COVID AUs

how Peter Marks and Deborah Birx might have gotten us vaccines a lot sooner

I want to tell two counterfactual histories about how our COVID-19 response circa mid-2020 could have gone, with 3 heroes:

1. Dr. Peter Marks (head of the FDA’s CBER)

2. Dr. Bob Kadlec (head of HHS’s pandemic response), and

3. Dr. Deborah Birx (head of the COVID task force)

Most of this information comes from Brendan Borrell's excellent and comprehensive account of Operation Warp Speed. I highly recommend it. You can also find some of this information in Dan Elton's account of the Pfizer vaccine delay.

If we think back to March 2020, as COVID was spreading in the US, the experts were telling us we would need at least 12-18 months before a vaccine was approved and scaled up— if we were very lucky. Two key people came up with an idea to radically speed up this timeline: Peter Marks at FDA and Robert Kadlec, at HHS’s ASPR, who thought up Operation Warp Speed (OWS). OWS was a tremendous success, but there were two choices that could have yielded even better results.

Just test in animals!

The first was more radical: Marks and Kadlec, familiar with the "Animal Rule" at the FDA that allowed for approval of countermeasures (vaccines and drugs) based only on animal studies in some circumstances

¹

, proposed a comprehensive set of animal experiments to substitute for a Phase 3 trial.

I'm not sure how detailed their plans ever were for this, but I imagine the FDA would have required Phase 1 and 2 human trials for safety and dosing and then a set of well-done challenge studies in a few different animal models. Then they probably would have approved the vaccine(s) only for particularly high-risk individuals while phase 3 studies were conducted in the general population. Here are the key book quotes

insistent, however, that the Operation Warp Speed portfolio be selected, at least in part, based on the animal trials he had set up. This would give the government a truly comparative picture of all the vaccines it was funding to ensure it was making the best investments. Marks’s model for the animal trials could also ensure enough validated animal data to justify deploying the vaccines before the clinical trials came to an end. (Location 2573)

scenario where they would run a more intensive set of animal studies than was usual for standard vaccine development. He wanted studies with multiple species: mice, hamsters, ferrets, and monkeys. The immunology results would then be cross-checked with the phase 1 safety trials in humans. This would allow greater confidence regarding which animal species were the best models for how humans would respond to the vaccine and suggest whether deploying a vaccine without the completion of a phase 3 trial was going to be an acceptable risk. (Location 2103)

Marks, in an inspired riff, took this idea farther, to a place not even Kadlec had anticipated. Marks realized that those animal studies could be part of a rigorous competition to evaluate the candidates of a sea of companies ... government would pay to ramp up manufacturing for those top candidates even while the vaccine was still being tested, so any that were granted the green light could start going into the arms of vulnerable adults immediately. (Location 2107)

For 70-80 year old’s, where COVID IFR was well above 5%, practically any vaccine with some efficacy that didn't outright kill people would have been a great idea. Plenty of other people had similar ideas because the cost-benefit of even mediocre vaccines was extremely obvious back then and any sensible prior of vaccines (which are usually quite safe, moderately effective, and almost never have cryptic long-term harms) would support that. Here is Avi Bitterman making a similar point in 06/2021.

For the general population, big phase 3 trials to pin down efficacy and safety would almost certainly have been required. I'm not sure how much that would have sped up approval and manufacturing, but it might have made a big difference. The big unknowns that could've made this scenario a wash are:

1. how quickly manufacturing could have scaled and

2. how high vaccine uptake would have been in the context of an approach that might be perceived as more experimental. 

Just send it to nursing homes!

