Adeno-associated virus (AAV) is the preferred vector for most in vivo gene therapies because it is non-pathogenic, persists as a stable episome, and delivers genes to non-dividing cells with durable expression, while adenovirus (Ad) is used mainly for oncolytic and vaccine applications because its large payload capacity and strong innate immune activation come at the cost of high immunogenicity and short-lived expression. In practical terms, AAV doses run roughly 100-fold lower per kilogram (about 1×10¹³ to 1×10¹⁴ vector genomes per kg) than first-generation Ad doses (about 1×10¹¹ to 1×10¹² infectious particles per kg), and that gap explains most of the safety, manufacturing, and redosing differences between the two platforms.
What Separates These Two Vectors at the Capsid Level
The single biggest mechanistic difference is not the transgene but the capsid and the DNA it carries. AAV is a 25-nanometer, non-enveloped parvovirus with a single-stranded DNA genome of about 4.7 kilobases. Adenovirus is a 90- to 100-nanometer, non-enveloped adenovirus with a double-stranded DNA genome of roughly 36 kilobases. That 4.7-versus-36 kb ratio is the root cause of nearly every trade-off downstream.
AAV enters cells through receptor-mediated endocytosis, escapes the endosome, and traffics to the nucleus, where its genome is converted to double-stranded DNA by host polymerases. In post-mitotic tissue such as neurons, photoreceptors, and hepatocytes, the vector persists as a concatemeric episome rather than integrating. Integration occurs at low frequency (estimated at 0.1 to 1 percent of transduced cells in some tissues) and is generally considered a low but non-zero genotoxicity risk rather than a designed feature.
Adenovirus works differently. Its capsid binds the coxsackie-adenovirus receptor and integrins, the virion uncoats at the nuclear pore, and its double-stranded genome enters the nucleus without needing host-mediated second-strand synthesis. Because adenoviral DNA is not efficiently maintained as a stable episome, expression from first-generation vectors fades within weeks as transduced cells turn over and the genome is diluted.
In plain terms: AAV is built to linger quietly, and adenovirus is built to deliver a loud, short pulse.
Quantified Comparison of Payload, Dose, and Expression
The table below reconciles the numbers that matter most for program design. It also resolves a common confusion: adenovirus dosing is usually reported in infectious particles or viral particles, while AAV dosing is reported in vector genomes. These are not interchangeable units, and comparing them directly without a conversion factor produces misleading statements. The dose ranges below are starting points for program planning, not clinical thresholds.
| Property | AAV | Adenovirus (first-generation) | Why it matters |
|---|---|---|---|
| Genome size | ~4.7 kb ssDNA | ~36 kb dsDNA | Ad fits large genes; AAV needs compact cassettes |
| Practical payload | ~4.4 kb with ITRs | ~8 kb in E1/E3-deleted, up to ~36 kb in helper-dependent | Large genes like dystrophin (11 kb) require Ad or split-AAV |
| Capsid diameter | 25 nm | 90–100 nm | AAV penetrates tissue better; Ad is cleared faster |
| Clinical dose range | ~1×10¹³–1×10¹⁴ vg/kg | ~1×10¹¹–1×10¹² infectious particles per dose | AAV requires 100–1,000× more capsids per patient |
| Expression onset | 1–4 weeks | 1–3 days | Ad suits acute indications |
| Expression duration | Months to years in non-dividing tissue | Days to weeks | AAV suits chronic disease |
| Pre-existing seroprevalence | 30–70 percent depending on serotype and region | 40–80 percent for common types | Both face neutralization barriers |
| Redosing | Difficult due to capsid immunity | Difficult for systemic use; possible for local delivery | Neither platform supports easy repeat systemic dosing |
| Inflammatory profile | Mild to moderate, dose-dependent | High (innate TLR and inflammasome activation) | Ad causes fever and hepatotoxicity more readily |
The dose reconciliation sentence worth remembering: AAV and Ad numbers overlap only when expressed in different units, so a 1×10¹⁴ vg/kg AAV dose and a 1×10¹² particle Ad dose can look similar in scale but differ by orders of magnitude in actual capsid burden per patient.
The Immunity Problem in Both Platforms
Capsid immunity is the shared bottleneck, but it behaves differently in each vector. For AAV, neutralizing antibodies against the capsid are the dominant barrier. Seroprevalence against AAV2 commonly sits around 30 to 70 percent depending on population and assay, and even low titers can blunt systemic liver transduction. This is why systemic AAV trials routinely screen out patients with high pre-existing neutralizing titers, typically using a cutoff in the range of 1:5 to 1:100 depending on the assay and program. That cutoff is not standardized across trials, and the lack of standardization is a real limitation.
Adenovirus faces a larger and faster immune response. Pre-existing antibodies plus memory T cells against common serotypes such as Ad5 mean that systemic dosing triggers a rapid innate response, followed by capsid-specific T-cell responses that clear transduced cells within days to weeks. This is the central reason first-generation Ad vectors fell out of favor for chronic genetic disease after early trials in the 1990s showed limited durability.
There is a non-obvious correction here. The 1999 death of Jesse Gelsinger in an adenoviral gene therapy trial is often cited as proof that adenovirus is simply too dangerous. That framing is incomplete. The dose was high, the vector was a first-generation E1/E3-deleted construct, and the patient had a complex underlying condition. The lesson was about dose escalation, immune monitoring, and trial design as much as about the vector class itself. Modern oncolytic adenoviruses use lower local doses and engineered capsids, and the safety profile looks different from that 1999 case.
