Foreword
Sterilizing-grade membrane filters, commonly used within the biopharmaceutical industry to sterilize fluids, require validation to assure these filters function as defined1,2. During validation the chosen membrane filter commonly undergoes a product bacteria challenge test under the process conditions specified for the production process. The objective of this test is to determine whether any product or process condition influences either the organism or the filter membrane matrix to a degree that the filter does not function appropriately, resulting in failure to retain the test organism.
Usually the test microorganism, Brevundimonas diminuta, is exposed to the product fluid and is then used to challenge the filter at a level of 107 per cm2 in accordance with ASTM F838-153. Before a challenge test is performed, a viability test is required to verify that the test microorganism survives the challenge fluid and/or process parameters. If this is not the case, an alternative challenge method can be used in as described in PDA Technical Report 262. Another route would be to determine whether or not the process stream shows a native bioburden surviving the fluid and process conditions. This indigenous bioburden would be used to challenge the filter at the highest achievable bioburden level possible.
PDA Technical Report 26 (1998) states in Section 6.2 “Filter membrane lots used for bacterial retention validation should have a pre-filtration, water wet, physical integrity test value at or near the filter manufacturer’s production limit.” This statement does not specify what the term “near” means. During a presentation at the PDA/FDA Conference in 1998, the question to define “near” was raised. The answer given by a panelist was 10%, however the answer was not supported by any test data and was simply a suggestion.
The 10% value was not a regulatory issue, presumably since there were no supporting test data available and retrospective validation work did not show the need for such a specification. However, multiple filter end-users have been asked by regulatory reviewers why the 10% specification has not been considered when conducting bacteria challenge validation studies and some reviewers have, erroneously, started enforcing the 10% limit suggestion.
Bubble Point Limit Definitions
Most product bacteria challenge tests are performed for end-users by validation laboratories, commonly as part of a filter manufacturer’s services. Service laboratory staff are very experienced in conducting bacteria challenge studies and these laboratories are well equipped for conducting small-scale simulations necessary to mimic productionscale processing conditions.
To be able to fulfill the statement “at or near” the filter manufacturer’s production limit, the practice has been to try to manufacture “validation membrane samples” which may have a bubble point within 10% of the limit. To achieve this goal, validation test membranes may or may not be produced as part of commercial production of the same membrane. In some cases, manufacturing such “low bubble point” membranes may require deviation from the specifications of commercial membranes.
Commercially available sterilizing-grade filter membranes are deliberately produced within well-defined specification settings and frequently have bubble point values higher than the membranes required for validation purposes, i.e., validation test membranes that may be within in the 10% tolerance level.
Figure 1 represents a typical non-destructive bubble point integrity test correlation to bacteria challenge tests according to the ASTM F838-15 standard methodology and illustrates the resulting limits for filter manufacturing and end-user product release.
Figure 1. Example depiction of the bubble point values evaluated and usedIn this example, hundreds of membrane filters are integrity tested and then subjected to the bacteria challenge test according to the ASTM F838-15 standard. The results of these multiple microbial retention tests, compared with their respective integrity test values are used to establish a “microbial breakthrough point” (red line). This point represents the lowest possible bubble point value which produced a sterile filtrate using B. diminuta under ASTM F838-15 standard conditions. Below this bubble point value the filtrate most commonly is nonsterile. Filter manufacturers do not recommend the use of this bubble point as the integrity test limit for the enduser. The end-user’s minimum allowable bubble point is set by the filter manufacturer with an appropriate safety margin (first safety margin) included (green line).
Since the manufacturers do not want to supply end-users with filters with integrity test values at the limit line, the filter manufacturer may establish an internal production release value (blue line). Any filter with a bubble point below this production release limit will be discarded by the manufacturer and is not shipped to any end-user. The enduser therefore will most commonly measure bubble point values above the production release criteria, i.e., within the light blue box range. However, steam cycles or procedure can, for example, result in bubble point values which are below the internal production release criteria (blue line), but still above the minimum allowable bubble point used by the end-user and qualified by the filter manufacturer (green line).
