Medicine and therapeutics

What counts as a gene therapy

The phrase covers two different jobs. One is gene addition: putting a working copy of a gene into cells that carry a faulty version, without touching the faulty copy. The other is gene editing: changing the sequence that is already there. Both are regulated as advanced therapy medicinal products in the European Union and as biologics in the United States, and both are usually given once rather than repeatedly.

There is a second split that matters just as much. In an ex vivo treatment, cells are taken out of the patient, modified in a facility and returned. In an in vivo treatment, the modifying agent is given directly — into the eye, the bloodstream or a muscle — and the editing happens inside the body. Ex vivo work gives you a chance to check the product before it goes back in. In vivo work skips the manufacturing bottleneck but leaves you with far less control.

The delivery problem

Getting a gene into the right cells is the part that keeps failing, and it shapes everything else. Adeno-associated virus, usually shortened to AAV, is the workhorse for in vivo delivery: different serotypes favour different tissues, the genome stays mostly outside the chromosomes, and expression can persist for years in cells that do not divide much. The constraint is size. AAV packages only a few kilobases of DNA, which is why large genes such as dystrophin need shortened versions rather than the full sequence.

Lentiviral vectors, derived from HIV, integrate into the genome and carry more cargo, which suits ex vivo work on blood stem cells. Integration is also the historic worry: early trials using gamma-retroviral vectors for severe combined immunodeficiency produced leukaemias when the vector landed near a growth-promoting gene, and vector design since then has been built around reducing that risk. Lipid nanoparticles, familiar from mRNA vaccines, are the non-viral option; they go to the liver readily and elsewhere with difficulty.

Immunity complicates all of it. Many people already carry antibodies to common AAV serotypes, which can exclude them from treatment, and the immune response to a high dose is a known safety concern rather than a theoretical one.

Adding, cutting, rewriting

CRISPR-Cas9 is the best-known editing tool: a short guide RNA directs the Cas9 nuclease to a matching sequence and it cuts both strands. What happens next is the cell's own repair, which is messy — small insertions and deletions that usually knock a gene out. That is genuinely useful when disabling something is the goal, which is how the sickle cell and beta thalassaemia treatment Casgevy works: it disrupts a regulatory region so that patients' cells make fetal haemoglobin again.

Base editing converts one DNA letter into another without a double-strand break. Prime editing writes a short new sequence in using a reverse transcriptase fused to a nicking Cas protein. Both were developed to reduce the unpredictability of repair, and both trade some flexibility for control. Off-target activity is assessed with sequencing-based assays as part of the regulatory package; it is a known question, not an ignored one.

Engineered cells

CAR-T therapy sits slightly apart. A patient's T cells are collected, given a chimeric antigen receptor that recognises a marker on cancer cells, grown up and infused back. Kymriah and Yescarta were the first two approved in the United States and Europe, both aimed at B-cell cancers, both targeting CD19. Before the infusion the patient receives lymphodepleting chemotherapy so the engineered cells have room to expand, and the characteristic side effects — cytokine release syndrome and neurological toxicity — are managed in specialist centres.

The trade-offs here are logistical as much as biological. Autologous products are made from one person's cells for that person, which means a vein-to-vein turnaround and no shelf stock. Allogeneic approaches, made from donor cells, would allow inventory but have to solve rejection and graft-versus-host problems first. Solid tumours remain much harder than blood cancers because there is rarely a clean target that healthy tissue does not also carry.

Approval and access

In the United States these products are reviewed by the FDA's Center for Biologics Evaluation and Research; in the European Union the Committee for Advanced Therapies prepares the assessment for the EMA; in the United Kingdom the MHRA licenses them. Long-term follow-up obligations are standard, because a one-off treatment that alters a genome has to be watched for years rather than months.

Payment is the unresolved part. A single administration with an upfront price far above any conventional medicine does not fit budgets built around monthly prescriptions, and health systems have responded with instalment and outcome-linked arrangements. If you want the current terms for a specific product in a specific country, the health technology assessment bodies publish them — NICE in England, IQWiG and G-BA in Germany, CADTH's successor agency in Canada — and those documents are the right source rather than a manufacturer's press page.

None of this is guidance for anyone's own situation. Eligibility for these treatments is narrow and decided by clinical teams; a specialist is the only sensible route to an answer.