Limitations of integrity tests
Filter membranes are often thought of as multiple cylindrical capillaries, in which bacterial retention is governed solely
by size exclusion of incident bacteria that are larger than the largest pores in the membrane. Under such a model, bubble
point-type tests can indicate the presence of excessively large pores or defects (i.e., holes, seal bypass) in the membrane
and forward flow tests can provide a quantitative measurement of flow that demonstrates the absence of excessively large pores
Retention of bacteria through microporous membranes, however, is not solely a function of size exclusion by cylindrical pores
smaller than the incident bacteria. Other properties of membranes can contribute to retention such as the shape and tortuosity
of the porous structure, thickness of the membrane (i.e., length of the flow path through the membrane pores from upstream
to downstream), and adsorptive forces, which may occur between the bacteria and the walls of any pores large enough for bacteria
insertion. Neither forward-flow nor bubble point-type tests are fully capable of detecting changes to these secondary retention
factors. Degradation of these conditions typically does not occur in a compatible validated membrane manufacturing process
and the limitation of integrity tests to detect deviations in these retention variables can be underappreciated.
Failure analysis of the re-used pleated filter cartridge ultimately identified the root cause of the filter penetration. The
damage incurred during the multiple re-use cycles was manifested by chemical degradation of the membrane, resulting in a thinning
that was localized at the pleat crests of the filter cartridge, as shown in Figure 1.
Figure 1: Diagrammatic cross-section of a pleated membrane filter showing locations of partial thickness membrane degradation
at pleat crests incurred during inadequate cleaning and resterilization. (ALL FIGURES ARE COURTESY OF THE AUTHOR.)
Localized chemical damage and thinning at pleat crests is typically indicative of partial drying of the filter whereby fluid
components capable of chemically attacking the membrane under hot steam conditions are concentrated at the pleat crests during
evaporation from those points. In this case, the partial degradation of the membrane's thickness was attributed to exposure
to hot caustic during the SIP resterilization phase of the re-use cycle. The presence of residual caustic prior to SIP was
attributed to insufficient rinse-out of the caustic cleaning agent, whereby subsequent evaporation of water from the cartridge
before resterilization caused increased concentration of caustic at the pleat crests during the SIP resterilization phase
of the re-use cycle. The elevated temperature from the steam on the residual concentrated caustic at the pleat crests then
caused accelerated chemical degradation of the membrane face surface at the aforementioned locations.
The compromise in membrane thickness in these localized regions was sufficient to enable bacterial penetration. However, because
the damage did not go all the way through the membrane (no holes or oversized pores), and the thin areas were limited to a
very small total area at the pleat crests, neither the bubble point-type test nor the forward-flow test measurements exceeded
their pass/fail limits.
As illustrated in Figure 2, the bubble point-type test can only detect full-thickness hole defects. Forward-flow diffusion-type
tests can provide values that relate, in principle, to membrane thickness, but they were unable to detect the limited thin
areas isolated at some of the pleat crests in this case. The thin areas were not detected because they did not elevate the
forward flow in excess of the flow test limit.
Figure 2: Membrane damage during re-use may be detectable or nondetectable by forward flow (FF) or bubble point-type (BP)
filter integrity tests. (ALL FIGURES ARE COURTESY OF THE AUTHOR.)