PDA Letter Article

PUPSIT and the Annex 1 Revision

by Tina Morris, PDA, Maik Jornitz, G-Con, Gabriele Gori, GSK, and Hal Baseman, ValSource

Filter Photo

An Update on Scientific Efforts to Develop a Risk-Based Position

The collaboration team leadership wishes to acknowledge the input and patient consideration by members of the Annex 1 Inspectors Working Group. Their insight has been most valuable in identifying technical areas to consider and making this effort more beneficial to the industry and regulators alike.

In December 2017, PDA and the Biophorum Operations Group began a collaborative effort to address industry and regulatory concerns related to pre-use, post-sterilization integrity testing (more commonly referred to as “PUPSIT”) of sterilizing grade filters used in aseptically processed biopharmaceutical products. BioPhorum agreed to coordinate the effort, along with encouraging participation by its member companies and industry thought leaders. PDA agreed to provide scientific guidance, laboratory support, technical experts, filter manufacturers participation and a venue for public communication and publication.

The overall objective of this collaboration is to perform a series of integrated and interdependent workstreams that will provide industry and regulators alike with information that can be used to make informed decisions on how best to control and prevent sterilizing grade filter failures or improve detection of failures. It is not anticipated that any individual workstream will result in definitive positions, but rather that the output of all workstreams are needed to form a valid, scientific position.

This article presents an update on the efforts of the joint PDA and BioPhorum collaboration workstreams—masking studies, historical data mining, filter manufacturing and use risk assessments and PUPSIT risk assessment and the development of a best practice guide. Due to the interdependence of workstreams and to avoid misinterpretation of results, the collaboration team has decided to publish output from all workstreams simultaneously. The team also thought it important to provide ongoing updates to our respective members on the progress and status of these efforts.


Since 1998, the EU Guidelines to Good Manufacturing Practice: Medicinal Products for Human and Veterinary Use, Annex 1 (Manufacture of Sterile Medicinal Products) or “Annex 1” has contained the requirement for verifying the integrity of a sterilizing grade filter before use and after its sterilization. The requirement remained in the 2008 revision and in the 2017 draft revision to Annex 1. While not a requirement by the U.S. FDA, EMA inspectors and some PIC/S inspectors have been increasingly expressing expectations for companies to employ this testing procedure.

What is PUPSIT?

It is a test of the integrity of the sterilizing filter and assembly, including filter housing, support structure and connection after that assembly has been installed and sterilized, and before it is used to filter product. One example of a sterilized filter system is a gamma irradiated sterilizing grade filtration bag and tubing assembly. The integrity test is usually performed using a bubble point, diffusive flow or pressure hold test.


Concerns have been raised that a sterilizing filter could develop certain flaws that would allow microbiological contamination to pass during filtration. The key is that flaws may be blocked or clogged by fluid contaminants or components during the filtration process and remain undiscovered during post-use integrity test. This phenomenon is sometimes referred to as “filter flaw masking.” If this were to happen, then the potential outcome and effect on product quality may be catastrophic. Therefore, the risk must be closely evaluated and addressed. It is expected that a post-sterilization, preuse test would allow to detect such flaws before use of the filter and as such eliminate that risk.

Why Not Use PUPSIT?

The concern is that the contamination/product deterioration risk associated with performing PUPSIT may greatly outweigh the risk of product contamination as a result of the masking effect. To test a filter that has been sterilized by current means, the sterile filtrate side of the sterilized filter must be under atmospheric pressure, requiring a fluid pathway to remove any wetting agent. The exposure of the downstream portions of the sterile product transport line poses a risk to maintaining the sterility of the filtered product. This, along with other risks, including additional interventions in the aseptic space, increased complexity of the filtration system, execution of this additional tests, failure of PUPSIT assembly components and stress on sterilized filter to perform the test, is greater than the remote likelihood of microbiological contamination from a flaw which can be masked during use of the filter that is not detected afterwards.

What is the Risk of Filter Flaw Masking?

For masking to occur, filter flaws must be small enough not to be uncovered during initial testing, yet able to become large enough to pass microbiological contamination after sterilization, and yet again, small enough to be closed by clogging. Additionally, the product being filtered must have the ability to clog the filter.