Bubble point limits vary among polymeric membrane materials and structures (symmetric versus asymmetric membranes). Therefore, every filter manufacturer establishes the minimum allowable bubble point for each manufactured membrane filter type. One will commonly not find the same bubble point value for filters from different manufacturers, even when the membrane polymer is the same. This does not mean, however, that one filter performs better than another.
Figure 1 shows that a 10% above the minimum allowable bubble point specification often does not represent the manufacturer’s production release criteria. Thus, it is difficult to manufacture membrane filters for validation purposes (within the 10% bracket) from routine filter production. Such filter membranes have to be either specially produced or might be recoverable from the membrane casting, for example from the start and end of the casting process. Neither represents the true production setting. In other instances, for example PTFE, which is produced using a stretching membrane production process, it is very difficult to produce membrane within the 10% specification, it would be a special cast possibly not representing commercially available lots.
Membrane Polymers and Manufacturing Processes
There are many types of sterilizing-grade filter membrane polymers, all of which have different properties, serve different applications and are produced in different ways. The table below lists common membrane filter cartridge polymers, properties and production process:
The production process determines the ease of creating a membrane sample which will be at or near the manufacturer’s bubble point. As previously stated, in the case of PTFE the stretching process makes it highly challenging; such samples cannot be obtained without changing the production process and therefore altering the membrane properties to a degree which would not be anything representative of the normal process requirements.
Membranes which are produced using a quenching or precipitation process often experience a similar fate and it is trying to obtain from a routine production process membrane samples which would represent membranes within the 10% limit. In this case, these lower bubble point membranes are only possible when small-scale casting machines or modified casting processes are used, which again does not commonly represent a fully specified production process. More typically, manufacturers that require membrane samples for validation work obtain such samples from the casting start or end (those segments are commonly discarded). For instance, lower bubble point membrane samples are created when the machines start running, i.e., the casting drums slowly start moving. The evaporation process also experiences a start-up and end period which could create membranes which might be used for validation purposes. Actually this casting process may present the best chance to obtain validation membrane samples. Thus, membranes which are at or near the manufacturer’s bubble point often are not regarded as representative of a full scale, well defined production lot, but may be the best choice for use in validation membrane studies.
Membrane Polymers Use
The physico-chemical property differences of the membrane polymers frequently dictate the specific applications of that membrane, which may obviate the need for a 10% specification. PTFE membranes are most often used as air filters, which are not subjected to productspecific bacteria challenge testing. In few instances PTFE membranes are used for corrosive and aggressive fluids, such as solvents, which are generally bactericidal. This means the only viable bioburden which will be found in such fluids are spores. Polyamide membranes are known to be used for solvent filtration; again fluids which generally only accommodate spores. A worst case spore could be Bacillus subtilis which has an approximate length of 0.87 – 1.4 micron and width of 0.41 - 0.58 micron; larger than B. diminuta, the typical challenge organism, with a length of 0.6 – 1 micron and width of 0.3 – 0.4 micron. This means that there is not necessarily a need to check whether an organism might penetrate when the minimum bubble point is reached; these organisms would be retained even when the bubble point is beneath the microbial breakthrough point or by a 0.45 μm rated filter, due to their large size and rigid structure.
PES tolerates a large pH range, and these filters are commonly used to filter highly basic or acidic solutions. Again, such aggressive pH values will probably contain only spores. Such high and low pH applications, though, are not the standard application for these filters. They are commonly used at neutral pH levels, which allow microbial proliferation. Filter materials like PES, PVDF and cellulose acetate are used in multiple applications to retain viable organisms, which might be affected by the fluid stream. In these instances one should understand whether Brevundimonas diminuta or any native bioburden would penetrate the membrane when the bubble point test limit is reached. In these instances, it would be advisable to obtain validation membranes which come close to the bubble point limits to verify the performance of such filter under circumstances which could influence microbial retention. It has been known for years that some fluid properties, for example ionic strength or osmolality, can cause shrinkage of organisms. Depending on the processing time such shrinkage could cause penetration of 0.2 μm-rated membrane filters. In these instances other, tighter rated filters, e.g., 0.1 μm-rated, should be considered.
