For gene therapy developers, the field is becoming increasingly modality-diverse. For decades, a useful shorthand for vector selection was “AAV for in vivo, lentivirus for ex vivo.” That framework still captures much of the current landscape. Adeno-associated virus (AAV) vectors have underpinned several approved in vivo gene therapies, including Luxturna, while lentiviral and other integrating viral vectors have played a central role in the development of genetically modified cell therapies, including CAR T cell products.
But the boundaries between these approaches are becoming less distinct. The expansion of CRISPR-based genome editing, prime editing, and other advanced genetic engineering strategies is increasing demand for delivery systems capable of transporting larger or more complex payloads. At the same time, the clinical validation and scale-up of lipid nanoparticles (LNPs) for mRNA vaccines have accelerated interest in non-viral delivery platforms, particularly for applications where transient expression, repeat dosing, or flexible cargo design could offer advantages over viral vectors.
This article compares the current state of viral and non-viral gene delivery, moving beyond basic definitions to examine the practical trade-offs that influence platform selection. We will explore cargo capacity, delivery efficiency, immunogenicity, manufacturing and cost of goods, scalability, and the regulatory considerations that are shaping the next generation of gene delivery technologies.
Viral vectors
Viral vectors remain the most clinically validated approach to gene delivery, leveraging viral mechanisms that have evolved to efficiently enter cells and deliver genetic material. This biological efficiency has made viral vectors foundational to the field, but it also introduces limitations related to cargo capacity, immunogenicity, redosing and manufacturing.
1. Adeno-associated virus (AAV)
AAV is one of the most established platforms for in vivo gene delivery and has been used to target tissues including the retina, central nervous system, and liver. Its relatively favorable safety profile, broad range of capsid serotypes, and ability to support long-term episomal persistence in many non-dividing cells make it well suited to gene replacement and gene addition strategies designed to provide durable expression after a single administration.
Advantages:
- Established tropism: Naturally occurring AAV serotypes, including AAV9 and AAVrh74, as well as engineered capsids, provide different tissue and cellular tropisms. This established body of preclinical and clinical experience can help inform capsid selection and development strategies.
- Clinical and regulatory precedent: Multiple AAV-based therapies have reached clinical use, providing developers with substantial precedent for the characterization, manufacturing and regulatory evaluation of AAV products.
Key limitations
- Limited cargo capacity: AAV has an effective packaging capacity of approximately 4.7 kb, restricting its use for genes or genetic engineering systems that exceed this limit. This is particularly challenging for large transgenes such as dystrophin, which is approximately 14 kb, as well as increasingly complex genome-editing systems. Dual- or multi-vector strategies can overcome some of these constraints, but they introduce additional challenges in vector manufacturing, characterization, dosing and coordinated delivery.
- Pre-existing immunity and redosing: Pre-existing neutralizing antibodies (NAbs) against AAV can exclude some patients from treatment or clinical trials, with prevalence varying substantially according to capsid serotype and patient population. In addition, immune responses to the vector can limit the feasibility of repeat administration. This creates an important consideration for therapies in which transgene expression may decline over time or where repeat dosing could otherwise be clinically desirable.
- Manufacturing complexity: AAV production generates a mixture of full, partially filled, and empty capsids. Separating genome-containing particles from empty and other product-related species is therefore an important downstream processing challenge. Achieving the required combination of purity, yield, consistency and scalability remains a significant contributor to AAV manufacturing costs.
2. Lentivirus (LV)
LVs are a well-established platform for ex vivo cell engineering, particularly in applications involving hematopoietic stem and progenitor cells (HSPCs) and T cells. Their ability to integrate the therapeutic transgene into the host-cell genome enables durable expression and, in proliferating cell populations, stable transmission of the transgene to daughter cells. This property has made LVs an important component of the development and manufacture of genetically modified cell therapies.
Advantages for developers
- Relatively high cargo capacity: LVs can accommodate payloads of approximately 8–10 kb, depending on vector design, providing greater capacity than AAV for larger or more complex constructs. This can enable multicistronic payloads incorporating elements such as chimeric antigen receptors (CARs), regulatory sequences and, in some cases, safety or control mechanisms.
Key limitations
- Integration-associated safety considerations: Because LVs integrate their genetic cargo into the host-cell genome, there is a theoretical risk that integration could alter the expression of endogenous genes, including genes involved in cell growth and proliferation. Modern self-inactivating (SIN) vector designs have substantially improved the safety profile of lentiviral vectors by reducing the potential for enhancer-mediated activation of nearby genes. Nevertheless, integration-site analysis and long-term follow-up remain important components of the clinical development of integrating vector-based therapies. For certain gene therapy products, regulatory guidance recommends long-term follow-up extending for many years after treatment.
