Our study "A highly virulent variant of HIV-1 circulating in the Netherlands" was published in Science in February 2022, here.
University of Oxford press release here.
Over 200 news outlets covered our finding. Some highlights: BBC World Service, AP, AFP (e.g. France24), Financial Times, NPR, NBC, Al Jazeera, The Times of India, Forbes, Nature, Scientific American, El País, Der Spiegel, UN News. UNAIDS echoed our statement that this finding emphasises the urgency of getting HIV treatment to millions of people.
We helped fact-checking teams debunk misinformation about this variant. Claims about links to SARS-CoV-2, COVID-19 or vaccines are completely false: this variant pre-dates them by about 20 years. Fact-checking by Reuters, AFP, AP, Full Fact, Euronews, Univision.
Twitter summaries here by principal investigator Christophe Fraser and here by lead analyst Chris Wymant. A webinar on our finding is here. Virtually the same webinar but with an introduction in French is here. Co-author François Blanquart wrote an accessible summary of finding in French here.
Main findings:
We discovered a variant of the HIV virus – the ‘VB variant’ – with higher virulence, i.e. having more damaging health impact than is normal for HIV.
The evolution that resulted in the VB variant took place during the late 1980s and 1990s in the Netherlands. It spread more quickly than other HIV variants during the 2000s, but its spread has been declining since around 2010.
Before starting treatment with antiretrovirals, individuals with the VB variant differ from individuals with other HIV variants. They have:
- a viral load (level of virus in the blood) 3.5 to 5.5 times higher;
- a CD4 cell decline – the hallmark of immune system damage by HIV – twice as fast, placing them at risk of much more rapidly developing AIDS;
- increased risk of transmitting the virus to others.
After starting treatment, individuals with the VB variant have similar immune system recovery and similar survival to individuals with other HIV variants. The VB variant does not have mutations that prevent successful treatment.
We identified 109 individuals with the VB variant, 107 of them in the Netherlands. The observed differences between these individuals and individuals with other HIV variants are likely caused by the virus, not by confounding differences between the individuals themselves.
Implications and interpretation:
Before this study, the genetics of the HIV virus were known to be relevant for virulence, implying that the evolution of a new variant could change its impact on health. Discovery of the VB variant demonstrated this, providing a rare example of the risk posed by viral virulence evolution.
Our finding emphasises the importance of World Health Organisation guidance that individuals at risk of acquiring HIV have access to regular testing to allow early diagnosis, followed by immediate treatment. This limits the amount of time HIV can damage an individual’s immune system and jeopardise their health. It also ensures that HIV is suppressed as quickly as possible, which prevents transmission to other individuals (“U=U”: undetectable equals untransmittable). These principles apply equally to the VB variant.
We consider that the evolution of this variant occurred in spite of widespread treatment in the Netherlands, not because of it. Treatment prevents transmission, and each averted transmission eliminates one opportunity for the virus to evolve into a more virulent form, as well as saving one person from HIV.
Studying exactly how this variant causes more immune system damage than other variants do could reveal new targets for next-generation drugs.
Press questions & answers with Chris Wymant:
What’s the background on HIV genetic variability?
HIV mutates so quickly that every individual has a virus which is different from everyone else’s, and indeed their virus changes over time. The large majority of these mutations make no difference. These different viruses can be grouped, in much the same way that a brother and sister have different DNA, but it’s very similar and we can group them into the same family. At a higher level we could group them into the same ethnicity. The highest-level grouping for HIV is into ‘subtypes’, and these are strongly correlated with geography. For example in Africa, subtypes A, C and D are most common; in Europe and the US, subtype B is most common. These differences between regions were established when HIV first began spreading round the world in earnest in the mid-twentieth century, and have changed little since.
What’s the background on current treatment?
