A drug built in six months for one baby

Most rare genetic diseases have no therapy, because a drug built for a few hundred patients at most cannot pay for itself. The result is a long tail of conditions that are untreatable not because the biology is intractable but because the economics do not work. A case reported in the New England Journal of Medicine on 15 May 2025 attacks that problem from the other end: build the medicine for one patient, and accept that the patient is the whole market.

The patient, an infant identified in the paper only by initials, was born with a severe urea-cycle disorder, an inborn error of metabolism in which a faulty enzyme lets nitrogenous waste build up in the blood. Without intervention the condition is rapidly fatal. A team led by Kiran Musunuru of the University of Pennsylvania and Rebecca Ahrens-Nicklas of the Children’s Hospital of Philadelphia set out to correct the specific mutation the child carried. From diagnosis to infusion took about six months, which is unprecedented for a custom biological therapy.

The tool was an adenine base editor. Unlike classic CRISPR, which cuts both strands of DNA and relies on the cell’s error-prone repair, a base editor chemically converts one letter of the genetic code into another without breaking the double helix. The team packaged the editor’s messenger RNA and a guide RNA inside a lipid nanoparticle, the same delivery vehicle used in approved vaccines, and infused it. Because these nanoparticles naturally accumulate in the liver, the edit landed in the hepatocytes where the faulty enzyme is made. The baby received doses at roughly seven and again at eight months of age.

What makes the case a precedent rather than a curiosity is the speed. A conventional gene therapy moves through years of chemistry, manufacturing and regulatory design before it reaches a patient. Here the mutation was sequenced, an editor designed against it, the construct built, formulated and infused inside a window short enough that the disease had not yet progressed beyond treatment. The authors report that the infant’s symptoms improved and metabolic markers moved in the right direction, and the accompanying editorial frames the work as the first N-of-1 gene-editing study.

The limitations are real and should not be papered over. A single patient is not a trial; the early improvement could yet fade, and the long-term safety of an in-vivo edit in an infant is unknown. The therapy is somatic, so the change is confined to this child and will not be inherited, which is the ethical point but also means it scales one person at a time. And six months is possible only because the team already had the delivery platform, the editor chemistry and the regulatory pathway in place; a less prepared institution could not repeat it on that clock.

That said, the logic generalizes. If a nanoparticle liver delivery is reusable and an adenine editor can address a large class of point mutations, then every patient with a single-gene liver disorder becomes a candidate for a bespoke therapy designed in months. The cost per patient would remain enormous, but so does the cost of losing a child, and the arithmetic shifts when the alternative is no treatment at all.

The paper’s deeper claim is about how drug development is organized. Rare diseases have been left untreated because markets are too small, but a personalized workflow attacks the fixed cost of creating each therapy rather than trying to spread it over many patients. A handful of such cases will not remake medicine, yet as a proof of the assembly line, the one-child paper may matter more than many larger trials.