Minimal antibody response in Imbokodo compels change of course for HIV vaccine research

It is 13 years now since RV144 became the only HIV vaccine ever to produce a (marginally) positive result in a large efficacy trial. And it’s eight years since aidsmap.com said that the vaccine’s results were “real – and could be made to work better.”

An analysis presented yesterday at the 24th International AIDS Conference (AIDS 2022) of the immune responses created by the Imbokodo vaccine, one of the successors to RV144, confirmed that the first half of that statement was true: it is possible to induce an antibody response to HIV with a vaccine, and the one produced in Imbokodo was essentially the same as that produced by RV144.

Unfortunately, the second half has not come to pass: no efficacy trial since has produced a result even as positive as the 31% reduction in infections seen in RV144. In fact, there have only been two such trials: Uhambo, which closed in February 2020, and Imbokodo, which closed in August 2021. Neither worked.

Imbokodo’s companion study, Mosaico, is still continuing among gay and bisexual men and transgender women and is not due to finish till early 2024. It uses a different and broader variety of HIV antigens (viral proteins aimed at eliciting an immune response) and it may yet produce a positive result. It is fair to say, however, that expectations are not high.

About Imbokodo

Imbokodo recruited 2600 young women in five southern African countries. There were 14% fewer infections in the women given the vaccine than in women given a placebo, but this did not reached statistical significance (in other words, it could have been due to chance).

Nonetheless, this slight hint of efficacy gave researchers hope that finding a correlate of protection in at least some participants might not be a futile task.

It is relatively easy to find what’s called a correlate of risk in a vaccine study. You look to see whether vaccine recipients who did not acquire the infection were more likely to have a certain characteristic than people who did acquire it.

Dr Avi Kenny of the University of Washington and colleagues compared 270 women in the Imbokodo trial who stayed HIV negative and with 54 women who acquired HIV. They soon found a single correlate of risk. Women with antibodies that reacted more strongly to two specific parts of the HIV envelope protein (the ‘knobs’ HIV uses to lock on to and infect cells) called the V1 and V2 loops were about 30% less likely to become infected.

This is more or less the same as the correlate of risk seen in RV144, showing that, while the results in Imbokodo were so weak they did not reach statistical significance, they were not necessarily unreal.

A correlate of risk, however, does not tell us anything about whether it is the actual immune reaction produced by the vaccine that is protective. This is because there are many factors that are not vaccine-generated that can provide some protection. Some are host factors: if the study population has more of the more sluggish types of immune receptor that mean cells get infected more slowly, it will influence the efficacy results. So will variety in the virus, and indeed one of the factors involved in RV144 is that it was more effective in people with the less pathogenic ‘tier 1’ type viruses, of which there are fewer in southern Africa.

So instead, you have to obtain a correlate of protection. Instead of separating out people who are infected and not infected, and then working out what type of immune response they have, you have to sort people by their type of immune response, and then find out if any of these predict efficacy.

The two simplest kinds of immune response did not predict efficacy in this trial. The blood levels of anti-HIV antibodies had no relationship to efficacy. Nor did the efficiency of antibodies in alerting other parts of the immune system, such as CD8 cells, to get active, in the process called antibody-dependent cellular phagocytosis (ADCP).

The only factor that predicted efficacy was, as before, sensitivity to the V1-V2 loop – and not just sensitivity but the breadth of that sensitivity. The vaccine worked better if this facet of the immune response was sensitive to a variety of different viruses with different conformations. It worked even better if a particular section of the antibody protein showed that sensitivity.

In figures, in the small number of women who developed antibodies that responded to 30 times the average number of different V1-V2 epitopes, vaccine efficacy was as much as 50%, rather than 14% (an epitope is the particular bit of an antigen that antibodies react to).

The problem was that women with this breadth of immunity formed a very small proportion of the trial population. This was also the case in RV144.

In other words, 13 years on, we are at essentially the same place with efficacy, though we know a lot more about why that efficacy is poor. We know that the vaccines we have do generate a degree of immunity to HIV; but in the large majority of people, that response is too weak, and above all too specific, to translate into useful efficacy.

Using mRNA vaccine technology in HIV

How do scientists generate a better response? And why haven’t they up till now? In a seminar about using messenger RNA (mRNA) technology to create HIV vaccines, leading prevention research Professor Lynn Morris of the University of Witwatersrand in Johannesburg asked the same thing.

Although COVID-19 vaccines benefited from research stretching back over 20 years, she asked how come effective vaccines were produced within a year of the discovery of the virus – yet we are nowhere near an HIV vaccine after more than 30 years?

Morris showed a picture of the SARS-CoV-2 virus and its spike protein, which mRNA vaccines generate and the immune system reacts to. The spike protein is a really prominent lump protruding from the viral membrane, all the better to infect cells rapidly and cause the fulminant destruction seen in COVID. She contrasted it with the HIV envelope spikes, which are fewer in number than COVID; coated with sticky sugar molecules that disguise them and impede access; and which contain not one but several different antigens, all well hidden, which different antibodies need to react to. In addition, HIV’s ability to generate variants makes COVID’s look lame.

"Researchers are not yet at the stage of creating useful bNAb responses in humans."

This means that the immune response a vaccine needs to generate in order to work against HIV must be as broad and varied as the virus; it must be stronger than any immune response HIV creates in nature; and it needs to be able to latch onto components of the HIV envelope that are both transient in conformation and well hidden.

In the fullness of time, about 10-30% of people with HIV do develop antibodies that have those qualities. These are called broadly neutralising antibodies (bNAbs). The problem is that bNAbs develop in response to a virus that has already outrun them in complexity.

We now know that, as a drug administered by infusion, cocktails of bNAbs can work as both treatment and prevention. But for them to work as a vaccine, they need to develop in the bodies of people who have never seen HIV.

