Gene Therapy
Gene Therapy
Delivering a working gene to replace a broken one — vectors, doses, durable cures
Gene therapy delivers a functional gene to replace a defective one. AAV, lentivirus, and LNP carry the payload to target tissue. Zolgensma cures SMA in infants with one IV infusion — $2.1M per dose, most expensive drug ever launched.
- AAV cargo limit~4.7 kb single-stranded DNA
- Lentivirus cargo~8 kb; integrates into host genome
- First approval (US)Luxturna · Dec 2017 · RPE65 retinal dystrophy
- Zolgensma price$2.1M single dose · most expensive drug ever
- Hemgenix price$3.5M single dose (hemophilia B, 2022)
- Approved products15+ (2026) across rare disease, hematology, ophthalmology
Interactive visualization
Press play, or step through manually. The visualization is yours to drive — try it before reading on.
Watch the 60-second explainer
A condensed visual walkthrough — narrated, captioned, under a minute.
Worked example — a Zolgensma infusion at 6 months old
A baby born with biallelic SMN1 deletion is identified at newborn screening (now standard in all 50 US states). She has zero copies of functional SMN1 and two copies of SMN2 — predicting SMA type 1, historically lethal by age 2 without ventilator support. The family elects Zolgensma at 6 months of age, weight 7.2 kg.
Dose. 1.1 × 1014 vector genomes per kg → 7.9 × 1014 vg total. That is roughly 1015 AAV9 capsids in a 50 mL IV infusion over one hour. Pre-dose serology confirms no neutralizing antibodies against AAV9 (anti-AAV9 NAb titer ≤1:50 required).
Prophylactic steroid. Prednisolone 1 mg/kg/day starts the day before dosing and tapers over 1-2 months — suppresses the T-cell response to AAV9 capsid that can otherwise transiently elevate transaminases.
Distribution. AAV9 capsids circulate and bind cell surface receptors broadly. AAV9's distinguishing property is crossing the blood-brain barrier in young patients — capsids reach motor neurons in the spinal cord and brainstem at sufficient copy number to transduce most. Liver, heart, skeletal muscle also receive vector but the motor neuron target is what matters.
Transduction. AAV9 capsids enter motor neurons via endocytosis, traffic to the nucleus, uncoat, and convert single-stranded DNA into double-stranded episomes. The SMN1 cDNA driven by a chicken-beta-actin/CMV hybrid promoter begins to produce SMN protein within days.
Outcome. By week 4-8, motor function stabilizes and then improves. By month 18-24, treated infants who would otherwise be ventilator-dependent are sitting, some standing, some walking. SPR1NT and STR1VE outcomes show 100% survival without permanent ventilation in dosed infants and motor milestone achievement well beyond untreated SMA1 natural history. The transgene persists episomally in non-dividing motor neurons — likely for life.
From 1990 to today
- 1990. First gene therapy trial — Ashanti DeSilva, ADA-SCID, retroviral transduction of T cells (French Anderson, NIH).
- 1999. Jesse Gelsinger dies in OTC deficiency adenoviral gene therapy trial at Penn — massive immune response. Trial suspensions, FDA reforms, decade-long setback.
- 2000-2003. First X-SCID gene therapy cures most patients in French/UK trials — but some develop leukemia from retroviral insertional mutagenesis activating LMO2. Field shifts to self-inactivating lentivirus.
- 2012. Glybera (alipogene tiparvovec, AAV1-LPL) — first gene therapy approved in Europe (lipoprotein lipase deficiency). Withdrawn 2017 due to low uptake.
- December 2017. Luxturna (voretigene neparvovec) — first FDA-approved gene therapy in the US.
- May 2019. Zolgensma approved for SMA — first IV systemic AAV gene therapy.
- 2022. Hemgenix for hemophilia B; Skysona for ALD; Zynteglo for beta-thalassemia.
- 2023. Casgevy and Lyfgenia for sickle cell; Roctavian for hemophilia A; Elevidys for Duchenne.
- 2024. Lenmeldy ($4.25M) for metachromatic leukodystrophy — currently the most expensive drug ever approved.
