Job Baer

These Long-Acting Antibodies Kept HIV Undetectable for Months — and Seemed to Train the Immune System

Subtitle: Rockefeller, Imperial, and Oxford’s RIO trial follow-up in Nature Medicine shows dual long-acting bNAbs working with patients’ own antibodies — intact reservoir half-life ~36 weeks, and one participant still off daily ART past 160 weeks.

Antiretroviral therapy turned HIV from a death sentence into a chronic disease you can live with. It did not turn it into something you can stop thinking about. Most people living with HIV still take daily pills for life, because a tiny pool of intact latent virus — the reservoir — can reignite infection the moment drug pressure lifts. That reservoir is why “cure” has stayed a word scientists use carefully, and why every credible strategy that might shrink it or teach the body to control it without daily drugs is news.

On September 17, 2026, Rockefeller University amplified a Nature Medicine analysis of secondary and exploratory outcomes from the phase II RIO trial — a Rockefeller–Imperial College London–University of Oxford collaboration funded in part by the Gates Foundation and Rockefeller’s Stavros Niarchos Institute of Infectious Diseases. The clinical punch line was already shareable: one or two infusions of long-acting broadly neutralizing antibodies (bNAbs) could keep HIV undetectable for months after a monitored pause in daily ART. The new paper asks the harder question: how — and why some participants stayed quiet long after the infused antibodies should have washed out of their blood.

Lead voice on the mechanism work is Michel C. Nussenzweig, head of Rockefeller’s Laboratory of Molecular Immunology, with postdoctoral associate Marcílio J. Fumagalli among the front-line interpreters of the reservoir and antibody data. The paper — “Broadly neutralizing antibodies in adult males living with HIV undergoing analytical treatment interruption: secondary and exploratory outcomes of the phase II randomized controlled RIO trial” — sits at DOI 10.1038/s41591-026-04644-8. The primary time-to-rebound results live in a companion Lancet HIV report on the same dual-antibody regimen (3BNC117-LS + 10-1074-LS).

What RIO actually did

RIO randomized 68 men living with HIV to receive either the two long-acting bNAbs or placebo, then paused daily ART under close monitoring — an analytical treatment interruption. Per Rockefeller’s institutional framing of the earlier clinical readouts:

  • A single dose allowed most participants receiving bNAbs to keep HIV at undetectable levels for up to ~20 weeks without daily ART.
  • Those who maintained suppression could take a second dose at 20 weeks.
  • Half of that second-dose group remained undetectable at 48 weeks.
  • About one third remained undetectable at 72 weeks.

The Lancet HIV primary analysis sharpens the early contrast: at 20 weeks, roughly 75% of bNAb recipients versus ~11% of placebo recipients had avoided viral rebound (hazard ratio reported near 0.09). Fourteen bNAb recipients avoided rebound through 72 weeks; seven remained ART-free at 96 weeks in the longer follow-up tallies circulating with that report. This is not “everyone cured.” It is a clear, randomized signal that two long-acting antibodies can buy months — and for a minority, years — of drug-free viral control under trial conditions.

Control was not only binary undetectable-or-not. Among 29 bNAb recipients who eventually had detectable virus, 38% showed fluctuating viremia — HIV rising and falling at low levels without meeting the trial’s criteria to restart ART — for as long as 58 weeks. That pattern is awkward for a simple “antibody still circulating” story. It is friendlier to an immune-system story.

The participant past three years

The number that travels on social: one participant at 160 weeks — more than three years — still not needing ART. Fumagalli’s Rockefeller quote is blunt: after that long, circulating bNAbs would be gone. “They’ve been washed out. And yet, he still does not need ART.”

That single case is not a population cure rate. It is proof-of-possibility that something beyond passive antibody pharmacokinetics is sometimes in play. Nussenzweig’s framing in the September 17 release: the antibodies appear to be teaching the immune system to control HIV — the patient’s own immune activity acting almost as a second component that, together with the bNAbs, corners the virus.

The reservoir that usually takes years to fade

Daily ART freezes most replication but leaves intact proviruses sitting for years. Historical estimates put the half-life of that intact reservoir on the order of four to seven years under ART. In the RIO samples, intact proviruses after bNAb therapy declined with an estimated half-life of about 36 weeks — roughly 0.75 years. Defective copies, which dominate the genetic junk drawer of the reservoir, did not show the same meaningful decline. Initial reservoir size did not cleanly predict how long someone stayed off ART. Rebound viruses often failed to match the dominant intact proviruses sampled in blood, hinting at unsampled tissue sources or deeply dormant clones.

