Automation & Digitisation
Cell Therapy
Raw Materials
Regulatory & Standards
Supply Chain

Achieving Automation in Cell Therapy, Selection, Sampling and Washing

19 October 2021
The latest automation working group centred on modularity and flexibility in closed systems recently focussed on how to achieve greater automation in the cell therapy development and manufacturing process.

With the group’s last meeting focusing on tech innovation and modularity, the latest meeting brought together a discussion on how to achieve greater automation in cell therapy, selection, sampling, and washing. Progress was discussed and the group brought experience from different fields together to identify gaps and opportunities.

“We must focus on automation at scale, and we should make distinctions between onset connectivity and sampling in-process; they’re different views on the future of automation.”


R&D is active in the robotics space, with a current aim of sub-contracting kitting to progress towards automated warehouses. We’re looking at sanitised boxes and aim to obtain specifications from the FDA on material airlocks. With that, materials can enter labs and manufacturing areas without human involvement. From there, we can reach a point where labs run 24/7, with skeleton staffs at night monitoring automated work that has been ‘setup and left’ by day-staff.

Robots can aid us with labour-intensive tasks like sampling bioreactors. They also can help us automate feeding materials into quality control (QC). Ultimately, we’re aiming for end-to-end automation of an entire facility. In reality, this will mean starting from scratch; we won’t be able to retrofit facilities as automated design principles are completely different.


It’s important to have physical connectivity. Keeping this closed is ideal, and we need better connectivity with data. Right now, digitalisation of manufacturing data is very limited. If we can improve the breadth of data we can feed into and connect with, we should.

We must focus on automation at scale, and we should make distinctions between onset connectivity and sampling in-process; they’re different views on the future of automation.

Cell Therapy

An issue here is that adherent cells, particularly IPSCs, don’t like being split. This is problematic because cell numbers are limited. We don’t usually have enough for automated production and there are manual steps to take before production can begin.

We must be mindful that adding steps means new elements and work, which adds vulnerability to a process. The less we have, the better.

Concentration and Washing

Is it enough to only be matching key process properties coming out or are companies, particularly therapy companies, concerned with what is happening to cells? It boils down to CQAs and how they map against what is being done. If the modality of processing doesn’t influence phenomena in or out of the cell, this will impact the CQAs.

Cell Selection

What are we seeing in terms of selection and enrichment? For selection, how do we make this more amenable to different systems, and can we improve connectivity across more than just one workflow? How do we ensure flexibility that can accommodate new tech as it comes in?

Right now, it seems like we are moving towards a defined population of CD4 and CD8. It’s a catch-22 situation for selection and definition; it’s driven by knowing the efficacy of a product. Not many have made it to clinic.

Ultimately, we need flexibility from tech providers while being able to drive down to the specificity we need, and we need to be able to handle diverse cell input ranges. The onus is on therapy providers to concentrate requirements, which may be why we see groups making do with existing systems instead of driving innovation.

Labs and Samples

We need to think of factories as a whole, not just a process. For Grade-C labs, we’re making progress with automation and have a prototype small lab that can handle frozen vials. These are entered, heated if necessary, and sanitised to produce an output suitable for processing. Miniaturising isolators would help, particularly as some autologous products are packed in vials that have sanitation concerns.

When taking biopsies and samples directly from patients, we find that tech solutions for closing processes early on are completely lacking. This may be because it is driven by the clinical user and the industry aligns with their preferences. We need something that can receive samples and be closed at that point.

Cell Markers

How many markers do we need to select for and what does that mean for a process? For connection, one extra step adds more than that to the process. For designing end-to-end, every step makes connections that can go wrong. It’s hard for tech providers here too; upfront investment is high. As a goal, we need more optionality in how to progress.

It’s also difficult for therapeutic companies when working with regulatory bodies. It would help to have an assurance that processes aren’t changing cell phenotypes; we need proof that this isn’t changed after a wash. Handling the regulatory aspect is important as we develop processes for this.

Lastly, we need to consider adapting and maximising processes to a larger scale. New tech is rarely adapted to high cell numbers; once comparability assessment is done, it’s often poorly adapted to the scale it will eventually end up at.

Allogeneic Gaps

Right now, we’re challenged by washing large volumes of cells without impacting their health. A workaround is performing depletion or selection at the end of a process to achieve desired clinical outcomes. Selecting the most potent cells at the end of a process would be ideal.

We’re also seeing an appetite to move away from beads and towards upfront drug-based selection. This is pre-clinical at the moment. A promising thought is moving towards drug-based manipulation as a new modality of enrichment and separation.

When exploring platforms, therapeutic providers will focus mostly on supply chain capability. Driving characteristics focus not on the mechanism of action, but the aspects around it. We’re limited by issues on consumables and the time to process, and scalability and redundancy are ongoing concerns. As we learn how processes respond and come to know products better, it’ll be easier to identify gaps. We must stay mindful of the regulatory framework, as any findings must be defended.

It’s also key for us to focus on automation at scale, and we should make distinctions between onset connectivity and sampling in-process; they’re different views on the future of automation.

This session was moderated by:

This session was hosted in partnership with ScaleReady on Tuesday 14th September.

Discover more with our Automation Specialist Group in the network.
Join today.