- Manufacturing and product stability: LVs are enveloped viral particles and can be sensitive to physical and chemical stresses encountered during processing and storage. Parameters such as shear, temperature, pH, and formulation conditions can affect vector integrity and infectivity. This can complicate downstream processing, filtration and concentration and contribute to lower recovery compared with more robust viral particles such as AAV. The relatively low volumetric productivity of some LV manufacturing systems and the challenges associated with scaling production can also contribute substantially to manufacturing costs.
Non-viral vectors
Non-viral delivery systems offer a fundamentally different approach to genetic medicine. Rather than relying on viral entry and trafficking mechanisms, these platforms use synthetic materials to package and deliver nucleic acids. Their potential advantages include greater cargo flexibility, scalable manufacturing, and, in some applications, the possibility of repeat administration. These benefits, however, can come with challenges in delivery efficiency, tissue targeting, and intracellular trafficking.
1. Lipid nanoparticles (LNPs)
The clinical validation of LNPs through mRNA COVID-19 vaccines has accelerated their development beyond prophylactic vaccination and into therapeutic applications, including in vivo gene editing. LNPs are particularly well suited to transient delivery of nucleic acids such as mRNA, making them attractive for genome-editing applications in which a short period of protein expression is sufficient to achieve the desired edit.
For example, LNPs can deliver mRNA encoding a genome-editing enzyme such as Cas9 alongside the appropriate guide RNA. Transient expression can then limit the duration of exposure to the editing machinery, although it does not by itself eliminate the risk of off-target editing.
Advantages for developers
- Cargo flexibility: LNPs can accommodate a range of nucleic acid cargos, including mRNA, siRNA, and other RNA modalities. Their capacity and formulation flexibility can also provide advantages for larger or more complex payloads that are difficult to package into conventional viral vectors.
- Potential for repeat administration: Unlike viral vectors, LNPs do not contain viral capsid proteins that can drive anti-vector immune responses. This creates the potential for repeat dosing, although immune responses to individual formulation components, including polyethylene glycol (PEG), and innate immune activation can still affect tolerability and dosing strategies.
- Scalable manufacturing: LNPs can be produced through controlled mixing processes, including microfluidic and impingement-jet technologies, that enable relatively reproducible particle formation and are compatible with continuous or large-scale manufacturing approaches. The established manufacturing infrastructure for mRNA-LNP vaccines has further accelerated development of the platform.
Key limitations
- Biodistribution and tissue targeting: Conventional LNP formulations have a strong propensity to accumulate in the liver following systemic administration. While this property is advantageous for liver-directed therapies, it presents a significant challenge for applications targeting extrahepatic tissues such as the central nervous system, skeletal muscle, and lung. Researchers are therefore developing new ionizable lipids, lipid compositions, and targeting strategies to alter LNP biodistribution and improve delivery to specific cell types.
- Endosomal escape: Following cellular uptake, a large proportion of LNP cargo remains trapped within endosomal compartments and is ultimately degraded. Efficient endosomal escape therefore remains one of the major barriers to intracellular delivery. Improving escape efficiency while maintaining particle stability and minimizing cellular toxicity is a central focus of LNP formulation research.
- Formulation-dependent tolerability: Although LNPs avoid some of the immunological limitations associated with viral vectors, their lipid components can activate innate immune pathways or contribute to dose-limiting toxicities. The composition, dose and route of administration therefore need to be optimized for each therapeutic application.
2. Emerging Technology
Beyond established viral and non-viral delivery platforms, researchers are developing a range of engineered biological and biomimetic systems that aim to combine the delivery efficiency of viral particles with some of the safety and engineering advantages of non-viral approaches. Virus-like particles (VLPs) and extracellular vesicles (EVs), including exosomes, are two such approaches attracting increasing interest.
- Virus-like particles (VLPs): VLPs retain structural features of viruses that enable efficient cellular entry but lack viral genetic material. This makes them attractive for transient delivery of genome-editing components, including RNA and ribonucleoprotein complexes. Key challenges include cargo loading, tissue targeting, and scalable manufacturing.
- Exosomes: EVs are naturally produced membrane-bound particles that transport proteins and nucleic acids between cells. Their biological origin makes them promising delivery vehicles for RNA and other therapeutic cargos, but efficient cargo loading, manufacturing consistency, and scalable production remain significant challenges.
