The impact

Labour-intensive processes
The cell and gene therapy space is one of the most exciting and transformative areas of medicine. But the highly sophisticated manufacturing process is characterised by many manual, labour-intensive procedures. It goes without saying that such therapy – costing around $500,000 per single dose – must be sterile to be safe. But standard testing in expensive clean rooms and labs with skilled staff can take up to two weeks. After conferring with a number of manufacturers, we established the unmet need for an on-line, continuous contamination detection system.

Real-time 24/7 feedback
The resulting, fully automated PureSentry system uses the speed and sensitivity of PROPHESEE event-based vision to deliver intelligent contamination monitoring with real-time 24/7 feedback. The technology is cost-effective enough for every batch to be continuously monitored, rather than sampled at a small number of timepoints. As well as providing peace of mind for doctors and patients alike, this utility could broaden manufacturers’ insight into how and why contamination occurs.
The challenge

Multidisciplinary inhouse team
We assembled a multidisciplinary inhouse team with expertise across cell biology, fluidics, optics, software and detection algorithms. The driving imperative was to harness those skills to create a viable, integrated system – all the while keeping costs to a minimum. Label-free, non-destructive testing – to get more doses to patients – was key. The lack of scalability inherent in manual testing also had to be addressed. Solutions were provided by the closed fluidic loop system and the PROPHESEE event-based camera allowing for the simultaneous, sample automation and batch protection. This was particularly suited to continuous and rapid detection as the camera output has minimal redundant data. With just the important data, we were able to train AI algorithms swiftly.

Event-based imaging
Ours was the first application of the PROPHESEE camera for microbe and human cell monitoring. Its event-based imaging is neuromorphic – like human eyes, it responds to scene changes and has a high dynamic range. This enables the detection of living cells without the need for stains or dyes. Low contrast cells are notoriously difficult to see under a microscope, but the event-based vision is well suited to identifying them as they flow within a microfluidic chip. The data is key to delivering real-time feedback on a low-cost system, with each frame being analysed in less than 20 milliseconds.

Bespoke cell therapy simulant
In addition to hardware and software, it was critical to develop a bespoke but credible cell therapy simulant. The team liaised with cell therapy research organisations to ensure the development reflected industry standard cell therapy growth medium and culture vessels. We also confirmed that our model microbes matched the concerns of cell and gene therapy manufacturers. Our recently upgraded mammalian cell culture facilities were put to work, and we used a human T cell line, as well as a live bacterial culture, to feed into our imaging experiments. Having a cell culture facility onsite – separate from a dedicated microbiology lab – allows the team to iterate more quickly as they can control parameters like cell concentration.

Addressing user experience
With the key technological components in place, the team embraced a further key challenge: how to address user experience. More specifically, how to deliver a novel camera technology and automated sampling system to a user in a biomanufacturing environment. The solution came in the shape of a lab prototype system with a touchscreen graphical user interface. The peristaltic pump is quick to load and connect without compromising the sample, while easy-to-use controls were created for configuring the camera and accessing live plots for the analysed data.

Ready to accelerate development
We believe that cell and gene therapy represents an exciting future. Cures are now available for a number of cancers, including leukaemia, based on reprogramming the cells of a patient. As researchers continue to break new ground, a string of further cures is increasingly likely. Accelerating development while reducing cost is vital – and PureSentry is ready to play a part. Not only is the technology scalable, but it can be retrofitted to existing manufacturing platforms and workflows.
The next step for the Cambridge Consultants team is to forge a collaboration, perhaps with a cell therapy manufacturer, an equipment vendor or scientific instrument specialist whose customers need faster more rigorous contamination detection in order to grow their business and succeed. Our ambition is to help develop a commercial system for market launch.
Tales of the team
"The problem of rapid microbial testing is well known. We believe that every batch should be continuously monitored as this will enable faster release of life-saving therapies, getting precision medicine to more patients in less time."
Matthew Jones, Project manager

“Cell therapy developers face huge production risks, with up to 30% of batches failing to meet specifications. This technology has the potential to save significant costs for companies in this sector.”
James Hallinan, Business developer

“Without the enthusiasm and creativity of the team it would have been a slow road to success. Developing PureSentry has been fascinating and challenging – and the outcome has potential to change the way we analyse liquid biologics.”
Josh Gibson, Technical lead

“It was an amazing and rewarding experience, because we were able to create a system to analyse cell samples in real time. I relished learning the new skills and cell culture techniques which helped the team achieve their goal.”
Rae Freestone, Lead laboratory technician

“The nature of the data meant we had to rethink many conventional AI techniques. We developed a model to reduce the typical bottlenecks of data collection and labelling while learning the complex, long-term characteristics of contaminated and sterile samples.”
George Barnett, Machine learning engineer
