The 26th International AIDS Conference (AIDS 2026) in Rio de Janeiro, Brazil, featured few new scientific breakthroughs in terms of vaccines or cures for HIV. But as well as featuring two more cases of cure following stem-cell transplants, a range of studies were presented that shed new light on how we might achieve a less drastic and more accessible form of cure, or at least enable people not to have to take regular antiretroviral therapy (ART).

In one of the most well-received plenary talks at the conference, Professor Michel Nussenzweig, head of the Laboratory of Molecular Immunology at the Rockefeller University in New York, reviewed several studies that used broadly neutralising antibodies (bnAbs) as potential cure agents. They have enabled a significant proportion of participants to maintain undetectable or low viral loads for prolonged periods without taking ART – in other words, to be in remission.

The proportion of people in these studies experiencing prolonged remission off ART is higher than what had been seen before. Nussenzweig said that the most interesting scientific findings have come from delving into what, exactly, was happening within the immune responses of people who achieved remission, and how they differed from those in people who ‘rebounded’ as soon as ART was withdrawn.

These studies have turned upside down several assumptions about how we might achieve a generalisable cure for HIV.

Nussenzweig said that HIV ensures its persistence by hiding within a few CD4 cells that don’t die, but clone themselves, copying their contents, including the HIV DNA spliced into their genes (so-called proviral DNA). It has been assumed that reducing the size of this ‘reservoir’ of cells was central to achieving a cure. The recent studies using bnAbs, however, have found no link between the size of the reservoir as such and the likelihood of achieving remission.

It had been thought that this reservoir of cloned cells was invisible to the immune system. In fact, the cloned cells do on occasion spit out intact viruses and thus become visible. What this means is that while the number of cells in the body containing defective proviral DNA stays constant or may even increase, the proportion containing DNA capable of producing intact virus halves roughly every five years, so the frequency with which they emit infectious viral particles may decrease too. However, because the immune system recognises and suppresses previous viral clones, the variety of intact viral strains produced may increase.

Secondly, bnAbs have been used in cure experiments precisely because it was assumed that the ordinary, non-neutralising antibodies most people make in response to HIV are ineffective and play no useful part in fighting the virus. This is why HIV is a lifelong infection, it was assumed.

BnAbs are uncommon, hyper-mutated antibodies that are produced by about 5-10% of people with HIV, but only after at least 2-3 years of infection. They have the potential ability to surround and block (neutralise) the infective capacity of HIV viral particles, which ordinary antibodies cannot do. BnAbs have been used in many studies as allogenic drugs – meaning derived from other people and not self-generated. Studies like AMP have investigated them as a kind of sophisticated PrEP for HIV-negative people; studies like RIO have examined them as part of a possible HIV cure for people with HIV. Even more sophisticated studies have investigated whether vaccines can induce HIV-negative people’s immune systems to produce their own bnAbs, in advance of exposure to HIV.

It has been found, however, that autologous (self-generated) antibodies in fact do form part of an important chain of immune-system signals that can initiate an effective response to HIV.

Antibodies are Y-shaped molecules; the two branches of the Y are the ones that attach to viruses and other invaders. The stem of the Y attaches to cells that have other parts to play in the immune system – to macrophages and monocytes, the big white cells that can physically ingest bacteria, viruses and virus-infected cells, and to natural-killer cells, an ‘early response’ force that destroys virally infected cells in the body using enzymes. (These processes, called ADCP and ADCC, are discussed further in this report, which shows how a new class of drugs could give autologous antibodies a helping hand in the response to HIV.)

Finally, HIV was thought to cause immunodeficiency by annihilating CD4 cells, while also driving CD8 cells into a permanently activated, exhausted state. It did this not just by directly killing cells but by forcing them to stay in a constantly alerted state of ‘terminal differentiation’ – a senescent inflammatory state from which there is no turning back.

This year, however, it has been reported that in people who achieve a degree of longer-term ART-free remission in the bnAb studies, a certain proportion of CD8 cells have a quality known as ‘stemness’ – this is a capacity to proliferate and respond to varied viral infections that is characteristic of CD8 cells newly differentiating from the stem cells in people’s bone marrow and lymphatic system.

Cells with this quality may exist as a small pre-existing population of CD8 cells that have preserved their potential to fight HIV. Given the right conditions, in response to very early signs of HIV rebound in people taken off ART and to antibody responses to the rebound, they are able to proliferate and differentiate.

This is important because, while antibodies respond to present infections, CD4 and CD8 cells (T-cells) ‘remember’ earlier infections – or vaccines, which are mock infections.

Nussenzweig put it this way: “Antibodies are stupid without T-cells telling them what to do.” That means the T-cells have to recognise new clonal variants of HIV as needing a changed response, so that new autologous antibodies with sensitivity to them can be generated.

aidsmap reported on this newly discovered ‘stemness’ quality in CD8 cells in two studies, while Nussenzweig cited a third, from a team at Harvard. They report that “HIV-specific CD8 T cells in post-treatment controllers exhibited a stemlike memory phenotype, characterized by high TCF-1 (T cell factor 1) expression.”

