors have often used longer duration of trials “because the FDA requires it”. On the contrary, N.A. Khin and other colleagues from the Division of Psychiatry Products of the FDA (8) have shown in their pooled analyses of 81 trials of antidepressants that 6-week trials had a higher success rate than 8-week trials (55% vs. 42%), and trials as short as four weeks are considered acceptable by the FDA.

Finally, adaptive designs have been mentioned as critical innovations by the FDA Director of Psychiatry Products at that time (9), and novel study designs aimed at reducing the placebo response, such as the sequential parallel comparison design, have been used in Phase II and Phase III antidepressant trials (10), with FDA statisticians having published on new methods to analyze them (11).

While it is important to identify factors that may have contributed to rising placebo response in depression trials, there is no evidence that the regulatory agencies per se played any role in this. On the contrary, we, as a field, have justified our conservative stances by using the anticipation of a negative FDA response as a reason for holding on to obsolete standard designs and methodologies.