Digital Twins in CAR-T Manufacturing

The ability to engineer a patient’s own immune cells outside of their body and reinfuse them to fight cancer is nothing short of a lifesaving revolution. This is what CAR-T cell therapies have delivered to over 30,000 patients since 2017. Complex cell and gene therapies are here to stay. In the next five years, an estimated two million more patients will be eligible for these potentially curative CAR-T treatments alone.

Sadly, the scale required to achieve global demand is not attainable under current manufacturing models. These processes are a blend of people and automation islands requiring specialized skillsets that are intricate, expensive, and time consuming to scale. Every touchpoint in the manufacturing process introduces risks, such as contamination, inconsistency, or outright failure. These issues drive up costs and prolong timelines. 

I think we can do something about these challenges. If we are to make CAR-T therapies truly accessible, we must fundamentally rethink how we approach their manufacture. Enter the digital twin.

Imagine a computerized system that feeds from all information pertaining to the making of these complex therapies, the composition of the patient’s blood, the ingredients of the medium that will expand the therapeutic cells, the process parameters that control the physics, chemistry, biochemistry, and biology of the entire manufacturing process. Such a system, designed to be an accurate in silico replica of the physical process, could break through the bottlenecks of existing approaches.

How? By using a technology platform that incorporates real-time monitoring. We can throw state-of-the-art process analytical technology at the digital twin challenge to capture, in real-time, all those important parameters. For the data that cannot be captured instantaneously, we can deploy streamlined processes to generate at-line and off-line analytics, a hybrid of devices and digital tools that rapidly generates this vast amount of data and transforms them into actionable formats. Like a photograph capturing a scene in an instant, we can capture all that we can during the manufacture of a CAR-T product; a snapshot for multiple products and multiple scenarios of the process, multiple times, quickly.

Why? The digital twin feeds on this data. Its purpose is to use all relevant associations in the dataset, all correlations, all hidden rules, to drive instructions to the manufacturing platform and make it produce the best quality product as quickly and as efficiently as possible. However (and this is where we are with CAR-T manufacture), those associations need to be worked out first. By collaborating with UCL’s Stephen Goldrick, the Cell and Gene Therapy Catapult hopes to develop advanced process models for CAR-T manufacturing. These modelling endeavors will help identify what is important to control, and a digital platform to simulate a wealth of scenarios, giving us options to identify which will optimize the manufacturing process.

Smooth sailing from here? Not quite. How do we close the loop? How can the digital twin bring virtual instructions to life in the real world? Dynamic process control requires a seamless blend of cutting-edge integrated technologies. We must track the biological fingerprint of the patient’s cells as critical quality attributes. A solid good manufacturing process (GMP) is essential to ensure that all data are adequate and directed to the digital twin in accordance with regulatory standards, using the right channels. This is critical because most CAR-T manufacture issues stem from poor cell quality and characterization, as well as process variability and limited control strategies. Then, the process must unfold under the guidance of the digital twin, with commands and instructions flowing smoothly through communication pipelines that link the twin to all the machines. Luckily, there exist a few standard communication protocols and data formats that enable this, but few technologies have yet been connected at scale – especially in a GMP setting – and even fewer involve cutting-edge sensors and a digital brain running the entire operation.  

Developing a digital twin for CAR-T manufacturing is a marathon, not a sprint. I envision a future where digital twins optimize protocols through modelling and simulation, enabling faster decision making and reducing the need for extensive manual monitoring. The automation of manufacturing and analytical technologies, combined with the insights offered by digital twins, will allow the production of CAR-T therapies at a lower cost and greater scale. Patients will benefit from quicker access to more affordable therapies.

A thriving industry delivering life-changing therapies to the world through powerful collaboration, sharing knowledge, and making breakthroughs. Right now, the CGT Catapult is expanding these capabilities through partnerships, innovative deployment of next-generation equipment and facilities, and access to unrivalled ATMP datasets. The goal is to inspire stakeholders from all fields to push technologies forward, establish industry standards, and drive collaborative efforts with the aim of digitalizing the sector. Greater patient access to these lifesaving therapies cannot be achieved at scale without digitalization, and digital twins are central to making that vision a reality.