The gene therapy field is having its hardest year, and the fallout is reshaping how the next generation of treatments will be built. Three patient deaths in 2025 linked to adeno-associated virus (AAV) vectors, a US Food and Drug Administration clinical hold on Intellia Therapeutics, and a boxed liver-toxicity warning on Sarepta Therapeutics’ Elevidys have pushed developers toward an alternative they once dismissed as a sideshow. Lipid nanoparticles, the Sleeping Beauty transposon, and electroporation are now drawing investor money and clinical activity that would have looked unthinkable five years ago.
Sarepta’s roughly $3.2 million-per-dose AAV therapy delandistrogene moxeparvovec sits at the center of the storm. The FDA’s July 2025 revised indication and boxed warning for Elevidys followed two non-ambulatory Duchenne muscular dystrophy boys dying of acute liver failure. Weeks later, the agency’s order requesting suspension of Elevidys distribution followed a third death, this one in a related limb-girdle program. The label now restricts the drug to ambulatory boys aged four and older.
The Elevidys story is not alone. In late October 2025, Intellia Therapeutics paused the Phase III MAGNITUDE trial of nexiguran ziclumeran after a participant suffered Grade 4 liver enzyme spikes; the FDA placed two pivotal CRISPR programs on hold on October 29, and the patient died on November 5. A month later, the first child dosed in Capsida Biotherapeutics’ CAP-002 trial for STXBP1 encephalopathy died, two months after receiving the AAV-delivered medicine.
The Viral Vector Crisis That Forced a Rethink
Each fatality tells a slightly different story, but the through-line is consistent. AAV cargo can overwhelm the liver, and the immune system reacts unpredictably. Onset of acute liver injury after Elevidys typically begins within 8 weeks of dosing, the FDA noted, and Sarepta now runs a 200-patient observational study to track long-term hepatic outcomes.
The 2025 setbacks read like a roll call of the field’s biggest names:
- Sarepta Elevidys: two pediatric DMD deaths from acute liver failure; boxed warning, non-ambulatory distribution paused.
- Sarepta limb-girdle program: a third liver-related death; clinical hold across multiple Sarepta gene therapy candidates.
- Intellia nexiguran ziclumeran: November 2025 death after Grade 4 transaminase elevation; FDA hold on Phase III MAGNITUDE and HAELO.
- Capsida CAP-002: first child dosed dies in the STXBP1 encephalopathy study; trial paused two months after dosing.
Ilya Yasny, partner at LanceBio Ventures, told Pharmaceutical Technology that the unpredictability and manufacturing complexity of viral vectors is precisely why his firm is now looking harder at non-viral delivery.
Inside venture circles the thesis has shifted: AAV’s commercial ceiling looks mispriced for the safety it carries. Several gene therapy biotech stocks sold off after fresh safety disclosures, and pipeline reviews accelerated.
Why Lipid Nanoparticles Pulled Ahead of the Pack
LNPs were a bench technology for two decades before Moderna and Pfizer turned them into household objects through SPIKEVAX and Comirnaty. Pieter Cullis, Michael Hope, and Thomas Madden’s team at the University of British Columbia laid the chemistry, but it was the COVID-19 mRNA vaccines that proved the platform could scale to billions of doses with manageable safety.
The therapeutic crossover was inevitable. Alnylam’s Onpattro became the first marketed RNAi therapy in August 2018, proving an LNP-delivered nucleic acid drug could clear the FDA. Six years on, Verve Therapeutics’ Heart-2 Phase 1b readout for VERVE-102 showed a single GalNAc-LNP base-editing infusion cut LDL cholesterol by a mean of 53% at 0.6 mg/kg, with no clinically significant liver enzyme changes. The Alnylam siRNA-LNP delivery platform overview traces how the same chemistry is now being repointed at non-liver tissues.
| Feature | AAV viral vectors | Lipid nanoparticles |
|---|---|---|
| Per-dose manufacturing cost | Up to $100,000 | Single-digit thousands at scale |
| Payload size limit | About 4.7 kilobases | No hard ceiling for mRNA cargo |
| Re-dosing | Largely one-shot due to anti-capsid immunity | Repeat dosing feasible |
| Approved gene therapies | Multiple, including Elevidys, Zolgensma, Luxturna | Onpattro (siRNA); none traditional yet |
A Sleeping Beauty That Finally Woke
Tamas Laufer, an industry-funded PhD student at University College London, is one of a small cohort betting that the Sleeping Beauty transposon will be the cheapest path to durable, integrating gene therapy. The system pairs a plasmid (or minicircle) carrying both the gene of interest and the transposon protein with an mRNA encoding a transposase enzyme. The transposase then grabs the cargo and pastes it into the recipient genome.
Cost is what excites Laufer most. Plasmids and mRNA can be made by standard microbial fermentation and chemical synthesis, dodging the GMP viral suspension culture that pushes AAV cost-of-goods toward six figures per dose. The system has cleared a meaningful clinical hurdle: the TranspoCART19 GMP scale-up paper in Molecular Therapy Methods and Clinical Development documented matched in vitro and in vivo antitumor activity versus lentiviral CD19 CAR-T, supporting the ongoing NCT06378190 lymphoma trial.
Venkata Indurthi, chief scientific officer at Aldevron, calls Sleeping Beauty promising in animal models and a credible option for diseases needing durable expression, though he cautions integration efficiency falls as cargo grows.
