Roche has been active in scaledown technology for some time. Marco Jenzsch, Ph.D., director of fermentation in pharma biotech production at Roche Diagnostics, recently presented a talk on a 2 L scaledown system that accurately simulates a 12,000 L cell culture production reactor. According to Dr. Jenzsch, such factors as cell culture growth, metabolism, and product yield/quality were accurately reproduced in the smaller system.
Roche performs scaledown exercises on every new process for which process development and process characterization cannot be performed in final production scale due to cost, timelines, or asset availability. Typical scenarios include process validation studies for risk mitigation during tech transfers and scale-up of development-stage and mature processes); troubleshooting for performance use tests of critical raw materials (such as protein hydrolysates and complex chemically defined media components); and modeling for continuous improvements throughout the product lifecycle.
“Scaledown is essential for process development and the approval of NBEs, since the design space of a production process cannot be assessed with runs at the commercial scale from an economic and strategic perspective,” Dr. Jenzsch explains.
Some process parameters are easier to model in scaledown mode than others, which makes understanding the true rate-limiting events in the large bioreactor critical. For a particular process, this might be mass transfer of oxygen or carbon dioxide, mixing, or shear forces. Key to establishing a successful scale down model is mimicking the operating windows of the cells for all of these regimes in the small scale bioreactor.
Due to the nonlinear behavior of hydrodynamics, some unit operations in large bioreactors are difficult to maintain constant at 2 L bench scale. Constant volumetric power input (Watts/m3 or Watts/kg) is used as the first scaledown. Calculating agitation speed using constant volumetric power input implies sufficient mixing, comparable oxygen mass transfer, and similar shear conditions at small scale compared to the at-scale process. “Liquid mixing is usually not an issue in small-scale bioreactors, but where mixing is a limiting step at manufacturing scale, it may need to be reflected in the scaledown model,” Dr. Jenzsch continues.
The most difficult parameter to match is the profile of dissolved CO2 across scales. In contrast to oxygen mass transfer, CO2 stripping depends only on gas throughput in large scale, whereas at small scale power input contributes to CO2 mass transfer as well.
“A fundamental prerequisite for [achieving a successful scaledown model] is characterization of the installed equipment and knowledge of the equipments’ influence on unit operations and cell culture performance,” Dr. Jenzsch notes.
Matching all engineering parameters matters little if cells don’t behave similarly at small scale to large scale, or if product quality is significantly different. Going a step further, quality can differ even when cell growth and metabolism are similar.
Demonstrating product equivalence therefore becomes critical when scaledown data is used in regulatory filings. According to Dr. Jenzsch, recent filings suggest that regulators are taking a close look at scaledown models.