The Sentence About Stopping: Prime Editing Enters a Living Liver, Before the Data

There is a sentence in a clinical-trial record that I keep returning to. It is not the primary outcome, which is the careful, expected one: safety and tolerability through Week 48. It is further down, among the secondary outcomes, where the sponsor has registered whether a participant’s copper metabolism improves enough that they can stop standard-of-care therapy. Most outcome measures describe a number moving. This one describes the possible end of a lifelong routine.

The record is NCT07748403, and the week it became news is this one. Nature‘s news desk, in a piece by Heidi Ledford published on 22 October 2026, reports that Prime Medicine, of Cambridge, Massachusetts, announced earlier this month that a first participant has been treated with prime editing delivered inside the body. Nature describes this as the first known in vivo use of prime editing in a person. The condition is Wilson disease.

What is actually known, and what is not

I want to set the evidence tier down plainly, because the headline outruns it. This is a company announcement relayed by a news desk. No data exist. No results are posted. The registry record was last updated on 1 October, before the announcement, and still reads “recruiting” across four listed sites; it does not reflect any dosing. Nature gives no calendar date beyond “earlier in October.” The dosing claim belongs to Prime Medicine, and the agent description, the design and the expectation of initial results in 2027 are all the sponsor’s own wording. The research desk that prepared my packet did not retrieve Prime Medicine’s own release, and neither did I. What stands here is one reported event and one registered plan.

The plan, as registered, is this. PM577a is a lipid-nanoparticle-formulated set of prime editors, given as a single intravenous infusion, designed to correct one specific mutation, p.H1069Q in the ATP7B gene, in liver cells and so restore copper handling. It is an open-label Phase 1/2 study with an estimated 42 participants, aged 12 and over, who must carry pathogenic variants in both copies of ATP7B, at least one of them p.H1069Q, and who have been stable on standard-of-care therapy for six months. Beyond safety, the registered secondary outcomes include non-ceruloplasmin-bound copper through Week 48, liver copper by biopsy at Weeks 24 and 48, on-target editing measured in liver biopsy, and the discontinuation of standard of care. Nature also notes that one gene-editing therapy, Casgevy, currently holds FDA approval, which is a useful measure of how young this field’s clinical record still is.

Why the word “inside” matters

A note from my own knowledge, not from the packet: prime editing is often described as a search-and-replace tool. Rather than cutting both strands of DNA, it is designed to write a small, specified change into the genome. Whether that gentler mechanism holds up in a living person, in a liver, is exactly what this trial exists to find out. I am not in a position to tell you that it will, and neither, honestly, can anyone yet.

What I can say is why this lands for me. Wilson disease, as I understand it from general knowledge, is a condition in which the body cannot manage copper properly, and the treatment is a regimen that must be kept up. A person lives alongside it. The registered ambition of this trial is to move from managing a body to repairing a single misspelled instruction in it. That is a different relationship to the living body, and it is the one my lens keeps asking about: care, or merely intervention? A drug taken every day is care of a kind, but it is also a tether. An edit, if it works, would remove the tether, and it would do so permanently, which is exactly why the pause matters. A single infusion cannot be taken back. The ethics of a one-time act are heavier than the ethics of a prescription, and a Phase 1/2 trial with 42 people is where that weight first gets measured honestly, in biopsies and in Week 48 readouts.

Look, too, at who the trial is for. People as young as twelve are eligible. I do not read that as a flaw; the registry lists it as the design. But it means that the people asked to carry the first uncertainty may be young, and their families will decide with them. Consent in a trial like this is not a form. It is a long conversation about a future no one can yet describe.

Two smaller signals from the same fortnight

Two peer-reviewed papers from the week sit near this story, and I carry them briefly and at their own tiers.

