There’s a lot of interesting antibody news to catch up on, from the early science to clinical trials. A previous post on this subject is here, with links to earlier background explanations, if you would like to catch up on the area. Here, for example, is a report from a large multi-center team in China characterizing two monoclonal antibodies against the coronavirus, both of which were derived from the B cells of recovered patients. The more potent of the two (CB6) was shown to bind to the RBD (receptor-binding domain) of the spike protein, in keeping with many other antibodies that have been described so far. That is, of course, the region of the Spike that recognizes the human ACE2 protein, which is the first step to viral entry and infection.
That CB6 antibody was evaluated in a rhesus monkey model of coronavirus infection – administration of the mAb at day 1 and 3 post-infection demonstrated a 1000x lowering of virus titer in the dosed animals compared to controls, which is good to see. A single dose pre-exposure was even more effective (see the preprint’s Figure 3a for more). On the other hand, the animals were still infected – this wasn’t sterilizing immunity that was conferred – and the dose of antibody was 50 mg/kg. That’s pretty high by therapeutic mAb standards – see, for example, Table 3 in this review of monoclonal antibody pharmacokinetics, where you’ll see that more typical doses are around 10 mg/kg. But near the end of that table, you’ll see Raxibacumab, which is a therapy against the anthrax toxin and dosed at 40 mg/kg. Then again, there’s also pavilizumab, a mAb used against RSV infection in infants and dosed at 15 mg/kg, so we’re going to have to watch the development of these to see what the dosages (and thus the manufacturing load) will be like. (See below for more on CB6).
And here’s another large effort from a group of Chinese institutions, characterizing 206 antibodies from eight recovered patients. These also show a very strong trend towards binding to the RBD and competing with ACE2 binding, and in fact that competition correlated with the activity of the antibodies in an in vitro assay. The competition assay was better at predicting activity as compared to sheer affinity. Note, though, as in that previous post details, that there are some fine details – not all the RBD-binding antibodies prevent the virus from binding to the human ACE2 protein, although they may well prevent the next step after that binding. This group obtained a crystal structure for the binding of one of their antibodies (P2B-2F6) and a Spike protein construct, and it does indeed keep ACE2 binding from taking place.
The eight patients studied all raised a different mixture of antibodies, as you would expect – there’s not much more of an individual characteristic that you can find than that. Some of the antibodies had apparently only shown up once during the response, while others had undergone a lot of clonal expansion – and among these, some of them had remained very close to their original form, while others had diverged into a set of related species. This means – and this is no surprise – that giving people “convalescent plasma” treatment from recovered coronavirus patients will necessarily be a variable sort of treatment, since every batch from every different donor will be a mixture of antibodies all its own.
And we’re starting to get more data about that mode of therapy as well. Here’s a study in 103 patients from another large multicenter effort in China. Unfortunately, that’s only about half the patients that they were trying to enroll – the pandemic diminished during the work – and the final study ended up statistically underpowered. There were trends toward improvement in the treated patients, which were more noticeable in the severely ill ones versus the even-worse-off critically ill ones. The other bit of encouraging news was that only two adverse events were reported, with no strong signs of antibody-dependent enhancement. But this is still the most well-controlled trial that we have; everything else is even fuzzier. Things are pointing in the right direction, but we need to know more.
We’ll be getting data from the more-easily-characterized monoclonal antibody trials, though, which have now begun. Eli Lilly and their partner Abcellera announced last week that they had started dosing human patients. Junshi Biosciences have also announced that they’ve begun human trials of their JS016, which from that press release appears to be the same CB6 antibody in that Nature paper described in the first paragraph above. They’re also working with Lilly outside of China. Regeneron’s first coronavirus mAb is just about to go into human patients (update: corrected this because they haven’t started yet). (Update 2, June 11: they have now!) Meanwhile, Vir Biotechnology (working with GSK) published on their own mAb work a while ago (in coronavirus time!) and are expected to begin human trials very shortly as well.
So there’s going to be quite a pileup in trial results soon. But the worry is that there will also be a pileup in the manufacturing capacity. Here’s an article at BioCentury about comments made by the heads of both Vir and Regeneron about that issue, and they’re warning that there may not be enough capacity anywhere to deliver the amounts of these agents that people expect. mAb manufacturing is nontrivial by any of the industrial routes, and by now all the people who can do it have been signed up by one player or another. Just as with vaccine production and rollout, we’re looking at some hard decisions later this year that are going to have to be taken relatively quickly (and on relatively thin data).
Which candidates will look the most successful? Keep in mind that success has several parts – efficacy against the virus, first of all, but also number of doses needed and the total amount of the actual agent that is being dosed. That, as mentioned above, is directly tied to manufacturing capacity. We could end up having to go with a slightly less efficacious treatment that can treat many more patients (and potential patients) versus a better one that has to be given at (say) five times the dose and would thus exceed the human race’s current capacity to produce it. Or versus a better-looking one (remember, these will be on limited data sets!) whose mode of production is just intrinsically more difficult. There will also be considerations about storage and shipment, stability of the production method, and many more: it’s going to be a hard call with a lot of variables involved.
Get ready for it later this summer and this fall, because this could get messy. There are surely going to be a lot of twists and turns, sudden reversals and surprises, and we should brace ourselves for the white water ahead. I think we’re going to come out of it with some real therapies – I really am optimistic about that – but there’s almost no way that it’s going to be a smooth process. . .