For more than a decade, gene editing was a technology of "someday." In 2026, that tense changed. On February 23, 2026, the U.S. Food and Drug Administration (FDA) released a draft regulatory framework meant to speed the development of individualized treatments for genetic diseases — gene editing included — aimed at a single patient [source: FDA, 2026]. And before that, at the Children's Hospital of Philadelphia (CHOP), an infant survived after receiving a CRISPR therapy designed for him alone [source: Children's Hospital of Philadelphia, 2025]. "A treatment for one person" stopped being a metaphor and became an actual event.
Why is the whole world watching this now? Gene editing has, until recently, focused on relatively common single-gene diseases such as sickle cell disease — because a market has to be large enough to recoup development costs. But as regulation and technology moved at the same time, diseases so rare that a patient may be the only one in the world began entering the realm of the treatable. This article separates what has been announced, what was actually measured, and what remains an unverified hope. In particular, it stresses that "a success in one person" and "a forecast that it will work in many" are two entirely different tiers of evidence.
A note on how this article reads its evidence. Four different kinds of statement appear below, and they are not interchangeable. A regulator's draft document is a proposal opened for comment, not a rule in force. A case report describes what happened to one identified patient, with a sample size of one. An approved product carries a record built from completed trials and a marketing authorization. A company plan or market forecast is a statement about the future made by a party with an interest in it. Wherever a number below describes a single patient, this article says so explicitly, because a result that is not labeled with its sample size reads like a rate.
Table of Contents
- In 2026, regulation moved toward "a treatment for one"
- Baby KJ — gene editing for a single patient becomes real
- The evolution of the gene scissors and 150 trials
- Between a single success and generalization — what has been proven
- Cost and equity — the real bottleneck of bespoke therapy
- What to watch
In 2026, regulation moved toward "a treatment for one"
Why the standard trial design breaks down
The backbone of conventional drug approval is the randomized controlled trial (RCT): hundreds or thousands of people split into drug and placebo arms to prove effect statistically. But in ultra-rare diseases with only a handful of patients worldwide, that method simply does not hold, because there is no one to form a control group with.
What the draft framework offers
The draft guidance the FDA issued on February 23, 2026, offers a way around that dead end. Its official name is the "Plausible Mechanism Framework for developing individualized therapies that target specific genetic conditions with a known biological cause" [source: FDA, 2026]. The gist is this. First, in place of a control arm, sponsors may use well-characterized natural-history data as an external comparator. Second, they confirm that the target gene was actually edited (target engagement). Third, multiple editing products aimed at different mutations across different patients can be bundled under a single master protocol and one application, and new variants can be added even after approval on the strength of the same mechanism [source: FDA, 2026].
The provision that changes the economics
That third provision is the crux. In gene-editing therapy, the target gene may be the same while the exact mutation to fix differs from patient to patient. If every case demanded a brand-new trial from scratch, "a treatment for one" would be economically impossible. This framework opens a path in which an editing "platform" is validated once, and per-mutation therapies are layered on top of it. One thing must be clear, however: this is draft guidance, not a settled rule. The comment period ran through April 27, 2026, and how the final version will be shaped is still open [source: FDA, 2026].
What it covers, and what status it has
The framework's full title states what it covers: "Considerations for the Use of the Plausible Mechanism Framework to Develop Individualized Therapies that Target Specific Genetic Conditions with Known Biological Cause." Its scope is not limited to gene editing. It also takes in RNA-based therapies — antisense oligonucleotides (ASOs) and small interfering RNA (siRNA) — and it is aimed at rare, severe, or life-threatening genetic conditions. The document additionally leaves open the possibility of extending the same logic to more common diseases [source: FDA, 2026].
It is worth being precise about the status of that document, because the distinction does real work later in this article. The FDA opened a 60-day public comment period on the framework, and it closed on April 27, 2026 [source: FDA, 2026]. Everything described above is therefore a proposal the agency has put forward and invited responses to — not a settled requirement a sponsor can build a filing on today. When this article says "the framework allows" something, read it as "the draft proposes to allow."
