Developing regulatory documents that span many countries, languages, and cultures is a formidable task, and, despite the best of intentions, often results in disagreement as to how to interpret the content. There are a number of examples from regulatory guidance documents that illustrate the challenges faced by the industry practitioners responsible for carrying out the requirements. EU GMP Annex 1, which is currently under revision, has a number of requirements that have been interpreted variously by regulators and firms—interpretations that may not have been the intent of the original authors.
For example, requirement 111 in Annex 1 states:
The final sterile filtration should be carried out as close as possible to the filling point.
This could be taken to mean that the final filter should be as physically close to the filling operation as equipment design permits, i.e., being 1m distant is preferable to being 5m distant. But how does that interpretation add any value to product quality, since the increased length of tubing post-filtration would be sterile, and contamination between the filter and the point of fill would be equally unacceptable and impactful regardless of the tubing length? Tubing, depending on materials of construction, can adsorb formulation components, but minimizing adsorption is clearly not the intention of this requirement. What is missing from the requirement is the underlying intent. Seemingly, this applies to sterility assurance, but it is unclear how it applies to this area. Perhaps the authors intended to minimize the potential for subsequent contamination after filtration by eliminating or minimizing the number of connections between final filtration and point of fill. If so, it would be helpful to practitioners and inspectors evaluating compliance if that intent was clearly stated, i.e., the line from final filter to the point of fill should be of continuous construction, where possible, in order to minimize the potential for ingress of contamination. Written justification must be available to support the adopted approach for those times when it is not possible for continuous construction.
Continuing with filtration, requirement 113 states:
The integrity of the sterilised filter should be verified before use and should be confirmed immediately after use by an appropriate method such as a bubble point, diffusive flow or pressure hold test.
In many situations, performing a filter integrity test on a sterilized filter prior to use may introduce unnecessary risk to the process due to the additional manipulations and connections required. While it has been suggested that product residue could plug a small hole in a sterilized filter, thereby, masking a nonintegral filter, no published data supports this theory. Perhaps, under unique circumstances, it could be possible (for instance, during aggregation of large protein molecules), but no evidence exists of it occurring in the vast majority of situations. Furthermore, if prefiltration is used to minimize bioburden prior to final sterile filtration, no fine particulate matter would be present to plug membrane defects. Adding liquid and gas piping, and the associated valves, vent filters and wetting fluid reservoirs to a filtration system that allows for post-sterilization integrity testing prior to use increases the potential risk of introducing contamination. It would, therefore, be beneficial to apply risk management to each situation and to eliminate this requirement where justified.
Another requirement that has led to divergent interpretations is requirement 117, which states:
Containers should be closed by appropriately validated methods. Containers closed by fusion, e.g. glass or plastic ampoules should be subject to 100% integrity testing. Samples of other containers should be checked for integrity according to appropriate procedures.
How this applies is very clear for containers closed by fusion, but rather vague for other containers. First, it is unclear whether the samples referenced are part of in-process checks or are required to be sampled and tested as part of each batch release. Second, the term “appropriate procedures” may be interpreted to include in-process checks (such as evaluation by camera) or checks of capping parameter (such as applied pressures). Requirement 117 has also been interpreted to mean a discrete container integrity test must be performed on samples of each finished batch. Since the capping equipment and process would have been validated to generate container/closure systems that meet container closure integrity (CCI) requirements, CCI testing per batch should not be required if process monitoring has ensured that validated parameters have been met continuously. Based on this rationale, the authors of Annex 1 may not have intended to require routine CCI testing, but the current wording certainly leads many to conclude such testing is required.
