Half the men in an HIV cure study stayed off treatment for nearly a year after a two-antibody shot

The day before the Conference on Retroviruses and Opportunistic Infections (CROI 2025) opened in San Francisco on Sunday, there was a community workshop on HIV cure science, at which Dr Michael Peluso, a local HIV physician and cure researcher, flagged up two studies to look out for.

“One is the RIO study,” he said. “This studied up to 70 people using a bnAb [broadly neutralising antibody] strategy. We don't know how many people achieved viral control in that study, but I think we're going to find out in a couple of days.

“The other is the FRESH/Gilead study, which is done in South African women living with HIV, most of whom have subtype C virus. This is a really different population and it’s critically important for the global cure agenda.”

RIO

The results from RIO were presented two days later. This study, conducted in the UK and Denmark, gave half of its 68 participants placebo saline infusions. The other half got infusions of long-lasting formulations of two antibodies that can neutralise HIV.

One is 3BNC117, which sticks to the part of HIV’s envelope ‘spikes’ that lock on to the CD4 receptor on T-cells. The other is 10-1074, which blocks a different part of the envelope, the V3 loop. Immediately after their infusions, all participants were taken off their antiretroviral therapy (ART) in an analytical treatment interruption (ATI).

Twenty weeks after starting the ATI, all but three of the 31 placebo recipients had been put back on ART because their HIV had reappeared. RIO’s definition of viral ‘rebound’ was either a viral load of over 1000 for at least six weeks, or two consecutive viral loads of more than 100,000 at least a week apart.

Interestingly, two of the remaining three placebo recipients have still not rebounded 120 weeks after starting their ATI, so are regarded as post-treatment controllers.

In the bnAb recipients, however, only 12 of the 34 had had a viral load ‘rebound’ by week 20. To put it the other way, 65% of the initial participants had maintained viral loads below 1000, or 75% of those still in the study (some dropped out). This may be the highest proportion of participants with delayed viral rebound yet seen in a bnAb study.

The percentage of participants rebounding at week 20 was the study’s primary endpoint, but at this point bnAb recipients who were still virally suppressed were given another shot of the two antibodies and continued without ART.

By the 48th week after stopping ART, 57% had still not rebounded, and by week 72 – over 14 months after stopping ART – 39% had still not. Seven participants (24%) had not had a single viral load reading over 50 by this point.

Chief investigator Professor Sarah Fidler of Imperial College London told the conference that 23.5% of bnAb recipients (eight people) had the typical viral rebound pattern of a rapidly increasing viral load within the first 20 weeks. Four of these ‘rapid rebounders’ had mutations in their HIV suggesting it would be resistant to 10-1074 (currently there is no assay for resistance to 3BNC117).

Forty-eight per cent of bnAb recipients had a delayed reappearance of their HIV, between 20 and 72 weeks after stopping ART. Not all of these were single rebound events – many participants had viral loads that oscillated between undetectability and quite high values, up to 55,000, before restarting.

This left seven people with complete viral suppression – which, so far, has lasted 120 weeks or nearly two years after stopping ART.

Interestingly, viral suppression cannot be explained solely by the levels of antibodies circulating in the participants. The two antibodies have half-lives of nine to ten weeks, so would have sub-therapeutic values by week 48.

Substudies found that the amount of intact proviral DNA in participants’ reservoir cells declined in all but one of the 22 participants in which it was measured, with a third of them having declines of over six logs (a millionfold decrease). At the same time, the amount of defective DNA incapable of producing live viruses increased slightly.

Furthermore, a study that used three different assays to measure T-cell responses to HIV in the participants who did not experience viral rebound found that the strength of these responses increased between week zero and week 36 after stopping ART. This is despite the fact that, because they remained virally suppressed, there were not enough HIV viral antigens circulating to stimulate such responses.

This indicates that the bnAbs, as had been hoped, did more than simply neutralise viral particles – they had a ‘vaccinal’ effect of strengthening immune responses.

