How CGT Leaders Are Reducing Failures, Scaling Processes, And Overcoming Regulatory Friction

By Tyler Menichiello, Chief Editor, Bioprocess Online

It’s a tough time for cell and gene therapies (CGT). Funding is scarce, with no immediate signs of improvement on the horizon. And despite the extraordinary innovation we’ve seen in recent years, there seems to be more pressure than ever to take these therapies to commercial, with almost no room for missteps. In this current environment, smaller biotech companies can’t afford to stall after Phase 1 — they need to be thinking about and scaling for commercial manufacturing in parallel to clinical development.
On the bright side, technology is evolving faster than you can say the “A” in AI, and there are more CDMOs with the CGT know-how to help than ever before. Yet, resources are preciously limited in this climate, and every decision — from who to partner with to what technologies to adopt — is high stakes.
These themes ran deep at this year’s BIO International Convention. Perhaps my favorite session of the week was the all-star, all-women panel of CGT executives who spoke about the challenges and pressure biotech companies face to scale. The panel, “Scalable Biomanufacturing for Living Medicines: Pioneering the Future Today,” was moderated by Marinna Madrid, Ph.D., co-founder and chief product officer at Cellino. It featured Julie O’Shaughnessy, Ph.D., COO at Vivodyne; Bruna Paulsen, Ph.D., VP of manufacturing and therapeutic development at Gameto; Kate Rochlin, Ph.D., COO at IN8bio; and Alexis Rovner, Ph.D., founder and CEO at 64x Bio.
These industry leaders shared their philosophy when it comes to reducing product failure, embracing AI, and building dynamic teams to stay ahead of the curve to bring curative CGT therapies to patients.
Reduce Failure And Timelines: Leaner Processes, Smarter Release Criteria
Reducing ATMP failure rates is an integral part of bringing these ATMPs to patients, O’Shaughnessy said early in the discussion, and an important part of that is having a highly reproducible product. The challenge many CGT companies run into, which Paulsen herself described at Gameto, is that of designing and validating potency assays that support reproducibility.
“Everyone assumes that there’s a ready-made assay for your asset to be tested with, and that is often not true,” Rochlin said. Having to build bespoke assays — or doing the work to co-opt existing assays — is a bottleneck that’s not talked about a lot, she said.
Another pitfall to bear in mind is when you’re defining a product’s release criteria. “You don’t want to fall into the very common product development trap of over-speccing your product,” said Madrid. Whatever you define in your release criteria has to be met, and adding unnecessary product specs can needlessly complicate and slow down product development. “If it’s not an important spec, then it probably shouldn’t be part of your release criteria to begin with,” she said.
“You should have as few release criteria as you possibly can,” Rochlin agreed, to avoid over-complicating products and shorten timelines. This philosophy also applies to her team’s process development. “The first thing we did was look at our process and say, ‘How many steps can we remove and have it be essentially the same process?’” she said. Simplifying processes and clearly defining release criteria — not waiting for the agency to dictate them for you — should help reduce the timeline to get into the clinic.
The requirements for clinical trials also need to be reconsidered when it comes to cell therapies, said Rochlin. Specifically, regulators should reconsider the need for fully randomized, controlled Phase 2 trials for diseases that have abundant real-world data on mechanisms and safety. She alluded to recent FDA guidance on this very issue.
“The cost of these trials, and the cost of having these bigger clinical cohorts, and controls, and processes that we can’t streamline is I think what’s killing a lot of cell therapy innovation,” Rochlin said. Reducing timelines and making it easier to conduct these trials will ultimately help more cell therapies make it through the clinic to pivotal trials as opposed to getting stuck in Phase 2.
Companies That Use AI Will Outcompete Those That Don’t
“Will AI replace jobs? I think, eventually, yes,” Rovner said as the conversation shifted to technology adoption. “But right now, we have these incredible tools at our disposal that are making us better able to make these medicines and find solutions to things, and it’s something we should be embracing.”
Perhaps when that day comes (the AI-workforce takeover), it will enable scientists to spend more time thinking about important problems and performing tasks that AI can’t, Rovner added.
While I selfishly hope this takeover doesn’t happen during my professional life, there’s no denying the growing power of AI, especially in biomanufacturing workflows. A big theme at BIO this year, as has been the case in other conferences and discussions I’ve attended, is the importance of good data in training and utilizing these tools.
“Language models are going to be hopeless without large, spectacular sets of data,” O’Shaughnessy said. She believes that with the right data, AI can provide insights to help cure disease and reduce the failure rate of ATMPs before they enter the clinic.
Rochlin agreed, suggesting that AI can be used to analyze complex genomic data to better understand diseases like glioblastoma. “How can we use AI to drive patient stratification, to understand relapse, to understand disease prognosis, progression, and all of those components?” she asked.
These tools can allow smaller companies to do what they otherwise wouldn’t have the capacity to, allowing them to compete with Big Pharma in an unprecedented way.
However, caution should be exercised when training teams on different AI models. Workers need to understand what is and what is not appropriate to be using it for, according to Rochlin. “You do have people sometimes inputting proprietary data or unpublished data into certain AI machines, which may be very helpful in helping you analyze it, but then that becomes part of the large language model and what it’s training on,” she said. “So, it is really important to not assume that people will know the difference.”
Training Biologists To Think Like Engineers
As the line between life sciences and technology blurs, building cross-disciplinary teams is increasingly more necessary and common. Innovation comes from thinking outside the box, said O’Shaughnessy, and solving the biggest problems in ATMP development and manufacturing requires pulling in engineers from other industries.
“At Vivodyne, we need the rigor of scientists, but those scientists also need to be engineers,” O’Shaughnessy said. “So, it’s marrying that biologist with the engineering mentality that’s going to allow us to be better, faster, [and] stronger.”
Madrid shared her philosophy on team building, which involves pulling experts together from diverse industries and backgrounds. “We have folks leading the AI team who came from the drone industry because they do a lot of image analysis,” she explained. “We have folks developing our fluid cassettes for cell culture from the gas and oil industry because there are a lot of really stellar fluid dynamics engineers from that industry.”
For legacy modalities (like antibodies), it’s fairly easy to find experts with the exact backgrounds you need for your pipeline. However, when you’re working with technologies that don’t have a lot of precedents, it’s almost impossible to hire people with the exact skillset you need, Rochlin said.
This is especially true when you consider how manufacturing processes differ among companies. Even if you find someone with experience manufacturing your type of cell therapy, “they probably did it very differently,” said Rochlin. When it comes to building a team, “You’re looking for people who have a certain mindset,” she said, people who want to solve problems and can do so creatively.
Adaptability is crucial for teams to succeed in this industry, and the panelists agree that skill building and investing in talent is the best way for companies to stay adaptable. “We’ve asked a lot of our biologists to learn Python,” Madrid said. “All the biologists love data, and if you learn Python, you can get better at asking questions of the data that you have.”
Diversity generally begets adaptability. This is true for teams, ecosystems, and cultures alike. Uniting people from disparate disciplines around a common goal will yield solutions that couldn’t be conceived otherwise. For smaller companies to survive this season of tight funding, adaptability — through hiring teachable problem-solvers — and a focus on reducing product risk and timelines to clinic offer their best chance at helping patients.