This is theoretically possible; the very existence of bNAbs shows that, somewhere in the body, there are the precursors of the B-cells, that when exposed to HIV components, have the capability to develop in such a way that they produce these highly mutated antibodies.

But to get an HIV-negative person’s immune system to do this implies ‘guiding’ their immune system along a focused pathway by giving them repeated doses of subtly different vaccines that would nudge B-cells to become capable of making bNAbs. An HIV vaccine might be more like a course of vaccines, which would end with B-cells being able to produce bNAbs when the real virus shows up.

In theory you could do this with vector vaccines like those used in the Imbokodo and Mosaico trials. They package HIV proteins inside vectors, the shells of harmless viruses that can get inside cells and induce them to make antibodies. But this might take a very long time.

This is where mRNA vaccines come in. They work differently, by issuing instructions to cells to make their own HIV proteins.

“In effect,” said Morris, “you get the body to make its own vaccine.”

This was how the Moderna and Pfizer vaccines worked, by getting cells to make large amounts of the SARS-CoV-2 spike protein (but not the rest of the virus), which the B-cells then made antibodies to.

The advantage of mRNA vaccines is that they can be made more easily, and can also easily be ‘tweaked’ to make the many different varieties of protein needed to generate a broad and deep enough bNAb response to work against its elusive viral target. (The disadvantage is that RNA is a fragile molecule that has to be kept cold, which might make distribution in low-income settings a problem.)

The mRNA HIV vaccine studies – so far and forthcoming

Professor Bill Schief is one of the most renowned vaccine scientists in the world; in his work at the Scripps Institute in California he has been involved in the development of a number of different vaccines. Now, in partnership with the International AIDS Vaccine Initiative (IAVI), Moderna, and other institutes, he is already taking the first steps to developing an mRNA vaccine for HIV. We reported the early stages from the IAS conference last year and the Scripps institute has an article explaining the concept here.

“What we now know about antibody development implies”, Schief told the conference, “that over the next 5-10 years we need to do a large number of small, phase I studies, all aimed at perfecting a course of mRNA vaccines that can induce bNAbs against HIV.”

A study in monkeys was presented at last year's Conference on Retroviruses and Opportunistic Infections (CROI). It proved that an mRNA vaccine could induce protective bNAbs. In seven monkeys given the vaccine, infection with a virulent human/monkey version of HIV (SHIV) was delayed by a month or more, and possibly by longer in two of them.

But the story of HIV vaccine development contains many vaccines that showed promise in animal studies but failed in humans; in any case, protection would need to last for longer than a month or two. Bill Schief emphasised that researchers are not yet at the stage of creating useful bNAb protection in humans. 

Before they manage that, they will they need to reverse-engineer the bNAb response in people who do produce them, and find out what types and conformations of HIV envelope proteins the jabs need to generate in order to induce similar responses, and which are useful. Early responses have already been seen.

The first study, IAVI G001, did not in fact use an mRNA vaccine: it used a self-assembling virus-like particle, similar to the technology used in the HPV vaccine, which has produced promising results in monkeys.

Its aim was to see if it could generate B-cells capable of making VRC01, which is one of the most studied bNAbs. This was used in AMP, so far the only large study using bNAb infusions as a form of PrEP.

The study finished in March 2020, but development – as with all HIV prevention studies – has been delayed due to COVID-19. In particular, partnership with Moderna was already on the table, but the firm diverted their research to making their highly successful COVID vaccine.

After one to two doses of this vaccine, all but one of the 18 recipients of both a low and a high dose produced a response indicating that their B-cells had become capable of producing a VRC-1-like antibody, though not necessarily a highly evolved one.

From 1 in 7000 to 1 in 1000 of the B-memory cells in their blood had developed this capability, which might not sound like a lot, but is a good start for a vaccine. Two doses of the vaccine did seem to be pushing the B-cells in the direction of being capable of making a more evolved set of antibodies.

The partnership with Moderna has restarted. The IAVI G002 study started in January in the US, while IAVI G003 started in June in Rwanda and South Africa. The vaccine used in these studies issues mRNA instructions to get cells to make particles of HIV protein that, it is hoped, will have an increased immune-eliciting capacity.

Another study called HVTN 302, with a different set of partners but also with Scripps and Moderna, also started in January in the US. This mRNA vaccine gets cells to produce the closest HIV-envelope protein look-alike yet developed.

Dr Sharon Riddler of the University of Pittsburgh introduced this study. One of several ‘tweaks’ in it is to include mRNA-induced proteins that include part of the ‘base’ that is attached to the viral membrane, all the better to look like the real thing.

Another variety of mRNA-generated protein has had the ability to attach to CD4 cells knocked out of it. This is because if the vaccine protein immediately latches onto CD4 cells it changes into a less immunogenic conformation. It would also restrict the ability of other sectors of the immune system, including the all-important B-cells, and the antigen-presenting dendritic cells, which are the ones that first ‘notice’ the HIV protein, to react.

Is this the right research pathway?

What if all this sophisticated work to develop effective antibodies does not work? Several audience members at the Imbokodo presentation and the mRNA satellite asked whether a ‘plan B’ was needed now that bNAb development is the new ‘plan A’.

In particular, does the emphasis on antibodies neglect the other, cellular, part of the immune system and in particular vaccines aimed at developing anti-HIV CD8 cell responses, which mop up already infected cells?

Morris said that the COVID vaccines had been able to generate T-cell and CD8 responses too, and other vaccines aimed at developing broad T-cell responses are being developed by other institutes.

But at least the shock of the failure of the apparently promising non-neutralising antibody vaccines appears to have produced a development pathway that recognises the complexity and sophistication of research needed in order to develop a vaccine that works against as slippery a foe as HIV.