Vector comparison
| Vector | Cargo capacity | Integrates? | Tropism | Duration | Best for |
|---|---|---|---|---|---|
| AAV2 | ~4.7 kb | Mostly episomal | Retina, CNS | Years to life (non-dividing) | Luxturna (subretinal) |
| AAV5 | ~4.7 kb | Episomal | Liver, lung | Years | Hemgenix, Roctavian |
| AAV8 | ~4.7 kb | Episomal | Liver (strong) | Years | Hepatic targets |
| AAV9 | ~4.7 kb | Episomal | CNS (BBB crossing in infants), heart, muscle | Years to life | Zolgensma |
| Lentivirus (SIN) | ~8 kb | Integrates (euchromatin) | Engineered ex vivo | Lifetime in modified cells | HSC therapies (Casgevy, Skysona) |
| LNP-mRNA | ~4-10 kb mRNA | No (cytoplasmic, transient) | Liver (IV); injection site (IM) | Days | mRNA vaccines, transient expression |
Where gene therapy is changing medicine
- Spinal muscular atrophy. Zolgensma transforms SMA1 from lethal to survivable.
- Inherited retinal dystrophies. Luxturna restores light response in RPE65 LCA.
- Hemophilia A & B. Hemgenix, Roctavian, Beqvez free patients from chronic clotting-factor infusions.
- Sickle cell & beta-thalassemia. Casgevy (CRISPR), Lyfgenia (lentivirus-BCL11A shRNA), Zynteglo (lentivirus-HBB-T87Q).
- Duchenne muscular dystrophy. Elevidys delivers a micro-dystrophin in AAVrh74.
- Leukodystrophies. Skysona (ALD), Lenmeldy (MLD) — ex vivo HSC gene therapy.
- Cancer. CAR-T cell therapies are gene therapies — ex vivo lentiviral CAR delivery.
- Cardiovascular. AAV-VEGF, AAV-AAT, base-editing PCSK9 trials underway.
Common misconceptions
- Gene therapy changes everyone's DNA. AAV is mostly episomal — most gene therapies do not modify the patient's genomic DNA at all.
- One dose lasts forever automatically. Durability depends on whether target cells divide — non-dividing neurons retain episomes for life; dividing cells dilute out episomes.
- You can always redose. Most patients develop neutralizing antibodies to AAV capsid; redosing is foreclosed without engineered capsids or antibody depletion.
- Gene therapy is the same as gene editing. Gene therapy adds or supplements; gene editing (CRISPR) changes the existing sequence.
- Approved means safe for everyone. Recipients are screened — body weight, AAV NAb titer, organ function. High-dose AAV has caused fatalities in some indications.
- $2-4M means it's too expensive ever. Cost-effectiveness vs. lifetime conventional therapy is often favorable; outcomes-based contracts and value pricing models are evolving.
Frequently asked questions
What is a gene therapy vector?
A vector is the delivery vehicle carrying the therapeutic gene into target cells. Viral vectors are stripped of replication genes and engineered to carry a transgene. Adeno-associated virus (AAV): single-stranded DNA, ~4.7 kb capacity, ~25 nm capsid, non-integrating (mostly episomal), strong tissue tropism by serotype, low immunogenicity. Lentivirus (HIV-based): ~8 kb capacity, integrates into the genome (preferentially active euchromatin), used for ex vivo modification of dividing cells. Adenovirus: large capacity (~36 kb), strong but transient expression, highly immunogenic — limits in vivo use. Non-viral: lipid nanoparticles (LNPs) deliver mRNA or plasmid DNA; naked DNA; electroporation. Choice depends on target tissue, cargo size, duration needed, and immune considerations.
How does Zolgensma cure spinal muscular atrophy?
Spinal muscular atrophy (SMA) is an autosomal recessive disease caused by biallelic deletion or mutation of SMN1, the gene encoding survival motor neuron protein. Without SMN, lower motor neurons in the spinal cord and brainstem die. SMA type 1 (the most severe) historically killed infants before age 2. Zolgensma (onasemnogene abeparvovec) is an AAV9 capsid carrying a functional SMN1 cDNA under a constitutive promoter. AAV9 crosses the blood-brain barrier well — given IV in infants ≤2 years (≤21 kg), it transduces motor neurons throughout the spinal cord. The transgene persists episomally in non-dividing neurons for years to a lifetime. A single infusion gives durable benefit. Price: $2.1 million per dose; outcomes-based contracts cover non-response. STR1VE and SPR1NT trials show survival without permanent ventilation and motor milestone acquisition in nearly all treated infants.
