FIT PRIORITIZATION CASE STUDY
The FMEA approach to determine the risk level associated with the integrity testing of filters was used with the ultimate
goal of optimizing cost and resources for FITs in a high run-rate routine manufacturing scenario. The FMEA session comprised
a cross-functional team including members from manufacturing, process development, and quality departments.
Each filter used throughout the purification and buffer preparation areas (liquid and vent filters) were evaluated based on
the following criteria:
- likelihood of failure occurring
- likelihood of failure being detected
- severity of a failure.
Occurrence is based on the number of failed integrity tests historically observed (based on filters previously integrity tested).
Steaming in place (SIP) of filters increases the probability of a breach because of potential issues with the SIP process.
If SIP is conducted for filters, the occurrence scores must be higher in comparison to filters for which SIP is not done.
Installation practices also could play a role. If more than one procedure is used for installation or if the training is not
adequate, FIT failure occurrence could be higher.
The primary way to detect a nonintegral filter is through a FIT. A large gross failure may be detected by monitoring differential
pressure across the filter, but any small breach is unlikely to be detected without an actual FIT being performed on the filter.
Therefore, filters that are not tested are likely to have low detectability scores.
The filters used in the FMEA case study presented here were routinely sterilized by SIP. The filters were installed using
procedures that are very similar to each other and the personnel involved in the installation were uniformly trained. This
resulted in the same occurrence rating across all of the filters evaluated in the case study. Similarly, it was assumed that
none are currently integrity tested and there are no other means to detect a filter breach. Therefore, the detectability of
a filter failure because of integrity breach is low and the same rating was applied across all filters evaluated. In a different
scenario, the occurrence and detection scores may not be the same if differences exist in SIP practices, installation procedures,
personnel training, or if alternate means for detecting a breach in associated filters exist.
The severity score of a filter failure is based on the criticality and risk associated with the particular filtration step.
In this case study, severity of failure varied widely across the filters evaluated. Filters validated for bacterial and viral
removal scored very high on the severity scale, whereas those used as guard filters stationed between unit operations scored
lower. Because the occurrence and detection scores were the same between filters, severity became the determining factor for
differentiation in risk. In a different scenario, all three parameters may vary from filter to filter. The risk level is then
determined by the product of the individual scores. This product is referred to as a risk prioritization number (RPN).
EVALUATION OF FILTERS BASED ON SEVERITY RATING OF A FAILING EVENT
In this case study, additional rigor was placed on assigning the severity score. A rating system for severity was developed
to assess each filter based on criticality of filter performance. The criticality was dependent on two factors: (1) filter
location (or the location where the filter is being vented if the filter physically is located away from where it is vented)
and (2) the processing step where the filter is being used. For this case study, to evaluate all filters from an equal starting
point, it was assumed that no filters currently are being integrity tested.
The filter location is important because filters serve as barriers against microorganisms entering tanks. The risk of contamination
from a breached filter varies depending on the particulate counts in the room in which the equipment is vented or exposed.
Buffer and product filters are typically housed and vented in a class 100,000 or 10,000 area while vent filters may or may
not be vented to classified areas.
Virus filtration is a validated filtration process that requires integrity testing of all associated filters. A failure closer
to the final DS or drug product poses a higher risk of affecting the product compared to a failure occurring further upstream
in the purification process where there are additional filtration or removal steps.
A risk rating system (scale of 1–5) addressing the two critical factors associated with severity (filter location and filtration
processing step) can be developed as shown in the example in Table 1. A rating of five is considered highest risk, while a
rating of one is considered lowest risk.
Table 1. Examples of critical factors with associated ratings
Table 2 shows an example of how the rating system is used to assess filters used in the buffer preparation and purification
areas in a manufacturing facility.
Table 2. Filtration step severity ratings for each critical factor (overall rating = location of filter x processing step)
A confirmation of filter integrity in the form of passing a FIT result is required for any filtration process that has been
validated for either sterility or virus removal. The virus filters and drug substance final filters, both validated for retention,
received overall severity ratings of 15 and 10, respectively. These filters typically are validated for virus or microbial
retention and require post-use integrity testing as a confirmation of either viral reduction or sterility. Therefore, in this
case study, an overall severity rating of 10 was chosen to be an appropriate cut-off to determine when a FIT must be performed.
This severity rating cut-off can vary depending on the level of risk a manufacturing facility is willing to take.