Key Challenges/Gaps/Mitigations
Despite the overall robustness and effectiveness of virus retentive
filters for the retention of large viruses (e.g. retroviruses) and small
viruses (e.g. parvoviruses), the retention of small viruses can be
highly variable and not always effective. While the newer generation
small virus retentive filters appear to have less variability
(Lute 2015) (Figure 2) there are still
concerns with the occasional less effective clearance values. To address
these concerns, the viral validation study is typically performed under
challenging, or worst-case, conditions. The science seems to indicate
that the target operating pressure, low pressure, a flow pause, and a
buffer chase should be included during each scale down viral validation
run. A pause typically occurs between the load and flush phase and is
different from running at low pressure because liquid only flows through
the filter during the low-pressure phase. There have been attempts to
perfectly match the pressure ramp up and ramp down dynamics during viral
filtration (Roush and Ma 2016), however
it would be impossible to do a comprehensive experiment covering all
ranges and combinations of pressure conditions and pauses that may occur
during routine manufacturing. High pressure is typically not considered
worst case as the limits are determined by filter structural robustness.
A variety of ways exist to set end point limit for a viral filter step.
For some early model filters, flux decay due to pore plugging
(Bolton, Cabatingan et al. 2005) and total
virus loaded per filter area was reported to decrease virus retention
(Lute, Bailey et al. 2007). Similarly, it
has been reported that mass of protein per filter area
(Soluk, Price et al. 2008,
Chen and Chen 2015,
Roush and Ma 2016,
Kreil and Roush 2018) or alternatively,
volume of protein solution per filter
(Chen and Chen 2015,
Roush and Ma 2016,
Kreil and Roush 2018) may be a worst case
condition. For instance, operating a low load concentration will provide
the highest volumetric (liter/m2) throughput and viral
loading but the lowest mass (gram/m2) throughput.
Conversely, operating using a high load concentration will provide the
lowest liter/m2 throughput and viral loading but the
highest gram/m2 throughput, given the tendency for
increased filter fouling at higher load concentrations, particularly
when using scaled down filter membranes. While many ways fouling may
occur and how end point limits are set, in-process controls are best
described in terms of liters per square meter of filter area.
Recent and emerging
applications of existing technology