What in vivo gene therapies are FDA-approved?
Luxturna (voretigene neparvovec, AAV2-RPE65) — 2017, subretinal injection for RPE65 Leber congenital amaurosis. Zolgensma (onasemnogene abeparvovec, AAV9-SMN1) — 2019, IV for SMA in infants. Hemgenix (etranacogene dezaparvovec, AAV5-Padua FIX) — 2022, single IV dose for hemophilia B; raises factor IX from <1% to ~30%. Roctavian (valoctocogene roxaparvovec, AAV5-FVIII) — 2023, hemophilia A. Elevidys (delandistrogene moxeparvovec, AAVrh74-microdystrophin) — 2023, Duchenne muscular dystrophy. Beqvez (fidanacogene elaparvovec, AAV-FIX-Padua) — 2024, hemophilia B. Casgevy and Lyfgenia (sickle cell) and others use ex vivo HSC modification. Lenmeldy (atidarsagene autotemcel) — 2024, ex vivo HSC therapy for metachromatic leukodystrophy.
Why is gene therapy so expensive?
Zolgensma at $2.1 million is the most expensive drug ever launched. Hemgenix: $3.5M. Lyfgenia: $3.1M. Elevidys: $3.2M. Casgevy: $2.2M. Lenmeldy: $4.25M. Reasons. Tiny patient population — SMA ~1 in 11,000 births. R&D and clinical trials must amortize over hundreds, not millions, of patients. Bespoke or low-throughput GMP manufacturing — viral vector production for one patient can cost hundreds of thousands. Single dose, lifetime benefit — value-based pricing argues the alternative is a lifetime of chronic therapy (nusinersen for SMA is ~$750k year 1, $375k/year thereafter). Outcomes-based contracts and installments increasingly defray the up-front shock. Whether high prices are sustainable as more therapies launch is an open policy question.
What about preexisting AAV immunity?
Wild-type AAV serotypes are common in humans — 30-70% of adults have neutralizing antibodies against AAV2, fewer against AAV5/8/9. Preexisting NAbs block re-administration and exclude many patients from AAV therapies. Mitigation strategies in development. Capsid engineering: directed evolution of capsids that evade NAbs (Voyager TRACER, AAVnerGene). Plasmapheresis or IdeS (imlifidase, a bacterial IgG protease) to deplete antibodies before dosing — Sarepta and others are testing. Empty capsid decoys. Immunosuppression at dosing. Alternative vectors when AAV is foreclosed (lentiviral, integrase-deficient lentivirus, non-viral). Redosing remains very difficult — most AAV therapies are one-shot.
What's the difference between in vivo and ex vivo gene therapy?
In vivo: the vector is injected directly into the patient (IV, intramuscular, subretinal, intrathecal). It must navigate to the right cells, evade pre-existing antibodies, and avoid off-target expression. Examples: Zolgensma (IV), Luxturna (subretinal), Hemgenix (IV). Ex vivo: cells are removed from the patient, modified in the lab, expanded, and reinfused. The vector never sees the patient's circulation. Examples: Casgevy and Lyfgenia (HSCs for sickle cell), Zynteglo (HSCs for beta-thalassemia), Skysona (HSCs for adrenoleukodystrophy), Lenmeldy (MLD), CAR-T cell products. Ex vivo allows precise quality control of modified cells and avoids systemic vector exposure but requires patient conditioning (chemotherapy) and complex manufacturing logistics.
What are the major safety concerns?
Insertional mutagenesis: integrating vectors (retro, lenti) can disrupt or activate genes near the integration site — caused leukemias in early X-SCID trials using gamma-retroviruses (LMO2 activation). Modern self-inactivating (SIN) lentivirus dramatically reduces but does not eliminate the risk. Capsid immune response: T-cell response to AAV capsid can transiently elevate transaminases — managed with steroid prophylaxis (Roctavian, Hemgenix). Complement activation in high-dose AAV: thrombotic microangiopathy seen at high systemic doses (Zolgensma >2×10¹⁴ vg/kg). Off-target expression: ubiquitous promoters express in non-target tissues; tissue-specific promoters and miRNA detargeting mitigate. Germline transmission: very low risk in current vectors but actively studied. Manufacturing variability.