Filtration is a key step in many stages of downstream processing, from product recovery to removal of viral contamination. The last few years have seen many developments in filtration science, including the increasing adoption of single-use filters. Sartorius Stedim Biotech recently hosted its sixth “European Downstream Technology Forum” at its headquarters in Goettingen, Germany, where experts discussed the latest developments and challenges in downstream processing.
Alexander Caliebe, Ph.D., head of production at Richter-Helm-Biologics, described how filtration techniques, such as crossflow filtration (CFF), play a role in improving purification. CFF is used in harvesting crude supernatant, protein dissolution, buffer exchange after refolding or chromatography, and removal of aggregates. There are big differences in these product streams in both physicochemical properties and purity of the product.
In one case involving purification of a chimeric protein made from a Fab fragment with an effector molecule, Richter-Helm-Biologics took on the production process after Phase I, intending to scale up to Phase II. It faced problems with high viscosity because of high cell density and needed to add antifoam, which led to low flux and fouling of membrane filters. The challenge of scaling up and improving this purification, without changing the techniques used, was achieved by preclarification with a depth filter.
“We then went on to make more changes,” Dr. Caliebe explained. “Viscosity and high load were still the main challenges so we went for in-line dilution. We were able to eliminate two CFF steps by introducing expanded bed absorption chromatography with no loss of product or quality. We also reduced the process time by 50 percent. But this was an unusual case. Cross flow filtration is still a powerful technique.”
Another study reviewed at the meeting involved a recombinant human cytokine that is produced as an inclusion body in E.coli. The original process involved depth filtration and 20-fold concentration with CFF, buffer exchange with CFF, and then dead-end filtration. Unfortunately, high salt concentration led to low-flux and high-buffer consumption. Rinsing the filter and better temperature control led to lower viscosity and better control of the process.
As a result, the purification became robust and efficient with 50% reduction in buffer consumption, reduction of membrane area by 20%, and no loss of product. “Here a small adjustment had a big effect.”
Jodee Lewis, technology engineer in Pfizer’s biomanufacturing services group, presented work on screening tangential flow filtration (TFF) filters. Ultrafiltration/diafiltration (UF/DF) is ubiquitous to purification of polysaccharide-protein conjugate vaccine intermediates such as Prev(e)nar® (a pneumococcal saccharide conjugate vaccine against Streptococcus pneumoniae). This product is made by conjugating pure polysaccharide and pure carrier and purifying the vaccine. Not surprisingly, it involves a complex manufacturing process with many purification processes.
The UF/DF operation is seen as a source of risk because it is so sensitive to change, therefore, experimental evaluation is required to understand and maybe mitigate, the impact of changes in ultrafilter characteristics as well as to qualify alternate or replacement ultrafilters prior to validation. The application of quality-by-design highlights the impact of small-scale work on regulatory communications throughout a product’s life cycle, Lewis noted.
Lewis discussed an ultrafilter that had to be replaced because the supplier was planning to discontinue the product. “Our strategy was to mitigate risk by looking at a small-scale evaluation and then scaling up.” Lewis’ group devised a small-scale model and then screened a narrow pool of 13 potential filters, thereby generating a large matrix of data.
One or two filters emerged from this exercise for further evaluation. The team then carried out design of experiments with risk assessment, which brought the choice down to one filter. The next stage was to see how this performed in scale up.
Lewis noted some roadblocks and recommendations. For example, analytical testing should be completed during evaluation so that there are no surprises in full scale. Water runs should be performed, although that is not always representative because of the viscosity of product streams. “Tech transfer of the process and knowledge sets a clear path for filter evaluation for submission.”