HIV vaccine trial failure is a “wake up call” for southern Africa, say researchers

Gus Cairns talks to Linda-Gail Bekker and Glenda Gray about the future of HIV vaccine research
Linda-Gail Bekker and Glenda Gray launching the HVTN 702 study at AIDS 2016 in Durban, South Africa. Photo ©International AIDS Society/Steve Forrest/Workers' Photos

“We have an answer – the regimen simply didn’t work. It feels really grim, though, today.”

This was the immediate and heartfelt reaction of one of the principal investigators of the first large-scale HIV efficacy vaccine trial to be launched for nine years – only one of three such 'phase III' trials in the last decade – when I emailed her on 3 February. That was the day it was announced that the HVTN 702 (Uhambo) trial would close, as the experimental vaccine had no efficacy. There had been no difference in the HIV infection rate between people receiving the vaccine and people receiving a placebo.

Dr Linda-Gail Bekker of Cape Town University – Chief Operating Officer of the Desmond Tutu HIV Foundation and former President of the International AIDS Society – was speaking for a lot of people in the HIV research and activist community. Although there was no certainty that the vaccine regimen involved would have any efficacy at all – which was the whole point of doing such a trial – there had been high hopes that it might.



How well something works (in a research study). See also ‘effectiveness’.


A harmless virus or bacteria used as a vaccine carrier to deliver pieces of a disease-causing organism (such as HIV) into the body’s cells to stimulate a protective immune response.


A pill or liquid which looks and tastes exactly like a real drug, but contains no active substance.

immune response

The immune response is how your body recognises and defends itself against bacteria, viruses and substances that appear foreign and harmful, and even dysfunctional cells.


A substance which forms the structure of most cells and enzymes.

The immune responses it had produced in earlier phase I trials (which look at safety and basic immunogenicity) and phase II trials (which evaluate indications of efficacy and decide which vaccine regimen to take forward) indicated that it might comfortably out-perform the original vaccine it was based on.

In the event, of the 5407 young South African women and men in the trial, 123 who were given a placebo caught HIV and 129 who were given the vaccine did so. This 4% annual infection rate was clearly the same in both arms – and also a reminder of how high HIV incidence remains in that country and its neighbours.

Why HIV is a challenge for a vaccine

Scientists have been trying to devise a vaccine against HIV for as long as we’ve known that this virus was the cause of AIDS.

But HIV is a very difficult organism to vaccinate against. It reproduces rapidly and imperfectly, meaning that it becomes resistant to the immune defences created by a vaccine just as fast as it develops drug resistance. HIV hijacks the body’s own cellular immune response and uses the T-cells that direct that response as its breeding ground – which means that the most basic types of vaccines, which use live but weakened versions of the same infection, might cause an infection themselves. It is also a retrovirus, which means that it hides itself away deep inside our own genetic code where it is invisible to the immune system.

Taken together, these factors mean that there is no natural immune response strong enough to prevent or contain infection, which is why HIV stays with you for life and needs lifelong therapy. And it also means that a vaccine needs to induce an immune response in its recipients that’s 'better than nature' – which is why the study of elite and post-treatment controllers, who manage to contain HIV infection spontaneously, is so important.

The search for a vaccine

The first two large efficacy trials, called Vax, were launched in 1998 and between them enrolled nearly 8000 people, in the US, Canada, the Netherlands and Thailand. They used a strategy that has been used ever since in most of the efficacy trials: a 'prime-boost' strategy consisting of two vaccines.

One is often a vector, a piece of DNA or protein wrapped inside the shell of another, harmless virus. This can enter cells and therefore alert the cellular part of the immune system, setting off a CD4 and CD8 cell response. The other is a ‘naked’ piece of HIV protein of DNA, but one tweaked to maximise or broaden the immune response, often by mixing epitopes – immune-stimulating sequences – from different varieties of HIV. This is designed to provoke the humoral part of the immune system and generate antibodies – though in practice, things are more complicated, as the presence of antibodies can also alert CD4 and CD8 cells.

Although the Vax trial did not work and ended in 2003, the HIV-protein component, called AIDSVAX, continued to be used in other trials, including HVTN 702.

A key study, STEP or HVTN 502, began to enroll 3000 people in December 2004. It was a trial solely of a vector vaccine, with the vector ‘shell’ based on an adenovirus, a viral type that often causes cold-like illnesses. In a foretaste of HVTN 702, the immune reactions it produced seemed promising and there was a general feeling of optimism that this vaccine might work.

It was therefore a blow when the STEP trial was terminated in March 2007. Worse, there seemed to be a slight excess of infections in people receiving the vaccine – 49 infections – versus those receiving the placebo – 33 infections.

