Today, I wrote a bit about AAV serotypes and doses, and target organs, inspired after reading Reporter’s notebook
CAP-002 case
clinical fatalities and the limits of IV-delivery
A child died in Capsida’s CAP-002 trial in September 2025, the first patient treated with an engineered, IV-administered, BBB-crossing AAV for STXBP1 encephalopathy.
Capsida Biotherapeutics voluntarily halted the SYNRGY trial (NCT06983158) after the first treated paediatric patient with developmental and epileptic encephalopathy (DEE) caused by STXBP1 mutations, died. The cause of death remains under investigation. CAP-002 is notable because its capsid was specifically engineered to cross the blood-brain barrier (BBB) via IV infusion, without intracranial injection. Regulators did not enforce the hold. However, Capsida self-imposed it while searching for the root cause.👍
Now, we should check some adverse events linked to IV AAVs.
Reported patient deaths linked to IV AAV by serotype
| Serotype | Drug / company | Indication | Cause of death | Status |
| AAV9 | Zolgensma (Novartis) | Spinal muscular atrophy (SMA) | Acute liver failure (ALF) | 2 deaths, 2022 (Zhang et al.) |
| AAV9 | Multiple high-dose trials | Various CNS / neuromuscular | Thrombotic microangiopathy (TMA) | Multiple cases (Zhang et al.) |
| AAV9 (NGN-40) | Neurogene | Rett syndrome | Rare hyperinflammatory syndrome | 1 deathPhase I/II (Joshua Silverwood) |
| AAV9-based (RP-A501) | Rocket Pharmaceuticals | Danon disease | Fatal acute systemic infection | 1 death, Phase II (Annabel Kartal Allen) |
| AAV8 | ASPIRO trial (AT132) | X-linked myotubular myopathy (XLMM) | Cholestatic liver failure | 4 deathsPhase I/II (Shieh et al.) |
| AAVrh74 | Elevidys (Sarepta) | Duchenne muscular dystrophy | Acute liver failure | 3 deaths (Annabel Kartal Allen) |
| Engineered AAV (CAP-002) | Capsida Biotherapeutics | STXBP1 encephalopathy | Unknown — under investigation | 1 death, Sept 2025 (Annabel Kartal Allen) |
What the AAV5 story teaches us
Now, we have to think about AAV5’s successful story. A decade of relative safety, why?
In August 2022, Europe approved the world’s first AAV5-based gene therapy for severe hemophilia A – Roctavian (valoctocogene roxaparvovec), developed by BioMarin. The FDA followed in June 2023. Since Phase 1/2 trials began in the mid-2010s, hundreds of patients have received a single IV infusion of AAV5, and to date, no patient deaths have been directly attributed to the therapy. In a field where other AAV programs have seen multiple fatalities, this is a notable record.
Hemophilia A is caused by a missing clotting protein (Factor VIII) that is normally made in the liver. So the therapeutic goal of Roctavian is simply to deliver a working gene to liver cells and the liver happens to be exactly where AAV naturally travels after an IV injection. The vector does not need to fight its way past any biological barriers. It goes where it was always going to go.
So what makes AAV5 different? The answer is not the stereotype itself [I think, and some others would agree with me although it is some part of stories].
The dose matters.
Because the liver is the natural destination for IV-delivered AAV5-based gene therapies, a relatively modest dose is enough to achieve a therapeutic effect in hemophilia. Roctavian is given at 4–6 × 10¹³ vg/kg, roughly half to a third of the doses used in CNS or muscle-targeting programs. That difference in dose is, in large part, the difference between a therapy with an acceptable safety profile and one that has killed patients.
Why does the target organ change everything?
Think of it this way. When you give AAV through an IV, the vector enters the bloodstream and travels throughout the body. The liver acts like a sponge as always. It absorbs a large fraction of whatever AAV is circulating, regardless of where the doctor wants it to go. This is simply how our body works.
For hemophilia, this is actually favorable: you want the gene delivered to liver cells, and the liver is already soaking it up. A dose of 4–6 × 10¹³ vg/kg is enough to transduce enough liver cells to restore Factor VIII production. The liver handles this dose without triggering a dangerous immune response in most patients.
But for diseases of the brain or muscle, the liver is an obstacle, not a destination. To get enough AAV past the blood-brain barrier or into muscle tissue, you have to flood the entire system with a much larger dose, often 2 to 10 times higher. The liver still absorbs most of it, now overwhelmed by vectors it was never meant to receive in such quantities. The result, in the worst cases, is acute liver failure, immune storms, or vascular damage.
The safety record of AAV5 in hemophilia is genuinely encouraging, but it would be a mistake to conclude that AAV5 is simply a “safe” serotype. The real lesson is more specific and more important: IV gene therapy works best and most safely when the target organ is the liver. AAV5 has never been tested at the doses that CNS or muscle delivery would require, so we simply do not know how it would perform in that context.
What the full clinical picture tells us is that the moment gene therapy asks AAV to travel beyond the liver via IV, the dose requirements climb into a range where serious toxicity and death become real risks regardless of which serotype is used. AAV8, AAV9, and AAVrh74 have all produced fatal outcomes in this higher dose regime. Engineered capsids designed to cross the blood-brain barrier represent the field’s attempt to break this trade-off, but as the September 2025 death in Capsida’s trial shows, even the most advanced designed next-generation vectors carry unknowns that only human trials can reveal [The dose not disclosed yet].
A lower dose of IV gene therapy targeting other organs might be successful, which underscores the need for more efficient AAV engineering or local injection to resolve these matters.
Life and Biology stories with Sooyoung and Others 