We now have a preprint with a great deal of data on the first mRNA coronavirus vaccine candidate from the Pfizer/BioNTech effort. This is actually the first real data set on any of the genetic vaccines, since Moderna’s paper on their Phase I trial has not yet appeared (all we had was a brief press release) and a brief press release is all we got from Inovio’s DNA vaccine work as well.
I’m very glad to see this – some of the other vaccine programs (Oxford/AZ and CanSino, for example) also have been very forthcoming with information, and it’s extremely important that all of the human trial data be made public. I realize that doing so opens the door for a lot of people who don’t understand human trial data at all to wander through it misinterpreting things – no doubt about it – but the dangers of that are far, far outweighed by the need for full disclosure. There are lots of qualified eyes out there, too, and for the public to have confidence in the vaccines that get approved we have to have everything out on the table with no room for suspicion. No one gets a pass: everyone publishes everything if they want to be taken seriously.
So what do we have today? It looks like good news. Recall that this effort actually has four different mRNA vaccine programs: two with modified RNA bases, one with extra uridine bases, and one in the “self-amplifying” category. This preprint covers the initial data on BNT162b1, which is one of the modified-base candidates (incorporating 1-methyl pseudouridine, which should cut down on the innate immune response attacking the mRNA vaccine itself and also increase protein production once it gets into the cell. And it encodes a trimer of the coronavirus Spike protein’s receptor-binding domain (RBD), the most common antigen that is being studied in all the vaccine programs worldwide. The trimerized variant uses a “foldon” protein scaffold that displays three of the RBDs simultaneously in a three-dimensional array – it’s a motif borrowed from a bacteriophage that has been used in vaccine production before.
In this study, 12 patients got 10 microgram doses at Day 1 and Day 21, 12 patients got two 30 microgram doses in the same way, and 12 got a single 100 microgram dose at the start. There were nine placebo-injection patients as a control, and a convalescent serum panel as a comparison group for antibody response. Male/female almost 50/50, ages from 19 to 54, mean 35 years old. No serious adverse reactions: one patient in the 100 microgram group reported severe pain at the injection site, and several others reported moderate pain or soreness after the first shot (by comparison, 2 of the 9 placebo patients reported the same). Pretty much everyone reported it after the second, which is to be expected. There were reports of headache, fatigue, and fever, which were dose-dependent and also more severe after the second dose, and in fact these are what limited the 100 microgram group to only one initial injection. This is also exactly what you expect from a vigorous immune response.
And that appears to be what the team got. The patients were profiled at Day 7, Day 21, Day 28, and Day 35 (with dosing, as mentioned, at Day 1 and Day 21). RBD-binding antibodies (IgG) were detected at Day 21 after the first dose, and were much stronger 7 days after the second dose (Day 28). In terms of just geometric-mean concentration of antibodies, the levels achieved after the first dose were 1.8x and 2.8x (in the 10µg and 30µg patients, respectively) the levels seen in the convalescent serum panel. After the second dose, these went up to 8x to 50x the convalescent levels (!) The antibody response was further profiled for titers of real neutralizing antibodies – at day 28 (seven days after the second dose) neutralizing antibodies were 1.8x and 2.8x (in the 10µg and 30µg patients, respectively) those seen in the convalescent patients.
That definitely sounds good, since the convalescent patients, remember, had already beaten the coronavirus infection with the levels they’re showing. One variable in this, which we’ll have to address, are the reports that antibody levels in such recovered patients may drop off faster than expected. You might want to try to standardize when you draw plasma from such a comparator group (compared to when they stop showing viral RNA, perhaps). As Matthew Herper noted, the antibody titers in the convalescent patients show a much wider spread than the treated patients from this study. Everyone will of course be monitoring the persistence of vaccine-induced antibodies as these trials go on as well, and we’ll get more data on the topic. Another thing that I hope we see plenty of data collection on in the trials is T-cell response – this recent preprint suggests that this might be an even more sensitive indicator than antibody titer, and that there are indeed people who develop such a response without ever really seroconverting (developing antibodies).
The FDA has published guidelines for such clinical trials, and they look reasonable. The gold standard will of course be protection against coronavirus infection versus a placebo control, and right behind that is the severity of disease in people who do get infected. Blood chemistry/immunology is not addressed in detail because we just don’t have enough predictive detail yet about antibody and T-cell responses (as this latest Pfizer/BioNTech manuscript also notes), but the guidance leaves open the possibility of approval on such markers if we get a clear enough picture later on. Any such surrogate endpoints will have to be tied directly to those clinical endpoints, of course.
But back to these latest results: I agree with the paper’s conclusion, which says that its findings “are encouraging and strongly support accelerated clinical development and at-risk manufacturing“. So far, so good, and remember, these folks have three more mRNA vaccines coming along simultaneously. I very much await Moderna’s paper on their Phase I mRNA results for comparison – it’s been six weeks since the press release, guys, an eternity in Covid-time. Bring on the data, everyone!