Overcoming Recovery Challenges In AAV Production
By Tyler Menichiello, Chief Editor, Bioprocess Online
The majority of in vivo gene therapies in development are built using AAVs, which are notoriously difficult to purify. Vector purity is critical for product quality and ensuring patient safety, and while teams are able to achieve high levels of purity in AAV production, this purity often comes at the expense of overall AAV recovery or product yield.
This inefficiency ultimately results in higher COGs and more expensive therapies.
Earlier this year, we hosted a Bioprocess Online Live event that explored the tradeoff and relationship between viral vector quality and quantity. This event, titled “Downstream Decisions To Maximize Vector Yields And Reduce COGs,” featured Nathalie Clément, Ph.D., senior VP of vector development at Siren Biotechnology, and Hung Trinh, Ph.D., CEO at Vertex Biopharm Consulting.
Afterwards, Clément joined me on the “Better Biopharma” podcast to continue the conversation, speaking from her experience in the AAV space. We talked about advancements in downstream AAV production, the challenge of purifying AAV capsids, and which unit operations she recommends combining to optimize vector recovery.
Below are excerpts from the full podcast episode, which you can find here (and wherever you get your podcasts).
The accompanying transcriptions have been edited for clarity.
Shifting Attention Away From Upstream
Historically, a lot of effort was focused on optimizing AAV yields upstream, but according to our event panelists, that seems to be changing. Now, Clément says, teams are starting to look downstream for ways to boost process performance and minimize losses.
Why do you think the industry is finally shifting its attention downstream? Do you think we’re finally seeing improvements in downstream recovery?
Clément: When I say the industry has spent a lot of time optimizing upstream, it’s actually pretty recent. We’re talking about the past 10 years. Before that, around the birth of [therapeutic] AAVs and the next 10-15 years after, it was upstream and downstream — about half and half at the beginning. After several years of seeing low recovery (around 5% to 10%), there was suddenly an interest in optimizing processes upstream.
The effort to improve upstream is still relatively recent; I would say it’s been emphasized for about the past five to eight years. We’ve now reached a point where vector yield has improved upstream along with vector quality, but downstream yield recovery hasn’t really improved much. It has improved, but we’re still at around 30% recovery in the end, at best.
As we say in the field, every drop of AAV is like gold, so we’ll take everything we can. An increase in recovery by 10% or 20% has a huge impact, but we can do more. Between 30% and the ideal, theoretical achievement of 100% recovery, there’s a lot of room for improvement.
To answer your question, the focus on downstream optimization goes back about five years. It’s always been a kind of back and forth. Until we get the balance where we’re satisfied with upstream and satisfied with downstream, it’s the role of CMC to decide when it’s time to stop development and move to GMP manufacturing for the clinic or scale up to commercial.
It’s always very challenging but also important for a company to say, “Now’s the time. We’ve invested enough in process development; we’ve optimized the best we could; now it’s time to stop, because the patients are waiting.”
Knowing When Pure Is Pure Enough
Chasing purity endlessly can have diminishing returns in terms of developmental and manufacturing costs. As Clément says, it’s up to CMC to decide when pure is pure enough. She explains how she looks at AAV purity, which CQAs matter the most to measure purity, and how indication or dosing requirements factor into these decisions.
Do you think the industry has become too focused on maximizing purity as opposed to recovery? At what point is pursuing higher purity not worth the squeeze?
Clément: That’s the million-dollar question! The best way I can answer that is: It’s pure enough when it’s safe enough.
There’s no specific metric for “pure.” In the certificate of analysis of clinical AAV material, there is a test called “purity,” and you’re getting a percentage, but that’s mostly just based on protein purity. You have other tests that measure things like residual DNA or residual benzonase, which are components of a purity assessment, but the field has yet to combine all of these measurements into one number to provide a percentage of “pure” that considers all these impurities.
That aside, it’s a case-by-case thing. It’s really about what your CMC scale looks like or what your commercial manufacturing plan looks like. It’s very different if you’re going for an ocular disease versus a systemic disease like Duchenne muscular dystrophy. For one, you’re adding 1011 or 1012 vector genome; for the other, it’s four log higher — 1015, 1016, and closer to 1017 — which is a huge difference.
