What Is Gene Therapy — and How Is It Curing Genetic Diseases?

Gene therapy treats disease by changing which genes a patient's cells carry or use. If a faulty gene causes the problem, the treatment adds a working copy, switches the faulty one off, or edits it. After a first human trial in 1990, a death in 1999, and years of modest results, the field now has FDA-approved products for spinal muscular atrophy, inherited blindness, sickle cell disease, and hemophilia. Some of those products carry price tags above $2 million, and a few have already failed commercially.

Educational content, not medical advice. Talk to your doctor before changing any medication, treatment, or routine. Information here reflects current research but every person's situation differs.

Key takeaways

• Gene therapy delivers genetic material, or a tool that edits it, into a patient's cells to treat disease. The FDA describes three broad moves: replace a disease-causing gene, turn off a malfunctioning gene, or add a new or altered gene.
• Most approved products use a viral vector, a virus stripped of its ability to cause disease and repurposed as a delivery truck. Others use lipid nanoparticles or cells edited in the lab and returned to the patient.
• Approved therapies exist for spinal muscular atrophy, one form of inherited retinal disease, sickle cell disease, beta-thalassemia, and hemophilia B, among others. List prices run from roughly $425,000 per eye (Luxturna) to $4.25 million (Lenmeldy).
• Gene therapy and DNA sequences in living cells with letter-level accuracy.">CRISPR overlap but are not synonyms. CRISPR is one way to do gene therapy, and most approved products still add a gene rather than edit one.
• The barriers now are immune reactions, liver and blood-cancer safety signals, durability, manufacturing, and payment. At least two hemophilia products were pulled from the market for lack of demand.

What gene therapy actually is

Many diseases trace back to a single gene that works badly or not at all. In sickle cell disease, a single-letter change in the gene for hemoglobin makes red blood cells deform. In spinal muscular atrophy (SMA), the motor neurons that control muscle lack enough of a protein called SMN because the SMN1 gene is missing or broken. In hemophilia, the gene for clotting Factor VIII or Factor IX is faulty, so blood does not clot properly. In one form of inherited blindness, both copies of a gene called RPE65 are defective, and the retina cannot recycle a molecule that light-sensing cells need.

A gene is a set of instructions, and a cell reads those instructions to build a protein. (For the underlying machinery, see How Do Genes Actually Work? and How Does DNA Work?.) If the instructions are missing, the obvious fix is to hand the cell a fresh copy. Think of a bakery whose only copy of a recipe was ruined by a spill. You can photocopy the recipe and slide it under the door, you can rewrite the smudged line in the original, or you can tell the bakers to ignore a recipe card that makes the wrong thing.

The hard part is not the idea. It is getting the instructions into the right cells, in a readable form, in enough of them, without the immune system attacking the delivery vehicle. Most of the history and most of the cost of gene therapy come from that logistics problem.

Disease	Gene involved	What goes wrong	Example therapy
Spinal muscular atrophy	SMN1	Motor neurons lack SMN protein	Zolgensma, Itvisma
RPE65-related retinal dystrophy	RPE65	Retinal cells cannot recycle a light-cycle molecule	Luxturna
Sickle cell disease	Hemoglobin beta gene	Hemoglobin distorts red blood cells	Casgevy, Lyfgenia
Hemophilia B	Factor IX gene	Blood lacks working Factor IX	Hemgenix
Hemophilia A	Factor VIII gene	Blood lacks working Factor VIII	Roctavian (withdrawn)

Four strategies: add, silence, edit, or instruct

Adding a working gene

Gene replacement, more accurately gene addition, delivers a functional copy of the gene to cells that carry the faulty version. The faulty gene stays where it is. The new copy sits alongside it and produces the missing protein. Most of the products approved so far work this way, including Zolgensma, Luxturna, and Hemgenix.

Silencing a gene

Some diseases come from a gene that makes a harmful protein rather than too little of a useful one. The treatment then delivers instructions that block that gene from producing its product. The FDA lists "turn off a disease-causing gene" as one of its three core approaches in its overview of gene therapy.

