Extreme Scale-Down Approaches for Rapid Chromatography Column Design and Scale-Up During Bioprocess Development - Measurement, Monitoring, Modelling, and Control of Bioprocesses - page 116

calibrated from runs with three cation exchange matrices in 1-mL pre-packed col-
umns to separate bovine serum albumin (BSA) and lactoferrin. The principles of the
Hutchinson et al. [ 11 ] study described above were also taken into account in the
prediction of scale-up profiles. The model was used to predict the laboratory- and
manufacturing-scale performance at different liquid velocities (at 50-100 % larger
than the 1-mL column), different bed heights (from 2.5 cm at 1-mL scale to 20-cm-
tall columns with volumes ranging between 40 mL and 160 L) and different sample
volumes (from 1 CV in the 1-mL column to 0.4 CV in larger columns). When the
model was used to simulate the scale-up chromatograms, there was tailing in the
experimental peaks which was not observed in the modelled peak, but in all cases the
location of the peaks (i.e. their retention times) were predicted well.
The decision as to whether to use mechanistic approaches or not depends upon
factors such as the understanding gained from using similar feedstocks and the
complexity of the separation challenge. Furthermore, the increased instance of
protein heterogeneity in 'real' bioprocess feedstocks increases the separation
difficulty and makes it harder to predict the scale-up column performance. Thus,
the empirical equations used to quantify axial dispersion and the film mass transfer
coefficient in the Gerontas et al. study for BSA and lactoferrin may be insufficient
to describe the adsorptive behaviour of more realistic feeds involving the sepa-
ration of a product from highly crude materials at high protein loading levels.
Furthermore, a great deal of computing power and user experience is needed to
achieve successful CFD-based scale-up predictions, whereas purely experimental
approaches offer greater simplicity and familiarity to bench-scale scientists.
3.4 Other Issues
Although millilitre-scale columns can provide highly valuable process development
data, issues of parallelisation and automation remain. A typical liquid pumping
system with a single set of ultraviolet (UV)/conductivity detectors can be used to
control only one column at a time. Although some automation is possible, there are
still many manual steps involved, such as resin packing (where pre-packed beds of
the required volume are unavailable). Ideally, the principle of miniaturisation could
also be taken further by reducing the resin volume. Automating multiple miniature
devices is a task more characteristic of microlitre scale-down, as described below.
4 Microlitre Scale-Down
4.1 Overview
Microlitre-scale chromatography takes the form of three different types of device:
batch incubation , pipette tips and miniaturised columns, and all three formats can
be used with many different off-the-shelf and custom resins. The formats differ in