If you’re going with a large-scale bioreactor, but your end product is going to be filled at a very low concentration for a very small volume to be administered to the patient, then I feel like you have more room to add steps if you want to reach a higher purity.
As I said, it’s case-by-case. Purity is assessed during your toxicology studies, which should tell you if your product is safe enough to start a clinical trial. Your human data are helpful in understanding and assessing product safety, and then adjustments are based on that.
Advancements In AAV Recovery
While today’s average of 30% total AAV recovery may seem low, it’s a lot better than the historically low 5% or 10%. Clément explains the historical advancements that allowed for this increase in recovery and shares her thoughts on where teams can focus to improve recovery further, beyond 30%.
What have been the biggest advancements that increased recovery percentage, and what do you think will get us past 30%?
Clément: There were big milestones in downstream optimization. To explain them, I probably have to provide a bit of history on how new AAV serotypes came along and drew a lot of attention for clinical purposes, specifically, AAV8 and AAV9 in the early 2000s. Before these serotypes, the field was working with the traditional AAV2, and there were a few groups working with AAV1 (talking specifically from the lens of therapeutic AAV development, not AAVs for research).
For clinical development, AAV2 and AAV1 were first, and they had their downstream processes pretty much locked down from the beginning. One major reason was that we knew the receptors for both AAV2 and AAV1, so we could immediately develop chromatography resins to capture these particles from a crude upstream lysate.
When the novel serotypes arrived (AAV7, AAV8, AAV9), we didn’t know the receptors, and suddenly we didn’t have capture resins to develop and purify these AAVs. Because scientists are never short of ideas, there were a lot of fun and powerful protocols put in place, but the field eventually started using capsid-specific antibodies in the capture step. To me, this was the most impactful addition to the downstream process.
These antibodies recognize formed AAV capsids and bind them extremely well. This step alone is 80% to 90% efficient in recovering AAVs, which represents a huge leap in improving downstream recovery. After the capture step comes polishing, which is usually an affinity chromatography step to separate full and empty capsids.
Better columns, resins, and protocols have been developed to be good enough, but I would say there’s still room for improvement.
Chromatography had a big impact, but another factor is filtration. During downstream, multiple filtration steps remove impurities like proteins, residual DNA, or residual RNA. A lot of the new serotypes were lost in these filters in ways that we could not really explain, so there’s been a lot of improvement there.
Altogether, I feel like the recovery of each step individually is better, but because we still have an average of four to six steps in downstream production, the combination or additive loss across all the steps still remains too high in my opinion.
Combining Downstream Unit Operations
While AAV recovery has improved over time for each purification step, the additive losses Clément referred to amount to a significant loss of product across the entire downstream process. To reconcile these losses, she suggests combining downstream unit operations wherever possible or even eliminating some altogether.
Which unit operations deserve the most attention in the name of optimizing AAV recovery downstream? Which do you think can be most effectively combined or eliminated altogether?
Clément: Combining unit operations is my motto. Reducing the number of steps helps remove some of the small losses all around; it’s easier, faster, and more efficient.
Two unit operations I would love to combine are the capture step, where you’re capturing most of the AAVs, and the separation step, where you’re separating full and empty. Right now, they’re different steps: It’s affinity chromatography with antibodies for the capture step, and it’s an anion exchange chromatography for the separation step.
Combine them! Capture, but try to capture mostly the full capsids. We’re not there yet, but I bet we will be one day.
The other unit operation(s) to reconsider are the filtration steps. Although recovery is usually pretty good, there’s always loss associated with filtration by even just the volume that’s lost or retained within the filters. It may not always be necessary to have filtration in between each step. I would challenge teams that have filtration at every step to really look at what they’re gaining. Does it add anything in the way of purity? Do you remove residuals when you’re filtering? If so, how much? Could you potentially skip a few of these filtration steps without impacting your chromatography?
If you liked these highlights, check out the full “Better Biopharma” episode with Nathalie Clément, Ph.D. Subscribe to be notified about new episodes, released every other Wednesday, wherever you get your podcasts!