The media has reported that the Trump Administration in the person of Mark Meadows, the Chief of Staff, had withheld new guidance to be issued by the FDA expanding or clarifying the regulatory requirements for a vaccine. That guidance was released and published Tuesday afternoon. The media has since reported that the new guidance makes a vaccine before Election Day impossible. Is this true? We think not.
The media point to two aspects of the new guidance: (1) the requirement that at least 5 persons in the placebo group have serious illness to demonstrate that the vaccine works to prevent serious illness; and (2) that the median follow-up in the safety cohort be at least two months.
We had previously suggested that an optimistic view of the current vaccines in development suggests that one or more companies may be able to announce vaccine approval before the election and, potentially, initiate vaccination shortly thereafter. Unlike therapeutic drugs, the rate limiting step for preventive vaccines is efficacy rather than safety. Safety populations for both are mandated at a minimum of 3,000 subjects. For a therapeutic, a patient population of 1,000 is generally sufficient. For a vaccine, an efficacy population of over 30,000 is generally required.
Briefly reviewing the Pfizer vaccine:
- The study started on July 27.
- By August 20, Pfizer announced that 11,000 subjects had been enrolled.
- The two-dose regimen (21 days apart) in that cohort would have been completed on September 10.
Now for some assumptions. If an average of 450 subjects were enrolled per day, the median 1500th subject of the 3,000 subject safety cohort would have completed two doses by August 21st and had a median follow-up of two months by October 21st – suspiciously close to FDA’s October 22nd Vaccine Advisory Panel meeting. FDA also stated in yesterday’s guidance that the agency would need two days from submission of data to a Vaccine Advisory Panel review.
That said, the FDA’s latest guidance is unclear as to whether the median two month follow-up is on the entire efficacy cohort of 30,000 (last enrolled in mid-September), which would not likely be until late December. The exact statement is:
“Data from Phase 3 studies should include a median follow-up duration of at least two months after completion of the full vaccination regimen to help provide adequate information to assess a vaccine’s benefit-risk profile.”
The relevant statement from the prior guidance is:
“The pre-licensure safety database for preventive vaccines for infectious diseases typically consists of at least 3,000 study participants vaccinated with the dosing regimen intended for licensure.”
My rationale for suggesting an October 21 date as the earliest feasible date is that, as noted, the safety population is the relevant population in which to assess the “risk” portion of the risk-benefit profile; that President Trump himself, per his Tweets, may have rapidly overturned Mr. Meadows’ hold after communication with Pfizer; the suspicious timing of the Vaccine Advisory Board; and, per FDA’s comments yesterday, Pfizer was already aware of the “new” requirements and has yet to withdraw its optimistic prediction.
Of course, this assumes that the most difficult hurdle – proving clinical efficacy – can be achieved by this vaccine. And that depends both on whether the vaccine does work and on the underlying prevalence of COVID-19 in the placebo population. Two significant unknowns that might be revealed by the next planned blinded efficacy analysis. In that regard, the media’s skepticism is well-founded.
The prevalence of serious disease in Brazil, one of the study sites, in September based on a variety of data from public pronouncements to a recent Lancet article, is likely around 0.12%. We pick September because, significantly, FDA’s new guidance permits exclusion of efficacy data from a period before both vaccinations in the series have taken effect – a request of the industry. That means that, among the 11,000 patients enrolled and with effective antibodies by September 20, there could be as many as 13 subjects with serious disease by October 21, making the 5 cases in the control arm a surmountable if close hurdle. Other sites may have a lower or high prevalence. To that end, the Brazil prevalence would be the general population. We suspect that Pfizer would be enrolling a substantial number of individuals at all sites who may be at higher risk such as healthcare workers, persons living in close communities, and the elderly and seeking early approval for those subjects with general population to follow.