With all of my posts here about vaccines and antibodies, it’s only fair that I spend some time talking about the potential toxicology problems with both of those. I remain optimistic about both categories, but it’s for sure that not all the candidates being advanced are going to make it through. So what sorts of things go wrong?
Well, efficacy, for one: some of them just don’t work so well against their targets. Efficacy is straightforward – well, relatively straightforward. Some of these antibodies and vaccines are just not going to work well enough, either on the absolute scale of not providing enough protection, or on the relative scale of not doing as much as other candidates. This is what Phase II and Phase III trials are for. We’re going to need to know how well these therapies protect people against coronavirus infection, and in the case of the monoclonal antibodies, we’ll also need to know how useful they are when given to people who are already infected.
You know all the arguing that has been going on about the various antiviral ideas, the repurposed drugs and so on? Most of those are arguments about efficacy. And as we’ve seen, even though this is an answerable question, it’s not answerable without a lot of work. We’re going to need to know a new treatment’s effects in various age groups, across male and female patients, people with pre-existing conditions (and those taking various other forms of treatment for them), and so on. What happens if you’re over 70? Or if you have hypertension or asthma? Or are taking one of the therapies for rheumatoid arthritis, for example, that alters immune function? There’s a long list of such questions, and here’s hoping that the upcoming trials will be able to provide real answers to as many of them as possible.
I would not expect all the candidates being tested to be equally efficacious (see below). There’s no way that’s happening. But I don’t expect all of us handicappers to even be able pick the ones a priori that will work the best. Have you ever seen an NCAA basketball tournament play out exactly according to the seeding? Exactly. No, the only way we’re going to sort out efficacy is by clinical results, lots of them, obtained in as well-controlled a way as we can possibly manage. Predictions aren’t worth much at this point. I only hope the picture is reasonably clear once we’ve gotten to the end of the trials, although there is of course no guarantee of that, either.
What about safety? With the repurposed drugs that we’ve been seeing, safety has been overall less of a concern, since they already have a history in human patients. Now, whether that’s always enough, when you’re giving such drugs to people are are ill with a virus that we’re never encountered before, that’s a real question (and is part of the argument about hydroxychloroquine). But some of the things that are coming have never been into people at all. As with any investigational drug, we will have to cross our fingers, taking our best shots based on all the antibody work and vaccine development that has come before.
What does that history tell us to look out for? First, let’s talk about the monoclonal antibodies, where there’s one thing that we have going for us right at the start, in that we are not targeting a human protein. You may have noticed the advertisements for monoclonal antibody drugs tend to list a lot of side effects, but that’s because of their targets. The highest-profile mAb drugs are things modulating immune response and the cell cycle, which are never going to be touched without risk. Targeting any human cell proteins comes with risks, honestly – you can occasionally set off a cytokine-storm response (cytokine release syndrome, CRS) when this process goes awry. (This and some other problems are associated not with the variable target-binding region of the antibody, but with the more constant Fc part of the molecule). But when you’re trying to shut down a pathogen’s protein, well, it’s more full speed ahead.
That said, one thing you want to make sure of is that the antibodies you are administering attack the coronavirus protein and only the coronavirus protein – you don’t want to find yourself hitting a human cell-surface protein without realizing it. There are plenty of autoimmune examples where an inappropriate response is raised to a bystander protein (more on this later in the vaccine section), and you don’t want to jump-start such a thing with a nonselective antibody. There are other immune responses, lower than outright CRS, that you want to look out for, such as hypersensitivity (which can be immediate or delayed, site-of-injection or more general). The good news here is that the mAbs that are being tested are based on antibodies raised by recovered human patients, so that tells you that these were able to clear the coronavirus without apparently setting off anything else. That’s the biggest factor in the speed with which these new agents are moving into human trials – in many important ways, these antibodies have already been in people. Overall, you would expect some sort hypersensitivity, not a threatening level, is going to probably be the most common adverse event.
But the bad news is that human immune systems vary so widely that you still need to be alert for trouble – that, as you’ll see, is a general theme in all this. Such events are probably going to be low-frequency, but potentially severe, which is just what you don’t want in a clinical development program. Human dosing of the mAbs and the vaccines is going to be wide, fast, and jumpy – accelerated clinical trials with a huge number of patients, but dosed with great attention to any potential adverse events. We’re going to sort these things out by efficacy, of course, but it’s quite possible that we end up with more than one similarly-efficacious therapy and end up ranking them by the incidence of rare side effects.
So let’s talk about the vaccines, then. In this case, we don’t have the already-in-humans advantage that the mAbs have, or at least not to the same degree. What we do know – and it’s no small thing – is that it is possible to raise a useful immune response to the coronavirus, antibodies and T cells, which means that we just have to recapitulate that without the disease itself. That’s as opposed, say, to viral diseases like hepatitis C (or rabies!) where disease-clearing immune responses basically just don’t happen, so that’s good.