Another challenge facing both industry practitioners and regulators is ensuring that cGMPs are based on good science. Clearly stating the scientific principles being addressed in a document is extremely important, as demonstrated by filter-face air velocity in critical aseptic processing. The European Union specifies an air velocity range of 0.36 to 0.54 m/s, whereas the science-based requirement necessitates that air velocity be sufficient to sweep away particulate matter from the critical zone. Although the range in Annex 1 is acknowledged to be a guidance, historical practice—especially its incorporation into a central guidance document—essentially makes the given range a de facto requirement from which industry has been reluctant to stray. Industry may be criticized for accepting a stated guidance value as standard; however, experience with regulatory enforcement suggests that taking a conservative position may be necessary to avoid costly delays while changes are made to established facilities, should a company’s rationale not be accepted. It would be better to specify the purpose of measuring airflow velocities and require companies to justify their chosen ranges with sound scientific rationale.
While written with the best of intentions, there are other requirements in Annex 1 that can be interpreted in different ways. For example, requirement 18 states:
Where aseptic operations are performed monitoring should be frequent using methods such as settle plates, volumetric air and surface sampling (e.g. swabs and contact plates). Sampling methods used in operation should not interfere with zone protection.
This could be—and probably was intended to be—interpreted as meaning that the stated methods are just examples, and that other methods may be used where appropriate. Whether due to translation or simple confusion regarding intent, some interpret the requirement to mean all of the stated methods must be used. There are situations, however, in which alternative sampling methods may prove beneficial. For example, when operating a RABS filling machine, where a key focus is on minimizing or eliminating aseptic interventions (e.g., door openings), the use of liquid media jars instead of agar plates allows for long-term monitoring of the environment without concerns about media desiccation or the need for frequent media replacement. These features allow the use of liquid media jars to meet the second part of the Annex 1.18 requirement by minimizing the potential adverse impact on the operating environment. Of course, recent advances in technology allow for real-time monitoring of viable particulates, negating the need for growth media and the associated need to periodically change it. The application of risk assessment to minimize potential impact on operations is a central component of aseptic control strategy. Industry should always be permitted to implement the optimal strategy for their particular situation. This would require regulators to clearly state and emphasize the intent behind a regulation, avoiding specific approaches and leaving room for flexibility.
Periodically, firms are asked how often they sanitize or sterilize (the wording varies) the production, storage, and distribution system used for water for injection; in fact, it is occasionally recommended that they do so at least annually. This expectation, though not stated in regulatory guidance documents, is flawed for multiple reasons. One is that, while there are certain exotic microflora that can grow in aqueous environments at elevated temperatures, this only occurs in very specialized environments. These include hot geothermal springs high in sulfur that stabilizes proteins from heat denaturation. In another example, the biological indicator G. stearothermophilus can grow at elevated temperatures. But the organisms from both of these examples are unable to grow in pure water; instead, they require specialized, complex growth environments. This illustrates the danger that can arise when information is taken out of context, then extrapolated into something more complex and, ultimately, inaccurate. Industry would be better served if all involved with a firm’s water systems came together and collaborated on a science-based solution. Otherwise, hot water systems capable of withstanding sterilization temperatures and pressures, and including all associated safety features, will be designed and built even though this added complexity is unnecessary and does not contribute to product quality or patient safety; it may, in fact, adversely impact the long-term robustness of the system. Furthermore, if bioburden were able to grow in the system, annual sanitization or sterilization would not provide the desired level of control for batches produced throughout the year. Instead of mandating unnecessary design requirements, emphasis should be placed on implementing robust monitoring systems that provide data at appropriate frequencies. If the data shows the system to be under control, as would be expected for hot, circulating purified water, this supports the rationale that the system is self-sanitizing and no further action is required. As this example illustrates, it may be beneficial for regulatory guidance documents, such as Annex 1, to clearly indicate expectations in certain key areas where flawed approaches have been advocated or adopted.
Industry is greatly appreciative of the abundance of well thought-out regulatory guidance on a variety of topics. Given the complexity and criticality of sterile manufacturing, the heightened desire to ensure that future guidance documents (including revisions to existing ones) provide clear, scientific and risk-based requirements is understandable. While the above considerations could be beneficial to the development and lifecycle of all regulatory documents, an immediate opportunity to build improvements exists with the current revision to Annex 1. We are confident that the authoring body will work hard to ensure a common understanding of expectations based on sound principles and science by engaging the vast wealth of experience and expertise within the industry. If the well-worn phrase of “compliance trumps science” is rarely uttered when implementing the revised Annex I, we will all have achieved a significant milestone in our common goal of providing a sustainable drug supply to our common customers.