Editing a gene

Gene editing uses a molecular tool, most famously CRISPR-Cas9, to cut or rewrite DNA at a chosen spot. Casgevy, approved for sickle cell disease, edits a switch in a patient's blood stem cells so they make more fetal hemoglobin, which does not sickle. The FDA's approval notice for sickle cell gene therapies (see its December 8, 2023 announcement) describes Casgevy as using CRISPR/Cas9 genome editing for this purpose. For the editing tool itself, see What Is CRISPR and How Does Gene Editing Actually Work? and CRISPR Gene Editing in 2026.

Delivering instructions without touching DNA

Messenger RNA (mRNA) is a temporary copy of a gene's instructions. If you deliver mRNA for a therapeutic protein, cells make it for a while and then the message degrades, leaving the genome untouched. The mRNA vaccines used against COVID-19 relied on the same principle (see mRNA Vaccines: COVID-19 Breakthrough Explained). Whether such treatments count as gene therapy depends on the definition. They change what cells do without changing what cells are.

How the genetic cargo gets in

A gene by itself will not cross a cell membrane in useful amounts, so therapies rely on vectors. The FDA's overview lists plasmid DNA, viral vectors, bacterial vectors, gene-editing technology, and patient-derived cellular products as the main categories. Three of those dominate current practice.

Viral vectors

Viruses are experts at entering cells and delivering genetic material. Vector designers remove the genes that let the virus replicate or cause illness and drop in the therapeutic gene instead. (For how natural viruses do the same job, see How Viruses Work: From Infection to Replication.) The main types:

• Adeno-associated virus (AAV): a small virus that is generally not associated with human disease. Zolgensma, Luxturna, and Hemgenix are all AAV-based. AAV carries only a small payload, so large genes do not fit.
• Lentivirus: a relative of HIV with the dangerous genes removed. It can integrate its cargo into a cell's chromosomes, which makes it suited to dividing cells such as blood stem cells. Lyfgenia uses a lentiviral vector, according to the FDA.
• Adenovirus: a cold virus. It was used in early trials and still appears in some products, but its strong immune stimulation is a known problem, as the story below shows.

Lipid nanoparticles

Lipid nanoparticles are tiny fat bubbles that wrap nucleic acids and ferry them into cells, most reliably in the liver. They became widely known through mRNA vaccines. They are also the delivery system in the personalized editing therapy described later.

Ex vivo versus in vivo

Ex vivo ("outside the body") means doctors collect a patient's cells, modify them in a lab, and infuse them back. Casgevy and Lyfgenia work this way, using blood stem cells. According to the FDA, patients first undergo high-dose chemotherapy conditioning (called myeloablation) to clear space in the bone marrow, then receive the modified cells as a transplant. That step is a serious medical procedure in its own right.

In vivo ("in the living body") means the vector is injected or infused and does its work inside the patient. Zolgensma, Luxturna, and Hemgenix are in vivo. This avoids conditioning chemotherapy but gives doctors less control over which cells take up the cargo.

Approach	Typical vector	Main advantage	Main limit
In vivo, AAV	AAV capsid	Single infusion or injection, no cell harvest	Small cargo; antibodies may block a second dose
Ex vivo, lentivirus	Lentiviral vector	Cargo can be integrated into dividing cells	Requires chemotherapy conditioning; cancer monitoring
Ex vivo, CRISPR	Electroporated editing machinery	Permanent edit in stem cells	Same conditioning; complex manufacturing
In vivo, lipid nanoparticle	Lipid bubble with mRNA	Redosing possible in principle; fast to redesign	Reaches mainly the liver today

A short, hard history

The first approved human gene therapy trial began on September 14, 1990, at the National Institutes of Health Clinical Center. A four-year-old girl, Ashanthi DeSilva, who had a form of severe combined immunodeficiency caused by ADA deficiency, received her own T cells after they had been modified with a retroviral vector carrying a working ADA gene. W. French Anderson and R. Michael Blaese led the trial, according to a retrospective in GEN. The trial was small and its results were hard to interpret because patients also kept receiving enzyme replacement. It did show that the procedure was feasible.

