Computational fluid dynamics based digital twins of fixed bed bioreactors validate scaling principles for recombinant adeno‐associated virus gene therapy manufacturing

Gene therapy using recombinant adeno‐associated virus (rAAV) as delivery vehicles has garnered much interest in recent years. There are still significant gaps in our fundamental understanding of the manufacturing processes to deliver sufficient products. Manufacturing efforts of rAAV using HEK293 cells have commonly relied on fixed bed falling film bioreactors like the iCELLis®. We used computational fluid dynamics (CFD) to validate the operating conditions required for a predictive iCELLis® 500 scale‐down model. The small‐scale and at‐scale systems have different flow paths causing validation of the corresponding agitation rates required to achieve the same linear flow through the fixed bed across scales to be non‐trivial. Therefore, we used CFD to predict the theoretical scaling relationship. In addition, CFD could predict kLa differences between the two systems and the operating conditions required to match kLa between scales. We also confirmed that the location of DO control must be the same in both systems to achieve proper scaling. Experimental runs confirming the validity of the novel scale‐down model showed that based on the modifications to the iCELLis® Nano system, we achieved similar DO, key metabolite, pH, and GC titer trends in both systems. A novel scale‐down model for fixed bed bioreactor adeno‐associated virus (AAV)‐based gene therapy production was developed utilizing computational fluid dynamics‐based digital twins to predict scaling relationships. Hill and colleagues utilized these predictions to modify a commercially available scale‐down reactor to better match oxygen transfer, probe placement, agitation settings, and fixed bed gradients resulting in yields, metabolite trends, and online trends that better match the manufacturing scale.