Sarepta's $3.2M Gene Therapy Killed Kids. Now It's Trying a Fix.
Sarepta's $3 million gene therapy for Duchenne muscular dystrophy has killed children from liver failure. Now the company is betting on an enhanced immunosuppression regimen to fix its flagship product, with implications for the entire gene therapy field.
Imagine paying $3.2 million for a one-time treatment to save your child's life, only to learn it might destroy their liver.
That's the nightmare Sarepta Therapeutics is racing to solve. The company's flagship gene therapy, Elevidys, was supposed to be a breakthrough for boys with Duchenne muscular dystrophy (DMD), a brutal genetic disease that progressively strips away muscle function. It became the first FDA-approved gene therapy for DMD in June 2023. But post-marketing reports revealed something terrifying: children were dying from acute liver failure after receiving the treatment.
Now Sarepta is testing a new safety regimen built around an additional immunosuppressive drug called sirolimus. The goal is to tame the liver damage that's already killed at least two young DMD patients and forced the FDA to slap a Boxed Warning on the therapy's label. It's the highest-stakes science experiment in gene therapy right now, and the outcome will ripple far beyond one company.
The Deaths That Changed Everything
The trouble started surfacing in early 2025. The FDA flagged reports of fatal liver toxicity in its adverse event database. By summer, two non-ambulatory (wheelchair-bound) boys with DMD had died from acute liver failure after receiving Elevidys. A third death occurred in an adult patient with a related condition called Limb Girdle Muscular Dystrophy, treated with a similar Sarepta therapy using the same viral vector.
Think of gene therapy like a delivery truck. The "cargo" is a working copy of a gene the patient is missing. The "truck" is a modified virus (in this case, an AAV, or adeno-associated virus) that carries the gene into cells. The problem? The body sometimes treats the truck like an invader and launches an immune attack, particularly in the liver, where these viral particles tend to accumulate.
In Elevidys's case, the immune response was devastating. Patients showed dangerously elevated liver enzymes within about eight weeks of infusion. One patient even developed life-threatening blood clots in the veins around the intestines, leading to bowel damage and portal hypertension.
Sarepta suspended U.S. distribution for non-ambulatory patients in June 2025. By November, the FDA had restricted the therapy to ambulatory patients only (those who can still walk) and added the dreaded Boxed Warning, the most serious safety label the agency can require.
Get tomorrow's biotech intelligence before your competitors.
Join thousands of biotech professionals who start their day with our free, daily briefing.
The Sirolimus Bet
The original Elevidys protocol already included immunosuppression. Patients received systemic corticosteroids (typically prednisone) starting one day before infusion, continuing for at least 30 days, then tapering gradually. That clearly wasn't enough for the sickest patients.
Sarepta's new approach layers sirolimus on top of the existing corticosteroid regimen. Sirolimus is an immunosuppressant that works through a different mechanism than steroids; it's been used for decades in organ transplant patients to prevent rejection. The idea is that hitting the immune system from two angles might prevent it from going haywire against the viral delivery vehicle.
The FDA approved this enhanced protocol in November 2025 for a study called ENDEAVOR Cohort 8, an open-label Phase 1b trial enrolling roughly 25 non-ambulatory patients in the U.S. Crucially, sirolimus dosing begins 14 days before the Elevidys infusion and continues for 12 weeks afterward. That pre-treatment window is key: by the time the viral vector arrives, the immune system is already dialed down.
Primary endpoint data (measuring both the incidence of liver injury and whether the therapy still produces its target protein, dystrophin) is expected in the second half of 2026. If the results look good, they could pave the way for resuming commercial use in non-ambulatory patients.
Why Non-Ambulatory Patients Matter So Much
You might wonder: if Elevidys is still approved for ambulatory patients, why does Sarepta care so much about the non-ambulatory population?
Because DMD is progressive. Every ambulatory patient will eventually become non-ambulatory. Losing that indication doesn't just shrink the addressable market today; it puts a ceiling on the therapy's long-term relevance. Parents whose sons can still walk face an agonizing question: do we treat now, knowing the therapy might not be available later when the disease worsens?
There's also the revenue math. Elevidys costs $3.2 million per patient. The broader DMD treatment market was valued at around $3.9 billion in 2025. Sarepta needs every eligible patient it can get, especially as competitors like Regenxbio (planning a BLA submission for its rival therapy RGX-202 by 2026) close in.