Scientific 10% Tolerance Validity
Since there are no specific data supporting the “at or near 10%” statement, what supports the validity and enforcement of such suggestion? The reason to perform a bacteria challenge test on a filter membrane is to determine whether or not the challenge organism will penetrate the membrane having a specific bubble point. Therefore if the manufacturer’s minimum allowable bubble point is at 50 psi, the statement is included in PDA Technical Report No. 26 with the goal of determining whether or not the filter is retentive at 50 psi independently of whether an organism or filter matrix or filter retention has been influenced by the product or process. If one now allows a 10% tolerance, the filter can have a 55 psi bubble point and the end-user still can use the 50 psi minimum allowable bubble point given by the filter manufacturer. If one wants to play devil’s advocate, one could ask what happens within the 5 psi difference. Do we know whether or not the use of a membrane with this bubble point would lead to penetration?
Another question would be, how much does the filter pore matrix changes within 5 psi? The simple answer is that nobody knows, although some answers have to be obtained empirically, i.e., by process validation. The reality is that there are filter membrane production processes which do not allow changes to the process to create such borderline filter membranes, and that these specially manufactured membranes may even be outside the 10% tolerance. Following the process validation, if these membranes are now used in production processes, penetration problems may not even occur. This may be so due to the fluid these filters have been used in or this may be so because the 10% rule does not have a basis. The answer probably lies within a risk assessment basis, which should be used to determine whether or not the fluid to be filtered could influence the organism or filter and for what purpose the filter is used.
Risk Assessment Proposal
Within filtration applications, risks range from fluid properties to the filtration stage in which the filter is used. For example, some fluid properties either inhibit organism growth or are bactericidal. Some fluids contain enough nutrients to allow the organisms to become larger resulting in a low probability that the organism will penetrate a sterilizing-grade filter membrane. Other fluids, though, have properties which could potentially cause an organism alteration, for example shrinkage.
The other risk criterion is the position and function of the filter. A final filter before the filling line would be a most critical filter. Such filter is the final barrier for any potential organism within the drug product. The further up-stream the filter is the lower the risk, as such upstream filters are commonly followed by many separation or purification steps. For example, a media filter’s criticality is mainly measured by an economical value of a cell culture batch potentially contaminated, when the filter in question is not functioning properly. However, such filter will not have the same influence on the quality of the final drug product filled as the filter at the pointof- use.
The risk criteria fluid or application and filter position have to be taken into consideration when one judges a filtration step. The table above) aims to create an overview from an application basis. It does not claim to be complete.
As mentioned, the table only addresses the possible application or fluid stream, not the position. For this reason one could say that a buffer or high-purity water filter could potentially have a higher risk than a media filter. Nevertheless, both fluids are often upor mid-stream fluids, i.e., further sterilizing steps follow the use of these fluids. This does not mean that the filters used within these applications should receive special leniency, only that the criticality is somewhat lower than for a terminal sterilizing filter.
Conclusion
PDA Technical Report 26 stated that one of the membrane lots used for process validation bacteria challenge tests should be at or near the minimum allowable Bubble Point specified by the filter manufacturer. The term “near” has been described as a 10% tolerance to the minimum allowable bubble point. This tolerance should be used as a guide only.
Lately the 10% tolerance has been mistakenly enforced as a rule by regulatory reviewers resulting in multiple questions from the industry.
This paper attempts to explain the different rationales and interpretations in regard to the validity of the 10% rule. Since there is no scientific data basis for the 10% tolerance, this suggestion should not be enforced in drug manufacturing. Rather, a risk-based approach should be used to determine the product risk in regard to influences on the microorganism and/or filter matrix. If there is a potential risk that the fluid or process conditions might adversely influence the organism or filter matrix properties, a more thorough analysis must be performed to ensure the filter will perform as defined.
References
- Guidance for Industry, Sterile Drug Products Produced by Aseptic Processing – Current Good Manufacturing Practice, U.S Food and Drug Administration, 2004.
- Technical Report No. 26, Revised 2008, Sterilizing Filtration of Liquids, PDA J. Pharmaceutical Science and Technology, Vol.62, Number S5.
- F838-15 Standard Test Method for Determining Bacterial Retention of Membrane Filters Utilized for Liquid Filtration. American Society for Testing and Materials (ASTM): 2015.