Note that we’re going to be trying to raise this through several rather different mechanisms, although they all get to the same place in the end: an immune response which translates to protection against the real viral infection. I’ve done more detailed vaccine posts, but for reference, we have in one group administration of individual viral proteins, or of inactivated (“killed”) virus, or of viral-like particles. You give those direcly and let them set off the immune system. In the other group, you’re looking at co-opting the body’s protein production machinery to (in effect) dose yourself with such proteins. That includes weakened forms of the coronavirus, engineered versions of totally different infectious viruses carrying genetic material for coronavirus proteins rather than their original make-more-virus payloads, or administration of DNA or RNA for those to be taken up by cells and set off such protein production from that direction.
As mentioned, there will be efficacy differences between the different vaccine candidates, and these are just the sorts of things you look out for in a more “normal” development pathway. Some of them may require larger doses than others (potential manufacturing issues there), and some may require more shots and at different intervals, if so (logistic and manufacturing problems both). Some will have different immunological kinetics (the time it takes to develop a protective response after dosing), and some of them will produce longer-lasting response than others. (It should almost go without saying that that last one is something we’re just not going to know; there are no good ways to predict it). They may behave differently in different populations – different antibody and/or T-cell profiles with younger or older patients, among other things. There are plenty of things to sort out! A big advance that might come out of the Phase II trials (and associated studies) is a marker that can be used going forward, an immune correlate of protection. You would want to be able to say “OK, you have raised a titer of at least X of this particular type of antibody, so that means that yeah, you’re protected”. We don’t have that now (not enough data), but the hope is that we can get closer to that for the Phase III studies.
Past efficacy and on to adverse effects, one thing that has been extensively talked about is antibody-dependent enhancement (ADE), because this has been seen in other vaccine candidates (such as for dengue, where the reasons don’t really apply here) but also in attempts to develop vaccines against the closely-related SARS coronavirus. To be honest, I am getting less and less worried about that as time goes on. The various mAb and vaccine teams have been alert for signs of ADE all along the way, and so far we appear to be in the clear. This doesn’t mean that we can forget about the issue, but it’s not at the top of the worry list, either.
Those earlier worries about hypersensitivity and inappropriate immune responses do apply, though. The classic example intersecting with infectious disease therapy is Guillain-Barré syndrome (GBS). This can be set off after a (sometimes unnoticed) bacterial or viral infection, and involves an immune attack on the myelin sheaths of the nervous system. That’s obviously very bad indeed – although most patients recover, not all of them do, and some of them need intensive care while the myelin damage gets reversed. The exact mechanism of GBS has eluded discovery, despite intensive research – there’s obviously something about the myelin sheaths that are vulnerable to immunologic misidentification, but the details aren’t clear, and the vulnerability varies according to a person’s individual immune system in ways that we cannot yet screen for. So such things will be watched for very carefully indeed.
Even with that, we will not know the real risks of such rare events, because they’re rare. GBS, for example, might occur on the order of one or two people out of a million with the seasonal flu vaccine. You’re only going to see things like that after you’re out in a very large patient population. It’s just not possible to assess something like that in even a rather large clinical trial involving tens of thousands of people. That’s not just the situation with these new coronavirus vaccine candidates; that’s how it is in general. But in this case we’re talking about vaccines that may well be going into hundreds of millions of people as quickly as we can roll them out. If one vaccine candidate gives GBS to 100 people out of a hundred million dosed and another one affects only ten, there is absolutely no way that we will be able to be sure of that before dosing the hundred million people.
I do not want to provide ammunition for the anti-vaccine camp by going into these details, but we can’t ignore medical (and mathematical) reality. All drugs have side effects, and every therapy is a tradeoff. The levels that I’m talking about for something like Guillain-Barré (which is generally not fatal) are, in fact, completely acceptable for preventing a disease like this one. As you see from that link in the paragraph above, they’re acceptable for something like seasonal influenza, which is definitely not as big a problem as the COVID-19 epidemic. State and national governments have already been talking overtly about how many coronavirus deaths are acceptable to avoid shutdowns that harm their economies, and if you’re willing to make that tradeoff, you should jump at the chance to make this one. It’s a much better deal.
This is one of those times when it’s instructive to do the math, as they say. There are (for example) outright fatal immune responses to some drugs (not just vaccines), which are generally very difficult to predict. Think penicillin as the famous example, although there are others. But as shown here, the death rates for these across the population are in the same range as death by lightning bolts (one out of a few million). Now, one should reduce one’s risk of being hit by lightning by not being out in the open during severe thunderstorms, and one should reduce the risk of harming people with a vaccine through careful clinical testing. But (to put it gently) many populations are not showing themselves to be very good at risk reduction these days. I cannot help but picture a bunch of unmasked people in a crowded bar yelling at each other over the music about how they’re not going to let themselves be poisoned by any damned vaccine.
All the issues mentioned are about to become very real as the various antibodies and vaccines move into larger patient populations. Keeping track of everything will not be easy, but, well, here we go!