Thread by @michaelzlin

A recent study by Novavax suggests it may be able to provide broader protection across variants than other vaccines so far. That has interesting implications for vaccine design, if true.

This seems to have escaped analysis entirely, so we'll take a look.

medrxiv.org/content/10.110…

The study came out on 12/25. It contains the data behind Novavax's 12/22 announcement that 3 shots of their vaccine provided Omicron protection. That got into the news, but nobody came back on Christmas to write about the actual findings (I wonder why).
ir.novavax.com/2021-12-22-Nov…

Sorry, Novavax, but you might benefit from hiring a science communication consultant for advice on publication strategy and writing. Besides the poor decision to release your article on 12/25, assuring no one will analyze it, the abstract is a disaster. Because there isn't one.

Abstracts are for stating interesting findings in plain English. And this study has very interesting findings: just 2 doses of the Novavax vaccine elicit anti-Omicron Abs at levels predictive of substantial case protection, and 3 doses elicit Abs predictive of >95% protection.

Instead squeezed between numbers and funding, we get an "interpretation" that omits everything interesting

"incremental increase in reactogenicity" - 🥱
"enhanced immune responses" - uninformative
"for both the prototype strain and all variants" - just say Delta and Omicron

Anyway, let's get to the good results. Volunteers were immunized with 2 shots of NVX-CoV2373 3 weeks apart. Abs were measured 2wk and 6mo later. A booster was given at 6mo, then Abs measured 1mo later. There's waning, but the booster brings Ab levels back.

L/R: Abs/nAbs

That's expected. What's interesting are the levels of antibodies against each of the variants after 2 shots and after 3.

While Abs recognizing Omicron are lower than other variants, the difference is not that big. Anti-Omicron Abs seem higher at both times than other vaccines.

nAb levels look slightly worse; some of the vaccinated after 2 doses with lower total Abs don't show neutralization. But there is still some activity to all variants including Omicron. Would be great to have some numbers, right? Oh, they're in a separate table. Inconvenient.

They should have taken a few minutes to write the numbers on top of the chart, just as the de Vries paper I analyzed yesterday did

medrxiv.org/content/10.110…

So while I should be working on my paid job instead, I'll just do Novavax's job, and label the chart.

And fold increase is uninformative if you're not going to specify the denominator, and unnecessary if you specify the numerator, so let's just cross that out.

So much nicer now

So now we can ask what does it mean? I like to refer to my all-time favorite paper of the pandemic, that correlates nAb levels to % case protection. It was an early paper that has held up really well (double rarity). nature.com/articles/s4159…

So Novavax 2 doses protected against ancestral with GMT = 120 here. This corresponds to 4x of convalescent sera (CS) on the second graph. Then anti-Omicron GMT of 15 equates to 0.5 of CS, which in turn matches... J&J (Ad26CoV2S) nAb levels against ancestral, which had VE of 66%!

To repeat: 2 doses Novavax, made with ancestral spike protein (which as 95% efficacy in preventing ancestral cases) may have 66% VE against Omicron cases.

So ancestral Novavax against Omicron may be as efficacious as ancestral J&J's against ancestral!

The other vax are far below. In de Vries, you can see anti-Omicron nAb levels are 20/15/58/26 after "full" AZ/JJ/Mod/Pfizer. These are all well below the 163 for JJ anti-ancestral nAbs, which again gives 66% VE to ancestral. (Note panels B vs C reconfirms the Nat Med paper too)

Things get even better after a 3rd shot of Novavax. There the levels of anti-Omicron nAbs are 214; that's even higher than the 120 for anti-ancestral nAbs that gave 95% VE. So you might predict 3 doses of Novavax will give >95% VE against Omicron.

The "interpretation" said "For... all variants evaluated, immune responses following the booster were notably higher than those associated with high levels of efficacy in phase 3 studies"

Much too vague. Just say: nAb levels against Omicron exceeded those associated with 95% VE

Again we don't see this breadth with the other vaccines even after a booster. From the de Vries paper you can that, after 3xRNA or JJ+RNA, anti-Omicron nAb levels are an order of magnitude lower than against all other strains.

So what does it mean for vaccine design? Omicron evading antibodies produced by most vaccines is no surprise, because Omicron is mutated in most of the antibody recognition sites on spike. We knew it would be highly immunoevasive.

Yet Novavax appears different. How can that be?

We know there *are* regions in spike that can be targeted by antibodies and that are conserved across all strains. That's how the mAb sotrovimab works. In fact it was isolated from a COVID19 patient and chosen precisely because it binds a site conserved from SARSCoV2 to SARSCoV1!

It's been an open question whether we can make a vaccine that can elicit a significant response to that epitope. Perhaps that is what Novavax is doing. And why might it be doing that?

This is pure speculation, but it could be that Novavax's format as a protein nanoparticle or the adjuvant used promotes spike uptake and antigen presentation by APCs (DCs or macrophages) differently than when spike is expressed in muscle cells by the other vaccines.

Perhaps this different route of antigen uptake leads to a different propensity for conserved sequences to be presented, and that leads to a set of antibodies that is more broadly reactive across strains.

So just a hypothesis for now, but it's important to figure this out. If it is indeed the case that this format presents a broader array of epitopes, then that would suggest Novavax may be more robust to unforeseen variants, in which case it should perhaps be prioritized.

Wow, turns out Novavax also worked on Omicron in a true virus neutralization assay. This was in their 12/22 investor call (h/t @peteyreplies), but it's not reported in the paper (should be).

As above, after 2 shots, Omicron nAbs only 4x lower vs original. Similar after 3 shots.

This addresses a valid concern raised in a few responses, that the paper only reported a hACE2 binding inhibition assay to assess anti-Omicron neutralization ability. I described it as nAbs for short, since Abs that inhibit hACE2 binding should neutralize.

Still, hACE2 binding inhibition might not have been the same as a neutralization assay using real virus, as the outputs of two assays may not be related in a scalar way. So this result with real virus is useful to confirm the effect.

If anything, Omicron neutralization after 2 shots is better in the virus assay than in the hACE2 binding inhibition assay, as it's only 4x worse than original/prototype strain compared to 15x worse. The 4x was stated by the presenter (slides and audio at ir.novavax.com/events)

Thus the data suggest that Novavax might induce broader nAbs than other vax. For example, it seems as good on Omicron as Delta, and this was also not seen for other vax (e.g. de Vries paper above).

But it's speculative of course until we get clinical data.

And Novavax was approved in UK and Germany today. Besides a potentially broader antibody response, Novavax has the advantage of being stable in the refrigerator, so should be easier to distribute as well (important for outside US and EU).
theguardian.com/society/2022/f…