We could have universal COVID vaccines very soon — if we urgently reform the process

Bureaucratic and logistical roadblocks, not the science, stand in our way

Patrick Collison is a co-founder of Fast Grants and the CEO of Stripe.

Operation Warp Speed and the broader 2020 vaccine effort was a miracle. It took less than 12 months to go from a novel pathogen to vaccine shots in arms. The response wasn’t perfect – while significantly faster than normal, the FDA was conservative – they slowed the trials and (unlike the UK) did not permit the human challenge trials that could have substantially accelerated things further. But, on the whole, the outcome was very good and one that we can all be proud of.

Since then, our response has been far less impressive. The urgency that prevailed in 2020 has evaporated. As a result of new COVID variants, vaccines are significantly less effective, and case figures have correspondingly gone bananas. There were many more new COVID infections in the past week in the US than in the corresponding week 1 year ago and 2 years ago. So, how should we feel, and where are we going from here?

Does all of this matter?

It’s hard to say exactly how much COVID matters in a highly vaccinated population (as now exists in most developed countries), but four considerations stand out.

• Deaths aside, COVID continues to make very large numbers of people sick. Virtually everyone knows plenty of people who contracted COVID after being vaccinated and were out of commission for a week or more. This isn’t the worst thing ever, but it’s something.

• COVID continues to kill large numbers of people. Yes, death counts are down a lot from the peak, but vaccine efficacy is waning: protection against death has declined from ~95% to ~85% or less.¹ In the US, roughly as many vaccinated people in the US die from COVID every day as there are total deaths across the country in motor vehicle accidents. (The current 7-day average is 380 deaths per day, about one-third of which are in vaccinated people.)

• It’s hard to know how to estimate the Long COVID burden. According to ONS data in the UK, 400,000 people (0.6% of the population) report that their ability to go about their day-to-day activities has been "limited a lot". That number continues to rise pretty quickly, even as infections predominantly occur in vaccinated individuals. There continues to be debate on the nature of Long COVID. While I view the evidence for its existence as quite strong, even a 50% chance of the pessimistic case being broadly true is quite costly in expected value terms.

• We run an ongoing risk of more harmful variants emerging. Estimating the relevant probabilities isn’t easy, but a quickly mutating, relatively pathogenic virus running rampant is not good news in any world.

We’re clearly in a much better position than we were in early 2020, but COVID isn’t over and done with. Compared to a world without COVID, I think it’s fair to call the current status quo “very bad.”

What can we do about it?

Moderna and Pfizer/BioNTech are working on “bivalent” vaccines that combine original COVID spike proteins with those from Omicron. These will help remedy the structurally silly situation that prevails today, where we’re vaccinating people with a COVID variant that was outcompeted a long time ago (early 2021). While there are nuances (Omicron appears to generate less cross-immunity than the original strain), updating the spikes as is currently proposed will almost certainly yield better immunity against current COVID variants.

While these updated vaccines will buy us time, they’re unlikely to be a panacea. Moderna and Pfizer recently announced that their bivalent vaccine candidates have roughly 3x lower neutralizing efficacy against BA.4/5 (the current variants) than they do against BA.1 (original Omicron). These results are consistent with other data showing that BA.4/5 possess potent immune escape, even in people who were infected with the original Omicron variant. So, vaccinating people with BA.1 spikes probably won’t end this thing – nor, most likely, will any polyvalent combination of spikes from particular variants.

Potent immune escape by recent COVID variants

COVID will very likely continue to mutate and cause problems until we create vaccines that inhibit transmission (which probably means nasal/mucosal membrane vaccines) and vaccines that confer better overall immunity across all COVID-19 variants.²

There are plenty of groups working on this, including Akiko Iwasaki’s lab at Yale, David Veesler’s at University of Washington, and Pamela Bjorkman’s at Caltech. Nature surveyed some more efforts a few months ago.

The basic strategy for pan-variant protection is pretty clear: most involve some kind of “nanoparticle” that embeds parts of different spike proteins that will cause the immune system to prefer the parts that remain unchanged across different coronavirus variants. (“Highly conserved epitopes.”) Current vaccines use the full spike protein, which means that the immune system indexes on fairly random parts of the spike, including parts that might change quickly.