The key arguments for not using PUPSIT are:

  1. The risks of filter flaws are generally to be considered low, due to filter manufacturing, handling and use controls in place to prevent filter flaws and failures and use of presterilization testing. While these are not universally performed at user sites, they are always performed at filter manufacturer sites and would uncover any defects that may exist.
  2. Performing PUPSIT properly is a complex process, requires additional manipulations of a sterilized system and, therefore, comes attached with an elevated risk of errors introduced during routine operations which may be difficult to detect afterwards.
  3. There is insufficient scientific evidence that the phenomenon PUPSIT is designed to address---masking of flaws does occur.

Therefore, PUPSIT should not be performed, as the contamination risk it poses to sterile product outweighs its benefit.

Counter arguments presented to the above position include:

  1. Filter manufacturing, transport and usage are insufficiently controlled, and there is a risk that filter flaws could be masked, thus, presterilization integrity testing is not universal enough to control this risk.
  2. Performance of PUPSIT should not be considered complex nor burdensome and should not increase the risk of system contamination. To date, assessments aimed at identifying PUPSIT risks have been hampered by predetermined outcomes and bias.
  3. There is a lack of scientific evidence and data to show that blinding does not occur.

Therefore, PUPSIT should be performed to mitigate even a small risk of contamination.

The risk/benefit equation for the use of PUPSIT hinges on the comparison of these two sets of risks—the risk of loss of product sterility assurance from contamination due to the masking effect or the loss of product sterility assurance from contamination due to an incorrect/sub-optimal execution of PUPSIT.

It is important to note that opinions and positions among both industry and global regulators vary, without clear consensus on either side. One thing that both sides seem to have in common, however, is the recognition that not enough objective scientific knowledge is currently available, and that additional scientific knowledge and data would be very valuable.

The objective of the PDA/BioPhorum collaboration efforts is to provide objective, unbiased, scientific data both industry and regulators can use to make informed decisions on the need for and use of further control measures such as PUPSIT.

Four important paradigms form the basis of the collaborative effort:

  1. The effort must consider the opinions and concerns of industry and regulators on both sides of the debate.
  2. There is a need for unbiased, objective, risk-based, scientific data and evidence. To that end, we believe the PDA/BioPhorum collaboration is the most comprehensive scientific effort currently underway to establish evidence of the risks of aseptic filtration flaw masking, conditions under which masking may occur and best approaches to mitigating such risks and conditions.
  3. PUPSIT is not a process or an objective. It is a test intended to reduce risk associated with aseptic processing. With that in mind, the mandate of the collaborative effort is focused on sterile filtration quality risk management (SFQRM), rather than PUPSIT. What began with a focus on PUPSIT, now has become a comprehensive effort to improve sterility assurance through the establishment of a holistic, end-to-end control strategy for sterile filtration. SFQRM activities include the assessment of potential risks to sterile filtration and the development of control strategies to prevent filter failure and, where appropriate, the use of detection controls, including, but not limited to, PUPSIT.
  4. Efforts to reduce risk that are not effective may be harmful in that they divert resources and attention from efforts that are considered more effective. Therefore, a full understanding of relative risk and of the effectiveness of associated controls is essential.
  5. Working on the prevention of a failure is more effective than working on improving the detection of a failure.

Project Status and Update

The workstream deliverables are designed to identify the filter manufacturing, filter use process conditions and product characteristics required to create the risk of masking, understand the risk of masking and prevent such risks where they occur. The masking study and historical data analysis are designed to establish the worst-case product characteristics required for masking of filter flaws to occur. The filter manufacturing and use risk assessment identify conditions that pose risk of filter flaws.

If this is accomplished, companies should be able to use these findings as guidance to determine if their processes and products combine to pose a real risk of masking, and, if so, how best to control or change their filtration processes in order to prevent the occurrence and remove the risk. If their process conditions and fluid compositions do not pose such a risk either through proper controls or lack of flawing/masking conditions—manufacturers should be able to scientifically justify the basis of that assessment.

This approach encourages application and emphasis of quality risk management (QRM) principles and practices focused on prevention of those conditions that might result in flaws and masking rather than sole (or unbalanced) reliance on detection mechanisms, including testing. Risks across the value stream, from filter manufacture through filter use, have also been included in QRM efforts.

Workstream 1 – Masking Study

Description: The masking studies are a series of experiments where nonintegral filters are challenged with a biological fluid of two foulant concentrations at various blocking rates.