Combination antiretroviral treatment for HIV, often called ART or simply treatment, is usually highly effective. For an individual on successful treatment, the deterioration of the immune system towards AIDS is stopped, and transmission of their virus to other individuals is stopped. Current treatments are typically in the form of a single pill taken daily - a simplification of the first treatment ‘cocktails’ to be used against HIV. Other formulations are being developed such as injections that last a long time; these could be helpful for people in infrequent contact with the health system, such as in remote rural communities in sub-Saharan Africa. Just as there is an arms race between the HIV virus and the immune system within any given individual - each continually trying to beat the other - at a higher level, medical researchers strive to develop new treatments that combat HIV, but HIV is evolving and may begin to change to escape from the effects of treatment. Many countries affected by HIV perform monitoring for signs of such drug-resistance evolution. The worst-case scenario would be the emergence of a variant that combines high virulence, high transmissibility, and resistance to treatment. The variant we discovered has only the first two of these properties.
What’s the background of the BEEHIVE study?
HIV affects individuals in a remarkably variable way: some progress to AIDS within months, while others do not progress after decades; some have viral loads (levels of virus) thousands of times higher than others. Research by our team and others before the BEEHIVE project established that this variability is partly due to the virus, not only due to people’s immune systems varying in their ability to fight the virus. The BEEHIVE project, begun in 2014, was created to understand how changes in the virus, encoded in its genetics, cause differences in disease. The project brings together data from seven national HIV cohorts in Europe plus one in Uganda.
(For many years previously, people examined only particular parts of HIV’s genetic sequence, in particular the part where drug resistance mutations are to be found. For BEEHIVE we wanted to examine the full genetic sequence of the virus, and to do this for many viruses. This was a new problem, and we needed to spend a few years developing new computational methods to make sense of this kind of data. Fortunately, these new methods have accelerated our other projects, such as the PANGEA project which is funded specifically to use African viral sequences to better understand which policy approaches will be most effective in preventing HIV transmission.)
Pursuing the BEEHIVE project’s primary aim, we identified a large number of mutations that were correlated with a substantially higher viral load. These mutations were also correlated with each other: individuals tended to have a virus with all of these mutations or none of them. This collection of mutations defines a particular type of virus, what we later came to call the VB variant (because it is virulent and part of the subtype B group of viruses). Technically, we performed a ‘principal components analysis’: we found one direction in the space of all possible mutations that was correlated with viral load. 15 of the 17 individuals we found were from the Netherlands. To make sure this finding wasn’t just a statistical fluke, we worked with colleagues at Stichting HIV Monitoring to obtain much more data from the Netherlands; we found another 92 individuals with the variant. Again, we found they had a substantially higher viral load than individuals with other HIV viruses, and we also found that their CD4 cells declined twice as fast. HIV attacks CD4 cells, gradually impairing an individual’s immune system, and the rate of this decline measures how much damage the virus is causing.
How unusual is discovery of a new HIV variant?
HIV mutates so quickly that every individual has a virus which is different from everyone else’s, and indeed their virus changes over time. The large majority of these mutations make no difference. These different viruses can be grouped, in much the same way that a brother and sister have DNA that is not identical but very similar, and we can group them into the same family. At a higher level we could group them into the same ethnicity. The highest-level grouping for HIV is into ‘subtypes’, and these are strongly correlated with geography. For example in Africa, subtypes A, C and D are most common; in Europe and the US, subtype B is most common. We can define smaller groups inside the subtypes, sometimes called variants. So finding a new variant is normal, but finding a new variant with unusual properties is not. Especially one with increased virulence.
How might this variant have arisen?
New forms of existing viruses can appear in different ways (as is also being discussed for the Omicron variant of SARS-CoV-2). One possibility is ‘recombination’: when two different viruses meet in a single individual and splice together. For the VB variant, we found evidence against this: it looks like the collection of mutations in this virus arose one by one over time. Another possibility is that the virus passed through a long chain of transmission from one person to the next, and viral genetic sequence data was not available for any of these individuals. In such a case, we don’t see the intermediate steps in viral evolution, only the big change from before to after. A third possibility is an unusually long infection in one individual, without successful treatment to stop the virus replicating and evolving. We don’t have the data to tell whether the evolution that resulted in the VB variant occurred in many individuals or just one. That evolution happened mostly before 1992, and at that time most individuals with HIV did not have the genetic sequence of their virus determined.
