Gene therapies represent a very different paradigm than conventional therapeutics. Not only are most single-dose, curative treatments, many candidates in the clinic today and nearing commercialization target rare diseases with limited patient populations and no existing treatment options. Product development companies and patient advocacy groups are pushing strongly to get these products rapidly moved through the development cycle. In addition, it is essential for innovators to be first to receive approval in a particular indication in order to address the initial and greatest demand.

Time is of the essence, and consequently developers often compromise on their manufacturing processes—not from a safety perspective, but with respect to scalability and cost of goods—in ways that would not be acceptable for engineered proteins and monoclonal antibodies (mAbs).

Complicating the situation is the fact that adeno-associated viral (AAV) vectors are highly complex products comprising both capsid proteins and the genetic material contained within each capsid. Once administered to a patient, each vector must infect the correct cells, effectively translocate the genetic material to the nucleus, and then express the desired protein, which then must perform its intended function. This is a complex, multi-step process and multiple characteristics of the AAV therapeutic affect its ability to perform any one of these critical functions, on which potency depends. There is no one-size-fits all manufacturing solution for these complex products. Purity requirements, vector quantity, and other needs vary depending on the indication, size of the patient population, route of administration, and dosage required.

Academic origins
Issues with scalability, reproducibility and comparability have arisen because many of the production processes for AAV vectors used for gene therapy were initially developed by academic research centers in laboratory settings. They are typically labor-intensive with numerous manual manipulations making them impractical for GMP manufacturing at larger scale.

These academic processes typically utilize adherent cells in plastic multi-layer flasks that cannot be scaled up, and even scaling out can become impracticable beyond low vector requirements. Downstream processes routinely utilize ultracentrifugation, which achieves target purity levels but is also not scalable and can be challenging to validate. Additionally, this ultracentrifugation method results in a very high percentage of full capsids in the final product, setting a high bar for the product profile going forward that cannot be replicated by more scalable techniques.

Due to time pressures, however, companies often take their products into Phase 1 clinical trials using these unsuitable processes. When they realize success, they are faced with the challenge of moving to a new process that is more scalable and enables the reliable and robust manufacture of the larger quantities of material required for late-phase clinical trials and ultimately the market.

Process changes during drug development necessitate the ability to demonstrate comparability, which is often lacking.

Upstream production platform limitations
The most widely used method for AAV production is transient transfection in HEK293 cells using three plasmids and a transfection reagent. It is often preferred over other options because it provides the fastest route to the clinic. Additionally, this method has been demonstrated to be successful across all serotypes for both clinical and commercial products, implying general acceptance by the regulatory bodies.

Consequently, the choice to move forward with a transient-based production method is often perceived as low risk, especially when performed in suspension-based systems, which are considered more scalable than the adherent-based processes used in the aforementioned academic methods.

While far more robust and scalable, scalability remains an issue due to the inherent nature of the transfection process, which becomes significantly less efficient as scale increases. As a result, acceptable performance can only be achieved at 200 to 500 liters, which is well below commercial mAb processes. Additionally, vector yield and percentage full capsids can vary widely between serotypes and even transgenes with the transient system, further stressing the paradigm that what is a suitable process for one product may not be for another.

Downstream separation issues
Significant advances have also been achieved with respect to downstream purification of AAV vectors, with most serotypes being amenable to standard, well-established and scalable chromatography-based methods. Challenges remain regarding the separation of full and empty capsids, though. Because patients are dosed based on viral genomes (vgs), empty capsids contribute to the overall viral load without delivering the gene of interest (GOI), prompting immunogenicity concerns. Current ion exchange (IEX) chromatography methods are more scalable but only enable enrichment, rather than true separation, of full capsids because IEX relies on the very subtle charge difference between capsids containing DNA and empty capsids.

For this reason, there are some cases where ultracentrifugation, which is not readily scalable, is preferred if not required. This situation arises when an academic process taken into the clinic relied on ultracentrifugation and IEX chromatography cannot provide the same percent full capsids, prohibiting a process change during clinical development. It may also occur with certain indications (i.e., ocular) where only small volumes can be administered and the percent full capsids must be very high.

The presence of partially full capsids that may contain truncated transgenes, host-cell DNA, plasmid DNA and other genetic material presents an additional challenge, as these subspecies are even more difficult if not impossible to separate from full, intact-GOI-containing capsids using IEX. Further compounding the issue is a lack of high-throughput, sufficiently sensitive and accurate analytical tools for product characterization with respect to empty, partial and full capsids.

