Driving Down COGS: Pursuing Automation In A Diverse Advanced Therapeutic Development Pipeline

By Life Science Connect Editorial Staff

Automation for more traditional therapeutic modalities has, over the years, become incredibly standardized and widespread. But for cell and gene therapies, the diversity of the assets in the modern development pipeline has meant that most automation occurs piecemeal, much of it tailored to a specific program.

Yet automation is likely to play an increasingly important role in reducing costs and enabling greater standardization for cell therapy manufacturing. As more organizations move toward automating highly manual and repetitive tasks, cell therapy workflows are bound to see greater consistency and reproducibility, particularly as a program scales.

In a recent installment of Cell & Gene Live, experts from pioneering cell therapy companies sat down with Cell & Gene’s chief editor, Erin Harris, to explore how the industry can improve scalability and cost through standardization and automation. Speakers for the event included:

• Craig Beasley, chief technical officer, BlueRock Therapeutics
• Narinder Singh, chief technical officer, Arcellx

Different therapeutic focus areas often require different forms and levels of automation. A number of factors, such as whether a therapy must be produced autologously or can transition to allogeneic production, can hugely impact the degree of automation for a process. Done right, automation can improve cell therapy applications like CAR-T at every step, from collection to infusion.

Pipeline Progress And Automation

The degree of automation achievable for a workflow, as well as what that automation looks like, can vary widely depending on the modality or individual asset in question. Certain products cannot be transitioned, or transitioned easily, to suspension processes, for example, and the bulk of the automation for these processes therefore hinges on media handling and distribution. “For a more advanced operation, you can do more of what we classically think of as automating, things like measuring cell densities at variable times so that we meet the cell health requirements by feeding them at the right time and harvesting them after the right duration,” Beasley said. “There are very big differences for different therapeutic areas or cell types.”

Other applications designed to enable a more “plug-and-play” approach may be able to work toward a more standardized, robust process from the get-go, so that a universal cell product can be adapted to unique proteins while having the necessary automation and efficiencies to accelerate a molecule to the clinic and the market while improving patient accessibility. “At Arecellx, we have a very different approach to the pipeline,” Singh explained. “We are trying to build a universal cell product that can work with multiple different proteins – the ARC-Sparx concept.” By working to standardize and limit variability for its platform, Singh said Arcellx hopes to reach a point where the consistency and robustness of its technology can result in a universal CAR-T program that accelerates an asset’s journey to the clinic.  

Auto, Allo, And AI

Not every aspect of cell therapy manufacturing today is ready for automation. Organizations should never seek to automate a bad process – that is, processing steps that are not fully understood and for which all significant variables have been outlined are not good candidates for automation. Other, very low volume steps that may have a tight window of process design space are unlikely to benefit from automation until the technologies supporting them improve.

Ultimately, the goal of automating a process step is to simplify it and make it more scalable. “Automation comes from having a good process in place; a good process comes from having a good fundamental construct in place,” Singh said. This evaluation can be very different between autologous and allogeneic applications: automating visual inspection, for example, may streamline an allogeneic workflow, but for the small volumes and processing surrounding autologous applications, this type of automation is more likely to introduce unnecessary complexity.

As more organizations move toward automating highly manual and repetitive tasks, cell therapy workflows are bound to see greater consistency and reproducibility, particularly as a program scales. This trend exists alongside increased interest in artificial intelligence (AI) and machine learning (ML) techniques aimed at bolstering analytics and enabling more comprehensive, targeted automation. “From my point of view, AI is going to be most useful, firstly, in developing robust processes and giving key insights into what makes a robust, scalable, predictable process,” Beasley said. “From there, how does AI increase the ability to select good analytical techniques?” Then, he said, you can automate a process better, owing to that more nuanced process understanding. “Once you have good process understanding and can measure things that help you manipulate the process, automation can build on those things.”