Several days after making some headlines with a press release about the data, the Novavax vaccine effort has published on Medrxiv. The is the first look at human data we have at an approach using recombinant coronavirus proteins (plus an adjuvant, in this case a proprietary saponin natural product). The protein itself is produced in the good ol’ Sf9/ baculovirus cell expression system, a real workhorse over the years for recombinant proteins. I’m glad to see this one, because we need as many different techniques as we can get. So how well does it work?
This was done in 131 adults, with two injections three weeks apart. These were either 5 micrograms or 25 micrograms of protein, with and without the company’s proprietary adjuvant. As seen in some of the other trials, there were six patients in a “sentinal” group who underwent the first dosing with observation before the rest of the groups were vaccinated (five groups, 25/group, with a placebo cohort). The antigen itself is the full-length Spike protein, with the furin protease site mutated and two proline substitutions introduced to stabilize the conformation. That’s pretty much the protein that the J&J vaccine is causing cells to produce via its adenovirus vector, so in this case it’s just being injected directly.
The reactions to the vaccine itself look fine – the usual soreness at the injection point, and a couple of reports of mild transient fever. These were more noticeable in the adjuvant group, which basically tells you that the adjuvant is probably just doing its immunogenic thing. Overall, there seem to be no overt safety signs to cause any concern.
And indeed, the adjuvant does really help a great deal. At Day 21 (before the second injection), antibody levels against the Spike protein antigen were tenfold higher in the adjuvant group as opposed to the plain injections. The response was also stronger after the second injection, and by 14 days after that one (Day 35 overall) the antibody levels for the adjuvant groups were at least 100x over the non-adjuvant ones. So that settles that! Interestingly, the levels were quite similar for the 5 microgram and 25 microgram groups, so that should help with overall vaccine manufacturing as well.
The effect on neutralizing antibodies was even more striking – it’s adjuvant or bust when measured that way (which is really the key figure, anyway). After the second vaccination, the preprint reports that neutralizing antibodies were 4x those seen in convalescent serum of outpatient-treated coronavirus cases, and spanning the same range as hospitalized convalescent cases. 16 patients were selected at random from the treatment groups to check T-cell response, and these showed CD4+ cells that were heavily biased towards Th1. We don’t know how long either response lasts – this paper covers out to two weeks past the second dose, but you can be sure that these numbers are being collected.
These look like strong results, and I’m glad that this candidate is in human efficacy trials. That’s something to emphasize – we’re all (naturally enough) trying to make what calls we can based on the Phase I immunogenicity data and the non-human-primate challenge experiments. But what matters is real human data from out in the field via the Phase II/III clinical trials. Right now, we have several vaccines that look like they will have good chances of working (this one very much among them). And we’re going to sort them out the only way that they can be sorted.