There is a clear need for an optimized, charge-based chromatography method—or other disruptive technologies—that can truly separate empty and partial capsid subspecies from full capsids. Similarly, effective analytical tools must be developed for measuring the batch-to-batch variability in full/partial/empty capsid ratios (similar to what is needed for evaluating post-translational modifications with mAbs) so that relevant process parameters impacting process behavior can be identified and controlled.

Lack of appropriate analytical tools
Indeed, the need for advanced analytics is an overarching challenge facing all steps involved in the manufacture of AAV vector-based gene therapies. Unfortunately, current analytical methods are not sufficiently robust for assessing the comparability of products produced using different manufacturing processes, often prohibiting much-needed process changes. 

Existing methods, and even our understanding and agreement as a scientific community on what characteristics of the vector may truly impact safety and efficacy, are still evolving. This uncertainty makes monitoring the impact of process changes and potential batch-to-batch variability on vector product quality difficult using existing methods that have garnered widespread industry support.

Simply stated, platform methods and the surrounding regulatory expectations do not yet exist for AAV vectors like they do for traditional biologics. The drive for speed to clinic does not help the issue at hand, and often results in product developers going into a Phase 1 clinical trial without having a robust potency assay established.

When this very key measure of a product’s efficacy is not established, this approach may present an insurmountable challenge with regard to making process changes. The evolving regulatory landscape around potency assays for AAV therapeutics further compounds the issue, with recent feedback from FDA requiring that these assays not just demonstrate expression of the therapeutic transgene, but also the correct functional output as well.
   
For these reasons, there is significant aversion to changing processes even if they lack scalability and robustness.

Capacity and sourcing issues
Access to capacity and crucial raw materials has become an additional challenge in recent years as increasing numbers of gene therapy candidates progress to later development stages. This issue has been magnified greatly by the emergence of the COVID-19 pandemic and the need to produce billions of doses of vaccines and therapeutics to fight the SARS-CoV-2 virus, which often utilize some of the same capacity, expertise, and raw materials as AAV vectors.

Limited capacity for viral vector production, as well as plasmid DNA and other raw materials (serum for adherent processes, transfection reagents, etc.), and long lead times for single-use equipment components combined with a limited pool of experienced and skilled operators are also constraining AAV gene therapy developers, whether they are startups or established biotechs. Some companies have elected to establish inhouse capabilities to overcome these challenges.

Even so, there are very few players with the full gamut of AAV gene therapy development, production, and testing expertise under one roof. Most companies will likely rely on a hybrid approach with both inhouse and outsourced manufacturing capacity. A hybrid approach offers additional appeal in managing sometimes very robust pipelines and fluctuations in material needs that is unlike traditional biologics, which follow a more predictable cadence of material needs for clinical and commercial production.

Solutions for the immediate term
Despite these challenges, today’s processes are significantly improved over the academic production platforms of the past. For the immediate term, it is important to focus on existing production processes that are more appropriate for GMP manufacturing, scale up and commercial launch.

Transient transfection is the fastest and most widely used and regulatory-validated approach, and the scalability limitations for suspension cell culture can be successfully overcome with scaling out and pooling strategies.

Fixed-bed bioreactor options for scalable adherent transient transfection are also now available. The iCELLis Bioreactor platform from Pall Biotech has been used for the commercial production of approved products, and the introduction of the scale-X fixed bed bioreactor portfolio from Univercells Technologies supports the viability of this approach, which in theory has the potential to offer greater scalability than the iCELLis Bioreactor with a comparable growth substrate for the cells.

Co-infection methods using the insect-based baculovirus expression vector system (BEVS) and human-derived herpes simplex virus (HSV) Type 1 systems are ultimately more scalable than transient transfection. Active infection methods, in fact, have the potential to be efficient even at the 2,000L scale. Additionally, these methods often result in higher yields by at least 10-fold and higher percentages of full capsids out of the bioreactor, increasing the feasibility of IEX methods to achieve an acceptable purity profile.

They have issues, though. Co-infection methods require generation of recombinant viral stocks, which can add up to one year to the development timeline. In addition, BEVS only affords high infectivity and potency results for one or two AAV serotypes and doesn’t always produce AAV vectors with suitable product quality profiles. HSV, meanwhile is a large enveloped virus that is challenging to manufacture in its own right and cannot be sterile filtered. The HSV production platforms also results in a different residuals profile that must be robustly controlled to ensure it does not impact patient immune responses.