Comparative Cheat Sheet
For a quick reference during your CMC strategy meetings, here is how the three major platforms stack up in the current environment.
Feature | AAV (Viral) | Lentivirus (Viral) | LNP (Non-Viral) |
|---|---|---|---|
Primary Use Case | In vivo gene replacement (CNS, Eye, Liver) | Ex vivo cell therapy (CAR-T, HSCs) | Gene editing (CRISPR/mRNA), Vaccines |
Cargo Capacity | ~4.7 kb (Strict) | ~10 kb (Moderate) | Flexible / High |
Genetic Persistence | Episomal (Long-term in non-dividing) | Integrated (Permanent in dividing) | Transient (Ideal for editing) |
Immunogenicity | High (NAbs prevent re-dosing) | Low (Use is mostly ex vivo) | Low (Re-dosable) |
Manufacturing COGS | High (Complex cell culture & purification) | High (Shear sensitivity, low yield) | Low to Medium (Chemical synthesis) |
Key CMC Bottleneck | Empty/Full Capsid Separation | Viral Stability & Titer | Lipid Purity & Microfluidic Fouling |
The regulatory & manufacturing landscape
The choice between viral and non-viral delivery is increasingly influenced not only by biological performance, but also by regulatory requirements, manufacturing complexity, and cost.
1. A changing regulatory landscape
Regulators are continuing to refine their approaches to gene and cell therapy development as new delivery technologies enter the clinic.
- Regulatory pathways and platform approaches: The FDA’s Regenerative Medicine Advanced Therapy (RMAT) designation can provide eligible developers with opportunities for increased interaction with the agency and expedited development pathways. More broadly, regulators and developers are exploring how knowledge gained from established delivery platforms can inform the development of new products. However, the extent to which manufacturing and preclinical data can be leveraged across products remains dependent on the specific platform, product and regulatory context.
- Post-market data: Viral vectors, particularly integrating vectors, can require extended post-treatment monitoring because of potential long-term risks associated with genomic integration. For non-integrating delivery systems such as LNPs, the duration and scope of follow-up are instead informed by factors including the persistence of the therapeutic effect, the nature of the cargo, and potential long-term biological effects. For in vivo genome-editing therapies, this includes assessment of potential off-target editing and other genomic changes.
- Potency assays: As delivery technologies become more sophisticated, potency testing increasingly needs to reflect the product’s mechanism of action rather than relying solely on measurements of transgene expression. Depending on the modality, assays may need to evaluate multiple aspects of activity, such as cellular uptake, intracellular trafficking, endosomal escape, and functional delivery of the therapeutic cargo.
2. Manufacturing
As advanced delivery platforms mature, manufacturing is becoming an increasingly important determinant of product performance, consistency, and cost. In this context, the principle that “process is product” is particularly relevant for complex biologics and gene delivery systems.
- Viral: Manufacturers are optimizing production systems to improve productivity and reduce reliance on resource-intensive processes such as transient plasmid transfection. Stable producer and packaging cell lines are being developed alongside improvements in upstream culture conditions and downstream purification, including chromatography technologies designed to improve the separation of full and empty AAV capsids.
- Non-viral: For LNPs, greater attention is being placed on the quality, purity, and consistency of lipid raw materials. Variations or impurities in lipid components can affect particle formation, stability, potency and tolerability. Advanced analytical techniques, including NMR spectroscopy and LC-MS, can be used to characterize lipid identity, purity and composition and support tighter control of critical raw materials and the final formulation.
Conclusion
There is no universally optimal gene delivery platform. The choice between viral and non-viral systems depends on a combination of factors, including the target tissue and cell type, cargo size and composition, required duration of expression, dosing strategy, immunogenicity, manufacturing requirements, and cost of goods.
Delivery platform selection is a product-specific optimization problem. The most effective approach is not necessarily the technology with the highest delivery efficiency in isolation, but the one that provides the appropriate balance of efficacy, safety, durability, manufacturability, and scalability for a particular therapeutic application.
As the field expands, developers are increasingly able to select delivery technologies according to the biology of the disease rather than committing to a single platform across an entire pipeline. AAV, lentiviral vectors, LNPs, and emerging delivery systems will each have applications where their particular properties provide a meaningful advantage.
The future of genetic medicine is therefore unlikely to be defined by a single winning delivery technology. Instead, continued innovation across viral and non-viral platforms will expand the range of genetic medicines that can be delivered safely, effectively, and at commercial scale.