This wasn’t a random result. Scientists found in 2008 that previous exposure to the adenovirus used as the vaccine vector actually made people’s cells more receptive, rather than less, to subsequent HIV infection. Two other efficacy trials, PAVE and Phambili, that used a similar vaccine to STEP, were also stopped.

"It was encouraging that there was an effective antibody response, though there were a couple of warning signs that it was not that strong."

This marked a low point in the morale of HIV vaccine developers. Pioneering virologist David Baltimore, who discovered reverse transcriptase, described HIV vaccine development in early 2008 as a “sad topic” and said: “We are no closer to a vaccine now than we were…the day HIV was discovered.”

At the time RV-144, an efficacy trial using the AIDSVAX protein, was ongoing in Thailand. The study had been widely criticised, because some researchers saw RV-144 as just a repeat of a strategy that had already failed. Therefore the news in September 2009 that the RV-144 study had produced a positive result was greeted therefore almost with disbelief. Our headline, An unpopular vaccine study produces surprising result, pretty much sums up the feelings at the time.

RV-144 was huge, with 12,000 participants, and incidence low, with 125 HIV infections, but there were 31% fewer infections in the vaccine arm than the placebo arm (51 versus 74) – and this was (just) statistically significant.

Later investigations showed that this was not a fluke. It took time to re-run aspects of RV-144 but in a follow-up study, RV-305, original trial volunteers given a new dose of vaccine produced strong antibody responses to particular parts of HIV’s envelope protein and a South African follow-up study, HVTN 097, showed that these responses could be strengthened. A particular kind of antibody response called IgG seemed to be crucial.

HVTN 097 was the first of a series of phase I and phase II trials, conducted by the same research team, that would eventually lead up to HVTN 702.

It was encouraging that there was an effective antibody response, though there were a couple of warning signs that it was not that strong: it only seemed to neutralise the least virulent ‘tier 1’ viruses and it tailed off rapidly with time. If RV-144 had been stopped at the end of its first year, it would seem to have had 60% efficacy.

Nonetheless, the fact that a vaccine had produced any kind of positive result had a galvanising effect on the field, in the same way that a single person with HIV cured galvanised cure research. So when the immune system responses seen in a subsequent trial, HVTN 100, comfortably exceeded the criteria for starting a phase III efficacy study, HVTN 702/Uhambo (the vaccine trial which has just closed) started to enrol participants.

At the time of the launch in November 2016, commented: “The signs are very promising that the HVTN 702 trial may show significant efficacy, but we have been unpleasantly surprised before, and may not know the full results for four years.”

Which takes us to today…

"Your head tells you this is good science…but your heart is so invested in hope this will be The One"

So says Linda-Gail Bekker of her disappointment when getting a call on the morning of Monday 3February from the Data and Safety Monitoring board of the HVTN 702/Uhambo study. I talked to Linda and her Co-Principal Investigator Professor Glenda Gray of the University of the Witwatersrand, who is also President of the South African Medical Research Council.

The Data and Safety Monitoring Board (DSMB) is the only body in any large randomised clinical trial that sees the unblinded data while the trial is running – meaning that, in this case, they know who had been given the vaccine and who the placebo.

“The DSMB was scheduled to have a look at the data every six months, or after every 50 HIV infections,” Bekker says. “They had a look at the end of last year, but then scheduled another look on 23rd January, so we knew something was up.”

“If our vaccine was only as efficacious as it was in RV-144, then for every infection that one stopped, ours would have had to stop fourteen."

There are only three reasons why a DSMB stops a trial, she explained. For safety, for efficacy, or for futility.

Stopping for safety means that the DSMB has evidence that the intervention is actually causing harm. For example, there are significantly more infections, or serious side effects, in the vaccine arm than the placebo arm (this could have happened in STEP, say, if there had been a much larger number of infections in people given the vaccine).

Stopping for efficacy is almost the opposite case; it means your intervention has been so effective in stopping disease that it would be unethical to keep giving half your trial subjects a placebo. This is what happened, for instance, when the PROUD PrEP study was stopped early because it was preventing 86% of infections.

Stopping for futility means that you have got to the stage where it has become clear that there is no chance of proving that your intervention is making a difference, for good or for bad. It does not necessarily imply it makes no difference at all; just that in order to prove any difference, your trial would have to enrol an unfeasible number of people or be continued for an unrealistic length of time. This can be a harder judgement call than stopping for the other two reasons. For instance, the Bangkok Tenofovir Study of PrEP in injecting drug users was extended because HIV incidence was lower than expected, so it took longer for the lower number of infections in people given PrEP to become statistically significant.

“In the case of HVTN 702, the protocol said we had to have at least 60% of participants in the trial for at least 18 months before we could even look at futility,” says Bekker. “We know we were getting infections, of course, because you’d see them in the clinic, but that didn’t mean we weren’t getting efficacy.”