Downstream purification issues have been discussed above. IEX chromatography is more scalable than ultracentrifugation, but in some cases the latter cannot be avoided. True separation of full capsids in a scalable and reproducible manner and without massive product loss remains a long-term goal.

Solutions for the longer term
The most robust and scalable solutions for AAV vector manufacturing leverage packaging and producer cell lines. Packaging cell lines contain the rep and cap components and only the GOI must be transfected into the cell. Producer cell lines have both the rep and cap components and the GOI integrated into the host cell genome.

As with coinfection methods, establishment of packaging and producer cell lines adds significant time to the early development stage. The need to be first to market therefore limits interest in these approaches currently. Over the last few years, however, the use of packaging and producer cell lines has begun to increase, which in turn has led to advances in supporting technologies.

Because processes based on packaging/producer cell lines are not limited by the nature of the transfection process, they offer more scalability. It is important, however, not to compare them directly to Chinese Hamster Ovary (CHO) cells for mAb cell culture. Mammalian cells lend themselves much more easily to the production of antibodies.

The design of packaging/producer cells for AAV vectors is quite different because cells do not want to make viral proteins, as they are toxic to the cells. A switch must be incorporated into the cells that keeps the viral proteins turned off until the right moment in the production process. As a result, there is an added layer of complexity involved. Currently those switches rely on antibiotics or viruses, and “leaky” cells may have some low level of production of the viral proteins, negatively impacting the expansion of healthy cells.

Even so, packaging cell lines might provide an interesting route for getting to market quickly. Initial investment in time and resources would be required to develop a packaging cell line for a particular AAV serotype. Once that was accomplished, however, only the GOI would need to be transfected to develop new therapeutic candidates. Since many drug developers platform on a single serotype, this could be an appealing option.
Once success for a gene therapy was demonstrated in humans, the packaging cell line could then be converted into a producer cell line. With this approach, it would be possible for all but the first of the platform pipeline products to reach the clinic quickly and still have the ability to move forward with a stable, reliable producer cell line. This approach also involves minimal process changes since the host cell line remains the same, reducing any potential comparability hurdles.

Of course, even if really disruptive production and purification technologies are successfully implemented for AAV vector manufacturing, it won’t be possible to fully leverage them unless the analytics needed to reliably prove comparability are also developed. The industry as a whole must get aligned on what needs to be measured and how.

It cannot be stressed enough that putting off development of potency assays until after Phase 1 clinical studies can leave drug developers scrambling to put good methods into place when their products are successful and move to the late-stage development phase. The need to demonstrate correct protein expression and correct protein activity is a very complex challenge that can be compounded if left until late in the development cycle. The lack of a good potency assay also makes the burden of demonstrating comparability even higher, and practically ensures that no process changes will be made regardless of the need for optimization.

Progress to be made
Gene therapy developers’ long-range success depends on establishing scalable, commercially viable manufacturing processes. Unfortunately, current AAV manufacturing platforms are not suitable to support the expansion of gene therapy past monogenic, rare diseases to the treatment of diseases with larger patient populations. We have an amazing platform in our hands to directly treat the genetic basis of disease and we, as an industry, must push for more scalable and cost-effective methods in order for it to be affordable to all. 

To enable truly robust, reproducible, large-scale AAV vector manufacturing chains, disruptive technologies will be required to dramatically increase upstream production yields and achieve downstream separation of products from product-related impurities such as partial/empty capsids. While there has been some progress, it needs to go further, to clearly define the challenges to enabling the upstream and downstream platform processes that are desperately needed. Appropriate analytics are equally critical in enabling the implementation of these advances.

Since inception, The Center for Breakthrough Medicines (CBM) has actively aligned with partners across the industry to develop more knowledge and understanding of the challenges, existing solutions, and future potential solutions for advancing novel therapies. The goal is to build a toolbox of technologies and services around the critical quality attributes that impact the efficacy, safety, and stability of next-generation medicines that will allow rapid development and commercialization of advanced technologies such as AAV vectors for gene therapy. 

---

Sybil Danby is Senior Vice President, Business Development & Strategy at The Center for Breakthrough Medicines. Prior to joining The Center for Breakthrough Medicines, Ms. Danby spent over three years with responsibility for the strategic business development of Paragon Bioservices, playing an active role in grooming the company for acquisition by Catalent Pharma Solutions for $1.2B in 2019. She previously spent two years at Pall Corporation as a driver of the development and sales of disruptive manufacturing technologies, and just under nine years at GSK in both the manufacturing and R&D organizations working on clinical and commercial monoclonal antibody products.