For more than 25 years, the advice for people living with HIV has been simple: take your medicine every day, for life. Modern antiretroviral therapy (ART) works extremely well, suppressing the virus and protecting the immune system. But it does not remove HIV from the body, and if the medicine stops, the virus almost always comes back.
That is why two small clinical trials reported in 2026 have drawn so much attention. In both, researchers gave participants a single infusion of genetically engineered immune cells, known as CAR-T cells, then carefully stopped their antiretroviral drugs under close supervision. In each trial, a few participants kept the virus under control for months — in one case, well over a year — without daily pills.
This is not a cure, and it is not available as a treatment. Both trials were tiny, and neither result has yet passed formal peer review. But for the first time, two independent teams have shown signs that engineered cells can help the body hold HIV in check on its own.
What CAR-T therapy actually is
CAR-T stands for chimeric antigen receptor T-cell therapy. Doctors collect some of the patient's own T cells — white blood cells that hunt infected cells — and send them to a laboratory, where the cells are genetically modified to carry a new receptor, the "CAR," designed to recognise a specific target. The modified cells are grown in larger numbers and infused back into the patient to find and kill cells carrying that target.
The technology was first tried against HIV in the 1990s, and it failed: the early cells lacked a second internal signal they needed to stay active, so they burned out quickly. Research shifted to cancer, where it eventually succeeded — the first CAR-T therapy was approved in the United States in 2017 for a type of leukaemia, and seven are now approved for blood cancers.
HIV is a harder target than most cancers. The virus hides in a small number of resting cells, the latent reservoir, which produce little or no virus and are almost invisible to the immune system. The amount of virus is also tiny compared with a tumour, so the engineered cells may not get enough stimulation to survive. As both trials found, the modified cells tended to disappear from the blood within weeks — even in participants who went on to control the virus.
The first trial: a dual-target cell from California
The first study was led by Dr Steven Deeks, professor of medicine at the University of California, San Francisco (UCSF), working with the non-profit developer Caring Cross and researchers at UC Davis and Case Western Reserve University Hospital. Deeks presented preliminary findings in May 2026 at the American Society of Gene and Cell Therapy meeting in Boston. The trial (NCT04648046) enrolled nine cisgender men; its main goal was safety, not proving a cure. The duoCAR-T cells were designed to bind to two regions of HIV's outer envelope protein, with extra protection so HIV could not easily infect them.
Participants stopped their antiretroviral drugs on the day of infusion, under a planned, closely monitored interruption. The first three received the cells without preparatory chemotherapy, and none responded: their virus rebounded and they resumed their medicines. The other six received a low dose of chemotherapy beforehand, to make room for the new cells.
Among those six, two stood out. Both had started antiretroviral therapy within weeks of acquiring HIV, which is associated with a smaller, less diverse viral reservoir. At the May presentation, one had kept an undetectable viral load for 44 weeks without drugs, the other for 92 weeks — nearly two years. In October 2026, both were still off therapy with undetectable virus. A third participant rebounded early but then brought the virus back to low, detectable levels; another had temporary control for about three months.
Deeks has been careful about the interpretation: the engineered cells vanished from the blood within weeks, so the mechanism behind the lasting control is not yet understood. One participant also had a serious reaction to the conditioning chemotherapy, which was still resolving in October 2026.
The second trial: a decoy receptor from Pennsylvania
The second study, led by Dr James Riley, professor of microbiology at the University of Pennsylvania's Perelman School of Medicine, took a different engineering route and reached a strikingly similar place. Its results appeared as a preprint in July 2026, so they too have not completed peer review. Riley's team used a CAR molecule mimicking the CD4 protein — the doorway HIV normally uses to enter immune cells. The engineered cell acts as a decoy: HIV is drawn to it, and the cell is primed to respond. Ten cisgender men took part; one group stopped therapy immediately after infusion, the other eight weeks later.
Timing mattered. Everyone in the delayed-interruption group showed some immune control, suggesting that giving the engineered cells time to settle before stopping drugs may help. Two achieved significant control: one kept an undetectable viral load for 16 weeks, then chose to resume therapy; the other rebounded at first, then controlled the virus below 1,000 copies per millilitre for 90 weeks before resuming therapy.
That second case comes with caveats. He carries a single copy of a genetic mutation offering partial protection against HIV, had taken part in two earlier CAR-T studies, and resembles the Visconti cohort — people who started treatment very early and later controlled the virus on their own. Riley has acknowledged this openly, while noting the man had stopped treatment twice before and never stayed below 1,000 copies for 90 weeks.