The second possibility was less radical but also potentially high-impact. The unfairly maligned Dr. Deborah Birx had a clever idea in Summer 2020: have the FDA issue a compassionate use exemption (also known as “expanded access”) for the vaccines for elderly people, perhaps only nursing home residents. Mortality rates for nursing home residents were staggeringly high, reaching 15-20% in some places. Practically any vaccine was a no-brainer back then. From the book (bold mine):

Kadlec, who had sat in on those Warp Speed meetings about vaccine distribution, was aligned with Deborah Birx in his belief that the first doses needed to get to the elderly as soon as possible. (Location 3870)

Birx shared a home with her ninety-year-old parents, and she said she was watching the coming winter peak with increasing alarm. She had just come back from a COVID-19 outbreak in North Dakota and feared for the country. She felt that people in their seventies and eighties, whose risk of death from COVID-19 was sixty times that of people in their thirties, should be prioritized. (Location 3725)

Birx was feeling increasingly frustrated that she hadn’t been given another opportunity to weigh in on the CDC’s administration plans since that meeting in July. In fact, she had been so concerned about the elderly that she had recently asked Fauci to encourage Moderna’s Stéphane Bancel to file a compassionate use application with the FDA that would get the vaccine to nursing-home residents before there was any proof that it worked in people. Moderna had declined. A similar request to Pfizer was also turned down. She chalked it up to the fact that no company wanted to be seen as doing Trump’s bidding prior to the election. (Location 3728)

However, when she brought this idea up to Moderna and Pfizer, they declined. She attributes this to companies being unwilling to "do Trump's bidding" so close to the election. This was a big missed opportunity. Blunting the impact of COVID on our most vulnerable could have made a big difference in the large winter waves.

Similar kinds of exemptions have been used in more ambiguous situations: convalescent plasma as an example. And they were also used in the AIDS crisis, when AZT was shown to be so effective that a trial was stopped early and a compassionate use exemption was granted to 4k+ patients while the FDA fully reviewed the drug and approved it. So Birx's proposal had precedent and would not have been some radical destruction of FDA authority. I cover that story here—ctrl+f “AZT” to skip to it. I quote below:

Phase 2 trials which had begun in February 1986 were halted early in September 1986 due to clear signs of treatment success, and AZT was officially submitted to the FDA for approval in December 1986. Eileen Cooper, a rising young star at the FDA, was in charge of reviewing it, and had been reviewing the AZT data for months before the official submission date. Even before the most militant AIDS activists had begun pressuring the FDA, she had been discussing with others on ways to speed and streamline the approval process. 

She took two important steps. First, in September 1986 she had released AZT for compassionate use to 4000+ AIDS patients, which likely saved many lives. Second, she sought the support of the FDA's Advisory Committee on Infective Drug Products in a January 1987 meeting, which would symbolically back up the FDA's decision to approve AZT on the basis of a single prematurely ended clinical trial. This would achieve two contradictory goals: the rapid release of a likely effective drug to suffering patients; and satisfy the consumer protection and public health voices that generally urged caution.

Since a classic anti-vax talking point is that the COVID vaccines were rushed, I want to take a moment to run down different gradations of "rushed" and show why the COVID vaccine approvals were not actually all that rushed.

Run vs walk vs standing still

1. RadVac, not FDA approved, “biohacker” vaccine, no animal data.

2. Peter Marks’s original plan: approved on the basis of the Animal rule + human studies to assess safety and dosing.

3. Birx’s nursing home triage plan: older adults in vulnerable settings access the vaccines before trial results were done under a compassionate use or emergency authorization program.

4. EUA on the basis of interim trial results that showed safety and efficacy, age recommendation that moves down over months, in-line with cost-benefit [what actually happened]

5. full 30k person trial must finish (multiple years by historical vaccine standards) before approval, no interim looks at trial data.

I don't know how close these ideas came to being enacted. My speculative guess is that the Mark’s original plan was unlikely and Birx’s plan was very realistic. I think that a less politically polarized situation (even a pandemic in 2019 instead of 2020, close to an election) could have easily led to the latter.

Villains?

I think much of the responsibility lies in the media, who fear-mongered, based on little evidence, that Trump would force the approval of a dangerous vaccine.

The lack of message discipline on Trump's part, with his rambling speculations on sunlight, bleach, and hydroxychloroquine, allowed the media to pounce on his remarks. Trump officials who didn't push aggressively on these issues may also bear some responsibility— Alex Azar, for instance, is noted repeatedly in Borrel's book to care a ton about "process", so perhaps he might have opposed these more zany ideas from Birx and Marks. On the other hand, he did approve OWS in the 1st place, so he is probably more open-minded than I think.