The objective of the first phase of the study is to see if an extremely high foulant concentration fluid and blockage rate, well beyond commercially feasible conditions, can mask the filter flaw that the post-use integrity test does not detect it.

The second phase of the study will include a series of experiments at foulant concentrations and foulant rates closer to upper edge commercial conditions to determine which conditions favour masking and which do not pose a risk of filter masking.

In Phase 1, ten-inch filter cartridges were collected with flaws that resulted in slight integrity failures, indicating relatively small defects that might be prone to clogging. In Phase 2, 47 mm flat discs with a laser drilled 10-micron flaw will be used. Laser drilling of filer discs was necessary, because it was difficult to find an adequate supply of non-integral filters with gradual non-catastrophic flaws. The typical filter integrity flaws are catastrophic and not appropriate for the masking study. The gradual non-catastrophic flaw failures needed for the study are rare and not reusable. Therefore, the laser drilled flaws were the best option.

Status: The first part of the study has been completed successfully at PDA’s Training and Research Institute in Bethesda, Md. Two of 24 ten-inch flawed filters were blocked to a degree that the flaws were masked, and the filters passed the post-use integrity test. It is important to note that to replicate the masking effect, the concentrations of organic materials needed to be increased to a level well above what would be attempted, encountered or practical in commercial pharmaceutical manufacturing. These conditions do not represent a typical filtration process or fluid; these extreme conditions were used to determine whether masking is at all possible. The trial result created a scientific basis needed as a foundation for the second phase trial work.

Next steps include repeating experiments at conditions very similar to typical commercial usage, using commercially flawed filters and laboratory flawed, laser drilled disc filters. The goal is to collect and analyze the data, then submit the study and data in combination with other findings to the PDA Journal of Pharmaceutical Technology for peer review and publication.

Discussion: It was difficult to find flawed filters appropriate for this study due apparently to the rare occurrence of gradual or minor flaws. The flawed filters were all discovered using standard pre-sterilization integrity testing at the filter manufacturer, which proved to be valuable in uncovering filter flaws prior to sterilization.

Masking of flawed filters resulting in passed post-use integrity tests can occur under extreme, non-commercial conditions under which sterilizing grade filters are not typically subjected as these are conditions that would cause premature filter fouling and blockage, rendering the filtration process unusable.

Preliminary tests performed, using lower foulant concentration to better simulate real commercial manufacturing operations showed that the flawed filters at various blocking rates were detected as flawed by the post-use integrity test. This may indicate that masking of flawed filters under most commercial conditions would not occur.

Still, the studies must be completed, and the results evaluated in conjunction with other workstream results. In addition, the correlation between laboratory produced flaws and commercially occurring flaws needs to be better established.

Early indications directionally suggest that the study results will provide guidance to industry to determine if their filtration conditions pose risk of binding.

Workstream 2 – Data Mining

Description: The data mining effort is an extensive analysis of historical data from standard bacterial challenge tests of 0.2µ and 0.45µ filters generated for a variety of products as part of filtration process validation studies. The objective is to understand the propensity of product fluids to increase the bubble point between pre- and post-testing and establish definitions of known and potential failure modes and as well as conditions under which pre-use failed filters may be masked.

Status: Initial data mining and analysis is complete. The results of the study and data in combination with masking study data will be submitted to the PDA Journal of Pharmaceutical Technology for peer-reviewed publication.

Discussion: To date approximately 2,080 data sets have been collected and statistically analysed from a multitude of different 0.2µ and 0.45µ filters used with different fluid streams. The data mining showed that a pre- and post-use bubble point value shift is extremely rare. The entire dataset is currently undergoing additional statistical analysis, including review of the outliers found and potential reasons for them.

Workstream 3 – FTA Risk Assessments

Description: Expert-facilitated risk assessments were performed by filter manufacturers and filter users on various aspects of filter manufacturing, transport, handling and use. The tool selected was fault tree analysis (FTA), designed to identify failure modes, evaluate the effectiveness of control measures and, where needed, help identify additional control measures to reduce risk of failure.

