CRISPR and Beyond: What Precision Gene Editing Can—and Cannot—Do in Medicine

Precision gene editing has crossed an important line from experimental biology into regulated medicine, but the reality is more specific—and more complicated—than the phrase “rewriting DNA” suggests. The clearest recent example came on July 1, 2026, when the U.S. Food and Drug Administration expanded the indication for Casgevy, a CRISPR/Cas9-edited cell therapy, to patients age 2 and older with sickle cell disease with recurrent vaso-occlusive crises or transfusion-dependent beta thalassemia. The FDA’s July 2026 announcement is a useful marker of how far the field has come.

It is equally important not to overread that milestone. Casgevy edits a patient’s blood-forming stem cells outside the body and then returns those cells after conditioning treatment. It is not a universal injection that finds and repairs any mutation in any organ. Meanwhile, base editing, prime editing, and direct in vivo CRISPR are moving through clinical development at different speeds.

What is verified today: CRISPR is a medicine, but the first model is ex vivo

Verified: Casgevy is an FDA-approved, autologous genome-edited hematopoietic stem cell therapy. The current FDA Casgevy page lists sickle cell disease with recurrent vaso-occlusive crises and transfusion-dependent beta thalassemia for patients age 2 and older. The therapy uses CRISPR/Cas9 editing in collected blood stem cells to increase fetal hemoglobin after those cells are returned to the patient.

Context matters: this treatment does not simply “repair the sickle mutation.” Its therapeutic strategy changes regulation of hemoglobin production. It also involves stem-cell collection, manufacturing, myeloablative conditioning, infusion, engraftment, and specialized follow-up. That means the editing step can be precise while the overall treatment remains medically intensive.

Action: when evaluating any gene-editing treatment, ask two separate questions: “What DNA change is being made?” and “What does the entire treatment process require?” The second question often determines who can realistically receive the therapy.

Misconception: CRISPR is one tool that can make any desired DNA change

“CRISPR” is often used as shorthand for a family of programmable editing systems. Classical CRISPR nucleases, base editors, and prime editors solve different problems. Their usefulness depends on the mutation, target tissue, required edit, delivery method, and tolerance for unintended changes.

Base editing can avoid a double-strand break, but “no cut” does not mean “no risk”

Verified: the original 2016 base-editing work showed programmable conversion of a target base without a DNA double-strand break or donor template. The foundational paper is indexed by the U.S. National Library of Medicine at PubMed. Clinical evidence has now moved beyond cell culture. A 2026 New England Journal of Medicine study reported phase 1–2 results from base-edited blood stem cells for sickle cell disease and concluded that the findings supported further investigation. Another 2026 NEJM study of VERVE-102 reported dose-dependent, sustained reductions in PCSK9 and LDL cholesterol after in vivo base editing.

Context matters: these are different therapeutic problems. One edits stem cells outside the body and reinfuses them; the other delivers an editor to the liver in vivo. Success in one tissue does not prove that the same delivery system will work in the brain, lung, skeletal muscle, or other difficult targets.

Action: look for the target organ and delivery vehicle before comparing headline results from two gene-editing programs. “Base editing” alone does not make two therapies equivalent.

Prime editing expands the menu of possible edits, but it is earlier in clinical development

Verified: prime editing was introduced in a 2019 Nature study as a programmable system capable in human cells of targeted substitutions, insertions, and deletions without requiring a double-strand break or a separate donor DNA template. In 2026, a first-in-human program using PM359 for p47phox autosomal recessive chronic granulomatous disease is listed on ClinicalTrials.gov.

Unknown: early clinical entry does not yet tell us how broadly prime editing will outperform nuclease editing or base editing in safety, durability, manufacturing, cost, or therapeutic effect. Those comparisons will require disease-specific data and longer follow-up.

Action: treat “can make more types of edits” as a technical capability, not as proof that the technology is already the best treatment for more diseases.

Misconception: precision means unintended editing has been solved

Verified: regulators still treat off-target editing, unintended on-target changes, and genome integrity as central safety questions. In April 2026, the FDA issued a draft guidance on next-generation sequencing for genome-editing safety assessment. It specifically addresses risks including off-target editing and unintended changes to the genome. The July 2026 Casgevy approval letter also describes postmarketing requirements related to secondary malignancies and off-target effects; the FDA approval letter is the primary source.

Context matters: “off-target” is not the only category of risk. A therapy can hit the intended DNA site but still create an unwanted local outcome, edit only a fraction of relevant cells, trigger immune responses to a delivery component, or require conditioning that carries its own risks. A highly accurate editor can therefore still be part of a high-complexity therapy.

Action: when reading a study, look beyond the single percentage labeled “editing efficiency.” Check how investigators measured off-target sites, large genomic changes, adverse events, cell engraftment or tissue exposure, and how long patients were followed.

