The model should achieve the same level of separation as at large scale. Aim for a similar elution profile. Deviations that
cannot be avoided should be analyzed carefully with regard to influence on the results. Step yields and the quality attributes
of the product (host cell proteins, host cell contaminants, product related variants, and aggregates) should be comparable
to full-scale. Height-equivalent theoretical plates (HETP) and asymmetry values should also be within the acceptable range
of that for the large-scale columns (although not necessarily identical).
Take data and evaluate the importance of other parameters that indicate column performance, including pool volume, peak retention
time, and peak shape. System dead volume and unavoidable scale differences such as header effects and pressure vs. flow characteristics
may result in anomalies with regard to these parameters. You must be able to explain these scale-dependent differences.
Figure 2. Example of a Successful Scale-up but an Unsuccessful Scale-down for a Chromatography Step.4 Feedstock is an aqueous solution. AE-HPLC = Anion exchange high-pressure liquid chromatography
If significant differences exist between the two scales, make a careful determination of the impact of these differences and
if necessary, redesign the small-scale model. In any case, having a qualified scale-down model prior to performing small-scale
cycling studies is necessary, and using a scale-down model that is flawed will lead to unreliable lifespan study results and
a waste of time and resources.
Table 1 presents an example of a successful scale-down of a cation-exchange chromatography step and its use for measuring
column lifespan.3 This column is performing purification in production of a monoclonal antibody. The input values are roughly 3,000 ppm Chinese
Hamster Ovary Proteins (CHOP), 3 ppm DNA, and <7.8 ppm Protein A. At both scales the step yields are comparable and the key
functions of the column step, which are reducing CHOP and DNA and clearing Protein A, are equivalent at both scales.
Figure 1 presents another comparison of the chromatographic profiles at small and large scale. The profile (absorbance) and
the resolution between the product and the different impurities are similar across scales. Impurity concentrations were higher
in the feed material for the small-scale run due to under-performance of the preceding step.
Figure 3. Comparison of Performance of a Depth Filtration Step at Small-scale and Large-scale.5
Figure 2 illustrates an example of a successful scale-up from lab-scale to pilot-scale facility as the column performance
improved both in terms of step yield and pool purity.4 However, if one were to evaluate the performance in the lab with respect to qualifying the scale-down model, differences
in step yield and pool purity would necessitate modification of the scale-down model.
The unit operation of filtration is perhaps the most common unit operation that is used in bioprocesses. It is used as depth
filtration for separation of particulates (cell debris) from the process stream; as normal flow filtration (NFF) for separation
of impurities (viruses, host cell proteins, DNA) from the process stream; as ultrafiltration (UF) for concentration of the
feed stream; and as diafiltration (DF) for buffer exchange. Here is a review of general guidelines to consider while scaling
down a filtration step. Many of these guidelines apply to all formats, but some may be specific for a certain format. It is
up to the reader's discretion to apply these appropriately.