Thread by @michaelzlin

Finally, Sanofi and GSK announce their results. VE = 58%. Patient dosing and followup occurred during Delta and Omicron, so that seems fairly good, as best as we can resolve with the statistics (resolution not great, confidence interval 27-77%)

sanofi.com/en/media-room/…

You'll hear 2 more numbers in the press, but these aren't real statistics: 75% protection against moderate-severe disease and 100% protection against hospitalization. Again, not real stats, no CIs, due to low numbers, e.g. 0 vs 4! You might call those figures "anecdotal"

Some people never learn though. You might have thought by now the @NYTimes would train their headline writers to not repeat just the most optimistic unsupported take from the manufacturers. Speaking of COI... why do we let the entity who will directly profit write the headlines?

I've been waiting for these results to understand whether a simple protein + adjuvant vaccine, indeed a tried and true method, would be effective. In an ironic/annoying twist of fate, every effective vaccine up to now has incorporated some newfangled inconvenient technology.

Pfizer-BioNTech and Moderna use RNA which requires frozen storage, AZ and J&J use adenovirus which hasn't been used at a wide scale before (and we now know has a thrombocytopenic thrombosis issue), and Novavax puts their proteins into lipid nanoparticles that are hard to make.

Each of those manufacturers claimed that their technology was responsible for their high efficacy, and again the press dutifully amplified that claim. Indeed so far inactivated virus vax were less effective, and GSK-Sanofi's traditional protein vaccine initially didn't work.

However I've been a holdout on the scientific jury. The issue with the inactivated virus vax might be dosing or adjuvant, and indeed the GSK-Sanofi vax problem was apparently solved by a higher dose. I covered this earlier.

https://twitter.com/michaelzlin/status/1455230211692269568

Sanofi-GSK had seen that their initial dose produced inadequate immunity in >50yo. In general, immune function is lower with age, so a drop-off is expected. But perhaps the drop-off was just a bit too much for Sanofi-GSK compared to the other vaccines.

Today's release actually provides some nice data that speaks to the age effect. You can see below in the gray bars (after primary vax series) that Sanofi-GSK is closer to Pfizer than J&J or AZ (note log scale) for 18-55yo, but is worse than all other vax for ≥56yo

So perhaps Sanofi-GSK's naked protein+adjuvant efficacy is more sensitive to immune function than the other vax types. That would mean the approach works, but dosing needed optimization through trial and error. The other vax were lucky not to run into similar issues.

In the above graph, the gray bars were after primary vax and waning (tested on ancestral virus). It's unclear how long after vaccination those samples were taken, so we don't know what peak titers might be.

The same caveat applies to the fold increase after using Sanofi-GSK as booster. Those are large ratios but we don't know the baseline, so impossible to compare to other results.

Too bad vax makers have settled on fold boosting as a useful number. It's completely meaningless.

Novavax had peak antibody levels between Pfizer and Moderna using half the amount of antigen. If Sanofi-GSK peak is also below Pfizer peak (again, can't say for sure), that would suggest Novavax's adjuvant is better, or its nanoparticle helps, or both.

https://twitter.com/michaelzlin/status/1455230213776834564

So perhaps Sanofi-GSK is still underdosed, but I'm sure they have their reasons for not going higher.

Overall shows that while the science of vaccine-making is reliable in the long term, in the short term there is some luck and art to it.

Aha, headline changed. Both are from front page, 8:26 (left) vs 10:35 PT (right)