Misconception: one dramatic patient result proves a reusable cure platform

Verified: in 2025, investigators reported a patient-specific in vivo base-editing therapy for an infant with severe CPS1 deficiency. The original NEJM report described two infusions at approximately 7 and 8 months of age, improved tolerance of dietary protein, reduced need for a nitrogen-scavenger medication during early follow-up, and no serious adverse events in the reported observation period. The authors explicitly stated that longer follow-up was warranted.

What that establishes: it is powerful evidence that a personalized editor can be designed, manufactured, reviewed, and delivered on a clinically meaningful timeline for a single patient.

What it does not establish: it does not prove long-term safety, durable benefit across a population, or that every ultra-rare mutation can be handled with the same speed. Different variants may require different guide designs, editor chemistries, delivery strategies, assays, and manufacturing controls.

Action: interpret N-of-1 success as a feasibility milestone. For a treatment decision or policy conclusion, wait for mutation-specific evidence, longer follow-up, and a clear regulatory pathway.

The biggest bottleneck may be delivery, not the editor

Verified: gene editors must reach enough of the correct cells while limiting exposure elsewhere. The liver is comparatively accessible to several delivery systems, which helps explain why multiple in vivo programs target liver-produced proteins. Other tissues remain harder. The NIH Targeted Genome Editor Delivery Challenge explicitly highlights difficult delivery problems, including crossing the blood-brain barrier.

A major in vivo CRISPR program illustrates how far the field has progressed without proving that every tissue is equally tractable. The phase 3 MAGNITUDE study of NTLA-2001 for transthyretin amyloidosis with cardiomyopathy is listed as recruiting on ClinicalTrials.gov, with a single infusion being evaluated against placebo.

Action: for any proposed in vivo therapy, identify the delivery vehicle, target cell type, route of administration, and whether the target tissue has been reached effectively in humans—not just in animal models.

What could speed precision editing over the next few years?

The future is unlikely to be a single winner called “CRISPR 2.0.” Progress is more likely to come from reusable platform knowledge: validated delivery systems, better guide design, standardized off-target assays, manufacturing processes, and regulatory approaches that can be reused across related products.

Verified regulatory signal: in June 2026, the FDA published a draft guidance on leveraging prior knowledge for genome-editing gene therapies. The draft discusses how public and platform knowledge may help make development more efficient, including for rare diseases. Because it is draft guidance, it should not be treated as a final rule or guaranteed shortcut.

Unknown: how much platform reuse will reduce development time and cost in practice will depend on how similar products truly are in editor components, target cells, manufacturing, delivery, dose, and disease biology.

Action: watch for evidence of validated platforms rather than promises of “one platform for thousands of diseases.” The strongest sign of scalability will be repeated success across related products with shared components and comparable safety testing.

Heritable editing is not simply the next clinical step

A common misunderstanding is that successful treatment of somatic cells naturally leads to editing embryos for inherited disease. The ethical and regulatory boundary is much sharper than that.

Verified: the World Health Organization distinguishes somatic genome editing from germline and heritable editing. Its current human genome editing guidance continues to emphasize greater safety and ethical concerns for heritable editing and cites its position that proceeding with clinical applications of human germline genome editing would be irresponsible at this time.

Action: separate claims about treating a patient’s somatic cells from claims about editing embryos or changes intended to pass to future generations. They are scientifically, ethically, and legally different categories.

How to judge a precision gene-editing claim

• Check regulatory status. “FDA approved,” “FDA cleared to begin a trial,” and “registered on ClinicalTrials.gov” are not the same thing.
• Identify ex vivo or in vivo editing. Editing collected cells outside the body has different delivery and safety challenges from sending an editor directly into an organ.
• Match the editor to the mutation. A nuclease, base editor, and prime editor can solve different classes of sequence changes.
• Read the original evidence. Prefer the regulator, trial registry, or peer-reviewed original study over a press release summarizing it.
• Look at follow-up length. A strong early biomarker or symptom improvement does not answer questions about durability over years or decades.
• Ask what else the treatment requires. Conditioning chemotherapy, stem-cell collection, hospitalization, immunosuppression, or specialized monitoring may matter as much as the editor itself.

The likely future: better matching, not magical editing

Precision gene editing is becoming a real therapeutic modality, not a futuristic metaphor. By 2026, CRISPR has an expanding FDA-approved use in blood disorders, base editing has produced human clinical data in both ex vivo and in vivo settings, prime editing has entered clinical development, and personalized editing has shown that an N-of-1 medicine can be built for a devastating rare disease.

The remaining questions are exactly the ones that matter in medicine: Which patients benefit? How durable is the effect? Which tissues can be reached safely? What unintended changes occur? Can manufacturing and monitoring scale? And can access expand without weakening safety standards?

The most useful way to follow the field is therefore not to ask whether CRISPR “works.” It already does in defined settings. Ask which editing technology, delivered to which cells, for which disease, with what evidence, for how long. That is where the future of precision gene editing will be decided.