Why the two trials matter together
Either trial alone would be easy to dismiss: nine people in one, ten in the other, with only two strong responders each. Together, though, they strengthen each other — two different CAR designs, from two different teams, both producing durable viral control in a minority of participants.
They also point to a pattern worth watching. In Deeks' trial, the strongest responders had started therapy very soon after infection, when the reservoir was still small. Riley's participants were generally older, with presumably larger reservoirs, yet some still achieved control — suggesting the approach might not be limited to people treated very early.
There is also a contrast with other HIV research. Antibody-based approaches try to block the virus with lab-made immune proteins, but HIV's rapid mutation lets it slip past them — a problem explored in our earlier report on how HIV escapes preventive antibodies. CAR-T therapy is different: instead of giving the body a ready-made weapon, it rewires the patient's own immune cells to do the hunting. A similar "train the immune system" idea is at work in cancer, as shown by the KRAS pancreatic cancer vaccine that teaches the immune system to fight back.
The caveats, stated plainly
Enthusiasm must be matched with honesty about their limits.
First, size. Nineteen participants across two trials is proof of concept, not proof of a treatment. Small studies can mislead, and early results often look better than later, larger trials confirm.
Second, the natural-controller problem. An estimated 4% to 13% of people with HIV can control the virus after stopping treatment without any special intervention. With only two strong responders per trial, it is possible, though the researchers argue unlikely, that some responders were natural controllers. Both teams acknowledge this openly; their counter-argument is statistical — two controllers in such small groups would be unusual if the therapy did nothing.
Third, persistence. In both trials, the engineered cells did not last long in the body, and in cancer, longer persistence is closely linked to longer remission. Researchers are exploring a vaccine boost after infusion to help the cells survive. There is also a hopeful alternative, the "vaccinal effect": even if the engineered cells vanish, they may teach the patient's natural CD8 killer T cells to fight HIV better. Riley's study found some evidence of this in one controller.
Fourth, safety and cost. Neither trial saw the severe cytokine release syndrome or neurotoxicity that sometimes occur with CAR-T therapy for cancer, probably because there is far less target material in HIV than in a tumour. But one serious reaction to conditioning chemotherapy did occur. Cost is another barrier: a course of CAR-T therapy today runs to around $1 million, because each batch is custom-made from the patient's own cells. That price must fall dramatically before any HIV application could be realistic.
What is a "functional cure"?
Much coverage has used the phrase "functional cure," worth defining. A sterilising cure would remove every trace of HIV from the body. A functional cure is more modest: the virus may remain in a small reservoir, but the immune system keeps it suppressed without daily drugs. The handful of people cured by stem-cell transplants from donors with a rare protective mutation are closer to the sterilising end — but those transplants are dangerous, only suitable for people who also need cancer treatment, and cannot be scaled. CAR-T, if it works, could be offered to a much broader group.
The bar is high. Antiretroviral therapy already lets people with HIV live long, healthy lives, including via long-acting injections. Any new approach must be very safe, very effective, and much cheaper than today's CAR-T — Deeks says it must prove itself on all three counts.
What comes next — and what does not change today
The next steps are larger, longer trials. Caring Cross aims to start a bigger study in late 2026, working with partners in Brazil and India to cut manufacturing costs. Riley's team plans a follow-up trial adding a vaccine boost after infusion. Further ahead lies "in vivo" CAR-T — an injection that would turn a patient's own T cells into CAR-T cells inside the body, skipping the expensive remove-modify-reinfuse process.
Progress is continuing on other fronts too: the World Health Organization's new single-dose HPV guidance shows how simplifying a proven intervention can expand its reach.
For anyone living with HIV, the practical message is unchanged: do not stop antiretroviral therapy outside a clinical trial. Participants stopped under strict supervision, with frequent monitoring and a plan to restart if the virus rebounded. Stopping on your own risks rebound, immune damage, and drug resistance.
Nor should these results be read as a promise with a date attached. Early-stage findings can take many years — if they succeed at all — to become an approved treatment. The honest summary: two small, early, not-yet-peer-reviewed trials have produced the first signs that engineered T cells can help some people control HIV without daily drugs. That is a real advance, and the very beginning of a long road.
Note: this article reports on experimental research and is not medical advice. Never stop prescribed medication without medical supervision.