TCF-1 is a protein inside cells that switches genes on and off. Cells with high levels of it are held back from becoming short-lived 'effector' cells and so retain the ability to keep responding to new or changed viruses – which is what 'stemness' means in this context.

This third study went further in finding a difference between elite controllers (the 1-2% of people who seem to have an innate capacity to suppress HIV without needing ART) – and these post-treatment controllers who have responded to bnAb treatment. The former seem to have been lucky enough to have, from the start, a narrow but effective response to a particular, indispensable part of their HIV. But post-treatment controllers seem to develop a broad range of responses to numerous related strains.

‘Blips and dips’ in viral load mark antibody responses to new HIV variants

How this happens was investigated by Rockefeller researchers Dr Marcilio Fumagalli and Anna Kaczynska. They looked at reservoir changes and antibody sensitivity in two trials.

In the randomised RIO study, infusions of the two bnAbs teropavimab (3BNC-117LS) and zinlirvimab (10-1074LS) were given to 34 people and saline placebos to another 34, who then stopped ART. In a second phase, they gave the two bnAbs to the original placebo recipients and then, after a pause to let antibody levels decline, also took them off ART.

RIO was a landmark as the first randomised study in humans in which more than half its participants were able to maintain a prolonged time off ART before their HIV rebounded. In a post-study analysis, time off ART was not correlated with the original size of people’s reservoir of proviral DNA, nor with teropavimab’s efficacy in neutralising HIV. It was correlated with 10-1074’s efficacy, but also with the efficacy of autologous antibodies at baseline.

Nussenzweig commented that the participants in RIO had been diagnosed with HIV soon after acquiring it, had started ART less than six months after their diagnosis, and had been on ART for one to five years before RIO. So they had a relatively small reservoir of cells containing proviral DNA and had what Nussenzweig called “some level of immune health”, including those pre-existing autogenic antibodies.

The MCA-1031 study almost replicated RIO. However, it recruited 28 people with chronic HIV who had been on ART for an average of 10 years and had a wide range of ages. It differed from RIO in two other ways. Firstly there was no placebo arm. Secondly, during the first 24 weeks of their ART interruption, participants were given a dose every three weeks of Anktiva (N-803), a superagonist or magnifier of the cytokine (immune messaging chemical) IL-15, which stimulates natural killer cell and CD8 cell proliferation. Nussenzweig himself did not seem to think the N-803 was crucial (“Don’t ask me exactly how it helps the immune system,” he said).

The results were strikingly similar to RIO. Ten participants’ HIV rebounded early, just as if they had had no bnAbs. But 16 people (57%) were able to stay off ART for longer than 24 weeks. Of this 16, 11 eventually restarted ART before the study end date of 72 weeks, though two had undetectable viral loads when they did so. This left five who had not restarted ART more than 72 weeks after stopping it, one of whom has stayed off for nearly three years.

Four of the people who stayed off ART had ‘blips’ where their HIV viral load temporarily rose to several hundred to several thousand copies before subsiding to undetectability again. During blips, participants had their HIV genotyped to see if it had become resistant to either the bnAbs or the autologous anti-HIV antibodies measured at baseline. A remarkable pattern was uncovered. Each time there was a viral blip, it became clear that a different genetic clone of HIV was emerging from different cells – but was being suppressed by antibodies. “It was like ‘Whack-a-Mole”’, Nussenzweig said. “A new virus would emerge and then quickly disappear.”

Nussenzweig was especially interested in the fifth person who stayed off ART – their viral load had hovered between a hundred and a thousand.

Over a space of 38 weeks, this participant’s HIV developed a degree of resistance to the bnAbs they were given, especially to zinlirvimab. The autologous antibodies measured at baseline, which had limited efficacy against HIV, had lost efficacy by week 38. But by then, in this participant and others, new autologous antibodies had appeared that were more effective than the ones assayed at baseline.

“The immune system in these patients is evolving over time,” Nussenzweig said. As a result, T-cells capable of directing fresh anti-HIV responses were also active. In fact, the T-cells in early rebounders and post-treatment controllers sorted into two very distinct populations, with the former non-responsive and the latter responsive to multiple different HIV components.

What this means for a cure

Nussenzweig concluded that enhanced post-treatment control is associated with:

• Pre-existing stemlike CD8+ T cells. Can we increase these with therapeutic vaccination?
• Pre-existing autologous antibodies. Can we also increase these by vaccination?
• A less diverse, rather than a smaller, starting reservoir. Can we develop sequence-based methods to predict who would benefit most from bnAb therapy?

“We finally understand what we have to do in order to achieve a cure,” Nussenzweig concluded.