Was there anything special about the Netherlands with regards to this virus evolving?
The Netherlands has had a strong programme of HIV care and monitoring for many years, getting antiretroviral treatment to as many people as possible as soon as possible. Successful treatment not only improves the health of the person receiving it, but prevents them passing the virus on to others, benefitting their health as well. This is also good for us on another level: it denies the virus an opportunity to evolve into something more virulent. As we quote in our article: “viruses cannot mutate if they cannot replicate” (anonymous), and “the best way to stop it changing is to stop it” (Marc Lipsitch). We therefore consider that the virus arose in spite of the strong programme of treatment in the Netherlands, and not because of it. The other side of the coin is that the excellent monitoring in the Netherlands made it more likely to detect such a variant once it appeared.
Why has the variant been declining since around 2010?
This is most likely a by-product of the strong efforts in the Netherlands to decrease transmission of any HIV, regardless what variant it is.
VB variant: present in other countries?
In our data, we also detected one individual with the variant in Switzerland and one in Belgium. We searched publicly available viral genetic sequences from around the world for matches - effectively ‘googling’ the DNA - but found none. There are probably at least a few more people elsewhere who have not been detected yet, perhaps more. By making the genetic sequence of the VB variant openly available, we are allowing other investigators in different countries to check their own private data.
Study limitations?
We described many different properties of this new variant. However, we didn’t explain the mechanism through which it has higher virulence.
(Our research group - Pathogen Dynamics, in the Big Data Institute, Nuffield Department of Medicine, University of Oxford, led by Professor Christophe Fraser - has a wide range of expertise in the evolutionary epidemiology of viruses. We do the experiments to determine the genetic sequences of viruses, the bioinformatics to make sense of that data, the phylogenetics to understand evolution, and mathematical modelling to bring it all together to answer certain questions of interest such as health intervention effectiveness.)
We don’t have expertise in cellular biology, and so we present this finding to pass the baton to people who can study the interaction of this virus with immune system cells in controlled experimental conditions.
Also, as explained above, our conclusion regarding increased transmissibility was arrived at more indirectly than our conclusions regarding higher viral load and CD4 decline, which could be obtained using relatively simple statistics. This is an almost universal caveat to such studies. Though we can count viral particles and CD4 cells in the blood, we usually cannot count how many times each person transmitted the virus because we don’t know who passed on the virus to whom. To say anything about transmission generally requires making a set of assumptions about how the viral genetic sequences relate to what we want to know, through an evolutionary history.
Uses of the finding?
We hope that experimental study of how this variant attacks immune system cells, and how this differs from normal, could tell us something we don’t know that applies to HIV generally and not just this variant. Better understanding of this process could lead to new ways to stop it or slow it down, i.e. to new treatment.
The VB variant of HIV provides a concrete example for the study of virulence evolution for viruses generally, which has been widely studied through theoretical means with few real-world examples to consider.
We found that on average, individuals with this variant would be expected to progress from diagnosis to ‘advanced HIV’ in 9 months, if they do not start treatment and if diagnosed in their thirties (the progression is faster still if they are older). Advanced HIV is known to cause long-term health problems. This finding therefore serves as an important reminder of World Health Organisation guidelines on testing and treatment: testing should be regular for those at risk of acquiring the virus, and treatment should be started immediately when someone is diagnosed.
How worrying is this finding?
The public needn’t be worried. Finding this variant emphasises the importance of guidance that was already in place: that individuals at risk of acquiring HIV have access to regular testing to allow early diagnosis, followed by immediate treatment. This limits the amount of time HIV can damage an individual’s immune system and jeopardise their health. It also ensures that HIV is suppressed as quickly as possible, which prevents transmission to other individuals (“U=U”: undetectable equals untransmittable). These principles apply equally to the VB variant.
A take away for the general public?