3 Techniques for Drafting User-Friendly Guidance Docs
Writing requirements documents for mass readership is always a challenge. A mass audience has varying levels of experience, resulting in an array of interpretations. When the topic is regulatory cGMP guidance, the resulting differences of interpretation often lead to confusion for both regulators and industry practitioners. This can add complexity to the manufacturing processes involved—something the original authors may not have intended. Of course, no one in the industry, including regulators, can be a subject matter expert in all areas. This leaves many in the industry to develop their own interpretations of regulatory requirements without the background to understand the requirements in context or the knowledge to consider alternative approaches.
We can all work toward helping develop clearer and more robust regulatory guidance documents, but first we need to analyze the challenges. Regulatory documents that are easy to read and interpret are hard to write. This is because careful thought must be applied, potential ambiguities identified and removed, sentence structure reviewed by a diverse set of reviewers, and wording revised to ensure a common understanding of the intended meaning. In addition, the requirements must be embedded in good science, which enables the key principles to be understood and applied in a variety of situations. Any examples provided should be clearly be labelled as examples and should include language stating that other approaches may be equally acceptable as long as they address the underlying principle of the requirement. This helps encourage and support adoption of new technologies as they become available after the document has been published.
As acknowledged, it is very difficult to draft regulations or guidance documents that will be universally interpreted as the authors intended. There are, however, three practices that could continuously increase universal understanding at the time the document is published, over the lifetime of the document, and as new technologies and scientific advancements are adopted:
- The document should be developed and written in a way that aids understanding, ensuring all requirements are well-defined and the underlying intent is clear and supported by scientific principles. If examples are given of ways to meet expectations, it should be made apparent these are only examples and other approaches may be acceptable. This is no easy task as it requires multiple reviews with as diverse an audience as possible. The additional time and effort required to do so should be viewed as an investment in the future for better understanding, implementation, and compliance.
- There needs to be a system to provide clarifications after publication since there may be unexpected nuances in interpretations of requirements over time. One such system might be periodic Q&A documents, such as those currently published by some regulatory agencies. Q&A documents work best when there is a defined mechanism for directly submitting questions and receiving responses in a timely manner. The intent is not to ask regulators whether a particular approach is acceptable, but rather to clarify exactly what are the intended expectations. By clarifying the latter, the former can be more easily developed.
- There should be a mechanism for discussing a topic in an informal environment with subject matter experts if an agency’s inspector(s) objects to a firm’s interpretation of a requirement. Where this is currently practiced, it tends to be a closed discussion by inspectors within the agency. By allowing a firm’s representatives to be included, and focusing discussion on the science and underlying principles, a more substantive discussion can be take place, leading to an enhanced understanding of the issue. All involved would then benefit from some thought-provoking discussion.
Note: The content of this article and the opinions contained in it are personal to the authors and do not necessarily reflect the position of the companies for which they work.
About the Authors
Geert Vandenbossche has been in the pharmaceutical/medical device business since 1994 in a variety of roles.
Marc Besson, PhD, has over 30 years of experience in the pharmaceutical industry and has worked at various director level positions within the industry.
Gabriele Gori has been in the pharmaceutical/medical device business since 1994 in a variety of roles of increasing responsibility. He is Vice President and Global Head of Audit and Quality Risk Management at GSK Vaccines.
Jette Christensen works at Novo Nordisk A/S where she has held several different positions. She currently serves as Scientific Director within Compliance and QMS in the Diabetes Finish Products section.
Gerry Morris, PhD, works within the Global Quality Systems group at Eli Lilly and Company focusing in the areas of production, validation and sterility assurance.