The number of companies developing cell therapies has dramatically increased over the last few years, according to a 2022 market analysis report by Grand View Research.1 This is due, the report claims, to growing funding for clinical trials and the success of products such as Gilead Sciences’ Yescarta chimeric antigen receptor (CAR) T-cell therapy. According to a report by Mordor Intelligence,2 in June 2021, Gilead announced that Yescarta improved event-free survival by 60% in second-line relapsed or refractory large B-cell lymphoma patients.3 Other positive results using cell therapies for cancer, as well as rare diseases, are boosting the development of therapies in this field.
However, cell therapies, defined as those where a patient is injected with live cells to replace diseased or dysfunctional ones, remain challenging (and expensive) to manufacture, with a tendency toward bespoke processes. To meet the potential of cell therapies over the next few years, companies that currently have products in early development or in clinical trials will need to develop robust and scalable processes for commercial production.
Scalability issues were addressed by speakers at the cell therapy manufacturing strand of the Bioprocessing Summit Europe last March.4 Since then, several of these speakers have taken the opportunity to share their thoughts with GEN. In this article, GEN summarizes their insights, which range from optimizing vector design to controlling bioreactor conditions to fully integrating workflows.
Using AI for optimization
“When I think about cell therapy, what drives me to get up every morning is how we capture the value of powerful tools such as CRISPR,” says Claire Aldridge, PhD, chief strategy officer at Form Bio. “Part of that is bringing sophisticated software into the biotechnology space.”
In 2022, Form Bio was spun out of Colossal Biosciences, a company attempting to de-extinct the woolly mammoth and dodo.5 The new entity was created to independently commercialize Colossal’s computational bioscience tools. “Due to our relationships, we went after gene therapy first,” Aldridge says. “But the broad solution we’ve built is applicable to anything in molecular medicine or advanced therapeutics.”
Form Bio uses an artificial intelligence (AI) program based on DNABERT, a complex neural network that is something like a genomic version of ChatGPT, the renowned natural learning language model.6 Form Bio notes that DNABERT is akin to Google’s DeepMind AlphaFold2, which can accurately predict the three-dimensional structures of proteins.
For CAR T-cell therapies, which are manufactured by engineering patients’ T cells using viral vectors,7 Form Bio’s AI can help optimize the structure of the vector to replicate more efficiently. “Constructs that are coding for the CAR proteins or T-cell receptors have what we call hotspots,” she explains. “They create secondary or tertiary structures that cause the replication machinery to fall off in certain places.”
If these truncated constructs amount to only a partial gene, or don’t work at all, the viral vector will fail to replicate efficiently and give rise to a heavily contaminated product. It may be possible, however, to optimize the construct while retaining the amino acid sequence. Such a construct,
Aldridge says, will be more conducive to full reads. “You can create a construct that loves to reproduce itself,” she adds, “and then you can have a product of higher quality.”
Stirring up manufacturing
Form Bio’s AI-based solution is designed to help the company’s customers optimize their cell therapy manufacturing processes. Other solutions for improving product yields are available from other companies. For example, a solution from Eppendorf involves the use of process control technology in stir-tank bioreactors. This solution is helping Hannover Medical School overcome difficulties in generating enough stem cells for clinical trials.
“Stir tanks have been a well-characterized system for decades,” says Philipp Nold, PhD, infield application and stem cell specialist, Eppendorf. “But for stem cell processes, it’s been happening only in the last 10–20 years.” Eppendorf, he claims, was one of the first companies to offer stir-tank mini-bioreactor technology. He adds that this technology can facilitate the development of processes for the manufacture of stem cell therapy products. During early development, milliliter-scale bioreactors are often favored.
Nold points out that stem cell manufacturing is often done in flasks during research and development, with the cells attached to the surface of the flask. This has many downsides. For example, it relies on manual handling, there’s little control over pH, feeding, and other parameters, and scaling up is difficult. As such, Eppendorf offers smaller bioreactors, which use a stir tank that suspends the cells in culture and thereby allows more cells to be cultured in the same three-dimensional volume.
“When people have ideas for products, they often start in the most inexpensive way possible, in two dimensions,” Nold relates. “By moving from two-dimensional expansion of those stem cells to a three-dimensional environment, they’re able to scale up in terms of cell numbers. Production in a three-dimensional environment is reproducible and more standardized, and it can produce cells of higher quality.”