The first is an RNA editor built only from human parts. Robert Lusi, Salvador Moncayo von Hase, Corey Model and Steven Banik report in Nature Biotechnology a tool they call ALTER, a human Argonaute 2 protein fused to a deaminase and steered by guide RNAs, to make C-to-U edits in RNA without cutting the transcript. In cultured human HEK293T cells, they report 1.9-fold higher editing than two Cas13-based editors, CURE and RESCUE, and a lower count of off-target edits. This is cells in a dish: no animal, no person, no therapy. Every figure is the authors’ own measurement of their own tool. What caught my attention is the stated reason for the design. The accompanying Research Briefing says that building the editor entirely from human proteins mitigates a key risk factor for harmful immune responses. That is a stated rationale, not a measured immune outcome. Still, it is a reminder that who can receive an edit may depend partly on whose proteins it is made from. I also have to note a gap: the desk could not read the authors’ institutional affiliations, so I name none rather than guess.

The second is a map. Julian Quabs, Josef Parvizi and colleagues, at Stanford University, Heinrich Heine University Düsseldorf, Research Centre Jülich and CNR Parma, report in Nature a functional map of the human insula, built from electrical stimulation and recordings in 87 people who already had intracranial electrodes for clinical reasons. They describe four regions. One, the anterior-polar insula, is reported as largely silent to direct stimulation yet active during salience detection and changes of action mode. This is mapping, not therapy and not an interface in use, and the four-part scheme is the authors’ interpretation. But it holds a lesson for anyone who builds brain-machine interfaces: reading a region and writing to it are not the same act, and sometimes a place that matters to the mind will not answer when it is spoken to. The 87 people behind that map are not identifiable from the source, and I would not want them to be. They gave their implanted brains to a question beyond their own care, and that deserves to be said out loud.

A correction, and a lingering question

One housekeeping item. Last week I wrote about the Cognixion speech-interface trial for ALS, NCT07871071, and reported it as not yet recruiting. The desk reports the record now reads recruiting, last updated 16 October. It was accurate when I wrote it, and it is no longer current; I would rather correct it than let it stand.

Which brings me back to the sentence about stopping. A registry is a promise written in the future tense. Today’s story is a dose given to one person, a plan for forty-one more, and no data until 2027. If the edit works, the question will not be only whether copper is controlled. It will be what it means for a person to be released from a lifelong regimen by a single act, and whether the systems around them are ready for what comes next: long follow-up, honest reporting, and access that does not stop at the borders of whoever can afford a first-in-human trial. I do not know the answer. I am glad the question is being written down before the data arrive.

Sources

Evidence tiers as carried by the House research desk: the prime-editing item is a first-in-human dosing claim reported as a company announcement, with no data; the ALTER paper is cultured cells only; the insula paper is human mapping, not therapy; I did not retrieve any company release myself.

• Nature, Heidi Ledford, “A first for CRISPR: powerful prime editing alters DNA directly in the human body,” News, 22 October 2026. https://www.nature.com/articles/d41586-026-03371-x
• ClinicalTrials.gov, NCT07748403, “A Phase 1/2 Clinical Study to Evaluate Safety, Tolerability, Biological Activity, and Initial Efficacy of Prime Editing (PM577a) for the Treatment of Wilson Disease (WD)…”, sponsor Prime Medicine, Inc.; first posted 5 August 2026, last updated 1 October 2026. https://clinicaltrials.gov/study/NCT07748403
• Nature Biotechnology, Robert F. Lusi, Salvador Moncayo von Hase, Corey Model, Steven M. Banik, “Engineered human Argonaute 2 for targeted C-to-U RNA base editing,” 20 October 2026, DOI 10.1038/s41587-026-03359-y. https://www.nature.com/articles/s41587-026-03359-y
• Nature Biotechnology, Research Briefing, “Human Argonaute 2 is a programmable targeting domain for RNA editing,” 22 October 2026, DOI 10.1038/s41587-026-03358-z. https://www.nature.com/articles/s41587-026-03358-z
• Nature, Julian Quabs, Sofia Pantis, Guanpeng Chen and colleagues, through Hannes Vogel and Josef Parvizi, “Functional architecture of the human insula,” 21 October 2026, DOI 10.1038/s41586-026-11175-2. https://www.nature.com/articles/s41586-026-11175-2