Thread by @michaelzlin

Was going to do the J&J news later, as it's actually the same info I presented yesterday, but need to correct some mistakes about it that are going around Twitter already.

Most importantly, J&J did *not* report its vaccine is 85% effective on Delta.

jnj.com/positive-new-d…

The press release was a bit confusing. In the section on Delta, it says "in the ENSEMBLE trial, Johnson & Johnson’s single-dose COVID-19 vaccine was 85 percent effective against severe/critical disease and demonstrated protection against hospitalization and death."

You might think that some trial demonstrated 85% efficacy against Delta, but that's not correct actually. ENSEMBLE was the trial finished in January that showed 85% efficacy against severe disease, back when Beta and Zeta were going around. That's mentioned in the next sentence.

What did happen is that J&J drew blood from 8 people from the completed ENSEMBLE trial, and tested their antibodies for neutralization of Delta virus. Those antibodies and T cells worked better on Delta than Beta. That's the new result.

This is the same info I discussed yesterday. Basically neutralization by J&J-elicited antibodies is better on Delta than on Beta. This is good news; I had expected the neutralization to be similar.

https://twitter.com/michaelzlin/status/1410449193215266822

The interesting thing we knew the J&J vaccine is that it gave the same result essentially on ancestral, Alpha, and Beta strains despite varying ability of antibodies raised against ancestral spike protein to recognize Alpha and Beta spike proteins.

That is, in people, efficacy was 64-72% vs moderate-severe disease and 82-86% vs severe disease across the 3 strains, even though antibody neutralization strength varied >4x. As Ab levels are not so high with this vax anyway, it's likely T cells provide most of the protection.

Yesterday I guessed that since vaccine-evoked Ab activity on Delta is in between ancestral and Beta, that predicted efficacy in people would be as well. So that puts predicted efficacy within that tiny range of 64-72% symptomatic 82-86% severe.

There is one caveat which I mentioned elsewhere. The in vitro assays where Delta did better than Beta use normalized amounts of virus, but it's believed Delta can bind to and enter cells more quickly, and thus can build up to higher levels.
nytimes.com/2021/06/12/wor…

The higher levels of Delta in the body wasn't seen with previous variants, so whether that impacts vaccine efficacy in people (as in the petri dish) is still unknown.

Another thing to keep in mind is that the way we measure vaccines is always relative to unvaccinated. That means a strain can create objectively worse outcomes by making both vaxxed and unvaxxed sicker, without affecting a vaccine's relative efficacy.

Now Delta nearly doubles the chance of getting ill in unvaxxed people (because R0 seems to be doubled vs ancestral strain). Delta could then actually double the chance of a vaxxed person getting sick without changing a measured vaccine efficacy number.

So if we still take 64-72% as the likely efficacy of J&J vaccine for symptomatic disease, but we also realize Delta is twice as contagious as before, we might still be concerned about the ability of vaccinated people to get sick and pass the disease on to close contacts.

Biorxiv paper up, h/t to @Eric_T1990
biorxiv.org/content/10.110…

Also relevant is this preprint from June 22, describing Delta transmission chains in HCWs who received the AZ vaccine. Delta replicated to 4x higher nasopharyngeal levels than prior variants, and breakthrough clusters were larger.

researchsquare.com/article/rs-637…