Electroporation’s Believers and Doubters
Electroporation is the brute-force option. Brief electrical pulses crack open cell membranes, the genetic payload slips in, and the membrane reseals. Vertex Pharmaceuticals and CRISPR Therapeutics’ Casgevy manufacturing process document describes the entire CRISPR/Cas9 edit completing in roughly six days, with no virus involved.
That footprint helps explain Casgevy’s $2.2 million list price for a single autologous course. Every patient still needs a dedicated batch, but the process skips viral cleanrooms entirely, the part of biologics manufacturing that has historically gated capacity.
PulseSight Therapeutics is testing electroporation in a different setting. The Paris-based biotech delivers a DNA plasmid into the ciliary muscle of the eye for dry age-related macular degeneration. PulseSight’s April 2026 update on the PST-611 Phase I trial confirmed dosing across two cohorts in six patients, with a Phase II proof-of-concept study targeted next.
Yasny is not buying it across the board. In his Pharmaceutical Technology interview, he said:
“LNPs hold great potential, but are still at the beginning of the road.”
He prefers intravenous routes that distribute predictably to target organs and is openly skeptical of electroporation outside the autologous cell setting.
The Cost Math Behind Every Decision
Manufacturing economics is the part of the conversation drugmakers prefer to keep out of press releases. Roland Berger analysts pegged AAV cost-of-goods for an investigational DMD therapy at as much as $100,000 per dose, and the industry’s stretch goal of $10,000 per dose still implies a tenfold compression nobody has demonstrated yet.
The numbers framing the strategic shift:
- $850,000 to $3.5 million: the price band for currently approved AAV gene therapies.
- $100,000: estimated upper-bound AAV cost-of-goods per dose under conventional production.
- 500 liters: typical AAV batch size ceiling, beyond which yields decline sharply.
- Billions of doses: LNP output already proven at scale during the COVID-19 vaccine rollout.
Why Viral Vectors Are Not Going Anywhere
For all the LNP momentum, every expert interviewed for this piece warned that viral and non-viral approaches will live alongside each other. Indurthi noted that immunogenicity is sometimes driven by the nucleic acid payload itself, not the viral capsid, so swapping delivery does not automatically erase the immune problem.
Indurthi told Pharmaceutical Technology that LNPs are “more tunable and controllable” than viral vectors, but added that the right choice depends on the disease being targeted. Some indications, especially those needing single-shot durable expression in tissues LNPs struggle to reach, still favor AAV.
Judith Greciet, the CEO at PulseSight, agrees. Each modality, she said, will keep distinct roles across diseases and patient populations. The May 2025 NIH and Children’s Hospital of Philadelphia base-editing infusion that rescued a 7-month-old with carbamoyl phosphate synthetase I (CPS1) deficiency, delivered via LNP, sits at one extreme of what non-viral can already do; Casgevy and Verve’s clean Heart-2 readout sit alongside it as proof of concept.
Frequently Asked Questions
How is non-viral gene therapy different from AAV gene therapy?
Non-viral gene therapy delivers DNA, mRNA, or gene-editing tools using lipid nanoparticles, electroporation, or transposons instead of an engineered virus. The practical difference is that non-viral methods can usually be re-dosed, manufactured at much higher scale, and produced for a fraction of viral cost-of-goods, though most do not yet match AAV’s ability to reach a single target tissue with one infusion.
Is Elevidys still available for my child?
Yes, but only for ambulatory Duchenne muscular dystrophy boys aged four and older. The FDA narrowed the indication in July 2025 after two non-ambulatory pediatric deaths from acute liver failure. Sarepta paused distribution for non-ambulatory patients and the label now carries a boxed liver-injury warning. If you were waiting on a non-ambulatory dose, ask your treating neurologist about the 200-patient post-marketing observational study and Phase I non-viral DMD candidates.
Are non-viral gene therapies actually safer than AAV?
Not automatically, and not yet proven across diseases. LNPs avoid the high-dose hepatic immunogenicity that has triggered AAV deaths, but as Aldevron’s Venkata Indurthi notes, the nucleic acid payload itself can drive immune reactions regardless of delivery vehicle. Verve-102’s clean liver enzyme profile through 0.6 mg/kg dosing is the most encouraging non-viral safety signal in a systemic gene-editing therapy to date.
When will the next non-viral gene therapy reach the US market?
The most likely near-term candidates are Verve Therapeutics’ VERVE-102 for familial hypercholesterolemia and Casgevy label expansions, with Verve targeting Phase II initiation through 2026 and a possible Eli Lilly opt-in decision. PulseSight’s PST-611 for geographic atrophy is on track for Phase II this year. Approval timelines for any new traditional gene therapy delivered non-virally are most likely 2028 or later.
Why does AAV gene therapy cost so much?
AAV’s price reflects expensive manufacturing. Producing the virus in mammalian cells under GMP conditions can cost up to $100,000 per dose, batch sizes are capped near 500 liters, and approved AAV therapies sell for between $850,000 and $3.5 million. LNP and plasmid-based non-viral methods piggyback on chemistry and microbial fermentation, both already industrialized, which is why the long-run cost gap could be ten-to-one or wider.
Drug developers are spreading their bets across LNPs, electroporation, and transposons because nobody yet has the data to know which approach scales safely to which disease. The decade-long AAV monopoly on gene therapy delivery is breaking, just not all at once. By the close of 2026, the field will know whether the early non-viral safety signals can survive scaling and longer follow-up.
Leave a Comment