“We started to get spooked because we were told to be on standby for a phone call,” says Glenda Gray. “We were ‘angsty’ in the run-up, but I thought we were mentally prepared.

“When they told us it was futility – well, my first thought was ‘At least it didn’t do any harm’. But then – well, I started to think of all the people with hopes invested in this trial, all the conference calls, all the meetings. It’s hard not to find it disappointing.

“You also got all the nay-sayers coming out and saying ‘You see, we said it wouldn’t work.’ But we had good reasons for taking it forward – we saw at least as much immunogenicity, at least as high an immune response, as RV-144.”

So given that, why did this vaccine make no difference?

“It’s much too early to say, but I think it will probably turn out to be a matter of the simply much higher incidence in among our participants,” says Glenda Gray. “Our incidence of HIV infection, at 4% a year, was fourteen times higher than in RV-144, at 0.3%.”

“If our vaccine was only as efficacious as it was in RV-144, then for every infection that one stopped, ours would have had to stop fourteen. So any efficacy it had would simply be drowned out by the much higher number of infections. We had a pellet gun, where in South Africa, you need a bazooka!

“One thing we do know is that it’s not because efficacy faded over time. Because we were worried it might, we had booster doses at 12 and 18 months. None of the immunogenicity seemed to be waning.”

Linda-Gail Bekker notes that one consolation for the failure of the vaccine is that, with 260 HIV infections, there will be plenty of data on how the vaccine interacted with the virus. They will be looking for a phenomenon called 'viral pressure': this is a general term to find out whether, despite not protecting people from HIV, the immune response to the vaccine forced HIV to make changes in its make-up to escape the immune system’s surveillance. This could bring about results such as a lower viral load, but even if it doesn’t, it provides data on how to enhance those changes and make them deep enough to repel or contain infection.

Ongoing vaccine studies

"The good news is that the Imbokodo and Mosaico trials use totally different strategies to Uhambo.”

HVTN 702/Uhambo was only the first of several current and forthcoming vaccine efficacy trials. The furthest along is HVTN 705/Imbokodo, which is due to finish in 2022, a five-country southern African study in which 2600 women aged 18-35 are given a vaccine or placebo.

This has been followed by HVTN 706/Mosaico, which started last autumn, which will recruit 3800 cisgender men and transgender people aged 18-60, who have sex with men and/or transgender people, in eight countries spread across three continents. This is due to finish in 2023.

“The good news,” says Glenda Gray, "is that these trials use totally different strategies to Uhambo.”

Both studies use four shots involving two vaccines. The first vaccine, given in all four shots, is dubbed Ad26.Mos4.HIV, and uses an engineered adenovirus vector, a harmless relative of the common cold virus, to deliver a so-called 'mosaic' of optimised HIV immunogens, or antigens that stimulate immune responses. (The vector is from the same viral family as was used in the STEP study, but people are screened for pre-existing immune responses to it, to avoid STEP’s adverse responses.)

The second vaccine, which is added as a boost to the third and fourth vector shots, consists of HIV proteins without a vector. In Mosaico, this protein similarly contains a mosaic of sequences from different viruses – hence the study’s name.

The 'mosaic' is designed to elicit immune responses to both clade C, the most common variety of HIV in southern Africa, and clade B, the primary one in the global epidemic in gay and bisexual men and trans people.

In the studies (such as APPROACH) leading up to these two phase III trials, strong immune responses were seen and a monkey-adapted version of the vaccine protected two-thirds of a group of 72 rhesus monkeys from infection with six successive exposures of a highly infectious laboratory-developed SIV/HIV virus.

However, we have learned not to expect human results to mimic monkey ones. The fact that the vaccine protected 94% of monkeys to any one exposure to HIV reminds us that any HIV vaccine will have to be extremely potent to fend off multiple exposures.

Further into the future

There are many other experimental vaccines at lab-test and safety-study stage queuing up to prove their worth, but vaccine development can take a long time.

Just one illustration was a novel live-attenuated vaccine using Cytomegalovirus (CMV) as its vector, which produced promising results as long ago as 2013. The exciting, but risky thing about this vaccine was that, whereas conventional vectors get into cells but do not reproduce, the CMV vaccine mimicked an ongoing infection – producing a much broader immune response to it, and a state in which HIV infection was as much cured, post-infection, as prevented. The immune response seen also seemed to be very long-lasting.

It only worked for 59% of the monkeys, though, and manufacturing a safe version for humans proved to be dauntingly difficult. But it was announced last year that a version suitable for clinical trials was at last ready. Even if this jumps through all the hurdles to a full efficacy trial, however, it will have taken at least 15 years from the first study in animals to being available.