A second guidance, pointing the other way
Regulation moved on the safety side too. On April 14, 2026, the FDA issued a separate draft guidance standardizing how to assess the off-target risks of gene editing — edits at unintended sites — using next-generation sequencing (NGS) [source: FDA, 2026]. A framework that opens speed and a guidance that tightens safety arrived side by side in the same year.
That second document has a name of its own: "Safety Assessment of Genome Editing in Human Gene Therapy Products Using Next-Generation Sequencing." It sets out how to use NGS to assess not only off-target edits but damage to genome integrity, and it expands a guidance the agency had issued in January 2024. Its own comment period ran to July 14, 2026 [source: FDA, 2026]. Both of the 2026 documents in this section are drafts. Regulation moved in two directions at once — one toward speed, one toward safety — and neither move is final.
Baby KJ — gene editing for a single patient becomes real
The disease
What stunned the world before any regulatory document did was an infant. KJ Muldoon was born with severe carbamoyl phosphate synthetase 1 (CPS1) deficiency, a urea-cycle disorder in which the liver cannot convert ammonia — a byproduct of protein metabolism — into urea for excretion. As toxic ammonia accumulates, it damages the brain; the condition is an ultra-rare, life-threatening disease [source: Children's Hospital of Philadelphia, 2025].
How the therapy was built
A joint CHOP and Penn Medicine team designed a base-editing therapy tailored to KJ's specific mutation. Base editing is a branch of CRISPR that rewrites a single letter (a point mutation) without cutting the DNA double strand. The team loaded this editing tool into lipid nanoparticles (LNPs) and infused it intravenously; the particles entered liver cells and corrected the gene directly inside the body — so-called in vivo editing. The time from diagnosis to manufactured therapy was about six months [source: Children's Hospital of Philadelphia, 2025].
The work was led by Kiran Musunuru of Penn Medicine and Rebecca Ahrens-Nicklas of the Children's Hospital of Philadelphia, with the two institutions running it jointly [source: Children's Hospital of Philadelphia, 2025]. That pairing is worth noticing when reading the result. Designing a mutation-specific editor is a laboratory problem; dosing an infant with a urea-cycle disorder is a hospital problem; and the six-month figure covers both happening at once. The clock in that number starts at diagnosis, not at the moment a design was settled on, so it is not a manufacturing time on its own.
What was measured, and over what window
KJ received his first dose in February 2025, at around six to seven months of age, followed by additional doses over the next several months. The results were encouraging. After a total of 307 days in the hospital he went home; following treatment he tolerated more dietary protein and could cut back on the medicine that clears ammonia. He reached developmental milestones such as sitting up, and even when he caught a rhinovirus his ammonia did not spike and he recovered [source: Children's Hospital of Philadelphia, 2025]. The case was published in the medical journal NEJM on May 15, 2025, and presented the same day at the American Society of Gene & Cell Therapy (ASGCT) [source: NEJM, 2025]. Rebecca Ahrens-Nicklas, who led the work, called the "initial findings quite promising" [source: Children's Hospital of Philadelphia, 2025].
Two details are worth holding onto. First, published accounts of the case differ on whether the additional dosing amounted to two doses or three, which is why the careful formulation is the one used above — a first dose, then additional doses over the following months — rather than a specific count. Second, the observation window is short. The first dose was given in February 2025, and the case was published on May 15, 2025, the same day it was presented at the ASGCT meeting in New Orleans [source: Children's Hospital of Philadelphia, 2025]. The reported follow-up is measured in months, not years, so nothing in it speaks to durability or to long-term safety.
What generalizes from a sample of one
One point must be made clear here. The genuinely reusable achievement in KJ's case is the method (the platform) — rapid design, LNP delivery, base editing — not proof that the specific edit used on KJ will work in another patient. The success of a single patient (N-of-1) is astonishing in itself, but it is a different kind of evidence from a clinical trial that establishes statistical efficacy.
Put the arithmetic plainly: the sample size is one. A single case has no control arm and yields no response rate, because a rate needs more than one patient to be a rate at all. What KJ's case demonstrates is that the sequence — diagnose, design an editor for that individual's mutation, manufacture it, deliver it in vivo, observe — can be completed inside a hospital's timeframe for a living infant. That is a substantial thing to have demonstrated. It is not the same thing as showing the approach helps patients in general, and the distance between those two statements is what the next section is about.