A Problem Bigger Than Sarepta
Liver toxicity isn't just Sarepta's headache. It's a class-wide issue for AAV gene therapies, and the field is littered with cautionary tales.
Consider Novartis's Zolgensma, an AAV9 gene therapy for spinal muscular atrophy that costs $2.1 million per dose. In March 2026, researchers at University College London published a case study of a child who developed severe hepatitis seven weeks after treatment. They found manufacturing contaminants (residual DNA fragments from the production process) in about 5% of the child's liver cells, triggering an immune response.
Then there's the X-linked myotubular myopathy (XLMTM) program, where an AAV8-based therapy caused four fatal liver failures despite showing real efficacy. Preclinical mouse models had completely missed the risk, exposing a fundamental gap in how the field predicts human toxicity.
The pattern is clear: high doses of AAV, especially in pediatric patients with neuromuscular diseases, carry serious hepatotoxic risk. The viral capsid (the protein shell of the delivery truck) provokes immune responses. Manufacturing impurities add fuel to the fire. And current animal models don't reliably predict who will be fine and who will end up in liver failure.
The Tightrope Walk
Sarepta's sirolimus strategy is essentially a calculated gamble. Add enough immunosuppression to protect the liver, but not so much that you leave vulnerable children defenseless against infections. It's like trying to turn down the volume on a fire alarm without also muting the smoke detector.
The company's own label already warns that immunosuppression increases infection risk. Stacking sirolimus on top of corticosteroids only amplifies that concern. These are children whose muscles are already failing them; a serious infection could be catastrophic.
There's also the efficacy question. Immunosuppression could theoretically dampen the very immune processes that help gene therapy work. If sirolimus blunts the body's response too effectively, dystrophin expression (the protein Elevidys is designed to produce) might suffer. That's why the ENDEAVOR trial is measuring both safety and dystrophin levels at 12 weeks: Sarepta needs to prove the fix doesn't break the therapy.
What Three-Year Data Tells Us (and What It Doesn't)
On the bright side, the ambulatory patient population has held up well. Three-year data from the Phase 3 EMBARK trial showed durable motor function benefits, including a 70%-plus reduction in functional decline on key movement tests compared to external controls. No new safety signals emerged in that group.
But "no problems in ambulatory patients" doesn't guarantee safety in non-ambulatory ones. The sicker population has weaker muscles (including respiratory muscles), less mobility, and potentially different immune profiles. Extrapolating safety from one group to the other is like assuming a car that handles well on dry pavement will be fine on ice.
The Regulatory Reckoning
The FDA isn't just watching Sarepta. It's requiring a post-marketing study of 200 patients as part of the updated label requirements. The agency also mandated enhanced monitoring protocols: liver function tests before infusion, weekly for at least the first three months, and as needed thereafter. Patients must stay near a healthcare facility for at least two months post-treatment.
For the broader gene therapy industry, the message is sobering. Accelerated approvals based on surrogate endpoints (like protein expression rather than long-term clinical outcomes) carry real risk. These safety events underscore that skepticism.
Expect the FDA to apply tighter scrutiny to every AAV gene therapy filing going forward. Companies like Regenxbio, pursuing DMD gene therapy through the accelerated approval pathway, will face harder questions about liver monitoring and immunosuppression protocols.
What Happens Next
The second half of 2026 will be decisive. If ENDEAVOR Cohort 8 shows that sirolimus meaningfully reduces liver injury while preserving dystrophin production, Sarepta could petition to resume commercial dosing in non-ambulatory patients. That would be a genuine scientific triumph, proof that the field can engineer around AAV's most dangerous flaw.
If the data disappoints, the implications extend well beyond one company or one disease. It would signal that AAV-based gene therapy has a hard ceiling on safety, one that no amount of immunosuppression can fix. And for the families watching from the sidelines, it would mean the $3.2 million miracle they were promised still comes with a price no parent should have to pay.
The gene therapy revolution was supposed to be a story about cures. Right now, it's a story about whether we can make those cures safe enough to use.
Clinical & Regulatory4 min read
The Gene Therapy the FDA Rejected Twice Just Got Approved
The FDA approved Rocket Pharma's gene therapy Kresladi for a rare immune disorder that kills 75% of untreated kids before age two. The twist: the agency had already rejected it twice, and the drug's clinical results were never the problem.