The Jesse Gelsinger case

On September 13, 1999, an 18-year-old named Jesse Gelsinger, who had a partial form of ornithine transcarbamoylase (OTC) deficiency that he controlled with diet and drugs, received an adenoviral vector infused into his hepatic artery at the University of Pennsylvania. He died four days later from a severe immune reaction, according to a CMAJ account of the case.

The FDA suspended the trial, citing inadequate staff training, missing operating procedures, and inadequate informed consent. It later halted the university's other gene therapy trials and reviewed dozens of trials nationally: 28 of 69 under way, with 13 requiring remedial action. Afterward, regulators introduced a gene therapy trial monitoring plan, created safety symposia to share adverse event data, and barred investigators directly involved in trials from holding equity in sponsoring companies. The family sued in September 2000 and settled about six weeks later, in early November, according to The Daily Pennsylvanian.

The case is why modern vectors are chosen for low immune stimulation, why doses are escalated cautiously, and why long-term follow-up is expected. The FDA's overview points to its guidance on long-term follow-up after administration of human gene therapy products.

What is approved and what it does

Dozens of cell and gene therapy products now appear on the FDA's list of approved cellular and gene therapy products, which was current as of September 17, 2026 when we checked it. It includes Zolgensma, Luxturna, Casgevy, Lyfgenia, Hemgenix, Roctavian, Beqvez, Elevidys, Zynteglo, Skysona, Lenmeldy, Kebilidi, Vyjuvek, Adstiladrin, Itvisma, Encelto, and Zevaskyn, among others. Being on the list does not mean a product is on the market, as two entries below show.

Therapy	Condition	Approach	Reported list price	Status
Zolgensma	SMA, children under 2	In vivo AAV	About $2.1 million	Marketed
Itvisma	SMA, ages 2 and older	In vivo AAV, intrathecal	Not confirmed in sources used	Approved November 2025
Luxturna	RPE65 retinal dystrophy	In vivo AAV, subretinal	$425,000 per eye	Marketed
Casgevy	Sickle cell disease, beta-thalassemia	Ex vivo CRISPR	$2.2 million	Marketed
Lyfgenia	Sickle cell disease	Ex vivo lentivirus	$3.1 million	Marketed
Hemgenix	Hemophilia B	In vivo AAV	$3.5 million	Marketed
Beqvez	Hemophilia B	In vivo AAV	$3.5 million	Discontinued by Pfizer, 2025
Roctavian	Hemophilia A	In vivo AAV	$2.9 million	BioMarin moved to withdraw it
Lenmeldy	Metachromatic leukodystrophy	Ex vivo	$4.25 million	Marketed (2024 price)

Prices are list prices reported by Pharmaceutical Technology and PharmaVoice (August 30, 2024); net prices after rebates and negotiated agreements can differ.

Zolgensma and Itvisma for SMA

Zolgensma delivers a working SMN1 gene using an AAV vector, as a single intravenous infusion dosed by body weight. It is approved for children up to age 2 in the United States. The label carries a boxed warning for serious liver injury and acute liver failure, with fatal cases reported, and clinicians give corticosteroids before and after the infusion. It is given once in part because the body develops antibodies against the viral carrier.

In November 2025, the FDA approved a second product from the same maker, Itvisma (onasemnogene abeparvovec-brve), for patients 2 years and older with confirmed SMN1 mutations. It is a one-time intrathecal injection, which means it goes into the fluid around the spinal cord, and the dose does not depend on age or weight. Approval rested mainly on the phase 3 STEER study, with the open-label STRENGTH study as support, and the FDA noted possible higher risk of liver and heart toxicity in adults with preexisting conditions.

Luxturna for inherited blindness

Luxturna (voretigene neparvovec) was an early landmark: a one-time gene therapy for a form of inherited blindness caused by mutations in both copies of RPE65. The FDA's summary of regulatory action is dated December 18, 2017. It is injected under the retina, where an AAV vector delivers a working RPE65 gene to retinal cells. The pivotal phase 3 trial, as summarized by Health Canada, compared 21 treated patients with 10 controls and measured performance on a multi-luminance mobility test, a course that tests how well a person navigates at different light levels. The treated group improved significantly more at one year. Because each eye is treated separately, the $425,000 price is per eye.