In RIO, the participants were all cisgender men, with an average age of just under 40 and an average CD4 count at enrolment of 800 (the minimum CD4 count allowed was 500). They had all started ART, mainly integrase-inhibitor-based, soon after infection.

FRESH

Apart from also having HIV, the participants in the second study, from the FRESH cohort in South Africa, could not have been more different.

FRESH (Females Rising through Education, Support and Health), established over a decade ago, combines epidemiology, cure research and community empowerment in young women in KwaZulu-Natal, the area with the world’s highest general-population HIV incidence. FRESH involves a cohort of initially HIV-negative young women who it tests for HIV very intensively, with RNA tests twice a week. This enables those who test positive to start ART at the earliest possible stage – at so-called Fiebig stage 1, within a couple of weeks of infection, when people first test positive for HIV RNA but have not yet become HIV antibody positive.

The study presented at CROI shows that it is possible to conduct sophisticated cure research in a lower-income African setting. As well as extending cure research to a disadvantaged minority, it also investigates whether the kind of cure intervention that worked in RIO’s predominantly male, middle-aged and gay participants – who mainly had HIV subtype B – also works in a young female population with subtype C.

The FRESH study used different antibodies to RIO: long-lasting versions of VRC07-523, another CD4 binding site bnAb, and CAP256V2, which blocks the V1-V2 section of HIV’s envelope protein. It was designed as an open-label study without a placebo, and primarily as a safety study, as this particular pair of bnAbs has not been tested in people with HIV before.

There were 20 participants, all but three of whom had their HIV diagnosed at Fiebig stage 1 and who started ART within three days of diagnosis. They remained on ART for an average of seven years (minimum 1.7 years). Because they had started ART so early, the majority had had a peak viral load no higher than 5000 and none of them has ever had a CD4 count below 500.

The participants were tested for susceptibility to both bnAbs. Eleven were sensitive to both, seven to VRC07 alone, and two to CAP256 alone.

They were given an infusion of each antibody at the end of the first week of the study. In addition to the bnAbs, they also took an oral immune-enhancing drug called vesatolimod, starting a week before the antibody infusion and then every two weeks for 18 weeks. Five weeks after the bnAb infusion, they stopped their ART.

The primary safety and viral suppression data was collected at week 20, two weeks after their last vesatolimod pill and 14 weeks into the ATI. Still, the ATI was continued (in those who did not need to resume ART) until its 42nd week, and then optionally extended in those still without viral rebound to its 54th week, over a year after stopping ART.

The criteria for restarting ART were similar to RIO’s – a viral load over 1000 for eight consecutive weeks, or a viral load of over 100,000 in two tests a week apart.

In terms of safety, one person did have a severe immune reaction to the bnAbs – a so-called cytokine release – that meant she had to discontinue the study. Most other adverse events were injection site reactions.

Ten participants needed to restart ART owing to the reappearance of their HIV before week 48 of the study. Another six had a delayed viral load rebound and although some did have moderate viral loads, they were able to delay restarting ART until week 48, the end of the per-protocol ATI.

Four participants, however, continued their ATI past that timepoint and have maintained either undetectable viral loads (in two cases) or viral loads that have never risen above 2000 in the other two, now for a median of 1.5 years.

Principal investigator Professor Thumbi Ndung’u told the conference that while 12 participants had typical viral rebounds where their viral load rose steadily, eight had an oscillating viral rebound where their HIV appeared and then disappeared repeatedly, a pattern also seen in RIO. This suggests a highly dynamic competition between the participants’ HIV and the vaccine-like effect of the bnAbs.

In the FRESH study, concentrations of the two antibodies had declined to their IC₅₀, the level where they only neutralised half the virus, by 22 weeks after being given VRC07 and 31 weeks after CAP256, so viral suppression after this was probably due to the vaccine-like effect of the bnAbs.

At the cure workshop, Michael Peluso commented that these two studies exemplified the most promising current area of cure research – of “antibodies that can be manufactured to target the most important areas of the virus, to either neutralise that virus itself or to improve immune responses against the virus.”