Thread by @michaelzlin

Antibodies elicited by the Pfizer/BioNTech RNA vax are 6x less potent in neutralizing B.1.617 (delta variant that devastated India).

This lowers predicted efficacy for symptomatic COVID19 to ~70%.

This compares to 2x drop in neutralizing activity for Covaxin.

I discuss in 🧵

https://twitter.com/TheLancet/status/1400581563314757633

The advantage of the RNA vax is their simplicity. They encode only the spike protein because it's the viral protein targeted by most of the antibodies that prevent viral entry. Another way to put it is Abs to the right place on spike are sufficient to block entry. (image Siemens)

The adenovirus vax also encode only spike but express it from Ad. Both types elicit high levels of neutralizing antibodies (nAbs) after just one dose. RNA is better here; in fact 1 dose of RNA gives you similar nAb levels to natural infection.

https://twitter.com/michaelzlin/status/1396222629430714371

But immunizing just to spike limits the nAb repertoire. There are only a few places where Abs tend to bind to spike, and it seems binding to a loop containing E484 actually accounts for much neutralizing activity, as E484K drops neutralizing activity 3x.

thelancet.com/journals/lanmi…

The delta variant has both E484K to give that 3x lower neutralization, and another mutation P681R that is believed to enhance membrane fusion and thus the rate of viral entry. Overall neutralization activity of antisera is down ~6x (today's Lancet paper)

A landmark paper in May related nAb levels to % efficacy in trials (in terms of % protection from symptomatic disease). Using their efficacy-vs-nAb curve, and with Ab levels at 3x of convalescent sera, the 6x drop in nAb potency predicts efficacy at ~70%.
doi.org/10.1038/s41591…

The Chile study I posted above measured nAb levels post-Pfizer/BioNTech at 10x convalescent sera, so it's not clear if 3x or 10x is more common. If 10x, then efficacy would be predicted to remain >85% even with a 6-fold loss of potency against the delta strain.

And now this is rather preliminary and verification is required, but an April preprint reported that nAbs elicited by Covaxin, the whole-killed/inactivated virus vaccine, showed only a 2x dropoff in potency against delta.

biorxiv.org/content/10.110…

That's potentially interesting because whole killed/inactivated virus vaccines present the entire virus. In addition to spike, there are two other surface proteins, E and M. Anti-spike is certainly sufficient for neutralization, but perhaps anti-E and anti-S can contribute too.

So, to wildly speculate: in whole-virus vaccines, maybe you get a mix of the anti-spike (predominantly anti-E484), anti-E, and anti-M. Then if you lose the spike binding with E484K, you still get some neutralizing activity through E and M. Requires more study to know for sure.

The other thing that whole-virus vaccines present are the intracellular viral antigens. Dendritic cells (DCs) route some phagocytosed antigens to MHC-I via cross-presentation, so they can activate CD8 T cells with TCRs that recognize those viral antigens.
en.wikipedia.org/wiki/Cross-pre…

Intracellular viral antigens presented on MHC-I by infected cells then induce killing by these activated cytotoxic CD8+ T cells (CTLs). Dogma is that DC cross-presentation is not reliable enough as a vaccine mechanism, but this is not always true.
journals.asm.org/doi/full/10.11…

Now whole-virus vaccines such as Coronavac, Sinovac, and Covaxin do seem to have very little activity after the first dose, in contrast to RNA and Ad vaccines. Immune activation is just not as strong with dead virus and adjuvant as RNA or Ad.

https://mobile.twitter.com/michaelzlin/status/1396222629430714371

However after dose 2, the whole-virus vax appear similar to convalescent sera. If they can maintain 50% neutralizing potency on new variants, then after dose 2, they will still be able to prevent symptoms by >50% and severe cases by >80%.

Adding a correction to main thread: Delta/B.1.617.2 has T478K; Kappa/B.1.617.1 has E484K.

They are on the same loop and both resist neutralizaing antibodies elicited to wild-type spike, so analyses and implications are unchanged.

Thx @Multi__X

nature.com/articles/s4159…

Also the 2x hit on Covaxin-elicited neutralization mentioned above was with Kappa/B.1.617.1. So for comparison we should look at RNA vaccine-elicited nAbs to B.1.617.1. They are down 7x.

We don't know Covaxin result on B.1.617.2; will be important to get

biorxiv.org/content/10.110…

To add a little background, anti-M antibodies are seen after coronavirus infections. Anti-E antibodies have not been seen so far.
nature.com/articles/s4146…

Meant "perhaps anti-E and anti-M can contribute too." Some evidence for anti-M.

In trials the Pfizer/BioNTech vax was estimated to be 88% effective against Delta in UK. That's very good news, and suggests the vax elicits very high nAb levels and/or T cell responses so that the 6x drop in nAb potency is not hindering efficacy too much

deccanherald.com/national/vacci…

Reference for the 88% efficacy finding (also in Lancet paper): medrxiv.org/content/10.110…

3a is another coronavirus transmembrane protein that, in SARSCoV1, was found to be targeted by neutralizing antibodies. It appears nobody has published anything on 3a of SARSCoV2 yet (S is clearly the right protein to target if you have to pick one)

febs.onlinelibrary.wiley.com/doi/full/10.10…