Are Placebo-Controlled Studies Still Ethical For The Development of COVID-19 Therapeutics?
The COVID-19 pandemic has been a moving target over the last two years. Changes in the virus that causes COVID-19 (SARS-CoV-2), such as the emergence of viral variants, have necessitated changes to surveillance protocols and medical patient management. The available COVID-19 therapeutics are currently limited to monoclonal antibody cocktails. But recently, Merck presented promising results from the trials of a ribonucleoside analog (molnupiravir) that indicates a 50% reduction in the risk of hospitalization or death among patients receiving molnupiravir compared to a placebo. Until just recently, the FDA agreed that placebo controls were appropriate for the evaluation of potential therapeutic candidates. Recent FDA feedback has indicated that the use of a placebo is now considered unethical, and the FDA expects that new drug candidates will be evaluated against an active treatment using a non-inferiority study trial design. The FDA offers meetings to discuss trial design.
Placebos are used in clinical trials to conceal whether a treatment is being given or not, and therefore, to control for the psychosomatic effects of offering treatment. The dominant arguments in opposition to placebo-controlled trials, in the face of proven effective treatment, are that they 1) violate the therapeutic obligation of physicians to offer optimal medical care and 2) lack both scientific and clinical merit. The changing FDA policy on placebo-controlled studies to demonstrate efficacy of a new drug candidate has changed the approach from demonstrating superiority against placebo to a non-inferiority approach using either monoclonal antibodies or molnupiravir as a positive control.
For COVID therapeutics, European regulators still accept the ethics of placebo-controls as none of the monoclonal antibody drugs are available in Europe and are in short supply even in the US (country of origin). The European regulatory authorities continue to allow placebo-controlled studies of COVID-19 therapeutics, and the distinction in the US complicates the development of a single international development plan (IDP) or even a single protocol for the US and Europe, much less a coherent strategy for development. The FDA is being pressed to address this conundrum since the use of active controls in the US and placebo elsewhere would not be optimal.
The nature of the pandemic has changed with the emergence of SARS-CoV-2 variants, such as the Delta variant which may be less sensitive to monoclonal antibody drugs. Delta is far more contagious than previous variants and may result in widespread disease, impacting younger adults and children who seemed to have a vague resistance to earlier viral variants. Regeneron has already withdrawn the Emergency Use Authorization (EUA) for one of its monoclonal antibody (mAb) drugs because of changes in the spike protein of SARS-CoV-2.
A EUA is typically granted based on interim data from clinical trials. Once the pivotal studies have been completed, the FDA receives the full dossier to review, and full approval may be granted pending evaluation of the data package. This was the situation with the monoclonal antibody drugs from Regeneron, Eli Lilly, and with molnupiravir from Merck.
Should active controls (non-placebos) be required when the armamentarium of drugs to treat COVID-19 is so limited? Do we need to know which drug is superior? Or at this stage do we just need to know a new drug candidate is safe and effective?
All three COVID-19 vaccines available in the US (Pfizer, Moderna, and J&J) were based on placebo-controlled superiority clinical trial design. Though each of these vaccines has its niche, if one vaccine were compared to another, we know now that the Moderna vaccine appears to provide a greater amount of protection from disease than either the Pfizer or the J&J vaccine. In a head-to-head comparison based strictly on efficacy, Moderna would have been the leader (although the margin would be small) and the other products might not be available. Likewise, molnupiravir was developed against placebo, despite the availability of the monoclonal antibodies from Regeneron and Eli Lilly. Merck reported that molnupiravir reduces death and hospitalization by 50%. The Regeneron and Eli Lilly therapeutic antibodies are reported to be 85% effective against the same endpoint. Using a non-inferiority design and a 20% non-inferiority margin had molnupiravir been studied in a non-inferiority study, it might have failed despite its effect on the progression of severe disease and death.
It may be that SARS-CoV-2 mutates and becomes resistant to molnupiravir much in the same way that drug resistance was a major issue in the earlier days of the HIV epidemic. We will need additional drugs with different targets and modes of action if SARS-CoV-2 becomes resistant to first-line therapy. A single drug, even if it is highly effective, is not sufficient against a target that mutates as readily as SARS-CoV-2. Combinations of antivirals were required in order to control HIV disease and may be required to best treat COVID-19.
Non-inferior, active study design (non-placebo) may become the new normal, and approval of molnupiravir may be an industry game-changer. But non-inferiority is often difficult to demonstrate. A paradigm for the selection of the appropriate comparator, as described above would be important. Furthermore, non-inferiority boundaries often become problematic – select too narrow a margin, and the demonstration of efficacy can take too long and require too many participants; select margins that are too wide, and the exercise becomes pointless.
Once multiple therapeutics are available, identifying the appropriate control to employ for future drug candidates is still in question. For example, should a biological like a monoclonal antibody be used as a positive control against a small molecule like a nucleoside? Or is another nucleoside inhibitor of the viral RNA-dependent RNA polymerase (RdRp) a better control? How will the appropriate active comparator be selected? It would seem to make sense that the active control should be of the same drug class as the drug candidate: antibodies for biologicals and nucleoside/nucleotide analogs for small molecules; RdRp inhibitors should be compared to RdRp inhibitors; protease inhibitors to other protease inhibitors once we have multiple drugs for each class.
In conclusion, we have not yet arrived at the point where we are able to determine if non-inferiority clinical trial designs provide benefit over placebo designs. We also don’t yet know about the synergy between multiple drugs. Once we have accumulated a sufficient arsenal of different agents, we can begin to ask these questions. But for now, we are left with the ethics of developing multiple agents for a single indication and the need to consider active control designs as opposed to placebo-controlled designs.