Adenovirus vs. AAV: Where Adenovirus Still Wins
Adenovirus is not obsolete. It holds specific advantages that AAV cannot match.
- Large payload capacity. Helper-dependent adenoviral vectors can carry up to about 36 kb, which is useful for large genes, multiple transgenes, or regulatory elements that do not fit in AAV.
- Rapid, high-level expression. Adenoviral transduction produces high protein levels within days, which suits acute indications such as cancer immunotherapy or vaccine antigen delivery.
- Oncolytic selectivity. Engineered adenoviruses can replicate selectively in tumor cells, amplifying the dose locally and lysing tumor tissue directly.
- Local delivery. Intratumoral or intravesical delivery reduces systemic exposure and sidesteps the worst of the innate immune response.
- Manufacturing speed. Adenoviral vectors can be produced at scale relatively quickly for surge or outbreak scenarios.
AAV wins for chronic, systemic, and non-dividing tissue targets: liver, central nervous system, retina, and muscle. That is why approved AAV products cluster around monogenic diseases with a clear tissue target and a compact gene, while adenoviral products cluster around oncology and infectious disease.
The 2026 Landscape and the Timeline of Vector Development
The field has moved through distinct eras, and understanding that timeline prevents the mistake of treating "gene therapy" as a single technology.
| Period | Milestone | Platform focus |
|---|---|---|
| 1990–1995 | First clinical gene therapy trials | Retrovirus and early adenovirus |
| 1999 | Gelsinger trial death | Adenovirus safety crisis |
| 2000–2010 | AAV serotype discovery (AAV8, AAV9) | AAV tissue targeting expands |
| 2012 | First AAV product approval in Europe (lipoprotein lipase deficiency) | AAV enters clinic |
| 2017–2019 | AAV approvals for inherited retinal disease and spinal muscular atrophy | AAV commercialization |
| 2022–2023 | Hemophilia B AAV approval and high-dose hepatotoxicity scrutiny | AAV dose ceiling recognized |
| 2024–2026 | Capsid engineering, immune suppression protocols, redosing research | Both platforms, with AAV dominant |
The current period is defined less by new vector classes and more by dose optimization and immune management. The field has learned that higher AAV doses do not scale linearly in benefit but do scale in complement activation and liver enzyme elevation. This is why clinical programs increasingly target the lowest effective dose rather than the maximum tolerated dose.
Common Misconceptions About AAV and Adenovirus
Several beliefs circulate widely and are worth correcting.
The first is that AAV integrates into the genome and is therefore inherently risky. Most AAV DNA persists episomally, and integration is a low-frequency event. The risk is real but small, and it is managed through long-term follow-up rather than by avoiding the platform.
The second is that AAV is completely non-immunogenic. It is non-pathogenic, which is not the same thing. Capsid-directed humoral and cellular responses occur, and high-dose systemic AAV can trigger complement activation and thrombotic microangiopathy, a serious and sometimes fatal complication documented in neuromuscular and hepatic programs.
The third is that adenovirus is too dangerous to use. First-generation systemic adenovirus at high dose is indeed problematic, but localized and engineered adenoviral vectors are in active clinical use for cancer.
The fourth, and the most consequential for program design, is that a higher dose always means better efficacy. For AAV, the dose-response curve flattens, and beyond a threshold the added capsids increase immunogenicity without proportional benefit. For adenovirus, higher doses increase innate immune activation and can trigger cytokine release.
In plain terms: more vector is not more therapy past a certain point, and both platforms have a ceiling.
What to Actually Do When Choosing a Vector
The choice between AAV and adenovirus should follow from the indication, not from preference. A concrete decision sequence looks like this:
- Size the cassette first. If the transgene plus regulatory elements exceeds about 4.4 kb, AAV requires split-vector or dual-vector strategies, and adenovirus becomes more attractive.
- Define the target tissue and its turnover rate. Non-dividing tissue with a need for years of expression favors AAV. Rapid turnover or acute need favors adenovirus.
- Check seroprevalence constraints. Screen for neutralizing antibodies and decide whether to exclude high-titer patients, switch serotypes, or add transient immunosuppression.
- Model the dose and the immune ceiling together. Set a maximum tolerable capsid burden early, not after dose escalation.
- Plan the manufacturing route. AAV requires baculovirus or HEK293 suspension platforms; adenovirus requires permissive cell lines and rigorous replication-competent adenovirus testing.
- Build a long-term follow-up plan. Both platforms require monitoring for insertional risk and delayed immune events, with AAV requiring longer horizons.
Limitations and What Remains Uncertain
The most important limitation is that neither platform solves redosing. Systemic readministration of AAV is blocked by capsid antibodies, and systemic readministration of adenovirus is blocked by both antibodies and memory T cells. Research into capsid switching, immune depletion, and vector shielding is active but not yet standard of care.
A second limitation is manufacturing. AAV yields remain a bottleneck, and empty capsid ratios can vary widely between batches, which affects effective dose. Adenovirus manufacturing is more established but requires careful control of replication-competent adenovirus contamination.
A third limitation is that the long-term durability data for AAV beyond about 10 years remain sparse. The episomal persistence model predicts durability, but prediction is not the same as observation. Children treated with AAV in early trials are now the key cohort for answering this question.
In plain terms: the biology strongly favors AAV for chronic disease, but the field is still learning where the durability curve actually flattens, and that answer will determine which indications remain viable for each platform.