Somewhere in Valinhos, in the Brazilian state of Sao Paulo, there are cells that began life inside a child's wobbly front tooth. The donors were healthy children aged 5 to 12 who gave up deciduous teeth, the ones families tuck under pillows. From the soft pulp inside, Cellavita, working with the Instituto Butantan, grew a cell line it calls NestaCell: cryopreserved stem cells suspended in saline, expressing nestin and secreting brain-derived neurotrophic factor, dripped into a vein over an hour. The premise is not to replace the neurons Huntington's destroys. It is to bathe them in something that might keep them alive a little longer.
That premise has now travelled through two published human studies, both authored by the people who make and run the product, both appearing in Stem Cell Research & Therapy last year. The first-in-human phase I study, published on 4 November 2025 and registered back in April 2016, gave infusions to six men, three at one million cells per kilogram and three at two million, with the number of infusions ranging from four to twenty-six across five years. It was open label, with no control group, so its hints of stabilisation on the Unified Huntington's Disease Rating Scale are exactly that: hints. What it did establish was tolerability over an unusually long exposure.
The phase II trial, known as ADORE-DH and published on 6 August 2025, is the one that produced a number worth arguing about. Forty-nine people were screened and 35 randomised, at a ratio of two to two to one, to the lower dose, the higher dose, or placebo, receiving nine intravenous infusions across eleven months at a single Brazilian site. The primary endpoint was change in the UHDRS Total Motor Score, and both doses separated from placebo with a reported p value of 0.005; the higher dose also showed a benefit on Total Functional Capacity, the scale that tracks whether a person can still work, manage money, and live at home, at p equals 0.011. No serious adverse event was judged related to treatment. The MRI signal that everyone wanted, slower loss of white and grey matter, moved in the right direction but did not reach statistical significance, and the authors say so plainly.
Set that against what Huntington's does. The phase II paper notes that people typically decline by around three points a year on the motor score, which is why its authors argue that even holding steady would matter. But 35 people at one centre over eleven months, in a trial funded by the product's developer, is a signal, not a settled fact. Small single-site trials with the sponsor's own staff as investigators are precisely the design that larger, multicentre, independent replication exists to check. That is not a slight on the researchers; it is the reason they themselves called for a phase III.
Which brings us to the registry entry at the top of this story. NCT06097780, acronym STAR, sponsored by Azidus Brasil, was first posted on 24 October 2023: a randomised, double-blind, placebo-controlled phase III in early and moderate Huntington's disease, 120 participants, multiple doses, efficacy measured at one year. Its estimated start date was 8 June 2024, primary completion 8 September 2025, completion 9 February 2026. All three dates have now passed. The record's last update was submitted on 18 October 2023, its last known recruitment status was "not yet recruiting", and its overall status reads "unknown". A stale registry entry is not proof that a trial failed, stalled, or quietly started without updating its paperwork; registries go unmaintained for dull administrative reasons all the time. But for a family reading about banked baby teeth and hoping to enrol, the honest position today is that the confirmatory trial has not reported, and the public record cannot say where it stands.
For readers who have followed Huntington's through antisense oligonucleotides and gene therapies, this is a different bet entirely: not silencing the mutant huntingtin message, but delivering cells that, in rodent work, reached the brain in small numbers (about two percent detectable three days after infusion in mice, with no evidence of long-term engraftment in immunodeficient animals) and raised markers including BDNF in a chemically induced rat model. Rodent findings are rodent findings. The human evidence remains six open-label patients and 35 randomised ones, and the trial designed to turn that into an answer is, for now, a page on a website that nobody has edited in three years.