They shouldn't be concerned, for a start (see answer above). This is an example of something that thankfully seems to be rare: viruses or bacteria evolving into a form that's more damaging to our health. We know that prevention is better than cure even just thinking short term: if I take actions to avoid getting an infectious disease, to prevent passing it on once I've got it, I've improved the health of people who otherwise would have become ill. However, prevention is also better than cure when we think about evolutionary epidemiology: each infection that we prevent denies the pathogen an opportunity to evolve into something worse. As we quote in our article: “viruses cannot mutate if they cannot replicate” (anonymous), and “the best way to stop it changing is to stop it” (Marc Lipsitch).
Might there be more variants not yet discovered?
Absolutely, for HIV and for pathogens causing other infectious diseases. This shows the value in routinely collecting genetic sequence data for pathogens.
Are your findings a reminder that the next SARS-CoV-2 variant could be more severe?
Yes. Some people say that SARS-CoV-2 will evolve to become milder, as though this can be taken for granted; it cannot. Much theoretical work has considered the balancing factors that determine the direction of evolution of virulence for pathogens (see the first part of this excellent tweet, the end of which we quote in the article https://twitter.com/mlipsitch/status/1228734360716791815), but we have very few real-world examples from which to draw general conclusions. It is said, correctly, that it is not in any virus' interest to kill the people it infects so quickly that they cannot pass it on to others. However, there is a lot of room for the virus to manoeuvre before reaching that extreme stage. As it explores the space of possible mutations, if it finds a combination that increases transmissibility and causes more damage at the same time, such a variant is expected to outcompete others and spread widely. It's uncertain how likely this is, but it's possible, and very much not in our interests. Hence, "the best way to stop it changing is to stop it".
Why did it take so long to discover this variant after it evolved?
The scientific history is relevant here. HIV affects individuals in a remarkably variable way: some progress to AIDS within months, while others do not progress after decades; some have viral loads (levels of virus) thousands of times higher than others. It was previously thought that this was only due to people’s immune systems varying in their ability to fight the virus. Publishing an article in 2007, Christophe Fraser began arguing that the virus itself was relevant too: that not all viruses are equal in their tendency to cause either severe or mild HIV infections. This idea took some years to become accepted. In 2014 the case was clearer and Christophe set up the BEEHIVE project to understand which changes in the virus’ genetics are relevant for virulence. The project generated viral genetic sequence data on a new scale, requiring a few years’ work to develop new computational methods capable of handling the data. And finally this specific study has taken several years to complete: we describe the VB variant from a lot of different angles, and it has been a busy two years for virus epidemiologists.
Changes in HIV virulence have been reported before. Why is this different?
We refer to examples in the paper that show overall population-level trends in virulence; what distinguishes our study is attribution of a change in virulence to an individual viral variant. When many viruses in many different individuals separately acquire mutations, this could produce an overall trend at the population level without the emergence of any new variant. It would also give us nothing concrete to focus our attention on to understand virulence. Hopefully, future experimental studies of the VB variant can explain how it interacts with CD4 cells differently to produce the more rapid decline we observed. And that understanding could tell us something about HIV-induced immune system damage more generally. Basic science in this area forms the basis for the development of next-generation drugs, and continual development of new drugs is prudent given that viruses can also evolve to become drug resistant. Thankfully, this variant has not.
How did you determine that the VB variant is more transmissible?
We estimated transmissibility using phylogenetic methods. These methods involve constructing trees, very similar to normal family trees for humans, that show how closely related different individuals, viruses, organisms etc. are to each other. One such tree represents our best guess at the evolutionary history behind an observed set of organisms - like inferring things about grandparents based only on seeing the grandchildren. A phylogenetic tree for the viruses in our dataset showed that the individuals with the VB variant had viruses that were unusually closely related to each other, i.e. there was little evolution happening between individuals acquiring the virus and passing it on. That’s a signal that the virus is moving from person to person quickly.
“The VB variant first arose during the late 1980s and 1990s in the Netherlands. It spread more quickly than other HIV variants during the 2000s, but its spread has been declining since around 2010,” - press release. How did you determine this timeline?
We arrived at these conclusions using phylogenetic methods (see answer above). Reconstructing the evolutionary history behind the viruses we did see allows us to infer things about past viruses we didn’t see. Regarding the declining spread since around 2010, we also looked at the simpler metric of numbers of individuals diagnosed with this variant each year.