When stir-tank bioreactors were introduced, there were concerns that the devices would impart a stirring motion damaging to fragile stem cells. To address these concerns, users of stir-tank bioreactors began developing mitigation techniques. An example of successful mitigation comes from Hannover Medical School. There, as Nold points out, investigators used Eppendorf equipment, a CellXpert C170i incubator and a 60–250 mL DASbox Mini Bioreactor system, to successfully expand human pluripotent stem cells in suspension culture.
By constantly adding fresh media to the stir tank and simultaneously removing the same amount of used media, the scientists sustained a perfusion process that produced 47% more induced pluripotent stem cells than would have been produced in a fed-batch system. In addition, the researchers experimented with pH and oxygen control, optimizing feeding with glucose. Finally, the scientists looked more closely at the effects of stirring.
They discovered that if you stir too fast, the sheer is too great to permit the formation of suspended aggregates of growing cells. If you stir too slowly, the aggregates get too large, fall to the bottom, and die. By using process control, quality-by-design experiments, and an in silico model, the researchers managed to optimize their production parameters until they had 10 times the output compared to what they attained at the start of the process.
“That’s the power of knowing your process and parameters, and of using process control,” Nold declares. “You get a nice result with many more [stem] cells.”
“Over the past few years, cell therapy products have been approved, but manufacturing processes still lack the robustness and scalability to meet availability and affordability requirements,” explains Alexander Nikolay, PhD, strategic product manager, Miltenyi Biotec. He suggests that integrating instruments could help standardize and automate cell therapy manufacturing and make processes more robust.
He claims that part of the reason for high costs is that many manufacturers have daisy-chained instruments from other biomanufacturing fields. This, he says, doesn’t tend to work well, as it prevents full integration of processes and often requires extensive manual handling with associated human error.
Miltenyi Biotec’s approach has been to offer a single, automated cell processing platform. According to Nikolay, Miltenyi Biotec introduced the CliniMACS Prodigy platform in 2012/2013, originally for cellular products, such as autologous stem cells for cardiac repair, but Stefan Miltenyi, PhD, the company’s founder, realized that the platform could be used in the emerging field of CAR T cells. “Miltenyi is a unique character with a sense for therapeutic approaches, biological needs, and technological possibilities,” Nikolay observes.
The platform can perform multiple steps of cell processing in a single closed system, including cell preparation, selection, enrichment, and culturing. Moreover, when equipped with add-on modules, the platform can perform transfection and formulation. Nikolay argues that the platform is flexible enough to manufacture both autologous and allogeneic (that is, both patient-cell-based and donor-cell-based) cell therapies, and that it’s suitable for most commercial manufacturing of cell therapies. He concedes, however, that it wouldn’t work for cell quantities needed for the inoculation of vast cultivation volumes—for example, those accommodated by a 200 L bioreactor.
“We’ve got many ClimiMACS Prodigy customers making allogeneic cell therapies in combination with half-liter cultivation systems, and that’s sufficient at a clinical scale,” Nikolay asserts. “At a commercial scale, the ideal production volumes remain to be identified. You’ve got to ask: Do you really take the risk to manufacture therapies for hundreds of people in a single 200 L bioreactor? Is that a good idea? And do we have suitable instruments in the field? That remains to be seen.”
1. Grand View Research. Cell Therapy Market Size, Share & Trends Analysis Report, 2023–2030. March 2023.
2. Mordor Intelligence. Cell Therapy Market (2023–2028). January 2023.
3. Gilead. Kite Announces Yescarta CAR T-Cell Therapy Improved Event-Free Survival by 60% over Chemotherapy Plus Stem Cell Transplant in Second-Line Relapsed or Refractory Large B-cell Lymphoma [press release]. June 28, 2021.
4. Bioprocessing Summit Europe. March 14–16, 2023.
5. Colossal Biosciences. Species: De-Extinction for the Survival of our Planet [web page].
6. Qin C, Zhang A, Zhan Z, et al. Is ChatGPT a General-Purpose Natural Language Processing Task Solver? arXiv. February 8, 2023.
7. Levine BL, Miskin J, Wonnacott K, Keir C. Global Manufacturing of CAR T Cell Therapy. Mol. Ther. Methods Clin. Dev. 2017; 4: 92–101.