Caveat, said plainly: this is an association from secondary/exploratory analyses in a finite cohort, not a randomized proof that bNAbs alone “clear” the reservoir. Some participants experienced replication during interruption. Sample size limits how hard you can press causation. The half-life shift is still one of the most retellable quantitative claims in the September package — and one reason cure researchers care about this paper beyond the social-media weeks-off-ART headline.

The tricycle: two infused wheels, one homemade stabilizer

The surprise that may shape the next trial designs is autologous neutralizing antibodies — the patient’s own anti-HIV antibodies, long dismissed as too weak or too escape-prone to matter alone. Participants who already had them averaged about 108 weeks before restarting ART, versus about 27.5 weeks among those without them.

When virus eventually rebounded after bNAbs, it typically gained resistance to 10-1074-LS while remaining vulnerable to 3BNC117-LS. Rebound, in that reading, often begins when 3BNC117-LS levels fall and leave 10-1074-LS fighting alone. Nussenzweig’s metaphor: a tricycle. The two infused bNAbs are two wheels; the patient’s autologous antibodies are the stabilizing third. Most participants lacked that third wheel.

That is why the Rockefeller takeaways point to a concrete next strategy: vaccinate first to raise a useful autologous response, then administer bNAbs — and design antibody combinations that make single-antibody escape harder. “Not everybody responds. Not everybody does well,” Nussenzweig says. The present work supplies testable clues for making good outcomes more general.

What this is — and what it is not

It is not an HIV cure announcement. It is not a license to stop ART outside a trial. RIO enrolled adult males under intensive monitoring; results do not automatically generalize to women, children, people with different viral subtypes, or unsupervised treatment stops. bNAbs are expensive biologics; delivery, resistance, and access remain open industrial and ethical questions. The 160-week participant is a beacon, not a guarantee.

What it is: peer-reviewed clinical immunology showing that long-acting dual bNAbs can produce months-to-years of ART-free control in a randomized interruption setting, that intact reservoir decay can accelerate into a ~36-week half-life regime in these analyses, and that patient-made antibodies may be the missing stabilizer that separates short remissions from long ones. Primary mechanism paper: Fumagalli, Nussenzweig, et al., Nature Medicine (2026). Public anchors: Rockefeller University news and EurekAlert (September 17, 2026). Primary efficacy companion: Lancet HIV RIO time-to-rebound trial.

Why tonight’s shareable fact sticks

HIV therapy’s modern miracle was daily control. The unfinished miracle is durable control without daily pills — and, eventually, a smaller or quieter reservoir. RIO does not finish that story. It adds a plot twist Sunday audiences can hold: infused antibodies that seem to recruit the immune system, a reservoir half-life measured in months instead of years, a tricycle that needs a third wheel, and one person still quiet past 160 weeks.

Not a pancreatic liquid biopsy. Not a brain implant. Not a claim that pharmacies will replace ART next quarter. A clinical-immunology Sunday: two long-acting bNAbs, a monitored pause, and evidence that some immune systems learned — or were helped — to hold the line.

Sources

  1. Fumagalli, M. J., Nussenzweig, M. C., et al. (2026). Broadly neutralizing antibodies in adult males living with HIV undergoing analytical treatment interruption: secondary and exploratory outcomes of the phase II randomized controlled RIO trial. Nature Medicine. https://doi.org/10.1038/s41591-026-04644-8
  2. Rockefeller University. (2026, September 17). Teaching the immune system to control HIV / HIV antibody trial results explain why the drug works long-term. https://www.rockefeller.edu/news/40454-hiv-antibody-drug-trial/
  3. Rockefeller University / EurekAlert! (2026, September 17). Teaching the immune system to control HIV. https://www.eurekalert.org/news-releases/1144471
  4. RIO trial investigators. Time to HIV rebound after infusion of long-acting broadly neutralising antibodies 3BNC117-LS and 10-1074-LS and analytical treatment interruption (the RIO trial). The Lancet HIV. https://doi.org/10.1016/s2352-3018(26)00059-7

← All posts