Some Democratic politicians like Kamala Harris and Andrew Cuomo made some lukewarm comments on vaccine initially, which might be a minor factor.

The final straw was the incorrectly worded remark by Steve Hahn (acting FDA commissioner) overstating the benefit of convalescent plasma at a press conference, which turned out to not do anything, and which was subsequently heavily criticized by Eric Topol et al.

Steve Hahn did nothing wrong!

Here, I want to take a moment to defend the emergency use authorizations of convalescent plasma and hydroxychloroquine. The statutory basis of the EUA authority the FDA has comes from a law (Public Readiness and Emergency Preparedness Act) that states that in pandemic emergencies, the evidentiary standard for therapeutics should be lower than normal.

The point is that the FDA had every right to approve medical therapies as long as they had a shot of working and probably didn't cause harm. Convalescent plasma and hydroxychloroquine both fit that bill. They both had some observational data behind them and while they turned out to be useless (like the vast majority of drug candidates) for treating COVID, it doesn't seem as though there was a lot of iatrogenic harm. A gentle reminder: doctors can and do prescribe plenty of drugs off-label, often, unfortunately, off nothing more than some suggestive observational trials. Prescribing HCQ for COVID outside of an RCT is bad, but quite within accepted rules of medicine.

We Need Better RCTs

The real lesson from the HCQ/plasma fiasco, in my view, is that we need a much better clinical trial system in the US. It is an absolute disgrace that clinical trials require so much overhead in terms of logistics, staff, IRB, etc. that we have many more patients (I’m guessing ~ 10x as a lower bound) participating in uncontrolled observational “trials” than pragmatic RCTs. If a patient had wanted to take an experimental drug for COVID we should've been saying: "hey, we have no idea what might work, there's these 5 different treatments we're trying, you might get assigned a placebo, you'll help us advance science, the consent form is an index card, that’s it.".

I have no idea how we'd go about making a system for doing RCTs quickly and at scale, but someone in the US government should hire the UK people who made the RECOVERY trial work, hire David Boulware, who worked in the trial that showed Flouvaxamine probably works, and get this done. Or maybe this requires a truly unified healthcare system and is thus a political non-starter. Not sure.

Anyways, all these factors above in my view, led to an instinctive caution by FDA career scientists (I think the same ones that retired early under Biden) and pharma, who didn't want to be seen as yielding to Trump's pressure to approve a vaccine early and had just been roundly criticized for months by the media and prominent academics. Some quotes:

FDA hadn’t updated the guidance it had published back in June, but Warp Speed had heard that Gruber’s office was floating that forty-two-day number. Was this going to be put into some kind of public guidance?…

Azar called FDA commissioner Stephen Hahn and told him he had heard about the letters his agency had sent. “No, we didn’t,” Hahn said. “Yes, you did,” Azar said. “No, we didn’t,” Hahn said again. “We’re literally reading the letter right now,” Azar snapped. Hahn hung up and soon found that, sure enough, Peter Marks and the FDA vaccine leader Marion Gruber had sent those letters to manufacturers without telling him.

P.S. Eric Topol, who led the charge against a slightly earlier approval of the “Trump vaccine” on the basis of a flimsy distinction between median and average, is now the leading cheerleader for approving booster doses for all ages, mostly on the basis of observational data (and some immunogenicity data) alone. I think he’s probably right on the cost-benefit calculation, especially for older folks, and Omicron has probably proved him more right, but the case was far more clear in November 2020, and yet his position then was “we need more data.” Very confusing!

Edited (12/20) to add the relevant “Gruber” quotes from First Shots

1. For some diseases like anthrax and smallpox it'd be unethical and/or dangerous (since nobody gets a smallpox vaccine nowadays and it is incredibly contagious) to do human challenge trials for new drugs, and there aren't enough cases in the wild for a regular RCT. This is where the Animal Rule comes in.