Model-assisted, Self-optimising Unit Operations for Accelerated Bioprocess Optimisation

Biopharmaceutical innovation saves lives, and it saves cost. Advances in the manufacturing of biopharmaceuticals reduce costs for healthcare, ensure drug quality, and generally make drugs cheaper and more widely available for the broader population. In this project, Lonza Biologics Ltd (Lonza) and UCL Biochemical Engineering (UCL-BE) are partnering to create a novel approach to improve biomanufacturing of biopharmaceuticals to address these needs. Lonza has been at the forefront of the development of Biologics for 40 years pioneering production in mammalian cell culture. Lonza’s work as a contract development and manufacturing organisation (CDMO) has brought hundreds of products to the clinic and supported the UK manufacturing sector. Better medicine means curing diseases instead of treating them, it means fewer side effects because of higher specificity, but it also means more complex molecules to be manufactured. Indeed, Lonza is increasingly being asked to manufacture more complex biologics (a diverse category of large complex biological molecules, such as monoclonal antibodies) which then require more specialised and advanced manufacturing processes. The development of these new processes is costly and slows the products progress to the clinic. UCL-BE and Lonza are proposing to address this challenge and unite their complementary strengths to develop a fundamentally new approach which will aid in reducing the time-to-market or time-to-clinic of these more advanced medicines. Thus, ultimately, this will aid to reduce cost and make advanced medicines more widely available. The new approach will be achieved using cutting-edge technologies: microfluidics (i.e. manipulation of tiny volumes in channels), modern analytical methods (known as Process Analytical Technologies, PAT) and machine learning (or Artificial Intelligence). By combining these three technologies in a smart way, we envisage new platforms for the development and optimisation of biomanufacturing processes. In these new platforms, small volumes of proteins and reagents will be rapidly mixed together using microfluidics, the interaction of the proteins with the reagents (i.e. buffer and medium) analysed with PAT, and the data will be processed with machine learning methods. More specifically: Continuous microfluidic buffer/medium preparation (‘recipe’) will enable the vast process space to be rapidly explored. Microfluidic models of unit operations will provide a continuous representative product stream from which the impact of the changes to the process conditions can be observed. Integration of spectroscopic protein measurement to the microfluidic devices will enable quantification of the impact of the changes to the recipe. The resulting data stream from the microfluidic device will be analysed using machine learning. The ML model will predict the most promising direction within the recipe space to explore next and update the recipe to the microfluidic mixer. Therefore, ultimately, we will have self-optimising platform units, each resolving a manufacturing process problem autonomously and thus a much-accelerated approach to finding the best possible manufacturing process conditions. This project will create three prototypes which will specifically tackle pressing issues of biopharmaceutical processes: protein aggregation (affects the safety of medicines), protein refolding (affects their therapeutic activity), and optimal culture medium compositions (to create large quantities cost-effectively). UCL-BE will take the leadership in the development of the technologies, Lonza will take leadership in their application to the three stated examples. Together, the project will deliver proof-of-concept on three related microfluidic prototypes ready to be adopted by Lonza in their operations and to expedite the progression of new biologics to the clinic.