HIV vaccines mainly seem to work by inducing the body to make antibodies that fight off infection. So far we have not yet developed a vaccine in human trials that induces the body to make the 'broadly neutralising' antibodies that are the best answer to the sheer genetic diversity of HIV.

However, they are being given as 'passive immunisation', i.e. as a novel kind of long-lasting drug given by infusion. Trials such as the AMP study may revolutionise HIV treatment, even if they are not vaccines, especially if they can be given alongside long-lasting injectable or oral HIV drugs, and last month it was announced that a study combining a broadly neutralising antibody infusion with the injectable formulation of the antiretroviral drug cabotegravir will start in the US.

However, these are essentially still refinements of HIV treatment. A vaccine that generates a broadly neutralising antibody response is likely years ahead. We now know that it’s possible, though, at least in monkeys. In January last year, scientists finally managed to get a vaccine given to monkeys to induce the animals’ immune system to make their own broadly neutralising antibodies. This was a major scientific breakthrough, but even so it only completely protected two out of 12 monkeys. This is only the start of a long research path to making something effective, cheap and practical enough to be the highly effective HIV vaccine we need.

If your vaccine fails, try PrEP

What to do in the interim? “At the moment, like it or not, maybe PrEP is the answer,” says Glenda Gray. “We do need something, because despite successes in certain localities, just stepping up HIV treatment has made no difference at all to the 4% annual HIV incidence we saw in our Uhambo participants. This rate of infection has persisted for two decades.”

She points out that none of huge cluster-randomised trials of enhanced testing and treatment that we have seen in Africa, such as PopART, and SEARCH, have changed that convincingly, partly because they did not reach the people at highest risk, partly because HIV stigma deters infected people from seeking care, and partly because the people who transmit HIV are more mobile than had been supposed.

“We are intensifying PrEP provision at all 14 of our Uhambo clinic sites, and this is happening at the same time as an intensified national rollout.”

It is indicative that SEARCH, a test-and-treat study in Kenya and Uganda that produced some of the more positive results, has now included PrEP in its next phase.

“We need to put much more energy and funding into rolling PrEP out properly,” says Linda-Gail Bekker. “I am now hearing more knowledge and acceptance of it within the community – our young women when told about PrEP were very enthusiastic – but I am very concerned about less momentum within the health system. The poor adherence we have seen can be addressed by community knowledge, but that counts for nothing if innovative and effective provision solutions are not developed by the health system.”

She points out, for instance, that though the new injectable HIV drugs may have promise within PrEP “they will still be provided within treatment settings, and there will need to be cold-chain storage of them. Oral PrEP, however, you can send to people anywhere by courier.”

She acknowledges that creating demand for PrEP needs to go a lot further still but adds that the failure of Uhambo may put new urgency into this demand. “We are intensifying PrEP provision at all 14 of our Uhambo clinic sites, and this is happening at the same time as an intensified national rollout,” she says.

Glenda Gray is slightly more sceptical about whether enough people will take it properly to make a difference. She acknowledges that both oral and injectable contraception are in wide use in young women in Africa, but “That’s because even if you have a 4% of getting HIV within a year, you have a 30% chance of getting pregnant.”

Maybe, as SEARCH is beginning to find, appealing to young men may be part of the answer.

Another is – well, why not combine a vaccine with PrEP? If the first generation of HIV vaccines are only going to be effective in situations of low HIV incidence, then why not bring HIV incidence in higher-risk individuals down to the lower level by giving them PrEP too?

This is the concept that will be used in the next vaccine trial to start. PrEPVacc will give either a placebo or one of two versions of a vaccine – one just a naked-DNA vaccine, the other that plus a vector – to at least 1668 women and men aged 18-40 in South Africa, Mozambique, Uganda and Tanzania, and will further randomise participants to take either of the two versions of PrEP we have available – emtricitabine plus either tenofovir disoproxil fumarate (TDF) or tenofovir alafenamide (TAF). People won't have to take PrEP to stay in the trial, but will be encouraged to. PrEPVacc is due to start within the next two months, and will last two years.

“Even if PrEP is not our silver bullet, though,” says Glenda Gray, “It could be a critical component of bringing our HIV epidemic here under control. And it is not under control. I think a lot of the time we are inured to the gravity of our HIV epidemic in southern Africa.

“One positive aspect of the closure of the Uhambo trial is that at least it is a wake-up call.”


Update: This study was subsequently published in a peer-reviewed journal:

Gray G et al. Vaccine Efficacy of ALVAC-HIV and Bivalent Subtype C gp120–MF59 in Adults. The New England Journal of Medicine, 384: 1089-1100, March 2021.

DOI: 10.1056/NEJMoa2031499