Broadly neutralising antibody treatment: is pre-existing resistance its Achilles heel?

Last week’s Conference on Retroviruses and Opportunistic Infections (CROI 2022) heard about two early phase I studies in which people with HIV who had unsuppressed viral loads were given combinations of broadly neutralising antibodies (bNAbs) instead of standard antiretroviral therapy (ART).   

The two studies only included ten people in total. One study gave participants a combination of three antibodies and the other study two long-acting ones. Between the two studies, three participants were able to achieve and maintain an undetectable viral load for periods of several months on bNAb treatment alone. However, six others (one was lost to follow-up) saw their viral loads initially decline, but return to baseline level after two to four weeks.

In both studies, the crucial predictor of virological failure was whether people had HIV with pre-existing resistance to one or more of the bNAbs. The results suggest that, in the absence of bNAbs and combinations of bNAbs that are even broader in their ability to neutralise many different strains of HIV, resistance tests will be required to determine who may be able to benefit from them. 

Background

Broadly neutralising antibodies to HIV eventually develop after years of infection in 10-30% of chronically infected people. The ‘neutralising’ means that they can physically prevent viruses infecting cells by attaching to them and acting as a barrier. The ‘broadly’ means they act against a variety of viral strains – strains that develop over time, inside people with chronic infection.

bNAbs persist in the body at therapeutic concentrations for weeks or months, so could be used for injectable treatment or PrEP (regular medication to prevent HIV infection).

Additionally, if bNAbs could be given at an early stage of infection or even beforehand, as PrEP or, even better, if they could be generated in response to a vaccine, they could even cure HIV or put it into long-term remission. This is because bNAbs don’t just block HIV, but also recruit other parts of the immune system to fight it. This may have happened in one person whose case was presented last week.

bNAbs have been have been trialled as PrEP and as treatment, including recently in children. Experimental vaccines that induce the immune system to make them are also under study. In a 2018 study, three successive doses of a dual-bNAb combination enabled nine out of eleven people to maintain an undetectable viral load for between two and seven months off ART.

However, that study pre-screened participants for resistance to the two antibodies used (and, even so, two turned out to have undetected resistance). In the two studies presented at CROI 2022, participants were not pre-screened – because both studies aimed to see how long a single shot of a bNAb combination suppressed HIV in people not on ART, and the degree to which this was affected by resistance.

The Ragon study – three antibodies

In the first study, a team led by Dr Boris Juelg of the Ragon Institute gave a combination of three antibodies to four people who were not taking ART. The antibodies were PGDM 1400 and PGT 121, both mentioned in this report, and VRC07-523-LS, mentioned here (LS signifies the long-acting formulation). The three antibodies attach to three different parts of the HIV envelope protein. 

After a dose-escalation safety study giving the triple combination or a placebo to four HIV-negative people, it was then given to four people with HIV who were not on ART and who had a viral load ranging from 10,000 to 100,000.

It was quickly apparent that the three antibodies together did not last as long in HIV-positive people with detectable viral load as they had in the HIV-negative participants. Whereas the half-life of the single antibodies in HIV-negative people was in the region of 17-20 days, it was only eleven days in people with HIV. This is explained by there being virus around to lock on to, so there are fewer free-floating antibody molecules in the blood.

Seven days after receiving their triple-antibody infusion, all participants experienced a drop in their viral load averaging 1.76 logs (a 57-fold drop). One of the four participants left the study and was lost to follow-up after 28 days, and their viral load seemed still to be decreasing at this point. Excluding that participant, the average maximum viral load drop to nadir (lowest point) before it rebounded was 2.04 logs (110-fold).

The baseline viral load in the three participants was closely related to the nadir. In the participant starting with a viral load of 100,000, it dropped to about 700. The other two had started with lower viral loads, around 10,000, which dropped to about 70 in one and under 50 (undetectable) in the other.

What most distinguished the three participants, though, was the time it took for their viral load to rebound to baseline. In the first participant, it only took 10 days; in the second, 20 days. But in the third it took 70 days. This participant’s viral load then rebounded back to the low thousands. But the last time it was measured, 175 days (nearly six months) into the study, it had again become undetectable off ART.

Resistance testing at baseline, and again after the first rebound, showed that the first participant with the shortest and smallest viral load drop had virus that was quite susceptible to PGDM 1400 and VRC07-523-LS, but quite resistant to PGT 121, so they were only benefiting from two antibodies. Their post-rebound virus was completely resistant to both PGDM 1400 and PGT 121, and had developed some resistance to VRC07-523-LS as well, within a matter of six weeks. At this point, with a viral load of about 70,000, they started ART.

The second participant turned out to have virus that was completely resistant to PGDM 1400 and PGT 121 from the start. They were therefore only benefiting from VRC07-523-LS, though sensitivity to that bNAb reduced during rebound too. They chose to stay off ART despite a post-rebound viral load averaging about 5000.

The participant with prolonged undetectability turned out to have virus that was very sensitive to all three antibodies. They preserved that sensitivity to VRC07-523-LS but worryingly, at day 80, when their viral load was only about 700, some of the viral strains they carried were already completely resistant to PGDM 1400 and PGT 121.

