February 3, 2021

At the moment, we are getting an astonishing insight into the action of evolution by natural selection. It is operating in just the way that Charles Darwin would recognise, especially if he was given a copy of Mendel’s paper about the peas; but it is happening at a speed which means we can see it happening. Instead of species changing over decades or centuries or millennia, it’s happening in viruses in days and weeks.

In the UK, we are now suddenly very worried about the “South African variant1”, or B.1.351; it is a different thing to the “UK variant”, B.1.1.7, which emerged late last year and which I wrote about over Christmas.

Both the South African strain and the UK strain have several mutations which make them genetically distinct from the earlier SARS-Cov2 virus. But they share one particular mutation, called N501Y. That seems to be what makes the UK one more transmissible – around 50% more likely to jump from one person to another – and the indications are that the South African strain is similarly effective. (There’s an argument over whether it’s more deadly, as well: as far as I understand it, the evidence is too unclear to draw any real conclusions – but as Adam Kucharski points out, being more transmissible is a far bigger problem than being more deadly anyway.)

What has people alarmed about the South African strain, though, is that various vaccines, and presumably pre-existing immunity from earlier infection, seem to be less effective against it. It has another mutation, E484K, which changes the shape of the spike protein; that appears to be what is driving it. All existing vaccines use the spike protein to prime the immune system to be ready for the real virus when it arrives.

That’s why it’s bad news that several cases of B.1.351 which couldn’t be traced back to South Africa have been found in Britain, suggesting community transmission. 

The Government is rolling out mass testing in nine postcodes around England — which, for avoidance of doubt, is good and necessary. I also wish that we had done more to establish rapid testing and hotel quarantine at borders weeks ago, as Yvette Cooper MP pointed out in a Liaison Committee meeting. But the scary thing, in a way, is not the South African variant itself at all. The scary thing is that the E484K mutation has arisen – twice, independently – in the UK strain as well. Ewan Birney, the deputy director-general of the European Molecular Biology Laboratory, called this “spooky”.

There’s a thing in biology called “convergent evolution”, where evolution hits upon similar solutions to a problem. The eye has evolved several times independently, because it is useful for lots of different kinds of animals to be able to detect light and use it to navigate the world. The common ancestor of birds and bats was flightless, but both evolved wings where their forelimbs used to be, because flight is useful for some lifestyles. When, in evolution, you see two species independently coming up with some similar systems, it’s usually evidence that that system is a solution to some evolutionary problem, not just a random quirk.

So the fact that the E484K mutation has appeared three times now, always so far in conjunction with the N501Y mutation, suggests that it is, indeed, evolutionarily useful. I’ve seen one suggestion that the mutation in the spike protein helps it “grip” the ACE2 receptor on the outside of human cells; that would be one possible evolutionary advantage it could bestow.

The other possibility is that the E484K mutation is just better at infecting people who already have some immunity, and that gives it an advantage.

For the record, it does seem to be better at infecting people who already have immunity. Again, the data is quite sparse, but it seems that those 90% effective vaccines drop down to 70% or so when pitted against the South African strain. Novavax goes down from 86% to 60%. Johnson & Johnson goes from 72% to 57%. The Ox/AZ, Pfizer and other vaccines will probably have similar numbers. 

It’s worth noting that the “efficacy” of a vaccine means how good it is at preventing symptomatic disease. It appears that they are still extremely effective at stopping severe disease, hospitalisations and death, although the data is much more sparse on those things simply because they are much rarer outcomes.

In the UK, about 15% have had the vaccine and perhaps a similar number have had the disease itself. If the E484K-mutant forms of the virus find it significantly easier to infect those people, then those new forms will have an evolutionary advantage over the older strains. At the moment this advantage is quite weak, depending on how much protection against transmission you get from immunity, but it will get stronger. And if the mutation happens to arise again, which it almost certainly will since viruses mutate all the time, then it will tend to spread.

Of course, “how much protection against transmission you get from immunity” is a key question. You might naively think – I certainly did – that if immunity reduces your viral load and prevents your symptoms, it’s a safe bet that it stops you from spreading the disease. But, as Dr Al Edwards, a professor in biomedical technology at the University of Reading, points out, it’s not as simple as that. Lots of viruses (including cold coronaviruses) spread very easily without symptoms, and you could reduce viral load an awful lot while still having many billions of virus particles in your system.