The evolution of the gene scissors and 150 trials
Where the trial count comes from
KJ's case stands out, but behind it is a widened clinical landscape. The specialist tracker CRISPR Medicine News monitors roughly 250 trials involving gene-editing candidates, of which more than 150 are currently active [source: Innovative Genomics Institute, 2026]. It is worth keeping in mind that this figure is an estimate from a specialist tracker, not an official count from a regulator.
It is worth knowing how that count is assembled. The figure comes from CRISPR Medicine News, a specialist tracking service, and reaches wider circulation through the Innovative Genomics Institute's "CRISPR Clinical Trials: A 2026 Update" [source: Innovative Genomics Institute, 2026]. Two things follow from that. A tracker's tally is a compilation, so it can move as entries are added or reclassified, and it is not a regulator's official register. And "more than 150" is a floor rather than a precise figure — the honest reading is at least that many, out of roughly 250 being monitored.
From cutting to rewriting
The technology itself has evolved. Where the original CRISPR-Cas9 works by "cutting and repairing" DNA, base editing swaps a single letter without a double-strand break, improving the safety profile. Going further, prime editing — a more versatile "search-and-replace" approach — is entering early trials [source: Innovative Genomics Institute, 2026]. The higher the precision, the wider the door for bespoke therapies aimed at a specific point mutation, as in KJ's case.
The two techniques sit at different stages of maturity, and that difference is easy to flatten. Base editing already has candidates in the clinic; Beam Therapeutics is among the companies that have taken it into trials. Prime editing, the more versatile search-and-replace approach, is at the point of entering Phase 1 [source: Innovative Genomics Institute, 2026]. Phase 1 is the earliest stage of human testing, where the questions are dose and safety rather than efficacy. Both belong in the story of where editing is heading; neither is a treatment on the shelf today.
The approved precedent, and its limits
At the head of this current is a precedent already on the market. Casgevy (exagamglogene autotemcel) is the first CRISPR medicine approved in the United States, developed by Vertex and CRISPR Therapeutics. It was approved for sickle cell disease on December 8, 2023, and for transfusion-dependent beta thalassemia on January 16, 2024 [source: Vertex, 2023]. But Casgevy differs in kind from KJ's approach: it is an ex vivo, autologous therapy in which the patient's cells are removed, edited, and returned, requiring infrastructure on the level of a bone-marrow transplant. That is exactly why KJ's in vivo approach draws attention — in principle it has room to apply more broadly.
Two boundaries around that precedent are worth stating. Casgevy's approvals cover patients aged 12 and older [source: Vertex, 2023]. Separately, a Phase III trial in children aged 5 to 11 is under way, with a regulatory filing expected in the first half of 2026 — that second item is a plan on a developer's timetable, not an approval, and it is listed here as such. The distinction matters whenever a headline compresses "trial under way" and "available treatment" into a single sentence.
Between a single success and generalization — what has been proven
Sorting the claims by tier
Let us sort the story so far by tier. The verified facts are these: gene editing was used to treat one infant with an ultra-rare disease, and he improved clinically; a regulator issued a draft framework for such individualized therapies; and more than 150 editing trials are under way.
Sorted into a ledger, the claims in this article fall into four columns. Measured and peer-reviewed: one infant's clinical course, published in NEJM, with a sample size of one and a follow-up measured in months. Proposed: the FDA's February 2026 framework and its April 2026 safety guidance, both drafts, both past their comment deadlines. Completed: Casgevy's two approvals, in December 2023 and January 2024, for patients aged 12 and older. Compiled or projected: a specialist tracker's count of active trials, a developer's pediatric filing timetable, and the market's expectation that costs will fall. Nothing in that fourth column should be read with the weight of the first.
Where the evidence stops
What is not yet verified is just as clear. First, whether the edit that worked for KJ will produce the same effect in another CPS1-deficiency patient with a different mutation is a separate question. Second, whether short-term improvement in a single patient carries over into long-term safety and durable effect needs more time. Third, the off-target risk of gene editing — edits at unintended sites — remains enough of a concern that the FDA saw fit to issue a separate safety guidance [source: FDA, 2026].