Casgevy and Lyfgenia for sickle cell disease

The FDA approved both on December 8, 2023, for patients 12 and older. Casgevy uses CRISPR/Cas9 to edit a patient's blood stem cells so they raise fetal hemoglobin, which keeps red cells from sickling. In the trial reported by the FDA, 29 of 31 evaluable patients (93.5%) were free of severe vaso-occlusive crises for at least 12 consecutive months. The UK regulator, the MHRA, had authorized Casgevy in November 2023 for sickle cell disease and transfusion-dependent beta-thalassemia. In the thalassemia trial data summarized in an MHRA regulatory assessment, 39 of 42 patients in the primary efficacy analysis (those followed for at least 16 months) went at least 12 consecutive months without red blood cell transfusions. A later data cut presented by Vertex in June 2024 put the figure at 49 of 52 evaluable patients (94.2%), according to the company's EHA 2024 announcement.

Lyfgenia takes a different route. A lentiviral vector adds a gene for a modified hemoglobin, HbAT87Q, that resists sickling. In its trial, 28 of 32 patients (88%) had complete resolution of vaso-occlusive events between 6 and 18 months after infusion. The label carries a boxed warning: blood cancer has occurred in patients treated with Lyfgenia, and patients need lifelong monitoring.

Hemgenix for hemophilia B

Hemgenix (etranacogene dezaparvovec) was approved on November 22, 2022 for adults with hemophilia B who meet certain criteria. In the phase 3 HOPE-B trial of 54 adults, the adjusted annualized bleeding rate fell 64% at 18 months compared with a six-month run-in period, according to the manufacturer's announcement. Mean Factor IX activity was about 39% of normal at 6 months and about 37% at 18 months. A small amount of working factor, far short of 100%, is enough to change the clinical picture because the baseline in severe hemophilia is near zero.

What "cure" does and does not mean here

Headlines say "cure." Trial reports say things like "free of severe events for 12 months" or "did not need transfusions for 12 months." Those are strong results, but they are not the same as a proven lifelong cure, because most of these therapies have only a few years of follow-up. The distinction matters in practice.

Three facts help calibrate the claims:

1. Durability varies by design. An AAV vector usually delivers its cargo as a loose piece of DNA that sits in the nucleus separate from the chromosomes. In cells that rarely divide, such as neurons or liver cells, that can last for years. In cells that divide often, the copy is diluted and lost. An edit made in blood stem cells is passed on to their daughter cells, so it should persist as long as those stem cells do.
2. One dose may mean one dose. Because patients can develop antibodies against the viral carrier, a second AAV dose may not work, which is part of why makers describe these products as single-administration. The label for Zolgensma notes that repeat administration has not been evaluated.
3. Cure is relative to the disease. Zolgensma does not restore motor neurons that have already died. The FDA label notes that safety and effectiveness have not been evaluated in advanced SMA, including complete limb paralysis. Early treatment matters.

Where the commercial picture is troubled

A technical success is not the same as a viable product. Hemophilia shows the gap.

Pfizer announced in February 2025 that it would end development and sale of Beqvez, a hemophilia B gene therapy approved in 2024, citing the limited interest patients and doctors had shown in hemophilia gene therapies. Fierce Pharma reported that no patients appeared to have received it commercially since its approval, and that the product carried a $3.5 million list price.

BioMarin followed a similar path with Roctavian. In October 2025 it said it would pursue divestiture. On February 23, 2026, after failing to find a qualified buyer, it announced it would voluntarily withdraw the product from the market, citing commercial reasons rather than safety or efficacy, with charges of roughly $240 million reported for the fourth quarter of 2025. In 2024, reporting had already noted that BioMarin treated only five patients in one quarter and restricted access to three countries.