We already have quite good evidence that such approaches can work based on initial mouse and non-human primate data, where robustly protective responses have been generated against a broad variety of sarbecoviruses. (The virus that causes COVID is part of this family.) Furthermore, the underlying nanoparticle platform used in one group’s work, David Veesler’s nanoparticle-based vaccine, has been validated in humans, and just finished a successful Phase 3 trial in South Korea. (The nanoparticle in this trial used a part of the spike protein from the original COVID strain, and isn’t itself a pan-variant vaccine.) But simply embedding slightly different proteins on its surface stands an excellent chance of conferring much broader protection.

We (Fast Grants) are in touch with a number of these groups. Despite excellent technology and promising early results in animal models, we estimate that the very earliest we will have access to these vaccines in humans is 2024. These groups need to run primate trials, then run human clinical trials, and then ramp manufacturing and distribution. Beyond having to jump through a lot of hoops, we’ve observed that they’re frequently tripped up by stupid things outside of their control, any one of which may hold their work back by months. (One group’s monkeys have been delayed by US Customs, which will push the start of their primate trial back ‘till September. Another is struggling to obtain necessary adjuvants. Multiple groups are unable to get access to current mRNA vaccines for research purposes because of legal barriers.) All groups we’ve interacted with are underfunded compared to what would be ideal.

Broadly speaking, the holdups involve some combination of logistical challenges and regulatory requirements, and the intersection between both. (You don’t in principle have to run a primate trial, but the FDA makes it harder to run a human trial if you don’t. You don’t in principle need to use “acute infection” as a trial endpoint; you could also use neutralizing antibody titers, which would be much faster and simpler.)

To speed things up:

• We should lower the barrier for human clinical trials and use simpler endpoints. For many vaccine candidates, we could run human trials concurrent with primate trials (once basic safety data has been obtained). In humans, we don’t need to repeat Phase I trials for platforms that have already been validated and derisked. (In this vein, the FDA’s recent announcement about not requiring trials for updated platforms was encouraging.)

• We should help these groups to scale manufacturing faster. Operation Warp Speed itself cost $10 billion; a second incarnation, with a tenth of that budget, could almost certainly accomplish a great deal.

Overall, we the federal government should empower someone to intervene (as it did with Operation Warp Speed). Private actors can’t change FDA policy. In our view it is probably true that, with competent execution, we could roll out pan-variant COVID vaccines before the end of 2022. Actually making that happen would require significant and coordinated logistical, regulatory, and administrative action. However, it would by no means be impossible. Not having pan-variant vaccines in 2022 is best thought of as a choice.

Why am I writing this?

As part of Fast Grants, we’ve tried to do what we can (including funding a number of these groups), but the impediments from here are mostly outside of our control. We’re therefore keen to elevate the prominence of the important work that these groups are doing and to highlight the impact that acting with greater urgency could have. Pan-variant vaccines are an important and tractable opportunity for us to save a lot of lives, suffering, and economic losses.

More broadly, I’m interested in getting things done quickly – especially things that are important for society. When we’re civilizationally incompetent, it often means something like “a bus lane takes decades to build instead of weeks”. In this case, however, it could easily mean that hundreds of thousands of people die needlessly.

Pan-variant vaccines are a $1 trillion dollar bill on the sidewalk. Some combination of a lack of basic institutional seriousness and self-imposed straightjackets mean that we will probably get these vaccines much later (perhaps years later) than we otherwise could. It is intellectually internally consistent to believe that COVID doesn’t matter and that we consequently shouldn’t care about any of this. But it is not consistent to think that COVID matters and that this situation is anything other than crazy.

1. An earlier version said 75%, but that was confusing since I didn’t separate out boosted and non-boosted populations. The latest data from the UK claims 50% vaccine efficacy against death after two shots but 87.6% vaccine efficacy against death after being boosted. (Other UK data shows much lower efficacy even after being boosted, but that isn’t corrected for incidental COVID infections.) Taiwan reports 83% efficacy against death in boosted populations. Washington State data shows 76% efficacy in the 65+ group, but that data doesn’t separate boosted and non-boosted populations. I think a reasonable conclusion is that there’s very substantial waning from the original efficacy without a booster, and that actual post-booster protection is somewhere in the 75%–90% range, with waning kicking in after 3–6 months.

2.  The other possibility would be creating some kind of super effective, cheap, safe, side effect-free, and mutation-resistant therapeutic that renders infection inconsequential. That’s probably harder, though.