Three specific gaps sit behind that phrase. There is no response rate to cite, because one patient cannot produce one. There is no long-term safety record, because the published follow-up spans months rather than years, and durability is a question time answers rather than one a case report answers. And off-target editing — the risk that the tool changes a site it was not aimed at — is enough of an open problem that the FDA wrote a guidance on how to measure it, a guidance that is itself still a draft [source: FDA, 2026]. None of this diminishes what was achieved. It describes where the evidence currently stops.
The crux is this: what opened is a path to build and evaluate individualized therapies, not a demonstration that individualized therapies are broadly effective and safe. Between a method's door opening and its results generalizing lies a great deal of evidence still to be filled in. The more sensational the subject, the easier it is to blur that distinction.
Cost and equity — the real bottleneck of bespoke therapy
Even with technology and regulation in place, a question remains: who gets to receive this treatment?
The prices that get reported
Gene therapies today are generally reported to cost in the millions of dollars per patient (roughly $2–3.5 million, some higher) [source: AJMC, 2026]. Individualized therapies carry an even more fundamental problem. Because each is a "one-off" product demanding its own biology and its own manufacturing process, economies of scale struggle to kick in. If the market is a single patient, there is no one to spread development and manufacturing costs across — which is also why payers hesitate [source: AJMC, 2026].
The reported range has an upper edge worth naming. Gene therapies are generally described in the range of roughly $2 million to $3.5 million per patient, with some reported as high as about $4.25 million, and annual US spending on gene therapy has been estimated at about $25.3 billion for 2026 [source: AJMC, 2026]. Two cautions attach to those numbers. They describe approved gene therapies as a category, not the cost of a one-patient individualized therapy, which is a different manufacturing situation altogether. And the spending figure is an estimate for a year, not an audited total.
Where the access gap falls
Equity compounds it. A therapy like Casgevy, which needs transplant-level infrastructure, is realistically hard to access in under-resourced regions. The fact that a large share of people with sickle cell disease live in places with the weakest health systems, such as sub-Saharan Africa, shows how the access gap can widen the further the technology races ahead [source: AJMC, 2026]. Add the bias that large genomic databases skew toward people of European ancestry, and even the accuracy of "personalization" can tilt in favor of particular populations.
The scale of that gap can be stated more concretely. About 80% of people with sickle cell disease live in sub-Saharan Africa [source: AJMC, 2026] — which is to say the disease with the first US-approved CRISPR treatment is concentrated in the region least able to supply the transplant-level infrastructure that treatment requires. A therapy can be approved, effective, and out of reach for most of the people who have the condition it treats, all at once; those three things are not in contradiction.
Reading a forecast as a forecast
So caution is warranted. Forecasts of the "thanks to in vivo editing or platformization, it will soon get cheap and universal" variety are largely estimates laced with market expectation, and they have not been independently verified. That manufacturing becomes theoretically easier and that real-world price and access fall accordingly are two separate matters. To date, cost and scalability remain the field's largest unresolved bottleneck.
What to watch
To sum up, 2026 is the year "gene therapy for one" drew near across three axes at once — regulation, technology, and an actual case. The verified facts are clear: baby KJ improved with a bespoke base edit, the FDA issued a draft framework for individualized therapies, and editing trials passed 150. At the same time, the open questions are clear too. A single success is not yet generalization, off-target safety is still a managed concern, and "cheap, universal bespoke therapy" is a forecast, not a settled fact.
What would change the picture
Three things are worth watching from here. First, how far the final FDA framework turns natural-history data and master protocols into actual approvals. Second, whether cases after KJ move beyond a single success to be reproduced across many patients and to accumulate long-term safety. Third, whether the advantages of in vivo editing translate into real price drops and equitable access. The door has clearly opened. How the efficacy and equity beyond it get filled in will be the next chapter of the story.
Four questions for the next headline
For anyone reading the next headline in this field, four questions do most of the work. How many patients — one, or a cohort? Over how long — months, or years? Is the regulatory document a draft or a final rule? And is the product approved, in trials, or on a company's timetable? Each of the claims in this article answers those four questions differently, and the answers, rather than the announcement, are what tell you how much weight a result can carry.