The objective is to provide examples and guidance to help filter manufacturers and users define a robust QRM approach that can be employed to determine risks of sterile filter failure and define a control strategy to provide improve filter integrity and sterility assurance, including both preventive and detection (e.g., PUPSIT) controls based on scientific understanding and the associated risk. This work will use a systematic approach to assess the end-to-end risk to the sterile filtration manufacturing and use process in several generic process examples.

Status: The FTA have been completed by the filter manufacturer community for common cartridge filter manufacturing scenario, with an additional FTA in progress for a simple assembled system. It is anticipated that the example FTAs will be prepared by BioPhorum and PDA, and then published by PDA as part of an overall report or points to consider document in combination with other deliverables in 2019.

Discussion: With failure modes identified, the FTA can be used to help filter manufacturers and users determine, implement and evaluate controls to reduce the risk of filter flaws. Data will continue to be incorporated into the filter use FTA on an as needed basis to ensure all risks and their magnitude are adequately addressed. These data will help inform risk identification and analysis for those instances in which specific detection controls, such as PUPSIT, are necessary for inclusion in the overall filtration control strategy.

Filter manufacturing failure modes and associated control measures were identified. Sterilizing grade filters are tested prior to release by the filter supplier providing further manufacturing failure mode control. Failure modes associated with filter packaging, shipping, transportation, handling and sterilization failure modes were also addressed.

Data will help inform risk identification and analysis for those instances in which specific detection controls, such as PUPSIT, are necessary for inclusion in the overall filtration control strategy. Examples may also be used by filter manufacturers as a guide for process improvement, and by users, as a guide for audit and selection of filter manufacturers.

Workstream 4 – PUPSIT Best Practice Points to Consider

Description: The objective of the Points to Consider is to assemble industry experts with extensive PUPSIT-related experience to prepare a comprehensive best practice guide for the selection, design and use of PUPSIT under varying circumstances that incorporates information from an adjunct PUPSIT failure mode and effects analysis (FMEA) risk assessment effort and BPOG surveys to present guidance on decision making criteria, best practices and points to consider for those using, or contemplating the use of, PUPSIT. This effort will assist companies to be successful in recognizing, planning for, training on and addressing the complexity of PUPSIT.

Status: The Points to Consider and FMEA have been completed and are under review for PDA publication.

Discussion: The best practice guide identified design and use considerations that required consideration in the FMEA. And the FMEA identified areas of risk and need for controls that required consideration in the guide.

The outcome should make the readers of the guide aware of the care, control and training required to design, install and maintain a proper PUPSIT system and successfully control and perform the PUPSIT process. Design and operation must take into consideration the potential risk posed by exposure of the downstream filter assembly to the atmosphere, the additional risk of component malfunction or improper operation, the increased interventions and activities occurring in the aseptic area and stress placed on the sterilized filter during the spot sterilization testing. This effort also made clear that there are a multitude of complexities, which must be further evaluated, e.g., drying pressures and time for water wetted filters and product and filter adverse effects for product wetted filters. PUPSIT is not easy but requires in-depth process and filtration knowledge and a well- thought-out step plan to avoid product quality compromises.

Ancillary Research

As the development of this deliverable progressed, it became apparent there was a need for additional information and research into the operational effort and risk associated with PUPSIT. To that end, two surveys were proposed and are in process that will cover 1) probability or likelihood of aseptic process failure leading to loss of product sterility, and 2) efforts required to plan and implement PUPSIT (e.g., resources and investments needed).

These surveys should provide evidence of the level of complexity, effort and risk associated with the implementation of PUPSIT.


As mentioned earlier, it is important to wait for all workstream activities to be completed, results analyzed, reviewed and compared to other supporting workstream results and analysis before drawing and stating conclusions. The following, however, are fairly certain:

  1. Companies and regulators should use a nonbiased risk-based approach drawing on scientific evidence from objective data when making decisions related to the feasibility and validity of controls to prevent aseptic process failures. That effort should be commensurate with process, product and patient risk.
  2. Where little to no scientific evidence or data is available, that should be a signal to industry and regulators to obtain or develop the missing evidence. Where the evidence supports or promotes prescribed or alternative approaches that are beneficial, or even more suitable, for the process/product in scope, those approaches should be considered and adopted.
  3. Where evidence is presented, results published and positions offered, industry leaders, suppliers and regulators should welcome commentary and discussion on how to best interpret and communicate these results.
  4. Stay tuned!