Hemophilia patients have an effective alternative in regular preventive treatment. A one-time therapy with an unknown long-term horizon, a multimillion-dollar price, and the possibility of liver side effects is a harder decision than a drug taken on a schedule. That is a lesson about medicine as much as economics: a cure has to beat not just the disease but the existing standard of care.

Why it costs so much

A $2 million price for a single treatment looks absurd until you list what goes into it, and even then it remains contested.

• Manufacturing: Producing clinical-grade viral vectors is slow and exacting, and each batch needs extensive quality control. Ex vivo products are made one patient at a time in specialized facilities.
• Small populations: Many of these diseases affect few people, so development costs are spread over a small number of patients.
• Value-based pricing: Companies argue that a one-time cure replaces decades of chronic care and price against those savings. PharmaVoice notes that prices are framed around savings over a patient's journey, though high price tags have not guaranteed strong uptake.
• Research history: Decades of work, including failed programs, are recovered through the products that succeed.

Governments and insurers are experimenting with payment models. In 2025, the Centers for Medicare & Medicaid Services announced that 33 states plus the District of Columbia and Puerto Rico would take part in its Cell and Gene Therapy Access Model for sickle cell disease. The jurisdictions represent roughly 84% of Medicaid beneficiaries with the disease. The model ties payment for Casgevy and Lyfgenia to health outcomes, with states receiving rebates if a therapy underperforms. Whether such deals scale to other diseases remains to be seen. The wider drug-pricing context is covered in Why Insulin Costs So Much in America.

The hard problems that remain

Immune responses

The immune system is built to attack foreign viruses, and a vector is a foreign virus. Pre-existing antibodies against a vector can block treatment, and new ones can block redosing. Strong immune reactions can also injure the patient, as the Gelsinger case showed. (The basics of how the body spots threats are in How the Immune System Recognizes Threats and Antibody Structure.)

Organ toxicity and cancer signals

High vector doses delivered to the liver can injure it, which is why Zolgensma carries its boxed warning. Integrating vectors raise a different concern: Lyfgenia's boxed warning for blood cancer is a direct example, and the FDA advises long-term follow-up for gene therapy recipients.

The conditioning problem

Ex vivo therapies for blood disorders require chemotherapy to make room for the modified cells. That is manageable for a trial participant at a specialized center and a heavy burden for a patient in a low-income country. Safer conditioning is an active research goal.

Access

A therapy that exists but cannot be delivered does not cure anyone. The first sickle cell approvals came in the United Kingdom and the United States, where treatment requires specialist transplant centers and funding. Many people with the disease live in places that have neither. That is why programs such as the CMS access model, described above, matter, and why cheaper manufacturing and simpler delivery are active goals.

Consider what a treatment center must have on hand: a facility to collect stem cells, a lab to process them, a transplant unit to give high-dose chemotherapy, and months of follow-up care. Each is a bottleneck. Even in wealthy health systems, the first patients treated with Casgevy waited through cell collection and manufacturing before infusion; PharmaVoice reported that as of mid-July 2024, only 20 patients had had their cells collected.

Germline editing

Everything above targets somatic cells, meaning body cells whose changes end with the patient. Editing eggs, sperm, or embryos would create heritable changes, and that is prohibited or heavily restricted in many countries. Approved therapies do not alter the germline, so a patient's future children would not inherit the treatment.

What is coming next

Personalized editing for ultra-rare disease

In 2025, a team at Children's Hospital of Philadelphia and Penn Medicine treated an infant known as KJ, who had severe carbamoyl phosphate synthetase 1 (CPS1) deficiency, a urea cycle disorder in which ammonia builds up to toxic levels. The treatment was a base editor, a CRISPR variant that changes a single DNA letter without cutting both strands, delivered by lipid nanoparticles to the liver. Reports say the team designed and made it in about six months. KJ received three infusions between February and April 2025. His physician reported that he tolerated treatment well, could eat more protein, and needed fewer nitrogen-scavenging drugs, while also saying that the treatment is not a cure, per CHOP's one-year update. One patient proves feasibility, not safety or efficacy in general, but it points to a model where a platform is reused and only the guide sequence is customized.

