(660e) Systematic Development of a Scale-Down Model for a 25,000 L Fed-Batch CHO Culture Process

We
will present a typical process of developing and fine-tuning a 5 L bioreactor model
for a 25,000 L scale CHO fed-batch process. The process was first scaled-down
according to engineering principles (same power input per volume and vvm).  We were able to reproduce most of the culture
parameters in 5 L bioreactors such as specific growth rate, peak viable cell
density, titer and specific productivity. However differences in certain
culture indicators (such as viability and lactate metabolic shift) and critical
quality attributes (such as glycoform profile) were observed
between 5 L and 25,000 L bioreactors.

We
carried out a systematic analysis and step-wise experimentations to identify
key cell culture parameters (such as CO₂ stripping and base
consumption) that contributed to the observed differences. Intrinsically, small
scale bioreactors are superior at mixing and O₂/CO₂
transfer and base required for pH control is often a few fold
less than that in large scale bioreactors. In experiments, cell inoculum and
media were obtained from manufacturing operations to avoid culture source and
materials variation. Controlled experiments showed that the glycoform
profile differences at 25,000 L scale was mainly due to decrease in cell
viability. We were able to fine tune the scale-down model to match the large
scale culture process closely by artificially re-creating the 25,000 L
osmolality trend in 5 L bioreactors using 1 N NaOH. Protein quality attributes
of specific interest, glycoform profiles, were produced
at the same levels as seen in 25,000 L bioreactors after the culture processes
were aligned. We will discuss the impact of osmolality and pCO₂ on
cell viability and protein quality attributes. Relationships between pCO_2,
lactate, osmolality and base consumption will be presented. Applications
of the improved scale-down model in cell culture performance evaluations (such
as the impact of cell population doubling length) will also be discussed.