The three men came in with HIV still measurable in their blood, somewhere between 17,600 and 19,600 copies in every milliliter, and they went home the same way. In between, for two weeks, a single intravenous dose of a molecule David Ho’s laboratory at Columbia had spent years engineering drove the virus down and held it there. Then, as the antibody drained out of them, it came back.

The molecule, 10E8.4/iMab, is a bispecific antibody: one arm grabs the virus at a stretch of its envelope it cannot easily change, the other blocks CD4, the receptor HIV uses to get into a cell. The phase 1 trial published July 7 in Nature Medicine calls it safe and well tolerated, and it was. Across all 54 volunteers, the great majority HIV-negative and only three of them carrying detectable virus, the worst that happened was fatigue, headaches, sore injection sites, and a few transient rashes that cleared on their own. No serious adverse events, nothing grade 3 or higher, no deaths. That phrase is what every clean first-in-human study earns. It means the drug did not hurt the people who got it. It says nothing about whether the drug works.

On whether it works, the trial offers three men and fourteen days. Registered as NCT03875209 and run at Columbia in Manhattan and the Orlando Immunology Center in Florida, it was built around safety and pharmacokinetics; the viremic treatment group was a small add-on, and then COVID-19 arrived, paused the study, and thinned the one arm that could show antiviral activity at all. What is left is a clean signal in three people: viral loads that dropped sharply after the dose, bottomed out, and then rebounded as the antibody washed out of the body. The antiviral readout was set at day 14. By day 21 the virus was back.

The rebound virus carried no mutations conferring resistance to either arm of the antibody. The virus did not evade the molecule. It outlasted it. The problem was never breadth. It was exposure.

That points straight at the drug’s design. The iMab arm is a version of ibalizumab, the CD4-binding antibody already sold as Trogarzo for multidrug-resistant HIV, and it clings to CD4 so tightly that it clears itself out. In the people who had the virus, it saturated the receptor at far lower concentrations and washed out about twice as fast, dropping its terminal half-life from 21 days in the uninfected volunteers to 12 days in those with HIV. It also drew anti-drug antibodies, an immune response against the therapeutic itself that the investigators tracked as a prespecified outcome and that wears a molecule like this down over repeat dosing. None of that sinks a phase 1 program. All of it is the sort of thing “safe and well tolerated” is built to walk past.

It helps to know why this antibody exists, because the field has stood here before. From 2016 to 2020 the Antibody Mediated Prevention trials tested VRC01, an earlier broadly neutralizing antibody, as HIV prevention in more than 4,600 people across two continents. The result, reported in the New England Journal of Medicine in 2021, was a lesson dressed as a disappointment. VRC01 prevented about 75 percent of infections caused by the strains most sensitive to it and showed no significant protection overall, because most of the virus in circulation was not that sensitive. One antibody, however elegant, could not cover a virus that mutates for a living.

Bispecifics like 10E8.4/iMab are the answer to that miss: two grips instead of one, enough engineered breadth to neutralize nearly every strain of HIV in a dish. It neutralizes nearly every strain of HIV in a dish. It has not yet prevented or cleared a single infection in a person.

The distance between the dish and the patient is where the money goes, and it goes where it has gone for fifteen years. The trial was paid for by the National Institute of Allergy and Infectious Diseases and the Bill & Melinda Gates Foundation, the two funders behind nearly every chapter of the broadly neutralizing antibody story, the VRC01 prevention trials that failed included. The pattern is not fraud. It is momentum. A large, well-capitalized research apparatus committed to the antibody bet keeps financing the next, better version, each one first proven safe in healthy volunteers, each producing an encouraging phase 1 readout, each still years from an answer on whether it prevents or treats anything. A companion NIH antibody, VRC07-523LS, ran its own phase 1 in people without HIV and posted its own clean safety numbers; the follow-up to this trial, now enrolling in Mbeya, Tanzania, combines the two.

Ho’s laboratory has built the most broadly neutralizing antibody it has ever made, and the paper’s authors call the phase 1 results encouraging. By the standard of a first-in-human study, they are. But the three men who came in viremic went home viremic, their virus back within a week of receiving the most sophisticated molecule the program has produced. The breadth was never the hard part. Keeping it in the blood is.

Sources

  1. Nature Medicine – Bispecific 10E8.4/iMab broadly neutralizing antibody in people with or without HIV-1: a partially randomized phase 1 trial (2026)
  2. i-Base – VRC01 antibody prevents only the minority of HIV infections: AMP study results (75.4% against sensitive viruses)
  3. Nature Reviews Drug Discovery – Broadly neutralizing antibody misses in HIV prophylaxis trial (2021)
  4. BMC Immunology – Pharmacokinetic interaction assessment of the HIV broadly neutralizing antibody VRC07-523LS across three phase 1 trials (2025)
  5. New England Journal of Medicine – Corey L, Gilbert PB, Juraska M, et al. “Two randomized trials of neutralizing antibodies to prevent HIV-1 acquisition.” 2021;384(11):1003–1014.