Better delivery and AI-assisted design

Many next-generation programs focus on delivery: capsids engineered to reach specific tissues, and nanoparticles that reach beyond the liver. Machine learning is increasingly used to design editors and guide sequences (see AI Meets CRISPR).

Beyond rare disease

Current approvals concentrate on rare single-gene disorders because the genetics are simplest and the unmet need is greatest. Extending the approach to common conditions such as high cholesterol is a stated goal of several programs, though approvals for those are not established. Related fields, including stem-cell medicine, overlap with this one; see How Stem Cells Work and What They Can Actually Treat.

Common misunderstandings

"Gene therapy rewrites all of your DNA." It does not. Treatments target particular cells, and most leave the DNA in other tissues unchanged.

"It is the same as CRISPR." CRISPR is one tool. Zolgensma, Luxturna, and Hemgenix do not use it.

"Approved means available." Beqvez and Roctavian show that regulatory approval and a working market are separate things.

"One dose, cured forever." Follow-up for most products is still measured in a few years, and durability differs by design.

"It will be cheap soon." Prices fall slowly, and the cost drivers, especially manufacturing and small patient numbers, are structural.

Frequently asked questions

Q: Is gene therapy the same as CRISPR? A: No. Gene therapy is the broad idea of treating disease by changing genes or gene activity. CRISPR is one tool for editing DNA. Casgevy uses it, but Zolgensma, Luxturna, and Hemgenix deliver a working gene copy through viral vectors without editing the patient's DNA.

Q: Will my children inherit the changes? A: Not from approved therapies. They modify somatic cells, which are body cells, rather than eggs or sperm. Germline editing, which could be inherited, is restricted or prohibited in many countries.

Q: Which diseases can gene therapy treat? A: Single-gene disorders are the most suitable, and approved examples include SMA, RPE65-related retinal dystrophy, sickle cell disease, beta-thalassemia, and hemophilia B. Diseases driven by many genes plus environment are much harder to target.

Q: Is gene therapy permanent? A: It depends on design. Edited blood stem cells pass the change to their descendants, so benefits may last. AAV cargo mostly stays outside chromosomes and can fade in tissues where cells divide. Long-term follow-up is still accumulating for most products.

Q: Why did some approved gene therapies get pulled from the market? A: Weak demand. Pfizer discontinued Beqvez in 2025 after limited uptake, and BioMarin moved to withdraw Roctavian after failing to find a buyer. Patients and doctors often preferred existing preventive treatments to a costly one-time therapy.

Q: What are the main risks? A: They include immune reactions to the vector, liver injury at high doses, and, for some integrating products, blood cancer. Zolgensma and Lyfgenia carry boxed warnings for liver injury and blood cancer respectively, and conditioning chemotherapy adds its own risks for ex vivo products.

The takeaway

Gene therapy has moved from a failed idea with a famous death attached to a set of approved treatments with measurable results: children with SMA treated with a single infusion, patients with sickle cell disease free of severe crises for a year or more in trials, and hemophilia patients with a fraction of normal clotting factor. The remaining problems are less about whether the biology works and more about whether it can be delivered safely, paid for, and made worth choosing over existing care. The Beqvez and Roctavian withdrawals show that technical success does not guarantee a viable product, while the KJ case suggests that the next phase may be about speed and customization rather than one drug per disease.

Sources

• FDA: What is gene therapy?
• FDA: Approved cellular and gene therapy products
• FDA: Approval of first gene therapies for sickle cell disease, December 8, 2023
• CMAJ: Account of the Gelsinger case
• PharmaVoice: World's priciest drug (August 30, 2024)
• CHOP: One-year anniversary of the first personalized CRISPR therapy

Related Reading

• What Is CRISPR and How Does Gene Editing Actually Work?
• Why Aging Is a Disease — And Could We Actually Cure It?
• The Science of Sleep — Why You Need It and What Happens Without It
• Why Do We Yawn — and Why Is It Contagious?
• How Do Genes Actually Work?