Annoyingly, there isn’t much data on the impact of vaccines on transmissibility. Most of the studies only looked at symptomatic disease, and got people to take a PCR test only if they showed symptoms. But the Oxford study has been getting people to do a swab every week (I know, I’m on it and it’s been really quite unpleasant). Early data from their trial suggests that two jabs reduces transmission by about 67%.

If that’s indicative, which I hope it is, then there is going to be significant and growing selection pressure on the virus to find ways around the vaccine immunity. The E484K mutation appears to be one such way.

That makes it all sound quite scary. I shall now give some reasons to be less scared. First, it’s only somewhat better at evading immunity: being vaccinated still gives you significant protection, especially (it seems) against severe disease. “We expect the potency to drop a bit with the new variants,” says Edwards, “but we don’t expect people to be completely immunologically naive.” Your body will still have a much improved chance of churning out a response quickly enough to stop you from getting really ill.

The other positive thing is that, having made all these vaccines, it’s relatively easy to simply make new ones. In the case of the mRNA vaccines, that’s especially true – they’re extremely plug-and-play, just put a new strand of RNA in and off they go. But, says Edwards, even the adenovirus ones like Ox/AZ are relatively easy to repurpose.

Of course, we’ll need to get better at testing them and licensing them. The Moderna vaccine was ready to go within a few days of the SARS-Cov2 genome being sequenced last January, but it took 11 months to be declared safe and effective. If we want new vaccines ready quickly in response to new strains, we’ll need to be a lot more nimble. The idea of human challenge trials has somewhat dropped off the radar lately, but it might be worth considering doing those in young, healthy volunteers each time you need a new version of a vaccine, because you can get safety and efficacy results so much faster if you don’t have to wait for people to get the disease naturally.

There’s another thing to consider. There are two things that drive evolution: selection and mutation. We’ve just been talking about selection pressures: that is, if there’s some mutation which is better at transmitting, or better at evading immunity, then it will tend to become more common in the population, because each virus will on average have more offspring than its non-mutant rivals. (That’s why the B.1.1.7 virus is so widespread in the UK now.)

But selection needs mutation to work upon. It might be that lions would be even better predators if they had wings and breathed fire, but if such a mutation never arises (which it won’t) then we’ll never find out. 

The way to keep the number of mutations down is to keep the number of cases down. It’s fairly linear: if you have twice as many infected people, all else being equal, you have about twice as many chances for a virus throwing up some dangerous new mutation.

(That is around the whole world, by the way. A new variant arising in Delhi or Rio de Janeiro will be here pretty quickly. We need to vaccinate everyone, everywhere, as quickly as possible.)

So that means that there is still a very strong case for keeping cases low, even as the deaths and hospitalisations drop. It will be enormously tempting to open up society again once the most vulnerable are vaccinated, but that will mean providing a huge opportunity for the virus to spread and mutate, while also piling the selection pressure on so those mutations will tend to become widespread. We’ll essentially be creating a perfect petri dish for mutant versions.

That might be a price worth paying; perhaps the economic gains (and therefore the real-life gains) will be sufficient to offset the risks, especially given the possibility of making new vaccines quickly. But in general, the lesson of this pandemic seems to have been that more relaxed restrictions backfire. And this government doesn’t have a great track record in actually thinking about the costs and benefits: it has a tendency to just put off eating the shit sandwich as long as possible. 

Testing for the “South African variant” (which I assume, and scientists seem to agree, means not just literally the South African ones, but any with the E484K mutation) is a good thing. Testing and quarantining at the border is a good thing, and keeping cases low is important. But this probably won’t be the last time we see the E484K mutation, or others like it, which help the virus partially escape immunity. Charles Darwin would understand.

  1. I’m not going to use the word “variant” any more, because, annoyingly, scientists use it to refer to two different things. Sometimes they use it to mean all the descendants of a particular virus further back in history. But sometimes they mean it to refer to a particular mutation. Instead, I’m going to refer to “strains” or “lineages”, and “mutations” or “changes”.