The performance of bNAbs depends on three factors; firstly, the number of different viral strains they can neutralise; secondly, the amount of time they can stay locked on to virus; and thirdly, the level they need to reach in the body in order to neutralise viruses. Although all participants retained some sensitivity to VRC07-523-LS, cell assays found that viral rebound still happened even with exposure to that bNAb, and happened at about the same level in all three participants – below 93 micrograms per millilitre (mcg/ml).

This was far higher than the levels suggested in laboratory assays. In a lab-dish assay that detects antibody efficacy by creating an artificial cell that glows when it is infected by HIV, the triple combination prevented 100% of infections by a panel of various viruses at a concentration of 10 mcg/ml, and 80% at 1 mcg/ml. Even if there was resistance to one of the three, 80% of viruses should be neutralised at a concentration of 10 mcg/ml.

Boris Juelg commented: “A fine balance of broad antiviral activity and high bNAb serum concentrations needs to be achieved in order to control the virus.”

In answer to questions, he added that where the rebound virus was coming from was also important. Antibodies could not necessarily penetrate the so-called ‘sanctuary’ sites in the body where viral replication restarted. “Just measuring the antibody level in the blood that confers full control in the lab may not be sufficient to predict control in vivo,” he said. 

The fact that the one bNAb with the long-lasting formulation did not, in this study at least, produce high-level resistance to it in HIV, may give hope that if it can be given with others that combine breadth of effect with a long half-life, better results might be obtained.

The Rockefeller study – two long-lasting antibodies

The second study, presented by Dr Marina Caskey of Rockefeller University, gave participants a double combination of two long-acting bNAbs, 10-1074-LS and 3BNC117-LS. Both were used in other bNAb studies at the conference, with the former given to children and the latter taken by the person who has now been off treatment for almost four years. Caskey's study used the LS formulation, which prolongs the antibodies' half-life – the time it took for their levels in blood to fall by 50% – from 24 and 18 days to 80 and 62 days respectively.

One infusion of the bNAb double therapy was given to six people with HIV, all male and aged 26-56, with an average CD4 count of 523, who had come off ART four weeks previously. By the time they received the bNAbs their viral load averaged 48,700, ranging from under 1000 to 125,000.

As in the other study, it was observed that the half-life of the bNAbs in people with replicating HIV was shorter than that seen in HIV-negative people.

Three participants had viral loads that fell then rebounded swiftly, reaching their nadir within seven days. The average viral load decline in these four participants was 1.4 logs or a 25-fold decline, and none reached a viral load below 1000. A fourth participant had a slower and deeper viral load fall and rebound, with a 2.5 log or 300-fold decline to a viral nadir of about 100. 

All these participants restarted ART, one at week six and the others at week eight. A post-hoc study looked at their resistance to the bNAbs at baseline, but post-rebound resistance was not measured in this study.

The two with the smallest decline in their viral load turned out to have had complete resistance to 10-1074 and low-level resistance to 3BNC117 at baseline. The other two had the opposite pattern, with high-level resistance to 3BNC117, but sensitivity to 10-1074, at baseline. 

This left two other participants. They were the two with the lowest baseline viral loads, of about 5000 and 1000. They both had about a 2 log (100-fold) viral load decline to a viral load of below 20. In one case (the person with the lower baseline viral load) this rebounded to baseline levels by week 20. In the other case, they stayed undetectable beyond the 24 weeks of the study and eventually rebounded at week 36, more than eight months after receiving their bNAbs. 

These two participants were completely sensitive to both antibodies. What this meant in practical terms, to give an illustration, is that they required one-sixth the level of 10-1074 to reduce viral replication compared with the two who had low-level resistance to it. 

Questions and implications

Marina Caskey was asked whether lower baseline viral load was a sign that someone would likely be sensitive to bNAbs. She said it might rather be the case that people with higher viral loads spent more time with replicating virus and antibodies both present in their system and might also have a wider variety of pre-existing viral strains. Both could set up the conditions for resistance.

Boris Juelg commented that this might imply that bNAbs as therapies might have a better place if used in people who already have an undetectable viral load on ART as intensifiers of viral suppression, and possibly to enable people with an already suppressed viral load to take breaks from ART. “Then they won’t have this huge wall of replicating virus to demolish”, he said.

Caskey said she agreed with Juelg that in vitro studies predicting the bNAb levels to which viruses were sensitive might underestimate the levels needed in vivo, with implications for dosing.

aidsmap.com asked the two presenters whether it was ever going to be possible to develop effective bNAb therapies that did not require people to be tested for resistance in advance, given that pre-existing resistance seemed so common.

Juelg said, “There are experimental therapies coming along like tri-specific antibodies [which attach to three different viral and cellular receptors at once]. 

“But clearly we can’t add more and more, and have a ten-antibody cocktail in a bid to neutralise every viral strain. We’re going to have to have a genotyping approach, as we already do for antiretrovirals, where we test in advance what mix of therapies are going to work, and that could include antibodies.

“But yes, we thought our cocktail in this study was exquisitely broad in its effect, yet we still saw high levels of existing and developing resistance. So the barriers to bNAb therapy right now are quite high.” 

Session chair Professor Joe Eron of the University of North Carolina commented that the state of current bNAb therapy resembled the early days of antiretroviral therapy, where the effectiveness of treatment was often rapidly compromised by resistance.

“There were people with low viral loads who did achieve viral suppression just on AZT plus ddI,” he said. “But we soon learned that for the majority, we needed more drugs of different classes to suppress HIV.”